diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,429970 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "eval_steps": 500,
+  "global_step": 61420,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "grad_norm": 15.371834754943848,
+      "learning_rate": 5.425935973955508e-09,
+      "loss": 1.2594,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 12.356569290161133,
+      "learning_rate": 1.0851871947911016e-08,
+      "loss": 0.921,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 7.090956687927246,
+      "learning_rate": 1.627780792186652e-08,
+      "loss": 1.058,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 7.8753862380981445,
+      "learning_rate": 2.1703743895822033e-08,
+      "loss": 1.1465,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 15.133658409118652,
+      "learning_rate": 2.7129679869777538e-08,
+      "loss": 1.296,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 9.461874961853027,
+      "learning_rate": 3.255561584373304e-08,
+      "loss": 0.9289,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 8.671860694885254,
+      "learning_rate": 3.798155181768856e-08,
+      "loss": 1.4044,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 12.060333251953125,
+      "learning_rate": 4.3407487791644066e-08,
+      "loss": 0.9927,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 10.893800735473633,
+      "learning_rate": 4.883342376559957e-08,
+      "loss": 0.6424,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 9.444788932800293,
+      "learning_rate": 5.4259359739555076e-08,
+      "loss": 0.9067,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 13.16620922088623,
+      "learning_rate": 5.968529571351058e-08,
+      "loss": 0.9862,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 10.028450012207031,
+      "learning_rate": 6.511123168746609e-08,
+      "loss": 1.0671,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 7.967885971069336,
+      "learning_rate": 7.05371676614216e-08,
+      "loss": 0.9417,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 6.34031343460083,
+      "learning_rate": 7.596310363537712e-08,
+      "loss": 0.9517,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 8.012462615966797,
+      "learning_rate": 8.138903960933262e-08,
+      "loss": 0.8472,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 7.280701637268066,
+      "learning_rate": 8.681497558328813e-08,
+      "loss": 0.8885,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 15.095734596252441,
+      "learning_rate": 9.224091155724363e-08,
+      "loss": 0.9859,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 7.94600248336792,
+      "learning_rate": 9.766684753119913e-08,
+      "loss": 0.8063,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 4.865286827087402,
+      "learning_rate": 1.0309278350515465e-07,
+      "loss": 0.6907,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 7.684237003326416,
+      "learning_rate": 1.0851871947911015e-07,
+      "loss": 1.1632,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 10.717279434204102,
+      "learning_rate": 1.1394465545306565e-07,
+      "loss": 1.1567,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 8.36603832244873,
+      "learning_rate": 1.1937059142702117e-07,
+      "loss": 0.8156,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 10.11129093170166,
+      "learning_rate": 1.2479652740097667e-07,
+      "loss": 1.2196,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 7.653027534484863,
+      "learning_rate": 1.3022246337493217e-07,
+      "loss": 0.9898,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 4.864245414733887,
+      "learning_rate": 1.356483993488877e-07,
+      "loss": 0.9251,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 12.48296070098877,
+      "learning_rate": 1.410743353228432e-07,
+      "loss": 1.2982,
+      "step": 26
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 9.0166015625,
+      "learning_rate": 1.4650027129679873e-07,
+      "loss": 1.1587,
+      "step": 27
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 10.998336791992188,
+      "learning_rate": 1.5192620727075423e-07,
+      "loss": 1.2123,
+      "step": 28
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 8.738945960998535,
+      "learning_rate": 1.5735214324470973e-07,
+      "loss": 1.3059,
+      "step": 29
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 8.197399139404297,
+      "learning_rate": 1.6277807921866523e-07,
+      "loss": 0.8028,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 13.09382152557373,
+      "learning_rate": 1.6820401519262074e-07,
+      "loss": 0.505,
+      "step": 31
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 7.789543628692627,
+      "learning_rate": 1.7362995116657626e-07,
+      "loss": 0.8178,
+      "step": 32
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 9.249221801757812,
+      "learning_rate": 1.7905588714053176e-07,
+      "loss": 1.2474,
+      "step": 33
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 10.528407096862793,
+      "learning_rate": 1.8448182311448727e-07,
+      "loss": 0.9393,
+      "step": 34
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 8.339982032775879,
+      "learning_rate": 1.8990775908844277e-07,
+      "loss": 1.0599,
+      "step": 35
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 7.184382915496826,
+      "learning_rate": 1.9533369506239827e-07,
+      "loss": 1.0046,
+      "step": 36
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 4.338878154754639,
+      "learning_rate": 2.007596310363538e-07,
+      "loss": 0.9092,
+      "step": 37
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 6.686618328094482,
+      "learning_rate": 2.061855670103093e-07,
+      "loss": 0.7125,
+      "step": 38
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 6.761378288269043,
+      "learning_rate": 2.116115029842648e-07,
+      "loss": 1.0185,
+      "step": 39
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 6.557549953460693,
+      "learning_rate": 2.170374389582203e-07,
+      "loss": 0.971,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 6.313762187957764,
+      "learning_rate": 2.224633749321758e-07,
+      "loss": 1.1721,
+      "step": 41
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 5.116060733795166,
+      "learning_rate": 2.278893109061313e-07,
+      "loss": 1.0162,
+      "step": 42
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 8.030807495117188,
+      "learning_rate": 2.3331524688008683e-07,
+      "loss": 0.7534,
+      "step": 43
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 5.137176990509033,
+      "learning_rate": 2.3874118285404234e-07,
+      "loss": 0.9844,
+      "step": 44
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 5.765946388244629,
+      "learning_rate": 2.4416711882799786e-07,
+      "loss": 1.0971,
+      "step": 45
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 6.685918807983398,
+      "learning_rate": 2.4959305480195334e-07,
+      "loss": 1.0107,
+      "step": 46
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 4.4795637130737305,
+      "learning_rate": 2.5501899077590887e-07,
+      "loss": 0.837,
+      "step": 47
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 5.672418117523193,
+      "learning_rate": 2.6044492674986434e-07,
+      "loss": 0.918,
+      "step": 48
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 5.3413238525390625,
+      "learning_rate": 2.6587086272381987e-07,
+      "loss": 0.8658,
+      "step": 49
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 11.203449249267578,
+      "learning_rate": 2.712967986977754e-07,
+      "loss": 1.1733,
+      "step": 50
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 6.669135570526123,
+      "learning_rate": 2.7672273467173087e-07,
+      "loss": 1.2795,
+      "step": 51
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 5.839574813842773,
+      "learning_rate": 2.821486706456864e-07,
+      "loss": 1.1082,
+      "step": 52
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 6.230474948883057,
+      "learning_rate": 2.8757460661964193e-07,
+      "loss": 1.1986,
+      "step": 53
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.4553885459899902,
+      "learning_rate": 2.9300054259359746e-07,
+      "loss": 0.808,
+      "step": 54
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 6.2141852378845215,
+      "learning_rate": 2.9842647856755293e-07,
+      "loss": 0.8696,
+      "step": 55
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 5.96108341217041,
+      "learning_rate": 3.0385241454150846e-07,
+      "loss": 1.1073,
+      "step": 56
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 6.334414005279541,
+      "learning_rate": 3.0927835051546394e-07,
+      "loss": 1.3079,
+      "step": 57
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.9598429203033447,
+      "learning_rate": 3.1470428648941946e-07,
+      "loss": 0.6406,
+      "step": 58
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.9963536262512207,
+      "learning_rate": 3.20130222463375e-07,
+      "loss": 0.4302,
+      "step": 59
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 5.815216541290283,
+      "learning_rate": 3.2555615843733047e-07,
+      "loss": 0.6482,
+      "step": 60
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 4.484065055847168,
+      "learning_rate": 3.30982094411286e-07,
+      "loss": 0.8467,
+      "step": 61
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 4.704000473022461,
+      "learning_rate": 3.3640803038524147e-07,
+      "loss": 0.8917,
+      "step": 62
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 4.691905498504639,
+      "learning_rate": 3.41833966359197e-07,
+      "loss": 1.0423,
+      "step": 63
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 5.5040388107299805,
+      "learning_rate": 3.472599023331525e-07,
+      "loss": 0.6352,
+      "step": 64
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 4.177995681762695,
+      "learning_rate": 3.52685838307108e-07,
+      "loss": 0.4842,
+      "step": 65
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 4.089771270751953,
+      "learning_rate": 3.5811177428106353e-07,
+      "loss": 0.7314,
+      "step": 66
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 4.073811054229736,
+      "learning_rate": 3.63537710255019e-07,
+      "loss": 0.7359,
+      "step": 67
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.90808367729187,
+      "learning_rate": 3.6896364622897453e-07,
+      "loss": 0.714,
+      "step": 68
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.487375020980835,
+      "learning_rate": 3.7438958220293006e-07,
+      "loss": 0.5542,
+      "step": 69
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 4.748147964477539,
+      "learning_rate": 3.7981551817688554e-07,
+      "loss": 1.0365,
+      "step": 70
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 4.82234001159668,
+      "learning_rate": 3.8524145415084106e-07,
+      "loss": 1.162,
+      "step": 71
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 4.148904323577881,
+      "learning_rate": 3.9066739012479654e-07,
+      "loss": 0.7869,
+      "step": 72
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.4781196117401123,
+      "learning_rate": 3.9609332609875207e-07,
+      "loss": 0.5746,
+      "step": 73
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 4.093063831329346,
+      "learning_rate": 4.015192620727076e-07,
+      "loss": 0.9664,
+      "step": 74
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 5.973780632019043,
+      "learning_rate": 4.0694519804666307e-07,
+      "loss": 0.7862,
+      "step": 75
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.7620902061462402,
+      "learning_rate": 4.123711340206186e-07,
+      "loss": 0.6891,
+      "step": 76
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 5.833592891693115,
+      "learning_rate": 4.177970699945741e-07,
+      "loss": 0.7078,
+      "step": 77
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.8975329399108887,
+      "learning_rate": 4.232230059685296e-07,
+      "loss": 0.4625,
+      "step": 78
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 12.770061492919922,
+      "learning_rate": 4.286489419424851e-07,
+      "loss": 0.8073,
+      "step": 79
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.8481180667877197,
+      "learning_rate": 4.340748779164406e-07,
+      "loss": 0.699,
+      "step": 80
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.8937346935272217,
+      "learning_rate": 4.3950081389039613e-07,
+      "loss": 0.73,
+      "step": 81
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 5.725331783294678,
+      "learning_rate": 4.449267498643516e-07,
+      "loss": 0.8829,
+      "step": 82
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 4.200120449066162,
+      "learning_rate": 4.5035268583830714e-07,
+      "loss": 0.7938,
+      "step": 83
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.511835813522339,
+      "learning_rate": 4.557786218122626e-07,
+      "loss": 0.725,
+      "step": 84
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.752042531967163,
+      "learning_rate": 4.6120455778621814e-07,
+      "loss": 0.836,
+      "step": 85
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.4256694316864014,
+      "learning_rate": 4.6663049376017367e-07,
+      "loss": 0.5914,
+      "step": 86
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 5.5191121101379395,
+      "learning_rate": 4.7205642973412914e-07,
+      "loss": 0.5677,
+      "step": 87
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.767040729522705,
+      "learning_rate": 4.774823657080847e-07,
+      "loss": 0.6785,
+      "step": 88
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 4.508133411407471,
+      "learning_rate": 4.829083016820401e-07,
+      "loss": 0.8336,
+      "step": 89
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.7141010761260986,
+      "learning_rate": 4.883342376559957e-07,
+      "loss": 0.6149,
+      "step": 90
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.0337352752685547,
+      "learning_rate": 4.937601736299512e-07,
+      "loss": 0.8425,
+      "step": 91
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 5.225765228271484,
+      "learning_rate": 4.991861096039067e-07,
+      "loss": 0.6158,
+      "step": 92
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 4.366479873657227,
+      "learning_rate": 5.046120455778622e-07,
+      "loss": 0.9179,
+      "step": 93
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 4.034251689910889,
+      "learning_rate": 5.100379815518177e-07,
+      "loss": 0.8247,
+      "step": 94
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 4.242748737335205,
+      "learning_rate": 5.154639175257732e-07,
+      "loss": 0.7987,
+      "step": 95
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.093703508377075,
+      "learning_rate": 5.208898534997287e-07,
+      "loss": 0.6058,
+      "step": 96
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.3239848613739014,
+      "learning_rate": 5.263157894736843e-07,
+      "loss": 0.592,
+      "step": 97
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 4.976627826690674,
+      "learning_rate": 5.317417254476397e-07,
+      "loss": 0.8057,
+      "step": 98
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.923651933670044,
+      "learning_rate": 5.371676614215952e-07,
+      "loss": 0.4013,
+      "step": 99
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.4130847454071045,
+      "learning_rate": 5.425935973955508e-07,
+      "loss": 0.6575,
+      "step": 100
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 5.592776298522949,
+      "learning_rate": 5.480195333695063e-07,
+      "loss": 0.7694,
+      "step": 101
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.6353678703308105,
+      "learning_rate": 5.534454693434617e-07,
+      "loss": 0.5989,
+      "step": 102
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.547201633453369,
+      "learning_rate": 5.588714053174173e-07,
+      "loss": 0.7909,
+      "step": 103
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 4.21688175201416,
+      "learning_rate": 5.642973412913728e-07,
+      "loss": 0.5366,
+      "step": 104
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 6.484134197235107,
+      "learning_rate": 5.697232772653284e-07,
+      "loss": 0.8336,
+      "step": 105
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 8.352890014648438,
+      "learning_rate": 5.751492132392839e-07,
+      "loss": 0.8611,
+      "step": 106
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 4.994505405426025,
+      "learning_rate": 5.805751492132393e-07,
+      "loss": 0.66,
+      "step": 107
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 4.283807277679443,
+      "learning_rate": 5.860010851871949e-07,
+      "loss": 0.8314,
+      "step": 108
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 4.079714775085449,
+      "learning_rate": 5.914270211611504e-07,
+      "loss": 0.7093,
+      "step": 109
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.160902500152588,
+      "learning_rate": 5.968529571351059e-07,
+      "loss": 0.4582,
+      "step": 110
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.681600332260132,
+      "learning_rate": 6.022788931090613e-07,
+      "loss": 0.7353,
+      "step": 111
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.037034273147583,
+      "learning_rate": 6.077048290830169e-07,
+      "loss": 0.578,
+      "step": 112
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 8.535654067993164,
+      "learning_rate": 6.131307650569724e-07,
+      "loss": 0.8646,
+      "step": 113
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.4741251468658447,
+      "learning_rate": 6.185567010309279e-07,
+      "loss": 0.6526,
+      "step": 114
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 6.1712517738342285,
+      "learning_rate": 6.239826370048835e-07,
+      "loss": 1.041,
+      "step": 115
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.368293285369873,
+      "learning_rate": 6.294085729788389e-07,
+      "loss": 0.5776,
+      "step": 116
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.548980474472046,
+      "learning_rate": 6.348345089527944e-07,
+      "loss": 0.4786,
+      "step": 117
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 7.603933811187744,
+      "learning_rate": 6.4026044492675e-07,
+      "loss": 0.8517,
+      "step": 118
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.7347373962402344,
+      "learning_rate": 6.456863809007054e-07,
+      "loss": 0.7802,
+      "step": 119
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.7638661861419678,
+      "learning_rate": 6.511123168746609e-07,
+      "loss": 0.9552,
+      "step": 120
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 4.985202312469482,
+      "learning_rate": 6.565382528486164e-07,
+      "loss": 0.3918,
+      "step": 121
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 4.231565475463867,
+      "learning_rate": 6.61964188822572e-07,
+      "loss": 1.04,
+      "step": 122
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.8643250465393066,
+      "learning_rate": 6.673901247965274e-07,
+      "loss": 0.8315,
+      "step": 123
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 4.459576606750488,
+      "learning_rate": 6.728160607704829e-07,
+      "loss": 0.4271,
+      "step": 124
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 4.693436145782471,
+      "learning_rate": 6.782419967444384e-07,
+      "loss": 0.9151,
+      "step": 125
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 5.344035625457764,
+      "learning_rate": 6.83667932718394e-07,
+      "loss": 0.7512,
+      "step": 126
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.3537466526031494,
+      "learning_rate": 6.890938686923495e-07,
+      "loss": 0.8622,
+      "step": 127
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.3347628116607666,
+      "learning_rate": 6.94519804666305e-07,
+      "loss": 0.8757,
+      "step": 128
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 4.35154914855957,
+      "learning_rate": 6.999457406402604e-07,
+      "loss": 0.8706,
+      "step": 129
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 11.448994636535645,
+      "learning_rate": 7.05371676614216e-07,
+      "loss": 0.8356,
+      "step": 130
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.912825345993042,
+      "learning_rate": 7.107976125881715e-07,
+      "loss": 0.5707,
+      "step": 131
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.0762598514556885,
+      "learning_rate": 7.162235485621271e-07,
+      "loss": 0.669,
+      "step": 132
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 8.724101066589355,
+      "learning_rate": 7.216494845360824e-07,
+      "loss": 0.8553,
+      "step": 133
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.1038482189178467,
+      "learning_rate": 7.27075420510038e-07,
+      "loss": 0.3557,
+      "step": 134
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.787149429321289,
+      "learning_rate": 7.325013564839935e-07,
+      "loss": 0.6782,
+      "step": 135
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 5.186440467834473,
+      "learning_rate": 7.379272924579491e-07,
+      "loss": 0.5457,
+      "step": 136
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.4921646118164062,
+      "learning_rate": 7.433532284319046e-07,
+      "loss": 0.7378,
+      "step": 137
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.444856882095337,
+      "learning_rate": 7.487791644058601e-07,
+      "loss": 0.8868,
+      "step": 138
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.472759485244751,
+      "learning_rate": 7.542051003798156e-07,
+      "loss": 0.8469,
+      "step": 139
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.219094753265381,
+      "learning_rate": 7.596310363537711e-07,
+      "loss": 0.8289,
+      "step": 140
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.351083517074585,
+      "learning_rate": 7.650569723277267e-07,
+      "loss": 0.6347,
+      "step": 141
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.5376930236816406,
+      "learning_rate": 7.704829083016821e-07,
+      "loss": 0.6883,
+      "step": 142
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.9480340480804443,
+      "learning_rate": 7.759088442756377e-07,
+      "loss": 0.6235,
+      "step": 143
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.9969840049743652,
+      "learning_rate": 7.813347802495931e-07,
+      "loss": 0.5281,
+      "step": 144
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 4.158041954040527,
+      "learning_rate": 7.867607162235487e-07,
+      "loss": 0.6349,
+      "step": 145
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.960346221923828,
+      "learning_rate": 7.921866521975041e-07,
+      "loss": 0.7027,
+      "step": 146
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 7.530239105224609,
+      "learning_rate": 7.976125881714597e-07,
+      "loss": 0.6982,
+      "step": 147
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 4.1998419761657715,
+      "learning_rate": 8.030385241454152e-07,
+      "loss": 0.6756,
+      "step": 148
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 4.270851135253906,
+      "learning_rate": 8.084644601193707e-07,
+      "loss": 0.778,
+      "step": 149
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.1087286472320557,
+      "learning_rate": 8.138903960933261e-07,
+      "loss": 0.6865,
+      "step": 150
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 4.13147497177124,
+      "learning_rate": 8.193163320672817e-07,
+      "loss": 0.735,
+      "step": 151
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.015498399734497,
+      "learning_rate": 8.247422680412372e-07,
+      "loss": 0.5452,
+      "step": 152
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 6.516010284423828,
+      "learning_rate": 8.301682040151928e-07,
+      "loss": 0.9347,
+      "step": 153
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.852725028991699,
+      "learning_rate": 8.355941399891481e-07,
+      "loss": 0.7248,
+      "step": 154
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.230384588241577,
+      "learning_rate": 8.410200759631037e-07,
+      "loss": 0.5606,
+      "step": 155
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 4.222452640533447,
+      "learning_rate": 8.464460119370592e-07,
+      "loss": 0.6434,
+      "step": 156
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.617326021194458,
+      "learning_rate": 8.518719479110148e-07,
+      "loss": 0.4964,
+      "step": 157
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.123520851135254,
+      "learning_rate": 8.572978838849702e-07,
+      "loss": 0.9207,
+      "step": 158
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.0824763774871826,
+      "learning_rate": 8.627238198589257e-07,
+      "loss": 0.7462,
+      "step": 159
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.376128911972046,
+      "learning_rate": 8.681497558328812e-07,
+      "loss": 0.7398,
+      "step": 160
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.9979610443115234,
+      "learning_rate": 8.735756918068368e-07,
+      "loss": 0.771,
+      "step": 161
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.035191059112549,
+      "learning_rate": 8.790016277807923e-07,
+      "loss": 1.1388,
+      "step": 162
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.5608012676239014,
+      "learning_rate": 8.844275637547478e-07,
+      "loss": 0.7282,
+      "step": 163
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.1810874938964844,
+      "learning_rate": 8.898534997287032e-07,
+      "loss": 0.6564,
+      "step": 164
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 4.367995262145996,
+      "learning_rate": 8.952794357026588e-07,
+      "loss": 0.8969,
+      "step": 165
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 9.94670295715332,
+      "learning_rate": 9.007053716766143e-07,
+      "loss": 0.5883,
+      "step": 166
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.6719841957092285,
+      "learning_rate": 9.061313076505699e-07,
+      "loss": 0.6783,
+      "step": 167
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 5.901128768920898,
+      "learning_rate": 9.115572436245252e-07,
+      "loss": 0.7748,
+      "step": 168
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 6.112943172454834,
+      "learning_rate": 9.169831795984808e-07,
+      "loss": 0.8025,
+      "step": 169
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.561649799346924,
+      "learning_rate": 9.224091155724363e-07,
+      "loss": 0.7458,
+      "step": 170
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 4.796851634979248,
+      "learning_rate": 9.278350515463919e-07,
+      "loss": 0.77,
+      "step": 171
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 4.21577262878418,
+      "learning_rate": 9.332609875203473e-07,
+      "loss": 0.5877,
+      "step": 172
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 4.161910057067871,
+      "learning_rate": 9.386869234943029e-07,
+      "loss": 0.7476,
+      "step": 173
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.7636566162109375,
+      "learning_rate": 9.441128594682583e-07,
+      "loss": 0.7065,
+      "step": 174
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.977440357208252,
+      "learning_rate": 9.495387954422139e-07,
+      "loss": 0.6965,
+      "step": 175
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 6.042709827423096,
+      "learning_rate": 9.549647314161693e-07,
+      "loss": 0.9024,
+      "step": 176
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 5.176137447357178,
+      "learning_rate": 9.603906673901248e-07,
+      "loss": 0.7265,
+      "step": 177
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 7.255624771118164,
+      "learning_rate": 9.658166033640803e-07,
+      "loss": 0.5028,
+      "step": 178
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.463024139404297,
+      "learning_rate": 9.71242539338036e-07,
+      "loss": 0.8666,
+      "step": 179
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.0147836208343506,
+      "learning_rate": 9.766684753119915e-07,
+      "loss": 0.6129,
+      "step": 180
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 6.778669834136963,
+      "learning_rate": 9.82094411285947e-07,
+      "loss": 0.671,
+      "step": 181
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 4.820185661315918,
+      "learning_rate": 9.875203472599024e-07,
+      "loss": 0.4357,
+      "step": 182
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.986441135406494,
+      "learning_rate": 9.929462832338579e-07,
+      "loss": 0.9302,
+      "step": 183
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.545694351196289,
+      "learning_rate": 9.983722192078134e-07,
+      "loss": 0.926,
+      "step": 184
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.3778269290924072,
+      "learning_rate": 1.003798155181769e-06,
+      "loss": 0.7711,
+      "step": 185
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.9482665061950684,
+      "learning_rate": 1.0092240911557243e-06,
+      "loss": 0.6794,
+      "step": 186
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 4.666112899780273,
+      "learning_rate": 1.01465002712968e-06,
+      "loss": 0.7532,
+      "step": 187
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.1291956901550293,
+      "learning_rate": 1.0200759631036355e-06,
+      "loss": 0.3718,
+      "step": 188
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.556145668029785,
+      "learning_rate": 1.025501899077591e-06,
+      "loss": 0.6618,
+      "step": 189
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 4.946144104003906,
+      "learning_rate": 1.0309278350515464e-06,
+      "loss": 0.5772,
+      "step": 190
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 4.959903717041016,
+      "learning_rate": 1.0363537710255019e-06,
+      "loss": 0.6873,
+      "step": 191
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.858750581741333,
+      "learning_rate": 1.0417797069994574e-06,
+      "loss": 0.6438,
+      "step": 192
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.6952288150787354,
+      "learning_rate": 1.047205642973413e-06,
+      "loss": 0.7755,
+      "step": 193
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 4.808685302734375,
+      "learning_rate": 1.0526315789473685e-06,
+      "loss": 0.3638,
+      "step": 194
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.56955623626709,
+      "learning_rate": 1.058057514921324e-06,
+      "loss": 0.6696,
+      "step": 195
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.6701879501342773,
+      "learning_rate": 1.0634834508952795e-06,
+      "loss": 0.5953,
+      "step": 196
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.897343873977661,
+      "learning_rate": 1.068909386869235e-06,
+      "loss": 0.7039,
+      "step": 197
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.5778825283050537,
+      "learning_rate": 1.0743353228431904e-06,
+      "loss": 0.5143,
+      "step": 198
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.482462167739868,
+      "learning_rate": 1.0797612588171461e-06,
+      "loss": 0.8976,
+      "step": 199
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.5241072177886963,
+      "learning_rate": 1.0851871947911016e-06,
+      "loss": 0.4942,
+      "step": 200
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 4.40706729888916,
+      "learning_rate": 1.090613130765057e-06,
+      "loss": 0.8145,
+      "step": 201
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.9230847358703613,
+      "learning_rate": 1.0960390667390125e-06,
+      "loss": 0.7329,
+      "step": 202
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.135878801345825,
+      "learning_rate": 1.101465002712968e-06,
+      "loss": 0.625,
+      "step": 203
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.6452465057373047,
+      "learning_rate": 1.1068909386869235e-06,
+      "loss": 0.4136,
+      "step": 204
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 4.363352298736572,
+      "learning_rate": 1.1123168746608792e-06,
+      "loss": 0.6262,
+      "step": 205
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.56931734085083,
+      "learning_rate": 1.1177428106348347e-06,
+      "loss": 0.7115,
+      "step": 206
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 5.121286869049072,
+      "learning_rate": 1.1231687466087901e-06,
+      "loss": 0.7045,
+      "step": 207
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 4.835542678833008,
+      "learning_rate": 1.1285946825827456e-06,
+      "loss": 0.5639,
+      "step": 208
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 9.164677619934082,
+      "learning_rate": 1.134020618556701e-06,
+      "loss": 0.7106,
+      "step": 209
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 5.755659103393555,
+      "learning_rate": 1.1394465545306568e-06,
+      "loss": 0.6575,
+      "step": 210
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.0523009300231934,
+      "learning_rate": 1.144872490504612e-06,
+      "loss": 0.7643,
+      "step": 211
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.276766538619995,
+      "learning_rate": 1.1502984264785677e-06,
+      "loss": 0.6878,
+      "step": 212
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.301572799682617,
+      "learning_rate": 1.1557243624525232e-06,
+      "loss": 0.7783,
+      "step": 213
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.0199127197265625,
+      "learning_rate": 1.1611502984264787e-06,
+      "loss": 0.6402,
+      "step": 214
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 4.792694568634033,
+      "learning_rate": 1.1665762344004341e-06,
+      "loss": 0.5302,
+      "step": 215
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 12.404638290405273,
+      "learning_rate": 1.1720021703743898e-06,
+      "loss": 0.5868,
+      "step": 216
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.1622884273529053,
+      "learning_rate": 1.177428106348345e-06,
+      "loss": 0.7402,
+      "step": 217
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.898940086364746,
+      "learning_rate": 1.1828540423223008e-06,
+      "loss": 0.4536,
+      "step": 218
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 4.573077201843262,
+      "learning_rate": 1.1882799782962563e-06,
+      "loss": 0.4301,
+      "step": 219
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.4816877841949463,
+      "learning_rate": 1.1937059142702117e-06,
+      "loss": 0.5395,
+      "step": 220
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.22835636138916,
+      "learning_rate": 1.1991318502441672e-06,
+      "loss": 0.5924,
+      "step": 221
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 4.842958450317383,
+      "learning_rate": 1.2045577862181227e-06,
+      "loss": 0.7036,
+      "step": 222
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.0279359817504883,
+      "learning_rate": 1.2099837221920782e-06,
+      "loss": 0.7856,
+      "step": 223
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 7.356193542480469,
+      "learning_rate": 1.2154096581660338e-06,
+      "loss": 0.7839,
+      "step": 224
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 8.856727600097656,
+      "learning_rate": 1.2208355941399893e-06,
+      "loss": 0.6431,
+      "step": 225
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.4257330894470215,
+      "learning_rate": 1.2262615301139448e-06,
+      "loss": 0.4951,
+      "step": 226
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.518494129180908,
+      "learning_rate": 1.2316874660879003e-06,
+      "loss": 0.702,
+      "step": 227
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.1207435131073,
+      "learning_rate": 1.2371134020618557e-06,
+      "loss": 0.5134,
+      "step": 228
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.102266311645508,
+      "learning_rate": 1.2425393380358112e-06,
+      "loss": 0.8039,
+      "step": 229
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.8454771041870117,
+      "learning_rate": 1.247965274009767e-06,
+      "loss": 0.7231,
+      "step": 230
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.594512939453125,
+      "learning_rate": 1.2533912099837222e-06,
+      "loss": 0.8301,
+      "step": 231
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.8884832859039307,
+      "learning_rate": 1.2588171459576779e-06,
+      "loss": 0.3032,
+      "step": 232
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 4.236685276031494,
+      "learning_rate": 1.2642430819316333e-06,
+      "loss": 0.598,
+      "step": 233
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.707441806793213,
+      "learning_rate": 1.2696690179055888e-06,
+      "loss": 0.6128,
+      "step": 234
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.8691060543060303,
+      "learning_rate": 1.2750949538795443e-06,
+      "loss": 0.3471,
+      "step": 235
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 5.204645156860352,
+      "learning_rate": 1.2805208898535e-06,
+      "loss": 0.6032,
+      "step": 236
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.333425283432007,
+      "learning_rate": 1.2859468258274554e-06,
+      "loss": 0.647,
+      "step": 237
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 5.106776237487793,
+      "learning_rate": 1.2913727618014107e-06,
+      "loss": 0.6087,
+      "step": 238
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 5.070241928100586,
+      "learning_rate": 1.2967986977753664e-06,
+      "loss": 0.6607,
+      "step": 239
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.069248914718628,
+      "learning_rate": 1.3022246337493219e-06,
+      "loss": 0.8316,
+      "step": 240
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.195213794708252,
+      "learning_rate": 1.3076505697232776e-06,
+      "loss": 0.5055,
+      "step": 241
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.9183337688446045,
+      "learning_rate": 1.3130765056972328e-06,
+      "loss": 0.5378,
+      "step": 242
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.799111843109131,
+      "learning_rate": 1.3185024416711883e-06,
+      "loss": 0.6203,
+      "step": 243
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 18.015649795532227,
+      "learning_rate": 1.323928377645144e-06,
+      "loss": 0.5886,
+      "step": 244
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.061116933822632,
+      "learning_rate": 1.3293543136190995e-06,
+      "loss": 0.7877,
+      "step": 245
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.0735466480255127,
+      "learning_rate": 1.3347802495930547e-06,
+      "loss": 0.952,
+      "step": 246
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.26529860496521,
+      "learning_rate": 1.3402061855670104e-06,
+      "loss": 0.5465,
+      "step": 247
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.0604729652404785,
+      "learning_rate": 1.3456321215409659e-06,
+      "loss": 0.7526,
+      "step": 248
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.5846195220947266,
+      "learning_rate": 1.3510580575149216e-06,
+      "loss": 0.6271,
+      "step": 249
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 4.01673698425293,
+      "learning_rate": 1.3564839934888768e-06,
+      "loss": 0.6476,
+      "step": 250
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.9814507961273193,
+      "learning_rate": 1.3619099294628323e-06,
+      "loss": 0.8065,
+      "step": 251
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.107943296432495,
+      "learning_rate": 1.367335865436788e-06,
+      "loss": 0.7342,
+      "step": 252
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.112107992172241,
+      "learning_rate": 1.3727618014107435e-06,
+      "loss": 0.5157,
+      "step": 253
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 4.033182144165039,
+      "learning_rate": 1.378187737384699e-06,
+      "loss": 0.6917,
+      "step": 254
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.69408917427063,
+      "learning_rate": 1.3836136733586544e-06,
+      "loss": 0.8495,
+      "step": 255
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.867784261703491,
+      "learning_rate": 1.38903960933261e-06,
+      "loss": 0.7758,
+      "step": 256
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 4.176469326019287,
+      "learning_rate": 1.3944655453065656e-06,
+      "loss": 0.5771,
+      "step": 257
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 22.752195358276367,
+      "learning_rate": 1.3998914812805208e-06,
+      "loss": 0.5486,
+      "step": 258
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.8036952018737793,
+      "learning_rate": 1.4053174172544765e-06,
+      "loss": 0.6332,
+      "step": 259
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.3636116981506348,
+      "learning_rate": 1.410743353228432e-06,
+      "loss": 0.5194,
+      "step": 260
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.7531392574310303,
+      "learning_rate": 1.4161692892023877e-06,
+      "loss": 0.5995,
+      "step": 261
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.6327829360961914,
+      "learning_rate": 1.421595225176343e-06,
+      "loss": 0.592,
+      "step": 262
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 4.448137283325195,
+      "learning_rate": 1.4270211611502984e-06,
+      "loss": 0.6775,
+      "step": 263
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.274266004562378,
+      "learning_rate": 1.4324470971242541e-06,
+      "loss": 0.3491,
+      "step": 264
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 6.0098347663879395,
+      "learning_rate": 1.4378730330982096e-06,
+      "loss": 0.8819,
+      "step": 265
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.488999366760254,
+      "learning_rate": 1.4432989690721649e-06,
+      "loss": 0.6517,
+      "step": 266
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 6.709583282470703,
+      "learning_rate": 1.4487249050461205e-06,
+      "loss": 0.8956,
+      "step": 267
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.7399163246154785,
+      "learning_rate": 1.454150841020076e-06,
+      "loss": 0.6327,
+      "step": 268
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 4.65533447265625,
+      "learning_rate": 1.4595767769940317e-06,
+      "loss": 0.322,
+      "step": 269
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.7928905487060547,
+      "learning_rate": 1.465002712967987e-06,
+      "loss": 0.5108,
+      "step": 270
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 7.98970365524292,
+      "learning_rate": 1.4704286489419424e-06,
+      "loss": 0.6488,
+      "step": 271
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.7206504344940186,
+      "learning_rate": 1.4758545849158981e-06,
+      "loss": 0.8236,
+      "step": 272
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 5.408740043640137,
+      "learning_rate": 1.4812805208898536e-06,
+      "loss": 0.687,
+      "step": 273
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.4224631786346436,
+      "learning_rate": 1.4867064568638093e-06,
+      "loss": 0.7225,
+      "step": 274
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.591111660003662,
+      "learning_rate": 1.4921323928377646e-06,
+      "loss": 0.7755,
+      "step": 275
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.407456159591675,
+      "learning_rate": 1.4975583288117202e-06,
+      "loss": 0.7949,
+      "step": 276
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.3843929767608643,
+      "learning_rate": 1.5029842647856757e-06,
+      "loss": 0.5093,
+      "step": 277
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.206838846206665,
+      "learning_rate": 1.5084102007596312e-06,
+      "loss": 0.6643,
+      "step": 278
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.228017807006836,
+      "learning_rate": 1.5138361367335867e-06,
+      "loss": 0.754,
+      "step": 279
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.552446126937866,
+      "learning_rate": 1.5192620727075421e-06,
+      "loss": 0.4749,
+      "step": 280
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.6136231422424316,
+      "learning_rate": 1.5246880086814978e-06,
+      "loss": 0.6584,
+      "step": 281
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 5.786938190460205,
+      "learning_rate": 1.5301139446554533e-06,
+      "loss": 0.859,
+      "step": 282
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.442307949066162,
+      "learning_rate": 1.5355398806294086e-06,
+      "loss": 0.7121,
+      "step": 283
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.5286612510681152,
+      "learning_rate": 1.5409658166033643e-06,
+      "loss": 0.55,
+      "step": 284
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.8725640773773193,
+      "learning_rate": 1.5463917525773197e-06,
+      "loss": 0.6119,
+      "step": 285
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.4791553020477295,
+      "learning_rate": 1.5518176885512754e-06,
+      "loss": 0.6905,
+      "step": 286
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 6.832215785980225,
+      "learning_rate": 1.5572436245252307e-06,
+      "loss": 0.7107,
+      "step": 287
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 4.966464519500732,
+      "learning_rate": 1.5626695604991862e-06,
+      "loss": 0.6959,
+      "step": 288
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 7.6618523597717285,
+      "learning_rate": 1.5680954964731418e-06,
+      "loss": 0.6934,
+      "step": 289
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.3799707889556885,
+      "learning_rate": 1.5735214324470973e-06,
+      "loss": 0.5284,
+      "step": 290
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.5896246433258057,
+      "learning_rate": 1.5789473684210526e-06,
+      "loss": 0.4095,
+      "step": 291
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.724632740020752,
+      "learning_rate": 1.5843733043950083e-06,
+      "loss": 0.651,
+      "step": 292
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 4.162586688995361,
+      "learning_rate": 1.5897992403689637e-06,
+      "loss": 0.4756,
+      "step": 293
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 8.5640869140625,
+      "learning_rate": 1.5952251763429194e-06,
+      "loss": 0.618,
+      "step": 294
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.8819756507873535,
+      "learning_rate": 1.6006511123168747e-06,
+      "loss": 0.6721,
+      "step": 295
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.756126880645752,
+      "learning_rate": 1.6060770482908304e-06,
+      "loss": 0.5006,
+      "step": 296
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 4.57651948928833,
+      "learning_rate": 1.6115029842647859e-06,
+      "loss": 0.7717,
+      "step": 297
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.628199577331543,
+      "learning_rate": 1.6169289202387413e-06,
+      "loss": 0.7147,
+      "step": 298
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 4.3042497634887695,
+      "learning_rate": 1.6223548562126968e-06,
+      "loss": 0.591,
+      "step": 299
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 6.391009330749512,
+      "learning_rate": 1.6277807921866523e-06,
+      "loss": 0.7601,
+      "step": 300
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.3290622234344482,
+      "learning_rate": 1.633206728160608e-06,
+      "loss": 0.5672,
+      "step": 301
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 5.606815814971924,
+      "learning_rate": 1.6386326641345634e-06,
+      "loss": 0.6644,
+      "step": 302
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 4.325339317321777,
+      "learning_rate": 1.6440586001085187e-06,
+      "loss": 0.9948,
+      "step": 303
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.500839948654175,
+      "learning_rate": 1.6494845360824744e-06,
+      "loss": 0.4706,
+      "step": 304
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 7.549091339111328,
+      "learning_rate": 1.6549104720564299e-06,
+      "loss": 0.6543,
+      "step": 305
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 5.705379486083984,
+      "learning_rate": 1.6603364080303856e-06,
+      "loss": 0.5044,
+      "step": 306
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 4.182437896728516,
+      "learning_rate": 1.6657623440043408e-06,
+      "loss": 0.7038,
+      "step": 307
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 8.380183219909668,
+      "learning_rate": 1.6711882799782963e-06,
+      "loss": 0.549,
+      "step": 308
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.8437917232513428,
+      "learning_rate": 1.676614215952252e-06,
+      "loss": 0.3944,
+      "step": 309
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 9.131180763244629,
+      "learning_rate": 1.6820401519262075e-06,
+      "loss": 0.641,
+      "step": 310
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.959563970565796,
+      "learning_rate": 1.6874660879001627e-06,
+      "loss": 0.6897,
+      "step": 311
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.452960968017578,
+      "learning_rate": 1.6928920238741184e-06,
+      "loss": 0.6304,
+      "step": 312
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.8465678691864014,
+      "learning_rate": 1.6983179598480739e-06,
+      "loss": 0.7346,
+      "step": 313
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 7.761960983276367,
+      "learning_rate": 1.7037438958220296e-06,
+      "loss": 0.6177,
+      "step": 314
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 6.788555145263672,
+      "learning_rate": 1.7091698317959848e-06,
+      "loss": 0.5473,
+      "step": 315
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.939141273498535,
+      "learning_rate": 1.7145957677699403e-06,
+      "loss": 0.6591,
+      "step": 316
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.6228227615356445,
+      "learning_rate": 1.720021703743896e-06,
+      "loss": 0.5817,
+      "step": 317
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.7759315967559814,
+      "learning_rate": 1.7254476397178515e-06,
+      "loss": 0.5915,
+      "step": 318
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.4559326171875,
+      "learning_rate": 1.730873575691807e-06,
+      "loss": 0.6732,
+      "step": 319
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.355924367904663,
+      "learning_rate": 1.7362995116657624e-06,
+      "loss": 0.6706,
+      "step": 320
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.957627058029175,
+      "learning_rate": 1.7417254476397181e-06,
+      "loss": 0.5706,
+      "step": 321
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.7507693767547607,
+      "learning_rate": 1.7471513836136736e-06,
+      "loss": 0.5895,
+      "step": 322
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.19152307510376,
+      "learning_rate": 1.7525773195876288e-06,
+      "loss": 0.9993,
+      "step": 323
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.577357769012451,
+      "learning_rate": 1.7580032555615845e-06,
+      "loss": 0.5782,
+      "step": 324
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.262109279632568,
+      "learning_rate": 1.76342919153554e-06,
+      "loss": 0.8454,
+      "step": 325
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.579437017440796,
+      "learning_rate": 1.7688551275094957e-06,
+      "loss": 0.6651,
+      "step": 326
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.8444461822509766,
+      "learning_rate": 1.774281063483451e-06,
+      "loss": 0.5886,
+      "step": 327
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.7981789112091064,
+      "learning_rate": 1.7797069994574064e-06,
+      "loss": 0.3619,
+      "step": 328
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.3352444171905518,
+      "learning_rate": 1.7851329354313621e-06,
+      "loss": 0.5856,
+      "step": 329
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.719818115234375,
+      "learning_rate": 1.7905588714053176e-06,
+      "loss": 0.6232,
+      "step": 330
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.966998815536499,
+      "learning_rate": 1.7959848073792729e-06,
+      "loss": 0.4183,
+      "step": 331
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.2060697078704834,
+      "learning_rate": 1.8014107433532285e-06,
+      "loss": 0.9811,
+      "step": 332
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.9264025688171387,
+      "learning_rate": 1.806836679327184e-06,
+      "loss": 0.6312,
+      "step": 333
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.3180699348449707,
+      "learning_rate": 1.8122626153011397e-06,
+      "loss": 0.5869,
+      "step": 334
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 5.751833915710449,
+      "learning_rate": 1.817688551275095e-06,
+      "loss": 0.9063,
+      "step": 335
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.6618735790252686,
+      "learning_rate": 1.8231144872490504e-06,
+      "loss": 0.7777,
+      "step": 336
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 5.5454511642456055,
+      "learning_rate": 1.8285404232230061e-06,
+      "loss": 0.505,
+      "step": 337
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.7108943462371826,
+      "learning_rate": 1.8339663591969616e-06,
+      "loss": 0.3365,
+      "step": 338
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 9.951203346252441,
+      "learning_rate": 1.839392295170917e-06,
+      "loss": 0.6296,
+      "step": 339
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.6205601692199707,
+      "learning_rate": 1.8448182311448726e-06,
+      "loss": 0.4654,
+      "step": 340
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.3543922901153564,
+      "learning_rate": 1.8502441671188282e-06,
+      "loss": 0.7806,
+      "step": 341
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.55720591545105,
+      "learning_rate": 1.8556701030927837e-06,
+      "loss": 0.5116,
+      "step": 342
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 5.139833927154541,
+      "learning_rate": 1.8610960390667394e-06,
+      "loss": 0.5571,
+      "step": 343
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.8540804386138916,
+      "learning_rate": 1.8665219750406947e-06,
+      "loss": 0.6903,
+      "step": 344
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.9035048484802246,
+      "learning_rate": 1.8719479110146501e-06,
+      "loss": 0.6364,
+      "step": 345
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.4458696842193604,
+      "learning_rate": 1.8773738469886058e-06,
+      "loss": 0.6207,
+      "step": 346
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 6.895814418792725,
+      "learning_rate": 1.8827997829625613e-06,
+      "loss": 0.8627,
+      "step": 347
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.035292625427246,
+      "learning_rate": 1.8882257189365166e-06,
+      "loss": 0.7052,
+      "step": 348
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.5902762413024902,
+      "learning_rate": 1.8936516549104723e-06,
+      "loss": 0.7126,
+      "step": 349
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.525807857513428,
+      "learning_rate": 1.8990775908844277e-06,
+      "loss": 0.5551,
+      "step": 350
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.187638282775879,
+      "learning_rate": 1.9045035268583834e-06,
+      "loss": 0.4732,
+      "step": 351
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 15.85193157196045,
+      "learning_rate": 1.9099294628323387e-06,
+      "loss": 0.5462,
+      "step": 352
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.7592999935150146,
+      "learning_rate": 1.915355398806294e-06,
+      "loss": 0.9046,
+      "step": 353
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.5824947357177734,
+      "learning_rate": 1.9207813347802496e-06,
+      "loss": 0.3728,
+      "step": 354
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.3437206745147705,
+      "learning_rate": 1.9262072707542055e-06,
+      "loss": 0.8698,
+      "step": 355
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.600663423538208,
+      "learning_rate": 1.9316332067281606e-06,
+      "loss": 0.8544,
+      "step": 356
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 6.385979175567627,
+      "learning_rate": 1.937059142702116e-06,
+      "loss": 0.5908,
+      "step": 357
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.5942306518554688,
+      "learning_rate": 1.942485078676072e-06,
+      "loss": 0.557,
+      "step": 358
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.983466386795044,
+      "learning_rate": 1.9479110146500274e-06,
+      "loss": 0.6193,
+      "step": 359
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.260362148284912,
+      "learning_rate": 1.953336950623983e-06,
+      "loss": 0.6939,
+      "step": 360
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.9321560859680176,
+      "learning_rate": 1.9587628865979384e-06,
+      "loss": 0.5211,
+      "step": 361
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.981595754623413,
+      "learning_rate": 1.964188822571894e-06,
+      "loss": 0.6211,
+      "step": 362
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 6.179776668548584,
+      "learning_rate": 1.9696147585458493e-06,
+      "loss": 0.5095,
+      "step": 363
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.9027509689331055,
+      "learning_rate": 1.975040694519805e-06,
+      "loss": 0.5792,
+      "step": 364
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.4844770431518555,
+      "learning_rate": 1.9804666304937603e-06,
+      "loss": 0.6,
+      "step": 365
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 8.19779109954834,
+      "learning_rate": 1.9858925664677158e-06,
+      "loss": 0.6319,
+      "step": 366
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.3586792945861816,
+      "learning_rate": 1.9913185024416717e-06,
+      "loss": 0.7228,
+      "step": 367
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.631184101104736,
+      "learning_rate": 1.9967444384156267e-06,
+      "loss": 0.5206,
+      "step": 368
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.184823989868164,
+      "learning_rate": 2.002170374389582e-06,
+      "loss": 0.5722,
+      "step": 369
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.9428372383117676,
+      "learning_rate": 2.007596310363538e-06,
+      "loss": 0.6282,
+      "step": 370
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.26615571975708,
+      "learning_rate": 2.0130222463374936e-06,
+      "loss": 0.5273,
+      "step": 371
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.9514057636260986,
+      "learning_rate": 2.0184481823114486e-06,
+      "loss": 0.787,
+      "step": 372
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.745385646820068,
+      "learning_rate": 2.0238741182854045e-06,
+      "loss": 0.7583,
+      "step": 373
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.4597623348236084,
+      "learning_rate": 2.02930005425936e-06,
+      "loss": 0.4416,
+      "step": 374
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.602121353149414,
+      "learning_rate": 2.0347259902333155e-06,
+      "loss": 0.4884,
+      "step": 375
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 6.237534046173096,
+      "learning_rate": 2.040151926207271e-06,
+      "loss": 0.588,
+      "step": 376
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.285388946533203,
+      "learning_rate": 2.0455778621812264e-06,
+      "loss": 0.8182,
+      "step": 377
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.696200370788574,
+      "learning_rate": 2.051003798155182e-06,
+      "loss": 0.6455,
+      "step": 378
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.974485874176025,
+      "learning_rate": 2.0564297341291374e-06,
+      "loss": 0.6629,
+      "step": 379
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.0515005588531494,
+      "learning_rate": 2.061855670103093e-06,
+      "loss": 0.681,
+      "step": 380
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 5.143970489501953,
+      "learning_rate": 2.0672816060770483e-06,
+      "loss": 0.6407,
+      "step": 381
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.127249002456665,
+      "learning_rate": 2.0727075420510038e-06,
+      "loss": 0.7737,
+      "step": 382
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 5.232067108154297,
+      "learning_rate": 2.0781334780249597e-06,
+      "loss": 0.6523,
+      "step": 383
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.4227945804595947,
+      "learning_rate": 2.0835594139989147e-06,
+      "loss": 0.5219,
+      "step": 384
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.9718236923217773,
+      "learning_rate": 2.0889853499728706e-06,
+      "loss": 0.6654,
+      "step": 385
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.9642481803894043,
+      "learning_rate": 2.094411285946826e-06,
+      "loss": 0.6542,
+      "step": 386
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.4693069458007812,
+      "learning_rate": 2.0998372219207816e-06,
+      "loss": 0.8682,
+      "step": 387
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 9.878421783447266,
+      "learning_rate": 2.105263157894737e-06,
+      "loss": 0.6592,
+      "step": 388
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.890127658843994,
+      "learning_rate": 2.1106890938686925e-06,
+      "loss": 0.5626,
+      "step": 389
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.8250300884246826,
+      "learning_rate": 2.116115029842648e-06,
+      "loss": 0.4389,
+      "step": 390
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 5.2143378257751465,
+      "learning_rate": 2.1215409658166035e-06,
+      "loss": 0.7067,
+      "step": 391
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.4625725746154785,
+      "learning_rate": 2.126966901790559e-06,
+      "loss": 0.3636,
+      "step": 392
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.4403457641601562,
+      "learning_rate": 2.1323928377645144e-06,
+      "loss": 0.8364,
+      "step": 393
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.6373183727264404,
+      "learning_rate": 2.13781877373847e-06,
+      "loss": 0.5235,
+      "step": 394
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 5.444168567657471,
+      "learning_rate": 2.143244709712426e-06,
+      "loss": 0.5659,
+      "step": 395
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.37904691696167,
+      "learning_rate": 2.148670645686381e-06,
+      "loss": 0.585,
+      "step": 396
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.448920726776123,
+      "learning_rate": 2.1540965816603363e-06,
+      "loss": 0.6126,
+      "step": 397
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.608491897583008,
+      "learning_rate": 2.1595225176342922e-06,
+      "loss": 0.7482,
+      "step": 398
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.902923107147217,
+      "learning_rate": 2.1649484536082477e-06,
+      "loss": 0.6722,
+      "step": 399
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 6.259256839752197,
+      "learning_rate": 2.170374389582203e-06,
+      "loss": 0.4327,
+      "step": 400
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.689056873321533,
+      "learning_rate": 2.1758003255561587e-06,
+      "loss": 0.6657,
+      "step": 401
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.0474588871002197,
+      "learning_rate": 2.181226261530114e-06,
+      "loss": 0.5642,
+      "step": 402
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 9.10236930847168,
+      "learning_rate": 2.1866521975040696e-06,
+      "loss": 0.5046,
+      "step": 403
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.938553810119629,
+      "learning_rate": 2.192078133478025e-06,
+      "loss": 0.5229,
+      "step": 404
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 7.207427978515625,
+      "learning_rate": 2.1975040694519806e-06,
+      "loss": 0.6129,
+      "step": 405
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 6.5248308181762695,
+      "learning_rate": 2.202930005425936e-06,
+      "loss": 0.5766,
+      "step": 406
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.953747510910034,
+      "learning_rate": 2.208355941399892e-06,
+      "loss": 0.5484,
+      "step": 407
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.827918529510498,
+      "learning_rate": 2.213781877373847e-06,
+      "loss": 0.6662,
+      "step": 408
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.183138608932495,
+      "learning_rate": 2.2192078133478025e-06,
+      "loss": 0.5623,
+      "step": 409
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.935085773468018,
+      "learning_rate": 2.2246337493217584e-06,
+      "loss": 0.6752,
+      "step": 410
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 5.551487445831299,
+      "learning_rate": 2.230059685295714e-06,
+      "loss": 0.7318,
+      "step": 411
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.670315742492676,
+      "learning_rate": 2.2354856212696693e-06,
+      "loss": 0.433,
+      "step": 412
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.14329195022583,
+      "learning_rate": 2.2409115572436248e-06,
+      "loss": 0.4756,
+      "step": 413
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.6799099445343018,
+      "learning_rate": 2.2463374932175803e-06,
+      "loss": 0.6651,
+      "step": 414
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.7152655124664307,
+      "learning_rate": 2.2517634291915357e-06,
+      "loss": 0.3045,
+      "step": 415
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.643199920654297,
+      "learning_rate": 2.257189365165491e-06,
+      "loss": 0.4029,
+      "step": 416
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.194269895553589,
+      "learning_rate": 2.2626153011394467e-06,
+      "loss": 0.6456,
+      "step": 417
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.4642317295074463,
+      "learning_rate": 2.268041237113402e-06,
+      "loss": 0.5155,
+      "step": 418
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.443725347518921,
+      "learning_rate": 2.2734671730873576e-06,
+      "loss": 0.7352,
+      "step": 419
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.638228416442871,
+      "learning_rate": 2.2788931090613135e-06,
+      "loss": 0.8598,
+      "step": 420
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 5.12514591217041,
+      "learning_rate": 2.2843190450352686e-06,
+      "loss": 0.8424,
+      "step": 421
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.006558418273926,
+      "learning_rate": 2.289744981009224e-06,
+      "loss": 0.7655,
+      "step": 422
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.166045665740967,
+      "learning_rate": 2.29517091698318e-06,
+      "loss": 0.583,
+      "step": 423
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.150792360305786,
+      "learning_rate": 2.3005968529571354e-06,
+      "loss": 0.7711,
+      "step": 424
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.9197535514831543,
+      "learning_rate": 2.306022788931091e-06,
+      "loss": 0.8838,
+      "step": 425
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.893321990966797,
+      "learning_rate": 2.3114487249050464e-06,
+      "loss": 0.7887,
+      "step": 426
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.3150782585144043,
+      "learning_rate": 2.316874660879002e-06,
+      "loss": 0.8309,
+      "step": 427
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.8468427658081055,
+      "learning_rate": 2.3223005968529573e-06,
+      "loss": 0.7662,
+      "step": 428
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.4848697185516357,
+      "learning_rate": 2.327726532826913e-06,
+      "loss": 0.8137,
+      "step": 429
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 5.247725963592529,
+      "learning_rate": 2.3331524688008683e-06,
+      "loss": 0.7843,
+      "step": 430
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.408827304840088,
+      "learning_rate": 2.3385784047748238e-06,
+      "loss": 0.43,
+      "step": 431
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.770543098449707,
+      "learning_rate": 2.3440043407487797e-06,
+      "loss": 0.842,
+      "step": 432
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.6275100708007812,
+      "learning_rate": 2.3494302767227347e-06,
+      "loss": 0.6038,
+      "step": 433
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.2764031887054443,
+      "learning_rate": 2.35485621269669e-06,
+      "loss": 0.4078,
+      "step": 434
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 6.396345615386963,
+      "learning_rate": 2.360282148670646e-06,
+      "loss": 0.3047,
+      "step": 435
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.3492114543914795,
+      "learning_rate": 2.3657080846446016e-06,
+      "loss": 0.6566,
+      "step": 436
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 5.194329261779785,
+      "learning_rate": 2.3711340206185566e-06,
+      "loss": 0.609,
+      "step": 437
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.8630571365356445,
+      "learning_rate": 2.3765599565925125e-06,
+      "loss": 0.8648,
+      "step": 438
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.5452165603637695,
+      "learning_rate": 2.381985892566468e-06,
+      "loss": 0.8501,
+      "step": 439
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.106873035430908,
+      "learning_rate": 2.3874118285404235e-06,
+      "loss": 0.711,
+      "step": 440
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 5.326449394226074,
+      "learning_rate": 2.392837764514379e-06,
+      "loss": 0.8282,
+      "step": 441
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.511202573776245,
+      "learning_rate": 2.3982637004883344e-06,
+      "loss": 0.8343,
+      "step": 442
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.6755363941192627,
+      "learning_rate": 2.40368963646229e-06,
+      "loss": 0.5608,
+      "step": 443
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 5.037632465362549,
+      "learning_rate": 2.4091155724362454e-06,
+      "loss": 0.8351,
+      "step": 444
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 5.035647392272949,
+      "learning_rate": 2.414541508410201e-06,
+      "loss": 0.7698,
+      "step": 445
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 6.3375444412231445,
+      "learning_rate": 2.4199674443841563e-06,
+      "loss": 0.8296,
+      "step": 446
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 6.8158416748046875,
+      "learning_rate": 2.425393380358112e-06,
+      "loss": 0.5854,
+      "step": 447
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.764331340789795,
+      "learning_rate": 2.4308193163320677e-06,
+      "loss": 0.6879,
+      "step": 448
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.987307548522949,
+      "learning_rate": 2.4362452523060227e-06,
+      "loss": 0.5181,
+      "step": 449
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.997915267944336,
+      "learning_rate": 2.4416711882799786e-06,
+      "loss": 0.7254,
+      "step": 450
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.4239003658294678,
+      "learning_rate": 2.447097124253934e-06,
+      "loss": 0.5806,
+      "step": 451
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.094815731048584,
+      "learning_rate": 2.4525230602278896e-06,
+      "loss": 0.6247,
+      "step": 452
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.446176528930664,
+      "learning_rate": 2.457948996201845e-06,
+      "loss": 0.435,
+      "step": 453
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 6.7582783699035645,
+      "learning_rate": 2.4633749321758005e-06,
+      "loss": 0.9588,
+      "step": 454
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 6.360816478729248,
+      "learning_rate": 2.468800868149756e-06,
+      "loss": 0.4779,
+      "step": 455
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 6.760409832000732,
+      "learning_rate": 2.4742268041237115e-06,
+      "loss": 0.8545,
+      "step": 456
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.1859235763549805,
+      "learning_rate": 2.479652740097667e-06,
+      "loss": 0.7358,
+      "step": 457
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.023052930831909,
+      "learning_rate": 2.4850786760716224e-06,
+      "loss": 0.3929,
+      "step": 458
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.373061180114746,
+      "learning_rate": 2.490504612045578e-06,
+      "loss": 0.7449,
+      "step": 459
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.676642656326294,
+      "learning_rate": 2.495930548019534e-06,
+      "loss": 0.8057,
+      "step": 460
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.9362690448760986,
+      "learning_rate": 2.501356483993489e-06,
+      "loss": 0.8042,
+      "step": 461
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.4391543865203857,
+      "learning_rate": 2.5067824199674443e-06,
+      "loss": 0.4311,
+      "step": 462
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 5.472496032714844,
+      "learning_rate": 2.5122083559414002e-06,
+      "loss": 0.6041,
+      "step": 463
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.6325483322143555,
+      "learning_rate": 2.5176342919153557e-06,
+      "loss": 0.4104,
+      "step": 464
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.2301833629608154,
+      "learning_rate": 2.523060227889311e-06,
+      "loss": 0.4346,
+      "step": 465
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.6823203563690186,
+      "learning_rate": 2.5284861638632667e-06,
+      "loss": 0.5335,
+      "step": 466
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.160337209701538,
+      "learning_rate": 2.5339120998372217e-06,
+      "loss": 0.6199,
+      "step": 467
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 8.221220970153809,
+      "learning_rate": 2.5393380358111776e-06,
+      "loss": 0.5958,
+      "step": 468
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.8410539627075195,
+      "learning_rate": 2.544763971785133e-06,
+      "loss": 0.5723,
+      "step": 469
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.891374111175537,
+      "learning_rate": 2.5501899077590886e-06,
+      "loss": 0.6953,
+      "step": 470
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 5.165709495544434,
+      "learning_rate": 2.555615843733044e-06,
+      "loss": 0.4371,
+      "step": 471
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 7.283283233642578,
+      "learning_rate": 2.561041779707e-06,
+      "loss": 0.5983,
+      "step": 472
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.7218959331512451,
+      "learning_rate": 2.5664677156809554e-06,
+      "loss": 0.5159,
+      "step": 473
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.860637903213501,
+      "learning_rate": 2.571893651654911e-06,
+      "loss": 0.6497,
+      "step": 474
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.6093974113464355,
+      "learning_rate": 2.577319587628866e-06,
+      "loss": 0.8487,
+      "step": 475
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.4497270584106445,
+      "learning_rate": 2.5827455236028214e-06,
+      "loss": 0.6184,
+      "step": 476
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.8384809494018555,
+      "learning_rate": 2.588171459576777e-06,
+      "loss": 0.6126,
+      "step": 477
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.1921048164367676,
+      "learning_rate": 2.5935973955507328e-06,
+      "loss": 0.5804,
+      "step": 478
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.5850296020507812,
+      "learning_rate": 2.5990233315246883e-06,
+      "loss": 0.6242,
+      "step": 479
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.009633541107178,
+      "learning_rate": 2.6044492674986437e-06,
+      "loss": 0.6232,
+      "step": 480
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.089764595031738,
+      "learning_rate": 2.609875203472599e-06,
+      "loss": 0.7529,
+      "step": 481
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 6.3803558349609375,
+      "learning_rate": 2.615301139446555e-06,
+      "loss": 0.7396,
+      "step": 482
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.518465042114258,
+      "learning_rate": 2.6207270754205106e-06,
+      "loss": 0.5699,
+      "step": 483
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.091625690460205,
+      "learning_rate": 2.6261530113944656e-06,
+      "loss": 0.347,
+      "step": 484
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.701561450958252,
+      "learning_rate": 2.631578947368421e-06,
+      "loss": 0.7027,
+      "step": 485
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.7645835876464844,
+      "learning_rate": 2.6370048833423766e-06,
+      "loss": 0.5931,
+      "step": 486
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.4653618335723877,
+      "learning_rate": 2.6424308193163325e-06,
+      "loss": 0.692,
+      "step": 487
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.409841775894165,
+      "learning_rate": 2.647856755290288e-06,
+      "loss": 0.56,
+      "step": 488
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.7647156715393066,
+      "learning_rate": 2.6532826912642434e-06,
+      "loss": 0.686,
+      "step": 489
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.72574520111084,
+      "learning_rate": 2.658708627238199e-06,
+      "loss": 0.8847,
+      "step": 490
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 6.593405723571777,
+      "learning_rate": 2.6641345632121544e-06,
+      "loss": 0.7098,
+      "step": 491
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.206516742706299,
+      "learning_rate": 2.6695604991861094e-06,
+      "loss": 0.6993,
+      "step": 492
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.31909441947937,
+      "learning_rate": 2.6749864351600653e-06,
+      "loss": 0.2194,
+      "step": 493
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.922511100769043,
+      "learning_rate": 2.680412371134021e-06,
+      "loss": 0.6528,
+      "step": 494
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.7614450454711914,
+      "learning_rate": 2.6858383071079763e-06,
+      "loss": 0.6817,
+      "step": 495
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 7.144618988037109,
+      "learning_rate": 2.6912642430819318e-06,
+      "loss": 0.705,
+      "step": 496
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.564678192138672,
+      "learning_rate": 2.6966901790558877e-06,
+      "loss": 0.7391,
+      "step": 497
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.81742000579834,
+      "learning_rate": 2.702116115029843e-06,
+      "loss": 0.658,
+      "step": 498
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 5.687780857086182,
+      "learning_rate": 2.7075420510037986e-06,
+      "loss": 0.8419,
+      "step": 499
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.521057605743408,
+      "learning_rate": 2.7129679869777537e-06,
+      "loss": 0.6188,
+      "step": 500
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 6.0916032791137695,
+      "learning_rate": 2.718393922951709e-06,
+      "loss": 0.6987,
+      "step": 501
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 6.832183361053467,
+      "learning_rate": 2.7238198589256646e-06,
+      "loss": 0.6174,
+      "step": 502
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.9268059730529785,
+      "learning_rate": 2.7292457948996205e-06,
+      "loss": 0.7871,
+      "step": 503
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.4391541481018066,
+      "learning_rate": 2.734671730873576e-06,
+      "loss": 0.823,
+      "step": 504
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.789524555206299,
+      "learning_rate": 2.7400976668475315e-06,
+      "loss": 0.4945,
+      "step": 505
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.33085036277771,
+      "learning_rate": 2.745523602821487e-06,
+      "loss": 0.4812,
+      "step": 506
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.542304039001465,
+      "learning_rate": 2.750949538795443e-06,
+      "loss": 0.6613,
+      "step": 507
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.1349079608917236,
+      "learning_rate": 2.756375474769398e-06,
+      "loss": 0.8186,
+      "step": 508
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.910358905792236,
+      "learning_rate": 2.7618014107433534e-06,
+      "loss": 0.5314,
+      "step": 509
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.898296356201172,
+      "learning_rate": 2.767227346717309e-06,
+      "loss": 0.6026,
+      "step": 510
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.7290122509002686,
+      "learning_rate": 2.7726532826912643e-06,
+      "loss": 0.8299,
+      "step": 511
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 6.941473960876465,
+      "learning_rate": 2.77807921866522e-06,
+      "loss": 0.5959,
+      "step": 512
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.126974582672119,
+      "learning_rate": 2.7835051546391757e-06,
+      "loss": 0.7685,
+      "step": 513
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.4374797344207764,
+      "learning_rate": 2.788931090613131e-06,
+      "loss": 0.6528,
+      "step": 514
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.161393165588379,
+      "learning_rate": 2.7943570265870866e-06,
+      "loss": 0.8627,
+      "step": 515
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.266381025314331,
+      "learning_rate": 2.7997829625610417e-06,
+      "loss": 0.5197,
+      "step": 516
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.9556145668029785,
+      "learning_rate": 2.805208898534997e-06,
+      "loss": 0.7441,
+      "step": 517
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.324782371520996,
+      "learning_rate": 2.810634834508953e-06,
+      "loss": 0.5486,
+      "step": 518
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.6474103927612305,
+      "learning_rate": 2.8160607704829085e-06,
+      "loss": 0.5472,
+      "step": 519
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 14.287810325622559,
+      "learning_rate": 2.821486706456864e-06,
+      "loss": 0.563,
+      "step": 520
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.5945661067962646,
+      "learning_rate": 2.8269126424308195e-06,
+      "loss": 0.9987,
+      "step": 521
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.526186227798462,
+      "learning_rate": 2.8323385784047754e-06,
+      "loss": 0.4299,
+      "step": 522
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.336894989013672,
+      "learning_rate": 2.837764514378731e-06,
+      "loss": 0.6811,
+      "step": 523
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.6765902042388916,
+      "learning_rate": 2.843190450352686e-06,
+      "loss": 0.7039,
+      "step": 524
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 9.413053512573242,
+      "learning_rate": 2.8486163863266414e-06,
+      "loss": 0.7124,
+      "step": 525
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.6481852531433105,
+      "learning_rate": 2.854042322300597e-06,
+      "loss": 0.5863,
+      "step": 526
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.1243183612823486,
+      "learning_rate": 2.8594682582745523e-06,
+      "loss": 0.4241,
+      "step": 527
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 5.649373531341553,
+      "learning_rate": 2.8648941942485082e-06,
+      "loss": 0.8415,
+      "step": 528
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.963700294494629,
+      "learning_rate": 2.8703201302224637e-06,
+      "loss": 0.5865,
+      "step": 529
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 5.258522033691406,
+      "learning_rate": 2.875746066196419e-06,
+      "loss": 0.7103,
+      "step": 530
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.3650219440460205,
+      "learning_rate": 2.8811720021703747e-06,
+      "loss": 0.5282,
+      "step": 531
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.6232759952545166,
+      "learning_rate": 2.8865979381443297e-06,
+      "loss": 0.4613,
+      "step": 532
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 10.552885055541992,
+      "learning_rate": 2.8920238741182856e-06,
+      "loss": 0.5432,
+      "step": 533
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.1861660480499268,
+      "learning_rate": 2.897449810092241e-06,
+      "loss": 0.5081,
+      "step": 534
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.0140583515167236,
+      "learning_rate": 2.9028757460661966e-06,
+      "loss": 0.6793,
+      "step": 535
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.722212076187134,
+      "learning_rate": 2.908301682040152e-06,
+      "loss": 0.7429,
+      "step": 536
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.10899019241333,
+      "learning_rate": 2.913727618014108e-06,
+      "loss": 0.3228,
+      "step": 537
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 11.692066192626953,
+      "learning_rate": 2.9191535539880634e-06,
+      "loss": 0.6989,
+      "step": 538
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 5.75980806350708,
+      "learning_rate": 2.924579489962019e-06,
+      "loss": 0.7452,
+      "step": 539
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 6.4372687339782715,
+      "learning_rate": 2.930005425935974e-06,
+      "loss": 0.95,
+      "step": 540
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.0145132541656494,
+      "learning_rate": 2.9354313619099294e-06,
+      "loss": 0.7712,
+      "step": 541
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 5.343160629272461,
+      "learning_rate": 2.940857297883885e-06,
+      "loss": 0.6508,
+      "step": 542
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.744476795196533,
+      "learning_rate": 2.9462832338578408e-06,
+      "loss": 0.7488,
+      "step": 543
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.3671207427978516,
+      "learning_rate": 2.9517091698317963e-06,
+      "loss": 0.5251,
+      "step": 544
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.416987895965576,
+      "learning_rate": 2.9571351058057517e-06,
+      "loss": 0.668,
+      "step": 545
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.492126703262329,
+      "learning_rate": 2.9625610417797072e-06,
+      "loss": 0.7112,
+      "step": 546
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.771510601043701,
+      "learning_rate": 2.967986977753663e-06,
+      "loss": 0.438,
+      "step": 547
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.357903003692627,
+      "learning_rate": 2.9734129137276186e-06,
+      "loss": 0.5099,
+      "step": 548
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 7.1840596199035645,
+      "learning_rate": 2.9788388497015736e-06,
+      "loss": 0.6134,
+      "step": 549
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 9.533690452575684,
+      "learning_rate": 2.984264785675529e-06,
+      "loss": 1.0784,
+      "step": 550
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.8269598484039307,
+      "learning_rate": 2.9896907216494846e-06,
+      "loss": 0.6375,
+      "step": 551
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.0468032360076904,
+      "learning_rate": 2.9951166576234405e-06,
+      "loss": 0.5177,
+      "step": 552
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.90291690826416,
+      "learning_rate": 3.000542593597396e-06,
+      "loss": 0.5296,
+      "step": 553
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 6.419053077697754,
+      "learning_rate": 3.0059685295713514e-06,
+      "loss": 0.7483,
+      "step": 554
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.6837029457092285,
+      "learning_rate": 3.011394465545307e-06,
+      "loss": 0.405,
+      "step": 555
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.12681770324707,
+      "learning_rate": 3.0168204015192624e-06,
+      "loss": 0.809,
+      "step": 556
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.9505882263183594,
+      "learning_rate": 3.0222463374932174e-06,
+      "loss": 0.8399,
+      "step": 557
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.520677089691162,
+      "learning_rate": 3.0276722734671733e-06,
+      "loss": 0.6047,
+      "step": 558
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.7374749183654785,
+      "learning_rate": 3.033098209441129e-06,
+      "loss": 0.643,
+      "step": 559
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.929847478866577,
+      "learning_rate": 3.0385241454150843e-06,
+      "loss": 0.8534,
+      "step": 560
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.561032772064209,
+      "learning_rate": 3.0439500813890398e-06,
+      "loss": 0.4575,
+      "step": 561
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 6.397409915924072,
+      "learning_rate": 3.0493760173629957e-06,
+      "loss": 0.5463,
+      "step": 562
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.934955596923828,
+      "learning_rate": 3.054801953336951e-06,
+      "loss": 0.7137,
+      "step": 563
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.053652763366699,
+      "learning_rate": 3.0602278893109066e-06,
+      "loss": 0.5274,
+      "step": 564
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.2374649047851562,
+      "learning_rate": 3.0656538252848617e-06,
+      "loss": 0.3657,
+      "step": 565
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 6.124660015106201,
+      "learning_rate": 3.071079761258817e-06,
+      "loss": 0.6066,
+      "step": 566
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.4570655822753906,
+      "learning_rate": 3.0765056972327726e-06,
+      "loss": 0.7976,
+      "step": 567
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 5.434204578399658,
+      "learning_rate": 3.0819316332067285e-06,
+      "loss": 0.4039,
+      "step": 568
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.928964614868164,
+      "learning_rate": 3.087357569180684e-06,
+      "loss": 0.6852,
+      "step": 569
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.842745542526245,
+      "learning_rate": 3.0927835051546395e-06,
+      "loss": 0.5126,
+      "step": 570
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.637947082519531,
+      "learning_rate": 3.098209441128595e-06,
+      "loss": 0.5221,
+      "step": 571
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 5.171610355377197,
+      "learning_rate": 3.103635377102551e-06,
+      "loss": 0.4703,
+      "step": 572
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.9962360858917236,
+      "learning_rate": 3.109061313076506e-06,
+      "loss": 0.5633,
+      "step": 573
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.8839361667633057,
+      "learning_rate": 3.1144872490504614e-06,
+      "loss": 0.3391,
+      "step": 574
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.454033136367798,
+      "learning_rate": 3.119913185024417e-06,
+      "loss": 0.8572,
+      "step": 575
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 10.574701309204102,
+      "learning_rate": 3.1253391209983723e-06,
+      "loss": 0.819,
+      "step": 576
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.2963690757751465,
+      "learning_rate": 3.1307650569723282e-06,
+      "loss": 0.5599,
+      "step": 577
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.550469875335693,
+      "learning_rate": 3.1361909929462837e-06,
+      "loss": 0.6393,
+      "step": 578
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.706594228744507,
+      "learning_rate": 3.141616928920239e-06,
+      "loss": 0.8332,
+      "step": 579
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.224299430847168,
+      "learning_rate": 3.1470428648941946e-06,
+      "loss": 0.6952,
+      "step": 580
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.5631625652313232,
+      "learning_rate": 3.1524688008681497e-06,
+      "loss": 0.5827,
+      "step": 581
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.646390676498413,
+      "learning_rate": 3.157894736842105e-06,
+      "loss": 0.7634,
+      "step": 582
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.090953350067139,
+      "learning_rate": 3.163320672816061e-06,
+      "loss": 0.5706,
+      "step": 583
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.2332234382629395,
+      "learning_rate": 3.1687466087900165e-06,
+      "loss": 0.6623,
+      "step": 584
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 19.10224723815918,
+      "learning_rate": 3.174172544763972e-06,
+      "loss": 0.6534,
+      "step": 585
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.5347118377685547,
+      "learning_rate": 3.1795984807379275e-06,
+      "loss": 0.7061,
+      "step": 586
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.3615777492523193,
+      "learning_rate": 3.1850244167118834e-06,
+      "loss": 0.6922,
+      "step": 587
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.8111391067504883,
+      "learning_rate": 3.190450352685839e-06,
+      "loss": 0.5144,
+      "step": 588
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 11.927906036376953,
+      "learning_rate": 3.195876288659794e-06,
+      "loss": 0.6778,
+      "step": 589
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.32316517829895,
+      "learning_rate": 3.2013022246337494e-06,
+      "loss": 0.5924,
+      "step": 590
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.9122865200042725,
+      "learning_rate": 3.206728160607705e-06,
+      "loss": 0.3164,
+      "step": 591
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 7.80124568939209,
+      "learning_rate": 3.2121540965816608e-06,
+      "loss": 0.8358,
+      "step": 592
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 6.983058452606201,
+      "learning_rate": 3.2175800325556162e-06,
+      "loss": 0.795,
+      "step": 593
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 16.85896110534668,
+      "learning_rate": 3.2230059685295717e-06,
+      "loss": 0.68,
+      "step": 594
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.5477426052093506,
+      "learning_rate": 3.228431904503527e-06,
+      "loss": 0.3729,
+      "step": 595
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.4073872566223145,
+      "learning_rate": 3.2338578404774827e-06,
+      "loss": 0.5078,
+      "step": 596
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.413442850112915,
+      "learning_rate": 3.2392837764514377e-06,
+      "loss": 0.6131,
+      "step": 597
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 7.412133693695068,
+      "learning_rate": 3.2447097124253936e-06,
+      "loss": 0.5112,
+      "step": 598
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 5.138059139251709,
+      "learning_rate": 3.250135648399349e-06,
+      "loss": 0.456,
+      "step": 599
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.566953659057617,
+      "learning_rate": 3.2555615843733046e-06,
+      "loss": 0.6328,
+      "step": 600
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.605090141296387,
+      "learning_rate": 3.26098752034726e-06,
+      "loss": 0.6016,
+      "step": 601
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.141775608062744,
+      "learning_rate": 3.266413456321216e-06,
+      "loss": 0.6054,
+      "step": 602
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.5525505542755127,
+      "learning_rate": 3.2718393922951714e-06,
+      "loss": 0.6273,
+      "step": 603
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 7.644792079925537,
+      "learning_rate": 3.277265328269127e-06,
+      "loss": 0.4226,
+      "step": 604
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 8.977129936218262,
+      "learning_rate": 3.282691264243082e-06,
+      "loss": 0.633,
+      "step": 605
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 7.189108848571777,
+      "learning_rate": 3.2881172002170374e-06,
+      "loss": 0.5315,
+      "step": 606
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 5.684975624084473,
+      "learning_rate": 3.293543136190993e-06,
+      "loss": 0.8671,
+      "step": 607
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 8.830580711364746,
+      "learning_rate": 3.298969072164949e-06,
+      "loss": 0.4746,
+      "step": 608
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.3558552265167236,
+      "learning_rate": 3.3043950081389043e-06,
+      "loss": 0.6896,
+      "step": 609
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.8798203468322754,
+      "learning_rate": 3.3098209441128597e-06,
+      "loss": 0.5871,
+      "step": 610
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.0817947387695312,
+      "learning_rate": 3.3152468800868152e-06,
+      "loss": 0.7144,
+      "step": 611
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.4752776622772217,
+      "learning_rate": 3.320672816060771e-06,
+      "loss": 0.8516,
+      "step": 612
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.9830877780914307,
+      "learning_rate": 3.326098752034726e-06,
+      "loss": 0.6609,
+      "step": 613
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.722573757171631,
+      "learning_rate": 3.3315246880086816e-06,
+      "loss": 0.8455,
+      "step": 614
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.2305538654327393,
+      "learning_rate": 3.336950623982637e-06,
+      "loss": 0.6966,
+      "step": 615
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.231105089187622,
+      "learning_rate": 3.3423765599565926e-06,
+      "loss": 0.5561,
+      "step": 616
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.241834878921509,
+      "learning_rate": 3.3478024959305485e-06,
+      "loss": 0.6858,
+      "step": 617
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.3830490112304688,
+      "learning_rate": 3.353228431904504e-06,
+      "loss": 0.5823,
+      "step": 618
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 6.515814781188965,
+      "learning_rate": 3.3586543678784594e-06,
+      "loss": 0.6229,
+      "step": 619
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.4859941005706787,
+      "learning_rate": 3.364080303852415e-06,
+      "loss": 0.4493,
+      "step": 620
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.841228723526001,
+      "learning_rate": 3.3695062398263704e-06,
+      "loss": 0.4637,
+      "step": 621
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.490614652633667,
+      "learning_rate": 3.3749321758003254e-06,
+      "loss": 0.7432,
+      "step": 622
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 5.034761905670166,
+      "learning_rate": 3.3803581117742813e-06,
+      "loss": 0.556,
+      "step": 623
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 6.398545265197754,
+      "learning_rate": 3.385784047748237e-06,
+      "loss": 0.6388,
+      "step": 624
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.8244869709014893,
+      "learning_rate": 3.3912099837221923e-06,
+      "loss": 0.7801,
+      "step": 625
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.9833526611328125,
+      "learning_rate": 3.3966359196961478e-06,
+      "loss": 0.514,
+      "step": 626
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.331082820892334,
+      "learning_rate": 3.4020618556701037e-06,
+      "loss": 0.3137,
+      "step": 627
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 5.496692657470703,
+      "learning_rate": 3.407487791644059e-06,
+      "loss": 0.7433,
+      "step": 628
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.3837697505950928,
+      "learning_rate": 3.4129137276180146e-06,
+      "loss": 0.5627,
+      "step": 629
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 6.360167980194092,
+      "learning_rate": 3.4183396635919697e-06,
+      "loss": 0.5472,
+      "step": 630
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.2160558700561523,
+      "learning_rate": 3.423765599565925e-06,
+      "loss": 0.6665,
+      "step": 631
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 6.714569568634033,
+      "learning_rate": 3.4291915355398806e-06,
+      "loss": 0.526,
+      "step": 632
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 7.493886470794678,
+      "learning_rate": 3.4346174715138365e-06,
+      "loss": 0.6489,
+      "step": 633
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.0544168949127197,
+      "learning_rate": 3.440043407487792e-06,
+      "loss": 0.734,
+      "step": 634
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.6884641647338867,
+      "learning_rate": 3.4454693434617475e-06,
+      "loss": 0.8494,
+      "step": 635
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.23567795753479,
+      "learning_rate": 3.450895279435703e-06,
+      "loss": 0.4155,
+      "step": 636
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.425140857696533,
+      "learning_rate": 3.456321215409659e-06,
+      "loss": 0.3907,
+      "step": 637
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.3101117610931396,
+      "learning_rate": 3.461747151383614e-06,
+      "loss": 0.5058,
+      "step": 638
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.480876445770264,
+      "learning_rate": 3.4671730873575694e-06,
+      "loss": 0.8928,
+      "step": 639
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.211671829223633,
+      "learning_rate": 3.472599023331525e-06,
+      "loss": 0.3608,
+      "step": 640
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.0141637325286865,
+      "learning_rate": 3.4780249593054803e-06,
+      "loss": 0.6065,
+      "step": 641
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.9783501625061035,
+      "learning_rate": 3.4834508952794362e-06,
+      "loss": 0.6506,
+      "step": 642
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 6.977810859680176,
+      "learning_rate": 3.4888768312533917e-06,
+      "loss": 0.6548,
+      "step": 643
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.6588103771209717,
+      "learning_rate": 3.494302767227347e-06,
+      "loss": 0.6203,
+      "step": 644
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.5979905128479004,
+      "learning_rate": 3.4997287032013026e-06,
+      "loss": 0.8082,
+      "step": 645
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 11.069365501403809,
+      "learning_rate": 3.5051546391752577e-06,
+      "loss": 0.7052,
+      "step": 646
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.639838695526123,
+      "learning_rate": 3.510580575149213e-06,
+      "loss": 0.7519,
+      "step": 647
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.052438735961914,
+      "learning_rate": 3.516006511123169e-06,
+      "loss": 0.7952,
+      "step": 648
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.482348918914795,
+      "learning_rate": 3.5214324470971245e-06,
+      "loss": 0.8459,
+      "step": 649
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.397634983062744,
+      "learning_rate": 3.52685838307108e-06,
+      "loss": 0.7213,
+      "step": 650
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.997109889984131,
+      "learning_rate": 3.5322843190450355e-06,
+      "loss": 0.4837,
+      "step": 651
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.269824504852295,
+      "learning_rate": 3.5377102550189914e-06,
+      "loss": 0.5747,
+      "step": 652
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 5.910425186157227,
+      "learning_rate": 3.543136190992947e-06,
+      "loss": 0.8262,
+      "step": 653
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.488101005554199,
+      "learning_rate": 3.548562126966902e-06,
+      "loss": 0.4164,
+      "step": 654
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 7.880976676940918,
+      "learning_rate": 3.5539880629408574e-06,
+      "loss": 0.658,
+      "step": 655
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 6.965835094451904,
+      "learning_rate": 3.559413998914813e-06,
+      "loss": 0.5227,
+      "step": 656
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.596857786178589,
+      "learning_rate": 3.5648399348887688e-06,
+      "loss": 0.6466,
+      "step": 657
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 6.5818963050842285,
+      "learning_rate": 3.5702658708627242e-06,
+      "loss": 0.5671,
+      "step": 658
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 5.714598178863525,
+      "learning_rate": 3.5756918068366797e-06,
+      "loss": 0.502,
+      "step": 659
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.006031513214111,
+      "learning_rate": 3.581117742810635e-06,
+      "loss": 0.7606,
+      "step": 660
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.7842965126037598,
+      "learning_rate": 3.5865436787845907e-06,
+      "loss": 0.4146,
+      "step": 661
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 9.799056053161621,
+      "learning_rate": 3.5919696147585457e-06,
+      "loss": 0.8146,
+      "step": 662
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.944117546081543,
+      "learning_rate": 3.5973955507325016e-06,
+      "loss": 0.6431,
+      "step": 663
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.253502130508423,
+      "learning_rate": 3.602821486706457e-06,
+      "loss": 0.4178,
+      "step": 664
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.278943061828613,
+      "learning_rate": 3.6082474226804126e-06,
+      "loss": 0.8092,
+      "step": 665
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 11.072491645812988,
+      "learning_rate": 3.613673358654368e-06,
+      "loss": 0.4353,
+      "step": 666
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.292236804962158,
+      "learning_rate": 3.619099294628324e-06,
+      "loss": 0.4748,
+      "step": 667
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.021798133850098,
+      "learning_rate": 3.6245252306022794e-06,
+      "loss": 0.6181,
+      "step": 668
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.6391448974609375,
+      "learning_rate": 3.629951166576235e-06,
+      "loss": 0.676,
+      "step": 669
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.899008750915527,
+      "learning_rate": 3.63537710255019e-06,
+      "loss": 0.5926,
+      "step": 670
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.1783127784729,
+      "learning_rate": 3.6408030385241454e-06,
+      "loss": 0.7113,
+      "step": 671
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.2804505825042725,
+      "learning_rate": 3.646228974498101e-06,
+      "loss": 0.4476,
+      "step": 672
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.259552001953125,
+      "learning_rate": 3.651654910472057e-06,
+      "loss": 0.7647,
+      "step": 673
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.1471548080444336,
+      "learning_rate": 3.6570808464460123e-06,
+      "loss": 0.9114,
+      "step": 674
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.9130606651306152,
+      "learning_rate": 3.6625067824199677e-06,
+      "loss": 0.3963,
+      "step": 675
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.058765888214111,
+      "learning_rate": 3.6679327183939232e-06,
+      "loss": 0.783,
+      "step": 676
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.6345887184143066,
+      "learning_rate": 3.673358654367879e-06,
+      "loss": 0.8038,
+      "step": 677
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.9325718879699707,
+      "learning_rate": 3.678784590341834e-06,
+      "loss": 0.6888,
+      "step": 678
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.4944450855255127,
+      "learning_rate": 3.6842105263157896e-06,
+      "loss": 0.547,
+      "step": 679
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.9259848594665527,
+      "learning_rate": 3.689636462289745e-06,
+      "loss": 0.5316,
+      "step": 680
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.845626354217529,
+      "learning_rate": 3.6950623982637006e-06,
+      "loss": 0.8265,
+      "step": 681
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.280968189239502,
+      "learning_rate": 3.7004883342376565e-06,
+      "loss": 0.4911,
+      "step": 682
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 6.173648357391357,
+      "learning_rate": 3.705914270211612e-06,
+      "loss": 0.3757,
+      "step": 683
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.729910850524902,
+      "learning_rate": 3.7113402061855674e-06,
+      "loss": 0.717,
+      "step": 684
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 5.336638927459717,
+      "learning_rate": 3.716766142159523e-06,
+      "loss": 0.7997,
+      "step": 685
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.4942779541015625,
+      "learning_rate": 3.722192078133479e-06,
+      "loss": 0.6724,
+      "step": 686
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 23.758148193359375,
+      "learning_rate": 3.7276180141074334e-06,
+      "loss": 0.9579,
+      "step": 687
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.6573519706726074,
+      "learning_rate": 3.7330439500813893e-06,
+      "loss": 0.753,
+      "step": 688
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.439124584197998,
+      "learning_rate": 3.738469886055345e-06,
+      "loss": 0.6701,
+      "step": 689
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 5.706178665161133,
+      "learning_rate": 3.7438958220293003e-06,
+      "loss": 0.7403,
+      "step": 690
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 7.338541507720947,
+      "learning_rate": 3.7493217580032558e-06,
+      "loss": 0.6098,
+      "step": 691
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.903101921081543,
+      "learning_rate": 3.7547476939772117e-06,
+      "loss": 0.6394,
+      "step": 692
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.0162651538848877,
+      "learning_rate": 3.760173629951167e-06,
+      "loss": 0.6008,
+      "step": 693
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 9.352497100830078,
+      "learning_rate": 3.7655995659251226e-06,
+      "loss": 0.3842,
+      "step": 694
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.1430256366729736,
+      "learning_rate": 3.7710255018990777e-06,
+      "loss": 0.6051,
+      "step": 695
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 7.965341567993164,
+      "learning_rate": 3.776451437873033e-06,
+      "loss": 0.6125,
+      "step": 696
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.9353039264678955,
+      "learning_rate": 3.781877373846989e-06,
+      "loss": 0.4905,
+      "step": 697
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 5.226293563842773,
+      "learning_rate": 3.7873033098209445e-06,
+      "loss": 0.5976,
+      "step": 698
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 7.95682954788208,
+      "learning_rate": 3.7927292457949e-06,
+      "loss": 0.6675,
+      "step": 699
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 5.938877105712891,
+      "learning_rate": 3.7981551817688555e-06,
+      "loss": 0.5275,
+      "step": 700
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 11.931106567382812,
+      "learning_rate": 3.803581117742811e-06,
+      "loss": 0.7462,
+      "step": 701
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.7707386016845703,
+      "learning_rate": 3.809007053716767e-06,
+      "loss": 0.8025,
+      "step": 702
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.8362505435943604,
+      "learning_rate": 3.814432989690722e-06,
+      "loss": 0.5847,
+      "step": 703
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.801651954650879,
+      "learning_rate": 3.819858925664677e-06,
+      "loss": 0.8027,
+      "step": 704
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.092442989349365,
+      "learning_rate": 3.825284861638633e-06,
+      "loss": 0.4992,
+      "step": 705
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.886186599731445,
+      "learning_rate": 3.830710797612588e-06,
+      "loss": 0.6284,
+      "step": 706
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.403731346130371,
+      "learning_rate": 3.836136733586544e-06,
+      "loss": 0.6075,
+      "step": 707
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 7.999245643615723,
+      "learning_rate": 3.841562669560499e-06,
+      "loss": 0.5219,
+      "step": 708
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 6.004327297210693,
+      "learning_rate": 3.846988605534456e-06,
+      "loss": 0.5833,
+      "step": 709
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.1722559928894043,
+      "learning_rate": 3.852414541508411e-06,
+      "loss": 0.8112,
+      "step": 710
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.478994369506836,
+      "learning_rate": 3.857840477482366e-06,
+      "loss": 0.6536,
+      "step": 711
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 8.532600402832031,
+      "learning_rate": 3.863266413456321e-06,
+      "loss": 0.917,
+      "step": 712
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.2792649269104004,
+      "learning_rate": 3.868692349430277e-06,
+      "loss": 0.8043,
+      "step": 713
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.07094407081604,
+      "learning_rate": 3.874118285404232e-06,
+      "loss": 0.8769,
+      "step": 714
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.5107150077819824,
+      "learning_rate": 3.8795442213781884e-06,
+      "loss": 0.4569,
+      "step": 715
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 5.8087568283081055,
+      "learning_rate": 3.884970157352144e-06,
+      "loss": 0.7869,
+      "step": 716
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.036777973175049,
+      "learning_rate": 3.890396093326099e-06,
+      "loss": 0.47,
+      "step": 717
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.7074899673461914,
+      "learning_rate": 3.895822029300055e-06,
+      "loss": 0.7299,
+      "step": 718
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.5969338417053223,
+      "learning_rate": 3.9012479652740095e-06,
+      "loss": 1.1114,
+      "step": 719
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 6.4888596534729,
+      "learning_rate": 3.906673901247966e-06,
+      "loss": 0.6167,
+      "step": 720
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.2016541957855225,
+      "learning_rate": 3.912099837221921e-06,
+      "loss": 0.5552,
+      "step": 721
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.657111883163452,
+      "learning_rate": 3.917525773195877e-06,
+      "loss": 0.6724,
+      "step": 722
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.291457414627075,
+      "learning_rate": 3.922951709169832e-06,
+      "loss": 0.6604,
+      "step": 723
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.5446839332580566,
+      "learning_rate": 3.928377645143788e-06,
+      "loss": 0.7161,
+      "step": 724
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.9065346717834473,
+      "learning_rate": 3.933803581117743e-06,
+      "loss": 0.5252,
+      "step": 725
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.347808361053467,
+      "learning_rate": 3.939229517091699e-06,
+      "loss": 0.4491,
+      "step": 726
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.4501278400421143,
+      "learning_rate": 3.944655453065654e-06,
+      "loss": 0.7285,
+      "step": 727
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 8.243633270263672,
+      "learning_rate": 3.95008138903961e-06,
+      "loss": 0.4732,
+      "step": 728
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.671466588973999,
+      "learning_rate": 3.955507325013565e-06,
+      "loss": 0.1834,
+      "step": 729
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.9374115467071533,
+      "learning_rate": 3.9609332609875206e-06,
+      "loss": 0.4564,
+      "step": 730
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.069698810577393,
+      "learning_rate": 3.966359196961476e-06,
+      "loss": 0.605,
+      "step": 731
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 5.08325719833374,
+      "learning_rate": 3.9717851329354315e-06,
+      "loss": 0.5688,
+      "step": 732
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.5298423767089844,
+      "learning_rate": 3.977211068909387e-06,
+      "loss": 0.5917,
+      "step": 733
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.8950157165527344,
+      "learning_rate": 3.982637004883343e-06,
+      "loss": 0.2869,
+      "step": 734
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.334839820861816,
+      "learning_rate": 3.988062940857298e-06,
+      "loss": 0.7682,
+      "step": 735
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.9833765029907227,
+      "learning_rate": 3.993488876831253e-06,
+      "loss": 0.7067,
+      "step": 736
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.288568019866943,
+      "learning_rate": 3.998914812805209e-06,
+      "loss": 0.5819,
+      "step": 737
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.143844127655029,
+      "learning_rate": 4.004340748779164e-06,
+      "loss": 0.7815,
+      "step": 738
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.390470027923584,
+      "learning_rate": 4.00976668475312e-06,
+      "loss": 0.3212,
+      "step": 739
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.510004758834839,
+      "learning_rate": 4.015192620727076e-06,
+      "loss": 0.4628,
+      "step": 740
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.720365285873413,
+      "learning_rate": 4.020618556701032e-06,
+      "loss": 0.5452,
+      "step": 741
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 11.226015090942383,
+      "learning_rate": 4.026044492674987e-06,
+      "loss": 0.6159,
+      "step": 742
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.6108076572418213,
+      "learning_rate": 4.031470428648942e-06,
+      "loss": 0.63,
+      "step": 743
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.208956480026245,
+      "learning_rate": 4.036896364622897e-06,
+      "loss": 0.5544,
+      "step": 744
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 5.13582706451416,
+      "learning_rate": 4.0423223005968535e-06,
+      "loss": 0.5957,
+      "step": 745
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.928217887878418,
+      "learning_rate": 4.047748236570809e-06,
+      "loss": 0.7243,
+      "step": 746
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.2266054153442383,
+      "learning_rate": 4.0531741725447645e-06,
+      "loss": 0.5294,
+      "step": 747
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.4554672241210938,
+      "learning_rate": 4.05860010851872e-06,
+      "loss": 0.7607,
+      "step": 748
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.5808000564575195,
+      "learning_rate": 4.0640260444926754e-06,
+      "loss": 0.6752,
+      "step": 749
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.918306827545166,
+      "learning_rate": 4.069451980466631e-06,
+      "loss": 0.619,
+      "step": 750
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.909853458404541,
+      "learning_rate": 4.074877916440586e-06,
+      "loss": 0.6379,
+      "step": 751
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.4120168685913086,
+      "learning_rate": 4.080303852414542e-06,
+      "loss": 0.5112,
+      "step": 752
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.382999897003174,
+      "learning_rate": 4.085729788388497e-06,
+      "loss": 0.6293,
+      "step": 753
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 5.9883880615234375,
+      "learning_rate": 4.091155724362453e-06,
+      "loss": 0.6103,
+      "step": 754
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.7289187908172607,
+      "learning_rate": 4.096581660336408e-06,
+      "loss": 0.5467,
+      "step": 755
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.091078758239746,
+      "learning_rate": 4.102007596310364e-06,
+      "loss": 0.4749,
+      "step": 756
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.6641273498535156,
+      "learning_rate": 4.107433532284319e-06,
+      "loss": 0.6718,
+      "step": 757
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 5.8359150886535645,
+      "learning_rate": 4.112859468258275e-06,
+      "loss": 0.7193,
+      "step": 758
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 5.1827802658081055,
+      "learning_rate": 4.118285404232231e-06,
+      "loss": 0.5662,
+      "step": 759
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.326897382736206,
+      "learning_rate": 4.123711340206186e-06,
+      "loss": 0.8102,
+      "step": 760
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.0198025703430176,
+      "learning_rate": 4.129137276180141e-06,
+      "loss": 0.7375,
+      "step": 761
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 5.452925682067871,
+      "learning_rate": 4.134563212154097e-06,
+      "loss": 0.7959,
+      "step": 762
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.27078914642334,
+      "learning_rate": 4.139989148128052e-06,
+      "loss": 0.617,
+      "step": 763
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.79369854927063,
+      "learning_rate": 4.1454150841020076e-06,
+      "loss": 0.8043,
+      "step": 764
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.28739595413208,
+      "learning_rate": 4.150841020075964e-06,
+      "loss": 0.929,
+      "step": 765
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.379513740539551,
+      "learning_rate": 4.156266956049919e-06,
+      "loss": 0.76,
+      "step": 766
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 5.691065311431885,
+      "learning_rate": 4.161692892023875e-06,
+      "loss": 0.7556,
+      "step": 767
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.1914637088775635,
+      "learning_rate": 4.1671188279978295e-06,
+      "loss": 0.8463,
+      "step": 768
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.207441806793213,
+      "learning_rate": 4.172544763971785e-06,
+      "loss": 0.6804,
+      "step": 769
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 6.778805255889893,
+      "learning_rate": 4.177970699945741e-06,
+      "loss": 0.8812,
+      "step": 770
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 6.756714820861816,
+      "learning_rate": 4.183396635919697e-06,
+      "loss": 0.763,
+      "step": 771
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.452974557876587,
+      "learning_rate": 4.188822571893652e-06,
+      "loss": 0.5614,
+      "step": 772
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.933640241622925,
+      "learning_rate": 4.194248507867608e-06,
+      "loss": 0.6713,
+      "step": 773
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.463021755218506,
+      "learning_rate": 4.199674443841563e-06,
+      "loss": 0.7487,
+      "step": 774
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.363323211669922,
+      "learning_rate": 4.205100379815519e-06,
+      "loss": 0.7641,
+      "step": 775
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 5.951618671417236,
+      "learning_rate": 4.210526315789474e-06,
+      "loss": 0.4909,
+      "step": 776
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.426722526550293,
+      "learning_rate": 4.21595225176343e-06,
+      "loss": 0.6382,
+      "step": 777
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.020448923110962,
+      "learning_rate": 4.221378187737385e-06,
+      "loss": 0.6295,
+      "step": 778
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.450845241546631,
+      "learning_rate": 4.2268041237113405e-06,
+      "loss": 0.6007,
+      "step": 779
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.7759389877319336,
+      "learning_rate": 4.232230059685296e-06,
+      "loss": 0.4306,
+      "step": 780
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 7.129498481750488,
+      "learning_rate": 4.2376559956592515e-06,
+      "loss": 0.7989,
+      "step": 781
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 8.155008316040039,
+      "learning_rate": 4.243081931633207e-06,
+      "loss": 0.7419,
+      "step": 782
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.1365890502929688,
+      "learning_rate": 4.2485078676071624e-06,
+      "loss": 0.5366,
+      "step": 783
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.794005870819092,
+      "learning_rate": 4.253933803581118e-06,
+      "loss": 0.6965,
+      "step": 784
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.6746833324432373,
+      "learning_rate": 4.259359739555073e-06,
+      "loss": 0.6305,
+      "step": 785
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 16.42200469970703,
+      "learning_rate": 4.264785675529029e-06,
+      "loss": 0.7213,
+      "step": 786
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.839621543884277,
+      "learning_rate": 4.270211611502984e-06,
+      "loss": 0.5698,
+      "step": 787
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.8825671672821045,
+      "learning_rate": 4.27563754747694e-06,
+      "loss": 0.7024,
+      "step": 788
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.825300931930542,
+      "learning_rate": 4.281063483450896e-06,
+      "loss": 0.8573,
+      "step": 789
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.338937759399414,
+      "learning_rate": 4.286489419424852e-06,
+      "loss": 0.7754,
+      "step": 790
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 5.879049777984619,
+      "learning_rate": 4.291915355398807e-06,
+      "loss": 0.8019,
+      "step": 791
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 5.414849281311035,
+      "learning_rate": 4.297341291372762e-06,
+      "loss": 0.7278,
+      "step": 792
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.2873919010162354,
+      "learning_rate": 4.302767227346717e-06,
+      "loss": 0.6107,
+      "step": 793
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.3709733486175537,
+      "learning_rate": 4.308193163320673e-06,
+      "loss": 0.9672,
+      "step": 794
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.797672986984253,
+      "learning_rate": 4.313619099294629e-06,
+      "loss": 0.524,
+      "step": 795
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 14.877348899841309,
+      "learning_rate": 4.3190450352685845e-06,
+      "loss": 0.4773,
+      "step": 796
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.5393142700195312,
+      "learning_rate": 4.32447097124254e-06,
+      "loss": 0.8058,
+      "step": 797
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.350130558013916,
+      "learning_rate": 4.329896907216495e-06,
+      "loss": 0.5261,
+      "step": 798
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.752315044403076,
+      "learning_rate": 4.335322843190451e-06,
+      "loss": 0.6303,
+      "step": 799
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 7.792935848236084,
+      "learning_rate": 4.340748779164406e-06,
+      "loss": 0.5894,
+      "step": 800
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.6394410133361816,
+      "learning_rate": 4.346174715138362e-06,
+      "loss": 0.6447,
+      "step": 801
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.1037557125091553,
+      "learning_rate": 4.351600651112317e-06,
+      "loss": 0.6311,
+      "step": 802
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.3567466735839844,
+      "learning_rate": 4.357026587086273e-06,
+      "loss": 0.595,
+      "step": 803
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.4556756019592285,
+      "learning_rate": 4.362452523060228e-06,
+      "loss": 0.5944,
+      "step": 804
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.8539531230926514,
+      "learning_rate": 4.367878459034184e-06,
+      "loss": 0.723,
+      "step": 805
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.8225693702697754,
+      "learning_rate": 4.373304395008139e-06,
+      "loss": 0.544,
+      "step": 806
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.644408702850342,
+      "learning_rate": 4.378730330982095e-06,
+      "loss": 0.6824,
+      "step": 807
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 5.215960502624512,
+      "learning_rate": 4.38415626695605e-06,
+      "loss": 0.548,
+      "step": 808
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.4944775104522705,
+      "learning_rate": 4.389582202930006e-06,
+      "loss": 0.701,
+      "step": 809
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.312279224395752,
+      "learning_rate": 4.395008138903961e-06,
+      "loss": 0.4421,
+      "step": 810
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.0247931480407715,
+      "learning_rate": 4.400434074877917e-06,
+      "loss": 0.7292,
+      "step": 811
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 8.552018165588379,
+      "learning_rate": 4.405860010851872e-06,
+      "loss": 1.0416,
+      "step": 812
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 8.131552696228027,
+      "learning_rate": 4.4112859468258275e-06,
+      "loss": 0.7975,
+      "step": 813
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.283381223678589,
+      "learning_rate": 4.416711882799784e-06,
+      "loss": 0.3943,
+      "step": 814
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.371232509613037,
+      "learning_rate": 4.422137818773739e-06,
+      "loss": 0.6545,
+      "step": 815
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.502869129180908,
+      "learning_rate": 4.427563754747694e-06,
+      "loss": 0.4831,
+      "step": 816
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.036629676818848,
+      "learning_rate": 4.4329896907216494e-06,
+      "loss": 0.8399,
+      "step": 817
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.1447577476501465,
+      "learning_rate": 4.438415626695605e-06,
+      "loss": 0.6102,
+      "step": 818
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 7.4203081130981445,
+      "learning_rate": 4.44384156266956e-06,
+      "loss": 0.9307,
+      "step": 819
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.116830348968506,
+      "learning_rate": 4.449267498643517e-06,
+      "loss": 0.4256,
+      "step": 820
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.942903518676758,
+      "learning_rate": 4.454693434617472e-06,
+      "loss": 0.6828,
+      "step": 821
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 5.02323055267334,
+      "learning_rate": 4.460119370591428e-06,
+      "loss": 0.5474,
+      "step": 822
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.395946502685547,
+      "learning_rate": 4.465545306565383e-06,
+      "loss": 0.3356,
+      "step": 823
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 8.982601165771484,
+      "learning_rate": 4.470971242539339e-06,
+      "loss": 0.6693,
+      "step": 824
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.5975074768066406,
+      "learning_rate": 4.476397178513294e-06,
+      "loss": 0.6471,
+      "step": 825
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.8070173263549805,
+      "learning_rate": 4.4818231144872496e-06,
+      "loss": 0.7088,
+      "step": 826
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.383444786071777,
+      "learning_rate": 4.487249050461205e-06,
+      "loss": 0.7303,
+      "step": 827
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.7668216228485107,
+      "learning_rate": 4.4926749864351605e-06,
+      "loss": 0.7405,
+      "step": 828
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.0867185592651367,
+      "learning_rate": 4.498100922409116e-06,
+      "loss": 0.4097,
+      "step": 829
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.322266578674316,
+      "learning_rate": 4.5035268583830715e-06,
+      "loss": 0.5752,
+      "step": 830
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.3338537216186523,
+      "learning_rate": 4.508952794357027e-06,
+      "loss": 0.4062,
+      "step": 831
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.7063639163970947,
+      "learning_rate": 4.514378730330982e-06,
+      "loss": 0.6439,
+      "step": 832
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.7095694541931152,
+      "learning_rate": 4.519804666304938e-06,
+      "loss": 0.7759,
+      "step": 833
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.3259871006011963,
+      "learning_rate": 4.525230602278893e-06,
+      "loss": 0.6432,
+      "step": 834
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.2152621746063232,
+      "learning_rate": 4.530656538252849e-06,
+      "loss": 0.6368,
+      "step": 835
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.822168827056885,
+      "learning_rate": 4.536082474226804e-06,
+      "loss": 0.576,
+      "step": 836
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.267298460006714,
+      "learning_rate": 4.54150841020076e-06,
+      "loss": 0.8201,
+      "step": 837
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.179721355438232,
+      "learning_rate": 4.546934346174715e-06,
+      "loss": 0.8379,
+      "step": 838
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.195823669433594,
+      "learning_rate": 4.552360282148672e-06,
+      "loss": 0.6778,
+      "step": 839
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.0107595920562744,
+      "learning_rate": 4.557786218122627e-06,
+      "loss": 0.5926,
+      "step": 840
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.113497495651245,
+      "learning_rate": 4.563212154096582e-06,
+      "loss": 0.8589,
+      "step": 841
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.0695724487304688,
+      "learning_rate": 4.568638090070537e-06,
+      "loss": 0.6922,
+      "step": 842
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.2507712841033936,
+      "learning_rate": 4.574064026044493e-06,
+      "loss": 0.3902,
+      "step": 843
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.6303491592407227,
+      "learning_rate": 4.579489962018448e-06,
+      "loss": 0.7807,
+      "step": 844
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.725531578063965,
+      "learning_rate": 4.5849158979924044e-06,
+      "loss": 0.6712,
+      "step": 845
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.725103378295898,
+      "learning_rate": 4.59034183396636e-06,
+      "loss": 0.5069,
+      "step": 846
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.343108892440796,
+      "learning_rate": 4.595767769940315e-06,
+      "loss": 0.7059,
+      "step": 847
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.9581310749053955,
+      "learning_rate": 4.601193705914271e-06,
+      "loss": 0.4683,
+      "step": 848
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.789189338684082,
+      "learning_rate": 4.6066196418882255e-06,
+      "loss": 0.6233,
+      "step": 849
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 5.69543981552124,
+      "learning_rate": 4.612045577862182e-06,
+      "loss": 0.473,
+      "step": 850
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.84273099899292,
+      "learning_rate": 4.617471513836137e-06,
+      "loss": 0.7282,
+      "step": 851
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.8862271308898926,
+      "learning_rate": 4.622897449810093e-06,
+      "loss": 0.8111,
+      "step": 852
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 6.219770431518555,
+      "learning_rate": 4.628323385784048e-06,
+      "loss": 0.6959,
+      "step": 853
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.4632461071014404,
+      "learning_rate": 4.633749321758004e-06,
+      "loss": 0.6583,
+      "step": 854
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.928844928741455,
+      "learning_rate": 4.639175257731959e-06,
+      "loss": 0.6998,
+      "step": 855
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.2666690349578857,
+      "learning_rate": 4.644601193705915e-06,
+      "loss": 0.5223,
+      "step": 856
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.4983229637146,
+      "learning_rate": 4.65002712967987e-06,
+      "loss": 0.5665,
+      "step": 857
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 27.633943557739258,
+      "learning_rate": 4.655453065653826e-06,
+      "loss": 0.4826,
+      "step": 858
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.1238951683044434,
+      "learning_rate": 4.660879001627781e-06,
+      "loss": 0.4423,
+      "step": 859
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 5.803679466247559,
+      "learning_rate": 4.6663049376017366e-06,
+      "loss": 0.6378,
+      "step": 860
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.6100168228149414,
+      "learning_rate": 4.671730873575692e-06,
+      "loss": 0.6688,
+      "step": 861
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.3023622035980225,
+      "learning_rate": 4.6771568095496475e-06,
+      "loss": 0.542,
+      "step": 862
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.373150587081909,
+      "learning_rate": 4.682582745523603e-06,
+      "loss": 0.7246,
+      "step": 863
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.5568575859069824,
+      "learning_rate": 4.688008681497559e-06,
+      "loss": 0.4183,
+      "step": 864
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.2585971355438232,
+      "learning_rate": 4.693434617471514e-06,
+      "loss": 0.4144,
+      "step": 865
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.4728810787200928,
+      "learning_rate": 4.698860553445469e-06,
+      "loss": 0.7077,
+      "step": 866
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 5.078420639038086,
+      "learning_rate": 4.704286489419425e-06,
+      "loss": 0.4062,
+      "step": 867
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.323285102844238,
+      "learning_rate": 4.70971242539338e-06,
+      "loss": 0.6701,
+      "step": 868
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.1915574073791504,
+      "learning_rate": 4.715138361367336e-06,
+      "loss": 0.4704,
+      "step": 869
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.899224281311035,
+      "learning_rate": 4.720564297341292e-06,
+      "loss": 0.7814,
+      "step": 870
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.9451205730438232,
+      "learning_rate": 4.725990233315248e-06,
+      "loss": 0.5384,
+      "step": 871
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.9175024032592773,
+      "learning_rate": 4.731416169289203e-06,
+      "loss": 0.7567,
+      "step": 872
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.8251075744628906,
+      "learning_rate": 4.736842105263158e-06,
+      "loss": 0.2792,
+      "step": 873
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.623182773590088,
+      "learning_rate": 4.742268041237113e-06,
+      "loss": 0.7475,
+      "step": 874
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 6.4025468826293945,
+      "learning_rate": 4.7476939772110695e-06,
+      "loss": 0.6443,
+      "step": 875
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.206711530685425,
+      "learning_rate": 4.753119913185025e-06,
+      "loss": 0.7049,
+      "step": 876
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.9310905933380127,
+      "learning_rate": 4.7585458491589805e-06,
+      "loss": 0.6323,
+      "step": 877
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.2996132373809814,
+      "learning_rate": 4.763971785132936e-06,
+      "loss": 0.5707,
+      "step": 878
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 7.368367671966553,
+      "learning_rate": 4.7693977211068914e-06,
+      "loss": 0.7056,
+      "step": 879
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.714535713195801,
+      "learning_rate": 4.774823657080847e-06,
+      "loss": 0.3978,
+      "step": 880
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 5.501245975494385,
+      "learning_rate": 4.780249593054802e-06,
+      "loss": 0.5281,
+      "step": 881
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.6306827068328857,
+      "learning_rate": 4.785675529028758e-06,
+      "loss": 0.7321,
+      "step": 882
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.6908600330352783,
+      "learning_rate": 4.791101465002713e-06,
+      "loss": 0.535,
+      "step": 883
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 5.707081317901611,
+      "learning_rate": 4.796527400976669e-06,
+      "loss": 0.7201,
+      "step": 884
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.062439918518066,
+      "learning_rate": 4.801953336950624e-06,
+      "loss": 0.8054,
+      "step": 885
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.935929775238037,
+      "learning_rate": 4.80737927292458e-06,
+      "loss": 0.6339,
+      "step": 886
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.510267734527588,
+      "learning_rate": 4.812805208898535e-06,
+      "loss": 0.5249,
+      "step": 887
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 9.213675498962402,
+      "learning_rate": 4.818231144872491e-06,
+      "loss": 0.6645,
+      "step": 888
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.4570794105529785,
+      "learning_rate": 4.823657080846447e-06,
+      "loss": 0.6403,
+      "step": 889
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.818037986755371,
+      "learning_rate": 4.829083016820402e-06,
+      "loss": 0.5603,
+      "step": 890
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 5.980953693389893,
+      "learning_rate": 4.834508952794357e-06,
+      "loss": 0.8698,
+      "step": 891
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 6.694294452667236,
+      "learning_rate": 4.839934888768313e-06,
+      "loss": 0.645,
+      "step": 892
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.459018230438232,
+      "learning_rate": 4.845360824742268e-06,
+      "loss": 0.9187,
+      "step": 893
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.6163034439086914,
+      "learning_rate": 4.850786760716224e-06,
+      "loss": 0.6423,
+      "step": 894
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 5.394880771636963,
+      "learning_rate": 4.85621269669018e-06,
+      "loss": 0.6548,
+      "step": 895
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.767407178878784,
+      "learning_rate": 4.861638632664135e-06,
+      "loss": 0.8054,
+      "step": 896
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.2701334953308105,
+      "learning_rate": 4.867064568638091e-06,
+      "loss": 0.5139,
+      "step": 897
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.611332893371582,
+      "learning_rate": 4.8724905046120455e-06,
+      "loss": 0.6821,
+      "step": 898
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.9304702281951904,
+      "learning_rate": 4.877916440586001e-06,
+      "loss": 0.51,
+      "step": 899
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.970105171203613,
+      "learning_rate": 4.883342376559957e-06,
+      "loss": 0.459,
+      "step": 900
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 7.6786298751831055,
+      "learning_rate": 4.888768312533913e-06,
+      "loss": 0.6359,
+      "step": 901
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.7789368629455566,
+      "learning_rate": 4.894194248507868e-06,
+      "loss": 0.4543,
+      "step": 902
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.5078048706054688,
+      "learning_rate": 4.899620184481824e-06,
+      "loss": 0.7577,
+      "step": 903
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.8860385417938232,
+      "learning_rate": 4.905046120455779e-06,
+      "loss": 0.6852,
+      "step": 904
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.5292203426361084,
+      "learning_rate": 4.910472056429735e-06,
+      "loss": 0.6998,
+      "step": 905
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.797824382781982,
+      "learning_rate": 4.91589799240369e-06,
+      "loss": 0.7208,
+      "step": 906
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 10.600814819335938,
+      "learning_rate": 4.921323928377646e-06,
+      "loss": 0.6613,
+      "step": 907
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.6808223724365234,
+      "learning_rate": 4.926749864351601e-06,
+      "loss": 0.7154,
+      "step": 908
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.2864975929260254,
+      "learning_rate": 4.9321758003255565e-06,
+      "loss": 0.7308,
+      "step": 909
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.434258222579956,
+      "learning_rate": 4.937601736299512e-06,
+      "loss": 0.7392,
+      "step": 910
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.2451131343841553,
+      "learning_rate": 4.9430276722734675e-06,
+      "loss": 0.6917,
+      "step": 911
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.7815537452697754,
+      "learning_rate": 4.948453608247423e-06,
+      "loss": 0.5964,
+      "step": 912
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.1892220973968506,
+      "learning_rate": 4.9538795442213784e-06,
+      "loss": 0.5332,
+      "step": 913
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 6.466815948486328,
+      "learning_rate": 4.959305480195334e-06,
+      "loss": 0.626,
+      "step": 914
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.7657382488250732,
+      "learning_rate": 4.964731416169289e-06,
+      "loss": 0.585,
+      "step": 915
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.838573932647705,
+      "learning_rate": 4.970157352143245e-06,
+      "loss": 0.7488,
+      "step": 916
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.375037670135498,
+      "learning_rate": 4.9755832881172e-06,
+      "loss": 0.8012,
+      "step": 917
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.4752655029296875,
+      "learning_rate": 4.981009224091156e-06,
+      "loss": 0.6792,
+      "step": 918
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.145828723907471,
+      "learning_rate": 4.986435160065112e-06,
+      "loss": 0.6382,
+      "step": 919
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.470599412918091,
+      "learning_rate": 4.991861096039068e-06,
+      "loss": 0.2694,
+      "step": 920
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.795828104019165,
+      "learning_rate": 4.997287032013023e-06,
+      "loss": 0.6174,
+      "step": 921
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.2545642852783203,
+      "learning_rate": 5.002712967986978e-06,
+      "loss": 0.6911,
+      "step": 922
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 7.837539196014404,
+      "learning_rate": 5.008138903960934e-06,
+      "loss": 0.6055,
+      "step": 923
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.1727139949798584,
+      "learning_rate": 5.013564839934889e-06,
+      "loss": 0.8335,
+      "step": 924
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.7007558345794678,
+      "learning_rate": 5.018990775908845e-06,
+      "loss": 0.8044,
+      "step": 925
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.6702747344970703,
+      "learning_rate": 5.0244167118828005e-06,
+      "loss": 0.599,
+      "step": 926
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.2851052284240723,
+      "learning_rate": 5.029842647856755e-06,
+      "loss": 0.7615,
+      "step": 927
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.9072768688201904,
+      "learning_rate": 5.035268583830711e-06,
+      "loss": 0.5969,
+      "step": 928
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.721240758895874,
+      "learning_rate": 5.040694519804666e-06,
+      "loss": 0.8403,
+      "step": 929
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 4.2119059562683105,
+      "learning_rate": 5.046120455778622e-06,
+      "loss": 0.8036,
+      "step": 930
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.143915891647339,
+      "learning_rate": 5.051546391752578e-06,
+      "loss": 0.5627,
+      "step": 931
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.2761573791503906,
+      "learning_rate": 5.056972327726533e-06,
+      "loss": 0.5732,
+      "step": 932
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.4594907760620117,
+      "learning_rate": 5.062398263700489e-06,
+      "loss": 0.4011,
+      "step": 933
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 4.818331718444824,
+      "learning_rate": 5.0678241996744434e-06,
+      "loss": 0.7867,
+      "step": 934
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 4.317580223083496,
+      "learning_rate": 5.0732501356484e-06,
+      "loss": 0.5554,
+      "step": 935
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.7615418434143066,
+      "learning_rate": 5.078676071622355e-06,
+      "loss": 0.7891,
+      "step": 936
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.041290283203125,
+      "learning_rate": 5.084102007596311e-06,
+      "loss": 0.7339,
+      "step": 937
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.7969727516174316,
+      "learning_rate": 5.089527943570266e-06,
+      "loss": 0.4355,
+      "step": 938
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 6.072770595550537,
+      "learning_rate": 5.0949538795442225e-06,
+      "loss": 0.8728,
+      "step": 939
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.5737686157226562,
+      "learning_rate": 5.100379815518177e-06,
+      "loss": 0.5538,
+      "step": 940
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.259178876876831,
+      "learning_rate": 5.1058057514921334e-06,
+      "loss": 0.7582,
+      "step": 941
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.3195197582244873,
+      "learning_rate": 5.111231687466088e-06,
+      "loss": 0.6451,
+      "step": 942
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 5.140257835388184,
+      "learning_rate": 5.1166576234400435e-06,
+      "loss": 0.5023,
+      "step": 943
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.4777350425720215,
+      "learning_rate": 5.122083559414e-06,
+      "loss": 0.4194,
+      "step": 944
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.057615041732788,
+      "learning_rate": 5.1275094953879545e-06,
+      "loss": 0.6877,
+      "step": 945
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 6.439277648925781,
+      "learning_rate": 5.132935431361911e-06,
+      "loss": 0.6504,
+      "step": 946
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.5714573860168457,
+      "learning_rate": 5.1383613673358654e-06,
+      "loss": 0.6014,
+      "step": 947
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 7.274555206298828,
+      "learning_rate": 5.143787303309822e-06,
+      "loss": 0.4928,
+      "step": 948
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.8445560932159424,
+      "learning_rate": 5.149213239283776e-06,
+      "loss": 0.5271,
+      "step": 949
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.688668727874756,
+      "learning_rate": 5.154639175257732e-06,
+      "loss": 0.6744,
+      "step": 950
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.4404191970825195,
+      "learning_rate": 5.160065111231688e-06,
+      "loss": 0.4221,
+      "step": 951
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.897892713546753,
+      "learning_rate": 5.165491047205643e-06,
+      "loss": 0.4898,
+      "step": 952
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.0412979125976562,
+      "learning_rate": 5.170916983179599e-06,
+      "loss": 0.5941,
+      "step": 953
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.799191474914551,
+      "learning_rate": 5.176342919153554e-06,
+      "loss": 0.628,
+      "step": 954
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 4.111303329467773,
+      "learning_rate": 5.18176885512751e-06,
+      "loss": 0.728,
+      "step": 955
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 7.789258003234863,
+      "learning_rate": 5.1871947911014656e-06,
+      "loss": 0.6343,
+      "step": 956
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.773531198501587,
+      "learning_rate": 5.192620727075421e-06,
+      "loss": 0.6305,
+      "step": 957
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 4.984939098358154,
+      "learning_rate": 5.1980466630493765e-06,
+      "loss": 0.7891,
+      "step": 958
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 4.911314487457275,
+      "learning_rate": 5.203472599023331e-06,
+      "loss": 0.6506,
+      "step": 959
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.2919745445251465,
+      "learning_rate": 5.2088985349972875e-06,
+      "loss": 0.8109,
+      "step": 960
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.8731908798217773,
+      "learning_rate": 5.214324470971243e-06,
+      "loss": 0.5223,
+      "step": 961
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 4.296701431274414,
+      "learning_rate": 5.219750406945198e-06,
+      "loss": 0.4801,
+      "step": 962
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.2811548709869385,
+      "learning_rate": 5.225176342919154e-06,
+      "loss": 0.4953,
+      "step": 963
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.871769905090332,
+      "learning_rate": 5.23060227889311e-06,
+      "loss": 0.5501,
+      "step": 964
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.2494380474090576,
+      "learning_rate": 5.236028214867065e-06,
+      "loss": 0.7161,
+      "step": 965
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 5.040472507476807,
+      "learning_rate": 5.241454150841021e-06,
+      "loss": 0.8141,
+      "step": 966
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.2697200775146484,
+      "learning_rate": 5.246880086814976e-06,
+      "loss": 0.5392,
+      "step": 967
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.365818977355957,
+      "learning_rate": 5.252306022788931e-06,
+      "loss": 0.6396,
+      "step": 968
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.2329885959625244,
+      "learning_rate": 5.257731958762888e-06,
+      "loss": 0.4158,
+      "step": 969
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.7800276279449463,
+      "learning_rate": 5.263157894736842e-06,
+      "loss": 0.6316,
+      "step": 970
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.386408805847168,
+      "learning_rate": 5.2685838307107985e-06,
+      "loss": 0.7363,
+      "step": 971
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.5896975994110107,
+      "learning_rate": 5.274009766684753e-06,
+      "loss": 0.9082,
+      "step": 972
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.316164970397949,
+      "learning_rate": 5.2794357026587095e-06,
+      "loss": 0.7487,
+      "step": 973
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.8726134300231934,
+      "learning_rate": 5.284861638632665e-06,
+      "loss": 0.7897,
+      "step": 974
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.907311201095581,
+      "learning_rate": 5.29028757460662e-06,
+      "loss": 0.6278,
+      "step": 975
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.261852264404297,
+      "learning_rate": 5.295713510580576e-06,
+      "loss": 0.7576,
+      "step": 976
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.6784780025482178,
+      "learning_rate": 5.3011394465545305e-06,
+      "loss": 0.8284,
+      "step": 977
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 5.913921356201172,
+      "learning_rate": 5.306565382528487e-06,
+      "loss": 0.6854,
+      "step": 978
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 4.079206466674805,
+      "learning_rate": 5.3119913185024415e-06,
+      "loss": 0.5942,
+      "step": 979
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.1555593013763428,
+      "learning_rate": 5.317417254476398e-06,
+      "loss": 0.5841,
+      "step": 980
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.273078441619873,
+      "learning_rate": 5.322843190450353e-06,
+      "loss": 0.5848,
+      "step": 981
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 5.945539474487305,
+      "learning_rate": 5.328269126424309e-06,
+      "loss": 0.4274,
+      "step": 982
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.578887939453125,
+      "learning_rate": 5.333695062398264e-06,
+      "loss": 0.2532,
+      "step": 983
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.625138521194458,
+      "learning_rate": 5.339120998372219e-06,
+      "loss": 0.49,
+      "step": 984
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 4.376350402832031,
+      "learning_rate": 5.344546934346175e-06,
+      "loss": 0.7198,
+      "step": 985
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 5.992015838623047,
+      "learning_rate": 5.349972870320131e-06,
+      "loss": 0.5928,
+      "step": 986
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 8.179695129394531,
+      "learning_rate": 5.355398806294086e-06,
+      "loss": 0.5535,
+      "step": 987
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 5.4003214836120605,
+      "learning_rate": 5.360824742268042e-06,
+      "loss": 0.5701,
+      "step": 988
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.7540395259857178,
+      "learning_rate": 5.366250678241998e-06,
+      "loss": 0.6282,
+      "step": 989
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.5697216987609863,
+      "learning_rate": 5.3716766142159526e-06,
+      "loss": 0.7583,
+      "step": 990
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.545787811279297,
+      "learning_rate": 5.377102550189908e-06,
+      "loss": 0.7029,
+      "step": 991
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.65226674079895,
+      "learning_rate": 5.3825284861638635e-06,
+      "loss": 0.4493,
+      "step": 992
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.0949764251708984,
+      "learning_rate": 5.387954422137819e-06,
+      "loss": 0.5254,
+      "step": 993
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 4.386163711547852,
+      "learning_rate": 5.393380358111775e-06,
+      "loss": 0.5282,
+      "step": 994
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.624309539794922,
+      "learning_rate": 5.39880629408573e-06,
+      "loss": 0.6347,
+      "step": 995
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.8887248039245605,
+      "learning_rate": 5.404232230059686e-06,
+      "loss": 0.4345,
+      "step": 996
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.199934720993042,
+      "learning_rate": 5.409658166033641e-06,
+      "loss": 0.9015,
+      "step": 997
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 4.7688212394714355,
+      "learning_rate": 5.415084102007597e-06,
+      "loss": 0.5475,
+      "step": 998
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 4.1086859703063965,
+      "learning_rate": 5.420510037981553e-06,
+      "loss": 0.7079,
+      "step": 999
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.125293493270874,
+      "learning_rate": 5.425935973955507e-06,
+      "loss": 0.8196,
+      "step": 1000
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 15.194843292236328,
+      "learning_rate": 5.431361909929464e-06,
+      "loss": 0.7716,
+      "step": 1001
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 5.1534318923950195,
+      "learning_rate": 5.436787845903418e-06,
+      "loss": 0.7589,
+      "step": 1002
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 7.023813247680664,
+      "learning_rate": 5.442213781877375e-06,
+      "loss": 0.577,
+      "step": 1003
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.8910460472106934,
+      "learning_rate": 5.447639717851329e-06,
+      "loss": 0.5365,
+      "step": 1004
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 4.217067241668701,
+      "learning_rate": 5.4530656538252855e-06,
+      "loss": 0.6054,
+      "step": 1005
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.57936429977417,
+      "learning_rate": 5.458491589799241e-06,
+      "loss": 0.645,
+      "step": 1006
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.434431314468384,
+      "learning_rate": 5.463917525773196e-06,
+      "loss": 0.4907,
+      "step": 1007
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.0243542194366455,
+      "learning_rate": 5.469343461747152e-06,
+      "loss": 0.8411,
+      "step": 1008
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.279352903366089,
+      "learning_rate": 5.474769397721107e-06,
+      "loss": 0.4921,
+      "step": 1009
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.886573314666748,
+      "learning_rate": 5.480195333695063e-06,
+      "loss": 0.5686,
+      "step": 1010
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.525362730026245,
+      "learning_rate": 5.485621269669018e-06,
+      "loss": 0.7517,
+      "step": 1011
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.814878225326538,
+      "learning_rate": 5.491047205642974e-06,
+      "loss": 0.6901,
+      "step": 1012
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.8291828632354736,
+      "learning_rate": 5.496473141616929e-06,
+      "loss": 0.4056,
+      "step": 1013
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 7.255096435546875,
+      "learning_rate": 5.501899077590886e-06,
+      "loss": 0.5623,
+      "step": 1014
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.671079397201538,
+      "learning_rate": 5.50732501356484e-06,
+      "loss": 0.4147,
+      "step": 1015
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.6801934242248535,
+      "learning_rate": 5.512750949538796e-06,
+      "loss": 0.8386,
+      "step": 1016
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.33061146736145,
+      "learning_rate": 5.518176885512751e-06,
+      "loss": 0.6409,
+      "step": 1017
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.4882287979125977,
+      "learning_rate": 5.523602821486707e-06,
+      "loss": 0.5734,
+      "step": 1018
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.1629364490509033,
+      "learning_rate": 5.529028757460663e-06,
+      "loss": 0.7125,
+      "step": 1019
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.084812879562378,
+      "learning_rate": 5.534454693434618e-06,
+      "loss": 0.342,
+      "step": 1020
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.927363872528076,
+      "learning_rate": 5.539880629408574e-06,
+      "loss": 0.6896,
+      "step": 1021
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 6.810591697692871,
+      "learning_rate": 5.545306565382529e-06,
+      "loss": 0.5127,
+      "step": 1022
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.0228190422058105,
+      "learning_rate": 5.550732501356484e-06,
+      "loss": 0.5156,
+      "step": 1023
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 4.034255027770996,
+      "learning_rate": 5.55615843733044e-06,
+      "loss": 0.5598,
+      "step": 1024
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.663712739944458,
+      "learning_rate": 5.561584373304395e-06,
+      "loss": 0.6655,
+      "step": 1025
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.6595386266708374,
+      "learning_rate": 5.567010309278351e-06,
+      "loss": 0.3943,
+      "step": 1026
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.686746835708618,
+      "learning_rate": 5.572436245252306e-06,
+      "loss": 0.5234,
+      "step": 1027
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 5.864205360412598,
+      "learning_rate": 5.577862181226262e-06,
+      "loss": 0.6837,
+      "step": 1028
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.6032421588897705,
+      "learning_rate": 5.583288117200217e-06,
+      "loss": 0.7976,
+      "step": 1029
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.5887198448181152,
+      "learning_rate": 5.588714053174173e-06,
+      "loss": 0.6589,
+      "step": 1030
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.4341421127319336,
+      "learning_rate": 5.594139989148129e-06,
+      "loss": 0.6009,
+      "step": 1031
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 5.691181659698486,
+      "learning_rate": 5.599565925122083e-06,
+      "loss": 0.4671,
+      "step": 1032
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 5.243782043457031,
+      "learning_rate": 5.60499186109604e-06,
+      "loss": 0.7647,
+      "step": 1033
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.2113559246063232,
+      "learning_rate": 5.610417797069994e-06,
+      "loss": 0.8182,
+      "step": 1034
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.301596164703369,
+      "learning_rate": 5.615843733043951e-06,
+      "loss": 0.5645,
+      "step": 1035
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.0480844974517822,
+      "learning_rate": 5.621269669017906e-06,
+      "loss": 0.4979,
+      "step": 1036
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.8702404499053955,
+      "learning_rate": 5.626695604991862e-06,
+      "loss": 0.6489,
+      "step": 1037
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.3505444526672363,
+      "learning_rate": 5.632121540965817e-06,
+      "loss": 0.5682,
+      "step": 1038
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 14.890772819519043,
+      "learning_rate": 5.637547476939773e-06,
+      "loss": 0.5755,
+      "step": 1039
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.6638479232788086,
+      "learning_rate": 5.642973412913728e-06,
+      "loss": 0.717,
+      "step": 1040
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.9236245155334473,
+      "learning_rate": 5.6483993488876835e-06,
+      "loss": 0.723,
+      "step": 1041
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 4.328904628753662,
+      "learning_rate": 5.653825284861639e-06,
+      "loss": 0.7505,
+      "step": 1042
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.9011709690093994,
+      "learning_rate": 5.6592512208355944e-06,
+      "loss": 0.5609,
+      "step": 1043
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 4.281935214996338,
+      "learning_rate": 5.664677156809551e-06,
+      "loss": 0.6863,
+      "step": 1044
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.855281114578247,
+      "learning_rate": 5.670103092783505e-06,
+      "loss": 0.5035,
+      "step": 1045
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 4.194986820220947,
+      "learning_rate": 5.675529028757462e-06,
+      "loss": 0.6814,
+      "step": 1046
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 5.076362133026123,
+      "learning_rate": 5.680954964731416e-06,
+      "loss": 0.5674,
+      "step": 1047
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.4739065170288086,
+      "learning_rate": 5.686380900705372e-06,
+      "loss": 0.6944,
+      "step": 1048
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 6.045414447784424,
+      "learning_rate": 5.691806836679328e-06,
+      "loss": 0.5535,
+      "step": 1049
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 4.8709869384765625,
+      "learning_rate": 5.697232772653283e-06,
+      "loss": 0.6559,
+      "step": 1050
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 4.769818305969238,
+      "learning_rate": 5.702658708627239e-06,
+      "loss": 0.6484,
+      "step": 1051
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.222123622894287,
+      "learning_rate": 5.708084644601194e-06,
+      "loss": 0.5716,
+      "step": 1052
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.9934520721435547,
+      "learning_rate": 5.71351058057515e-06,
+      "loss": 0.4692,
+      "step": 1053
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.642552137374878,
+      "learning_rate": 5.718936516549105e-06,
+      "loss": 0.6336,
+      "step": 1054
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.5027830600738525,
+      "learning_rate": 5.724362452523061e-06,
+      "loss": 0.4213,
+      "step": 1055
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.8761857748031616,
+      "learning_rate": 5.7297883884970165e-06,
+      "loss": 0.734,
+      "step": 1056
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 7.075722694396973,
+      "learning_rate": 5.735214324470971e-06,
+      "loss": 0.6836,
+      "step": 1057
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 14.917378425598145,
+      "learning_rate": 5.7406402604449274e-06,
+      "loss": 0.6506,
+      "step": 1058
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.7742667198181152,
+      "learning_rate": 5.746066196418882e-06,
+      "loss": 0.4602,
+      "step": 1059
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.2447128295898438,
+      "learning_rate": 5.751492132392838e-06,
+      "loss": 0.6733,
+      "step": 1060
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.779898166656494,
+      "learning_rate": 5.756918068366794e-06,
+      "loss": 0.496,
+      "step": 1061
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 7.912144184112549,
+      "learning_rate": 5.762344004340749e-06,
+      "loss": 0.6841,
+      "step": 1062
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.738574266433716,
+      "learning_rate": 5.767769940314705e-06,
+      "loss": 0.525,
+      "step": 1063
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 5.774019718170166,
+      "learning_rate": 5.7731958762886594e-06,
+      "loss": 0.6278,
+      "step": 1064
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.45694637298584,
+      "learning_rate": 5.778621812262616e-06,
+      "loss": 0.5328,
+      "step": 1065
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 4.132791042327881,
+      "learning_rate": 5.784047748236571e-06,
+      "loss": 0.7961,
+      "step": 1066
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.002523422241211,
+      "learning_rate": 5.789473684210527e-06,
+      "loss": 0.4263,
+      "step": 1067
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 5.823241233825684,
+      "learning_rate": 5.794899620184482e-06,
+      "loss": 0.7751,
+      "step": 1068
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.247157335281372,
+      "learning_rate": 5.8003255561584385e-06,
+      "loss": 0.6931,
+      "step": 1069
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 4.318613529205322,
+      "learning_rate": 5.805751492132393e-06,
+      "loss": 0.6061,
+      "step": 1070
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.133953332901001,
+      "learning_rate": 5.8111774281063494e-06,
+      "loss": 0.7953,
+      "step": 1071
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.5063648223876953,
+      "learning_rate": 5.816603364080304e-06,
+      "loss": 0.6003,
+      "step": 1072
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 6.553199291229248,
+      "learning_rate": 5.8220293000542596e-06,
+      "loss": 0.6977,
+      "step": 1073
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.0796878337860107,
+      "learning_rate": 5.827455236028216e-06,
+      "loss": 0.6676,
+      "step": 1074
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.9613583087921143,
+      "learning_rate": 5.8328811720021705e-06,
+      "loss": 0.8029,
+      "step": 1075
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.706794261932373,
+      "learning_rate": 5.838307107976127e-06,
+      "loss": 0.5434,
+      "step": 1076
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.5657103061676025,
+      "learning_rate": 5.8437330439500815e-06,
+      "loss": 0.5693,
+      "step": 1077
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.177999258041382,
+      "learning_rate": 5.849158979924038e-06,
+      "loss": 0.4933,
+      "step": 1078
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 4.150742530822754,
+      "learning_rate": 5.854584915897993e-06,
+      "loss": 0.8401,
+      "step": 1079
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.9067752361297607,
+      "learning_rate": 5.860010851871948e-06,
+      "loss": 0.323,
+      "step": 1080
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 6.660613059997559,
+      "learning_rate": 5.865436787845904e-06,
+      "loss": 0.6979,
+      "step": 1081
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 5.292941570281982,
+      "learning_rate": 5.870862723819859e-06,
+      "loss": 0.6922,
+      "step": 1082
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 5.279029846191406,
+      "learning_rate": 5.876288659793815e-06,
+      "loss": 0.7976,
+      "step": 1083
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 4.702332973480225,
+      "learning_rate": 5.88171459576777e-06,
+      "loss": 0.8171,
+      "step": 1084
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.1755917072296143,
+      "learning_rate": 5.887140531741726e-06,
+      "loss": 0.7644,
+      "step": 1085
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 4.543763160705566,
+      "learning_rate": 5.8925664677156816e-06,
+      "loss": 0.5208,
+      "step": 1086
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 6.596467971801758,
+      "learning_rate": 5.897992403689637e-06,
+      "loss": 0.6832,
+      "step": 1087
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 4.444796085357666,
+      "learning_rate": 5.9034183396635925e-06,
+      "loss": 0.7684,
+      "step": 1088
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.0946319103240967,
+      "learning_rate": 5.908844275637547e-06,
+      "loss": 0.5673,
+      "step": 1089
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 7.804619312286377,
+      "learning_rate": 5.9142702116115035e-06,
+      "loss": 0.7015,
+      "step": 1090
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.8358397483825684,
+      "learning_rate": 5.919696147585459e-06,
+      "loss": 0.6708,
+      "step": 1091
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.2248034477233887,
+      "learning_rate": 5.9251220835594144e-06,
+      "loss": 0.457,
+      "step": 1092
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.7658538818359375,
+      "learning_rate": 5.93054801953337e-06,
+      "loss": 0.4876,
+      "step": 1093
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.9382903575897217,
+      "learning_rate": 5.935973955507326e-06,
+      "loss": 0.5735,
+      "step": 1094
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.3250603675842285,
+      "learning_rate": 5.941399891481281e-06,
+      "loss": 0.624,
+      "step": 1095
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 8.46684741973877,
+      "learning_rate": 5.946825827455237e-06,
+      "loss": 0.7118,
+      "step": 1096
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.778815507888794,
+      "learning_rate": 5.952251763429192e-06,
+      "loss": 0.5428,
+      "step": 1097
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.020547389984131,
+      "learning_rate": 5.957677699403147e-06,
+      "loss": 0.7303,
+      "step": 1098
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.674469232559204,
+      "learning_rate": 5.963103635377104e-06,
+      "loss": 0.9148,
+      "step": 1099
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 4.323952674865723,
+      "learning_rate": 5.968529571351058e-06,
+      "loss": 0.7202,
+      "step": 1100
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.617783546447754,
+      "learning_rate": 5.9739555073250145e-06,
+      "loss": 0.394,
+      "step": 1101
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.4379208087921143,
+      "learning_rate": 5.979381443298969e-06,
+      "loss": 0.8167,
+      "step": 1102
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.350220203399658,
+      "learning_rate": 5.9848073792729255e-06,
+      "loss": 0.5446,
+      "step": 1103
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 9.553943634033203,
+      "learning_rate": 5.990233315246881e-06,
+      "loss": 0.6818,
+      "step": 1104
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.431882381439209,
+      "learning_rate": 5.995659251220836e-06,
+      "loss": 0.505,
+      "step": 1105
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.2992334365844727,
+      "learning_rate": 6.001085187194792e-06,
+      "loss": 0.599,
+      "step": 1106
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 4.481942653656006,
+      "learning_rate": 6.0065111231687466e-06,
+      "loss": 0.6444,
+      "step": 1107
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.1871204376220703,
+      "learning_rate": 6.011937059142703e-06,
+      "loss": 0.6134,
+      "step": 1108
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.2505013942718506,
+      "learning_rate": 6.0173629951166575e-06,
+      "loss": 0.5266,
+      "step": 1109
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 5.2910356521606445,
+      "learning_rate": 6.022788931090614e-06,
+      "loss": 0.7897,
+      "step": 1110
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.6516878604888916,
+      "learning_rate": 6.028214867064569e-06,
+      "loss": 0.7768,
+      "step": 1111
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.201444625854492,
+      "learning_rate": 6.033640803038525e-06,
+      "loss": 0.4566,
+      "step": 1112
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.49961256980896,
+      "learning_rate": 6.03906673901248e-06,
+      "loss": 0.6047,
+      "step": 1113
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 4.760199546813965,
+      "learning_rate": 6.044492674986435e-06,
+      "loss": 0.6276,
+      "step": 1114
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 7.341987133026123,
+      "learning_rate": 6.049918610960391e-06,
+      "loss": 0.5782,
+      "step": 1115
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.3826181888580322,
+      "learning_rate": 6.055344546934347e-06,
+      "loss": 0.599,
+      "step": 1116
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 5.937858581542969,
+      "learning_rate": 6.060770482908302e-06,
+      "loss": 0.7963,
+      "step": 1117
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 6.149051189422607,
+      "learning_rate": 6.066196418882258e-06,
+      "loss": 0.5509,
+      "step": 1118
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.0571579933166504,
+      "learning_rate": 6.071622354856214e-06,
+      "loss": 0.7111,
+      "step": 1119
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.514007091522217,
+      "learning_rate": 6.077048290830169e-06,
+      "loss": 0.5589,
+      "step": 1120
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.4780893325805664,
+      "learning_rate": 6.082474226804124e-06,
+      "loss": 0.5968,
+      "step": 1121
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.8979148864746094,
+      "learning_rate": 6.0879001627780795e-06,
+      "loss": 0.7154,
+      "step": 1122
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 6.631845951080322,
+      "learning_rate": 6.093326098752035e-06,
+      "loss": 0.5347,
+      "step": 1123
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 4.053477764129639,
+      "learning_rate": 6.098752034725991e-06,
+      "loss": 0.3978,
+      "step": 1124
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.8863768577575684,
+      "learning_rate": 6.104177970699946e-06,
+      "loss": 0.5315,
+      "step": 1125
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 5.053970813751221,
+      "learning_rate": 6.109603906673902e-06,
+      "loss": 0.6602,
+      "step": 1126
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.9914281368255615,
+      "learning_rate": 6.115029842647857e-06,
+      "loss": 0.6807,
+      "step": 1127
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 7.713101863861084,
+      "learning_rate": 6.120455778621813e-06,
+      "loss": 0.3631,
+      "step": 1128
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.1445531845092773,
+      "learning_rate": 6.125881714595769e-06,
+      "loss": 0.7357,
+      "step": 1129
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.6748316287994385,
+      "learning_rate": 6.131307650569723e-06,
+      "loss": 0.5448,
+      "step": 1130
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 5.355565071105957,
+      "learning_rate": 6.13673358654368e-06,
+      "loss": 0.633,
+      "step": 1131
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.8665246963500977,
+      "learning_rate": 6.142159522517634e-06,
+      "loss": 0.7961,
+      "step": 1132
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 5.300385475158691,
+      "learning_rate": 6.147585458491591e-06,
+      "loss": 0.5167,
+      "step": 1133
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 10.268411636352539,
+      "learning_rate": 6.153011394465545e-06,
+      "loss": 0.7822,
+      "step": 1134
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.020124673843384,
+      "learning_rate": 6.1584373304395016e-06,
+      "loss": 0.8636,
+      "step": 1135
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.9787023067474365,
+      "learning_rate": 6.163863266413457e-06,
+      "loss": 0.9103,
+      "step": 1136
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 9.243511199951172,
+      "learning_rate": 6.169289202387412e-06,
+      "loss": 0.627,
+      "step": 1137
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 4.832414627075195,
+      "learning_rate": 6.174715138361368e-06,
+      "loss": 0.8408,
+      "step": 1138
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.1459033489227295,
+      "learning_rate": 6.180141074335323e-06,
+      "loss": 0.752,
+      "step": 1139
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 4.272872447967529,
+      "learning_rate": 6.185567010309279e-06,
+      "loss": 0.858,
+      "step": 1140
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.563739776611328,
+      "learning_rate": 6.190992946283234e-06,
+      "loss": 0.7079,
+      "step": 1141
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.649425983428955,
+      "learning_rate": 6.19641888225719e-06,
+      "loss": 0.7804,
+      "step": 1142
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 4.193507194519043,
+      "learning_rate": 6.201844818231145e-06,
+      "loss": 0.8553,
+      "step": 1143
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 4.8613362312316895,
+      "learning_rate": 6.207270754205102e-06,
+      "loss": 0.6049,
+      "step": 1144
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.5905776023864746,
+      "learning_rate": 6.212696690179056e-06,
+      "loss": 0.6066,
+      "step": 1145
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.7909939289093018,
+      "learning_rate": 6.218122626153012e-06,
+      "loss": 0.7311,
+      "step": 1146
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 5.251087665557861,
+      "learning_rate": 6.223548562126967e-06,
+      "loss": 0.7404,
+      "step": 1147
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 4.241470813751221,
+      "learning_rate": 6.228974498100923e-06,
+      "loss": 0.5386,
+      "step": 1148
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.7220520973205566,
+      "learning_rate": 6.234400434074879e-06,
+      "loss": 0.9037,
+      "step": 1149
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.3040854930877686,
+      "learning_rate": 6.239826370048834e-06,
+      "loss": 0.7241,
+      "step": 1150
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.650104522705078,
+      "learning_rate": 6.24525230602279e-06,
+      "loss": 0.5129,
+      "step": 1151
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 7.976061820983887,
+      "learning_rate": 6.250678241996745e-06,
+      "loss": 0.5651,
+      "step": 1152
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.840526819229126,
+      "learning_rate": 6.2561041779707e-06,
+      "loss": 0.6478,
+      "step": 1153
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 5.168181419372559,
+      "learning_rate": 6.2615301139446564e-06,
+      "loss": 0.3855,
+      "step": 1154
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 4.327991485595703,
+      "learning_rate": 6.266956049918611e-06,
+      "loss": 0.6262,
+      "step": 1155
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.398496389389038,
+      "learning_rate": 6.272381985892567e-06,
+      "loss": 0.4104,
+      "step": 1156
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.074653387069702,
+      "learning_rate": 6.277807921866522e-06,
+      "loss": 0.6655,
+      "step": 1157
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.932647228240967,
+      "learning_rate": 6.283233857840478e-06,
+      "loss": 0.6187,
+      "step": 1158
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.833096981048584,
+      "learning_rate": 6.288659793814433e-06,
+      "loss": 0.6294,
+      "step": 1159
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 4.003586292266846,
+      "learning_rate": 6.294085729788389e-06,
+      "loss": 0.4759,
+      "step": 1160
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.091545820236206,
+      "learning_rate": 6.299511665762345e-06,
+      "loss": 0.7189,
+      "step": 1161
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 5.039397716522217,
+      "learning_rate": 6.304937601736299e-06,
+      "loss": 0.6658,
+      "step": 1162
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.310519218444824,
+      "learning_rate": 6.310363537710256e-06,
+      "loss": 0.6146,
+      "step": 1163
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.153041362762451,
+      "learning_rate": 6.31578947368421e-06,
+      "loss": 0.6323,
+      "step": 1164
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 6.128571510314941,
+      "learning_rate": 6.321215409658167e-06,
+      "loss": 0.5143,
+      "step": 1165
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.689112424850464,
+      "learning_rate": 6.326641345632122e-06,
+      "loss": 0.5168,
+      "step": 1166
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.0676400661468506,
+      "learning_rate": 6.332067281606078e-06,
+      "loss": 0.7586,
+      "step": 1167
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.344195604324341,
+      "learning_rate": 6.337493217580033e-06,
+      "loss": 0.5171,
+      "step": 1168
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.530937671661377,
+      "learning_rate": 6.342919153553989e-06,
+      "loss": 0.6388,
+      "step": 1169
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.9433897733688354,
+      "learning_rate": 6.348345089527944e-06,
+      "loss": 0.8505,
+      "step": 1170
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.7580201625823975,
+      "learning_rate": 6.3537710255018995e-06,
+      "loss": 0.5817,
+      "step": 1171
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.3945603370666504,
+      "learning_rate": 6.359196961475855e-06,
+      "loss": 0.8437,
+      "step": 1172
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.68473482131958,
+      "learning_rate": 6.3646228974498105e-06,
+      "loss": 0.7295,
+      "step": 1173
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.198878049850464,
+      "learning_rate": 6.370048833423767e-06,
+      "loss": 1.0066,
+      "step": 1174
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.4883673191070557,
+      "learning_rate": 6.375474769397721e-06,
+      "loss": 0.6854,
+      "step": 1175
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.6124439239501953,
+      "learning_rate": 6.380900705371678e-06,
+      "loss": 0.8031,
+      "step": 1176
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.2789535522460938,
+      "learning_rate": 6.386326641345632e-06,
+      "loss": 0.8677,
+      "step": 1177
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.0934436321258545,
+      "learning_rate": 6.391752577319588e-06,
+      "loss": 0.5336,
+      "step": 1178
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.0118303298950195,
+      "learning_rate": 6.397178513293544e-06,
+      "loss": 0.6862,
+      "step": 1179
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.924478530883789,
+      "learning_rate": 6.402604449267499e-06,
+      "loss": 1.0289,
+      "step": 1180
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 7.919103622436523,
+      "learning_rate": 6.408030385241455e-06,
+      "loss": 0.7926,
+      "step": 1181
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 6.348391056060791,
+      "learning_rate": 6.41345632121541e-06,
+      "loss": 0.7172,
+      "step": 1182
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 5.094282627105713,
+      "learning_rate": 6.418882257189366e-06,
+      "loss": 0.5469,
+      "step": 1183
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.9936764240264893,
+      "learning_rate": 6.4243081931633215e-06,
+      "loss": 0.6928,
+      "step": 1184
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.862697124481201,
+      "learning_rate": 6.429734129137277e-06,
+      "loss": 0.5294,
+      "step": 1185
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.7370822429656982,
+      "learning_rate": 6.4351600651112325e-06,
+      "loss": 0.525,
+      "step": 1186
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.5831215381622314,
+      "learning_rate": 6.440586001085187e-06,
+      "loss": 0.612,
+      "step": 1187
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 4.420978546142578,
+      "learning_rate": 6.4460119370591434e-06,
+      "loss": 0.5354,
+      "step": 1188
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.385829210281372,
+      "learning_rate": 6.451437873033098e-06,
+      "loss": 0.5806,
+      "step": 1189
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.53128719329834,
+      "learning_rate": 6.456863809007054e-06,
+      "loss": 0.528,
+      "step": 1190
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.1413426399230957,
+      "learning_rate": 6.46228974498101e-06,
+      "loss": 0.7264,
+      "step": 1191
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.365420341491699,
+      "learning_rate": 6.467715680954965e-06,
+      "loss": 0.5251,
+      "step": 1192
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.6035990715026855,
+      "learning_rate": 6.473141616928921e-06,
+      "loss": 0.6835,
+      "step": 1193
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 5.570479393005371,
+      "learning_rate": 6.4785675529028754e-06,
+      "loss": 0.564,
+      "step": 1194
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 10.998910903930664,
+      "learning_rate": 6.483993488876832e-06,
+      "loss": 0.6346,
+      "step": 1195
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.947542667388916,
+      "learning_rate": 6.489419424850787e-06,
+      "loss": 0.6155,
+      "step": 1196
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.7886481285095215,
+      "learning_rate": 6.494845360824743e-06,
+      "loss": 0.8676,
+      "step": 1197
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 4.498384475708008,
+      "learning_rate": 6.500271296798698e-06,
+      "loss": 0.6395,
+      "step": 1198
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 4.280331134796143,
+      "learning_rate": 6.5056972327726545e-06,
+      "loss": 0.8377,
+      "step": 1199
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 5.155703067779541,
+      "learning_rate": 6.511123168746609e-06,
+      "loss": 0.5226,
+      "step": 1200
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.7306747436523438,
+      "learning_rate": 6.5165491047205655e-06,
+      "loss": 0.5092,
+      "step": 1201
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.85154390335083,
+      "learning_rate": 6.52197504069452e-06,
+      "loss": 0.512,
+      "step": 1202
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.809274435043335,
+      "learning_rate": 6.5274009766684756e-06,
+      "loss": 0.8711,
+      "step": 1203
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.894843339920044,
+      "learning_rate": 6.532826912642432e-06,
+      "loss": 0.5884,
+      "step": 1204
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.0596368312835693,
+      "learning_rate": 6.5382528486163865e-06,
+      "loss": 0.4837,
+      "step": 1205
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 4.0510945320129395,
+      "learning_rate": 6.543678784590343e-06,
+      "loss": 0.4961,
+      "step": 1206
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.943944215774536,
+      "learning_rate": 6.5491047205642975e-06,
+      "loss": 0.6766,
+      "step": 1207
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.596524238586426,
+      "learning_rate": 6.554530656538254e-06,
+      "loss": 0.8416,
+      "step": 1208
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.6808276176452637,
+      "learning_rate": 6.559956592512209e-06,
+      "loss": 0.4251,
+      "step": 1209
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.236543893814087,
+      "learning_rate": 6.565382528486164e-06,
+      "loss": 0.7151,
+      "step": 1210
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 5.790587902069092,
+      "learning_rate": 6.57080846446012e-06,
+      "loss": 0.7718,
+      "step": 1211
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 4.123170375823975,
+      "learning_rate": 6.576234400434075e-06,
+      "loss": 0.4881,
+      "step": 1212
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 5.571085453033447,
+      "learning_rate": 6.581660336408031e-06,
+      "loss": 0.4195,
+      "step": 1213
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.789781093597412,
+      "learning_rate": 6.587086272381986e-06,
+      "loss": 0.7257,
+      "step": 1214
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.121239423751831,
+      "learning_rate": 6.592512208355942e-06,
+      "loss": 0.5901,
+      "step": 1215
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.183011770248413,
+      "learning_rate": 6.597938144329898e-06,
+      "loss": 0.4941,
+      "step": 1216
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 4.111740589141846,
+      "learning_rate": 6.603364080303853e-06,
+      "loss": 0.4003,
+      "step": 1217
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 4.807558059692383,
+      "learning_rate": 6.6087900162778085e-06,
+      "loss": 0.5964,
+      "step": 1218
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 4.004347324371338,
+      "learning_rate": 6.614215952251763e-06,
+      "loss": 0.6162,
+      "step": 1219
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 4.085539817810059,
+      "learning_rate": 6.6196418882257195e-06,
+      "loss": 0.5692,
+      "step": 1220
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.229004383087158,
+      "learning_rate": 6.625067824199675e-06,
+      "loss": 0.7215,
+      "step": 1221
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.8712236881256104,
+      "learning_rate": 6.6304937601736304e-06,
+      "loss": 0.4116,
+      "step": 1222
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 4.089196681976318,
+      "learning_rate": 6.635919696147586e-06,
+      "loss": 0.6058,
+      "step": 1223
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.3997013568878174,
+      "learning_rate": 6.641345632121542e-06,
+      "loss": 0.6761,
+      "step": 1224
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.3803699016571045,
+      "learning_rate": 6.646771568095497e-06,
+      "loss": 0.6785,
+      "step": 1225
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.6293785572052,
+      "learning_rate": 6.652197504069452e-06,
+      "loss": 0.6334,
+      "step": 1226
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.5737454891204834,
+      "learning_rate": 6.657623440043408e-06,
+      "loss": 0.5941,
+      "step": 1227
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.636007070541382,
+      "learning_rate": 6.663049376017363e-06,
+      "loss": 0.5765,
+      "step": 1228
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.7105650901794434,
+      "learning_rate": 6.66847531199132e-06,
+      "loss": 0.5396,
+      "step": 1229
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 5.024198532104492,
+      "learning_rate": 6.673901247965274e-06,
+      "loss": 0.5858,
+      "step": 1230
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.4454450607299805,
+      "learning_rate": 6.6793271839392306e-06,
+      "loss": 0.8433,
+      "step": 1231
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.17295503616333,
+      "learning_rate": 6.684753119913185e-06,
+      "loss": 0.5421,
+      "step": 1232
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.5798251628875732,
+      "learning_rate": 6.6901790558871415e-06,
+      "loss": 0.5199,
+      "step": 1233
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.3191440105438232,
+      "learning_rate": 6.695604991861097e-06,
+      "loss": 0.7578,
+      "step": 1234
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.8877041339874268,
+      "learning_rate": 6.701030927835052e-06,
+      "loss": 0.927,
+      "step": 1235
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.712905168533325,
+      "learning_rate": 6.706456863809008e-06,
+      "loss": 0.4409,
+      "step": 1236
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.7640960216522217,
+      "learning_rate": 6.7118827997829626e-06,
+      "loss": 0.6171,
+      "step": 1237
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 7.316420555114746,
+      "learning_rate": 6.717308735756919e-06,
+      "loss": 0.8698,
+      "step": 1238
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.652040958404541,
+      "learning_rate": 6.7227346717308735e-06,
+      "loss": 0.7146,
+      "step": 1239
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 4.067009449005127,
+      "learning_rate": 6.72816060770483e-06,
+      "loss": 0.548,
+      "step": 1240
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.9526920318603516,
+      "learning_rate": 6.733586543678785e-06,
+      "loss": 0.4649,
+      "step": 1241
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 4.916018962860107,
+      "learning_rate": 6.739012479652741e-06,
+      "loss": 0.5418,
+      "step": 1242
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.6896910667419434,
+      "learning_rate": 6.744438415626696e-06,
+      "loss": 0.4544,
+      "step": 1243
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 4.948587417602539,
+      "learning_rate": 6.749864351600651e-06,
+      "loss": 0.564,
+      "step": 1244
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 4.15958309173584,
+      "learning_rate": 6.755290287574607e-06,
+      "loss": 0.6793,
+      "step": 1245
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.4453413486480713,
+      "learning_rate": 6.760716223548563e-06,
+      "loss": 0.7035,
+      "step": 1246
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.460646629333496,
+      "learning_rate": 6.766142159522518e-06,
+      "loss": 0.7852,
+      "step": 1247
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.741234540939331,
+      "learning_rate": 6.771568095496474e-06,
+      "loss": 0.3241,
+      "step": 1248
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.273031234741211,
+      "learning_rate": 6.77699403147043e-06,
+      "loss": 0.4064,
+      "step": 1249
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 11.615177154541016,
+      "learning_rate": 6.782419967444385e-06,
+      "loss": 0.7113,
+      "step": 1250
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.9624757766723633,
+      "learning_rate": 6.78784590341834e-06,
+      "loss": 0.8612,
+      "step": 1251
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 4.339876174926758,
+      "learning_rate": 6.7932718393922955e-06,
+      "loss": 0.2841,
+      "step": 1252
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 4.763699054718018,
+      "learning_rate": 6.798697775366251e-06,
+      "loss": 0.6049,
+      "step": 1253
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 4.7203049659729,
+      "learning_rate": 6.804123711340207e-06,
+      "loss": 0.6856,
+      "step": 1254
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.020017623901367,
+      "learning_rate": 6.809549647314162e-06,
+      "loss": 0.6532,
+      "step": 1255
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 4.751757621765137,
+      "learning_rate": 6.814975583288118e-06,
+      "loss": 0.4109,
+      "step": 1256
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.127495288848877,
+      "learning_rate": 6.820401519262073e-06,
+      "loss": 0.8183,
+      "step": 1257
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.429730176925659,
+      "learning_rate": 6.825827455236029e-06,
+      "loss": 0.8977,
+      "step": 1258
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.6612565517425537,
+      "learning_rate": 6.831253391209985e-06,
+      "loss": 0.6458,
+      "step": 1259
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 4.95210599899292,
+      "learning_rate": 6.836679327183939e-06,
+      "loss": 0.6,
+      "step": 1260
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.297334671020508,
+      "learning_rate": 6.842105263157896e-06,
+      "loss": 0.6936,
+      "step": 1261
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.7167515754699707,
+      "learning_rate": 6.84753119913185e-06,
+      "loss": 0.725,
+      "step": 1262
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 5.757476806640625,
+      "learning_rate": 6.852957135105807e-06,
+      "loss": 0.6548,
+      "step": 1263
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.7190253734588623,
+      "learning_rate": 6.858383071079761e-06,
+      "loss": 0.4885,
+      "step": 1264
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.630627393722534,
+      "learning_rate": 6.8638090070537176e-06,
+      "loss": 0.7447,
+      "step": 1265
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.203411817550659,
+      "learning_rate": 6.869234943027673e-06,
+      "loss": 0.4916,
+      "step": 1266
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.3304708003997803,
+      "learning_rate": 6.874660879001628e-06,
+      "loss": 0.6979,
+      "step": 1267
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.0000789165496826,
+      "learning_rate": 6.880086814975584e-06,
+      "loss": 0.5687,
+      "step": 1268
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.5509650707244873,
+      "learning_rate": 6.885512750949539e-06,
+      "loss": 0.7664,
+      "step": 1269
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 4.241332530975342,
+      "learning_rate": 6.890938686923495e-06,
+      "loss": 0.6765,
+      "step": 1270
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 8.189007759094238,
+      "learning_rate": 6.89636462289745e-06,
+      "loss": 0.6321,
+      "step": 1271
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 4.284722328186035,
+      "learning_rate": 6.901790558871406e-06,
+      "loss": 0.4478,
+      "step": 1272
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.9488792419433594,
+      "learning_rate": 6.907216494845361e-06,
+      "loss": 0.6501,
+      "step": 1273
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 6.189094543457031,
+      "learning_rate": 6.912642430819318e-06,
+      "loss": 0.6566,
+      "step": 1274
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.033494234085083,
+      "learning_rate": 6.918068366793272e-06,
+      "loss": 0.7194,
+      "step": 1275
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.5696420669555664,
+      "learning_rate": 6.923494302767228e-06,
+      "loss": 0.7571,
+      "step": 1276
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.521697521209717,
+      "learning_rate": 6.928920238741183e-06,
+      "loss": 0.5547,
+      "step": 1277
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 5.089188575744629,
+      "learning_rate": 6.934346174715139e-06,
+      "loss": 0.8408,
+      "step": 1278
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 6.1960368156433105,
+      "learning_rate": 6.939772110689095e-06,
+      "loss": 0.5006,
+      "step": 1279
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.9891421794891357,
+      "learning_rate": 6.94519804666305e-06,
+      "loss": 0.7431,
+      "step": 1280
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 15.00875473022461,
+      "learning_rate": 6.950623982637006e-06,
+      "loss": 0.7056,
+      "step": 1281
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.9528839588165283,
+      "learning_rate": 6.956049918610961e-06,
+      "loss": 0.9026,
+      "step": 1282
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.464416742324829,
+      "learning_rate": 6.961475854584916e-06,
+      "loss": 0.7109,
+      "step": 1283
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.903050422668457,
+      "learning_rate": 6.9669017905588724e-06,
+      "loss": 0.6448,
+      "step": 1284
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.815108299255371,
+      "learning_rate": 6.972327726532827e-06,
+      "loss": 0.4169,
+      "step": 1285
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.0152981281280518,
+      "learning_rate": 6.977753662506783e-06,
+      "loss": 0.5152,
+      "step": 1286
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.8931784629821777,
+      "learning_rate": 6.983179598480738e-06,
+      "loss": 0.3693,
+      "step": 1287
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.4913506507873535,
+      "learning_rate": 6.988605534454694e-06,
+      "loss": 0.6184,
+      "step": 1288
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.7973790168762207,
+      "learning_rate": 6.99403147042865e-06,
+      "loss": 0.6932,
+      "step": 1289
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.9112679958343506,
+      "learning_rate": 6.999457406402605e-06,
+      "loss": 0.7492,
+      "step": 1290
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.467797040939331,
+      "learning_rate": 7.004883342376561e-06,
+      "loss": 0.5588,
+      "step": 1291
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 4.158562660217285,
+      "learning_rate": 7.010309278350515e-06,
+      "loss": 0.6558,
+      "step": 1292
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.7063517570495605,
+      "learning_rate": 7.015735214324472e-06,
+      "loss": 0.7649,
+      "step": 1293
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.4013237953186035,
+      "learning_rate": 7.021161150298426e-06,
+      "loss": 0.7432,
+      "step": 1294
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.4330825805664062,
+      "learning_rate": 7.026587086272383e-06,
+      "loss": 0.7922,
+      "step": 1295
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.961975574493408,
+      "learning_rate": 7.032013022246338e-06,
+      "loss": 0.4614,
+      "step": 1296
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 4.681619167327881,
+      "learning_rate": 7.037438958220294e-06,
+      "loss": 0.5899,
+      "step": 1297
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.4078571796417236,
+      "learning_rate": 7.042864894194249e-06,
+      "loss": 0.6349,
+      "step": 1298
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.4141430854797363,
+      "learning_rate": 7.048290830168205e-06,
+      "loss": 0.6102,
+      "step": 1299
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.8289666175842285,
+      "learning_rate": 7.05371676614216e-06,
+      "loss": 0.6827,
+      "step": 1300
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.2078070640563965,
+      "learning_rate": 7.0591427021161155e-06,
+      "loss": 0.7,
+      "step": 1301
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.527787208557129,
+      "learning_rate": 7.064568638090071e-06,
+      "loss": 0.7334,
+      "step": 1302
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.264819383621216,
+      "learning_rate": 7.0699945740640265e-06,
+      "loss": 0.453,
+      "step": 1303
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.954998016357422,
+      "learning_rate": 7.075420510037983e-06,
+      "loss": 0.7192,
+      "step": 1304
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.055337905883789,
+      "learning_rate": 7.080846446011937e-06,
+      "loss": 0.6999,
+      "step": 1305
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 4.294620990753174,
+      "learning_rate": 7.086272381985894e-06,
+      "loss": 0.551,
+      "step": 1306
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 5.027046203613281,
+      "learning_rate": 7.091698317959848e-06,
+      "loss": 0.5073,
+      "step": 1307
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.887040853500366,
+      "learning_rate": 7.097124253933804e-06,
+      "loss": 0.6651,
+      "step": 1308
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 4.286401271820068,
+      "learning_rate": 7.10255018990776e-06,
+      "loss": 0.704,
+      "step": 1309
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.692870855331421,
+      "learning_rate": 7.107976125881715e-06,
+      "loss": 0.6533,
+      "step": 1310
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.2863478660583496,
+      "learning_rate": 7.113402061855671e-06,
+      "loss": 0.5369,
+      "step": 1311
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 4.1267008781433105,
+      "learning_rate": 7.118827997829626e-06,
+      "loss": 0.6038,
+      "step": 1312
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 4.9931864738464355,
+      "learning_rate": 7.124253933803582e-06,
+      "loss": 0.4917,
+      "step": 1313
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.20092511177063,
+      "learning_rate": 7.1296798697775375e-06,
+      "loss": 0.6018,
+      "step": 1314
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.859004020690918,
+      "learning_rate": 7.135105805751493e-06,
+      "loss": 0.629,
+      "step": 1315
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.375049591064453,
+      "learning_rate": 7.1405317417254485e-06,
+      "loss": 0.5856,
+      "step": 1316
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 4.1077752113342285,
+      "learning_rate": 7.145957677699403e-06,
+      "loss": 0.5357,
+      "step": 1317
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.7477622032165527,
+      "learning_rate": 7.1513836136733594e-06,
+      "loss": 0.7491,
+      "step": 1318
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.6888675689697266,
+      "learning_rate": 7.156809549647314e-06,
+      "loss": 1.0002,
+      "step": 1319
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.521578073501587,
+      "learning_rate": 7.16223548562127e-06,
+      "loss": 0.4227,
+      "step": 1320
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 5.107117176055908,
+      "learning_rate": 7.167661421595226e-06,
+      "loss": 0.6597,
+      "step": 1321
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.823690176010132,
+      "learning_rate": 7.173087357569181e-06,
+      "loss": 0.6545,
+      "step": 1322
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.5705995559692383,
+      "learning_rate": 7.178513293543137e-06,
+      "loss": 0.6449,
+      "step": 1323
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 5.939917087554932,
+      "learning_rate": 7.1839392295170914e-06,
+      "loss": 0.514,
+      "step": 1324
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.67993426322937,
+      "learning_rate": 7.189365165491048e-06,
+      "loss": 0.811,
+      "step": 1325
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 5.33483362197876,
+      "learning_rate": 7.194791101465003e-06,
+      "loss": 0.6939,
+      "step": 1326
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 5.075522422790527,
+      "learning_rate": 7.200217037438959e-06,
+      "loss": 0.7234,
+      "step": 1327
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.888251304626465,
+      "learning_rate": 7.205642973412914e-06,
+      "loss": 0.3375,
+      "step": 1328
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 7.5106306076049805,
+      "learning_rate": 7.2110689093868705e-06,
+      "loss": 0.7354,
+      "step": 1329
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 4.956221103668213,
+      "learning_rate": 7.216494845360825e-06,
+      "loss": 0.6381,
+      "step": 1330
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.6413190364837646,
+      "learning_rate": 7.2219207813347815e-06,
+      "loss": 0.8211,
+      "step": 1331
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.3850910663604736,
+      "learning_rate": 7.227346717308736e-06,
+      "loss": 0.6383,
+      "step": 1332
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 9.20187759399414,
+      "learning_rate": 7.2327726532826916e-06,
+      "loss": 0.6378,
+      "step": 1333
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.2498435974121094,
+      "learning_rate": 7.238198589256648e-06,
+      "loss": 0.5991,
+      "step": 1334
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 19.218461990356445,
+      "learning_rate": 7.2436245252306025e-06,
+      "loss": 0.8101,
+      "step": 1335
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.602914810180664,
+      "learning_rate": 7.249050461204559e-06,
+      "loss": 0.6391,
+      "step": 1336
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.795907735824585,
+      "learning_rate": 7.2544763971785135e-06,
+      "loss": 0.7137,
+      "step": 1337
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.533869743347168,
+      "learning_rate": 7.25990233315247e-06,
+      "loss": 0.5105,
+      "step": 1338
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.132204532623291,
+      "learning_rate": 7.265328269126425e-06,
+      "loss": 0.7509,
+      "step": 1339
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.381194829940796,
+      "learning_rate": 7.27075420510038e-06,
+      "loss": 0.7345,
+      "step": 1340
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.760641574859619,
+      "learning_rate": 7.276180141074336e-06,
+      "loss": 0.663,
+      "step": 1341
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.159606695175171,
+      "learning_rate": 7.281606077048291e-06,
+      "loss": 0.7009,
+      "step": 1342
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.558579921722412,
+      "learning_rate": 7.287032013022247e-06,
+      "loss": 0.66,
+      "step": 1343
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 5.249468803405762,
+      "learning_rate": 7.292457948996202e-06,
+      "loss": 0.8478,
+      "step": 1344
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.7617037296295166,
+      "learning_rate": 7.297883884970158e-06,
+      "loss": 0.897,
+      "step": 1345
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 4.501162528991699,
+      "learning_rate": 7.303309820944114e-06,
+      "loss": 0.6189,
+      "step": 1346
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 4.289915561676025,
+      "learning_rate": 7.308735756918069e-06,
+      "loss": 0.9818,
+      "step": 1347
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 4.9679155349731445,
+      "learning_rate": 7.3141616928920245e-06,
+      "loss": 0.6598,
+      "step": 1348
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.2692723274230957,
+      "learning_rate": 7.319587628865979e-06,
+      "loss": 0.5206,
+      "step": 1349
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.4765522480010986,
+      "learning_rate": 7.3250135648399355e-06,
+      "loss": 0.8165,
+      "step": 1350
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.757632255554199,
+      "learning_rate": 7.330439500813891e-06,
+      "loss": 0.5332,
+      "step": 1351
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 7.03855562210083,
+      "learning_rate": 7.3358654367878464e-06,
+      "loss": 0.8059,
+      "step": 1352
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.323133945465088,
+      "learning_rate": 7.341291372761802e-06,
+      "loss": 0.7119,
+      "step": 1353
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 5.188218116760254,
+      "learning_rate": 7.346717308735758e-06,
+      "loss": 0.7073,
+      "step": 1354
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.304656982421875,
+      "learning_rate": 7.352143244709713e-06,
+      "loss": 0.7213,
+      "step": 1355
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 6.52625036239624,
+      "learning_rate": 7.357569180683668e-06,
+      "loss": 0.6743,
+      "step": 1356
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.309302568435669,
+      "learning_rate": 7.362995116657624e-06,
+      "loss": 0.4599,
+      "step": 1357
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.8236660957336426,
+      "learning_rate": 7.368421052631579e-06,
+      "loss": 0.5347,
+      "step": 1358
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.7798922061920166,
+      "learning_rate": 7.373846988605536e-06,
+      "loss": 0.5714,
+      "step": 1359
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.269421100616455,
+      "learning_rate": 7.37927292457949e-06,
+      "loss": 0.7089,
+      "step": 1360
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.9079270362854004,
+      "learning_rate": 7.3846988605534466e-06,
+      "loss": 0.7103,
+      "step": 1361
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.7706875801086426,
+      "learning_rate": 7.390124796527401e-06,
+      "loss": 0.697,
+      "step": 1362
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 4.040767192840576,
+      "learning_rate": 7.3955507325013575e-06,
+      "loss": 0.4783,
+      "step": 1363
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 6.137670516967773,
+      "learning_rate": 7.400976668475313e-06,
+      "loss": 0.5496,
+      "step": 1364
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 4.040045261383057,
+      "learning_rate": 7.406402604449268e-06,
+      "loss": 0.6058,
+      "step": 1365
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.9494400024414062,
+      "learning_rate": 7.411828540423224e-06,
+      "loss": 0.7654,
+      "step": 1366
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.336550712585449,
+      "learning_rate": 7.4172544763971786e-06,
+      "loss": 0.664,
+      "step": 1367
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.6216776371002197,
+      "learning_rate": 7.422680412371135e-06,
+      "loss": 0.4269,
+      "step": 1368
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 11.670979499816895,
+      "learning_rate": 7.4281063483450895e-06,
+      "loss": 0.8144,
+      "step": 1369
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.524662971496582,
+      "learning_rate": 7.433532284319046e-06,
+      "loss": 0.733,
+      "step": 1370
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.737217903137207,
+      "learning_rate": 7.438958220293001e-06,
+      "loss": 0.5503,
+      "step": 1371
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.9097986221313477,
+      "learning_rate": 7.444384156266958e-06,
+      "loss": 0.4949,
+      "step": 1372
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.2270941734313965,
+      "learning_rate": 7.449810092240912e-06,
+      "loss": 0.5712,
+      "step": 1373
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 6.657175064086914,
+      "learning_rate": 7.455236028214867e-06,
+      "loss": 0.5808,
+      "step": 1374
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.475111246109009,
+      "learning_rate": 7.460661964188823e-06,
+      "loss": 0.5014,
+      "step": 1375
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.074215888977051,
+      "learning_rate": 7.466087900162779e-06,
+      "loss": 0.9727,
+      "step": 1376
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.4700329303741455,
+      "learning_rate": 7.471513836136734e-06,
+      "loss": 0.8445,
+      "step": 1377
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.065434217453003,
+      "learning_rate": 7.47693977211069e-06,
+      "loss": 0.5937,
+      "step": 1378
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.3778772354125977,
+      "learning_rate": 7.482365708084646e-06,
+      "loss": 0.6142,
+      "step": 1379
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.55171799659729,
+      "learning_rate": 7.487791644058601e-06,
+      "loss": 0.7004,
+      "step": 1380
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.6132402420043945,
+      "learning_rate": 7.493217580032556e-06,
+      "loss": 0.5329,
+      "step": 1381
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.5221195220947266,
+      "learning_rate": 7.4986435160065115e-06,
+      "loss": 0.7604,
+      "step": 1382
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.972565174102783,
+      "learning_rate": 7.504069451980467e-06,
+      "loss": 0.5723,
+      "step": 1383
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.7618329524993896,
+      "learning_rate": 7.509495387954423e-06,
+      "loss": 0.6025,
+      "step": 1384
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.7557613849639893,
+      "learning_rate": 7.514921323928378e-06,
+      "loss": 0.8428,
+      "step": 1385
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.3230855464935303,
+      "learning_rate": 7.520347259902334e-06,
+      "loss": 0.5789,
+      "step": 1386
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 6.875540256500244,
+      "learning_rate": 7.525773195876289e-06,
+      "loss": 0.6555,
+      "step": 1387
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.0399508476257324,
+      "learning_rate": 7.531199131850245e-06,
+      "loss": 0.6975,
+      "step": 1388
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.279181957244873,
+      "learning_rate": 7.536625067824201e-06,
+      "loss": 0.6861,
+      "step": 1389
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.613536834716797,
+      "learning_rate": 7.542051003798155e-06,
+      "loss": 0.8263,
+      "step": 1390
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.9458751678466797,
+      "learning_rate": 7.547476939772112e-06,
+      "loss": 0.5209,
+      "step": 1391
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.668384075164795,
+      "learning_rate": 7.552902875746066e-06,
+      "loss": 0.5076,
+      "step": 1392
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.4541103839874268,
+      "learning_rate": 7.558328811720023e-06,
+      "loss": 0.398,
+      "step": 1393
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 5.059971809387207,
+      "learning_rate": 7.563754747693978e-06,
+      "loss": 0.7925,
+      "step": 1394
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.909433603286743,
+      "learning_rate": 7.5691806836679336e-06,
+      "loss": 0.5471,
+      "step": 1395
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.1498095989227295,
+      "learning_rate": 7.574606619641889e-06,
+      "loss": 0.7005,
+      "step": 1396
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.604809284210205,
+      "learning_rate": 7.580032555615844e-06,
+      "loss": 0.6547,
+      "step": 1397
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.287306547164917,
+      "learning_rate": 7.5854584915898e-06,
+      "loss": 0.6694,
+      "step": 1398
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.3926162719726562,
+      "learning_rate": 7.590884427563755e-06,
+      "loss": 0.8206,
+      "step": 1399
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 6.934560775756836,
+      "learning_rate": 7.596310363537711e-06,
+      "loss": 0.5561,
+      "step": 1400
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.7501566410064697,
+      "learning_rate": 7.601736299511666e-06,
+      "loss": 0.6422,
+      "step": 1401
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 5.045010566711426,
+      "learning_rate": 7.607162235485622e-06,
+      "loss": 0.4186,
+      "step": 1402
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.9396231174468994,
+      "learning_rate": 7.612588171459577e-06,
+      "loss": 0.6038,
+      "step": 1403
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.633975028991699,
+      "learning_rate": 7.618014107433534e-06,
+      "loss": 0.6934,
+      "step": 1404
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.3050458431243896,
+      "learning_rate": 7.623440043407488e-06,
+      "loss": 0.6291,
+      "step": 1405
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 4.012405872344971,
+      "learning_rate": 7.628865979381444e-06,
+      "loss": 0.6655,
+      "step": 1406
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.9644083976745605,
+      "learning_rate": 7.6342919153554e-06,
+      "loss": 0.7648,
+      "step": 1407
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.7710330486297607,
+      "learning_rate": 7.639717851329355e-06,
+      "loss": 0.88,
+      "step": 1408
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.592569351196289,
+      "learning_rate": 7.645143787303311e-06,
+      "loss": 0.4755,
+      "step": 1409
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 6.541748046875,
+      "learning_rate": 7.650569723277266e-06,
+      "loss": 0.7364,
+      "step": 1410
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.326277256011963,
+      "learning_rate": 7.655995659251222e-06,
+      "loss": 0.6076,
+      "step": 1411
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 4.490007400512695,
+      "learning_rate": 7.661421595225177e-06,
+      "loss": 0.334,
+      "step": 1412
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.813875675201416,
+      "learning_rate": 7.666847531199131e-06,
+      "loss": 0.6353,
+      "step": 1413
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.2838144302368164,
+      "learning_rate": 7.672273467173088e-06,
+      "loss": 0.6185,
+      "step": 1414
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 5.407020092010498,
+      "learning_rate": 7.677699403147044e-06,
+      "loss": 0.6388,
+      "step": 1415
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.0353493690490723,
+      "learning_rate": 7.683125339120999e-06,
+      "loss": 0.7071,
+      "step": 1416
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 13.673770904541016,
+      "learning_rate": 7.688551275094955e-06,
+      "loss": 0.7315,
+      "step": 1417
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.7871224880218506,
+      "learning_rate": 7.693977211068911e-06,
+      "loss": 0.3528,
+      "step": 1418
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.2348694801330566,
+      "learning_rate": 7.699403147042866e-06,
+      "loss": 0.8601,
+      "step": 1419
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 4.167965412139893,
+      "learning_rate": 7.704829083016822e-06,
+      "loss": 0.4419,
+      "step": 1420
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.3072705268859863,
+      "learning_rate": 7.710255018990777e-06,
+      "loss": 0.6276,
+      "step": 1421
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.7482333183288574,
+      "learning_rate": 7.715680954964731e-06,
+      "loss": 0.7351,
+      "step": 1422
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.765777826309204,
+      "learning_rate": 7.721106890938688e-06,
+      "loss": 0.8954,
+      "step": 1423
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 4.647091388702393,
+      "learning_rate": 7.726532826912642e-06,
+      "loss": 0.4668,
+      "step": 1424
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.107078790664673,
+      "learning_rate": 7.731958762886599e-06,
+      "loss": 0.9692,
+      "step": 1425
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.9581375122070312,
+      "learning_rate": 7.737384698860553e-06,
+      "loss": 0.6275,
+      "step": 1426
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.5190773010253906,
+      "learning_rate": 7.74281063483451e-06,
+      "loss": 0.621,
+      "step": 1427
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.9839683771133423,
+      "learning_rate": 7.748236570808464e-06,
+      "loss": 0.7798,
+      "step": 1428
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.631398916244507,
+      "learning_rate": 7.75366250678242e-06,
+      "loss": 0.4523,
+      "step": 1429
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.441793918609619,
+      "learning_rate": 7.759088442756377e-06,
+      "loss": 0.5433,
+      "step": 1430
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.611877918243408,
+      "learning_rate": 7.764514378730332e-06,
+      "loss": 0.6856,
+      "step": 1431
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 11.646836280822754,
+      "learning_rate": 7.769940314704288e-06,
+      "loss": 0.4193,
+      "step": 1432
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.2883810997009277,
+      "learning_rate": 7.775366250678242e-06,
+      "loss": 0.6369,
+      "step": 1433
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 4.409623622894287,
+      "learning_rate": 7.780792186652199e-06,
+      "loss": 0.8283,
+      "step": 1434
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.456385374069214,
+      "learning_rate": 7.786218122626153e-06,
+      "loss": 0.432,
+      "step": 1435
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.380134105682373,
+      "learning_rate": 7.79164405860011e-06,
+      "loss": 0.6328,
+      "step": 1436
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.609989881515503,
+      "learning_rate": 7.797069994574064e-06,
+      "loss": 0.7564,
+      "step": 1437
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.683284044265747,
+      "learning_rate": 7.802495930548019e-06,
+      "loss": 0.7774,
+      "step": 1438
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.107571840286255,
+      "learning_rate": 7.807921866521975e-06,
+      "loss": 0.5837,
+      "step": 1439
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.894989252090454,
+      "learning_rate": 7.813347802495932e-06,
+      "loss": 0.4062,
+      "step": 1440
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.324984312057495,
+      "learning_rate": 7.818773738469886e-06,
+      "loss": 0.6393,
+      "step": 1441
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.64013934135437,
+      "learning_rate": 7.824199674443843e-06,
+      "loss": 0.7962,
+      "step": 1442
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 6.596169471740723,
+      "learning_rate": 7.829625610417799e-06,
+      "loss": 0.7037,
+      "step": 1443
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.2699460983276367,
+      "learning_rate": 7.835051546391754e-06,
+      "loss": 0.6423,
+      "step": 1444
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.5525519847869873,
+      "learning_rate": 7.84047748236571e-06,
+      "loss": 0.7017,
+      "step": 1445
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.0448691844940186,
+      "learning_rate": 7.845903418339664e-06,
+      "loss": 0.5163,
+      "step": 1446
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.8108365535736084,
+      "learning_rate": 7.851329354313619e-06,
+      "loss": 0.609,
+      "step": 1447
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.7213988304138184,
+      "learning_rate": 7.856755290287575e-06,
+      "loss": 0.7229,
+      "step": 1448
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.359011173248291,
+      "learning_rate": 7.86218122626153e-06,
+      "loss": 0.4862,
+      "step": 1449
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.3876731395721436,
+      "learning_rate": 7.867607162235486e-06,
+      "loss": 0.7281,
+      "step": 1450
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.030688762664795,
+      "learning_rate": 7.873033098209441e-06,
+      "loss": 0.693,
+      "step": 1451
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 4.4743218421936035,
+      "learning_rate": 7.878459034183397e-06,
+      "loss": 0.7859,
+      "step": 1452
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.365260362625122,
+      "learning_rate": 7.883884970157352e-06,
+      "loss": 0.802,
+      "step": 1453
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.6854827404022217,
+      "learning_rate": 7.889310906131308e-06,
+      "loss": 0.6901,
+      "step": 1454
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.0094058513641357,
+      "learning_rate": 7.894736842105265e-06,
+      "loss": 0.8551,
+      "step": 1455
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.3197476863861084,
+      "learning_rate": 7.90016277807922e-06,
+      "loss": 0.8205,
+      "step": 1456
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.9004082679748535,
+      "learning_rate": 7.905588714053176e-06,
+      "loss": 0.5164,
+      "step": 1457
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.925859212875366,
+      "learning_rate": 7.91101465002713e-06,
+      "loss": 0.852,
+      "step": 1458
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 21.569921493530273,
+      "learning_rate": 7.916440586001087e-06,
+      "loss": 0.8664,
+      "step": 1459
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 4.73588228225708,
+      "learning_rate": 7.921866521975041e-06,
+      "loss": 0.5778,
+      "step": 1460
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.7643990516662598,
+      "learning_rate": 7.927292457948997e-06,
+      "loss": 0.7498,
+      "step": 1461
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.939589500427246,
+      "learning_rate": 7.932718393922952e-06,
+      "loss": 0.7223,
+      "step": 1462
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 5.048740386962891,
+      "learning_rate": 7.938144329896907e-06,
+      "loss": 0.8057,
+      "step": 1463
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.4294140338897705,
+      "learning_rate": 7.943570265870863e-06,
+      "loss": 0.394,
+      "step": 1464
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.7532644271850586,
+      "learning_rate": 7.94899620184482e-06,
+      "loss": 0.6126,
+      "step": 1465
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.927168130874634,
+      "learning_rate": 7.954422137818774e-06,
+      "loss": 0.6903,
+      "step": 1466
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 10.049833297729492,
+      "learning_rate": 7.95984807379273e-06,
+      "loss": 0.6335,
+      "step": 1467
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.551985502243042,
+      "learning_rate": 7.965274009766687e-06,
+      "loss": 0.723,
+      "step": 1468
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.6544108390808105,
+      "learning_rate": 7.970699945740641e-06,
+      "loss": 0.6671,
+      "step": 1469
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.3908348083496094,
+      "learning_rate": 7.976125881714596e-06,
+      "loss": 0.7982,
+      "step": 1470
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.8969805240631104,
+      "learning_rate": 7.981551817688552e-06,
+      "loss": 0.6821,
+      "step": 1471
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.5227231979370117,
+      "learning_rate": 7.986977753662507e-06,
+      "loss": 0.8545,
+      "step": 1472
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.862189292907715,
+      "learning_rate": 7.992403689636463e-06,
+      "loss": 0.6773,
+      "step": 1473
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.355382204055786,
+      "learning_rate": 7.997829625610418e-06,
+      "loss": 0.4751,
+      "step": 1474
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 4.434351921081543,
+      "learning_rate": 8.003255561584374e-06,
+      "loss": 0.8137,
+      "step": 1475
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.4361910820007324,
+      "learning_rate": 8.008681497558329e-06,
+      "loss": 0.9705,
+      "step": 1476
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.160565137863159,
+      "learning_rate": 8.014107433532285e-06,
+      "loss": 0.6936,
+      "step": 1477
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.263735771179199,
+      "learning_rate": 8.01953336950624e-06,
+      "loss": 0.478,
+      "step": 1478
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.2625317573547363,
+      "learning_rate": 8.024959305480196e-06,
+      "loss": 0.4808,
+      "step": 1479
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.325014591217041,
+      "learning_rate": 8.030385241454152e-06,
+      "loss": 0.6413,
+      "step": 1480
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.318685531616211,
+      "learning_rate": 8.035811177428107e-06,
+      "loss": 0.4859,
+      "step": 1481
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.4808013439178467,
+      "learning_rate": 8.041237113402063e-06,
+      "loss": 0.4131,
+      "step": 1482
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.849252223968506,
+      "learning_rate": 8.046663049376018e-06,
+      "loss": 0.7264,
+      "step": 1483
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 8.66748332977295,
+      "learning_rate": 8.052088985349974e-06,
+      "loss": 0.6374,
+      "step": 1484
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.0576670169830322,
+      "learning_rate": 8.057514921323929e-06,
+      "loss": 0.5576,
+      "step": 1485
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.6494531631469727,
+      "learning_rate": 8.062940857297883e-06,
+      "loss": 0.6363,
+      "step": 1486
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.208310127258301,
+      "learning_rate": 8.06836679327184e-06,
+      "loss": 0.7234,
+      "step": 1487
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.997885227203369,
+      "learning_rate": 8.073792729245794e-06,
+      "loss": 0.6248,
+      "step": 1488
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.640810966491699,
+      "learning_rate": 8.07921866521975e-06,
+      "loss": 0.7233,
+      "step": 1489
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.315563917160034,
+      "learning_rate": 8.084644601193707e-06,
+      "loss": 0.6073,
+      "step": 1490
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.4852662086486816,
+      "learning_rate": 8.090070537167662e-06,
+      "loss": 0.5999,
+      "step": 1491
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.149168014526367,
+      "learning_rate": 8.095496473141618e-06,
+      "loss": 0.6614,
+      "step": 1492
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 4.529680252075195,
+      "learning_rate": 8.100922409115574e-06,
+      "loss": 0.5695,
+      "step": 1493
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 4.64592981338501,
+      "learning_rate": 8.106348345089529e-06,
+      "loss": 0.4803,
+      "step": 1494
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.944103479385376,
+      "learning_rate": 8.111774281063484e-06,
+      "loss": 0.7921,
+      "step": 1495
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.1000289916992188,
+      "learning_rate": 8.11720021703744e-06,
+      "loss": 0.8457,
+      "step": 1496
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.9898102283477783,
+      "learning_rate": 8.122626153011395e-06,
+      "loss": 0.4933,
+      "step": 1497
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.3147990703582764,
+      "learning_rate": 8.128052088985351e-06,
+      "loss": 0.4374,
+      "step": 1498
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 6.379023551940918,
+      "learning_rate": 8.133478024959306e-06,
+      "loss": 0.5447,
+      "step": 1499
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.968632698059082,
+      "learning_rate": 8.138903960933262e-06,
+      "loss": 0.7912,
+      "step": 1500
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 4.210444450378418,
+      "learning_rate": 8.144329896907216e-06,
+      "loss": 0.5957,
+      "step": 1501
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.105250835418701,
+      "learning_rate": 8.149755832881173e-06,
+      "loss": 0.5822,
+      "step": 1502
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 4.735480785369873,
+      "learning_rate": 8.155181768855127e-06,
+      "loss": 0.8957,
+      "step": 1503
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 5.078051567077637,
+      "learning_rate": 8.160607704829084e-06,
+      "loss": 0.507,
+      "step": 1504
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.87186336517334,
+      "learning_rate": 8.16603364080304e-06,
+      "loss": 0.5713,
+      "step": 1505
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 4.53760290145874,
+      "learning_rate": 8.171459576776995e-06,
+      "loss": 0.4717,
+      "step": 1506
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.893239974975586,
+      "learning_rate": 8.176885512750951e-06,
+      "loss": 0.7988,
+      "step": 1507
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.6046764850616455,
+      "learning_rate": 8.182311448724906e-06,
+      "loss": 0.632,
+      "step": 1508
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.6955738067626953,
+      "learning_rate": 8.187737384698862e-06,
+      "loss": 0.7488,
+      "step": 1509
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.7368991374969482,
+      "learning_rate": 8.193163320672817e-06,
+      "loss": 0.6084,
+      "step": 1510
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.997152805328369,
+      "learning_rate": 8.198589256646771e-06,
+      "loss": 0.6385,
+      "step": 1511
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 4.0200066566467285,
+      "learning_rate": 8.204015192620728e-06,
+      "loss": 0.6225,
+      "step": 1512
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.4916207790374756,
+      "learning_rate": 8.209441128594682e-06,
+      "loss": 0.7352,
+      "step": 1513
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.4970197677612305,
+      "learning_rate": 8.214867064568638e-06,
+      "loss": 0.5813,
+      "step": 1514
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 4.337362289428711,
+      "learning_rate": 8.220293000542595e-06,
+      "loss": 0.6918,
+      "step": 1515
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 4.847332954406738,
+      "learning_rate": 8.22571893651655e-06,
+      "loss": 0.686,
+      "step": 1516
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.7489535808563232,
+      "learning_rate": 8.231144872490506e-06,
+      "loss": 0.7557,
+      "step": 1517
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 4.141211032867432,
+      "learning_rate": 8.236570808464462e-06,
+      "loss": 1.0719,
+      "step": 1518
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.427903175354004,
+      "learning_rate": 8.241996744438417e-06,
+      "loss": 0.7526,
+      "step": 1519
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.0507521629333496,
+      "learning_rate": 8.247422680412371e-06,
+      "loss": 0.7375,
+      "step": 1520
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 4.2449259757995605,
+      "learning_rate": 8.252848616386328e-06,
+      "loss": 0.7138,
+      "step": 1521
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.9555078744888306,
+      "learning_rate": 8.258274552360282e-06,
+      "loss": 0.4355,
+      "step": 1522
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.9440877437591553,
+      "learning_rate": 8.263700488334239e-06,
+      "loss": 0.812,
+      "step": 1523
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.4716742038726807,
+      "learning_rate": 8.269126424308193e-06,
+      "loss": 0.6876,
+      "step": 1524
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 4.8773698806762695,
+      "learning_rate": 8.27455236028215e-06,
+      "loss": 0.5889,
+      "step": 1525
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.822770357131958,
+      "learning_rate": 8.279978296256104e-06,
+      "loss": 0.613,
+      "step": 1526
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 4.6890363693237305,
+      "learning_rate": 8.28540423223006e-06,
+      "loss": 0.745,
+      "step": 1527
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 8.431660652160645,
+      "learning_rate": 8.290830168204015e-06,
+      "loss": 0.807,
+      "step": 1528
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 4.413090229034424,
+      "learning_rate": 8.296256104177971e-06,
+      "loss": 0.8141,
+      "step": 1529
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.762481212615967,
+      "learning_rate": 8.301682040151928e-06,
+      "loss": 0.6302,
+      "step": 1530
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 4.398227691650391,
+      "learning_rate": 8.307107976125882e-06,
+      "loss": 0.8306,
+      "step": 1531
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.011493682861328,
+      "learning_rate": 8.312533912099839e-06,
+      "loss": 0.6356,
+      "step": 1532
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 4.768395900726318,
+      "learning_rate": 8.317959848073793e-06,
+      "loss": 0.7218,
+      "step": 1533
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 4.478621006011963,
+      "learning_rate": 8.32338578404775e-06,
+      "loss": 0.6441,
+      "step": 1534
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 4.361572265625,
+      "learning_rate": 8.328811720021704e-06,
+      "loss": 0.7204,
+      "step": 1535
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.729358673095703,
+      "learning_rate": 8.334237655995659e-06,
+      "loss": 1.0624,
+      "step": 1536
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.556766986846924,
+      "learning_rate": 8.339663591969615e-06,
+      "loss": 0.4656,
+      "step": 1537
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 4.8690900802612305,
+      "learning_rate": 8.34508952794357e-06,
+      "loss": 0.7513,
+      "step": 1538
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 10.056615829467773,
+      "learning_rate": 8.350515463917526e-06,
+      "loss": 0.7187,
+      "step": 1539
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.2189900875091553,
+      "learning_rate": 8.355941399891483e-06,
+      "loss": 0.5623,
+      "step": 1540
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.8501635789871216,
+      "learning_rate": 8.361367335865437e-06,
+      "loss": 0.7608,
+      "step": 1541
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 5.297530651092529,
+      "learning_rate": 8.366793271839393e-06,
+      "loss": 0.6177,
+      "step": 1542
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.518074631690979,
+      "learning_rate": 8.372219207813348e-06,
+      "loss": 0.5094,
+      "step": 1543
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 7.765463352203369,
+      "learning_rate": 8.377645143787304e-06,
+      "loss": 0.6193,
+      "step": 1544
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.7826647758483887,
+      "learning_rate": 8.383071079761259e-06,
+      "loss": 0.8159,
+      "step": 1545
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 4.480681419372559,
+      "learning_rate": 8.388497015735215e-06,
+      "loss": 0.5975,
+      "step": 1546
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.9133834838867188,
+      "learning_rate": 8.39392295170917e-06,
+      "loss": 0.6089,
+      "step": 1547
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.500969648361206,
+      "learning_rate": 8.399348887683126e-06,
+      "loss": 0.5764,
+      "step": 1548
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.5525829792022705,
+      "learning_rate": 8.404774823657081e-06,
+      "loss": 0.705,
+      "step": 1549
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.30619740486145,
+      "learning_rate": 8.410200759631037e-06,
+      "loss": 0.6515,
+      "step": 1550
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.472036838531494,
+      "learning_rate": 8.415626695604992e-06,
+      "loss": 0.5612,
+      "step": 1551
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.277198553085327,
+      "learning_rate": 8.421052631578948e-06,
+      "loss": 0.688,
+      "step": 1552
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.9110567569732666,
+      "learning_rate": 8.426478567552905e-06,
+      "loss": 0.7586,
+      "step": 1553
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.969254732131958,
+      "learning_rate": 8.43190450352686e-06,
+      "loss": 0.7102,
+      "step": 1554
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.119349718093872,
+      "learning_rate": 8.437330439500816e-06,
+      "loss": 0.8528,
+      "step": 1555
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.0607800483703613,
+      "learning_rate": 8.44275637547477e-06,
+      "loss": 0.5727,
+      "step": 1556
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.906374454498291,
+      "learning_rate": 8.448182311448726e-06,
+      "loss": 0.9149,
+      "step": 1557
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.3334720134735107,
+      "learning_rate": 8.453608247422681e-06,
+      "loss": 0.6462,
+      "step": 1558
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.4116694927215576,
+      "learning_rate": 8.459034183396636e-06,
+      "loss": 0.6352,
+      "step": 1559
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 4.680532455444336,
+      "learning_rate": 8.464460119370592e-06,
+      "loss": 0.767,
+      "step": 1560
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 4.20576810836792,
+      "learning_rate": 8.469886055344547e-06,
+      "loss": 0.6331,
+      "step": 1561
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.85015606880188,
+      "learning_rate": 8.475311991318503e-06,
+      "loss": 0.4114,
+      "step": 1562
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.2137696743011475,
+      "learning_rate": 8.480737927292458e-06,
+      "loss": 0.6407,
+      "step": 1563
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 7.33692741394043,
+      "learning_rate": 8.486163863266414e-06,
+      "loss": 0.5151,
+      "step": 1564
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.54156231880188,
+      "learning_rate": 8.49158979924037e-06,
+      "loss": 0.9002,
+      "step": 1565
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.8098008632659912,
+      "learning_rate": 8.497015735214325e-06,
+      "loss": 0.5897,
+      "step": 1566
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.9605817794799805,
+      "learning_rate": 8.502441671188281e-06,
+      "loss": 0.7081,
+      "step": 1567
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.5724668502807617,
+      "learning_rate": 8.507867607162236e-06,
+      "loss": 0.4892,
+      "step": 1568
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.8526188135147095,
+      "learning_rate": 8.513293543136192e-06,
+      "loss": 0.5117,
+      "step": 1569
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 4.1612396240234375,
+      "learning_rate": 8.518719479110147e-06,
+      "loss": 0.8204,
+      "step": 1570
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.4727396965026855,
+      "learning_rate": 8.524145415084103e-06,
+      "loss": 0.6713,
+      "step": 1571
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.457425832748413,
+      "learning_rate": 8.529571351058058e-06,
+      "loss": 0.7146,
+      "step": 1572
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 5.01406717300415,
+      "learning_rate": 8.534997287032014e-06,
+      "loss": 0.9632,
+      "step": 1573
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.236159563064575,
+      "learning_rate": 8.540423223005969e-06,
+      "loss": 0.6179,
+      "step": 1574
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.4786059856414795,
+      "learning_rate": 8.545849158979925e-06,
+      "loss": 0.626,
+      "step": 1575
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.773244857788086,
+      "learning_rate": 8.55127509495388e-06,
+      "loss": 0.6269,
+      "step": 1576
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 5.55796480178833,
+      "learning_rate": 8.556701030927836e-06,
+      "loss": 0.6514,
+      "step": 1577
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.75199556350708,
+      "learning_rate": 8.562126966901792e-06,
+      "loss": 0.7356,
+      "step": 1578
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 6.321151256561279,
+      "learning_rate": 8.567552902875747e-06,
+      "loss": 0.544,
+      "step": 1579
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.008065700531006,
+      "learning_rate": 8.572978838849703e-06,
+      "loss": 0.6918,
+      "step": 1580
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.7969329357147217,
+      "learning_rate": 8.578404774823658e-06,
+      "loss": 0.5678,
+      "step": 1581
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.73311710357666,
+      "learning_rate": 8.583830710797614e-06,
+      "loss": 0.4851,
+      "step": 1582
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.033398151397705,
+      "learning_rate": 8.589256646771569e-06,
+      "loss": 0.7422,
+      "step": 1583
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.2019944190979004,
+      "learning_rate": 8.594682582745523e-06,
+      "loss": 0.8691,
+      "step": 1584
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.2525527477264404,
+      "learning_rate": 8.60010851871948e-06,
+      "loss": 0.8085,
+      "step": 1585
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.793503522872925,
+      "learning_rate": 8.605534454693434e-06,
+      "loss": 0.9418,
+      "step": 1586
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 19.351463317871094,
+      "learning_rate": 8.61096039066739e-06,
+      "loss": 0.667,
+      "step": 1587
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.599546432495117,
+      "learning_rate": 8.616386326641345e-06,
+      "loss": 0.7652,
+      "step": 1588
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.288710117340088,
+      "learning_rate": 8.621812262615302e-06,
+      "loss": 0.3993,
+      "step": 1589
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.359984874725342,
+      "learning_rate": 8.627238198589258e-06,
+      "loss": 0.5974,
+      "step": 1590
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.280367851257324,
+      "learning_rate": 8.632664134563213e-06,
+      "loss": 0.7677,
+      "step": 1591
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.4289612770080566,
+      "learning_rate": 8.638090070537169e-06,
+      "loss": 0.5204,
+      "step": 1592
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 5.496277332305908,
+      "learning_rate": 8.643516006511124e-06,
+      "loss": 0.7425,
+      "step": 1593
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.9376285076141357,
+      "learning_rate": 8.64894194248508e-06,
+      "loss": 0.7806,
+      "step": 1594
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.195709705352783,
+      "learning_rate": 8.654367878459035e-06,
+      "loss": 0.7854,
+      "step": 1595
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 4.196096897125244,
+      "learning_rate": 8.65979381443299e-06,
+      "loss": 0.3757,
+      "step": 1596
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.839325428009033,
+      "learning_rate": 8.665219750406945e-06,
+      "loss": 0.9268,
+      "step": 1597
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.3311495780944824,
+      "learning_rate": 8.670645686380902e-06,
+      "loss": 0.5235,
+      "step": 1598
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.479508399963379,
+      "learning_rate": 8.676071622354856e-06,
+      "loss": 0.6393,
+      "step": 1599
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.2020657062530518,
+      "learning_rate": 8.681497558328813e-06,
+      "loss": 0.7994,
+      "step": 1600
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 4.070367336273193,
+      "learning_rate": 8.686923494302767e-06,
+      "loss": 0.7329,
+      "step": 1601
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 5.764654159545898,
+      "learning_rate": 8.692349430276724e-06,
+      "loss": 0.5923,
+      "step": 1602
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.2577452659606934,
+      "learning_rate": 8.69777536625068e-06,
+      "loss": 0.4969,
+      "step": 1603
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.2851030826568604,
+      "learning_rate": 8.703201302224635e-06,
+      "loss": 0.7162,
+      "step": 1604
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.7374284267425537,
+      "learning_rate": 8.708627238198591e-06,
+      "loss": 0.6868,
+      "step": 1605
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 5.6402411460876465,
+      "learning_rate": 8.714053174172546e-06,
+      "loss": 0.5169,
+      "step": 1606
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 22.067996978759766,
+      "learning_rate": 8.719479110146502e-06,
+      "loss": 0.7113,
+      "step": 1607
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.9884141683578491,
+      "learning_rate": 8.724905046120457e-06,
+      "loss": 0.6997,
+      "step": 1608
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.520179033279419,
+      "learning_rate": 8.730330982094411e-06,
+      "loss": 0.8827,
+      "step": 1609
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.4168524742126465,
+      "learning_rate": 8.735756918068367e-06,
+      "loss": 0.6734,
+      "step": 1610
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.353336811065674,
+      "learning_rate": 8.741182854042322e-06,
+      "loss": 0.4698,
+      "step": 1611
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.7258753776550293,
+      "learning_rate": 8.746608790016278e-06,
+      "loss": 0.5496,
+      "step": 1612
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 4.6685638427734375,
+      "learning_rate": 8.752034725990233e-06,
+      "loss": 0.569,
+      "step": 1613
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.110361337661743,
+      "learning_rate": 8.75746066196419e-06,
+      "loss": 0.7309,
+      "step": 1614
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.081204414367676,
+      "learning_rate": 8.762886597938146e-06,
+      "loss": 0.6599,
+      "step": 1615
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 11.263636589050293,
+      "learning_rate": 8.7683125339121e-06,
+      "loss": 0.8472,
+      "step": 1616
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 4.172140121459961,
+      "learning_rate": 8.773738469886057e-06,
+      "loss": 0.7038,
+      "step": 1617
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.809682607650757,
+      "learning_rate": 8.779164405860011e-06,
+      "loss": 0.6681,
+      "step": 1618
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.1165435314178467,
+      "learning_rate": 8.784590341833968e-06,
+      "loss": 0.6167,
+      "step": 1619
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.507364511489868,
+      "learning_rate": 8.790016277807922e-06,
+      "loss": 0.7828,
+      "step": 1620
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.300699234008789,
+      "learning_rate": 8.795442213781879e-06,
+      "loss": 0.4259,
+      "step": 1621
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.2152037620544434,
+      "learning_rate": 8.800868149755833e-06,
+      "loss": 0.5279,
+      "step": 1622
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 4.158199787139893,
+      "learning_rate": 8.80629408572979e-06,
+      "loss": 0.509,
+      "step": 1623
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.4654717445373535,
+      "learning_rate": 8.811720021703744e-06,
+      "loss": 0.4978,
+      "step": 1624
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.6880171298980713,
+      "learning_rate": 8.8171459576777e-06,
+      "loss": 0.7761,
+      "step": 1625
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 6.904397964477539,
+      "learning_rate": 8.822571893651655e-06,
+      "loss": 0.7586,
+      "step": 1626
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.8515472412109375,
+      "learning_rate": 8.827997829625611e-06,
+      "loss": 0.6877,
+      "step": 1627
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.1278750896453857,
+      "learning_rate": 8.833423765599568e-06,
+      "loss": 0.5946,
+      "step": 1628
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.497729539871216,
+      "learning_rate": 8.838849701573522e-06,
+      "loss": 0.8664,
+      "step": 1629
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.9784185886383057,
+      "learning_rate": 8.844275637547479e-06,
+      "loss": 0.6403,
+      "step": 1630
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.8998186588287354,
+      "learning_rate": 8.849701573521433e-06,
+      "loss": 0.3724,
+      "step": 1631
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 7.086496353149414,
+      "learning_rate": 8.855127509495388e-06,
+      "loss": 0.6342,
+      "step": 1632
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.9404988288879395,
+      "learning_rate": 8.860553445469344e-06,
+      "loss": 0.6388,
+      "step": 1633
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.761119842529297,
+      "learning_rate": 8.865979381443299e-06,
+      "loss": 0.9229,
+      "step": 1634
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.415125846862793,
+      "learning_rate": 8.871405317417255e-06,
+      "loss": 0.7087,
+      "step": 1635
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.8416762351989746,
+      "learning_rate": 8.87683125339121e-06,
+      "loss": 0.7792,
+      "step": 1636
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 12.57605266571045,
+      "learning_rate": 8.882257189365166e-06,
+      "loss": 0.9724,
+      "step": 1637
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.9395081996917725,
+      "learning_rate": 8.88768312533912e-06,
+      "loss": 0.6125,
+      "step": 1638
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 4.442663192749023,
+      "learning_rate": 8.893109061313077e-06,
+      "loss": 0.4959,
+      "step": 1639
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.9020819664001465,
+      "learning_rate": 8.898534997287033e-06,
+      "loss": 0.4843,
+      "step": 1640
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.7524020671844482,
+      "learning_rate": 8.903960933260988e-06,
+      "loss": 0.9918,
+      "step": 1641
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.0136983394622803,
+      "learning_rate": 8.909386869234944e-06,
+      "loss": 0.5111,
+      "step": 1642
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.891387462615967,
+      "learning_rate": 8.914812805208899e-06,
+      "loss": 0.7894,
+      "step": 1643
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 7.276500701904297,
+      "learning_rate": 8.920238741182855e-06,
+      "loss": 0.691,
+      "step": 1644
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.4108893871307373,
+      "learning_rate": 8.92566467715681e-06,
+      "loss": 0.464,
+      "step": 1645
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.9464688301086426,
+      "learning_rate": 8.931090613130766e-06,
+      "loss": 0.6682,
+      "step": 1646
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.438748598098755,
+      "learning_rate": 8.936516549104721e-06,
+      "loss": 0.4118,
+      "step": 1647
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.830687999725342,
+      "learning_rate": 8.941942485078677e-06,
+      "loss": 0.7356,
+      "step": 1648
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 4.771071434020996,
+      "learning_rate": 8.947368421052632e-06,
+      "loss": 0.4738,
+      "step": 1649
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.3508925437927246,
+      "learning_rate": 8.952794357026588e-06,
+      "loss": 0.5062,
+      "step": 1650
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.8003196716308594,
+      "learning_rate": 8.958220293000543e-06,
+      "loss": 0.4459,
+      "step": 1651
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.118889331817627,
+      "learning_rate": 8.963646228974499e-06,
+      "loss": 0.7182,
+      "step": 1652
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.310014247894287,
+      "learning_rate": 8.969072164948455e-06,
+      "loss": 0.6017,
+      "step": 1653
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 97.1465835571289,
+      "learning_rate": 8.97449810092241e-06,
+      "loss": 0.7392,
+      "step": 1654
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.269399642944336,
+      "learning_rate": 8.979924036896366e-06,
+      "loss": 0.8543,
+      "step": 1655
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.5339903831481934,
+      "learning_rate": 8.985349972870321e-06,
+      "loss": 0.4446,
+      "step": 1656
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 4.240148544311523,
+      "learning_rate": 8.990775908844276e-06,
+      "loss": 0.7743,
+      "step": 1657
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 5.013742923736572,
+      "learning_rate": 8.996201844818232e-06,
+      "loss": 0.656,
+      "step": 1658
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.200817584991455,
+      "learning_rate": 9.001627780792187e-06,
+      "loss": 0.6406,
+      "step": 1659
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.5108213424682617,
+      "learning_rate": 9.007053716766143e-06,
+      "loss": 0.5028,
+      "step": 1660
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.3077211380004883,
+      "learning_rate": 9.012479652740098e-06,
+      "loss": 0.4,
+      "step": 1661
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 4.07047176361084,
+      "learning_rate": 9.017905588714054e-06,
+      "loss": 0.646,
+      "step": 1662
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.326136350631714,
+      "learning_rate": 9.023331524688009e-06,
+      "loss": 0.6931,
+      "step": 1663
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 4.6263813972473145,
+      "learning_rate": 9.028757460661965e-06,
+      "loss": 0.6142,
+      "step": 1664
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.937803030014038,
+      "learning_rate": 9.034183396635921e-06,
+      "loss": 0.8193,
+      "step": 1665
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 6.77678918838501,
+      "learning_rate": 9.039609332609876e-06,
+      "loss": 0.4182,
+      "step": 1666
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.146634101867676,
+      "learning_rate": 9.045035268583832e-06,
+      "loss": 0.6352,
+      "step": 1667
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.93564510345459,
+      "learning_rate": 9.050461204557787e-06,
+      "loss": 0.7955,
+      "step": 1668
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.8898751735687256,
+      "learning_rate": 9.055887140531743e-06,
+      "loss": 0.6936,
+      "step": 1669
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.4953198432922363,
+      "learning_rate": 9.061313076505698e-06,
+      "loss": 0.6811,
+      "step": 1670
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.047318935394287,
+      "learning_rate": 9.066739012479654e-06,
+      "loss": 0.6516,
+      "step": 1671
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.65636944770813,
+      "learning_rate": 9.072164948453609e-06,
+      "loss": 0.703,
+      "step": 1672
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.543917179107666,
+      "learning_rate": 9.077590884427563e-06,
+      "loss": 0.7124,
+      "step": 1673
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 4.660201549530029,
+      "learning_rate": 9.08301682040152e-06,
+      "loss": 0.6377,
+      "step": 1674
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.4524049758911133,
+      "learning_rate": 9.088442756375476e-06,
+      "loss": 0.5613,
+      "step": 1675
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.697133779525757,
+      "learning_rate": 9.09386869234943e-06,
+      "loss": 0.822,
+      "step": 1676
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 4.7428436279296875,
+      "learning_rate": 9.099294628323387e-06,
+      "loss": 0.9276,
+      "step": 1677
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.894984722137451,
+      "learning_rate": 9.104720564297343e-06,
+      "loss": 0.5152,
+      "step": 1678
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.3514325618743896,
+      "learning_rate": 9.110146500271298e-06,
+      "loss": 0.8001,
+      "step": 1679
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.612095355987549,
+      "learning_rate": 9.115572436245254e-06,
+      "loss": 0.5938,
+      "step": 1680
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 5.0532612800598145,
+      "learning_rate": 9.120998372219209e-06,
+      "loss": 0.6525,
+      "step": 1681
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.609055757522583,
+      "learning_rate": 9.126424308193163e-06,
+      "loss": 0.6809,
+      "step": 1682
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.9472272396087646,
+      "learning_rate": 9.13185024416712e-06,
+      "loss": 0.5882,
+      "step": 1683
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 4.99237585067749,
+      "learning_rate": 9.137276180141074e-06,
+      "loss": 0.5872,
+      "step": 1684
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.7243523597717285,
+      "learning_rate": 9.14270211611503e-06,
+      "loss": 0.7167,
+      "step": 1685
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 6.094716548919678,
+      "learning_rate": 9.148128052088985e-06,
+      "loss": 0.9357,
+      "step": 1686
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 4.222879886627197,
+      "learning_rate": 9.153553988062942e-06,
+      "loss": 0.7102,
+      "step": 1687
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.5418386459350586,
+      "learning_rate": 9.158979924036896e-06,
+      "loss": 0.782,
+      "step": 1688
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.140627145767212,
+      "learning_rate": 9.164405860010853e-06,
+      "loss": 0.6412,
+      "step": 1689
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.128467082977295,
+      "learning_rate": 9.169831795984809e-06,
+      "loss": 0.8287,
+      "step": 1690
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.929133415222168,
+      "learning_rate": 9.175257731958764e-06,
+      "loss": 0.6704,
+      "step": 1691
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.4044675827026367,
+      "learning_rate": 9.18068366793272e-06,
+      "loss": 0.7582,
+      "step": 1692
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.7450830936431885,
+      "learning_rate": 9.186109603906674e-06,
+      "loss": 0.5268,
+      "step": 1693
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.4473824501037598,
+      "learning_rate": 9.19153553988063e-06,
+      "loss": 0.6475,
+      "step": 1694
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 4.007517337799072,
+      "learning_rate": 9.196961475854585e-06,
+      "loss": 0.4139,
+      "step": 1695
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 4.543822288513184,
+      "learning_rate": 9.202387411828542e-06,
+      "loss": 0.8157,
+      "step": 1696
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 6.487241268157959,
+      "learning_rate": 9.207813347802496e-06,
+      "loss": 0.8445,
+      "step": 1697
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.979092836380005,
+      "learning_rate": 9.213239283776451e-06,
+      "loss": 0.4568,
+      "step": 1698
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.1116373538970947,
+      "learning_rate": 9.218665219750407e-06,
+      "loss": 0.6718,
+      "step": 1699
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.879091501235962,
+      "learning_rate": 9.224091155724364e-06,
+      "loss": 0.5992,
+      "step": 1700
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.5810892581939697,
+      "learning_rate": 9.229517091698318e-06,
+      "loss": 0.6119,
+      "step": 1701
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.221078634262085,
+      "learning_rate": 9.234943027672275e-06,
+      "loss": 0.5106,
+      "step": 1702
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 8.434675216674805,
+      "learning_rate": 9.240368963646231e-06,
+      "loss": 0.5885,
+      "step": 1703
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.419058084487915,
+      "learning_rate": 9.245794899620186e-06,
+      "loss": 0.7735,
+      "step": 1704
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.386651039123535,
+      "learning_rate": 9.25122083559414e-06,
+      "loss": 0.8587,
+      "step": 1705
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.3596432209014893,
+      "learning_rate": 9.256646771568096e-06,
+      "loss": 0.6949,
+      "step": 1706
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 4.073123931884766,
+      "learning_rate": 9.262072707542051e-06,
+      "loss": 0.7257,
+      "step": 1707
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.9874444007873535,
+      "learning_rate": 9.267498643516007e-06,
+      "loss": 0.4301,
+      "step": 1708
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.9584319591522217,
+      "learning_rate": 9.272924579489962e-06,
+      "loss": 0.5682,
+      "step": 1709
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.786191940307617,
+      "learning_rate": 9.278350515463918e-06,
+      "loss": 0.8334,
+      "step": 1710
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.6433796882629395,
+      "learning_rate": 9.283776451437873e-06,
+      "loss": 0.7495,
+      "step": 1711
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 4.192656993865967,
+      "learning_rate": 9.28920238741183e-06,
+      "loss": 0.5737,
+      "step": 1712
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 4.992117881774902,
+      "learning_rate": 9.294628323385784e-06,
+      "loss": 0.455,
+      "step": 1713
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.976194381713867,
+      "learning_rate": 9.30005425935974e-06,
+      "loss": 0.861,
+      "step": 1714
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.3122808933258057,
+      "learning_rate": 9.305480195333697e-06,
+      "loss": 0.4991,
+      "step": 1715
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.815927743911743,
+      "learning_rate": 9.310906131307651e-06,
+      "loss": 0.519,
+      "step": 1716
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.5537590980529785,
+      "learning_rate": 9.316332067281608e-06,
+      "loss": 0.5207,
+      "step": 1717
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.640167713165283,
+      "learning_rate": 9.321758003255562e-06,
+      "loss": 0.6613,
+      "step": 1718
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.242067337036133,
+      "learning_rate": 9.327183939229519e-06,
+      "loss": 0.5778,
+      "step": 1719
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.8758444786071777,
+      "learning_rate": 9.332609875203473e-06,
+      "loss": 0.4604,
+      "step": 1720
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 6.906590938568115,
+      "learning_rate": 9.33803581117743e-06,
+      "loss": 0.6167,
+      "step": 1721
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.655463695526123,
+      "learning_rate": 9.343461747151384e-06,
+      "loss": 0.6578,
+      "step": 1722
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.908991575241089,
+      "learning_rate": 9.348887683125339e-06,
+      "loss": 0.8432,
+      "step": 1723
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.5538487434387207,
+      "learning_rate": 9.354313619099295e-06,
+      "loss": 0.656,
+      "step": 1724
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.0368072986602783,
+      "learning_rate": 9.359739555073251e-06,
+      "loss": 0.9236,
+      "step": 1725
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.182832717895508,
+      "learning_rate": 9.365165491047206e-06,
+      "loss": 0.6787,
+      "step": 1726
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 4.960565567016602,
+      "learning_rate": 9.370591427021162e-06,
+      "loss": 0.8423,
+      "step": 1727
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.1211769580841064,
+      "learning_rate": 9.376017362995119e-06,
+      "loss": 0.8897,
+      "step": 1728
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.281097173690796,
+      "learning_rate": 9.381443298969073e-06,
+      "loss": 0.6827,
+      "step": 1729
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 4.726823329925537,
+      "learning_rate": 9.386869234943028e-06,
+      "loss": 0.6382,
+      "step": 1730
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.9552600383758545,
+      "learning_rate": 9.392295170916984e-06,
+      "loss": 0.4621,
+      "step": 1731
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 6.016453266143799,
+      "learning_rate": 9.397721106890939e-06,
+      "loss": 0.7661,
+      "step": 1732
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.6233997344970703,
+      "learning_rate": 9.403147042864895e-06,
+      "loss": 0.6608,
+      "step": 1733
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 6.995673179626465,
+      "learning_rate": 9.40857297883885e-06,
+      "loss": 0.7057,
+      "step": 1734
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 4.487472057342529,
+      "learning_rate": 9.413998914812806e-06,
+      "loss": 0.4997,
+      "step": 1735
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 4.529114246368408,
+      "learning_rate": 9.41942485078676e-06,
+      "loss": 0.627,
+      "step": 1736
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.9340765476226807,
+      "learning_rate": 9.424850786760717e-06,
+      "loss": 0.8103,
+      "step": 1737
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 5.830662727355957,
+      "learning_rate": 9.430276722734672e-06,
+      "loss": 0.6037,
+      "step": 1738
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.4757239818573,
+      "learning_rate": 9.435702658708628e-06,
+      "loss": 0.7293,
+      "step": 1739
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.404325485229492,
+      "learning_rate": 9.441128594682584e-06,
+      "loss": 0.7054,
+      "step": 1740
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 4.333992958068848,
+      "learning_rate": 9.446554530656539e-06,
+      "loss": 0.6715,
+      "step": 1741
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.895162343978882,
+      "learning_rate": 9.451980466630495e-06,
+      "loss": 0.7227,
+      "step": 1742
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.952202796936035,
+      "learning_rate": 9.45740640260445e-06,
+      "loss": 0.6112,
+      "step": 1743
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.1890549659729004,
+      "learning_rate": 9.462832338578406e-06,
+      "loss": 0.6377,
+      "step": 1744
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.8928492069244385,
+      "learning_rate": 9.468258274552361e-06,
+      "loss": 0.6164,
+      "step": 1745
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 4.169036865234375,
+      "learning_rate": 9.473684210526315e-06,
+      "loss": 0.6156,
+      "step": 1746
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.6523630619049072,
+      "learning_rate": 9.479110146500272e-06,
+      "loss": 0.6503,
+      "step": 1747
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.9472200870513916,
+      "learning_rate": 9.484536082474226e-06,
+      "loss": 0.5971,
+      "step": 1748
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 5.772794246673584,
+      "learning_rate": 9.489962018448183e-06,
+      "loss": 0.5531,
+      "step": 1749
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.6502020359039307,
+      "learning_rate": 9.495387954422139e-06,
+      "loss": 0.9651,
+      "step": 1750
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.137848138809204,
+      "learning_rate": 9.500813890396094e-06,
+      "loss": 0.5925,
+      "step": 1751
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.4244492053985596,
+      "learning_rate": 9.50623982637005e-06,
+      "loss": 0.8444,
+      "step": 1752
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.5253188610076904,
+      "learning_rate": 9.511665762344006e-06,
+      "loss": 0.7891,
+      "step": 1753
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.0059850215911865,
+      "learning_rate": 9.517091698317961e-06,
+      "loss": 0.7275,
+      "step": 1754
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 5.406351089477539,
+      "learning_rate": 9.522517634291916e-06,
+      "loss": 0.654,
+      "step": 1755
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.4339001178741455,
+      "learning_rate": 9.527943570265872e-06,
+      "loss": 0.7752,
+      "step": 1756
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.544710874557495,
+      "learning_rate": 9.533369506239827e-06,
+      "loss": 0.6697,
+      "step": 1757
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.563178539276123,
+      "learning_rate": 9.538795442213783e-06,
+      "loss": 0.9266,
+      "step": 1758
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.0594966411590576,
+      "learning_rate": 9.544221378187738e-06,
+      "loss": 0.8094,
+      "step": 1759
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 8.39729118347168,
+      "learning_rate": 9.549647314161694e-06,
+      "loss": 0.793,
+      "step": 1760
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.4687299728393555,
+      "learning_rate": 9.555073250135648e-06,
+      "loss": 0.6825,
+      "step": 1761
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.525848150253296,
+      "learning_rate": 9.560499186109605e-06,
+      "loss": 0.564,
+      "step": 1762
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 4.174354076385498,
+      "learning_rate": 9.565925122083561e-06,
+      "loss": 0.9108,
+      "step": 1763
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.854282855987549,
+      "learning_rate": 9.571351058057516e-06,
+      "loss": 0.873,
+      "step": 1764
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 4.618694305419922,
+      "learning_rate": 9.576776994031472e-06,
+      "loss": 0.6983,
+      "step": 1765
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.341549873352051,
+      "learning_rate": 9.582202930005427e-06,
+      "loss": 0.7548,
+      "step": 1766
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.332468032836914,
+      "learning_rate": 9.587628865979383e-06,
+      "loss": 0.6754,
+      "step": 1767
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 4.051456928253174,
+      "learning_rate": 9.593054801953338e-06,
+      "loss": 0.7722,
+      "step": 1768
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.6788055896759033,
+      "learning_rate": 9.598480737927294e-06,
+      "loss": 0.5777,
+      "step": 1769
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.9763264656066895,
+      "learning_rate": 9.603906673901249e-06,
+      "loss": 0.6669,
+      "step": 1770
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 7.09104061126709,
+      "learning_rate": 9.609332609875203e-06,
+      "loss": 0.785,
+      "step": 1771
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.9005157947540283,
+      "learning_rate": 9.61475854584916e-06,
+      "loss": 0.8804,
+      "step": 1772
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.9945383071899414,
+      "learning_rate": 9.620184481823114e-06,
+      "loss": 0.4325,
+      "step": 1773
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.656789779663086,
+      "learning_rate": 9.62561041779707e-06,
+      "loss": 0.6423,
+      "step": 1774
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 4.308764457702637,
+      "learning_rate": 9.631036353771027e-06,
+      "loss": 0.4256,
+      "step": 1775
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.8866770267486572,
+      "learning_rate": 9.636462289744981e-06,
+      "loss": 0.5113,
+      "step": 1776
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.756580114364624,
+      "learning_rate": 9.641888225718938e-06,
+      "loss": 0.6815,
+      "step": 1777
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.408679962158203,
+      "learning_rate": 9.647314161692894e-06,
+      "loss": 0.7424,
+      "step": 1778
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.6676297187805176,
+      "learning_rate": 9.652740097666849e-06,
+      "loss": 0.6109,
+      "step": 1779
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.1593780517578125,
+      "learning_rate": 9.658166033640803e-06,
+      "loss": 0.7153,
+      "step": 1780
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.323209762573242,
+      "learning_rate": 9.66359196961476e-06,
+      "loss": 0.3249,
+      "step": 1781
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 6.187316417694092,
+      "learning_rate": 9.669017905588714e-06,
+      "loss": 0.5032,
+      "step": 1782
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 5.965433597564697,
+      "learning_rate": 9.67444384156267e-06,
+      "loss": 0.7485,
+      "step": 1783
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.778571367263794,
+      "learning_rate": 9.679869777536625e-06,
+      "loss": 0.7275,
+      "step": 1784
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.5031983852386475,
+      "learning_rate": 9.685295713510582e-06,
+      "loss": 0.4372,
+      "step": 1785
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.2518508434295654,
+      "learning_rate": 9.690721649484536e-06,
+      "loss": 0.6687,
+      "step": 1786
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 4.471490859985352,
+      "learning_rate": 9.696147585458493e-06,
+      "loss": 0.7348,
+      "step": 1787
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 4.4109039306640625,
+      "learning_rate": 9.701573521432449e-06,
+      "loss": 0.4799,
+      "step": 1788
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.30936861038208,
+      "learning_rate": 9.706999457406403e-06,
+      "loss": 0.4332,
+      "step": 1789
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 4.088038921356201,
+      "learning_rate": 9.71242539338036e-06,
+      "loss": 0.6195,
+      "step": 1790
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 4.082264423370361,
+      "learning_rate": 9.717851329354314e-06,
+      "loss": 0.5352,
+      "step": 1791
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.2775166034698486,
+      "learning_rate": 9.72327726532827e-06,
+      "loss": 0.6742,
+      "step": 1792
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.653582811355591,
+      "learning_rate": 9.728703201302225e-06,
+      "loss": 0.7298,
+      "step": 1793
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.7703022956848145,
+      "learning_rate": 9.734129137276182e-06,
+      "loss": 0.611,
+      "step": 1794
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.2875425815582275,
+      "learning_rate": 9.739555073250136e-06,
+      "loss": 0.8995,
+      "step": 1795
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.164185047149658,
+      "learning_rate": 9.744981009224091e-06,
+      "loss": 0.8419,
+      "step": 1796
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.454408645629883,
+      "learning_rate": 9.750406945198047e-06,
+      "loss": 0.527,
+      "step": 1797
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 5.212310314178467,
+      "learning_rate": 9.755832881172002e-06,
+      "loss": 0.6337,
+      "step": 1798
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 6.680805206298828,
+      "learning_rate": 9.761258817145958e-06,
+      "loss": 0.7109,
+      "step": 1799
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.4182300567626953,
+      "learning_rate": 9.766684753119915e-06,
+      "loss": 0.856,
+      "step": 1800
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.5877883434295654,
+      "learning_rate": 9.77211068909387e-06,
+      "loss": 0.6187,
+      "step": 1801
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.424466848373413,
+      "learning_rate": 9.777536625067825e-06,
+      "loss": 0.5687,
+      "step": 1802
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.955275535583496,
+      "learning_rate": 9.78296256104178e-06,
+      "loss": 0.7171,
+      "step": 1803
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.830179452896118,
+      "learning_rate": 9.788388497015736e-06,
+      "loss": 0.8626,
+      "step": 1804
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.7441303730010986,
+      "learning_rate": 9.793814432989691e-06,
+      "loss": 0.862,
+      "step": 1805
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.3291001319885254,
+      "learning_rate": 9.799240368963647e-06,
+      "loss": 0.8376,
+      "step": 1806
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.1355652809143066,
+      "learning_rate": 9.804666304937602e-06,
+      "loss": 0.8709,
+      "step": 1807
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.960632085800171,
+      "learning_rate": 9.810092240911558e-06,
+      "loss": 1.1718,
+      "step": 1808
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.047135829925537,
+      "learning_rate": 9.815518176885513e-06,
+      "loss": 0.7516,
+      "step": 1809
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.9338934421539307,
+      "learning_rate": 9.82094411285947e-06,
+      "loss": 0.6045,
+      "step": 1810
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.249512195587158,
+      "learning_rate": 9.826370048833424e-06,
+      "loss": 0.5102,
+      "step": 1811
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.8852379322052,
+      "learning_rate": 9.83179598480738e-06,
+      "loss": 0.698,
+      "step": 1812
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 5.112749099731445,
+      "learning_rate": 9.837221920781337e-06,
+      "loss": 0.8115,
+      "step": 1813
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.1928462982177734,
+      "learning_rate": 9.842647856755291e-06,
+      "loss": 0.5809,
+      "step": 1814
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.0574264526367188,
+      "learning_rate": 9.848073792729248e-06,
+      "loss": 0.7249,
+      "step": 1815
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.7749717235565186,
+      "learning_rate": 9.853499728703202e-06,
+      "loss": 0.2828,
+      "step": 1816
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.920943260192871,
+      "learning_rate": 9.858925664677158e-06,
+      "loss": 0.7663,
+      "step": 1817
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.2379109859466553,
+      "learning_rate": 9.864351600651113e-06,
+      "loss": 0.7444,
+      "step": 1818
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.661921977996826,
+      "learning_rate": 9.869777536625068e-06,
+      "loss": 0.624,
+      "step": 1819
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 5.578468322753906,
+      "learning_rate": 9.875203472599024e-06,
+      "loss": 0.6554,
+      "step": 1820
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.5147907733917236,
+      "learning_rate": 9.880629408572979e-06,
+      "loss": 0.4755,
+      "step": 1821
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.776432752609253,
+      "learning_rate": 9.886055344546935e-06,
+      "loss": 0.7682,
+      "step": 1822
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.3738656044006348,
+      "learning_rate": 9.89148128052089e-06,
+      "loss": 1.0733,
+      "step": 1823
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 4.594886779785156,
+      "learning_rate": 9.896907216494846e-06,
+      "loss": 0.6723,
+      "step": 1824
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.096754550933838,
+      "learning_rate": 9.902333152468802e-06,
+      "loss": 0.7331,
+      "step": 1825
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.498154640197754,
+      "learning_rate": 9.907759088442757e-06,
+      "loss": 0.6258,
+      "step": 1826
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.476153612136841,
+      "learning_rate": 9.913185024416713e-06,
+      "loss": 0.6899,
+      "step": 1827
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 5.060966968536377,
+      "learning_rate": 9.918610960390668e-06,
+      "loss": 0.5489,
+      "step": 1828
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.252438545227051,
+      "learning_rate": 9.924036896364624e-06,
+      "loss": 0.763,
+      "step": 1829
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 4.415834903717041,
+      "learning_rate": 9.929462832338579e-06,
+      "loss": 0.6687,
+      "step": 1830
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.4314076900482178,
+      "learning_rate": 9.934888768312535e-06,
+      "loss": 0.8018,
+      "step": 1831
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 4.767874717712402,
+      "learning_rate": 9.94031470428649e-06,
+      "loss": 0.6868,
+      "step": 1832
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.444445848464966,
+      "learning_rate": 9.945740640260446e-06,
+      "loss": 0.4949,
+      "step": 1833
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.2775983810424805,
+      "learning_rate": 9.9511665762344e-06,
+      "loss": 0.3642,
+      "step": 1834
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 6.116227626800537,
+      "learning_rate": 9.956592512208357e-06,
+      "loss": 0.7161,
+      "step": 1835
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.0087554454803467,
+      "learning_rate": 9.962018448182312e-06,
+      "loss": 0.7487,
+      "step": 1836
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 10.492149353027344,
+      "learning_rate": 9.967444384156268e-06,
+      "loss": 0.5916,
+      "step": 1837
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.0715484619140625,
+      "learning_rate": 9.972870320130224e-06,
+      "loss": 0.4598,
+      "step": 1838
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.3793256282806396,
+      "learning_rate": 9.978296256104179e-06,
+      "loss": 0.3799,
+      "step": 1839
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.3124024868011475,
+      "learning_rate": 9.983722192078135e-06,
+      "loss": 0.6886,
+      "step": 1840
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.3899738788604736,
+      "learning_rate": 9.98914812805209e-06,
+      "loss": 0.4609,
+      "step": 1841
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.5008697509765625,
+      "learning_rate": 9.994574064026046e-06,
+      "loss": 0.9251,
+      "step": 1842
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 4.515383243560791,
+      "learning_rate": 1e-05,
+      "loss": 0.4588,
+      "step": 1843
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.3006043434143066,
+      "learning_rate": 9.999999993048437e-06,
+      "loss": 0.9247,
+      "step": 1844
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 4.554304599761963,
+      "learning_rate": 9.999999972193747e-06,
+      "loss": 0.396,
+      "step": 1845
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.7557485103607178,
+      "learning_rate": 9.99999993743593e-06,
+      "loss": 0.4951,
+      "step": 1846
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.0979342460632324,
+      "learning_rate": 9.999999888774985e-06,
+      "loss": 0.584,
+      "step": 1847
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 11.040298461914062,
+      "learning_rate": 9.999999826210915e-06,
+      "loss": 0.4738,
+      "step": 1848
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.9289350509643555,
+      "learning_rate": 9.999999749743717e-06,
+      "loss": 0.6776,
+      "step": 1849
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.2187752723693848,
+      "learning_rate": 9.999999659373394e-06,
+      "loss": 0.5943,
+      "step": 1850
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.3365907669067383,
+      "learning_rate": 9.999999555099943e-06,
+      "loss": 0.6296,
+      "step": 1851
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 4.140570163726807,
+      "learning_rate": 9.999999436923368e-06,
+      "loss": 0.9248,
+      "step": 1852
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.7536301612854004,
+      "learning_rate": 9.999999304843668e-06,
+      "loss": 0.5894,
+      "step": 1853
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.0149245262145996,
+      "learning_rate": 9.999999158860843e-06,
+      "loss": 0.6597,
+      "step": 1854
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.228769540786743,
+      "learning_rate": 9.999998998974893e-06,
+      "loss": 0.4648,
+      "step": 1855
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.1621456146240234,
+      "learning_rate": 9.99999882518582e-06,
+      "loss": 0.546,
+      "step": 1856
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 4.2027764320373535,
+      "learning_rate": 9.99999863749362e-06,
+      "loss": 0.5033,
+      "step": 1857
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.5926151275634766,
+      "learning_rate": 9.9999984358983e-06,
+      "loss": 0.7833,
+      "step": 1858
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.246143102645874,
+      "learning_rate": 9.999998220399855e-06,
+      "loss": 0.6178,
+      "step": 1859
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.7528457641601562,
+      "learning_rate": 9.99999799099829e-06,
+      "loss": 0.5954,
+      "step": 1860
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.2709076404571533,
+      "learning_rate": 9.999997747693602e-06,
+      "loss": 0.8972,
+      "step": 1861
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.735793352127075,
+      "learning_rate": 9.999997490485795e-06,
+      "loss": 0.7919,
+      "step": 1862
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.505697011947632,
+      "learning_rate": 9.999997219374866e-06,
+      "loss": 0.5833,
+      "step": 1863
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.2934417724609375,
+      "learning_rate": 9.99999693436082e-06,
+      "loss": 0.7329,
+      "step": 1864
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.6806840896606445,
+      "learning_rate": 9.999996635443654e-06,
+      "loss": 0.9,
+      "step": 1865
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 6.125889301300049,
+      "learning_rate": 9.99999632262337e-06,
+      "loss": 0.8755,
+      "step": 1866
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.3255372047424316,
+      "learning_rate": 9.99999599589997e-06,
+      "loss": 0.7171,
+      "step": 1867
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 4.132991790771484,
+      "learning_rate": 9.999995655273455e-06,
+      "loss": 0.5898,
+      "step": 1868
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.025501251220703,
+      "learning_rate": 9.999995300743824e-06,
+      "loss": 0.8131,
+      "step": 1869
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.704444408416748,
+      "learning_rate": 9.999994932311079e-06,
+      "loss": 0.7314,
+      "step": 1870
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.1663126945495605,
+      "learning_rate": 9.999994549975222e-06,
+      "loss": 0.5969,
+      "step": 1871
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.631037473678589,
+      "learning_rate": 9.999994153736254e-06,
+      "loss": 0.7718,
+      "step": 1872
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.447580099105835,
+      "learning_rate": 9.999993743594175e-06,
+      "loss": 0.5932,
+      "step": 1873
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.386505365371704,
+      "learning_rate": 9.999993319548983e-06,
+      "loss": 0.6443,
+      "step": 1874
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.78316330909729,
+      "learning_rate": 9.999992881600686e-06,
+      "loss": 0.7457,
+      "step": 1875
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.51059627532959,
+      "learning_rate": 9.999992429749281e-06,
+      "loss": 0.6944,
+      "step": 1876
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 4.292200565338135,
+      "learning_rate": 9.999991963994771e-06,
+      "loss": 0.5197,
+      "step": 1877
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.972071647644043,
+      "learning_rate": 9.999991484337156e-06,
+      "loss": 0.5308,
+      "step": 1878
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.052499532699585,
+      "learning_rate": 9.999990990776436e-06,
+      "loss": 0.574,
+      "step": 1879
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 4.062962532043457,
+      "learning_rate": 9.999990483312616e-06,
+      "loss": 0.6002,
+      "step": 1880
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 4.205875873565674,
+      "learning_rate": 9.999989961945695e-06,
+      "loss": 0.6741,
+      "step": 1881
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.252249240875244,
+      "learning_rate": 9.999989426675673e-06,
+      "loss": 0.6623,
+      "step": 1882
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.8825366497039795,
+      "learning_rate": 9.999988877502555e-06,
+      "loss": 0.7642,
+      "step": 1883
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.182142734527588,
+      "learning_rate": 9.999988314426343e-06,
+      "loss": 0.6479,
+      "step": 1884
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.687919855117798,
+      "learning_rate": 9.999987737447035e-06,
+      "loss": 0.7635,
+      "step": 1885
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.3320233821868896,
+      "learning_rate": 9.999987146564634e-06,
+      "loss": 0.7641,
+      "step": 1886
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 5.901373386383057,
+      "learning_rate": 9.99998654177914e-06,
+      "loss": 0.7024,
+      "step": 1887
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 11.913612365722656,
+      "learning_rate": 9.999985923090558e-06,
+      "loss": 0.6148,
+      "step": 1888
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.758005380630493,
+      "learning_rate": 9.999985290498889e-06,
+      "loss": 0.6397,
+      "step": 1889
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.892709255218506,
+      "learning_rate": 9.999984644004132e-06,
+      "loss": 0.7066,
+      "step": 1890
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.537569999694824,
+      "learning_rate": 9.999983983606294e-06,
+      "loss": 0.748,
+      "step": 1891
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.2337169647216797,
+      "learning_rate": 9.999983309305371e-06,
+      "loss": 0.4822,
+      "step": 1892
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.7192790508270264,
+      "learning_rate": 9.999982621101367e-06,
+      "loss": 0.6063,
+      "step": 1893
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 5.730989456176758,
+      "learning_rate": 9.999981918994285e-06,
+      "loss": 0.6039,
+      "step": 1894
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.7329015731811523,
+      "learning_rate": 9.999981202984128e-06,
+      "loss": 0.6571,
+      "step": 1895
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.5178816318511963,
+      "learning_rate": 9.999980473070896e-06,
+      "loss": 0.8414,
+      "step": 1896
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.2742934226989746,
+      "learning_rate": 9.99997972925459e-06,
+      "loss": 0.562,
+      "step": 1897
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.585723876953125,
+      "learning_rate": 9.999978971535214e-06,
+      "loss": 0.6633,
+      "step": 1898
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.844719886779785,
+      "learning_rate": 9.999978199912768e-06,
+      "loss": 0.698,
+      "step": 1899
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.4317800998687744,
+      "learning_rate": 9.999977414387257e-06,
+      "loss": 0.5141,
+      "step": 1900
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 4.768895149230957,
+      "learning_rate": 9.999976614958681e-06,
+      "loss": 0.7416,
+      "step": 1901
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.9366700649261475,
+      "learning_rate": 9.999975801627044e-06,
+      "loss": 0.6361,
+      "step": 1902
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 4.108584403991699,
+      "learning_rate": 9.999974974392348e-06,
+      "loss": 0.6512,
+      "step": 1903
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.8338496685028076,
+      "learning_rate": 9.999974133254594e-06,
+      "loss": 0.9405,
+      "step": 1904
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.470606565475464,
+      "learning_rate": 9.999973278213784e-06,
+      "loss": 0.5655,
+      "step": 1905
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.6650195121765137,
+      "learning_rate": 9.999972409269922e-06,
+      "loss": 0.6003,
+      "step": 1906
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 7.051216125488281,
+      "learning_rate": 9.999971526423011e-06,
+      "loss": 0.7899,
+      "step": 1907
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.960444450378418,
+      "learning_rate": 9.99997062967305e-06,
+      "loss": 1.0074,
+      "step": 1908
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.1542069911956787,
+      "learning_rate": 9.999969719020046e-06,
+      "loss": 0.7021,
+      "step": 1909
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.764927864074707,
+      "learning_rate": 9.999968794463997e-06,
+      "loss": 1.0001,
+      "step": 1910
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 4.5328474044799805,
+      "learning_rate": 9.999967856004909e-06,
+      "loss": 0.6864,
+      "step": 1911
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.65494966506958,
+      "learning_rate": 9.999966903642784e-06,
+      "loss": 0.7766,
+      "step": 1912
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.908613920211792,
+      "learning_rate": 9.999965937377622e-06,
+      "loss": 0.6876,
+      "step": 1913
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.857848644256592,
+      "learning_rate": 9.99996495720943e-06,
+      "loss": 0.5676,
+      "step": 1914
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.2573204040527344,
+      "learning_rate": 9.999963963138208e-06,
+      "loss": 0.6857,
+      "step": 1915
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 6.764009952545166,
+      "learning_rate": 9.999962955163959e-06,
+      "loss": 0.827,
+      "step": 1916
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 19.840639114379883,
+      "learning_rate": 9.999961933286686e-06,
+      "loss": 0.6341,
+      "step": 1917
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 4.046363830566406,
+      "learning_rate": 9.999960897506392e-06,
+      "loss": 0.6455,
+      "step": 1918
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.4658658504486084,
+      "learning_rate": 9.99995984782308e-06,
+      "loss": 0.8565,
+      "step": 1919
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 4.5952043533325195,
+      "learning_rate": 9.999958784236751e-06,
+      "loss": 0.6293,
+      "step": 1920
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 5.748356342315674,
+      "learning_rate": 9.999957706747411e-06,
+      "loss": 0.7658,
+      "step": 1921
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.024219274520874,
+      "learning_rate": 9.999956615355062e-06,
+      "loss": 0.5325,
+      "step": 1922
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.624185085296631,
+      "learning_rate": 9.999955510059706e-06,
+      "loss": 0.5726,
+      "step": 1923
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.7793099880218506,
+      "learning_rate": 9.999954390861348e-06,
+      "loss": 0.9698,
+      "step": 1924
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.3443424701690674,
+      "learning_rate": 9.999953257759989e-06,
+      "loss": 0.5412,
+      "step": 1925
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.996976375579834,
+      "learning_rate": 9.999952110755632e-06,
+      "loss": 0.7691,
+      "step": 1926
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.2687525749206543,
+      "learning_rate": 9.999950949848282e-06,
+      "loss": 0.5906,
+      "step": 1927
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.0645980834960938,
+      "learning_rate": 9.999949775037942e-06,
+      "loss": 0.5586,
+      "step": 1928
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.9746291637420654,
+      "learning_rate": 9.999948586324614e-06,
+      "loss": 0.8034,
+      "step": 1929
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 5.577073574066162,
+      "learning_rate": 9.999947383708301e-06,
+      "loss": 0.614,
+      "step": 1930
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.774261236190796,
+      "learning_rate": 9.99994616718901e-06,
+      "loss": 0.8572,
+      "step": 1931
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 8.652884483337402,
+      "learning_rate": 9.99994493676674e-06,
+      "loss": 0.8742,
+      "step": 1932
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.812756061553955,
+      "learning_rate": 9.999943692441496e-06,
+      "loss": 0.7759,
+      "step": 1933
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.2944910526275635,
+      "learning_rate": 9.999942434213282e-06,
+      "loss": 0.8238,
+      "step": 1934
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.956136703491211,
+      "learning_rate": 9.9999411620821e-06,
+      "loss": 0.8625,
+      "step": 1935
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.5341973304748535,
+      "learning_rate": 9.999939876047956e-06,
+      "loss": 0.6728,
+      "step": 1936
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 4.12306022644043,
+      "learning_rate": 9.999938576110853e-06,
+      "loss": 0.6806,
+      "step": 1937
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.1317317485809326,
+      "learning_rate": 9.999937262270794e-06,
+      "loss": 0.6,
+      "step": 1938
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 8.427379608154297,
+      "learning_rate": 9.999935934527781e-06,
+      "loss": 0.7363,
+      "step": 1939
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 12.257966995239258,
+      "learning_rate": 9.99993459288182e-06,
+      "loss": 0.7653,
+      "step": 1940
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 5.129399299621582,
+      "learning_rate": 9.999933237332914e-06,
+      "loss": 0.5688,
+      "step": 1941
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.1079742908477783,
+      "learning_rate": 9.999931867881064e-06,
+      "loss": 0.6326,
+      "step": 1942
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 7.311315059661865,
+      "learning_rate": 9.999930484526279e-06,
+      "loss": 0.5961,
+      "step": 1943
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.150193452835083,
+      "learning_rate": 9.999929087268561e-06,
+      "loss": 0.862,
+      "step": 1944
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.4593863487243652,
+      "learning_rate": 9.999927676107912e-06,
+      "loss": 0.7616,
+      "step": 1945
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 4.101261138916016,
+      "learning_rate": 9.999926251044338e-06,
+      "loss": 0.6491,
+      "step": 1946
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 4.009993076324463,
+      "learning_rate": 9.999924812077842e-06,
+      "loss": 0.6931,
+      "step": 1947
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.530946969985962,
+      "learning_rate": 9.999923359208426e-06,
+      "loss": 0.5332,
+      "step": 1948
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.9882895946502686,
+      "learning_rate": 9.999921892436098e-06,
+      "loss": 0.7288,
+      "step": 1949
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.4409921169281006,
+      "learning_rate": 9.99992041176086e-06,
+      "loss": 0.2678,
+      "step": 1950
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.3031256198883057,
+      "learning_rate": 9.999918917182717e-06,
+      "loss": 0.7668,
+      "step": 1951
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.7392584085464478,
+      "learning_rate": 9.999917408701671e-06,
+      "loss": 0.9298,
+      "step": 1952
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.0857975482940674,
+      "learning_rate": 9.999915886317729e-06,
+      "loss": 0.7141,
+      "step": 1953
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.20969820022583,
+      "learning_rate": 9.999914350030892e-06,
+      "loss": 0.5577,
+      "step": 1954
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.096522331237793,
+      "learning_rate": 9.999912799841169e-06,
+      "loss": 0.6569,
+      "step": 1955
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.6954081058502197,
+      "learning_rate": 9.99991123574856e-06,
+      "loss": 0.5766,
+      "step": 1956
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.8726295232772827,
+      "learning_rate": 9.999909657753071e-06,
+      "loss": 0.9125,
+      "step": 1957
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.587296962738037,
+      "learning_rate": 9.999908065854707e-06,
+      "loss": 0.6058,
+      "step": 1958
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 4.635515213012695,
+      "learning_rate": 9.99990646005347e-06,
+      "loss": 0.5168,
+      "step": 1959
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.2362184524536133,
+      "learning_rate": 9.999904840349368e-06,
+      "loss": 0.5406,
+      "step": 1960
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.1138532161712646,
+      "learning_rate": 9.999903206742403e-06,
+      "loss": 0.5257,
+      "step": 1961
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.029341459274292,
+      "learning_rate": 9.999901559232578e-06,
+      "loss": 0.6277,
+      "step": 1962
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 7.584851264953613,
+      "learning_rate": 9.999899897819902e-06,
+      "loss": 0.747,
+      "step": 1963
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 5.691926002502441,
+      "learning_rate": 9.999898222504378e-06,
+      "loss": 0.4836,
+      "step": 1964
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.8134994506835938,
+      "learning_rate": 9.999896533286009e-06,
+      "loss": 0.7492,
+      "step": 1965
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.3231394290924072,
+      "learning_rate": 9.9998948301648e-06,
+      "loss": 0.8472,
+      "step": 1966
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.781933546066284,
+      "learning_rate": 9.999893113140759e-06,
+      "loss": 0.7034,
+      "step": 1967
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.151702880859375,
+      "learning_rate": 9.999891382213886e-06,
+      "loss": 0.9363,
+      "step": 1968
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.9465811252593994,
+      "learning_rate": 9.999889637384187e-06,
+      "loss": 0.8347,
+      "step": 1969
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.0010054111480713,
+      "learning_rate": 9.99988787865167e-06,
+      "loss": 0.5015,
+      "step": 1970
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.5749926567077637,
+      "learning_rate": 9.999886106016338e-06,
+      "loss": 0.9349,
+      "step": 1971
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.385693073272705,
+      "learning_rate": 9.999884319478195e-06,
+      "loss": 0.7261,
+      "step": 1972
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.5567026138305664,
+      "learning_rate": 9.999882519037248e-06,
+      "loss": 0.6636,
+      "step": 1973
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 6.639320373535156,
+      "learning_rate": 9.999880704693498e-06,
+      "loss": 0.6981,
+      "step": 1974
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.9628942012786865,
+      "learning_rate": 9.999878876446956e-06,
+      "loss": 0.519,
+      "step": 1975
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.995103120803833,
+      "learning_rate": 9.999877034297622e-06,
+      "loss": 1.0712,
+      "step": 1976
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 6.27642297744751,
+      "learning_rate": 9.999875178245502e-06,
+      "loss": 0.7109,
+      "step": 1977
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.7704062461853027,
+      "learning_rate": 9.999873308290604e-06,
+      "loss": 0.6731,
+      "step": 1978
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 5.959547519683838,
+      "learning_rate": 9.99987142443293e-06,
+      "loss": 0.904,
+      "step": 1979
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 5.193602561950684,
+      "learning_rate": 9.999869526672487e-06,
+      "loss": 0.8841,
+      "step": 1980
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.732692241668701,
+      "learning_rate": 9.999867615009281e-06,
+      "loss": 0.633,
+      "step": 1981
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 5.306741237640381,
+      "learning_rate": 9.999865689443316e-06,
+      "loss": 0.7294,
+      "step": 1982
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.4366016387939453,
+      "learning_rate": 9.999863749974597e-06,
+      "loss": 0.4739,
+      "step": 1983
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.169332265853882,
+      "learning_rate": 9.999861796603129e-06,
+      "loss": 0.3768,
+      "step": 1984
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 4.43646764755249,
+      "learning_rate": 9.999859829328919e-06,
+      "loss": 0.5908,
+      "step": 1985
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.930811882019043,
+      "learning_rate": 9.99985784815197e-06,
+      "loss": 0.7182,
+      "step": 1986
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.955049514770508,
+      "learning_rate": 9.999855853072293e-06,
+      "loss": 0.707,
+      "step": 1987
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.695387363433838,
+      "learning_rate": 9.999853844089888e-06,
+      "loss": 0.6465,
+      "step": 1988
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.776087760925293,
+      "learning_rate": 9.999851821204761e-06,
+      "loss": 0.7952,
+      "step": 1989
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.522993564605713,
+      "learning_rate": 9.99984978441692e-06,
+      "loss": 0.776,
+      "step": 1990
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.7815260887145996,
+      "learning_rate": 9.99984773372637e-06,
+      "loss": 0.8309,
+      "step": 1991
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 4.273420333862305,
+      "learning_rate": 9.999845669133116e-06,
+      "loss": 0.7644,
+      "step": 1992
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 4.789883136749268,
+      "learning_rate": 9.999843590637166e-06,
+      "loss": 0.7235,
+      "step": 1993
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.318136215209961,
+      "learning_rate": 9.999841498238523e-06,
+      "loss": 0.9917,
+      "step": 1994
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.7531468868255615,
+      "learning_rate": 9.999839391937193e-06,
+      "loss": 0.5834,
+      "step": 1995
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 6.112000465393066,
+      "learning_rate": 9.999837271733184e-06,
+      "loss": 0.6841,
+      "step": 1996
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.1181437969207764,
+      "learning_rate": 9.999835137626499e-06,
+      "loss": 0.6043,
+      "step": 1997
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.2957441806793213,
+      "learning_rate": 9.999832989617147e-06,
+      "loss": 0.9037,
+      "step": 1998
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.9189624786376953,
+      "learning_rate": 9.99983082770513e-06,
+      "loss": 0.7586,
+      "step": 1999
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.1382248401641846,
+      "learning_rate": 9.99982865189046e-06,
+      "loss": 0.6028,
+      "step": 2000
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 10.056156158447266,
+      "learning_rate": 9.999826462173136e-06,
+      "loss": 0.5389,
+      "step": 2001
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.5580976009368896,
+      "learning_rate": 9.99982425855317e-06,
+      "loss": 0.8006,
+      "step": 2002
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.1691291332244873,
+      "learning_rate": 9.999822041030565e-06,
+      "loss": 0.532,
+      "step": 2003
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.8623478412628174,
+      "learning_rate": 9.999819809605327e-06,
+      "loss": 0.793,
+      "step": 2004
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.3109211921691895,
+      "learning_rate": 9.999817564277464e-06,
+      "loss": 0.5426,
+      "step": 2005
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 4.058021068572998,
+      "learning_rate": 9.999815305046981e-06,
+      "loss": 0.6925,
+      "step": 2006
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 5.732763290405273,
+      "learning_rate": 9.999813031913885e-06,
+      "loss": 0.6195,
+      "step": 2007
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.975738763809204,
+      "learning_rate": 9.999810744878183e-06,
+      "loss": 0.7132,
+      "step": 2008
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 8.483386993408203,
+      "learning_rate": 9.999808443939879e-06,
+      "loss": 0.4338,
+      "step": 2009
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.508713960647583,
+      "learning_rate": 9.999806129098978e-06,
+      "loss": 0.5838,
+      "step": 2010
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.0596866607666016,
+      "learning_rate": 9.999803800355493e-06,
+      "loss": 0.6112,
+      "step": 2011
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.0505290031433105,
+      "learning_rate": 9.999801457709425e-06,
+      "loss": 0.5945,
+      "step": 2012
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 6.89737606048584,
+      "learning_rate": 9.999799101160783e-06,
+      "loss": 0.6677,
+      "step": 2013
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.443990468978882,
+      "learning_rate": 9.999796730709572e-06,
+      "loss": 0.6536,
+      "step": 2014
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.4202961921691895,
+      "learning_rate": 9.9997943463558e-06,
+      "loss": 0.8173,
+      "step": 2015
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.171118974685669,
+      "learning_rate": 9.999791948099471e-06,
+      "loss": 0.7859,
+      "step": 2016
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.8628451824188232,
+      "learning_rate": 9.999789535940596e-06,
+      "loss": 0.5901,
+      "step": 2017
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.1360177993774414,
+      "learning_rate": 9.999787109879177e-06,
+      "loss": 0.7615,
+      "step": 2018
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.2618205547332764,
+      "learning_rate": 9.999784669915224e-06,
+      "loss": 0.5487,
+      "step": 2019
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 5.658918857574463,
+      "learning_rate": 9.999782216048743e-06,
+      "loss": 0.9501,
+      "step": 2020
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.4553675651550293,
+      "learning_rate": 9.99977974827974e-06,
+      "loss": 0.7687,
+      "step": 2021
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.0767951011657715,
+      "learning_rate": 9.999777266608224e-06,
+      "loss": 0.5579,
+      "step": 2022
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.016173839569092,
+      "learning_rate": 9.9997747710342e-06,
+      "loss": 0.5622,
+      "step": 2023
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.365206003189087,
+      "learning_rate": 9.999772261557674e-06,
+      "loss": 0.4651,
+      "step": 2024
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.3888893127441406,
+      "learning_rate": 9.999769738178655e-06,
+      "loss": 0.644,
+      "step": 2025
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 4.084037780761719,
+      "learning_rate": 9.999767200897148e-06,
+      "loss": 0.6539,
+      "step": 2026
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.474219560623169,
+      "learning_rate": 9.999764649713163e-06,
+      "loss": 0.6827,
+      "step": 2027
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.4283528327941895,
+      "learning_rate": 9.999762084626706e-06,
+      "loss": 0.7608,
+      "step": 2028
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.779902458190918,
+      "learning_rate": 9.999759505637783e-06,
+      "loss": 0.8083,
+      "step": 2029
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.9722800254821777,
+      "learning_rate": 9.999756912746402e-06,
+      "loss": 1.052,
+      "step": 2030
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.1768105030059814,
+      "learning_rate": 9.999754305952569e-06,
+      "loss": 0.6233,
+      "step": 2031
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.7593259811401367,
+      "learning_rate": 9.999751685256293e-06,
+      "loss": 0.6085,
+      "step": 2032
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 4.1510009765625,
+      "learning_rate": 9.99974905065758e-06,
+      "loss": 0.8367,
+      "step": 2033
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.922394037246704,
+      "learning_rate": 9.999746402156438e-06,
+      "loss": 0.618,
+      "step": 2034
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.196125030517578,
+      "learning_rate": 9.999743739752875e-06,
+      "loss": 0.7117,
+      "step": 2035
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.5382087230682373,
+      "learning_rate": 9.999741063446897e-06,
+      "loss": 0.4579,
+      "step": 2036
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.464616298675537,
+      "learning_rate": 9.999738373238511e-06,
+      "loss": 0.9218,
+      "step": 2037
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 6.80758810043335,
+      "learning_rate": 9.999735669127726e-06,
+      "loss": 0.6212,
+      "step": 2038
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.0510752201080322,
+      "learning_rate": 9.999732951114551e-06,
+      "loss": 0.6966,
+      "step": 2039
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.25075626373291,
+      "learning_rate": 9.999730219198988e-06,
+      "loss": 0.8312,
+      "step": 2040
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.23608660697937,
+      "learning_rate": 9.999727473381052e-06,
+      "loss": 0.7662,
+      "step": 2041
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.30147385597229,
+      "learning_rate": 9.999724713660744e-06,
+      "loss": 0.9522,
+      "step": 2042
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.376685619354248,
+      "learning_rate": 9.999721940038076e-06,
+      "loss": 0.8466,
+      "step": 2043
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.302940607070923,
+      "learning_rate": 9.999719152513055e-06,
+      "loss": 0.7793,
+      "step": 2044
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 5.625790596008301,
+      "learning_rate": 9.999716351085686e-06,
+      "loss": 0.8068,
+      "step": 2045
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.366903781890869,
+      "learning_rate": 9.99971353575598e-06,
+      "loss": 0.6186,
+      "step": 2046
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 4.667184352874756,
+      "learning_rate": 9.999710706523943e-06,
+      "loss": 0.685,
+      "step": 2047
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 5.232559680938721,
+      "learning_rate": 9.999707863389583e-06,
+      "loss": 0.5966,
+      "step": 2048
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.5140938758850098,
+      "learning_rate": 9.999705006352909e-06,
+      "loss": 0.7867,
+      "step": 2049
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.952310562133789,
+      "learning_rate": 9.99970213541393e-06,
+      "loss": 0.7979,
+      "step": 2050
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.120920181274414,
+      "learning_rate": 9.99969925057265e-06,
+      "loss": 0.6555,
+      "step": 2051
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.131239891052246,
+      "learning_rate": 9.99969635182908e-06,
+      "loss": 0.3411,
+      "step": 2052
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.337313175201416,
+      "learning_rate": 9.999693439183227e-06,
+      "loss": 0.7887,
+      "step": 2053
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.2587759494781494,
+      "learning_rate": 9.9996905126351e-06,
+      "loss": 0.5906,
+      "step": 2054
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.014998197555542,
+      "learning_rate": 9.999687572184706e-06,
+      "loss": 0.65,
+      "step": 2055
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.6224591732025146,
+      "learning_rate": 9.999684617832055e-06,
+      "loss": 0.8307,
+      "step": 2056
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.7038650512695312,
+      "learning_rate": 9.999681649577153e-06,
+      "loss": 0.5628,
+      "step": 2057
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.527803659439087,
+      "learning_rate": 9.999678667420009e-06,
+      "loss": 0.787,
+      "step": 2058
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.4924204349517822,
+      "learning_rate": 9.999675671360632e-06,
+      "loss": 0.5151,
+      "step": 2059
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 4.464627265930176,
+      "learning_rate": 9.99967266139903e-06,
+      "loss": 0.4647,
+      "step": 2060
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.8122942447662354,
+      "learning_rate": 9.999669637535211e-06,
+      "loss": 0.6382,
+      "step": 2061
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.7971107959747314,
+      "learning_rate": 9.999666599769184e-06,
+      "loss": 0.6518,
+      "step": 2062
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.6076838970184326,
+      "learning_rate": 9.999663548100957e-06,
+      "loss": 0.5214,
+      "step": 2063
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.8105340003967285,
+      "learning_rate": 9.999660482530537e-06,
+      "loss": 0.6278,
+      "step": 2064
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.3558733463287354,
+      "learning_rate": 9.999657403057937e-06,
+      "loss": 0.7966,
+      "step": 2065
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.2570362091064453,
+      "learning_rate": 9.999654309683161e-06,
+      "loss": 0.6453,
+      "step": 2066
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 4.119926929473877,
+      "learning_rate": 9.99965120240622e-06,
+      "loss": 0.8743,
+      "step": 2067
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.092893123626709,
+      "learning_rate": 9.999648081227122e-06,
+      "loss": 0.6048,
+      "step": 2068
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.008148670196533,
+      "learning_rate": 9.999644946145874e-06,
+      "loss": 0.6658,
+      "step": 2069
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.5009684562683105,
+      "learning_rate": 9.999641797162487e-06,
+      "loss": 0.5151,
+      "step": 2070
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.446073293685913,
+      "learning_rate": 9.99963863427697e-06,
+      "loss": 0.7587,
+      "step": 2071
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.9923348426818848,
+      "learning_rate": 9.999635457489331e-06,
+      "loss": 0.6571,
+      "step": 2072
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 5.326996803283691,
+      "learning_rate": 9.999632266799577e-06,
+      "loss": 0.7112,
+      "step": 2073
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.490452527999878,
+      "learning_rate": 9.99962906220772e-06,
+      "loss": 0.857,
+      "step": 2074
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.9019696712493896,
+      "learning_rate": 9.999625843713767e-06,
+      "loss": 0.8646,
+      "step": 2075
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.560948610305786,
+      "learning_rate": 9.999622611317727e-06,
+      "loss": 0.7702,
+      "step": 2076
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.3899881839752197,
+      "learning_rate": 9.999619365019611e-06,
+      "loss": 0.5306,
+      "step": 2077
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.7077136039733887,
+      "learning_rate": 9.999616104819425e-06,
+      "loss": 0.722,
+      "step": 2078
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 4.807618141174316,
+      "learning_rate": 9.99961283071718e-06,
+      "loss": 0.5407,
+      "step": 2079
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 4.047152042388916,
+      "learning_rate": 9.999609542712884e-06,
+      "loss": 1.0155,
+      "step": 2080
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.9619033336639404,
+      "learning_rate": 9.999606240806547e-06,
+      "loss": 0.6269,
+      "step": 2081
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.607969045639038,
+      "learning_rate": 9.99960292499818e-06,
+      "loss": 0.7976,
+      "step": 2082
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 11.363327026367188,
+      "learning_rate": 9.999599595287788e-06,
+      "loss": 0.9076,
+      "step": 2083
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.2333900928497314,
+      "learning_rate": 9.999596251675382e-06,
+      "loss": 0.4522,
+      "step": 2084
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.724616765975952,
+      "learning_rate": 9.999592894160972e-06,
+      "loss": 0.7112,
+      "step": 2085
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 5.364461421966553,
+      "learning_rate": 9.999589522744569e-06,
+      "loss": 0.7368,
+      "step": 2086
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 6.629299163818359,
+      "learning_rate": 9.999586137426178e-06,
+      "loss": 0.7219,
+      "step": 2087
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 6.615974426269531,
+      "learning_rate": 9.999582738205813e-06,
+      "loss": 0.5449,
+      "step": 2088
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.039095878601074,
+      "learning_rate": 9.999579325083479e-06,
+      "loss": 0.6876,
+      "step": 2089
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 4.197049140930176,
+      "learning_rate": 9.999575898059188e-06,
+      "loss": 0.6782,
+      "step": 2090
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.1557669639587402,
+      "learning_rate": 9.999572457132951e-06,
+      "loss": 0.7027,
+      "step": 2091
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.3117752075195312,
+      "learning_rate": 9.999569002304775e-06,
+      "loss": 0.7342,
+      "step": 2092
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.1465225219726562,
+      "learning_rate": 9.99956553357467e-06,
+      "loss": 0.6354,
+      "step": 2093
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.154412269592285,
+      "learning_rate": 9.999562050942648e-06,
+      "loss": 0.3835,
+      "step": 2094
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.7934696674346924,
+      "learning_rate": 9.999558554408716e-06,
+      "loss": 0.8116,
+      "step": 2095
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.1345551013946533,
+      "learning_rate": 9.999555043972882e-06,
+      "loss": 0.4561,
+      "step": 2096
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 4.554568767547607,
+      "learning_rate": 9.999551519635161e-06,
+      "loss": 0.6723,
+      "step": 2097
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.411243438720703,
+      "learning_rate": 9.99954798139556e-06,
+      "loss": 0.8263,
+      "step": 2098
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 5.634403228759766,
+      "learning_rate": 9.999544429254089e-06,
+      "loss": 0.8312,
+      "step": 2099
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.965597152709961,
+      "learning_rate": 9.999540863210757e-06,
+      "loss": 0.7781,
+      "step": 2100
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 11.849372863769531,
+      "learning_rate": 9.999537283265574e-06,
+      "loss": 0.6896,
+      "step": 2101
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.6157267093658447,
+      "learning_rate": 9.999533689418552e-06,
+      "loss": 0.6345,
+      "step": 2102
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 5.975534439086914,
+      "learning_rate": 9.999530081669699e-06,
+      "loss": 0.7622,
+      "step": 2103
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.213728427886963,
+      "learning_rate": 9.999526460019027e-06,
+      "loss": 0.731,
+      "step": 2104
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.835432291030884,
+      "learning_rate": 9.999522824466543e-06,
+      "loss": 0.8857,
+      "step": 2105
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.613295793533325,
+      "learning_rate": 9.99951917501226e-06,
+      "loss": 0.7414,
+      "step": 2106
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.4613170623779297,
+      "learning_rate": 9.999515511656185e-06,
+      "loss": 0.6572,
+      "step": 2107
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.409348726272583,
+      "learning_rate": 9.999511834398333e-06,
+      "loss": 0.8532,
+      "step": 2108
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.6485257148742676,
+      "learning_rate": 9.999508143238711e-06,
+      "loss": 0.4509,
+      "step": 2109
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 4.249114513397217,
+      "learning_rate": 9.999504438177329e-06,
+      "loss": 0.6197,
+      "step": 2110
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.1623482704162598,
+      "learning_rate": 9.999500719214198e-06,
+      "loss": 0.3425,
+      "step": 2111
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.3212602138519287,
+      "learning_rate": 9.999496986349329e-06,
+      "loss": 0.9059,
+      "step": 2112
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.6485536098480225,
+      "learning_rate": 9.999493239582731e-06,
+      "loss": 0.5166,
+      "step": 2113
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.4223177433013916,
+      "learning_rate": 9.999489478914415e-06,
+      "loss": 0.7881,
+      "step": 2114
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.1933507919311523,
+      "learning_rate": 9.999485704344391e-06,
+      "loss": 0.6644,
+      "step": 2115
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.403372287750244,
+      "learning_rate": 9.999481915872671e-06,
+      "loss": 0.3894,
+      "step": 2116
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.6598877906799316,
+      "learning_rate": 9.999478113499265e-06,
+      "loss": 0.6222,
+      "step": 2117
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.0712664127349854,
+      "learning_rate": 9.999474297224183e-06,
+      "loss": 0.2463,
+      "step": 2118
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 4.999281406402588,
+      "learning_rate": 9.999470467047436e-06,
+      "loss": 0.7237,
+      "step": 2119
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.056410789489746,
+      "learning_rate": 9.999466622969034e-06,
+      "loss": 0.7748,
+      "step": 2120
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 4.403641700744629,
+      "learning_rate": 9.99946276498899e-06,
+      "loss": 0.6373,
+      "step": 2121
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.5178165435791016,
+      "learning_rate": 9.99945889310731e-06,
+      "loss": 0.5466,
+      "step": 2122
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 4.697458744049072,
+      "learning_rate": 9.99945500732401e-06,
+      "loss": 0.6032,
+      "step": 2123
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.2319352626800537,
+      "learning_rate": 9.999451107639096e-06,
+      "loss": 0.7519,
+      "step": 2124
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.6656506061553955,
+      "learning_rate": 9.999447194052583e-06,
+      "loss": 0.5952,
+      "step": 2125
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.4986517429351807,
+      "learning_rate": 9.999443266564481e-06,
+      "loss": 0.7065,
+      "step": 2126
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.1854188442230225,
+      "learning_rate": 9.999439325174798e-06,
+      "loss": 0.6263,
+      "step": 2127
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.2141976356506348,
+      "learning_rate": 9.999435369883547e-06,
+      "loss": 0.5878,
+      "step": 2128
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.2675986289978027,
+      "learning_rate": 9.999431400690741e-06,
+      "loss": 0.596,
+      "step": 2129
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.7816368341445923,
+      "learning_rate": 9.999427417596388e-06,
+      "loss": 0.5988,
+      "step": 2130
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 8.482349395751953,
+      "learning_rate": 9.9994234206005e-06,
+      "loss": 0.8262,
+      "step": 2131
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 4.113472938537598,
+      "learning_rate": 9.99941940970309e-06,
+      "loss": 0.6369,
+      "step": 2132
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.4035372734069824,
+      "learning_rate": 9.999415384904166e-06,
+      "loss": 0.7168,
+      "step": 2133
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 4.203887939453125,
+      "learning_rate": 9.999411346203741e-06,
+      "loss": 0.7857,
+      "step": 2134
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 4.412224769592285,
+      "learning_rate": 9.999407293601827e-06,
+      "loss": 0.6275,
+      "step": 2135
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.720393180847168,
+      "learning_rate": 9.999403227098431e-06,
+      "loss": 0.9403,
+      "step": 2136
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.2427926063537598,
+      "learning_rate": 9.999399146693569e-06,
+      "loss": 0.6412,
+      "step": 2137
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 4.489736557006836,
+      "learning_rate": 9.99939505238725e-06,
+      "loss": 0.8259,
+      "step": 2138
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.451462984085083,
+      "learning_rate": 9.999390944179489e-06,
+      "loss": 0.5576,
+      "step": 2139
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.750375986099243,
+      "learning_rate": 9.999386822070293e-06,
+      "loss": 0.6335,
+      "step": 2140
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.7506189346313477,
+      "learning_rate": 9.999382686059674e-06,
+      "loss": 0.8632,
+      "step": 2141
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 4.124522686004639,
+      "learning_rate": 9.999378536147645e-06,
+      "loss": 0.7942,
+      "step": 2142
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.078187942504883,
+      "learning_rate": 9.999374372334218e-06,
+      "loss": 0.6431,
+      "step": 2143
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.300943374633789,
+      "learning_rate": 9.999370194619405e-06,
+      "loss": 0.5212,
+      "step": 2144
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.250591993331909,
+      "learning_rate": 9.999366003003213e-06,
+      "loss": 0.7085,
+      "step": 2145
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.0694336891174316,
+      "learning_rate": 9.999361797485658e-06,
+      "loss": 0.4815,
+      "step": 2146
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.166572332382202,
+      "learning_rate": 9.999357578066753e-06,
+      "loss": 0.723,
+      "step": 2147
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 4.81805944442749,
+      "learning_rate": 9.999353344746503e-06,
+      "loss": 0.8398,
+      "step": 2148
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.31432843208313,
+      "learning_rate": 9.999349097524929e-06,
+      "loss": 0.4999,
+      "step": 2149
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.709561824798584,
+      "learning_rate": 9.999344836402035e-06,
+      "loss": 0.3843,
+      "step": 2150
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.6533647775650024,
+      "learning_rate": 9.999340561377835e-06,
+      "loss": 0.379,
+      "step": 2151
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.3392324447631836,
+      "learning_rate": 9.999336272452345e-06,
+      "loss": 0.8234,
+      "step": 2152
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.0622384548187256,
+      "learning_rate": 9.999331969625571e-06,
+      "loss": 0.768,
+      "step": 2153
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.058258056640625,
+      "learning_rate": 9.999327652897529e-06,
+      "loss": 0.6505,
+      "step": 2154
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.932298183441162,
+      "learning_rate": 9.999323322268229e-06,
+      "loss": 0.7722,
+      "step": 2155
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.837571620941162,
+      "learning_rate": 9.999318977737682e-06,
+      "loss": 0.7922,
+      "step": 2156
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.789717197418213,
+      "learning_rate": 9.999314619305904e-06,
+      "loss": 0.9278,
+      "step": 2157
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.7754180431365967,
+      "learning_rate": 9.999310246972905e-06,
+      "loss": 0.8819,
+      "step": 2158
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 5.011226177215576,
+      "learning_rate": 9.999305860738696e-06,
+      "loss": 0.7568,
+      "step": 2159
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 5.085333824157715,
+      "learning_rate": 9.99930146060329e-06,
+      "loss": 0.5318,
+      "step": 2160
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.949202060699463,
+      "learning_rate": 9.999297046566698e-06,
+      "loss": 0.6417,
+      "step": 2161
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.626429796218872,
+      "learning_rate": 9.999292618628937e-06,
+      "loss": 1.0071,
+      "step": 2162
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.8137905597686768,
+      "learning_rate": 9.999288176790015e-06,
+      "loss": 0.6581,
+      "step": 2163
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.0909223556518555,
+      "learning_rate": 9.999283721049943e-06,
+      "loss": 0.8574,
+      "step": 2164
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 5.201725482940674,
+      "learning_rate": 9.999279251408737e-06,
+      "loss": 0.5571,
+      "step": 2165
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 5.251007556915283,
+      "learning_rate": 9.999274767866409e-06,
+      "loss": 0.7003,
+      "step": 2166
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.8635973930358887,
+      "learning_rate": 9.99927027042297e-06,
+      "loss": 0.8721,
+      "step": 2167
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.432316780090332,
+      "learning_rate": 9.999265759078434e-06,
+      "loss": 0.6123,
+      "step": 2168
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.6180191040039062,
+      "learning_rate": 9.999261233832813e-06,
+      "loss": 0.5964,
+      "step": 2169
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.593207359313965,
+      "learning_rate": 9.999256694686117e-06,
+      "loss": 0.739,
+      "step": 2170
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 4.7271318435668945,
+      "learning_rate": 9.999252141638362e-06,
+      "loss": 0.9272,
+      "step": 2171
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 4.420246124267578,
+      "learning_rate": 9.999247574689562e-06,
+      "loss": 0.7434,
+      "step": 2172
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.095398426055908,
+      "learning_rate": 9.999242993839724e-06,
+      "loss": 0.5467,
+      "step": 2173
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.7015695571899414,
+      "learning_rate": 9.999238399088865e-06,
+      "loss": 0.6447,
+      "step": 2174
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 5.316882610321045,
+      "learning_rate": 9.999233790436997e-06,
+      "loss": 0.8601,
+      "step": 2175
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.516925573348999,
+      "learning_rate": 9.999229167884133e-06,
+      "loss": 0.7716,
+      "step": 2176
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.720428705215454,
+      "learning_rate": 9.999224531430285e-06,
+      "loss": 0.5194,
+      "step": 2177
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 4.2161126136779785,
+      "learning_rate": 9.999219881075467e-06,
+      "loss": 0.7089,
+      "step": 2178
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.494047164916992,
+      "learning_rate": 9.999215216819689e-06,
+      "loss": 0.855,
+      "step": 2179
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.848047137260437,
+      "learning_rate": 9.999210538662967e-06,
+      "loss": 0.7621,
+      "step": 2180
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.524925708770752,
+      "learning_rate": 9.999205846605314e-06,
+      "loss": 0.2972,
+      "step": 2181
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.6056230068206787,
+      "learning_rate": 9.999201140646743e-06,
+      "loss": 0.615,
+      "step": 2182
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.2057085037231445,
+      "learning_rate": 9.999196420787266e-06,
+      "loss": 0.7504,
+      "step": 2183
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 4.7222900390625,
+      "learning_rate": 9.999191687026895e-06,
+      "loss": 0.7105,
+      "step": 2184
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 4.189956188201904,
+      "learning_rate": 9.999186939365646e-06,
+      "loss": 0.5767,
+      "step": 2185
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.1509594917297363,
+      "learning_rate": 9.99918217780353e-06,
+      "loss": 0.6313,
+      "step": 2186
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.403707265853882,
+      "learning_rate": 9.999177402340563e-06,
+      "loss": 0.6817,
+      "step": 2187
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.0305988788604736,
+      "learning_rate": 9.999172612976755e-06,
+      "loss": 0.7151,
+      "step": 2188
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.219076156616211,
+      "learning_rate": 9.99916780971212e-06,
+      "loss": 0.6626,
+      "step": 2189
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 6.965869903564453,
+      "learning_rate": 9.999162992546674e-06,
+      "loss": 0.8651,
+      "step": 2190
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.747011423110962,
+      "learning_rate": 9.999158161480427e-06,
+      "loss": 0.6097,
+      "step": 2191
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.9084596633911133,
+      "learning_rate": 9.999153316513394e-06,
+      "loss": 0.8693,
+      "step": 2192
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.7473244667053223,
+      "learning_rate": 9.99914845764559e-06,
+      "loss": 0.8518,
+      "step": 2193
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.2930843830108643,
+      "learning_rate": 9.999143584877024e-06,
+      "loss": 0.9322,
+      "step": 2194
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.0747244358062744,
+      "learning_rate": 9.999138698207713e-06,
+      "loss": 0.7191,
+      "step": 2195
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.9059133529663086,
+      "learning_rate": 9.999133797637671e-06,
+      "loss": 0.8603,
+      "step": 2196
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.2359986305236816,
+      "learning_rate": 9.99912888316691e-06,
+      "loss": 0.8353,
+      "step": 2197
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.48022198677063,
+      "learning_rate": 9.999123954795445e-06,
+      "loss": 0.8504,
+      "step": 2198
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.1939961910247803,
+      "learning_rate": 9.999119012523288e-06,
+      "loss": 0.4703,
+      "step": 2199
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.5465586185455322,
+      "learning_rate": 9.999114056350456e-06,
+      "loss": 0.7254,
+      "step": 2200
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.103858470916748,
+      "learning_rate": 9.999109086276958e-06,
+      "loss": 0.6049,
+      "step": 2201
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.9196436405181885,
+      "learning_rate": 9.999104102302813e-06,
+      "loss": 0.3389,
+      "step": 2202
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 4.4124979972839355,
+      "learning_rate": 9.99909910442803e-06,
+      "loss": 0.5155,
+      "step": 2203
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.6265804767608643,
+      "learning_rate": 9.999094092652625e-06,
+      "loss": 0.8932,
+      "step": 2204
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.1815831661224365,
+      "learning_rate": 9.999089066976613e-06,
+      "loss": 0.5856,
+      "step": 2205
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.0886261463165283,
+      "learning_rate": 9.999084027400005e-06,
+      "loss": 0.7506,
+      "step": 2206
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.1136956214904785,
+      "learning_rate": 9.99907897392282e-06,
+      "loss": 0.8434,
+      "step": 2207
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.788163900375366,
+      "learning_rate": 9.999073906545067e-06,
+      "loss": 0.3878,
+      "step": 2208
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.0515074729919434,
+      "learning_rate": 9.999068825266765e-06,
+      "loss": 0.5612,
+      "step": 2209
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.4659323692321777,
+      "learning_rate": 9.99906373008792e-06,
+      "loss": 0.8594,
+      "step": 2210
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.326300621032715,
+      "learning_rate": 9.999058621008557e-06,
+      "loss": 0.7978,
+      "step": 2211
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.999626874923706,
+      "learning_rate": 9.999053498028682e-06,
+      "loss": 0.5904,
+      "step": 2212
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 4.344348430633545,
+      "learning_rate": 9.999048361148311e-06,
+      "loss": 0.6936,
+      "step": 2213
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.498560667037964,
+      "learning_rate": 9.999043210367463e-06,
+      "loss": 0.8263,
+      "step": 2214
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.4556305408477783,
+      "learning_rate": 9.999038045686144e-06,
+      "loss": 0.7659,
+      "step": 2215
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.145453453063965,
+      "learning_rate": 9.999032867104375e-06,
+      "loss": 0.5337,
+      "step": 2216
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.9035520553588867,
+      "learning_rate": 9.999027674622168e-06,
+      "loss": 0.9004,
+      "step": 2217
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.0888140201568604,
+      "learning_rate": 9.99902246823954e-06,
+      "loss": 0.663,
+      "step": 2218
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 6.90838623046875,
+      "learning_rate": 9.9990172479565e-06,
+      "loss": 0.7951,
+      "step": 2219
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 5.869106292724609,
+      "learning_rate": 9.999012013773066e-06,
+      "loss": 0.7731,
+      "step": 2220
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.2807095050811768,
+      "learning_rate": 9.999006765689253e-06,
+      "loss": 0.7343,
+      "step": 2221
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 5.185286045074463,
+      "learning_rate": 9.999001503705074e-06,
+      "loss": 0.4103,
+      "step": 2222
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 6.4357123374938965,
+      "learning_rate": 9.998996227820546e-06,
+      "loss": 0.5993,
+      "step": 2223
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.431074857711792,
+      "learning_rate": 9.998990938035682e-06,
+      "loss": 0.7238,
+      "step": 2224
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.649057388305664,
+      "learning_rate": 9.998985634350495e-06,
+      "loss": 0.8818,
+      "step": 2225
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.3886704444885254,
+      "learning_rate": 9.998980316765003e-06,
+      "loss": 0.6909,
+      "step": 2226
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 7.116087436676025,
+      "learning_rate": 9.99897498527922e-06,
+      "loss": 0.6762,
+      "step": 2227
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.744434356689453,
+      "learning_rate": 9.99896963989316e-06,
+      "loss": 0.8131,
+      "step": 2228
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 19.455434799194336,
+      "learning_rate": 9.998964280606838e-06,
+      "loss": 0.5704,
+      "step": 2229
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.8697071075439453,
+      "learning_rate": 9.99895890742027e-06,
+      "loss": 0.8365,
+      "step": 2230
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.3585424423217773,
+      "learning_rate": 9.998953520333466e-06,
+      "loss": 0.369,
+      "step": 2231
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.176269769668579,
+      "learning_rate": 9.998948119346447e-06,
+      "loss": 0.5447,
+      "step": 2232
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.231536865234375,
+      "learning_rate": 9.998942704459227e-06,
+      "loss": 0.7664,
+      "step": 2233
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 5.446402072906494,
+      "learning_rate": 9.99893727567182e-06,
+      "loss": 0.7476,
+      "step": 2234
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.6506731510162354,
+      "learning_rate": 9.99893183298424e-06,
+      "loss": 0.6733,
+      "step": 2235
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.3435111045837402,
+      "learning_rate": 9.998926376396505e-06,
+      "loss": 0.6126,
+      "step": 2236
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.3620171546936035,
+      "learning_rate": 9.998920905908627e-06,
+      "loss": 0.6195,
+      "step": 2237
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.0226831436157227,
+      "learning_rate": 9.998915421520621e-06,
+      "loss": 0.7215,
+      "step": 2238
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.705857276916504,
+      "learning_rate": 9.998909923232507e-06,
+      "loss": 0.6429,
+      "step": 2239
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.168921709060669,
+      "learning_rate": 9.998904411044294e-06,
+      "loss": 0.4782,
+      "step": 2240
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 5.353969573974609,
+      "learning_rate": 9.998898884956003e-06,
+      "loss": 0.5277,
+      "step": 2241
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 4.417496204376221,
+      "learning_rate": 9.998893344967647e-06,
+      "loss": 0.6513,
+      "step": 2242
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.8335055112838745,
+      "learning_rate": 9.99888779107924e-06,
+      "loss": 0.5805,
+      "step": 2243
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 10.985512733459473,
+      "learning_rate": 9.998882223290798e-06,
+      "loss": 0.7118,
+      "step": 2244
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 4.367556095123291,
+      "learning_rate": 9.998876641602339e-06,
+      "loss": 0.503,
+      "step": 2245
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.5587050914764404,
+      "learning_rate": 9.998871046013876e-06,
+      "loss": 0.5563,
+      "step": 2246
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 4.0878143310546875,
+      "learning_rate": 9.998865436525425e-06,
+      "loss": 0.7305,
+      "step": 2247
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.8920819759368896,
+      "learning_rate": 9.998859813137e-06,
+      "loss": 0.7335,
+      "step": 2248
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 4.993102073669434,
+      "learning_rate": 9.998854175848622e-06,
+      "loss": 0.9119,
+      "step": 2249
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.147606134414673,
+      "learning_rate": 9.998848524660301e-06,
+      "loss": 0.7254,
+      "step": 2250
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 5.231389999389648,
+      "learning_rate": 9.998842859572055e-06,
+      "loss": 0.5757,
+      "step": 2251
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.4254770278930664,
+      "learning_rate": 9.9988371805839e-06,
+      "loss": 0.3007,
+      "step": 2252
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 8.143065452575684,
+      "learning_rate": 9.998831487695851e-06,
+      "loss": 0.6691,
+      "step": 2253
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 4.009747505187988,
+      "learning_rate": 9.998825780907924e-06,
+      "loss": 0.6603,
+      "step": 2254
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.7255263328552246,
+      "learning_rate": 9.998820060220137e-06,
+      "loss": 0.6281,
+      "step": 2255
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.9168651103973389,
+      "learning_rate": 9.998814325632502e-06,
+      "loss": 0.6744,
+      "step": 2256
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.5584867000579834,
+      "learning_rate": 9.998808577145038e-06,
+      "loss": 0.7962,
+      "step": 2257
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.0704188346862793,
+      "learning_rate": 9.99880281475776e-06,
+      "loss": 0.6357,
+      "step": 2258
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.1976561546325684,
+      "learning_rate": 9.998797038470683e-06,
+      "loss": 0.7354,
+      "step": 2259
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.6853723526000977,
+      "learning_rate": 9.998791248283824e-06,
+      "loss": 0.6482,
+      "step": 2260
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.6285929679870605,
+      "learning_rate": 9.9987854441972e-06,
+      "loss": 0.6657,
+      "step": 2261
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.1904711723327637,
+      "learning_rate": 9.998779626210827e-06,
+      "loss": 0.5429,
+      "step": 2262
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.5708837509155273,
+      "learning_rate": 9.998773794324719e-06,
+      "loss": 0.4797,
+      "step": 2263
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.1574268341064453,
+      "learning_rate": 9.998767948538895e-06,
+      "loss": 0.5258,
+      "step": 2264
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 4.378274440765381,
+      "learning_rate": 9.99876208885337e-06,
+      "loss": 0.6256,
+      "step": 2265
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.033135175704956,
+      "learning_rate": 9.998756215268158e-06,
+      "loss": 0.6495,
+      "step": 2266
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.8987693786621094,
+      "learning_rate": 9.998750327783278e-06,
+      "loss": 0.6506,
+      "step": 2267
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.6709632873535156,
+      "learning_rate": 9.998744426398747e-06,
+      "loss": 0.7968,
+      "step": 2268
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.67696213722229,
+      "learning_rate": 9.99873851111458e-06,
+      "loss": 0.6636,
+      "step": 2269
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 4.509652614593506,
+      "learning_rate": 9.998732581930793e-06,
+      "loss": 0.9443,
+      "step": 2270
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.135424852371216,
+      "learning_rate": 9.998726638847405e-06,
+      "loss": 0.4683,
+      "step": 2271
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.936391830444336,
+      "learning_rate": 9.998720681864429e-06,
+      "loss": 0.5692,
+      "step": 2272
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 9.963854789733887,
+      "learning_rate": 9.998714710981885e-06,
+      "loss": 0.5214,
+      "step": 2273
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 4.840737342834473,
+      "learning_rate": 9.998708726199786e-06,
+      "loss": 0.7624,
+      "step": 2274
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.4725818634033203,
+      "learning_rate": 9.998702727518152e-06,
+      "loss": 0.6141,
+      "step": 2275
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.7145965099334717,
+      "learning_rate": 9.998696714936999e-06,
+      "loss": 0.8303,
+      "step": 2276
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.604273557662964,
+      "learning_rate": 9.99869068845634e-06,
+      "loss": 0.5691,
+      "step": 2277
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.2835097312927246,
+      "learning_rate": 9.998684648076198e-06,
+      "loss": 0.8209,
+      "step": 2278
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 4.250868797302246,
+      "learning_rate": 9.998678593796584e-06,
+      "loss": 0.7017,
+      "step": 2279
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.9416240453720093,
+      "learning_rate": 9.99867252561752e-06,
+      "loss": 0.6202,
+      "step": 2280
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.8165327310562134,
+      "learning_rate": 9.998666443539018e-06,
+      "loss": 0.8274,
+      "step": 2281
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.896899938583374,
+      "learning_rate": 9.998660347561098e-06,
+      "loss": 0.6647,
+      "step": 2282
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.378368377685547,
+      "learning_rate": 9.998654237683776e-06,
+      "loss": 0.6454,
+      "step": 2283
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 5.322161674499512,
+      "learning_rate": 9.99864811390707e-06,
+      "loss": 0.457,
+      "step": 2284
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 4.9773268699646,
+      "learning_rate": 9.998641976230996e-06,
+      "loss": 0.838,
+      "step": 2285
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.755091428756714,
+      "learning_rate": 9.99863582465557e-06,
+      "loss": 0.3306,
+      "step": 2286
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.1279726028442383,
+      "learning_rate": 9.998629659180811e-06,
+      "loss": 0.5542,
+      "step": 2287
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 5.521897315979004,
+      "learning_rate": 9.998623479806735e-06,
+      "loss": 0.8281,
+      "step": 2288
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.7776331901550293,
+      "learning_rate": 9.998617286533361e-06,
+      "loss": 0.6065,
+      "step": 2289
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.521122932434082,
+      "learning_rate": 9.998611079360703e-06,
+      "loss": 0.4756,
+      "step": 2290
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 5.071190357208252,
+      "learning_rate": 9.99860485828878e-06,
+      "loss": 0.5573,
+      "step": 2291
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 4.895202159881592,
+      "learning_rate": 9.998598623317611e-06,
+      "loss": 0.9585,
+      "step": 2292
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.946471691131592,
+      "learning_rate": 9.99859237444721e-06,
+      "loss": 0.724,
+      "step": 2293
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.964797258377075,
+      "learning_rate": 9.998586111677599e-06,
+      "loss": 0.8384,
+      "step": 2294
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.3324267864227295,
+      "learning_rate": 9.99857983500879e-06,
+      "loss": 0.5804,
+      "step": 2295
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.9297789335250854,
+      "learning_rate": 9.998573544440804e-06,
+      "loss": 0.769,
+      "step": 2296
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.7176527976989746,
+      "learning_rate": 9.998567239973657e-06,
+      "loss": 0.8561,
+      "step": 2297
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 4.601107120513916,
+      "learning_rate": 9.998560921607368e-06,
+      "loss": 0.7184,
+      "step": 2298
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.0124411582946777,
+      "learning_rate": 9.998554589341952e-06,
+      "loss": 0.6154,
+      "step": 2299
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.565195083618164,
+      "learning_rate": 9.998548243177429e-06,
+      "loss": 0.6283,
+      "step": 2300
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.1958167552948,
+      "learning_rate": 9.998541883113816e-06,
+      "loss": 0.5266,
+      "step": 2301
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.890773057937622,
+      "learning_rate": 9.99853550915113e-06,
+      "loss": 0.8948,
+      "step": 2302
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.0992376804351807,
+      "learning_rate": 9.99852912128939e-06,
+      "loss": 0.7264,
+      "step": 2303
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.6977009773254395,
+      "learning_rate": 9.998522719528613e-06,
+      "loss": 0.587,
+      "step": 2304
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.4394919872283936,
+      "learning_rate": 9.998516303868817e-06,
+      "loss": 0.7999,
+      "step": 2305
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.8315428495407104,
+      "learning_rate": 9.99850987431002e-06,
+      "loss": 0.8275,
+      "step": 2306
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.0458881855010986,
+      "learning_rate": 9.998503430852237e-06,
+      "loss": 0.4718,
+      "step": 2307
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 5.48071813583374,
+      "learning_rate": 9.998496973495491e-06,
+      "loss": 0.607,
+      "step": 2308
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.8755178451538086,
+      "learning_rate": 9.998490502239796e-06,
+      "loss": 0.6115,
+      "step": 2309
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.0265519618988037,
+      "learning_rate": 9.99848401708517e-06,
+      "loss": 0.5447,
+      "step": 2310
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.7945003509521484,
+      "learning_rate": 9.998477518031635e-06,
+      "loss": 0.4798,
+      "step": 2311
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 4.552859783172607,
+      "learning_rate": 9.998471005079206e-06,
+      "loss": 0.7684,
+      "step": 2312
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.6524665355682373,
+      "learning_rate": 9.998464478227901e-06,
+      "loss": 0.7323,
+      "step": 2313
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.1074657440185547,
+      "learning_rate": 9.99845793747774e-06,
+      "loss": 0.6683,
+      "step": 2314
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 5.745067119598389,
+      "learning_rate": 9.99845138282874e-06,
+      "loss": 0.5183,
+      "step": 2315
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.8627209663391113,
+      "learning_rate": 9.998444814280917e-06,
+      "loss": 0.5169,
+      "step": 2316
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.7670977115631104,
+      "learning_rate": 9.998438231834293e-06,
+      "loss": 0.5305,
+      "step": 2317
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 5.2547926902771,
+      "learning_rate": 9.998431635488883e-06,
+      "loss": 0.7874,
+      "step": 2318
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.783992052078247,
+      "learning_rate": 9.99842502524471e-06,
+      "loss": 0.7796,
+      "step": 2319
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.4300177097320557,
+      "learning_rate": 9.998418401101786e-06,
+      "loss": 0.6694,
+      "step": 2320
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.760251522064209,
+      "learning_rate": 9.998411763060136e-06,
+      "loss": 0.6697,
+      "step": 2321
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.2280068397521973,
+      "learning_rate": 9.998405111119773e-06,
+      "loss": 0.642,
+      "step": 2322
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.179610252380371,
+      "learning_rate": 9.99839844528072e-06,
+      "loss": 0.6786,
+      "step": 2323
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.8904941082000732,
+      "learning_rate": 9.998391765542992e-06,
+      "loss": 0.7976,
+      "step": 2324
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.9454841613769531,
+      "learning_rate": 9.998385071906609e-06,
+      "loss": 0.8941,
+      "step": 2325
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.739271640777588,
+      "learning_rate": 9.99837836437159e-06,
+      "loss": 0.5819,
+      "step": 2326
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 5.125761032104492,
+      "learning_rate": 9.998371642937954e-06,
+      "loss": 0.7099,
+      "step": 2327
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.8274855613708496,
+      "learning_rate": 9.998364907605717e-06,
+      "loss": 0.6527,
+      "step": 2328
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.7370078563690186,
+      "learning_rate": 9.998358158374901e-06,
+      "loss": 0.764,
+      "step": 2329
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.801102638244629,
+      "learning_rate": 9.998351395245522e-06,
+      "loss": 0.9206,
+      "step": 2330
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.005437135696411,
+      "learning_rate": 9.9983446182176e-06,
+      "loss": 0.5407,
+      "step": 2331
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 4.002213478088379,
+      "learning_rate": 9.998337827291157e-06,
+      "loss": 0.6872,
+      "step": 2332
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.3886682987213135,
+      "learning_rate": 9.998331022466208e-06,
+      "loss": 0.6614,
+      "step": 2333
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.4190421104431152,
+      "learning_rate": 9.99832420374277e-06,
+      "loss": 0.5715,
+      "step": 2334
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.1074931621551514,
+      "learning_rate": 9.998317371120867e-06,
+      "loss": 0.7777,
+      "step": 2335
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.970705270767212,
+      "learning_rate": 9.998310524600516e-06,
+      "loss": 0.7151,
+      "step": 2336
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.000718832015991,
+      "learning_rate": 9.998303664181736e-06,
+      "loss": 0.7458,
+      "step": 2337
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.293745279312134,
+      "learning_rate": 9.998296789864545e-06,
+      "loss": 0.6592,
+      "step": 2338
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.192145824432373,
+      "learning_rate": 9.998289901648963e-06,
+      "loss": 0.5103,
+      "step": 2339
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.2038590908050537,
+      "learning_rate": 9.99828299953501e-06,
+      "loss": 0.5788,
+      "step": 2340
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.256362199783325,
+      "learning_rate": 9.998276083522703e-06,
+      "loss": 0.9407,
+      "step": 2341
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.0111734867095947,
+      "learning_rate": 9.998269153612063e-06,
+      "loss": 0.8083,
+      "step": 2342
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.3749139308929443,
+      "learning_rate": 9.99826220980311e-06,
+      "loss": 0.602,
+      "step": 2343
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.8804460763931274,
+      "learning_rate": 9.998255252095862e-06,
+      "loss": 0.3382,
+      "step": 2344
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.2915751934051514,
+      "learning_rate": 9.998248280490337e-06,
+      "loss": 0.8837,
+      "step": 2345
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.7703049182891846,
+      "learning_rate": 9.998241294986558e-06,
+      "loss": 0.6091,
+      "step": 2346
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.6140244007110596,
+      "learning_rate": 9.99823429558454e-06,
+      "loss": 0.7066,
+      "step": 2347
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.8537994623184204,
+      "learning_rate": 9.998227282284308e-06,
+      "loss": 0.6476,
+      "step": 2348
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.814802646636963,
+      "learning_rate": 9.998220255085876e-06,
+      "loss": 0.7344,
+      "step": 2349
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.6345293521881104,
+      "learning_rate": 9.998213213989267e-06,
+      "loss": 0.8114,
+      "step": 2350
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.0291359424591064,
+      "learning_rate": 9.998206158994498e-06,
+      "loss": 0.5484,
+      "step": 2351
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.6802453994750977,
+      "learning_rate": 9.99819909010159e-06,
+      "loss": 0.8121,
+      "step": 2352
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.2015795707702637,
+      "learning_rate": 9.998192007310564e-06,
+      "loss": 0.6868,
+      "step": 2353
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 4.051909923553467,
+      "learning_rate": 9.998184910621438e-06,
+      "loss": 0.7841,
+      "step": 2354
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.2535042762756348,
+      "learning_rate": 9.998177800034232e-06,
+      "loss": 0.5625,
+      "step": 2355
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.932908296585083,
+      "learning_rate": 9.998170675548965e-06,
+      "loss": 0.531,
+      "step": 2356
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.4695560932159424,
+      "learning_rate": 9.99816353716566e-06,
+      "loss": 0.7147,
+      "step": 2357
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.9111645221710205,
+      "learning_rate": 9.998156384884332e-06,
+      "loss": 0.5823,
+      "step": 2358
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 5.243614673614502,
+      "learning_rate": 9.998149218705004e-06,
+      "loss": 0.764,
+      "step": 2359
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.818870782852173,
+      "learning_rate": 9.998142038627697e-06,
+      "loss": 0.4908,
+      "step": 2360
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.2932114601135254,
+      "learning_rate": 9.998134844652429e-06,
+      "loss": 0.6797,
+      "step": 2361
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.1475207805633545,
+      "learning_rate": 9.99812763677922e-06,
+      "loss": 0.6407,
+      "step": 2362
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.2938132286071777,
+      "learning_rate": 9.998120415008088e-06,
+      "loss": 0.4918,
+      "step": 2363
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 4.134765625,
+      "learning_rate": 9.998113179339058e-06,
+      "loss": 0.7811,
+      "step": 2364
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.4514904022216797,
+      "learning_rate": 9.998105929772149e-06,
+      "loss": 0.8173,
+      "step": 2365
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.9383299350738525,
+      "learning_rate": 9.998098666307376e-06,
+      "loss": 0.6853,
+      "step": 2366
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.921654462814331,
+      "learning_rate": 9.998091388944765e-06,
+      "loss": 0.7461,
+      "step": 2367
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.204385280609131,
+      "learning_rate": 9.998084097684333e-06,
+      "loss": 0.6445,
+      "step": 2368
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 4.028822422027588,
+      "learning_rate": 9.998076792526104e-06,
+      "loss": 0.7018,
+      "step": 2369
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.703731060028076,
+      "learning_rate": 9.998069473470093e-06,
+      "loss": 0.508,
+      "step": 2370
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.7228903770446777,
+      "learning_rate": 9.998062140516325e-06,
+      "loss": 0.6349,
+      "step": 2371
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.6034953594207764,
+      "learning_rate": 9.998054793664817e-06,
+      "loss": 0.3688,
+      "step": 2372
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.371004343032837,
+      "learning_rate": 9.998047432915592e-06,
+      "loss": 0.8155,
+      "step": 2373
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.9055308103561401,
+      "learning_rate": 9.99804005826867e-06,
+      "loss": 0.57,
+      "step": 2374
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.4685003757476807,
+      "learning_rate": 9.99803266972407e-06,
+      "loss": 0.5999,
+      "step": 2375
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.586294174194336,
+      "learning_rate": 9.998025267281812e-06,
+      "loss": 0.7885,
+      "step": 2376
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.203449249267578,
+      "learning_rate": 9.998017850941921e-06,
+      "loss": 0.7416,
+      "step": 2377
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.158529758453369,
+      "learning_rate": 9.998010420704413e-06,
+      "loss": 0.7105,
+      "step": 2378
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 8.301596641540527,
+      "learning_rate": 9.99800297656931e-06,
+      "loss": 0.8446,
+      "step": 2379
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.993047833442688,
+      "learning_rate": 9.997995518536634e-06,
+      "loss": 0.315,
+      "step": 2380
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 4.638898849487305,
+      "learning_rate": 9.997988046606404e-06,
+      "loss": 0.5646,
+      "step": 2381
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.1348886489868164,
+      "learning_rate": 9.997980560778642e-06,
+      "loss": 0.6527,
+      "step": 2382
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.288569450378418,
+      "learning_rate": 9.997973061053368e-06,
+      "loss": 0.7911,
+      "step": 2383
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.475388288497925,
+      "learning_rate": 9.997965547430603e-06,
+      "loss": 0.7487,
+      "step": 2384
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.8207732439041138,
+      "learning_rate": 9.997958019910368e-06,
+      "loss": 0.5571,
+      "step": 2385
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.0296802520751953,
+      "learning_rate": 9.997950478492684e-06,
+      "loss": 0.7831,
+      "step": 2386
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.487433433532715,
+      "learning_rate": 9.997942923177573e-06,
+      "loss": 0.5818,
+      "step": 2387
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.615429639816284,
+      "learning_rate": 9.997935353965055e-06,
+      "loss": 0.7015,
+      "step": 2388
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.949256658554077,
+      "learning_rate": 9.997927770855151e-06,
+      "loss": 0.8156,
+      "step": 2389
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.1467814445495605,
+      "learning_rate": 9.99792017384788e-06,
+      "loss": 0.6142,
+      "step": 2390
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.798363208770752,
+      "learning_rate": 9.997912562943268e-06,
+      "loss": 0.4983,
+      "step": 2391
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.24670147895813,
+      "learning_rate": 9.997904938141331e-06,
+      "loss": 0.8914,
+      "step": 2392
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.365083694458008,
+      "learning_rate": 9.997897299442095e-06,
+      "loss": 0.7149,
+      "step": 2393
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.470991373062134,
+      "learning_rate": 9.997889646845578e-06,
+      "loss": 0.5042,
+      "step": 2394
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.351489782333374,
+      "learning_rate": 9.997881980351802e-06,
+      "loss": 0.576,
+      "step": 2395
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.216141700744629,
+      "learning_rate": 9.997874299960787e-06,
+      "loss": 0.8049,
+      "step": 2396
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.9753530025482178,
+      "learning_rate": 9.997866605672559e-06,
+      "loss": 0.7626,
+      "step": 2397
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.5735771656036377,
+      "learning_rate": 9.997858897487134e-06,
+      "loss": 0.6297,
+      "step": 2398
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.510312557220459,
+      "learning_rate": 9.997851175404535e-06,
+      "loss": 0.5519,
+      "step": 2399
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 8.068066596984863,
+      "learning_rate": 9.997843439424786e-06,
+      "loss": 0.7834,
+      "step": 2400
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 4.352299213409424,
+      "learning_rate": 9.997835689547906e-06,
+      "loss": 0.9112,
+      "step": 2401
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 6.537634372711182,
+      "learning_rate": 9.997827925773918e-06,
+      "loss": 0.5858,
+      "step": 2402
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.5389347076416016,
+      "learning_rate": 9.997820148102841e-06,
+      "loss": 0.5001,
+      "step": 2403
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.0798709392547607,
+      "learning_rate": 9.9978123565347e-06,
+      "loss": 0.5042,
+      "step": 2404
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.908233642578125,
+      "learning_rate": 9.997804551069514e-06,
+      "loss": 0.5434,
+      "step": 2405
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.7648677825927734,
+      "learning_rate": 9.997796731707306e-06,
+      "loss": 0.7498,
+      "step": 2406
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.0996110439300537,
+      "learning_rate": 9.997788898448096e-06,
+      "loss": 0.495,
+      "step": 2407
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.5373785495758057,
+      "learning_rate": 9.99778105129191e-06,
+      "loss": 0.6105,
+      "step": 2408
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.1499903202056885,
+      "learning_rate": 9.997773190238766e-06,
+      "loss": 0.4859,
+      "step": 2409
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.583080768585205,
+      "learning_rate": 9.997765315288686e-06,
+      "loss": 0.7782,
+      "step": 2410
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.98801326751709,
+      "learning_rate": 9.997757426441694e-06,
+      "loss": 0.5979,
+      "step": 2411
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.7612712383270264,
+      "learning_rate": 9.997749523697811e-06,
+      "loss": 0.6936,
+      "step": 2412
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.2996184825897217,
+      "learning_rate": 9.997741607057058e-06,
+      "loss": 0.8008,
+      "step": 2413
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.8855743408203125,
+      "learning_rate": 9.997733676519458e-06,
+      "loss": 0.4645,
+      "step": 2414
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.9630112648010254,
+      "learning_rate": 9.997725732085033e-06,
+      "loss": 0.5697,
+      "step": 2415
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 5.507529258728027,
+      "learning_rate": 9.997717773753805e-06,
+      "loss": 0.7043,
+      "step": 2416
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.9707212448120117,
+      "learning_rate": 9.997709801525796e-06,
+      "loss": 0.8097,
+      "step": 2417
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.543890953063965,
+      "learning_rate": 9.997701815401027e-06,
+      "loss": 0.7906,
+      "step": 2418
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.5376429557800293,
+      "learning_rate": 9.997693815379523e-06,
+      "loss": 0.7017,
+      "step": 2419
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.203206777572632,
+      "learning_rate": 9.997685801461304e-06,
+      "loss": 0.617,
+      "step": 2420
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.8605120182037354,
+      "learning_rate": 9.997677773646393e-06,
+      "loss": 0.6315,
+      "step": 2421
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.9936718940734863,
+      "learning_rate": 9.997669731934813e-06,
+      "loss": 0.5835,
+      "step": 2422
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.728971004486084,
+      "learning_rate": 9.997661676326585e-06,
+      "loss": 0.9191,
+      "step": 2423
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.3281147480010986,
+      "learning_rate": 9.997653606821733e-06,
+      "loss": 0.8532,
+      "step": 2424
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.338402509689331,
+      "learning_rate": 9.997645523420278e-06,
+      "loss": 0.4723,
+      "step": 2425
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.8179874420166016,
+      "learning_rate": 9.997637426122242e-06,
+      "loss": 1.0331,
+      "step": 2426
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.777113676071167,
+      "learning_rate": 9.997629314927648e-06,
+      "loss": 0.7087,
+      "step": 2427
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.605560064315796,
+      "learning_rate": 9.997621189836522e-06,
+      "loss": 0.7153,
+      "step": 2428
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.465752601623535,
+      "learning_rate": 9.997613050848882e-06,
+      "loss": 0.527,
+      "step": 2429
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.7224807739257812,
+      "learning_rate": 9.99760489796475e-06,
+      "loss": 0.4836,
+      "step": 2430
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.680896282196045,
+      "learning_rate": 9.997596731184154e-06,
+      "loss": 0.5245,
+      "step": 2431
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.070366382598877,
+      "learning_rate": 9.997588550507112e-06,
+      "loss": 0.5436,
+      "step": 2432
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.5502500534057617,
+      "learning_rate": 9.997580355933649e-06,
+      "loss": 0.4493,
+      "step": 2433
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.0087156295776367,
+      "learning_rate": 9.997572147463788e-06,
+      "loss": 0.731,
+      "step": 2434
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 5.137729167938232,
+      "learning_rate": 9.99756392509755e-06,
+      "loss": 0.7663,
+      "step": 2435
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.8421638011932373,
+      "learning_rate": 9.99755568883496e-06,
+      "loss": 0.7003,
+      "step": 2436
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.164013385772705,
+      "learning_rate": 9.997547438676038e-06,
+      "loss": 0.6283,
+      "step": 2437
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.6635756492614746,
+      "learning_rate": 9.99753917462081e-06,
+      "loss": 0.5207,
+      "step": 2438
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 9.711877822875977,
+      "learning_rate": 9.997530896669296e-06,
+      "loss": 0.9118,
+      "step": 2439
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.3730757236480713,
+      "learning_rate": 9.997522604821522e-06,
+      "loss": 0.9887,
+      "step": 2440
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.0530691146850586,
+      "learning_rate": 9.997514299077508e-06,
+      "loss": 0.8248,
+      "step": 2441
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.7111244201660156,
+      "learning_rate": 9.997505979437281e-06,
+      "loss": 0.723,
+      "step": 2442
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.0522539615631104,
+      "learning_rate": 9.997497645900863e-06,
+      "loss": 0.8075,
+      "step": 2443
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.415803909301758,
+      "learning_rate": 9.997489298468273e-06,
+      "loss": 0.6689,
+      "step": 2444
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.5986337661743164,
+      "learning_rate": 9.997480937139539e-06,
+      "loss": 0.7837,
+      "step": 2445
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.154524564743042,
+      "learning_rate": 9.997472561914683e-06,
+      "loss": 0.5979,
+      "step": 2446
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.6785242557525635,
+      "learning_rate": 9.997464172793727e-06,
+      "loss": 0.5543,
+      "step": 2447
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 4.1352691650390625,
+      "learning_rate": 9.997455769776696e-06,
+      "loss": 0.677,
+      "step": 2448
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.8730921745300293,
+      "learning_rate": 9.997447352863611e-06,
+      "loss": 0.8075,
+      "step": 2449
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.5251030921936035,
+      "learning_rate": 9.997438922054499e-06,
+      "loss": 0.6937,
+      "step": 2450
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.422480583190918,
+      "learning_rate": 9.997430477349382e-06,
+      "loss": 0.6275,
+      "step": 2451
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.673586845397949,
+      "learning_rate": 9.99742201874828e-06,
+      "loss": 0.9013,
+      "step": 2452
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.2055821418762207,
+      "learning_rate": 9.997413546251221e-06,
+      "loss": 0.5465,
+      "step": 2453
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.4112589359283447,
+      "learning_rate": 9.997405059858227e-06,
+      "loss": 0.6243,
+      "step": 2454
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.607334852218628,
+      "learning_rate": 9.997396559569322e-06,
+      "loss": 0.584,
+      "step": 2455
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.808384895324707,
+      "learning_rate": 9.99738804538453e-06,
+      "loss": 0.5583,
+      "step": 2456
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.076622724533081,
+      "learning_rate": 9.997379517303871e-06,
+      "loss": 0.6683,
+      "step": 2457
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.8510994911193848,
+      "learning_rate": 9.997370975327374e-06,
+      "loss": 0.6491,
+      "step": 2458
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.790198802947998,
+      "learning_rate": 9.99736241945506e-06,
+      "loss": 0.7138,
+      "step": 2459
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.563037395477295,
+      "learning_rate": 9.997353849686952e-06,
+      "loss": 0.8482,
+      "step": 2460
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 4.3980584144592285,
+      "learning_rate": 9.997345266023077e-06,
+      "loss": 0.7332,
+      "step": 2461
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.4939634799957275,
+      "learning_rate": 9.997336668463455e-06,
+      "loss": 0.4776,
+      "step": 2462
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 6.386855602264404,
+      "learning_rate": 9.997328057008112e-06,
+      "loss": 0.6559,
+      "step": 2463
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.958503246307373,
+      "learning_rate": 9.997319431657074e-06,
+      "loss": 0.748,
+      "step": 2464
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.567197799682617,
+      "learning_rate": 9.997310792410361e-06,
+      "loss": 0.626,
+      "step": 2465
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.096446990966797,
+      "learning_rate": 9.997302139267999e-06,
+      "loss": 0.5315,
+      "step": 2466
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.8467438220977783,
+      "learning_rate": 9.99729347223001e-06,
+      "loss": 0.5639,
+      "step": 2467
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.7513759136199951,
+      "learning_rate": 9.997284791296422e-06,
+      "loss": 0.4599,
+      "step": 2468
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.910083770751953,
+      "learning_rate": 9.997276096467257e-06,
+      "loss": 0.6721,
+      "step": 2469
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.7829220294952393,
+      "learning_rate": 9.997267387742538e-06,
+      "loss": 0.4848,
+      "step": 2470
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 14.15453815460205,
+      "learning_rate": 9.997258665122292e-06,
+      "loss": 0.6653,
+      "step": 2471
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.0211074352264404,
+      "learning_rate": 9.99724992860654e-06,
+      "loss": 0.8348,
+      "step": 2472
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.7559874057769775,
+      "learning_rate": 9.997241178195309e-06,
+      "loss": 0.785,
+      "step": 2473
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.9688799381256104,
+      "learning_rate": 9.997232413888622e-06,
+      "loss": 0.442,
+      "step": 2474
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.863654375076294,
+      "learning_rate": 9.997223635686503e-06,
+      "loss": 0.672,
+      "step": 2475
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.7482361793518066,
+      "learning_rate": 9.997214843588979e-06,
+      "loss": 0.4844,
+      "step": 2476
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 4.07444953918457,
+      "learning_rate": 9.99720603759607e-06,
+      "loss": 0.6646,
+      "step": 2477
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.0583553314208984,
+      "learning_rate": 9.997197217707803e-06,
+      "loss": 0.7497,
+      "step": 2478
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 5.020679950714111,
+      "learning_rate": 9.997188383924204e-06,
+      "loss": 0.7151,
+      "step": 2479
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.25519061088562,
+      "learning_rate": 9.997179536245296e-06,
+      "loss": 0.7138,
+      "step": 2480
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.0481953620910645,
+      "learning_rate": 9.997170674671103e-06,
+      "loss": 0.7512,
+      "step": 2481
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.5815116167068481,
+      "learning_rate": 9.99716179920165e-06,
+      "loss": 0.6411,
+      "step": 2482
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 4.026105880737305,
+      "learning_rate": 9.997152909836964e-06,
+      "loss": 0.8442,
+      "step": 2483
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.2594923973083496,
+      "learning_rate": 9.997144006577067e-06,
+      "loss": 0.5238,
+      "step": 2484
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.987982153892517,
+      "learning_rate": 9.997135089421985e-06,
+      "loss": 0.7442,
+      "step": 2485
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.28783917427063,
+      "learning_rate": 9.997126158371741e-06,
+      "loss": 0.6936,
+      "step": 2486
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.4160397052764893,
+      "learning_rate": 9.997117213426361e-06,
+      "loss": 0.3882,
+      "step": 2487
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 4.498936176300049,
+      "learning_rate": 9.99710825458587e-06,
+      "loss": 0.9005,
+      "step": 2488
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.1266753673553467,
+      "learning_rate": 9.997099281850295e-06,
+      "loss": 0.7941,
+      "step": 2489
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.828308582305908,
+      "learning_rate": 9.997090295219656e-06,
+      "loss": 0.7144,
+      "step": 2490
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.894697427749634,
+      "learning_rate": 9.997081294693981e-06,
+      "loss": 0.7711,
+      "step": 2491
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.8211236000061035,
+      "learning_rate": 9.997072280273297e-06,
+      "loss": 0.8291,
+      "step": 2492
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.27034330368042,
+      "learning_rate": 9.997063251957626e-06,
+      "loss": 0.7108,
+      "step": 2493
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 4.13437032699585,
+      "learning_rate": 9.997054209746995e-06,
+      "loss": 0.5516,
+      "step": 2494
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.222393274307251,
+      "learning_rate": 9.997045153641427e-06,
+      "loss": 0.6899,
+      "step": 2495
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.565964698791504,
+      "learning_rate": 9.997036083640947e-06,
+      "loss": 0.6145,
+      "step": 2496
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.3537421226501465,
+      "learning_rate": 9.997026999745583e-06,
+      "loss": 0.7292,
+      "step": 2497
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.507730007171631,
+      "learning_rate": 9.997017901955359e-06,
+      "loss": 0.8431,
+      "step": 2498
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.648172616958618,
+      "learning_rate": 9.997008790270301e-06,
+      "loss": 0.8698,
+      "step": 2499
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.2437806129455566,
+      "learning_rate": 9.996999664690432e-06,
+      "loss": 0.6724,
+      "step": 2500
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.230693817138672,
+      "learning_rate": 9.99699052521578e-06,
+      "loss": 0.7744,
+      "step": 2501
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.180644989013672,
+      "learning_rate": 9.996981371846368e-06,
+      "loss": 0.6431,
+      "step": 2502
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.9525821208953857,
+      "learning_rate": 9.996972204582224e-06,
+      "loss": 0.7331,
+      "step": 2503
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.4159724712371826,
+      "learning_rate": 9.996963023423371e-06,
+      "loss": 0.9527,
+      "step": 2504
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 9.651532173156738,
+      "learning_rate": 9.996953828369836e-06,
+      "loss": 0.8559,
+      "step": 2505
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.7058987617492676,
+      "learning_rate": 9.996944619421645e-06,
+      "loss": 0.6134,
+      "step": 2506
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.915374279022217,
+      "learning_rate": 9.996935396578822e-06,
+      "loss": 0.8172,
+      "step": 2507
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.2413952350616455,
+      "learning_rate": 9.996926159841394e-06,
+      "loss": 0.6481,
+      "step": 2508
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.802027940750122,
+      "learning_rate": 9.996916909209387e-06,
+      "loss": 0.712,
+      "step": 2509
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.430445909500122,
+      "learning_rate": 9.996907644682826e-06,
+      "loss": 0.8439,
+      "step": 2510
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.7709977626800537,
+      "learning_rate": 9.996898366261736e-06,
+      "loss": 0.6064,
+      "step": 2511
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.943485975265503,
+      "learning_rate": 9.996889073946143e-06,
+      "loss": 1.0502,
+      "step": 2512
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.3337221145629883,
+      "learning_rate": 9.996879767736074e-06,
+      "loss": 0.8974,
+      "step": 2513
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.124190330505371,
+      "learning_rate": 9.996870447631553e-06,
+      "loss": 0.8573,
+      "step": 2514
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.8293139934539795,
+      "learning_rate": 9.996861113632608e-06,
+      "loss": 0.7013,
+      "step": 2515
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.416358232498169,
+      "learning_rate": 9.996851765739267e-06,
+      "loss": 0.6085,
+      "step": 2516
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.9665849208831787,
+      "learning_rate": 9.99684240395155e-06,
+      "loss": 0.4238,
+      "step": 2517
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.6275947093963623,
+      "learning_rate": 9.996833028269487e-06,
+      "loss": 0.4967,
+      "step": 2518
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 4.669879913330078,
+      "learning_rate": 9.9968236386931e-06,
+      "loss": 0.7091,
+      "step": 2519
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.042257308959961,
+      "learning_rate": 9.996814235222422e-06,
+      "loss": 0.6284,
+      "step": 2520
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.7065348625183105,
+      "learning_rate": 9.996804817857475e-06,
+      "loss": 0.7815,
+      "step": 2521
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.3149473667144775,
+      "learning_rate": 9.996795386598285e-06,
+      "loss": 0.8671,
+      "step": 2522
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.9100165367126465,
+      "learning_rate": 9.99678594144488e-06,
+      "loss": 1.0354,
+      "step": 2523
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.543867826461792,
+      "learning_rate": 9.996776482397283e-06,
+      "loss": 0.5438,
+      "step": 2524
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.3052828311920166,
+      "learning_rate": 9.996767009455524e-06,
+      "loss": 0.8409,
+      "step": 2525
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.2156434059143066,
+      "learning_rate": 9.996757522619628e-06,
+      "loss": 0.6889,
+      "step": 2526
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.306609630584717,
+      "learning_rate": 9.99674802188962e-06,
+      "loss": 0.7698,
+      "step": 2527
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.8903422355651855,
+      "learning_rate": 9.996738507265529e-06,
+      "loss": 0.3329,
+      "step": 2528
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.756007671356201,
+      "learning_rate": 9.996728978747379e-06,
+      "loss": 0.6272,
+      "step": 2529
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.321042060852051,
+      "learning_rate": 9.996719436335198e-06,
+      "loss": 0.6733,
+      "step": 2530
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.362917423248291,
+      "learning_rate": 9.996709880029012e-06,
+      "loss": 0.6939,
+      "step": 2531
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.6748054027557373,
+      "learning_rate": 9.996700309828847e-06,
+      "loss": 0.6316,
+      "step": 2532
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.4862618446350098,
+      "learning_rate": 9.99669072573473e-06,
+      "loss": 0.6902,
+      "step": 2533
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 5.56859827041626,
+      "learning_rate": 9.996681127746689e-06,
+      "loss": 0.8789,
+      "step": 2534
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.7637970447540283,
+      "learning_rate": 9.996671515864748e-06,
+      "loss": 0.7771,
+      "step": 2535
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.4900522232055664,
+      "learning_rate": 9.996661890088938e-06,
+      "loss": 0.7921,
+      "step": 2536
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.6472420692443848,
+      "learning_rate": 9.996652250419281e-06,
+      "loss": 0.6705,
+      "step": 2537
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 5.0868940353393555,
+      "learning_rate": 9.996642596855808e-06,
+      "loss": 0.6124,
+      "step": 2538
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.803370237350464,
+      "learning_rate": 9.996632929398542e-06,
+      "loss": 0.8207,
+      "step": 2539
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.6216092109680176,
+      "learning_rate": 9.996623248047512e-06,
+      "loss": 0.7223,
+      "step": 2540
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.4025566577911377,
+      "learning_rate": 9.996613552802743e-06,
+      "loss": 0.517,
+      "step": 2541
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.14577317237854,
+      "learning_rate": 9.996603843664265e-06,
+      "loss": 0.4636,
+      "step": 2542
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.5876803398132324,
+      "learning_rate": 9.996594120632103e-06,
+      "loss": 0.7627,
+      "step": 2543
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.643618583679199,
+      "learning_rate": 9.996584383706286e-06,
+      "loss": 0.6262,
+      "step": 2544
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.3933656215667725,
+      "learning_rate": 9.996574632886837e-06,
+      "loss": 0.7144,
+      "step": 2545
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 5.109819412231445,
+      "learning_rate": 9.996564868173788e-06,
+      "loss": 0.5524,
+      "step": 2546
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 4.205166816711426,
+      "learning_rate": 9.996555089567163e-06,
+      "loss": 0.7872,
+      "step": 2547
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 5.526734352111816,
+      "learning_rate": 9.99654529706699e-06,
+      "loss": 0.5787,
+      "step": 2548
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.6370627880096436,
+      "learning_rate": 9.996535490673295e-06,
+      "loss": 0.6829,
+      "step": 2549
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.762272357940674,
+      "learning_rate": 9.996525670386108e-06,
+      "loss": 0.913,
+      "step": 2550
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.557110548019409,
+      "learning_rate": 9.996515836205457e-06,
+      "loss": 0.7214,
+      "step": 2551
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.4807653427124023,
+      "learning_rate": 9.996505988131364e-06,
+      "loss": 0.5803,
+      "step": 2552
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.469139814376831,
+      "learning_rate": 9.996496126163859e-06,
+      "loss": 0.6963,
+      "step": 2553
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.3668227195739746,
+      "learning_rate": 9.996486250302972e-06,
+      "loss": 0.7508,
+      "step": 2554
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.258150815963745,
+      "learning_rate": 9.996476360548728e-06,
+      "loss": 0.7814,
+      "step": 2555
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.276724100112915,
+      "learning_rate": 9.996466456901154e-06,
+      "loss": 0.6533,
+      "step": 2556
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 4.237672805786133,
+      "learning_rate": 9.99645653936028e-06,
+      "loss": 0.5184,
+      "step": 2557
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.215913772583008,
+      "learning_rate": 9.996446607926131e-06,
+      "loss": 0.4212,
+      "step": 2558
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.585186243057251,
+      "learning_rate": 9.996436662598736e-06,
+      "loss": 0.6237,
+      "step": 2559
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 5.626017093658447,
+      "learning_rate": 9.996426703378124e-06,
+      "loss": 0.9069,
+      "step": 2560
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.945868492126465,
+      "learning_rate": 9.99641673026432e-06,
+      "loss": 0.7235,
+      "step": 2561
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.2516446113586426,
+      "learning_rate": 9.996406743257352e-06,
+      "loss": 0.5397,
+      "step": 2562
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.618283271789551,
+      "learning_rate": 9.996396742357248e-06,
+      "loss": 0.7017,
+      "step": 2563
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.9479823112487793,
+      "learning_rate": 9.996386727564039e-06,
+      "loss": 0.5724,
+      "step": 2564
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.7775095701217651,
+      "learning_rate": 9.996376698877746e-06,
+      "loss": 0.7497,
+      "step": 2565
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.5899484157562256,
+      "learning_rate": 9.996366656298404e-06,
+      "loss": 0.889,
+      "step": 2566
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.847632646560669,
+      "learning_rate": 9.996356599826039e-06,
+      "loss": 0.4255,
+      "step": 2567
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.637850046157837,
+      "learning_rate": 9.996346529460675e-06,
+      "loss": 0.9096,
+      "step": 2568
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 4.721668243408203,
+      "learning_rate": 9.996336445202345e-06,
+      "loss": 0.5883,
+      "step": 2569
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.3958325386047363,
+      "learning_rate": 9.996326347051075e-06,
+      "loss": 0.7592,
+      "step": 2570
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 5.269902229309082,
+      "learning_rate": 9.996316235006892e-06,
+      "loss": 0.6991,
+      "step": 2571
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.3510122299194336,
+      "learning_rate": 9.996306109069825e-06,
+      "loss": 0.5992,
+      "step": 2572
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.3948113918304443,
+      "learning_rate": 9.996295969239904e-06,
+      "loss": 0.9251,
+      "step": 2573
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.469532012939453,
+      "learning_rate": 9.996285815517155e-06,
+      "loss": 0.747,
+      "step": 2574
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.479520797729492,
+      "learning_rate": 9.996275647901606e-06,
+      "loss": 0.4351,
+      "step": 2575
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.6190643310546875,
+      "learning_rate": 9.996265466393287e-06,
+      "loss": 0.4298,
+      "step": 2576
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.824856162071228,
+      "learning_rate": 9.996255270992225e-06,
+      "loss": 0.5736,
+      "step": 2577
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.412135601043701,
+      "learning_rate": 9.996245061698448e-06,
+      "loss": 0.6989,
+      "step": 2578
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.493058443069458,
+      "learning_rate": 9.996234838511986e-06,
+      "loss": 0.6982,
+      "step": 2579
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.5981719493865967,
+      "learning_rate": 9.996224601432868e-06,
+      "loss": 0.5001,
+      "step": 2580
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 4.444066047668457,
+      "learning_rate": 9.996214350461118e-06,
+      "loss": 0.6589,
+      "step": 2581
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 4.460372447967529,
+      "learning_rate": 9.996204085596769e-06,
+      "loss": 0.6785,
+      "step": 2582
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 6.4233198165893555,
+      "learning_rate": 9.996193806839848e-06,
+      "loss": 0.1706,
+      "step": 2583
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.3511433601379395,
+      "learning_rate": 9.996183514190383e-06,
+      "loss": 0.8931,
+      "step": 2584
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.4414963722229004,
+      "learning_rate": 9.996173207648405e-06,
+      "loss": 0.4741,
+      "step": 2585
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.8997206687927246,
+      "learning_rate": 9.996162887213937e-06,
+      "loss": 0.6271,
+      "step": 2586
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.9519381523132324,
+      "learning_rate": 9.996152552887017e-06,
+      "loss": 0.7203,
+      "step": 2587
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.5618228912353516,
+      "learning_rate": 9.996142204667665e-06,
+      "loss": 0.8561,
+      "step": 2588
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.432427406311035,
+      "learning_rate": 9.996131842555914e-06,
+      "loss": 0.6282,
+      "step": 2589
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.9075076580047607,
+      "learning_rate": 9.996121466551792e-06,
+      "loss": 1.0085,
+      "step": 2590
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.554550886154175,
+      "learning_rate": 9.996111076655329e-06,
+      "loss": 0.6352,
+      "step": 2591
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 4.216426849365234,
+      "learning_rate": 9.99610067286655e-06,
+      "loss": 0.7395,
+      "step": 2592
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.176440715789795,
+      "learning_rate": 9.996090255185487e-06,
+      "loss": 0.8342,
+      "step": 2593
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.9444363117218018,
+      "learning_rate": 9.996079823612171e-06,
+      "loss": 0.6024,
+      "step": 2594
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.34096097946167,
+      "learning_rate": 9.996069378146627e-06,
+      "loss": 0.709,
+      "step": 2595
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 13.058948516845703,
+      "learning_rate": 9.996058918788885e-06,
+      "loss": 0.668,
+      "step": 2596
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.4837028980255127,
+      "learning_rate": 9.996048445538975e-06,
+      "loss": 0.8027,
+      "step": 2597
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.497065544128418,
+      "learning_rate": 9.996037958396928e-06,
+      "loss": 0.7037,
+      "step": 2598
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 4.034827709197998,
+      "learning_rate": 9.996027457362768e-06,
+      "loss": 0.6932,
+      "step": 2599
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.0214850902557373,
+      "learning_rate": 9.996016942436529e-06,
+      "loss": 0.7409,
+      "step": 2600
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.5961804389953613,
+      "learning_rate": 9.996006413618238e-06,
+      "loss": 0.6043,
+      "step": 2601
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.5819337368011475,
+      "learning_rate": 9.995995870907925e-06,
+      "loss": 0.816,
+      "step": 2602
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 4.707671165466309,
+      "learning_rate": 9.995985314305617e-06,
+      "loss": 0.7805,
+      "step": 2603
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.86713445186615,
+      "learning_rate": 9.995974743811347e-06,
+      "loss": 0.6386,
+      "step": 2604
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 4.148024082183838,
+      "learning_rate": 9.995964159425142e-06,
+      "loss": 0.6221,
+      "step": 2605
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.235164165496826,
+      "learning_rate": 9.995953561147034e-06,
+      "loss": 0.605,
+      "step": 2606
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 4.15742301940918,
+      "learning_rate": 9.995942948977049e-06,
+      "loss": 0.6499,
+      "step": 2607
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.6686553955078125,
+      "learning_rate": 9.995932322915219e-06,
+      "loss": 0.8337,
+      "step": 2608
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.6802525520324707,
+      "learning_rate": 9.995921682961574e-06,
+      "loss": 0.9034,
+      "step": 2609
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.483980178833008,
+      "learning_rate": 9.99591102911614e-06,
+      "loss": 0.7491,
+      "step": 2610
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.7549383640289307,
+      "learning_rate": 9.99590036137895e-06,
+      "loss": 0.6705,
+      "step": 2611
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.8075218200683594,
+      "learning_rate": 9.995889679750032e-06,
+      "loss": 0.3869,
+      "step": 2612
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 9.190667152404785,
+      "learning_rate": 9.995878984229417e-06,
+      "loss": 0.7431,
+      "step": 2613
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.6200437545776367,
+      "learning_rate": 9.995868274817134e-06,
+      "loss": 0.7944,
+      "step": 2614
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.4475884437561035,
+      "learning_rate": 9.995857551513212e-06,
+      "loss": 0.7523,
+      "step": 2615
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.353013753890991,
+      "learning_rate": 9.995846814317683e-06,
+      "loss": 0.6326,
+      "step": 2616
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.2702853679656982,
+      "learning_rate": 9.995836063230574e-06,
+      "loss": 0.5773,
+      "step": 2617
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.315307378768921,
+      "learning_rate": 9.995825298251917e-06,
+      "loss": 0.6334,
+      "step": 2618
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.550778388977051,
+      "learning_rate": 9.995814519381742e-06,
+      "loss": 0.8126,
+      "step": 2619
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.916840076446533,
+      "learning_rate": 9.995803726620077e-06,
+      "loss": 0.6856,
+      "step": 2620
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 4.0156660079956055,
+      "learning_rate": 9.995792919966953e-06,
+      "loss": 0.7826,
+      "step": 2621
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.9298222064971924,
+      "learning_rate": 9.995782099422402e-06,
+      "loss": 0.7354,
+      "step": 2622
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.7575442790985107,
+      "learning_rate": 9.995771264986452e-06,
+      "loss": 0.6825,
+      "step": 2623
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.8365306854248047,
+      "learning_rate": 9.995760416659134e-06,
+      "loss": 0.5427,
+      "step": 2624
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.8398046493530273,
+      "learning_rate": 9.995749554440476e-06,
+      "loss": 0.5446,
+      "step": 2625
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.8123704195022583,
+      "learning_rate": 9.99573867833051e-06,
+      "loss": 0.8903,
+      "step": 2626
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.1899795532226562,
+      "learning_rate": 9.99572778832927e-06,
+      "loss": 0.5777,
+      "step": 2627
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.4705896377563477,
+      "learning_rate": 9.995716884436779e-06,
+      "loss": 0.6258,
+      "step": 2628
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.0592446327209473,
+      "learning_rate": 9.995705966653071e-06,
+      "loss": 0.6077,
+      "step": 2629
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.0794200897216797,
+      "learning_rate": 9.995695034978176e-06,
+      "loss": 0.5766,
+      "step": 2630
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.8642516136169434,
+      "learning_rate": 9.995684089412125e-06,
+      "loss": 0.7517,
+      "step": 2631
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.0395309925079346,
+      "learning_rate": 9.995673129954948e-06,
+      "loss": 0.6534,
+      "step": 2632
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 5.719278812408447,
+      "learning_rate": 9.995662156606677e-06,
+      "loss": 0.576,
+      "step": 2633
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.200354814529419,
+      "learning_rate": 9.99565116936734e-06,
+      "loss": 0.7858,
+      "step": 2634
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.38458514213562,
+      "learning_rate": 9.995640168236968e-06,
+      "loss": 0.5509,
+      "step": 2635
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.5668578147888184,
+      "learning_rate": 9.995629153215591e-06,
+      "loss": 0.6926,
+      "step": 2636
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.214226484298706,
+      "learning_rate": 9.995618124303243e-06,
+      "loss": 0.6028,
+      "step": 2637
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.469146251678467,
+      "learning_rate": 9.99560708149995e-06,
+      "loss": 0.471,
+      "step": 2638
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.2102622985839844,
+      "learning_rate": 9.995596024805746e-06,
+      "loss": 0.6653,
+      "step": 2639
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 4.287530899047852,
+      "learning_rate": 9.995584954220663e-06,
+      "loss": 0.7232,
+      "step": 2640
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.2767562866210938,
+      "learning_rate": 9.995573869744729e-06,
+      "loss": 0.7253,
+      "step": 2641
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.375154495239258,
+      "learning_rate": 9.995562771377974e-06,
+      "loss": 0.7839,
+      "step": 2642
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.412611961364746,
+      "learning_rate": 9.99555165912043e-06,
+      "loss": 0.424,
+      "step": 2643
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 11.70103931427002,
+      "learning_rate": 9.995540532972128e-06,
+      "loss": 0.6338,
+      "step": 2644
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.1534554958343506,
+      "learning_rate": 9.995529392933101e-06,
+      "loss": 0.5925,
+      "step": 2645
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.8911540508270264,
+      "learning_rate": 9.995518239003377e-06,
+      "loss": 0.6834,
+      "step": 2646
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.9050830602645874,
+      "learning_rate": 9.995507071182989e-06,
+      "loss": 0.4481,
+      "step": 2647
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.321720838546753,
+      "learning_rate": 9.995495889471965e-06,
+      "loss": 0.3918,
+      "step": 2648
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.2863194942474365,
+      "learning_rate": 9.99548469387034e-06,
+      "loss": 0.5119,
+      "step": 2649
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.601431131362915,
+      "learning_rate": 9.995473484378143e-06,
+      "loss": 0.3884,
+      "step": 2650
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.3339881896972656,
+      "learning_rate": 9.995462260995405e-06,
+      "loss": 0.483,
+      "step": 2651
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.008679151535034,
+      "learning_rate": 9.995451023722161e-06,
+      "loss": 0.563,
+      "step": 2652
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 4.704974174499512,
+      "learning_rate": 9.995439772558435e-06,
+      "loss": 0.6467,
+      "step": 2653
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.223635196685791,
+      "learning_rate": 9.995428507504264e-06,
+      "loss": 0.6354,
+      "step": 2654
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.8425599336624146,
+      "learning_rate": 9.995417228559678e-06,
+      "loss": 0.7797,
+      "step": 2655
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.544985055923462,
+      "learning_rate": 9.995405935724706e-06,
+      "loss": 0.7911,
+      "step": 2656
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.186382293701172,
+      "learning_rate": 9.995394628999385e-06,
+      "loss": 0.7034,
+      "step": 2657
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.3003475666046143,
+      "learning_rate": 9.99538330838374e-06,
+      "loss": 0.7788,
+      "step": 2658
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.340061664581299,
+      "learning_rate": 9.995371973877805e-06,
+      "loss": 0.5375,
+      "step": 2659
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 4.997055530548096,
+      "learning_rate": 9.995360625481613e-06,
+      "loss": 0.8023,
+      "step": 2660
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.8844261169433594,
+      "learning_rate": 9.995349263195195e-06,
+      "loss": 0.5879,
+      "step": 2661
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 4.966362476348877,
+      "learning_rate": 9.995337887018583e-06,
+      "loss": 0.7691,
+      "step": 2662
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.409059762954712,
+      "learning_rate": 9.995326496951804e-06,
+      "loss": 0.8734,
+      "step": 2663
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.5392870903015137,
+      "learning_rate": 9.995315092994897e-06,
+      "loss": 0.8837,
+      "step": 2664
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 5.357706069946289,
+      "learning_rate": 9.995303675147887e-06,
+      "loss": 0.9014,
+      "step": 2665
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 4.679743766784668,
+      "learning_rate": 9.995292243410812e-06,
+      "loss": 0.6528,
+      "step": 2666
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 5.798387050628662,
+      "learning_rate": 9.995280797783698e-06,
+      "loss": 0.7233,
+      "step": 2667
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.1288223266601562,
+      "learning_rate": 9.99526933826658e-06,
+      "loss": 0.6986,
+      "step": 2668
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.93623685836792,
+      "learning_rate": 9.995257864859489e-06,
+      "loss": 0.7373,
+      "step": 2669
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.6159831285476685,
+      "learning_rate": 9.995246377562457e-06,
+      "loss": 0.7587,
+      "step": 2670
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.3687796592712402,
+      "learning_rate": 9.995234876375517e-06,
+      "loss": 0.7427,
+      "step": 2671
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.590646505355835,
+      "learning_rate": 9.9952233612987e-06,
+      "loss": 0.6224,
+      "step": 2672
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 13.40497875213623,
+      "learning_rate": 9.995211832332039e-06,
+      "loss": 0.781,
+      "step": 2673
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.2031614780426025,
+      "learning_rate": 9.995200289475563e-06,
+      "loss": 0.4678,
+      "step": 2674
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.1858670711517334,
+      "learning_rate": 9.995188732729306e-06,
+      "loss": 0.8082,
+      "step": 2675
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.874131917953491,
+      "learning_rate": 9.995177162093303e-06,
+      "loss": 0.6381,
+      "step": 2676
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.1367716789245605,
+      "learning_rate": 9.99516557756758e-06,
+      "loss": 0.7631,
+      "step": 2677
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.993584156036377,
+      "learning_rate": 9.995153979152178e-06,
+      "loss": 0.7014,
+      "step": 2678
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.5102341175079346,
+      "learning_rate": 9.995142366847118e-06,
+      "loss": 0.8088,
+      "step": 2679
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 4.5388875007629395,
+      "learning_rate": 9.995130740652442e-06,
+      "loss": 0.6395,
+      "step": 2680
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.8532278537750244,
+      "learning_rate": 9.995119100568178e-06,
+      "loss": 0.8037,
+      "step": 2681
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.4949851036071777,
+      "learning_rate": 9.995107446594358e-06,
+      "loss": 0.663,
+      "step": 2682
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.140599250793457,
+      "learning_rate": 9.995095778731017e-06,
+      "loss": 0.6524,
+      "step": 2683
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.494354486465454,
+      "learning_rate": 9.995084096978184e-06,
+      "loss": 0.6817,
+      "step": 2684
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.226050615310669,
+      "learning_rate": 9.995072401335895e-06,
+      "loss": 0.7881,
+      "step": 2685
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.9774057865142822,
+      "learning_rate": 9.995060691804179e-06,
+      "loss": 0.6681,
+      "step": 2686
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 5.0059895515441895,
+      "learning_rate": 9.995048968383071e-06,
+      "loss": 0.7745,
+      "step": 2687
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.132692575454712,
+      "learning_rate": 9.995037231072605e-06,
+      "loss": 0.8836,
+      "step": 2688
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.293931007385254,
+      "learning_rate": 9.995025479872811e-06,
+      "loss": 0.7322,
+      "step": 2689
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.394345998764038,
+      "learning_rate": 9.995013714783722e-06,
+      "loss": 0.6711,
+      "step": 2690
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.8737716674804688,
+      "learning_rate": 9.995001935805369e-06,
+      "loss": 0.7276,
+      "step": 2691
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 12.187643051147461,
+      "learning_rate": 9.994990142937788e-06,
+      "loss": 0.5992,
+      "step": 2692
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.052612781524658,
+      "learning_rate": 9.994978336181012e-06,
+      "loss": 0.7236,
+      "step": 2693
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.3990654945373535,
+      "learning_rate": 9.99496651553507e-06,
+      "loss": 0.5484,
+      "step": 2694
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.1947391033172607,
+      "learning_rate": 9.994954681e-06,
+      "loss": 0.5079,
+      "step": 2695
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.6672263145446777,
+      "learning_rate": 9.994942832575831e-06,
+      "loss": 0.6854,
+      "step": 2696
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.0762033462524414,
+      "learning_rate": 9.994930970262596e-06,
+      "loss": 0.6656,
+      "step": 2697
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.0251760482788086,
+      "learning_rate": 9.99491909406033e-06,
+      "loss": 0.4764,
+      "step": 2698
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.6664655208587646,
+      "learning_rate": 9.994907203969065e-06,
+      "loss": 0.7858,
+      "step": 2699
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.8282389640808105,
+      "learning_rate": 9.994895299988835e-06,
+      "loss": 0.6215,
+      "step": 2700
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 5.42211389541626,
+      "learning_rate": 9.994883382119671e-06,
+      "loss": 0.5937,
+      "step": 2701
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.9677648544311523,
+      "learning_rate": 9.994871450361608e-06,
+      "loss": 0.7581,
+      "step": 2702
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.184814691543579,
+      "learning_rate": 9.994859504714681e-06,
+      "loss": 0.8502,
+      "step": 2703
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 4.215241432189941,
+      "learning_rate": 9.994847545178916e-06,
+      "loss": 0.652,
+      "step": 2704
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.1814897060394287,
+      "learning_rate": 9.994835571754355e-06,
+      "loss": 0.4691,
+      "step": 2705
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.779849052429199,
+      "learning_rate": 9.994823584441025e-06,
+      "loss": 0.4748,
+      "step": 2706
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.6497888565063477,
+      "learning_rate": 9.994811583238964e-06,
+      "loss": 0.45,
+      "step": 2707
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.6090826988220215,
+      "learning_rate": 9.994799568148202e-06,
+      "loss": 0.6729,
+      "step": 2708
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.1825618743896484,
+      "learning_rate": 9.994787539168772e-06,
+      "loss": 0.8644,
+      "step": 2709
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.654388904571533,
+      "learning_rate": 9.99477549630071e-06,
+      "loss": 0.3966,
+      "step": 2710
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.503450393676758,
+      "learning_rate": 9.99476343954405e-06,
+      "loss": 0.554,
+      "step": 2711
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.840984582901001,
+      "learning_rate": 9.994751368898822e-06,
+      "loss": 0.6292,
+      "step": 2712
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.985067844390869,
+      "learning_rate": 9.994739284365062e-06,
+      "loss": 0.3178,
+      "step": 2713
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.221440553665161,
+      "learning_rate": 9.994727185942804e-06,
+      "loss": 0.5436,
+      "step": 2714
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 8.249119758605957,
+      "learning_rate": 9.99471507363208e-06,
+      "loss": 0.5555,
+      "step": 2715
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 4.351923942565918,
+      "learning_rate": 9.994702947432924e-06,
+      "loss": 0.7913,
+      "step": 2716
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.7971537113189697,
+      "learning_rate": 9.994690807345371e-06,
+      "loss": 0.9926,
+      "step": 2717
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.5323433876037598,
+      "learning_rate": 9.994678653369454e-06,
+      "loss": 0.7044,
+      "step": 2718
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 5.369791507720947,
+      "learning_rate": 9.994666485505207e-06,
+      "loss": 0.7721,
+      "step": 2719
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.665408134460449,
+      "learning_rate": 9.994654303752663e-06,
+      "loss": 0.8907,
+      "step": 2720
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.2572875022888184,
+      "learning_rate": 9.994642108111856e-06,
+      "loss": 0.5592,
+      "step": 2721
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.353106737136841,
+      "learning_rate": 9.994629898582819e-06,
+      "loss": 0.6288,
+      "step": 2722
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.677119731903076,
+      "learning_rate": 9.99461767516559e-06,
+      "loss": 0.8205,
+      "step": 2723
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.0483360290527344,
+      "learning_rate": 9.994605437860199e-06,
+      "loss": 0.7365,
+      "step": 2724
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 4.493021011352539,
+      "learning_rate": 9.994593186666681e-06,
+      "loss": 0.8297,
+      "step": 2725
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.2535033226013184,
+      "learning_rate": 9.994580921585072e-06,
+      "loss": 0.6435,
+      "step": 2726
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.5165798664093018,
+      "learning_rate": 9.994568642615403e-06,
+      "loss": 0.5988,
+      "step": 2727
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.1000866889953613,
+      "learning_rate": 9.994556349757709e-06,
+      "loss": 0.3651,
+      "step": 2728
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 6.427459716796875,
+      "learning_rate": 9.994544043012027e-06,
+      "loss": 0.5985,
+      "step": 2729
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.8575778007507324,
+      "learning_rate": 9.994531722378387e-06,
+      "loss": 0.7606,
+      "step": 2730
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.867769718170166,
+      "learning_rate": 9.994519387856826e-06,
+      "loss": 0.7083,
+      "step": 2731
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.2109720706939697,
+      "learning_rate": 9.994507039447378e-06,
+      "loss": 0.6424,
+      "step": 2732
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.882136821746826,
+      "learning_rate": 9.994494677150075e-06,
+      "loss": 0.6689,
+      "step": 2733
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.447798013687134,
+      "learning_rate": 9.994482300964956e-06,
+      "loss": 0.8951,
+      "step": 2734
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.6781530380249023,
+      "learning_rate": 9.994469910892051e-06,
+      "loss": 0.5002,
+      "step": 2735
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.695030450820923,
+      "learning_rate": 9.994457506931397e-06,
+      "loss": 0.6561,
+      "step": 2736
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.770738124847412,
+      "learning_rate": 9.994445089083026e-06,
+      "loss": 0.8929,
+      "step": 2737
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.452932834625244,
+      "learning_rate": 9.994432657346976e-06,
+      "loss": 0.6575,
+      "step": 2738
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.1609368324279785,
+      "learning_rate": 9.99442021172328e-06,
+      "loss": 0.6022,
+      "step": 2739
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.9859631061553955,
+      "learning_rate": 9.99440775221197e-06,
+      "loss": 0.6452,
+      "step": 2740
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 4.416324138641357,
+      "learning_rate": 9.994395278813085e-06,
+      "loss": 0.6566,
+      "step": 2741
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.7797131538391113,
+      "learning_rate": 9.994382791526657e-06,
+      "loss": 0.4925,
+      "step": 2742
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.7478162050247192,
+      "learning_rate": 9.994370290352721e-06,
+      "loss": 0.5766,
+      "step": 2743
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.9453072547912598,
+      "learning_rate": 9.994357775291314e-06,
+      "loss": 0.6857,
+      "step": 2744
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.605677366256714,
+      "learning_rate": 9.994345246342467e-06,
+      "loss": 0.8281,
+      "step": 2745
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.276371479034424,
+      "learning_rate": 9.994332703506217e-06,
+      "loss": 0.6474,
+      "step": 2746
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.0688316822052,
+      "learning_rate": 9.994320146782602e-06,
+      "loss": 0.7367,
+      "step": 2747
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.8654041290283203,
+      "learning_rate": 9.99430757617165e-06,
+      "loss": 0.5684,
+      "step": 2748
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.8744187355041504,
+      "learning_rate": 9.994294991673398e-06,
+      "loss": 0.6293,
+      "step": 2749
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.691484212875366,
+      "learning_rate": 9.994282393287886e-06,
+      "loss": 0.6771,
+      "step": 2750
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.5739870071411133,
+      "learning_rate": 9.994269781015144e-06,
+      "loss": 0.7496,
+      "step": 2751
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.301246166229248,
+      "learning_rate": 9.99425715485521e-06,
+      "loss": 0.8098,
+      "step": 2752
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.9378421306610107,
+      "learning_rate": 9.994244514808116e-06,
+      "loss": 0.7674,
+      "step": 2753
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.1492691040039062,
+      "learning_rate": 9.994231860873899e-06,
+      "loss": 0.8251,
+      "step": 2754
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 7.170670509338379,
+      "learning_rate": 9.994219193052595e-06,
+      "loss": 0.7166,
+      "step": 2755
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 4.86409330368042,
+      "learning_rate": 9.994206511344239e-06,
+      "loss": 0.5425,
+      "step": 2756
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.852001905441284,
+      "learning_rate": 9.994193815748865e-06,
+      "loss": 0.8667,
+      "step": 2757
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.4560344219207764,
+      "learning_rate": 9.994181106266507e-06,
+      "loss": 0.8116,
+      "step": 2758
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.7596282958984375,
+      "learning_rate": 9.994168382897204e-06,
+      "loss": 0.4901,
+      "step": 2759
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.4330029487609863,
+      "learning_rate": 9.99415564564099e-06,
+      "loss": 0.8452,
+      "step": 2760
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.2967772483825684,
+      "learning_rate": 9.994142894497898e-06,
+      "loss": 0.8188,
+      "step": 2761
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.3767051696777344,
+      "learning_rate": 9.994130129467966e-06,
+      "loss": 0.7585,
+      "step": 2762
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.837024688720703,
+      "learning_rate": 9.99411735055123e-06,
+      "loss": 0.5468,
+      "step": 2763
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.6576483249664307,
+      "learning_rate": 9.994104557747725e-06,
+      "loss": 0.5932,
+      "step": 2764
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.419189453125,
+      "learning_rate": 9.994091751057486e-06,
+      "loss": 0.6923,
+      "step": 2765
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.210679531097412,
+      "learning_rate": 9.994078930480547e-06,
+      "loss": 0.6746,
+      "step": 2766
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.7618520259857178,
+      "learning_rate": 9.994066096016946e-06,
+      "loss": 0.7673,
+      "step": 2767
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.1244187355041504,
+      "learning_rate": 9.99405324766672e-06,
+      "loss": 0.591,
+      "step": 2768
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.9455008506774902,
+      "learning_rate": 9.9940403854299e-06,
+      "loss": 0.5405,
+      "step": 2769
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 5.899669170379639,
+      "learning_rate": 9.994027509306525e-06,
+      "loss": 0.6089,
+      "step": 2770
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.8886165618896484,
+      "learning_rate": 9.994014619296631e-06,
+      "loss": 0.8769,
+      "step": 2771
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.1105706691741943,
+      "learning_rate": 9.994001715400252e-06,
+      "loss": 0.4288,
+      "step": 2772
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.2016820907592773,
+      "learning_rate": 9.993988797617426e-06,
+      "loss": 0.7671,
+      "step": 2773
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.7771518230438232,
+      "learning_rate": 9.993975865948189e-06,
+      "loss": 0.5605,
+      "step": 2774
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.335186243057251,
+      "learning_rate": 9.993962920392574e-06,
+      "loss": 0.6605,
+      "step": 2775
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.625105619430542,
+      "learning_rate": 9.99394996095062e-06,
+      "loss": 0.3878,
+      "step": 2776
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.9849460124969482,
+      "learning_rate": 9.99393698762236e-06,
+      "loss": 0.6718,
+      "step": 2777
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.4134681224823,
+      "learning_rate": 9.993924000407834e-06,
+      "loss": 0.4379,
+      "step": 2778
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.952117681503296,
+      "learning_rate": 9.993910999307077e-06,
+      "loss": 0.7659,
+      "step": 2779
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.4597859382629395,
+      "learning_rate": 9.993897984320123e-06,
+      "loss": 0.8073,
+      "step": 2780
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.6119933128356934,
+      "learning_rate": 9.993884955447009e-06,
+      "loss": 0.7885,
+      "step": 2781
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 6.194363117218018,
+      "learning_rate": 9.993871912687772e-06,
+      "loss": 1.0103,
+      "step": 2782
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.1386516094207764,
+      "learning_rate": 9.993858856042448e-06,
+      "loss": 0.6072,
+      "step": 2783
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.224351406097412,
+      "learning_rate": 9.993845785511075e-06,
+      "loss": 0.47,
+      "step": 2784
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.944216251373291,
+      "learning_rate": 9.993832701093683e-06,
+      "loss": 0.7491,
+      "step": 2785
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.4547953605651855,
+      "learning_rate": 9.993819602790317e-06,
+      "loss": 0.9027,
+      "step": 2786
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.55621600151062,
+      "learning_rate": 9.99380649060101e-06,
+      "loss": 0.5374,
+      "step": 2787
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.79577374458313,
+      "learning_rate": 9.993793364525796e-06,
+      "loss": 0.4195,
+      "step": 2788
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.563861131668091,
+      "learning_rate": 9.993780224564714e-06,
+      "loss": 0.6494,
+      "step": 2789
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.8254218101501465,
+      "learning_rate": 9.993767070717799e-06,
+      "loss": 0.8103,
+      "step": 2790
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 4.032562732696533,
+      "learning_rate": 9.993753902985088e-06,
+      "loss": 0.6241,
+      "step": 2791
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 6.333394527435303,
+      "learning_rate": 9.99374072136662e-06,
+      "loss": 0.7781,
+      "step": 2792
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.0975887775421143,
+      "learning_rate": 9.993727525862429e-06,
+      "loss": 0.8499,
+      "step": 2793
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.572589635848999,
+      "learning_rate": 9.993714316472552e-06,
+      "loss": 0.6334,
+      "step": 2794
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.371410608291626,
+      "learning_rate": 9.993701093197026e-06,
+      "loss": 0.7165,
+      "step": 2795
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.419827938079834,
+      "learning_rate": 9.99368785603589e-06,
+      "loss": 0.8206,
+      "step": 2796
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 4.6715264320373535,
+      "learning_rate": 9.993674604989176e-06,
+      "loss": 0.7203,
+      "step": 2797
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 8.355217933654785,
+      "learning_rate": 9.993661340056925e-06,
+      "loss": 0.741,
+      "step": 2798
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.5038905143737793,
+      "learning_rate": 9.993648061239173e-06,
+      "loss": 1.044,
+      "step": 2799
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.4246039390563965,
+      "learning_rate": 9.993634768535954e-06,
+      "loss": 0.4854,
+      "step": 2800
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.6116254329681396,
+      "learning_rate": 9.99362146194731e-06,
+      "loss": 0.7398,
+      "step": 2801
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 4.0506110191345215,
+      "learning_rate": 9.993608141473275e-06,
+      "loss": 0.4848,
+      "step": 2802
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 10.51820182800293,
+      "learning_rate": 9.993594807113884e-06,
+      "loss": 0.5858,
+      "step": 2803
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.670117139816284,
+      "learning_rate": 9.99358145886918e-06,
+      "loss": 0.5474,
+      "step": 2804
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.12894868850708,
+      "learning_rate": 9.993568096739195e-06,
+      "loss": 0.7726,
+      "step": 2805
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 5.370707988739014,
+      "learning_rate": 9.993554720723966e-06,
+      "loss": 0.8523,
+      "step": 2806
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 4.565191268920898,
+      "learning_rate": 9.993541330823534e-06,
+      "loss": 0.5722,
+      "step": 2807
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.033546209335327,
+      "learning_rate": 9.993527927037934e-06,
+      "loss": 0.7292,
+      "step": 2808
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.6311919689178467,
+      "learning_rate": 9.993514509367203e-06,
+      "loss": 0.8378,
+      "step": 2809
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.379049062728882,
+      "learning_rate": 9.99350107781138e-06,
+      "loss": 0.4897,
+      "step": 2810
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.8828821182250977,
+      "learning_rate": 9.9934876323705e-06,
+      "loss": 0.6458,
+      "step": 2811
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.0239059925079346,
+      "learning_rate": 9.993474173044602e-06,
+      "loss": 0.6479,
+      "step": 2812
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 5.824314117431641,
+      "learning_rate": 9.993460699833721e-06,
+      "loss": 0.5072,
+      "step": 2813
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.230727195739746,
+      "learning_rate": 9.993447212737898e-06,
+      "loss": 0.5448,
+      "step": 2814
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.0918266773223877,
+      "learning_rate": 9.99343371175717e-06,
+      "loss": 0.6336,
+      "step": 2815
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.0021491050720215,
+      "learning_rate": 9.993420196891572e-06,
+      "loss": 0.9026,
+      "step": 2816
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.761603832244873,
+      "learning_rate": 9.993406668141144e-06,
+      "loss": 0.5186,
+      "step": 2817
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 5.575868606567383,
+      "learning_rate": 9.993393125505921e-06,
+      "loss": 0.6512,
+      "step": 2818
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.199352979660034,
+      "learning_rate": 9.993379568985943e-06,
+      "loss": 0.3481,
+      "step": 2819
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.5691990852355957,
+      "learning_rate": 9.993365998581247e-06,
+      "loss": 0.6759,
+      "step": 2820
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.350090980529785,
+      "learning_rate": 9.99335241429187e-06,
+      "loss": 0.6893,
+      "step": 2821
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 4.054163455963135,
+      "learning_rate": 9.993338816117853e-06,
+      "loss": 0.7887,
+      "step": 2822
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 4.131401538848877,
+      "learning_rate": 9.993325204059229e-06,
+      "loss": 0.5973,
+      "step": 2823
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 5.7513580322265625,
+      "learning_rate": 9.993311578116038e-06,
+      "loss": 0.7041,
+      "step": 2824
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.620929718017578,
+      "learning_rate": 9.99329793828832e-06,
+      "loss": 0.6317,
+      "step": 2825
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.918776512145996,
+      "learning_rate": 9.993284284576108e-06,
+      "loss": 0.5252,
+      "step": 2826
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.623322010040283,
+      "learning_rate": 9.993270616979445e-06,
+      "loss": 0.7067,
+      "step": 2827
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.16457200050354,
+      "learning_rate": 9.993256935498367e-06,
+      "loss": 0.6572,
+      "step": 2828
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.7086689472198486,
+      "learning_rate": 9.993243240132913e-06,
+      "loss": 0.5979,
+      "step": 2829
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.728606700897217,
+      "learning_rate": 9.993229530883118e-06,
+      "loss": 0.7167,
+      "step": 2830
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.765265941619873,
+      "learning_rate": 9.993215807749023e-06,
+      "loss": 0.9549,
+      "step": 2831
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.6423394680023193,
+      "learning_rate": 9.993202070730665e-06,
+      "loss": 0.708,
+      "step": 2832
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.3615870475769043,
+      "learning_rate": 9.993188319828082e-06,
+      "loss": 0.5299,
+      "step": 2833
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.187349319458008,
+      "learning_rate": 9.993174555041314e-06,
+      "loss": 0.561,
+      "step": 2834
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 4.661136150360107,
+      "learning_rate": 9.9931607763704e-06,
+      "loss": 0.672,
+      "step": 2835
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.655280351638794,
+      "learning_rate": 9.993146983815372e-06,
+      "loss": 0.7212,
+      "step": 2836
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.4024155139923096,
+      "learning_rate": 9.993133177376275e-06,
+      "loss": 0.89,
+      "step": 2837
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.8030943870544434,
+      "learning_rate": 9.993119357053145e-06,
+      "loss": 0.6351,
+      "step": 2838
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.7512426376342773,
+      "learning_rate": 9.99310552284602e-06,
+      "loss": 0.5682,
+      "step": 2839
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.2942912578582764,
+      "learning_rate": 9.99309167475494e-06,
+      "loss": 0.7283,
+      "step": 2840
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.4400441646575928,
+      "learning_rate": 9.99307781277994e-06,
+      "loss": 0.4451,
+      "step": 2841
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.634674072265625,
+      "learning_rate": 9.993063936921064e-06,
+      "loss": 0.5784,
+      "step": 2842
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 6.285053730010986,
+      "learning_rate": 9.993050047178345e-06,
+      "loss": 0.5081,
+      "step": 2843
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.5068068504333496,
+      "learning_rate": 9.993036143551826e-06,
+      "loss": 0.6326,
+      "step": 2844
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.9335391521453857,
+      "learning_rate": 9.993022226041544e-06,
+      "loss": 0.7644,
+      "step": 2845
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 4.194652080535889,
+      "learning_rate": 9.993008294647537e-06,
+      "loss": 0.7699,
+      "step": 2846
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.867828607559204,
+      "learning_rate": 9.992994349369846e-06,
+      "loss": 0.5958,
+      "step": 2847
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 4.531279563903809,
+      "learning_rate": 9.992980390208507e-06,
+      "loss": 0.6234,
+      "step": 2848
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.9134654998779297,
+      "learning_rate": 9.992966417163558e-06,
+      "loss": 0.6476,
+      "step": 2849
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.2247607707977295,
+      "learning_rate": 9.992952430235042e-06,
+      "loss": 0.7629,
+      "step": 2850
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.841979503631592,
+      "learning_rate": 9.992938429422996e-06,
+      "loss": 0.6704,
+      "step": 2851
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.5065271854400635,
+      "learning_rate": 9.992924414727458e-06,
+      "loss": 0.8205,
+      "step": 2852
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.129704236984253,
+      "learning_rate": 9.992910386148466e-06,
+      "loss": 0.8357,
+      "step": 2853
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 4.298375606536865,
+      "learning_rate": 9.992896343686062e-06,
+      "loss": 1.0262,
+      "step": 2854
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.8073269128799438,
+      "learning_rate": 9.992882287340284e-06,
+      "loss": 0.666,
+      "step": 2855
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 7.318272113800049,
+      "learning_rate": 9.992868217111171e-06,
+      "loss": 0.7672,
+      "step": 2856
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.2605955600738525,
+      "learning_rate": 9.992854132998762e-06,
+      "loss": 0.7442,
+      "step": 2857
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.5132033824920654,
+      "learning_rate": 9.992840035003096e-06,
+      "loss": 0.6748,
+      "step": 2858
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.679725170135498,
+      "learning_rate": 9.99282592312421e-06,
+      "loss": 0.7448,
+      "step": 2859
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.342440605163574,
+      "learning_rate": 9.992811797362148e-06,
+      "loss": 0.4428,
+      "step": 2860
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 4.611356735229492,
+      "learning_rate": 9.992797657716945e-06,
+      "loss": 0.773,
+      "step": 2861
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 4.034618854522705,
+      "learning_rate": 9.992783504188644e-06,
+      "loss": 0.6754,
+      "step": 2862
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.962397575378418,
+      "learning_rate": 9.99276933677728e-06,
+      "loss": 0.5588,
+      "step": 2863
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.2719624042510986,
+      "learning_rate": 9.992755155482896e-06,
+      "loss": 0.8347,
+      "step": 2864
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.63777232170105,
+      "learning_rate": 9.992740960305531e-06,
+      "loss": 0.7321,
+      "step": 2865
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 4.445666313171387,
+      "learning_rate": 9.992726751245223e-06,
+      "loss": 0.6252,
+      "step": 2866
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.022531270980835,
+      "learning_rate": 9.992712528302011e-06,
+      "loss": 0.8698,
+      "step": 2867
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.06821608543396,
+      "learning_rate": 9.992698291475938e-06,
+      "loss": 0.4672,
+      "step": 2868
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.5280938148498535,
+      "learning_rate": 9.99268404076704e-06,
+      "loss": 0.7414,
+      "step": 2869
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.4655606746673584,
+      "learning_rate": 9.992669776175358e-06,
+      "loss": 0.7214,
+      "step": 2870
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.9455643892288208,
+      "learning_rate": 9.992655497700932e-06,
+      "loss": 0.8799,
+      "step": 2871
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.7536349296569824,
+      "learning_rate": 9.992641205343802e-06,
+      "loss": 0.5915,
+      "step": 2872
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 5.463235855102539,
+      "learning_rate": 9.992626899104006e-06,
+      "loss": 0.8459,
+      "step": 2873
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.8971407413482666,
+      "learning_rate": 9.992612578981584e-06,
+      "loss": 0.4122,
+      "step": 2874
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.4887521266937256,
+      "learning_rate": 9.99259824497658e-06,
+      "loss": 0.9552,
+      "step": 2875
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.2941577434539795,
+      "learning_rate": 9.992583897089028e-06,
+      "loss": 0.4363,
+      "step": 2876
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.9514448642730713,
+      "learning_rate": 9.99256953531897e-06,
+      "loss": 0.7778,
+      "step": 2877
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.2805612087249756,
+      "learning_rate": 9.992555159666448e-06,
+      "loss": 0.5671,
+      "step": 2878
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 4.2931976318359375,
+      "learning_rate": 9.992540770131501e-06,
+      "loss": 0.7527,
+      "step": 2879
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.5170135498046875,
+      "learning_rate": 9.992526366714165e-06,
+      "loss": 0.8031,
+      "step": 2880
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 5.6111040115356445,
+      "learning_rate": 9.992511949414487e-06,
+      "loss": 0.462,
+      "step": 2881
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.375164031982422,
+      "learning_rate": 9.9924975182325e-06,
+      "loss": 0.5797,
+      "step": 2882
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.866734027862549,
+      "learning_rate": 9.99248307316825e-06,
+      "loss": 0.667,
+      "step": 2883
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 5.50405740737915,
+      "learning_rate": 9.992468614221775e-06,
+      "loss": 0.62,
+      "step": 2884
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.4789628982543945,
+      "learning_rate": 9.992454141393115e-06,
+      "loss": 0.5843,
+      "step": 2885
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.6093902587890625,
+      "learning_rate": 9.992439654682308e-06,
+      "loss": 0.7252,
+      "step": 2886
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.293285608291626,
+      "learning_rate": 9.9924251540894e-06,
+      "loss": 0.5087,
+      "step": 2887
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 4.091044902801514,
+      "learning_rate": 9.992410639614425e-06,
+      "loss": 0.6335,
+      "step": 2888
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.2205917835235596,
+      "learning_rate": 9.992396111257427e-06,
+      "loss": 0.6751,
+      "step": 2889
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.2955284118652344,
+      "learning_rate": 9.992381569018447e-06,
+      "loss": 0.8695,
+      "step": 2890
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.4008026123046875,
+      "learning_rate": 9.992367012897524e-06,
+      "loss": 0.7331,
+      "step": 2891
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.4990625381469727,
+      "learning_rate": 9.992352442894697e-06,
+      "loss": 0.5908,
+      "step": 2892
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.7456729412078857,
+      "learning_rate": 9.992337859010009e-06,
+      "loss": 0.7203,
+      "step": 2893
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.1481616497039795,
+      "learning_rate": 9.992323261243498e-06,
+      "loss": 0.6994,
+      "step": 2894
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 5.392573833465576,
+      "learning_rate": 9.992308649595208e-06,
+      "loss": 0.7004,
+      "step": 2895
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.8863072395324707,
+      "learning_rate": 9.992294024065176e-06,
+      "loss": 0.5309,
+      "step": 2896
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 6.850286483764648,
+      "learning_rate": 9.992279384653445e-06,
+      "loss": 0.6184,
+      "step": 2897
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.5136985778808594,
+      "learning_rate": 9.992264731360057e-06,
+      "loss": 0.6745,
+      "step": 2898
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.291095495223999,
+      "learning_rate": 9.992250064185049e-06,
+      "loss": 0.7076,
+      "step": 2899
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.1713502407073975,
+      "learning_rate": 9.992235383128465e-06,
+      "loss": 0.6026,
+      "step": 2900
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.5613207817077637,
+      "learning_rate": 9.992220688190343e-06,
+      "loss": 0.5865,
+      "step": 2901
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.187192678451538,
+      "learning_rate": 9.992205979370726e-06,
+      "loss": 0.661,
+      "step": 2902
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.5979228019714355,
+      "learning_rate": 9.992191256669656e-06,
+      "loss": 0.7654,
+      "step": 2903
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.136216402053833,
+      "learning_rate": 9.992176520087171e-06,
+      "loss": 1.0194,
+      "step": 2904
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.427950382232666,
+      "learning_rate": 9.992161769623312e-06,
+      "loss": 0.7208,
+      "step": 2905
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.4663748741149902,
+      "learning_rate": 9.992147005278122e-06,
+      "loss": 0.618,
+      "step": 2906
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.140611171722412,
+      "learning_rate": 9.992132227051642e-06,
+      "loss": 0.6408,
+      "step": 2907
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 4.257504940032959,
+      "learning_rate": 9.992117434943912e-06,
+      "loss": 0.7976,
+      "step": 2908
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.9434642791748047,
+      "learning_rate": 9.992102628954973e-06,
+      "loss": 0.8845,
+      "step": 2909
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.6198570728302,
+      "learning_rate": 9.992087809084867e-06,
+      "loss": 0.6508,
+      "step": 2910
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.0199637413024902,
+      "learning_rate": 9.992072975333635e-06,
+      "loss": 0.7888,
+      "step": 2911
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.7758872509002686,
+      "learning_rate": 9.992058127701319e-06,
+      "loss": 0.7572,
+      "step": 2912
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 4.272189140319824,
+      "learning_rate": 9.992043266187959e-06,
+      "loss": 0.9928,
+      "step": 2913
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.739971399307251,
+      "learning_rate": 9.992028390793595e-06,
+      "loss": 0.6915,
+      "step": 2914
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.523587465286255,
+      "learning_rate": 9.992013501518271e-06,
+      "loss": 0.6669,
+      "step": 2915
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.946023941040039,
+      "learning_rate": 9.991998598362027e-06,
+      "loss": 0.6247,
+      "step": 2916
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.4887783527374268,
+      "learning_rate": 9.991983681324907e-06,
+      "loss": 0.783,
+      "step": 2917
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.7470450401306152,
+      "learning_rate": 9.991968750406948e-06,
+      "loss": 0.8877,
+      "step": 2918
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.1424524784088135,
+      "learning_rate": 9.991953805608196e-06,
+      "loss": 0.5543,
+      "step": 2919
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.2992448806762695,
+      "learning_rate": 9.99193884692869e-06,
+      "loss": 0.7029,
+      "step": 2920
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.1463518142700195,
+      "learning_rate": 9.991923874368472e-06,
+      "loss": 0.441,
+      "step": 2921
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.0868799686431885,
+      "learning_rate": 9.991908887927585e-06,
+      "loss": 0.4922,
+      "step": 2922
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 6.255990505218506,
+      "learning_rate": 9.991893887606065e-06,
+      "loss": 0.6301,
+      "step": 2923
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.756728410720825,
+      "learning_rate": 9.991878873403963e-06,
+      "loss": 0.5456,
+      "step": 2924
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.729233980178833,
+      "learning_rate": 9.991863845321313e-06,
+      "loss": 0.6734,
+      "step": 2925
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 4.119256496429443,
+      "learning_rate": 9.991848803358161e-06,
+      "loss": 0.4637,
+      "step": 2926
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.623368501663208,
+      "learning_rate": 9.991833747514547e-06,
+      "loss": 0.9824,
+      "step": 2927
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.400261878967285,
+      "learning_rate": 9.991818677790512e-06,
+      "loss": 0.5552,
+      "step": 2928
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 6.542472839355469,
+      "learning_rate": 9.9918035941861e-06,
+      "loss": 0.7721,
+      "step": 2929
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 4.582450866699219,
+      "learning_rate": 9.991788496701354e-06,
+      "loss": 0.5404,
+      "step": 2930
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 5.1144609451293945,
+      "learning_rate": 9.991773385336312e-06,
+      "loss": 0.7316,
+      "step": 2931
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 5.349917888641357,
+      "learning_rate": 9.991758260091021e-06,
+      "loss": 0.7858,
+      "step": 2932
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.965827226638794,
+      "learning_rate": 9.991743120965518e-06,
+      "loss": 0.5484,
+      "step": 2933
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.288670539855957,
+      "learning_rate": 9.991727967959845e-06,
+      "loss": 0.5588,
+      "step": 2934
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.9020028114318848,
+      "learning_rate": 9.99171280107405e-06,
+      "loss": 0.3725,
+      "step": 2935
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.1520330905914307,
+      "learning_rate": 9.991697620308172e-06,
+      "loss": 0.7046,
+      "step": 2936
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.247208833694458,
+      "learning_rate": 9.99168242566225e-06,
+      "loss": 0.7237,
+      "step": 2937
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.070176601409912,
+      "learning_rate": 9.991667217136331e-06,
+      "loss": 0.7614,
+      "step": 2938
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.047189474105835,
+      "learning_rate": 9.991651994730455e-06,
+      "loss": 0.565,
+      "step": 2939
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.4759933948516846,
+      "learning_rate": 9.991636758444664e-06,
+      "loss": 0.8128,
+      "step": 2940
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.1523261070251465,
+      "learning_rate": 9.991621508279003e-06,
+      "loss": 0.6788,
+      "step": 2941
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.5174198150634766,
+      "learning_rate": 9.991606244233511e-06,
+      "loss": 0.7228,
+      "step": 2942
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 4.822750091552734,
+      "learning_rate": 9.991590966308233e-06,
+      "loss": 0.7877,
+      "step": 2943
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.509505033493042,
+      "learning_rate": 9.99157567450321e-06,
+      "loss": 0.4361,
+      "step": 2944
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.6541147232055664,
+      "learning_rate": 9.991560368818484e-06,
+      "loss": 0.5748,
+      "step": 2945
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 7.594227313995361,
+      "learning_rate": 9.991545049254099e-06,
+      "loss": 0.8164,
+      "step": 2946
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.395604133605957,
+      "learning_rate": 9.991529715810098e-06,
+      "loss": 0.7677,
+      "step": 2947
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 4.638432025909424,
+      "learning_rate": 9.991514368486521e-06,
+      "loss": 0.6821,
+      "step": 2948
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.6283493041992188,
+      "learning_rate": 9.991499007283412e-06,
+      "loss": 0.5665,
+      "step": 2949
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.0745348930358887,
+      "learning_rate": 9.991483632200815e-06,
+      "loss": 0.628,
+      "step": 2950
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.221940755844116,
+      "learning_rate": 9.991468243238774e-06,
+      "loss": 0.8184,
+      "step": 2951
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.7110071182250977,
+      "learning_rate": 9.991452840397328e-06,
+      "loss": 0.6086,
+      "step": 2952
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.252732992172241,
+      "learning_rate": 9.99143742367652e-06,
+      "loss": 0.5985,
+      "step": 2953
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.2529447078704834,
+      "learning_rate": 9.991421993076398e-06,
+      "loss": 0.7561,
+      "step": 2954
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.7104122638702393,
+      "learning_rate": 9.991406548596997e-06,
+      "loss": 0.5865,
+      "step": 2955
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.6529078483581543,
+      "learning_rate": 9.991391090238366e-06,
+      "loss": 0.6229,
+      "step": 2956
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.2499256134033203,
+      "learning_rate": 9.991375618000548e-06,
+      "loss": 0.5306,
+      "step": 2957
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.138967275619507,
+      "learning_rate": 9.991360131883583e-06,
+      "loss": 0.7131,
+      "step": 2958
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.442899465560913,
+      "learning_rate": 9.991344631887515e-06,
+      "loss": 0.7273,
+      "step": 2959
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 4.6864752769470215,
+      "learning_rate": 9.991329118012387e-06,
+      "loss": 0.4651,
+      "step": 2960
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.062824249267578,
+      "learning_rate": 9.991313590258242e-06,
+      "loss": 0.5198,
+      "step": 2961
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 5.5290656089782715,
+      "learning_rate": 9.991298048625126e-06,
+      "loss": 0.8453,
+      "step": 2962
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.5340914726257324,
+      "learning_rate": 9.991282493113077e-06,
+      "loss": 0.6163,
+      "step": 2963
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 4.791538715362549,
+      "learning_rate": 9.991266923722144e-06,
+      "loss": 0.7318,
+      "step": 2964
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.6932268142700195,
+      "learning_rate": 9.991251340452365e-06,
+      "loss": 0.3806,
+      "step": 2965
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.737328052520752,
+      "learning_rate": 9.991235743303787e-06,
+      "loss": 0.5098,
+      "step": 2966
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.155716896057129,
+      "learning_rate": 9.991220132276453e-06,
+      "loss": 0.569,
+      "step": 2967
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 17.85845184326172,
+      "learning_rate": 9.991204507370404e-06,
+      "loss": 0.6454,
+      "step": 2968
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.5561022758483887,
+      "learning_rate": 9.991188868585686e-06,
+      "loss": 0.8947,
+      "step": 2969
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.5876545906066895,
+      "learning_rate": 9.99117321592234e-06,
+      "loss": 0.9637,
+      "step": 2970
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.300630807876587,
+      "learning_rate": 9.991157549380412e-06,
+      "loss": 0.6277,
+      "step": 2971
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.493624687194824,
+      "learning_rate": 9.991141868959946e-06,
+      "loss": 0.8609,
+      "step": 2972
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 26.6833553314209,
+      "learning_rate": 9.991126174660983e-06,
+      "loss": 0.6897,
+      "step": 2973
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.6392003297805786,
+      "learning_rate": 9.991110466483567e-06,
+      "loss": 0.424,
+      "step": 2974
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.776448965072632,
+      "learning_rate": 9.991094744427744e-06,
+      "loss": 0.8816,
+      "step": 2975
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.445464611053467,
+      "learning_rate": 9.991079008493556e-06,
+      "loss": 0.6343,
+      "step": 2976
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.4207024574279785,
+      "learning_rate": 9.991063258681047e-06,
+      "loss": 0.4986,
+      "step": 2977
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.053776741027832,
+      "learning_rate": 9.991047494990259e-06,
+      "loss": 0.6015,
+      "step": 2978
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.9248101711273193,
+      "learning_rate": 9.991031717421239e-06,
+      "loss": 0.647,
+      "step": 2979
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.1757168769836426,
+      "learning_rate": 9.991015925974028e-06,
+      "loss": 0.7171,
+      "step": 2980
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.472036123275757,
+      "learning_rate": 9.991000120648674e-06,
+      "loss": 0.9341,
+      "step": 2981
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.9403889179229736,
+      "learning_rate": 9.990984301445217e-06,
+      "loss": 0.5836,
+      "step": 2982
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.8253488540649414,
+      "learning_rate": 9.990968468363703e-06,
+      "loss": 0.7046,
+      "step": 2983
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.406768560409546,
+      "learning_rate": 9.990952621404173e-06,
+      "loss": 0.6114,
+      "step": 2984
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.9014086723327637,
+      "learning_rate": 9.990936760566677e-06,
+      "loss": 0.7804,
+      "step": 2985
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.3881900310516357,
+      "learning_rate": 9.990920885851253e-06,
+      "loss": 0.6951,
+      "step": 2986
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.30470609664917,
+      "learning_rate": 9.99090499725795e-06,
+      "loss": 0.5634,
+      "step": 2987
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.3532512187957764,
+      "learning_rate": 9.990889094786806e-06,
+      "loss": 0.7068,
+      "step": 2988
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.012234926223755,
+      "learning_rate": 9.990873178437872e-06,
+      "loss": 0.6852,
+      "step": 2989
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.4101247787475586,
+      "learning_rate": 9.990857248211188e-06,
+      "loss": 0.6556,
+      "step": 2990
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.5446622371673584,
+      "learning_rate": 9.9908413041068e-06,
+      "loss": 0.6264,
+      "step": 2991
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.4343128204345703,
+      "learning_rate": 9.990825346124753e-06,
+      "loss": 0.6347,
+      "step": 2992
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.787863850593567,
+      "learning_rate": 9.99080937426509e-06,
+      "loss": 0.7673,
+      "step": 2993
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.306513547897339,
+      "learning_rate": 9.990793388527853e-06,
+      "loss": 0.7822,
+      "step": 2994
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.173886299133301,
+      "learning_rate": 9.990777388913092e-06,
+      "loss": 0.5305,
+      "step": 2995
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.4892499446868896,
+      "learning_rate": 9.990761375420848e-06,
+      "loss": 0.7357,
+      "step": 2996
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.3467400074005127,
+      "learning_rate": 9.990745348051166e-06,
+      "loss": 0.5773,
+      "step": 2997
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.3291733264923096,
+      "learning_rate": 9.990729306804092e-06,
+      "loss": 0.6663,
+      "step": 2998
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.4334731101989746,
+      "learning_rate": 9.990713251679667e-06,
+      "loss": 0.5908,
+      "step": 2999
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.3349673748016357,
+      "learning_rate": 9.99069718267794e-06,
+      "loss": 0.6593,
+      "step": 3000
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.0699594020843506,
+      "learning_rate": 9.990681099798954e-06,
+      "loss": 0.4838,
+      "step": 3001
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.6403846740722656,
+      "learning_rate": 9.99066500304275e-06,
+      "loss": 0.5654,
+      "step": 3002
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.51013445854187,
+      "learning_rate": 9.99064889240938e-06,
+      "loss": 0.429,
+      "step": 3003
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.4295032024383545,
+      "learning_rate": 9.990632767898884e-06,
+      "loss": 0.511,
+      "step": 3004
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.62453293800354,
+      "learning_rate": 9.990616629511308e-06,
+      "loss": 0.6667,
+      "step": 3005
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.3855793476104736,
+      "learning_rate": 9.990600477246695e-06,
+      "loss": 0.6118,
+      "step": 3006
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.504160165786743,
+      "learning_rate": 9.990584311105091e-06,
+      "loss": 0.6699,
+      "step": 3007
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.181154251098633,
+      "learning_rate": 9.990568131086544e-06,
+      "loss": 0.746,
+      "step": 3008
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.150712728500366,
+      "learning_rate": 9.990551937191097e-06,
+      "loss": 0.8987,
+      "step": 3009
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.8256878852844238,
+      "learning_rate": 9.990535729418793e-06,
+      "loss": 0.6238,
+      "step": 3010
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.6362557411193848,
+      "learning_rate": 9.990519507769678e-06,
+      "loss": 0.7634,
+      "step": 3011
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.1953370571136475,
+      "learning_rate": 9.9905032722438e-06,
+      "loss": 0.5003,
+      "step": 3012
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.1995513439178467,
+      "learning_rate": 9.9904870228412e-06,
+      "loss": 0.8395,
+      "step": 3013
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 4.859455108642578,
+      "learning_rate": 9.990470759561925e-06,
+      "loss": 0.7184,
+      "step": 3014
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.2524561882019043,
+      "learning_rate": 9.990454482406022e-06,
+      "loss": 0.757,
+      "step": 3015
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 5.149326324462891,
+      "learning_rate": 9.990438191373532e-06,
+      "loss": 0.6085,
+      "step": 3016
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.6205978393554688,
+      "learning_rate": 9.990421886464505e-06,
+      "loss": 0.5157,
+      "step": 3017
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.389133930206299,
+      "learning_rate": 9.990405567678983e-06,
+      "loss": 0.463,
+      "step": 3018
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 4.669370651245117,
+      "learning_rate": 9.990389235017013e-06,
+      "loss": 0.4342,
+      "step": 3019
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.160144567489624,
+      "learning_rate": 9.99037288847864e-06,
+      "loss": 0.7232,
+      "step": 3020
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.719576358795166,
+      "learning_rate": 9.990356528063909e-06,
+      "loss": 0.6317,
+      "step": 3021
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.4894144535064697,
+      "learning_rate": 9.990340153772868e-06,
+      "loss": 0.5832,
+      "step": 3022
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.6393210887908936,
+      "learning_rate": 9.990323765605559e-06,
+      "loss": 0.5358,
+      "step": 3023
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.1693387031555176,
+      "learning_rate": 9.990307363562029e-06,
+      "loss": 0.6351,
+      "step": 3024
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.5316121578216553,
+      "learning_rate": 9.990290947642322e-06,
+      "loss": 0.434,
+      "step": 3025
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.29416561126709,
+      "learning_rate": 9.990274517846488e-06,
+      "loss": 0.7532,
+      "step": 3026
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.116729736328125,
+      "learning_rate": 9.990258074174568e-06,
+      "loss": 0.7236,
+      "step": 3027
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.714175224304199,
+      "learning_rate": 9.990241616626611e-06,
+      "loss": 0.9135,
+      "step": 3028
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 60.87980270385742,
+      "learning_rate": 9.990225145202662e-06,
+      "loss": 1.1308,
+      "step": 3029
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 6.079155921936035,
+      "learning_rate": 9.990208659902764e-06,
+      "loss": 0.566,
+      "step": 3030
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.2985055446624756,
+      "learning_rate": 9.990192160726966e-06,
+      "loss": 0.8888,
+      "step": 3031
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.9574636220932007,
+      "learning_rate": 9.990175647675313e-06,
+      "loss": 0.6881,
+      "step": 3032
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.9910106658935547,
+      "learning_rate": 9.990159120747851e-06,
+      "loss": 0.6945,
+      "step": 3033
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.6371097564697266,
+      "learning_rate": 9.990142579944625e-06,
+      "loss": 0.6541,
+      "step": 3034
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.9253350496292114,
+      "learning_rate": 9.990126025265684e-06,
+      "loss": 0.5166,
+      "step": 3035
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.5460262298583984,
+      "learning_rate": 9.99010945671107e-06,
+      "loss": 0.7003,
+      "step": 3036
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 6.381752014160156,
+      "learning_rate": 9.990092874280833e-06,
+      "loss": 0.8512,
+      "step": 3037
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.7999508380889893,
+      "learning_rate": 9.990076277975014e-06,
+      "loss": 0.6784,
+      "step": 3038
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.2944064140319824,
+      "learning_rate": 9.990059667793665e-06,
+      "loss": 0.8142,
+      "step": 3039
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.320148229598999,
+      "learning_rate": 9.990043043736828e-06,
+      "loss": 0.6426,
+      "step": 3040
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.590998888015747,
+      "learning_rate": 9.990026405804551e-06,
+      "loss": 0.8137,
+      "step": 3041
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 4.143613338470459,
+      "learning_rate": 9.990009753996878e-06,
+      "loss": 0.6871,
+      "step": 3042
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.750943899154663,
+      "learning_rate": 9.98999308831386e-06,
+      "loss": 0.6717,
+      "step": 3043
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.045504570007324,
+      "learning_rate": 9.989976408755539e-06,
+      "loss": 0.7755,
+      "step": 3044
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 14.126935005187988,
+      "learning_rate": 9.989959715321964e-06,
+      "loss": 0.7738,
+      "step": 3045
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.0904572010040283,
+      "learning_rate": 9.98994300801318e-06,
+      "loss": 0.4741,
+      "step": 3046
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.9179174900054932,
+      "learning_rate": 9.989926286829232e-06,
+      "loss": 0.6746,
+      "step": 3047
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.2798783779144287,
+      "learning_rate": 9.989909551770171e-06,
+      "loss": 0.4479,
+      "step": 3048
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.5107126235961914,
+      "learning_rate": 9.989892802836039e-06,
+      "loss": 0.6773,
+      "step": 3049
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.349120855331421,
+      "learning_rate": 9.989876040026884e-06,
+      "loss": 0.4938,
+      "step": 3050
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.415134906768799,
+      "learning_rate": 9.989859263342755e-06,
+      "loss": 0.6804,
+      "step": 3051
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.4420206546783447,
+      "learning_rate": 9.989842472783695e-06,
+      "loss": 0.5255,
+      "step": 3052
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.8580286502838135,
+      "learning_rate": 9.989825668349753e-06,
+      "loss": 0.6361,
+      "step": 3053
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.849514961242676,
+      "learning_rate": 9.989808850040974e-06,
+      "loss": 0.642,
+      "step": 3054
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.456101655960083,
+      "learning_rate": 9.989792017857406e-06,
+      "loss": 0.7413,
+      "step": 3055
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.190394401550293,
+      "learning_rate": 9.989775171799097e-06,
+      "loss": 0.4411,
+      "step": 3056
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.44746732711792,
+      "learning_rate": 9.989758311866092e-06,
+      "loss": 0.6246,
+      "step": 3057
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.1259820461273193,
+      "learning_rate": 9.989741438058438e-06,
+      "loss": 0.5783,
+      "step": 3058
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.1320786476135254,
+      "learning_rate": 9.989724550376182e-06,
+      "loss": 1.0354,
+      "step": 3059
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.789658546447754,
+      "learning_rate": 9.989707648819371e-06,
+      "loss": 0.6672,
+      "step": 3060
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.9483556747436523,
+      "learning_rate": 9.989690733388052e-06,
+      "loss": 0.7457,
+      "step": 3061
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.7332231998443604,
+      "learning_rate": 9.989673804082275e-06,
+      "loss": 0.8739,
+      "step": 3062
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.4250617027282715,
+      "learning_rate": 9.989656860902082e-06,
+      "loss": 0.4951,
+      "step": 3063
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.863433837890625,
+      "learning_rate": 9.989639903847523e-06,
+      "loss": 0.6622,
+      "step": 3064
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.858032703399658,
+      "learning_rate": 9.989622932918643e-06,
+      "loss": 0.4925,
+      "step": 3065
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.5834615230560303,
+      "learning_rate": 9.989605948115493e-06,
+      "loss": 0.7031,
+      "step": 3066
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 4.498581886291504,
+      "learning_rate": 9.989588949438116e-06,
+      "loss": 0.8397,
+      "step": 3067
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.8805809020996094,
+      "learning_rate": 9.989571936886564e-06,
+      "loss": 0.579,
+      "step": 3068
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.0009689331054688,
+      "learning_rate": 9.98955491046088e-06,
+      "loss": 0.42,
+      "step": 3069
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.222130298614502,
+      "learning_rate": 9.989537870161112e-06,
+      "loss": 0.5793,
+      "step": 3070
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.8727452754974365,
+      "learning_rate": 9.98952081598731e-06,
+      "loss": 0.8104,
+      "step": 3071
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 4.561255931854248,
+      "learning_rate": 9.989503747939518e-06,
+      "loss": 0.7069,
+      "step": 3072
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 4.047937870025635,
+      "learning_rate": 9.989486666017786e-06,
+      "loss": 0.4254,
+      "step": 3073
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 6.159420013427734,
+      "learning_rate": 9.989469570222162e-06,
+      "loss": 0.504,
+      "step": 3074
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.511953830718994,
+      "learning_rate": 9.989452460552691e-06,
+      "loss": 0.7498,
+      "step": 3075
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.3275301456451416,
+      "learning_rate": 9.989435337009421e-06,
+      "loss": 0.7429,
+      "step": 3076
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.9527667760849,
+      "learning_rate": 9.989418199592401e-06,
+      "loss": 0.6185,
+      "step": 3077
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.4435410499572754,
+      "learning_rate": 9.989401048301678e-06,
+      "loss": 0.7459,
+      "step": 3078
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.146987199783325,
+      "learning_rate": 9.9893838831373e-06,
+      "loss": 0.5871,
+      "step": 3079
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.0766496658325195,
+      "learning_rate": 9.989366704099315e-06,
+      "loss": 0.7716,
+      "step": 3080
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.0861449241638184,
+      "learning_rate": 9.989349511187769e-06,
+      "loss": 0.7302,
+      "step": 3081
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.4279816150665283,
+      "learning_rate": 9.989332304402713e-06,
+      "loss": 0.6358,
+      "step": 3082
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.2136685848236084,
+      "learning_rate": 9.989315083744191e-06,
+      "loss": 0.8933,
+      "step": 3083
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.8885183334350586,
+      "learning_rate": 9.989297849212254e-06,
+      "loss": 0.6359,
+      "step": 3084
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.936143159866333,
+      "learning_rate": 9.989280600806947e-06,
+      "loss": 0.7309,
+      "step": 3085
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.3642070293426514,
+      "learning_rate": 9.989263338528322e-06,
+      "loss": 0.5869,
+      "step": 3086
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.707812786102295,
+      "learning_rate": 9.989246062376424e-06,
+      "loss": 0.5645,
+      "step": 3087
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.2277700901031494,
+      "learning_rate": 9.989228772351302e-06,
+      "loss": 0.4911,
+      "step": 3088
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.043829917907715,
+      "learning_rate": 9.989211468453002e-06,
+      "loss": 0.6706,
+      "step": 3089
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.729379892349243,
+      "learning_rate": 9.989194150681575e-06,
+      "loss": 0.6358,
+      "step": 3090
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.8420450687408447,
+      "learning_rate": 9.989176819037069e-06,
+      "loss": 0.6213,
+      "step": 3091
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.7586522102355957,
+      "learning_rate": 9.98915947351953e-06,
+      "loss": 0.8122,
+      "step": 3092
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.572478771209717,
+      "learning_rate": 9.989142114129008e-06,
+      "loss": 0.7615,
+      "step": 3093
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.8502798080444336,
+      "learning_rate": 9.98912474086555e-06,
+      "loss": 0.6803,
+      "step": 3094
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.751263380050659,
+      "learning_rate": 9.989107353729205e-06,
+      "loss": 0.8028,
+      "step": 3095
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.9258484840393066,
+      "learning_rate": 9.989089952720024e-06,
+      "loss": 0.5872,
+      "step": 3096
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.7159086465835571,
+      "learning_rate": 9.98907253783805e-06,
+      "loss": 0.7344,
+      "step": 3097
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 8.038412094116211,
+      "learning_rate": 9.989055109083335e-06,
+      "loss": 0.6851,
+      "step": 3098
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.655643939971924,
+      "learning_rate": 9.989037666455928e-06,
+      "loss": 0.7343,
+      "step": 3099
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.765597343444824,
+      "learning_rate": 9.989020209955873e-06,
+      "loss": 0.613,
+      "step": 3100
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 4.480127811431885,
+      "learning_rate": 9.989002739583224e-06,
+      "loss": 0.863,
+      "step": 3101
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 7.261947154998779,
+      "learning_rate": 9.988985255338027e-06,
+      "loss": 0.6591,
+      "step": 3102
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.7570772171020508,
+      "learning_rate": 9.988967757220331e-06,
+      "loss": 0.3724,
+      "step": 3103
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.3604443073272705,
+      "learning_rate": 9.988950245230184e-06,
+      "loss": 0.5623,
+      "step": 3104
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.5519628524780273,
+      "learning_rate": 9.988932719367636e-06,
+      "loss": 0.5662,
+      "step": 3105
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 5.14898157119751,
+      "learning_rate": 9.988915179632735e-06,
+      "loss": 0.5119,
+      "step": 3106
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.969959259033203,
+      "learning_rate": 9.98889762602553e-06,
+      "loss": 0.8394,
+      "step": 3107
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.8228018283843994,
+      "learning_rate": 9.988880058546069e-06,
+      "loss": 0.9032,
+      "step": 3108
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.645017385482788,
+      "learning_rate": 9.988862477194401e-06,
+      "loss": 0.6656,
+      "step": 3109
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.6133978366851807,
+      "learning_rate": 9.988844881970577e-06,
+      "loss": 0.6677,
+      "step": 3110
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.9054338932037354,
+      "learning_rate": 9.988827272874643e-06,
+      "loss": 0.987,
+      "step": 3111
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.968764543533325,
+      "learning_rate": 9.988809649906649e-06,
+      "loss": 0.7544,
+      "step": 3112
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 4.818802356719971,
+      "learning_rate": 9.988792013066643e-06,
+      "loss": 0.8556,
+      "step": 3113
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.8839964866638184,
+      "learning_rate": 9.988774362354678e-06,
+      "loss": 0.6036,
+      "step": 3114
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.0281012058258057,
+      "learning_rate": 9.9887566977708e-06,
+      "loss": 0.7385,
+      "step": 3115
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.805588960647583,
+      "learning_rate": 9.988739019315056e-06,
+      "loss": 0.6042,
+      "step": 3116
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 4.697748184204102,
+      "learning_rate": 9.9887213269875e-06,
+      "loss": 0.5739,
+      "step": 3117
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.548151969909668,
+      "learning_rate": 9.988703620788178e-06,
+      "loss": 0.8157,
+      "step": 3118
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.3708906173706055,
+      "learning_rate": 9.98868590071714e-06,
+      "loss": 0.5276,
+      "step": 3119
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.5023305416107178,
+      "learning_rate": 9.988668166774435e-06,
+      "loss": 0.5533,
+      "step": 3120
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.0885567665100098,
+      "learning_rate": 9.988650418960114e-06,
+      "loss": 0.9128,
+      "step": 3121
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.39717435836792,
+      "learning_rate": 9.988632657274225e-06,
+      "loss": 0.8247,
+      "step": 3122
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.7205305099487305,
+      "learning_rate": 9.988614881716815e-06,
+      "loss": 0.8163,
+      "step": 3123
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.7779978513717651,
+      "learning_rate": 9.988597092287937e-06,
+      "loss": 0.6458,
+      "step": 3124
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.8078625202178955,
+      "learning_rate": 9.98857928898764e-06,
+      "loss": 0.8659,
+      "step": 3125
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.969067096710205,
+      "learning_rate": 9.988561471815972e-06,
+      "loss": 0.8412,
+      "step": 3126
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.7576310634613037,
+      "learning_rate": 9.988543640772983e-06,
+      "loss": 0.7239,
+      "step": 3127
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 4.010254383087158,
+      "learning_rate": 9.988525795858723e-06,
+      "loss": 0.481,
+      "step": 3128
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 4.483575820922852,
+      "learning_rate": 9.988507937073243e-06,
+      "loss": 1.0361,
+      "step": 3129
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.404714822769165,
+      "learning_rate": 9.98849006441659e-06,
+      "loss": 0.7435,
+      "step": 3130
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.490868330001831,
+      "learning_rate": 9.988472177888816e-06,
+      "loss": 0.7703,
+      "step": 3131
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 4.960220813751221,
+      "learning_rate": 9.988454277489967e-06,
+      "loss": 0.7877,
+      "step": 3132
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.7960000038146973,
+      "learning_rate": 9.988436363220097e-06,
+      "loss": 0.5459,
+      "step": 3133
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.9057888984680176,
+      "learning_rate": 9.988418435079253e-06,
+      "loss": 0.5662,
+      "step": 3134
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.5225672721862793,
+      "learning_rate": 9.988400493067486e-06,
+      "loss": 0.6941,
+      "step": 3135
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.6032121181488037,
+      "learning_rate": 9.98838253718485e-06,
+      "loss": 0.7793,
+      "step": 3136
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.621954917907715,
+      "learning_rate": 9.988364567431387e-06,
+      "loss": 0.5555,
+      "step": 3137
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.0574498176574707,
+      "learning_rate": 9.988346583807152e-06,
+      "loss": 0.366,
+      "step": 3138
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.9505294561386108,
+      "learning_rate": 9.988328586312193e-06,
+      "loss": 0.6314,
+      "step": 3139
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 4.3266987800598145,
+      "learning_rate": 9.98831057494656e-06,
+      "loss": 0.7169,
+      "step": 3140
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.794405937194824,
+      "learning_rate": 9.988292549710307e-06,
+      "loss": 0.8254,
+      "step": 3141
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.642714262008667,
+      "learning_rate": 9.98827451060348e-06,
+      "loss": 0.8641,
+      "step": 3142
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.6087844371795654,
+      "learning_rate": 9.988256457626128e-06,
+      "loss": 0.8816,
+      "step": 3143
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.0330746173858643,
+      "learning_rate": 9.988238390778306e-06,
+      "loss": 0.4579,
+      "step": 3144
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.8280937671661377,
+      "learning_rate": 9.988220310060061e-06,
+      "loss": 0.7824,
+      "step": 3145
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.8886781930923462,
+      "learning_rate": 9.988202215471444e-06,
+      "loss": 0.6828,
+      "step": 3146
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.9344215393066406,
+      "learning_rate": 9.988184107012504e-06,
+      "loss": 0.4258,
+      "step": 3147
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 4.494489669799805,
+      "learning_rate": 9.988165984683294e-06,
+      "loss": 0.7128,
+      "step": 3148
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.7168185710906982,
+      "learning_rate": 9.988147848483863e-06,
+      "loss": 0.5777,
+      "step": 3149
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.884040117263794,
+      "learning_rate": 9.98812969841426e-06,
+      "loss": 0.6863,
+      "step": 3150
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.9739036560058594,
+      "learning_rate": 9.988111534474539e-06,
+      "loss": 0.692,
+      "step": 3151
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.786329984664917,
+      "learning_rate": 9.988093356664747e-06,
+      "loss": 0.6731,
+      "step": 3152
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.3707497119903564,
+      "learning_rate": 9.988075164984937e-06,
+      "loss": 0.7658,
+      "step": 3153
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.314815044403076,
+      "learning_rate": 9.988056959435158e-06,
+      "loss": 0.9853,
+      "step": 3154
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.1170082092285156,
+      "learning_rate": 9.98803874001546e-06,
+      "loss": 0.8369,
+      "step": 3155
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 4.90825080871582,
+      "learning_rate": 9.988020506725895e-06,
+      "loss": 0.7558,
+      "step": 3156
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.6014742851257324,
+      "learning_rate": 9.988002259566515e-06,
+      "loss": 0.6613,
+      "step": 3157
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.7362189292907715,
+      "learning_rate": 9.987983998537368e-06,
+      "loss": 0.6959,
+      "step": 3158
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.3160903453826904,
+      "learning_rate": 9.987965723638506e-06,
+      "loss": 0.6755,
+      "step": 3159
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.5562381744384766,
+      "learning_rate": 9.987947434869981e-06,
+      "loss": 0.7588,
+      "step": 3160
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.6529381275177,
+      "learning_rate": 9.987929132231842e-06,
+      "loss": 0.8229,
+      "step": 3161
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.1958651542663574,
+      "learning_rate": 9.987910815724142e-06,
+      "loss": 0.7186,
+      "step": 3162
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.9590907096862793,
+      "learning_rate": 9.987892485346929e-06,
+      "loss": 0.5706,
+      "step": 3163
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 4.1133198738098145,
+      "learning_rate": 9.987874141100256e-06,
+      "loss": 0.6155,
+      "step": 3164
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 7.903247833251953,
+      "learning_rate": 9.987855782984173e-06,
+      "loss": 0.7445,
+      "step": 3165
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.8684051036834717,
+      "learning_rate": 9.987837410998732e-06,
+      "loss": 0.5246,
+      "step": 3166
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 4.254024505615234,
+      "learning_rate": 9.987819025143983e-06,
+      "loss": 0.7696,
+      "step": 3167
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.3265318870544434,
+      "learning_rate": 9.98780062541998e-06,
+      "loss": 0.9698,
+      "step": 3168
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.376141309738159,
+      "learning_rate": 9.987782211826769e-06,
+      "loss": 0.4252,
+      "step": 3169
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.8105573654174805,
+      "learning_rate": 9.987763784364407e-06,
+      "loss": 0.4933,
+      "step": 3170
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 4.013579368591309,
+      "learning_rate": 9.98774534303294e-06,
+      "loss": 0.9545,
+      "step": 3171
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.0299489498138428,
+      "learning_rate": 9.987726887832422e-06,
+      "loss": 0.8219,
+      "step": 3172
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.203456163406372,
+      "learning_rate": 9.987708418762906e-06,
+      "loss": 0.8387,
+      "step": 3173
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.574601173400879,
+      "learning_rate": 9.987689935824439e-06,
+      "loss": 0.9128,
+      "step": 3174
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.9348626136779785,
+      "learning_rate": 9.987671439017075e-06,
+      "loss": 0.6875,
+      "step": 3175
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.5429797172546387,
+      "learning_rate": 9.987652928340866e-06,
+      "loss": 0.6929,
+      "step": 3176
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.2479138374328613,
+      "learning_rate": 9.987634403795861e-06,
+      "loss": 0.7163,
+      "step": 3177
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.9058282375335693,
+      "learning_rate": 9.987615865382114e-06,
+      "loss": 0.2864,
+      "step": 3178
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.7099831104278564,
+      "learning_rate": 9.987597313099677e-06,
+      "loss": 0.7546,
+      "step": 3179
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.8683013916015625,
+      "learning_rate": 9.987578746948599e-06,
+      "loss": 1.0019,
+      "step": 3180
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.519052743911743,
+      "learning_rate": 9.987560166928933e-06,
+      "loss": 0.5403,
+      "step": 3181
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.97070050239563,
+      "learning_rate": 9.987541573040731e-06,
+      "loss": 0.6873,
+      "step": 3182
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.926159620285034,
+      "learning_rate": 9.987522965284045e-06,
+      "loss": 0.6832,
+      "step": 3183
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.252084732055664,
+      "learning_rate": 9.987504343658925e-06,
+      "loss": 0.8719,
+      "step": 3184
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.906484603881836,
+      "learning_rate": 9.987485708165424e-06,
+      "loss": 0.6782,
+      "step": 3185
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.8668594360351562,
+      "learning_rate": 9.987467058803592e-06,
+      "loss": 0.4631,
+      "step": 3186
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.9885512590408325,
+      "learning_rate": 9.987448395573485e-06,
+      "loss": 0.7858,
+      "step": 3187
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.9904325008392334,
+      "learning_rate": 9.98742971847515e-06,
+      "loss": 0.6994,
+      "step": 3188
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.3817920684814453,
+      "learning_rate": 9.987411027508644e-06,
+      "loss": 0.5961,
+      "step": 3189
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.4873273372650146,
+      "learning_rate": 9.987392322674015e-06,
+      "loss": 0.3745,
+      "step": 3190
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 4.7754411697387695,
+      "learning_rate": 9.987373603971315e-06,
+      "loss": 0.3788,
+      "step": 3191
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.994687557220459,
+      "learning_rate": 9.987354871400599e-06,
+      "loss": 0.7827,
+      "step": 3192
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.7869505882263184,
+      "learning_rate": 9.987336124961915e-06,
+      "loss": 0.7766,
+      "step": 3193
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.0987205505371094,
+      "learning_rate": 9.98731736465532e-06,
+      "loss": 0.8675,
+      "step": 3194
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.3270463943481445,
+      "learning_rate": 9.987298590480863e-06,
+      "loss": 0.6491,
+      "step": 3195
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.3472838401794434,
+      "learning_rate": 9.987279802438598e-06,
+      "loss": 0.4336,
+      "step": 3196
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.201019048690796,
+      "learning_rate": 9.987261000528574e-06,
+      "loss": 0.8665,
+      "step": 3197
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 4.042012691497803,
+      "learning_rate": 9.987242184750847e-06,
+      "loss": 0.8468,
+      "step": 3198
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.543644428253174,
+      "learning_rate": 9.987223355105467e-06,
+      "loss": 0.6602,
+      "step": 3199
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.8565609455108643,
+      "learning_rate": 9.987204511592488e-06,
+      "loss": 0.6205,
+      "step": 3200
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 6.213814735412598,
+      "learning_rate": 9.98718565421196e-06,
+      "loss": 0.7494,
+      "step": 3201
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.057827949523926,
+      "learning_rate": 9.987166782963938e-06,
+      "loss": 0.6066,
+      "step": 3202
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.2165656089782715,
+      "learning_rate": 9.987147897848473e-06,
+      "loss": 0.6129,
+      "step": 3203
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.712508201599121,
+      "learning_rate": 9.987128998865619e-06,
+      "loss": 0.6083,
+      "step": 3204
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.059591054916382,
+      "learning_rate": 9.987110086015425e-06,
+      "loss": 0.6908,
+      "step": 3205
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.8987178802490234,
+      "learning_rate": 9.98709115929795e-06,
+      "loss": 0.6781,
+      "step": 3206
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.8319272994995117,
+      "learning_rate": 9.98707221871324e-06,
+      "loss": 0.6812,
+      "step": 3207
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.753355860710144,
+      "learning_rate": 9.98705326426135e-06,
+      "loss": 0.7426,
+      "step": 3208
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.352468490600586,
+      "learning_rate": 9.987034295942333e-06,
+      "loss": 0.8715,
+      "step": 3209
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 4.152082443237305,
+      "learning_rate": 9.987015313756244e-06,
+      "loss": 0.8152,
+      "step": 3210
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.78731369972229,
+      "learning_rate": 9.986996317703133e-06,
+      "loss": 0.6152,
+      "step": 3211
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.7681329250335693,
+      "learning_rate": 9.986977307783052e-06,
+      "loss": 0.5933,
+      "step": 3212
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.0685513019561768,
+      "learning_rate": 9.986958283996056e-06,
+      "loss": 0.7606,
+      "step": 3213
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.8713595867156982,
+      "learning_rate": 9.986939246342197e-06,
+      "loss": 0.6555,
+      "step": 3214
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.6696693897247314,
+      "learning_rate": 9.986920194821527e-06,
+      "loss": 0.5874,
+      "step": 3215
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.096205472946167,
+      "learning_rate": 9.9869011294341e-06,
+      "loss": 0.5113,
+      "step": 3216
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.5451698303222656,
+      "learning_rate": 9.986882050179972e-06,
+      "loss": 0.5319,
+      "step": 3217
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.03359055519104,
+      "learning_rate": 9.986862957059192e-06,
+      "loss": 0.4548,
+      "step": 3218
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.133364677429199,
+      "learning_rate": 9.986843850071813e-06,
+      "loss": 0.7264,
+      "step": 3219
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.216437816619873,
+      "learning_rate": 9.986824729217889e-06,
+      "loss": 0.8121,
+      "step": 3220
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.798689365386963,
+      "learning_rate": 9.986805594497475e-06,
+      "loss": 0.6083,
+      "step": 3221
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.9517624378204346,
+      "learning_rate": 9.986786445910623e-06,
+      "loss": 0.5954,
+      "step": 3222
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.0680999755859375,
+      "learning_rate": 9.986767283457386e-06,
+      "loss": 0.5618,
+      "step": 3223
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.2594096660614014,
+      "learning_rate": 9.986748107137816e-06,
+      "loss": 0.6086,
+      "step": 3224
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.281731605529785,
+      "learning_rate": 9.986728916951968e-06,
+      "loss": 0.69,
+      "step": 3225
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.496000051498413,
+      "learning_rate": 9.986709712899896e-06,
+      "loss": 0.7135,
+      "step": 3226
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 5.793482303619385,
+      "learning_rate": 9.986690494981651e-06,
+      "loss": 0.8092,
+      "step": 3227
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.699971914291382,
+      "learning_rate": 9.986671263197288e-06,
+      "loss": 0.543,
+      "step": 3228
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.007490873336792,
+      "learning_rate": 9.986652017546861e-06,
+      "loss": 0.6581,
+      "step": 3229
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.7637014389038086,
+      "learning_rate": 9.986632758030423e-06,
+      "loss": 0.7935,
+      "step": 3230
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.9089152812957764,
+      "learning_rate": 9.986613484648026e-06,
+      "loss": 0.6773,
+      "step": 3231
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.966829538345337,
+      "learning_rate": 9.986594197399726e-06,
+      "loss": 0.6625,
+      "step": 3232
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.0835351943969727,
+      "learning_rate": 9.986574896285574e-06,
+      "loss": 0.6423,
+      "step": 3233
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.2547953128814697,
+      "learning_rate": 9.986555581305629e-06,
+      "loss": 0.7721,
+      "step": 3234
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.0498902797698975,
+      "learning_rate": 9.986536252459937e-06,
+      "loss": 0.7485,
+      "step": 3235
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.792515516281128,
+      "learning_rate": 9.986516909748556e-06,
+      "loss": 0.4691,
+      "step": 3236
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 4.658959865570068,
+      "learning_rate": 9.986497553171541e-06,
+      "loss": 0.8892,
+      "step": 3237
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.8273756504058838,
+      "learning_rate": 9.986478182728944e-06,
+      "loss": 0.4076,
+      "step": 3238
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.2352852821350098,
+      "learning_rate": 9.98645879842082e-06,
+      "loss": 0.7244,
+      "step": 3239
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.577613353729248,
+      "learning_rate": 9.98643940024722e-06,
+      "loss": 0.6313,
+      "step": 3240
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.6026707887649536,
+      "learning_rate": 9.9864199882082e-06,
+      "loss": 0.7093,
+      "step": 3241
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.713475227355957,
+      "learning_rate": 9.986400562303815e-06,
+      "loss": 0.5551,
+      "step": 3242
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.2626655101776123,
+      "learning_rate": 9.986381122534118e-06,
+      "loss": 0.7303,
+      "step": 3243
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.700915575027466,
+      "learning_rate": 9.986361668899164e-06,
+      "loss": 0.6447,
+      "step": 3244
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.2679641246795654,
+      "learning_rate": 9.986342201399005e-06,
+      "loss": 0.4099,
+      "step": 3245
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 6.464822769165039,
+      "learning_rate": 9.986322720033697e-06,
+      "loss": 0.8825,
+      "step": 3246
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.0443108081817627,
+      "learning_rate": 9.986303224803292e-06,
+      "loss": 0.951,
+      "step": 3247
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.0992982387542725,
+      "learning_rate": 9.986283715707846e-06,
+      "loss": 0.6579,
+      "step": 3248
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.0233075618743896,
+      "learning_rate": 9.986264192747414e-06,
+      "loss": 0.8249,
+      "step": 3249
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.598234176635742,
+      "learning_rate": 9.986244655922048e-06,
+      "loss": 0.7985,
+      "step": 3250
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.4641757011413574,
+      "learning_rate": 9.986225105231803e-06,
+      "loss": 0.8833,
+      "step": 3251
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.2685723304748535,
+      "learning_rate": 9.986205540676737e-06,
+      "loss": 0.5963,
+      "step": 3252
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.1634535789489746,
+      "learning_rate": 9.986185962256897e-06,
+      "loss": 0.7482,
+      "step": 3253
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.8742709159851074,
+      "learning_rate": 9.986166369972346e-06,
+      "loss": 0.6369,
+      "step": 3254
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.324087142944336,
+      "learning_rate": 9.98614676382313e-06,
+      "loss": 0.7617,
+      "step": 3255
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.4881484508514404,
+      "learning_rate": 9.98612714380931e-06,
+      "loss": 0.7625,
+      "step": 3256
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.8422812223434448,
+      "learning_rate": 9.986107509930938e-06,
+      "loss": 0.529,
+      "step": 3257
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.7659270763397217,
+      "learning_rate": 9.98608786218807e-06,
+      "loss": 0.8621,
+      "step": 3258
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.9630138874053955,
+      "learning_rate": 9.986068200580758e-06,
+      "loss": 0.9609,
+      "step": 3259
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.7575013637542725,
+      "learning_rate": 9.986048525109058e-06,
+      "loss": 0.6669,
+      "step": 3260
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.7672350406646729,
+      "learning_rate": 9.986028835773025e-06,
+      "loss": 0.8431,
+      "step": 3261
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 4.323107719421387,
+      "learning_rate": 9.986009132572715e-06,
+      "loss": 0.4234,
+      "step": 3262
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.6942366361618042,
+      "learning_rate": 9.985989415508179e-06,
+      "loss": 0.6124,
+      "step": 3263
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.793155312538147,
+      "learning_rate": 9.985969684579478e-06,
+      "loss": 0.5955,
+      "step": 3264
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.8981071710586548,
+      "learning_rate": 9.98594993978666e-06,
+      "loss": 0.5766,
+      "step": 3265
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.576871633529663,
+      "learning_rate": 9.985930181129784e-06,
+      "loss": 0.6277,
+      "step": 3266
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.4029502868652344,
+      "learning_rate": 9.985910408608903e-06,
+      "loss": 0.8728,
+      "step": 3267
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.717719554901123,
+      "learning_rate": 9.985890622224072e-06,
+      "loss": 0.4982,
+      "step": 3268
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 4.1273064613342285,
+      "learning_rate": 9.98587082197535e-06,
+      "loss": 0.507,
+      "step": 3269
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.6767334938049316,
+      "learning_rate": 9.985851007862788e-06,
+      "loss": 0.5356,
+      "step": 3270
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.9583176374435425,
+      "learning_rate": 9.98583117988644e-06,
+      "loss": 0.7755,
+      "step": 3271
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.214855432510376,
+      "learning_rate": 9.985811338046365e-06,
+      "loss": 0.688,
+      "step": 3272
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.198092460632324,
+      "learning_rate": 9.985791482342618e-06,
+      "loss": 0.8365,
+      "step": 3273
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 6.211399555206299,
+      "learning_rate": 9.98577161277525e-06,
+      "loss": 0.8852,
+      "step": 3274
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.599203586578369,
+      "learning_rate": 9.985751729344319e-06,
+      "loss": 0.5624,
+      "step": 3275
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.717592239379883,
+      "learning_rate": 9.985731832049881e-06,
+      "loss": 0.4698,
+      "step": 3276
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.2492730617523193,
+      "learning_rate": 9.985711920891991e-06,
+      "loss": 0.7767,
+      "step": 3277
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.8513996601104736,
+      "learning_rate": 9.985691995870704e-06,
+      "loss": 0.7785,
+      "step": 3278
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.92121958732605,
+      "learning_rate": 9.985672056986073e-06,
+      "loss": 0.5487,
+      "step": 3279
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.9368679523468018,
+      "learning_rate": 9.985652104238157e-06,
+      "loss": 0.6417,
+      "step": 3280
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.9540303945541382,
+      "learning_rate": 9.98563213762701e-06,
+      "loss": 0.6388,
+      "step": 3281
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.9146744012832642,
+      "learning_rate": 9.985612157152688e-06,
+      "loss": 0.6259,
+      "step": 3282
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.9084689617156982,
+      "learning_rate": 9.985592162815246e-06,
+      "loss": 0.7549,
+      "step": 3283
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.9683191776275635,
+      "learning_rate": 9.98557215461474e-06,
+      "loss": 0.8496,
+      "step": 3284
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.1155824661254883,
+      "learning_rate": 9.985552132551225e-06,
+      "loss": 0.6064,
+      "step": 3285
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.522690534591675,
+      "learning_rate": 9.985532096624759e-06,
+      "loss": 0.6664,
+      "step": 3286
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.8059816360473633,
+      "learning_rate": 9.985512046835393e-06,
+      "loss": 0.7883,
+      "step": 3287
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.2218270301818848,
+      "learning_rate": 9.985491983183187e-06,
+      "loss": 0.6253,
+      "step": 3288
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.840894937515259,
+      "learning_rate": 9.985471905668197e-06,
+      "loss": 0.7934,
+      "step": 3289
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.436723470687866,
+      "learning_rate": 9.985451814290474e-06,
+      "loss": 0.8269,
+      "step": 3290
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.8268656730651855,
+      "learning_rate": 9.98543170905008e-06,
+      "loss": 0.7466,
+      "step": 3291
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.1131367683410645,
+      "learning_rate": 9.985411589947066e-06,
+      "loss": 0.6757,
+      "step": 3292
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.5474679470062256,
+      "learning_rate": 9.985391456981491e-06,
+      "loss": 0.7908,
+      "step": 3293
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 6.9083991050720215,
+      "learning_rate": 9.985371310153409e-06,
+      "loss": 0.8464,
+      "step": 3294
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.678229808807373,
+      "learning_rate": 9.985351149462879e-06,
+      "loss": 0.6962,
+      "step": 3295
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.6489622592926025,
+      "learning_rate": 9.985330974909953e-06,
+      "loss": 0.7527,
+      "step": 3296
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.797145128250122,
+      "learning_rate": 9.98531078649469e-06,
+      "loss": 0.4947,
+      "step": 3297
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.9275083541870117,
+      "learning_rate": 9.985290584217145e-06,
+      "loss": 0.6861,
+      "step": 3298
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.2464776039123535,
+      "learning_rate": 9.985270368077375e-06,
+      "loss": 0.609,
+      "step": 3299
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.003035545349121,
+      "learning_rate": 9.985250138075436e-06,
+      "loss": 0.7658,
+      "step": 3300
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.117811918258667,
+      "learning_rate": 9.985229894211381e-06,
+      "loss": 0.7445,
+      "step": 3301
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.875149965286255,
+      "learning_rate": 9.985209636485271e-06,
+      "loss": 0.7814,
+      "step": 3302
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 12.891106605529785,
+      "learning_rate": 9.985189364897161e-06,
+      "loss": 0.6204,
+      "step": 3303
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.684260129928589,
+      "learning_rate": 9.985169079447107e-06,
+      "loss": 0.6225,
+      "step": 3304
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.8384792804718018,
+      "learning_rate": 9.985148780135164e-06,
+      "loss": 0.7049,
+      "step": 3305
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.7271790504455566,
+      "learning_rate": 9.985128466961392e-06,
+      "loss": 0.7522,
+      "step": 3306
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.091621160507202,
+      "learning_rate": 9.985108139925844e-06,
+      "loss": 0.6431,
+      "step": 3307
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 4.352491855621338,
+      "learning_rate": 9.985087799028576e-06,
+      "loss": 0.7032,
+      "step": 3308
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.3549299240112305,
+      "learning_rate": 9.985067444269647e-06,
+      "loss": 0.6795,
+      "step": 3309
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.9134936332702637,
+      "learning_rate": 9.985047075649114e-06,
+      "loss": 0.7049,
+      "step": 3310
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.390875816345215,
+      "learning_rate": 9.985026693167032e-06,
+      "loss": 0.526,
+      "step": 3311
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 4.033547878265381,
+      "learning_rate": 9.985006296823458e-06,
+      "loss": 1.1253,
+      "step": 3312
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.6885135173797607,
+      "learning_rate": 9.98498588661845e-06,
+      "loss": 0.7652,
+      "step": 3313
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 5.069367408752441,
+      "learning_rate": 9.984965462552063e-06,
+      "loss": 0.6391,
+      "step": 3314
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.1515350341796875,
+      "learning_rate": 9.984945024624355e-06,
+      "loss": 0.9606,
+      "step": 3315
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.3633527755737305,
+      "learning_rate": 9.98492457283538e-06,
+      "loss": 0.7218,
+      "step": 3316
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.5144333839416504,
+      "learning_rate": 9.9849041071852e-06,
+      "loss": 0.4609,
+      "step": 3317
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.5305771827697754,
+      "learning_rate": 9.984883627673867e-06,
+      "loss": 0.4731,
+      "step": 3318
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.430243492126465,
+      "learning_rate": 9.984863134301441e-06,
+      "loss": 0.8164,
+      "step": 3319
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.9094847440719604,
+      "learning_rate": 9.984842627067979e-06,
+      "loss": 0.7588,
+      "step": 3320
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.913540840148926,
+      "learning_rate": 9.984822105973535e-06,
+      "loss": 0.5556,
+      "step": 3321
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 5.247519493103027,
+      "learning_rate": 9.98480157101817e-06,
+      "loss": 0.8025,
+      "step": 3322
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.51750111579895,
+      "learning_rate": 9.984781022201937e-06,
+      "loss": 0.6713,
+      "step": 3323
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.20505952835083,
+      "learning_rate": 9.984760459524896e-06,
+      "loss": 0.7576,
+      "step": 3324
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.604830026626587,
+      "learning_rate": 9.984739882987105e-06,
+      "loss": 0.5617,
+      "step": 3325
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.720501661300659,
+      "learning_rate": 9.984719292588619e-06,
+      "loss": 0.6215,
+      "step": 3326
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.005876302719116,
+      "learning_rate": 9.984698688329496e-06,
+      "loss": 0.8275,
+      "step": 3327
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.9100604057312012,
+      "learning_rate": 9.984678070209793e-06,
+      "loss": 0.506,
+      "step": 3328
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.649533748626709,
+      "learning_rate": 9.984657438229567e-06,
+      "loss": 0.475,
+      "step": 3329
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.4277875423431396,
+      "learning_rate": 9.984636792388876e-06,
+      "loss": 0.4393,
+      "step": 3330
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.457655429840088,
+      "learning_rate": 9.984616132687777e-06,
+      "loss": 0.5877,
+      "step": 3331
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.768907070159912,
+      "learning_rate": 9.984595459126329e-06,
+      "loss": 0.5217,
+      "step": 3332
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.7635234594345093,
+      "learning_rate": 9.984574771704588e-06,
+      "loss": 0.4382,
+      "step": 3333
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 4.25911283493042,
+      "learning_rate": 9.984554070422612e-06,
+      "loss": 0.7635,
+      "step": 3334
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.303863048553467,
+      "learning_rate": 9.98453335528046e-06,
+      "loss": 0.8539,
+      "step": 3335
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.610388994216919,
+      "learning_rate": 9.984512626278184e-06,
+      "loss": 0.5414,
+      "step": 3336
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.0839946269989014,
+      "learning_rate": 9.984491883415848e-06,
+      "loss": 0.7436,
+      "step": 3337
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 342.8569030761719,
+      "learning_rate": 9.984471126693506e-06,
+      "loss": 0.6736,
+      "step": 3338
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.3918912410736084,
+      "learning_rate": 9.984450356111219e-06,
+      "loss": 0.5694,
+      "step": 3339
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.2574024200439453,
+      "learning_rate": 9.984429571669043e-06,
+      "loss": 0.9208,
+      "step": 3340
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.3346495628356934,
+      "learning_rate": 9.984408773367033e-06,
+      "loss": 0.6554,
+      "step": 3341
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 7.016981601715088,
+      "learning_rate": 9.98438796120525e-06,
+      "loss": 0.8118,
+      "step": 3342
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 4.068885326385498,
+      "learning_rate": 9.984367135183753e-06,
+      "loss": 0.9419,
+      "step": 3343
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 4.273824214935303,
+      "learning_rate": 9.984346295302598e-06,
+      "loss": 0.9111,
+      "step": 3344
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.1321280002593994,
+      "learning_rate": 9.984325441561842e-06,
+      "loss": 0.7681,
+      "step": 3345
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.3607325553894043,
+      "learning_rate": 9.984304573961546e-06,
+      "loss": 0.7522,
+      "step": 3346
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.7311952114105225,
+      "learning_rate": 9.984283692501765e-06,
+      "loss": 0.7205,
+      "step": 3347
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.7653379440307617,
+      "learning_rate": 9.984262797182558e-06,
+      "loss": 0.488,
+      "step": 3348
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.9595682621002197,
+      "learning_rate": 9.984241888003982e-06,
+      "loss": 0.5139,
+      "step": 3349
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 4.062949180603027,
+      "learning_rate": 9.984220964966098e-06,
+      "loss": 0.6686,
+      "step": 3350
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.6199963092803955,
+      "learning_rate": 9.984200028068963e-06,
+      "loss": 0.8008,
+      "step": 3351
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.600680351257324,
+      "learning_rate": 9.984179077312632e-06,
+      "loss": 0.7241,
+      "step": 3352
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.246264934539795,
+      "learning_rate": 9.984158112697168e-06,
+      "loss": 0.8065,
+      "step": 3353
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 4.358322620391846,
+      "learning_rate": 9.984137134222629e-06,
+      "loss": 0.6189,
+      "step": 3354
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.3732712268829346,
+      "learning_rate": 9.984116141889071e-06,
+      "loss": 0.4319,
+      "step": 3355
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.652158260345459,
+      "learning_rate": 9.984095135696551e-06,
+      "loss": 0.6526,
+      "step": 3356
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.8789558410644531,
+      "learning_rate": 9.984074115645131e-06,
+      "loss": 0.741,
+      "step": 3357
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.590014696121216,
+      "learning_rate": 9.984053081734867e-06,
+      "loss": 0.6925,
+      "step": 3358
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.57784366607666,
+      "learning_rate": 9.98403203396582e-06,
+      "loss": 0.5469,
+      "step": 3359
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.2204246520996094,
+      "learning_rate": 9.984010972338047e-06,
+      "loss": 0.798,
+      "step": 3360
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.0572478771209717,
+      "learning_rate": 9.983989896851606e-06,
+      "loss": 0.4814,
+      "step": 3361
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.106677293777466,
+      "learning_rate": 9.983968807506557e-06,
+      "loss": 0.6975,
+      "step": 3362
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.715909242630005,
+      "learning_rate": 9.983947704302957e-06,
+      "loss": 0.8348,
+      "step": 3363
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 9.343684196472168,
+      "learning_rate": 9.983926587240863e-06,
+      "loss": 0.58,
+      "step": 3364
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.1319382190704346,
+      "learning_rate": 9.98390545632034e-06,
+      "loss": 0.6053,
+      "step": 3365
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.8428609371185303,
+      "learning_rate": 9.983884311541441e-06,
+      "loss": 0.8442,
+      "step": 3366
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 5.0500006675720215,
+      "learning_rate": 9.983863152904228e-06,
+      "loss": 0.6472,
+      "step": 3367
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.044254779815674,
+      "learning_rate": 9.983841980408757e-06,
+      "loss": 0.5118,
+      "step": 3368
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.3447256088256836,
+      "learning_rate": 9.98382079405509e-06,
+      "loss": 0.8581,
+      "step": 3369
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.3378331661224365,
+      "learning_rate": 9.983799593843281e-06,
+      "loss": 0.4825,
+      "step": 3370
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.0938560962677,
+      "learning_rate": 9.983778379773397e-06,
+      "loss": 0.8237,
+      "step": 3371
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.088860034942627,
+      "learning_rate": 9.98375715184549e-06,
+      "loss": 0.8389,
+      "step": 3372
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.78253436088562,
+      "learning_rate": 9.983735910059622e-06,
+      "loss": 0.8607,
+      "step": 3373
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.9484955072402954,
+      "learning_rate": 9.98371465441585e-06,
+      "loss": 0.5789,
+      "step": 3374
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.1790878772735596,
+      "learning_rate": 9.983693384914234e-06,
+      "loss": 0.6619,
+      "step": 3375
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.0953683853149414,
+      "learning_rate": 9.983672101554835e-06,
+      "loss": 0.7212,
+      "step": 3376
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.0747036933898926,
+      "learning_rate": 9.983650804337712e-06,
+      "loss": 0.7151,
+      "step": 3377
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.688206911087036,
+      "learning_rate": 9.983629493262921e-06,
+      "loss": 0.5566,
+      "step": 3378
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.039431095123291,
+      "learning_rate": 9.983608168330525e-06,
+      "loss": 0.5179,
+      "step": 3379
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.3227124214172363,
+      "learning_rate": 9.98358682954058e-06,
+      "loss": 0.5172,
+      "step": 3380
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.093971014022827,
+      "learning_rate": 9.983565476893147e-06,
+      "loss": 0.5848,
+      "step": 3381
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 4.142402648925781,
+      "learning_rate": 9.983544110388287e-06,
+      "loss": 0.8091,
+      "step": 3382
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.37668514251709,
+      "learning_rate": 9.983522730026056e-06,
+      "loss": 0.7951,
+      "step": 3383
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.5374679565429688,
+      "learning_rate": 9.983501335806516e-06,
+      "loss": 0.6848,
+      "step": 3384
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 4.980534553527832,
+      "learning_rate": 9.983479927729726e-06,
+      "loss": 0.5557,
+      "step": 3385
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.9783973693847656,
+      "learning_rate": 9.983458505795745e-06,
+      "loss": 0.5549,
+      "step": 3386
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 4.714261054992676,
+      "learning_rate": 9.983437070004632e-06,
+      "loss": 0.7672,
+      "step": 3387
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.5693461894989014,
+      "learning_rate": 9.983415620356447e-06,
+      "loss": 0.5547,
+      "step": 3388
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.861042022705078,
+      "learning_rate": 9.98339415685125e-06,
+      "loss": 0.9952,
+      "step": 3389
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.0691230297088623,
+      "learning_rate": 9.983372679489102e-06,
+      "loss": 0.7215,
+      "step": 3390
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.232139825820923,
+      "learning_rate": 9.98335118827006e-06,
+      "loss": 0.5436,
+      "step": 3391
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.680415630340576,
+      "learning_rate": 9.983329683194187e-06,
+      "loss": 0.6661,
+      "step": 3392
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.2963900566101074,
+      "learning_rate": 9.983308164261539e-06,
+      "loss": 0.7293,
+      "step": 3393
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.0748422145843506,
+      "learning_rate": 9.98328663147218e-06,
+      "loss": 0.4467,
+      "step": 3394
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.7246487140655518,
+      "learning_rate": 9.983265084826166e-06,
+      "loss": 0.7955,
+      "step": 3395
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.831449031829834,
+      "learning_rate": 9.983243524323558e-06,
+      "loss": 0.6366,
+      "step": 3396
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 10.131364822387695,
+      "learning_rate": 9.983221949964417e-06,
+      "loss": 0.6894,
+      "step": 3397
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.5181221961975098,
+      "learning_rate": 9.983200361748805e-06,
+      "loss": 0.6569,
+      "step": 3398
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.038588523864746,
+      "learning_rate": 9.983178759676775e-06,
+      "loss": 0.6577,
+      "step": 3399
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.3070175647735596,
+      "learning_rate": 9.983157143748395e-06,
+      "loss": 0.7272,
+      "step": 3400
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.8280680179595947,
+      "learning_rate": 9.983135513963719e-06,
+      "loss": 0.6582,
+      "step": 3401
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.575939416885376,
+      "learning_rate": 9.983113870322814e-06,
+      "loss": 0.7014,
+      "step": 3402
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.067483425140381,
+      "learning_rate": 9.983092212825733e-06,
+      "loss": 0.7242,
+      "step": 3403
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.4595186710357666,
+      "learning_rate": 9.98307054147254e-06,
+      "loss": 0.7784,
+      "step": 3404
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.9480671882629395,
+      "learning_rate": 9.983048856263293e-06,
+      "loss": 0.6328,
+      "step": 3405
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.3372104167938232,
+      "learning_rate": 9.983027157198056e-06,
+      "loss": 0.6411,
+      "step": 3406
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.0310981273651123,
+      "learning_rate": 9.983005444276886e-06,
+      "loss": 0.6185,
+      "step": 3407
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.2166032791137695,
+      "learning_rate": 9.982983717499845e-06,
+      "loss": 0.6912,
+      "step": 3408
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.1295347213745117,
+      "learning_rate": 9.982961976866993e-06,
+      "loss": 0.6597,
+      "step": 3409
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.2018139362335205,
+      "learning_rate": 9.98294022237839e-06,
+      "loss": 0.7266,
+      "step": 3410
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.7537981271743774,
+      "learning_rate": 9.982918454034098e-06,
+      "loss": 0.7899,
+      "step": 3411
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.973320722579956,
+      "learning_rate": 9.982896671834175e-06,
+      "loss": 0.3917,
+      "step": 3412
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 4.737184047698975,
+      "learning_rate": 9.982874875778684e-06,
+      "loss": 0.5964,
+      "step": 3413
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 4.1899333000183105,
+      "learning_rate": 9.982853065867683e-06,
+      "loss": 0.3624,
+      "step": 3414
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.170452356338501,
+      "learning_rate": 9.982831242101236e-06,
+      "loss": 0.6315,
+      "step": 3415
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.2382423877716064,
+      "learning_rate": 9.982809404479401e-06,
+      "loss": 0.5883,
+      "step": 3416
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.7378652095794678,
+      "learning_rate": 9.98278755300224e-06,
+      "loss": 0.8889,
+      "step": 3417
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 4.339071750640869,
+      "learning_rate": 9.982765687669813e-06,
+      "loss": 0.8253,
+      "step": 3418
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.002535343170166,
+      "learning_rate": 9.982743808482182e-06,
+      "loss": 0.4768,
+      "step": 3419
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.576392412185669,
+      "learning_rate": 9.982721915439409e-06,
+      "loss": 0.6572,
+      "step": 3420
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.3850650787353516,
+      "learning_rate": 9.98270000854155e-06,
+      "loss": 0.7985,
+      "step": 3421
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.9330695867538452,
+      "learning_rate": 9.982678087788668e-06,
+      "loss": 0.5428,
+      "step": 3422
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.6583309173583984,
+      "learning_rate": 9.982656153180828e-06,
+      "loss": 0.6485,
+      "step": 3423
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.0271801948547363,
+      "learning_rate": 9.982634204718085e-06,
+      "loss": 0.5919,
+      "step": 3424
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.52197527885437,
+      "learning_rate": 9.982612242400505e-06,
+      "loss": 0.7436,
+      "step": 3425
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.8082964420318604,
+      "learning_rate": 9.982590266228145e-06,
+      "loss": 0.6464,
+      "step": 3426
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.0719752311706543,
+      "learning_rate": 9.982568276201068e-06,
+      "loss": 0.497,
+      "step": 3427
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.9891061782836914,
+      "learning_rate": 9.982546272319336e-06,
+      "loss": 0.535,
+      "step": 3428
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 13.787482261657715,
+      "learning_rate": 9.98252425458301e-06,
+      "loss": 0.7736,
+      "step": 3429
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.972060203552246,
+      "learning_rate": 9.98250222299215e-06,
+      "loss": 0.7453,
+      "step": 3430
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 6.102320671081543,
+      "learning_rate": 9.982480177546817e-06,
+      "loss": 0.8846,
+      "step": 3431
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.3937196731567383,
+      "learning_rate": 9.982458118247073e-06,
+      "loss": 0.4022,
+      "step": 3432
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.3296308517456055,
+      "learning_rate": 9.982436045092981e-06,
+      "loss": 0.6807,
+      "step": 3433
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.250089168548584,
+      "learning_rate": 9.9824139580846e-06,
+      "loss": 0.7706,
+      "step": 3434
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.936387777328491,
+      "learning_rate": 9.982391857221991e-06,
+      "loss": 0.5984,
+      "step": 3435
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.5393145084381104,
+      "learning_rate": 9.982369742505219e-06,
+      "loss": 0.8165,
+      "step": 3436
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.603873610496521,
+      "learning_rate": 9.982347613934342e-06,
+      "loss": 0.722,
+      "step": 3437
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.940314769744873,
+      "learning_rate": 9.982325471509422e-06,
+      "loss": 0.8595,
+      "step": 3438
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.345637083053589,
+      "learning_rate": 9.982303315230522e-06,
+      "loss": 0.6416,
+      "step": 3439
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.788813591003418,
+      "learning_rate": 9.982281145097703e-06,
+      "loss": 0.5647,
+      "step": 3440
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.945655584335327,
+      "learning_rate": 9.982258961111027e-06,
+      "loss": 0.9777,
+      "step": 3441
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.751065731048584,
+      "learning_rate": 9.982236763270555e-06,
+      "loss": 0.8118,
+      "step": 3442
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.8954765796661377,
+      "learning_rate": 9.982214551576348e-06,
+      "loss": 0.8282,
+      "step": 3443
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 31.95616340637207,
+      "learning_rate": 9.982192326028467e-06,
+      "loss": 0.5719,
+      "step": 3444
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.8944883346557617,
+      "learning_rate": 9.98217008662698e-06,
+      "loss": 0.7339,
+      "step": 3445
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.894197463989258,
+      "learning_rate": 9.982147833371943e-06,
+      "loss": 0.8218,
+      "step": 3446
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.3261632919311523,
+      "learning_rate": 9.982125566263418e-06,
+      "loss": 0.5312,
+      "step": 3447
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.161005735397339,
+      "learning_rate": 9.982103285301468e-06,
+      "loss": 0.8592,
+      "step": 3448
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 4.0191450119018555,
+      "learning_rate": 9.982080990486157e-06,
+      "loss": 0.6708,
+      "step": 3449
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.1468183994293213,
+      "learning_rate": 9.982058681817542e-06,
+      "loss": 0.65,
+      "step": 3450
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.195082664489746,
+      "learning_rate": 9.98203635929569e-06,
+      "loss": 0.7279,
+      "step": 3451
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.2992918491363525,
+      "learning_rate": 9.982014022920661e-06,
+      "loss": 0.456,
+      "step": 3452
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.6474735736846924,
+      "learning_rate": 9.981991672692517e-06,
+      "loss": 0.8729,
+      "step": 3453
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.994572639465332,
+      "learning_rate": 9.981969308611322e-06,
+      "loss": 0.5132,
+      "step": 3454
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.5059516429901123,
+      "learning_rate": 9.981946930677135e-06,
+      "loss": 0.6963,
+      "step": 3455
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.3024587631225586,
+      "learning_rate": 9.98192453889002e-06,
+      "loss": 0.7173,
+      "step": 3456
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 4.882111549377441,
+      "learning_rate": 9.98190213325004e-06,
+      "loss": 0.5794,
+      "step": 3457
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.1067705154418945,
+      "learning_rate": 9.981879713757257e-06,
+      "loss": 0.8767,
+      "step": 3458
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.517571210861206,
+      "learning_rate": 9.98185728041173e-06,
+      "loss": 0.7029,
+      "step": 3459
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.9745192527770996,
+      "learning_rate": 9.981834833213526e-06,
+      "loss": 0.9054,
+      "step": 3460
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.542639970779419,
+      "learning_rate": 9.981812372162705e-06,
+      "loss": 0.7248,
+      "step": 3461
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.8860034942626953,
+      "learning_rate": 9.981789897259331e-06,
+      "loss": 0.7188,
+      "step": 3462
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.324209690093994,
+      "learning_rate": 9.981767408503465e-06,
+      "loss": 0.7043,
+      "step": 3463
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.9265947341918945,
+      "learning_rate": 9.98174490589517e-06,
+      "loss": 0.552,
+      "step": 3464
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.2542505264282227,
+      "learning_rate": 9.981722389434509e-06,
+      "loss": 0.5996,
+      "step": 3465
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.5412585735321045,
+      "learning_rate": 9.981699859121543e-06,
+      "loss": 0.5808,
+      "step": 3466
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.0973236560821533,
+      "learning_rate": 9.981677314956337e-06,
+      "loss": 0.4909,
+      "step": 3467
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.266613483428955,
+      "learning_rate": 9.981654756938952e-06,
+      "loss": 0.7287,
+      "step": 3468
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.2474164962768555,
+      "learning_rate": 9.981632185069452e-06,
+      "loss": 0.5594,
+      "step": 3469
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.823833703994751,
+      "learning_rate": 9.981609599347899e-06,
+      "loss": 0.6503,
+      "step": 3470
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.5020058155059814,
+      "learning_rate": 9.981586999774356e-06,
+      "loss": 0.6021,
+      "step": 3471
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.2294819355010986,
+      "learning_rate": 9.981564386348886e-06,
+      "loss": 0.6551,
+      "step": 3472
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.958102226257324,
+      "learning_rate": 9.981541759071548e-06,
+      "loss": 0.5716,
+      "step": 3473
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.792191505432129,
+      "learning_rate": 9.981519117942414e-06,
+      "loss": 0.5214,
+      "step": 3474
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.888420343399048,
+      "learning_rate": 9.981496462961538e-06,
+      "loss": 0.4682,
+      "step": 3475
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 4.397500514984131,
+      "learning_rate": 9.981473794128986e-06,
+      "loss": 0.8664,
+      "step": 3476
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.1609058380126953,
+      "learning_rate": 9.981451111444825e-06,
+      "loss": 0.6735,
+      "step": 3477
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.215841770172119,
+      "learning_rate": 9.981428414909112e-06,
+      "loss": 0.6522,
+      "step": 3478
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.840550661087036,
+      "learning_rate": 9.981405704521911e-06,
+      "loss": 0.6616,
+      "step": 3479
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.6874607801437378,
+      "learning_rate": 9.98138298028329e-06,
+      "loss": 0.6972,
+      "step": 3480
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.1558656692504883,
+      "learning_rate": 9.981360242193308e-06,
+      "loss": 0.8269,
+      "step": 3481
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.285309314727783,
+      "learning_rate": 9.981337490252028e-06,
+      "loss": 0.9189,
+      "step": 3482
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.4036316871643066,
+      "learning_rate": 9.981314724459515e-06,
+      "loss": 0.7089,
+      "step": 3483
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.067988634109497,
+      "learning_rate": 9.981291944815833e-06,
+      "loss": 0.7621,
+      "step": 3484
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.9156025648117065,
+      "learning_rate": 9.981269151321042e-06,
+      "loss": 0.715,
+      "step": 3485
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.8201165199279785,
+      "learning_rate": 9.981246343975209e-06,
+      "loss": 0.5687,
+      "step": 3486
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.569707155227661,
+      "learning_rate": 9.981223522778396e-06,
+      "loss": 0.6504,
+      "step": 3487
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.9533228874206543,
+      "learning_rate": 9.981200687730667e-06,
+      "loss": 0.8503,
+      "step": 3488
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.254316806793213,
+      "learning_rate": 9.981177838832083e-06,
+      "loss": 0.7508,
+      "step": 3489
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.0339090824127197,
+      "learning_rate": 9.981154976082712e-06,
+      "loss": 0.8446,
+      "step": 3490
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.0137248039245605,
+      "learning_rate": 9.981132099482613e-06,
+      "loss": 0.9507,
+      "step": 3491
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.250774383544922,
+      "learning_rate": 9.981109209031853e-06,
+      "loss": 0.5466,
+      "step": 3492
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.122335195541382,
+      "learning_rate": 9.981086304730493e-06,
+      "loss": 0.7197,
+      "step": 3493
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.8012499809265137,
+      "learning_rate": 9.981063386578597e-06,
+      "loss": 0.7494,
+      "step": 3494
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.342252731323242,
+      "learning_rate": 9.981040454576233e-06,
+      "loss": 0.6622,
+      "step": 3495
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.345247745513916,
+      "learning_rate": 9.981017508723458e-06,
+      "loss": 0.6416,
+      "step": 3496
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.8663456439971924,
+      "learning_rate": 9.98099454902034e-06,
+      "loss": 0.8594,
+      "step": 3497
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.9397547245025635,
+      "learning_rate": 9.980971575466943e-06,
+      "loss": 0.6669,
+      "step": 3498
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.121931552886963,
+      "learning_rate": 9.980948588063331e-06,
+      "loss": 0.6001,
+      "step": 3499
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.8701441287994385,
+      "learning_rate": 9.980925586809565e-06,
+      "loss": 0.8887,
+      "step": 3500
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.450753688812256,
+      "learning_rate": 9.980902571705711e-06,
+      "loss": 0.6408,
+      "step": 3501
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.0144124031066895,
+      "learning_rate": 9.980879542751833e-06,
+      "loss": 0.6102,
+      "step": 3502
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.460660696029663,
+      "learning_rate": 9.980856499947994e-06,
+      "loss": 0.6649,
+      "step": 3503
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.5337107181549072,
+      "learning_rate": 9.98083344329426e-06,
+      "loss": 0.7603,
+      "step": 3504
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.253499746322632,
+      "learning_rate": 9.980810372790694e-06,
+      "loss": 0.5092,
+      "step": 3505
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.507361650466919,
+      "learning_rate": 9.980787288437362e-06,
+      "loss": 1.0672,
+      "step": 3506
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.310314178466797,
+      "learning_rate": 9.980764190234324e-06,
+      "loss": 0.5413,
+      "step": 3507
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.062837600708008,
+      "learning_rate": 9.980741078181649e-06,
+      "loss": 0.7238,
+      "step": 3508
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.071605682373047,
+      "learning_rate": 9.980717952279395e-06,
+      "loss": 0.9977,
+      "step": 3509
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.786724090576172,
+      "learning_rate": 9.980694812527633e-06,
+      "loss": 0.6852,
+      "step": 3510
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.238255262374878,
+      "learning_rate": 9.980671658926425e-06,
+      "loss": 0.6803,
+      "step": 3511
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.6585047245025635,
+      "learning_rate": 9.980648491475833e-06,
+      "loss": 0.6592,
+      "step": 3512
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.0655291080474854,
+      "learning_rate": 9.980625310175924e-06,
+      "loss": 0.4604,
+      "step": 3513
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.8297321796417236,
+      "learning_rate": 9.980602115026764e-06,
+      "loss": 0.4905,
+      "step": 3514
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.244433879852295,
+      "learning_rate": 9.980578906028411e-06,
+      "loss": 0.5977,
+      "step": 3515
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.073072910308838,
+      "learning_rate": 9.980555683180936e-06,
+      "loss": 0.5896,
+      "step": 3516
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.423395872116089,
+      "learning_rate": 9.980532446484403e-06,
+      "loss": 0.8223,
+      "step": 3517
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.0884714126586914,
+      "learning_rate": 9.980509195938873e-06,
+      "loss": 0.7558,
+      "step": 3518
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.832733154296875,
+      "learning_rate": 9.980485931544414e-06,
+      "loss": 1.0007,
+      "step": 3519
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 4.3397321701049805,
+      "learning_rate": 9.98046265330109e-06,
+      "loss": 0.7336,
+      "step": 3520
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 8.131771087646484,
+      "learning_rate": 9.980439361208962e-06,
+      "loss": 0.6475,
+      "step": 3521
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.6323654651641846,
+      "learning_rate": 9.980416055268099e-06,
+      "loss": 0.6176,
+      "step": 3522
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.4670615196228027,
+      "learning_rate": 9.980392735478566e-06,
+      "loss": 0.6808,
+      "step": 3523
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.5311710834503174,
+      "learning_rate": 9.980369401840425e-06,
+      "loss": 0.651,
+      "step": 3524
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.3096048831939697,
+      "learning_rate": 9.980346054353744e-06,
+      "loss": 0.8875,
+      "step": 3525
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.2419919967651367,
+      "learning_rate": 9.980322693018585e-06,
+      "loss": 0.644,
+      "step": 3526
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.112290382385254,
+      "learning_rate": 9.980299317835014e-06,
+      "loss": 0.4026,
+      "step": 3527
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.417490005493164,
+      "learning_rate": 9.980275928803097e-06,
+      "loss": 0.6464,
+      "step": 3528
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.828677177429199,
+      "learning_rate": 9.980252525922898e-06,
+      "loss": 0.5525,
+      "step": 3529
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.291064739227295,
+      "learning_rate": 9.980229109194484e-06,
+      "loss": 0.7057,
+      "step": 3530
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.240563154220581,
+      "learning_rate": 9.980205678617915e-06,
+      "loss": 0.6375,
+      "step": 3531
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.4357492923736572,
+      "learning_rate": 9.980182234193264e-06,
+      "loss": 0.478,
+      "step": 3532
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.572381019592285,
+      "learning_rate": 9.980158775920587e-06,
+      "loss": 0.7814,
+      "step": 3533
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.4386966228485107,
+      "learning_rate": 9.980135303799958e-06,
+      "loss": 0.8709,
+      "step": 3534
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.7413471937179565,
+      "learning_rate": 9.980111817831437e-06,
+      "loss": 0.8243,
+      "step": 3535
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.7799594402313232,
+      "learning_rate": 9.980088318015091e-06,
+      "loss": 0.681,
+      "step": 3536
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.5409812927246094,
+      "learning_rate": 9.980064804350984e-06,
+      "loss": 0.7655,
+      "step": 3537
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.84238600730896,
+      "learning_rate": 9.980041276839183e-06,
+      "loss": 0.5871,
+      "step": 3538
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.659134864807129,
+      "learning_rate": 9.980017735479752e-06,
+      "loss": 0.6829,
+      "step": 3539
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.4547297954559326,
+      "learning_rate": 9.979994180272758e-06,
+      "loss": 0.8127,
+      "step": 3540
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.7372876405715942,
+      "learning_rate": 9.979970611218266e-06,
+      "loss": 0.753,
+      "step": 3541
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.3938207626342773,
+      "learning_rate": 9.97994702831634e-06,
+      "loss": 0.6174,
+      "step": 3542
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.189692258834839,
+      "learning_rate": 9.979923431567048e-06,
+      "loss": 0.7839,
+      "step": 3543
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.2269346714019775,
+      "learning_rate": 9.979899820970453e-06,
+      "loss": 0.8761,
+      "step": 3544
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.1041016578674316,
+      "learning_rate": 9.979876196526625e-06,
+      "loss": 0.5403,
+      "step": 3545
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.5612576007843018,
+      "learning_rate": 9.979852558235625e-06,
+      "loss": 0.6044,
+      "step": 3546
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.1712605953216553,
+      "learning_rate": 9.979828906097519e-06,
+      "loss": 0.7665,
+      "step": 3547
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.3986270427703857,
+      "learning_rate": 9.979805240112378e-06,
+      "loss": 0.6865,
+      "step": 3548
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.6519455909729004,
+      "learning_rate": 9.979781560280262e-06,
+      "loss": 0.6727,
+      "step": 3549
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 5.15714693069458,
+      "learning_rate": 9.97975786660124e-06,
+      "loss": 0.7668,
+      "step": 3550
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.8090579509735107,
+      "learning_rate": 9.979734159075374e-06,
+      "loss": 0.8357,
+      "step": 3551
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.1142823696136475,
+      "learning_rate": 9.979710437702735e-06,
+      "loss": 0.5102,
+      "step": 3552
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.4036307334899902,
+      "learning_rate": 9.979686702483385e-06,
+      "loss": 0.8644,
+      "step": 3553
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.0286645889282227,
+      "learning_rate": 9.979662953417393e-06,
+      "loss": 0.7129,
+      "step": 3554
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.6456925868988037,
+      "learning_rate": 9.979639190504823e-06,
+      "loss": 0.6389,
+      "step": 3555
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.7423763275146484,
+      "learning_rate": 9.979615413745745e-06,
+      "loss": 0.5031,
+      "step": 3556
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.9926793575286865,
+      "learning_rate": 9.979591623140219e-06,
+      "loss": 0.8793,
+      "step": 3557
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.424612045288086,
+      "learning_rate": 9.979567818688314e-06,
+      "loss": 0.5947,
+      "step": 3558
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.2286839485168457,
+      "learning_rate": 9.979544000390096e-06,
+      "loss": 0.837,
+      "step": 3559
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.863800525665283,
+      "learning_rate": 9.979520168245634e-06,
+      "loss": 0.4956,
+      "step": 3560
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.7116200923919678,
+      "learning_rate": 9.979496322254988e-06,
+      "loss": 0.6333,
+      "step": 3561
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.8033348321914673,
+      "learning_rate": 9.979472462418232e-06,
+      "loss": 0.5884,
+      "step": 3562
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.4497928619384766,
+      "learning_rate": 9.979448588735426e-06,
+      "loss": 0.7448,
+      "step": 3563
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.47465443611145,
+      "learning_rate": 9.979424701206639e-06,
+      "loss": 0.7225,
+      "step": 3564
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 4.87502384185791,
+      "learning_rate": 9.97940079983194e-06,
+      "loss": 0.9011,
+      "step": 3565
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.1782925128936768,
+      "learning_rate": 9.979376884611389e-06,
+      "loss": 0.8149,
+      "step": 3566
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.6329140663146973,
+      "learning_rate": 9.979352955545057e-06,
+      "loss": 0.7959,
+      "step": 3567
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.105551242828369,
+      "learning_rate": 9.97932901263301e-06,
+      "loss": 0.4477,
+      "step": 3568
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 4.23583459854126,
+      "learning_rate": 9.979305055875316e-06,
+      "loss": 1.0173,
+      "step": 3569
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.6194695234298706,
+      "learning_rate": 9.97928108527204e-06,
+      "loss": 0.6937,
+      "step": 3570
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.5802161693573,
+      "learning_rate": 9.979257100823245e-06,
+      "loss": 0.7421,
+      "step": 3571
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 4.299648284912109,
+      "learning_rate": 9.979233102529004e-06,
+      "loss": 0.9568,
+      "step": 3572
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.2298734188079834,
+      "learning_rate": 9.979209090389382e-06,
+      "loss": 0.6108,
+      "step": 3573
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 4.644710063934326,
+      "learning_rate": 9.979185064404441e-06,
+      "loss": 0.6086,
+      "step": 3574
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.357924699783325,
+      "learning_rate": 9.979161024574255e-06,
+      "loss": 0.5849,
+      "step": 3575
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 5.88523530960083,
+      "learning_rate": 9.979136970898886e-06,
+      "loss": 0.5745,
+      "step": 3576
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.6590576171875,
+      "learning_rate": 9.979112903378402e-06,
+      "loss": 0.6778,
+      "step": 3577
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.9564272165298462,
+      "learning_rate": 9.979088822012871e-06,
+      "loss": 0.6872,
+      "step": 3578
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.187812328338623,
+      "learning_rate": 9.979064726802359e-06,
+      "loss": 0.7262,
+      "step": 3579
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.0349984169006348,
+      "learning_rate": 9.979040617746934e-06,
+      "loss": 0.5326,
+      "step": 3580
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.766411781311035,
+      "learning_rate": 9.97901649484666e-06,
+      "loss": 0.6896,
+      "step": 3581
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.060682773590088,
+      "learning_rate": 9.978992358101608e-06,
+      "loss": 0.5389,
+      "step": 3582
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 136.82875061035156,
+      "learning_rate": 9.978968207511843e-06,
+      "loss": 0.5497,
+      "step": 3583
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.2821266651153564,
+      "learning_rate": 9.978944043077435e-06,
+      "loss": 0.5035,
+      "step": 3584
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 18.555753707885742,
+      "learning_rate": 9.978919864798446e-06,
+      "loss": 0.7099,
+      "step": 3585
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.021468162536621,
+      "learning_rate": 9.978895672674947e-06,
+      "loss": 0.5858,
+      "step": 3586
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.968886137008667,
+      "learning_rate": 9.978871466707003e-06,
+      "loss": 0.6433,
+      "step": 3587
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.9090369939804077,
+      "learning_rate": 9.978847246894683e-06,
+      "loss": 0.6929,
+      "step": 3588
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.9020183086395264,
+      "learning_rate": 9.978823013238054e-06,
+      "loss": 0.5503,
+      "step": 3589
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.44067120552063,
+      "learning_rate": 9.978798765737183e-06,
+      "loss": 0.528,
+      "step": 3590
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.2626991271972656,
+      "learning_rate": 9.978774504392139e-06,
+      "loss": 0.8394,
+      "step": 3591
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.9647483825683594,
+      "learning_rate": 9.978750229202989e-06,
+      "loss": 0.4909,
+      "step": 3592
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.086660623550415,
+      "learning_rate": 9.978725940169798e-06,
+      "loss": 0.5726,
+      "step": 3593
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.9207215309143066,
+      "learning_rate": 9.978701637292635e-06,
+      "loss": 0.743,
+      "step": 3594
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.856440782546997,
+      "learning_rate": 9.978677320571567e-06,
+      "loss": 0.6996,
+      "step": 3595
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.629014492034912,
+      "learning_rate": 9.978652990006666e-06,
+      "loss": 0.8507,
+      "step": 3596
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.189675807952881,
+      "learning_rate": 9.978628645597992e-06,
+      "loss": 0.719,
+      "step": 3597
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 8.352169036865234,
+      "learning_rate": 9.97860428734562e-06,
+      "loss": 0.6012,
+      "step": 3598
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.788486957550049,
+      "learning_rate": 9.978579915249612e-06,
+      "loss": 0.7792,
+      "step": 3599
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.9901015758514404,
+      "learning_rate": 9.978555529310038e-06,
+      "loss": 0.5931,
+      "step": 3600
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.354952812194824,
+      "learning_rate": 9.97853112952697e-06,
+      "loss": 0.5621,
+      "step": 3601
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.9258184432983398,
+      "learning_rate": 9.978506715900469e-06,
+      "loss": 0.6177,
+      "step": 3602
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.4588751792907715,
+      "learning_rate": 9.978482288430606e-06,
+      "loss": 0.8869,
+      "step": 3603
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.714600086212158,
+      "learning_rate": 9.978457847117449e-06,
+      "loss": 0.6816,
+      "step": 3604
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.2456817626953125,
+      "learning_rate": 9.978433391961064e-06,
+      "loss": 0.7194,
+      "step": 3605
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.4026238918304443,
+      "learning_rate": 9.978408922961524e-06,
+      "loss": 0.7516,
+      "step": 3606
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.146012544631958,
+      "learning_rate": 9.97838444011889e-06,
+      "loss": 0.7765,
+      "step": 3607
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.2292749881744385,
+      "learning_rate": 9.978359943433237e-06,
+      "loss": 0.5985,
+      "step": 3608
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 4.933447360992432,
+      "learning_rate": 9.978335432904629e-06,
+      "loss": 0.3298,
+      "step": 3609
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.367793321609497,
+      "learning_rate": 9.978310908533136e-06,
+      "loss": 0.713,
+      "step": 3610
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.2819504737854004,
+      "learning_rate": 9.978286370318824e-06,
+      "loss": 0.5957,
+      "step": 3611
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.731274127960205,
+      "learning_rate": 9.978261818261761e-06,
+      "loss": 0.6674,
+      "step": 3612
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 5.235377311706543,
+      "learning_rate": 9.97823725236202e-06,
+      "loss": 0.4995,
+      "step": 3613
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 4.98893404006958,
+      "learning_rate": 9.978212672619665e-06,
+      "loss": 0.9597,
+      "step": 3614
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.2417280673980713,
+      "learning_rate": 9.978188079034765e-06,
+      "loss": 0.7436,
+      "step": 3615
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.9428927898406982,
+      "learning_rate": 9.978163471607387e-06,
+      "loss": 0.6891,
+      "step": 3616
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.892913341522217,
+      "learning_rate": 9.978138850337605e-06,
+      "loss": 0.6185,
+      "step": 3617
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 5.157970428466797,
+      "learning_rate": 9.97811421522548e-06,
+      "loss": 0.7974,
+      "step": 3618
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.3128910064697266,
+      "learning_rate": 9.978089566271087e-06,
+      "loss": 0.2932,
+      "step": 3619
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.3022711277008057,
+      "learning_rate": 9.97806490347449e-06,
+      "loss": 0.8777,
+      "step": 3620
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.3755972385406494,
+      "learning_rate": 9.97804022683576e-06,
+      "loss": 0.7099,
+      "step": 3621
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.131641387939453,
+      "learning_rate": 9.978015536354967e-06,
+      "loss": 0.7616,
+      "step": 3622
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.426455020904541,
+      "learning_rate": 9.977990832032174e-06,
+      "loss": 0.48,
+      "step": 3623
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.6370534896850586,
+      "learning_rate": 9.977966113867454e-06,
+      "loss": 0.5155,
+      "step": 3624
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.812230110168457,
+      "learning_rate": 9.977941381860876e-06,
+      "loss": 0.7581,
+      "step": 3625
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.0023112297058105,
+      "learning_rate": 9.977916636012509e-06,
+      "loss": 0.7804,
+      "step": 3626
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.9835927486419678,
+      "learning_rate": 9.977891876322418e-06,
+      "loss": 0.6717,
+      "step": 3627
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.1064810752868652,
+      "learning_rate": 9.977867102790677e-06,
+      "loss": 0.7088,
+      "step": 3628
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.9508297443389893,
+      "learning_rate": 9.97784231541735e-06,
+      "loss": 0.5641,
+      "step": 3629
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.3290419578552246,
+      "learning_rate": 9.97781751420251e-06,
+      "loss": 0.7104,
+      "step": 3630
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.840822219848633,
+      "learning_rate": 9.977792699146224e-06,
+      "loss": 0.886,
+      "step": 3631
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.6463003158569336,
+      "learning_rate": 9.977767870248561e-06,
+      "loss": 0.7539,
+      "step": 3632
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.7746840715408325,
+      "learning_rate": 9.977743027509589e-06,
+      "loss": 0.6299,
+      "step": 3633
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.559340000152588,
+      "learning_rate": 9.97771817092938e-06,
+      "loss": 0.5622,
+      "step": 3634
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.847151756286621,
+      "learning_rate": 9.977693300508001e-06,
+      "loss": 0.6579,
+      "step": 3635
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.7969768047332764,
+      "learning_rate": 9.977668416245521e-06,
+      "loss": 0.7009,
+      "step": 3636
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.1810660362243652,
+      "learning_rate": 9.97764351814201e-06,
+      "loss": 0.7393,
+      "step": 3637
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.0396718978881836,
+      "learning_rate": 9.977618606197538e-06,
+      "loss": 0.736,
+      "step": 3638
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.7962048053741455,
+      "learning_rate": 9.977593680412173e-06,
+      "loss": 0.4566,
+      "step": 3639
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.0573816299438477,
+      "learning_rate": 9.977568740785984e-06,
+      "loss": 0.8776,
+      "step": 3640
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.409297466278076,
+      "learning_rate": 9.977543787319042e-06,
+      "loss": 0.6276,
+      "step": 3641
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.3210043907165527,
+      "learning_rate": 9.977518820011415e-06,
+      "loss": 0.4523,
+      "step": 3642
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.7684789896011353,
+      "learning_rate": 9.977493838863172e-06,
+      "loss": 0.4482,
+      "step": 3643
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.1641600131988525,
+      "learning_rate": 9.977468843874383e-06,
+      "loss": 0.8718,
+      "step": 3644
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.9683432579040527,
+      "learning_rate": 9.97744383504512e-06,
+      "loss": 0.7073,
+      "step": 3645
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.420806884765625,
+      "learning_rate": 9.977418812375449e-06,
+      "loss": 0.6433,
+      "step": 3646
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.6400957107543945,
+      "learning_rate": 9.97739377586544e-06,
+      "loss": 0.5893,
+      "step": 3647
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.8689873218536377,
+      "learning_rate": 9.977368725515164e-06,
+      "loss": 0.5695,
+      "step": 3648
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.2624287605285645,
+      "learning_rate": 9.97734366132469e-06,
+      "loss": 0.571,
+      "step": 3649
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.070312976837158,
+      "learning_rate": 9.977318583294088e-06,
+      "loss": 0.6492,
+      "step": 3650
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.0867767333984375,
+      "learning_rate": 9.977293491423428e-06,
+      "loss": 0.7165,
+      "step": 3651
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.16910982131958,
+      "learning_rate": 9.977268385712778e-06,
+      "loss": 0.6575,
+      "step": 3652
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.4595937728881836,
+      "learning_rate": 9.97724326616221e-06,
+      "loss": 0.5952,
+      "step": 3653
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.881995916366577,
+      "learning_rate": 9.977218132771794e-06,
+      "loss": 0.3776,
+      "step": 3654
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.490525722503662,
+      "learning_rate": 9.977192985541597e-06,
+      "loss": 0.5517,
+      "step": 3655
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.899027705192566,
+      "learning_rate": 9.977167824471692e-06,
+      "loss": 0.8772,
+      "step": 3656
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.8545897006988525,
+      "learning_rate": 9.977142649562148e-06,
+      "loss": 0.5694,
+      "step": 3657
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 5.33769416809082,
+      "learning_rate": 9.977117460813035e-06,
+      "loss": 0.6804,
+      "step": 3658
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.647467017173767,
+      "learning_rate": 9.97709225822442e-06,
+      "loss": 0.6207,
+      "step": 3659
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.8089241981506348,
+      "learning_rate": 9.97706704179638e-06,
+      "loss": 0.8756,
+      "step": 3660
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.6917307376861572,
+      "learning_rate": 9.977041811528979e-06,
+      "loss": 0.4971,
+      "step": 3661
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.968839406967163,
+      "learning_rate": 9.977016567422289e-06,
+      "loss": 0.8888,
+      "step": 3662
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.4896445274353027,
+      "learning_rate": 9.97699130947638e-06,
+      "loss": 0.5329,
+      "step": 3663
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.9950387477874756,
+      "learning_rate": 9.976966037691323e-06,
+      "loss": 0.8045,
+      "step": 3664
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.6733176708221436,
+      "learning_rate": 9.976940752067188e-06,
+      "loss": 0.6387,
+      "step": 3665
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.7589049339294434,
+      "learning_rate": 9.976915452604046e-06,
+      "loss": 0.8073,
+      "step": 3666
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.4362006187438965,
+      "learning_rate": 9.976890139301966e-06,
+      "loss": 0.7332,
+      "step": 3667
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.9918148517608643,
+      "learning_rate": 9.976864812161018e-06,
+      "loss": 0.6739,
+      "step": 3668
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.735534906387329,
+      "learning_rate": 9.976839471181275e-06,
+      "loss": 0.8576,
+      "step": 3669
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.709618330001831,
+      "learning_rate": 9.976814116362804e-06,
+      "loss": 0.4586,
+      "step": 3670
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.9404255151748657,
+      "learning_rate": 9.976788747705677e-06,
+      "loss": 0.485,
+      "step": 3671
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.66515851020813,
+      "learning_rate": 9.976763365209968e-06,
+      "loss": 0.7406,
+      "step": 3672
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.7089500427246094,
+      "learning_rate": 9.97673796887574e-06,
+      "loss": 0.7776,
+      "step": 3673
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 4.811774253845215,
+      "learning_rate": 9.976712558703073e-06,
+      "loss": 0.8395,
+      "step": 3674
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.6569656133651733,
+      "learning_rate": 9.976687134692028e-06,
+      "loss": 0.8335,
+      "step": 3675
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.318418502807617,
+      "learning_rate": 9.976661696842683e-06,
+      "loss": 0.5262,
+      "step": 3676
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 5.593687534332275,
+      "learning_rate": 9.976636245155107e-06,
+      "loss": 0.7594,
+      "step": 3677
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 4.200995922088623,
+      "learning_rate": 9.976610779629368e-06,
+      "loss": 0.3035,
+      "step": 3678
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.9753963947296143,
+      "learning_rate": 9.976585300265537e-06,
+      "loss": 0.7085,
+      "step": 3679
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.417578935623169,
+      "learning_rate": 9.97655980706369e-06,
+      "loss": 0.7622,
+      "step": 3680
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.0804903507232666,
+      "learning_rate": 9.976534300023893e-06,
+      "loss": 0.6816,
+      "step": 3681
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.894677758216858,
+      "learning_rate": 9.97650877914622e-06,
+      "loss": 0.5393,
+      "step": 3682
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.101207733154297,
+      "learning_rate": 9.976483244430737e-06,
+      "loss": 0.8646,
+      "step": 3683
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.6395976543426514,
+      "learning_rate": 9.97645769587752e-06,
+      "loss": 0.8009,
+      "step": 3684
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.874534845352173,
+      "learning_rate": 9.97643213348664e-06,
+      "loss": 0.8317,
+      "step": 3685
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.0471208095550537,
+      "learning_rate": 9.976406557258164e-06,
+      "loss": 0.6705,
+      "step": 3686
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 4.011102199554443,
+      "learning_rate": 9.976380967192167e-06,
+      "loss": 0.7345,
+      "step": 3687
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.2089614868164062,
+      "learning_rate": 9.976355363288718e-06,
+      "loss": 0.6926,
+      "step": 3688
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.6612720489501953,
+      "learning_rate": 9.97632974554789e-06,
+      "loss": 0.4577,
+      "step": 3689
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.6904802322387695,
+      "learning_rate": 9.97630411396975e-06,
+      "loss": 0.529,
+      "step": 3690
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.216040849685669,
+      "learning_rate": 9.976278468554377e-06,
+      "loss": 0.6832,
+      "step": 3691
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.0817925930023193,
+      "learning_rate": 9.976252809301834e-06,
+      "loss": 0.5226,
+      "step": 3692
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 11.932365417480469,
+      "learning_rate": 9.976227136212198e-06,
+      "loss": 0.9655,
+      "step": 3693
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.128492593765259,
+      "learning_rate": 9.976201449285539e-06,
+      "loss": 0.7467,
+      "step": 3694
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.4942777156829834,
+      "learning_rate": 9.976175748521926e-06,
+      "loss": 0.7221,
+      "step": 3695
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.6561737060546875,
+      "learning_rate": 9.976150033921434e-06,
+      "loss": 0.7156,
+      "step": 3696
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.9570060968399048,
+      "learning_rate": 9.976124305484131e-06,
+      "loss": 0.9941,
+      "step": 3697
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 6.067004203796387,
+      "learning_rate": 9.976098563210093e-06,
+      "loss": 0.4573,
+      "step": 3698
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.116687297821045,
+      "learning_rate": 9.976072807099387e-06,
+      "loss": 0.6288,
+      "step": 3699
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.4403395652770996,
+      "learning_rate": 9.976047037152087e-06,
+      "loss": 0.552,
+      "step": 3700
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.7145259380340576,
+      "learning_rate": 9.976021253368264e-06,
+      "loss": 0.539,
+      "step": 3701
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 9.725775718688965,
+      "learning_rate": 9.97599545574799e-06,
+      "loss": 0.642,
+      "step": 3702
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 11.375762939453125,
+      "learning_rate": 9.97596964429134e-06,
+      "loss": 0.5582,
+      "step": 3703
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.2492434978485107,
+      "learning_rate": 9.975943818998377e-06,
+      "loss": 0.5355,
+      "step": 3704
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.875170946121216,
+      "learning_rate": 9.975917979869183e-06,
+      "loss": 0.6021,
+      "step": 3705
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 4.665700435638428,
+      "learning_rate": 9.975892126903822e-06,
+      "loss": 0.9433,
+      "step": 3706
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.781388521194458,
+      "learning_rate": 9.975866260102369e-06,
+      "loss": 0.5539,
+      "step": 3707
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.389758825302124,
+      "learning_rate": 9.975840379464897e-06,
+      "loss": 0.8178,
+      "step": 3708
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.9549753665924072,
+      "learning_rate": 9.975814484991477e-06,
+      "loss": 0.8221,
+      "step": 3709
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.656981945037842,
+      "learning_rate": 9.975788576682181e-06,
+      "loss": 0.5721,
+      "step": 3710
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.1381070613861084,
+      "learning_rate": 9.97576265453708e-06,
+      "loss": 0.554,
+      "step": 3711
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.709489345550537,
+      "learning_rate": 9.975736718556249e-06,
+      "loss": 0.5666,
+      "step": 3712
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.0105457305908203,
+      "learning_rate": 9.975710768739756e-06,
+      "loss": 0.4262,
+      "step": 3713
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.8108980655670166,
+      "learning_rate": 9.975684805087675e-06,
+      "loss": 0.6573,
+      "step": 3714
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.081630229949951,
+      "learning_rate": 9.97565882760008e-06,
+      "loss": 0.5912,
+      "step": 3715
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.537766695022583,
+      "learning_rate": 9.975632836277042e-06,
+      "loss": 0.7161,
+      "step": 3716
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.2274160385131836,
+      "learning_rate": 9.975606831118631e-06,
+      "loss": 0.6685,
+      "step": 3717
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.2893693447113037,
+      "learning_rate": 9.975580812124924e-06,
+      "loss": 0.6663,
+      "step": 3718
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 4.2745490074157715,
+      "learning_rate": 9.975554779295989e-06,
+      "loss": 0.8416,
+      "step": 3719
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.2005834579467773,
+      "learning_rate": 9.975528732631901e-06,
+      "loss": 0.4414,
+      "step": 3720
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.853288412094116,
+      "learning_rate": 9.975502672132731e-06,
+      "loss": 0.6705,
+      "step": 3721
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.7972253561019897,
+      "learning_rate": 9.97547659779855e-06,
+      "loss": 0.7487,
+      "step": 3722
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.8834946155548096,
+      "learning_rate": 9.975450509629436e-06,
+      "loss": 0.518,
+      "step": 3723
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.3464925289154053,
+      "learning_rate": 9.975424407625456e-06,
+      "loss": 0.514,
+      "step": 3724
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.754010796546936,
+      "learning_rate": 9.975398291786684e-06,
+      "loss": 0.6003,
+      "step": 3725
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.6400017738342285,
+      "learning_rate": 9.975372162113193e-06,
+      "loss": 0.7646,
+      "step": 3726
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.9934513568878174,
+      "learning_rate": 9.975346018605057e-06,
+      "loss": 0.7682,
+      "step": 3727
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.9391727447509766,
+      "learning_rate": 9.975319861262346e-06,
+      "loss": 0.4781,
+      "step": 3728
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.019893169403076,
+      "learning_rate": 9.975293690085137e-06,
+      "loss": 0.5894,
+      "step": 3729
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.8853511810302734,
+      "learning_rate": 9.975267505073499e-06,
+      "loss": 0.555,
+      "step": 3730
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.5027146339416504,
+      "learning_rate": 9.975241306227506e-06,
+      "loss": 0.6806,
+      "step": 3731
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.560786247253418,
+      "learning_rate": 9.975215093547228e-06,
+      "loss": 0.843,
+      "step": 3732
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.013843297958374,
+      "learning_rate": 9.975188867032743e-06,
+      "loss": 0.7709,
+      "step": 3733
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.0927743911743164,
+      "learning_rate": 9.975162626684121e-06,
+      "loss": 0.6719,
+      "step": 3734
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 5.118415832519531,
+      "learning_rate": 9.975136372501437e-06,
+      "loss": 0.7361,
+      "step": 3735
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.5292913913726807,
+      "learning_rate": 9.97511010448476e-06,
+      "loss": 0.6351,
+      "step": 3736
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.4496426582336426,
+      "learning_rate": 9.975083822634167e-06,
+      "loss": 0.6756,
+      "step": 3737
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.492560863494873,
+      "learning_rate": 9.97505752694973e-06,
+      "loss": 0.6108,
+      "step": 3738
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.8396565914154053,
+      "learning_rate": 9.97503121743152e-06,
+      "loss": 0.6407,
+      "step": 3739
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.845879554748535,
+      "learning_rate": 9.975004894079613e-06,
+      "loss": 0.7872,
+      "step": 3740
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.638867974281311,
+      "learning_rate": 9.974978556894081e-06,
+      "loss": 0.4284,
+      "step": 3741
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.6435070037841797,
+      "learning_rate": 9.974952205874997e-06,
+      "loss": 0.8386,
+      "step": 3742
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.1898176670074463,
+      "learning_rate": 9.974925841022435e-06,
+      "loss": 0.7064,
+      "step": 3743
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.6857253313064575,
+      "learning_rate": 9.974899462336467e-06,
+      "loss": 0.7826,
+      "step": 3744
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.8746755123138428,
+      "learning_rate": 9.974873069817168e-06,
+      "loss": 0.4461,
+      "step": 3745
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.2015273571014404,
+      "learning_rate": 9.97484666346461e-06,
+      "loss": 0.669,
+      "step": 3746
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.3270580768585205,
+      "learning_rate": 9.974820243278868e-06,
+      "loss": 0.6822,
+      "step": 3747
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.4184277057647705,
+      "learning_rate": 9.974793809260015e-06,
+      "loss": 0.678,
+      "step": 3748
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.7970008850097656,
+      "learning_rate": 9.974767361408123e-06,
+      "loss": 0.8373,
+      "step": 3749
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.4413492679595947,
+      "learning_rate": 9.974740899723266e-06,
+      "loss": 0.7356,
+      "step": 3750
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.2031073570251465,
+      "learning_rate": 9.974714424205518e-06,
+      "loss": 0.6683,
+      "step": 3751
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.291048526763916,
+      "learning_rate": 9.974687934854954e-06,
+      "loss": 0.606,
+      "step": 3752
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.6833908557891846,
+      "learning_rate": 9.974661431671645e-06,
+      "loss": 0.7316,
+      "step": 3753
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.1096954345703125,
+      "learning_rate": 9.974634914655669e-06,
+      "loss": 0.6221,
+      "step": 3754
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.6486732959747314,
+      "learning_rate": 9.974608383807094e-06,
+      "loss": 0.6972,
+      "step": 3755
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.528184175491333,
+      "learning_rate": 9.974581839125999e-06,
+      "loss": 0.8253,
+      "step": 3756
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.797696828842163,
+      "learning_rate": 9.974555280612452e-06,
+      "loss": 0.5808,
+      "step": 3757
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.119694471359253,
+      "learning_rate": 9.974528708266534e-06,
+      "loss": 0.7479,
+      "step": 3758
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.325930595397949,
+      "learning_rate": 9.974502122088311e-06,
+      "loss": 0.8144,
+      "step": 3759
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.188891649246216,
+      "learning_rate": 9.974475522077865e-06,
+      "loss": 0.695,
+      "step": 3760
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.9182713031768799,
+      "learning_rate": 9.974448908235264e-06,
+      "loss": 0.9952,
+      "step": 3761
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.5274534225463867,
+      "learning_rate": 9.974422280560584e-06,
+      "loss": 0.7376,
+      "step": 3762
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.2499780654907227,
+      "learning_rate": 9.9743956390539e-06,
+      "loss": 0.6289,
+      "step": 3763
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.3574535846710205,
+      "learning_rate": 9.974368983715286e-06,
+      "loss": 0.7838,
+      "step": 3764
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 5.020464897155762,
+      "learning_rate": 9.974342314544814e-06,
+      "loss": 0.7232,
+      "step": 3765
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 5.626638412475586,
+      "learning_rate": 9.97431563154256e-06,
+      "loss": 0.8449,
+      "step": 3766
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 4.400149345397949,
+      "learning_rate": 9.974288934708597e-06,
+      "loss": 1.0755,
+      "step": 3767
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.1565070152282715,
+      "learning_rate": 9.974262224043e-06,
+      "loss": 0.65,
+      "step": 3768
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.4424386024475098,
+      "learning_rate": 9.974235499545844e-06,
+      "loss": 0.7928,
+      "step": 3769
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.797603726387024,
+      "learning_rate": 9.9742087612172e-06,
+      "loss": 0.7287,
+      "step": 3770
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.703873634338379,
+      "learning_rate": 9.974182009057148e-06,
+      "loss": 1.0604,
+      "step": 3771
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.0475800037384033,
+      "learning_rate": 9.974155243065757e-06,
+      "loss": 0.7385,
+      "step": 3772
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.623098850250244,
+      "learning_rate": 9.974128463243105e-06,
+      "loss": 0.7033,
+      "step": 3773
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.8810820579528809,
+      "learning_rate": 9.974101669589265e-06,
+      "loss": 0.6639,
+      "step": 3774
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 4.896986484527588,
+      "learning_rate": 9.97407486210431e-06,
+      "loss": 0.9564,
+      "step": 3775
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.111880302429199,
+      "learning_rate": 9.974048040788317e-06,
+      "loss": 0.4006,
+      "step": 3776
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.644360065460205,
+      "learning_rate": 9.97402120564136e-06,
+      "loss": 0.84,
+      "step": 3777
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.4542136192321777,
+      "learning_rate": 9.973994356663514e-06,
+      "loss": 0.6808,
+      "step": 3778
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.7901477813720703,
+      "learning_rate": 9.973967493854853e-06,
+      "loss": 0.8959,
+      "step": 3779
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.369626998901367,
+      "learning_rate": 9.97394061721545e-06,
+      "loss": 0.5154,
+      "step": 3780
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 4.61347770690918,
+      "learning_rate": 9.973913726745383e-06,
+      "loss": 0.9299,
+      "step": 3781
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.3635056018829346,
+      "learning_rate": 9.973886822444726e-06,
+      "loss": 0.8016,
+      "step": 3782
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.740110158920288,
+      "learning_rate": 9.97385990431355e-06,
+      "loss": 0.4467,
+      "step": 3783
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.3781542778015137,
+      "learning_rate": 9.973832972351934e-06,
+      "loss": 0.8115,
+      "step": 3784
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 5.398865222930908,
+      "learning_rate": 9.973806026559952e-06,
+      "loss": 0.9034,
+      "step": 3785
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.3938660621643066,
+      "learning_rate": 9.973779066937678e-06,
+      "loss": 0.2864,
+      "step": 3786
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.188871145248413,
+      "learning_rate": 9.97375209348519e-06,
+      "loss": 0.766,
+      "step": 3787
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 4.464680194854736,
+      "learning_rate": 9.973725106202559e-06,
+      "loss": 0.5546,
+      "step": 3788
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 4.621243000030518,
+      "learning_rate": 9.973698105089862e-06,
+      "loss": 0.3996,
+      "step": 3789
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.055593729019165,
+      "learning_rate": 9.973671090147174e-06,
+      "loss": 0.5204,
+      "step": 3790
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.5086941719055176,
+      "learning_rate": 9.973644061374568e-06,
+      "loss": 0.5967,
+      "step": 3791
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 5.251680850982666,
+      "learning_rate": 9.973617018772124e-06,
+      "loss": 0.8125,
+      "step": 3792
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.0903940200805664,
+      "learning_rate": 9.973589962339912e-06,
+      "loss": 0.7557,
+      "step": 3793
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.254047393798828,
+      "learning_rate": 9.97356289207801e-06,
+      "loss": 0.5505,
+      "step": 3794
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.809465169906616,
+      "learning_rate": 9.973535807986495e-06,
+      "loss": 0.5504,
+      "step": 3795
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.8319377899169922,
+      "learning_rate": 9.973508710065437e-06,
+      "loss": 0.5033,
+      "step": 3796
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.899940252304077,
+      "learning_rate": 9.973481598314914e-06,
+      "loss": 0.8391,
+      "step": 3797
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.5848865509033203,
+      "learning_rate": 9.973454472735005e-06,
+      "loss": 0.4595,
+      "step": 3798
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.273179292678833,
+      "learning_rate": 9.97342733332578e-06,
+      "loss": 0.7055,
+      "step": 3799
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.474071979522705,
+      "learning_rate": 9.973400180087316e-06,
+      "loss": 0.5703,
+      "step": 3800
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.672763705253601,
+      "learning_rate": 9.973373013019692e-06,
+      "loss": 0.6098,
+      "step": 3801
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.761753559112549,
+      "learning_rate": 9.973345832122978e-06,
+      "loss": 0.6875,
+      "step": 3802
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.357099771499634,
+      "learning_rate": 9.973318637397254e-06,
+      "loss": 0.705,
+      "step": 3803
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 4.879098415374756,
+      "learning_rate": 9.973291428842593e-06,
+      "loss": 0.5616,
+      "step": 3804
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.497554302215576,
+      "learning_rate": 9.973264206459071e-06,
+      "loss": 0.8909,
+      "step": 3805
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.74503755569458,
+      "learning_rate": 9.973236970246764e-06,
+      "loss": 0.6945,
+      "step": 3806
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.00895357131958,
+      "learning_rate": 9.97320972020575e-06,
+      "loss": 0.7449,
+      "step": 3807
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.8982889652252197,
+      "learning_rate": 9.9731824563361e-06,
+      "loss": 0.6576,
+      "step": 3808
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.2316644191741943,
+      "learning_rate": 9.973155178637897e-06,
+      "loss": 0.8063,
+      "step": 3809
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.6909942626953125,
+      "learning_rate": 9.973127887111209e-06,
+      "loss": 0.7386,
+      "step": 3810
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.948543667793274,
+      "learning_rate": 9.973100581756117e-06,
+      "loss": 0.8688,
+      "step": 3811
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.0939242839813232,
+      "learning_rate": 9.973073262572693e-06,
+      "loss": 0.5615,
+      "step": 3812
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.4219212532043457,
+      "learning_rate": 9.973045929561016e-06,
+      "loss": 0.8766,
+      "step": 3813
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.6563968658447266,
+      "learning_rate": 9.973018582721163e-06,
+      "loss": 0.754,
+      "step": 3814
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.3267593383789062,
+      "learning_rate": 9.972991222053207e-06,
+      "loss": 0.4221,
+      "step": 3815
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 4.378140449523926,
+      "learning_rate": 9.972963847557225e-06,
+      "loss": 0.7624,
+      "step": 3816
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 6.112624645233154,
+      "learning_rate": 9.972936459233294e-06,
+      "loss": 0.6307,
+      "step": 3817
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.7670228481292725,
+      "learning_rate": 9.97290905708149e-06,
+      "loss": 0.4799,
+      "step": 3818
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.907700300216675,
+      "learning_rate": 9.972881641101888e-06,
+      "loss": 0.5069,
+      "step": 3819
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.1140296459198,
+      "learning_rate": 9.972854211294564e-06,
+      "loss": 0.6312,
+      "step": 3820
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.699693441390991,
+      "learning_rate": 9.972826767659599e-06,
+      "loss": 0.8347,
+      "step": 3821
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 6.531801700592041,
+      "learning_rate": 9.972799310197063e-06,
+      "loss": 0.7437,
+      "step": 3822
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.729631185531616,
+      "learning_rate": 9.972771838907034e-06,
+      "loss": 0.8119,
+      "step": 3823
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.0224225521087646,
+      "learning_rate": 9.972744353789591e-06,
+      "loss": 0.9065,
+      "step": 3824
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 4.499367713928223,
+      "learning_rate": 9.972716854844808e-06,
+      "loss": 0.6958,
+      "step": 3825
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.463103771209717,
+      "learning_rate": 9.972689342072763e-06,
+      "loss": 0.742,
+      "step": 3826
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.3875696659088135,
+      "learning_rate": 9.972661815473532e-06,
+      "loss": 0.7803,
+      "step": 3827
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.0331664085388184,
+      "learning_rate": 9.97263427504719e-06,
+      "loss": 0.7638,
+      "step": 3828
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.7435216903686523,
+      "learning_rate": 9.972606720793816e-06,
+      "loss": 0.6967,
+      "step": 3829
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.498595714569092,
+      "learning_rate": 9.972579152713484e-06,
+      "loss": 0.6737,
+      "step": 3830
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.608379364013672,
+      "learning_rate": 9.972551570806275e-06,
+      "loss": 0.855,
+      "step": 3831
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.921924114227295,
+      "learning_rate": 9.97252397507226e-06,
+      "loss": 0.8963,
+      "step": 3832
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.1254303455352783,
+      "learning_rate": 9.972496365511519e-06,
+      "loss": 0.9715,
+      "step": 3833
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.3466129302978516,
+      "learning_rate": 9.97246874212413e-06,
+      "loss": 0.4125,
+      "step": 3834
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.021043539047241,
+      "learning_rate": 9.972441104910167e-06,
+      "loss": 0.9692,
+      "step": 3835
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.8108437061309814,
+      "learning_rate": 9.972413453869706e-06,
+      "loss": 0.6378,
+      "step": 3836
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 8.145825386047363,
+      "learning_rate": 9.97238578900283e-06,
+      "loss": 0.5183,
+      "step": 3837
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.7786908149719238,
+      "learning_rate": 9.97235811030961e-06,
+      "loss": 0.5427,
+      "step": 3838
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.79986572265625,
+      "learning_rate": 9.972330417790124e-06,
+      "loss": 1.1554,
+      "step": 3839
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.252885341644287,
+      "learning_rate": 9.97230271144445e-06,
+      "loss": 0.7874,
+      "step": 3840
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 4.299147129058838,
+      "learning_rate": 9.972274991272666e-06,
+      "loss": 0.5859,
+      "step": 3841
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 5.343671798706055,
+      "learning_rate": 9.972247257274848e-06,
+      "loss": 0.4994,
+      "step": 3842
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.1269779205322266,
+      "learning_rate": 9.97221950945107e-06,
+      "loss": 0.6556,
+      "step": 3843
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.342836380004883,
+      "learning_rate": 9.972191747801416e-06,
+      "loss": 0.7341,
+      "step": 3844
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.349949836730957,
+      "learning_rate": 9.972163972325957e-06,
+      "loss": 0.6357,
+      "step": 3845
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 4.353829383850098,
+      "learning_rate": 9.972136183024773e-06,
+      "loss": 0.9023,
+      "step": 3846
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.206719398498535,
+      "learning_rate": 9.972108379897942e-06,
+      "loss": 0.6512,
+      "step": 3847
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.5205299854278564,
+      "learning_rate": 9.972080562945538e-06,
+      "loss": 0.5759,
+      "step": 3848
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.381349802017212,
+      "learning_rate": 9.972052732167644e-06,
+      "loss": 0.6018,
+      "step": 3849
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.2207531929016113,
+      "learning_rate": 9.97202488756433e-06,
+      "loss": 0.5246,
+      "step": 3850
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.7933077812194824,
+      "learning_rate": 9.97199702913568e-06,
+      "loss": 0.9672,
+      "step": 3851
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 4.247515678405762,
+      "learning_rate": 9.971969156881768e-06,
+      "loss": 0.7567,
+      "step": 3852
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 4.012975215911865,
+      "learning_rate": 9.971941270802671e-06,
+      "loss": 0.8738,
+      "step": 3853
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.616586446762085,
+      "learning_rate": 9.971913370898471e-06,
+      "loss": 0.6424,
+      "step": 3854
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.4110260009765625,
+      "learning_rate": 9.97188545716924e-06,
+      "loss": 0.7199,
+      "step": 3855
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.341869831085205,
+      "learning_rate": 9.97185752961506e-06,
+      "loss": 0.8896,
+      "step": 3856
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.8668421506881714,
+      "learning_rate": 9.971829588236004e-06,
+      "loss": 0.3683,
+      "step": 3857
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.2359795570373535,
+      "learning_rate": 9.971801633032154e-06,
+      "loss": 0.5485,
+      "step": 3858
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.8901482820510864,
+      "learning_rate": 9.971773664003586e-06,
+      "loss": 0.7552,
+      "step": 3859
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.8386952877044678,
+      "learning_rate": 9.971745681150377e-06,
+      "loss": 0.446,
+      "step": 3860
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.255673408508301,
+      "learning_rate": 9.971717684472608e-06,
+      "loss": 0.5935,
+      "step": 3861
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.236131191253662,
+      "learning_rate": 9.971689673970353e-06,
+      "loss": 0.6295,
+      "step": 3862
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.038128137588501,
+      "learning_rate": 9.97166164964369e-06,
+      "loss": 0.714,
+      "step": 3863
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.052793264389038,
+      "learning_rate": 9.971633611492702e-06,
+      "loss": 0.8634,
+      "step": 3864
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.2114360332489014,
+      "learning_rate": 9.971605559517459e-06,
+      "loss": 0.7932,
+      "step": 3865
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.9463855028152466,
+      "learning_rate": 9.971577493718046e-06,
+      "loss": 0.4927,
+      "step": 3866
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.131751298904419,
+      "learning_rate": 9.971549414094538e-06,
+      "loss": 0.7729,
+      "step": 3867
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.032057762145996,
+      "learning_rate": 9.971521320647014e-06,
+      "loss": 0.9345,
+      "step": 3868
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.5219485759735107,
+      "learning_rate": 9.971493213375552e-06,
+      "loss": 0.6304,
+      "step": 3869
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.1253600120544434,
+      "learning_rate": 9.971465092280228e-06,
+      "loss": 0.7176,
+      "step": 3870
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.2418711185455322,
+      "learning_rate": 9.971436957361122e-06,
+      "loss": 0.8839,
+      "step": 3871
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.211885929107666,
+      "learning_rate": 9.971408808618313e-06,
+      "loss": 0.6929,
+      "step": 3872
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.442063331604004,
+      "learning_rate": 9.97138064605188e-06,
+      "loss": 0.8306,
+      "step": 3873
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.200986385345459,
+      "learning_rate": 9.971352469661896e-06,
+      "loss": 0.6214,
+      "step": 3874
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.460914373397827,
+      "learning_rate": 9.971324279448447e-06,
+      "loss": 1.0254,
+      "step": 3875
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 6.03832483291626,
+      "learning_rate": 9.971296075411605e-06,
+      "loss": 0.5648,
+      "step": 3876
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.14740252494812,
+      "learning_rate": 9.971267857551452e-06,
+      "loss": 0.7569,
+      "step": 3877
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 4.520980358123779,
+      "learning_rate": 9.971239625868066e-06,
+      "loss": 0.6467,
+      "step": 3878
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.203348398208618,
+      "learning_rate": 9.971211380361523e-06,
+      "loss": 0.786,
+      "step": 3879
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.119554042816162,
+      "learning_rate": 9.971183121031905e-06,
+      "loss": 0.8881,
+      "step": 3880
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.896586298942566,
+      "learning_rate": 9.971154847879287e-06,
+      "loss": 0.7098,
+      "step": 3881
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.616016149520874,
+      "learning_rate": 9.97112656090375e-06,
+      "loss": 0.8645,
+      "step": 3882
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.26440167427063,
+      "learning_rate": 9.971098260105374e-06,
+      "loss": 0.7132,
+      "step": 3883
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.64650821685791,
+      "learning_rate": 9.971069945484236e-06,
+      "loss": 0.7047,
+      "step": 3884
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.430497169494629,
+      "learning_rate": 9.971041617040412e-06,
+      "loss": 0.5924,
+      "step": 3885
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.55257248878479,
+      "learning_rate": 9.971013274773987e-06,
+      "loss": 0.7187,
+      "step": 3886
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.890657663345337,
+      "learning_rate": 9.970984918685034e-06,
+      "loss": 0.7933,
+      "step": 3887
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.903430461883545,
+      "learning_rate": 9.970956548773634e-06,
+      "loss": 0.7516,
+      "step": 3888
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.4492945671081543,
+      "learning_rate": 9.970928165039867e-06,
+      "loss": 0.8861,
+      "step": 3889
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.6132919788360596,
+      "learning_rate": 9.970899767483811e-06,
+      "loss": 0.7973,
+      "step": 3890
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.9517310857772827,
+      "learning_rate": 9.970871356105543e-06,
+      "loss": 0.9329,
+      "step": 3891
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.8086196184158325,
+      "learning_rate": 9.970842930905144e-06,
+      "loss": 0.5619,
+      "step": 3892
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.7771625518798828,
+      "learning_rate": 9.970814491882696e-06,
+      "loss": 0.4226,
+      "step": 3893
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.721126079559326,
+      "learning_rate": 9.970786039038273e-06,
+      "loss": 0.7502,
+      "step": 3894
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 5.256905555725098,
+      "learning_rate": 9.970757572371957e-06,
+      "loss": 1.0422,
+      "step": 3895
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 4.699684143066406,
+      "learning_rate": 9.970729091883824e-06,
+      "loss": 0.6769,
+      "step": 3896
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.2484400272369385,
+      "learning_rate": 9.970700597573958e-06,
+      "loss": 0.6844,
+      "step": 3897
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.234903335571289,
+      "learning_rate": 9.970672089442435e-06,
+      "loss": 0.7189,
+      "step": 3898
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.457644462585449,
+      "learning_rate": 9.970643567489334e-06,
+      "loss": 0.5252,
+      "step": 3899
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.6445472240448,
+      "learning_rate": 9.970615031714736e-06,
+      "loss": 0.6908,
+      "step": 3900
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.9583120346069336,
+      "learning_rate": 9.97058648211872e-06,
+      "loss": 0.4263,
+      "step": 3901
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.667388439178467,
+      "learning_rate": 9.970557918701365e-06,
+      "loss": 0.593,
+      "step": 3902
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.554481029510498,
+      "learning_rate": 9.970529341462749e-06,
+      "loss": 0.7011,
+      "step": 3903
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.2807583808898926,
+      "learning_rate": 9.970500750402955e-06,
+      "loss": 0.517,
+      "step": 3904
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.1628146171569824,
+      "learning_rate": 9.970472145522058e-06,
+      "loss": 0.8032,
+      "step": 3905
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.8127901554107666,
+      "learning_rate": 9.970443526820142e-06,
+      "loss": 0.6563,
+      "step": 3906
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.728240966796875,
+      "learning_rate": 9.970414894297284e-06,
+      "loss": 0.5357,
+      "step": 3907
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 4.064976692199707,
+      "learning_rate": 9.970386247953564e-06,
+      "loss": 0.6282,
+      "step": 3908
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.8554067611694336,
+      "learning_rate": 9.970357587789062e-06,
+      "loss": 0.6558,
+      "step": 3909
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 5.537252902984619,
+      "learning_rate": 9.970328913803857e-06,
+      "loss": 0.5367,
+      "step": 3910
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.472449541091919,
+      "learning_rate": 9.970300225998031e-06,
+      "loss": 1.0406,
+      "step": 3911
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.684934377670288,
+      "learning_rate": 9.970271524371659e-06,
+      "loss": 0.6917,
+      "step": 3912
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.245370864868164,
+      "learning_rate": 9.970242808924826e-06,
+      "loss": 0.6023,
+      "step": 3913
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 6.657118797302246,
+      "learning_rate": 9.970214079657609e-06,
+      "loss": 0.83,
+      "step": 3914
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.2296838760375977,
+      "learning_rate": 9.970185336570089e-06,
+      "loss": 0.8675,
+      "step": 3915
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.3348402976989746,
+      "learning_rate": 9.970156579662345e-06,
+      "loss": 0.6525,
+      "step": 3916
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.0080409049987793,
+      "learning_rate": 9.970127808934457e-06,
+      "loss": 0.678,
+      "step": 3917
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.9911986589431763,
+      "learning_rate": 9.970099024386506e-06,
+      "loss": 0.4206,
+      "step": 3918
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.436818838119507,
+      "learning_rate": 9.970070226018571e-06,
+      "loss": 0.7023,
+      "step": 3919
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.9307820796966553,
+      "learning_rate": 9.970041413830733e-06,
+      "loss": 0.7603,
+      "step": 3920
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.4194693565368652,
+      "learning_rate": 9.97001258782307e-06,
+      "loss": 0.5394,
+      "step": 3921
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.1480159759521484,
+      "learning_rate": 9.969983747995666e-06,
+      "loss": 0.767,
+      "step": 3922
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.3390469551086426,
+      "learning_rate": 9.9699548943486e-06,
+      "loss": 0.7383,
+      "step": 3923
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.09304141998291,
+      "learning_rate": 9.969926026881948e-06,
+      "loss": 0.5776,
+      "step": 3924
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.777330160140991,
+      "learning_rate": 9.969897145595795e-06,
+      "loss": 0.6774,
+      "step": 3925
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.8633687496185303,
+      "learning_rate": 9.96986825049022e-06,
+      "loss": 0.6783,
+      "step": 3926
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.9001917839050293,
+      "learning_rate": 9.969839341565303e-06,
+      "loss": 0.816,
+      "step": 3927
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.5527403354644775,
+      "learning_rate": 9.969810418821124e-06,
+      "loss": 0.8832,
+      "step": 3928
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.108999252319336,
+      "learning_rate": 9.969781482257765e-06,
+      "loss": 0.4406,
+      "step": 3929
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.7146761417388916,
+      "learning_rate": 9.969752531875304e-06,
+      "loss": 0.5052,
+      "step": 3930
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.810484766960144,
+      "learning_rate": 9.969723567673824e-06,
+      "loss": 0.4653,
+      "step": 3931
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.3675122261047363,
+      "learning_rate": 9.969694589653405e-06,
+      "loss": 0.5136,
+      "step": 3932
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.6662850379943848,
+      "learning_rate": 9.969665597814126e-06,
+      "loss": 0.7453,
+      "step": 3933
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.257901668548584,
+      "learning_rate": 9.969636592156068e-06,
+      "loss": 0.5587,
+      "step": 3934
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.2130281925201416,
+      "learning_rate": 9.969607572679314e-06,
+      "loss": 0.5888,
+      "step": 3935
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 4.467640399932861,
+      "learning_rate": 9.969578539383942e-06,
+      "loss": 0.5955,
+      "step": 3936
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.974034547805786,
+      "learning_rate": 9.969549492270034e-06,
+      "loss": 0.6595,
+      "step": 3937
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.441192626953125,
+      "learning_rate": 9.969520431337669e-06,
+      "loss": 0.5839,
+      "step": 3938
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.1673343181610107,
+      "learning_rate": 9.96949135658693e-06,
+      "loss": 0.5697,
+      "step": 3939
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 4.1500420570373535,
+      "learning_rate": 9.9694622680179e-06,
+      "loss": 0.6577,
+      "step": 3940
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.0677120685577393,
+      "learning_rate": 9.969433165630654e-06,
+      "loss": 0.7005,
+      "step": 3941
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.9492485523223877,
+      "learning_rate": 9.969404049425278e-06,
+      "loss": 0.798,
+      "step": 3942
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.7881077527999878,
+      "learning_rate": 9.969374919401848e-06,
+      "loss": 0.6115,
+      "step": 3943
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.0191996097564697,
+      "learning_rate": 9.96934577556045e-06,
+      "loss": 0.5714,
+      "step": 3944
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.7252823114395142,
+      "learning_rate": 9.969316617901163e-06,
+      "loss": 0.5383,
+      "step": 3945
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.7567944526672363,
+      "learning_rate": 9.969287446424067e-06,
+      "loss": 0.401,
+      "step": 3946
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 4.432502746582031,
+      "learning_rate": 9.969258261129246e-06,
+      "loss": 0.8504,
+      "step": 3947
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.777108669281006,
+      "learning_rate": 9.969229062016778e-06,
+      "loss": 0.8289,
+      "step": 3948
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 10.332745552062988,
+      "learning_rate": 9.969199849086746e-06,
+      "loss": 0.4988,
+      "step": 3949
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.4951565265655518,
+      "learning_rate": 9.96917062233923e-06,
+      "loss": 0.6451,
+      "step": 3950
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.994532346725464,
+      "learning_rate": 9.969141381774312e-06,
+      "loss": 0.7585,
+      "step": 3951
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.468444347381592,
+      "learning_rate": 9.969112127392073e-06,
+      "loss": 0.7731,
+      "step": 3952
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.6327571868896484,
+      "learning_rate": 9.969082859192596e-06,
+      "loss": 0.8077,
+      "step": 3953
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.3770973682403564,
+      "learning_rate": 9.96905357717596e-06,
+      "loss": 0.8193,
+      "step": 3954
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.586155652999878,
+      "learning_rate": 9.969024281342247e-06,
+      "loss": 0.4542,
+      "step": 3955
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.177166700363159,
+      "learning_rate": 9.96899497169154e-06,
+      "loss": 0.7319,
+      "step": 3956
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.1785876750946045,
+      "learning_rate": 9.968965648223921e-06,
+      "loss": 0.7832,
+      "step": 3957
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 4.539702415466309,
+      "learning_rate": 9.968936310939468e-06,
+      "loss": 0.7404,
+      "step": 3958
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.3462371826171875,
+      "learning_rate": 9.968906959838264e-06,
+      "loss": 0.4725,
+      "step": 3959
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.7818541526794434,
+      "learning_rate": 9.968877594920391e-06,
+      "loss": 0.3843,
+      "step": 3960
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 4.6651129722595215,
+      "learning_rate": 9.968848216185932e-06,
+      "loss": 0.7067,
+      "step": 3961
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.821744203567505,
+      "learning_rate": 9.968818823634968e-06,
+      "loss": 0.8762,
+      "step": 3962
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.5393378734588623,
+      "learning_rate": 9.968789417267578e-06,
+      "loss": 0.9168,
+      "step": 3963
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.735696792602539,
+      "learning_rate": 9.968759997083848e-06,
+      "loss": 0.8391,
+      "step": 3964
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.564404010772705,
+      "learning_rate": 9.968730563083858e-06,
+      "loss": 0.7478,
+      "step": 3965
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.538638114929199,
+      "learning_rate": 9.968701115267687e-06,
+      "loss": 0.6424,
+      "step": 3966
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.6856112480163574,
+      "learning_rate": 9.96867165363542e-06,
+      "loss": 0.707,
+      "step": 3967
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.5986056327819824,
+      "learning_rate": 9.96864217818714e-06,
+      "loss": 0.6001,
+      "step": 3968
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.4259185791015625,
+      "learning_rate": 9.968612688922927e-06,
+      "loss": 0.7469,
+      "step": 3969
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.8894288539886475,
+      "learning_rate": 9.968583185842864e-06,
+      "loss": 0.6124,
+      "step": 3970
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 4.6563262939453125,
+      "learning_rate": 9.968553668947033e-06,
+      "loss": 0.5042,
+      "step": 3971
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.9561402797698975,
+      "learning_rate": 9.968524138235513e-06,
+      "loss": 0.7049,
+      "step": 3972
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.055198907852173,
+      "learning_rate": 9.96849459370839e-06,
+      "loss": 0.6603,
+      "step": 3973
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.7426459789276123,
+      "learning_rate": 9.968465035365744e-06,
+      "loss": 0.5414,
+      "step": 3974
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.534325361251831,
+      "learning_rate": 9.96843546320766e-06,
+      "loss": 0.6176,
+      "step": 3975
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.6432058811187744,
+      "learning_rate": 9.968405877234217e-06,
+      "loss": 0.6603,
+      "step": 3976
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.8504152297973633,
+      "learning_rate": 9.968376277445497e-06,
+      "loss": 0.872,
+      "step": 3977
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.300048351287842,
+      "learning_rate": 9.968346663841586e-06,
+      "loss": 0.7749,
+      "step": 3978
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.6725716590881348,
+      "learning_rate": 9.968317036422563e-06,
+      "loss": 0.3613,
+      "step": 3979
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.650291919708252,
+      "learning_rate": 9.968287395188512e-06,
+      "loss": 0.6826,
+      "step": 3980
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.9833977222442627,
+      "learning_rate": 9.968257740139513e-06,
+      "loss": 0.6544,
+      "step": 3981
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.645655393600464,
+      "learning_rate": 9.968228071275655e-06,
+      "loss": 0.7324,
+      "step": 3982
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 4.809957027435303,
+      "learning_rate": 9.968198388597012e-06,
+      "loss": 0.7019,
+      "step": 3983
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.8969138860702515,
+      "learning_rate": 9.96816869210367e-06,
+      "loss": 0.6406,
+      "step": 3984
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.0458438396453857,
+      "learning_rate": 9.968138981795714e-06,
+      "loss": 0.5414,
+      "step": 3985
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.674910545349121,
+      "learning_rate": 9.968109257673225e-06,
+      "loss": 0.641,
+      "step": 3986
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.935020685195923,
+      "learning_rate": 9.968079519736282e-06,
+      "loss": 0.7267,
+      "step": 3987
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.972424030303955,
+      "learning_rate": 9.968049767984974e-06,
+      "loss": 0.6828,
+      "step": 3988
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.5772042274475098,
+      "learning_rate": 9.96802000241938e-06,
+      "loss": 0.6026,
+      "step": 3989
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.637876510620117,
+      "learning_rate": 9.967990223039583e-06,
+      "loss": 0.7004,
+      "step": 3990
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.1788065433502197,
+      "learning_rate": 9.967960429845667e-06,
+      "loss": 0.9092,
+      "step": 3991
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 5.296181678771973,
+      "learning_rate": 9.967930622837712e-06,
+      "loss": 0.6403,
+      "step": 3992
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.9833406209945679,
+      "learning_rate": 9.967900802015806e-06,
+      "loss": 0.9128,
+      "step": 3993
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.0373029708862305,
+      "learning_rate": 9.967870967380027e-06,
+      "loss": 0.6693,
+      "step": 3994
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 6.670356750488281,
+      "learning_rate": 9.96784111893046e-06,
+      "loss": 0.5671,
+      "step": 3995
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.157532215118408,
+      "learning_rate": 9.967811256667188e-06,
+      "loss": 0.6405,
+      "step": 3996
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 5.246817588806152,
+      "learning_rate": 9.967781380590293e-06,
+      "loss": 1.023,
+      "step": 3997
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 5.523828983306885,
+      "learning_rate": 9.967751490699861e-06,
+      "loss": 0.4297,
+      "step": 3998
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 4.292252540588379,
+      "learning_rate": 9.96772158699597e-06,
+      "loss": 0.8012,
+      "step": 3999
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.4258880615234375,
+      "learning_rate": 9.967691669478709e-06,
+      "loss": 0.7873,
+      "step": 4000
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.314318895339966,
+      "learning_rate": 9.967661738148157e-06,
+      "loss": 0.5629,
+      "step": 4001
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.771599531173706,
+      "learning_rate": 9.9676317930044e-06,
+      "loss": 0.6983,
+      "step": 4002
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.4301416873931885,
+      "learning_rate": 9.96760183404752e-06,
+      "loss": 0.6555,
+      "step": 4003
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.5322494506835938,
+      "learning_rate": 9.967571861277598e-06,
+      "loss": 0.7352,
+      "step": 4004
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.3271846771240234,
+      "learning_rate": 9.967541874694722e-06,
+      "loss": 0.9166,
+      "step": 4005
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.452552556991577,
+      "learning_rate": 9.967511874298973e-06,
+      "loss": 0.6147,
+      "step": 4006
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.2021210193634033,
+      "learning_rate": 9.967481860090431e-06,
+      "loss": 0.6263,
+      "step": 4007
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.5176851749420166,
+      "learning_rate": 9.967451832069185e-06,
+      "loss": 0.8335,
+      "step": 4008
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.39801287651062,
+      "learning_rate": 9.967421790235317e-06,
+      "loss": 0.7339,
+      "step": 4009
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.3780808448791504,
+      "learning_rate": 9.96739173458891e-06,
+      "loss": 0.6143,
+      "step": 4010
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.967478036880493,
+      "learning_rate": 9.967361665130046e-06,
+      "loss": 0.6323,
+      "step": 4011
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.1726694107055664,
+      "learning_rate": 9.96733158185881e-06,
+      "loss": 0.79,
+      "step": 4012
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.705636739730835,
+      "learning_rate": 9.967301484775285e-06,
+      "loss": 0.5481,
+      "step": 4013
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 4.372578144073486,
+      "learning_rate": 9.967271373879558e-06,
+      "loss": 0.8029,
+      "step": 4014
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.085279941558838,
+      "learning_rate": 9.967241249171708e-06,
+      "loss": 0.9561,
+      "step": 4015
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.5960452556610107,
+      "learning_rate": 9.967211110651822e-06,
+      "loss": 0.6863,
+      "step": 4016
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.9464104175567627,
+      "learning_rate": 9.967180958319982e-06,
+      "loss": 0.5683,
+      "step": 4017
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.656466484069824,
+      "learning_rate": 9.967150792176273e-06,
+      "loss": 0.4651,
+      "step": 4018
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.3452932834625244,
+      "learning_rate": 9.967120612220777e-06,
+      "loss": 0.7321,
+      "step": 4019
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.9890931844711304,
+      "learning_rate": 9.96709041845358e-06,
+      "loss": 0.4933,
+      "step": 4020
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.9313642978668213,
+      "learning_rate": 9.967060210874767e-06,
+      "loss": 0.8084,
+      "step": 4021
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.0983145236968994,
+      "learning_rate": 9.967029989484418e-06,
+      "loss": 0.8471,
+      "step": 4022
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.9165459871292114,
+      "learning_rate": 9.96699975428262e-06,
+      "loss": 0.4876,
+      "step": 4023
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 4.821770668029785,
+      "learning_rate": 9.966969505269458e-06,
+      "loss": 0.4538,
+      "step": 4024
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.2749853134155273,
+      "learning_rate": 9.966939242445013e-06,
+      "loss": 0.6488,
+      "step": 4025
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 4.426034927368164,
+      "learning_rate": 9.96690896580937e-06,
+      "loss": 0.8362,
+      "step": 4026
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.9618165493011475,
+      "learning_rate": 9.966878675362614e-06,
+      "loss": 0.6922,
+      "step": 4027
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.174215078353882,
+      "learning_rate": 9.96684837110483e-06,
+      "loss": 0.7909,
+      "step": 4028
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.9749529361724854,
+      "learning_rate": 9.9668180530361e-06,
+      "loss": 0.9373,
+      "step": 4029
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.354905366897583,
+      "learning_rate": 9.966787721156511e-06,
+      "loss": 0.5163,
+      "step": 4030
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.8091943264007568,
+      "learning_rate": 9.966757375466144e-06,
+      "loss": 0.8971,
+      "step": 4031
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.2038869857788086,
+      "learning_rate": 9.966727015965088e-06,
+      "loss": 0.6408,
+      "step": 4032
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.806178331375122,
+      "learning_rate": 9.966696642653424e-06,
+      "loss": 0.8539,
+      "step": 4033
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.5644127130508423,
+      "learning_rate": 9.966666255531237e-06,
+      "loss": 0.6482,
+      "step": 4034
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 10.507744789123535,
+      "learning_rate": 9.966635854598611e-06,
+      "loss": 0.5038,
+      "step": 4035
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.6611838340759277,
+      "learning_rate": 9.966605439855632e-06,
+      "loss": 0.6236,
+      "step": 4036
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.9602950811386108,
+      "learning_rate": 9.966575011302381e-06,
+      "loss": 0.6884,
+      "step": 4037
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.973750591278076,
+      "learning_rate": 9.96654456893895e-06,
+      "loss": 0.6416,
+      "step": 4038
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.8051164150238037,
+      "learning_rate": 9.966514112765417e-06,
+      "loss": 0.6383,
+      "step": 4039
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.529250144958496,
+      "learning_rate": 9.966483642781867e-06,
+      "loss": 0.7428,
+      "step": 4040
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.389925479888916,
+      "learning_rate": 9.966453158988389e-06,
+      "loss": 0.6987,
+      "step": 4041
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 4.683279991149902,
+      "learning_rate": 9.966422661385065e-06,
+      "loss": 0.6824,
+      "step": 4042
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.5078986883163452,
+      "learning_rate": 9.966392149971979e-06,
+      "loss": 0.654,
+      "step": 4043
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.403317451477051,
+      "learning_rate": 9.966361624749216e-06,
+      "loss": 0.7568,
+      "step": 4044
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.7981009483337402,
+      "learning_rate": 9.966331085716864e-06,
+      "loss": 0.642,
+      "step": 4045
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.6248183250427246,
+      "learning_rate": 9.966300532875003e-06,
+      "loss": 0.761,
+      "step": 4046
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.75470232963562,
+      "learning_rate": 9.966269966223724e-06,
+      "loss": 0.911,
+      "step": 4047
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.547236919403076,
+      "learning_rate": 9.966239385763107e-06,
+      "loss": 0.7905,
+      "step": 4048
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.347013473510742,
+      "learning_rate": 9.966208791493239e-06,
+      "loss": 0.6976,
+      "step": 4049
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 4.241486549377441,
+      "learning_rate": 9.966178183414203e-06,
+      "loss": 0.7769,
+      "step": 4050
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.0746638774871826,
+      "learning_rate": 9.966147561526087e-06,
+      "loss": 0.7728,
+      "step": 4051
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.5018370151519775,
+      "learning_rate": 9.966116925828976e-06,
+      "loss": 0.7329,
+      "step": 4052
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.522106170654297,
+      "learning_rate": 9.966086276322953e-06,
+      "loss": 0.952,
+      "step": 4053
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.8648127317428589,
+      "learning_rate": 9.966055613008106e-06,
+      "loss": 0.6833,
+      "step": 4054
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.9746077060699463,
+      "learning_rate": 9.966024935884518e-06,
+      "loss": 0.7514,
+      "step": 4055
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.8120373487472534,
+      "learning_rate": 9.965994244952273e-06,
+      "loss": 0.6601,
+      "step": 4056
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.9866089820861816,
+      "learning_rate": 9.96596354021146e-06,
+      "loss": 0.6274,
+      "step": 4057
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.9800595045089722,
+      "learning_rate": 9.965932821662164e-06,
+      "loss": 0.8876,
+      "step": 4058
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 5.269138336181641,
+      "learning_rate": 9.965902089304467e-06,
+      "loss": 0.7095,
+      "step": 4059
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 4.210580825805664,
+      "learning_rate": 9.965871343138456e-06,
+      "loss": 0.6477,
+      "step": 4060
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.3894851207733154,
+      "learning_rate": 9.965840583164218e-06,
+      "loss": 0.7082,
+      "step": 4061
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.3803329467773438,
+      "learning_rate": 9.965809809381838e-06,
+      "loss": 0.8158,
+      "step": 4062
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.378286123275757,
+      "learning_rate": 9.965779021791401e-06,
+      "loss": 0.8302,
+      "step": 4063
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.593971014022827,
+      "learning_rate": 9.965748220392992e-06,
+      "loss": 0.7794,
+      "step": 4064
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.6844124794006348,
+      "learning_rate": 9.9657174051867e-06,
+      "loss": 0.6305,
+      "step": 4065
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.154966354370117,
+      "learning_rate": 9.965686576172604e-06,
+      "loss": 0.75,
+      "step": 4066
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.6250200271606445,
+      "learning_rate": 9.965655733350796e-06,
+      "loss": 0.7568,
+      "step": 4067
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.7773396968841553,
+      "learning_rate": 9.965624876721361e-06,
+      "loss": 1.0502,
+      "step": 4068
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.040354013442993,
+      "learning_rate": 9.96559400628438e-06,
+      "loss": 0.7365,
+      "step": 4069
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.738600254058838,
+      "learning_rate": 9.965563122039945e-06,
+      "loss": 0.8814,
+      "step": 4070
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.0988821983337402,
+      "learning_rate": 9.965532223988138e-06,
+      "loss": 0.7381,
+      "step": 4071
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.128789186477661,
+      "learning_rate": 9.965501312129047e-06,
+      "loss": 0.3886,
+      "step": 4072
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.9301047325134277,
+      "learning_rate": 9.965470386462756e-06,
+      "loss": 0.7776,
+      "step": 4073
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.8379037380218506,
+      "learning_rate": 9.965439446989352e-06,
+      "loss": 0.7463,
+      "step": 4074
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.313584804534912,
+      "learning_rate": 9.965408493708921e-06,
+      "loss": 0.6564,
+      "step": 4075
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.4122514724731445,
+      "learning_rate": 9.965377526621549e-06,
+      "loss": 0.6695,
+      "step": 4076
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 4.1665358543396,
+      "learning_rate": 9.965346545727324e-06,
+      "loss": 0.7986,
+      "step": 4077
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 4.040792942047119,
+      "learning_rate": 9.965315551026327e-06,
+      "loss": 0.5818,
+      "step": 4078
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.01973032951355,
+      "learning_rate": 9.96528454251865e-06,
+      "loss": 0.514,
+      "step": 4079
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.043067216873169,
+      "learning_rate": 9.965253520204377e-06,
+      "loss": 0.8407,
+      "step": 4080
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.4845314025878906,
+      "learning_rate": 9.965222484083593e-06,
+      "loss": 0.971,
+      "step": 4081
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.610020399093628,
+      "learning_rate": 9.965191434156384e-06,
+      "loss": 0.6657,
+      "step": 4082
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.0256717205047607,
+      "learning_rate": 9.965160370422839e-06,
+      "loss": 0.4293,
+      "step": 4083
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 6.773927688598633,
+      "learning_rate": 9.965129292883042e-06,
+      "loss": 0.7143,
+      "step": 4084
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.170433521270752,
+      "learning_rate": 9.965098201537082e-06,
+      "loss": 0.8091,
+      "step": 4085
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 14.0272216796875,
+      "learning_rate": 9.965067096385043e-06,
+      "loss": 0.6966,
+      "step": 4086
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.1861355304718018,
+      "learning_rate": 9.965035977427013e-06,
+      "loss": 0.4617,
+      "step": 4087
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.0759973526000977,
+      "learning_rate": 9.965004844663077e-06,
+      "loss": 0.8594,
+      "step": 4088
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.385244369506836,
+      "learning_rate": 9.964973698093323e-06,
+      "loss": 0.6419,
+      "step": 4089
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.6298112869262695,
+      "learning_rate": 9.964942537717836e-06,
+      "loss": 0.7495,
+      "step": 4090
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.463477373123169,
+      "learning_rate": 9.964911363536705e-06,
+      "loss": 0.6913,
+      "step": 4091
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 5.683345317840576,
+      "learning_rate": 9.964880175550016e-06,
+      "loss": 0.7705,
+      "step": 4092
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.213413953781128,
+      "learning_rate": 9.964848973757854e-06,
+      "loss": 0.9449,
+      "step": 4093
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.348139762878418,
+      "learning_rate": 9.964817758160307e-06,
+      "loss": 0.8236,
+      "step": 4094
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.218090772628784,
+      "learning_rate": 9.964786528757463e-06,
+      "loss": 0.5524,
+      "step": 4095
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.9766305685043335,
+      "learning_rate": 9.964755285549406e-06,
+      "loss": 0.8164,
+      "step": 4096
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.8570424318313599,
+      "learning_rate": 9.964724028536224e-06,
+      "loss": 0.6521,
+      "step": 4097
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.816025733947754,
+      "learning_rate": 9.964692757718005e-06,
+      "loss": 0.6279,
+      "step": 4098
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.5865626335144043,
+      "learning_rate": 9.964661473094835e-06,
+      "loss": 0.6083,
+      "step": 4099
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.7129228115081787,
+      "learning_rate": 9.964630174666801e-06,
+      "loss": 0.6651,
+      "step": 4100
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.1239233016967773,
+      "learning_rate": 9.96459886243399e-06,
+      "loss": 0.7327,
+      "step": 4101
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.556074857711792,
+      "learning_rate": 9.96456753639649e-06,
+      "loss": 0.7854,
+      "step": 4102
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.150684356689453,
+      "learning_rate": 9.964536196554388e-06,
+      "loss": 0.747,
+      "step": 4103
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.088534116744995,
+      "learning_rate": 9.96450484290777e-06,
+      "loss": 0.6724,
+      "step": 4104
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.4339237213134766,
+      "learning_rate": 9.964473475456724e-06,
+      "loss": 0.4838,
+      "step": 4105
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.493379592895508,
+      "learning_rate": 9.964442094201337e-06,
+      "loss": 0.7242,
+      "step": 4106
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.4669618606567383,
+      "learning_rate": 9.964410699141696e-06,
+      "loss": 0.5909,
+      "step": 4107
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.349329948425293,
+      "learning_rate": 9.96437929027789e-06,
+      "loss": 0.7934,
+      "step": 4108
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.922513008117676,
+      "learning_rate": 9.964347867610002e-06,
+      "loss": 0.5034,
+      "step": 4109
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.6577670574188232,
+      "learning_rate": 9.964316431138124e-06,
+      "loss": 0.7755,
+      "step": 4110
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.2898049354553223,
+      "learning_rate": 9.964284980862342e-06,
+      "loss": 0.4364,
+      "step": 4111
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.8028844594955444,
+      "learning_rate": 9.964253516782743e-06,
+      "loss": 0.6456,
+      "step": 4112
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.8760316371917725,
+      "learning_rate": 9.964222038899414e-06,
+      "loss": 0.7019,
+      "step": 4113
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.6807326078414917,
+      "learning_rate": 9.964190547212443e-06,
+      "loss": 0.8227,
+      "step": 4114
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.469268321990967,
+      "learning_rate": 9.964159041721919e-06,
+      "loss": 0.6381,
+      "step": 4115
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 11.171354293823242,
+      "learning_rate": 9.964127522427927e-06,
+      "loss": 0.716,
+      "step": 4116
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.871593713760376,
+      "learning_rate": 9.964095989330557e-06,
+      "loss": 0.5962,
+      "step": 4117
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.1708123683929443,
+      "learning_rate": 9.964064442429894e-06,
+      "loss": 0.8019,
+      "step": 4118
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.941433072090149,
+      "learning_rate": 9.964032881726028e-06,
+      "loss": 0.7423,
+      "step": 4119
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.327138662338257,
+      "learning_rate": 9.964001307219047e-06,
+      "loss": 0.7829,
+      "step": 4120
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.155722141265869,
+      "learning_rate": 9.963969718909038e-06,
+      "loss": 0.5581,
+      "step": 4121
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.296255350112915,
+      "learning_rate": 9.963938116796088e-06,
+      "loss": 0.5201,
+      "step": 4122
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.1049489974975586,
+      "learning_rate": 9.963906500880286e-06,
+      "loss": 0.5513,
+      "step": 4123
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.5424633026123047,
+      "learning_rate": 9.96387487116172e-06,
+      "loss": 0.5374,
+      "step": 4124
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.1684632301330566,
+      "learning_rate": 9.963843227640478e-06,
+      "loss": 0.8645,
+      "step": 4125
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.6077632904052734,
+      "learning_rate": 9.963811570316644e-06,
+      "loss": 0.4316,
+      "step": 4126
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.2994701862335205,
+      "learning_rate": 9.963779899190314e-06,
+      "loss": 0.8104,
+      "step": 4127
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 5.770123481750488,
+      "learning_rate": 9.963748214261569e-06,
+      "loss": 0.9081,
+      "step": 4128
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.9905736446380615,
+      "learning_rate": 9.963716515530499e-06,
+      "loss": 0.5937,
+      "step": 4129
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.7328028678894043,
+      "learning_rate": 9.963684802997194e-06,
+      "loss": 0.8371,
+      "step": 4130
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.136484146118164,
+      "learning_rate": 9.96365307666174e-06,
+      "loss": 0.8065,
+      "step": 4131
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.5462191104888916,
+      "learning_rate": 9.96362133652423e-06,
+      "loss": 0.5822,
+      "step": 4132
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.3265600204467773,
+      "learning_rate": 9.963589582584745e-06,
+      "loss": 0.5824,
+      "step": 4133
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.8612644672393799,
+      "learning_rate": 9.963557814843376e-06,
+      "loss": 0.5609,
+      "step": 4134
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 4.563834190368652,
+      "learning_rate": 9.963526033300214e-06,
+      "loss": 0.9078,
+      "step": 4135
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.8214573860168457,
+      "learning_rate": 9.963494237955345e-06,
+      "loss": 0.8426,
+      "step": 4136
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.894415020942688,
+      "learning_rate": 9.963462428808859e-06,
+      "loss": 0.5696,
+      "step": 4137
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 9.98132038116455,
+      "learning_rate": 9.963430605860841e-06,
+      "loss": 0.8943,
+      "step": 4138
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.0428414344787598,
+      "learning_rate": 9.963398769111384e-06,
+      "loss": 0.53,
+      "step": 4139
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.705104112625122,
+      "learning_rate": 9.963366918560574e-06,
+      "loss": 0.6377,
+      "step": 4140
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.3193845748901367,
+      "learning_rate": 9.963335054208498e-06,
+      "loss": 0.4421,
+      "step": 4141
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.641770839691162,
+      "learning_rate": 9.963303176055249e-06,
+      "loss": 0.634,
+      "step": 4142
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.057893991470337,
+      "learning_rate": 9.963271284100913e-06,
+      "loss": 0.5999,
+      "step": 4143
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.549633741378784,
+      "learning_rate": 9.963239378345577e-06,
+      "loss": 0.7169,
+      "step": 4144
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.993129253387451,
+      "learning_rate": 9.963207458789332e-06,
+      "loss": 0.5732,
+      "step": 4145
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.1561503410339355,
+      "learning_rate": 9.963175525432267e-06,
+      "loss": 0.6056,
+      "step": 4146
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 4.462357521057129,
+      "learning_rate": 9.96314357827447e-06,
+      "loss": 0.9051,
+      "step": 4147
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.041468858718872,
+      "learning_rate": 9.963111617316031e-06,
+      "loss": 0.6258,
+      "step": 4148
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.6355953216552734,
+      "learning_rate": 9.963079642557036e-06,
+      "loss": 0.7943,
+      "step": 4149
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.95989990234375,
+      "learning_rate": 9.963047653997577e-06,
+      "loss": 0.613,
+      "step": 4150
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.467931032180786,
+      "learning_rate": 9.963015651637739e-06,
+      "loss": 0.5674,
+      "step": 4151
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.31902813911438,
+      "learning_rate": 9.962983635477617e-06,
+      "loss": 0.447,
+      "step": 4152
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.0820953845977783,
+      "learning_rate": 9.962951605517295e-06,
+      "loss": 0.6835,
+      "step": 4153
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.383979558944702,
+      "learning_rate": 9.962919561756865e-06,
+      "loss": 0.5906,
+      "step": 4154
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.86838698387146,
+      "learning_rate": 9.962887504196413e-06,
+      "loss": 0.5061,
+      "step": 4155
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.092585325241089,
+      "learning_rate": 9.962855432836031e-06,
+      "loss": 0.6795,
+      "step": 4156
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.124666690826416,
+      "learning_rate": 9.962823347675807e-06,
+      "loss": 0.6791,
+      "step": 4157
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.922250986099243,
+      "learning_rate": 9.96279124871583e-06,
+      "loss": 0.696,
+      "step": 4158
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.2063117027282715,
+      "learning_rate": 9.962759135956191e-06,
+      "loss": 0.8459,
+      "step": 4159
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.336318254470825,
+      "learning_rate": 9.962727009396978e-06,
+      "loss": 0.5757,
+      "step": 4160
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.4707329273223877,
+      "learning_rate": 9.96269486903828e-06,
+      "loss": 0.5957,
+      "step": 4161
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 4.810973167419434,
+      "learning_rate": 9.962662714880185e-06,
+      "loss": 1.0375,
+      "step": 4162
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.2198312282562256,
+      "learning_rate": 9.962630546922785e-06,
+      "loss": 0.934,
+      "step": 4163
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.1595187187194824,
+      "learning_rate": 9.96259836516617e-06,
+      "loss": 0.6202,
+      "step": 4164
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.245544195175171,
+      "learning_rate": 9.962566169610427e-06,
+      "loss": 0.6394,
+      "step": 4165
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 4.148227214813232,
+      "learning_rate": 9.962533960255645e-06,
+      "loss": 0.6769,
+      "step": 4166
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.8139610290527344,
+      "learning_rate": 9.962501737101916e-06,
+      "loss": 0.5431,
+      "step": 4167
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.0185177326202393,
+      "learning_rate": 9.962469500149328e-06,
+      "loss": 0.5364,
+      "step": 4168
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.998375177383423,
+      "learning_rate": 9.962437249397973e-06,
+      "loss": 0.6416,
+      "step": 4169
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.6662890911102295,
+      "learning_rate": 9.962404984847938e-06,
+      "loss": 0.5455,
+      "step": 4170
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.040706157684326,
+      "learning_rate": 9.962372706499312e-06,
+      "loss": 0.772,
+      "step": 4171
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.4740700721740723,
+      "learning_rate": 9.96234041435219e-06,
+      "loss": 0.8263,
+      "step": 4172
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.0132017135620117,
+      "learning_rate": 9.962308108406656e-06,
+      "loss": 0.6403,
+      "step": 4173
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.912207007408142,
+      "learning_rate": 9.962275788662803e-06,
+      "loss": 0.6307,
+      "step": 4174
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.3906166553497314,
+      "learning_rate": 9.96224345512072e-06,
+      "loss": 0.9499,
+      "step": 4175
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.3247761726379395,
+      "learning_rate": 9.962211107780495e-06,
+      "loss": 0.8506,
+      "step": 4176
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.1472578048706055,
+      "learning_rate": 9.962178746642222e-06,
+      "loss": 0.8167,
+      "step": 4177
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.149146318435669,
+      "learning_rate": 9.962146371705987e-06,
+      "loss": 0.7409,
+      "step": 4178
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.648944139480591,
+      "learning_rate": 9.962113982971884e-06,
+      "loss": 0.8023,
+      "step": 4179
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.7051212787628174,
+      "learning_rate": 9.962081580439998e-06,
+      "loss": 0.7401,
+      "step": 4180
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.772208571434021,
+      "learning_rate": 9.962049164110424e-06,
+      "loss": 0.9832,
+      "step": 4181
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.5189850330352783,
+      "learning_rate": 9.96201673398325e-06,
+      "loss": 0.7614,
+      "step": 4182
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.5415680408477783,
+      "learning_rate": 9.961984290058566e-06,
+      "loss": 0.6693,
+      "step": 4183
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.862574815750122,
+      "learning_rate": 9.961951832336463e-06,
+      "loss": 0.7369,
+      "step": 4184
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.8046135902404785,
+      "learning_rate": 9.961919360817029e-06,
+      "loss": 0.5421,
+      "step": 4185
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.278655767440796,
+      "learning_rate": 9.961886875500358e-06,
+      "loss": 0.6176,
+      "step": 4186
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.1569483280181885,
+      "learning_rate": 9.961854376386537e-06,
+      "loss": 0.7383,
+      "step": 4187
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.2315449714660645,
+      "learning_rate": 9.961821863475658e-06,
+      "loss": 0.9026,
+      "step": 4188
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.90385365486145,
+      "learning_rate": 9.961789336767813e-06,
+      "loss": 0.7673,
+      "step": 4189
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.9811304807662964,
+      "learning_rate": 9.961756796263088e-06,
+      "loss": 0.5328,
+      "step": 4190
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.763513445854187,
+      "learning_rate": 9.961724241961577e-06,
+      "loss": 0.6854,
+      "step": 4191
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.6635560989379883,
+      "learning_rate": 9.96169167386337e-06,
+      "loss": 0.7275,
+      "step": 4192
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.9395370483398438,
+      "learning_rate": 9.961659091968557e-06,
+      "loss": 0.4594,
+      "step": 4193
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.225831985473633,
+      "learning_rate": 9.961626496277228e-06,
+      "loss": 0.6286,
+      "step": 4194
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.48808217048645,
+      "learning_rate": 9.961593886789476e-06,
+      "loss": 0.8074,
+      "step": 4195
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 4.248265743255615,
+      "learning_rate": 9.961561263505388e-06,
+      "loss": 0.8959,
+      "step": 4196
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.8165781497955322,
+      "learning_rate": 9.961528626425057e-06,
+      "loss": 0.5318,
+      "step": 4197
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 5.135164737701416,
+      "learning_rate": 9.961495975548575e-06,
+      "loss": 0.5475,
+      "step": 4198
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.0998694896698,
+      "learning_rate": 9.96146331087603e-06,
+      "loss": 0.7311,
+      "step": 4199
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 4.030489921569824,
+      "learning_rate": 9.961430632407515e-06,
+      "loss": 0.6109,
+      "step": 4200
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.063354015350342,
+      "learning_rate": 9.961397940143121e-06,
+      "loss": 0.7199,
+      "step": 4201
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.89033579826355,
+      "learning_rate": 9.961365234082935e-06,
+      "loss": 0.6116,
+      "step": 4202
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.1797971725463867,
+      "learning_rate": 9.961332514227053e-06,
+      "loss": 0.7142,
+      "step": 4203
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.5174825191497803,
+      "learning_rate": 9.961299780575563e-06,
+      "loss": 0.771,
+      "step": 4204
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.109442710876465,
+      "learning_rate": 9.961267033128558e-06,
+      "loss": 0.7653,
+      "step": 4205
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.0197534561157227,
+      "learning_rate": 9.961234271886126e-06,
+      "loss": 0.591,
+      "step": 4206
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.1631898880004883,
+      "learning_rate": 9.961201496848363e-06,
+      "loss": 0.9339,
+      "step": 4207
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.9946482181549072,
+      "learning_rate": 9.961168708015353e-06,
+      "loss": 0.8138,
+      "step": 4208
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.296081781387329,
+      "learning_rate": 9.961135905387194e-06,
+      "loss": 0.7937,
+      "step": 4209
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.234243869781494,
+      "learning_rate": 9.961103088963976e-06,
+      "loss": 0.655,
+      "step": 4210
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.2589752674102783,
+      "learning_rate": 9.961070258745787e-06,
+      "loss": 0.4116,
+      "step": 4211
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.4142675399780273,
+      "learning_rate": 9.96103741473272e-06,
+      "loss": 0.6861,
+      "step": 4212
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 4.434939861297607,
+      "learning_rate": 9.961004556924865e-06,
+      "loss": 0.645,
+      "step": 4213
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.5229711532592773,
+      "learning_rate": 9.960971685322315e-06,
+      "loss": 0.7718,
+      "step": 4214
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 4.280956268310547,
+      "learning_rate": 9.960938799925165e-06,
+      "loss": 0.8638,
+      "step": 4215
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.9856510162353516,
+      "learning_rate": 9.960905900733498e-06,
+      "loss": 0.5071,
+      "step": 4216
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.3093149662017822,
+      "learning_rate": 9.960872987747412e-06,
+      "loss": 0.6516,
+      "step": 4217
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.2740814685821533,
+      "learning_rate": 9.960840060966996e-06,
+      "loss": 0.646,
+      "step": 4218
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.6405436992645264,
+      "learning_rate": 9.960807120392344e-06,
+      "loss": 0.7695,
+      "step": 4219
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.345813512802124,
+      "learning_rate": 9.960774166023545e-06,
+      "loss": 0.5583,
+      "step": 4220
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.9342050552368164,
+      "learning_rate": 9.960741197860691e-06,
+      "loss": 0.7381,
+      "step": 4221
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.9291741847991943,
+      "learning_rate": 9.960708215903873e-06,
+      "loss": 0.7036,
+      "step": 4222
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.9539326429367065,
+      "learning_rate": 9.960675220153185e-06,
+      "loss": 0.7379,
+      "step": 4223
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.2943739891052246,
+      "learning_rate": 9.960642210608717e-06,
+      "loss": 0.559,
+      "step": 4224
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.182102680206299,
+      "learning_rate": 9.960609187270562e-06,
+      "loss": 0.7412,
+      "step": 4225
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.2537667751312256,
+      "learning_rate": 9.96057615013881e-06,
+      "loss": 0.8304,
+      "step": 4226
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.0077147483825684,
+      "learning_rate": 9.960543099213555e-06,
+      "loss": 0.7577,
+      "step": 4227
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.3090662956237793,
+      "learning_rate": 9.960510034494888e-06,
+      "loss": 0.7826,
+      "step": 4228
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.3472557067871094,
+      "learning_rate": 9.9604769559829e-06,
+      "loss": 0.7417,
+      "step": 4229
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 4.331810474395752,
+      "learning_rate": 9.960443863677685e-06,
+      "loss": 0.8661,
+      "step": 4230
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.7373998165130615,
+      "learning_rate": 9.960410757579334e-06,
+      "loss": 0.6233,
+      "step": 4231
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 4.546141624450684,
+      "learning_rate": 9.960377637687937e-06,
+      "loss": 0.7895,
+      "step": 4232
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.10659122467041,
+      "learning_rate": 9.96034450400359e-06,
+      "loss": 0.454,
+      "step": 4233
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.1499085426330566,
+      "learning_rate": 9.960311356526381e-06,
+      "loss": 0.5765,
+      "step": 4234
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.778010129928589,
+      "learning_rate": 9.960278195256405e-06,
+      "loss": 0.8049,
+      "step": 4235
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.7123115062713623,
+      "learning_rate": 9.960245020193755e-06,
+      "loss": 0.7003,
+      "step": 4236
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.3868327140808105,
+      "learning_rate": 9.96021183133852e-06,
+      "loss": 0.6183,
+      "step": 4237
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.3773093223571777,
+      "learning_rate": 9.960178628690796e-06,
+      "loss": 0.8746,
+      "step": 4238
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.5245072841644287,
+      "learning_rate": 9.96014541225067e-06,
+      "loss": 0.8152,
+      "step": 4239
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.3266477584838867,
+      "learning_rate": 9.96011218201824e-06,
+      "loss": 0.7115,
+      "step": 4240
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.408111333847046,
+      "learning_rate": 9.960078937993597e-06,
+      "loss": 0.5759,
+      "step": 4241
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.773261308670044,
+      "learning_rate": 9.960045680176831e-06,
+      "loss": 0.7152,
+      "step": 4242
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 4.212543964385986,
+      "learning_rate": 9.960012408568037e-06,
+      "loss": 0.8297,
+      "step": 4243
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.9050710201263428,
+      "learning_rate": 9.959979123167305e-06,
+      "loss": 0.6697,
+      "step": 4244
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.8523976802825928,
+      "learning_rate": 9.95994582397473e-06,
+      "loss": 0.52,
+      "step": 4245
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.845640778541565,
+      "learning_rate": 9.959912510990405e-06,
+      "loss": 0.7113,
+      "step": 4246
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.371216297149658,
+      "learning_rate": 9.95987918421442e-06,
+      "loss": 0.5454,
+      "step": 4247
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.2688872814178467,
+      "learning_rate": 9.95984584364687e-06,
+      "loss": 0.7655,
+      "step": 4248
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 4.245182514190674,
+      "learning_rate": 9.959812489287845e-06,
+      "loss": 0.8323,
+      "step": 4249
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.8868913650512695,
+      "learning_rate": 9.959779121137441e-06,
+      "loss": 0.6497,
+      "step": 4250
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.9050774574279785,
+      "learning_rate": 9.959745739195748e-06,
+      "loss": 0.7584,
+      "step": 4251
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.7407269477844238,
+      "learning_rate": 9.959712343462861e-06,
+      "loss": 0.448,
+      "step": 4252
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.8896427154541016,
+      "learning_rate": 9.959678933938872e-06,
+      "loss": 0.561,
+      "step": 4253
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.166759729385376,
+      "learning_rate": 9.959645510623873e-06,
+      "loss": 0.7249,
+      "step": 4254
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.866291046142578,
+      "learning_rate": 9.959612073517958e-06,
+      "loss": 0.6276,
+      "step": 4255
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.4367849826812744,
+      "learning_rate": 9.959578622621219e-06,
+      "loss": 0.6761,
+      "step": 4256
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.389683246612549,
+      "learning_rate": 9.959545157933752e-06,
+      "loss": 0.8057,
+      "step": 4257
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.4163401126861572,
+      "learning_rate": 9.959511679455648e-06,
+      "loss": 0.5679,
+      "step": 4258
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.0791049003601074,
+      "learning_rate": 9.959478187186997e-06,
+      "loss": 0.9145,
+      "step": 4259
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.1793885231018066,
+      "learning_rate": 9.959444681127897e-06,
+      "loss": 0.7098,
+      "step": 4260
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.1749589443206787,
+      "learning_rate": 9.959411161278438e-06,
+      "loss": 0.7126,
+      "step": 4261
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.2853195667266846,
+      "learning_rate": 9.959377627638716e-06,
+      "loss": 0.8113,
+      "step": 4262
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.3096327781677246,
+      "learning_rate": 9.959344080208821e-06,
+      "loss": 0.6213,
+      "step": 4263
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 4.879098415374756,
+      "learning_rate": 9.959310518988847e-06,
+      "loss": 0.6041,
+      "step": 4264
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.0756514072418213,
+      "learning_rate": 9.959276943978891e-06,
+      "loss": 0.7094,
+      "step": 4265
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.1437478065490723,
+      "learning_rate": 9.959243355179042e-06,
+      "loss": 0.9678,
+      "step": 4266
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.52341890335083,
+      "learning_rate": 9.959209752589396e-06,
+      "loss": 0.6663,
+      "step": 4267
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.0826334953308105,
+      "learning_rate": 9.959176136210044e-06,
+      "loss": 0.7031,
+      "step": 4268
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.172523260116577,
+      "learning_rate": 9.959142506041082e-06,
+      "loss": 0.528,
+      "step": 4269
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.48612117767334,
+      "learning_rate": 9.959108862082603e-06,
+      "loss": 0.4973,
+      "step": 4270
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 4.109440803527832,
+      "learning_rate": 9.959075204334699e-06,
+      "loss": 0.7,
+      "step": 4271
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.7127625942230225,
+      "learning_rate": 9.959041532797464e-06,
+      "loss": 0.6143,
+      "step": 4272
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.1177780628204346,
+      "learning_rate": 9.959007847470994e-06,
+      "loss": 0.6411,
+      "step": 4273
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.5394604206085205,
+      "learning_rate": 9.958974148355379e-06,
+      "loss": 0.8676,
+      "step": 4274
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.2470245361328125,
+      "learning_rate": 9.958940435450715e-06,
+      "loss": 0.4826,
+      "step": 4275
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.8863602876663208,
+      "learning_rate": 9.958906708757097e-06,
+      "loss": 0.7777,
+      "step": 4276
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.8547558784484863,
+      "learning_rate": 9.958872968274615e-06,
+      "loss": 0.7724,
+      "step": 4277
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.7033488750457764,
+      "learning_rate": 9.958839214003366e-06,
+      "loss": 0.4822,
+      "step": 4278
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.6643908023834229,
+      "learning_rate": 9.958805445943442e-06,
+      "loss": 0.5119,
+      "step": 4279
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.2920703887939453,
+      "learning_rate": 9.958771664094938e-06,
+      "loss": 0.7247,
+      "step": 4280
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 4.185944080352783,
+      "learning_rate": 9.958737868457949e-06,
+      "loss": 0.5411,
+      "step": 4281
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.761450171470642,
+      "learning_rate": 9.958704059032566e-06,
+      "loss": 0.6195,
+      "step": 4282
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.840770721435547,
+      "learning_rate": 9.958670235818883e-06,
+      "loss": 0.7328,
+      "step": 4283
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.535762071609497,
+      "learning_rate": 9.958636398816998e-06,
+      "loss": 0.6258,
+      "step": 4284
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.4563181400299072,
+      "learning_rate": 9.958602548027004e-06,
+      "loss": 0.615,
+      "step": 4285
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.772088885307312,
+      "learning_rate": 9.958568683448992e-06,
+      "loss": 0.8082,
+      "step": 4286
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.1635782718658447,
+      "learning_rate": 9.958534805083057e-06,
+      "loss": 0.7264,
+      "step": 4287
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.751779794692993,
+      "learning_rate": 9.958500912929296e-06,
+      "loss": 0.6306,
+      "step": 4288
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.7404630184173584,
+      "learning_rate": 9.9584670069878e-06,
+      "loss": 0.6568,
+      "step": 4289
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.2358126640319824,
+      "learning_rate": 9.958433087258666e-06,
+      "loss": 0.6385,
+      "step": 4290
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.0649592876434326,
+      "learning_rate": 9.958399153741987e-06,
+      "loss": 0.9118,
+      "step": 4291
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.7391057014465332,
+      "learning_rate": 9.958365206437857e-06,
+      "loss": 0.4487,
+      "step": 4292
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.3765909671783447,
+      "learning_rate": 9.95833124534637e-06,
+      "loss": 0.6518,
+      "step": 4293
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.636833429336548,
+      "learning_rate": 9.958297270467623e-06,
+      "loss": 0.9138,
+      "step": 4294
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.7810895442962646,
+      "learning_rate": 9.958263281801708e-06,
+      "loss": 0.6815,
+      "step": 4295
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.072493076324463,
+      "learning_rate": 9.958229279348718e-06,
+      "loss": 0.8939,
+      "step": 4296
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.694897174835205,
+      "learning_rate": 9.958195263108752e-06,
+      "loss": 0.4146,
+      "step": 4297
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.983981132507324,
+      "learning_rate": 9.958161233081904e-06,
+      "loss": 0.853,
+      "step": 4298
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.4437789916992188,
+      "learning_rate": 9.958127189268264e-06,
+      "loss": 0.6458,
+      "step": 4299
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.0323734283447266,
+      "learning_rate": 9.95809313166793e-06,
+      "loss": 0.6252,
+      "step": 4300
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.0483272075653076,
+      "learning_rate": 9.958059060280999e-06,
+      "loss": 0.7653,
+      "step": 4301
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.6274259090423584,
+      "learning_rate": 9.95802497510756e-06,
+      "loss": 0.5498,
+      "step": 4302
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.432832717895508,
+      "learning_rate": 9.957990876147712e-06,
+      "loss": 0.5599,
+      "step": 4303
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.218311309814453,
+      "learning_rate": 9.957956763401549e-06,
+      "loss": 0.8902,
+      "step": 4304
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.31227707862854,
+      "learning_rate": 9.957922636869165e-06,
+      "loss": 0.606,
+      "step": 4305
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.5144450664520264,
+      "learning_rate": 9.957888496550657e-06,
+      "loss": 0.6314,
+      "step": 4306
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.8615596294403076,
+      "learning_rate": 9.957854342446116e-06,
+      "loss": 0.551,
+      "step": 4307
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.6775660514831543,
+      "learning_rate": 9.95782017455564e-06,
+      "loss": 0.6054,
+      "step": 4308
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.635499119758606,
+      "learning_rate": 9.957785992879322e-06,
+      "loss": 0.5182,
+      "step": 4309
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.4244167804718018,
+      "learning_rate": 9.957751797417261e-06,
+      "loss": 0.5728,
+      "step": 4310
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.6702594757080078,
+      "learning_rate": 9.957717588169549e-06,
+      "loss": 0.5939,
+      "step": 4311
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 4.568820476531982,
+      "learning_rate": 9.957683365136281e-06,
+      "loss": 0.6324,
+      "step": 4312
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.3833696842193604,
+      "learning_rate": 9.957649128317553e-06,
+      "loss": 0.4586,
+      "step": 4313
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.990870714187622,
+      "learning_rate": 9.957614877713459e-06,
+      "loss": 0.8495,
+      "step": 4314
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.1335089206695557,
+      "learning_rate": 9.957580613324097e-06,
+      "loss": 0.5241,
+      "step": 4315
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.7090611457824707,
+      "learning_rate": 9.95754633514956e-06,
+      "loss": 0.6372,
+      "step": 4316
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.9443564414978027,
+      "learning_rate": 9.957512043189943e-06,
+      "loss": 0.4062,
+      "step": 4317
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 6.8100104331970215,
+      "learning_rate": 9.957477737445342e-06,
+      "loss": 0.7483,
+      "step": 4318
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.158851385116577,
+      "learning_rate": 9.957443417915853e-06,
+      "loss": 0.9579,
+      "step": 4319
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.2600674629211426,
+      "learning_rate": 9.957409084601571e-06,
+      "loss": 0.5366,
+      "step": 4320
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.164746046066284,
+      "learning_rate": 9.957374737502593e-06,
+      "loss": 0.9195,
+      "step": 4321
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.295663595199585,
+      "learning_rate": 9.957340376619009e-06,
+      "loss": 0.9882,
+      "step": 4322
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.5598580837249756,
+      "learning_rate": 9.957306001950922e-06,
+      "loss": 0.8342,
+      "step": 4323
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.2428929805755615,
+      "learning_rate": 9.957271613498423e-06,
+      "loss": 0.51,
+      "step": 4324
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.240332841873169,
+      "learning_rate": 9.957237211261608e-06,
+      "loss": 0.6201,
+      "step": 4325
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.8827261924743652,
+      "learning_rate": 9.957202795240573e-06,
+      "loss": 0.717,
+      "step": 4326
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 4.321632385253906,
+      "learning_rate": 9.957168365435415e-06,
+      "loss": 0.6687,
+      "step": 4327
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.6981258392333984,
+      "learning_rate": 9.95713392184623e-06,
+      "loss": 0.9753,
+      "step": 4328
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.6819491386413574,
+      "learning_rate": 9.95709946447311e-06,
+      "loss": 0.7142,
+      "step": 4329
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.456742286682129,
+      "learning_rate": 9.957064993316154e-06,
+      "loss": 0.6172,
+      "step": 4330
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.195719003677368,
+      "learning_rate": 9.957030508375458e-06,
+      "loss": 0.7876,
+      "step": 4331
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.4106650352478027,
+      "learning_rate": 9.956996009651116e-06,
+      "loss": 0.4063,
+      "step": 4332
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.8110711574554443,
+      "learning_rate": 9.956961497143227e-06,
+      "loss": 0.687,
+      "step": 4333
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.655051350593567,
+      "learning_rate": 9.956926970851884e-06,
+      "loss": 0.5794,
+      "step": 4334
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.822110652923584,
+      "learning_rate": 9.956892430777183e-06,
+      "loss": 0.5311,
+      "step": 4335
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.411543130874634,
+      "learning_rate": 9.956857876919223e-06,
+      "loss": 0.8413,
+      "step": 4336
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.9796642065048218,
+      "learning_rate": 9.956823309278096e-06,
+      "loss": 0.4436,
+      "step": 4337
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.140577793121338,
+      "learning_rate": 9.9567887278539e-06,
+      "loss": 0.704,
+      "step": 4338
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 4.671708106994629,
+      "learning_rate": 9.956754132646734e-06,
+      "loss": 0.7013,
+      "step": 4339
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.494694948196411,
+      "learning_rate": 9.95671952365669e-06,
+      "loss": 0.9459,
+      "step": 4340
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.043595314025879,
+      "learning_rate": 9.956684900883867e-06,
+      "loss": 0.9394,
+      "step": 4341
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.6816718578338623,
+      "learning_rate": 9.95665026432836e-06,
+      "loss": 0.5212,
+      "step": 4342
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.6351635456085205,
+      "learning_rate": 9.956615613990263e-06,
+      "loss": 0.5587,
+      "step": 4343
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 5.226381778717041,
+      "learning_rate": 9.956580949869677e-06,
+      "loss": 0.7237,
+      "step": 4344
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.6003406047821045,
+      "learning_rate": 9.956546271966697e-06,
+      "loss": 1.0897,
+      "step": 4345
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.665383219718933,
+      "learning_rate": 9.956511580281417e-06,
+      "loss": 0.7698,
+      "step": 4346
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.5199904441833496,
+      "learning_rate": 9.956476874813933e-06,
+      "loss": 0.6171,
+      "step": 4347
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.658006429672241,
+      "learning_rate": 9.956442155564347e-06,
+      "loss": 0.8699,
+      "step": 4348
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.786522150039673,
+      "learning_rate": 9.95640742253275e-06,
+      "loss": 0.6305,
+      "step": 4349
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.444585084915161,
+      "learning_rate": 9.956372675719241e-06,
+      "loss": 0.7078,
+      "step": 4350
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.532688856124878,
+      "learning_rate": 9.956337915123918e-06,
+      "loss": 0.3968,
+      "step": 4351
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.347348213195801,
+      "learning_rate": 9.956303140746875e-06,
+      "loss": 0.7621,
+      "step": 4352
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 5.492105960845947,
+      "learning_rate": 9.956268352588209e-06,
+      "loss": 0.7181,
+      "step": 4353
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.4065916538238525,
+      "learning_rate": 9.956233550648018e-06,
+      "loss": 0.2697,
+      "step": 4354
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.727410316467285,
+      "learning_rate": 9.956198734926396e-06,
+      "loss": 0.6101,
+      "step": 4355
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.201371431350708,
+      "learning_rate": 9.956163905423446e-06,
+      "loss": 1.0129,
+      "step": 4356
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.274209976196289,
+      "learning_rate": 9.956129062139257e-06,
+      "loss": 0.7201,
+      "step": 4357
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 4.19360876083374,
+      "learning_rate": 9.956094205073932e-06,
+      "loss": 0.673,
+      "step": 4358
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.6469430923461914,
+      "learning_rate": 9.956059334227565e-06,
+      "loss": 0.3803,
+      "step": 4359
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.030027389526367,
+      "learning_rate": 9.956024449600253e-06,
+      "loss": 0.9864,
+      "step": 4360
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.1024460792541504,
+      "learning_rate": 9.955989551192095e-06,
+      "loss": 0.8749,
+      "step": 4361
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.4566783905029297,
+      "learning_rate": 9.955954639003185e-06,
+      "loss": 0.6172,
+      "step": 4362
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.2548952102661133,
+      "learning_rate": 9.955919713033623e-06,
+      "loss": 0.6949,
+      "step": 4363
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.128054141998291,
+      "learning_rate": 9.955884773283503e-06,
+      "loss": 0.5448,
+      "step": 4364
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 4.686114311218262,
+      "learning_rate": 9.955849819752926e-06,
+      "loss": 0.6982,
+      "step": 4365
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.746959924697876,
+      "learning_rate": 9.955814852441987e-06,
+      "loss": 0.883,
+      "step": 4366
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.741231918334961,
+      "learning_rate": 9.955779871350782e-06,
+      "loss": 0.8242,
+      "step": 4367
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.3496451377868652,
+      "learning_rate": 9.955744876479411e-06,
+      "loss": 0.6743,
+      "step": 4368
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.25701904296875,
+      "learning_rate": 9.95570986782797e-06,
+      "loss": 0.8652,
+      "step": 4369
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.243532419204712,
+      "learning_rate": 9.955674845396555e-06,
+      "loss": 0.6868,
+      "step": 4370
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.9704519510269165,
+      "learning_rate": 9.955639809185264e-06,
+      "loss": 0.8855,
+      "step": 4371
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.082423686981201,
+      "learning_rate": 9.955604759194197e-06,
+      "loss": 0.475,
+      "step": 4372
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.205739736557007,
+      "learning_rate": 9.955569695423449e-06,
+      "loss": 0.8835,
+      "step": 4373
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.1090500354766846,
+      "learning_rate": 9.955534617873118e-06,
+      "loss": 0.5142,
+      "step": 4374
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.8841286897659302,
+      "learning_rate": 9.955499526543302e-06,
+      "loss": 0.6697,
+      "step": 4375
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.016406536102295,
+      "learning_rate": 9.955464421434099e-06,
+      "loss": 0.7041,
+      "step": 4376
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.499962568283081,
+      "learning_rate": 9.955429302545602e-06,
+      "loss": 0.8386,
+      "step": 4377
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.443427085876465,
+      "learning_rate": 9.955394169877917e-06,
+      "loss": 0.878,
+      "step": 4378
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.0615124702453613,
+      "learning_rate": 9.955359023431134e-06,
+      "loss": 0.5091,
+      "step": 4379
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.765887975692749,
+      "learning_rate": 9.955323863205356e-06,
+      "loss": 0.3562,
+      "step": 4380
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.9288345575332642,
+      "learning_rate": 9.955288689200677e-06,
+      "loss": 0.8778,
+      "step": 4381
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.7143924236297607,
+      "learning_rate": 9.955253501417198e-06,
+      "loss": 0.7405,
+      "step": 4382
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.771982192993164,
+      "learning_rate": 9.955218299855015e-06,
+      "loss": 0.6506,
+      "step": 4383
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.6850194931030273,
+      "learning_rate": 9.955183084514226e-06,
+      "loss": 0.6056,
+      "step": 4384
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 4.061385154724121,
+      "learning_rate": 9.955147855394929e-06,
+      "loss": 0.9289,
+      "step": 4385
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.632709503173828,
+      "learning_rate": 9.955112612497222e-06,
+      "loss": 0.8538,
+      "step": 4386
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.5530284643173218,
+      "learning_rate": 9.955077355821202e-06,
+      "loss": 0.5461,
+      "step": 4387
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.0520315170288086,
+      "learning_rate": 9.955042085366969e-06,
+      "loss": 0.7086,
+      "step": 4388
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.5742175579071045,
+      "learning_rate": 9.95500680113462e-06,
+      "loss": 0.6351,
+      "step": 4389
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.8654009103775024,
+      "learning_rate": 9.954971503124253e-06,
+      "loss": 0.5856,
+      "step": 4390
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.585601806640625,
+      "learning_rate": 9.954936191335968e-06,
+      "loss": 0.4777,
+      "step": 4391
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.8292558193206787,
+      "learning_rate": 9.954900865769862e-06,
+      "loss": 0.4958,
+      "step": 4392
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.427245855331421,
+      "learning_rate": 9.95486552642603e-06,
+      "loss": 0.8529,
+      "step": 4393
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.6497583389282227,
+      "learning_rate": 9.954830173304574e-06,
+      "loss": 0.61,
+      "step": 4394
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.43811297416687,
+      "learning_rate": 9.954794806405592e-06,
+      "loss": 0.4722,
+      "step": 4395
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.3818750381469727,
+      "learning_rate": 9.95475942572918e-06,
+      "loss": 0.7714,
+      "step": 4396
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.2397964000701904,
+      "learning_rate": 9.954724031275442e-06,
+      "loss": 0.9462,
+      "step": 4397
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.254883050918579,
+      "learning_rate": 9.95468862304447e-06,
+      "loss": 0.9557,
+      "step": 4398
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.6621493101119995,
+      "learning_rate": 9.954653201036365e-06,
+      "loss": 0.3988,
+      "step": 4399
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.5306999683380127,
+      "learning_rate": 9.954617765251226e-06,
+      "loss": 0.7431,
+      "step": 4400
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.8947768211364746,
+      "learning_rate": 9.95458231568915e-06,
+      "loss": 0.7469,
+      "step": 4401
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.459313154220581,
+      "learning_rate": 9.954546852350237e-06,
+      "loss": 0.7084,
+      "step": 4402
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.7316043376922607,
+      "learning_rate": 9.954511375234587e-06,
+      "loss": 0.6769,
+      "step": 4403
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.7871296405792236,
+      "learning_rate": 9.954475884342295e-06,
+      "loss": 0.7392,
+      "step": 4404
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.337103843688965,
+      "learning_rate": 9.954440379673462e-06,
+      "loss": 0.6482,
+      "step": 4405
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 4.503066062927246,
+      "learning_rate": 9.954404861228186e-06,
+      "loss": 0.4051,
+      "step": 4406
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.3295071125030518,
+      "learning_rate": 9.954369329006566e-06,
+      "loss": 0.5511,
+      "step": 4407
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.504920244216919,
+      "learning_rate": 9.954333783008702e-06,
+      "loss": 0.629,
+      "step": 4408
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.6946189403533936,
+      "learning_rate": 9.954298223234691e-06,
+      "loss": 0.8388,
+      "step": 4409
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.4540042877197266,
+      "learning_rate": 9.954262649684633e-06,
+      "loss": 0.7094,
+      "step": 4410
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.3355612754821777,
+      "learning_rate": 9.954227062358626e-06,
+      "loss": 0.9043,
+      "step": 4411
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.3449013233184814,
+      "learning_rate": 9.95419146125677e-06,
+      "loss": 0.6491,
+      "step": 4412
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.8953137397766113,
+      "learning_rate": 9.954155846379162e-06,
+      "loss": 0.6205,
+      "step": 4413
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.3652048110961914,
+      "learning_rate": 9.954120217725904e-06,
+      "loss": 0.7975,
+      "step": 4414
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.030034065246582,
+      "learning_rate": 9.954084575297091e-06,
+      "loss": 0.7777,
+      "step": 4415
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 4.555300712585449,
+      "learning_rate": 9.954048919092826e-06,
+      "loss": 0.7579,
+      "step": 4416
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 4.515816688537598,
+      "learning_rate": 9.954013249113208e-06,
+      "loss": 0.5933,
+      "step": 4417
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.7940139770507812,
+      "learning_rate": 9.953977565358334e-06,
+      "loss": 0.3236,
+      "step": 4418
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.532424211502075,
+      "learning_rate": 9.953941867828304e-06,
+      "loss": 0.6848,
+      "step": 4419
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.1353037357330322,
+      "learning_rate": 9.95390615652322e-06,
+      "loss": 0.6695,
+      "step": 4420
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.2381327152252197,
+      "learning_rate": 9.953870431443175e-06,
+      "loss": 0.645,
+      "step": 4421
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.3253085613250732,
+      "learning_rate": 9.953834692588276e-06,
+      "loss": 0.461,
+      "step": 4422
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.1000590324401855,
+      "learning_rate": 9.953798939958615e-06,
+      "loss": 0.5813,
+      "step": 4423
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 5.710429668426514,
+      "learning_rate": 9.953763173554296e-06,
+      "loss": 0.5603,
+      "step": 4424
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.5738437175750732,
+      "learning_rate": 9.953727393375418e-06,
+      "loss": 0.581,
+      "step": 4425
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.55318546295166,
+      "learning_rate": 9.953691599422079e-06,
+      "loss": 0.6369,
+      "step": 4426
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.3656585216522217,
+      "learning_rate": 9.953655791694378e-06,
+      "loss": 0.7691,
+      "step": 4427
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.1484055519104004,
+      "learning_rate": 9.953619970192418e-06,
+      "loss": 0.7824,
+      "step": 4428
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.6562983989715576,
+      "learning_rate": 9.953584134916296e-06,
+      "loss": 0.8219,
+      "step": 4429
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.153705596923828,
+      "learning_rate": 9.953548285866112e-06,
+      "loss": 0.8205,
+      "step": 4430
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.9729610681533813,
+      "learning_rate": 9.953512423041967e-06,
+      "loss": 0.7145,
+      "step": 4431
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.255657911300659,
+      "learning_rate": 9.953476546443958e-06,
+      "loss": 0.6855,
+      "step": 4432
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.8260226249694824,
+      "learning_rate": 9.953440656072188e-06,
+      "loss": 0.6435,
+      "step": 4433
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.1711134910583496,
+      "learning_rate": 9.953404751926754e-06,
+      "loss": 0.9468,
+      "step": 4434
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 4.624855041503906,
+      "learning_rate": 9.953368834007757e-06,
+      "loss": 0.8864,
+      "step": 4435
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.2353267669677734,
+      "learning_rate": 9.953332902315297e-06,
+      "loss": 0.7424,
+      "step": 4436
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.9670426845550537,
+      "learning_rate": 9.953296956849474e-06,
+      "loss": 0.7578,
+      "step": 4437
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 4.347574234008789,
+      "learning_rate": 9.953260997610388e-06,
+      "loss": 0.7004,
+      "step": 4438
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 4.582129955291748,
+      "learning_rate": 9.953225024598138e-06,
+      "loss": 0.8405,
+      "step": 4439
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.486527919769287,
+      "learning_rate": 9.953189037812824e-06,
+      "loss": 0.6292,
+      "step": 4440
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 4.106201171875,
+      "learning_rate": 9.953153037254548e-06,
+      "loss": 0.7993,
+      "step": 4441
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.3615517616271973,
+      "learning_rate": 9.953117022923408e-06,
+      "loss": 0.7377,
+      "step": 4442
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.9672743082046509,
+      "learning_rate": 9.953080994819507e-06,
+      "loss": 0.6773,
+      "step": 4443
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.532397747039795,
+      "learning_rate": 9.953044952942941e-06,
+      "loss": 0.8041,
+      "step": 4444
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.549219846725464,
+      "learning_rate": 9.953008897293814e-06,
+      "loss": 0.8533,
+      "step": 4445
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.955871820449829,
+      "learning_rate": 9.952972827872223e-06,
+      "loss": 0.5205,
+      "step": 4446
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.1470766067504883,
+      "learning_rate": 9.952936744678271e-06,
+      "loss": 0.8183,
+      "step": 4447
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.9886462688446045,
+      "learning_rate": 9.952900647712059e-06,
+      "loss": 0.6661,
+      "step": 4448
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.9696447849273682,
+      "learning_rate": 9.952864536973683e-06,
+      "loss": 0.5293,
+      "step": 4449
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.276066541671753,
+      "learning_rate": 9.952828412463247e-06,
+      "loss": 0.6717,
+      "step": 4450
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 4.220001220703125,
+      "learning_rate": 9.95279227418085e-06,
+      "loss": 0.5114,
+      "step": 4451
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 5.267399787902832,
+      "learning_rate": 9.952756122126595e-06,
+      "loss": 0.7852,
+      "step": 4452
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.772671699523926,
+      "learning_rate": 9.952719956300578e-06,
+      "loss": 0.5839,
+      "step": 4453
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.294400453567505,
+      "learning_rate": 9.952683776702903e-06,
+      "loss": 0.803,
+      "step": 4454
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.558297872543335,
+      "learning_rate": 9.95264758333367e-06,
+      "loss": 0.6162,
+      "step": 4455
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.446107864379883,
+      "learning_rate": 9.95261137619298e-06,
+      "loss": 0.745,
+      "step": 4456
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.7180423736572266,
+      "learning_rate": 9.952575155280933e-06,
+      "loss": 0.7152,
+      "step": 4457
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.8293907642364502,
+      "learning_rate": 9.952538920597629e-06,
+      "loss": 0.4353,
+      "step": 4458
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.599339008331299,
+      "learning_rate": 9.95250267214317e-06,
+      "loss": 0.7456,
+      "step": 4459
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.936233639717102,
+      "learning_rate": 9.952466409917656e-06,
+      "loss": 0.6057,
+      "step": 4460
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.4535601139068604,
+      "learning_rate": 9.952430133921188e-06,
+      "loss": 0.5415,
+      "step": 4461
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.02602219581604,
+      "learning_rate": 9.952393844153868e-06,
+      "loss": 0.6254,
+      "step": 4462
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.0099830627441406,
+      "learning_rate": 9.952357540615794e-06,
+      "loss": 0.5452,
+      "step": 4463
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.586475133895874,
+      "learning_rate": 9.95232122330707e-06,
+      "loss": 0.7535,
+      "step": 4464
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.6873605251312256,
+      "learning_rate": 9.952284892227797e-06,
+      "loss": 0.8442,
+      "step": 4465
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.8214168548583984,
+      "learning_rate": 9.952248547378073e-06,
+      "loss": 0.7895,
+      "step": 4466
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.2530717849731445,
+      "learning_rate": 9.952212188758003e-06,
+      "loss": 0.8073,
+      "step": 4467
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.8450177907943726,
+      "learning_rate": 9.952175816367684e-06,
+      "loss": 0.7903,
+      "step": 4468
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.342724561691284,
+      "learning_rate": 9.95213943020722e-06,
+      "loss": 0.7286,
+      "step": 4469
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.6935484409332275,
+      "learning_rate": 9.952103030276712e-06,
+      "loss": 0.9296,
+      "step": 4470
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.24713397026062,
+      "learning_rate": 9.95206661657626e-06,
+      "loss": 0.7979,
+      "step": 4471
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.2316782474517822,
+      "learning_rate": 9.952030189105966e-06,
+      "loss": 0.6617,
+      "step": 4472
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.5178160667419434,
+      "learning_rate": 9.951993747865931e-06,
+      "loss": 0.6294,
+      "step": 4473
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.746028184890747,
+      "learning_rate": 9.951957292856256e-06,
+      "loss": 0.4257,
+      "step": 4474
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.3325016498565674,
+      "learning_rate": 9.951920824077043e-06,
+      "loss": 0.5288,
+      "step": 4475
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.4893383979797363,
+      "learning_rate": 9.951884341528392e-06,
+      "loss": 0.7454,
+      "step": 4476
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 4.6685791015625,
+      "learning_rate": 9.951847845210409e-06,
+      "loss": 0.8485,
+      "step": 4477
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.4772868156433105,
+      "learning_rate": 9.951811335123189e-06,
+      "loss": 0.7605,
+      "step": 4478
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.803125262260437,
+      "learning_rate": 9.95177481126684e-06,
+      "loss": 0.5092,
+      "step": 4479
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.893618583679199,
+      "learning_rate": 9.951738273641457e-06,
+      "loss": 0.7283,
+      "step": 4480
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.1514058113098145,
+      "learning_rate": 9.951701722247146e-06,
+      "loss": 0.5831,
+      "step": 4481
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 4.157179355621338,
+      "learning_rate": 9.951665157084008e-06,
+      "loss": 0.6931,
+      "step": 4482
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.9646263122558594,
+      "learning_rate": 9.951628578152143e-06,
+      "loss": 0.7227,
+      "step": 4483
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 4.407310485839844,
+      "learning_rate": 9.951591985451655e-06,
+      "loss": 0.5165,
+      "step": 4484
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.1635289192199707,
+      "learning_rate": 9.951555378982642e-06,
+      "loss": 0.7056,
+      "step": 4485
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.614683151245117,
+      "learning_rate": 9.951518758745211e-06,
+      "loss": 0.5345,
+      "step": 4486
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.1402785778045654,
+      "learning_rate": 9.951482124739461e-06,
+      "loss": 0.6828,
+      "step": 4487
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.006190776824951,
+      "learning_rate": 9.951445476965493e-06,
+      "loss": 0.6348,
+      "step": 4488
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.627002477645874,
+      "learning_rate": 9.951408815423412e-06,
+      "loss": 0.8206,
+      "step": 4489
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 8.536477088928223,
+      "learning_rate": 9.951372140113316e-06,
+      "loss": 0.9891,
+      "step": 4490
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.0050947666168213,
+      "learning_rate": 9.951335451035308e-06,
+      "loss": 0.7959,
+      "step": 4491
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.8893439769744873,
+      "learning_rate": 9.951298748189493e-06,
+      "loss": 0.8679,
+      "step": 4492
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.369466781616211,
+      "learning_rate": 9.95126203157597e-06,
+      "loss": 0.6106,
+      "step": 4493
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.7835898399353027,
+      "learning_rate": 9.951225301194844e-06,
+      "loss": 0.6582,
+      "step": 4494
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.298527956008911,
+      "learning_rate": 9.951188557046213e-06,
+      "loss": 0.7227,
+      "step": 4495
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.0831174850463867,
+      "learning_rate": 9.951151799130181e-06,
+      "loss": 0.5646,
+      "step": 4496
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.4117093086242676,
+      "learning_rate": 9.951115027446852e-06,
+      "loss": 0.4047,
+      "step": 4497
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.2198257446289062,
+      "learning_rate": 9.951078241996327e-06,
+      "loss": 0.5015,
+      "step": 4498
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.6581311225891113,
+      "learning_rate": 9.951041442778708e-06,
+      "loss": 0.7886,
+      "step": 4499
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.528676748275757,
+      "learning_rate": 9.951004629794097e-06,
+      "loss": 0.7428,
+      "step": 4500
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.1060614585876465,
+      "learning_rate": 9.950967803042598e-06,
+      "loss": 0.7116,
+      "step": 4501
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.0921225547790527,
+      "learning_rate": 9.95093096252431e-06,
+      "loss": 0.4549,
+      "step": 4502
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.4001049995422363,
+      "learning_rate": 9.950894108239339e-06,
+      "loss": 0.6146,
+      "step": 4503
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.380741596221924,
+      "learning_rate": 9.950857240187785e-06,
+      "loss": 0.8684,
+      "step": 4504
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.9658496379852295,
+      "learning_rate": 9.950820358369753e-06,
+      "loss": 0.8523,
+      "step": 4505
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.649883508682251,
+      "learning_rate": 9.950783462785343e-06,
+      "loss": 0.8401,
+      "step": 4506
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.6973693370819092,
+      "learning_rate": 9.95074655343466e-06,
+      "loss": 0.6844,
+      "step": 4507
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.44763445854187,
+      "learning_rate": 9.950709630317805e-06,
+      "loss": 0.9913,
+      "step": 4508
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 4.923973083496094,
+      "learning_rate": 9.95067269343488e-06,
+      "loss": 0.7618,
+      "step": 4509
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.339350938796997,
+      "learning_rate": 9.95063574278599e-06,
+      "loss": 0.5921,
+      "step": 4510
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.4222025871276855,
+      "learning_rate": 9.950598778371237e-06,
+      "loss": 0.7497,
+      "step": 4511
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.0100347995758057,
+      "learning_rate": 9.950561800190723e-06,
+      "loss": 0.5217,
+      "step": 4512
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.0708365440368652,
+      "learning_rate": 9.95052480824455e-06,
+      "loss": 0.6346,
+      "step": 4513
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.0591070652008057,
+      "learning_rate": 9.950487802532823e-06,
+      "loss": 0.7577,
+      "step": 4514
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.205476999282837,
+      "learning_rate": 9.950450783055645e-06,
+      "loss": 0.562,
+      "step": 4515
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.144158363342285,
+      "learning_rate": 9.950413749813115e-06,
+      "loss": 0.7761,
+      "step": 4516
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 4.236269950866699,
+      "learning_rate": 9.950376702805342e-06,
+      "loss": 0.6009,
+      "step": 4517
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.1379470825195312,
+      "learning_rate": 9.950339642032424e-06,
+      "loss": 0.9859,
+      "step": 4518
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.3920419216156006,
+      "learning_rate": 9.950302567494466e-06,
+      "loss": 0.6515,
+      "step": 4519
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.7413592338562012,
+      "learning_rate": 9.950265479191571e-06,
+      "loss": 0.6444,
+      "step": 4520
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.1671833992004395,
+      "learning_rate": 9.950228377123842e-06,
+      "loss": 0.6947,
+      "step": 4521
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.198517322540283,
+      "learning_rate": 9.950191261291382e-06,
+      "loss": 0.8171,
+      "step": 4522
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.3514716625213623,
+      "learning_rate": 9.950154131694297e-06,
+      "loss": 0.9022,
+      "step": 4523
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.773061752319336,
+      "learning_rate": 9.950116988332685e-06,
+      "loss": 0.7971,
+      "step": 4524
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.90751314163208,
+      "learning_rate": 9.950079831206652e-06,
+      "loss": 0.7406,
+      "step": 4525
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.0999362468719482,
+      "learning_rate": 9.950042660316304e-06,
+      "loss": 0.6619,
+      "step": 4526
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 4.0861921310424805,
+      "learning_rate": 9.95000547566174e-06,
+      "loss": 0.7753,
+      "step": 4527
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.201368570327759,
+      "learning_rate": 9.949968277243064e-06,
+      "loss": 0.7025,
+      "step": 4528
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.8712977170944214,
+      "learning_rate": 9.949931065060381e-06,
+      "loss": 0.8544,
+      "step": 4529
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.562293767929077,
+      "learning_rate": 9.949893839113794e-06,
+      "loss": 0.8689,
+      "step": 4530
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.904276132583618,
+      "learning_rate": 9.949856599403406e-06,
+      "loss": 0.8032,
+      "step": 4531
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.546116352081299,
+      "learning_rate": 9.949819345929324e-06,
+      "loss": 0.7547,
+      "step": 4532
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.9465208053588867,
+      "learning_rate": 9.949782078691646e-06,
+      "loss": 0.8694,
+      "step": 4533
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.5947093963623047,
+      "learning_rate": 9.949744797690479e-06,
+      "loss": 0.5884,
+      "step": 4534
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.349161148071289,
+      "learning_rate": 9.949707502925926e-06,
+      "loss": 0.6244,
+      "step": 4535
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 7.712183952331543,
+      "learning_rate": 9.949670194398092e-06,
+      "loss": 0.8141,
+      "step": 4536
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.113443613052368,
+      "learning_rate": 9.949632872107077e-06,
+      "loss": 0.8183,
+      "step": 4537
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.1710920333862305,
+      "learning_rate": 9.949595536052986e-06,
+      "loss": 0.7862,
+      "step": 4538
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.187605619430542,
+      "learning_rate": 9.949558186235928e-06,
+      "loss": 0.4913,
+      "step": 4539
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.629940986633301,
+      "learning_rate": 9.949520822656e-06,
+      "loss": 0.8286,
+      "step": 4540
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.512603282928467,
+      "learning_rate": 9.94948344531331e-06,
+      "loss": 0.7307,
+      "step": 4541
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.075310468673706,
+      "learning_rate": 9.949446054207961e-06,
+      "loss": 0.8987,
+      "step": 4542
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.569091558456421,
+      "learning_rate": 9.949408649340055e-06,
+      "loss": 0.8482,
+      "step": 4543
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.6625571250915527,
+      "learning_rate": 9.949371230709698e-06,
+      "loss": 0.5871,
+      "step": 4544
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.8315281867980957,
+      "learning_rate": 9.949333798316994e-06,
+      "loss": 0.9843,
+      "step": 4545
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.417480945587158,
+      "learning_rate": 9.949296352162048e-06,
+      "loss": 0.6274,
+      "step": 4546
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.346991539001465,
+      "learning_rate": 9.949258892244962e-06,
+      "loss": 0.8221,
+      "step": 4547
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.476606845855713,
+      "learning_rate": 9.949221418565839e-06,
+      "loss": 0.7429,
+      "step": 4548
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.528369188308716,
+      "learning_rate": 9.949183931124789e-06,
+      "loss": 0.7602,
+      "step": 4549
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.3488376140594482,
+      "learning_rate": 9.94914642992191e-06,
+      "loss": 0.8141,
+      "step": 4550
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.702948570251465,
+      "learning_rate": 9.94910891495731e-06,
+      "loss": 0.7664,
+      "step": 4551
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.052417516708374,
+      "learning_rate": 9.949071386231091e-06,
+      "loss": 0.8885,
+      "step": 4552
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.238112449645996,
+      "learning_rate": 9.949033843743359e-06,
+      "loss": 0.9429,
+      "step": 4553
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.8093466758728027,
+      "learning_rate": 9.948996287494217e-06,
+      "loss": 0.8713,
+      "step": 4554
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.2707746028900146,
+      "learning_rate": 9.948958717483772e-06,
+      "loss": 0.7323,
+      "step": 4555
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.264005661010742,
+      "learning_rate": 9.948921133712126e-06,
+      "loss": 0.7127,
+      "step": 4556
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.6044180393218994,
+      "learning_rate": 9.948883536179383e-06,
+      "loss": 0.5957,
+      "step": 4557
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 4.045000076293945,
+      "learning_rate": 9.948845924885651e-06,
+      "loss": 0.6261,
+      "step": 4558
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.082382917404175,
+      "learning_rate": 9.94880829983103e-06,
+      "loss": 0.8431,
+      "step": 4559
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.563427209854126,
+      "learning_rate": 9.948770661015628e-06,
+      "loss": 0.6856,
+      "step": 4560
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.1807518005371094,
+      "learning_rate": 9.948733008439549e-06,
+      "loss": 0.796,
+      "step": 4561
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.0324490070343018,
+      "learning_rate": 9.948695342102897e-06,
+      "loss": 0.5508,
+      "step": 4562
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.4410178661346436,
+      "learning_rate": 9.948657662005779e-06,
+      "loss": 0.538,
+      "step": 4563
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.7726789712905884,
+      "learning_rate": 9.948619968148295e-06,
+      "loss": 0.4139,
+      "step": 4564
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.048436164855957,
+      "learning_rate": 9.948582260530552e-06,
+      "loss": 0.7647,
+      "step": 4565
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 4.5688676834106445,
+      "learning_rate": 9.948544539152658e-06,
+      "loss": 0.613,
+      "step": 4566
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.054417610168457,
+      "learning_rate": 9.948506804014715e-06,
+      "loss": 0.4895,
+      "step": 4567
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.528533697128296,
+      "learning_rate": 9.948469055116829e-06,
+      "loss": 0.8218,
+      "step": 4568
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.494943618774414,
+      "learning_rate": 9.948431292459102e-06,
+      "loss": 0.5794,
+      "step": 4569
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.703901767730713,
+      "learning_rate": 9.948393516041644e-06,
+      "loss": 0.5933,
+      "step": 4570
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.3636105060577393,
+      "learning_rate": 9.948355725864555e-06,
+      "loss": 0.7646,
+      "step": 4571
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.2633779048919678,
+      "learning_rate": 9.948317921927943e-06,
+      "loss": 0.7097,
+      "step": 4572
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.993544101715088,
+      "learning_rate": 9.948280104231913e-06,
+      "loss": 0.5803,
+      "step": 4573
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.514707088470459,
+      "learning_rate": 9.94824227277657e-06,
+      "loss": 0.8737,
+      "step": 4574
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.443721055984497,
+      "learning_rate": 9.948204427562018e-06,
+      "loss": 0.6632,
+      "step": 4575
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.9067559242248535,
+      "learning_rate": 9.948166568588363e-06,
+      "loss": 0.4312,
+      "step": 4576
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.0406272411346436,
+      "learning_rate": 9.948128695855712e-06,
+      "loss": 0.9741,
+      "step": 4577
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.7141332626342773,
+      "learning_rate": 9.948090809364167e-06,
+      "loss": 0.6359,
+      "step": 4578
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.7380709648132324,
+      "learning_rate": 9.948052909113836e-06,
+      "loss": 0.6884,
+      "step": 4579
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.328913927078247,
+      "learning_rate": 9.948014995104823e-06,
+      "loss": 0.5997,
+      "step": 4580
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 4.084225654602051,
+      "learning_rate": 9.947977067337233e-06,
+      "loss": 0.8317,
+      "step": 4581
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.8457109928131104,
+      "learning_rate": 9.947939125811174e-06,
+      "loss": 0.6249,
+      "step": 4582
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.551089286804199,
+      "learning_rate": 9.947901170526751e-06,
+      "loss": 0.8354,
+      "step": 4583
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.9500795602798462,
+      "learning_rate": 9.947863201484065e-06,
+      "loss": 0.6613,
+      "step": 4584
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.331857442855835,
+      "learning_rate": 9.947825218683227e-06,
+      "loss": 0.8244,
+      "step": 4585
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.0119147300720215,
+      "learning_rate": 9.94778722212434e-06,
+      "loss": 0.8913,
+      "step": 4586
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.0987935066223145,
+      "learning_rate": 9.94774921180751e-06,
+      "loss": 0.7531,
+      "step": 4587
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.360445976257324,
+      "learning_rate": 9.947711187732843e-06,
+      "loss": 0.7085,
+      "step": 4588
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.222137212753296,
+      "learning_rate": 9.947673149900445e-06,
+      "loss": 0.4789,
+      "step": 4589
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.7723000049591064,
+      "learning_rate": 9.947635098310422e-06,
+      "loss": 0.5617,
+      "step": 4590
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.9535109996795654,
+      "learning_rate": 9.947597032962878e-06,
+      "loss": 0.5075,
+      "step": 4591
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.8720703125,
+      "learning_rate": 9.947558953857921e-06,
+      "loss": 0.7404,
+      "step": 4592
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.0715529918670654,
+      "learning_rate": 9.947520860995656e-06,
+      "loss": 0.7351,
+      "step": 4593
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 6.1314496994018555,
+      "learning_rate": 9.947482754376188e-06,
+      "loss": 0.4575,
+      "step": 4594
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.8292670249938965,
+      "learning_rate": 9.947444633999624e-06,
+      "loss": 0.5852,
+      "step": 4595
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.1071181297302246,
+      "learning_rate": 9.94740649986607e-06,
+      "loss": 0.5374,
+      "step": 4596
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.709031343460083,
+      "learning_rate": 9.947368351975632e-06,
+      "loss": 0.4208,
+      "step": 4597
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.4364869594573975,
+      "learning_rate": 9.947330190328414e-06,
+      "loss": 0.7205,
+      "step": 4598
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.6601388454437256,
+      "learning_rate": 9.947292014924528e-06,
+      "loss": 0.9066,
+      "step": 4599
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.091409206390381,
+      "learning_rate": 9.947253825764074e-06,
+      "loss": 0.546,
+      "step": 4600
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.4492545127868652,
+      "learning_rate": 9.94721562284716e-06,
+      "loss": 0.4125,
+      "step": 4601
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.211904525756836,
+      "learning_rate": 9.947177406173893e-06,
+      "loss": 0.7988,
+      "step": 4602
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 4.3205366134643555,
+      "learning_rate": 9.94713917574438e-06,
+      "loss": 0.8678,
+      "step": 4603
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.4642856121063232,
+      "learning_rate": 9.947100931558724e-06,
+      "loss": 0.5019,
+      "step": 4604
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.792256236076355,
+      "learning_rate": 9.947062673617035e-06,
+      "loss": 0.7741,
+      "step": 4605
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.251046895980835,
+      "learning_rate": 9.947024401919418e-06,
+      "loss": 0.6682,
+      "step": 4606
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.963545322418213,
+      "learning_rate": 9.946986116465978e-06,
+      "loss": 0.4072,
+      "step": 4607
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.312361001968384,
+      "learning_rate": 9.946947817256825e-06,
+      "loss": 0.7833,
+      "step": 4608
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.410613536834717,
+      "learning_rate": 9.946909504292062e-06,
+      "loss": 0.6843,
+      "step": 4609
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 5.584275245666504,
+      "learning_rate": 9.946871177571796e-06,
+      "loss": 0.8453,
+      "step": 4610
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.885301113128662,
+      "learning_rate": 9.946832837096135e-06,
+      "loss": 0.9159,
+      "step": 4611
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.407822608947754,
+      "learning_rate": 9.946794482865185e-06,
+      "loss": 0.7287,
+      "step": 4612
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.133981227874756,
+      "learning_rate": 9.946756114879054e-06,
+      "loss": 0.7486,
+      "step": 4613
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.447777271270752,
+      "learning_rate": 9.946717733137847e-06,
+      "loss": 0.4422,
+      "step": 4614
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.7374529838562012,
+      "learning_rate": 9.94667933764167e-06,
+      "loss": 0.6231,
+      "step": 4615
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.023871898651123,
+      "learning_rate": 9.946640928390631e-06,
+      "loss": 0.4273,
+      "step": 4616
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.1126744747161865,
+      "learning_rate": 9.946602505384837e-06,
+      "loss": 0.6569,
+      "step": 4617
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.9421789646148682,
+      "learning_rate": 9.946564068624394e-06,
+      "loss": 0.4884,
+      "step": 4618
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.8040287494659424,
+      "learning_rate": 9.946525618109409e-06,
+      "loss": 0.6714,
+      "step": 4619
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.982764482498169,
+      "learning_rate": 9.94648715383999e-06,
+      "loss": 0.7646,
+      "step": 4620
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.8689751625061035,
+      "learning_rate": 9.946448675816243e-06,
+      "loss": 0.894,
+      "step": 4621
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.756438732147217,
+      "learning_rate": 9.946410184038275e-06,
+      "loss": 0.7302,
+      "step": 4622
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.853548765182495,
+      "learning_rate": 9.946371678506196e-06,
+      "loss": 0.706,
+      "step": 4623
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.9505815505981445,
+      "learning_rate": 9.946333159220105e-06,
+      "loss": 0.7806,
+      "step": 4624
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.046072244644165,
+      "learning_rate": 9.946294626180118e-06,
+      "loss": 0.6803,
+      "step": 4625
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.860387086868286,
+      "learning_rate": 9.946256079386338e-06,
+      "loss": 0.9337,
+      "step": 4626
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.3342537879943848,
+      "learning_rate": 9.946217518838873e-06,
+      "loss": 0.4803,
+      "step": 4627
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 4.24758243560791,
+      "learning_rate": 9.94617894453783e-06,
+      "loss": 0.5504,
+      "step": 4628
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.9020922183990479,
+      "learning_rate": 9.946140356483315e-06,
+      "loss": 0.7841,
+      "step": 4629
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.14253830909729,
+      "learning_rate": 9.946101754675438e-06,
+      "loss": 0.674,
+      "step": 4630
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.599039316177368,
+      "learning_rate": 9.946063139114304e-06,
+      "loss": 0.6765,
+      "step": 4631
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.0437023639678955,
+      "learning_rate": 9.946024509800022e-06,
+      "loss": 0.5209,
+      "step": 4632
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.345757007598877,
+      "learning_rate": 9.945985866732697e-06,
+      "loss": 0.6988,
+      "step": 4633
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.3846237659454346,
+      "learning_rate": 9.94594720991244e-06,
+      "loss": 0.6792,
+      "step": 4634
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 4.679755210876465,
+      "learning_rate": 9.945908539339357e-06,
+      "loss": 0.8125,
+      "step": 4635
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 5.546121120452881,
+      "learning_rate": 9.945869855013552e-06,
+      "loss": 0.8209,
+      "step": 4636
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.2210514545440674,
+      "learning_rate": 9.94583115693514e-06,
+      "loss": 0.5731,
+      "step": 4637
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.809272289276123,
+      "learning_rate": 9.945792445104222e-06,
+      "loss": 0.6236,
+      "step": 4638
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.2250163555145264,
+      "learning_rate": 9.945753719520908e-06,
+      "loss": 0.8267,
+      "step": 4639
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.6673760414123535,
+      "learning_rate": 9.945714980185304e-06,
+      "loss": 0.3835,
+      "step": 4640
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.7369391918182373,
+      "learning_rate": 9.94567622709752e-06,
+      "loss": 0.6526,
+      "step": 4641
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.910027027130127,
+      "learning_rate": 9.945637460257666e-06,
+      "loss": 0.7401,
+      "step": 4642
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 4.854179859161377,
+      "learning_rate": 9.945598679665845e-06,
+      "loss": 0.5917,
+      "step": 4643
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.448167324066162,
+      "learning_rate": 9.945559885322165e-06,
+      "loss": 0.5604,
+      "step": 4644
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 4.740459442138672,
+      "learning_rate": 9.945521077226739e-06,
+      "loss": 0.618,
+      "step": 4645
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.41033935546875,
+      "learning_rate": 9.945482255379669e-06,
+      "loss": 0.8692,
+      "step": 4646
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.1755359172821045,
+      "learning_rate": 9.945443419781067e-06,
+      "loss": 0.5107,
+      "step": 4647
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.2115490436553955,
+      "learning_rate": 9.945404570431039e-06,
+      "loss": 0.5854,
+      "step": 4648
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 5.1247711181640625,
+      "learning_rate": 9.945365707329693e-06,
+      "loss": 0.36,
+      "step": 4649
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.5223841667175293,
+      "learning_rate": 9.945326830477138e-06,
+      "loss": 0.7247,
+      "step": 4650
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.972712755203247,
+      "learning_rate": 9.94528793987348e-06,
+      "loss": 0.7066,
+      "step": 4651
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 4.509586334228516,
+      "learning_rate": 9.94524903551883e-06,
+      "loss": 0.7307,
+      "step": 4652
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.533968687057495,
+      "learning_rate": 9.945210117413295e-06,
+      "loss": 0.5931,
+      "step": 4653
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.2008323669433594,
+      "learning_rate": 9.945171185556984e-06,
+      "loss": 0.8136,
+      "step": 4654
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.4068641662597656,
+      "learning_rate": 9.945132239950004e-06,
+      "loss": 0.7301,
+      "step": 4655
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.174527645111084,
+      "learning_rate": 9.945093280592464e-06,
+      "loss": 0.5797,
+      "step": 4656
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.5407164096832275,
+      "learning_rate": 9.945054307484471e-06,
+      "loss": 0.7318,
+      "step": 4657
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.7248928546905518,
+      "learning_rate": 9.945015320626135e-06,
+      "loss": 0.9504,
+      "step": 4658
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.4711976051330566,
+      "learning_rate": 9.944976320017563e-06,
+      "loss": 0.4835,
+      "step": 4659
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.649125814437866,
+      "learning_rate": 9.944937305658866e-06,
+      "loss": 0.8127,
+      "step": 4660
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.0824337005615234,
+      "learning_rate": 9.94489827755015e-06,
+      "loss": 0.7719,
+      "step": 4661
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.0519838333129883,
+      "learning_rate": 9.944859235691524e-06,
+      "loss": 0.9005,
+      "step": 4662
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.771104574203491,
+      "learning_rate": 9.944820180083097e-06,
+      "loss": 0.5574,
+      "step": 4663
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 4.899678707122803,
+      "learning_rate": 9.944781110724977e-06,
+      "loss": 0.6417,
+      "step": 4664
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.29390287399292,
+      "learning_rate": 9.944742027617273e-06,
+      "loss": 0.817,
+      "step": 4665
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.4888014793395996,
+      "learning_rate": 9.944702930760094e-06,
+      "loss": 0.5599,
+      "step": 4666
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 4.1209917068481445,
+      "learning_rate": 9.944663820153551e-06,
+      "loss": 0.8271,
+      "step": 4667
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 4.483585357666016,
+      "learning_rate": 9.944624695797746e-06,
+      "loss": 1.0647,
+      "step": 4668
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.9455513954162598,
+      "learning_rate": 9.944585557692795e-06,
+      "loss": 0.9036,
+      "step": 4669
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 5.525146961212158,
+      "learning_rate": 9.944546405838803e-06,
+      "loss": 0.55,
+      "step": 4670
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 5.041517734527588,
+      "learning_rate": 9.94450724023588e-06,
+      "loss": 0.5778,
+      "step": 4671
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.6247007846832275,
+      "learning_rate": 9.944468060884134e-06,
+      "loss": 0.757,
+      "step": 4672
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.3139383792877197,
+      "learning_rate": 9.944428867783675e-06,
+      "loss": 0.5071,
+      "step": 4673
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.6173553466796875,
+      "learning_rate": 9.944389660934612e-06,
+      "loss": 0.6607,
+      "step": 4674
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.8237907886505127,
+      "learning_rate": 9.944350440337051e-06,
+      "loss": 0.6567,
+      "step": 4675
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.841888666152954,
+      "learning_rate": 9.944311205991107e-06,
+      "loss": 0.7558,
+      "step": 4676
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 10.141938209533691,
+      "learning_rate": 9.944271957896883e-06,
+      "loss": 0.483,
+      "step": 4677
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.8440228700637817,
+      "learning_rate": 9.944232696054493e-06,
+      "loss": 0.4582,
+      "step": 4678
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.025592803955078,
+      "learning_rate": 9.944193420464043e-06,
+      "loss": 0.8055,
+      "step": 4679
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.8692013025283813,
+      "learning_rate": 9.944154131125643e-06,
+      "loss": 0.7342,
+      "step": 4680
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 4.1667914390563965,
+      "learning_rate": 9.944114828039404e-06,
+      "loss": 0.4966,
+      "step": 4681
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.5101912021636963,
+      "learning_rate": 9.944075511205433e-06,
+      "loss": 0.7283,
+      "step": 4682
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 4.1637372970581055,
+      "learning_rate": 9.94403618062384e-06,
+      "loss": 0.6963,
+      "step": 4683
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.076295852661133,
+      "learning_rate": 9.943996836294734e-06,
+      "loss": 0.8629,
+      "step": 4684
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.958876132965088,
+      "learning_rate": 9.943957478218225e-06,
+      "loss": 0.7864,
+      "step": 4685
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.9847218990325928,
+      "learning_rate": 9.943918106394422e-06,
+      "loss": 0.5904,
+      "step": 4686
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.6468656063079834,
+      "learning_rate": 9.943878720823435e-06,
+      "loss": 0.6544,
+      "step": 4687
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.078571081161499,
+      "learning_rate": 9.943839321505373e-06,
+      "loss": 0.65,
+      "step": 4688
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.162763833999634,
+      "learning_rate": 9.943799908440346e-06,
+      "loss": 0.7859,
+      "step": 4689
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.801325559616089,
+      "learning_rate": 9.943760481628463e-06,
+      "loss": 0.702,
+      "step": 4690
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 4.304174423217773,
+      "learning_rate": 9.943721041069836e-06,
+      "loss": 0.7935,
+      "step": 4691
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 4.0729899406433105,
+      "learning_rate": 9.94368158676457e-06,
+      "loss": 0.5779,
+      "step": 4692
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.644196033477783,
+      "learning_rate": 9.94364211871278e-06,
+      "loss": 0.4956,
+      "step": 4693
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.1807126998901367,
+      "learning_rate": 9.943602636914573e-06,
+      "loss": 0.7632,
+      "step": 4694
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.0283539295196533,
+      "learning_rate": 9.943563141370058e-06,
+      "loss": 0.7234,
+      "step": 4695
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 4.660898208618164,
+      "learning_rate": 9.943523632079347e-06,
+      "loss": 0.5959,
+      "step": 4696
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.5611698627471924,
+      "learning_rate": 9.943484109042546e-06,
+      "loss": 0.7191,
+      "step": 4697
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.2964773178100586,
+      "learning_rate": 9.94344457225977e-06,
+      "loss": 0.6537,
+      "step": 4698
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 6.2816290855407715,
+      "learning_rate": 9.943405021731126e-06,
+      "loss": 0.6136,
+      "step": 4699
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.1827609539031982,
+      "learning_rate": 9.943365457456724e-06,
+      "loss": 0.5548,
+      "step": 4700
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.857268810272217,
+      "learning_rate": 9.943325879436674e-06,
+      "loss": 0.6805,
+      "step": 4701
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.3766090869903564,
+      "learning_rate": 9.943286287671088e-06,
+      "loss": 0.7076,
+      "step": 4702
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.9535250663757324,
+      "learning_rate": 9.943246682160072e-06,
+      "loss": 0.6428,
+      "step": 4703
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.323967456817627,
+      "learning_rate": 9.943207062903742e-06,
+      "loss": 0.8888,
+      "step": 4704
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.334831476211548,
+      "learning_rate": 9.943167429902202e-06,
+      "loss": 0.5202,
+      "step": 4705
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.7344274520874023,
+      "learning_rate": 9.943127783155569e-06,
+      "loss": 0.5632,
+      "step": 4706
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.3822836875915527,
+      "learning_rate": 9.943088122663947e-06,
+      "loss": 0.6992,
+      "step": 4707
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 4.396145343780518,
+      "learning_rate": 9.943048448427447e-06,
+      "loss": 0.6412,
+      "step": 4708
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.0655229091644287,
+      "learning_rate": 9.943008760446183e-06,
+      "loss": 0.5082,
+      "step": 4709
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.602306842803955,
+      "learning_rate": 9.942969058720264e-06,
+      "loss": 0.6325,
+      "step": 4710
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.6490092277526855,
+      "learning_rate": 9.942929343249797e-06,
+      "loss": 0.7201,
+      "step": 4711
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.9806100130081177,
+      "learning_rate": 9.942889614034898e-06,
+      "loss": 0.8183,
+      "step": 4712
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 4.182120323181152,
+      "learning_rate": 9.942849871075672e-06,
+      "loss": 0.7144,
+      "step": 4713
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 5.893625736236572,
+      "learning_rate": 9.942810114372235e-06,
+      "loss": 0.5219,
+      "step": 4714
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 4.655343532562256,
+      "learning_rate": 9.942770343924693e-06,
+      "loss": 0.9202,
+      "step": 4715
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.4372458457946777,
+      "learning_rate": 9.942730559733159e-06,
+      "loss": 0.7854,
+      "step": 4716
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.9744207859039307,
+      "learning_rate": 9.942690761797743e-06,
+      "loss": 0.8308,
+      "step": 4717
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.1525051593780518,
+      "learning_rate": 9.942650950118555e-06,
+      "loss": 0.4787,
+      "step": 4718
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.7108402252197266,
+      "learning_rate": 9.942611124695705e-06,
+      "loss": 0.8274,
+      "step": 4719
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 4.8674139976501465,
+      "learning_rate": 9.942571285529307e-06,
+      "loss": 0.8957,
+      "step": 4720
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.9887425899505615,
+      "learning_rate": 9.942531432619469e-06,
+      "loss": 0.6599,
+      "step": 4721
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.524495840072632,
+      "learning_rate": 9.942491565966303e-06,
+      "loss": 0.7614,
+      "step": 4722
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.3130147457122803,
+      "learning_rate": 9.942451685569919e-06,
+      "loss": 0.693,
+      "step": 4723
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 4.1479997634887695,
+      "learning_rate": 9.942411791430428e-06,
+      "loss": 0.6855,
+      "step": 4724
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.4121291637420654,
+      "learning_rate": 9.942371883547944e-06,
+      "loss": 0.463,
+      "step": 4725
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.554008722305298,
+      "learning_rate": 9.942331961922571e-06,
+      "loss": 0.6548,
+      "step": 4726
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.6849687099456787,
+      "learning_rate": 9.942292026554428e-06,
+      "loss": 0.3343,
+      "step": 4727
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.053102970123291,
+      "learning_rate": 9.94225207744362e-06,
+      "loss": 0.7139,
+      "step": 4728
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.6069819927215576,
+      "learning_rate": 9.94221211459026e-06,
+      "loss": 0.7934,
+      "step": 4729
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.401580333709717,
+      "learning_rate": 9.942172137994462e-06,
+      "loss": 0.8708,
+      "step": 4730
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.4391226768493652,
+      "learning_rate": 9.942132147656332e-06,
+      "loss": 0.4095,
+      "step": 4731
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.124600887298584,
+      "learning_rate": 9.942092143575985e-06,
+      "loss": 0.6936,
+      "step": 4732
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.6588718891143799,
+      "learning_rate": 9.942052125753531e-06,
+      "loss": 0.6359,
+      "step": 4733
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.8401436805725098,
+      "learning_rate": 9.942012094189081e-06,
+      "loss": 0.8401,
+      "step": 4734
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.1051290035247803,
+      "learning_rate": 9.941972048882748e-06,
+      "loss": 0.7534,
+      "step": 4735
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.586423635482788,
+      "learning_rate": 9.94193198983464e-06,
+      "loss": 0.742,
+      "step": 4736
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.849745273590088,
+      "learning_rate": 9.941891917044873e-06,
+      "loss": 0.5316,
+      "step": 4737
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 6.242190361022949,
+      "learning_rate": 9.941851830513555e-06,
+      "loss": 0.5965,
+      "step": 4738
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 4.125624179840088,
+      "learning_rate": 9.941811730240797e-06,
+      "loss": 0.5205,
+      "step": 4739
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.7347781658172607,
+      "learning_rate": 9.941771616226712e-06,
+      "loss": 0.6438,
+      "step": 4740
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.763796806335449,
+      "learning_rate": 9.941731488471414e-06,
+      "loss": 0.4772,
+      "step": 4741
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.6348023414611816,
+      "learning_rate": 9.94169134697501e-06,
+      "loss": 0.7883,
+      "step": 4742
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.5779857635498047,
+      "learning_rate": 9.941651191737613e-06,
+      "loss": 0.6507,
+      "step": 4743
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.5443007946014404,
+      "learning_rate": 9.941611022759335e-06,
+      "loss": 0.5592,
+      "step": 4744
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.573268175125122,
+      "learning_rate": 9.941570840040288e-06,
+      "loss": 0.727,
+      "step": 4745
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.7651493549346924,
+      "learning_rate": 9.941530643580585e-06,
+      "loss": 0.905,
+      "step": 4746
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.794656276702881,
+      "learning_rate": 9.941490433380337e-06,
+      "loss": 0.5698,
+      "step": 4747
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.688727617263794,
+      "learning_rate": 9.941450209439653e-06,
+      "loss": 0.7649,
+      "step": 4748
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.2670795917510986,
+      "learning_rate": 9.94140997175865e-06,
+      "loss": 0.7925,
+      "step": 4749
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.5244317054748535,
+      "learning_rate": 9.941369720337433e-06,
+      "loss": 0.7028,
+      "step": 4750
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.457014322280884,
+      "learning_rate": 9.94132945517612e-06,
+      "loss": 0.7626,
+      "step": 4751
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.8331663608551025,
+      "learning_rate": 9.941289176274821e-06,
+      "loss": 0.5747,
+      "step": 4752
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.1504878997802734,
+      "learning_rate": 9.941248883633648e-06,
+      "loss": 0.8105,
+      "step": 4753
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 4.11173152923584,
+      "learning_rate": 9.941208577252712e-06,
+      "loss": 0.7027,
+      "step": 4754
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.4263999462127686,
+      "learning_rate": 9.941168257132126e-06,
+      "loss": 0.3799,
+      "step": 4755
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.6036832332611084,
+      "learning_rate": 9.941127923272002e-06,
+      "loss": 0.7222,
+      "step": 4756
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 4.403662204742432,
+      "learning_rate": 9.941087575672453e-06,
+      "loss": 0.7493,
+      "step": 4757
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.1314969062805176,
+      "learning_rate": 9.94104721433359e-06,
+      "loss": 0.5911,
+      "step": 4758
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.8811066150665283,
+      "learning_rate": 9.941006839255526e-06,
+      "loss": 0.507,
+      "step": 4759
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.184332847595215,
+      "learning_rate": 9.940966450438375e-06,
+      "loss": 0.8152,
+      "step": 4760
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.2195260524749756,
+      "learning_rate": 9.940926047882243e-06,
+      "loss": 0.5803,
+      "step": 4761
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.0854926109313965,
+      "learning_rate": 9.94088563158725e-06,
+      "loss": 0.4666,
+      "step": 4762
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.3200793266296387,
+      "learning_rate": 9.940845201553502e-06,
+      "loss": 0.7114,
+      "step": 4763
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.018603801727295,
+      "learning_rate": 9.940804757781117e-06,
+      "loss": 0.9535,
+      "step": 4764
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.6532866954803467,
+      "learning_rate": 9.940764300270203e-06,
+      "loss": 0.8961,
+      "step": 4765
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 4.099346160888672,
+      "learning_rate": 9.940723829020875e-06,
+      "loss": 0.7943,
+      "step": 4766
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 5.446796894073486,
+      "learning_rate": 9.940683344033244e-06,
+      "loss": 0.7903,
+      "step": 4767
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.7425291538238525,
+      "learning_rate": 9.940642845307425e-06,
+      "loss": 0.6585,
+      "step": 4768
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.002866268157959,
+      "learning_rate": 9.94060233284353e-06,
+      "loss": 0.6296,
+      "step": 4769
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.805201768875122,
+      "learning_rate": 9.940561806641669e-06,
+      "loss": 0.404,
+      "step": 4770
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 4.394008636474609,
+      "learning_rate": 9.940521266701955e-06,
+      "loss": 0.6734,
+      "step": 4771
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.7364718914031982,
+      "learning_rate": 9.940480713024504e-06,
+      "loss": 0.8456,
+      "step": 4772
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.2327568531036377,
+      "learning_rate": 9.940440145609427e-06,
+      "loss": 0.5742,
+      "step": 4773
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.4060614109039307,
+      "learning_rate": 9.940399564456836e-06,
+      "loss": 0.5791,
+      "step": 4774
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.6963497400283813,
+      "learning_rate": 9.940358969566845e-06,
+      "loss": 0.6374,
+      "step": 4775
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.6222610473632812,
+      "learning_rate": 9.940318360939566e-06,
+      "loss": 0.9379,
+      "step": 4776
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.8584446907043457,
+      "learning_rate": 9.940277738575112e-06,
+      "loss": 0.9715,
+      "step": 4777
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.7951335906982422,
+      "learning_rate": 9.940237102473598e-06,
+      "loss": 0.9516,
+      "step": 4778
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.5517141819000244,
+      "learning_rate": 9.940196452635133e-06,
+      "loss": 0.761,
+      "step": 4779
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.986991286277771,
+      "learning_rate": 9.940155789059832e-06,
+      "loss": 0.5946,
+      "step": 4780
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.0720741748809814,
+      "learning_rate": 9.940115111747811e-06,
+      "loss": 0.7803,
+      "step": 4781
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 5.724656581878662,
+      "learning_rate": 9.94007442069918e-06,
+      "loss": 0.6649,
+      "step": 4782
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.2298851013183594,
+      "learning_rate": 9.940033715914051e-06,
+      "loss": 0.5213,
+      "step": 4783
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 4.54960823059082,
+      "learning_rate": 9.93999299739254e-06,
+      "loss": 1.0584,
+      "step": 4784
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.7671704292297363,
+      "learning_rate": 9.939952265134758e-06,
+      "loss": 0.6908,
+      "step": 4785
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.9470040798187256,
+      "learning_rate": 9.939911519140821e-06,
+      "loss": 0.7331,
+      "step": 4786
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.1642627716064453,
+      "learning_rate": 9.939870759410839e-06,
+      "loss": 0.7543,
+      "step": 4787
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 5.188140869140625,
+      "learning_rate": 9.939829985944928e-06,
+      "loss": 0.4608,
+      "step": 4788
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.3960108757019043,
+      "learning_rate": 9.939789198743199e-06,
+      "loss": 0.6135,
+      "step": 4789
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.896331310272217,
+      "learning_rate": 9.939748397805768e-06,
+      "loss": 0.6931,
+      "step": 4790
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.57184100151062,
+      "learning_rate": 9.939707583132748e-06,
+      "loss": 0.5671,
+      "step": 4791
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 4.094829082489014,
+      "learning_rate": 9.939666754724248e-06,
+      "loss": 0.6705,
+      "step": 4792
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.699341058731079,
+      "learning_rate": 9.93962591258039e-06,
+      "loss": 0.8717,
+      "step": 4793
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.7310584783554077,
+      "learning_rate": 9.93958505670128e-06,
+      "loss": 0.7585,
+      "step": 4794
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 9.73387336730957,
+      "learning_rate": 9.939544187087034e-06,
+      "loss": 0.6579,
+      "step": 4795
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 4.972394943237305,
+      "learning_rate": 9.939503303737768e-06,
+      "loss": 0.3127,
+      "step": 4796
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.9744744300842285,
+      "learning_rate": 9.93946240665359e-06,
+      "loss": 0.8362,
+      "step": 4797
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.3554437160491943,
+      "learning_rate": 9.93942149583462e-06,
+      "loss": 0.7534,
+      "step": 4798
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.8490102291107178,
+      "learning_rate": 9.93938057128097e-06,
+      "loss": 0.597,
+      "step": 4799
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 5.100355625152588,
+      "learning_rate": 9.939339632992753e-06,
+      "loss": 0.7771,
+      "step": 4800
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 4.287455081939697,
+      "learning_rate": 9.93929868097008e-06,
+      "loss": 0.7327,
+      "step": 4801
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.299839735031128,
+      "learning_rate": 9.939257715213071e-06,
+      "loss": 0.7682,
+      "step": 4802
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.062032461166382,
+      "learning_rate": 9.939216735721835e-06,
+      "loss": 0.6458,
+      "step": 4803
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.053208827972412,
+      "learning_rate": 9.939175742496487e-06,
+      "loss": 0.7405,
+      "step": 4804
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 4.989010334014893,
+      "learning_rate": 9.939134735537142e-06,
+      "loss": 0.5051,
+      "step": 4805
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.9760043621063232,
+      "learning_rate": 9.939093714843913e-06,
+      "loss": 0.701,
+      "step": 4806
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 4.316423416137695,
+      "learning_rate": 9.939052680416916e-06,
+      "loss": 0.7838,
+      "step": 4807
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 29.480920791625977,
+      "learning_rate": 9.939011632256262e-06,
+      "loss": 0.6487,
+      "step": 4808
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.31325626373291,
+      "learning_rate": 9.938970570362068e-06,
+      "loss": 0.5853,
+      "step": 4809
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.9220821857452393,
+      "learning_rate": 9.938929494734446e-06,
+      "loss": 0.7775,
+      "step": 4810
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.0472142696380615,
+      "learning_rate": 9.938888405373514e-06,
+      "loss": 0.6335,
+      "step": 4811
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.6999175548553467,
+      "learning_rate": 9.938847302279381e-06,
+      "loss": 0.7964,
+      "step": 4812
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.407194137573242,
+      "learning_rate": 9.938806185452165e-06,
+      "loss": 0.5601,
+      "step": 4813
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 4.2607831954956055,
+      "learning_rate": 9.938765054891978e-06,
+      "loss": 0.7826,
+      "step": 4814
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.4789698123931885,
+      "learning_rate": 9.938723910598936e-06,
+      "loss": 0.6536,
+      "step": 4815
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.2805449962615967,
+      "learning_rate": 9.938682752573152e-06,
+      "loss": 0.8211,
+      "step": 4816
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.873492956161499,
+      "learning_rate": 9.938641580814743e-06,
+      "loss": 0.7913,
+      "step": 4817
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.5789523124694824,
+      "learning_rate": 9.93860039532382e-06,
+      "loss": 0.6823,
+      "step": 4818
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 4.312883377075195,
+      "learning_rate": 9.938559196100502e-06,
+      "loss": 0.7434,
+      "step": 4819
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.4683022499084473,
+      "learning_rate": 9.938517983144897e-06,
+      "loss": 0.6845,
+      "step": 4820
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.3557472229003906,
+      "learning_rate": 9.938476756457127e-06,
+      "loss": 0.7704,
+      "step": 4821
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.8638062477111816,
+      "learning_rate": 9.938435516037302e-06,
+      "loss": 0.9696,
+      "step": 4822
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 6.566242694854736,
+      "learning_rate": 9.938394261885538e-06,
+      "loss": 0.4878,
+      "step": 4823
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.474410057067871,
+      "learning_rate": 9.938352994001949e-06,
+      "loss": 0.7486,
+      "step": 4824
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 7.0824480056762695,
+      "learning_rate": 9.93831171238665e-06,
+      "loss": 0.6299,
+      "step": 4825
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.927986979484558,
+      "learning_rate": 9.938270417039757e-06,
+      "loss": 0.7812,
+      "step": 4826
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 5.225847244262695,
+      "learning_rate": 9.938229107961383e-06,
+      "loss": 0.4718,
+      "step": 4827
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.7584376335144043,
+      "learning_rate": 9.938187785151644e-06,
+      "loss": 0.687,
+      "step": 4828
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.07785701751709,
+      "learning_rate": 9.938146448610655e-06,
+      "loss": 0.748,
+      "step": 4829
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.8771920204162598,
+      "learning_rate": 9.93810509833853e-06,
+      "loss": 1.0998,
+      "step": 4830
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.865978956222534,
+      "learning_rate": 9.938063734335386e-06,
+      "loss": 0.6985,
+      "step": 4831
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.536501407623291,
+      "learning_rate": 9.938022356601336e-06,
+      "loss": 0.9137,
+      "step": 4832
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.4668126106262207,
+      "learning_rate": 9.937980965136495e-06,
+      "loss": 0.8041,
+      "step": 4833
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 5.119095325469971,
+      "learning_rate": 9.93793955994098e-06,
+      "loss": 0.7987,
+      "step": 4834
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.48612642288208,
+      "learning_rate": 9.937898141014902e-06,
+      "loss": 0.5956,
+      "step": 4835
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.359579086303711,
+      "learning_rate": 9.937856708358382e-06,
+      "loss": 0.6578,
+      "step": 4836
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.722825527191162,
+      "learning_rate": 9.93781526197153e-06,
+      "loss": 0.5925,
+      "step": 4837
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.9225635528564453,
+      "learning_rate": 9.937773801854464e-06,
+      "loss": 0.5909,
+      "step": 4838
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.674502372741699,
+      "learning_rate": 9.9377323280073e-06,
+      "loss": 0.7013,
+      "step": 4839
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.290163516998291,
+      "learning_rate": 9.937690840430152e-06,
+      "loss": 0.5854,
+      "step": 4840
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.3197197914123535,
+      "learning_rate": 9.937649339123136e-06,
+      "loss": 0.6722,
+      "step": 4841
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.4067749977111816,
+      "learning_rate": 9.937607824086364e-06,
+      "loss": 0.5321,
+      "step": 4842
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.2726757526397705,
+      "learning_rate": 9.937566295319958e-06,
+      "loss": 0.6984,
+      "step": 4843
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.5345215797424316,
+      "learning_rate": 9.937524752824029e-06,
+      "loss": 0.6878,
+      "step": 4844
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.1060311794281006,
+      "learning_rate": 9.937483196598691e-06,
+      "loss": 0.7562,
+      "step": 4845
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.007070779800415,
+      "learning_rate": 9.937441626644063e-06,
+      "loss": 0.8019,
+      "step": 4846
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.2871108055114746,
+      "learning_rate": 9.937400042960258e-06,
+      "loss": 0.6532,
+      "step": 4847
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.149951457977295,
+      "learning_rate": 9.937358445547397e-06,
+      "loss": 0.7408,
+      "step": 4848
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.7524912357330322,
+      "learning_rate": 9.937316834405588e-06,
+      "loss": 0.7063,
+      "step": 4849
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.3978970050811768,
+      "learning_rate": 9.937275209534954e-06,
+      "loss": 0.7161,
+      "step": 4850
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.0537638664245605,
+      "learning_rate": 9.937233570935605e-06,
+      "loss": 0.596,
+      "step": 4851
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.4391610622406006,
+      "learning_rate": 9.937191918607659e-06,
+      "loss": 0.6516,
+      "step": 4852
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.553863525390625,
+      "learning_rate": 9.93715025255123e-06,
+      "loss": 0.833,
+      "step": 4853
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.9166500568389893,
+      "learning_rate": 9.937108572766439e-06,
+      "loss": 0.8594,
+      "step": 4854
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.600532293319702,
+      "learning_rate": 9.937066879253397e-06,
+      "loss": 0.7846,
+      "step": 4855
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.467719316482544,
+      "learning_rate": 9.93702517201222e-06,
+      "loss": 1.0705,
+      "step": 4856
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 6.692129135131836,
+      "learning_rate": 9.936983451043027e-06,
+      "loss": 0.8687,
+      "step": 4857
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.926990032196045,
+      "learning_rate": 9.936941716345933e-06,
+      "loss": 0.6567,
+      "step": 4858
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.3520660400390625,
+      "learning_rate": 9.936899967921053e-06,
+      "loss": 0.6944,
+      "step": 4859
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.381664991378784,
+      "learning_rate": 9.936858205768504e-06,
+      "loss": 0.6422,
+      "step": 4860
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.6670252084732056,
+      "learning_rate": 9.9368164298884e-06,
+      "loss": 0.7056,
+      "step": 4861
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 7.80296516418457,
+      "learning_rate": 9.936774640280862e-06,
+      "loss": 0.7904,
+      "step": 4862
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.655374526977539,
+      "learning_rate": 9.936732836946e-06,
+      "loss": 0.5526,
+      "step": 4863
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.583256483078003,
+      "learning_rate": 9.936691019883935e-06,
+      "loss": 0.819,
+      "step": 4864
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.3790178298950195,
+      "learning_rate": 9.93664918909478e-06,
+      "loss": 0.6654,
+      "step": 4865
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.6367682218551636,
+      "learning_rate": 9.936607344578654e-06,
+      "loss": 0.9086,
+      "step": 4866
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.9779093265533447,
+      "learning_rate": 9.936565486335673e-06,
+      "loss": 0.5991,
+      "step": 4867
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.6489477157592773,
+      "learning_rate": 9.936523614365951e-06,
+      "loss": 0.5098,
+      "step": 4868
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.0509889125823975,
+      "learning_rate": 9.936481728669609e-06,
+      "loss": 0.9679,
+      "step": 4869
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.668741464614868,
+      "learning_rate": 9.936439829246758e-06,
+      "loss": 1.0485,
+      "step": 4870
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.699411392211914,
+      "learning_rate": 9.936397916097517e-06,
+      "loss": 0.7993,
+      "step": 4871
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 4.227519989013672,
+      "learning_rate": 9.936355989222004e-06,
+      "loss": 0.7116,
+      "step": 4872
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 4.1218581199646,
+      "learning_rate": 9.936314048620333e-06,
+      "loss": 0.6967,
+      "step": 4873
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 5.346779823303223,
+      "learning_rate": 9.936272094292623e-06,
+      "loss": 0.3652,
+      "step": 4874
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.8935911655426025,
+      "learning_rate": 9.936230126238989e-06,
+      "loss": 0.5784,
+      "step": 4875
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.3196020126342773,
+      "learning_rate": 9.936188144459548e-06,
+      "loss": 0.7157,
+      "step": 4876
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.3909223079681396,
+      "learning_rate": 9.936146148954417e-06,
+      "loss": 0.476,
+      "step": 4877
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.8672361373901367,
+      "learning_rate": 9.936104139723713e-06,
+      "loss": 0.6501,
+      "step": 4878
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.082914113998413,
+      "learning_rate": 9.936062116767551e-06,
+      "loss": 0.4996,
+      "step": 4879
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.6024014949798584,
+      "learning_rate": 9.936020080086051e-06,
+      "loss": 0.7661,
+      "step": 4880
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.056346893310547,
+      "learning_rate": 9.935978029679327e-06,
+      "loss": 0.8923,
+      "step": 4881
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.210667848587036,
+      "learning_rate": 9.935935965547498e-06,
+      "loss": 0.4422,
+      "step": 4882
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.016380548477173,
+      "learning_rate": 9.93589388769068e-06,
+      "loss": 0.6186,
+      "step": 4883
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 4.823781967163086,
+      "learning_rate": 9.93585179610899e-06,
+      "loss": 0.7787,
+      "step": 4884
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.410459041595459,
+      "learning_rate": 9.935809690802545e-06,
+      "loss": 0.6895,
+      "step": 4885
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 5.4224653244018555,
+      "learning_rate": 9.935767571771462e-06,
+      "loss": 0.8185,
+      "step": 4886
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.644118070602417,
+      "learning_rate": 9.935725439015859e-06,
+      "loss": 0.6855,
+      "step": 4887
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.1932597160339355,
+      "learning_rate": 9.935683292535853e-06,
+      "loss": 0.6235,
+      "step": 4888
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.671781301498413,
+      "learning_rate": 9.93564113233156e-06,
+      "loss": 0.833,
+      "step": 4889
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.421252489089966,
+      "learning_rate": 9.935598958403097e-06,
+      "loss": 0.7503,
+      "step": 4890
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 5.323977947235107,
+      "learning_rate": 9.935556770750585e-06,
+      "loss": 0.8516,
+      "step": 4891
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.230517625808716,
+      "learning_rate": 9.935514569374136e-06,
+      "loss": 0.6973,
+      "step": 4892
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.9828591346740723,
+      "learning_rate": 9.93547235427387e-06,
+      "loss": 0.7101,
+      "step": 4893
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.3473474979400635,
+      "learning_rate": 9.935430125449905e-06,
+      "loss": 0.9473,
+      "step": 4894
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.6060497760772705,
+      "learning_rate": 9.935387882902357e-06,
+      "loss": 0.8559,
+      "step": 4895
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.161952257156372,
+      "learning_rate": 9.935345626631345e-06,
+      "loss": 0.9521,
+      "step": 4896
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.090167999267578,
+      "learning_rate": 9.935303356636984e-06,
+      "loss": 0.5039,
+      "step": 4897
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.02063250541687,
+      "learning_rate": 9.935261072919396e-06,
+      "loss": 0.7451,
+      "step": 4898
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.6669096946716309,
+      "learning_rate": 9.935218775478694e-06,
+      "loss": 0.5976,
+      "step": 4899
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.9520132541656494,
+      "learning_rate": 9.935176464314997e-06,
+      "loss": 0.6149,
+      "step": 4900
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.570223093032837,
+      "learning_rate": 9.935134139428424e-06,
+      "loss": 0.7021,
+      "step": 4901
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.693883180618286,
+      "learning_rate": 9.935091800819092e-06,
+      "loss": 0.8034,
+      "step": 4902
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 4.516499042510986,
+      "learning_rate": 9.935049448487118e-06,
+      "loss": 0.415,
+      "step": 4903
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.7486746311187744,
+      "learning_rate": 9.93500708243262e-06,
+      "loss": 0.8073,
+      "step": 4904
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 4.14466667175293,
+      "learning_rate": 9.934964702655714e-06,
+      "loss": 0.6693,
+      "step": 4905
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.720160484313965,
+      "learning_rate": 9.934922309156523e-06,
+      "loss": 0.641,
+      "step": 4906
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.893693208694458,
+      "learning_rate": 9.934879901935161e-06,
+      "loss": 0.796,
+      "step": 4907
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.626133918762207,
+      "learning_rate": 9.934837480991747e-06,
+      "loss": 0.7041,
+      "step": 4908
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 4.275939464569092,
+      "learning_rate": 9.934795046326397e-06,
+      "loss": 0.5645,
+      "step": 4909
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 11.041194915771484,
+      "learning_rate": 9.93475259793923e-06,
+      "loss": 0.7224,
+      "step": 4910
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 6.1357102394104,
+      "learning_rate": 9.934710135830367e-06,
+      "loss": 0.6238,
+      "step": 4911
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.003180980682373,
+      "learning_rate": 9.934667659999922e-06,
+      "loss": 0.7123,
+      "step": 4912
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.433199882507324,
+      "learning_rate": 9.934625170448016e-06,
+      "loss": 0.4869,
+      "step": 4913
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 4.282390117645264,
+      "learning_rate": 9.934582667174766e-06,
+      "loss": 0.8406,
+      "step": 4914
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 4.817854404449463,
+      "learning_rate": 9.934540150180289e-06,
+      "loss": 0.7507,
+      "step": 4915
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.5407426357269287,
+      "learning_rate": 9.934497619464705e-06,
+      "loss": 0.515,
+      "step": 4916
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.511848211288452,
+      "learning_rate": 9.934455075028132e-06,
+      "loss": 0.6822,
+      "step": 4917
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.4513437747955322,
+      "learning_rate": 9.934412516870686e-06,
+      "loss": 0.3282,
+      "step": 4918
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.735989570617676,
+      "learning_rate": 9.934369944992488e-06,
+      "loss": 0.7704,
+      "step": 4919
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.2055611610412598,
+      "learning_rate": 9.934327359393657e-06,
+      "loss": 0.7919,
+      "step": 4920
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.4499130249023438,
+      "learning_rate": 9.934284760074309e-06,
+      "loss": 0.5465,
+      "step": 4921
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.9830611944198608,
+      "learning_rate": 9.934242147034566e-06,
+      "loss": 0.7103,
+      "step": 4922
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 5.4250946044921875,
+      "learning_rate": 9.93419952027454e-06,
+      "loss": 0.678,
+      "step": 4923
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.721278667449951,
+      "learning_rate": 9.934156879794357e-06,
+      "loss": 0.5455,
+      "step": 4924
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.530137062072754,
+      "learning_rate": 9.934114225594129e-06,
+      "loss": 0.5689,
+      "step": 4925
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.3103673458099365,
+      "learning_rate": 9.93407155767398e-06,
+      "loss": 0.5232,
+      "step": 4926
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.591428279876709,
+      "learning_rate": 9.934028876034026e-06,
+      "loss": 0.7396,
+      "step": 4927
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.0410327911376953,
+      "learning_rate": 9.933986180674386e-06,
+      "loss": 0.9327,
+      "step": 4928
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 4.835996627807617,
+      "learning_rate": 9.933943471595178e-06,
+      "loss": 0.6531,
+      "step": 4929
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.7721893787384033,
+      "learning_rate": 9.93390074879652e-06,
+      "loss": 0.8247,
+      "step": 4930
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.98089337348938,
+      "learning_rate": 9.933858012278535e-06,
+      "loss": 0.4575,
+      "step": 4931
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.2885725498199463,
+      "learning_rate": 9.933815262041339e-06,
+      "loss": 0.624,
+      "step": 4932
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.5990495681762695,
+      "learning_rate": 9.933772498085051e-06,
+      "loss": 0.714,
+      "step": 4933
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.9491169452667236,
+      "learning_rate": 9.933729720409787e-06,
+      "loss": 0.6091,
+      "step": 4934
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.605902910232544,
+      "learning_rate": 9.933686929015672e-06,
+      "loss": 0.5617,
+      "step": 4935
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.825575590133667,
+      "learning_rate": 9.933644123902822e-06,
+      "loss": 0.8036,
+      "step": 4936
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.0045948028564453,
+      "learning_rate": 9.933601305071354e-06,
+      "loss": 0.8927,
+      "step": 4937
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.892977476119995,
+      "learning_rate": 9.93355847252139e-06,
+      "loss": 0.6786,
+      "step": 4938
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 4.7176899909973145,
+      "learning_rate": 9.933515626253047e-06,
+      "loss": 0.6825,
+      "step": 4939
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 4.354426383972168,
+      "learning_rate": 9.933472766266447e-06,
+      "loss": 0.8519,
+      "step": 4940
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.9682655334472656,
+      "learning_rate": 9.933429892561705e-06,
+      "loss": 0.3876,
+      "step": 4941
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 8.819275856018066,
+      "learning_rate": 9.933387005138945e-06,
+      "loss": 0.7965,
+      "step": 4942
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.1937005519866943,
+      "learning_rate": 9.933344103998282e-06,
+      "loss": 0.7296,
+      "step": 4943
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.541494846343994,
+      "learning_rate": 9.93330118913984e-06,
+      "loss": 0.6958,
+      "step": 4944
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 5.398994445800781,
+      "learning_rate": 9.933258260563732e-06,
+      "loss": 0.8676,
+      "step": 4945
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 5.65027379989624,
+      "learning_rate": 9.933215318270081e-06,
+      "loss": 0.5617,
+      "step": 4946
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.6663742065429688,
+      "learning_rate": 9.933172362259009e-06,
+      "loss": 0.5937,
+      "step": 4947
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.8763439655303955,
+      "learning_rate": 9.93312939253063e-06,
+      "loss": 0.8834,
+      "step": 4948
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.1309473514556885,
+      "learning_rate": 9.933086409085068e-06,
+      "loss": 0.5844,
+      "step": 4949
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 4.116574287414551,
+      "learning_rate": 9.93304341192244e-06,
+      "loss": 1.0992,
+      "step": 4950
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.14701509475708,
+      "learning_rate": 9.933000401042865e-06,
+      "loss": 0.8278,
+      "step": 4951
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.7456977367401123,
+      "learning_rate": 9.932957376446465e-06,
+      "loss": 0.6411,
+      "step": 4952
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.598299264907837,
+      "learning_rate": 9.93291433813336e-06,
+      "loss": 0.6058,
+      "step": 4953
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.486292600631714,
+      "learning_rate": 9.932871286103665e-06,
+      "loss": 0.6498,
+      "step": 4954
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.419755458831787,
+      "learning_rate": 9.932828220357504e-06,
+      "loss": 1.0548,
+      "step": 4955
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 4.770900249481201,
+      "learning_rate": 9.932785140894996e-06,
+      "loss": 0.7491,
+      "step": 4956
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.81362247467041,
+      "learning_rate": 9.93274204771626e-06,
+      "loss": 0.5516,
+      "step": 4957
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.7567355632781982,
+      "learning_rate": 9.932698940821418e-06,
+      "loss": 0.5146,
+      "step": 4958
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.2203497886657715,
+      "learning_rate": 9.932655820210586e-06,
+      "loss": 0.6997,
+      "step": 4959
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.7803337574005127,
+      "learning_rate": 9.932612685883887e-06,
+      "loss": 0.7156,
+      "step": 4960
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.727784514427185,
+      "learning_rate": 9.93256953784144e-06,
+      "loss": 0.6368,
+      "step": 4961
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.945722818374634,
+      "learning_rate": 9.932526376083363e-06,
+      "loss": 0.5998,
+      "step": 4962
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.417147636413574,
+      "learning_rate": 9.93248320060978e-06,
+      "loss": 0.6998,
+      "step": 4963
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.283292531967163,
+      "learning_rate": 9.932440011420807e-06,
+      "loss": 0.5482,
+      "step": 4964
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.711599588394165,
+      "learning_rate": 9.93239680851657e-06,
+      "loss": 0.6445,
+      "step": 4965
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 5.078585147857666,
+      "learning_rate": 9.932353591897182e-06,
+      "loss": 0.5077,
+      "step": 4966
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.2783665657043457,
+      "learning_rate": 9.932310361562767e-06,
+      "loss": 0.4895,
+      "step": 4967
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.212223529815674,
+      "learning_rate": 9.932267117513444e-06,
+      "loss": 0.6497,
+      "step": 4968
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.6675775051116943,
+      "learning_rate": 9.932223859749333e-06,
+      "loss": 0.629,
+      "step": 4969
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.9854074716567993,
+      "learning_rate": 9.932180588270557e-06,
+      "loss": 0.3744,
+      "step": 4970
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.267977476119995,
+      "learning_rate": 9.932137303077234e-06,
+      "loss": 0.4659,
+      "step": 4971
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.0486366748809814,
+      "learning_rate": 9.932094004169485e-06,
+      "loss": 0.6096,
+      "step": 4972
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.277287483215332,
+      "learning_rate": 9.93205069154743e-06,
+      "loss": 0.9333,
+      "step": 4973
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.995577096939087,
+      "learning_rate": 9.93200736521119e-06,
+      "loss": 0.8104,
+      "step": 4974
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.3078582286834717,
+      "learning_rate": 9.931964025160882e-06,
+      "loss": 0.9002,
+      "step": 4975
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.97863507270813,
+      "learning_rate": 9.931920671396632e-06,
+      "loss": 0.4651,
+      "step": 4976
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 4.975131511688232,
+      "learning_rate": 9.93187730391856e-06,
+      "loss": 0.6006,
+      "step": 4977
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 4.1438889503479,
+      "learning_rate": 9.931833922726782e-06,
+      "loss": 0.7269,
+      "step": 4978
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.6131551265716553,
+      "learning_rate": 9.931790527821423e-06,
+      "loss": 0.8927,
+      "step": 4979
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 4.084049701690674,
+      "learning_rate": 9.931747119202601e-06,
+      "loss": 0.5136,
+      "step": 4980
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.3867239952087402,
+      "learning_rate": 9.931703696870438e-06,
+      "loss": 0.5083,
+      "step": 4981
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.2235071659088135,
+      "learning_rate": 9.931660260825055e-06,
+      "loss": 0.6485,
+      "step": 4982
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.761897087097168,
+      "learning_rate": 9.931616811066572e-06,
+      "loss": 0.6795,
+      "step": 4983
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.756002902984619,
+      "learning_rate": 9.93157334759511e-06,
+      "loss": 0.8577,
+      "step": 4984
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.208031415939331,
+      "learning_rate": 9.93152987041079e-06,
+      "loss": 0.6424,
+      "step": 4985
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.014347553253174,
+      "learning_rate": 9.931486379513733e-06,
+      "loss": 0.8485,
+      "step": 4986
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.3192176818847656,
+      "learning_rate": 9.931442874904058e-06,
+      "loss": 0.3212,
+      "step": 4987
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.188042640686035,
+      "learning_rate": 9.93139935658189e-06,
+      "loss": 0.489,
+      "step": 4988
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.112644910812378,
+      "learning_rate": 9.931355824547346e-06,
+      "loss": 1.0884,
+      "step": 4989
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.822946310043335,
+      "learning_rate": 9.931312278800551e-06,
+      "loss": 0.6556,
+      "step": 4990
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.9980179071426392,
+      "learning_rate": 9.931268719341623e-06,
+      "loss": 0.6394,
+      "step": 4991
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.453519821166992,
+      "learning_rate": 9.931225146170682e-06,
+      "loss": 0.85,
+      "step": 4992
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 4.280611038208008,
+      "learning_rate": 9.931181559287853e-06,
+      "loss": 0.4794,
+      "step": 4993
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.3378021717071533,
+      "learning_rate": 9.931137958693255e-06,
+      "loss": 0.8017,
+      "step": 4994
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.715811014175415,
+      "learning_rate": 9.93109434438701e-06,
+      "loss": 0.4017,
+      "step": 4995
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.4418694972991943,
+      "learning_rate": 9.931050716369239e-06,
+      "loss": 0.6563,
+      "step": 4996
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.582324743270874,
+      "learning_rate": 9.931007074640062e-06,
+      "loss": 0.8116,
+      "step": 4997
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.576108694076538,
+      "learning_rate": 9.930963419199601e-06,
+      "loss": 0.5264,
+      "step": 4998
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.5787014961242676,
+      "learning_rate": 9.93091975004798e-06,
+      "loss": 0.5389,
+      "step": 4999
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.5763747692108154,
+      "learning_rate": 9.930876067185318e-06,
+      "loss": 0.7297,
+      "step": 5000
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.957322120666504,
+      "learning_rate": 9.930832370611737e-06,
+      "loss": 0.6303,
+      "step": 5001
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.633720874786377,
+      "learning_rate": 9.930788660327357e-06,
+      "loss": 0.7632,
+      "step": 5002
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.8324297666549683,
+      "learning_rate": 9.9307449363323e-06,
+      "loss": 0.7588,
+      "step": 5003
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.3123631477355957,
+      "learning_rate": 9.93070119862669e-06,
+      "loss": 0.6664,
+      "step": 5004
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.52852463722229,
+      "learning_rate": 9.930657447210647e-06,
+      "loss": 0.7953,
+      "step": 5005
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.4718894958496094,
+      "learning_rate": 9.930613682084294e-06,
+      "loss": 0.5718,
+      "step": 5006
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.7321010828018188,
+      "learning_rate": 9.930569903247749e-06,
+      "loss": 0.606,
+      "step": 5007
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.4652037620544434,
+      "learning_rate": 9.930526110701137e-06,
+      "loss": 1.0167,
+      "step": 5008
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.8746321201324463,
+      "learning_rate": 9.93048230444458e-06,
+      "loss": 0.7156,
+      "step": 5009
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.136599063873291,
+      "learning_rate": 9.930438484478197e-06,
+      "loss": 0.6715,
+      "step": 5010
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 5.536617279052734,
+      "learning_rate": 9.930394650802113e-06,
+      "loss": 0.5692,
+      "step": 5011
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.7528560161590576,
+      "learning_rate": 9.930350803416446e-06,
+      "loss": 0.5917,
+      "step": 5012
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 5.484799861907959,
+      "learning_rate": 9.930306942321323e-06,
+      "loss": 0.8251,
+      "step": 5013
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.568582534790039,
+      "learning_rate": 9.930263067516862e-06,
+      "loss": 0.8526,
+      "step": 5014
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.907667636871338,
+      "learning_rate": 9.930219179003187e-06,
+      "loss": 0.5459,
+      "step": 5015
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.4546990394592285,
+      "learning_rate": 9.93017527678042e-06,
+      "loss": 0.7717,
+      "step": 5016
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.214000701904297,
+      "learning_rate": 9.93013136084868e-06,
+      "loss": 0.67,
+      "step": 5017
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.9403631687164307,
+      "learning_rate": 9.930087431208093e-06,
+      "loss": 0.7024,
+      "step": 5018
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.6191396713256836,
+      "learning_rate": 9.930043487858781e-06,
+      "loss": 0.806,
+      "step": 5019
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.8874199390411377,
+      "learning_rate": 9.929999530800864e-06,
+      "loss": 0.6241,
+      "step": 5020
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.33552622795105,
+      "learning_rate": 9.929955560034464e-06,
+      "loss": 0.5294,
+      "step": 5021
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.191209316253662,
+      "learning_rate": 9.929911575559706e-06,
+      "loss": 0.8272,
+      "step": 5022
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.3124566078186035,
+      "learning_rate": 9.92986757737671e-06,
+      "loss": 0.5329,
+      "step": 5023
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.5524048805236816,
+      "learning_rate": 9.9298235654856e-06,
+      "loss": 0.7752,
+      "step": 5024
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.6115522384643555,
+      "learning_rate": 9.929779539886495e-06,
+      "loss": 0.4786,
+      "step": 5025
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.446820020675659,
+      "learning_rate": 9.929735500579521e-06,
+      "loss": 1.0068,
+      "step": 5026
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.8586182594299316,
+      "learning_rate": 9.9296914475648e-06,
+      "loss": 0.8927,
+      "step": 5027
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.2425544261932373,
+      "learning_rate": 9.929647380842453e-06,
+      "loss": 0.7528,
+      "step": 5028
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.8999874591827393,
+      "learning_rate": 9.929603300412604e-06,
+      "loss": 0.6202,
+      "step": 5029
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.6841254234313965,
+      "learning_rate": 9.929559206275374e-06,
+      "loss": 0.8337,
+      "step": 5030
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.4472568035125732,
+      "learning_rate": 9.929515098430886e-06,
+      "loss": 0.6757,
+      "step": 5031
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.6195919513702393,
+      "learning_rate": 9.929470976879264e-06,
+      "loss": 0.5168,
+      "step": 5032
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.0308427810668945,
+      "learning_rate": 9.92942684162063e-06,
+      "loss": 0.5955,
+      "step": 5033
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.1136159896850586,
+      "learning_rate": 9.929382692655107e-06,
+      "loss": 0.732,
+      "step": 5034
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.7828314304351807,
+      "learning_rate": 9.929338529982817e-06,
+      "loss": 0.7861,
+      "step": 5035
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 7.820859432220459,
+      "learning_rate": 9.929294353603884e-06,
+      "loss": 0.664,
+      "step": 5036
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.2964558601379395,
+      "learning_rate": 9.92925016351843e-06,
+      "loss": 0.399,
+      "step": 5037
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.4033355712890625,
+      "learning_rate": 9.929205959726576e-06,
+      "loss": 0.9095,
+      "step": 5038
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.924217939376831,
+      "learning_rate": 9.929161742228447e-06,
+      "loss": 0.5556,
+      "step": 5039
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 4.023550987243652,
+      "learning_rate": 9.929117511024167e-06,
+      "loss": 0.8459,
+      "step": 5040
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.3544113636016846,
+      "learning_rate": 9.929073266113856e-06,
+      "loss": 0.7592,
+      "step": 5041
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.5031239986419678,
+      "learning_rate": 9.92902900749764e-06,
+      "loss": 1.103,
+      "step": 5042
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 4.279013156890869,
+      "learning_rate": 9.92898473517564e-06,
+      "loss": 0.6842,
+      "step": 5043
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.938129186630249,
+      "learning_rate": 9.92894044914798e-06,
+      "loss": 0.7449,
+      "step": 5044
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 5.307285308837891,
+      "learning_rate": 9.928896149414784e-06,
+      "loss": 0.844,
+      "step": 5045
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.0495026111602783,
+      "learning_rate": 9.928851835976174e-06,
+      "loss": 0.6257,
+      "step": 5046
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.5495152473449707,
+      "learning_rate": 9.928807508832273e-06,
+      "loss": 0.6373,
+      "step": 5047
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.456958532333374,
+      "learning_rate": 9.928763167983204e-06,
+      "loss": 0.6907,
+      "step": 5048
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 4.992599010467529,
+      "learning_rate": 9.928718813429093e-06,
+      "loss": 0.8979,
+      "step": 5049
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.988525629043579,
+      "learning_rate": 9.928674445170059e-06,
+      "loss": 0.7208,
+      "step": 5050
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.1751954555511475,
+      "learning_rate": 9.928630063206228e-06,
+      "loss": 0.5768,
+      "step": 5051
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.8459694385528564,
+      "learning_rate": 9.928585667537725e-06,
+      "loss": 0.6141,
+      "step": 5052
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.3850889205932617,
+      "learning_rate": 9.92854125816467e-06,
+      "loss": 0.6531,
+      "step": 5053
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.936178207397461,
+      "learning_rate": 9.928496835087188e-06,
+      "loss": 0.7016,
+      "step": 5054
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.8585410118103027,
+      "learning_rate": 9.928452398305404e-06,
+      "loss": 0.6971,
+      "step": 5055
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.3691494464874268,
+      "learning_rate": 9.928407947819438e-06,
+      "loss": 0.6053,
+      "step": 5056
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.789057731628418,
+      "learning_rate": 9.928363483629418e-06,
+      "loss": 0.8659,
+      "step": 5057
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.4866621494293213,
+      "learning_rate": 9.928319005735463e-06,
+      "loss": 0.8684,
+      "step": 5058
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.0236380100250244,
+      "learning_rate": 9.9282745141377e-06,
+      "loss": 0.6316,
+      "step": 5059
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.496001720428467,
+      "learning_rate": 9.928230008836252e-06,
+      "loss": 0.5771,
+      "step": 5060
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.7682809829711914,
+      "learning_rate": 9.928185489831242e-06,
+      "loss": 0.6491,
+      "step": 5061
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.6567916870117188,
+      "learning_rate": 9.928140957122795e-06,
+      "loss": 0.6994,
+      "step": 5062
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.710017204284668,
+      "learning_rate": 9.928096410711034e-06,
+      "loss": 0.5842,
+      "step": 5063
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.7925469875335693,
+      "learning_rate": 9.928051850596083e-06,
+      "loss": 0.4465,
+      "step": 5064
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.6578948497772217,
+      "learning_rate": 9.928007276778066e-06,
+      "loss": 0.6384,
+      "step": 5065
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.258044719696045,
+      "learning_rate": 9.927962689257106e-06,
+      "loss": 0.7133,
+      "step": 5066
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.2778573036193848,
+      "learning_rate": 9.927918088033328e-06,
+      "loss": 0.5698,
+      "step": 5067
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.25273060798645,
+      "learning_rate": 9.927873473106858e-06,
+      "loss": 0.7827,
+      "step": 5068
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.721853733062744,
+      "learning_rate": 9.927828844477816e-06,
+      "loss": 0.7793,
+      "step": 5069
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.0710296630859375,
+      "learning_rate": 9.927784202146327e-06,
+      "loss": 0.7391,
+      "step": 5070
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.9418630599975586,
+      "learning_rate": 9.927739546112519e-06,
+      "loss": 0.8386,
+      "step": 5071
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.8739757537841797,
+      "learning_rate": 9.92769487637651e-06,
+      "loss": 0.7369,
+      "step": 5072
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.2800984382629395,
+      "learning_rate": 9.92765019293843e-06,
+      "loss": 0.4838,
+      "step": 5073
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.579214572906494,
+      "learning_rate": 9.927605495798398e-06,
+      "loss": 0.7845,
+      "step": 5074
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 4.019787788391113,
+      "learning_rate": 9.927560784956544e-06,
+      "loss": 0.7423,
+      "step": 5075
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 4.365840911865234,
+      "learning_rate": 9.927516060412987e-06,
+      "loss": 0.5932,
+      "step": 5076
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.599111795425415,
+      "learning_rate": 9.927471322167857e-06,
+      "loss": 0.5865,
+      "step": 5077
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.93662691116333,
+      "learning_rate": 9.927426570221273e-06,
+      "loss": 0.6765,
+      "step": 5078
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 4.0615692138671875,
+      "learning_rate": 9.927381804573361e-06,
+      "loss": 0.5179,
+      "step": 5079
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.788639545440674,
+      "learning_rate": 9.927337025224247e-06,
+      "loss": 0.5202,
+      "step": 5080
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.6178621053695679,
+      "learning_rate": 9.927292232174052e-06,
+      "loss": 0.6441,
+      "step": 5081
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 4.6954522132873535,
+      "learning_rate": 9.927247425422907e-06,
+      "loss": 0.5481,
+      "step": 5082
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.368957281112671,
+      "learning_rate": 9.927202604970932e-06,
+      "loss": 0.7128,
+      "step": 5083
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.042433023452759,
+      "learning_rate": 9.92715777081825e-06,
+      "loss": 0.5399,
+      "step": 5084
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.224992275238037,
+      "learning_rate": 9.92711292296499e-06,
+      "loss": 0.7915,
+      "step": 5085
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.2250237464904785,
+      "learning_rate": 9.927068061411274e-06,
+      "loss": 1.0009,
+      "step": 5086
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.7039949893951416,
+      "learning_rate": 9.927023186157228e-06,
+      "loss": 0.7977,
+      "step": 5087
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.215552568435669,
+      "learning_rate": 9.926978297202975e-06,
+      "loss": 0.5282,
+      "step": 5088
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.508875608444214,
+      "learning_rate": 9.926933394548641e-06,
+      "loss": 0.8972,
+      "step": 5089
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.0307364463806152,
+      "learning_rate": 9.926888478194351e-06,
+      "loss": 0.8845,
+      "step": 5090
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.29669189453125,
+      "learning_rate": 9.926843548140232e-06,
+      "loss": 0.6441,
+      "step": 5091
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.1953887939453125,
+      "learning_rate": 9.926798604386405e-06,
+      "loss": 0.7346,
+      "step": 5092
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.030541181564331,
+      "learning_rate": 9.926753646932997e-06,
+      "loss": 0.8846,
+      "step": 5093
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.2012195587158203,
+      "learning_rate": 9.926708675780132e-06,
+      "loss": 0.7979,
+      "step": 5094
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.162709951400757,
+      "learning_rate": 9.926663690927936e-06,
+      "loss": 0.6003,
+      "step": 5095
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 4.251805305480957,
+      "learning_rate": 9.926618692376535e-06,
+      "loss": 0.8279,
+      "step": 5096
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.152068853378296,
+      "learning_rate": 9.926573680126052e-06,
+      "loss": 0.4902,
+      "step": 5097
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.5504825115203857,
+      "learning_rate": 9.926528654176612e-06,
+      "loss": 0.7624,
+      "step": 5098
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.3513407707214355,
+      "learning_rate": 9.926483614528343e-06,
+      "loss": 0.5592,
+      "step": 5099
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.9877923727035522,
+      "learning_rate": 9.926438561181368e-06,
+      "loss": 0.5944,
+      "step": 5100
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.9802225828170776,
+      "learning_rate": 9.926393494135814e-06,
+      "loss": 0.448,
+      "step": 5101
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.5974042415618896,
+      "learning_rate": 9.926348413391804e-06,
+      "loss": 1.0504,
+      "step": 5102
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 22.067502975463867,
+      "learning_rate": 9.926303318949465e-06,
+      "loss": 0.7906,
+      "step": 5103
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.5339720249176025,
+      "learning_rate": 9.926258210808922e-06,
+      "loss": 0.6021,
+      "step": 5104
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.9420578479766846,
+      "learning_rate": 9.9262130889703e-06,
+      "loss": 0.9165,
+      "step": 5105
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.188633680343628,
+      "learning_rate": 9.926167953433726e-06,
+      "loss": 0.7196,
+      "step": 5106
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.781874656677246,
+      "learning_rate": 9.926122804199321e-06,
+      "loss": 0.9095,
+      "step": 5107
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.4837300777435303,
+      "learning_rate": 9.926077641267216e-06,
+      "loss": 0.5922,
+      "step": 5108
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.1114089488983154,
+      "learning_rate": 9.926032464637535e-06,
+      "loss": 0.5834,
+      "step": 5109
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.328660488128662,
+      "learning_rate": 9.925987274310403e-06,
+      "loss": 0.6437,
+      "step": 5110
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.2155251502990723,
+      "learning_rate": 9.925942070285944e-06,
+      "loss": 0.5745,
+      "step": 5111
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.209895610809326,
+      "learning_rate": 9.925896852564288e-06,
+      "loss": 0.6971,
+      "step": 5112
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.6237664222717285,
+      "learning_rate": 9.925851621145554e-06,
+      "loss": 0.9159,
+      "step": 5113
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.960177183151245,
+      "learning_rate": 9.925806376029876e-06,
+      "loss": 0.6657,
+      "step": 5114
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.020977735519409,
+      "learning_rate": 9.925761117217373e-06,
+      "loss": 0.8519,
+      "step": 5115
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.6451737880706787,
+      "learning_rate": 9.925715844708177e-06,
+      "loss": 0.8317,
+      "step": 5116
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 4.367865562438965,
+      "learning_rate": 9.925670558502407e-06,
+      "loss": 0.507,
+      "step": 5117
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.0329976081848145,
+      "learning_rate": 9.925625258600194e-06,
+      "loss": 0.7257,
+      "step": 5118
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 4.62069034576416,
+      "learning_rate": 9.925579945001662e-06,
+      "loss": 0.5273,
+      "step": 5119
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.1495518684387207,
+      "learning_rate": 9.925534617706939e-06,
+      "loss": 0.4355,
+      "step": 5120
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.467236280441284,
+      "learning_rate": 9.925489276716147e-06,
+      "loss": 0.5078,
+      "step": 5121
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.044579267501831,
+      "learning_rate": 9.925443922029415e-06,
+      "loss": 0.617,
+      "step": 5122
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.3694419860839844,
+      "learning_rate": 9.925398553646868e-06,
+      "loss": 0.7447,
+      "step": 5123
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.002105712890625,
+      "learning_rate": 9.925353171568634e-06,
+      "loss": 0.7734,
+      "step": 5124
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 4.499125003814697,
+      "learning_rate": 9.925307775794838e-06,
+      "loss": 0.5673,
+      "step": 5125
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.4837615489959717,
+      "learning_rate": 9.925262366325604e-06,
+      "loss": 0.6425,
+      "step": 5126
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 4.117613792419434,
+      "learning_rate": 9.925216943161063e-06,
+      "loss": 1.0357,
+      "step": 5127
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.0863912105560303,
+      "learning_rate": 9.925171506301337e-06,
+      "loss": 0.7866,
+      "step": 5128
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.6477110385894775,
+      "learning_rate": 9.925126055746554e-06,
+      "loss": 0.6302,
+      "step": 5129
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.1700448989868164,
+      "learning_rate": 9.925080591496842e-06,
+      "loss": 0.6008,
+      "step": 5130
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.870365619659424,
+      "learning_rate": 9.925035113552325e-06,
+      "loss": 0.3832,
+      "step": 5131
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.6513774394989014,
+      "learning_rate": 9.92498962191313e-06,
+      "loss": 0.6863,
+      "step": 5132
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.313088893890381,
+      "learning_rate": 9.924944116579385e-06,
+      "loss": 0.599,
+      "step": 5133
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.412574529647827,
+      "learning_rate": 9.924898597551214e-06,
+      "loss": 0.4492,
+      "step": 5134
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 4.702681541442871,
+      "learning_rate": 9.924853064828746e-06,
+      "loss": 0.5427,
+      "step": 5135
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.7246718406677246,
+      "learning_rate": 9.924807518412105e-06,
+      "loss": 0.6591,
+      "step": 5136
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.6388258934020996,
+      "learning_rate": 9.92476195830142e-06,
+      "loss": 0.7762,
+      "step": 5137
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.7309064865112305,
+      "learning_rate": 9.924716384496817e-06,
+      "loss": 0.795,
+      "step": 5138
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.9373788833618164,
+      "learning_rate": 9.924670796998421e-06,
+      "loss": 0.6901,
+      "step": 5139
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.8521860837936401,
+      "learning_rate": 9.924625195806363e-06,
+      "loss": 0.5267,
+      "step": 5140
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.5739023685455322,
+      "learning_rate": 9.924579580920766e-06,
+      "loss": 0.5947,
+      "step": 5141
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.541027545928955,
+      "learning_rate": 9.924533952341757e-06,
+      "loss": 0.6363,
+      "step": 5142
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.5294325351715088,
+      "learning_rate": 9.924488310069465e-06,
+      "loss": 0.5346,
+      "step": 5143
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.252758502960205,
+      "learning_rate": 9.924442654104017e-06,
+      "loss": 0.7906,
+      "step": 5144
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.5329549312591553,
+      "learning_rate": 9.924396984445537e-06,
+      "loss": 0.6153,
+      "step": 5145
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 5.319924354553223,
+      "learning_rate": 9.924351301094153e-06,
+      "loss": 0.681,
+      "step": 5146
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.887121558189392,
+      "learning_rate": 9.924305604049992e-06,
+      "loss": 0.4739,
+      "step": 5147
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.7710494995117188,
+      "learning_rate": 9.924259893313185e-06,
+      "loss": 0.9231,
+      "step": 5148
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.5748281478881836,
+      "learning_rate": 9.924214168883854e-06,
+      "loss": 0.7967,
+      "step": 5149
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.408464193344116,
+      "learning_rate": 9.924168430762129e-06,
+      "loss": 0.6471,
+      "step": 5150
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.4912109375,
+      "learning_rate": 9.924122678948136e-06,
+      "loss": 0.5015,
+      "step": 5151
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.766112804412842,
+      "learning_rate": 9.924076913442002e-06,
+      "loss": 0.9513,
+      "step": 5152
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.8248211145401,
+      "learning_rate": 9.924031134243855e-06,
+      "loss": 0.834,
+      "step": 5153
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 4.19196891784668,
+      "learning_rate": 9.923985341353821e-06,
+      "loss": 0.573,
+      "step": 5154
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.7586174011230469,
+      "learning_rate": 9.923939534772029e-06,
+      "loss": 0.4192,
+      "step": 5155
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.5428831577301025,
+      "learning_rate": 9.923893714498607e-06,
+      "loss": 0.7269,
+      "step": 5156
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.7399275302886963,
+      "learning_rate": 9.92384788053368e-06,
+      "loss": 0.5401,
+      "step": 5157
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.4319934844970703,
+      "learning_rate": 9.923802032877376e-06,
+      "loss": 0.832,
+      "step": 5158
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.628629207611084,
+      "learning_rate": 9.923756171529824e-06,
+      "loss": 0.5709,
+      "step": 5159
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.438973903656006,
+      "learning_rate": 9.923710296491149e-06,
+      "loss": 0.5498,
+      "step": 5160
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.2544162273406982,
+      "learning_rate": 9.923664407761482e-06,
+      "loss": 0.4944,
+      "step": 5161
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.6145195960998535,
+      "learning_rate": 9.923618505340948e-06,
+      "loss": 0.9537,
+      "step": 5162
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.4863476753234863,
+      "learning_rate": 9.923572589229675e-06,
+      "loss": 0.4889,
+      "step": 5163
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.97817325592041,
+      "learning_rate": 9.92352665942779e-06,
+      "loss": 0.9921,
+      "step": 5164
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.7940115928649902,
+      "learning_rate": 9.923480715935425e-06,
+      "loss": 0.7403,
+      "step": 5165
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.745572805404663,
+      "learning_rate": 9.923434758752702e-06,
+      "loss": 0.8995,
+      "step": 5166
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.3688669204711914,
+      "learning_rate": 9.923388787879752e-06,
+      "loss": 1.0065,
+      "step": 5167
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.1828482151031494,
+      "learning_rate": 9.923342803316703e-06,
+      "loss": 0.832,
+      "step": 5168
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.761110782623291,
+      "learning_rate": 9.92329680506368e-06,
+      "loss": 0.607,
+      "step": 5169
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.2234294414520264,
+      "learning_rate": 9.923250793120815e-06,
+      "loss": 0.6909,
+      "step": 5170
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.1360888481140137,
+      "learning_rate": 9.923204767488233e-06,
+      "loss": 0.7699,
+      "step": 5171
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.6141297817230225,
+      "learning_rate": 9.923158728166061e-06,
+      "loss": 0.5623,
+      "step": 5172
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 4.417358875274658,
+      "learning_rate": 9.923112675154432e-06,
+      "loss": 0.7783,
+      "step": 5173
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.9720656871795654,
+      "learning_rate": 9.923066608453468e-06,
+      "loss": 0.7347,
+      "step": 5174
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.2716636657714844,
+      "learning_rate": 9.923020528063302e-06,
+      "loss": 0.5248,
+      "step": 5175
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 4.939866065979004,
+      "learning_rate": 9.922974433984058e-06,
+      "loss": 0.6594,
+      "step": 5176
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.085662841796875,
+      "learning_rate": 9.922928326215869e-06,
+      "loss": 0.8468,
+      "step": 5177
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.7351481914520264,
+      "learning_rate": 9.92288220475886e-06,
+      "loss": 0.4141,
+      "step": 5178
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.8614726066589355,
+      "learning_rate": 9.922836069613158e-06,
+      "loss": 0.3856,
+      "step": 5179
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.2806694507598877,
+      "learning_rate": 9.922789920778894e-06,
+      "loss": 0.5352,
+      "step": 5180
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.7084696292877197,
+      "learning_rate": 9.922743758256195e-06,
+      "loss": 0.9718,
+      "step": 5181
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.202918291091919,
+      "learning_rate": 9.92269758204519e-06,
+      "loss": 0.6126,
+      "step": 5182
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 6.672900199890137,
+      "learning_rate": 9.922651392146007e-06,
+      "loss": 0.8748,
+      "step": 5183
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.139254331588745,
+      "learning_rate": 9.922605188558776e-06,
+      "loss": 0.6355,
+      "step": 5184
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.2972488403320312,
+      "learning_rate": 9.922558971283623e-06,
+      "loss": 0.746,
+      "step": 5185
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.337085247039795,
+      "learning_rate": 9.922512740320675e-06,
+      "loss": 0.7111,
+      "step": 5186
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.6582080125808716,
+      "learning_rate": 9.922466495670066e-06,
+      "loss": 0.7643,
+      "step": 5187
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.9090211391448975,
+      "learning_rate": 9.922420237331923e-06,
+      "loss": 0.4903,
+      "step": 5188
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.5342812538146973,
+      "learning_rate": 9.92237396530637e-06,
+      "loss": 0.7845,
+      "step": 5189
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.0929031372070312,
+      "learning_rate": 9.92232767959354e-06,
+      "loss": 0.7379,
+      "step": 5190
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.159802198410034,
+      "learning_rate": 9.922281380193562e-06,
+      "loss": 0.7183,
+      "step": 5191
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.9023311138153076,
+      "learning_rate": 9.922235067106562e-06,
+      "loss": 0.6143,
+      "step": 5192
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.1655654907226562,
+      "learning_rate": 9.92218874033267e-06,
+      "loss": 0.6602,
+      "step": 5193
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.4895589351654053,
+      "learning_rate": 9.922142399872016e-06,
+      "loss": 0.5069,
+      "step": 5194
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.1787209510803223,
+      "learning_rate": 9.922096045724728e-06,
+      "loss": 0.7588,
+      "step": 5195
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.072150230407715,
+      "learning_rate": 9.922049677890932e-06,
+      "loss": 0.7942,
+      "step": 5196
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.5180182456970215,
+      "learning_rate": 9.922003296370764e-06,
+      "loss": 0.7477,
+      "step": 5197
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.830657720565796,
+      "learning_rate": 9.921956901164345e-06,
+      "loss": 0.6774,
+      "step": 5198
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.4544200897216797,
+      "learning_rate": 9.92191049227181e-06,
+      "loss": 0.7021,
+      "step": 5199
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.9498653411865234,
+      "learning_rate": 9.921864069693285e-06,
+      "loss": 0.7648,
+      "step": 5200
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.33778977394104,
+      "learning_rate": 9.921817633428899e-06,
+      "loss": 0.7053,
+      "step": 5201
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.5293729305267334,
+      "learning_rate": 9.921771183478783e-06,
+      "loss": 0.821,
+      "step": 5202
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.055203437805176,
+      "learning_rate": 9.921724719843065e-06,
+      "loss": 0.437,
+      "step": 5203
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 16.449525833129883,
+      "learning_rate": 9.921678242521874e-06,
+      "loss": 0.6941,
+      "step": 5204
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 4.471606731414795,
+      "learning_rate": 9.921631751515339e-06,
+      "loss": 0.5256,
+      "step": 5205
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 4.613789081573486,
+      "learning_rate": 9.92158524682359e-06,
+      "loss": 0.9499,
+      "step": 5206
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.2380542755126953,
+      "learning_rate": 9.921538728446756e-06,
+      "loss": 0.6407,
+      "step": 5207
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.4118669033050537,
+      "learning_rate": 9.921492196384968e-06,
+      "loss": 0.7569,
+      "step": 5208
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 5.8473334312438965,
+      "learning_rate": 9.921445650638353e-06,
+      "loss": 0.6783,
+      "step": 5209
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 4.457103729248047,
+      "learning_rate": 9.921399091207041e-06,
+      "loss": 0.6866,
+      "step": 5210
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.762277603149414,
+      "learning_rate": 9.921352518091161e-06,
+      "loss": 0.9174,
+      "step": 5211
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.954636335372925,
+      "learning_rate": 9.921305931290843e-06,
+      "loss": 0.6807,
+      "step": 5212
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 4.432750701904297,
+      "learning_rate": 9.92125933080622e-06,
+      "loss": 0.8913,
+      "step": 5213
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.9754891395568848,
+      "learning_rate": 9.921212716637414e-06,
+      "loss": 0.5915,
+      "step": 5214
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.904010772705078,
+      "learning_rate": 9.921166088784562e-06,
+      "loss": 0.7094,
+      "step": 5215
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.6761181354522705,
+      "learning_rate": 9.921119447247788e-06,
+      "loss": 0.7506,
+      "step": 5216
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 4.2059645652771,
+      "learning_rate": 9.921072792027226e-06,
+      "loss": 0.8876,
+      "step": 5217
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.7371163368225098,
+      "learning_rate": 9.921026123123005e-06,
+      "loss": 0.5876,
+      "step": 5218
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.0917882919311523,
+      "learning_rate": 9.920979440535252e-06,
+      "loss": 0.6435,
+      "step": 5219
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.4527981281280518,
+      "learning_rate": 9.9209327442641e-06,
+      "loss": 0.658,
+      "step": 5220
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 4.612520217895508,
+      "learning_rate": 9.920886034309676e-06,
+      "loss": 0.7396,
+      "step": 5221
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 4.044054985046387,
+      "learning_rate": 9.920839310672112e-06,
+      "loss": 0.7246,
+      "step": 5222
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 4.635966777801514,
+      "learning_rate": 9.920792573351538e-06,
+      "loss": 0.7822,
+      "step": 5223
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.772749900817871,
+      "learning_rate": 9.920745822348082e-06,
+      "loss": 0.6158,
+      "step": 5224
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.7612998485565186,
+      "learning_rate": 9.920699057661875e-06,
+      "loss": 0.4777,
+      "step": 5225
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.5494272708892822,
+      "learning_rate": 9.92065227929305e-06,
+      "loss": 0.5968,
+      "step": 5226
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.9027795791625977,
+      "learning_rate": 9.920605487241732e-06,
+      "loss": 0.4785,
+      "step": 5227
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.0286808013916016,
+      "learning_rate": 9.920558681508055e-06,
+      "loss": 0.6014,
+      "step": 5228
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.8186120986938477,
+      "learning_rate": 9.920511862092146e-06,
+      "loss": 1.0519,
+      "step": 5229
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.916095018386841,
+      "learning_rate": 9.920465028994136e-06,
+      "loss": 0.6214,
+      "step": 5230
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.5244600772857666,
+      "learning_rate": 9.920418182214158e-06,
+      "loss": 0.6602,
+      "step": 5231
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.2152256965637207,
+      "learning_rate": 9.92037132175234e-06,
+      "loss": 0.7538,
+      "step": 5232
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.5862960815429688,
+      "learning_rate": 9.92032444760881e-06,
+      "loss": 0.4923,
+      "step": 5233
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 8.77050495147705,
+      "learning_rate": 9.920277559783705e-06,
+      "loss": 0.6781,
+      "step": 5234
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.4828710556030273,
+      "learning_rate": 9.920230658277148e-06,
+      "loss": 0.7275,
+      "step": 5235
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 4.132396221160889,
+      "learning_rate": 9.920183743089274e-06,
+      "loss": 0.6285,
+      "step": 5236
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 4.056614875793457,
+      "learning_rate": 9.920136814220213e-06,
+      "loss": 0.895,
+      "step": 5237
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.9756898880004883,
+      "learning_rate": 9.920089871670093e-06,
+      "loss": 0.708,
+      "step": 5238
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.6245298385620117,
+      "learning_rate": 9.920042915439048e-06,
+      "loss": 0.6074,
+      "step": 5239
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.103755235671997,
+      "learning_rate": 9.919995945527205e-06,
+      "loss": 0.586,
+      "step": 5240
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.5785024166107178,
+      "learning_rate": 9.919948961934697e-06,
+      "loss": 0.7284,
+      "step": 5241
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.109410285949707,
+      "learning_rate": 9.919901964661653e-06,
+      "loss": 0.4538,
+      "step": 5242
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.1471798419952393,
+      "learning_rate": 9.919854953708206e-06,
+      "loss": 0.7991,
+      "step": 5243
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 4.783324718475342,
+      "learning_rate": 9.919807929074485e-06,
+      "loss": 0.7011,
+      "step": 5244
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 4.174393653869629,
+      "learning_rate": 9.919760890760622e-06,
+      "loss": 0.9343,
+      "step": 5245
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.8799155950546265,
+      "learning_rate": 9.919713838766747e-06,
+      "loss": 0.7005,
+      "step": 5246
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.0774171352386475,
+      "learning_rate": 9.919666773092988e-06,
+      "loss": 0.7648,
+      "step": 5247
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.4154216051101685,
+      "learning_rate": 9.919619693739482e-06,
+      "loss": 0.7934,
+      "step": 5248
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.7711689472198486,
+      "learning_rate": 9.919572600706355e-06,
+      "loss": 0.8126,
+      "step": 5249
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 4.5128021240234375,
+      "learning_rate": 9.91952549399374e-06,
+      "loss": 0.7052,
+      "step": 5250
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 5.7295331954956055,
+      "learning_rate": 9.919478373601768e-06,
+      "loss": 0.5706,
+      "step": 5251
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.6471796035766602,
+      "learning_rate": 9.919431239530569e-06,
+      "loss": 0.6397,
+      "step": 5252
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 4.002257823944092,
+      "learning_rate": 9.919384091780273e-06,
+      "loss": 0.1961,
+      "step": 5253
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.920124053955078,
+      "learning_rate": 9.919336930351013e-06,
+      "loss": 0.4985,
+      "step": 5254
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.57126522064209,
+      "learning_rate": 9.919289755242923e-06,
+      "loss": 0.6879,
+      "step": 5255
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.1220693588256836,
+      "learning_rate": 9.919242566456128e-06,
+      "loss": 0.7209,
+      "step": 5256
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 4.1659464836120605,
+      "learning_rate": 9.919195363990763e-06,
+      "loss": 0.613,
+      "step": 5257
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.8642420768737793,
+      "learning_rate": 9.919148147846957e-06,
+      "loss": 0.7491,
+      "step": 5258
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.9596731662750244,
+      "learning_rate": 9.919100918024845e-06,
+      "loss": 0.5957,
+      "step": 5259
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.429723024368286,
+      "learning_rate": 9.919053674524556e-06,
+      "loss": 0.7098,
+      "step": 5260
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.910616397857666,
+      "learning_rate": 9.91900641734622e-06,
+      "loss": 0.8802,
+      "step": 5261
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.9715631008148193,
+      "learning_rate": 9.91895914648997e-06,
+      "loss": 0.6637,
+      "step": 5262
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.857480049133301,
+      "learning_rate": 9.918911861955939e-06,
+      "loss": 0.843,
+      "step": 5263
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.7290210723876953,
+      "learning_rate": 9.918864563744256e-06,
+      "loss": 0.7586,
+      "step": 5264
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 4.303942680358887,
+      "learning_rate": 9.918817251855052e-06,
+      "loss": 0.6889,
+      "step": 5265
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.49204421043396,
+      "learning_rate": 9.918769926288462e-06,
+      "loss": 0.5964,
+      "step": 5266
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.861846446990967,
+      "learning_rate": 9.918722587044614e-06,
+      "loss": 0.8084,
+      "step": 5267
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 7.109946250915527,
+      "learning_rate": 9.91867523412364e-06,
+      "loss": 0.7389,
+      "step": 5268
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.3494606018066406,
+      "learning_rate": 9.918627867525675e-06,
+      "loss": 0.7614,
+      "step": 5269
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.078080892562866,
+      "learning_rate": 9.918580487250848e-06,
+      "loss": 0.5414,
+      "step": 5270
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.1972455978393555,
+      "learning_rate": 9.91853309329929e-06,
+      "loss": 0.6002,
+      "step": 5271
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.967834234237671,
+      "learning_rate": 9.918485685671135e-06,
+      "loss": 0.6073,
+      "step": 5272
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.231969118118286,
+      "learning_rate": 9.918438264366515e-06,
+      "loss": 0.6351,
+      "step": 5273
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 8.602578163146973,
+      "learning_rate": 9.918390829385559e-06,
+      "loss": 0.4191,
+      "step": 5274
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 4.043108940124512,
+      "learning_rate": 9.918343380728402e-06,
+      "loss": 0.3496,
+      "step": 5275
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.5573408603668213,
+      "learning_rate": 9.918295918395172e-06,
+      "loss": 0.624,
+      "step": 5276
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.286186456680298,
+      "learning_rate": 9.918248442386005e-06,
+      "loss": 0.4161,
+      "step": 5277
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.715458869934082,
+      "learning_rate": 9.918200952701032e-06,
+      "loss": 0.6388,
+      "step": 5278
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.5681252479553223,
+      "learning_rate": 9.918153449340383e-06,
+      "loss": 0.639,
+      "step": 5279
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 4.493866443634033,
+      "learning_rate": 9.918105932304192e-06,
+      "loss": 0.7501,
+      "step": 5280
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.9152679443359375,
+      "learning_rate": 9.918058401592592e-06,
+      "loss": 0.5691,
+      "step": 5281
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.486222267150879,
+      "learning_rate": 9.918010857205713e-06,
+      "loss": 0.7439,
+      "step": 5282
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.5750603675842285,
+      "learning_rate": 9.91796329914369e-06,
+      "loss": 1.0317,
+      "step": 5283
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.836880922317505,
+      "learning_rate": 9.91791572740665e-06,
+      "loss": 0.7244,
+      "step": 5284
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.9766838550567627,
+      "learning_rate": 9.91786814199473e-06,
+      "loss": 0.6066,
+      "step": 5285
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.1924564838409424,
+      "learning_rate": 9.917820542908063e-06,
+      "loss": 0.3943,
+      "step": 5286
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.5244054794311523,
+      "learning_rate": 9.917772930146776e-06,
+      "loss": 0.6761,
+      "step": 5287
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.7056140899658203,
+      "learning_rate": 9.917725303711007e-06,
+      "loss": 0.7478,
+      "step": 5288
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 5.662286281585693,
+      "learning_rate": 9.917677663600887e-06,
+      "loss": 0.6643,
+      "step": 5289
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.821402072906494,
+      "learning_rate": 9.917630009816546e-06,
+      "loss": 0.6632,
+      "step": 5290
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.6993694305419922,
+      "learning_rate": 9.917582342358117e-06,
+      "loss": 0.7501,
+      "step": 5291
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.362483501434326,
+      "learning_rate": 9.917534661225736e-06,
+      "loss": 0.778,
+      "step": 5292
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.6102371215820312,
+      "learning_rate": 9.917486966419532e-06,
+      "loss": 0.7936,
+      "step": 5293
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.325394630432129,
+      "learning_rate": 9.917439257939638e-06,
+      "loss": 0.8452,
+      "step": 5294
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.7984466552734375,
+      "learning_rate": 9.91739153578619e-06,
+      "loss": 0.7211,
+      "step": 5295
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.667282819747925,
+      "learning_rate": 9.917343799959317e-06,
+      "loss": 0.8121,
+      "step": 5296
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.7063825130462646,
+      "learning_rate": 9.917296050459153e-06,
+      "loss": 0.4928,
+      "step": 5297
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.9320454597473145,
+      "learning_rate": 9.91724828728583e-06,
+      "loss": 0.7173,
+      "step": 5298
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.1436190605163574,
+      "learning_rate": 9.917200510439481e-06,
+      "loss": 0.7375,
+      "step": 5299
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.0903241634368896,
+      "learning_rate": 9.917152719920242e-06,
+      "loss": 0.6496,
+      "step": 5300
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.446190118789673,
+      "learning_rate": 9.917104915728242e-06,
+      "loss": 0.7243,
+      "step": 5301
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.5614514350891113,
+      "learning_rate": 9.917057097863614e-06,
+      "loss": 0.5792,
+      "step": 5302
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.1147842407226562,
+      "learning_rate": 9.917009266326494e-06,
+      "loss": 0.5796,
+      "step": 5303
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.0360000133514404,
+      "learning_rate": 9.916961421117012e-06,
+      "loss": 0.9179,
+      "step": 5304
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.0958282947540283,
+      "learning_rate": 9.916913562235304e-06,
+      "loss": 0.7957,
+      "step": 5305
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 6.990150451660156,
+      "learning_rate": 9.9168656896815e-06,
+      "loss": 0.4513,
+      "step": 5306
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.4062929153442383,
+      "learning_rate": 9.916817803455734e-06,
+      "loss": 0.6542,
+      "step": 5307
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 5.503777980804443,
+      "learning_rate": 9.91676990355814e-06,
+      "loss": 0.7515,
+      "step": 5308
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 5.037022113800049,
+      "learning_rate": 9.916721989988851e-06,
+      "loss": 0.6578,
+      "step": 5309
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.603595018386841,
+      "learning_rate": 9.916674062747999e-06,
+      "loss": 0.7791,
+      "step": 5310
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 5.721814155578613,
+      "learning_rate": 9.91662612183572e-06,
+      "loss": 0.7081,
+      "step": 5311
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.6823019981384277,
+      "learning_rate": 9.916578167252145e-06,
+      "loss": 0.6498,
+      "step": 5312
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.6145825386047363,
+      "learning_rate": 9.916530198997407e-06,
+      "loss": 0.785,
+      "step": 5313
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.355909824371338,
+      "learning_rate": 9.916482217071642e-06,
+      "loss": 0.6334,
+      "step": 5314
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.9312424659729004,
+      "learning_rate": 9.916434221474979e-06,
+      "loss": 0.615,
+      "step": 5315
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.614100217819214,
+      "learning_rate": 9.916386212207556e-06,
+      "loss": 0.6347,
+      "step": 5316
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.0300800800323486,
+      "learning_rate": 9.916338189269503e-06,
+      "loss": 0.491,
+      "step": 5317
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.142808198928833,
+      "learning_rate": 9.916290152660956e-06,
+      "loss": 0.7181,
+      "step": 5318
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.643930673599243,
+      "learning_rate": 9.91624210238205e-06,
+      "loss": 0.4756,
+      "step": 5319
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.629486083984375,
+      "learning_rate": 9.916194038432914e-06,
+      "loss": 0.8055,
+      "step": 5320
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.396794319152832,
+      "learning_rate": 9.916145960813685e-06,
+      "loss": 0.7075,
+      "step": 5321
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.1053338050842285,
+      "learning_rate": 9.916097869524496e-06,
+      "loss": 0.4246,
+      "step": 5322
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.3868918418884277,
+      "learning_rate": 9.91604976456548e-06,
+      "loss": 0.7741,
+      "step": 5323
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.3521530628204346,
+      "learning_rate": 9.91600164593677e-06,
+      "loss": 0.8284,
+      "step": 5324
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.6226651668548584,
+      "learning_rate": 9.915953513638503e-06,
+      "loss": 0.6917,
+      "step": 5325
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.460408926010132,
+      "learning_rate": 9.91590536767081e-06,
+      "loss": 0.5755,
+      "step": 5326
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.2083475589752197,
+      "learning_rate": 9.915857208033826e-06,
+      "loss": 0.7433,
+      "step": 5327
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.1281630992889404,
+      "learning_rate": 9.915809034727684e-06,
+      "loss": 0.5781,
+      "step": 5328
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.665128469467163,
+      "learning_rate": 9.915760847752518e-06,
+      "loss": 0.9447,
+      "step": 5329
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 4.074838161468506,
+      "learning_rate": 9.915712647108464e-06,
+      "loss": 0.844,
+      "step": 5330
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.682117223739624,
+      "learning_rate": 9.915664432795652e-06,
+      "loss": 0.7365,
+      "step": 5331
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.988886833190918,
+      "learning_rate": 9.915616204814223e-06,
+      "loss": 0.6056,
+      "step": 5332
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.4413514137268066,
+      "learning_rate": 9.915567963164304e-06,
+      "loss": 0.5912,
+      "step": 5333
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.8156046867370605,
+      "learning_rate": 9.915519707846032e-06,
+      "loss": 0.8579,
+      "step": 5334
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 15.050984382629395,
+      "learning_rate": 9.915471438859541e-06,
+      "loss": 0.7002,
+      "step": 5335
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.7258856296539307,
+      "learning_rate": 9.915423156204966e-06,
+      "loss": 1.0727,
+      "step": 5336
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.384880542755127,
+      "learning_rate": 9.915374859882441e-06,
+      "loss": 0.5649,
+      "step": 5337
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.927659511566162,
+      "learning_rate": 9.915326549892098e-06,
+      "loss": 0.6185,
+      "step": 5338
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.6920201778411865,
+      "learning_rate": 9.915278226234075e-06,
+      "loss": 0.8557,
+      "step": 5339
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.547335624694824,
+      "learning_rate": 9.915229888908503e-06,
+      "loss": 0.7688,
+      "step": 5340
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.9113314151763916,
+      "learning_rate": 9.91518153791552e-06,
+      "loss": 0.6009,
+      "step": 5341
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.4327104091644287,
+      "learning_rate": 9.915133173255256e-06,
+      "loss": 0.576,
+      "step": 5342
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.808893918991089,
+      "learning_rate": 9.91508479492785e-06,
+      "loss": 0.9623,
+      "step": 5343
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 6.47015380859375,
+      "learning_rate": 9.915036402933433e-06,
+      "loss": 0.4744,
+      "step": 5344
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.8927366733551025,
+      "learning_rate": 9.914987997272142e-06,
+      "loss": 0.7763,
+      "step": 5345
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.3465397357940674,
+      "learning_rate": 9.91493957794411e-06,
+      "loss": 0.6678,
+      "step": 5346
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.2926104068756104,
+      "learning_rate": 9.914891144949473e-06,
+      "loss": 0.537,
+      "step": 5347
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.1570467948913574,
+      "learning_rate": 9.914842698288363e-06,
+      "loss": 0.701,
+      "step": 5348
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.1392788887023926,
+      "learning_rate": 9.914794237960919e-06,
+      "loss": 0.6775,
+      "step": 5349
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 4.7054290771484375,
+      "learning_rate": 9.91474576396727e-06,
+      "loss": 0.8276,
+      "step": 5350
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.515754461288452,
+      "learning_rate": 9.914697276307558e-06,
+      "loss": 0.5897,
+      "step": 5351
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.775911569595337,
+      "learning_rate": 9.914648774981912e-06,
+      "loss": 0.5361,
+      "step": 5352
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.54242205619812,
+      "learning_rate": 9.91460025999047e-06,
+      "loss": 0.6528,
+      "step": 5353
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.6432323455810547,
+      "learning_rate": 9.914551731333364e-06,
+      "loss": 0.8904,
+      "step": 5354
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.417612075805664,
+      "learning_rate": 9.914503189010731e-06,
+      "loss": 0.6783,
+      "step": 5355
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.7319285869598389,
+      "learning_rate": 9.914454633022706e-06,
+      "loss": 0.8634,
+      "step": 5356
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.385931968688965,
+      "learning_rate": 9.914406063369424e-06,
+      "loss": 0.6116,
+      "step": 5357
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.4644904136657715,
+      "learning_rate": 9.91435748005102e-06,
+      "loss": 0.5004,
+      "step": 5358
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.9581161737442017,
+      "learning_rate": 9.914308883067627e-06,
+      "loss": 0.9049,
+      "step": 5359
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 4.016529083251953,
+      "learning_rate": 9.914260272419385e-06,
+      "loss": 0.865,
+      "step": 5360
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.8166937828063965,
+      "learning_rate": 9.914211648106424e-06,
+      "loss": 0.5111,
+      "step": 5361
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 4.9022016525268555,
+      "learning_rate": 9.91416301012888e-06,
+      "loss": 0.6868,
+      "step": 5362
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.947340965270996,
+      "learning_rate": 9.914114358486893e-06,
+      "loss": 0.9067,
+      "step": 5363
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.8818180561065674,
+      "learning_rate": 9.914065693180593e-06,
+      "loss": 0.6915,
+      "step": 5364
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 4.5084428787231445,
+      "learning_rate": 9.914017014210116e-06,
+      "loss": 0.8225,
+      "step": 5365
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.834318161010742,
+      "learning_rate": 9.9139683215756e-06,
+      "loss": 0.6687,
+      "step": 5366
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.34153413772583,
+      "learning_rate": 9.91391961527718e-06,
+      "loss": 0.7527,
+      "step": 5367
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.7408273220062256,
+      "learning_rate": 9.913870895314987e-06,
+      "loss": 0.5463,
+      "step": 5368
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.7726633548736572,
+      "learning_rate": 9.913822161689164e-06,
+      "loss": 0.6502,
+      "step": 5369
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.3466899394989014,
+      "learning_rate": 9.913773414399839e-06,
+      "loss": 0.6039,
+      "step": 5370
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.8615334033966064,
+      "learning_rate": 9.913724653447152e-06,
+      "loss": 0.684,
+      "step": 5371
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.7877583503723145,
+      "learning_rate": 9.913675878831236e-06,
+      "loss": 0.8908,
+      "step": 5372
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.2049901485443115,
+      "learning_rate": 9.913627090552232e-06,
+      "loss": 0.6106,
+      "step": 5373
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.735384702682495,
+      "learning_rate": 9.913578288610269e-06,
+      "loss": 0.7082,
+      "step": 5374
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.0123469829559326,
+      "learning_rate": 9.913529473005485e-06,
+      "loss": 0.5049,
+      "step": 5375
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.252136468887329,
+      "learning_rate": 9.913480643738017e-06,
+      "loss": 0.7213,
+      "step": 5376
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.33750057220459,
+      "learning_rate": 9.913431800808e-06,
+      "loss": 0.6245,
+      "step": 5377
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.25045108795166,
+      "learning_rate": 9.91338294421557e-06,
+      "loss": 0.5186,
+      "step": 5378
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 4.188529014587402,
+      "learning_rate": 9.913334073960862e-06,
+      "loss": 0.9934,
+      "step": 5379
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.417461395263672,
+      "learning_rate": 9.913285190044013e-06,
+      "loss": 0.7591,
+      "step": 5380
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.32053804397583,
+      "learning_rate": 9.913236292465158e-06,
+      "loss": 0.6567,
+      "step": 5381
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.5329782962799072,
+      "learning_rate": 9.913187381224433e-06,
+      "loss": 0.7913,
+      "step": 5382
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.830043315887451,
+      "learning_rate": 9.913138456321976e-06,
+      "loss": 0.7845,
+      "step": 5383
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.899848461151123,
+      "learning_rate": 9.913089517757922e-06,
+      "loss": 0.6671,
+      "step": 5384
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.536496162414551,
+      "learning_rate": 9.913040565532404e-06,
+      "loss": 0.9461,
+      "step": 5385
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.7892701625823975,
+      "learning_rate": 9.912991599645562e-06,
+      "loss": 0.888,
+      "step": 5386
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 4.132125377655029,
+      "learning_rate": 9.91294262009753e-06,
+      "loss": 0.8516,
+      "step": 5387
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.829312801361084,
+      "learning_rate": 9.912893626888446e-06,
+      "loss": 0.6431,
+      "step": 5388
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.1381144523620605,
+      "learning_rate": 9.912844620018444e-06,
+      "loss": 0.8256,
+      "step": 5389
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.6444268226623535,
+      "learning_rate": 9.912795599487664e-06,
+      "loss": 0.8,
+      "step": 5390
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 5.542173862457275,
+      "learning_rate": 9.912746565296236e-06,
+      "loss": 0.5383,
+      "step": 5391
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 5.638376712799072,
+      "learning_rate": 9.912697517444304e-06,
+      "loss": 0.6567,
+      "step": 5392
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.8813674449920654,
+      "learning_rate": 9.912648455931998e-06,
+      "loss": 0.5782,
+      "step": 5393
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.544952869415283,
+      "learning_rate": 9.912599380759458e-06,
+      "loss": 0.5644,
+      "step": 5394
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.2147715091705322,
+      "learning_rate": 9.91255029192682e-06,
+      "loss": 0.7626,
+      "step": 5395
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.6778457164764404,
+      "learning_rate": 9.912501189434217e-06,
+      "loss": 0.8433,
+      "step": 5396
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.0974647998809814,
+      "learning_rate": 9.912452073281791e-06,
+      "loss": 0.6437,
+      "step": 5397
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.3269519805908203,
+      "learning_rate": 9.912402943469677e-06,
+      "loss": 0.5421,
+      "step": 5398
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.046971559524536,
+      "learning_rate": 9.912353799998009e-06,
+      "loss": 0.4884,
+      "step": 5399
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.319805860519409,
+      "learning_rate": 9.912304642866925e-06,
+      "loss": 0.5535,
+      "step": 5400
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.4468116760253906,
+      "learning_rate": 9.912255472076563e-06,
+      "loss": 0.6868,
+      "step": 5401
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.77028489112854,
+      "learning_rate": 9.912206287627057e-06,
+      "loss": 0.6897,
+      "step": 5402
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.998310923576355,
+      "learning_rate": 9.912157089518549e-06,
+      "loss": 0.7093,
+      "step": 5403
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.5981762409210205,
+      "learning_rate": 9.91210787775117e-06,
+      "loss": 0.7925,
+      "step": 5404
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.7316888570785522,
+      "learning_rate": 9.912058652325058e-06,
+      "loss": 0.6138,
+      "step": 5405
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.80198073387146,
+      "learning_rate": 9.912009413240352e-06,
+      "loss": 0.9701,
+      "step": 5406
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.6475913524627686,
+      "learning_rate": 9.911960160497187e-06,
+      "loss": 0.694,
+      "step": 5407
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 4.863465309143066,
+      "learning_rate": 9.911910894095702e-06,
+      "loss": 0.7535,
+      "step": 5408
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.7312525510787964,
+      "learning_rate": 9.911861614036034e-06,
+      "loss": 0.8481,
+      "step": 5409
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.959958791732788,
+      "learning_rate": 9.911812320318316e-06,
+      "loss": 0.7666,
+      "step": 5410
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.511225938796997,
+      "learning_rate": 9.91176301294269e-06,
+      "loss": 0.7143,
+      "step": 5411
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.8734902143478394,
+      "learning_rate": 9.91171369190929e-06,
+      "loss": 0.4354,
+      "step": 5412
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 7.787213325500488,
+      "learning_rate": 9.911664357218256e-06,
+      "loss": 0.7055,
+      "step": 5413
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.9821077585220337,
+      "learning_rate": 9.911615008869722e-06,
+      "loss": 0.4947,
+      "step": 5414
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.991072416305542,
+      "learning_rate": 9.911565646863826e-06,
+      "loss": 0.4594,
+      "step": 5415
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.753063440322876,
+      "learning_rate": 9.911516271200707e-06,
+      "loss": 0.5915,
+      "step": 5416
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.2070252895355225,
+      "learning_rate": 9.911466881880501e-06,
+      "loss": 0.7473,
+      "step": 5417
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.105334520339966,
+      "learning_rate": 9.911417478903344e-06,
+      "loss": 0.6071,
+      "step": 5418
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.6315536499023438,
+      "learning_rate": 9.911368062269377e-06,
+      "loss": 0.6588,
+      "step": 5419
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.394829750061035,
+      "learning_rate": 9.911318631978735e-06,
+      "loss": 0.5768,
+      "step": 5420
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 5.870758056640625,
+      "learning_rate": 9.911269188031554e-06,
+      "loss": 0.4407,
+      "step": 5421
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.262794256210327,
+      "learning_rate": 9.911219730427974e-06,
+      "loss": 0.7704,
+      "step": 5422
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 4.415421485900879,
+      "learning_rate": 9.911170259168132e-06,
+      "loss": 0.6975,
+      "step": 5423
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 5.090567588806152,
+      "learning_rate": 9.911120774252165e-06,
+      "loss": 0.9801,
+      "step": 5424
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.7796475887298584,
+      "learning_rate": 9.911071275680211e-06,
+      "loss": 0.6027,
+      "step": 5425
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 4.744011878967285,
+      "learning_rate": 9.911021763452408e-06,
+      "loss": 0.6306,
+      "step": 5426
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.8435797691345215,
+      "learning_rate": 9.910972237568893e-06,
+      "loss": 0.645,
+      "step": 5427
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.284518003463745,
+      "learning_rate": 9.910922698029803e-06,
+      "loss": 0.724,
+      "step": 5428
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.865811824798584,
+      "learning_rate": 9.910873144835278e-06,
+      "loss": 0.7087,
+      "step": 5429
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.6782426834106445,
+      "learning_rate": 9.910823577985454e-06,
+      "loss": 0.8757,
+      "step": 5430
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.338423490524292,
+      "learning_rate": 9.910773997480469e-06,
+      "loss": 0.5729,
+      "step": 5431
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.591352701187134,
+      "learning_rate": 9.910724403320461e-06,
+      "loss": 0.6676,
+      "step": 5432
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.1741769313812256,
+      "learning_rate": 9.910674795505569e-06,
+      "loss": 0.4937,
+      "step": 5433
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.481224775314331,
+      "learning_rate": 9.910625174035929e-06,
+      "loss": 0.5024,
+      "step": 5434
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.1623475551605225,
+      "learning_rate": 9.91057553891168e-06,
+      "loss": 0.6653,
+      "step": 5435
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.183617353439331,
+      "learning_rate": 9.91052589013296e-06,
+      "loss": 0.7153,
+      "step": 5436
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 4.389906406402588,
+      "learning_rate": 9.910476227699907e-06,
+      "loss": 0.6869,
+      "step": 5437
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.1152002811431885,
+      "learning_rate": 9.91042655161266e-06,
+      "loss": 0.4769,
+      "step": 5438
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.6599197387695312,
+      "learning_rate": 9.910376861871354e-06,
+      "loss": 0.6567,
+      "step": 5439
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.270524501800537,
+      "learning_rate": 9.910327158476132e-06,
+      "loss": 0.7298,
+      "step": 5440
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.033132791519165,
+      "learning_rate": 9.910277441427127e-06,
+      "loss": 0.8551,
+      "step": 5441
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.664698839187622,
+      "learning_rate": 9.910227710724483e-06,
+      "loss": 0.6643,
+      "step": 5442
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.8927135467529297,
+      "learning_rate": 9.910177966368333e-06,
+      "loss": 0.6924,
+      "step": 5443
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 4.7649922370910645,
+      "learning_rate": 9.910128208358818e-06,
+      "loss": 0.7399,
+      "step": 5444
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.5114731788635254,
+      "learning_rate": 9.910078436696076e-06,
+      "loss": 0.6036,
+      "step": 5445
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.6532108783721924,
+      "learning_rate": 9.910028651380244e-06,
+      "loss": 0.7884,
+      "step": 5446
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.9575512409210205,
+      "learning_rate": 9.909978852411464e-06,
+      "loss": 0.7087,
+      "step": 5447
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 6.984023571014404,
+      "learning_rate": 9.90992903978987e-06,
+      "loss": 0.9718,
+      "step": 5448
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.5009400844573975,
+      "learning_rate": 9.909879213515604e-06,
+      "loss": 0.3511,
+      "step": 5449
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.607396364212036,
+      "learning_rate": 9.909829373588802e-06,
+      "loss": 0.6996,
+      "step": 5450
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.181567907333374,
+      "learning_rate": 9.909779520009605e-06,
+      "loss": 0.3587,
+      "step": 5451
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.374143362045288,
+      "learning_rate": 9.90972965277815e-06,
+      "loss": 0.8672,
+      "step": 5452
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.4932985305786133,
+      "learning_rate": 9.909679771894576e-06,
+      "loss": 0.626,
+      "step": 5453
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 4.262042999267578,
+      "learning_rate": 9.909629877359022e-06,
+      "loss": 0.8717,
+      "step": 5454
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.2390661239624023,
+      "learning_rate": 9.909579969171625e-06,
+      "loss": 0.6595,
+      "step": 5455
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.0740368366241455,
+      "learning_rate": 9.909530047332527e-06,
+      "loss": 0.6004,
+      "step": 5456
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.9996933937072754,
+      "learning_rate": 9.909480111841865e-06,
+      "loss": 0.8953,
+      "step": 5457
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 6.487476825714111,
+      "learning_rate": 9.909430162699777e-06,
+      "loss": 0.5996,
+      "step": 5458
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.5490176677703857,
+      "learning_rate": 9.909380199906403e-06,
+      "loss": 0.6426,
+      "step": 5459
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.5623180866241455,
+      "learning_rate": 9.909330223461881e-06,
+      "loss": 0.8248,
+      "step": 5460
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.61197566986084,
+      "learning_rate": 9.909280233366353e-06,
+      "loss": 0.4585,
+      "step": 5461
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.9677224159240723,
+      "learning_rate": 9.909230229619955e-06,
+      "loss": 0.5793,
+      "step": 5462
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.2408957481384277,
+      "learning_rate": 9.909180212222826e-06,
+      "loss": 0.5961,
+      "step": 5463
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.4056944847106934,
+      "learning_rate": 9.909130181175106e-06,
+      "loss": 0.6307,
+      "step": 5464
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.502270460128784,
+      "learning_rate": 9.909080136476933e-06,
+      "loss": 0.5416,
+      "step": 5465
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.8084592819213867,
+      "learning_rate": 9.90903007812845e-06,
+      "loss": 0.7635,
+      "step": 5466
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.7718217372894287,
+      "learning_rate": 9.908980006129791e-06,
+      "loss": 0.9543,
+      "step": 5467
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.233266592025757,
+      "learning_rate": 9.908929920481099e-06,
+      "loss": 0.4757,
+      "step": 5468
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 4.709874629974365,
+      "learning_rate": 9.90887982118251e-06,
+      "loss": 0.7781,
+      "step": 5469
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 4.187461853027344,
+      "learning_rate": 9.908829708234166e-06,
+      "loss": 0.7989,
+      "step": 5470
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.8258793354034424,
+      "learning_rate": 9.908779581636205e-06,
+      "loss": 0.5336,
+      "step": 5471
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.0009219646453857,
+      "learning_rate": 9.908729441388769e-06,
+      "loss": 0.442,
+      "step": 5472
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.6204779148101807,
+      "learning_rate": 9.908679287491993e-06,
+      "loss": 0.744,
+      "step": 5473
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.181673288345337,
+      "learning_rate": 9.908629119946018e-06,
+      "loss": 0.702,
+      "step": 5474
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 9.494287490844727,
+      "learning_rate": 9.908578938750986e-06,
+      "loss": 0.8612,
+      "step": 5475
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.9965548515319824,
+      "learning_rate": 9.908528743907035e-06,
+      "loss": 0.707,
+      "step": 5476
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.716226100921631,
+      "learning_rate": 9.908478535414303e-06,
+      "loss": 0.8547,
+      "step": 5477
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.2533740997314453,
+      "learning_rate": 9.908428313272933e-06,
+      "loss": 0.6705,
+      "step": 5478
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.2253856658935547,
+      "learning_rate": 9.90837807748306e-06,
+      "loss": 0.4429,
+      "step": 5479
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.7155673503875732,
+      "learning_rate": 9.908327828044829e-06,
+      "loss": 0.9211,
+      "step": 5480
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.6310176849365234,
+      "learning_rate": 9.908277564958375e-06,
+      "loss": 0.7216,
+      "step": 5481
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.131291389465332,
+      "learning_rate": 9.90822728822384e-06,
+      "loss": 0.8306,
+      "step": 5482
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.3691213130950928,
+      "learning_rate": 9.908176997841363e-06,
+      "loss": 0.7712,
+      "step": 5483
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.4671406745910645,
+      "learning_rate": 9.908126693811086e-06,
+      "loss": 0.7189,
+      "step": 5484
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.1580541133880615,
+      "learning_rate": 9.908076376133146e-06,
+      "loss": 0.7363,
+      "step": 5485
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.1955406665802,
+      "learning_rate": 9.908026044807686e-06,
+      "loss": 0.8474,
+      "step": 5486
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.3951938152313232,
+      "learning_rate": 9.907975699834842e-06,
+      "loss": 0.509,
+      "step": 5487
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.450359344482422,
+      "learning_rate": 9.907925341214758e-06,
+      "loss": 0.7611,
+      "step": 5488
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.4011173248291016,
+      "learning_rate": 9.907874968947571e-06,
+      "loss": 0.6131,
+      "step": 5489
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.4999630451202393,
+      "learning_rate": 9.907824583033422e-06,
+      "loss": 0.6258,
+      "step": 5490
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.7228872776031494,
+      "learning_rate": 9.907774183472453e-06,
+      "loss": 0.5721,
+      "step": 5491
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.7757785320281982,
+      "learning_rate": 9.9077237702648e-06,
+      "loss": 0.8849,
+      "step": 5492
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.922917366027832,
+      "learning_rate": 9.907673343410609e-06,
+      "loss": 0.8371,
+      "step": 5493
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.63482928276062,
+      "learning_rate": 9.907622902910015e-06,
+      "loss": 0.4612,
+      "step": 5494
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.415576219558716,
+      "learning_rate": 9.907572448763158e-06,
+      "loss": 0.9792,
+      "step": 5495
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.385510206222534,
+      "learning_rate": 9.907521980970183e-06,
+      "loss": 0.4918,
+      "step": 5496
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.926492929458618,
+      "learning_rate": 9.907471499531228e-06,
+      "loss": 0.6448,
+      "step": 5497
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.2717297077178955,
+      "learning_rate": 9.907421004446432e-06,
+      "loss": 0.3932,
+      "step": 5498
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.8856420516967773,
+      "learning_rate": 9.907370495715938e-06,
+      "loss": 0.7035,
+      "step": 5499
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.317945718765259,
+      "learning_rate": 9.907319973339884e-06,
+      "loss": 0.8184,
+      "step": 5500
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 5.481766700744629,
+      "learning_rate": 9.907269437318412e-06,
+      "loss": 0.6652,
+      "step": 5501
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.1780588626861572,
+      "learning_rate": 9.90721888765166e-06,
+      "loss": 0.816,
+      "step": 5502
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.0211384296417236,
+      "learning_rate": 9.907168324339773e-06,
+      "loss": 0.4949,
+      "step": 5503
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.7901332378387451,
+      "learning_rate": 9.907117747382889e-06,
+      "loss": 0.4485,
+      "step": 5504
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.5724880695343018,
+      "learning_rate": 9.907067156781147e-06,
+      "loss": 0.6856,
+      "step": 5505
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 5.424927711486816,
+      "learning_rate": 9.907016552534691e-06,
+      "loss": 0.433,
+      "step": 5506
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 4.499792098999023,
+      "learning_rate": 9.90696593464366e-06,
+      "loss": 0.5723,
+      "step": 5507
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 6.002030372619629,
+      "learning_rate": 9.906915303108193e-06,
+      "loss": 0.6489,
+      "step": 5508
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.407867431640625,
+      "learning_rate": 9.906864657928434e-06,
+      "loss": 0.6402,
+      "step": 5509
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.037461519241333,
+      "learning_rate": 9.906813999104523e-06,
+      "loss": 0.563,
+      "step": 5510
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.091780662536621,
+      "learning_rate": 9.906763326636601e-06,
+      "loss": 0.6213,
+      "step": 5511
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.9798340797424316,
+      "learning_rate": 9.906712640524808e-06,
+      "loss": 0.8485,
+      "step": 5512
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.0386929512023926,
+      "learning_rate": 9.906661940769284e-06,
+      "loss": 0.6718,
+      "step": 5513
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 6.870899200439453,
+      "learning_rate": 9.906611227370172e-06,
+      "loss": 0.501,
+      "step": 5514
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.950782060623169,
+      "learning_rate": 9.906560500327612e-06,
+      "loss": 0.6521,
+      "step": 5515
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.3619766235351562,
+      "learning_rate": 9.906509759641746e-06,
+      "loss": 0.7899,
+      "step": 5516
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.6483019590377808,
+      "learning_rate": 9.906459005312715e-06,
+      "loss": 0.7974,
+      "step": 5517
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.4790382385253906,
+      "learning_rate": 9.906408237340657e-06,
+      "loss": 0.441,
+      "step": 5518
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 4.981144905090332,
+      "learning_rate": 9.906357455725718e-06,
+      "loss": 0.7943,
+      "step": 5519
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.0199718475341797,
+      "learning_rate": 9.906306660468035e-06,
+      "loss": 0.5927,
+      "step": 5520
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.0413615703582764,
+      "learning_rate": 9.906255851567752e-06,
+      "loss": 0.6949,
+      "step": 5521
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.205746650695801,
+      "learning_rate": 9.90620502902501e-06,
+      "loss": 0.7579,
+      "step": 5522
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.6882283687591553,
+      "learning_rate": 9.90615419283995e-06,
+      "loss": 0.6453,
+      "step": 5523
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.7200615406036377,
+      "learning_rate": 9.906103343012712e-06,
+      "loss": 0.3281,
+      "step": 5524
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.871185541152954,
+      "learning_rate": 9.906052479543438e-06,
+      "loss": 0.7816,
+      "step": 5525
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 5.741092681884766,
+      "learning_rate": 9.90600160243227e-06,
+      "loss": 0.8631,
+      "step": 5526
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.708817958831787,
+      "learning_rate": 9.90595071167935e-06,
+      "loss": 0.7984,
+      "step": 5527
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.483785390853882,
+      "learning_rate": 9.90589980728482e-06,
+      "loss": 0.6873,
+      "step": 5528
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.1653199195861816,
+      "learning_rate": 9.90584888924882e-06,
+      "loss": 0.5892,
+      "step": 5529
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.794067859649658,
+      "learning_rate": 9.90579795757149e-06,
+      "loss": 0.51,
+      "step": 5530
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 6.141931533813477,
+      "learning_rate": 9.905747012252975e-06,
+      "loss": 0.9254,
+      "step": 5531
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.7705590724945068,
+      "learning_rate": 9.905696053293415e-06,
+      "loss": 0.5271,
+      "step": 5532
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.3027265071868896,
+      "learning_rate": 9.905645080692953e-06,
+      "loss": 0.6514,
+      "step": 5533
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.891833543777466,
+      "learning_rate": 9.90559409445173e-06,
+      "loss": 0.7456,
+      "step": 5534
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 4.753790855407715,
+      "learning_rate": 9.905543094569885e-06,
+      "loss": 0.8351,
+      "step": 5535
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.4266226291656494,
+      "learning_rate": 9.905492081047563e-06,
+      "loss": 0.5891,
+      "step": 5536
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.6506192684173584,
+      "learning_rate": 9.905441053884908e-06,
+      "loss": 0.685,
+      "step": 5537
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.136531352996826,
+      "learning_rate": 9.905390013082057e-06,
+      "loss": 0.5494,
+      "step": 5538
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.8757495880126953,
+      "learning_rate": 9.905338958639152e-06,
+      "loss": 0.6411,
+      "step": 5539
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.777541399002075,
+      "learning_rate": 9.90528789055634e-06,
+      "loss": 0.62,
+      "step": 5540
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.8197999000549316,
+      "learning_rate": 9.905236808833758e-06,
+      "loss": 0.9225,
+      "step": 5541
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.774916887283325,
+      "learning_rate": 9.905185713471552e-06,
+      "loss": 0.5864,
+      "step": 5542
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.1797420978546143,
+      "learning_rate": 9.905134604469861e-06,
+      "loss": 0.7377,
+      "step": 5543
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.9308298826217651,
+      "learning_rate": 9.905083481828828e-06,
+      "loss": 0.5791,
+      "step": 5544
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.2289116382598877,
+      "learning_rate": 9.905032345548595e-06,
+      "loss": 0.5622,
+      "step": 5545
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.5712640285491943,
+      "learning_rate": 9.904981195629305e-06,
+      "loss": 0.6995,
+      "step": 5546
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.506898880004883,
+      "learning_rate": 9.9049300320711e-06,
+      "loss": 0.5698,
+      "step": 5547
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.0836668014526367,
+      "learning_rate": 9.904878854874121e-06,
+      "loss": 0.716,
+      "step": 5548
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.991621732711792,
+      "learning_rate": 9.904827664038512e-06,
+      "loss": 0.7275,
+      "step": 5549
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.930353879928589,
+      "learning_rate": 9.904776459564415e-06,
+      "loss": 0.5746,
+      "step": 5550
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 5.021697998046875,
+      "learning_rate": 9.904725241451971e-06,
+      "loss": 0.6078,
+      "step": 5551
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.578420639038086,
+      "learning_rate": 9.904674009701325e-06,
+      "loss": 0.8106,
+      "step": 5552
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.748788595199585,
+      "learning_rate": 9.904622764312616e-06,
+      "loss": 0.8612,
+      "step": 5553
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.7312769889831543,
+      "learning_rate": 9.90457150528599e-06,
+      "loss": 0.4851,
+      "step": 5554
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.265639543533325,
+      "learning_rate": 9.904520232621588e-06,
+      "loss": 0.6079,
+      "step": 5555
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.028778076171875,
+      "learning_rate": 9.904468946319551e-06,
+      "loss": 0.7144,
+      "step": 5556
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.3869779109954834,
+      "learning_rate": 9.904417646380025e-06,
+      "loss": 0.5159,
+      "step": 5557
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 4.695724010467529,
+      "learning_rate": 9.90436633280315e-06,
+      "loss": 0.5504,
+      "step": 5558
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.386338472366333,
+      "learning_rate": 9.90431500558907e-06,
+      "loss": 0.6444,
+      "step": 5559
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.6703672409057617,
+      "learning_rate": 9.904263664737925e-06,
+      "loss": 0.9073,
+      "step": 5560
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.793233633041382,
+      "learning_rate": 9.904212310249861e-06,
+      "loss": 0.8417,
+      "step": 5561
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.060633420944214,
+      "learning_rate": 9.904160942125021e-06,
+      "loss": 0.4272,
+      "step": 5562
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.5360753536224365,
+      "learning_rate": 9.904109560363544e-06,
+      "loss": 0.4703,
+      "step": 5563
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.427032232284546,
+      "learning_rate": 9.904058164965578e-06,
+      "loss": 0.8036,
+      "step": 5564
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.338034152984619,
+      "learning_rate": 9.90400675593126e-06,
+      "loss": 0.671,
+      "step": 5565
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.0912926197052,
+      "learning_rate": 9.903955333260741e-06,
+      "loss": 0.4924,
+      "step": 5566
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.3456435203552246,
+      "learning_rate": 9.903903896954157e-06,
+      "loss": 0.913,
+      "step": 5567
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.7583820819854736,
+      "learning_rate": 9.903852447011652e-06,
+      "loss": 0.8047,
+      "step": 5568
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.359361171722412,
+      "learning_rate": 9.903800983433373e-06,
+      "loss": 0.8921,
+      "step": 5569
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 5.171976089477539,
+      "learning_rate": 9.903749506219458e-06,
+      "loss": 0.7825,
+      "step": 5570
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 5.154469966888428,
+      "learning_rate": 9.903698015370054e-06,
+      "loss": 0.5518,
+      "step": 5571
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.1635608673095703,
+      "learning_rate": 9.903646510885303e-06,
+      "loss": 0.7234,
+      "step": 5572
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.867529273033142,
+      "learning_rate": 9.903594992765347e-06,
+      "loss": 0.7461,
+      "step": 5573
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 16.346664428710938,
+      "learning_rate": 9.903543461010331e-06,
+      "loss": 0.631,
+      "step": 5574
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.981362819671631,
+      "learning_rate": 9.903491915620398e-06,
+      "loss": 0.4387,
+      "step": 5575
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.539763927459717,
+      "learning_rate": 9.90344035659569e-06,
+      "loss": 0.6175,
+      "step": 5576
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.703094720840454,
+      "learning_rate": 9.903388783936351e-06,
+      "loss": 1.0225,
+      "step": 5577
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.7980916500091553,
+      "learning_rate": 9.903337197642526e-06,
+      "loss": 0.6995,
+      "step": 5578
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.7916929721832275,
+      "learning_rate": 9.903285597714355e-06,
+      "loss": 0.4273,
+      "step": 5579
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.7603919506073,
+      "learning_rate": 9.903233984151985e-06,
+      "loss": 0.6157,
+      "step": 5580
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 14.496685028076172,
+      "learning_rate": 9.903182356955558e-06,
+      "loss": 0.633,
+      "step": 5581
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.1015219688415527,
+      "learning_rate": 9.903130716125217e-06,
+      "loss": 0.8438,
+      "step": 5582
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 4.123937606811523,
+      "learning_rate": 9.903079061661108e-06,
+      "loss": 0.6323,
+      "step": 5583
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.909902811050415,
+      "learning_rate": 9.90302739356337e-06,
+      "loss": 1.046,
+      "step": 5584
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.7994298934936523,
+      "learning_rate": 9.902975711832152e-06,
+      "loss": 0.7928,
+      "step": 5585
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 8.283931732177734,
+      "learning_rate": 9.902924016467595e-06,
+      "loss": 0.7381,
+      "step": 5586
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 4.471031665802002,
+      "learning_rate": 9.902872307469843e-06,
+      "loss": 0.3282,
+      "step": 5587
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.669034719467163,
+      "learning_rate": 9.902820584839037e-06,
+      "loss": 0.6546,
+      "step": 5588
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.7508738040924072,
+      "learning_rate": 9.902768848575327e-06,
+      "loss": 0.7039,
+      "step": 5589
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.2225658893585205,
+      "learning_rate": 9.902717098678851e-06,
+      "loss": 0.5199,
+      "step": 5590
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.7548884153366089,
+      "learning_rate": 9.902665335149756e-06,
+      "loss": 0.652,
+      "step": 5591
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.4047703742980957,
+      "learning_rate": 9.902613557988185e-06,
+      "loss": 0.4303,
+      "step": 5592
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.052260398864746,
+      "learning_rate": 9.902561767194284e-06,
+      "loss": 0.6669,
+      "step": 5593
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.110483169555664,
+      "learning_rate": 9.902509962768194e-06,
+      "loss": 0.735,
+      "step": 5594
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.1449759006500244,
+      "learning_rate": 9.90245814471006e-06,
+      "loss": 0.6974,
+      "step": 5595
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 4.060351371765137,
+      "learning_rate": 9.902406313020028e-06,
+      "loss": 0.7933,
+      "step": 5596
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.060338258743286,
+      "learning_rate": 9.902354467698238e-06,
+      "loss": 0.6662,
+      "step": 5597
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.0951340198516846,
+      "learning_rate": 9.902302608744841e-06,
+      "loss": 0.8288,
+      "step": 5598
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 4.071040153503418,
+      "learning_rate": 9.902250736159973e-06,
+      "loss": 0.5738,
+      "step": 5599
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.054691791534424,
+      "learning_rate": 9.902198849943782e-06,
+      "loss": 0.5836,
+      "step": 5600
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.1700541973114014,
+      "learning_rate": 9.902146950096413e-06,
+      "loss": 0.9035,
+      "step": 5601
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.6268258094787598,
+      "learning_rate": 9.902095036618011e-06,
+      "loss": 0.5678,
+      "step": 5602
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.5759034156799316,
+      "learning_rate": 9.90204310950872e-06,
+      "loss": 0.6282,
+      "step": 5603
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.0372774600982666,
+      "learning_rate": 9.90199116876868e-06,
+      "loss": 0.7139,
+      "step": 5604
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.744577169418335,
+      "learning_rate": 9.901939214398041e-06,
+      "loss": 0.657,
+      "step": 5605
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 5.053682327270508,
+      "learning_rate": 9.901887246396946e-06,
+      "loss": 0.6421,
+      "step": 5606
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.732414722442627,
+      "learning_rate": 9.901835264765538e-06,
+      "loss": 0.6428,
+      "step": 5607
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.0227036476135254,
+      "learning_rate": 9.901783269503962e-06,
+      "loss": 0.6457,
+      "step": 5608
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.072253465652466,
+      "learning_rate": 9.901731260612363e-06,
+      "loss": 0.5866,
+      "step": 5609
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.9805423021316528,
+      "learning_rate": 9.901679238090886e-06,
+      "loss": 0.752,
+      "step": 5610
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 4.264787197113037,
+      "learning_rate": 9.901627201939676e-06,
+      "loss": 0.7218,
+      "step": 5611
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.9708251953125,
+      "learning_rate": 9.901575152158876e-06,
+      "loss": 0.5585,
+      "step": 5612
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.880293369293213,
+      "learning_rate": 9.901523088748632e-06,
+      "loss": 0.5707,
+      "step": 5613
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 4.653345108032227,
+      "learning_rate": 9.901471011709088e-06,
+      "loss": 0.6117,
+      "step": 5614
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.5992354154586792,
+      "learning_rate": 9.901418921040388e-06,
+      "loss": 0.6724,
+      "step": 5615
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.471317768096924,
+      "learning_rate": 9.901366816742682e-06,
+      "loss": 0.6008,
+      "step": 5616
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.054232358932495,
+      "learning_rate": 9.901314698816106e-06,
+      "loss": 0.3998,
+      "step": 5617
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.9162938594818115,
+      "learning_rate": 9.901262567260815e-06,
+      "loss": 0.5921,
+      "step": 5618
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.488992929458618,
+      "learning_rate": 9.901210422076946e-06,
+      "loss": 0.5073,
+      "step": 5619
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 4.911495685577393,
+      "learning_rate": 9.901158263264646e-06,
+      "loss": 0.6415,
+      "step": 5620
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.0316050052642822,
+      "learning_rate": 9.901106090824062e-06,
+      "loss": 0.5424,
+      "step": 5621
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 4.165033340454102,
+      "learning_rate": 9.901053904755338e-06,
+      "loss": 0.8112,
+      "step": 5622
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.216003894805908,
+      "learning_rate": 9.90100170505862e-06,
+      "loss": 0.8717,
+      "step": 5623
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.9071249961853027,
+      "learning_rate": 9.90094949173405e-06,
+      "loss": 0.5876,
+      "step": 5624
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.344538450241089,
+      "learning_rate": 9.900897264781776e-06,
+      "loss": 0.4038,
+      "step": 5625
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.220817804336548,
+      "learning_rate": 9.900845024201945e-06,
+      "loss": 0.7409,
+      "step": 5626
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.5286126136779785,
+      "learning_rate": 9.900792769994698e-06,
+      "loss": 0.4846,
+      "step": 5627
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.0642175674438477,
+      "learning_rate": 9.900740502160182e-06,
+      "loss": 0.5959,
+      "step": 5628
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.9860949516296387,
+      "learning_rate": 9.900688220698543e-06,
+      "loss": 0.5585,
+      "step": 5629
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.46372652053833,
+      "learning_rate": 9.900635925609926e-06,
+      "loss": 0.3837,
+      "step": 5630
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.8984785079956055,
+      "learning_rate": 9.900583616894477e-06,
+      "loss": 0.7439,
+      "step": 5631
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.132972240447998,
+      "learning_rate": 9.900531294552339e-06,
+      "loss": 0.4172,
+      "step": 5632
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.0545248985290527,
+      "learning_rate": 9.90047895858366e-06,
+      "loss": 0.6639,
+      "step": 5633
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.50276255607605,
+      "learning_rate": 9.900426608988585e-06,
+      "loss": 0.8149,
+      "step": 5634
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.7042551040649414,
+      "learning_rate": 9.90037424576726e-06,
+      "loss": 0.7081,
+      "step": 5635
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.667182683944702,
+      "learning_rate": 9.90032186891983e-06,
+      "loss": 0.3566,
+      "step": 5636
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.0425968170166016,
+      "learning_rate": 9.900269478446438e-06,
+      "loss": 0.6221,
+      "step": 5637
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.593531608581543,
+      "learning_rate": 9.900217074347233e-06,
+      "loss": 0.6971,
+      "step": 5638
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.3186068534851074,
+      "learning_rate": 9.900164656622362e-06,
+      "loss": 0.9519,
+      "step": 5639
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.379880905151367,
+      "learning_rate": 9.900112225271968e-06,
+      "loss": 0.6985,
+      "step": 5640
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.7353529930114746,
+      "learning_rate": 9.900059780296197e-06,
+      "loss": 0.5732,
+      "step": 5641
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.0289113521575928,
+      "learning_rate": 9.900007321695196e-06,
+      "loss": 0.6405,
+      "step": 5642
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.337996482849121,
+      "learning_rate": 9.89995484946911e-06,
+      "loss": 0.5332,
+      "step": 5643
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.3779239654541016,
+      "learning_rate": 9.899902363618085e-06,
+      "loss": 0.868,
+      "step": 5644
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.0645039081573486,
+      "learning_rate": 9.899849864142266e-06,
+      "loss": 0.8382,
+      "step": 5645
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 4.327507495880127,
+      "learning_rate": 9.899797351041802e-06,
+      "loss": 0.7107,
+      "step": 5646
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 4.333005905151367,
+      "learning_rate": 9.899744824316836e-06,
+      "loss": 0.6573,
+      "step": 5647
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 7.974786758422852,
+      "learning_rate": 9.899692283967515e-06,
+      "loss": 0.6406,
+      "step": 5648
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.2700889110565186,
+      "learning_rate": 9.899639729993986e-06,
+      "loss": 0.5892,
+      "step": 5649
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.2471091747283936,
+      "learning_rate": 9.899587162396394e-06,
+      "loss": 0.5098,
+      "step": 5650
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.280163288116455,
+      "learning_rate": 9.899534581174886e-06,
+      "loss": 0.6062,
+      "step": 5651
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.223684549331665,
+      "learning_rate": 9.899481986329606e-06,
+      "loss": 0.4702,
+      "step": 5652
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.7774057388305664,
+      "learning_rate": 9.899429377860703e-06,
+      "loss": 0.9061,
+      "step": 5653
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 4.448142051696777,
+      "learning_rate": 9.899376755768323e-06,
+      "loss": 0.6677,
+      "step": 5654
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 4.848208427429199,
+      "learning_rate": 9.899324120052611e-06,
+      "loss": 0.649,
+      "step": 5655
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 8.634325981140137,
+      "learning_rate": 9.899271470713714e-06,
+      "loss": 0.4763,
+      "step": 5656
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.5209012031555176,
+      "learning_rate": 9.899218807751778e-06,
+      "loss": 0.591,
+      "step": 5657
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.631528615951538,
+      "learning_rate": 9.89916613116695e-06,
+      "loss": 1.0285,
+      "step": 5658
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.5461313724517822,
+      "learning_rate": 9.899113440959378e-06,
+      "loss": 0.8022,
+      "step": 5659
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 8.895642280578613,
+      "learning_rate": 9.899060737129205e-06,
+      "loss": 0.2079,
+      "step": 5660
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.1554527282714844,
+      "learning_rate": 9.899008019676579e-06,
+      "loss": 0.7324,
+      "step": 5661
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.7772915363311768,
+      "learning_rate": 9.898955288601647e-06,
+      "loss": 0.7065,
+      "step": 5662
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.6776797771453857,
+      "learning_rate": 9.898902543904556e-06,
+      "loss": 0.7701,
+      "step": 5663
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.0653936862945557,
+      "learning_rate": 9.898849785585452e-06,
+      "loss": 0.6309,
+      "step": 5664
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.071284294128418,
+      "learning_rate": 9.898797013644482e-06,
+      "loss": 0.5416,
+      "step": 5665
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.326805591583252,
+      "learning_rate": 9.898744228081793e-06,
+      "loss": 0.5969,
+      "step": 5666
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 4.2542724609375,
+      "learning_rate": 9.898691428897531e-06,
+      "loss": 0.8043,
+      "step": 5667
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.0021700859069824,
+      "learning_rate": 9.898638616091843e-06,
+      "loss": 0.8835,
+      "step": 5668
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.0964772701263428,
+      "learning_rate": 9.898585789664876e-06,
+      "loss": 0.7283,
+      "step": 5669
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.1288225650787354,
+      "learning_rate": 9.898532949616777e-06,
+      "loss": 0.8623,
+      "step": 5670
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.335294246673584,
+      "learning_rate": 9.898480095947693e-06,
+      "loss": 0.6586,
+      "step": 5671
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 4.090083599090576,
+      "learning_rate": 9.898427228657773e-06,
+      "loss": 0.7844,
+      "step": 5672
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 4.451943397521973,
+      "learning_rate": 9.89837434774716e-06,
+      "loss": 0.5459,
+      "step": 5673
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.621966600418091,
+      "learning_rate": 9.898321453216002e-06,
+      "loss": 0.6654,
+      "step": 5674
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.9480575323104858,
+      "learning_rate": 9.898268545064447e-06,
+      "loss": 0.6972,
+      "step": 5675
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.8121678829193115,
+      "learning_rate": 9.898215623292644e-06,
+      "loss": 0.6975,
+      "step": 5676
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.324329137802124,
+      "learning_rate": 9.898162687900736e-06,
+      "loss": 0.6071,
+      "step": 5677
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.69538950920105,
+      "learning_rate": 9.898109738888875e-06,
+      "loss": 0.8386,
+      "step": 5678
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.1222057342529297,
+      "learning_rate": 9.898056776257202e-06,
+      "loss": 0.4457,
+      "step": 5679
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.3166229724884033,
+      "learning_rate": 9.898003800005872e-06,
+      "loss": 0.662,
+      "step": 5680
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.20500111579895,
+      "learning_rate": 9.897950810135026e-06,
+      "loss": 0.4089,
+      "step": 5681
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.4464333057403564,
+      "learning_rate": 9.897897806644814e-06,
+      "loss": 0.4778,
+      "step": 5682
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.0677103996276855,
+      "learning_rate": 9.897844789535384e-06,
+      "loss": 0.7488,
+      "step": 5683
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.5516719818115234,
+      "learning_rate": 9.897791758806882e-06,
+      "loss": 0.6532,
+      "step": 5684
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.9956930875778198,
+      "learning_rate": 9.897738714459456e-06,
+      "loss": 0.5836,
+      "step": 5685
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 4.132537364959717,
+      "learning_rate": 9.897685656493252e-06,
+      "loss": 0.907,
+      "step": 5686
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.7502492666244507,
+      "learning_rate": 9.89763258490842e-06,
+      "loss": 0.5577,
+      "step": 5687
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.4160265922546387,
+      "learning_rate": 9.897579499705106e-06,
+      "loss": 0.8081,
+      "step": 5688
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.5317630767822266,
+      "learning_rate": 9.897526400883458e-06,
+      "loss": 0.8372,
+      "step": 5689
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.5760903358459473,
+      "learning_rate": 9.897473288443624e-06,
+      "loss": 0.8223,
+      "step": 5690
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.5244600772857666,
+      "learning_rate": 9.897420162385751e-06,
+      "loss": 0.6662,
+      "step": 5691
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.4398093223571777,
+      "learning_rate": 9.89736702270999e-06,
+      "loss": 0.4478,
+      "step": 5692
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.924233913421631,
+      "learning_rate": 9.897313869416482e-06,
+      "loss": 0.5081,
+      "step": 5693
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.549790620803833,
+      "learning_rate": 9.89726070250538e-06,
+      "loss": 0.7871,
+      "step": 5694
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.6660610437393188,
+      "learning_rate": 9.89720752197683e-06,
+      "loss": 0.7648,
+      "step": 5695
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.1816234588623047,
+      "learning_rate": 9.897154327830982e-06,
+      "loss": 0.6489,
+      "step": 5696
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.858778715133667,
+      "learning_rate": 9.897101120067982e-06,
+      "loss": 0.5393,
+      "step": 5697
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.0171897411346436,
+      "learning_rate": 9.897047898687976e-06,
+      "loss": 0.6162,
+      "step": 5698
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.343590259552002,
+      "learning_rate": 9.896994663691118e-06,
+      "loss": 0.6912,
+      "step": 5699
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.095966100692749,
+      "learning_rate": 9.89694141507755e-06,
+      "loss": 0.7218,
+      "step": 5700
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.435171127319336,
+      "learning_rate": 9.896888152847423e-06,
+      "loss": 0.6964,
+      "step": 5701
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.994044303894043,
+      "learning_rate": 9.896834877000884e-06,
+      "loss": 0.802,
+      "step": 5702
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.0707650184631348,
+      "learning_rate": 9.896781587538083e-06,
+      "loss": 0.8069,
+      "step": 5703
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 5.167571544647217,
+      "learning_rate": 9.896728284459166e-06,
+      "loss": 0.2431,
+      "step": 5704
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.973063349723816,
+      "learning_rate": 9.896674967764279e-06,
+      "loss": 0.7496,
+      "step": 5705
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.0189456939697266,
+      "learning_rate": 9.896621637453576e-06,
+      "loss": 0.6728,
+      "step": 5706
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.550327777862549,
+      "learning_rate": 9.896568293527203e-06,
+      "loss": 0.8447,
+      "step": 5707
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.309067964553833,
+      "learning_rate": 9.896514935985307e-06,
+      "loss": 0.5533,
+      "step": 5708
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.372873067855835,
+      "learning_rate": 9.896461564828038e-06,
+      "loss": 0.6566,
+      "step": 5709
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.9239044189453125,
+      "learning_rate": 9.896408180055543e-06,
+      "loss": 0.5759,
+      "step": 5710
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.437976121902466,
+      "learning_rate": 9.896354781667972e-06,
+      "loss": 0.6941,
+      "step": 5711
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.7505759000778198,
+      "learning_rate": 9.896301369665471e-06,
+      "loss": 0.6787,
+      "step": 5712
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.8364412784576416,
+      "learning_rate": 9.896247944048192e-06,
+      "loss": 0.9567,
+      "step": 5713
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.3928661346435547,
+      "learning_rate": 9.896194504816279e-06,
+      "loss": 0.6645,
+      "step": 5714
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.3032498359680176,
+      "learning_rate": 9.896141051969885e-06,
+      "loss": 0.9394,
+      "step": 5715
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.461472511291504,
+      "learning_rate": 9.896087585509158e-06,
+      "loss": 0.7988,
+      "step": 5716
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.9562864303588867,
+      "learning_rate": 9.896034105434243e-06,
+      "loss": 0.7814,
+      "step": 5717
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.7128477096557617,
+      "learning_rate": 9.895980611745292e-06,
+      "loss": 0.6317,
+      "step": 5718
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 6.597471714019775,
+      "learning_rate": 9.895927104442452e-06,
+      "loss": 0.5028,
+      "step": 5719
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 5.403994083404541,
+      "learning_rate": 9.895873583525877e-06,
+      "loss": 0.6589,
+      "step": 5720
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.994823694229126,
+      "learning_rate": 9.895820048995708e-06,
+      "loss": 0.5894,
+      "step": 5721
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.9926351308822632,
+      "learning_rate": 9.895766500852098e-06,
+      "loss": 0.7215,
+      "step": 5722
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.6208319664001465,
+      "learning_rate": 9.895712939095195e-06,
+      "loss": 0.87,
+      "step": 5723
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.810133218765259,
+      "learning_rate": 9.89565936372515e-06,
+      "loss": 0.6334,
+      "step": 5724
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 4.3894944190979,
+      "learning_rate": 9.895605774742108e-06,
+      "loss": 0.8633,
+      "step": 5725
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 4.269012451171875,
+      "learning_rate": 9.895552172146222e-06,
+      "loss": 1.132,
+      "step": 5726
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.243083953857422,
+      "learning_rate": 9.895498555937639e-06,
+      "loss": 0.8399,
+      "step": 5727
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.2762606143951416,
+      "learning_rate": 9.89544492611651e-06,
+      "loss": 0.6811,
+      "step": 5728
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.0707952976226807,
+      "learning_rate": 9.89539128268298e-06,
+      "loss": 0.8516,
+      "step": 5729
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.607327938079834,
+      "learning_rate": 9.8953376256372e-06,
+      "loss": 0.737,
+      "step": 5730
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.6530447006225586,
+      "learning_rate": 9.895283954979323e-06,
+      "loss": 0.4304,
+      "step": 5731
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.2019670009613037,
+      "learning_rate": 9.895230270709492e-06,
+      "loss": 0.5611,
+      "step": 5732
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.9353106021881104,
+      "learning_rate": 9.895176572827861e-06,
+      "loss": 0.4963,
+      "step": 5733
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.027273416519165,
+      "learning_rate": 9.895122861334578e-06,
+      "loss": 0.5946,
+      "step": 5734
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.775656223297119,
+      "learning_rate": 9.895069136229791e-06,
+      "loss": 0.7324,
+      "step": 5735
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.6381311416625977,
+      "learning_rate": 9.895015397513652e-06,
+      "loss": 0.4196,
+      "step": 5736
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.1842756271362305,
+      "learning_rate": 9.894961645186308e-06,
+      "loss": 0.6532,
+      "step": 5737
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.7206883430480957,
+      "learning_rate": 9.894907879247907e-06,
+      "loss": 0.5845,
+      "step": 5738
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.3864269256591797,
+      "learning_rate": 9.894854099698605e-06,
+      "loss": 0.7704,
+      "step": 5739
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 4.21544075012207,
+      "learning_rate": 9.894800306538546e-06,
+      "loss": 0.6676,
+      "step": 5740
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.30391263961792,
+      "learning_rate": 9.89474649976788e-06,
+      "loss": 0.5489,
+      "step": 5741
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.645632743835449,
+      "learning_rate": 9.89469267938676e-06,
+      "loss": 0.8965,
+      "step": 5742
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.8866909742355347,
+      "learning_rate": 9.89463884539533e-06,
+      "loss": 0.7364,
+      "step": 5743
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.413134813308716,
+      "learning_rate": 9.894584997793744e-06,
+      "loss": 0.657,
+      "step": 5744
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.8344613313674927,
+      "learning_rate": 9.894531136582152e-06,
+      "loss": 0.6816,
+      "step": 5745
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.2988827228546143,
+      "learning_rate": 9.894477261760701e-06,
+      "loss": 0.711,
+      "step": 5746
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 4.464077949523926,
+      "learning_rate": 9.894423373329542e-06,
+      "loss": 0.6349,
+      "step": 5747
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.877345561981201,
+      "learning_rate": 9.894369471288827e-06,
+      "loss": 0.5232,
+      "step": 5748
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.16689395904541,
+      "learning_rate": 9.894315555638703e-06,
+      "loss": 1.0011,
+      "step": 5749
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.647481918334961,
+      "learning_rate": 9.89426162637932e-06,
+      "loss": 0.7427,
+      "step": 5750
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.931361436843872,
+      "learning_rate": 9.89420768351083e-06,
+      "loss": 0.8916,
+      "step": 5751
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.9115426540374756,
+      "learning_rate": 9.894153727033382e-06,
+      "loss": 0.5454,
+      "step": 5752
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.6084437370300293,
+      "learning_rate": 9.894099756947124e-06,
+      "loss": 0.7668,
+      "step": 5753
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.2728207111358643,
+      "learning_rate": 9.89404577325221e-06,
+      "loss": 0.4526,
+      "step": 5754
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.5894854068756104,
+      "learning_rate": 9.893991775948788e-06,
+      "loss": 0.3978,
+      "step": 5755
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.4764223098754883,
+      "learning_rate": 9.893937765037007e-06,
+      "loss": 0.816,
+      "step": 5756
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 4.181276798248291,
+      "learning_rate": 9.893883740517018e-06,
+      "loss": 0.4858,
+      "step": 5757
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.573711633682251,
+      "learning_rate": 9.893829702388972e-06,
+      "loss": 0.8115,
+      "step": 5758
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.150517225265503,
+      "learning_rate": 9.893775650653022e-06,
+      "loss": 0.8016,
+      "step": 5759
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.6075289249420166,
+      "learning_rate": 9.893721585309312e-06,
+      "loss": 0.6972,
+      "step": 5760
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.3493099212646484,
+      "learning_rate": 9.893667506357996e-06,
+      "loss": 0.7291,
+      "step": 5761
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.568305015563965,
+      "learning_rate": 9.893613413799224e-06,
+      "loss": 0.5676,
+      "step": 5762
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 4.9472455978393555,
+      "learning_rate": 9.893559307633145e-06,
+      "loss": 0.7233,
+      "step": 5763
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.3323659896850586,
+      "learning_rate": 9.893505187859913e-06,
+      "loss": 0.6745,
+      "step": 5764
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.5237529277801514,
+      "learning_rate": 9.893451054479676e-06,
+      "loss": 0.6559,
+      "step": 5765
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 4.468761920928955,
+      "learning_rate": 9.893396907492583e-06,
+      "loss": 0.973,
+      "step": 5766
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.097193717956543,
+      "learning_rate": 9.893342746898786e-06,
+      "loss": 0.7677,
+      "step": 5767
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 5.5652666091918945,
+      "learning_rate": 9.893288572698438e-06,
+      "loss": 0.758,
+      "step": 5768
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.7221264839172363,
+      "learning_rate": 9.893234384891687e-06,
+      "loss": 0.6657,
+      "step": 5769
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 4.617040634155273,
+      "learning_rate": 9.893180183478684e-06,
+      "loss": 0.6905,
+      "step": 5770
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 4.515186309814453,
+      "learning_rate": 9.89312596845958e-06,
+      "loss": 0.8015,
+      "step": 5771
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.0877983570098877,
+      "learning_rate": 9.893071739834525e-06,
+      "loss": 0.8046,
+      "step": 5772
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.952972412109375,
+      "learning_rate": 9.89301749760367e-06,
+      "loss": 0.6258,
+      "step": 5773
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.9957472085952759,
+      "learning_rate": 9.892963241767168e-06,
+      "loss": 0.8507,
+      "step": 5774
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.227186441421509,
+      "learning_rate": 9.892908972325168e-06,
+      "loss": 0.7216,
+      "step": 5775
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.3771347999572754,
+      "learning_rate": 9.89285468927782e-06,
+      "loss": 0.7432,
+      "step": 5776
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.4796504974365234,
+      "learning_rate": 9.892800392625276e-06,
+      "loss": 0.7796,
+      "step": 5777
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 5.291471004486084,
+      "learning_rate": 9.892746082367688e-06,
+      "loss": 0.42,
+      "step": 5778
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.288733959197998,
+      "learning_rate": 9.892691758505207e-06,
+      "loss": 0.5472,
+      "step": 5779
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.7458713054656982,
+      "learning_rate": 9.892637421037981e-06,
+      "loss": 0.628,
+      "step": 5780
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.031679391860962,
+      "learning_rate": 9.892583069966164e-06,
+      "loss": 0.7697,
+      "step": 5781
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.9806342124938965,
+      "learning_rate": 9.892528705289905e-06,
+      "loss": 0.8002,
+      "step": 5782
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.01236891746521,
+      "learning_rate": 9.892474327009357e-06,
+      "loss": 0.4842,
+      "step": 5783
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.873644232749939,
+      "learning_rate": 9.892419935124673e-06,
+      "loss": 0.5275,
+      "step": 5784
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.4703269004821777,
+      "learning_rate": 9.892365529635998e-06,
+      "loss": 0.552,
+      "step": 5785
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 4.560686111450195,
+      "learning_rate": 9.89231111054349e-06,
+      "loss": 1.1471,
+      "step": 5786
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.7106003761291504,
+      "learning_rate": 9.892256677847298e-06,
+      "loss": 0.6947,
+      "step": 5787
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.206026792526245,
+      "learning_rate": 9.892202231547571e-06,
+      "loss": 0.4662,
+      "step": 5788
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.859790325164795,
+      "learning_rate": 9.892147771644462e-06,
+      "loss": 0.8658,
+      "step": 5789
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.6561849117279053,
+      "learning_rate": 9.892093298138125e-06,
+      "loss": 0.6522,
+      "step": 5790
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.249877691268921,
+      "learning_rate": 9.892038811028707e-06,
+      "loss": 0.579,
+      "step": 5791
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.3032314777374268,
+      "learning_rate": 9.891984310316362e-06,
+      "loss": 0.861,
+      "step": 5792
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.470933675765991,
+      "learning_rate": 9.891929796001244e-06,
+      "loss": 0.6764,
+      "step": 5793
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 4.364970684051514,
+      "learning_rate": 9.891875268083498e-06,
+      "loss": 0.8166,
+      "step": 5794
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.209364175796509,
+      "learning_rate": 9.891820726563282e-06,
+      "loss": 0.6928,
+      "step": 5795
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 6.203551292419434,
+      "learning_rate": 9.891766171440742e-06,
+      "loss": 0.7802,
+      "step": 5796
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.560518264770508,
+      "learning_rate": 9.891711602716035e-06,
+      "loss": 0.8728,
+      "step": 5797
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.601914405822754,
+      "learning_rate": 9.89165702038931e-06,
+      "loss": 0.8757,
+      "step": 5798
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.423628330230713,
+      "learning_rate": 9.89160242446072e-06,
+      "loss": 0.541,
+      "step": 5799
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.871353268623352,
+      "learning_rate": 9.891547814930416e-06,
+      "loss": 0.5643,
+      "step": 5800
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.224196672439575,
+      "learning_rate": 9.891493191798548e-06,
+      "loss": 0.5488,
+      "step": 5801
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 4.279468059539795,
+      "learning_rate": 9.891438555065273e-06,
+      "loss": 0.9854,
+      "step": 5802
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.4426255226135254,
+      "learning_rate": 9.891383904730738e-06,
+      "loss": 0.909,
+      "step": 5803
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.7255048751831055,
+      "learning_rate": 9.891329240795096e-06,
+      "loss": 0.7843,
+      "step": 5804
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.231184959411621,
+      "learning_rate": 9.891274563258502e-06,
+      "loss": 0.6744,
+      "step": 5805
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.5047690868377686,
+      "learning_rate": 9.891219872121102e-06,
+      "loss": 0.7093,
+      "step": 5806
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.5767393112182617,
+      "learning_rate": 9.891165167383055e-06,
+      "loss": 0.6457,
+      "step": 5807
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.28110671043396,
+      "learning_rate": 9.89111044904451e-06,
+      "loss": 0.6435,
+      "step": 5808
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.2924556732177734,
+      "learning_rate": 9.891055717105618e-06,
+      "loss": 0.8531,
+      "step": 5809
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.6808340549468994,
+      "learning_rate": 9.891000971566532e-06,
+      "loss": 0.7496,
+      "step": 5810
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.7494049072265625,
+      "learning_rate": 9.890946212427406e-06,
+      "loss": 1.0673,
+      "step": 5811
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.490849733352661,
+      "learning_rate": 9.890891439688388e-06,
+      "loss": 0.7216,
+      "step": 5812
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.3393394947052,
+      "learning_rate": 9.890836653349637e-06,
+      "loss": 0.6245,
+      "step": 5813
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.084430456161499,
+      "learning_rate": 9.890781853411298e-06,
+      "loss": 0.7457,
+      "step": 5814
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.3684518337249756,
+      "learning_rate": 9.890727039873528e-06,
+      "loss": 0.4564,
+      "step": 5815
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.781179428100586,
+      "learning_rate": 9.890672212736479e-06,
+      "loss": 0.7515,
+      "step": 5816
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.8723647594451904,
+      "learning_rate": 9.8906173720003e-06,
+      "loss": 0.602,
+      "step": 5817
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.2037289142608643,
+      "learning_rate": 9.890562517665149e-06,
+      "loss": 0.6059,
+      "step": 5818
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.5231597423553467,
+      "learning_rate": 9.890507649731175e-06,
+      "loss": 0.5821,
+      "step": 5819
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 4.0534515380859375,
+      "learning_rate": 9.89045276819853e-06,
+      "loss": 0.9974,
+      "step": 5820
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 4.745471477508545,
+      "learning_rate": 9.89039787306737e-06,
+      "loss": 0.9109,
+      "step": 5821
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.40822172164917,
+      "learning_rate": 9.890342964337845e-06,
+      "loss": 0.6136,
+      "step": 5822
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.0757429599761963,
+      "learning_rate": 9.890288042010105e-06,
+      "loss": 0.7798,
+      "step": 5823
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.5449204444885254,
+      "learning_rate": 9.89023310608431e-06,
+      "loss": 0.6751,
+      "step": 5824
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.7557904720306396,
+      "learning_rate": 9.890178156560607e-06,
+      "loss": 0.984,
+      "step": 5825
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.2415647506713867,
+      "learning_rate": 9.89012319343915e-06,
+      "loss": 0.6733,
+      "step": 5826
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.8476989269256592,
+      "learning_rate": 9.890068216720093e-06,
+      "loss": 0.9996,
+      "step": 5827
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.2303996086120605,
+      "learning_rate": 9.890013226403588e-06,
+      "loss": 0.4748,
+      "step": 5828
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.0059804916381836,
+      "learning_rate": 9.889958222489789e-06,
+      "loss": 0.4222,
+      "step": 5829
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.8973820209503174,
+      "learning_rate": 9.889903204978846e-06,
+      "loss": 0.7183,
+      "step": 5830
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.8253672122955322,
+      "learning_rate": 9.889848173870915e-06,
+      "loss": 0.7425,
+      "step": 5831
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.455864191055298,
+      "learning_rate": 9.889793129166147e-06,
+      "loss": 0.7849,
+      "step": 5832
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.8249244689941406,
+      "learning_rate": 9.889738070864698e-06,
+      "loss": 0.8703,
+      "step": 5833
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.3328473567962646,
+      "learning_rate": 9.88968299896672e-06,
+      "loss": 0.856,
+      "step": 5834
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.788722515106201,
+      "learning_rate": 9.889627913472362e-06,
+      "loss": 0.79,
+      "step": 5835
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.862301826477051,
+      "learning_rate": 9.889572814381781e-06,
+      "loss": 0.6425,
+      "step": 5836
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.178025960922241,
+      "learning_rate": 9.889517701695132e-06,
+      "loss": 0.587,
+      "step": 5837
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.475822925567627,
+      "learning_rate": 9.889462575412565e-06,
+      "loss": 0.3935,
+      "step": 5838
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.774669647216797,
+      "learning_rate": 9.889407435534233e-06,
+      "loss": 0.6874,
+      "step": 5839
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 13.186452865600586,
+      "learning_rate": 9.889352282060292e-06,
+      "loss": 0.4774,
+      "step": 5840
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.8560025691986084,
+      "learning_rate": 9.889297114990893e-06,
+      "loss": 0.7011,
+      "step": 5841
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.4506897926330566,
+      "learning_rate": 9.88924193432619e-06,
+      "loss": 0.5982,
+      "step": 5842
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 4.080769062042236,
+      "learning_rate": 9.889186740066337e-06,
+      "loss": 0.7282,
+      "step": 5843
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.806323289871216,
+      "learning_rate": 9.889131532211488e-06,
+      "loss": 0.7377,
+      "step": 5844
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.490158796310425,
+      "learning_rate": 9.889076310761795e-06,
+      "loss": 0.9397,
+      "step": 5845
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.044772148132324,
+      "learning_rate": 9.889021075717414e-06,
+      "loss": 0.7369,
+      "step": 5846
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.3831746578216553,
+      "learning_rate": 9.888965827078493e-06,
+      "loss": 0.5519,
+      "step": 5847
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.8714749813079834,
+      "learning_rate": 9.888910564845193e-06,
+      "loss": 0.79,
+      "step": 5848
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 4.439332962036133,
+      "learning_rate": 9.888855289017663e-06,
+      "loss": 0.5673,
+      "step": 5849
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.2942283153533936,
+      "learning_rate": 9.888799999596058e-06,
+      "loss": 0.7026,
+      "step": 5850
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.6556684970855713,
+      "learning_rate": 9.88874469658053e-06,
+      "loss": 0.7631,
+      "step": 5851
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.2587099075317383,
+      "learning_rate": 9.888689379971237e-06,
+      "loss": 0.6568,
+      "step": 5852
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.9412877559661865,
+      "learning_rate": 9.888634049768329e-06,
+      "loss": 0.4415,
+      "step": 5853
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.233982801437378,
+      "learning_rate": 9.88857870597196e-06,
+      "loss": 0.7599,
+      "step": 5854
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.7967654466629028,
+      "learning_rate": 9.888523348582286e-06,
+      "loss": 0.6565,
+      "step": 5855
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.0983753204345703,
+      "learning_rate": 9.888467977599461e-06,
+      "loss": 0.5587,
+      "step": 5856
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.437866449356079,
+      "learning_rate": 9.888412593023636e-06,
+      "loss": 0.6008,
+      "step": 5857
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 7.709596633911133,
+      "learning_rate": 9.888357194854967e-06,
+      "loss": 0.6919,
+      "step": 5858
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.3957314491271973,
+      "learning_rate": 9.888301783093607e-06,
+      "loss": 1.008,
+      "step": 5859
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.160332441329956,
+      "learning_rate": 9.888246357739714e-06,
+      "loss": 0.6835,
+      "step": 5860
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.155271053314209,
+      "learning_rate": 9.888190918793437e-06,
+      "loss": 0.5669,
+      "step": 5861
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.704835057258606,
+      "learning_rate": 9.888135466254932e-06,
+      "loss": 0.7039,
+      "step": 5862
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.4074020385742188,
+      "learning_rate": 9.888080000124354e-06,
+      "loss": 0.854,
+      "step": 5863
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.3534181118011475,
+      "learning_rate": 9.888024520401857e-06,
+      "loss": 0.8487,
+      "step": 5864
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.855041980743408,
+      "learning_rate": 9.887969027087595e-06,
+      "loss": 0.8014,
+      "step": 5865
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.696758270263672,
+      "learning_rate": 9.88791352018172e-06,
+      "loss": 0.6041,
+      "step": 5866
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.8155555725097656,
+      "learning_rate": 9.887857999684391e-06,
+      "loss": 0.8482,
+      "step": 5867
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.9209821224212646,
+      "learning_rate": 9.88780246559576e-06,
+      "loss": 0.8362,
+      "step": 5868
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.1784379482269287,
+      "learning_rate": 9.887746917915979e-06,
+      "loss": 0.4885,
+      "step": 5869
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.213289260864258,
+      "learning_rate": 9.887691356645207e-06,
+      "loss": 0.6809,
+      "step": 5870
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.6773500442504883,
+      "learning_rate": 9.887635781783596e-06,
+      "loss": 0.6052,
+      "step": 5871
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.0377297401428223,
+      "learning_rate": 9.887580193331298e-06,
+      "loss": 0.8798,
+      "step": 5872
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.713418960571289,
+      "learning_rate": 9.887524591288475e-06,
+      "loss": 0.8217,
+      "step": 5873
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.8435423374176025,
+      "learning_rate": 9.887468975655274e-06,
+      "loss": 0.8369,
+      "step": 5874
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.8635740280151367,
+      "learning_rate": 9.887413346431854e-06,
+      "loss": 0.6994,
+      "step": 5875
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.108349561691284,
+      "learning_rate": 9.887357703618367e-06,
+      "loss": 0.8423,
+      "step": 5876
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.689546585083008,
+      "learning_rate": 9.88730204721497e-06,
+      "loss": 0.7333,
+      "step": 5877
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.3846800327301025,
+      "learning_rate": 9.887246377221817e-06,
+      "loss": 0.7972,
+      "step": 5878
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 4.484243869781494,
+      "learning_rate": 9.887190693639063e-06,
+      "loss": 0.7973,
+      "step": 5879
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 16.499584197998047,
+      "learning_rate": 9.887134996466861e-06,
+      "loss": 1.0613,
+      "step": 5880
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.5310020446777344,
+      "learning_rate": 9.887079285705368e-06,
+      "loss": 0.6388,
+      "step": 5881
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.7836410999298096,
+      "learning_rate": 9.88702356135474e-06,
+      "loss": 0.8012,
+      "step": 5882
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.1494877338409424,
+      "learning_rate": 9.886967823415128e-06,
+      "loss": 0.7713,
+      "step": 5883
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.8607420921325684,
+      "learning_rate": 9.88691207188669e-06,
+      "loss": 0.7726,
+      "step": 5884
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 4.659762859344482,
+      "learning_rate": 9.88685630676958e-06,
+      "loss": 0.7136,
+      "step": 5885
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.899570345878601,
+      "learning_rate": 9.886800528063953e-06,
+      "loss": 0.8251,
+      "step": 5886
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 4.685173988342285,
+      "learning_rate": 9.886744735769964e-06,
+      "loss": 0.7471,
+      "step": 5887
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.10506010055542,
+      "learning_rate": 9.88668892988777e-06,
+      "loss": 0.5898,
+      "step": 5888
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.6061742305755615,
+      "learning_rate": 9.886633110417524e-06,
+      "loss": 0.8423,
+      "step": 5889
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.6596287488937378,
+      "learning_rate": 9.886577277359382e-06,
+      "loss": 0.4938,
+      "step": 5890
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.748676300048828,
+      "learning_rate": 9.886521430713498e-06,
+      "loss": 0.526,
+      "step": 5891
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 5.318597793579102,
+      "learning_rate": 9.886465570480031e-06,
+      "loss": 0.5711,
+      "step": 5892
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.4317328929901123,
+      "learning_rate": 9.886409696659131e-06,
+      "loss": 0.7743,
+      "step": 5893
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.939866304397583,
+      "learning_rate": 9.886353809250957e-06,
+      "loss": 0.7247,
+      "step": 5894
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 5.059201240539551,
+      "learning_rate": 9.886297908255664e-06,
+      "loss": 0.7473,
+      "step": 5895
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.7422027587890625,
+      "learning_rate": 9.886241993673406e-06,
+      "loss": 0.5419,
+      "step": 5896
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.832843780517578,
+      "learning_rate": 9.886186065504341e-06,
+      "loss": 0.6858,
+      "step": 5897
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 4.025873184204102,
+      "learning_rate": 9.886130123748622e-06,
+      "loss": 0.7684,
+      "step": 5898
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.665663242340088,
+      "learning_rate": 9.886074168406405e-06,
+      "loss": 0.4075,
+      "step": 5899
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.9584455490112305,
+      "learning_rate": 9.886018199477847e-06,
+      "loss": 0.6798,
+      "step": 5900
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 4.254816055297852,
+      "learning_rate": 9.885962216963103e-06,
+      "loss": 0.6389,
+      "step": 5901
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 6.1664299964904785,
+      "learning_rate": 9.885906220862328e-06,
+      "loss": 0.7432,
+      "step": 5902
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 5.8676838874816895,
+      "learning_rate": 9.885850211175678e-06,
+      "loss": 0.7754,
+      "step": 5903
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.806553840637207,
+      "learning_rate": 9.885794187903309e-06,
+      "loss": 0.8123,
+      "step": 5904
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.9109318256378174,
+      "learning_rate": 9.885738151045376e-06,
+      "loss": 0.9239,
+      "step": 5905
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.1444826126098633,
+      "learning_rate": 9.885682100602035e-06,
+      "loss": 0.6491,
+      "step": 5906
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 4.46528959274292,
+      "learning_rate": 9.885626036573443e-06,
+      "loss": 0.7925,
+      "step": 5907
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.5507302284240723,
+      "learning_rate": 9.885569958959756e-06,
+      "loss": 0.703,
+      "step": 5908
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.1128532886505127,
+      "learning_rate": 9.885513867761129e-06,
+      "loss": 0.7173,
+      "step": 5909
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.3916399478912354,
+      "learning_rate": 9.885457762977717e-06,
+      "loss": 0.7412,
+      "step": 5910
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.5074243545532227,
+      "learning_rate": 9.885401644609677e-06,
+      "loss": 0.5435,
+      "step": 5911
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.517775058746338,
+      "learning_rate": 9.885345512657167e-06,
+      "loss": 0.6997,
+      "step": 5912
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.9650800228118896,
+      "learning_rate": 9.88528936712034e-06,
+      "loss": 0.7445,
+      "step": 5913
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.5632059574127197,
+      "learning_rate": 9.885233207999354e-06,
+      "loss": 0.8291,
+      "step": 5914
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.0585150718688965,
+      "learning_rate": 9.885177035294364e-06,
+      "loss": 0.2736,
+      "step": 5915
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.5787246227264404,
+      "learning_rate": 9.885120849005525e-06,
+      "loss": 0.6123,
+      "step": 5916
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.1819562911987305,
+      "learning_rate": 9.885064649132997e-06,
+      "loss": 0.5913,
+      "step": 5917
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.6628801822662354,
+      "learning_rate": 9.885008435676934e-06,
+      "loss": 0.4502,
+      "step": 5918
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.6656923294067383,
+      "learning_rate": 9.884952208637493e-06,
+      "loss": 0.5134,
+      "step": 5919
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 4.875051021575928,
+      "learning_rate": 9.884895968014828e-06,
+      "loss": 0.8812,
+      "step": 5920
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.4153621196746826,
+      "learning_rate": 9.884839713809099e-06,
+      "loss": 0.4946,
+      "step": 5921
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.8333683013916016,
+      "learning_rate": 9.884783446020458e-06,
+      "loss": 0.9395,
+      "step": 5922
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 6.488482475280762,
+      "learning_rate": 9.884727164649067e-06,
+      "loss": 0.7906,
+      "step": 5923
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.5683443546295166,
+      "learning_rate": 9.884670869695078e-06,
+      "loss": 0.6753,
+      "step": 5924
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 11.966947555541992,
+      "learning_rate": 9.88461456115865e-06,
+      "loss": 0.6067,
+      "step": 5925
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.8655219078063965,
+      "learning_rate": 9.884558239039938e-06,
+      "loss": 0.7791,
+      "step": 5926
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.663022041320801,
+      "learning_rate": 9.884501903339101e-06,
+      "loss": 0.664,
+      "step": 5927
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.727905035018921,
+      "learning_rate": 9.884445554056292e-06,
+      "loss": 0.5038,
+      "step": 5928
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.25201416015625,
+      "learning_rate": 9.884389191191672e-06,
+      "loss": 0.5013,
+      "step": 5929
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.682039260864258,
+      "learning_rate": 9.884332814745395e-06,
+      "loss": 0.7937,
+      "step": 5930
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.027737855911255,
+      "learning_rate": 9.884276424717616e-06,
+      "loss": 0.6385,
+      "step": 5931
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.8985092639923096,
+      "learning_rate": 9.884220021108495e-06,
+      "loss": 0.5557,
+      "step": 5932
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 4.680704116821289,
+      "learning_rate": 9.884163603918187e-06,
+      "loss": 0.83,
+      "step": 5933
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.273074150085449,
+      "learning_rate": 9.884107173146851e-06,
+      "loss": 0.9395,
+      "step": 5934
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.5003483295440674,
+      "learning_rate": 9.884050728794641e-06,
+      "loss": 0.6976,
+      "step": 5935
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.4733184576034546,
+      "learning_rate": 9.883994270861718e-06,
+      "loss": 0.7462,
+      "step": 5936
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 5.815114974975586,
+      "learning_rate": 9.883937799348235e-06,
+      "loss": 0.6467,
+      "step": 5937
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.888779401779175,
+      "learning_rate": 9.883881314254351e-06,
+      "loss": 0.6469,
+      "step": 5938
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.44940185546875,
+      "learning_rate": 9.883824815580223e-06,
+      "loss": 0.6582,
+      "step": 5939
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.7084473371505737,
+      "learning_rate": 9.883768303326006e-06,
+      "loss": 0.6325,
+      "step": 5940
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.078176975250244,
+      "learning_rate": 9.88371177749186e-06,
+      "loss": 0.5794,
+      "step": 5941
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 4.3804168701171875,
+      "learning_rate": 9.883655238077941e-06,
+      "loss": 0.6934,
+      "step": 5942
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.156381368637085,
+      "learning_rate": 9.883598685084406e-06,
+      "loss": 0.5328,
+      "step": 5943
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.0982322692871094,
+      "learning_rate": 9.883542118511411e-06,
+      "loss": 0.7185,
+      "step": 5944
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.2506906986236572,
+      "learning_rate": 9.883485538359118e-06,
+      "loss": 0.6172,
+      "step": 5945
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.8816235065460205,
+      "learning_rate": 9.88342894462768e-06,
+      "loss": 0.5167,
+      "step": 5946
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.7267563343048096,
+      "learning_rate": 9.883372337317254e-06,
+      "loss": 0.6022,
+      "step": 5947
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.9930875301361084,
+      "learning_rate": 9.883315716427999e-06,
+      "loss": 0.8439,
+      "step": 5948
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.7666372060775757,
+      "learning_rate": 9.883259081960075e-06,
+      "loss": 0.5395,
+      "step": 5949
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.8711211681365967,
+      "learning_rate": 9.883202433913635e-06,
+      "loss": 0.7531,
+      "step": 5950
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.8121347427368164,
+      "learning_rate": 9.883145772288837e-06,
+      "loss": 0.5452,
+      "step": 5951
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 33.59147262573242,
+      "learning_rate": 9.883089097085843e-06,
+      "loss": 0.6528,
+      "step": 5952
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.7732219696044922,
+      "learning_rate": 9.883032408304805e-06,
+      "loss": 0.4971,
+      "step": 5953
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 6.716513156890869,
+      "learning_rate": 9.882975705945883e-06,
+      "loss": 0.6146,
+      "step": 5954
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.6855695247650146,
+      "learning_rate": 9.882918990009236e-06,
+      "loss": 0.574,
+      "step": 5955
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 4.10124397277832,
+      "learning_rate": 9.88286226049502e-06,
+      "loss": 0.6232,
+      "step": 5956
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.8444766998291016,
+      "learning_rate": 9.882805517403393e-06,
+      "loss": 0.4307,
+      "step": 5957
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 21.989425659179688,
+      "learning_rate": 9.882748760734512e-06,
+      "loss": 0.6894,
+      "step": 5958
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.1699914932250977,
+      "learning_rate": 9.882691990488537e-06,
+      "loss": 0.5897,
+      "step": 5959
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.0632407665252686,
+      "learning_rate": 9.882635206665626e-06,
+      "loss": 0.9043,
+      "step": 5960
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.720484495162964,
+      "learning_rate": 9.882578409265934e-06,
+      "loss": 0.5419,
+      "step": 5961
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.3918888568878174,
+      "learning_rate": 9.882521598289621e-06,
+      "loss": 0.4401,
+      "step": 5962
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.437554359436035,
+      "learning_rate": 9.882464773736844e-06,
+      "loss": 0.4005,
+      "step": 5963
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.615347146987915,
+      "learning_rate": 9.882407935607762e-06,
+      "loss": 0.7814,
+      "step": 5964
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.214789628982544,
+      "learning_rate": 9.882351083902533e-06,
+      "loss": 0.6737,
+      "step": 5965
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 5.444109916687012,
+      "learning_rate": 9.882294218621313e-06,
+      "loss": 0.6998,
+      "step": 5966
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.6786413192749023,
+      "learning_rate": 9.882237339764263e-06,
+      "loss": 0.4984,
+      "step": 5967
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.6398215293884277,
+      "learning_rate": 9.882180447331539e-06,
+      "loss": 0.7156,
+      "step": 5968
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.3680951595306396,
+      "learning_rate": 9.882123541323301e-06,
+      "loss": 0.5854,
+      "step": 5969
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.325399398803711,
+      "learning_rate": 9.882066621739705e-06,
+      "loss": 0.7126,
+      "step": 5970
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.6943728923797607,
+      "learning_rate": 9.882009688580912e-06,
+      "loss": 0.8107,
+      "step": 5971
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.419508695602417,
+      "learning_rate": 9.881952741847078e-06,
+      "loss": 0.6394,
+      "step": 5972
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.0220484733581543,
+      "learning_rate": 9.881895781538361e-06,
+      "loss": 0.5236,
+      "step": 5973
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.9937191009521484,
+      "learning_rate": 9.881838807654924e-06,
+      "loss": 0.5402,
+      "step": 5974
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 4.588347434997559,
+      "learning_rate": 9.881781820196921e-06,
+      "loss": 0.8449,
+      "step": 5975
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.956793785095215,
+      "learning_rate": 9.88172481916451e-06,
+      "loss": 0.8937,
+      "step": 5976
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.4888904094696045,
+      "learning_rate": 9.881667804557852e-06,
+      "loss": 0.7239,
+      "step": 5977
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 6.113325119018555,
+      "learning_rate": 9.881610776377104e-06,
+      "loss": 0.6297,
+      "step": 5978
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.3906989097595215,
+      "learning_rate": 9.881553734622426e-06,
+      "loss": 0.7752,
+      "step": 5979
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 4.538681507110596,
+      "learning_rate": 9.881496679293975e-06,
+      "loss": 0.8554,
+      "step": 5980
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.1613576412200928,
+      "learning_rate": 9.88143961039191e-06,
+      "loss": 0.6432,
+      "step": 5981
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.4906721115112305,
+      "learning_rate": 9.88138252791639e-06,
+      "loss": 0.6862,
+      "step": 5982
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.5588560104370117,
+      "learning_rate": 9.881325431867576e-06,
+      "loss": 0.7296,
+      "step": 5983
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.547081470489502,
+      "learning_rate": 9.881268322245623e-06,
+      "loss": 0.57,
+      "step": 5984
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.8894765377044678,
+      "learning_rate": 9.88121119905069e-06,
+      "loss": 0.6393,
+      "step": 5985
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 4.274303436279297,
+      "learning_rate": 9.88115406228294e-06,
+      "loss": 0.6598,
+      "step": 5986
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 5.80501127243042,
+      "learning_rate": 9.881096911942527e-06,
+      "loss": 0.7727,
+      "step": 5987
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.16504168510437,
+      "learning_rate": 9.881039748029613e-06,
+      "loss": 0.9381,
+      "step": 5988
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 12.453754425048828,
+      "learning_rate": 9.880982570544356e-06,
+      "loss": 0.6769,
+      "step": 5989
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.8750874996185303,
+      "learning_rate": 9.880925379486914e-06,
+      "loss": 0.6464,
+      "step": 5990
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 4.090942859649658,
+      "learning_rate": 9.880868174857448e-06,
+      "loss": 0.8438,
+      "step": 5991
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.3380298614501953,
+      "learning_rate": 9.880810956656115e-06,
+      "loss": 0.7336,
+      "step": 5992
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.6387882232666016,
+      "learning_rate": 9.880753724883075e-06,
+      "loss": 0.5343,
+      "step": 5993
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.3405373096466064,
+      "learning_rate": 9.880696479538488e-06,
+      "loss": 0.5495,
+      "step": 5994
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.8460261821746826,
+      "learning_rate": 9.880639220622513e-06,
+      "loss": 0.7914,
+      "step": 5995
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 5.149903774261475,
+      "learning_rate": 9.880581948135308e-06,
+      "loss": 1.108,
+      "step": 5996
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.532992124557495,
+      "learning_rate": 9.880524662077034e-06,
+      "loss": 0.7269,
+      "step": 5997
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 7.78752326965332,
+      "learning_rate": 9.880467362447847e-06,
+      "loss": 0.654,
+      "step": 5998
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 4.237514019012451,
+      "learning_rate": 9.880410049247909e-06,
+      "loss": 0.7268,
+      "step": 5999
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 6.108941555023193,
+      "learning_rate": 9.88035272247738e-06,
+      "loss": 0.4855,
+      "step": 6000
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.1356329917907715,
+      "learning_rate": 9.880295382136418e-06,
+      "loss": 0.8755,
+      "step": 6001
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 7.8125715255737305,
+      "learning_rate": 9.880238028225182e-06,
+      "loss": 0.5368,
+      "step": 6002
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.810467481613159,
+      "learning_rate": 9.880180660743832e-06,
+      "loss": 0.9209,
+      "step": 6003
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 11.737865447998047,
+      "learning_rate": 9.880123279692529e-06,
+      "loss": 0.6889,
+      "step": 6004
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.9575504064559937,
+      "learning_rate": 9.880065885071431e-06,
+      "loss": 0.6943,
+      "step": 6005
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.8094608783721924,
+      "learning_rate": 9.880008476880698e-06,
+      "loss": 0.6406,
+      "step": 6006
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.9376437664031982,
+      "learning_rate": 9.879951055120489e-06,
+      "loss": 0.4032,
+      "step": 6007
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.668034076690674,
+      "learning_rate": 9.879893619790964e-06,
+      "loss": 0.6174,
+      "step": 6008
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.4152684211730957,
+      "learning_rate": 9.879836170892284e-06,
+      "loss": 0.7746,
+      "step": 6009
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.7216012477874756,
+      "learning_rate": 9.879778708424608e-06,
+      "loss": 0.8521,
+      "step": 6010
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 4.605466842651367,
+      "learning_rate": 9.879721232388092e-06,
+      "loss": 0.5165,
+      "step": 6011
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 6.506742000579834,
+      "learning_rate": 9.879663742782904e-06,
+      "loss": 0.62,
+      "step": 6012
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 8.72323989868164,
+      "learning_rate": 9.879606239609196e-06,
+      "loss": 0.5801,
+      "step": 6013
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 6.2944512367248535,
+      "learning_rate": 9.879548722867132e-06,
+      "loss": 0.6633,
+      "step": 6014
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 9.942667961120605,
+      "learning_rate": 9.879491192556869e-06,
+      "loss": 0.6347,
+      "step": 6015
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.4106152057647705,
+      "learning_rate": 9.87943364867857e-06,
+      "loss": 0.6167,
+      "step": 6016
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.531442880630493,
+      "learning_rate": 9.879376091232396e-06,
+      "loss": 0.5601,
+      "step": 6017
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.9892746210098267,
+      "learning_rate": 9.879318520218503e-06,
+      "loss": 0.9698,
+      "step": 6018
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.2281558513641357,
+      "learning_rate": 9.879260935637053e-06,
+      "loss": 0.5832,
+      "step": 6019
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.2599048614501953,
+      "learning_rate": 9.879203337488208e-06,
+      "loss": 0.8215,
+      "step": 6020
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.6903109550476074,
+      "learning_rate": 9.879145725772124e-06,
+      "loss": 0.5618,
+      "step": 6021
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.5310401916503906,
+      "learning_rate": 9.879088100488965e-06,
+      "loss": 0.8565,
+      "step": 6022
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.8369650840759277,
+      "learning_rate": 9.879030461638887e-06,
+      "loss": 0.7381,
+      "step": 6023
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.4731075763702393,
+      "learning_rate": 9.878972809222055e-06,
+      "loss": 0.6175,
+      "step": 6024
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.0577213764190674,
+      "learning_rate": 9.878915143238629e-06,
+      "loss": 0.5425,
+      "step": 6025
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.9783103466033936,
+      "learning_rate": 9.878857463688766e-06,
+      "loss": 0.7065,
+      "step": 6026
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.2086212635040283,
+      "learning_rate": 9.878799770572628e-06,
+      "loss": 0.6331,
+      "step": 6027
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.2013533115386963,
+      "learning_rate": 9.878742063890376e-06,
+      "loss": 0.6941,
+      "step": 6028
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 4.485550403594971,
+      "learning_rate": 9.878684343642168e-06,
+      "loss": 0.7647,
+      "step": 6029
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.486067295074463,
+      "learning_rate": 9.878626609828169e-06,
+      "loss": 0.4737,
+      "step": 6030
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.552276849746704,
+      "learning_rate": 9.878568862448537e-06,
+      "loss": 0.6278,
+      "step": 6031
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.078230381011963,
+      "learning_rate": 9.878511101503431e-06,
+      "loss": 0.5608,
+      "step": 6032
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.978761672973633,
+      "learning_rate": 9.878453326993014e-06,
+      "loss": 0.6687,
+      "step": 6033
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 5.749993324279785,
+      "learning_rate": 9.878395538917445e-06,
+      "loss": 0.7456,
+      "step": 6034
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.148780107498169,
+      "learning_rate": 9.878337737276886e-06,
+      "loss": 0.7053,
+      "step": 6035
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 4.028290271759033,
+      "learning_rate": 9.878279922071497e-06,
+      "loss": 0.6551,
+      "step": 6036
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.2602741718292236,
+      "learning_rate": 9.87822209330144e-06,
+      "loss": 0.8606,
+      "step": 6037
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.0440406799316406,
+      "learning_rate": 9.878164250966873e-06,
+      "loss": 0.5747,
+      "step": 6038
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 5.373800754547119,
+      "learning_rate": 9.87810639506796e-06,
+      "loss": 0.779,
+      "step": 6039
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.0703039169311523,
+      "learning_rate": 9.87804852560486e-06,
+      "loss": 0.7584,
+      "step": 6040
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.5992372035980225,
+      "learning_rate": 9.877990642577736e-06,
+      "loss": 0.7019,
+      "step": 6041
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.2907326221466064,
+      "learning_rate": 9.877932745986745e-06,
+      "loss": 0.777,
+      "step": 6042
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.2294323444366455,
+      "learning_rate": 9.87787483583205e-06,
+      "loss": 0.4196,
+      "step": 6043
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.494499444961548,
+      "learning_rate": 9.877816912113813e-06,
+      "loss": 0.7435,
+      "step": 6044
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 4.693303108215332,
+      "learning_rate": 9.877758974832196e-06,
+      "loss": 0.3974,
+      "step": 6045
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.4188554286956787,
+      "learning_rate": 9.877701023987358e-06,
+      "loss": 0.4252,
+      "step": 6046
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.757127285003662,
+      "learning_rate": 9.87764305957946e-06,
+      "loss": 0.6543,
+      "step": 6047
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.2806854248046875,
+      "learning_rate": 9.877585081608666e-06,
+      "loss": 0.6506,
+      "step": 6048
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.9667880535125732,
+      "learning_rate": 9.877527090075132e-06,
+      "loss": 0.7291,
+      "step": 6049
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.073418617248535,
+      "learning_rate": 9.877469084979023e-06,
+      "loss": 0.726,
+      "step": 6050
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.2606847286224365,
+      "learning_rate": 9.8774110663205e-06,
+      "loss": 0.7884,
+      "step": 6051
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.163151502609253,
+      "learning_rate": 9.877353034099723e-06,
+      "loss": 0.7641,
+      "step": 6052
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.7985594272613525,
+      "learning_rate": 9.877294988316857e-06,
+      "loss": 0.4037,
+      "step": 6053
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 4.745019912719727,
+      "learning_rate": 9.877236928972058e-06,
+      "loss": 0.5761,
+      "step": 6054
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 4.19899320602417,
+      "learning_rate": 9.877178856065492e-06,
+      "loss": 0.8865,
+      "step": 6055
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 12.377310752868652,
+      "learning_rate": 9.877120769597318e-06,
+      "loss": 0.6307,
+      "step": 6056
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 6.384642124176025,
+      "learning_rate": 9.877062669567698e-06,
+      "loss": 0.9051,
+      "step": 6057
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 6.276153564453125,
+      "learning_rate": 9.877004555976793e-06,
+      "loss": 0.7926,
+      "step": 6058
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.953953266143799,
+      "learning_rate": 9.876946428824766e-06,
+      "loss": 0.5858,
+      "step": 6059
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.9770872592926025,
+      "learning_rate": 9.87688828811178e-06,
+      "loss": 0.6145,
+      "step": 6060
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.242762804031372,
+      "learning_rate": 9.87683013383799e-06,
+      "loss": 0.7248,
+      "step": 6061
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.490490436553955,
+      "learning_rate": 9.876771966003565e-06,
+      "loss": 0.5698,
+      "step": 6062
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.2606687545776367,
+      "learning_rate": 9.876713784608664e-06,
+      "loss": 0.628,
+      "step": 6063
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.2440037727355957,
+      "learning_rate": 9.87665558965345e-06,
+      "loss": 0.8327,
+      "step": 6064
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.068167209625244,
+      "learning_rate": 9.876597381138082e-06,
+      "loss": 0.5533,
+      "step": 6065
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.7290289402008057,
+      "learning_rate": 9.876539159062724e-06,
+      "loss": 0.8073,
+      "step": 6066
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.9387948513031006,
+      "learning_rate": 9.876480923427536e-06,
+      "loss": 0.7984,
+      "step": 6067
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.4563746452331543,
+      "learning_rate": 9.876422674232684e-06,
+      "loss": 0.7478,
+      "step": 6068
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 4.2552595138549805,
+      "learning_rate": 9.876364411478325e-06,
+      "loss": 0.549,
+      "step": 6069
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.6157491207122803,
+      "learning_rate": 9.876306135164624e-06,
+      "loss": 0.5683,
+      "step": 6070
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.8636786937713623,
+      "learning_rate": 9.876247845291741e-06,
+      "loss": 0.8974,
+      "step": 6071
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.565389394760132,
+      "learning_rate": 9.876189541859843e-06,
+      "loss": 0.8117,
+      "step": 6072
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 4.594772815704346,
+      "learning_rate": 9.876131224869085e-06,
+      "loss": 0.6524,
+      "step": 6073
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.3207812309265137,
+      "learning_rate": 9.876072894319634e-06,
+      "loss": 0.7118,
+      "step": 6074
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.1069977283477783,
+      "learning_rate": 9.876014550211652e-06,
+      "loss": 0.8547,
+      "step": 6075
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 4.395376205444336,
+      "learning_rate": 9.8759561925453e-06,
+      "loss": 0.7173,
+      "step": 6076
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.9471566677093506,
+      "learning_rate": 9.875897821320739e-06,
+      "loss": 0.7819,
+      "step": 6077
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 5.138659477233887,
+      "learning_rate": 9.875839436538134e-06,
+      "loss": 0.4658,
+      "step": 6078
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 4.920341968536377,
+      "learning_rate": 9.875781038197645e-06,
+      "loss": 0.7883,
+      "step": 6079
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.9042294025421143,
+      "learning_rate": 9.875722626299435e-06,
+      "loss": 0.7837,
+      "step": 6080
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.7520580291748047,
+      "learning_rate": 9.87566420084367e-06,
+      "loss": 0.6676,
+      "step": 6081
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.8863534927368164,
+      "learning_rate": 9.875605761830506e-06,
+      "loss": 0.6976,
+      "step": 6082
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.3961524963378906,
+      "learning_rate": 9.875547309260109e-06,
+      "loss": 0.6639,
+      "step": 6083
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 4.950158596038818,
+      "learning_rate": 9.875488843132643e-06,
+      "loss": 0.6681,
+      "step": 6084
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 4.893917083740234,
+      "learning_rate": 9.875430363448269e-06,
+      "loss": 0.8225,
+      "step": 6085
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.4783198833465576,
+      "learning_rate": 9.875371870207148e-06,
+      "loss": 0.7822,
+      "step": 6086
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.936004400253296,
+      "learning_rate": 9.875313363409444e-06,
+      "loss": 0.6663,
+      "step": 6087
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.53000545501709,
+      "learning_rate": 9.875254843055322e-06,
+      "loss": 0.7633,
+      "step": 6088
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.351088762283325,
+      "learning_rate": 9.875196309144941e-06,
+      "loss": 0.6289,
+      "step": 6089
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.523812770843506,
+      "learning_rate": 9.875137761678465e-06,
+      "loss": 0.7909,
+      "step": 6090
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.6983824968338013,
+      "learning_rate": 9.875079200656058e-06,
+      "loss": 0.5877,
+      "step": 6091
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.8230834007263184,
+      "learning_rate": 9.875020626077882e-06,
+      "loss": 0.7018,
+      "step": 6092
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.9409327507019043,
+      "learning_rate": 9.874962037944098e-06,
+      "loss": 0.7093,
+      "step": 6093
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 4.345630168914795,
+      "learning_rate": 9.874903436254874e-06,
+      "loss": 0.7933,
+      "step": 6094
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.458157777786255,
+      "learning_rate": 9.874844821010368e-06,
+      "loss": 0.6735,
+      "step": 6095
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.201463222503662,
+      "learning_rate": 9.874786192210745e-06,
+      "loss": 0.4551,
+      "step": 6096
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.8790674209594727,
+      "learning_rate": 9.874727549856167e-06,
+      "loss": 0.7386,
+      "step": 6097
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.110517740249634,
+      "learning_rate": 9.8746688939468e-06,
+      "loss": 0.7631,
+      "step": 6098
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.9272379875183105,
+      "learning_rate": 9.874610224482802e-06,
+      "loss": 0.746,
+      "step": 6099
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.9459962844848633,
+      "learning_rate": 9.87455154146434e-06,
+      "loss": 0.7151,
+      "step": 6100
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.382002353668213,
+      "learning_rate": 9.874492844891577e-06,
+      "loss": 0.9348,
+      "step": 6101
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.9847800731658936,
+      "learning_rate": 9.874434134764674e-06,
+      "loss": 0.4071,
+      "step": 6102
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.3064613342285156,
+      "learning_rate": 9.874375411083797e-06,
+      "loss": 0.4866,
+      "step": 6103
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.965808391571045,
+      "learning_rate": 9.87431667384911e-06,
+      "loss": 0.9562,
+      "step": 6104
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.290388822555542,
+      "learning_rate": 9.87425792306077e-06,
+      "loss": 0.6267,
+      "step": 6105
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.4432742595672607,
+      "learning_rate": 9.874199158718946e-06,
+      "loss": 0.5431,
+      "step": 6106
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 4.271233081817627,
+      "learning_rate": 9.874140380823803e-06,
+      "loss": 1.0127,
+      "step": 6107
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.1702961921691895,
+      "learning_rate": 9.874081589375497e-06,
+      "loss": 0.9127,
+      "step": 6108
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.4382410049438477,
+      "learning_rate": 9.874022784374199e-06,
+      "loss": 0.7808,
+      "step": 6109
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.4130141735076904,
+      "learning_rate": 9.873963965820068e-06,
+      "loss": 0.6466,
+      "step": 6110
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.4160170555114746,
+      "learning_rate": 9.87390513371327e-06,
+      "loss": 0.5394,
+      "step": 6111
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.8049943447113037,
+      "learning_rate": 9.873846288053968e-06,
+      "loss": 0.4503,
+      "step": 6112
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.571155071258545,
+      "learning_rate": 9.873787428842323e-06,
+      "loss": 0.4806,
+      "step": 6113
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.269192934036255,
+      "learning_rate": 9.873728556078504e-06,
+      "loss": 0.557,
+      "step": 6114
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.377516269683838,
+      "learning_rate": 9.873669669762668e-06,
+      "loss": 0.6961,
+      "step": 6115
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.5203909873962402,
+      "learning_rate": 9.873610769894985e-06,
+      "loss": 0.3691,
+      "step": 6116
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.7279044389724731,
+      "learning_rate": 9.873551856475615e-06,
+      "loss": 0.7461,
+      "step": 6117
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.642704963684082,
+      "learning_rate": 9.873492929504725e-06,
+      "loss": 0.6622,
+      "step": 6118
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.9618360996246338,
+      "learning_rate": 9.873433988982475e-06,
+      "loss": 0.8038,
+      "step": 6119
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.1401660442352295,
+      "learning_rate": 9.873375034909032e-06,
+      "loss": 0.5751,
+      "step": 6120
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.5567777156829834,
+      "learning_rate": 9.873316067284558e-06,
+      "loss": 0.5686,
+      "step": 6121
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.4029715061187744,
+      "learning_rate": 9.873257086109215e-06,
+      "loss": 0.817,
+      "step": 6122
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.3360283374786377,
+      "learning_rate": 9.873198091383175e-06,
+      "loss": 0.6323,
+      "step": 6123
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.509542465209961,
+      "learning_rate": 9.87313908310659e-06,
+      "loss": 1.0171,
+      "step": 6124
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.3567447662353516,
+      "learning_rate": 9.873080061279636e-06,
+      "loss": 0.6808,
+      "step": 6125
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 4.7884626388549805,
+      "learning_rate": 9.87302102590247e-06,
+      "loss": 0.5371,
+      "step": 6126
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 4.454724311828613,
+      "learning_rate": 9.872961976975259e-06,
+      "loss": 1.0016,
+      "step": 6127
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.741450071334839,
+      "learning_rate": 9.872902914498165e-06,
+      "loss": 0.5613,
+      "step": 6128
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.4882686138153076,
+      "learning_rate": 9.872843838471352e-06,
+      "loss": 0.3764,
+      "step": 6129
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.4670603275299072,
+      "learning_rate": 9.872784748894988e-06,
+      "loss": 0.6899,
+      "step": 6130
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 7.418478012084961,
+      "learning_rate": 9.872725645769235e-06,
+      "loss": 0.3035,
+      "step": 6131
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.311957836151123,
+      "learning_rate": 9.872666529094257e-06,
+      "loss": 0.6758,
+      "step": 6132
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.282191276550293,
+      "learning_rate": 9.872607398870219e-06,
+      "loss": 0.8518,
+      "step": 6133
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.4157562255859375,
+      "learning_rate": 9.872548255097284e-06,
+      "loss": 0.7409,
+      "step": 6134
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.318953037261963,
+      "learning_rate": 9.872489097775618e-06,
+      "loss": 0.9207,
+      "step": 6135
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.225126028060913,
+      "learning_rate": 9.872429926905386e-06,
+      "loss": 0.5727,
+      "step": 6136
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 7.993923187255859,
+      "learning_rate": 9.87237074248675e-06,
+      "loss": 0.7284,
+      "step": 6137
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 7.760264873504639,
+      "learning_rate": 9.872311544519877e-06,
+      "loss": 0.6928,
+      "step": 6138
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.763357400894165,
+      "learning_rate": 9.872252333004932e-06,
+      "loss": 0.6755,
+      "step": 6139
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.959035634994507,
+      "learning_rate": 9.872193107942077e-06,
+      "loss": 0.7609,
+      "step": 6140
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.0490753650665283,
+      "learning_rate": 9.87213386933148e-06,
+      "loss": 0.6186,
+      "step": 6141
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.566539764404297,
+      "learning_rate": 9.872074617173302e-06,
+      "loss": 0.7064,
+      "step": 6142
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 4.413809299468994,
+      "learning_rate": 9.87201535146771e-06,
+      "loss": 0.875,
+      "step": 6143
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.751133441925049,
+      "learning_rate": 9.871956072214867e-06,
+      "loss": 0.4157,
+      "step": 6144
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.7908103466033936,
+      "learning_rate": 9.871896779414943e-06,
+      "loss": 0.7098,
+      "step": 6145
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 4.0822625160217285,
+      "learning_rate": 9.871837473068095e-06,
+      "loss": 0.6972,
+      "step": 6146
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.750717878341675,
+      "learning_rate": 9.871778153174495e-06,
+      "loss": 0.8322,
+      "step": 6147
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.8299615383148193,
+      "learning_rate": 9.871718819734303e-06,
+      "loss": 0.5692,
+      "step": 6148
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 4.202718257904053,
+      "learning_rate": 9.871659472747687e-06,
+      "loss": 0.7172,
+      "step": 6149
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.0893146991729736,
+      "learning_rate": 9.87160011221481e-06,
+      "loss": 0.7066,
+      "step": 6150
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.4666366577148438,
+      "learning_rate": 9.871540738135838e-06,
+      "loss": 0.6847,
+      "step": 6151
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.0381336212158203,
+      "learning_rate": 9.87148135051094e-06,
+      "loss": 0.4439,
+      "step": 6152
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.5080487728118896,
+      "learning_rate": 9.871421949340273e-06,
+      "loss": 0.6179,
+      "step": 6153
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.5379767417907715,
+      "learning_rate": 9.871362534624007e-06,
+      "loss": 0.8844,
+      "step": 6154
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.5111968517303467,
+      "learning_rate": 9.871303106362306e-06,
+      "loss": 0.6913,
+      "step": 6155
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.178269624710083,
+      "learning_rate": 9.871243664555337e-06,
+      "loss": 0.8363,
+      "step": 6156
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.0398168563842773,
+      "learning_rate": 9.871184209203264e-06,
+      "loss": 0.818,
+      "step": 6157
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.733241319656372,
+      "learning_rate": 9.871124740306254e-06,
+      "loss": 0.7995,
+      "step": 6158
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.6129233837127686,
+      "learning_rate": 9.87106525786447e-06,
+      "loss": 0.7875,
+      "step": 6159
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.694861650466919,
+      "learning_rate": 9.871005761878077e-06,
+      "loss": 0.7668,
+      "step": 6160
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.018501043319702,
+      "learning_rate": 9.870946252347242e-06,
+      "loss": 0.9175,
+      "step": 6161
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 4.216161727905273,
+      "learning_rate": 9.87088672927213e-06,
+      "loss": 0.7946,
+      "step": 6162
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.6989147663116455,
+      "learning_rate": 9.87082719265291e-06,
+      "loss": 0.7191,
+      "step": 6163
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.0004355907440186,
+      "learning_rate": 9.87076764248974e-06,
+      "loss": 0.6643,
+      "step": 6164
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.9027352333068848,
+      "learning_rate": 9.870708078782792e-06,
+      "loss": 0.4907,
+      "step": 6165
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.728907823562622,
+      "learning_rate": 9.87064850153223e-06,
+      "loss": 0.693,
+      "step": 6166
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.8585052490234375,
+      "learning_rate": 9.870588910738216e-06,
+      "loss": 0.8046,
+      "step": 6167
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.5388965606689453,
+      "learning_rate": 9.870529306400922e-06,
+      "loss": 0.704,
+      "step": 6168
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 4.803625583648682,
+      "learning_rate": 9.87046968852051e-06,
+      "loss": 0.6783,
+      "step": 6169
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.4362905025482178,
+      "learning_rate": 9.870410057097144e-06,
+      "loss": 0.7964,
+      "step": 6170
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.2207863330841064,
+      "learning_rate": 9.870350412130995e-06,
+      "loss": 0.8054,
+      "step": 6171
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.911907196044922,
+      "learning_rate": 9.870290753622225e-06,
+      "loss": 0.851,
+      "step": 6172
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 4.408554553985596,
+      "learning_rate": 9.870231081571e-06,
+      "loss": 0.8028,
+      "step": 6173
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.6755123138427734,
+      "learning_rate": 9.870171395977488e-06,
+      "loss": 0.8295,
+      "step": 6174
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.1166481971740723,
+      "learning_rate": 9.870111696841853e-06,
+      "loss": 0.7194,
+      "step": 6175
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.4546988010406494,
+      "learning_rate": 9.870051984164261e-06,
+      "loss": 0.8786,
+      "step": 6176
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.242637872695923,
+      "learning_rate": 9.86999225794488e-06,
+      "loss": 0.6666,
+      "step": 6177
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.0924997329711914,
+      "learning_rate": 9.869932518183875e-06,
+      "loss": 0.6381,
+      "step": 6178
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.4948790073394775,
+      "learning_rate": 9.869872764881412e-06,
+      "loss": 0.7367,
+      "step": 6179
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.999211311340332,
+      "learning_rate": 9.869812998037657e-06,
+      "loss": 0.7931,
+      "step": 6180
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.8477847576141357,
+      "learning_rate": 9.869753217652776e-06,
+      "loss": 0.761,
+      "step": 6181
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 7.579611778259277,
+      "learning_rate": 9.869693423726936e-06,
+      "loss": 0.5844,
+      "step": 6182
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.2848927974700928,
+      "learning_rate": 9.869633616260303e-06,
+      "loss": 0.7921,
+      "step": 6183
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.2805399894714355,
+      "learning_rate": 9.869573795253043e-06,
+      "loss": 0.7811,
+      "step": 6184
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.6787970066070557,
+      "learning_rate": 9.869513960705321e-06,
+      "loss": 0.3534,
+      "step": 6185
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.2999796867370605,
+      "learning_rate": 9.869454112617305e-06,
+      "loss": 0.6043,
+      "step": 6186
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.815166711807251,
+      "learning_rate": 9.869394250989163e-06,
+      "loss": 0.5709,
+      "step": 6187
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.9800591468811035,
+      "learning_rate": 9.869334375821058e-06,
+      "loss": 1.0054,
+      "step": 6188
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.8805139064788818,
+      "learning_rate": 9.86927448711316e-06,
+      "loss": 0.5438,
+      "step": 6189
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.851407766342163,
+      "learning_rate": 9.869214584865632e-06,
+      "loss": 0.6738,
+      "step": 6190
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.9850878715515137,
+      "learning_rate": 9.869154669078643e-06,
+      "loss": 0.7338,
+      "step": 6191
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.4504263401031494,
+      "learning_rate": 9.869094739752359e-06,
+      "loss": 0.8539,
+      "step": 6192
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.776113510131836,
+      "learning_rate": 9.869034796886946e-06,
+      "loss": 0.6146,
+      "step": 6193
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.768092632293701,
+      "learning_rate": 9.868974840482573e-06,
+      "loss": 0.4864,
+      "step": 6194
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.4165027141571045,
+      "learning_rate": 9.868914870539404e-06,
+      "loss": 0.6477,
+      "step": 6195
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.0622315406799316,
+      "learning_rate": 9.868854887057606e-06,
+      "loss": 0.6529,
+      "step": 6196
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.1311604976654053,
+      "learning_rate": 9.868794890037347e-06,
+      "loss": 0.7203,
+      "step": 6197
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 6.971319675445557,
+      "learning_rate": 9.868734879478792e-06,
+      "loss": 1.022,
+      "step": 6198
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.2775185108184814,
+      "learning_rate": 9.86867485538211e-06,
+      "loss": 0.6805,
+      "step": 6199
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.353850841522217,
+      "learning_rate": 9.868614817747468e-06,
+      "loss": 0.5397,
+      "step": 6200
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.6061859130859375,
+      "learning_rate": 9.86855476657503e-06,
+      "loss": 0.6806,
+      "step": 6201
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.974679470062256,
+      "learning_rate": 9.868494701864969e-06,
+      "loss": 0.8741,
+      "step": 6202
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.3003060817718506,
+      "learning_rate": 9.868434623617445e-06,
+      "loss": 0.5532,
+      "step": 6203
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.9005130529403687,
+      "learning_rate": 9.868374531832627e-06,
+      "loss": 0.7666,
+      "step": 6204
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 5.500938415527344,
+      "learning_rate": 9.868314426510686e-06,
+      "loss": 0.7755,
+      "step": 6205
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.366834878921509,
+      "learning_rate": 9.868254307651783e-06,
+      "loss": 0.6068,
+      "step": 6206
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 4.3673906326293945,
+      "learning_rate": 9.868194175256092e-06,
+      "loss": 0.5699,
+      "step": 6207
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.9566688537597656,
+      "learning_rate": 9.868134029323773e-06,
+      "loss": 0.6732,
+      "step": 6208
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.042996644973755,
+      "learning_rate": 9.868073869854999e-06,
+      "loss": 0.4626,
+      "step": 6209
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.0510215759277344,
+      "learning_rate": 9.868013696849937e-06,
+      "loss": 0.5498,
+      "step": 6210
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 7.058084487915039,
+      "learning_rate": 9.86795351030875e-06,
+      "loss": 0.7945,
+      "step": 6211
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.6927404403686523,
+      "learning_rate": 9.867893310231609e-06,
+      "loss": 0.691,
+      "step": 6212
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.819547176361084,
+      "learning_rate": 9.86783309661868e-06,
+      "loss": 0.7971,
+      "step": 6213
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.3596739768981934,
+      "learning_rate": 9.867772869470129e-06,
+      "loss": 0.6574,
+      "step": 6214
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.1589086055755615,
+      "learning_rate": 9.867712628786126e-06,
+      "loss": 0.6697,
+      "step": 6215
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.672475814819336,
+      "learning_rate": 9.867652374566838e-06,
+      "loss": 0.6342,
+      "step": 6216
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.3931360244750977,
+      "learning_rate": 9.867592106812433e-06,
+      "loss": 0.5744,
+      "step": 6217
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.12910532951355,
+      "learning_rate": 9.867531825523077e-06,
+      "loss": 0.7894,
+      "step": 6218
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.147881269454956,
+      "learning_rate": 9.86747153069894e-06,
+      "loss": 1.071,
+      "step": 6219
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 4.464024066925049,
+      "learning_rate": 9.867411222340187e-06,
+      "loss": 0.4556,
+      "step": 6220
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.514955759048462,
+      "learning_rate": 9.867350900446985e-06,
+      "loss": 0.3387,
+      "step": 6221
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 8.53402042388916,
+      "learning_rate": 9.867290565019505e-06,
+      "loss": 0.6439,
+      "step": 6222
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.8002045154571533,
+      "learning_rate": 9.867230216057913e-06,
+      "loss": 0.7801,
+      "step": 6223
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 12.97934341430664,
+      "learning_rate": 9.867169853562378e-06,
+      "loss": 0.4705,
+      "step": 6224
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.938488483428955,
+      "learning_rate": 9.867109477533066e-06,
+      "loss": 0.6194,
+      "step": 6225
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 9.358681678771973,
+      "learning_rate": 9.867049087970146e-06,
+      "loss": 0.5495,
+      "step": 6226
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.3096582889556885,
+      "learning_rate": 9.866988684873786e-06,
+      "loss": 0.6583,
+      "step": 6227
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.2599401473999023,
+      "learning_rate": 9.866928268244154e-06,
+      "loss": 0.5304,
+      "step": 6228
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.1052896976470947,
+      "learning_rate": 9.866867838081417e-06,
+      "loss": 0.5974,
+      "step": 6229
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.8155081272125244,
+      "learning_rate": 9.866807394385743e-06,
+      "loss": 0.5431,
+      "step": 6230
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.956955671310425,
+      "learning_rate": 9.866746937157303e-06,
+      "loss": 0.5583,
+      "step": 6231
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.675105333328247,
+      "learning_rate": 9.86668646639626e-06,
+      "loss": 0.5897,
+      "step": 6232
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.3655755519866943,
+      "learning_rate": 9.866625982102786e-06,
+      "loss": 0.5166,
+      "step": 6233
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.485846996307373,
+      "learning_rate": 9.866565484277051e-06,
+      "loss": 0.5098,
+      "step": 6234
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.090494394302368,
+      "learning_rate": 9.866504972919218e-06,
+      "loss": 0.6982,
+      "step": 6235
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.392801523208618,
+      "learning_rate": 9.866444448029458e-06,
+      "loss": 0.8044,
+      "step": 6236
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.0044939517974854,
+      "learning_rate": 9.866383909607938e-06,
+      "loss": 0.7699,
+      "step": 6237
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.7676661014556885,
+      "learning_rate": 9.866323357654827e-06,
+      "loss": 0.7791,
+      "step": 6238
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.6638753414154053,
+      "learning_rate": 9.866262792170296e-06,
+      "loss": 0.5298,
+      "step": 6239
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 4.140051364898682,
+      "learning_rate": 9.86620221315451e-06,
+      "loss": 0.6886,
+      "step": 6240
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.3173391819000244,
+      "learning_rate": 9.86614162060764e-06,
+      "loss": 0.6817,
+      "step": 6241
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.269362688064575,
+      "learning_rate": 9.86608101452985e-06,
+      "loss": 0.7404,
+      "step": 6242
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.837207794189453,
+      "learning_rate": 9.866020394921314e-06,
+      "loss": 0.9296,
+      "step": 6243
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.996558666229248,
+      "learning_rate": 9.865959761782198e-06,
+      "loss": 0.6076,
+      "step": 6244
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.209221124649048,
+      "learning_rate": 9.865899115112669e-06,
+      "loss": 0.733,
+      "step": 6245
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.5449647903442383,
+      "learning_rate": 9.865838454912899e-06,
+      "loss": 0.7719,
+      "step": 6246
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.086958408355713,
+      "learning_rate": 9.865777781183054e-06,
+      "loss": 0.8322,
+      "step": 6247
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.1913321018218994,
+      "learning_rate": 9.865717093923305e-06,
+      "loss": 0.6895,
+      "step": 6248
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.5059895515441895,
+      "learning_rate": 9.86565639313382e-06,
+      "loss": 0.4837,
+      "step": 6249
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.4690003395080566,
+      "learning_rate": 9.865595678814766e-06,
+      "loss": 1.0637,
+      "step": 6250
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 4.173593044281006,
+      "learning_rate": 9.865534950966313e-06,
+      "loss": 0.587,
+      "step": 6251
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.423034429550171,
+      "learning_rate": 9.865474209588631e-06,
+      "loss": 0.5066,
+      "step": 6252
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.9459149837493896,
+      "learning_rate": 9.865413454681886e-06,
+      "loss": 0.7911,
+      "step": 6253
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.1724419593811035,
+      "learning_rate": 9.86535268624625e-06,
+      "loss": 0.389,
+      "step": 6254
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.1825129985809326,
+      "learning_rate": 9.86529190428189e-06,
+      "loss": 0.516,
+      "step": 6255
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.388089418411255,
+      "learning_rate": 9.865231108788978e-06,
+      "loss": 0.5227,
+      "step": 6256
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 4.091557025909424,
+      "learning_rate": 9.865170299767679e-06,
+      "loss": 0.472,
+      "step": 6257
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 4.454445838928223,
+      "learning_rate": 9.865109477218164e-06,
+      "loss": 0.7724,
+      "step": 6258
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.8055182695388794,
+      "learning_rate": 9.865048641140603e-06,
+      "loss": 0.6093,
+      "step": 6259
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.8567399978637695,
+      "learning_rate": 9.864987791535163e-06,
+      "loss": 0.7535,
+      "step": 6260
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.740567684173584,
+      "learning_rate": 9.864926928402014e-06,
+      "loss": 0.7998,
+      "step": 6261
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.1208510398864746,
+      "learning_rate": 9.864866051741326e-06,
+      "loss": 0.5958,
+      "step": 6262
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 4.1284589767456055,
+      "learning_rate": 9.864805161553269e-06,
+      "loss": 0.5636,
+      "step": 6263
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.8784897327423096,
+      "learning_rate": 9.864744257838011e-06,
+      "loss": 0.6223,
+      "step": 6264
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 4.149277687072754,
+      "learning_rate": 9.864683340595722e-06,
+      "loss": 0.4947,
+      "step": 6265
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.9632086753845215,
+      "learning_rate": 9.86462240982657e-06,
+      "loss": 0.5925,
+      "step": 6266
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.169137716293335,
+      "learning_rate": 9.864561465530726e-06,
+      "loss": 0.6677,
+      "step": 6267
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.2211732864379883,
+      "learning_rate": 9.864500507708357e-06,
+      "loss": 0.532,
+      "step": 6268
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.810567617416382,
+      "learning_rate": 9.864439536359636e-06,
+      "loss": 0.5113,
+      "step": 6269
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.9040515422821045,
+      "learning_rate": 9.864378551484732e-06,
+      "loss": 0.5966,
+      "step": 6270
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 4.923914909362793,
+      "learning_rate": 9.864317553083812e-06,
+      "loss": 0.7105,
+      "step": 6271
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.7987687587738037,
+      "learning_rate": 9.864256541157047e-06,
+      "loss": 0.6422,
+      "step": 6272
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.470407009124756,
+      "learning_rate": 9.864195515704608e-06,
+      "loss": 0.681,
+      "step": 6273
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.2206945419311523,
+      "learning_rate": 9.864134476726662e-06,
+      "loss": 0.6775,
+      "step": 6274
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.22849178314209,
+      "learning_rate": 9.864073424223382e-06,
+      "loss": 0.8723,
+      "step": 6275
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 4.562882900238037,
+      "learning_rate": 9.864012358194935e-06,
+      "loss": 0.6627,
+      "step": 6276
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.7925984859466553,
+      "learning_rate": 9.86395127864149e-06,
+      "loss": 0.7323,
+      "step": 6277
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 4.580107688903809,
+      "learning_rate": 9.863890185563221e-06,
+      "loss": 0.5319,
+      "step": 6278
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 4.786999225616455,
+      "learning_rate": 9.863829078960294e-06,
+      "loss": 0.9973,
+      "step": 6279
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 6.736896991729736,
+      "learning_rate": 9.863767958832882e-06,
+      "loss": 0.6205,
+      "step": 6280
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.075258493423462,
+      "learning_rate": 9.863706825181152e-06,
+      "loss": 0.8175,
+      "step": 6281
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.3129734992980957,
+      "learning_rate": 9.863645678005274e-06,
+      "loss": 0.5659,
+      "step": 6282
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 5.393900394439697,
+      "learning_rate": 9.863584517305423e-06,
+      "loss": 0.6253,
+      "step": 6283
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.477888584136963,
+      "learning_rate": 9.863523343081762e-06,
+      "loss": 0.6573,
+      "step": 6284
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 4.111656188964844,
+      "learning_rate": 9.863462155334466e-06,
+      "loss": 0.8703,
+      "step": 6285
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.0094058513641357,
+      "learning_rate": 9.863400954063703e-06,
+      "loss": 0.6794,
+      "step": 6286
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.245807647705078,
+      "learning_rate": 9.863339739269644e-06,
+      "loss": 0.6685,
+      "step": 6287
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.1380138397216797,
+      "learning_rate": 9.863278510952458e-06,
+      "loss": 0.7065,
+      "step": 6288
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.0260324478149414,
+      "learning_rate": 9.863217269112319e-06,
+      "loss": 0.7818,
+      "step": 6289
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.2397208213806152,
+      "learning_rate": 9.863156013749393e-06,
+      "loss": 0.7408,
+      "step": 6290
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.425929546356201,
+      "learning_rate": 9.863094744863853e-06,
+      "loss": 0.687,
+      "step": 6291
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.5405831336975098,
+      "learning_rate": 9.863033462455866e-06,
+      "loss": 0.4834,
+      "step": 6292
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 5.202689170837402,
+      "learning_rate": 9.862972166525605e-06,
+      "loss": 0.5099,
+      "step": 6293
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.320101022720337,
+      "learning_rate": 9.862910857073243e-06,
+      "loss": 0.5417,
+      "step": 6294
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.450934886932373,
+      "learning_rate": 9.862849534098945e-06,
+      "loss": 0.7951,
+      "step": 6295
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.0686748027801514,
+      "learning_rate": 9.862788197602883e-06,
+      "loss": 0.692,
+      "step": 6296
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.4508676528930664,
+      "learning_rate": 9.862726847585232e-06,
+      "loss": 0.443,
+      "step": 6297
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.713141441345215,
+      "learning_rate": 9.862665484046158e-06,
+      "loss": 0.731,
+      "step": 6298
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.6759049892425537,
+      "learning_rate": 9.862604106985832e-06,
+      "loss": 0.5453,
+      "step": 6299
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.102619171142578,
+      "learning_rate": 9.862542716404426e-06,
+      "loss": 0.7794,
+      "step": 6300
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.41750431060791,
+      "learning_rate": 9.86248131230211e-06,
+      "loss": 0.6635,
+      "step": 6301
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 5.124555587768555,
+      "learning_rate": 9.862419894679056e-06,
+      "loss": 0.831,
+      "step": 6302
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.254793882369995,
+      "learning_rate": 9.862358463535434e-06,
+      "loss": 0.7407,
+      "step": 6303
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.6154353618621826,
+      "learning_rate": 9.862297018871414e-06,
+      "loss": 0.7225,
+      "step": 6304
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.3689796924591064,
+      "learning_rate": 9.862235560687165e-06,
+      "loss": 0.7077,
+      "step": 6305
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.1295511722564697,
+      "learning_rate": 9.862174088982863e-06,
+      "loss": 0.4801,
+      "step": 6306
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.350677251815796,
+      "learning_rate": 9.862112603758677e-06,
+      "loss": 0.7011,
+      "step": 6307
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 5.098160743713379,
+      "learning_rate": 9.862051105014776e-06,
+      "loss": 0.6751,
+      "step": 6308
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.1253621578216553,
+      "learning_rate": 9.861989592751332e-06,
+      "loss": 0.4748,
+      "step": 6309
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 6.181213855743408,
+      "learning_rate": 9.861928066968516e-06,
+      "loss": 0.7593,
+      "step": 6310
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 5.030734539031982,
+      "learning_rate": 9.861866527666501e-06,
+      "loss": 0.4912,
+      "step": 6311
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.418199300765991,
+      "learning_rate": 9.861804974845456e-06,
+      "loss": 0.799,
+      "step": 6312
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.7451987266540527,
+      "learning_rate": 9.86174340850555e-06,
+      "loss": 0.6139,
+      "step": 6313
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 4.378519058227539,
+      "learning_rate": 9.86168182864696e-06,
+      "loss": 0.8988,
+      "step": 6314
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.5572043657302856,
+      "learning_rate": 9.861620235269852e-06,
+      "loss": 0.6621,
+      "step": 6315
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.112347364425659,
+      "learning_rate": 9.861558628374401e-06,
+      "loss": 0.4736,
+      "step": 6316
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.6509921550750732,
+      "learning_rate": 9.861497007960777e-06,
+      "loss": 0.6984,
+      "step": 6317
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.431792974472046,
+      "learning_rate": 9.861435374029149e-06,
+      "loss": 0.8043,
+      "step": 6318
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.587109327316284,
+      "learning_rate": 9.86137372657969e-06,
+      "loss": 0.8423,
+      "step": 6319
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.0146946907043457,
+      "learning_rate": 9.861312065612574e-06,
+      "loss": 0.5082,
+      "step": 6320
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.4333794116973877,
+      "learning_rate": 9.86125039112797e-06,
+      "loss": 0.7284,
+      "step": 6321
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.912436008453369,
+      "learning_rate": 9.861188703126048e-06,
+      "loss": 0.6635,
+      "step": 6322
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.7728917598724365,
+      "learning_rate": 9.861127001606982e-06,
+      "loss": 0.8308,
+      "step": 6323
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.0976316928863525,
+      "learning_rate": 9.861065286570943e-06,
+      "loss": 0.7839,
+      "step": 6324
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.2802631855010986,
+      "learning_rate": 9.861003558018103e-06,
+      "loss": 0.7013,
+      "step": 6325
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.247911214828491,
+      "learning_rate": 9.860941815948633e-06,
+      "loss": 0.7025,
+      "step": 6326
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.8992531299591064,
+      "learning_rate": 9.860880060362705e-06,
+      "loss": 0.7068,
+      "step": 6327
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.7486766576766968,
+      "learning_rate": 9.860818291260488e-06,
+      "loss": 0.5422,
+      "step": 6328
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 5.915833950042725,
+      "learning_rate": 9.86075650864216e-06,
+      "loss": 0.6692,
+      "step": 6329
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.501086950302124,
+      "learning_rate": 9.860694712507885e-06,
+      "loss": 0.5807,
+      "step": 6330
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.824805736541748,
+      "learning_rate": 9.860632902857842e-06,
+      "loss": 0.5936,
+      "step": 6331
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.7729697227478027,
+      "learning_rate": 9.860571079692198e-06,
+      "loss": 0.576,
+      "step": 6332
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.8483201265335083,
+      "learning_rate": 9.860509243011128e-06,
+      "loss": 0.765,
+      "step": 6333
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.8682713508605957,
+      "learning_rate": 9.860447392814802e-06,
+      "loss": 0.4867,
+      "step": 6334
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.251713991165161,
+      "learning_rate": 9.860385529103393e-06,
+      "loss": 0.786,
+      "step": 6335
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.436722993850708,
+      "learning_rate": 9.860323651877072e-06,
+      "loss": 0.8849,
+      "step": 6336
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.8456531763076782,
+      "learning_rate": 9.860261761136011e-06,
+      "loss": 0.484,
+      "step": 6337
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.883805274963379,
+      "learning_rate": 9.860199856880384e-06,
+      "loss": 0.7391,
+      "step": 6338
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.665539026260376,
+      "learning_rate": 9.860137939110362e-06,
+      "loss": 0.5335,
+      "step": 6339
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.5081446170806885,
+      "learning_rate": 9.860076007826115e-06,
+      "loss": 0.704,
+      "step": 6340
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.3789680004119873,
+      "learning_rate": 9.860014063027819e-06,
+      "loss": 0.6467,
+      "step": 6341
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.7780473232269287,
+      "learning_rate": 9.859952104715645e-06,
+      "loss": 0.7789,
+      "step": 6342
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.48848819732666,
+      "learning_rate": 9.859890132889764e-06,
+      "loss": 0.5283,
+      "step": 6343
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.095357894897461,
+      "learning_rate": 9.859828147550348e-06,
+      "loss": 0.4793,
+      "step": 6344
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.524714946746826,
+      "learning_rate": 9.859766148697571e-06,
+      "loss": 0.5327,
+      "step": 6345
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.9028325080871582,
+      "learning_rate": 9.859704136331605e-06,
+      "loss": 0.9481,
+      "step": 6346
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.707839250564575,
+      "learning_rate": 9.859642110452623e-06,
+      "loss": 0.792,
+      "step": 6347
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.7216700315475464,
+      "learning_rate": 9.859580071060797e-06,
+      "loss": 0.5906,
+      "step": 6348
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.253161907196045,
+      "learning_rate": 9.859518018156297e-06,
+      "loss": 0.7129,
+      "step": 6349
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.723923683166504,
+      "learning_rate": 9.8594559517393e-06,
+      "loss": 0.6902,
+      "step": 6350
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.690825343132019,
+      "learning_rate": 9.859393871809977e-06,
+      "loss": 0.7972,
+      "step": 6351
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.5978281497955322,
+      "learning_rate": 9.859331778368498e-06,
+      "loss": 0.6443,
+      "step": 6352
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.671810507774353,
+      "learning_rate": 9.859269671415039e-06,
+      "loss": 0.6171,
+      "step": 6353
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 8.3583984375,
+      "learning_rate": 9.859207550949771e-06,
+      "loss": 0.6343,
+      "step": 6354
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.802236557006836,
+      "learning_rate": 9.859145416972866e-06,
+      "loss": 0.7403,
+      "step": 6355
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 6.995109558105469,
+      "learning_rate": 9.8590832694845e-06,
+      "loss": 0.8946,
+      "step": 6356
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.137772560119629,
+      "learning_rate": 9.859021108484842e-06,
+      "loss": 0.6954,
+      "step": 6357
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.3298299312591553,
+      "learning_rate": 9.858958933974068e-06,
+      "loss": 0.5023,
+      "step": 6358
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.6992709636688232,
+      "learning_rate": 9.858896745952347e-06,
+      "loss": 0.8457,
+      "step": 6359
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.249117374420166,
+      "learning_rate": 9.858834544419857e-06,
+      "loss": 0.4474,
+      "step": 6360
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.520169734954834,
+      "learning_rate": 9.858772329376767e-06,
+      "loss": 0.5041,
+      "step": 6361
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.116837978363037,
+      "learning_rate": 9.85871010082325e-06,
+      "loss": 0.628,
+      "step": 6362
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.8780064582824707,
+      "learning_rate": 9.858647858759482e-06,
+      "loss": 0.7027,
+      "step": 6363
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.9680540561676025,
+      "learning_rate": 9.858585603185634e-06,
+      "loss": 0.749,
+      "step": 6364
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 4.248071670532227,
+      "learning_rate": 9.85852333410188e-06,
+      "loss": 0.8205,
+      "step": 6365
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.8356306552886963,
+      "learning_rate": 9.858461051508393e-06,
+      "loss": 0.9062,
+      "step": 6366
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.720778465270996,
+      "learning_rate": 9.858398755405344e-06,
+      "loss": 0.5403,
+      "step": 6367
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.4730217456817627,
+      "learning_rate": 9.858336445792908e-06,
+      "loss": 0.7371,
+      "step": 6368
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.8867502212524414,
+      "learning_rate": 9.858274122671261e-06,
+      "loss": 0.7135,
+      "step": 6369
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.355093002319336,
+      "learning_rate": 9.858211786040573e-06,
+      "loss": 0.6967,
+      "step": 6370
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.198974370956421,
+      "learning_rate": 9.858149435901017e-06,
+      "loss": 0.4922,
+      "step": 6371
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.9572560787200928,
+      "learning_rate": 9.858087072252767e-06,
+      "loss": 0.8347,
+      "step": 6372
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.269299030303955,
+      "learning_rate": 9.858024695095996e-06,
+      "loss": 0.79,
+      "step": 6373
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.283846139907837,
+      "learning_rate": 9.857962304430879e-06,
+      "loss": 0.6302,
+      "step": 6374
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.2829489707946777,
+      "learning_rate": 9.857899900257589e-06,
+      "loss": 0.6177,
+      "step": 6375
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.6503167152404785,
+      "learning_rate": 9.8578374825763e-06,
+      "loss": 0.6293,
+      "step": 6376
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.721311092376709,
+      "learning_rate": 9.857775051387184e-06,
+      "loss": 0.7133,
+      "step": 6377
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 4.121047496795654,
+      "learning_rate": 9.857712606690415e-06,
+      "loss": 0.8969,
+      "step": 6378
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.2070505619049072,
+      "learning_rate": 9.857650148486167e-06,
+      "loss": 0.7535,
+      "step": 6379
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 4.397424697875977,
+      "learning_rate": 9.857587676774613e-06,
+      "loss": 0.5999,
+      "step": 6380
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 5.991911888122559,
+      "learning_rate": 9.857525191555929e-06,
+      "loss": 0.6115,
+      "step": 6381
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.2611849308013916,
+      "learning_rate": 9.857462692830287e-06,
+      "loss": 0.7238,
+      "step": 6382
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.920992851257324,
+      "learning_rate": 9.85740018059786e-06,
+      "loss": 0.7605,
+      "step": 6383
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.939060926437378,
+      "learning_rate": 9.857337654858823e-06,
+      "loss": 0.7645,
+      "step": 6384
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.0857725143432617,
+      "learning_rate": 9.85727511561335e-06,
+      "loss": 0.5002,
+      "step": 6385
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 4.083180904388428,
+      "learning_rate": 9.857212562861614e-06,
+      "loss": 0.7354,
+      "step": 6386
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.7789013385772705,
+      "learning_rate": 9.857149996603787e-06,
+      "loss": 0.6673,
+      "step": 6387
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.601635217666626,
+      "learning_rate": 9.85708741684005e-06,
+      "loss": 0.6987,
+      "step": 6388
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.3322112560272217,
+      "learning_rate": 9.85702482357057e-06,
+      "loss": 0.7032,
+      "step": 6389
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.726170301437378,
+      "learning_rate": 9.856962216795523e-06,
+      "loss": 0.635,
+      "step": 6390
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.216111898422241,
+      "learning_rate": 9.856899596515084e-06,
+      "loss": 0.5098,
+      "step": 6391
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.21979022026062,
+      "learning_rate": 9.856836962729427e-06,
+      "loss": 0.4217,
+      "step": 6392
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.1579251289367676,
+      "learning_rate": 9.856774315438726e-06,
+      "loss": 0.8192,
+      "step": 6393
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.028367757797241,
+      "learning_rate": 9.856711654643154e-06,
+      "loss": 0.6882,
+      "step": 6394
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 6.555727005004883,
+      "learning_rate": 9.856648980342887e-06,
+      "loss": 0.5209,
+      "step": 6395
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.8434832096099854,
+      "learning_rate": 9.8565862925381e-06,
+      "loss": 0.5267,
+      "step": 6396
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.1113665103912354,
+      "learning_rate": 9.856523591228964e-06,
+      "loss": 0.6659,
+      "step": 6397
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 4.121848106384277,
+      "learning_rate": 9.856460876415655e-06,
+      "loss": 0.6238,
+      "step": 6398
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.1724226474761963,
+      "learning_rate": 9.856398148098348e-06,
+      "loss": 0.4379,
+      "step": 6399
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.8635447025299072,
+      "learning_rate": 9.856335406277217e-06,
+      "loss": 0.638,
+      "step": 6400
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.8815956115722656,
+      "learning_rate": 9.856272650952436e-06,
+      "loss": 0.7457,
+      "step": 6401
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 4.384608268737793,
+      "learning_rate": 9.85620988212418e-06,
+      "loss": 0.8551,
+      "step": 6402
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 17.506013870239258,
+      "learning_rate": 9.856147099792625e-06,
+      "loss": 0.5504,
+      "step": 6403
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.206085443496704,
+      "learning_rate": 9.856084303957942e-06,
+      "loss": 0.8547,
+      "step": 6404
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.390432834625244,
+      "learning_rate": 9.85602149462031e-06,
+      "loss": 0.6012,
+      "step": 6405
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.2531545162200928,
+      "learning_rate": 9.8559586717799e-06,
+      "loss": 0.6545,
+      "step": 6406
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.5380772352218628,
+      "learning_rate": 9.855895835436888e-06,
+      "loss": 0.4148,
+      "step": 6407
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.0656471252441406,
+      "learning_rate": 9.855832985591448e-06,
+      "loss": 0.6264,
+      "step": 6408
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.7947089672088623,
+      "learning_rate": 9.855770122243757e-06,
+      "loss": 0.8454,
+      "step": 6409
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.372342824935913,
+      "learning_rate": 9.855707245393987e-06,
+      "loss": 0.7642,
+      "step": 6410
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.4927213191986084,
+      "learning_rate": 9.855644355042315e-06,
+      "loss": 0.7596,
+      "step": 6411
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 4.16127347946167,
+      "learning_rate": 9.855581451188915e-06,
+      "loss": 0.633,
+      "step": 6412
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.3430824279785156,
+      "learning_rate": 9.855518533833962e-06,
+      "loss": 0.7063,
+      "step": 6413
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.0633420944213867,
+      "learning_rate": 9.85545560297763e-06,
+      "loss": 0.8602,
+      "step": 6414
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.1574888229370117,
+      "learning_rate": 9.855392658620096e-06,
+      "loss": 0.513,
+      "step": 6415
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.659722089767456,
+      "learning_rate": 9.855329700761533e-06,
+      "loss": 0.6415,
+      "step": 6416
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.2571942806243896,
+      "learning_rate": 9.855266729402118e-06,
+      "loss": 0.4804,
+      "step": 6417
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 4.061828136444092,
+      "learning_rate": 9.855203744542023e-06,
+      "loss": 0.9075,
+      "step": 6418
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.2319846153259277,
+      "learning_rate": 9.855140746181426e-06,
+      "loss": 0.6986,
+      "step": 6419
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.6430206298828125,
+      "learning_rate": 9.855077734320503e-06,
+      "loss": 0.6453,
+      "step": 6420
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.147684335708618,
+      "learning_rate": 9.855014708959428e-06,
+      "loss": 0.5295,
+      "step": 6421
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.6807353496551514,
+      "learning_rate": 9.854951670098373e-06,
+      "loss": 0.6858,
+      "step": 6422
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.9702211618423462,
+      "learning_rate": 9.854888617737517e-06,
+      "loss": 0.3956,
+      "step": 6423
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.8850677013397217,
+      "learning_rate": 9.854825551877035e-06,
+      "loss": 0.6405,
+      "step": 6424
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.105811357498169,
+      "learning_rate": 9.854762472517101e-06,
+      "loss": 0.4904,
+      "step": 6425
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.9188206195831299,
+      "learning_rate": 9.854699379657893e-06,
+      "loss": 0.3961,
+      "step": 6426
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.758521556854248,
+      "learning_rate": 9.854636273299582e-06,
+      "loss": 0.5424,
+      "step": 6427
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 5.242002964019775,
+      "learning_rate": 9.854573153442348e-06,
+      "loss": 0.8041,
+      "step": 6428
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.7337000370025635,
+      "learning_rate": 9.854510020086366e-06,
+      "loss": 0.4561,
+      "step": 6429
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.017549514770508,
+      "learning_rate": 9.854446873231806e-06,
+      "loss": 0.6905,
+      "step": 6430
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.5185463428497314,
+      "learning_rate": 9.854383712878851e-06,
+      "loss": 0.5865,
+      "step": 6431
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.59078311920166,
+      "learning_rate": 9.854320539027672e-06,
+      "loss": 0.4556,
+      "step": 6432
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.2151896953582764,
+      "learning_rate": 9.854257351678446e-06,
+      "loss": 0.6926,
+      "step": 6433
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.7103255987167358,
+      "learning_rate": 9.854194150831352e-06,
+      "loss": 0.4963,
+      "step": 6434
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 6.855143070220947,
+      "learning_rate": 9.854130936486558e-06,
+      "loss": 0.8884,
+      "step": 6435
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.954463481903076,
+      "learning_rate": 9.854067708644246e-06,
+      "loss": 0.8234,
+      "step": 6436
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.2852749824523926,
+      "learning_rate": 9.85400446730459e-06,
+      "loss": 0.5924,
+      "step": 6437
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.6397143602371216,
+      "learning_rate": 9.853941212467767e-06,
+      "loss": 0.5305,
+      "step": 6438
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.1311304569244385,
+      "learning_rate": 9.853877944133951e-06,
+      "loss": 0.4995,
+      "step": 6439
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.217071294784546,
+      "learning_rate": 9.853814662303319e-06,
+      "loss": 0.6586,
+      "step": 6440
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.8357627391815186,
+      "learning_rate": 9.853751366976045e-06,
+      "loss": 0.7594,
+      "step": 6441
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.8869566917419434,
+      "learning_rate": 9.853688058152309e-06,
+      "loss": 0.6952,
+      "step": 6442
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.1578001976013184,
+      "learning_rate": 9.853624735832284e-06,
+      "loss": 0.306,
+      "step": 6443
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.0115749835968018,
+      "learning_rate": 9.853561400016146e-06,
+      "loss": 0.1931,
+      "step": 6444
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.9157754182815552,
+      "learning_rate": 9.853498050704074e-06,
+      "loss": 0.8452,
+      "step": 6445
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 4.186073303222656,
+      "learning_rate": 9.85343468789624e-06,
+      "loss": 0.676,
+      "step": 6446
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.592141628265381,
+      "learning_rate": 9.853371311592823e-06,
+      "loss": 0.8159,
+      "step": 6447
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.739391803741455,
+      "learning_rate": 9.853307921793998e-06,
+      "loss": 0.8176,
+      "step": 6448
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.5026724338531494,
+      "learning_rate": 9.853244518499942e-06,
+      "loss": 0.9004,
+      "step": 6449
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.5048680305480957,
+      "learning_rate": 9.85318110171083e-06,
+      "loss": 0.7474,
+      "step": 6450
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 18.454097747802734,
+      "learning_rate": 9.853117671426841e-06,
+      "loss": 0.7238,
+      "step": 6451
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.015104293823242,
+      "learning_rate": 9.85305422764815e-06,
+      "loss": 0.8583,
+      "step": 6452
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.5954160690307617,
+      "learning_rate": 9.852990770374932e-06,
+      "loss": 0.3519,
+      "step": 6453
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.3215930461883545,
+      "learning_rate": 9.852927299607367e-06,
+      "loss": 0.9309,
+      "step": 6454
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.04293155670166,
+      "learning_rate": 9.852863815345627e-06,
+      "loss": 0.6667,
+      "step": 6455
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.140212297439575,
+      "learning_rate": 9.85280031758989e-06,
+      "loss": 0.4808,
+      "step": 6456
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.8011474609375,
+      "learning_rate": 9.852736806340336e-06,
+      "loss": 0.712,
+      "step": 6457
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.4598050117492676,
+      "learning_rate": 9.852673281597136e-06,
+      "loss": 0.8242,
+      "step": 6458
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 4.23762845993042,
+      "learning_rate": 9.85260974336047e-06,
+      "loss": 0.7172,
+      "step": 6459
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.2460591793060303,
+      "learning_rate": 9.852546191630515e-06,
+      "loss": 0.5717,
+      "step": 6460
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.8343333005905151,
+      "learning_rate": 9.852482626407446e-06,
+      "loss": 0.636,
+      "step": 6461
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.0903396606445312,
+      "learning_rate": 9.852419047691442e-06,
+      "loss": 0.6696,
+      "step": 6462
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.1976640224456787,
+      "learning_rate": 9.852355455482676e-06,
+      "loss": 0.4488,
+      "step": 6463
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.1578879356384277,
+      "learning_rate": 9.85229184978133e-06,
+      "loss": 0.7015,
+      "step": 6464
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.1700775623321533,
+      "learning_rate": 9.852228230587578e-06,
+      "loss": 0.4924,
+      "step": 6465
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.4525868892669678,
+      "learning_rate": 9.852164597901594e-06,
+      "loss": 0.7226,
+      "step": 6466
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.950935125350952,
+      "learning_rate": 9.85210095172356e-06,
+      "loss": 0.5951,
+      "step": 6467
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.1436057090759277,
+      "learning_rate": 9.852037292053651e-06,
+      "loss": 0.5306,
+      "step": 6468
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 6.412509918212891,
+      "learning_rate": 9.851973618892044e-06,
+      "loss": 0.735,
+      "step": 6469
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.755714178085327,
+      "learning_rate": 9.851909932238915e-06,
+      "loss": 0.4948,
+      "step": 6470
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.9643957614898682,
+      "learning_rate": 9.851846232094443e-06,
+      "loss": 0.5784,
+      "step": 6471
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.736665725708008,
+      "learning_rate": 9.851782518458803e-06,
+      "loss": 0.9524,
+      "step": 6472
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.702526807785034,
+      "learning_rate": 9.851718791332173e-06,
+      "loss": 0.641,
+      "step": 6473
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.295271158218384,
+      "learning_rate": 9.851655050714732e-06,
+      "loss": 0.5559,
+      "step": 6474
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 4.747699737548828,
+      "learning_rate": 9.851591296606655e-06,
+      "loss": 0.5687,
+      "step": 6475
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.7594659328460693,
+      "learning_rate": 9.85152752900812e-06,
+      "loss": 0.7291,
+      "step": 6476
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.503340005874634,
+      "learning_rate": 9.851463747919306e-06,
+      "loss": 0.7064,
+      "step": 6477
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.8158164024353027,
+      "learning_rate": 9.851399953340386e-06,
+      "loss": 0.6985,
+      "step": 6478
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 4.571846008300781,
+      "learning_rate": 9.851336145271541e-06,
+      "loss": 0.7852,
+      "step": 6479
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.108489751815796,
+      "learning_rate": 9.851272323712948e-06,
+      "loss": 0.8032,
+      "step": 6480
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.5742502212524414,
+      "learning_rate": 9.851208488664785e-06,
+      "loss": 0.7514,
+      "step": 6481
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 4.510971546173096,
+      "learning_rate": 9.851144640127228e-06,
+      "loss": 0.7139,
+      "step": 6482
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.3526010513305664,
+      "learning_rate": 9.851080778100456e-06,
+      "loss": 0.6758,
+      "step": 6483
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.2864444255828857,
+      "learning_rate": 9.851016902584642e-06,
+      "loss": 0.8028,
+      "step": 6484
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.3623385429382324,
+      "learning_rate": 9.850953013579968e-06,
+      "loss": 0.6587,
+      "step": 6485
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.2597737312316895,
+      "learning_rate": 9.850889111086615e-06,
+      "loss": 0.8108,
+      "step": 6486
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.71803617477417,
+      "learning_rate": 9.850825195104753e-06,
+      "loss": 0.7312,
+      "step": 6487
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.555290699005127,
+      "learning_rate": 9.850761265634564e-06,
+      "loss": 0.758,
+      "step": 6488
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.6676790714263916,
+      "learning_rate": 9.850697322676226e-06,
+      "loss": 0.6059,
+      "step": 6489
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.692635416984558,
+      "learning_rate": 9.850633366229913e-06,
+      "loss": 0.3555,
+      "step": 6490
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.0002291202545166,
+      "learning_rate": 9.850569396295808e-06,
+      "loss": 0.7633,
+      "step": 6491
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.2053773403167725,
+      "learning_rate": 9.850505412874088e-06,
+      "loss": 0.3078,
+      "step": 6492
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.052182197570801,
+      "learning_rate": 9.850441415964926e-06,
+      "loss": 0.5137,
+      "step": 6493
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.8644553422927856,
+      "learning_rate": 9.850377405568506e-06,
+      "loss": 0.5382,
+      "step": 6494
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.1439945697784424,
+      "learning_rate": 9.850313381685005e-06,
+      "loss": 0.5641,
+      "step": 6495
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.7959656715393066,
+      "learning_rate": 9.850249344314595e-06,
+      "loss": 0.797,
+      "step": 6496
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.8963217735290527,
+      "learning_rate": 9.850185293457461e-06,
+      "loss": 0.64,
+      "step": 6497
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.5064659118652344,
+      "learning_rate": 9.850121229113777e-06,
+      "loss": 0.8869,
+      "step": 6498
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.956324815750122,
+      "learning_rate": 9.850057151283725e-06,
+      "loss": 0.5228,
+      "step": 6499
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.581627607345581,
+      "learning_rate": 9.84999305996748e-06,
+      "loss": 0.8187,
+      "step": 6500
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 4.957623481750488,
+      "learning_rate": 9.849928955165221e-06,
+      "loss": 0.9558,
+      "step": 6501
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.2182748317718506,
+      "learning_rate": 9.849864836877127e-06,
+      "loss": 0.5498,
+      "step": 6502
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.552044153213501,
+      "learning_rate": 9.849800705103375e-06,
+      "loss": 0.6809,
+      "step": 6503
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.596264600753784,
+      "learning_rate": 9.849736559844145e-06,
+      "loss": 0.9743,
+      "step": 6504
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.6548402309417725,
+      "learning_rate": 9.849672401099613e-06,
+      "loss": 0.3423,
+      "step": 6505
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 6.291617393493652,
+      "learning_rate": 9.84960822886996e-06,
+      "loss": 0.6693,
+      "step": 6506
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.5043840408325195,
+      "learning_rate": 9.849544043155361e-06,
+      "loss": 0.4499,
+      "step": 6507
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 7.459723472595215,
+      "learning_rate": 9.849479843955999e-06,
+      "loss": 0.6272,
+      "step": 6508
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.6512303352355957,
+      "learning_rate": 9.84941563127205e-06,
+      "loss": 0.8361,
+      "step": 6509
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.0370354652404785,
+      "learning_rate": 9.849351405103692e-06,
+      "loss": 0.7715,
+      "step": 6510
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 4.735937118530273,
+      "learning_rate": 9.849287165451105e-06,
+      "loss": 0.8425,
+      "step": 6511
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.4681239128112793,
+      "learning_rate": 9.849222912314467e-06,
+      "loss": 0.613,
+      "step": 6512
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 5.6236653327941895,
+      "learning_rate": 9.849158645693955e-06,
+      "loss": 0.7709,
+      "step": 6513
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.4796142578125,
+      "learning_rate": 9.849094365589751e-06,
+      "loss": 0.5977,
+      "step": 6514
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.664750337600708,
+      "learning_rate": 9.849030072002033e-06,
+      "loss": 0.5524,
+      "step": 6515
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.5589962005615234,
+      "learning_rate": 9.848965764930975e-06,
+      "loss": 0.6126,
+      "step": 6516
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.408747434616089,
+      "learning_rate": 9.848901444376763e-06,
+      "loss": 0.7321,
+      "step": 6517
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 4.709187984466553,
+      "learning_rate": 9.848837110339573e-06,
+      "loss": 0.6446,
+      "step": 6518
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.5152101516723633,
+      "learning_rate": 9.848772762819581e-06,
+      "loss": 0.4638,
+      "step": 6519
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.147573232650757,
+      "learning_rate": 9.84870840181697e-06,
+      "loss": 0.9407,
+      "step": 6520
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.3317456245422363,
+      "learning_rate": 9.848644027331918e-06,
+      "loss": 0.658,
+      "step": 6521
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.9922981262207031,
+      "learning_rate": 9.848579639364602e-06,
+      "loss": 0.5712,
+      "step": 6522
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.7343854904174805,
+      "learning_rate": 9.848515237915202e-06,
+      "loss": 0.7181,
+      "step": 6523
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.8076226711273193,
+      "learning_rate": 9.848450822983897e-06,
+      "loss": 0.7731,
+      "step": 6524
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.2297909259796143,
+      "learning_rate": 9.848386394570868e-06,
+      "loss": 0.8509,
+      "step": 6525
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.541318655014038,
+      "learning_rate": 9.848321952676292e-06,
+      "loss": 0.5454,
+      "step": 6526
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.225287914276123,
+      "learning_rate": 9.848257497300349e-06,
+      "loss": 0.9024,
+      "step": 6527
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.348592519760132,
+      "learning_rate": 9.848193028443218e-06,
+      "loss": 0.839,
+      "step": 6528
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.705484390258789,
+      "learning_rate": 9.848128546105078e-06,
+      "loss": 0.7434,
+      "step": 6529
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 4.942009925842285,
+      "learning_rate": 9.84806405028611e-06,
+      "loss": 0.5287,
+      "step": 6530
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.5716065168380737,
+      "learning_rate": 9.847999540986491e-06,
+      "loss": 0.4787,
+      "step": 6531
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.405372381210327,
+      "learning_rate": 9.8479350182064e-06,
+      "loss": 0.3408,
+      "step": 6532
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 4.574134349822998,
+      "learning_rate": 9.84787048194602e-06,
+      "loss": 0.6005,
+      "step": 6533
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.697723627090454,
+      "learning_rate": 9.847805932205528e-06,
+      "loss": 0.7033,
+      "step": 6534
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.8319036960601807,
+      "learning_rate": 9.847741368985102e-06,
+      "loss": 1.0374,
+      "step": 6535
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.4493651390075684,
+      "learning_rate": 9.847676792284925e-06,
+      "loss": 0.7048,
+      "step": 6536
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.5941476821899414,
+      "learning_rate": 9.847612202105175e-06,
+      "loss": 0.7654,
+      "step": 6537
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.210841655731201,
+      "learning_rate": 9.84754759844603e-06,
+      "loss": 0.6628,
+      "step": 6538
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.5468318462371826,
+      "learning_rate": 9.847482981307672e-06,
+      "loss": 0.5041,
+      "step": 6539
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.691927194595337,
+      "learning_rate": 9.847418350690279e-06,
+      "loss": 0.5277,
+      "step": 6540
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.9812718629837036,
+      "learning_rate": 9.847353706594032e-06,
+      "loss": 0.5053,
+      "step": 6541
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.561436653137207,
+      "learning_rate": 9.847289049019111e-06,
+      "loss": 0.537,
+      "step": 6542
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 5.486660480499268,
+      "learning_rate": 9.847224377965695e-06,
+      "loss": 1.0037,
+      "step": 6543
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.347918748855591,
+      "learning_rate": 9.84715969343396e-06,
+      "loss": 0.8031,
+      "step": 6544
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.7159554958343506,
+      "learning_rate": 9.847094995424093e-06,
+      "loss": 0.591,
+      "step": 6545
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.19384765625,
+      "learning_rate": 9.847030283936272e-06,
+      "loss": 0.7032,
+      "step": 6546
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.3813936710357666,
+      "learning_rate": 9.846965558970674e-06,
+      "loss": 0.6963,
+      "step": 6547
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.491367816925049,
+      "learning_rate": 9.84690082052748e-06,
+      "loss": 0.6494,
+      "step": 6548
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.8371529579162598,
+      "learning_rate": 9.84683606860687e-06,
+      "loss": 0.6521,
+      "step": 6549
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.232607841491699,
+      "learning_rate": 9.846771303209026e-06,
+      "loss": 0.6724,
+      "step": 6550
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.227482318878174,
+      "learning_rate": 9.846706524334126e-06,
+      "loss": 0.5715,
+      "step": 6551
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.5610966682434082,
+      "learning_rate": 9.84664173198235e-06,
+      "loss": 0.59,
+      "step": 6552
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.274533987045288,
+      "learning_rate": 9.846576926153879e-06,
+      "loss": 0.628,
+      "step": 6553
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.0137033462524414,
+      "learning_rate": 9.846512106848894e-06,
+      "loss": 0.7269,
+      "step": 6554
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.9689992666244507,
+      "learning_rate": 9.846447274067574e-06,
+      "loss": 0.8798,
+      "step": 6555
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 4.531224250793457,
+      "learning_rate": 9.8463824278101e-06,
+      "loss": 0.9479,
+      "step": 6556
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.298461675643921,
+      "learning_rate": 9.846317568076653e-06,
+      "loss": 0.8203,
+      "step": 6557
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.9425851106643677,
+      "learning_rate": 9.846252694867411e-06,
+      "loss": 0.7608,
+      "step": 6558
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.7371480464935303,
+      "learning_rate": 9.846187808182556e-06,
+      "loss": 0.8547,
+      "step": 6559
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.8226354122161865,
+      "learning_rate": 9.846122908022268e-06,
+      "loss": 0.7134,
+      "step": 6560
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.425696849822998,
+      "learning_rate": 9.846057994386727e-06,
+      "loss": 0.7193,
+      "step": 6561
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.0382564067840576,
+      "learning_rate": 9.845993067276115e-06,
+      "loss": 0.7267,
+      "step": 6562
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 4.1115007400512695,
+      "learning_rate": 9.845928126690611e-06,
+      "loss": 0.8518,
+      "step": 6563
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.1108713150024414,
+      "learning_rate": 9.845863172630397e-06,
+      "loss": 0.8157,
+      "step": 6564
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.5779331922531128,
+      "learning_rate": 9.845798205095651e-06,
+      "loss": 0.6834,
+      "step": 6565
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.010535717010498,
+      "learning_rate": 9.845733224086557e-06,
+      "loss": 0.7662,
+      "step": 6566
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 4.414896011352539,
+      "learning_rate": 9.845668229603294e-06,
+      "loss": 0.6561,
+      "step": 6567
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.454423666000366,
+      "learning_rate": 9.845603221646043e-06,
+      "loss": 0.7291,
+      "step": 6568
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.34640371799469,
+      "learning_rate": 9.845538200214985e-06,
+      "loss": 0.5861,
+      "step": 6569
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.5045775175094604,
+      "learning_rate": 9.8454731653103e-06,
+      "loss": 0.4232,
+      "step": 6570
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.1353049278259277,
+      "learning_rate": 9.845408116932169e-06,
+      "loss": 0.7574,
+      "step": 6571
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.7905538082122803,
+      "learning_rate": 9.845343055080773e-06,
+      "loss": 0.7167,
+      "step": 6572
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.0874717235565186,
+      "learning_rate": 9.845277979756294e-06,
+      "loss": 0.6239,
+      "step": 6573
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.3202881813049316,
+      "learning_rate": 9.84521289095891e-06,
+      "loss": 0.7336,
+      "step": 6574
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 4.180565357208252,
+      "learning_rate": 9.845147788688806e-06,
+      "loss": 0.5997,
+      "step": 6575
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.4819416999816895,
+      "learning_rate": 9.84508267294616e-06,
+      "loss": 0.6192,
+      "step": 6576
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.6687531471252441,
+      "learning_rate": 9.845017543731155e-06,
+      "loss": 0.8774,
+      "step": 6577
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.7345194816589355,
+      "learning_rate": 9.844952401043971e-06,
+      "loss": 0.8469,
+      "step": 6578
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 4.967562198638916,
+      "learning_rate": 9.844887244884788e-06,
+      "loss": 0.739,
+      "step": 6579
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.1440837383270264,
+      "learning_rate": 9.84482207525379e-06,
+      "loss": 0.5757,
+      "step": 6580
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.617397665977478,
+      "learning_rate": 9.844756892151157e-06,
+      "loss": 0.5129,
+      "step": 6581
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.225231647491455,
+      "learning_rate": 9.844691695577069e-06,
+      "loss": 0.5771,
+      "step": 6582
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.8871773481369019,
+      "learning_rate": 9.844626485531708e-06,
+      "loss": 0.5515,
+      "step": 6583
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.0396323204040527,
+      "learning_rate": 9.844561262015257e-06,
+      "loss": 0.6953,
+      "step": 6584
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.729451060295105,
+      "learning_rate": 9.844496025027893e-06,
+      "loss": 0.6541,
+      "step": 6585
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.369542121887207,
+      "learning_rate": 9.844430774569804e-06,
+      "loss": 0.7025,
+      "step": 6586
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.9883414506912231,
+      "learning_rate": 9.844365510641167e-06,
+      "loss": 0.601,
+      "step": 6587
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.278422236442566,
+      "learning_rate": 9.844300233242164e-06,
+      "loss": 0.7298,
+      "step": 6588
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.369093418121338,
+      "learning_rate": 9.844234942372976e-06,
+      "loss": 0.6855,
+      "step": 6589
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.9131624698638916,
+      "learning_rate": 9.844169638033785e-06,
+      "loss": 0.7643,
+      "step": 6590
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.067636489868164,
+      "learning_rate": 9.844104320224774e-06,
+      "loss": 0.7142,
+      "step": 6591
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.6831297874450684,
+      "learning_rate": 9.844038988946123e-06,
+      "loss": 0.5071,
+      "step": 6592
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.1321065425872803,
+      "learning_rate": 9.843973644198013e-06,
+      "loss": 0.8193,
+      "step": 6593
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.5922341346740723,
+      "learning_rate": 9.843908285980628e-06,
+      "loss": 0.8551,
+      "step": 6594
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.013812780380249,
+      "learning_rate": 9.84384291429415e-06,
+      "loss": 0.6759,
+      "step": 6595
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.1630749702453613,
+      "learning_rate": 9.843777529138757e-06,
+      "loss": 0.6124,
+      "step": 6596
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 4.798533916473389,
+      "learning_rate": 9.843712130514634e-06,
+      "loss": 0.517,
+      "step": 6597
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.3660595417022705,
+      "learning_rate": 9.843646718421963e-06,
+      "loss": 0.5504,
+      "step": 6598
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.4456117153167725,
+      "learning_rate": 9.843581292860925e-06,
+      "loss": 0.5594,
+      "step": 6599
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.295067310333252,
+      "learning_rate": 9.8435158538317e-06,
+      "loss": 0.7283,
+      "step": 6600
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.793039560317993,
+      "learning_rate": 9.843450401334473e-06,
+      "loss": 0.9017,
+      "step": 6601
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 4.7627482414245605,
+      "learning_rate": 9.843384935369425e-06,
+      "loss": 0.8682,
+      "step": 6602
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.1398086547851562,
+      "learning_rate": 9.843319455936737e-06,
+      "loss": 0.6102,
+      "step": 6603
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.4874074459075928,
+      "learning_rate": 9.843253963036594e-06,
+      "loss": 0.6813,
+      "step": 6604
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.854961633682251,
+      "learning_rate": 9.843188456669175e-06,
+      "loss": 0.7843,
+      "step": 6605
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 4.805079460144043,
+      "learning_rate": 9.843122936834661e-06,
+      "loss": 0.7379,
+      "step": 6606
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.9142916202545166,
+      "learning_rate": 9.84305740353324e-06,
+      "loss": 0.9161,
+      "step": 6607
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.4466512203216553,
+      "learning_rate": 9.842991856765089e-06,
+      "loss": 0.8301,
+      "step": 6608
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.429001808166504,
+      "learning_rate": 9.84292629653039e-06,
+      "loss": 1.0743,
+      "step": 6609
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.6268529891967773,
+      "learning_rate": 9.84286072282933e-06,
+      "loss": 0.6909,
+      "step": 6610
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.8482575416564941,
+      "learning_rate": 9.842795135662087e-06,
+      "loss": 0.7684,
+      "step": 6611
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.6631295680999756,
+      "learning_rate": 9.842729535028845e-06,
+      "loss": 0.5812,
+      "step": 6612
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.069436550140381,
+      "learning_rate": 9.842663920929787e-06,
+      "loss": 0.7042,
+      "step": 6613
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 5.023177146911621,
+      "learning_rate": 9.842598293365095e-06,
+      "loss": 0.5105,
+      "step": 6614
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.7755112648010254,
+      "learning_rate": 9.84253265233495e-06,
+      "loss": 1.0415,
+      "step": 6615
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.9776408672332764,
+      "learning_rate": 9.842466997839537e-06,
+      "loss": 0.535,
+      "step": 6616
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.420137405395508,
+      "learning_rate": 9.842401329879036e-06,
+      "loss": 0.5236,
+      "step": 6617
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.2963571548461914,
+      "learning_rate": 9.842335648453631e-06,
+      "loss": 0.4652,
+      "step": 6618
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.0701091289520264,
+      "learning_rate": 9.842269953563506e-06,
+      "loss": 0.767,
+      "step": 6619
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.6331496238708496,
+      "learning_rate": 9.842204245208841e-06,
+      "loss": 0.7474,
+      "step": 6620
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 5.291462421417236,
+      "learning_rate": 9.842138523389821e-06,
+      "loss": 0.4588,
+      "step": 6621
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.2491395473480225,
+      "learning_rate": 9.842072788106627e-06,
+      "loss": 0.8039,
+      "step": 6622
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.2287242412567139,
+      "learning_rate": 9.842007039359443e-06,
+      "loss": 0.507,
+      "step": 6623
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.7783958911895752,
+      "learning_rate": 9.84194127714845e-06,
+      "loss": 0.4502,
+      "step": 6624
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.365858316421509,
+      "learning_rate": 9.841875501473832e-06,
+      "loss": 0.7923,
+      "step": 6625
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.3422367572784424,
+      "learning_rate": 9.841809712335773e-06,
+      "loss": 0.5808,
+      "step": 6626
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.4776829481124878,
+      "learning_rate": 9.841743909734456e-06,
+      "loss": 0.6384,
+      "step": 6627
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.743186116218567,
+      "learning_rate": 9.841678093670061e-06,
+      "loss": 0.5919,
+      "step": 6628
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.8302292823791504,
+      "learning_rate": 9.841612264142773e-06,
+      "loss": 0.41,
+      "step": 6629
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.1807658672332764,
+      "learning_rate": 9.841546421152777e-06,
+      "loss": 0.6794,
+      "step": 6630
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.073998212814331,
+      "learning_rate": 9.841480564700251e-06,
+      "loss": 0.435,
+      "step": 6631
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.46856951713562,
+      "learning_rate": 9.841414694785384e-06,
+      "loss": 0.4119,
+      "step": 6632
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.8971478939056396,
+      "learning_rate": 9.841348811408356e-06,
+      "loss": 0.6812,
+      "step": 6633
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.4868314266204834,
+      "learning_rate": 9.84128291456935e-06,
+      "loss": 0.4573,
+      "step": 6634
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.2518863677978516,
+      "learning_rate": 9.841217004268548e-06,
+      "loss": 0.871,
+      "step": 6635
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.1344621181488037,
+      "learning_rate": 9.841151080506137e-06,
+      "loss": 0.5108,
+      "step": 6636
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.4223484992980957,
+      "learning_rate": 9.841085143282298e-06,
+      "loss": 0.5218,
+      "step": 6637
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.91300630569458,
+      "learning_rate": 9.841019192597216e-06,
+      "loss": 0.5444,
+      "step": 6638
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.272470712661743,
+      "learning_rate": 9.840953228451072e-06,
+      "loss": 0.552,
+      "step": 6639
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.148958921432495,
+      "learning_rate": 9.84088725084405e-06,
+      "loss": 0.6041,
+      "step": 6640
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.155067205429077,
+      "learning_rate": 9.840821259776335e-06,
+      "loss": 0.5726,
+      "step": 6641
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.6630301475524902,
+      "learning_rate": 9.84075525524811e-06,
+      "loss": 0.4757,
+      "step": 6642
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.8021559715270996,
+      "learning_rate": 9.840689237259557e-06,
+      "loss": 0.6639,
+      "step": 6643
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.6579551696777344,
+      "learning_rate": 9.84062320581086e-06,
+      "loss": 0.5712,
+      "step": 6644
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.4118074178695679,
+      "learning_rate": 9.840557160902205e-06,
+      "loss": 0.5447,
+      "step": 6645
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.7971444129943848,
+      "learning_rate": 9.84049110253377e-06,
+      "loss": 0.7947,
+      "step": 6646
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.8617818355560303,
+      "learning_rate": 9.840425030705747e-06,
+      "loss": 0.9122,
+      "step": 6647
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.96147084236145,
+      "learning_rate": 9.840358945418314e-06,
+      "loss": 0.5997,
+      "step": 6648
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.8809800148010254,
+      "learning_rate": 9.840292846671655e-06,
+      "loss": 0.6945,
+      "step": 6649
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.447096109390259,
+      "learning_rate": 9.840226734465955e-06,
+      "loss": 0.7257,
+      "step": 6650
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.382270336151123,
+      "learning_rate": 9.840160608801399e-06,
+      "loss": 0.9607,
+      "step": 6651
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.6068116426467896,
+      "learning_rate": 9.840094469678168e-06,
+      "loss": 0.5461,
+      "step": 6652
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.6038230657577515,
+      "learning_rate": 9.840028317096449e-06,
+      "loss": 0.4981,
+      "step": 6653
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.8665261268615723,
+      "learning_rate": 9.839962151056422e-06,
+      "loss": 0.6389,
+      "step": 6654
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.959960699081421,
+      "learning_rate": 9.839895971558275e-06,
+      "loss": 0.8355,
+      "step": 6655
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.524541139602661,
+      "learning_rate": 9.83982977860219e-06,
+      "loss": 0.6253,
+      "step": 6656
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 6.266164779663086,
+      "learning_rate": 9.839763572188351e-06,
+      "loss": 0.8385,
+      "step": 6657
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.844533920288086,
+      "learning_rate": 9.839697352316944e-06,
+      "loss": 0.7498,
+      "step": 6658
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.216891050338745,
+      "learning_rate": 9.839631118988149e-06,
+      "loss": 0.4318,
+      "step": 6659
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.6398632526397705,
+      "learning_rate": 9.839564872202155e-06,
+      "loss": 0.5851,
+      "step": 6660
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.5248067378997803,
+      "learning_rate": 9.839498611959143e-06,
+      "loss": 0.5308,
+      "step": 6661
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.823120594024658,
+      "learning_rate": 9.8394323382593e-06,
+      "loss": 0.7973,
+      "step": 6662
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.2509820461273193,
+      "learning_rate": 9.839366051102807e-06,
+      "loss": 0.8214,
+      "step": 6663
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.0811169147491455,
+      "learning_rate": 9.83929975048985e-06,
+      "loss": 0.5157,
+      "step": 6664
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.617011070251465,
+      "learning_rate": 9.839233436420613e-06,
+      "loss": 0.6268,
+      "step": 6665
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.0596702098846436,
+      "learning_rate": 9.839167108895281e-06,
+      "loss": 0.6622,
+      "step": 6666
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.121561050415039,
+      "learning_rate": 9.839100767914038e-06,
+      "loss": 0.632,
+      "step": 6667
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.046576738357544,
+      "learning_rate": 9.839034413477068e-06,
+      "loss": 0.8915,
+      "step": 6668
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.8196709156036377,
+      "learning_rate": 9.838968045584558e-06,
+      "loss": 0.5971,
+      "step": 6669
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.5882437229156494,
+      "learning_rate": 9.838901664236689e-06,
+      "loss": 0.5208,
+      "step": 6670
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 5.108281135559082,
+      "learning_rate": 9.838835269433646e-06,
+      "loss": 0.7115,
+      "step": 6671
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.9125055074691772,
+      "learning_rate": 9.838768861175616e-06,
+      "loss": 0.526,
+      "step": 6672
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.691431760787964,
+      "learning_rate": 9.838702439462784e-06,
+      "loss": 0.7637,
+      "step": 6673
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 5.254922389984131,
+      "learning_rate": 9.838636004295332e-06,
+      "loss": 0.8235,
+      "step": 6674
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.7331695556640625,
+      "learning_rate": 9.838569555673445e-06,
+      "loss": 0.6386,
+      "step": 6675
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.8825018405914307,
+      "learning_rate": 9.838503093597309e-06,
+      "loss": 0.6333,
+      "step": 6676
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.3341493606567383,
+      "learning_rate": 9.838436618067108e-06,
+      "loss": 0.8099,
+      "step": 6677
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.153031587600708,
+      "learning_rate": 9.83837012908303e-06,
+      "loss": 0.853,
+      "step": 6678
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.3130242824554443,
+      "learning_rate": 9.838303626645254e-06,
+      "loss": 0.7326,
+      "step": 6679
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.1852059364318848,
+      "learning_rate": 9.83823711075397e-06,
+      "loss": 0.7182,
+      "step": 6680
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.1470577716827393,
+      "learning_rate": 9.838170581409361e-06,
+      "loss": 0.7551,
+      "step": 6681
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.0614750385284424,
+      "learning_rate": 9.838104038611611e-06,
+      "loss": 0.7204,
+      "step": 6682
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.187695264816284,
+      "learning_rate": 9.838037482360905e-06,
+      "loss": 0.7944,
+      "step": 6683
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 5.179729461669922,
+      "learning_rate": 9.83797091265743e-06,
+      "loss": 0.6108,
+      "step": 6684
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.4309260845184326,
+      "learning_rate": 9.837904329501372e-06,
+      "loss": 0.6289,
+      "step": 6685
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.2296135425567627,
+      "learning_rate": 9.837837732892913e-06,
+      "loss": 0.5991,
+      "step": 6686
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.038463592529297,
+      "learning_rate": 9.83777112283224e-06,
+      "loss": 0.7388,
+      "step": 6687
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.8504436016082764,
+      "learning_rate": 9.837704499319537e-06,
+      "loss": 0.7088,
+      "step": 6688
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.6244277954101562,
+      "learning_rate": 9.837637862354992e-06,
+      "loss": 0.9664,
+      "step": 6689
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.838383913040161,
+      "learning_rate": 9.837571211938786e-06,
+      "loss": 0.8253,
+      "step": 6690
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.932672381401062,
+      "learning_rate": 9.837504548071108e-06,
+      "loss": 0.5998,
+      "step": 6691
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.067300796508789,
+      "learning_rate": 9.837437870752141e-06,
+      "loss": 0.7967,
+      "step": 6692
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.067650318145752,
+      "learning_rate": 9.837371179982071e-06,
+      "loss": 0.5276,
+      "step": 6693
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.360990047454834,
+      "learning_rate": 9.837304475761085e-06,
+      "loss": 0.9295,
+      "step": 6694
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.8006207942962646,
+      "learning_rate": 9.83723775808937e-06,
+      "loss": 0.6825,
+      "step": 6695
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.2979817390441895,
+      "learning_rate": 9.837171026967103e-06,
+      "loss": 0.6504,
+      "step": 6696
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.3916432857513428,
+      "learning_rate": 9.83710428239448e-06,
+      "loss": 0.9817,
+      "step": 6697
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 4.130778789520264,
+      "learning_rate": 9.837037524371679e-06,
+      "loss": 0.7708,
+      "step": 6698
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.711703062057495,
+      "learning_rate": 9.836970752898891e-06,
+      "loss": 0.6954,
+      "step": 6699
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.1121363639831543,
+      "learning_rate": 9.8369039679763e-06,
+      "loss": 0.8777,
+      "step": 6700
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.453695774078369,
+      "learning_rate": 9.836837169604087e-06,
+      "loss": 0.5521,
+      "step": 6701
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.6767914295196533,
+      "learning_rate": 9.836770357782446e-06,
+      "loss": 0.54,
+      "step": 6702
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.8510372638702393,
+      "learning_rate": 9.836703532511555e-06,
+      "loss": 0.3782,
+      "step": 6703
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.2430410385131836,
+      "learning_rate": 9.836636693791606e-06,
+      "loss": 0.6402,
+      "step": 6704
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.8295676708221436,
+      "learning_rate": 9.83656984162278e-06,
+      "loss": 0.7089,
+      "step": 6705
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.17384934425354,
+      "learning_rate": 9.836502976005267e-06,
+      "loss": 0.505,
+      "step": 6706
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.069232940673828,
+      "learning_rate": 9.83643609693925e-06,
+      "loss": 0.5241,
+      "step": 6707
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.193343162536621,
+      "learning_rate": 9.836369204424917e-06,
+      "loss": 0.6726,
+      "step": 6708
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.1578121185302734,
+      "learning_rate": 9.836302298462452e-06,
+      "loss": 0.5114,
+      "step": 6709
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.1748149394989014,
+      "learning_rate": 9.836235379052043e-06,
+      "loss": 1.0047,
+      "step": 6710
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.0288548469543457,
+      "learning_rate": 9.836168446193875e-06,
+      "loss": 0.5409,
+      "step": 6711
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.7616932392120361,
+      "learning_rate": 9.836101499888134e-06,
+      "loss": 0.7012,
+      "step": 6712
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.3503942489624023,
+      "learning_rate": 9.836034540135006e-06,
+      "loss": 0.5148,
+      "step": 6713
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.5016064643859863,
+      "learning_rate": 9.835967566934676e-06,
+      "loss": 0.6032,
+      "step": 6714
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.7704781293869019,
+      "learning_rate": 9.835900580287334e-06,
+      "loss": 0.4429,
+      "step": 6715
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.5060250759124756,
+      "learning_rate": 9.835833580193164e-06,
+      "loss": 0.872,
+      "step": 6716
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.9433205127716064,
+      "learning_rate": 9.835766566652352e-06,
+      "loss": 0.7413,
+      "step": 6717
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.734121561050415,
+      "learning_rate": 9.835699539665084e-06,
+      "loss": 0.5994,
+      "step": 6718
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.1173815727233887,
+      "learning_rate": 9.835632499231548e-06,
+      "loss": 0.5266,
+      "step": 6719
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.265712261199951,
+      "learning_rate": 9.83556544535193e-06,
+      "loss": 0.6598,
+      "step": 6720
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 5.459347248077393,
+      "learning_rate": 9.835498378026414e-06,
+      "loss": 0.3744,
+      "step": 6721
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.9332456588745117,
+      "learning_rate": 9.835431297255188e-06,
+      "loss": 0.5865,
+      "step": 6722
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.4537148475646973,
+      "learning_rate": 9.835364203038441e-06,
+      "loss": 0.6796,
+      "step": 6723
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.758615493774414,
+      "learning_rate": 9.835297095376358e-06,
+      "loss": 0.7725,
+      "step": 6724
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.551060914993286,
+      "learning_rate": 9.835229974269124e-06,
+      "loss": 0.844,
+      "step": 6725
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.7266926765441895,
+      "learning_rate": 9.835162839716927e-06,
+      "loss": 0.4853,
+      "step": 6726
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.133965492248535,
+      "learning_rate": 9.835095691719954e-06,
+      "loss": 0.8009,
+      "step": 6727
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.480591058731079,
+      "learning_rate": 9.83502853027839e-06,
+      "loss": 0.4268,
+      "step": 6728
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 4.273105144500732,
+      "learning_rate": 9.834961355392422e-06,
+      "loss": 0.8698,
+      "step": 6729
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.030658721923828,
+      "learning_rate": 9.83489416706224e-06,
+      "loss": 0.8495,
+      "step": 6730
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.1352150440216064,
+      "learning_rate": 9.834826965288027e-06,
+      "loss": 0.9154,
+      "step": 6731
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.7054845094680786,
+      "learning_rate": 9.834759750069973e-06,
+      "loss": 0.6834,
+      "step": 6732
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.346000909805298,
+      "learning_rate": 9.834692521408262e-06,
+      "loss": 0.5894,
+      "step": 6733
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.846096992492676,
+      "learning_rate": 9.83462527930308e-06,
+      "loss": 0.8968,
+      "step": 6734
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.480766534805298,
+      "learning_rate": 9.83455802375462e-06,
+      "loss": 0.6516,
+      "step": 6735
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.53248929977417,
+      "learning_rate": 9.834490754763065e-06,
+      "loss": 0.5217,
+      "step": 6736
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.35558819770813,
+      "learning_rate": 9.8344234723286e-06,
+      "loss": 0.776,
+      "step": 6737
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 4.732817649841309,
+      "learning_rate": 9.834356176451417e-06,
+      "loss": 0.6055,
+      "step": 6738
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 4.251257419586182,
+      "learning_rate": 9.834288867131698e-06,
+      "loss": 0.9648,
+      "step": 6739
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.9423370361328125,
+      "learning_rate": 9.834221544369632e-06,
+      "loss": 0.976,
+      "step": 6740
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.835323691368103,
+      "learning_rate": 9.83415420816541e-06,
+      "loss": 0.7424,
+      "step": 6741
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.5592074394226074,
+      "learning_rate": 9.834086858519213e-06,
+      "loss": 0.6313,
+      "step": 6742
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.5820329189300537,
+      "learning_rate": 9.834019495431232e-06,
+      "loss": 0.6641,
+      "step": 6743
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.6262402534484863,
+      "learning_rate": 9.833952118901655e-06,
+      "loss": 0.7381,
+      "step": 6744
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 4.0198073387146,
+      "learning_rate": 9.833884728930668e-06,
+      "loss": 0.6621,
+      "step": 6745
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.1860694885253906,
+      "learning_rate": 9.833817325518459e-06,
+      "loss": 0.8298,
+      "step": 6746
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.6758925914764404,
+      "learning_rate": 9.833749908665212e-06,
+      "loss": 0.4508,
+      "step": 6747
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.195037603378296,
+      "learning_rate": 9.833682478371118e-06,
+      "loss": 0.6463,
+      "step": 6748
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.454906463623047,
+      "learning_rate": 9.833615034636366e-06,
+      "loss": 0.3593,
+      "step": 6749
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.1313486099243164,
+      "learning_rate": 9.83354757746114e-06,
+      "loss": 0.6908,
+      "step": 6750
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.9354510307312012,
+      "learning_rate": 9.833480106845627e-06,
+      "loss": 0.6551,
+      "step": 6751
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.6323964595794678,
+      "learning_rate": 9.83341262279002e-06,
+      "loss": 0.752,
+      "step": 6752
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.2022461891174316,
+      "learning_rate": 9.8333451252945e-06,
+      "loss": 0.7046,
+      "step": 6753
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.0544002056121826,
+      "learning_rate": 9.83327761435926e-06,
+      "loss": 0.6404,
+      "step": 6754
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.496732473373413,
+      "learning_rate": 9.833210089984484e-06,
+      "loss": 0.723,
+      "step": 6755
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.161555528640747,
+      "learning_rate": 9.833142552170362e-06,
+      "loss": 0.8254,
+      "step": 6756
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.014753580093384,
+      "learning_rate": 9.833075000917082e-06,
+      "loss": 0.7979,
+      "step": 6757
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.828209161758423,
+      "learning_rate": 9.833007436224829e-06,
+      "loss": 0.719,
+      "step": 6758
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.258118152618408,
+      "learning_rate": 9.832939858093795e-06,
+      "loss": 0.6621,
+      "step": 6759
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.222193479537964,
+      "learning_rate": 9.832872266524164e-06,
+      "loss": 0.7432,
+      "step": 6760
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.472580909729004,
+      "learning_rate": 9.832804661516126e-06,
+      "loss": 0.5161,
+      "step": 6761
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.3275811672210693,
+      "learning_rate": 9.832737043069868e-06,
+      "loss": 0.656,
+      "step": 6762
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.656928300857544,
+      "learning_rate": 9.83266941118558e-06,
+      "loss": 0.8571,
+      "step": 6763
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.6647205352783203,
+      "learning_rate": 9.83260176586345e-06,
+      "loss": 0.6507,
+      "step": 6764
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.7921175956726074,
+      "learning_rate": 9.832534107103661e-06,
+      "loss": 0.7494,
+      "step": 6765
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.5835000276565552,
+      "learning_rate": 9.832466434906409e-06,
+      "loss": 0.7522,
+      "step": 6766
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.425274133682251,
+      "learning_rate": 9.832398749271876e-06,
+      "loss": 0.8532,
+      "step": 6767
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.733950614929199,
+      "learning_rate": 9.832331050200252e-06,
+      "loss": 0.805,
+      "step": 6768
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 8.560796737670898,
+      "learning_rate": 9.832263337691725e-06,
+      "loss": 0.7523,
+      "step": 6769
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 4.377222061157227,
+      "learning_rate": 9.832195611746487e-06,
+      "loss": 0.7864,
+      "step": 6770
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.4969969987869263,
+      "learning_rate": 9.83212787236472e-06,
+      "loss": 0.6802,
+      "step": 6771
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.4983670711517334,
+      "learning_rate": 9.832060119546618e-06,
+      "loss": 0.7484,
+      "step": 6772
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.1491005420684814,
+      "learning_rate": 9.831992353292365e-06,
+      "loss": 0.7076,
+      "step": 6773
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.2163751125335693,
+      "learning_rate": 9.831924573602152e-06,
+      "loss": 0.4327,
+      "step": 6774
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.6522809267044067,
+      "learning_rate": 9.831856780476166e-06,
+      "loss": 0.5695,
+      "step": 6775
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 5.251799583435059,
+      "learning_rate": 9.831788973914598e-06,
+      "loss": 0.8581,
+      "step": 6776
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.7556276321411133,
+      "learning_rate": 9.831721153917633e-06,
+      "loss": 0.579,
+      "step": 6777
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 4.5130486488342285,
+      "learning_rate": 9.831653320485464e-06,
+      "loss": 0.852,
+      "step": 6778
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.3224055767059326,
+      "learning_rate": 9.831585473618274e-06,
+      "loss": 0.5684,
+      "step": 6779
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.9530045986175537,
+      "learning_rate": 9.831517613316258e-06,
+      "loss": 0.7402,
+      "step": 6780
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.626437187194824,
+      "learning_rate": 9.831449739579599e-06,
+      "loss": 0.8225,
+      "step": 6781
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.9116156101226807,
+      "learning_rate": 9.83138185240849e-06,
+      "loss": 0.4377,
+      "step": 6782
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.5563783645629883,
+      "learning_rate": 9.831313951803116e-06,
+      "loss": 0.6518,
+      "step": 6783
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.0855112075805664,
+      "learning_rate": 9.831246037763669e-06,
+      "loss": 0.5404,
+      "step": 6784
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.7412121295928955,
+      "learning_rate": 9.831178110290335e-06,
+      "loss": 0.842,
+      "step": 6785
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.0964763164520264,
+      "learning_rate": 9.831110169383307e-06,
+      "loss": 0.5626,
+      "step": 6786
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 4.089901447296143,
+      "learning_rate": 9.831042215042769e-06,
+      "loss": 0.7017,
+      "step": 6787
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.255007743835449,
+      "learning_rate": 9.830974247268914e-06,
+      "loss": 0.552,
+      "step": 6788
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.765189290046692,
+      "learning_rate": 9.83090626606193e-06,
+      "loss": 0.3471,
+      "step": 6789
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.408527135848999,
+      "learning_rate": 9.830838271422004e-06,
+      "loss": 0.7415,
+      "step": 6790
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.853224754333496,
+      "learning_rate": 9.830770263349327e-06,
+      "loss": 0.6694,
+      "step": 6791
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.142667770385742,
+      "learning_rate": 9.830702241844088e-06,
+      "loss": 0.5007,
+      "step": 6792
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.874393939971924,
+      "learning_rate": 9.830634206906473e-06,
+      "loss": 0.5234,
+      "step": 6793
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.2078936100006104,
+      "learning_rate": 9.830566158536676e-06,
+      "loss": 0.6475,
+      "step": 6794
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.137904644012451,
+      "learning_rate": 9.830498096734884e-06,
+      "loss": 0.7068,
+      "step": 6795
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.052722454071045,
+      "learning_rate": 9.830430021501286e-06,
+      "loss": 0.7514,
+      "step": 6796
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.5734636783599854,
+      "learning_rate": 9.830361932836072e-06,
+      "loss": 0.5969,
+      "step": 6797
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.696683883666992,
+      "learning_rate": 9.83029383073943e-06,
+      "loss": 0.3767,
+      "step": 6798
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.378533124923706,
+      "learning_rate": 9.830225715211552e-06,
+      "loss": 0.8273,
+      "step": 6799
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.5175769329071045,
+      "learning_rate": 9.830157586252626e-06,
+      "loss": 0.5084,
+      "step": 6800
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.5187008380889893,
+      "learning_rate": 9.83008944386284e-06,
+      "loss": 0.6714,
+      "step": 6801
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.3593993186950684,
+      "learning_rate": 9.830021288042385e-06,
+      "loss": 1.0078,
+      "step": 6802
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.5554561614990234,
+      "learning_rate": 9.829953118791448e-06,
+      "loss": 0.7141,
+      "step": 6803
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.370962142944336,
+      "learning_rate": 9.829884936110224e-06,
+      "loss": 0.582,
+      "step": 6804
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.017752170562744,
+      "learning_rate": 9.829816739998897e-06,
+      "loss": 0.6581,
+      "step": 6805
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.4858570098876953,
+      "learning_rate": 9.82974853045766e-06,
+      "loss": 0.4012,
+      "step": 6806
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.1408402919769287,
+      "learning_rate": 9.829680307486703e-06,
+      "loss": 0.9229,
+      "step": 6807
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.6374599933624268,
+      "learning_rate": 9.829612071086211e-06,
+      "loss": 0.6307,
+      "step": 6808
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.3801138401031494,
+      "learning_rate": 9.82954382125638e-06,
+      "loss": 0.7292,
+      "step": 6809
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.2071588039398193,
+      "learning_rate": 9.829475557997394e-06,
+      "loss": 0.7784,
+      "step": 6810
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.7443864345550537,
+      "learning_rate": 9.829407281309449e-06,
+      "loss": 0.6402,
+      "step": 6811
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.3523716926574707,
+      "learning_rate": 9.82933899119273e-06,
+      "loss": 0.6292,
+      "step": 6812
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.415167808532715,
+      "learning_rate": 9.829270687647427e-06,
+      "loss": 0.8326,
+      "step": 6813
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.3568246364593506,
+      "learning_rate": 9.829202370673731e-06,
+      "loss": 0.7585,
+      "step": 6814
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.970009446144104,
+      "learning_rate": 9.829134040271834e-06,
+      "loss": 0.7016,
+      "step": 6815
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.2548136711120605,
+      "learning_rate": 9.829065696441926e-06,
+      "loss": 0.8527,
+      "step": 6816
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.3970680236816406,
+      "learning_rate": 9.828997339184192e-06,
+      "loss": 0.5469,
+      "step": 6817
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.71450138092041,
+      "learning_rate": 9.828928968498827e-06,
+      "loss": 0.7236,
+      "step": 6818
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 4.18592643737793,
+      "learning_rate": 9.82886058438602e-06,
+      "loss": 0.8436,
+      "step": 6819
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.5436980724334717,
+      "learning_rate": 9.828792186845959e-06,
+      "loss": 0.7231,
+      "step": 6820
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.1862761974334717,
+      "learning_rate": 9.828723775878836e-06,
+      "loss": 0.605,
+      "step": 6821
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.4334805011749268,
+      "learning_rate": 9.82865535148484e-06,
+      "loss": 0.5465,
+      "step": 6822
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 6.9011688232421875,
+      "learning_rate": 9.828586913664167e-06,
+      "loss": 0.4929,
+      "step": 6823
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.2755184173583984,
+      "learning_rate": 9.828518462416999e-06,
+      "loss": 0.5978,
+      "step": 6824
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.020460844039917,
+      "learning_rate": 9.82844999774353e-06,
+      "loss": 0.7282,
+      "step": 6825
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.7050561904907227,
+      "learning_rate": 9.828381519643951e-06,
+      "loss": 0.5996,
+      "step": 6826
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.5821945667266846,
+      "learning_rate": 9.828313028118452e-06,
+      "loss": 0.7067,
+      "step": 6827
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.634316921234131,
+      "learning_rate": 9.828244523167224e-06,
+      "loss": 0.5524,
+      "step": 6828
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.0674898624420166,
+      "learning_rate": 9.828176004790454e-06,
+      "loss": 0.8988,
+      "step": 6829
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 4.382277011871338,
+      "learning_rate": 9.828107472988337e-06,
+      "loss": 0.4946,
+      "step": 6830
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.5531280040740967,
+      "learning_rate": 9.82803892776106e-06,
+      "loss": 0.5313,
+      "step": 6831
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.7680697441101074,
+      "learning_rate": 9.827970369108818e-06,
+      "loss": 0.5342,
+      "step": 6832
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.8460099697113037,
+      "learning_rate": 9.827901797031798e-06,
+      "loss": 0.6563,
+      "step": 6833
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.411649703979492,
+      "learning_rate": 9.827833211530191e-06,
+      "loss": 0.7775,
+      "step": 6834
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 5.9346842765808105,
+      "learning_rate": 9.82776461260419e-06,
+      "loss": 0.757,
+      "step": 6835
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.7458925247192383,
+      "learning_rate": 9.827696000253983e-06,
+      "loss": 0.6963,
+      "step": 6836
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.343477487564087,
+      "learning_rate": 9.827627374479762e-06,
+      "loss": 0.6966,
+      "step": 6837
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.9835824966430664,
+      "learning_rate": 9.827558735281718e-06,
+      "loss": 0.7071,
+      "step": 6838
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.88808012008667,
+      "learning_rate": 9.82749008266004e-06,
+      "loss": 0.9487,
+      "step": 6839
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.2156620025634766,
+      "learning_rate": 9.827421416614923e-06,
+      "loss": 0.6103,
+      "step": 6840
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 4.102158546447754,
+      "learning_rate": 9.827352737146554e-06,
+      "loss": 0.5203,
+      "step": 6841
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.212437629699707,
+      "learning_rate": 9.827284044255125e-06,
+      "loss": 1.0143,
+      "step": 6842
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.0948805809020996,
+      "learning_rate": 9.827215337940829e-06,
+      "loss": 0.7863,
+      "step": 6843
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.9844849109649658,
+      "learning_rate": 9.827146618203854e-06,
+      "loss": 0.4762,
+      "step": 6844
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.79947829246521,
+      "learning_rate": 9.827077885044394e-06,
+      "loss": 0.717,
+      "step": 6845
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.8677563667297363,
+      "learning_rate": 9.827009138462638e-06,
+      "loss": 0.7318,
+      "step": 6846
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.214419364929199,
+      "learning_rate": 9.826940378458776e-06,
+      "loss": 0.615,
+      "step": 6847
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.933790683746338,
+      "learning_rate": 9.826871605033003e-06,
+      "loss": 0.4815,
+      "step": 6848
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.2692179679870605,
+      "learning_rate": 9.826802818185508e-06,
+      "loss": 0.8267,
+      "step": 6849
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.3968136310577393,
+      "learning_rate": 9.826734017916483e-06,
+      "loss": 0.7921,
+      "step": 6850
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.370776891708374,
+      "learning_rate": 9.826665204226118e-06,
+      "loss": 0.9418,
+      "step": 6851
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 4.302151203155518,
+      "learning_rate": 9.826596377114605e-06,
+      "loss": 0.4815,
+      "step": 6852
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.1356003284454346,
+      "learning_rate": 9.826527536582135e-06,
+      "loss": 0.4104,
+      "step": 6853
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.5524961948394775,
+      "learning_rate": 9.826458682628902e-06,
+      "loss": 0.6649,
+      "step": 6854
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.2542624473571777,
+      "learning_rate": 9.826389815255095e-06,
+      "loss": 0.9347,
+      "step": 6855
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.9987646341323853,
+      "learning_rate": 9.826320934460904e-06,
+      "loss": 0.7446,
+      "step": 6856
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.5970709323883057,
+      "learning_rate": 9.826252040246525e-06,
+      "loss": 0.8334,
+      "step": 6857
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.9813363552093506,
+      "learning_rate": 9.826183132612147e-06,
+      "loss": 0.7081,
+      "step": 6858
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.482362985610962,
+      "learning_rate": 9.826114211557958e-06,
+      "loss": 0.849,
+      "step": 6859
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.8722407817840576,
+      "learning_rate": 9.826045277084156e-06,
+      "loss": 0.9481,
+      "step": 6860
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.3558108806610107,
+      "learning_rate": 9.82597632919093e-06,
+      "loss": 0.7671,
+      "step": 6861
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.189957857131958,
+      "learning_rate": 9.825907367878472e-06,
+      "loss": 0.6632,
+      "step": 6862
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.7320873737335205,
+      "learning_rate": 9.825838393146972e-06,
+      "loss": 0.5633,
+      "step": 6863
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.6729977130889893,
+      "learning_rate": 9.825769404996625e-06,
+      "loss": 0.8029,
+      "step": 6864
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.5427882671356201,
+      "learning_rate": 9.82570040342762e-06,
+      "loss": 0.6964,
+      "step": 6865
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.1017868518829346,
+      "learning_rate": 9.82563138844015e-06,
+      "loss": 0.5389,
+      "step": 6866
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.222954034805298,
+      "learning_rate": 9.825562360034406e-06,
+      "loss": 0.8696,
+      "step": 6867
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.069838285446167,
+      "learning_rate": 9.825493318210582e-06,
+      "loss": 0.6266,
+      "step": 6868
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.645556926727295,
+      "learning_rate": 9.825424262968868e-06,
+      "loss": 0.6692,
+      "step": 6869
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.2223451137542725,
+      "learning_rate": 9.825355194309459e-06,
+      "loss": 0.7918,
+      "step": 6870
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 5.202667236328125,
+      "learning_rate": 9.825286112232542e-06,
+      "loss": 0.6868,
+      "step": 6871
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.5612233877182007,
+      "learning_rate": 9.825217016738313e-06,
+      "loss": 0.6363,
+      "step": 6872
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.309011459350586,
+      "learning_rate": 9.825147907826962e-06,
+      "loss": 0.5159,
+      "step": 6873
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.299736976623535,
+      "learning_rate": 9.825078785498684e-06,
+      "loss": 0.759,
+      "step": 6874
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.2643203735351562,
+      "learning_rate": 9.825009649753668e-06,
+      "loss": 0.4884,
+      "step": 6875
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.9419126510620117,
+      "learning_rate": 9.824940500592109e-06,
+      "loss": 0.4654,
+      "step": 6876
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.3604090213775635,
+      "learning_rate": 9.824871338014197e-06,
+      "loss": 0.649,
+      "step": 6877
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.7437636852264404,
+      "learning_rate": 9.824802162020125e-06,
+      "loss": 0.8293,
+      "step": 6878
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.935065269470215,
+      "learning_rate": 9.824732972610086e-06,
+      "loss": 0.7041,
+      "step": 6879
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.439255475997925,
+      "learning_rate": 9.824663769784272e-06,
+      "loss": 0.6463,
+      "step": 6880
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.5245959758758545,
+      "learning_rate": 9.824594553542874e-06,
+      "loss": 0.71,
+      "step": 6881
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.3531479835510254,
+      "learning_rate": 9.824525323886089e-06,
+      "loss": 0.4189,
+      "step": 6882
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.0652899742126465,
+      "learning_rate": 9.824456080814104e-06,
+      "loss": 0.7329,
+      "step": 6883
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.841769218444824,
+      "learning_rate": 9.824386824327115e-06,
+      "loss": 0.7655,
+      "step": 6884
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.1387102603912354,
+      "learning_rate": 9.824317554425314e-06,
+      "loss": 0.6129,
+      "step": 6885
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.233443021774292,
+      "learning_rate": 9.824248271108891e-06,
+      "loss": 0.7168,
+      "step": 6886
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.639251947402954,
+      "learning_rate": 9.824178974378042e-06,
+      "loss": 0.6064,
+      "step": 6887
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.1004445552825928,
+      "learning_rate": 9.824109664232959e-06,
+      "loss": 0.5174,
+      "step": 6888
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.3647806644439697,
+      "learning_rate": 9.824040340673833e-06,
+      "loss": 0.7475,
+      "step": 6889
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.997006416320801,
+      "learning_rate": 9.823971003700858e-06,
+      "loss": 0.4847,
+      "step": 6890
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.9129650592803955,
+      "learning_rate": 9.823901653314227e-06,
+      "loss": 0.5566,
+      "step": 6891
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.858874559402466,
+      "learning_rate": 9.823832289514133e-06,
+      "loss": 0.7531,
+      "step": 6892
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.139023780822754,
+      "learning_rate": 9.82376291230077e-06,
+      "loss": 0.8015,
+      "step": 6893
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.789433479309082,
+      "learning_rate": 9.823693521674327e-06,
+      "loss": 0.7358,
+      "step": 6894
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.243367910385132,
+      "learning_rate": 9.823624117635e-06,
+      "loss": 0.6293,
+      "step": 6895
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.3240320682525635,
+      "learning_rate": 9.82355470018298e-06,
+      "loss": 0.5354,
+      "step": 6896
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.1010007858276367,
+      "learning_rate": 9.823485269318463e-06,
+      "loss": 0.4825,
+      "step": 6897
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.4032084941864014,
+      "learning_rate": 9.82341582504164e-06,
+      "loss": 0.5249,
+      "step": 6898
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 4.071831226348877,
+      "learning_rate": 9.823346367352704e-06,
+      "loss": 0.8745,
+      "step": 6899
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.046436071395874,
+      "learning_rate": 9.82327689625185e-06,
+      "loss": 0.9202,
+      "step": 6900
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.272949695587158,
+      "learning_rate": 9.823207411739268e-06,
+      "loss": 0.6791,
+      "step": 6901
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 4.308311462402344,
+      "learning_rate": 9.823137913815155e-06,
+      "loss": 0.5752,
+      "step": 6902
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.8686721324920654,
+      "learning_rate": 9.8230684024797e-06,
+      "loss": 0.937,
+      "step": 6903
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 11.778286933898926,
+      "learning_rate": 9.822998877733101e-06,
+      "loss": 0.7278,
+      "step": 6904
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.0991365909576416,
+      "learning_rate": 9.822929339575547e-06,
+      "loss": 0.7153,
+      "step": 6905
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.933966636657715,
+      "learning_rate": 9.822859788007233e-06,
+      "loss": 0.9395,
+      "step": 6906
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.8396594524383545,
+      "learning_rate": 9.822790223028354e-06,
+      "loss": 0.5639,
+      "step": 6907
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 5.04756498336792,
+      "learning_rate": 9.822720644639102e-06,
+      "loss": 0.5856,
+      "step": 6908
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 4.105938911437988,
+      "learning_rate": 9.82265105283967e-06,
+      "loss": 0.4909,
+      "step": 6909
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.4954910278320312,
+      "learning_rate": 9.822581447630251e-06,
+      "loss": 0.637,
+      "step": 6910
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.980234146118164,
+      "learning_rate": 9.82251182901104e-06,
+      "loss": 0.5192,
+      "step": 6911
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 4.299525737762451,
+      "learning_rate": 9.82244219698223e-06,
+      "loss": 1.0604,
+      "step": 6912
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.6228058338165283,
+      "learning_rate": 9.822372551544017e-06,
+      "loss": 0.6294,
+      "step": 6913
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.805793285369873,
+      "learning_rate": 9.82230289269659e-06,
+      "loss": 0.6604,
+      "step": 6914
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.4017043113708496,
+      "learning_rate": 9.822233220440148e-06,
+      "loss": 0.6556,
+      "step": 6915
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.295468330383301,
+      "learning_rate": 9.822163534774877e-06,
+      "loss": 0.5386,
+      "step": 6916
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.592895984649658,
+      "learning_rate": 9.82209383570098e-06,
+      "loss": 0.8726,
+      "step": 6917
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.794466018676758,
+      "learning_rate": 9.822024123218646e-06,
+      "loss": 0.6535,
+      "step": 6918
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.2532410621643066,
+      "learning_rate": 9.821954397328068e-06,
+      "loss": 0.6168,
+      "step": 6919
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.992045521736145,
+      "learning_rate": 9.821884658029442e-06,
+      "loss": 0.7871,
+      "step": 6920
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.0850419998168945,
+      "learning_rate": 9.821814905322959e-06,
+      "loss": 0.6255,
+      "step": 6921
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.5022189617156982,
+      "learning_rate": 9.821745139208816e-06,
+      "loss": 0.5464,
+      "step": 6922
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.7403451204299927,
+      "learning_rate": 9.821675359687208e-06,
+      "loss": 0.9586,
+      "step": 6923
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.639103889465332,
+      "learning_rate": 9.821605566758325e-06,
+      "loss": 0.8951,
+      "step": 6924
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.481231927871704,
+      "learning_rate": 9.821535760422363e-06,
+      "loss": 0.715,
+      "step": 6925
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.781284809112549,
+      "learning_rate": 9.821465940679517e-06,
+      "loss": 0.6848,
+      "step": 6926
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.0337135791778564,
+      "learning_rate": 9.821396107529982e-06,
+      "loss": 0.6867,
+      "step": 6927
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.237715005874634,
+      "learning_rate": 9.821326260973948e-06,
+      "loss": 0.8374,
+      "step": 6928
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 4.154456615447998,
+      "learning_rate": 9.821256401011611e-06,
+      "loss": 0.5389,
+      "step": 6929
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.8980631828308105,
+      "learning_rate": 9.821186527643169e-06,
+      "loss": 0.7006,
+      "step": 6930
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.1211421489715576,
+      "learning_rate": 9.821116640868813e-06,
+      "loss": 0.6592,
+      "step": 6931
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.569049596786499,
+      "learning_rate": 9.821046740688735e-06,
+      "loss": 0.8336,
+      "step": 6932
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.2318687438964844,
+      "learning_rate": 9.820976827103135e-06,
+      "loss": 0.6939,
+      "step": 6933
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.832871675491333,
+      "learning_rate": 9.820906900112203e-06,
+      "loss": 0.8115,
+      "step": 6934
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.6695573329925537,
+      "learning_rate": 9.820836959716135e-06,
+      "loss": 0.7041,
+      "step": 6935
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.3496177196502686,
+      "learning_rate": 9.820767005915126e-06,
+      "loss": 0.5469,
+      "step": 6936
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.771575689315796,
+      "learning_rate": 9.820697038709368e-06,
+      "loss": 0.6574,
+      "step": 6937
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.1557252407073975,
+      "learning_rate": 9.82062705809906e-06,
+      "loss": 0.7339,
+      "step": 6938
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.080122947692871,
+      "learning_rate": 9.820557064084393e-06,
+      "loss": 0.612,
+      "step": 6939
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.4959516525268555,
+      "learning_rate": 9.820487056665562e-06,
+      "loss": 0.5139,
+      "step": 6940
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.7449443340301514,
+      "learning_rate": 9.820417035842761e-06,
+      "loss": 0.6035,
+      "step": 6941
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.3779799938201904,
+      "learning_rate": 9.820347001616188e-06,
+      "loss": 0.6291,
+      "step": 6942
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 5.358976364135742,
+      "learning_rate": 9.820276953986037e-06,
+      "loss": 0.8944,
+      "step": 6943
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.4700119495391846,
+      "learning_rate": 9.820206892952498e-06,
+      "loss": 0.7586,
+      "step": 6944
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 5.603726387023926,
+      "learning_rate": 9.820136818515772e-06,
+      "loss": 0.7721,
+      "step": 6945
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.6735167503356934,
+      "learning_rate": 9.82006673067605e-06,
+      "loss": 0.751,
+      "step": 6946
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.9212092161178589,
+      "learning_rate": 9.819996629433529e-06,
+      "loss": 0.5317,
+      "step": 6947
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.136378526687622,
+      "learning_rate": 9.819926514788402e-06,
+      "loss": 0.859,
+      "step": 6948
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.731920599937439,
+      "learning_rate": 9.819856386740865e-06,
+      "loss": 0.4348,
+      "step": 6949
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.8901166915893555,
+      "learning_rate": 9.819786245291113e-06,
+      "loss": 0.6812,
+      "step": 6950
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.379034996032715,
+      "learning_rate": 9.819716090439341e-06,
+      "loss": 0.6738,
+      "step": 6951
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.6151304244995117,
+      "learning_rate": 9.819645922185745e-06,
+      "loss": 0.6578,
+      "step": 6952
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.163923501968384,
+      "learning_rate": 9.819575740530518e-06,
+      "loss": 0.7634,
+      "step": 6953
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.5155396461486816,
+      "learning_rate": 9.819505545473858e-06,
+      "loss": 0.8681,
+      "step": 6954
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.987344264984131,
+      "learning_rate": 9.819435337015957e-06,
+      "loss": 0.9467,
+      "step": 6955
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.799586296081543,
+      "learning_rate": 9.819365115157011e-06,
+      "loss": 0.7445,
+      "step": 6956
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.1731534004211426,
+      "learning_rate": 9.819294879897217e-06,
+      "loss": 0.7291,
+      "step": 6957
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.8670287132263184,
+      "learning_rate": 9.81922463123677e-06,
+      "loss": 0.9067,
+      "step": 6958
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.933591365814209,
+      "learning_rate": 9.819154369175863e-06,
+      "loss": 0.7522,
+      "step": 6959
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 5.913595676422119,
+      "learning_rate": 9.819084093714695e-06,
+      "loss": 0.7947,
+      "step": 6960
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.469595432281494,
+      "learning_rate": 9.819013804853459e-06,
+      "loss": 0.7952,
+      "step": 6961
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.961153268814087,
+      "learning_rate": 9.818943502592349e-06,
+      "loss": 0.8884,
+      "step": 6962
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.3837037086486816,
+      "learning_rate": 9.818873186931564e-06,
+      "loss": 0.5691,
+      "step": 6963
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.553955554962158,
+      "learning_rate": 9.818802857871297e-06,
+      "loss": 0.4445,
+      "step": 6964
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.4520537853240967,
+      "learning_rate": 9.818732515411744e-06,
+      "loss": 0.6848,
+      "step": 6965
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.742777109146118,
+      "learning_rate": 9.818662159553103e-06,
+      "loss": 0.9265,
+      "step": 6966
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.6403303146362305,
+      "learning_rate": 9.818591790295567e-06,
+      "loss": 0.7363,
+      "step": 6967
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.120030403137207,
+      "learning_rate": 9.81852140763933e-06,
+      "loss": 0.7527,
+      "step": 6968
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.08245587348938,
+      "learning_rate": 9.818451011584592e-06,
+      "loss": 0.8594,
+      "step": 6969
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.284409761428833,
+      "learning_rate": 9.818380602131547e-06,
+      "loss": 0.7053,
+      "step": 6970
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.612464427947998,
+      "learning_rate": 9.818310179280391e-06,
+      "loss": 0.7259,
+      "step": 6971
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.7176294326782227,
+      "learning_rate": 9.818239743031317e-06,
+      "loss": 0.6203,
+      "step": 6972
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 6.7485246658325195,
+      "learning_rate": 9.818169293384526e-06,
+      "loss": 0.7321,
+      "step": 6973
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.9406256675720215,
+      "learning_rate": 9.81809883034021e-06,
+      "loss": 0.6838,
+      "step": 6974
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.332914113998413,
+      "learning_rate": 9.818028353898566e-06,
+      "loss": 0.6583,
+      "step": 6975
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.8299094438552856,
+      "learning_rate": 9.817957864059789e-06,
+      "loss": 0.5947,
+      "step": 6976
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.5453898906707764,
+      "learning_rate": 9.817887360824076e-06,
+      "loss": 0.6996,
+      "step": 6977
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.9708415269851685,
+      "learning_rate": 9.817816844191624e-06,
+      "loss": 0.7326,
+      "step": 6978
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.247184991836548,
+      "learning_rate": 9.817746314162628e-06,
+      "loss": 0.8666,
+      "step": 6979
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.860668182373047,
+      "learning_rate": 9.817675770737285e-06,
+      "loss": 0.8336,
+      "step": 6980
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.9198129177093506,
+      "learning_rate": 9.81760521391579e-06,
+      "loss": 0.6901,
+      "step": 6981
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.482521891593933,
+      "learning_rate": 9.817534643698338e-06,
+      "loss": 0.7887,
+      "step": 6982
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.5471396446228027,
+      "learning_rate": 9.81746406008513e-06,
+      "loss": 0.4311,
+      "step": 6983
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 5.176010608673096,
+      "learning_rate": 9.817393463076357e-06,
+      "loss": 0.7159,
+      "step": 6984
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.138122081756592,
+      "learning_rate": 9.817322852672217e-06,
+      "loss": 0.6854,
+      "step": 6985
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.8576765060424805,
+      "learning_rate": 9.817252228872909e-06,
+      "loss": 0.5935,
+      "step": 6986
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.5128560066223145,
+      "learning_rate": 9.817181591678625e-06,
+      "loss": 0.6384,
+      "step": 6987
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.323042392730713,
+      "learning_rate": 9.817110941089565e-06,
+      "loss": 0.6491,
+      "step": 6988
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.3673605918884277,
+      "learning_rate": 9.817040277105923e-06,
+      "loss": 0.777,
+      "step": 6989
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.92597496509552,
+      "learning_rate": 9.816969599727897e-06,
+      "loss": 0.6114,
+      "step": 6990
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.9656803607940674,
+      "learning_rate": 9.816898908955683e-06,
+      "loss": 0.6689,
+      "step": 6991
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.6095232963562012,
+      "learning_rate": 9.816828204789477e-06,
+      "loss": 0.4277,
+      "step": 6992
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.0607070922851562,
+      "learning_rate": 9.816757487229475e-06,
+      "loss": 0.5087,
+      "step": 6993
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.6271326541900635,
+      "learning_rate": 9.816686756275878e-06,
+      "loss": 0.8978,
+      "step": 6994
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.141049861907959,
+      "learning_rate": 9.816616011928877e-06,
+      "loss": 0.8104,
+      "step": 6995
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.1980509757995605,
+      "learning_rate": 9.816545254188673e-06,
+      "loss": 0.5696,
+      "step": 6996
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 6.716854572296143,
+      "learning_rate": 9.81647448305546e-06,
+      "loss": 0.4653,
+      "step": 6997
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.6320230960845947,
+      "learning_rate": 9.816403698529435e-06,
+      "loss": 0.8942,
+      "step": 6998
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.0348386764526367,
+      "learning_rate": 9.816332900610797e-06,
+      "loss": 0.6624,
+      "step": 6999
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.274092435836792,
+      "learning_rate": 9.816262089299742e-06,
+      "loss": 0.7734,
+      "step": 7000
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.054629325866699,
+      "learning_rate": 9.816191264596463e-06,
+      "loss": 0.7563,
+      "step": 7001
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 4.6965651512146,
+      "learning_rate": 9.816120426501164e-06,
+      "loss": 0.7262,
+      "step": 7002
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.8923763036727905,
+      "learning_rate": 9.816049575014037e-06,
+      "loss": 0.9719,
+      "step": 7003
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 4.8521857261657715,
+      "learning_rate": 9.81597871013528e-06,
+      "loss": 0.4331,
+      "step": 7004
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.572178602218628,
+      "learning_rate": 9.81590783186509e-06,
+      "loss": 0.5188,
+      "step": 7005
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.326840877532959,
+      "learning_rate": 9.815836940203665e-06,
+      "loss": 0.6359,
+      "step": 7006
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 4.095197677612305,
+      "learning_rate": 9.8157660351512e-06,
+      "loss": 0.6771,
+      "step": 7007
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.8576183319091797,
+      "learning_rate": 9.815695116707895e-06,
+      "loss": 0.6736,
+      "step": 7008
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.4717159271240234,
+      "learning_rate": 9.815624184873947e-06,
+      "loss": 0.8261,
+      "step": 7009
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.9430952072143555,
+      "learning_rate": 9.81555323964955e-06,
+      "loss": 0.4217,
+      "step": 7010
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.6943764686584473,
+      "learning_rate": 9.815482281034905e-06,
+      "loss": 0.841,
+      "step": 7011
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.8790929317474365,
+      "learning_rate": 9.815411309030207e-06,
+      "loss": 0.6712,
+      "step": 7012
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 4.107418060302734,
+      "learning_rate": 9.815340323635653e-06,
+      "loss": 0.7776,
+      "step": 7013
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.403648614883423,
+      "learning_rate": 9.815269324851444e-06,
+      "loss": 0.6814,
+      "step": 7014
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.8657381534576416,
+      "learning_rate": 9.815198312677772e-06,
+      "loss": 0.6233,
+      "step": 7015
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.822944402694702,
+      "learning_rate": 9.815127287114839e-06,
+      "loss": 0.6121,
+      "step": 7016
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.2369229793548584,
+      "learning_rate": 9.81505624816284e-06,
+      "loss": 0.2824,
+      "step": 7017
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.8819384574890137,
+      "learning_rate": 9.814985195821975e-06,
+      "loss": 0.5633,
+      "step": 7018
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.2176358699798584,
+      "learning_rate": 9.81491413009244e-06,
+      "loss": 0.7247,
+      "step": 7019
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 4.881229400634766,
+      "learning_rate": 9.81484305097443e-06,
+      "loss": 0.6449,
+      "step": 7020
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.562047004699707,
+      "learning_rate": 9.814771958468146e-06,
+      "loss": 0.5574,
+      "step": 7021
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.3035309314727783,
+      "learning_rate": 9.814700852573784e-06,
+      "loss": 0.6953,
+      "step": 7022
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 4.399063587188721,
+      "learning_rate": 9.814629733291546e-06,
+      "loss": 0.6892,
+      "step": 7023
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.1096153259277344,
+      "learning_rate": 9.814558600621623e-06,
+      "loss": 0.5434,
+      "step": 7024
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.556901216506958,
+      "learning_rate": 9.814487454564217e-06,
+      "loss": 0.9458,
+      "step": 7025
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.429652690887451,
+      "learning_rate": 9.814416295119524e-06,
+      "loss": 0.9378,
+      "step": 7026
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.1522486209869385,
+      "learning_rate": 9.814345122287746e-06,
+      "loss": 0.4591,
+      "step": 7027
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.9147675037384033,
+      "learning_rate": 9.814273936069074e-06,
+      "loss": 0.7323,
+      "step": 7028
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.249228000640869,
+      "learning_rate": 9.814202736463711e-06,
+      "loss": 0.6196,
+      "step": 7029
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.2048068046569824,
+      "learning_rate": 9.814131523471855e-06,
+      "loss": 0.4987,
+      "step": 7030
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.1694443225860596,
+      "learning_rate": 9.814060297093701e-06,
+      "loss": 0.8606,
+      "step": 7031
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.3289477825164795,
+      "learning_rate": 9.81398905732945e-06,
+      "loss": 0.6992,
+      "step": 7032
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.3301680088043213,
+      "learning_rate": 9.813917804179298e-06,
+      "loss": 0.8698,
+      "step": 7033
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.2829601764678955,
+      "learning_rate": 9.813846537643443e-06,
+      "loss": 0.8161,
+      "step": 7034
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.346153497695923,
+      "learning_rate": 9.813775257722086e-06,
+      "loss": 0.658,
+      "step": 7035
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.8275742530822754,
+      "learning_rate": 9.813703964415422e-06,
+      "loss": 0.5998,
+      "step": 7036
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 4.292749404907227,
+      "learning_rate": 9.813632657723651e-06,
+      "loss": 0.66,
+      "step": 7037
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.4658265113830566,
+      "learning_rate": 9.81356133764697e-06,
+      "loss": 0.8079,
+      "step": 7038
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.766500234603882,
+      "learning_rate": 9.813490004185579e-06,
+      "loss": 0.7446,
+      "step": 7039
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.6296520233154297,
+      "learning_rate": 9.813418657339676e-06,
+      "loss": 0.6154,
+      "step": 7040
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 4.206751346588135,
+      "learning_rate": 9.813347297109457e-06,
+      "loss": 0.5088,
+      "step": 7041
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.4583654403686523,
+      "learning_rate": 9.813275923495124e-06,
+      "loss": 0.76,
+      "step": 7042
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.7366769313812256,
+      "learning_rate": 9.813204536496873e-06,
+      "loss": 0.8751,
+      "step": 7043
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.4505014419555664,
+      "learning_rate": 9.813133136114903e-06,
+      "loss": 0.7863,
+      "step": 7044
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.9803131818771362,
+      "learning_rate": 9.813061722349414e-06,
+      "loss": 0.3853,
+      "step": 7045
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.8700246810913086,
+      "learning_rate": 9.812990295200601e-06,
+      "loss": 0.8307,
+      "step": 7046
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.578777551651001,
+      "learning_rate": 9.812918854668668e-06,
+      "loss": 0.7368,
+      "step": 7047
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.3657262325286865,
+      "learning_rate": 9.812847400753809e-06,
+      "loss": 0.5547,
+      "step": 7048
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.991577386856079,
+      "learning_rate": 9.812775933456224e-06,
+      "loss": 0.6148,
+      "step": 7049
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.1124427318573,
+      "learning_rate": 9.812704452776113e-06,
+      "loss": 0.6126,
+      "step": 7050
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.9133756160736084,
+      "learning_rate": 9.812632958713673e-06,
+      "loss": 0.501,
+      "step": 7051
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.9006667137145996,
+      "learning_rate": 9.812561451269105e-06,
+      "loss": 0.6126,
+      "step": 7052
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.1419191360473633,
+      "learning_rate": 9.812489930442605e-06,
+      "loss": 0.7047,
+      "step": 7053
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.689033031463623,
+      "learning_rate": 9.812418396234373e-06,
+      "loss": 0.671,
+      "step": 7054
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.173401355743408,
+      "learning_rate": 9.812346848644609e-06,
+      "loss": 0.8408,
+      "step": 7055
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.0615041255950928,
+      "learning_rate": 9.812275287673511e-06,
+      "loss": 0.5936,
+      "step": 7056
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.0514540672302246,
+      "learning_rate": 9.812203713321279e-06,
+      "loss": 0.9347,
+      "step": 7057
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.6507866382598877,
+      "learning_rate": 9.81213212558811e-06,
+      "loss": 0.6317,
+      "step": 7058
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.5750362873077393,
+      "learning_rate": 9.812060524474205e-06,
+      "loss": 0.6175,
+      "step": 7059
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.6473400592803955,
+      "learning_rate": 9.811988909979761e-06,
+      "loss": 0.6933,
+      "step": 7060
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.3001811504364014,
+      "learning_rate": 9.811917282104979e-06,
+      "loss": 0.7229,
+      "step": 7061
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.174015760421753,
+      "learning_rate": 9.811845640850059e-06,
+      "loss": 0.6365,
+      "step": 7062
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 4.405415058135986,
+      "learning_rate": 9.811773986215196e-06,
+      "loss": 0.6646,
+      "step": 7063
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.4703900814056396,
+      "learning_rate": 9.811702318200595e-06,
+      "loss": 0.8773,
+      "step": 7064
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.6874582767486572,
+      "learning_rate": 9.811630636806452e-06,
+      "loss": 0.6663,
+      "step": 7065
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 4.271317005157471,
+      "learning_rate": 9.811558942032965e-06,
+      "loss": 0.864,
+      "step": 7066
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.4797608852386475,
+      "learning_rate": 9.811487233880337e-06,
+      "loss": 0.7363,
+      "step": 7067
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 4.15908145904541,
+      "learning_rate": 9.811415512348763e-06,
+      "loss": 0.8661,
+      "step": 7068
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 4.085733890533447,
+      "learning_rate": 9.811343777438447e-06,
+      "loss": 0.5543,
+      "step": 7069
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 4.40140438079834,
+      "learning_rate": 9.811272029149586e-06,
+      "loss": 0.7591,
+      "step": 7070
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.470323085784912,
+      "learning_rate": 9.81120026748238e-06,
+      "loss": 0.9586,
+      "step": 7071
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.740182876586914,
+      "learning_rate": 9.811128492437027e-06,
+      "loss": 0.5856,
+      "step": 7072
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.1903998851776123,
+      "learning_rate": 9.81105670401373e-06,
+      "loss": 0.6062,
+      "step": 7073
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 16.120651245117188,
+      "learning_rate": 9.810984902212684e-06,
+      "loss": 0.8568,
+      "step": 7074
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.6697511672973633,
+      "learning_rate": 9.810913087034092e-06,
+      "loss": 0.5981,
+      "step": 7075
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.1858105659484863,
+      "learning_rate": 9.810841258478155e-06,
+      "loss": 0.8933,
+      "step": 7076
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.974992275238037,
+      "learning_rate": 9.810769416545067e-06,
+      "loss": 0.7999,
+      "step": 7077
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.110858678817749,
+      "learning_rate": 9.810697561235035e-06,
+      "loss": 0.8164,
+      "step": 7078
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.0966832637786865,
+      "learning_rate": 9.810625692548252e-06,
+      "loss": 0.6676,
+      "step": 7079
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.403139591217041,
+      "learning_rate": 9.810553810484924e-06,
+      "loss": 0.5656,
+      "step": 7080
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.098750114440918,
+      "learning_rate": 9.810481915045246e-06,
+      "loss": 0.9032,
+      "step": 7081
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.1535699367523193,
+      "learning_rate": 9.81041000622942e-06,
+      "loss": 0.791,
+      "step": 7082
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 4.940645694732666,
+      "learning_rate": 9.810338084037648e-06,
+      "loss": 0.5872,
+      "step": 7083
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.8924384117126465,
+      "learning_rate": 9.810266148470124e-06,
+      "loss": 0.9094,
+      "step": 7084
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.479260206222534,
+      "learning_rate": 9.810194199527054e-06,
+      "loss": 0.6349,
+      "step": 7085
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.7791669368743896,
+      "learning_rate": 9.810122237208637e-06,
+      "loss": 0.4487,
+      "step": 7086
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.499742269515991,
+      "learning_rate": 9.81005026151507e-06,
+      "loss": 0.6642,
+      "step": 7087
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.468266725540161,
+      "learning_rate": 9.809978272446556e-06,
+      "loss": 0.505,
+      "step": 7088
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 4.430003643035889,
+      "learning_rate": 9.809906270003294e-06,
+      "loss": 0.8227,
+      "step": 7089
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 4.356377601623535,
+      "learning_rate": 9.809834254185485e-06,
+      "loss": 0.6793,
+      "step": 7090
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 5.438969612121582,
+      "learning_rate": 9.809762224993328e-06,
+      "loss": 0.6626,
+      "step": 7091
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 4.012573719024658,
+      "learning_rate": 9.809690182427025e-06,
+      "loss": 0.5918,
+      "step": 7092
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.6932923793792725,
+      "learning_rate": 9.809618126486775e-06,
+      "loss": 0.5113,
+      "step": 7093
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.065382957458496,
+      "learning_rate": 9.809546057172778e-06,
+      "loss": 0.7026,
+      "step": 7094
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.6024351119995117,
+      "learning_rate": 9.809473974485235e-06,
+      "loss": 0.2528,
+      "step": 7095
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.7200400829315186,
+      "learning_rate": 9.809401878424346e-06,
+      "loss": 0.5469,
+      "step": 7096
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.324216604232788,
+      "learning_rate": 9.809329768990315e-06,
+      "loss": 0.7755,
+      "step": 7097
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.5023176670074463,
+      "learning_rate": 9.809257646183336e-06,
+      "loss": 0.8027,
+      "step": 7098
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.4127304553985596,
+      "learning_rate": 9.809185510003616e-06,
+      "loss": 0.4484,
+      "step": 7099
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.120393753051758,
+      "learning_rate": 9.809113360451352e-06,
+      "loss": 0.5759,
+      "step": 7100
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.0169522762298584,
+      "learning_rate": 9.809041197526744e-06,
+      "loss": 0.7198,
+      "step": 7101
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.7707386016845703,
+      "learning_rate": 9.808969021229995e-06,
+      "loss": 0.7585,
+      "step": 7102
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.241682291030884,
+      "learning_rate": 9.808896831561304e-06,
+      "loss": 0.428,
+      "step": 7103
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.0567498207092285,
+      "learning_rate": 9.808824628520872e-06,
+      "loss": 0.8754,
+      "step": 7104
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.1422462463378906,
+      "learning_rate": 9.808752412108902e-06,
+      "loss": 0.4526,
+      "step": 7105
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 6.93939208984375,
+      "learning_rate": 9.808680182325592e-06,
+      "loss": 0.6008,
+      "step": 7106
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 4.232153415679932,
+      "learning_rate": 9.808607939171142e-06,
+      "loss": 0.8584,
+      "step": 7107
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.8404014110565186,
+      "learning_rate": 9.808535682645757e-06,
+      "loss": 0.6047,
+      "step": 7108
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.9572569131851196,
+      "learning_rate": 9.808463412749634e-06,
+      "loss": 0.5193,
+      "step": 7109
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.5120184421539307,
+      "learning_rate": 9.808391129482977e-06,
+      "loss": 0.7222,
+      "step": 7110
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.063344717025757,
+      "learning_rate": 9.808318832845984e-06,
+      "loss": 0.6364,
+      "step": 7111
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.097447395324707,
+      "learning_rate": 9.808246522838859e-06,
+      "loss": 0.5917,
+      "step": 7112
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.4078893661499023,
+      "learning_rate": 9.808174199461801e-06,
+      "loss": 0.4482,
+      "step": 7113
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.102203369140625,
+      "learning_rate": 9.808101862715011e-06,
+      "loss": 0.5027,
+      "step": 7114
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.1571197509765625,
+      "learning_rate": 9.80802951259869e-06,
+      "loss": 0.6447,
+      "step": 7115
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.8075902462005615,
+      "learning_rate": 9.807957149113043e-06,
+      "loss": 0.593,
+      "step": 7116
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.9331960678100586,
+      "learning_rate": 9.807884772258267e-06,
+      "loss": 0.5965,
+      "step": 7117
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.4960615634918213,
+      "learning_rate": 9.807812382034565e-06,
+      "loss": 0.6763,
+      "step": 7118
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.243201494216919,
+      "learning_rate": 9.807739978442136e-06,
+      "loss": 0.6723,
+      "step": 7119
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.1209800243377686,
+      "learning_rate": 9.807667561481186e-06,
+      "loss": 0.4648,
+      "step": 7120
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 7.654330253601074,
+      "learning_rate": 9.807595131151912e-06,
+      "loss": 0.7295,
+      "step": 7121
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 5.243066787719727,
+      "learning_rate": 9.807522687454516e-06,
+      "loss": 0.5447,
+      "step": 7122
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.306159257888794,
+      "learning_rate": 9.807450230389201e-06,
+      "loss": 0.7396,
+      "step": 7123
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.4391157627105713,
+      "learning_rate": 9.807377759956169e-06,
+      "loss": 0.4728,
+      "step": 7124
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 4.717736721038818,
+      "learning_rate": 9.807305276155618e-06,
+      "loss": 0.8206,
+      "step": 7125
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.312059164047241,
+      "learning_rate": 9.807232778987754e-06,
+      "loss": 0.7636,
+      "step": 7126
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 4.2767252922058105,
+      "learning_rate": 9.807160268452775e-06,
+      "loss": 0.5845,
+      "step": 7127
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.495781183242798,
+      "learning_rate": 9.807087744550886e-06,
+      "loss": 0.8199,
+      "step": 7128
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.2280704975128174,
+      "learning_rate": 9.807015207282285e-06,
+      "loss": 0.5466,
+      "step": 7129
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.7061679363250732,
+      "learning_rate": 9.806942656647175e-06,
+      "loss": 0.8265,
+      "step": 7130
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.444880962371826,
+      "learning_rate": 9.806870092645759e-06,
+      "loss": 0.9516,
+      "step": 7131
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.2068090438842773,
+      "learning_rate": 9.806797515278239e-06,
+      "loss": 0.6262,
+      "step": 7132
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.4793384075164795,
+      "learning_rate": 9.806724924544815e-06,
+      "loss": 0.6194,
+      "step": 7133
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.2224459648132324,
+      "learning_rate": 9.80665232044569e-06,
+      "loss": 0.8196,
+      "step": 7134
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.5741257667541504,
+      "learning_rate": 9.806579702981065e-06,
+      "loss": 0.5891,
+      "step": 7135
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.839141845703125,
+      "learning_rate": 9.806507072151143e-06,
+      "loss": 0.7827,
+      "step": 7136
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.139033079147339,
+      "learning_rate": 9.806434427956123e-06,
+      "loss": 0.7038,
+      "step": 7137
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.9362916946411133,
+      "learning_rate": 9.806361770396213e-06,
+      "loss": 0.6683,
+      "step": 7138
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.3091893196105957,
+      "learning_rate": 9.806289099471608e-06,
+      "loss": 0.6939,
+      "step": 7139
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 4.327014923095703,
+      "learning_rate": 9.806216415182516e-06,
+      "loss": 0.4016,
+      "step": 7140
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.3819150924682617,
+      "learning_rate": 9.806143717529136e-06,
+      "loss": 0.7425,
+      "step": 7141
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.7231414318084717,
+      "learning_rate": 9.80607100651167e-06,
+      "loss": 0.5707,
+      "step": 7142
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.1537117958068848,
+      "learning_rate": 9.805998282130322e-06,
+      "loss": 0.7194,
+      "step": 7143
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.8476948738098145,
+      "learning_rate": 9.805925544385292e-06,
+      "loss": 0.7287,
+      "step": 7144
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.9272701740264893,
+      "learning_rate": 9.805852793276785e-06,
+      "loss": 0.8128,
+      "step": 7145
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.873307466506958,
+      "learning_rate": 9.805780028805e-06,
+      "loss": 0.6586,
+      "step": 7146
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.8499832153320312,
+      "learning_rate": 9.805707250970142e-06,
+      "loss": 0.8151,
+      "step": 7147
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 4.38194465637207,
+      "learning_rate": 9.805634459772412e-06,
+      "loss": 0.6551,
+      "step": 7148
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.8502142429351807,
+      "learning_rate": 9.805561655212012e-06,
+      "loss": 0.7549,
+      "step": 7149
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.3853859901428223,
+      "learning_rate": 9.805488837289148e-06,
+      "loss": 0.5317,
+      "step": 7150
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.9517362117767334,
+      "learning_rate": 9.805416006004017e-06,
+      "loss": 0.8513,
+      "step": 7151
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.0789759159088135,
+      "learning_rate": 9.805343161356825e-06,
+      "loss": 0.6122,
+      "step": 7152
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 4.283178329467773,
+      "learning_rate": 9.805270303347774e-06,
+      "loss": 0.759,
+      "step": 7153
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.9530832767486572,
+      "learning_rate": 9.805197431977066e-06,
+      "loss": 0.6108,
+      "step": 7154
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.2648050785064697,
+      "learning_rate": 9.805124547244905e-06,
+      "loss": 0.6515,
+      "step": 7155
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.7876296043395996,
+      "learning_rate": 9.805051649151492e-06,
+      "loss": 0.7944,
+      "step": 7156
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.0059654712677,
+      "learning_rate": 9.804978737697031e-06,
+      "loss": 0.5588,
+      "step": 7157
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.885549545288086,
+      "learning_rate": 9.804905812881723e-06,
+      "loss": 0.3426,
+      "step": 7158
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.4895100593566895,
+      "learning_rate": 9.804832874705772e-06,
+      "loss": 0.7308,
+      "step": 7159
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.7209649085998535,
+      "learning_rate": 9.80475992316938e-06,
+      "loss": 0.5682,
+      "step": 7160
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.3138766288757324,
+      "learning_rate": 9.804686958272754e-06,
+      "loss": 0.5677,
+      "step": 7161
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.7915302515029907,
+      "learning_rate": 9.804613980016088e-06,
+      "loss": 0.5154,
+      "step": 7162
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.0084338188171387,
+      "learning_rate": 9.804540988399595e-06,
+      "loss": 0.9215,
+      "step": 7163
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.8886685371398926,
+      "learning_rate": 9.804467983423472e-06,
+      "loss": 0.5359,
+      "step": 7164
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.268425941467285,
+      "learning_rate": 9.804394965087923e-06,
+      "loss": 0.742,
+      "step": 7165
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.3944740295410156,
+      "learning_rate": 9.804321933393151e-06,
+      "loss": 0.9093,
+      "step": 7166
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.2946128845214844,
+      "learning_rate": 9.80424888833936e-06,
+      "loss": 0.9218,
+      "step": 7167
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.93890380859375,
+      "learning_rate": 9.804175829926753e-06,
+      "loss": 0.6891,
+      "step": 7168
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 4.218251705169678,
+      "learning_rate": 9.804102758155532e-06,
+      "loss": 0.6669,
+      "step": 7169
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.5349862575531006,
+      "learning_rate": 9.8040296730259e-06,
+      "loss": 0.3551,
+      "step": 7170
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.697869062423706,
+      "learning_rate": 9.803956574538063e-06,
+      "loss": 0.5772,
+      "step": 7171
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.7936875820159912,
+      "learning_rate": 9.803883462692221e-06,
+      "loss": 0.5484,
+      "step": 7172
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.843194007873535,
+      "learning_rate": 9.80381033748858e-06,
+      "loss": 0.5883,
+      "step": 7173
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.0677146911621094,
+      "learning_rate": 9.803737198927341e-06,
+      "loss": 0.2144,
+      "step": 7174
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.4814627170562744,
+      "learning_rate": 9.803664047008708e-06,
+      "loss": 0.6144,
+      "step": 7175
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.58327579498291,
+      "learning_rate": 9.803590881732887e-06,
+      "loss": 0.8981,
+      "step": 7176
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.923316478729248,
+      "learning_rate": 9.803517703100076e-06,
+      "loss": 0.6256,
+      "step": 7177
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.939862012863159,
+      "learning_rate": 9.803444511110486e-06,
+      "loss": 0.5793,
+      "step": 7178
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.5472497940063477,
+      "learning_rate": 9.80337130576431e-06,
+      "loss": 0.9208,
+      "step": 7179
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.348165988922119,
+      "learning_rate": 9.803298087061764e-06,
+      "loss": 0.6193,
+      "step": 7180
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.8066654205322266,
+      "learning_rate": 9.803224855003041e-06,
+      "loss": 0.6711,
+      "step": 7181
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.2918777465820312,
+      "learning_rate": 9.80315160958835e-06,
+      "loss": 0.6215,
+      "step": 7182
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.592313766479492,
+      "learning_rate": 9.803078350817897e-06,
+      "loss": 0.5586,
+      "step": 7183
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.0261712074279785,
+      "learning_rate": 9.803005078691878e-06,
+      "loss": 0.6795,
+      "step": 7184
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.084156036376953,
+      "learning_rate": 9.802931793210504e-06,
+      "loss": 0.6752,
+      "step": 7185
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.5728397369384766,
+      "learning_rate": 9.802858494373975e-06,
+      "loss": 0.8859,
+      "step": 7186
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.017026662826538,
+      "learning_rate": 9.802785182182495e-06,
+      "loss": 0.8832,
+      "step": 7187
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.93878173828125,
+      "learning_rate": 9.802711856636268e-06,
+      "loss": 0.7693,
+      "step": 7188
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.850766181945801,
+      "learning_rate": 9.8026385177355e-06,
+      "loss": 0.569,
+      "step": 7189
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.8610825538635254,
+      "learning_rate": 9.802565165480393e-06,
+      "loss": 0.6842,
+      "step": 7190
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.7485530376434326,
+      "learning_rate": 9.80249179987115e-06,
+      "loss": 0.793,
+      "step": 7191
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.6173573732376099,
+      "learning_rate": 9.802418420907979e-06,
+      "loss": 0.4315,
+      "step": 7192
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.393728494644165,
+      "learning_rate": 9.802345028591081e-06,
+      "loss": 0.5977,
+      "step": 7193
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.484616994857788,
+      "learning_rate": 9.802271622920659e-06,
+      "loss": 0.5419,
+      "step": 7194
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.349106788635254,
+      "learning_rate": 9.802198203896918e-06,
+      "loss": 0.5189,
+      "step": 7195
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.475785493850708,
+      "learning_rate": 9.802124771520065e-06,
+      "loss": 0.7917,
+      "step": 7196
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.2111682891845703,
+      "learning_rate": 9.802051325790301e-06,
+      "loss": 0.3955,
+      "step": 7197
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.4052369594573975,
+      "learning_rate": 9.801977866707831e-06,
+      "loss": 0.6836,
+      "step": 7198
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.9677708148956299,
+      "learning_rate": 9.801904394272859e-06,
+      "loss": 0.6025,
+      "step": 7199
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.3773690462112427,
+      "learning_rate": 9.80183090848559e-06,
+      "loss": 0.3761,
+      "step": 7200
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.1991212368011475,
+      "learning_rate": 9.801757409346228e-06,
+      "loss": 0.6773,
+      "step": 7201
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.2924787998199463,
+      "learning_rate": 9.801683896854976e-06,
+      "loss": 0.5979,
+      "step": 7202
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 4.754568576812744,
+      "learning_rate": 9.801610371012042e-06,
+      "loss": 0.9658,
+      "step": 7203
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.7664856910705566,
+      "learning_rate": 9.801536831817628e-06,
+      "loss": 0.7121,
+      "step": 7204
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.9780155420303345,
+      "learning_rate": 9.80146327927194e-06,
+      "loss": 0.679,
+      "step": 7205
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.777647018432617,
+      "learning_rate": 9.801389713375177e-06,
+      "loss": 0.7771,
+      "step": 7206
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.254093647003174,
+      "learning_rate": 9.801316134127551e-06,
+      "loss": 0.5411,
+      "step": 7207
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.3415627479553223,
+      "learning_rate": 9.801242541529263e-06,
+      "loss": 0.5723,
+      "step": 7208
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.7574222087860107,
+      "learning_rate": 9.801168935580519e-06,
+      "loss": 0.8022,
+      "step": 7209
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.8843841552734375,
+      "learning_rate": 9.801095316281521e-06,
+      "loss": 0.617,
+      "step": 7210
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.316100835800171,
+      "learning_rate": 9.801021683632476e-06,
+      "loss": 0.6703,
+      "step": 7211
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.8905550241470337,
+      "learning_rate": 9.800948037633588e-06,
+      "loss": 0.6971,
+      "step": 7212
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 5.789470195770264,
+      "learning_rate": 9.800874378285063e-06,
+      "loss": 0.6675,
+      "step": 7213
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.4001002311706543,
+      "learning_rate": 9.800800705587105e-06,
+      "loss": 0.6958,
+      "step": 7214
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.803591012954712,
+      "learning_rate": 9.800727019539918e-06,
+      "loss": 0.6754,
+      "step": 7215
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.846071481704712,
+      "learning_rate": 9.800653320143708e-06,
+      "loss": 0.686,
+      "step": 7216
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.6095597743988037,
+      "learning_rate": 9.800579607398678e-06,
+      "loss": 0.8325,
+      "step": 7217
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.5144903659820557,
+      "learning_rate": 9.800505881305036e-06,
+      "loss": 0.84,
+      "step": 7218
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.526962995529175,
+      "learning_rate": 9.800432141862985e-06,
+      "loss": 0.5019,
+      "step": 7219
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.0737335681915283,
+      "learning_rate": 9.80035838907273e-06,
+      "loss": 0.9315,
+      "step": 7220
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.52913236618042,
+      "learning_rate": 9.800284622934477e-06,
+      "loss": 0.51,
+      "step": 7221
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.1104981899261475,
+      "learning_rate": 9.80021084344843e-06,
+      "loss": 0.8747,
+      "step": 7222
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 4.709591865539551,
+      "learning_rate": 9.800137050614795e-06,
+      "loss": 0.645,
+      "step": 7223
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.9527699947357178,
+      "learning_rate": 9.800063244433778e-06,
+      "loss": 0.9307,
+      "step": 7224
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.9872725009918213,
+      "learning_rate": 9.799989424905583e-06,
+      "loss": 0.6629,
+      "step": 7225
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.260364294052124,
+      "learning_rate": 9.799915592030416e-06,
+      "loss": 0.5844,
+      "step": 7226
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.4418137073516846,
+      "learning_rate": 9.799841745808481e-06,
+      "loss": 0.5537,
+      "step": 7227
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.9723148345947266,
+      "learning_rate": 9.799767886239983e-06,
+      "loss": 0.5907,
+      "step": 7228
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 4.21924352645874,
+      "learning_rate": 9.79969401332513e-06,
+      "loss": 0.4877,
+      "step": 7229
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.4924099445343018,
+      "learning_rate": 9.799620127064127e-06,
+      "loss": 0.7229,
+      "step": 7230
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.0107436180114746,
+      "learning_rate": 9.799546227457177e-06,
+      "loss": 0.5475,
+      "step": 7231
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.2372658252716064,
+      "learning_rate": 9.799472314504485e-06,
+      "loss": 0.5793,
+      "step": 7232
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.510854959487915,
+      "learning_rate": 9.799398388206263e-06,
+      "loss": 0.7659,
+      "step": 7233
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 6.730306148529053,
+      "learning_rate": 9.799324448562708e-06,
+      "loss": 0.7975,
+      "step": 7234
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.9402215480804443,
+      "learning_rate": 9.79925049557403e-06,
+      "loss": 0.758,
+      "step": 7235
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.8261682987213135,
+      "learning_rate": 9.799176529240438e-06,
+      "loss": 0.8912,
+      "step": 7236
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 4.500394821166992,
+      "learning_rate": 9.79910254956213e-06,
+      "loss": 0.8687,
+      "step": 7237
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.945726156234741,
+      "learning_rate": 9.799028556539316e-06,
+      "loss": 0.5267,
+      "step": 7238
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.5626776218414307,
+      "learning_rate": 9.798954550172202e-06,
+      "loss": 0.7858,
+      "step": 7239
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.3616998195648193,
+      "learning_rate": 9.798880530460994e-06,
+      "loss": 0.7582,
+      "step": 7240
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.729145050048828,
+      "learning_rate": 9.798806497405897e-06,
+      "loss": 0.5031,
+      "step": 7241
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.5573511123657227,
+      "learning_rate": 9.798732451007115e-06,
+      "loss": 0.8367,
+      "step": 7242
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.176833152770996,
+      "learning_rate": 9.798658391264857e-06,
+      "loss": 0.4366,
+      "step": 7243
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 4.245896339416504,
+      "learning_rate": 9.798584318179327e-06,
+      "loss": 0.6539,
+      "step": 7244
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 4.8554768562316895,
+      "learning_rate": 9.798510231750732e-06,
+      "loss": 0.5863,
+      "step": 7245
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.59660005569458,
+      "learning_rate": 9.798436131979276e-06,
+      "loss": 0.6045,
+      "step": 7246
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.335000991821289,
+      "learning_rate": 9.798362018865169e-06,
+      "loss": 0.694,
+      "step": 7247
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 5.137150764465332,
+      "learning_rate": 9.798287892408614e-06,
+      "loss": 0.7761,
+      "step": 7248
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.181267023086548,
+      "learning_rate": 9.798213752609818e-06,
+      "loss": 0.7672,
+      "step": 7249
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.6612913608551025,
+      "learning_rate": 9.798139599468986e-06,
+      "loss": 0.5495,
+      "step": 7250
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.4767727851867676,
+      "learning_rate": 9.798065432986327e-06,
+      "loss": 0.7339,
+      "step": 7251
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.6849054098129272,
+      "learning_rate": 9.797991253162045e-06,
+      "loss": 0.5399,
+      "step": 7252
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.937450885772705,
+      "learning_rate": 9.797917059996345e-06,
+      "loss": 0.7958,
+      "step": 7253
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.476872682571411,
+      "learning_rate": 9.797842853489437e-06,
+      "loss": 0.958,
+      "step": 7254
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.3088996410369873,
+      "learning_rate": 9.797768633641523e-06,
+      "loss": 1.0533,
+      "step": 7255
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 4.205419063568115,
+      "learning_rate": 9.797694400452813e-06,
+      "loss": 0.9327,
+      "step": 7256
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.2838850021362305,
+      "learning_rate": 9.797620153923512e-06,
+      "loss": 0.5795,
+      "step": 7257
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.951317071914673,
+      "learning_rate": 9.797545894053829e-06,
+      "loss": 0.8455,
+      "step": 7258
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 4.552724838256836,
+      "learning_rate": 9.797471620843964e-06,
+      "loss": 0.5199,
+      "step": 7259
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.4352962970733643,
+      "learning_rate": 9.79739733429413e-06,
+      "loss": 0.5133,
+      "step": 7260
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.674475908279419,
+      "learning_rate": 9.797323034404532e-06,
+      "loss": 0.871,
+      "step": 7261
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 5.881767272949219,
+      "learning_rate": 9.797248721175372e-06,
+      "loss": 0.6769,
+      "step": 7262
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.369842767715454,
+      "learning_rate": 9.797174394606864e-06,
+      "loss": 0.7008,
+      "step": 7263
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 4.206986427307129,
+      "learning_rate": 9.797100054699209e-06,
+      "loss": 0.8824,
+      "step": 7264
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.9537158012390137,
+      "learning_rate": 9.797025701452618e-06,
+      "loss": 0.8309,
+      "step": 7265
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.2191104888916016,
+      "learning_rate": 9.796951334867293e-06,
+      "loss": 0.5591,
+      "step": 7266
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.8500038385391235,
+      "learning_rate": 9.796876954943446e-06,
+      "loss": 0.7873,
+      "step": 7267
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.111098289489746,
+      "learning_rate": 9.796802561681279e-06,
+      "loss": 0.7151,
+      "step": 7268
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.600609540939331,
+      "learning_rate": 9.796728155081001e-06,
+      "loss": 0.7619,
+      "step": 7269
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.8230764865875244,
+      "learning_rate": 9.796653735142821e-06,
+      "loss": 0.6442,
+      "step": 7270
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.6148691177368164,
+      "learning_rate": 9.796579301866941e-06,
+      "loss": 0.6169,
+      "step": 7271
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.7809405326843262,
+      "learning_rate": 9.796504855253573e-06,
+      "loss": 0.5848,
+      "step": 7272
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.577718734741211,
+      "learning_rate": 9.796430395302921e-06,
+      "loss": 0.7918,
+      "step": 7273
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.376962423324585,
+      "learning_rate": 9.796355922015192e-06,
+      "loss": 0.5401,
+      "step": 7274
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.241492509841919,
+      "learning_rate": 9.796281435390596e-06,
+      "loss": 0.894,
+      "step": 7275
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.511307716369629,
+      "learning_rate": 9.796206935429335e-06,
+      "loss": 0.3599,
+      "step": 7276
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.0859310626983643,
+      "learning_rate": 9.79613242213162e-06,
+      "loss": 0.6309,
+      "step": 7277
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.4826102256774902,
+      "learning_rate": 9.79605789549766e-06,
+      "loss": 0.7046,
+      "step": 7278
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.3164379596710205,
+      "learning_rate": 9.795983355527659e-06,
+      "loss": 0.6061,
+      "step": 7279
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.9072556495666504,
+      "learning_rate": 9.795908802221824e-06,
+      "loss": 0.9665,
+      "step": 7280
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.585387706756592,
+      "learning_rate": 9.795834235580362e-06,
+      "loss": 0.5408,
+      "step": 7281
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.6396005153656006,
+      "learning_rate": 9.795759655603482e-06,
+      "loss": 0.9222,
+      "step": 7282
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.308809280395508,
+      "learning_rate": 9.795685062291393e-06,
+      "loss": 0.7328,
+      "step": 7283
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.5274412631988525,
+      "learning_rate": 9.7956104556443e-06,
+      "loss": 0.7824,
+      "step": 7284
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.751554012298584,
+      "learning_rate": 9.795535835662408e-06,
+      "loss": 0.709,
+      "step": 7285
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.2196364402770996,
+      "learning_rate": 9.79546120234593e-06,
+      "loss": 0.7275,
+      "step": 7286
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.0870232582092285,
+      "learning_rate": 9.79538655569507e-06,
+      "loss": 0.6738,
+      "step": 7287
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.068866729736328,
+      "learning_rate": 9.795311895710036e-06,
+      "loss": 0.7919,
+      "step": 7288
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.424344062805176,
+      "learning_rate": 9.795237222391037e-06,
+      "loss": 0.6168,
+      "step": 7289
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.833613634109497,
+      "learning_rate": 9.795162535738278e-06,
+      "loss": 0.8731,
+      "step": 7290
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 4.098442554473877,
+      "learning_rate": 9.79508783575197e-06,
+      "loss": 0.725,
+      "step": 7291
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.8259824514389038,
+      "learning_rate": 9.795013122432318e-06,
+      "loss": 0.7986,
+      "step": 7292
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.293717861175537,
+      "learning_rate": 9.794938395779532e-06,
+      "loss": 0.7095,
+      "step": 7293
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.5458192825317383,
+      "learning_rate": 9.794863655793816e-06,
+      "loss": 0.7514,
+      "step": 7294
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.6338567733764648,
+      "learning_rate": 9.794788902475382e-06,
+      "loss": 0.6619,
+      "step": 7295
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.6471035480499268,
+      "learning_rate": 9.794714135824436e-06,
+      "loss": 0.9177,
+      "step": 7296
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.3193538188934326,
+      "learning_rate": 9.794639355841187e-06,
+      "loss": 0.7532,
+      "step": 7297
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.2670297622680664,
+      "learning_rate": 9.794564562525842e-06,
+      "loss": 0.8065,
+      "step": 7298
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.7581069469451904,
+      "learning_rate": 9.794489755878606e-06,
+      "loss": 0.7591,
+      "step": 7299
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 7.793868541717529,
+      "learning_rate": 9.794414935899694e-06,
+      "loss": 0.9211,
+      "step": 7300
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.109072685241699,
+      "learning_rate": 9.794340102589308e-06,
+      "loss": 0.634,
+      "step": 7301
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.558032751083374,
+      "learning_rate": 9.794265255947659e-06,
+      "loss": 0.61,
+      "step": 7302
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 4.637415409088135,
+      "learning_rate": 9.794190395974955e-06,
+      "loss": 0.6158,
+      "step": 7303
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.6574127674102783,
+      "learning_rate": 9.794115522671402e-06,
+      "loss": 0.6995,
+      "step": 7304
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.9316298961639404,
+      "learning_rate": 9.79404063603721e-06,
+      "loss": 0.8387,
+      "step": 7305
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.2662532329559326,
+      "learning_rate": 9.793965736072586e-06,
+      "loss": 0.6251,
+      "step": 7306
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.4650328159332275,
+      "learning_rate": 9.793890822777741e-06,
+      "loss": 0.8783,
+      "step": 7307
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.188904285430908,
+      "learning_rate": 9.793815896152881e-06,
+      "loss": 0.6961,
+      "step": 7308
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.1906867027282715,
+      "learning_rate": 9.793740956198213e-06,
+      "loss": 0.5513,
+      "step": 7309
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.021026849746704,
+      "learning_rate": 9.793666002913949e-06,
+      "loss": 0.5175,
+      "step": 7310
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.146517753601074,
+      "learning_rate": 9.793591036300296e-06,
+      "loss": 0.6446,
+      "step": 7311
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.6785333156585693,
+      "learning_rate": 9.79351605635746e-06,
+      "loss": 0.6025,
+      "step": 7312
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.1978492736816406,
+      "learning_rate": 9.793441063085654e-06,
+      "loss": 0.741,
+      "step": 7313
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.382594347000122,
+      "learning_rate": 9.793366056485082e-06,
+      "loss": 0.4429,
+      "step": 7314
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 4.139888286590576,
+      "learning_rate": 9.793291036555955e-06,
+      "loss": 0.7778,
+      "step": 7315
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.5183275938034058,
+      "learning_rate": 9.793216003298482e-06,
+      "loss": 0.4916,
+      "step": 7316
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.8077123165130615,
+      "learning_rate": 9.793140956712869e-06,
+      "loss": 0.6733,
+      "step": 7317
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.456178665161133,
+      "learning_rate": 9.79306589679933e-06,
+      "loss": 0.7115,
+      "step": 7318
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.5035555362701416,
+      "learning_rate": 9.792990823558068e-06,
+      "loss": 0.5709,
+      "step": 7319
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 4.006189823150635,
+      "learning_rate": 9.792915736989295e-06,
+      "loss": 0.4989,
+      "step": 7320
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.4504058361053467,
+      "learning_rate": 9.792840637093217e-06,
+      "loss": 0.9551,
+      "step": 7321
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.598517894744873,
+      "learning_rate": 9.792765523870047e-06,
+      "loss": 0.6625,
+      "step": 7322
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.206448554992676,
+      "learning_rate": 9.792690397319989e-06,
+      "loss": 0.5462,
+      "step": 7323
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.9711878299713135,
+      "learning_rate": 9.792615257443256e-06,
+      "loss": 0.6264,
+      "step": 7324
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.9803416728973389,
+      "learning_rate": 9.792540104240056e-06,
+      "loss": 0.7629,
+      "step": 7325
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.905200958251953,
+      "learning_rate": 9.792464937710595e-06,
+      "loss": 0.6905,
+      "step": 7326
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.270113468170166,
+      "learning_rate": 9.792389757855085e-06,
+      "loss": 0.8668,
+      "step": 7327
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.295624017715454,
+      "learning_rate": 9.792314564673736e-06,
+      "loss": 0.6175,
+      "step": 7328
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.8690788745880127,
+      "learning_rate": 9.792239358166754e-06,
+      "loss": 0.5977,
+      "step": 7329
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 4.183809757232666,
+      "learning_rate": 9.79216413833435e-06,
+      "loss": 0.7518,
+      "step": 7330
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.338395357131958,
+      "learning_rate": 9.792088905176734e-06,
+      "loss": 0.6831,
+      "step": 7331
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.832000494003296,
+      "learning_rate": 9.792013658694112e-06,
+      "loss": 0.6086,
+      "step": 7332
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.7417373657226562,
+      "learning_rate": 9.791938398886696e-06,
+      "loss": 0.8511,
+      "step": 7333
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.8841772079467773,
+      "learning_rate": 9.791863125754693e-06,
+      "loss": 0.5182,
+      "step": 7334
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.296555280685425,
+      "learning_rate": 9.791787839298314e-06,
+      "loss": 0.8254,
+      "step": 7335
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.278265953063965,
+      "learning_rate": 9.79171253951777e-06,
+      "loss": 0.7702,
+      "step": 7336
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.9733798503875732,
+      "learning_rate": 9.791637226413268e-06,
+      "loss": 0.74,
+      "step": 7337
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.009016990661621,
+      "learning_rate": 9.791561899985017e-06,
+      "loss": 0.6282,
+      "step": 7338
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.9041025638580322,
+      "learning_rate": 9.791486560233227e-06,
+      "loss": 0.7735,
+      "step": 7339
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.881369113922119,
+      "learning_rate": 9.791411207158109e-06,
+      "loss": 0.7068,
+      "step": 7340
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.5649060010910034,
+      "learning_rate": 9.79133584075987e-06,
+      "loss": 0.6946,
+      "step": 7341
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.2737393379211426,
+      "learning_rate": 9.791260461038721e-06,
+      "loss": 0.806,
+      "step": 7342
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 4.372560024261475,
+      "learning_rate": 9.791185067994873e-06,
+      "loss": 0.8071,
+      "step": 7343
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.9718929529190063,
+      "learning_rate": 9.791109661628533e-06,
+      "loss": 0.5243,
+      "step": 7344
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.767817258834839,
+      "learning_rate": 9.791034241939913e-06,
+      "loss": 0.6449,
+      "step": 7345
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.8812865018844604,
+      "learning_rate": 9.790958808929221e-06,
+      "loss": 0.3784,
+      "step": 7346
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.6963915824890137,
+      "learning_rate": 9.790883362596666e-06,
+      "loss": 0.5096,
+      "step": 7347
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.634212017059326,
+      "learning_rate": 9.79080790294246e-06,
+      "loss": 0.5038,
+      "step": 7348
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.6928497552871704,
+      "learning_rate": 9.790732429966811e-06,
+      "loss": 0.5796,
+      "step": 7349
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.4656333923339844,
+      "learning_rate": 9.790656943669932e-06,
+      "loss": 0.5173,
+      "step": 7350
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.218289852142334,
+      "learning_rate": 9.790581444052029e-06,
+      "loss": 0.7094,
+      "step": 7351
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.301715135574341,
+      "learning_rate": 9.790505931113313e-06,
+      "loss": 0.4098,
+      "step": 7352
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.5253148078918457,
+      "learning_rate": 9.790430404853996e-06,
+      "loss": 0.8095,
+      "step": 7353
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 4.782609939575195,
+      "learning_rate": 9.790354865274286e-06,
+      "loss": 0.6378,
+      "step": 7354
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.001030683517456,
+      "learning_rate": 9.790279312374393e-06,
+      "loss": 0.9,
+      "step": 7355
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.783369541168213,
+      "learning_rate": 9.790203746154528e-06,
+      "loss": 0.693,
+      "step": 7356
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 5.409219741821289,
+      "learning_rate": 9.790128166614902e-06,
+      "loss": 1.0151,
+      "step": 7357
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.885688066482544,
+      "learning_rate": 9.790052573755722e-06,
+      "loss": 0.864,
+      "step": 7358
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.5315823554992676,
+      "learning_rate": 9.789976967577203e-06,
+      "loss": 0.7639,
+      "step": 7359
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.0133700370788574,
+      "learning_rate": 9.78990134807955e-06,
+      "loss": 0.7419,
+      "step": 7360
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.3594846725463867,
+      "learning_rate": 9.789825715262977e-06,
+      "loss": 0.6244,
+      "step": 7361
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.729742169380188,
+      "learning_rate": 9.789750069127693e-06,
+      "loss": 0.6221,
+      "step": 7362
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.7814524173736572,
+      "learning_rate": 9.789674409673906e-06,
+      "loss": 0.6647,
+      "step": 7363
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.289201021194458,
+      "learning_rate": 9.789598736901832e-06,
+      "loss": 0.7462,
+      "step": 7364
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.3268063068389893,
+      "learning_rate": 9.789523050811675e-06,
+      "loss": 0.7086,
+      "step": 7365
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.3685715198516846,
+      "learning_rate": 9.789447351403651e-06,
+      "loss": 0.818,
+      "step": 7366
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.417041540145874,
+      "learning_rate": 9.789371638677967e-06,
+      "loss": 0.64,
+      "step": 7367
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.9740424156188965,
+      "learning_rate": 9.789295912634835e-06,
+      "loss": 0.5758,
+      "step": 7368
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.205576181411743,
+      "learning_rate": 9.789220173274465e-06,
+      "loss": 0.5109,
+      "step": 7369
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.1745617389678955,
+      "learning_rate": 9.789144420597069e-06,
+      "loss": 0.7719,
+      "step": 7370
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.6448782682418823,
+      "learning_rate": 9.789068654602855e-06,
+      "loss": 0.5842,
+      "step": 7371
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.7182449102401733,
+      "learning_rate": 9.788992875292035e-06,
+      "loss": 0.6852,
+      "step": 7372
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.706498622894287,
+      "learning_rate": 9.788917082664822e-06,
+      "loss": 0.8272,
+      "step": 7373
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.7294875383377075,
+      "learning_rate": 9.788841276721421e-06,
+      "loss": 0.6631,
+      "step": 7374
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.3685081005096436,
+      "learning_rate": 9.788765457462049e-06,
+      "loss": 0.8255,
+      "step": 7375
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.720353364944458,
+      "learning_rate": 9.788689624886913e-06,
+      "loss": 0.5692,
+      "step": 7376
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.1788084506988525,
+      "learning_rate": 9.788613778996225e-06,
+      "loss": 0.2912,
+      "step": 7377
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.397024154663086,
+      "learning_rate": 9.788537919790198e-06,
+      "loss": 0.7592,
+      "step": 7378
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.162006139755249,
+      "learning_rate": 9.788462047269038e-06,
+      "loss": 0.6985,
+      "step": 7379
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.9578404426574707,
+      "learning_rate": 9.78838616143296e-06,
+      "loss": 0.7735,
+      "step": 7380
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.791429281234741,
+      "learning_rate": 9.788310262282175e-06,
+      "loss": 0.741,
+      "step": 7381
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.6290078163146973,
+      "learning_rate": 9.788234349816891e-06,
+      "loss": 0.7277,
+      "step": 7382
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.2380433082580566,
+      "learning_rate": 9.788158424037322e-06,
+      "loss": 0.786,
+      "step": 7383
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.9606646299362183,
+      "learning_rate": 9.788082484943677e-06,
+      "loss": 0.5506,
+      "step": 7384
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.1900455951690674,
+      "learning_rate": 9.78800653253617e-06,
+      "loss": 0.5457,
+      "step": 7385
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.1278305053710938,
+      "learning_rate": 9.78793056681501e-06,
+      "loss": 0.4078,
+      "step": 7386
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.861875057220459,
+      "learning_rate": 9.787854587780408e-06,
+      "loss": 0.5296,
+      "step": 7387
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.72619891166687,
+      "learning_rate": 9.787778595432576e-06,
+      "loss": 0.6289,
+      "step": 7388
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.1101503372192383,
+      "learning_rate": 9.787702589771726e-06,
+      "loss": 0.615,
+      "step": 7389
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 4.468290328979492,
+      "learning_rate": 9.787626570798068e-06,
+      "loss": 0.7414,
+      "step": 7390
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.227215528488159,
+      "learning_rate": 9.787550538511814e-06,
+      "loss": 0.6243,
+      "step": 7391
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.5131053924560547,
+      "learning_rate": 9.787474492913176e-06,
+      "loss": 0.604,
+      "step": 7392
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.259465217590332,
+      "learning_rate": 9.787398434002364e-06,
+      "loss": 0.6654,
+      "step": 7393
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.397928476333618,
+      "learning_rate": 9.78732236177959e-06,
+      "loss": 0.6151,
+      "step": 7394
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 6.7900872230529785,
+      "learning_rate": 9.787246276245067e-06,
+      "loss": 0.6922,
+      "step": 7395
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.532179594039917,
+      "learning_rate": 9.787170177399006e-06,
+      "loss": 0.837,
+      "step": 7396
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.5988543033599854,
+      "learning_rate": 9.787094065241615e-06,
+      "loss": 0.5117,
+      "step": 7397
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.2041175365448,
+      "learning_rate": 9.787017939773112e-06,
+      "loss": 0.6412,
+      "step": 7398
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.1176352500915527,
+      "learning_rate": 9.786941800993705e-06,
+      "loss": 0.7047,
+      "step": 7399
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 6.1097822189331055,
+      "learning_rate": 9.786865648903605e-06,
+      "loss": 0.6669,
+      "step": 7400
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.480926513671875,
+      "learning_rate": 9.786789483503026e-06,
+      "loss": 0.8596,
+      "step": 7401
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.6145784854888916,
+      "learning_rate": 9.786713304792176e-06,
+      "loss": 0.6933,
+      "step": 7402
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.221074342727661,
+      "learning_rate": 9.786637112771271e-06,
+      "loss": 0.5231,
+      "step": 7403
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.8136749267578125,
+      "learning_rate": 9.786560907440523e-06,
+      "loss": 0.6853,
+      "step": 7404
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.7742772102355957,
+      "learning_rate": 9.78648468880014e-06,
+      "loss": 0.8432,
+      "step": 7405
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.920757532119751,
+      "learning_rate": 9.786408456850336e-06,
+      "loss": 0.6276,
+      "step": 7406
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.7339296340942383,
+      "learning_rate": 9.786332211591325e-06,
+      "loss": 0.582,
+      "step": 7407
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.2958245277404785,
+      "learning_rate": 9.786255953023317e-06,
+      "loss": 0.813,
+      "step": 7408
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.958211898803711,
+      "learning_rate": 9.786179681146522e-06,
+      "loss": 0.4575,
+      "step": 7409
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.7413145303726196,
+      "learning_rate": 9.786103395961154e-06,
+      "loss": 0.6632,
+      "step": 7410
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.461963415145874,
+      "learning_rate": 9.786027097467427e-06,
+      "loss": 0.6609,
+      "step": 7411
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 5.938239574432373,
+      "learning_rate": 9.785950785665552e-06,
+      "loss": 0.5746,
+      "step": 7412
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.06426739692688,
+      "learning_rate": 9.785874460555742e-06,
+      "loss": 0.6851,
+      "step": 7413
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.821960687637329,
+      "learning_rate": 9.785798122138205e-06,
+      "loss": 0.464,
+      "step": 7414
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.635783314704895,
+      "learning_rate": 9.785721770413157e-06,
+      "loss": 0.585,
+      "step": 7415
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.8115034103393555,
+      "learning_rate": 9.78564540538081e-06,
+      "loss": 0.5403,
+      "step": 7416
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.408121347427368,
+      "learning_rate": 9.785569027041375e-06,
+      "loss": 0.5761,
+      "step": 7417
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.714609146118164,
+      "learning_rate": 9.785492635395067e-06,
+      "loss": 0.7018,
+      "step": 7418
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.8858637809753418,
+      "learning_rate": 9.785416230442095e-06,
+      "loss": 0.7632,
+      "step": 7419
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.5844826698303223,
+      "learning_rate": 9.785339812182674e-06,
+      "loss": 0.4332,
+      "step": 7420
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.4391655921936035,
+      "learning_rate": 9.785263380617014e-06,
+      "loss": 0.58,
+      "step": 7421
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.5025131702423096,
+      "learning_rate": 9.785186935745333e-06,
+      "loss": 0.6362,
+      "step": 7422
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.1990301609039307,
+      "learning_rate": 9.785110477567836e-06,
+      "loss": 0.7375,
+      "step": 7423
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.6752785444259644,
+      "learning_rate": 9.78503400608474e-06,
+      "loss": 0.5439,
+      "step": 7424
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 5.838561058044434,
+      "learning_rate": 9.784957521296257e-06,
+      "loss": 0.6277,
+      "step": 7425
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.908050537109375,
+      "learning_rate": 9.784881023202602e-06,
+      "loss": 0.6729,
+      "step": 7426
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.309220314025879,
+      "learning_rate": 9.784804511803982e-06,
+      "loss": 0.5038,
+      "step": 7427
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.2263052463531494,
+      "learning_rate": 9.784727987100614e-06,
+      "loss": 0.5831,
+      "step": 7428
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.2140214443206787,
+      "learning_rate": 9.78465144909271e-06,
+      "loss": 0.9966,
+      "step": 7429
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.502368211746216,
+      "learning_rate": 9.784574897780483e-06,
+      "loss": 0.6071,
+      "step": 7430
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.8957210779190063,
+      "learning_rate": 9.784498333164146e-06,
+      "loss": 0.4512,
+      "step": 7431
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 5.903812408447266,
+      "learning_rate": 9.784421755243912e-06,
+      "loss": 0.6498,
+      "step": 7432
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 4.433905601501465,
+      "learning_rate": 9.784345164019992e-06,
+      "loss": 0.6148,
+      "step": 7433
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.657163381576538,
+      "learning_rate": 9.7842685594926e-06,
+      "loss": 0.7699,
+      "step": 7434
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 4.9655375480651855,
+      "learning_rate": 9.784191941661951e-06,
+      "loss": 0.8132,
+      "step": 7435
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.2121834754943848,
+      "learning_rate": 9.784115310528255e-06,
+      "loss": 0.561,
+      "step": 7436
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.3690109252929688,
+      "learning_rate": 9.784038666091728e-06,
+      "loss": 0.8558,
+      "step": 7437
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.0251779556274414,
+      "learning_rate": 9.78396200835258e-06,
+      "loss": 0.7265,
+      "step": 7438
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.824256181716919,
+      "learning_rate": 9.783885337311027e-06,
+      "loss": 0.5297,
+      "step": 7439
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.104525566101074,
+      "learning_rate": 9.78380865296728e-06,
+      "loss": 0.5757,
+      "step": 7440
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.211613416671753,
+      "learning_rate": 9.783731955321553e-06,
+      "loss": 0.5605,
+      "step": 7441
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.9968814849853516,
+      "learning_rate": 9.783655244374061e-06,
+      "loss": 0.7834,
+      "step": 7442
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.615251302719116,
+      "learning_rate": 9.783578520125015e-06,
+      "loss": 0.6872,
+      "step": 7443
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.230182647705078,
+      "learning_rate": 9.783501782574627e-06,
+      "loss": 0.7447,
+      "step": 7444
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.9343414306640625,
+      "learning_rate": 9.783425031723115e-06,
+      "loss": 0.552,
+      "step": 7445
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.417168378829956,
+      "learning_rate": 9.783348267570689e-06,
+      "loss": 0.5024,
+      "step": 7446
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.823106050491333,
+      "learning_rate": 9.783271490117565e-06,
+      "loss": 0.6136,
+      "step": 7447
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.0821099281311035,
+      "learning_rate": 9.783194699363952e-06,
+      "loss": 0.8687,
+      "step": 7448
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.132096767425537,
+      "learning_rate": 9.783117895310069e-06,
+      "loss": 0.599,
+      "step": 7449
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.166761875152588,
+      "learning_rate": 9.783041077956127e-06,
+      "loss": 0.6303,
+      "step": 7450
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.391310691833496,
+      "learning_rate": 9.782964247302337e-06,
+      "loss": 0.9517,
+      "step": 7451
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.071061611175537,
+      "learning_rate": 9.782887403348916e-06,
+      "loss": 0.5793,
+      "step": 7452
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.5604419708251953,
+      "learning_rate": 9.782810546096078e-06,
+      "loss": 0.7063,
+      "step": 7453
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.6691603660583496,
+      "learning_rate": 9.782733675544035e-06,
+      "loss": 0.6646,
+      "step": 7454
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.1959116458892822,
+      "learning_rate": 9.782656791693e-06,
+      "loss": 0.8733,
+      "step": 7455
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.8474647998809814,
+      "learning_rate": 9.782579894543189e-06,
+      "loss": 0.5228,
+      "step": 7456
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 25.5607967376709,
+      "learning_rate": 9.782502984094817e-06,
+      "loss": 0.5533,
+      "step": 7457
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.265592336654663,
+      "learning_rate": 9.782426060348093e-06,
+      "loss": 0.8623,
+      "step": 7458
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 4.214865207672119,
+      "learning_rate": 9.782349123303233e-06,
+      "loss": 0.5892,
+      "step": 7459
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.5861048698425293,
+      "learning_rate": 9.782272172960455e-06,
+      "loss": 0.8845,
+      "step": 7460
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.25632905960083,
+      "learning_rate": 9.782195209319967e-06,
+      "loss": 0.8309,
+      "step": 7461
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 5.253982067108154,
+      "learning_rate": 9.782118232381986e-06,
+      "loss": 0.6687,
+      "step": 7462
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.0248565673828125,
+      "learning_rate": 9.782041242146727e-06,
+      "loss": 0.5401,
+      "step": 7463
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.249356508255005,
+      "learning_rate": 9.781964238614401e-06,
+      "loss": 0.7147,
+      "step": 7464
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 5.925128936767578,
+      "learning_rate": 9.781887221785225e-06,
+      "loss": 0.4938,
+      "step": 7465
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.0781736373901367,
+      "learning_rate": 9.781810191659411e-06,
+      "loss": 0.5326,
+      "step": 7466
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.4757423400878906,
+      "learning_rate": 9.781733148237175e-06,
+      "loss": 0.7038,
+      "step": 7467
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.119798183441162,
+      "learning_rate": 9.78165609151873e-06,
+      "loss": 0.8563,
+      "step": 7468
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.3897523880004883,
+      "learning_rate": 9.78157902150429e-06,
+      "loss": 0.4887,
+      "step": 7469
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 4.231012344360352,
+      "learning_rate": 9.78150193819407e-06,
+      "loss": 0.5964,
+      "step": 7470
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.8090620040893555,
+      "learning_rate": 9.781424841588285e-06,
+      "loss": 0.7991,
+      "step": 7471
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.951362371444702,
+      "learning_rate": 9.781347731687149e-06,
+      "loss": 0.5406,
+      "step": 7472
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.822113275527954,
+      "learning_rate": 9.781270608490877e-06,
+      "loss": 0.2222,
+      "step": 7473
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.5169737339019775,
+      "learning_rate": 9.78119347199968e-06,
+      "loss": 0.3974,
+      "step": 7474
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.962068796157837,
+      "learning_rate": 9.781116322213777e-06,
+      "loss": 0.7185,
+      "step": 7475
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.5137062072753906,
+      "learning_rate": 9.78103915913338e-06,
+      "loss": 0.9668,
+      "step": 7476
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.5167016983032227,
+      "learning_rate": 9.780961982758703e-06,
+      "loss": 0.4508,
+      "step": 7477
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.5911378860473633,
+      "learning_rate": 9.780884793089963e-06,
+      "loss": 0.6811,
+      "step": 7478
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.8862242698669434,
+      "learning_rate": 9.780807590127372e-06,
+      "loss": 0.6576,
+      "step": 7479
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.8514257669448853,
+      "learning_rate": 9.780730373871147e-06,
+      "loss": 0.6683,
+      "step": 7480
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.9570776224136353,
+      "learning_rate": 9.780653144321503e-06,
+      "loss": 0.5272,
+      "step": 7481
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.496849775314331,
+      "learning_rate": 9.780575901478652e-06,
+      "loss": 0.6697,
+      "step": 7482
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.6160597801208496,
+      "learning_rate": 9.78049864534281e-06,
+      "loss": 0.8611,
+      "step": 7483
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.1091690063476562,
+      "learning_rate": 9.780421375914194e-06,
+      "loss": 0.4141,
+      "step": 7484
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.978627920150757,
+      "learning_rate": 9.780344093193015e-06,
+      "loss": 0.6266,
+      "step": 7485
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.0535807609558105,
+      "learning_rate": 9.780266797179489e-06,
+      "loss": 0.5937,
+      "step": 7486
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.4191219806671143,
+      "learning_rate": 9.780189487873835e-06,
+      "loss": 0.6711,
+      "step": 7487
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 5.461769104003906,
+      "learning_rate": 9.780112165276261e-06,
+      "loss": 0.5509,
+      "step": 7488
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.980301856994629,
+      "learning_rate": 9.780034829386989e-06,
+      "loss": 0.8871,
+      "step": 7489
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.2051005363464355,
+      "learning_rate": 9.779957480206227e-06,
+      "loss": 0.705,
+      "step": 7490
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.4251556396484375,
+      "learning_rate": 9.779880117734198e-06,
+      "loss": 0.9084,
+      "step": 7491
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.527507781982422,
+      "learning_rate": 9.779802741971112e-06,
+      "loss": 0.7776,
+      "step": 7492
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.290882110595703,
+      "learning_rate": 9.779725352917183e-06,
+      "loss": 0.6112,
+      "step": 7493
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.841724395751953,
+      "learning_rate": 9.77964795057263e-06,
+      "loss": 0.7649,
+      "step": 7494
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.96899151802063,
+      "learning_rate": 9.779570534937666e-06,
+      "loss": 0.5716,
+      "step": 7495
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.7301182746887207,
+      "learning_rate": 9.779493106012507e-06,
+      "loss": 0.9571,
+      "step": 7496
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.8921895027160645,
+      "learning_rate": 9.779415663797369e-06,
+      "loss": 0.5886,
+      "step": 7497
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 4.071241855621338,
+      "learning_rate": 9.779338208292465e-06,
+      "loss": 0.811,
+      "step": 7498
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.6010711193084717,
+      "learning_rate": 9.779260739498012e-06,
+      "loss": 0.6998,
+      "step": 7499
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.5680696964263916,
+      "learning_rate": 9.779183257414227e-06,
+      "loss": 0.7547,
+      "step": 7500
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.4736340045928955,
+      "learning_rate": 9.77910576204132e-06,
+      "loss": 0.6591,
+      "step": 7501
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.670401096343994,
+      "learning_rate": 9.779028253379514e-06,
+      "loss": 0.7804,
+      "step": 7502
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.113215208053589,
+      "learning_rate": 9.778950731429019e-06,
+      "loss": 0.7717,
+      "step": 7503
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.8197919130325317,
+      "learning_rate": 9.778873196190052e-06,
+      "loss": 0.7282,
+      "step": 7504
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.7148913145065308,
+      "learning_rate": 9.778795647662828e-06,
+      "loss": 0.9791,
+      "step": 7505
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 4.1785783767700195,
+      "learning_rate": 9.778718085847566e-06,
+      "loss": 0.6556,
+      "step": 7506
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.9742978811264038,
+      "learning_rate": 9.778640510744478e-06,
+      "loss": 0.6509,
+      "step": 7507
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 5.693828582763672,
+      "learning_rate": 9.778562922353781e-06,
+      "loss": 0.8641,
+      "step": 7508
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.03273606300354,
+      "learning_rate": 9.77848532067569e-06,
+      "loss": 0.8914,
+      "step": 7509
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.6090517044067383,
+      "learning_rate": 9.778407705710422e-06,
+      "loss": 0.7383,
+      "step": 7510
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.311795234680176,
+      "learning_rate": 9.778330077458191e-06,
+      "loss": 0.6199,
+      "step": 7511
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.007774591445923,
+      "learning_rate": 9.778252435919216e-06,
+      "loss": 0.5413,
+      "step": 7512
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.5081422328948975,
+      "learning_rate": 9.77817478109371e-06,
+      "loss": 0.5006,
+      "step": 7513
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.5399246215820312,
+      "learning_rate": 9.77809711298189e-06,
+      "loss": 0.7275,
+      "step": 7514
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.0261614322662354,
+      "learning_rate": 9.778019431583972e-06,
+      "loss": 0.6958,
+      "step": 7515
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.9648414850234985,
+      "learning_rate": 9.777941736900172e-06,
+      "loss": 0.7165,
+      "step": 7516
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.4317190647125244,
+      "learning_rate": 9.777864028930705e-06,
+      "loss": 0.9169,
+      "step": 7517
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.6621286869049072,
+      "learning_rate": 9.777786307675787e-06,
+      "loss": 0.3533,
+      "step": 7518
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.009432315826416,
+      "learning_rate": 9.777708573135638e-06,
+      "loss": 0.5081,
+      "step": 7519
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.3672029972076416,
+      "learning_rate": 9.777630825310469e-06,
+      "loss": 0.7203,
+      "step": 7520
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.031829595565796,
+      "learning_rate": 9.777553064200499e-06,
+      "loss": 0.3706,
+      "step": 7521
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.1019582748413086,
+      "learning_rate": 9.777475289805944e-06,
+      "loss": 0.5305,
+      "step": 7522
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.915730595588684,
+      "learning_rate": 9.777397502127019e-06,
+      "loss": 0.613,
+      "step": 7523
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.891251564025879,
+      "learning_rate": 9.777319701163944e-06,
+      "loss": 0.8067,
+      "step": 7524
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 5.838580131530762,
+      "learning_rate": 9.777241886916929e-06,
+      "loss": 0.494,
+      "step": 7525
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.647080183029175,
+      "learning_rate": 9.777164059386195e-06,
+      "loss": 0.6659,
+      "step": 7526
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.6503467559814453,
+      "learning_rate": 9.777086218571958e-06,
+      "loss": 0.679,
+      "step": 7527
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.1849052906036377,
+      "learning_rate": 9.777008364474433e-06,
+      "loss": 0.7855,
+      "step": 7528
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.0686490535736084,
+      "learning_rate": 9.776930497093838e-06,
+      "loss": 0.7893,
+      "step": 7529
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.0586419105529785,
+      "learning_rate": 9.776852616430387e-06,
+      "loss": 0.9434,
+      "step": 7530
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.99105167388916,
+      "learning_rate": 9.776774722484301e-06,
+      "loss": 0.5983,
+      "step": 7531
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.9219367504119873,
+      "learning_rate": 9.776696815255793e-06,
+      "loss": 0.641,
+      "step": 7532
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.9125545024871826,
+      "learning_rate": 9.77661889474508e-06,
+      "loss": 0.6176,
+      "step": 7533
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.971606492996216,
+      "learning_rate": 9.776540960952378e-06,
+      "loss": 0.6969,
+      "step": 7534
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.071565628051758,
+      "learning_rate": 9.776463013877907e-06,
+      "loss": 0.848,
+      "step": 7535
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.345099687576294,
+      "learning_rate": 9.776385053521882e-06,
+      "loss": 0.5137,
+      "step": 7536
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.713810920715332,
+      "learning_rate": 9.776307079884517e-06,
+      "loss": 0.7647,
+      "step": 7537
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.2314951419830322,
+      "learning_rate": 9.776229092966034e-06,
+      "loss": 0.3705,
+      "step": 7538
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.095036029815674,
+      "learning_rate": 9.776151092766646e-06,
+      "loss": 0.5179,
+      "step": 7539
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.4014276266098022,
+      "learning_rate": 9.77607307928657e-06,
+      "loss": 0.7508,
+      "step": 7540
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.715337038040161,
+      "learning_rate": 9.775995052526023e-06,
+      "loss": 0.5483,
+      "step": 7541
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.1796247959136963,
+      "learning_rate": 9.775917012485227e-06,
+      "loss": 0.6114,
+      "step": 7542
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.437361478805542,
+      "learning_rate": 9.775838959164391e-06,
+      "loss": 0.4902,
+      "step": 7543
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.213818073272705,
+      "learning_rate": 9.775760892563737e-06,
+      "loss": 0.4919,
+      "step": 7544
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.5395429134368896,
+      "learning_rate": 9.775682812683482e-06,
+      "loss": 0.705,
+      "step": 7545
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.201241970062256,
+      "learning_rate": 9.775604719523842e-06,
+      "loss": 0.6487,
+      "step": 7546
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.9662582874298096,
+      "learning_rate": 9.775526613085033e-06,
+      "loss": 0.8464,
+      "step": 7547
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.491908073425293,
+      "learning_rate": 9.775448493367275e-06,
+      "loss": 0.9623,
+      "step": 7548
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.06856107711792,
+      "learning_rate": 9.775370360370783e-06,
+      "loss": 0.772,
+      "step": 7549
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.818438768386841,
+      "learning_rate": 9.775292214095776e-06,
+      "loss": 0.6471,
+      "step": 7550
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.807140350341797,
+      "learning_rate": 9.775214054542468e-06,
+      "loss": 0.7683,
+      "step": 7551
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 4.1057820320129395,
+      "learning_rate": 9.775135881711079e-06,
+      "loss": 1.0745,
+      "step": 7552
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.7084341049194336,
+      "learning_rate": 9.775057695601826e-06,
+      "loss": 0.5882,
+      "step": 7553
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.1333415508270264,
+      "learning_rate": 9.774979496214928e-06,
+      "loss": 0.6662,
+      "step": 7554
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.6789634227752686,
+      "learning_rate": 9.7749012835506e-06,
+      "loss": 0.6288,
+      "step": 7555
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.179757595062256,
+      "learning_rate": 9.77482305760906e-06,
+      "loss": 0.9077,
+      "step": 7556
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.2645578384399414,
+      "learning_rate": 9.774744818390525e-06,
+      "loss": 0.7617,
+      "step": 7557
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 4.738199234008789,
+      "learning_rate": 9.774666565895213e-06,
+      "loss": 0.8162,
+      "step": 7558
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.4315989017486572,
+      "learning_rate": 9.774588300123343e-06,
+      "loss": 0.7206,
+      "step": 7559
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.6445086002349854,
+      "learning_rate": 9.774510021075132e-06,
+      "loss": 0.83,
+      "step": 7560
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.6921920776367188,
+      "learning_rate": 9.774431728750795e-06,
+      "loss": 0.6691,
+      "step": 7561
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.794558048248291,
+      "learning_rate": 9.774353423150556e-06,
+      "loss": 0.9188,
+      "step": 7562
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.351163864135742,
+      "learning_rate": 9.774275104274625e-06,
+      "loss": 0.8075,
+      "step": 7563
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.618090867996216,
+      "learning_rate": 9.774196772123225e-06,
+      "loss": 0.9055,
+      "step": 7564
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.9844651222229004,
+      "learning_rate": 9.774118426696572e-06,
+      "loss": 0.7561,
+      "step": 7565
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.764141082763672,
+      "learning_rate": 9.774040067994883e-06,
+      "loss": 0.8883,
+      "step": 7566
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.9024922847747803,
+      "learning_rate": 9.773961696018379e-06,
+      "loss": 0.9275,
+      "step": 7567
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.533108949661255,
+      "learning_rate": 9.773883310767274e-06,
+      "loss": 0.7065,
+      "step": 7568
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.3049776554107666,
+      "learning_rate": 9.77380491224179e-06,
+      "loss": 0.6842,
+      "step": 7569
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.091947555541992,
+      "learning_rate": 9.77372650044214e-06,
+      "loss": 0.6763,
+      "step": 7570
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.9844954013824463,
+      "learning_rate": 9.773648075368546e-06,
+      "loss": 0.5631,
+      "step": 7571
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.7570149898529053,
+      "learning_rate": 9.773569637021226e-06,
+      "loss": 0.6683,
+      "step": 7572
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.9345134496688843,
+      "learning_rate": 9.773491185400395e-06,
+      "loss": 0.4155,
+      "step": 7573
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.661294460296631,
+      "learning_rate": 9.773412720506274e-06,
+      "loss": 0.6845,
+      "step": 7574
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.88274884223938,
+      "learning_rate": 9.773334242339082e-06,
+      "loss": 0.6503,
+      "step": 7575
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.1847074031829834,
+      "learning_rate": 9.773255750899035e-06,
+      "loss": 0.6712,
+      "step": 7576
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.4258382320404053,
+      "learning_rate": 9.773177246186352e-06,
+      "loss": 0.8888,
+      "step": 7577
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.1966392993927,
+      "learning_rate": 9.77309872820125e-06,
+      "loss": 0.7955,
+      "step": 7578
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.8471016883850098,
+      "learning_rate": 9.773020196943948e-06,
+      "loss": 0.5882,
+      "step": 7579
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.164827585220337,
+      "learning_rate": 9.772941652414667e-06,
+      "loss": 0.6721,
+      "step": 7580
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 4.724477767944336,
+      "learning_rate": 9.77286309461362e-06,
+      "loss": 0.8181,
+      "step": 7581
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.6425535678863525,
+      "learning_rate": 9.772784523541033e-06,
+      "loss": 0.6777,
+      "step": 7582
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.56585693359375,
+      "learning_rate": 9.772705939197117e-06,
+      "loss": 0.6811,
+      "step": 7583
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.9523775577545166,
+      "learning_rate": 9.772627341582095e-06,
+      "loss": 0.8243,
+      "step": 7584
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.791434407234192,
+      "learning_rate": 9.772548730696184e-06,
+      "loss": 0.7317,
+      "step": 7585
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.216627836227417,
+      "learning_rate": 9.772470106539602e-06,
+      "loss": 0.5395,
+      "step": 7586
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 4.211348533630371,
+      "learning_rate": 9.772391469112569e-06,
+      "loss": 0.8303,
+      "step": 7587
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.508749485015869,
+      "learning_rate": 9.772312818415303e-06,
+      "loss": 0.6674,
+      "step": 7588
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.7360129356384277,
+      "learning_rate": 9.772234154448024e-06,
+      "loss": 0.2771,
+      "step": 7589
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 4.127920150756836,
+      "learning_rate": 9.772155477210947e-06,
+      "loss": 1.0214,
+      "step": 7590
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.3896780014038086,
+      "learning_rate": 9.772076786704296e-06,
+      "loss": 0.8727,
+      "step": 7591
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.3339126110076904,
+      "learning_rate": 9.771998082928287e-06,
+      "loss": 0.4967,
+      "step": 7592
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.4868396520614624,
+      "learning_rate": 9.771919365883136e-06,
+      "loss": 0.9437,
+      "step": 7593
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.0993080139160156,
+      "learning_rate": 9.771840635569068e-06,
+      "loss": 0.5532,
+      "step": 7594
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.3673105239868164,
+      "learning_rate": 9.771761891986297e-06,
+      "loss": 0.8947,
+      "step": 7595
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.845120429992676,
+      "learning_rate": 9.771683135135044e-06,
+      "loss": 0.8506,
+      "step": 7596
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.578479528427124,
+      "learning_rate": 9.771604365015527e-06,
+      "loss": 0.6087,
+      "step": 7597
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.611987352371216,
+      "learning_rate": 9.771525581627968e-06,
+      "loss": 0.5497,
+      "step": 7598
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.9798190593719482,
+      "learning_rate": 9.771446784972583e-06,
+      "loss": 0.5567,
+      "step": 7599
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.6140196323394775,
+      "learning_rate": 9.771367975049591e-06,
+      "loss": 0.7891,
+      "step": 7600
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.7578306198120117,
+      "learning_rate": 9.771289151859213e-06,
+      "loss": 0.8141,
+      "step": 7601
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.5260894298553467,
+      "learning_rate": 9.771210315401667e-06,
+      "loss": 0.8308,
+      "step": 7602
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.451289653778076,
+      "learning_rate": 9.771131465677171e-06,
+      "loss": 0.6635,
+      "step": 7603
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.425516128540039,
+      "learning_rate": 9.771052602685947e-06,
+      "loss": 0.7009,
+      "step": 7604
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.215132236480713,
+      "learning_rate": 9.770973726428213e-06,
+      "loss": 0.8829,
+      "step": 7605
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.7881593704223633,
+      "learning_rate": 9.770894836904188e-06,
+      "loss": 0.7937,
+      "step": 7606
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.7845622301101685,
+      "learning_rate": 9.770815934114093e-06,
+      "loss": 0.7899,
+      "step": 7607
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.5085272789001465,
+      "learning_rate": 9.770737018058145e-06,
+      "loss": 0.5634,
+      "step": 7608
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 4.822560787200928,
+      "learning_rate": 9.770658088736564e-06,
+      "loss": 0.5608,
+      "step": 7609
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.821927070617676,
+      "learning_rate": 9.77057914614957e-06,
+      "loss": 0.5809,
+      "step": 7610
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.9226858615875244,
+      "learning_rate": 9.770500190297382e-06,
+      "loss": 0.7495,
+      "step": 7611
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.761646032333374,
+      "learning_rate": 9.770421221180223e-06,
+      "loss": 0.7989,
+      "step": 7612
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.0176188945770264,
+      "learning_rate": 9.770342238798307e-06,
+      "loss": 0.8317,
+      "step": 7613
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.212280750274658,
+      "learning_rate": 9.770263243151857e-06,
+      "loss": 0.4058,
+      "step": 7614
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.478771209716797,
+      "learning_rate": 9.770184234241091e-06,
+      "loss": 0.5807,
+      "step": 7615
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.1640913486480713,
+      "learning_rate": 9.770105212066232e-06,
+      "loss": 0.8106,
+      "step": 7616
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.463195323944092,
+      "learning_rate": 9.770026176627494e-06,
+      "loss": 0.6736,
+      "step": 7617
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.8311265707015991,
+      "learning_rate": 9.769947127925102e-06,
+      "loss": 0.6459,
+      "step": 7618
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.1998984813690186,
+      "learning_rate": 9.769868065959274e-06,
+      "loss": 0.9674,
+      "step": 7619
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 10.10949993133545,
+      "learning_rate": 9.76978899073023e-06,
+      "loss": 0.8352,
+      "step": 7620
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.736834168434143,
+      "learning_rate": 9.76970990223819e-06,
+      "loss": 0.6354,
+      "step": 7621
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.8497538566589355,
+      "learning_rate": 9.769630800483372e-06,
+      "loss": 0.6229,
+      "step": 7622
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 4.595667839050293,
+      "learning_rate": 9.769551685465996e-06,
+      "loss": 0.9065,
+      "step": 7623
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.9010940790176392,
+      "learning_rate": 9.769472557186288e-06,
+      "loss": 0.7551,
+      "step": 7624
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.637052297592163,
+      "learning_rate": 9.769393415644459e-06,
+      "loss": 0.7763,
+      "step": 7625
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.6491496562957764,
+      "learning_rate": 9.769314260840736e-06,
+      "loss": 0.5431,
+      "step": 7626
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.2134897708892822,
+      "learning_rate": 9.769235092775336e-06,
+      "loss": 0.7741,
+      "step": 7627
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.8874974250793457,
+      "learning_rate": 9.76915591144848e-06,
+      "loss": 0.7115,
+      "step": 7628
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.091669797897339,
+      "learning_rate": 9.769076716860389e-06,
+      "loss": 0.6248,
+      "step": 7629
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.7812514305114746,
+      "learning_rate": 9.768997509011281e-06,
+      "loss": 0.7003,
+      "step": 7630
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.5162134170532227,
+      "learning_rate": 9.768918287901377e-06,
+      "loss": 0.5079,
+      "step": 7631
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.8107408285140991,
+      "learning_rate": 9.768839053530899e-06,
+      "loss": 0.609,
+      "step": 7632
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.2455170154571533,
+      "learning_rate": 9.768759805900065e-06,
+      "loss": 0.8057,
+      "step": 7633
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.1451268196105957,
+      "learning_rate": 9.768680545009097e-06,
+      "loss": 0.7457,
+      "step": 7634
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.642148971557617,
+      "learning_rate": 9.768601270858215e-06,
+      "loss": 0.7156,
+      "step": 7635
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.653190851211548,
+      "learning_rate": 9.768521983447638e-06,
+      "loss": 0.6609,
+      "step": 7636
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.2607529163360596,
+      "learning_rate": 9.76844268277759e-06,
+      "loss": 0.8961,
+      "step": 7637
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.675877332687378,
+      "learning_rate": 9.768363368848288e-06,
+      "loss": 0.615,
+      "step": 7638
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 4.433523654937744,
+      "learning_rate": 9.768284041659952e-06,
+      "loss": 0.6642,
+      "step": 7639
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.2063262462615967,
+      "learning_rate": 9.768204701212807e-06,
+      "loss": 0.4029,
+      "step": 7640
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.8257408142089844,
+      "learning_rate": 9.76812534750707e-06,
+      "loss": 0.6349,
+      "step": 7641
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.2062647342681885,
+      "learning_rate": 9.768045980542962e-06,
+      "loss": 0.5582,
+      "step": 7642
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 5.169865608215332,
+      "learning_rate": 9.767966600320704e-06,
+      "loss": 0.7032,
+      "step": 7643
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.611137866973877,
+      "learning_rate": 9.767887206840519e-06,
+      "loss": 0.5245,
+      "step": 7644
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.6400084495544434,
+      "learning_rate": 9.767807800102623e-06,
+      "loss": 0.8724,
+      "step": 7645
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.2187273502349854,
+      "learning_rate": 9.767728380107243e-06,
+      "loss": 0.7858,
+      "step": 7646
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.0928430557250977,
+      "learning_rate": 9.767648946854593e-06,
+      "loss": 0.7591,
+      "step": 7647
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.2144460678100586,
+      "learning_rate": 9.767569500344898e-06,
+      "loss": 0.9852,
+      "step": 7648
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 4.243098258972168,
+      "learning_rate": 9.767490040578379e-06,
+      "loss": 0.5534,
+      "step": 7649
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.6308672428131104,
+      "learning_rate": 9.767410567555256e-06,
+      "loss": 0.6202,
+      "step": 7650
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.112082004547119,
+      "learning_rate": 9.76733108127575e-06,
+      "loss": 0.4393,
+      "step": 7651
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.408879518508911,
+      "learning_rate": 9.767251581740083e-06,
+      "loss": 0.5315,
+      "step": 7652
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.4414143562316895,
+      "learning_rate": 9.767172068948474e-06,
+      "loss": 0.6006,
+      "step": 7653
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.094590187072754,
+      "learning_rate": 9.767092542901146e-06,
+      "loss": 0.9188,
+      "step": 7654
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.9163645505905151,
+      "learning_rate": 9.767013003598318e-06,
+      "loss": 0.7448,
+      "step": 7655
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.2811880111694336,
+      "learning_rate": 9.766933451040215e-06,
+      "loss": 0.6534,
+      "step": 7656
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.9628721475601196,
+      "learning_rate": 9.766853885227055e-06,
+      "loss": 0.6503,
+      "step": 7657
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.494154453277588,
+      "learning_rate": 9.76677430615906e-06,
+      "loss": 0.6934,
+      "step": 7658
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.1527369022369385,
+      "learning_rate": 9.76669471383645e-06,
+      "loss": 0.6791,
+      "step": 7659
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.233250379562378,
+      "learning_rate": 9.766615108259449e-06,
+      "loss": 0.7621,
+      "step": 7660
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.823665976524353,
+      "learning_rate": 9.766535489428276e-06,
+      "loss": 0.6052,
+      "step": 7661
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.273242712020874,
+      "learning_rate": 9.766455857343153e-06,
+      "loss": 0.5882,
+      "step": 7662
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.860063076019287,
+      "learning_rate": 9.766376212004305e-06,
+      "loss": 0.7067,
+      "step": 7663
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.533223867416382,
+      "learning_rate": 9.766296553411947e-06,
+      "loss": 0.4456,
+      "step": 7664
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.4826347827911377,
+      "learning_rate": 9.766216881566305e-06,
+      "loss": 0.7248,
+      "step": 7665
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.5125277042388916,
+      "learning_rate": 9.766137196467599e-06,
+      "loss": 0.5042,
+      "step": 7666
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.922305107116699,
+      "learning_rate": 9.766057498116051e-06,
+      "loss": 0.6797,
+      "step": 7667
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.2178659439086914,
+      "learning_rate": 9.765977786511882e-06,
+      "loss": 0.6059,
+      "step": 7668
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.0305917263031006,
+      "learning_rate": 9.765898061655315e-06,
+      "loss": 0.8846,
+      "step": 7669
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.049903392791748,
+      "learning_rate": 9.76581832354657e-06,
+      "loss": 0.8418,
+      "step": 7670
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.1058430671691895,
+      "learning_rate": 9.765738572185871e-06,
+      "loss": 1.0461,
+      "step": 7671
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.3662631511688232,
+      "learning_rate": 9.765658807573437e-06,
+      "loss": 0.7173,
+      "step": 7672
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.1993701457977295,
+      "learning_rate": 9.765579029709492e-06,
+      "loss": 0.8236,
+      "step": 7673
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.9172364473342896,
+      "learning_rate": 9.765499238594257e-06,
+      "loss": 0.5955,
+      "step": 7674
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.541961431503296,
+      "learning_rate": 9.765419434227952e-06,
+      "loss": 0.8334,
+      "step": 7675
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.7445414066314697,
+      "learning_rate": 9.765339616610803e-06,
+      "loss": 0.5421,
+      "step": 7676
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.116887092590332,
+      "learning_rate": 9.765259785743028e-06,
+      "loss": 0.5497,
+      "step": 7677
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.5129140615463257,
+      "learning_rate": 9.765179941624852e-06,
+      "loss": 0.644,
+      "step": 7678
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.84163761138916,
+      "learning_rate": 9.765100084256496e-06,
+      "loss": 0.9465,
+      "step": 7679
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.477912187576294,
+      "learning_rate": 9.76502021363818e-06,
+      "loss": 0.7882,
+      "step": 7680
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 4.030703067779541,
+      "learning_rate": 9.764940329770128e-06,
+      "loss": 0.8793,
+      "step": 7681
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.3784351348876953,
+      "learning_rate": 9.764860432652563e-06,
+      "loss": 0.7439,
+      "step": 7682
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.2768733501434326,
+      "learning_rate": 9.764780522285706e-06,
+      "loss": 0.5673,
+      "step": 7683
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.03304386138916,
+      "learning_rate": 9.764700598669778e-06,
+      "loss": 0.6535,
+      "step": 7684
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.7274088859558105,
+      "learning_rate": 9.764620661805005e-06,
+      "loss": 0.5336,
+      "step": 7685
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.151188850402832,
+      "learning_rate": 9.764540711691604e-06,
+      "loss": 0.8388,
+      "step": 7686
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.606787919998169,
+      "learning_rate": 9.764460748329802e-06,
+      "loss": 0.7879,
+      "step": 7687
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 12.123703956604004,
+      "learning_rate": 9.764380771719819e-06,
+      "loss": 0.7512,
+      "step": 7688
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.7082083225250244,
+      "learning_rate": 9.764300781861877e-06,
+      "loss": 0.8744,
+      "step": 7689
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.9788540601730347,
+      "learning_rate": 9.7642207787562e-06,
+      "loss": 0.886,
+      "step": 7690
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.5761911869049072,
+      "learning_rate": 9.764140762403009e-06,
+      "loss": 0.6124,
+      "step": 7691
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.4300484657287598,
+      "learning_rate": 9.764060732802529e-06,
+      "loss": 0.6937,
+      "step": 7692
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 6.063046455383301,
+      "learning_rate": 9.763980689954979e-06,
+      "loss": 0.9681,
+      "step": 7693
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.016963481903076,
+      "learning_rate": 9.763900633860584e-06,
+      "loss": 0.369,
+      "step": 7694
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.7504563331604004,
+      "learning_rate": 9.763820564519566e-06,
+      "loss": 0.3735,
+      "step": 7695
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.8774569034576416,
+      "learning_rate": 9.763740481932147e-06,
+      "loss": 0.6658,
+      "step": 7696
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.881917953491211,
+      "learning_rate": 9.763660386098551e-06,
+      "loss": 0.7067,
+      "step": 7697
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.300380229949951,
+      "learning_rate": 9.763580277019001e-06,
+      "loss": 0.5411,
+      "step": 7698
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.8630311489105225,
+      "learning_rate": 9.763500154693716e-06,
+      "loss": 0.7846,
+      "step": 7699
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 4.293248176574707,
+      "learning_rate": 9.763420019122923e-06,
+      "loss": 0.5506,
+      "step": 7700
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 4.526722431182861,
+      "learning_rate": 9.763339870306843e-06,
+      "loss": 0.6425,
+      "step": 7701
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.364234685897827,
+      "learning_rate": 9.7632597082457e-06,
+      "loss": 0.6201,
+      "step": 7702
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.0171430110931396,
+      "learning_rate": 9.763179532939714e-06,
+      "loss": 0.6952,
+      "step": 7703
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.7293853759765625,
+      "learning_rate": 9.763099344389112e-06,
+      "loss": 0.4351,
+      "step": 7704
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.5049455165863037,
+      "learning_rate": 9.763019142594115e-06,
+      "loss": 0.8169,
+      "step": 7705
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.3054866790771484,
+      "learning_rate": 9.762938927554945e-06,
+      "loss": 0.7022,
+      "step": 7706
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.912403106689453,
+      "learning_rate": 9.762858699271828e-06,
+      "loss": 0.9016,
+      "step": 7707
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.0678374767303467,
+      "learning_rate": 9.762778457744984e-06,
+      "loss": 0.895,
+      "step": 7708
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.3718347549438477,
+      "learning_rate": 9.762698202974637e-06,
+      "loss": 0.6093,
+      "step": 7709
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.6985890865325928,
+      "learning_rate": 9.762617934961012e-06,
+      "loss": 0.8583,
+      "step": 7710
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.1634151935577393,
+      "learning_rate": 9.762537653704328e-06,
+      "loss": 0.4643,
+      "step": 7711
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.9194835424423218,
+      "learning_rate": 9.762457359204814e-06,
+      "loss": 0.8522,
+      "step": 7712
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.813854932785034,
+      "learning_rate": 9.762377051462686e-06,
+      "loss": 0.7321,
+      "step": 7713
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.9068009853363037,
+      "learning_rate": 9.762296730478175e-06,
+      "loss": 0.4245,
+      "step": 7714
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.0881083011627197,
+      "learning_rate": 9.7622163962515e-06,
+      "loss": 0.6825,
+      "step": 7715
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.46598482131958,
+      "learning_rate": 9.762136048782885e-06,
+      "loss": 0.7859,
+      "step": 7716
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.8316686153411865,
+      "learning_rate": 9.762055688072552e-06,
+      "loss": 0.6755,
+      "step": 7717
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.952956199645996,
+      "learning_rate": 9.76197531412073e-06,
+      "loss": 0.6216,
+      "step": 7718
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.0327656269073486,
+      "learning_rate": 9.761894926927633e-06,
+      "loss": 0.6922,
+      "step": 7719
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 4.179234504699707,
+      "learning_rate": 9.761814526493493e-06,
+      "loss": 1.0342,
+      "step": 7720
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.7564501762390137,
+      "learning_rate": 9.76173411281853e-06,
+      "loss": 0.7535,
+      "step": 7721
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.2843081951141357,
+      "learning_rate": 9.761653685902969e-06,
+      "loss": 0.7868,
+      "step": 7722
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 4.083866596221924,
+      "learning_rate": 9.761573245747034e-06,
+      "loss": 0.8267,
+      "step": 7723
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.4585447311401367,
+      "learning_rate": 9.761492792350946e-06,
+      "loss": 0.8354,
+      "step": 7724
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.815127968788147,
+      "learning_rate": 9.76141232571493e-06,
+      "loss": 0.5995,
+      "step": 7725
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.031839609146118,
+      "learning_rate": 9.761331845839211e-06,
+      "loss": 0.5642,
+      "step": 7726
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 5.001364707946777,
+      "learning_rate": 9.76125135272401e-06,
+      "loss": 0.4789,
+      "step": 7727
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.3555257320404053,
+      "learning_rate": 9.761170846369555e-06,
+      "loss": 0.6016,
+      "step": 7728
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.757368803024292,
+      "learning_rate": 9.761090326776066e-06,
+      "loss": 0.7492,
+      "step": 7729
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 4.189624309539795,
+      "learning_rate": 9.761009793943768e-06,
+      "loss": 0.8542,
+      "step": 7730
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.0628931522369385,
+      "learning_rate": 9.760929247872885e-06,
+      "loss": 0.4858,
+      "step": 7731
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.3126931190490723,
+      "learning_rate": 9.760848688563643e-06,
+      "loss": 0.6345,
+      "step": 7732
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.4949874877929688,
+      "learning_rate": 9.760768116016264e-06,
+      "loss": 0.8227,
+      "step": 7733
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.961682081222534,
+      "learning_rate": 9.76068753023097e-06,
+      "loss": 0.6799,
+      "step": 7734
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.7658681869506836,
+      "learning_rate": 9.76060693120799e-06,
+      "loss": 0.8973,
+      "step": 7735
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.556661367416382,
+      "learning_rate": 9.760526318947546e-06,
+      "loss": 0.625,
+      "step": 7736
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.4849843978881836,
+      "learning_rate": 9.760445693449859e-06,
+      "loss": 0.4518,
+      "step": 7737
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.224066734313965,
+      "learning_rate": 9.760365054715157e-06,
+      "loss": 0.5534,
+      "step": 7738
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.1725668907165527,
+      "learning_rate": 9.760284402743663e-06,
+      "loss": 0.8608,
+      "step": 7739
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 4.109073638916016,
+      "learning_rate": 9.7602037375356e-06,
+      "loss": 0.7765,
+      "step": 7740
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.372227668762207,
+      "learning_rate": 9.760123059091195e-06,
+      "loss": 0.5942,
+      "step": 7741
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.338318109512329,
+      "learning_rate": 9.760042367410671e-06,
+      "loss": 0.5689,
+      "step": 7742
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 5.184192180633545,
+      "learning_rate": 9.759961662494253e-06,
+      "loss": 1.0443,
+      "step": 7743
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.161313056945801,
+      "learning_rate": 9.759880944342164e-06,
+      "loss": 0.7659,
+      "step": 7744
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.712099313735962,
+      "learning_rate": 9.75980021295463e-06,
+      "loss": 0.7926,
+      "step": 7745
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.2105400562286377,
+      "learning_rate": 9.759719468331875e-06,
+      "loss": 0.7495,
+      "step": 7746
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.9836077690124512,
+      "learning_rate": 9.75963871047412e-06,
+      "loss": 0.7175,
+      "step": 7747
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.2361807823181152,
+      "learning_rate": 9.759557939381595e-06,
+      "loss": 0.768,
+      "step": 7748
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.8361833095550537,
+      "learning_rate": 9.759477155054522e-06,
+      "loss": 0.6155,
+      "step": 7749
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.16914439201355,
+      "learning_rate": 9.759396357493128e-06,
+      "loss": 0.5844,
+      "step": 7750
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.19270658493042,
+      "learning_rate": 9.759315546697632e-06,
+      "loss": 0.7015,
+      "step": 7751
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.9896023273468018,
+      "learning_rate": 9.759234722668264e-06,
+      "loss": 0.6253,
+      "step": 7752
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.639908790588379,
+      "learning_rate": 9.759153885405249e-06,
+      "loss": 0.8188,
+      "step": 7753
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.673476219177246,
+      "learning_rate": 9.759073034908805e-06,
+      "loss": 0.6985,
+      "step": 7754
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 4.2505340576171875,
+      "learning_rate": 9.758992171179168e-06,
+      "loss": 0.6024,
+      "step": 7755
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.9130668640136719,
+      "learning_rate": 9.758911294216551e-06,
+      "loss": 0.5577,
+      "step": 7756
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.1787092685699463,
+      "learning_rate": 9.758830404021186e-06,
+      "loss": 0.8929,
+      "step": 7757
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.2460873126983643,
+      "learning_rate": 9.758749500593298e-06,
+      "loss": 0.7401,
+      "step": 7758
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 6.3512043952941895,
+      "learning_rate": 9.75866858393311e-06,
+      "loss": 0.6584,
+      "step": 7759
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.5867841243743896,
+      "learning_rate": 9.758587654040845e-06,
+      "loss": 0.7544,
+      "step": 7760
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.3613247871398926,
+      "learning_rate": 9.758506710916733e-06,
+      "loss": 0.5934,
+      "step": 7761
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.899266481399536,
+      "learning_rate": 9.758425754560996e-06,
+      "loss": 0.7746,
+      "step": 7762
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.4321370124816895,
+      "learning_rate": 9.758344784973858e-06,
+      "loss": 0.7266,
+      "step": 7763
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.9314219951629639,
+      "learning_rate": 9.758263802155547e-06,
+      "loss": 0.6937,
+      "step": 7764
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.420245409011841,
+      "learning_rate": 9.758182806106287e-06,
+      "loss": 0.6308,
+      "step": 7765
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.808030605316162,
+      "learning_rate": 9.758101796826303e-06,
+      "loss": 0.609,
+      "step": 7766
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 4.145461559295654,
+      "learning_rate": 9.758020774315818e-06,
+      "loss": 1.0073,
+      "step": 7767
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.8477673530578613,
+      "learning_rate": 9.757939738575061e-06,
+      "loss": 0.4767,
+      "step": 7768
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.601203680038452,
+      "learning_rate": 9.757858689604258e-06,
+      "loss": 0.7893,
+      "step": 7769
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.13643479347229,
+      "learning_rate": 9.75777762740363e-06,
+      "loss": 0.7307,
+      "step": 7770
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 4.166924476623535,
+      "learning_rate": 9.757696551973407e-06,
+      "loss": 0.6546,
+      "step": 7771
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.8358601331710815,
+      "learning_rate": 9.757615463313809e-06,
+      "loss": 0.7687,
+      "step": 7772
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.8416526317596436,
+      "learning_rate": 9.757534361425068e-06,
+      "loss": 0.675,
+      "step": 7773
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.559704065322876,
+      "learning_rate": 9.757453246307402e-06,
+      "loss": 0.5024,
+      "step": 7774
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.697685956954956,
+      "learning_rate": 9.757372117961043e-06,
+      "loss": 0.6418,
+      "step": 7775
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.8906528949737549,
+      "learning_rate": 9.757290976386215e-06,
+      "loss": 0.6745,
+      "step": 7776
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.99691641330719,
+      "learning_rate": 9.757209821583143e-06,
+      "loss": 0.6656,
+      "step": 7777
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.6194560527801514,
+      "learning_rate": 9.757128653552052e-06,
+      "loss": 0.567,
+      "step": 7778
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 4.505764961242676,
+      "learning_rate": 9.75704747229317e-06,
+      "loss": 0.6804,
+      "step": 7779
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.308842897415161,
+      "learning_rate": 9.756966277806717e-06,
+      "loss": 0.7153,
+      "step": 7780
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 7.476923942565918,
+      "learning_rate": 9.756885070092925e-06,
+      "loss": 0.5396,
+      "step": 7781
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.0618112087249756,
+      "learning_rate": 9.756803849152016e-06,
+      "loss": 0.636,
+      "step": 7782
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.164794921875,
+      "learning_rate": 9.75672261498422e-06,
+      "loss": 0.3523,
+      "step": 7783
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.1402535438537598,
+      "learning_rate": 9.756641367589759e-06,
+      "loss": 0.5889,
+      "step": 7784
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.8430259227752686,
+      "learning_rate": 9.756560106968862e-06,
+      "loss": 0.7802,
+      "step": 7785
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.9890581369400024,
+      "learning_rate": 9.756478833121751e-06,
+      "loss": 0.5628,
+      "step": 7786
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.1386473178863525,
+      "learning_rate": 9.756397546048655e-06,
+      "loss": 0.6175,
+      "step": 7787
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.415710926055908,
+      "learning_rate": 9.7563162457498e-06,
+      "loss": 0.9264,
+      "step": 7788
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 5.787433624267578,
+      "learning_rate": 9.75623493222541e-06,
+      "loss": 0.9061,
+      "step": 7789
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.7334706783294678,
+      "learning_rate": 9.756153605475712e-06,
+      "loss": 0.774,
+      "step": 7790
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.7766075134277344,
+      "learning_rate": 9.756072265500934e-06,
+      "loss": 0.7051,
+      "step": 7791
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.3677968978881836,
+      "learning_rate": 9.7559909123013e-06,
+      "loss": 0.6502,
+      "step": 7792
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.940485954284668,
+      "learning_rate": 9.755909545877036e-06,
+      "loss": 0.6521,
+      "step": 7793
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.8108127117156982,
+      "learning_rate": 9.75582816622837e-06,
+      "loss": 0.6902,
+      "step": 7794
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.5015931129455566,
+      "learning_rate": 9.755746773355527e-06,
+      "loss": 0.7662,
+      "step": 7795
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.240999698638916,
+      "learning_rate": 9.755665367258735e-06,
+      "loss": 0.531,
+      "step": 7796
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.098737955093384,
+      "learning_rate": 9.755583947938219e-06,
+      "loss": 0.9444,
+      "step": 7797
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.8000096082687378,
+      "learning_rate": 9.755502515394204e-06,
+      "loss": 1.082,
+      "step": 7798
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.2812302112579346,
+      "learning_rate": 9.75542106962692e-06,
+      "loss": 0.6284,
+      "step": 7799
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 5.380776882171631,
+      "learning_rate": 9.75533961063659e-06,
+      "loss": 0.339,
+      "step": 7800
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.7036292552947998,
+      "learning_rate": 9.755258138423442e-06,
+      "loss": 0.6508,
+      "step": 7801
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.3969857692718506,
+      "learning_rate": 9.755176652987703e-06,
+      "loss": 0.7322,
+      "step": 7802
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.0695247650146484,
+      "learning_rate": 9.755095154329598e-06,
+      "loss": 0.7755,
+      "step": 7803
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.3510582447052,
+      "learning_rate": 9.755013642449355e-06,
+      "loss": 0.6129,
+      "step": 7804
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.094139575958252,
+      "learning_rate": 9.754932117347202e-06,
+      "loss": 0.6791,
+      "step": 7805
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.6694806814193726,
+      "learning_rate": 9.754850579023363e-06,
+      "loss": 0.9219,
+      "step": 7806
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.5297437906265259,
+      "learning_rate": 9.754769027478064e-06,
+      "loss": 0.5084,
+      "step": 7807
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.9188735485076904,
+      "learning_rate": 9.754687462711537e-06,
+      "loss": 0.6013,
+      "step": 7808
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.405708074569702,
+      "learning_rate": 9.754605884724003e-06,
+      "loss": 0.4358,
+      "step": 7809
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.386044979095459,
+      "learning_rate": 9.75452429351569e-06,
+      "loss": 0.7536,
+      "step": 7810
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 4.414033889770508,
+      "learning_rate": 9.754442689086828e-06,
+      "loss": 0.7801,
+      "step": 7811
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.9828763008117676,
+      "learning_rate": 9.754361071437642e-06,
+      "loss": 0.9385,
+      "step": 7812
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.8622658252716064,
+      "learning_rate": 9.754279440568358e-06,
+      "loss": 0.5212,
+      "step": 7813
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.7295536994934082,
+      "learning_rate": 9.754197796479204e-06,
+      "loss": 0.675,
+      "step": 7814
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.3808966875076294,
+      "learning_rate": 9.754116139170407e-06,
+      "loss": 0.6108,
+      "step": 7815
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.4413866996765137,
+      "learning_rate": 9.754034468642194e-06,
+      "loss": 0.7018,
+      "step": 7816
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.971343755722046,
+      "learning_rate": 9.753952784894794e-06,
+      "loss": 0.8797,
+      "step": 7817
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.2411139011383057,
+      "learning_rate": 9.75387108792843e-06,
+      "loss": 0.5902,
+      "step": 7818
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.319924831390381,
+      "learning_rate": 9.75378937774333e-06,
+      "loss": 0.7838,
+      "step": 7819
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.5139780044555664,
+      "learning_rate": 9.753707654339725e-06,
+      "loss": 0.4487,
+      "step": 7820
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.087379217147827,
+      "learning_rate": 9.75362591771784e-06,
+      "loss": 0.3514,
+      "step": 7821
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 8.087340354919434,
+      "learning_rate": 9.7535441678779e-06,
+      "loss": 0.5191,
+      "step": 7822
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.5201377868652344,
+      "learning_rate": 9.753462404820136e-06,
+      "loss": 0.9206,
+      "step": 7823
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.9733492136001587,
+      "learning_rate": 9.753380628544774e-06,
+      "loss": 0.6524,
+      "step": 7824
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.979179859161377,
+      "learning_rate": 9.753298839052038e-06,
+      "loss": 0.5555,
+      "step": 7825
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 5.74252986907959,
+      "learning_rate": 9.753217036342162e-06,
+      "loss": 0.5807,
+      "step": 7826
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.3786728382110596,
+      "learning_rate": 9.753135220415368e-06,
+      "loss": 0.6095,
+      "step": 7827
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.8753061294555664,
+      "learning_rate": 9.753053391271887e-06,
+      "loss": 0.8803,
+      "step": 7828
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.5038156509399414,
+      "learning_rate": 9.752971548911945e-06,
+      "loss": 0.5188,
+      "step": 7829
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.612154483795166,
+      "learning_rate": 9.752889693335769e-06,
+      "loss": 0.7083,
+      "step": 7830
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.6590325832366943,
+      "learning_rate": 9.752807824543587e-06,
+      "loss": 0.8025,
+      "step": 7831
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 5.10548210144043,
+      "learning_rate": 9.752725942535628e-06,
+      "loss": 0.8161,
+      "step": 7832
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.65425968170166,
+      "learning_rate": 9.752644047312117e-06,
+      "loss": 0.349,
+      "step": 7833
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.8385125398635864,
+      "learning_rate": 9.752562138873284e-06,
+      "loss": 0.4622,
+      "step": 7834
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.9097341299057007,
+      "learning_rate": 9.752480217219355e-06,
+      "loss": 0.5947,
+      "step": 7835
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.598353862762451,
+      "learning_rate": 9.752398282350562e-06,
+      "loss": 0.6555,
+      "step": 7836
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.9501334428787231,
+      "learning_rate": 9.752316334267126e-06,
+      "loss": 0.4173,
+      "step": 7837
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.8772029876708984,
+      "learning_rate": 9.75223437296928e-06,
+      "loss": 0.8304,
+      "step": 7838
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.2866759300231934,
+      "learning_rate": 9.752152398457251e-06,
+      "loss": 0.6639,
+      "step": 7839
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.9250102043151855,
+      "learning_rate": 9.752070410731267e-06,
+      "loss": 0.6705,
+      "step": 7840
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.9574334621429443,
+      "learning_rate": 9.751988409791553e-06,
+      "loss": 0.8147,
+      "step": 7841
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.441373348236084,
+      "learning_rate": 9.751906395638341e-06,
+      "loss": 0.7337,
+      "step": 7842
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.3501195907592773,
+      "learning_rate": 9.751824368271858e-06,
+      "loss": 0.8236,
+      "step": 7843
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.0055270195007324,
+      "learning_rate": 9.751742327692331e-06,
+      "loss": 0.9259,
+      "step": 7844
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.3011536598205566,
+      "learning_rate": 9.751660273899988e-06,
+      "loss": 0.8703,
+      "step": 7845
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 4.0568718910217285,
+      "learning_rate": 9.751578206895058e-06,
+      "loss": 0.9201,
+      "step": 7846
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.8879077434539795,
+      "learning_rate": 9.751496126677769e-06,
+      "loss": 0.7496,
+      "step": 7847
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.382917642593384,
+      "learning_rate": 9.751414033248348e-06,
+      "loss": 0.6185,
+      "step": 7848
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.097034454345703,
+      "learning_rate": 9.751331926607026e-06,
+      "loss": 0.6971,
+      "step": 7849
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 4.904994964599609,
+      "learning_rate": 9.75124980675403e-06,
+      "loss": 0.6944,
+      "step": 7850
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.264645576477051,
+      "learning_rate": 9.751167673689588e-06,
+      "loss": 0.7432,
+      "step": 7851
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 4.383571624755859,
+      "learning_rate": 9.751085527413927e-06,
+      "loss": 0.8945,
+      "step": 7852
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.867931842803955,
+      "learning_rate": 9.751003367927279e-06,
+      "loss": 0.8971,
+      "step": 7853
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.741159200668335,
+      "learning_rate": 9.750921195229869e-06,
+      "loss": 0.5333,
+      "step": 7854
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 4.172741413116455,
+      "learning_rate": 9.750839009321928e-06,
+      "loss": 0.6548,
+      "step": 7855
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 6.189442157745361,
+      "learning_rate": 9.750756810203683e-06,
+      "loss": 0.8398,
+      "step": 7856
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.2167792320251465,
+      "learning_rate": 9.750674597875362e-06,
+      "loss": 0.6378,
+      "step": 7857
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.3609395027160645,
+      "learning_rate": 9.750592372337195e-06,
+      "loss": 0.7667,
+      "step": 7858
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.844980478286743,
+      "learning_rate": 9.750510133589412e-06,
+      "loss": 0.6303,
+      "step": 7859
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.948125123977661,
+      "learning_rate": 9.750427881632239e-06,
+      "loss": 0.8217,
+      "step": 7860
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.0760045051574707,
+      "learning_rate": 9.750345616465904e-06,
+      "loss": 0.8188,
+      "step": 7861
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 4.269561290740967,
+      "learning_rate": 9.750263338090639e-06,
+      "loss": 0.7256,
+      "step": 7862
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.5386619567871094,
+      "learning_rate": 9.750181046506671e-06,
+      "loss": 0.6813,
+      "step": 7863
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.9729955196380615,
+      "learning_rate": 9.750098741714227e-06,
+      "loss": 1.0115,
+      "step": 7864
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.0206122398376465,
+      "learning_rate": 9.75001642371354e-06,
+      "loss": 0.742,
+      "step": 7865
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.9424235820770264,
+      "learning_rate": 9.749934092504835e-06,
+      "loss": 0.6406,
+      "step": 7866
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.488325595855713,
+      "learning_rate": 9.749851748088346e-06,
+      "loss": 0.5609,
+      "step": 7867
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.281430721282959,
+      "learning_rate": 9.749769390464297e-06,
+      "loss": 0.6267,
+      "step": 7868
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.861443281173706,
+      "learning_rate": 9.749687019632916e-06,
+      "loss": 0.5737,
+      "step": 7869
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.135782241821289,
+      "learning_rate": 9.749604635594438e-06,
+      "loss": 0.7265,
+      "step": 7870
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.3788936138153076,
+      "learning_rate": 9.749522238349087e-06,
+      "loss": 0.7916,
+      "step": 7871
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.3252792358398438,
+      "learning_rate": 9.749439827897094e-06,
+      "loss": 0.8689,
+      "step": 7872
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.18875789642334,
+      "learning_rate": 9.749357404238689e-06,
+      "loss": 0.5721,
+      "step": 7873
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.2580747604370117,
+      "learning_rate": 9.7492749673741e-06,
+      "loss": 0.496,
+      "step": 7874
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.628424644470215,
+      "learning_rate": 9.749192517303556e-06,
+      "loss": 0.321,
+      "step": 7875
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.473475933074951,
+      "learning_rate": 9.749110054027287e-06,
+      "loss": 0.5429,
+      "step": 7876
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.617643117904663,
+      "learning_rate": 9.749027577545522e-06,
+      "loss": 0.6329,
+      "step": 7877
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.3574087619781494,
+      "learning_rate": 9.74894508785849e-06,
+      "loss": 0.8806,
+      "step": 7878
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.9722936153411865,
+      "learning_rate": 9.74886258496642e-06,
+      "loss": 0.6615,
+      "step": 7879
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.58544659614563,
+      "learning_rate": 9.748780068869543e-06,
+      "loss": 0.6733,
+      "step": 7880
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.611417055130005,
+      "learning_rate": 9.748697539568087e-06,
+      "loss": 0.8355,
+      "step": 7881
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.626835584640503,
+      "learning_rate": 9.748614997062283e-06,
+      "loss": 1.029,
+      "step": 7882
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.8415982723236084,
+      "learning_rate": 9.748532441352359e-06,
+      "loss": 0.7466,
+      "step": 7883
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.687450647354126,
+      "learning_rate": 9.748449872438546e-06,
+      "loss": 0.8123,
+      "step": 7884
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.6258280277252197,
+      "learning_rate": 9.74836729032107e-06,
+      "loss": 0.8405,
+      "step": 7885
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.965442657470703,
+      "learning_rate": 9.748284695000165e-06,
+      "loss": 0.6582,
+      "step": 7886
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.360511064529419,
+      "learning_rate": 9.74820208647606e-06,
+      "loss": 0.9904,
+      "step": 7887
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 4.133638858795166,
+      "learning_rate": 9.748119464748985e-06,
+      "loss": 0.4006,
+      "step": 7888
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.0025432109832764,
+      "learning_rate": 9.748036829819165e-06,
+      "loss": 0.8698,
+      "step": 7889
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.217472791671753,
+      "learning_rate": 9.747954181686834e-06,
+      "loss": 0.8524,
+      "step": 7890
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.3915817737579346,
+      "learning_rate": 9.747871520352222e-06,
+      "loss": 0.7416,
+      "step": 7891
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 13.668126106262207,
+      "learning_rate": 9.747788845815558e-06,
+      "loss": 0.6797,
+      "step": 7892
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.8710134029388428,
+      "learning_rate": 9.747706158077071e-06,
+      "loss": 0.655,
+      "step": 7893
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.783564329147339,
+      "learning_rate": 9.747623457136993e-06,
+      "loss": 0.6823,
+      "step": 7894
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.3009471893310547,
+      "learning_rate": 9.747540742995552e-06,
+      "loss": 0.6121,
+      "step": 7895
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.531944513320923,
+      "learning_rate": 9.747458015652977e-06,
+      "loss": 0.3957,
+      "step": 7896
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.7024383544921875,
+      "learning_rate": 9.7473752751095e-06,
+      "loss": 0.5507,
+      "step": 7897
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.8498543500900269,
+      "learning_rate": 9.747292521365354e-06,
+      "loss": 0.6283,
+      "step": 7898
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.1682121753692627,
+      "learning_rate": 9.747209754420764e-06,
+      "loss": 0.6387,
+      "step": 7899
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.816533088684082,
+      "learning_rate": 9.74712697427596e-06,
+      "loss": 0.4213,
+      "step": 7900
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.505401849746704,
+      "learning_rate": 9.747044180931176e-06,
+      "loss": 0.7632,
+      "step": 7901
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.3260252475738525,
+      "learning_rate": 9.746961374386642e-06,
+      "loss": 0.917,
+      "step": 7902
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.005814790725708,
+      "learning_rate": 9.746878554642584e-06,
+      "loss": 0.7101,
+      "step": 7903
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.4245223999023438,
+      "learning_rate": 9.746795721699236e-06,
+      "loss": 0.5897,
+      "step": 7904
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.3943450450897217,
+      "learning_rate": 9.746712875556828e-06,
+      "loss": 0.792,
+      "step": 7905
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.6807972192764282,
+      "learning_rate": 9.746630016215588e-06,
+      "loss": 0.5119,
+      "step": 7906
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.7673587799072266,
+      "learning_rate": 9.74654714367575e-06,
+      "loss": 0.7286,
+      "step": 7907
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.8953444957733154,
+      "learning_rate": 9.746464257937542e-06,
+      "loss": 0.7704,
+      "step": 7908
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.612577438354492,
+      "learning_rate": 9.746381359001193e-06,
+      "loss": 0.7434,
+      "step": 7909
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.6591384410858154,
+      "learning_rate": 9.746298446866937e-06,
+      "loss": 1.0561,
+      "step": 7910
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.6280100345611572,
+      "learning_rate": 9.746215521535004e-06,
+      "loss": 0.6788,
+      "step": 7911
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.1107337474823,
+      "learning_rate": 9.746132583005622e-06,
+      "loss": 0.6046,
+      "step": 7912
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.2951717376708984,
+      "learning_rate": 9.746049631279024e-06,
+      "loss": 0.6241,
+      "step": 7913
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.0463478565216064,
+      "learning_rate": 9.745966666355439e-06,
+      "loss": 0.5324,
+      "step": 7914
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.13370680809021,
+      "learning_rate": 9.745883688235101e-06,
+      "loss": 0.5618,
+      "step": 7915
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.4307056665420532,
+      "learning_rate": 9.745800696918235e-06,
+      "loss": 0.5119,
+      "step": 7916
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.2406699657440186,
+      "learning_rate": 9.745717692405077e-06,
+      "loss": 0.5967,
+      "step": 7917
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.149289131164551,
+      "learning_rate": 9.745634674695856e-06,
+      "loss": 0.5826,
+      "step": 7918
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 6.435347080230713,
+      "learning_rate": 9.745551643790803e-06,
+      "loss": 0.7513,
+      "step": 7919
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.249586582183838,
+      "learning_rate": 9.745468599690146e-06,
+      "loss": 0.743,
+      "step": 7920
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.149160146713257,
+      "learning_rate": 9.745385542394121e-06,
+      "loss": 0.6463,
+      "step": 7921
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.3108201026916504,
+      "learning_rate": 9.745302471902955e-06,
+      "loss": 0.6009,
+      "step": 7922
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.230402708053589,
+      "learning_rate": 9.745219388216882e-06,
+      "loss": 0.8005,
+      "step": 7923
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.387037992477417,
+      "learning_rate": 9.745136291336131e-06,
+      "loss": 0.8848,
+      "step": 7924
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.3885233402252197,
+      "learning_rate": 9.745053181260933e-06,
+      "loss": 0.7926,
+      "step": 7925
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.4441726207733154,
+      "learning_rate": 9.744970057991522e-06,
+      "loss": 0.5719,
+      "step": 7926
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.0285816192626953,
+      "learning_rate": 9.744886921528123e-06,
+      "loss": 0.6709,
+      "step": 7927
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.8063957691192627,
+      "learning_rate": 9.744803771870972e-06,
+      "loss": 0.8095,
+      "step": 7928
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.9321238994598389,
+      "learning_rate": 9.7447206090203e-06,
+      "loss": 0.7019,
+      "step": 7929
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.7231998443603516,
+      "learning_rate": 9.744637432976339e-06,
+      "loss": 0.7497,
+      "step": 7930
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.693543553352356,
+      "learning_rate": 9.744554243739317e-06,
+      "loss": 0.6806,
+      "step": 7931
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.5800986289978027,
+      "learning_rate": 9.744471041309467e-06,
+      "loss": 0.7657,
+      "step": 7932
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.2692313194274902,
+      "learning_rate": 9.74438782568702e-06,
+      "loss": 0.5516,
+      "step": 7933
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.3986308574676514,
+      "learning_rate": 9.744304596872207e-06,
+      "loss": 0.6742,
+      "step": 7934
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.6279137134552,
+      "learning_rate": 9.744221354865263e-06,
+      "loss": 0.6456,
+      "step": 7935
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.6003668308258057,
+      "learning_rate": 9.744138099666414e-06,
+      "loss": 0.6385,
+      "step": 7936
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.665456533432007,
+      "learning_rate": 9.744054831275894e-06,
+      "loss": 0.7032,
+      "step": 7937
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.954132080078125,
+      "learning_rate": 9.743971549693936e-06,
+      "loss": 0.8078,
+      "step": 7938
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.155665159225464,
+      "learning_rate": 9.74388825492077e-06,
+      "loss": 0.7938,
+      "step": 7939
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.6996753215789795,
+      "learning_rate": 9.743804946956627e-06,
+      "loss": 0.7117,
+      "step": 7940
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 4.509688854217529,
+      "learning_rate": 9.743721625801742e-06,
+      "loss": 0.4284,
+      "step": 7941
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.2071590423583984,
+      "learning_rate": 9.743638291456341e-06,
+      "loss": 0.6417,
+      "step": 7942
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.7249977588653564,
+      "learning_rate": 9.74355494392066e-06,
+      "loss": 0.8118,
+      "step": 7943
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.769202470779419,
+      "learning_rate": 9.743471583194932e-06,
+      "loss": 0.55,
+      "step": 7944
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.050330877304077,
+      "learning_rate": 9.743388209279382e-06,
+      "loss": 0.6983,
+      "step": 7945
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.6090407371520996,
+      "learning_rate": 9.743304822174251e-06,
+      "loss": 0.8774,
+      "step": 7946
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.443161964416504,
+      "learning_rate": 9.743221421879763e-06,
+      "loss": 0.7798,
+      "step": 7947
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.490389347076416,
+      "learning_rate": 9.743138008396155e-06,
+      "loss": 0.465,
+      "step": 7948
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.5586719512939453,
+      "learning_rate": 9.743054581723655e-06,
+      "loss": 0.6986,
+      "step": 7949
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.2091288566589355,
+      "learning_rate": 9.742971141862498e-06,
+      "loss": 0.6222,
+      "step": 7950
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.5611090660095215,
+      "learning_rate": 9.742887688812916e-06,
+      "loss": 0.6868,
+      "step": 7951
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.365009069442749,
+      "learning_rate": 9.74280422257514e-06,
+      "loss": 0.66,
+      "step": 7952
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.791562080383301,
+      "learning_rate": 9.742720743149402e-06,
+      "loss": 0.6499,
+      "step": 7953
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.093517303466797,
+      "learning_rate": 9.742637250535932e-06,
+      "loss": 0.6642,
+      "step": 7954
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.633668899536133,
+      "learning_rate": 9.742553744734966e-06,
+      "loss": 0.7242,
+      "step": 7955
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.8853797912597656,
+      "learning_rate": 9.742470225746735e-06,
+      "loss": 0.6694,
+      "step": 7956
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.831775188446045,
+      "learning_rate": 9.74238669357147e-06,
+      "loss": 0.6576,
+      "step": 7957
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.834179401397705,
+      "learning_rate": 9.742303148209405e-06,
+      "loss": 0.5101,
+      "step": 7958
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.408895492553711,
+      "learning_rate": 9.742219589660772e-06,
+      "loss": 0.5889,
+      "step": 7959
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.011000871658325,
+      "learning_rate": 9.7421360179258e-06,
+      "loss": 0.5401,
+      "step": 7960
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 7.700663089752197,
+      "learning_rate": 9.742052433004726e-06,
+      "loss": 0.4993,
+      "step": 7961
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.8253285884857178,
+      "learning_rate": 9.741968834897782e-06,
+      "loss": 0.5574,
+      "step": 7962
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.693087100982666,
+      "learning_rate": 9.741885223605198e-06,
+      "loss": 0.6864,
+      "step": 7963
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.1797821521759033,
+      "learning_rate": 9.741801599127208e-06,
+      "loss": 0.6914,
+      "step": 7964
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 4.180696964263916,
+      "learning_rate": 9.741717961464044e-06,
+      "loss": 0.623,
+      "step": 7965
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.7370296716690063,
+      "learning_rate": 9.741634310615938e-06,
+      "loss": 0.4158,
+      "step": 7966
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 4.1495866775512695,
+      "learning_rate": 9.741550646583122e-06,
+      "loss": 0.7665,
+      "step": 7967
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.017561435699463,
+      "learning_rate": 9.741466969365832e-06,
+      "loss": 0.4927,
+      "step": 7968
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.9647942781448364,
+      "learning_rate": 9.741383278964297e-06,
+      "loss": 0.7126,
+      "step": 7969
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.6579055786132812,
+      "learning_rate": 9.741299575378753e-06,
+      "loss": 0.4517,
+      "step": 7970
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.1457788944244385,
+      "learning_rate": 9.74121585860943e-06,
+      "loss": 0.836,
+      "step": 7971
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.2670655250549316,
+      "learning_rate": 9.741132128656561e-06,
+      "loss": 0.7514,
+      "step": 7972
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.043348550796509,
+      "learning_rate": 9.74104838552038e-06,
+      "loss": 0.5536,
+      "step": 7973
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.428164005279541,
+      "learning_rate": 9.74096462920112e-06,
+      "loss": 0.6459,
+      "step": 7974
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.8025050163269043,
+      "learning_rate": 9.740880859699014e-06,
+      "loss": 0.5411,
+      "step": 7975
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.8478292226791382,
+      "learning_rate": 9.740797077014292e-06,
+      "loss": 0.8791,
+      "step": 7976
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.0099546909332275,
+      "learning_rate": 9.740713281147191e-06,
+      "loss": 0.6214,
+      "step": 7977
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.3033194541931152,
+      "learning_rate": 9.740629472097944e-06,
+      "loss": 0.3339,
+      "step": 7978
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.9720135927200317,
+      "learning_rate": 9.74054564986678e-06,
+      "loss": 0.5866,
+      "step": 7979
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.5545730590820312,
+      "learning_rate": 9.740461814453935e-06,
+      "loss": 0.7759,
+      "step": 7980
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.5777971744537354,
+      "learning_rate": 9.740377965859641e-06,
+      "loss": 0.7286,
+      "step": 7981
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.9294064044952393,
+      "learning_rate": 9.740294104084131e-06,
+      "loss": 0.7047,
+      "step": 7982
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.875247001647949,
+      "learning_rate": 9.74021022912764e-06,
+      "loss": 0.6373,
+      "step": 7983
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.918940782546997,
+      "learning_rate": 9.7401263409904e-06,
+      "loss": 0.7115,
+      "step": 7984
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.8998641967773438,
+      "learning_rate": 9.740042439672644e-06,
+      "loss": 0.8468,
+      "step": 7985
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.4084835052490234,
+      "learning_rate": 9.739958525174607e-06,
+      "loss": 0.784,
+      "step": 7986
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.0089125633239746,
+      "learning_rate": 9.73987459749652e-06,
+      "loss": 0.4446,
+      "step": 7987
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.0791921615600586,
+      "learning_rate": 9.739790656638617e-06,
+      "loss": 0.8474,
+      "step": 7988
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.3235278129577637,
+      "learning_rate": 9.739706702601132e-06,
+      "loss": 0.5424,
+      "step": 7989
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.837167739868164,
+      "learning_rate": 9.739622735384297e-06,
+      "loss": 0.7129,
+      "step": 7990
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.1450212001800537,
+      "learning_rate": 9.73953875498835e-06,
+      "loss": 0.5271,
+      "step": 7991
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.171513795852661,
+      "learning_rate": 9.739454761413518e-06,
+      "loss": 0.5501,
+      "step": 7992
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.6654229164123535,
+      "learning_rate": 9.739370754660038e-06,
+      "loss": 0.8026,
+      "step": 7993
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.9191070795059204,
+      "learning_rate": 9.739286734728146e-06,
+      "loss": 0.6267,
+      "step": 7994
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.9893710613250732,
+      "learning_rate": 9.73920270161807e-06,
+      "loss": 0.7235,
+      "step": 7995
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.524308443069458,
+      "learning_rate": 9.739118655330047e-06,
+      "loss": 0.656,
+      "step": 7996
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.653045654296875,
+      "learning_rate": 9.739034595864311e-06,
+      "loss": 0.4858,
+      "step": 7997
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.954315423965454,
+      "learning_rate": 9.738950523221096e-06,
+      "loss": 0.8951,
+      "step": 7998
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.4986425638198853,
+      "learning_rate": 9.738866437400634e-06,
+      "loss": 0.7491,
+      "step": 7999
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.4359216690063477,
+      "learning_rate": 9.738782338403159e-06,
+      "loss": 0.6781,
+      "step": 8000
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.6871566772460938,
+      "learning_rate": 9.738698226228907e-06,
+      "loss": 0.5842,
+      "step": 8001
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.527756690979004,
+      "learning_rate": 9.738614100878109e-06,
+      "loss": 0.7966,
+      "step": 8002
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.8120081424713135,
+      "learning_rate": 9.738529962351e-06,
+      "loss": 0.5331,
+      "step": 8003
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 4.09879732131958,
+      "learning_rate": 9.738445810647816e-06,
+      "loss": 0.5401,
+      "step": 8004
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.9455033540725708,
+      "learning_rate": 9.738361645768788e-06,
+      "loss": 0.6193,
+      "step": 8005
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.1373305320739746,
+      "learning_rate": 9.738277467714153e-06,
+      "loss": 0.7891,
+      "step": 8006
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.5512168407440186,
+      "learning_rate": 9.738193276484141e-06,
+      "loss": 0.5678,
+      "step": 8007
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.9056892395019531,
+      "learning_rate": 9.738109072078989e-06,
+      "loss": 0.695,
+      "step": 8008
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.9166715145111084,
+      "learning_rate": 9.738024854498932e-06,
+      "loss": 0.4667,
+      "step": 8009
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.1952462196350098,
+      "learning_rate": 9.737940623744202e-06,
+      "loss": 0.5707,
+      "step": 8010
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 4.054886341094971,
+      "learning_rate": 9.737856379815034e-06,
+      "loss": 0.978,
+      "step": 8011
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.7589895725250244,
+      "learning_rate": 9.737772122711662e-06,
+      "loss": 0.8221,
+      "step": 8012
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.254411458969116,
+      "learning_rate": 9.737687852434322e-06,
+      "loss": 0.7243,
+      "step": 8013
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.0507893562316895,
+      "learning_rate": 9.737603568983243e-06,
+      "loss": 0.8895,
+      "step": 8014
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.6877700090408325,
+      "learning_rate": 9.737519272358667e-06,
+      "loss": 0.48,
+      "step": 8015
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 4.222691059112549,
+      "learning_rate": 9.737434962560824e-06,
+      "loss": 0.6804,
+      "step": 8016
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.8272614479064941,
+      "learning_rate": 9.737350639589948e-06,
+      "loss": 0.3899,
+      "step": 8017
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.805776596069336,
+      "learning_rate": 9.737266303446275e-06,
+      "loss": 0.5095,
+      "step": 8018
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.6464450359344482,
+      "learning_rate": 9.737181954130038e-06,
+      "loss": 0.8718,
+      "step": 8019
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.230653762817383,
+      "learning_rate": 9.737097591641474e-06,
+      "loss": 0.73,
+      "step": 8020
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.8444353342056274,
+      "learning_rate": 9.737013215980815e-06,
+      "loss": 0.6663,
+      "step": 8021
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.4017863273620605,
+      "learning_rate": 9.736928827148298e-06,
+      "loss": 0.6114,
+      "step": 8022
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.2687387466430664,
+      "learning_rate": 9.736844425144156e-06,
+      "loss": 0.598,
+      "step": 8023
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.737765312194824,
+      "learning_rate": 9.736760009968624e-06,
+      "loss": 0.7077,
+      "step": 8024
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.322732925415039,
+      "learning_rate": 9.736675581621937e-06,
+      "loss": 0.638,
+      "step": 8025
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.553743839263916,
+      "learning_rate": 9.73659114010433e-06,
+      "loss": 0.5241,
+      "step": 8026
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.739694595336914,
+      "learning_rate": 9.736506685416036e-06,
+      "loss": 0.8182,
+      "step": 8027
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.6212602853775024,
+      "learning_rate": 9.736422217557293e-06,
+      "loss": 0.5968,
+      "step": 8028
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.6037487983703613,
+      "learning_rate": 9.736337736528333e-06,
+      "loss": 0.8051,
+      "step": 8029
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.1240291595458984,
+      "learning_rate": 9.736253242329392e-06,
+      "loss": 0.6026,
+      "step": 8030
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.217376947402954,
+      "learning_rate": 9.736168734960704e-06,
+      "loss": 0.6991,
+      "step": 8031
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.512293577194214,
+      "learning_rate": 9.736084214422507e-06,
+      "loss": 0.4114,
+      "step": 8032
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.1409976482391357,
+      "learning_rate": 9.735999680715033e-06,
+      "loss": 0.4808,
+      "step": 8033
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.4754719734191895,
+      "learning_rate": 9.735915133838518e-06,
+      "loss": 0.5101,
+      "step": 8034
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.2610933780670166,
+      "learning_rate": 9.735830573793198e-06,
+      "loss": 0.642,
+      "step": 8035
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 5.722482204437256,
+      "learning_rate": 9.735746000579308e-06,
+      "loss": 0.9336,
+      "step": 8036
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.8294541835784912,
+      "learning_rate": 9.735661414197079e-06,
+      "loss": 0.6406,
+      "step": 8037
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 7.162172794342041,
+      "learning_rate": 9.735576814646752e-06,
+      "loss": 0.6649,
+      "step": 8038
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.109518051147461,
+      "learning_rate": 9.73549220192856e-06,
+      "loss": 0.6407,
+      "step": 8039
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.7692432403564453,
+      "learning_rate": 9.735407576042737e-06,
+      "loss": 0.4089,
+      "step": 8040
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.9114041328430176,
+      "learning_rate": 9.735322936989522e-06,
+      "loss": 0.743,
+      "step": 8041
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.6181859970092773,
+      "learning_rate": 9.735238284769145e-06,
+      "loss": 0.9178,
+      "step": 8042
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.340803861618042,
+      "learning_rate": 9.735153619381846e-06,
+      "loss": 0.574,
+      "step": 8043
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.340843677520752,
+      "learning_rate": 9.735068940827857e-06,
+      "loss": 0.376,
+      "step": 8044
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.926384210586548,
+      "learning_rate": 9.734984249107415e-06,
+      "loss": 0.5951,
+      "step": 8045
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.3297810554504395,
+      "learning_rate": 9.734899544220757e-06,
+      "loss": 0.7332,
+      "step": 8046
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.9739034175872803,
+      "learning_rate": 9.734814826168117e-06,
+      "loss": 0.7528,
+      "step": 8047
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 4.646662712097168,
+      "learning_rate": 9.73473009494973e-06,
+      "loss": 0.7658,
+      "step": 8048
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.4600107669830322,
+      "learning_rate": 9.734645350565832e-06,
+      "loss": 0.7064,
+      "step": 8049
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.011240005493164,
+      "learning_rate": 9.734560593016659e-06,
+      "loss": 0.6034,
+      "step": 8050
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.333536148071289,
+      "learning_rate": 9.734475822302448e-06,
+      "loss": 0.9593,
+      "step": 8051
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.6765217781066895,
+      "learning_rate": 9.734391038423432e-06,
+      "loss": 0.7756,
+      "step": 8052
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.4838922023773193,
+      "learning_rate": 9.734306241379847e-06,
+      "loss": 0.7092,
+      "step": 8053
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.8726792335510254,
+      "learning_rate": 9.734221431171932e-06,
+      "loss": 0.51,
+      "step": 8054
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.8437082767486572,
+      "learning_rate": 9.73413660779992e-06,
+      "loss": 0.7553,
+      "step": 8055
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.104936122894287,
+      "learning_rate": 9.734051771264047e-06,
+      "loss": 0.6164,
+      "step": 8056
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.9365615844726562,
+      "learning_rate": 9.733966921564551e-06,
+      "loss": 0.6259,
+      "step": 8057
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.7976548671722412,
+      "learning_rate": 9.733882058701664e-06,
+      "loss": 0.6522,
+      "step": 8058
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.3786532878875732,
+      "learning_rate": 9.733797182675627e-06,
+      "loss": 0.8703,
+      "step": 8059
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.69944429397583,
+      "learning_rate": 9.73371229348667e-06,
+      "loss": 0.6198,
+      "step": 8060
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 6.285220623016357,
+      "learning_rate": 9.733627391135034e-06,
+      "loss": 0.8213,
+      "step": 8061
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.1475799083709717,
+      "learning_rate": 9.733542475620954e-06,
+      "loss": 0.3981,
+      "step": 8062
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.148174285888672,
+      "learning_rate": 9.733457546944666e-06,
+      "loss": 0.5377,
+      "step": 8063
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.27129864692688,
+      "learning_rate": 9.733372605106405e-06,
+      "loss": 0.6787,
+      "step": 8064
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.9909954071044922,
+      "learning_rate": 9.733287650106408e-06,
+      "loss": 0.675,
+      "step": 8065
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.1307919025421143,
+      "learning_rate": 9.73320268194491e-06,
+      "loss": 0.5137,
+      "step": 8066
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.635437488555908,
+      "learning_rate": 9.733117700622152e-06,
+      "loss": 0.6712,
+      "step": 8067
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.834031343460083,
+      "learning_rate": 9.733032706138363e-06,
+      "loss": 0.5909,
+      "step": 8068
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.892627000808716,
+      "learning_rate": 9.732947698493785e-06,
+      "loss": 0.7245,
+      "step": 8069
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.5577735900878906,
+      "learning_rate": 9.732862677688653e-06,
+      "loss": 0.7784,
+      "step": 8070
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.5480377674102783,
+      "learning_rate": 9.732777643723202e-06,
+      "loss": 0.5843,
+      "step": 8071
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.2707370519638062,
+      "learning_rate": 9.732692596597669e-06,
+      "loss": 0.4607,
+      "step": 8072
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 4.670010566711426,
+      "learning_rate": 9.73260753631229e-06,
+      "loss": 0.7456,
+      "step": 8073
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.05838942527771,
+      "learning_rate": 9.732522462867302e-06,
+      "loss": 0.5888,
+      "step": 8074
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.3910958766937256,
+      "learning_rate": 9.732437376262945e-06,
+      "loss": 0.5684,
+      "step": 8075
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.203859567642212,
+      "learning_rate": 9.73235227649945e-06,
+      "loss": 0.6522,
+      "step": 8076
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.435966730117798,
+      "learning_rate": 9.732267163577056e-06,
+      "loss": 0.7547,
+      "step": 8077
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.286742687225342,
+      "learning_rate": 9.732182037496001e-06,
+      "loss": 0.5962,
+      "step": 8078
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.9290666580200195,
+      "learning_rate": 9.73209689825652e-06,
+      "loss": 0.7516,
+      "step": 8079
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.138613224029541,
+      "learning_rate": 9.73201174585885e-06,
+      "loss": 0.7623,
+      "step": 8080
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.9267518520355225,
+      "learning_rate": 9.731926580303228e-06,
+      "loss": 0.5144,
+      "step": 8081
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.8318417072296143,
+      "learning_rate": 9.731841401589892e-06,
+      "loss": 0.5933,
+      "step": 8082
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 4.033509254455566,
+      "learning_rate": 9.731756209719076e-06,
+      "loss": 0.5943,
+      "step": 8083
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.5922296047210693,
+      "learning_rate": 9.731671004691019e-06,
+      "loss": 0.7115,
+      "step": 8084
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.6887829303741455,
+      "learning_rate": 9.731585786505957e-06,
+      "loss": 0.8875,
+      "step": 8085
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.7754853963851929,
+      "learning_rate": 9.731500555164128e-06,
+      "loss": 0.7409,
+      "step": 8086
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.0233349800109863,
+      "learning_rate": 9.731415310665768e-06,
+      "loss": 0.7089,
+      "step": 8087
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.010394334793091,
+      "learning_rate": 9.731330053011116e-06,
+      "loss": 0.8075,
+      "step": 8088
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.2754158973693848,
+      "learning_rate": 9.731244782200406e-06,
+      "loss": 0.6214,
+      "step": 8089
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.6240415573120117,
+      "learning_rate": 9.731159498233877e-06,
+      "loss": 0.6487,
+      "step": 8090
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.437798261642456,
+      "learning_rate": 9.731074201111766e-06,
+      "loss": 0.2817,
+      "step": 8091
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.4640512466430664,
+      "learning_rate": 9.73098889083431e-06,
+      "loss": 0.5221,
+      "step": 8092
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.619375467300415,
+      "learning_rate": 9.730903567401745e-06,
+      "loss": 0.7467,
+      "step": 8093
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.2857682704925537,
+      "learning_rate": 9.730818230814312e-06,
+      "loss": 0.5907,
+      "step": 8094
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.0403189659118652,
+      "learning_rate": 9.730732881072243e-06,
+      "loss": 0.7522,
+      "step": 8095
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.9891854524612427,
+      "learning_rate": 9.73064751817578e-06,
+      "loss": 0.766,
+      "step": 8096
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.1245603561401367,
+      "learning_rate": 9.730562142125156e-06,
+      "loss": 0.6172,
+      "step": 8097
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.675814390182495,
+      "learning_rate": 9.730476752920613e-06,
+      "loss": 0.4771,
+      "step": 8098
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.798581600189209,
+      "learning_rate": 9.730391350562384e-06,
+      "loss": 0.7371,
+      "step": 8099
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.168785572052002,
+      "learning_rate": 9.730305935050712e-06,
+      "loss": 0.5839,
+      "step": 8100
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.51155948638916,
+      "learning_rate": 9.730220506385829e-06,
+      "loss": 0.5524,
+      "step": 8101
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.0767717361450195,
+      "learning_rate": 9.730135064567974e-06,
+      "loss": 0.7767,
+      "step": 8102
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.6516460180282593,
+      "learning_rate": 9.730049609597387e-06,
+      "loss": 0.631,
+      "step": 8103
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.040296792984009,
+      "learning_rate": 9.729964141474304e-06,
+      "loss": 0.9087,
+      "step": 8104
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.7618106603622437,
+      "learning_rate": 9.729878660198961e-06,
+      "loss": 0.7699,
+      "step": 8105
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 5.006500244140625,
+      "learning_rate": 9.729793165771599e-06,
+      "loss": 0.988,
+      "step": 8106
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.309130907058716,
+      "learning_rate": 9.729707658192452e-06,
+      "loss": 0.702,
+      "step": 8107
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.0511081218719482,
+      "learning_rate": 9.729622137461763e-06,
+      "loss": 0.7812,
+      "step": 8108
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.450977087020874,
+      "learning_rate": 9.729536603579765e-06,
+      "loss": 0.6899,
+      "step": 8109
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 4.953139781951904,
+      "learning_rate": 9.729451056546696e-06,
+      "loss": 0.7998,
+      "step": 8110
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.6880874633789062,
+      "learning_rate": 9.729365496362798e-06,
+      "loss": 0.6221,
+      "step": 8111
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.2547786235809326,
+      "learning_rate": 9.729279923028304e-06,
+      "loss": 0.8383,
+      "step": 8112
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.3132400512695312,
+      "learning_rate": 9.729194336543454e-06,
+      "loss": 0.6924,
+      "step": 8113
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.431792736053467,
+      "learning_rate": 9.729108736908488e-06,
+      "loss": 0.8242,
+      "step": 8114
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.8985469341278076,
+      "learning_rate": 9.729023124123641e-06,
+      "loss": 0.6438,
+      "step": 8115
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.6414897441864014,
+      "learning_rate": 9.728937498189152e-06,
+      "loss": 0.9763,
+      "step": 8116
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.4371612071990967,
+      "learning_rate": 9.72885185910526e-06,
+      "loss": 0.8069,
+      "step": 8117
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.577061176300049,
+      "learning_rate": 9.728766206872203e-06,
+      "loss": 0.6961,
+      "step": 8118
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.0516366958618164,
+      "learning_rate": 9.728680541490217e-06,
+      "loss": 0.6448,
+      "step": 8119
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.600722551345825,
+      "learning_rate": 9.728594862959542e-06,
+      "loss": 0.6718,
+      "step": 8120
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.959260940551758,
+      "learning_rate": 9.728509171280417e-06,
+      "loss": 0.4896,
+      "step": 8121
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 4.828045845031738,
+      "learning_rate": 9.728423466453078e-06,
+      "loss": 0.6014,
+      "step": 8122
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.2326033115386963,
+      "learning_rate": 9.728337748477768e-06,
+      "loss": 0.6844,
+      "step": 8123
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.0132994651794434,
+      "learning_rate": 9.728252017354717e-06,
+      "loss": 0.7888,
+      "step": 8124
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 4.547303676605225,
+      "learning_rate": 9.728166273084172e-06,
+      "loss": 0.5132,
+      "step": 8125
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.3586113452911377,
+      "learning_rate": 9.728080515666367e-06,
+      "loss": 0.5562,
+      "step": 8126
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.3833212852478027,
+      "learning_rate": 9.72799474510154e-06,
+      "loss": 0.8794,
+      "step": 8127
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.6160027980804443,
+      "learning_rate": 9.727908961389933e-06,
+      "loss": 0.715,
+      "step": 8128
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.283765435218811,
+      "learning_rate": 9.727823164531781e-06,
+      "loss": 0.5833,
+      "step": 8129
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 5.442291259765625,
+      "learning_rate": 9.727737354527325e-06,
+      "loss": 0.8222,
+      "step": 8130
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.140907049179077,
+      "learning_rate": 9.7276515313768e-06,
+      "loss": 0.6333,
+      "step": 8131
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.9312268495559692,
+      "learning_rate": 9.72756569508045e-06,
+      "loss": 0.5637,
+      "step": 8132
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 4.200614929199219,
+      "learning_rate": 9.727479845638508e-06,
+      "loss": 0.6563,
+      "step": 8133
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.387484312057495,
+      "learning_rate": 9.727393983051218e-06,
+      "loss": 0.8204,
+      "step": 8134
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.879565477371216,
+      "learning_rate": 9.727308107318815e-06,
+      "loss": 0.7299,
+      "step": 8135
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.2482495307922363,
+      "learning_rate": 9.72722221844154e-06,
+      "loss": 0.7111,
+      "step": 8136
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.288032054901123,
+      "learning_rate": 9.727136316419629e-06,
+      "loss": 0.7254,
+      "step": 8137
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 5.065507888793945,
+      "learning_rate": 9.727050401253324e-06,
+      "loss": 0.8283,
+      "step": 8138
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.189253330230713,
+      "learning_rate": 9.726964472942862e-06,
+      "loss": 0.7482,
+      "step": 8139
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.75866436958313,
+      "learning_rate": 9.726878531488483e-06,
+      "loss": 0.7963,
+      "step": 8140
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.145906448364258,
+      "learning_rate": 9.726792576890425e-06,
+      "loss": 0.5601,
+      "step": 8141
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.454735517501831,
+      "learning_rate": 9.726706609148929e-06,
+      "loss": 0.4543,
+      "step": 8142
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.37931227684021,
+      "learning_rate": 9.726620628264231e-06,
+      "loss": 0.7703,
+      "step": 8143
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.0912423133850098,
+      "learning_rate": 9.726534634236574e-06,
+      "loss": 0.5553,
+      "step": 8144
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.656435489654541,
+      "learning_rate": 9.726448627066192e-06,
+      "loss": 0.6474,
+      "step": 8145
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.1891839504241943,
+      "learning_rate": 9.726362606753327e-06,
+      "loss": 0.31,
+      "step": 8146
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.3149237632751465,
+      "learning_rate": 9.726276573298219e-06,
+      "loss": 0.6851,
+      "step": 8147
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.485307216644287,
+      "learning_rate": 9.726190526701107e-06,
+      "loss": 0.6781,
+      "step": 8148
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.4972989559173584,
+      "learning_rate": 9.72610446696223e-06,
+      "loss": 0.4641,
+      "step": 8149
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.137742042541504,
+      "learning_rate": 9.726018394081824e-06,
+      "loss": 0.736,
+      "step": 8150
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.9426517486572266,
+      "learning_rate": 9.725932308060134e-06,
+      "loss": 0.6519,
+      "step": 8151
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.1533615589141846,
+      "learning_rate": 9.725846208897397e-06,
+      "loss": 0.5981,
+      "step": 8152
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.460062026977539,
+      "learning_rate": 9.72576009659385e-06,
+      "loss": 0.8536,
+      "step": 8153
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.176224708557129,
+      "learning_rate": 9.725673971149736e-06,
+      "loss": 0.672,
+      "step": 8154
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.280243158340454,
+      "learning_rate": 9.725587832565292e-06,
+      "loss": 0.5948,
+      "step": 8155
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.002277135848999,
+      "learning_rate": 9.72550168084076e-06,
+      "loss": 0.5935,
+      "step": 8156
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.4269471168518066,
+      "learning_rate": 9.725415515976375e-06,
+      "loss": 0.6916,
+      "step": 8157
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.3870816230773926,
+      "learning_rate": 9.725329337972383e-06,
+      "loss": 0.5211,
+      "step": 8158
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.9059656858444214,
+      "learning_rate": 9.725243146829018e-06,
+      "loss": 0.8965,
+      "step": 8159
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.185955047607422,
+      "learning_rate": 9.725156942546523e-06,
+      "loss": 0.5903,
+      "step": 8160
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.8392040729522705,
+      "learning_rate": 9.725070725125137e-06,
+      "loss": 0.7125,
+      "step": 8161
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 6.114715099334717,
+      "learning_rate": 9.724984494565098e-06,
+      "loss": 0.8375,
+      "step": 8162
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.215020179748535,
+      "learning_rate": 9.724898250866648e-06,
+      "loss": 0.4743,
+      "step": 8163
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.624232292175293,
+      "learning_rate": 9.724811994030025e-06,
+      "loss": 0.7493,
+      "step": 8164
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.3765387535095215,
+      "learning_rate": 9.72472572405547e-06,
+      "loss": 0.585,
+      "step": 8165
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.21187424659729,
+      "learning_rate": 9.724639440943225e-06,
+      "loss": 0.6794,
+      "step": 8166
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.649609088897705,
+      "learning_rate": 9.724553144693525e-06,
+      "loss": 0.6908,
+      "step": 8167
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.620192050933838,
+      "learning_rate": 9.724466835306615e-06,
+      "loss": 0.7706,
+      "step": 8168
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.3717455863952637,
+      "learning_rate": 9.72438051278273e-06,
+      "loss": 0.5843,
+      "step": 8169
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.500971794128418,
+      "learning_rate": 9.724294177122114e-06,
+      "loss": 0.6101,
+      "step": 8170
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.949434518814087,
+      "learning_rate": 9.724207828325005e-06,
+      "loss": 0.787,
+      "step": 8171
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.9969654083251953,
+      "learning_rate": 9.724121466391643e-06,
+      "loss": 0.5223,
+      "step": 8172
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.722892165184021,
+      "learning_rate": 9.72403509132227e-06,
+      "loss": 0.4403,
+      "step": 8173
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.9125425815582275,
+      "learning_rate": 9.723948703117125e-06,
+      "loss": 0.4401,
+      "step": 8174
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 6.089273929595947,
+      "learning_rate": 9.723862301776448e-06,
+      "loss": 0.6575,
+      "step": 8175
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.1571993827819824,
+      "learning_rate": 9.72377588730048e-06,
+      "loss": 0.6692,
+      "step": 8176
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.005220413208008,
+      "learning_rate": 9.72368945968946e-06,
+      "loss": 0.7987,
+      "step": 8177
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 4.692904472351074,
+      "learning_rate": 9.72360301894363e-06,
+      "loss": 0.733,
+      "step": 8178
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.478661060333252,
+      "learning_rate": 9.72351656506323e-06,
+      "loss": 0.8294,
+      "step": 8179
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.5953519344329834,
+      "learning_rate": 9.723430098048498e-06,
+      "loss": 0.6994,
+      "step": 8180
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.5270648002624512,
+      "learning_rate": 9.72334361789968e-06,
+      "loss": 0.736,
+      "step": 8181
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.7765320539474487,
+      "learning_rate": 9.72325712461701e-06,
+      "loss": 0.5853,
+      "step": 8182
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.2432103157043457,
+      "learning_rate": 9.72317061820073e-06,
+      "loss": 0.7345,
+      "step": 8183
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.7395896911621094,
+      "learning_rate": 9.723084098651083e-06,
+      "loss": 0.7303,
+      "step": 8184
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.4114034175872803,
+      "learning_rate": 9.722997565968311e-06,
+      "loss": 0.6312,
+      "step": 8185
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.4138638973236084,
+      "learning_rate": 9.722911020152648e-06,
+      "loss": 0.5583,
+      "step": 8186
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.885321617126465,
+      "learning_rate": 9.72282446120434e-06,
+      "loss": 0.91,
+      "step": 8187
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.305093288421631,
+      "learning_rate": 9.722737889123628e-06,
+      "loss": 0.8283,
+      "step": 8188
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.8777828216552734,
+      "learning_rate": 9.722651303910751e-06,
+      "loss": 0.5798,
+      "step": 8189
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.886547327041626,
+      "learning_rate": 9.722564705565949e-06,
+      "loss": 0.5997,
+      "step": 8190
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.0537047386169434,
+      "learning_rate": 9.722478094089464e-06,
+      "loss": 0.9428,
+      "step": 8191
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.172670364379883,
+      "learning_rate": 9.722391469481535e-06,
+      "loss": 0.6132,
+      "step": 8192
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.5232386589050293,
+      "learning_rate": 9.722304831742406e-06,
+      "loss": 0.8303,
+      "step": 8193
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.414529800415039,
+      "learning_rate": 9.722218180872316e-06,
+      "loss": 0.6437,
+      "step": 8194
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.9614439010620117,
+      "learning_rate": 9.722131516871506e-06,
+      "loss": 0.5896,
+      "step": 8195
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.8361003398895264,
+      "learning_rate": 9.722044839740217e-06,
+      "loss": 0.4625,
+      "step": 8196
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.296550989151001,
+      "learning_rate": 9.72195814947869e-06,
+      "loss": 0.652,
+      "step": 8197
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.3301194906234741,
+      "learning_rate": 9.721871446087167e-06,
+      "loss": 0.6032,
+      "step": 8198
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.505058765411377,
+      "learning_rate": 9.721784729565888e-06,
+      "loss": 0.6062,
+      "step": 8199
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.54953670501709,
+      "learning_rate": 9.721697999915095e-06,
+      "loss": 0.7765,
+      "step": 8200
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.6362357139587402,
+      "learning_rate": 9.721611257135029e-06,
+      "loss": 0.8615,
+      "step": 8201
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 4.18079137802124,
+      "learning_rate": 9.72152450122593e-06,
+      "loss": 0.765,
+      "step": 8202
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.118089437484741,
+      "learning_rate": 9.721437732188039e-06,
+      "loss": 0.705,
+      "step": 8203
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.579409122467041,
+      "learning_rate": 9.7213509500216e-06,
+      "loss": 0.819,
+      "step": 8204
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.8639661073684692,
+      "learning_rate": 9.721264154726852e-06,
+      "loss": 0.7253,
+      "step": 8205
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.3015313148498535,
+      "learning_rate": 9.721177346304038e-06,
+      "loss": 0.7898,
+      "step": 8206
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.787886381149292,
+      "learning_rate": 9.721090524753398e-06,
+      "loss": 0.5105,
+      "step": 8207
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.655534267425537,
+      "learning_rate": 9.721003690075173e-06,
+      "loss": 0.6479,
+      "step": 8208
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.3287527561187744,
+      "learning_rate": 9.720916842269606e-06,
+      "loss": 0.8025,
+      "step": 8209
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 4.246278285980225,
+      "learning_rate": 9.720829981336937e-06,
+      "loss": 0.5423,
+      "step": 8210
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.7317392826080322,
+      "learning_rate": 9.720743107277408e-06,
+      "loss": 0.4938,
+      "step": 8211
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.7898566722869873,
+      "learning_rate": 9.720656220091262e-06,
+      "loss": 0.8357,
+      "step": 8212
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.9699839353561401,
+      "learning_rate": 9.720569319778739e-06,
+      "loss": 0.726,
+      "step": 8213
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.7392330169677734,
+      "learning_rate": 9.720482406340082e-06,
+      "loss": 0.7194,
+      "step": 8214
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.359384298324585,
+      "learning_rate": 9.72039547977553e-06,
+      "loss": 0.7568,
+      "step": 8215
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.3210489749908447,
+      "learning_rate": 9.720308540085328e-06,
+      "loss": 0.6224,
+      "step": 8216
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.8917617797851562,
+      "learning_rate": 9.720221587269716e-06,
+      "loss": 0.7938,
+      "step": 8217
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.9151942729949951,
+      "learning_rate": 9.720134621328935e-06,
+      "loss": 0.6279,
+      "step": 8218
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.7797751426696777,
+      "learning_rate": 9.720047642263229e-06,
+      "loss": 0.5811,
+      "step": 8219
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.108365058898926,
+      "learning_rate": 9.719960650072838e-06,
+      "loss": 0.4991,
+      "step": 8220
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.7100558280944824,
+      "learning_rate": 9.719873644758004e-06,
+      "loss": 0.7754,
+      "step": 8221
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.8397114276885986,
+      "learning_rate": 9.71978662631897e-06,
+      "loss": 0.6939,
+      "step": 8222
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.943815231323242,
+      "learning_rate": 9.719699594755978e-06,
+      "loss": 0.4394,
+      "step": 8223
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.877474784851074,
+      "learning_rate": 9.719612550069268e-06,
+      "loss": 0.8832,
+      "step": 8224
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.4432525634765625,
+      "learning_rate": 9.719525492259085e-06,
+      "loss": 0.6297,
+      "step": 8225
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.5890889167785645,
+      "learning_rate": 9.719438421325668e-06,
+      "loss": 0.7768,
+      "step": 8226
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.23319673538208,
+      "learning_rate": 9.719351337269263e-06,
+      "loss": 0.5374,
+      "step": 8227
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.2613518238067627,
+      "learning_rate": 9.71926424009011e-06,
+      "loss": 0.5778,
+      "step": 8228
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.2046937942504883,
+      "learning_rate": 9.719177129788448e-06,
+      "loss": 0.4717,
+      "step": 8229
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.8370651006698608,
+      "learning_rate": 9.719090006364525e-06,
+      "loss": 0.8165,
+      "step": 8230
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 5.89475154876709,
+      "learning_rate": 9.71900286981858e-06,
+      "loss": 0.6552,
+      "step": 8231
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.2228081226348877,
+      "learning_rate": 9.718915720150855e-06,
+      "loss": 0.3969,
+      "step": 8232
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.6042909622192383,
+      "learning_rate": 9.718828557361592e-06,
+      "loss": 0.6591,
+      "step": 8233
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.7660696506500244,
+      "learning_rate": 9.718741381451036e-06,
+      "loss": 0.6569,
+      "step": 8234
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 8.42495346069336,
+      "learning_rate": 9.718654192419427e-06,
+      "loss": 0.888,
+      "step": 8235
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.775739908218384,
+      "learning_rate": 9.71856699026701e-06,
+      "loss": 0.7272,
+      "step": 8236
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.501746892929077,
+      "learning_rate": 9.718479774994024e-06,
+      "loss": 0.8747,
+      "step": 8237
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.3549680709838867,
+      "learning_rate": 9.718392546600716e-06,
+      "loss": 0.4917,
+      "step": 8238
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.104820728302002,
+      "learning_rate": 9.718305305087324e-06,
+      "loss": 0.9118,
+      "step": 8239
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.8178417682647705,
+      "learning_rate": 9.718218050454092e-06,
+      "loss": 0.7935,
+      "step": 8240
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.7682676315307617,
+      "learning_rate": 9.718130782701264e-06,
+      "loss": 0.4142,
+      "step": 8241
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.323059558868408,
+      "learning_rate": 9.718043501829083e-06,
+      "loss": 0.4686,
+      "step": 8242
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.694541335105896,
+      "learning_rate": 9.717956207837788e-06,
+      "loss": 0.5247,
+      "step": 8243
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.103229522705078,
+      "learning_rate": 9.717868900727624e-06,
+      "loss": 1.0378,
+      "step": 8244
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.0053341388702393,
+      "learning_rate": 9.717781580498835e-06,
+      "loss": 0.7702,
+      "step": 8245
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.437047243118286,
+      "learning_rate": 9.717694247151663e-06,
+      "loss": 0.6727,
+      "step": 8246
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.6496875286102295,
+      "learning_rate": 9.71760690068635e-06,
+      "loss": 0.5592,
+      "step": 8247
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.067821979522705,
+      "learning_rate": 9.71751954110314e-06,
+      "loss": 0.696,
+      "step": 8248
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.2740249633789062,
+      "learning_rate": 9.717432168402275e-06,
+      "loss": 0.9569,
+      "step": 8249
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.0320048332214355,
+      "learning_rate": 9.717344782583999e-06,
+      "loss": 0.4627,
+      "step": 8250
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.8138675689697266,
+      "learning_rate": 9.717257383648554e-06,
+      "loss": 0.6357,
+      "step": 8251
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.437220335006714,
+      "learning_rate": 9.717169971596184e-06,
+      "loss": 0.7743,
+      "step": 8252
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 4.940139293670654,
+      "learning_rate": 9.71708254642713e-06,
+      "loss": 0.9291,
+      "step": 8253
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.8851783275604248,
+      "learning_rate": 9.716995108141637e-06,
+      "loss": 0.7776,
+      "step": 8254
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.01381516456604,
+      "learning_rate": 9.716907656739948e-06,
+      "loss": 0.8253,
+      "step": 8255
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.879437208175659,
+      "learning_rate": 9.716820192222304e-06,
+      "loss": 0.6623,
+      "step": 8256
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.237848997116089,
+      "learning_rate": 9.716732714588952e-06,
+      "loss": 0.5426,
+      "step": 8257
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.253751754760742,
+      "learning_rate": 9.716645223840133e-06,
+      "loss": 0.6843,
+      "step": 8258
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.9264121055603027,
+      "learning_rate": 9.71655771997609e-06,
+      "loss": 0.7903,
+      "step": 8259
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.1876254081726074,
+      "learning_rate": 9.716470202997069e-06,
+      "loss": 0.6441,
+      "step": 8260
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.0544626712799072,
+      "learning_rate": 9.716382672903309e-06,
+      "loss": 0.4158,
+      "step": 8261
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.869424819946289,
+      "learning_rate": 9.716295129695054e-06,
+      "loss": 0.4851,
+      "step": 8262
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.903773546218872,
+      "learning_rate": 9.716207573372551e-06,
+      "loss": 0.6987,
+      "step": 8263
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 4.972118377685547,
+      "learning_rate": 9.716120003936043e-06,
+      "loss": 0.9924,
+      "step": 8264
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.224799633026123,
+      "learning_rate": 9.71603242138577e-06,
+      "loss": 0.7267,
+      "step": 8265
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.1977109909057617,
+      "learning_rate": 9.715944825721976e-06,
+      "loss": 0.5853,
+      "step": 8266
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.479584217071533,
+      "learning_rate": 9.715857216944908e-06,
+      "loss": 0.6827,
+      "step": 8267
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.7435054779052734,
+      "learning_rate": 9.715769595054809e-06,
+      "loss": 0.8104,
+      "step": 8268
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.8979663848876953,
+      "learning_rate": 9.715681960051919e-06,
+      "loss": 0.4674,
+      "step": 8269
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.469883441925049,
+      "learning_rate": 9.715594311936486e-06,
+      "loss": 0.7235,
+      "step": 8270
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.6552841663360596,
+      "learning_rate": 9.71550665070875e-06,
+      "loss": 0.7562,
+      "step": 8271
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.631533145904541,
+      "learning_rate": 9.715418976368957e-06,
+      "loss": 0.6054,
+      "step": 8272
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.073263645172119,
+      "learning_rate": 9.715331288917351e-06,
+      "loss": 0.7466,
+      "step": 8273
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.5361411571502686,
+      "learning_rate": 9.715243588354175e-06,
+      "loss": 0.6827,
+      "step": 8274
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 4.31883430480957,
+      "learning_rate": 9.71515587467967e-06,
+      "loss": 0.7515,
+      "step": 8275
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 5.615947723388672,
+      "learning_rate": 9.715068147894087e-06,
+      "loss": 0.9084,
+      "step": 8276
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.4868640899658203,
+      "learning_rate": 9.714980407997664e-06,
+      "loss": 0.5105,
+      "step": 8277
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.4006011486053467,
+      "learning_rate": 9.714892654990646e-06,
+      "loss": 0.6599,
+      "step": 8278
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.190483331680298,
+      "learning_rate": 9.71480488887328e-06,
+      "loss": 0.6807,
+      "step": 8279
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.735722303390503,
+      "learning_rate": 9.714717109645806e-06,
+      "loss": 0.72,
+      "step": 8280
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.3455872535705566,
+      "learning_rate": 9.71462931730847e-06,
+      "loss": 1.0374,
+      "step": 8281
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.9445167779922485,
+      "learning_rate": 9.714541511861517e-06,
+      "loss": 0.7351,
+      "step": 8282
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.9652442932128906,
+      "learning_rate": 9.714453693305188e-06,
+      "loss": 0.6645,
+      "step": 8283
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.3821861743927002,
+      "learning_rate": 9.714365861639731e-06,
+      "loss": 0.572,
+      "step": 8284
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.338123321533203,
+      "learning_rate": 9.714278016865387e-06,
+      "loss": 0.7859,
+      "step": 8285
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.119450807571411,
+      "learning_rate": 9.714190158982406e-06,
+      "loss": 0.3638,
+      "step": 8286
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.388209819793701,
+      "learning_rate": 9.714102287991025e-06,
+      "loss": 0.5719,
+      "step": 8287
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.8205745220184326,
+      "learning_rate": 9.714014403891492e-06,
+      "loss": 0.5702,
+      "step": 8288
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.0523128509521484,
+      "learning_rate": 9.71392650668405e-06,
+      "loss": 0.6601,
+      "step": 8289
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.600731134414673,
+      "learning_rate": 9.713838596368945e-06,
+      "loss": 0.4723,
+      "step": 8290
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.4910330772399902,
+      "learning_rate": 9.713750672946421e-06,
+      "loss": 0.4747,
+      "step": 8291
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.0292389392852783,
+      "learning_rate": 9.713662736416722e-06,
+      "loss": 0.7514,
+      "step": 8292
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.127011299133301,
+      "learning_rate": 9.713574786780092e-06,
+      "loss": 0.7114,
+      "step": 8293
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.460104465484619,
+      "learning_rate": 9.713486824036778e-06,
+      "loss": 0.8494,
+      "step": 8294
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.7786458730697632,
+      "learning_rate": 9.713398848187022e-06,
+      "loss": 0.6286,
+      "step": 8295
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.1995513439178467,
+      "learning_rate": 9.713310859231068e-06,
+      "loss": 0.8864,
+      "step": 8296
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.4147205352783203,
+      "learning_rate": 9.713222857169164e-06,
+      "loss": 0.4777,
+      "step": 8297
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.1572773456573486,
+      "learning_rate": 9.713134842001551e-06,
+      "loss": 0.7698,
+      "step": 8298
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.0418074131011963,
+      "learning_rate": 9.713046813728477e-06,
+      "loss": 0.5487,
+      "step": 8299
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.8961198329925537,
+      "learning_rate": 9.712958772350185e-06,
+      "loss": 0.7509,
+      "step": 8300
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.911531686782837,
+      "learning_rate": 9.71287071786692e-06,
+      "loss": 0.5983,
+      "step": 8301
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 4.843429088592529,
+      "learning_rate": 9.712782650278929e-06,
+      "loss": 0.5045,
+      "step": 8302
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.6907896995544434,
+      "learning_rate": 9.712694569586451e-06,
+      "loss": 0.6985,
+      "step": 8303
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.9681429862976074,
+      "learning_rate": 9.712606475789739e-06,
+      "loss": 0.7651,
+      "step": 8304
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.6976068019866943,
+      "learning_rate": 9.712518368889032e-06,
+      "loss": 0.5069,
+      "step": 8305
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.827206611633301,
+      "learning_rate": 9.712430248884576e-06,
+      "loss": 0.7781,
+      "step": 8306
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.0151240825653076,
+      "learning_rate": 9.712342115776617e-06,
+      "loss": 0.947,
+      "step": 8307
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.0073440074920654,
+      "learning_rate": 9.7122539695654e-06,
+      "loss": 0.7309,
+      "step": 8308
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 5.7072553634643555,
+      "learning_rate": 9.712165810251171e-06,
+      "loss": 0.7931,
+      "step": 8309
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.8288465738296509,
+      "learning_rate": 9.712077637834173e-06,
+      "loss": 0.6324,
+      "step": 8310
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.0121548175811768,
+      "learning_rate": 9.711989452314651e-06,
+      "loss": 0.6823,
+      "step": 8311
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.280625820159912,
+      "learning_rate": 9.711901253692855e-06,
+      "loss": 0.6447,
+      "step": 8312
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.1625075340270996,
+      "learning_rate": 9.711813041969025e-06,
+      "loss": 0.6671,
+      "step": 8313
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.987356424331665,
+      "learning_rate": 9.711724817143407e-06,
+      "loss": 0.6223,
+      "step": 8314
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 5.125615119934082,
+      "learning_rate": 9.711636579216248e-06,
+      "loss": 0.878,
+      "step": 8315
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.714984655380249,
+      "learning_rate": 9.711548328187792e-06,
+      "loss": 0.4606,
+      "step": 8316
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.41750168800354,
+      "learning_rate": 9.711460064058287e-06,
+      "loss": 0.6504,
+      "step": 8317
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.8626508712768555,
+      "learning_rate": 9.711371786827975e-06,
+      "loss": 0.7958,
+      "step": 8318
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.9646859169006348,
+      "learning_rate": 9.711283496497103e-06,
+      "loss": 0.7624,
+      "step": 8319
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.4756278991699219,
+      "learning_rate": 9.711195193065918e-06,
+      "loss": 0.7562,
+      "step": 8320
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.969048023223877,
+      "learning_rate": 9.711106876534662e-06,
+      "loss": 0.3988,
+      "step": 8321
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.780036449432373,
+      "learning_rate": 9.711018546903583e-06,
+      "loss": 0.4035,
+      "step": 8322
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.7151734828948975,
+      "learning_rate": 9.710930204172929e-06,
+      "loss": 0.8889,
+      "step": 8323
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.8432581424713135,
+      "learning_rate": 9.71084184834294e-06,
+      "loss": 0.6533,
+      "step": 8324
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.236156702041626,
+      "learning_rate": 9.710753479413863e-06,
+      "loss": 0.8432,
+      "step": 8325
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 6.810915470123291,
+      "learning_rate": 9.710665097385949e-06,
+      "loss": 0.5719,
+      "step": 8326
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.8942373991012573,
+      "learning_rate": 9.710576702259437e-06,
+      "loss": 0.7007,
+      "step": 8327
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.4801394939422607,
+      "learning_rate": 9.710488294034579e-06,
+      "loss": 0.7104,
+      "step": 8328
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.2622199058532715,
+      "learning_rate": 9.710399872711615e-06,
+      "loss": 0.6447,
+      "step": 8329
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.1905713081359863,
+      "learning_rate": 9.710311438290794e-06,
+      "loss": 0.9012,
+      "step": 8330
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.7955844402313232,
+      "learning_rate": 9.710222990772362e-06,
+      "loss": 0.7005,
+      "step": 8331
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.6806702613830566,
+      "learning_rate": 9.710134530156565e-06,
+      "loss": 0.7284,
+      "step": 8332
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.424335241317749,
+      "learning_rate": 9.710046056443647e-06,
+      "loss": 0.6875,
+      "step": 8333
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.919081449508667,
+      "learning_rate": 9.709957569633856e-06,
+      "loss": 0.8482,
+      "step": 8334
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.8743185997009277,
+      "learning_rate": 9.709869069727437e-06,
+      "loss": 0.7693,
+      "step": 8335
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.5243513584136963,
+      "learning_rate": 9.709780556724637e-06,
+      "loss": 0.5882,
+      "step": 8336
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.743018627166748,
+      "learning_rate": 9.709692030625702e-06,
+      "loss": 0.5417,
+      "step": 8337
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.5217270851135254,
+      "learning_rate": 9.709603491430875e-06,
+      "loss": 0.7799,
+      "step": 8338
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.2899210453033447,
+      "learning_rate": 9.709514939140407e-06,
+      "loss": 0.7442,
+      "step": 8339
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 5.211493492126465,
+      "learning_rate": 9.709426373754542e-06,
+      "loss": 0.6035,
+      "step": 8340
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.7324997186660767,
+      "learning_rate": 9.709337795273528e-06,
+      "loss": 0.7488,
+      "step": 8341
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.1986403465270996,
+      "learning_rate": 9.709249203697608e-06,
+      "loss": 0.7547,
+      "step": 8342
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.9092367887496948,
+      "learning_rate": 9.70916059902703e-06,
+      "loss": 0.5695,
+      "step": 8343
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.3174611330032349,
+      "learning_rate": 9.709071981262042e-06,
+      "loss": 0.5527,
+      "step": 8344
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.3569705486297607,
+      "learning_rate": 9.708983350402887e-06,
+      "loss": 0.8558,
+      "step": 8345
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.8046224117279053,
+      "learning_rate": 9.708894706449815e-06,
+      "loss": 0.6188,
+      "step": 8346
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.7641067504882812,
+      "learning_rate": 9.708806049403069e-06,
+      "loss": 0.6365,
+      "step": 8347
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.302651882171631,
+      "learning_rate": 9.708717379262897e-06,
+      "loss": 0.7363,
+      "step": 8348
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.1205732822418213,
+      "learning_rate": 9.708628696029548e-06,
+      "loss": 0.5664,
+      "step": 8349
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.0981593132019043,
+      "learning_rate": 9.708539999703265e-06,
+      "loss": 0.5693,
+      "step": 8350
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 11.140937805175781,
+      "learning_rate": 9.708451290284298e-06,
+      "loss": 0.4954,
+      "step": 8351
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.495323419570923,
+      "learning_rate": 9.708362567772889e-06,
+      "loss": 0.5531,
+      "step": 8352
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 4.044354438781738,
+      "learning_rate": 9.70827383216929e-06,
+      "loss": 0.5369,
+      "step": 8353
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.114490270614624,
+      "learning_rate": 9.708185083473745e-06,
+      "loss": 0.5244,
+      "step": 8354
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.4630424976348877,
+      "learning_rate": 9.7080963216865e-06,
+      "loss": 0.7978,
+      "step": 8355
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.422542095184326,
+      "learning_rate": 9.708007546807803e-06,
+      "loss": 0.5942,
+      "step": 8356
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.9560203552246094,
+      "learning_rate": 9.7079187588379e-06,
+      "loss": 0.7266,
+      "step": 8357
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 4.005173683166504,
+      "learning_rate": 9.70782995777704e-06,
+      "loss": 0.6689,
+      "step": 8358
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.358992338180542,
+      "learning_rate": 9.707741143625467e-06,
+      "loss": 0.7943,
+      "step": 8359
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.735018491744995,
+      "learning_rate": 9.707652316383432e-06,
+      "loss": 0.7151,
+      "step": 8360
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.4044277667999268,
+      "learning_rate": 9.707563476051175e-06,
+      "loss": 0.8002,
+      "step": 8361
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.5900609493255615,
+      "learning_rate": 9.707474622628951e-06,
+      "loss": 0.5787,
+      "step": 8362
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.7084531784057617,
+      "learning_rate": 9.707385756117003e-06,
+      "loss": 0.6621,
+      "step": 8363
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.4634346961975098,
+      "learning_rate": 9.707296876515579e-06,
+      "loss": 0.532,
+      "step": 8364
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.4769299030303955,
+      "learning_rate": 9.707207983824924e-06,
+      "loss": 0.7197,
+      "step": 8365
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.2688910961151123,
+      "learning_rate": 9.707119078045288e-06,
+      "loss": 0.7502,
+      "step": 8366
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.224271535873413,
+      "learning_rate": 9.707030159176917e-06,
+      "loss": 0.6111,
+      "step": 8367
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.353963613510132,
+      "learning_rate": 9.70694122722006e-06,
+      "loss": 0.5907,
+      "step": 8368
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 5.650509357452393,
+      "learning_rate": 9.706852282174961e-06,
+      "loss": 0.7707,
+      "step": 8369
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.021406412124634,
+      "learning_rate": 9.70676332404187e-06,
+      "loss": 0.4916,
+      "step": 8370
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.109351873397827,
+      "learning_rate": 9.706674352821032e-06,
+      "loss": 0.7055,
+      "step": 8371
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.3032212257385254,
+      "learning_rate": 9.706585368512697e-06,
+      "loss": 0.9366,
+      "step": 8372
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.921657085418701,
+      "learning_rate": 9.706496371117111e-06,
+      "loss": 0.6693,
+      "step": 8373
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.5116872787475586,
+      "learning_rate": 9.706407360634521e-06,
+      "loss": 0.783,
+      "step": 8374
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.0091640949249268,
+      "learning_rate": 9.706318337065175e-06,
+      "loss": 0.6501,
+      "step": 8375
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 4.333288192749023,
+      "learning_rate": 9.706229300409321e-06,
+      "loss": 0.6799,
+      "step": 8376
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.8501498699188232,
+      "learning_rate": 9.706140250667208e-06,
+      "loss": 0.7261,
+      "step": 8377
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.9415712356567383,
+      "learning_rate": 9.706051187839081e-06,
+      "loss": 0.5685,
+      "step": 8378
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.4504878520965576,
+      "learning_rate": 9.705962111925188e-06,
+      "loss": 0.9544,
+      "step": 8379
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.3826584815979004,
+      "learning_rate": 9.705873022925778e-06,
+      "loss": 0.7287,
+      "step": 8380
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.8373987674713135,
+      "learning_rate": 9.705783920841096e-06,
+      "loss": 0.3718,
+      "step": 8381
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.764415979385376,
+      "learning_rate": 9.705694805671395e-06,
+      "loss": 0.5164,
+      "step": 8382
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.603548765182495,
+      "learning_rate": 9.705605677416916e-06,
+      "loss": 0.4291,
+      "step": 8383
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.9260319471359253,
+      "learning_rate": 9.705516536077912e-06,
+      "loss": 0.7987,
+      "step": 8384
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.278745412826538,
+      "learning_rate": 9.705427381654629e-06,
+      "loss": 0.619,
+      "step": 8385
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.0339877605438232,
+      "learning_rate": 9.705338214147316e-06,
+      "loss": 0.8379,
+      "step": 8386
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.0411369800567627,
+      "learning_rate": 9.70524903355622e-06,
+      "loss": 0.6181,
+      "step": 8387
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.708624839782715,
+      "learning_rate": 9.705159839881588e-06,
+      "loss": 0.5122,
+      "step": 8388
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 4.188806533813477,
+      "learning_rate": 9.70507063312367e-06,
+      "loss": 0.5508,
+      "step": 8389
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.749934196472168,
+      "learning_rate": 9.704981413282712e-06,
+      "loss": 0.6589,
+      "step": 8390
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.5263773202896118,
+      "learning_rate": 9.704892180358965e-06,
+      "loss": 0.5335,
+      "step": 8391
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.7892367839813232,
+      "learning_rate": 9.704802934352675e-06,
+      "loss": 0.6159,
+      "step": 8392
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.7387380599975586,
+      "learning_rate": 9.70471367526409e-06,
+      "loss": 0.7339,
+      "step": 8393
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.6943559646606445,
+      "learning_rate": 9.704624403093458e-06,
+      "loss": 0.7325,
+      "step": 8394
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.726130485534668,
+      "learning_rate": 9.704535117841029e-06,
+      "loss": 0.7791,
+      "step": 8395
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.790728211402893,
+      "learning_rate": 9.70444581950705e-06,
+      "loss": 0.6199,
+      "step": 8396
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.5848946571350098,
+      "learning_rate": 9.704356508091768e-06,
+      "loss": 0.4404,
+      "step": 8397
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.1624538898468018,
+      "learning_rate": 9.704267183595434e-06,
+      "loss": 0.4684,
+      "step": 8398
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.2001638412475586,
+      "learning_rate": 9.704177846018296e-06,
+      "loss": 0.6962,
+      "step": 8399
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.143789291381836,
+      "learning_rate": 9.7040884953606e-06,
+      "loss": 0.8436,
+      "step": 8400
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.640241861343384,
+      "learning_rate": 9.703999131622597e-06,
+      "loss": 0.6122,
+      "step": 8401
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.9122742414474487,
+      "learning_rate": 9.703909754804534e-06,
+      "loss": 0.8525,
+      "step": 8402
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.7152652740478516,
+      "learning_rate": 9.703820364906662e-06,
+      "loss": 0.5408,
+      "step": 8403
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.8109264373779297,
+      "learning_rate": 9.703730961929226e-06,
+      "loss": 0.7173,
+      "step": 8404
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.579677939414978,
+      "learning_rate": 9.703641545872477e-06,
+      "loss": 0.3605,
+      "step": 8405
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.4385852813720703,
+      "learning_rate": 9.703552116736661e-06,
+      "loss": 0.6705,
+      "step": 8406
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.1266703605651855,
+      "learning_rate": 9.70346267452203e-06,
+      "loss": 0.8113,
+      "step": 8407
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.671126127243042,
+      "learning_rate": 9.703373219228834e-06,
+      "loss": 0.6305,
+      "step": 8408
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.490079402923584,
+      "learning_rate": 9.703283750857316e-06,
+      "loss": 0.8014,
+      "step": 8409
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.96742582321167,
+      "learning_rate": 9.703194269407727e-06,
+      "loss": 0.5248,
+      "step": 8410
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.734543800354004,
+      "learning_rate": 9.703104774880317e-06,
+      "loss": 0.3832,
+      "step": 8411
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.2813985347747803,
+      "learning_rate": 9.703015267275335e-06,
+      "loss": 0.5203,
+      "step": 8412
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.6486895084381104,
+      "learning_rate": 9.70292574659303e-06,
+      "loss": 0.631,
+      "step": 8413
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.6489129066467285,
+      "learning_rate": 9.70283621283365e-06,
+      "loss": 0.3807,
+      "step": 8414
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.16610050201416,
+      "learning_rate": 9.702746665997443e-06,
+      "loss": 0.5579,
+      "step": 8415
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 4.545204162597656,
+      "learning_rate": 9.702657106084661e-06,
+      "loss": 0.8247,
+      "step": 8416
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.7873576879501343,
+      "learning_rate": 9.702567533095551e-06,
+      "loss": 0.7746,
+      "step": 8417
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 4.659178256988525,
+      "learning_rate": 9.702477947030361e-06,
+      "loss": 0.7482,
+      "step": 8418
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 5.495406150817871,
+      "learning_rate": 9.702388347889343e-06,
+      "loss": 0.6856,
+      "step": 8419
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.6311047077178955,
+      "learning_rate": 9.702298735672741e-06,
+      "loss": 0.7669,
+      "step": 8420
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.4477955102920532,
+      "learning_rate": 9.702209110380812e-06,
+      "loss": 0.6137,
+      "step": 8421
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.1728084087371826,
+      "learning_rate": 9.7021194720138e-06,
+      "loss": 0.6132,
+      "step": 8422
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.2271783351898193,
+      "learning_rate": 9.702029820571954e-06,
+      "loss": 0.6168,
+      "step": 8423
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.492985725402832,
+      "learning_rate": 9.701940156055524e-06,
+      "loss": 0.7045,
+      "step": 8424
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.6276800632476807,
+      "learning_rate": 9.701850478464762e-06,
+      "loss": 0.8801,
+      "step": 8425
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.8525733947753906,
+      "learning_rate": 9.701760787799914e-06,
+      "loss": 0.4953,
+      "step": 8426
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.007744789123535,
+      "learning_rate": 9.70167108406123e-06,
+      "loss": 0.5746,
+      "step": 8427
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.4600279331207275,
+      "learning_rate": 9.70158136724896e-06,
+      "loss": 0.4516,
+      "step": 8428
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.287597417831421,
+      "learning_rate": 9.701491637363354e-06,
+      "loss": 0.6382,
+      "step": 8429
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.2815957069396973,
+      "learning_rate": 9.701401894404662e-06,
+      "loss": 0.5556,
+      "step": 8430
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.3523502349853516,
+      "learning_rate": 9.701312138373131e-06,
+      "loss": 0.5032,
+      "step": 8431
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.3357045650482178,
+      "learning_rate": 9.701222369269012e-06,
+      "loss": 0.6918,
+      "step": 8432
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.375953197479248,
+      "learning_rate": 9.701132587092556e-06,
+      "loss": 0.8749,
+      "step": 8433
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 14.914790153503418,
+      "learning_rate": 9.701042791844011e-06,
+      "loss": 0.7041,
+      "step": 8434
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.490905284881592,
+      "learning_rate": 9.700952983523626e-06,
+      "loss": 0.7206,
+      "step": 8435
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 5.6625165939331055,
+      "learning_rate": 9.700863162131653e-06,
+      "loss": 0.7588,
+      "step": 8436
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.753077983856201,
+      "learning_rate": 9.700773327668341e-06,
+      "loss": 0.6986,
+      "step": 8437
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.6008225679397583,
+      "learning_rate": 9.700683480133938e-06,
+      "loss": 0.7098,
+      "step": 8438
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.6244418621063232,
+      "learning_rate": 9.700593619528696e-06,
+      "loss": 0.7979,
+      "step": 8439
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.777540445327759,
+      "learning_rate": 9.700503745852862e-06,
+      "loss": 0.6469,
+      "step": 8440
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.363130569458008,
+      "learning_rate": 9.70041385910669e-06,
+      "loss": 0.8213,
+      "step": 8441
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.712583541870117,
+      "learning_rate": 9.700323959290428e-06,
+      "loss": 0.6417,
+      "step": 8442
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.661867618560791,
+      "learning_rate": 9.700234046404324e-06,
+      "loss": 0.6895,
+      "step": 8443
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.9424842596054077,
+      "learning_rate": 9.700144120448631e-06,
+      "loss": 0.7314,
+      "step": 8444
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.0337741374969482,
+      "learning_rate": 9.7000541814236e-06,
+      "loss": 0.8658,
+      "step": 8445
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.6610010862350464,
+      "learning_rate": 9.699964229329476e-06,
+      "loss": 0.5681,
+      "step": 8446
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.205939531326294,
+      "learning_rate": 9.699874264166514e-06,
+      "loss": 0.7273,
+      "step": 8447
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.6461377143859863,
+      "learning_rate": 9.699784285934961e-06,
+      "loss": 0.765,
+      "step": 8448
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.252291440963745,
+      "learning_rate": 9.699694294635068e-06,
+      "loss": 0.7835,
+      "step": 8449
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.312413454055786,
+      "learning_rate": 9.699604290267087e-06,
+      "loss": 0.6926,
+      "step": 8450
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.580163836479187,
+      "learning_rate": 9.699514272831266e-06,
+      "loss": 0.7804,
+      "step": 8451
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.2371411323547363,
+      "learning_rate": 9.699424242327857e-06,
+      "loss": 0.6208,
+      "step": 8452
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.3570990562438965,
+      "learning_rate": 9.69933419875711e-06,
+      "loss": 0.6136,
+      "step": 8453
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.057544469833374,
+      "learning_rate": 9.699244142119275e-06,
+      "loss": 0.6012,
+      "step": 8454
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.537238597869873,
+      "learning_rate": 9.699154072414601e-06,
+      "loss": 0.7634,
+      "step": 8455
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 4.1347784996032715,
+      "learning_rate": 9.69906398964334e-06,
+      "loss": 0.9529,
+      "step": 8456
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.4276721477508545,
+      "learning_rate": 9.698973893805745e-06,
+      "loss": 0.56,
+      "step": 8457
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.64199161529541,
+      "learning_rate": 9.69888378490206e-06,
+      "loss": 0.6011,
+      "step": 8458
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.4632487297058105,
+      "learning_rate": 9.698793662932542e-06,
+      "loss": 0.5561,
+      "step": 8459
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.4693026542663574,
+      "learning_rate": 9.698703527897439e-06,
+      "loss": 0.7056,
+      "step": 8460
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.4978270530700684,
+      "learning_rate": 9.698613379797001e-06,
+      "loss": 0.7047,
+      "step": 8461
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.886923313140869,
+      "learning_rate": 9.698523218631479e-06,
+      "loss": 0.6381,
+      "step": 8462
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.9188034534454346,
+      "learning_rate": 9.698433044401125e-06,
+      "loss": 0.6106,
+      "step": 8463
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.6833927631378174,
+      "learning_rate": 9.698342857106188e-06,
+      "loss": 0.8329,
+      "step": 8464
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 4.0356364250183105,
+      "learning_rate": 9.69825265674692e-06,
+      "loss": 0.72,
+      "step": 8465
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.6328787803649902,
+      "learning_rate": 9.69816244332357e-06,
+      "loss": 0.7167,
+      "step": 8466
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.31754994392395,
+      "learning_rate": 9.698072216836394e-06,
+      "loss": 0.5112,
+      "step": 8467
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.6082332134246826,
+      "learning_rate": 9.697981977285636e-06,
+      "loss": 0.5976,
+      "step": 8468
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 7.156685829162598,
+      "learning_rate": 9.697891724671551e-06,
+      "loss": 0.7634,
+      "step": 8469
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.9430811405181885,
+      "learning_rate": 9.697801458994388e-06,
+      "loss": 0.989,
+      "step": 8470
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.4777050018310547,
+      "learning_rate": 9.697711180254399e-06,
+      "loss": 0.7034,
+      "step": 8471
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.039252996444702,
+      "learning_rate": 9.697620888451835e-06,
+      "loss": 0.7891,
+      "step": 8472
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.0594160556793213,
+      "learning_rate": 9.697530583586948e-06,
+      "loss": 0.6377,
+      "step": 8473
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.9441537857055664,
+      "learning_rate": 9.697440265659987e-06,
+      "loss": 0.6154,
+      "step": 8474
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.6355133056640625,
+      "learning_rate": 9.697349934671206e-06,
+      "loss": 0.7404,
+      "step": 8475
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 6.622068881988525,
+      "learning_rate": 9.697259590620854e-06,
+      "loss": 0.7393,
+      "step": 8476
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.380448341369629,
+      "learning_rate": 9.697169233509183e-06,
+      "loss": 0.5442,
+      "step": 8477
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 4.412491798400879,
+      "learning_rate": 9.697078863336441e-06,
+      "loss": 1.0284,
+      "step": 8478
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 4.503813743591309,
+      "learning_rate": 9.696988480102885e-06,
+      "loss": 0.5003,
+      "step": 8479
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.468491792678833,
+      "learning_rate": 9.696898083808763e-06,
+      "loss": 0.7934,
+      "step": 8480
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.8513259887695312,
+      "learning_rate": 9.696807674454329e-06,
+      "loss": 0.7753,
+      "step": 8481
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.6089463233947754,
+      "learning_rate": 9.696717252039829e-06,
+      "loss": 0.8859,
+      "step": 8482
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.592805862426758,
+      "learning_rate": 9.69662681656552e-06,
+      "loss": 0.7166,
+      "step": 8483
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.249783515930176,
+      "learning_rate": 9.696536368031651e-06,
+      "loss": 0.4521,
+      "step": 8484
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.8691282272338867,
+      "learning_rate": 9.696445906438472e-06,
+      "loss": 0.7328,
+      "step": 8485
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.494682788848877,
+      "learning_rate": 9.696355431786237e-06,
+      "loss": 0.5947,
+      "step": 8486
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.7251126766204834,
+      "learning_rate": 9.696264944075198e-06,
+      "loss": 0.8681,
+      "step": 8487
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.5100557804107666,
+      "learning_rate": 9.696174443305604e-06,
+      "loss": 0.4365,
+      "step": 8488
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.6036176681518555,
+      "learning_rate": 9.696083929477709e-06,
+      "loss": 0.6726,
+      "step": 8489
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.923187494277954,
+      "learning_rate": 9.695993402591764e-06,
+      "loss": 0.5218,
+      "step": 8490
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.5884671211242676,
+      "learning_rate": 9.695902862648019e-06,
+      "loss": 0.6738,
+      "step": 8491
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.3374505043029785,
+      "learning_rate": 9.695812309646728e-06,
+      "loss": 0.6989,
+      "step": 8492
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.132560968399048,
+      "learning_rate": 9.695721743588142e-06,
+      "loss": 0.7795,
+      "step": 8493
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.4108798503875732,
+      "learning_rate": 9.695631164472515e-06,
+      "loss": 0.5276,
+      "step": 8494
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.4171652793884277,
+      "learning_rate": 9.695540572300094e-06,
+      "loss": 0.5752,
+      "step": 8495
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.302659511566162,
+      "learning_rate": 9.695449967071134e-06,
+      "loss": 0.7788,
+      "step": 8496
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.4117727279663086,
+      "learning_rate": 9.695359348785887e-06,
+      "loss": 0.4275,
+      "step": 8497
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.4348405599594116,
+      "learning_rate": 9.695268717444605e-06,
+      "loss": 0.4268,
+      "step": 8498
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.9914956092834473,
+      "learning_rate": 9.69517807304754e-06,
+      "loss": 0.8032,
+      "step": 8499
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.278101921081543,
+      "learning_rate": 9.695087415594944e-06,
+      "loss": 0.524,
+      "step": 8500
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.5235567092895508,
+      "learning_rate": 9.694996745087066e-06,
+      "loss": 0.5038,
+      "step": 8501
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.883578062057495,
+      "learning_rate": 9.694906061524163e-06,
+      "loss": 0.4057,
+      "step": 8502
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.0694336891174316,
+      "learning_rate": 9.694815364906485e-06,
+      "loss": 0.8036,
+      "step": 8503
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.3006649017333984,
+      "learning_rate": 9.694724655234284e-06,
+      "loss": 0.8725,
+      "step": 8504
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 4.37893533706665,
+      "learning_rate": 9.694633932507812e-06,
+      "loss": 0.6197,
+      "step": 8505
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.3127639293670654,
+      "learning_rate": 9.694543196727321e-06,
+      "loss": 0.6834,
+      "step": 8506
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.2375175952911377,
+      "learning_rate": 9.694452447893066e-06,
+      "loss": 0.7185,
+      "step": 8507
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.7831709384918213,
+      "learning_rate": 9.694361686005296e-06,
+      "loss": 0.6241,
+      "step": 8508
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.2619271278381348,
+      "learning_rate": 9.694270911064265e-06,
+      "loss": 0.5372,
+      "step": 8509
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.829718589782715,
+      "learning_rate": 9.694180123070224e-06,
+      "loss": 0.7555,
+      "step": 8510
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.145846366882324,
+      "learning_rate": 9.694089322023428e-06,
+      "loss": 0.9458,
+      "step": 8511
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.9138565063476562,
+      "learning_rate": 9.693998507924127e-06,
+      "loss": 0.5993,
+      "step": 8512
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.2927048206329346,
+      "learning_rate": 9.693907680772577e-06,
+      "loss": 0.6332,
+      "step": 8513
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.7807469367980957,
+      "learning_rate": 9.693816840569024e-06,
+      "loss": 0.6903,
+      "step": 8514
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 4.061173439025879,
+      "learning_rate": 9.693725987313728e-06,
+      "loss": 0.7063,
+      "step": 8515
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.2061657905578613,
+      "learning_rate": 9.693635121006936e-06,
+      "loss": 0.7508,
+      "step": 8516
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.598576784133911,
+      "learning_rate": 9.693544241648906e-06,
+      "loss": 0.8397,
+      "step": 8517
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.1808102130889893,
+      "learning_rate": 9.693453349239884e-06,
+      "loss": 0.5981,
+      "step": 8518
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.0256807804107666,
+      "learning_rate": 9.69336244378013e-06,
+      "loss": 0.6547,
+      "step": 8519
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.371521472930908,
+      "learning_rate": 9.693271525269891e-06,
+      "loss": 0.7417,
+      "step": 8520
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 4.519982814788818,
+      "learning_rate": 9.693180593709424e-06,
+      "loss": 0.7171,
+      "step": 8521
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.3767354488372803,
+      "learning_rate": 9.693089649098978e-06,
+      "loss": 0.569,
+      "step": 8522
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.707240104675293,
+      "learning_rate": 9.692998691438809e-06,
+      "loss": 0.9134,
+      "step": 8523
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.0287249088287354,
+      "learning_rate": 9.692907720729168e-06,
+      "loss": 0.695,
+      "step": 8524
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.5800410509109497,
+      "learning_rate": 9.692816736970308e-06,
+      "loss": 0.3913,
+      "step": 8525
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.2157013416290283,
+      "learning_rate": 9.692725740162483e-06,
+      "loss": 0.7925,
+      "step": 8526
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.8421449661254883,
+      "learning_rate": 9.692634730305945e-06,
+      "loss": 0.5922,
+      "step": 8527
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.7649872303009033,
+      "learning_rate": 9.69254370740095e-06,
+      "loss": 0.6895,
+      "step": 8528
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 4.650245189666748,
+      "learning_rate": 9.692452671447747e-06,
+      "loss": 0.57,
+      "step": 8529
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.3605175018310547,
+      "learning_rate": 9.692361622446592e-06,
+      "loss": 0.7962,
+      "step": 8530
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.5794222354888916,
+      "learning_rate": 9.692270560397736e-06,
+      "loss": 0.7087,
+      "step": 8531
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.362921118736267,
+      "learning_rate": 9.692179485301435e-06,
+      "loss": 0.856,
+      "step": 8532
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.5189632177352905,
+      "learning_rate": 9.692088397157939e-06,
+      "loss": 0.7617,
+      "step": 8533
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.5551207065582275,
+      "learning_rate": 9.691997295967503e-06,
+      "loss": 0.7344,
+      "step": 8534
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 5.021712779998779,
+      "learning_rate": 9.69190618173038e-06,
+      "loss": 0.6456,
+      "step": 8535
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.2046358585357666,
+      "learning_rate": 9.691815054446824e-06,
+      "loss": 0.7948,
+      "step": 8536
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.277188777923584,
+      "learning_rate": 9.691723914117088e-06,
+      "loss": 0.8414,
+      "step": 8537
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.0639750957489014,
+      "learning_rate": 9.691632760741425e-06,
+      "loss": 0.5759,
+      "step": 8538
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.894906520843506,
+      "learning_rate": 9.691541594320089e-06,
+      "loss": 0.7725,
+      "step": 8539
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.0725934505462646,
+      "learning_rate": 9.691450414853334e-06,
+      "loss": 0.8445,
+      "step": 8540
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.908392906188965,
+      "learning_rate": 9.691359222341412e-06,
+      "loss": 0.9654,
+      "step": 8541
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.7009801864624023,
+      "learning_rate": 9.691268016784579e-06,
+      "loss": 0.65,
+      "step": 8542
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.7561163902282715,
+      "learning_rate": 9.691176798183084e-06,
+      "loss": 0.8722,
+      "step": 8543
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.8172004222869873,
+      "learning_rate": 9.691085566537186e-06,
+      "loss": 0.4813,
+      "step": 8544
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.449514627456665,
+      "learning_rate": 9.690994321847136e-06,
+      "loss": 0.6882,
+      "step": 8545
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.608024835586548,
+      "learning_rate": 9.690903064113188e-06,
+      "loss": 0.7426,
+      "step": 8546
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.594578504562378,
+      "learning_rate": 9.690811793335596e-06,
+      "loss": 0.4683,
+      "step": 8547
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.9098021984100342,
+      "learning_rate": 9.690720509514612e-06,
+      "loss": 0.5667,
+      "step": 8548
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.2226274013519287,
+      "learning_rate": 9.690629212650494e-06,
+      "loss": 0.6845,
+      "step": 8549
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.3224472999572754,
+      "learning_rate": 9.690537902743493e-06,
+      "loss": 0.5093,
+      "step": 8550
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.1206045150756836,
+      "learning_rate": 9.690446579793861e-06,
+      "loss": 0.6261,
+      "step": 8551
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.435007333755493,
+      "learning_rate": 9.690355243801856e-06,
+      "loss": 0.6335,
+      "step": 8552
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.239217519760132,
+      "learning_rate": 9.69026389476773e-06,
+      "loss": 0.7971,
+      "step": 8553
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.0969173908233643,
+      "learning_rate": 9.690172532691735e-06,
+      "loss": 0.706,
+      "step": 8554
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.5044755935668945,
+      "learning_rate": 9.690081157574129e-06,
+      "loss": 0.5427,
+      "step": 8555
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.222560405731201,
+      "learning_rate": 9.689989769415165e-06,
+      "loss": 0.8382,
+      "step": 8556
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.8606265783309937,
+      "learning_rate": 9.689898368215095e-06,
+      "loss": 0.5239,
+      "step": 8557
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.4502499103546143,
+      "learning_rate": 9.689806953974175e-06,
+      "loss": 0.5327,
+      "step": 8558
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.161897659301758,
+      "learning_rate": 9.68971552669266e-06,
+      "loss": 0.5277,
+      "step": 8559
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.4778330326080322,
+      "learning_rate": 9.6896240863708e-06,
+      "loss": 0.8448,
+      "step": 8560
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.1238691806793213,
+      "learning_rate": 9.689532633008853e-06,
+      "loss": 0.3981,
+      "step": 8561
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 4.952577590942383,
+      "learning_rate": 9.689441166607074e-06,
+      "loss": 0.8103,
+      "step": 8562
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.833482027053833,
+      "learning_rate": 9.689349687165715e-06,
+      "loss": 0.6967,
+      "step": 8563
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.6841444969177246,
+      "learning_rate": 9.689258194685031e-06,
+      "loss": 0.5979,
+      "step": 8564
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.971449375152588,
+      "learning_rate": 9.689166689165277e-06,
+      "loss": 0.4816,
+      "step": 8565
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.789078712463379,
+      "learning_rate": 9.689075170606708e-06,
+      "loss": 0.6571,
+      "step": 8566
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.2446227073669434,
+      "learning_rate": 9.688983639009575e-06,
+      "loss": 0.8162,
+      "step": 8567
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 4.367471218109131,
+      "learning_rate": 9.688892094374139e-06,
+      "loss": 0.4752,
+      "step": 8568
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.6817121505737305,
+      "learning_rate": 9.688800536700646e-06,
+      "loss": 0.5833,
+      "step": 8569
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 4.693740367889404,
+      "learning_rate": 9.688708965989358e-06,
+      "loss": 0.9253,
+      "step": 8570
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.0773627758026123,
+      "learning_rate": 9.688617382240526e-06,
+      "loss": 0.9454,
+      "step": 8571
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.3134138584136963,
+      "learning_rate": 9.688525785454406e-06,
+      "loss": 0.7744,
+      "step": 8572
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.728407382965088,
+      "learning_rate": 9.68843417563125e-06,
+      "loss": 0.7882,
+      "step": 8573
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 5.130764961242676,
+      "learning_rate": 9.688342552771318e-06,
+      "loss": 0.5485,
+      "step": 8574
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.325728178024292,
+      "learning_rate": 9.688250916874859e-06,
+      "loss": 0.5426,
+      "step": 8575
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.620842695236206,
+      "learning_rate": 9.688159267942131e-06,
+      "loss": 0.3528,
+      "step": 8576
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.4984564781188965,
+      "learning_rate": 9.68806760597339e-06,
+      "loss": 0.7129,
+      "step": 8577
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 9.943269729614258,
+      "learning_rate": 9.687975930968889e-06,
+      "loss": 0.6436,
+      "step": 8578
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.7159507274627686,
+      "learning_rate": 9.687884242928882e-06,
+      "loss": 0.6727,
+      "step": 8579
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.5547733306884766,
+      "learning_rate": 9.687792541853624e-06,
+      "loss": 0.7059,
+      "step": 8580
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.4181735515594482,
+      "learning_rate": 9.687700827743373e-06,
+      "loss": 0.831,
+      "step": 8581
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.901554822921753,
+      "learning_rate": 9.687609100598381e-06,
+      "loss": 0.5115,
+      "step": 8582
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.5406854152679443,
+      "learning_rate": 9.687517360418906e-06,
+      "loss": 0.7125,
+      "step": 8583
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.765300750732422,
+      "learning_rate": 9.687425607205199e-06,
+      "loss": 0.871,
+      "step": 8584
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.499636173248291,
+      "learning_rate": 9.687333840957519e-06,
+      "loss": 0.5748,
+      "step": 8585
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.4508039951324463,
+      "learning_rate": 9.687242061676117e-06,
+      "loss": 0.6128,
+      "step": 8586
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.5894341468811035,
+      "learning_rate": 9.687150269361252e-06,
+      "loss": 0.9094,
+      "step": 8587
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.3259055614471436,
+      "learning_rate": 9.687058464013177e-06,
+      "loss": 0.7158,
+      "step": 8588
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.637380838394165,
+      "learning_rate": 9.68696664563215e-06,
+      "loss": 0.4923,
+      "step": 8589
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.8957641124725342,
+      "learning_rate": 9.686874814218424e-06,
+      "loss": 0.7348,
+      "step": 8590
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.903425455093384,
+      "learning_rate": 9.686782969772254e-06,
+      "loss": 0.6049,
+      "step": 8591
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.7760183811187744,
+      "learning_rate": 9.686691112293897e-06,
+      "loss": 0.7592,
+      "step": 8592
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.583481550216675,
+      "learning_rate": 9.686599241783609e-06,
+      "loss": 0.8023,
+      "step": 8593
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 4.553149700164795,
+      "learning_rate": 9.686507358241642e-06,
+      "loss": 0.6256,
+      "step": 8594
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.0319178104400635,
+      "learning_rate": 9.686415461668254e-06,
+      "loss": 0.526,
+      "step": 8595
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.2887487411499023,
+      "learning_rate": 9.686323552063702e-06,
+      "loss": 0.5753,
+      "step": 8596
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.3976826667785645,
+      "learning_rate": 9.686231629428237e-06,
+      "loss": 0.5222,
+      "step": 8597
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 4.726877212524414,
+      "learning_rate": 9.68613969376212e-06,
+      "loss": 0.9209,
+      "step": 8598
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.9811477661132812,
+      "learning_rate": 9.686047745065604e-06,
+      "loss": 0.6148,
+      "step": 8599
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.2353596687316895,
+      "learning_rate": 9.685955783338943e-06,
+      "loss": 0.4887,
+      "step": 8600
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.623547077178955,
+      "learning_rate": 9.685863808582393e-06,
+      "loss": 0.6145,
+      "step": 8601
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.365023612976074,
+      "learning_rate": 9.685771820796214e-06,
+      "loss": 0.7081,
+      "step": 8602
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.568037509918213,
+      "learning_rate": 9.685679819980658e-06,
+      "loss": 0.6285,
+      "step": 8603
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.8305751085281372,
+      "learning_rate": 9.685587806135981e-06,
+      "loss": 0.5714,
+      "step": 8604
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.4701895713806152,
+      "learning_rate": 9.685495779262441e-06,
+      "loss": 0.5318,
+      "step": 8605
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.6648602485656738,
+      "learning_rate": 9.685403739360291e-06,
+      "loss": 0.7276,
+      "step": 8606
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.495849132537842,
+      "learning_rate": 9.685311686429788e-06,
+      "loss": 0.8366,
+      "step": 8607
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.6013097763061523,
+      "learning_rate": 9.68521962047119e-06,
+      "loss": 0.8647,
+      "step": 8608
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.9217917919158936,
+      "learning_rate": 9.685127541484752e-06,
+      "loss": 0.5867,
+      "step": 8609
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.5817956924438477,
+      "learning_rate": 9.685035449470727e-06,
+      "loss": 0.8016,
+      "step": 8610
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.214994192123413,
+      "learning_rate": 9.684943344429374e-06,
+      "loss": 0.6236,
+      "step": 8611
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.884723663330078,
+      "learning_rate": 9.68485122636095e-06,
+      "loss": 0.5125,
+      "step": 8612
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.534792900085449,
+      "learning_rate": 9.684759095265708e-06,
+      "loss": 0.6665,
+      "step": 8613
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.620908498764038,
+      "learning_rate": 9.684666951143906e-06,
+      "loss": 0.802,
+      "step": 8614
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.798059105873108,
+      "learning_rate": 9.684574793995802e-06,
+      "loss": 0.6946,
+      "step": 8615
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 4.034893989562988,
+      "learning_rate": 9.684482623821649e-06,
+      "loss": 0.8806,
+      "step": 8616
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.9328888654708862,
+      "learning_rate": 9.684390440621705e-06,
+      "loss": 0.7333,
+      "step": 8617
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.406341314315796,
+      "learning_rate": 9.684298244396227e-06,
+      "loss": 0.8492,
+      "step": 8618
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 4.696689128875732,
+      "learning_rate": 9.684206035145468e-06,
+      "loss": 0.8234,
+      "step": 8619
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.2266745567321777,
+      "learning_rate": 9.68411381286969e-06,
+      "loss": 0.8323,
+      "step": 8620
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.8934803009033203,
+      "learning_rate": 9.684021577569142e-06,
+      "loss": 0.6587,
+      "step": 8621
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 4.890803337097168,
+      "learning_rate": 9.683929329244087e-06,
+      "loss": 0.5846,
+      "step": 8622
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.7719197273254395,
+      "learning_rate": 9.68383706789478e-06,
+      "loss": 0.6055,
+      "step": 8623
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 9.24194049835205,
+      "learning_rate": 9.683744793521475e-06,
+      "loss": 0.7736,
+      "step": 8624
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.2205467224121094,
+      "learning_rate": 9.683652506124431e-06,
+      "loss": 0.581,
+      "step": 8625
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.981341004371643,
+      "learning_rate": 9.683560205703903e-06,
+      "loss": 0.7091,
+      "step": 8626
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.6466326713562012,
+      "learning_rate": 9.68346789226015e-06,
+      "loss": 0.4731,
+      "step": 8627
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.36142635345459,
+      "learning_rate": 9.683375565793427e-06,
+      "loss": 0.5392,
+      "step": 8628
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.9655691385269165,
+      "learning_rate": 9.683283226303993e-06,
+      "loss": 0.7853,
+      "step": 8629
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.1024038791656494,
+      "learning_rate": 9.683190873792099e-06,
+      "loss": 0.7711,
+      "step": 8630
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 4.922736167907715,
+      "learning_rate": 9.683098508258007e-06,
+      "loss": 0.738,
+      "step": 8631
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.8274636268615723,
+      "learning_rate": 9.683006129701973e-06,
+      "loss": 0.8007,
+      "step": 8632
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.237064838409424,
+      "learning_rate": 9.682913738124252e-06,
+      "loss": 0.5956,
+      "step": 8633
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.4712889194488525,
+      "learning_rate": 9.682821333525104e-06,
+      "loss": 0.5949,
+      "step": 8634
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.8050334453582764,
+      "learning_rate": 9.682728915904782e-06,
+      "loss": 0.7721,
+      "step": 8635
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.129350423812866,
+      "learning_rate": 9.682636485263545e-06,
+      "loss": 0.6684,
+      "step": 8636
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.7777423858642578,
+      "learning_rate": 9.682544041601652e-06,
+      "loss": 0.5158,
+      "step": 8637
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.7466925382614136,
+      "learning_rate": 9.682451584919357e-06,
+      "loss": 0.7788,
+      "step": 8638
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.109112501144409,
+      "learning_rate": 9.68235911521692e-06,
+      "loss": 0.686,
+      "step": 8639
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.2876830101013184,
+      "learning_rate": 9.682266632494595e-06,
+      "loss": 0.6429,
+      "step": 8640
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.7307682037353516,
+      "learning_rate": 9.68217413675264e-06,
+      "loss": 0.6477,
+      "step": 8641
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.106314182281494,
+      "learning_rate": 9.682081627991313e-06,
+      "loss": 0.6003,
+      "step": 8642
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.5891878604888916,
+      "learning_rate": 9.68198910621087e-06,
+      "loss": 0.4045,
+      "step": 8643
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 5.12224817276001,
+      "learning_rate": 9.68189657141157e-06,
+      "loss": 0.6464,
+      "step": 8644
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.8170878887176514,
+      "learning_rate": 9.681804023593671e-06,
+      "loss": 0.5049,
+      "step": 8645
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.037346363067627,
+      "learning_rate": 9.681711462757427e-06,
+      "loss": 0.6774,
+      "step": 8646
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.041187047958374,
+      "learning_rate": 9.681618888903097e-06,
+      "loss": 0.5909,
+      "step": 8647
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.793266534805298,
+      "learning_rate": 9.68152630203094e-06,
+      "loss": 0.6847,
+      "step": 8648
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 4.075845718383789,
+      "learning_rate": 9.681433702141211e-06,
+      "loss": 0.8244,
+      "step": 8649
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 6.1519036293029785,
+      "learning_rate": 9.68134108923417e-06,
+      "loss": 0.7799,
+      "step": 8650
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.556100368499756,
+      "learning_rate": 9.681248463310072e-06,
+      "loss": 0.585,
+      "step": 8651
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 4.3750433921813965,
+      "learning_rate": 9.681155824369176e-06,
+      "loss": 0.9257,
+      "step": 8652
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.7855687141418457,
+      "learning_rate": 9.68106317241174e-06,
+      "loss": 0.7713,
+      "step": 8653
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.9076309204101562,
+      "learning_rate": 9.680970507438019e-06,
+      "loss": 0.6546,
+      "step": 8654
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.3673980236053467,
+      "learning_rate": 9.680877829448273e-06,
+      "loss": 0.7737,
+      "step": 8655
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.5295188426971436,
+      "learning_rate": 9.680785138442761e-06,
+      "loss": 0.6385,
+      "step": 8656
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.179643392562866,
+      "learning_rate": 9.680692434421738e-06,
+      "loss": 0.7699,
+      "step": 8657
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.5533214807510376,
+      "learning_rate": 9.680599717385462e-06,
+      "loss": 0.7176,
+      "step": 8658
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.2168726921081543,
+      "learning_rate": 9.680506987334193e-06,
+      "loss": 0.6717,
+      "step": 8659
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.1277477741241455,
+      "learning_rate": 9.680414244268186e-06,
+      "loss": 0.4448,
+      "step": 8660
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.0630555152893066,
+      "learning_rate": 9.6803214881877e-06,
+      "loss": 0.6359,
+      "step": 8661
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.5844134092330933,
+      "learning_rate": 9.680228719092994e-06,
+      "loss": 0.5549,
+      "step": 8662
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.915596604347229,
+      "learning_rate": 9.680135936984326e-06,
+      "loss": 0.5725,
+      "step": 8663
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.71886944770813,
+      "learning_rate": 9.680043141861952e-06,
+      "loss": 0.8263,
+      "step": 8664
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.1928653717041016,
+      "learning_rate": 9.679950333726133e-06,
+      "loss": 0.5499,
+      "step": 8665
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 5.913621425628662,
+      "learning_rate": 9.679857512577124e-06,
+      "loss": 0.7937,
+      "step": 8666
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.7683799266815186,
+      "learning_rate": 9.679764678415184e-06,
+      "loss": 0.6308,
+      "step": 8667
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.3951416015625,
+      "learning_rate": 9.67967183124057e-06,
+      "loss": 0.8564,
+      "step": 8668
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.0699820518493652,
+      "learning_rate": 9.679578971053544e-06,
+      "loss": 0.5556,
+      "step": 8669
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.3092920780181885,
+      "learning_rate": 9.679486097854363e-06,
+      "loss": 0.4532,
+      "step": 8670
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.1714742183685303,
+      "learning_rate": 9.679393211643283e-06,
+      "loss": 0.8037,
+      "step": 8671
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.434023141860962,
+      "learning_rate": 9.679300312420563e-06,
+      "loss": 0.5715,
+      "step": 8672
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.528137683868408,
+      "learning_rate": 9.679207400186462e-06,
+      "loss": 0.6944,
+      "step": 8673
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.994072198867798,
+      "learning_rate": 9.679114474941237e-06,
+      "loss": 0.7061,
+      "step": 8674
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.3192298412323,
+      "learning_rate": 9.67902153668515e-06,
+      "loss": 0.6952,
+      "step": 8675
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.057745933532715,
+      "learning_rate": 9.678928585418454e-06,
+      "loss": 0.6353,
+      "step": 8676
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.300002336502075,
+      "learning_rate": 9.678835621141412e-06,
+      "loss": 0.4012,
+      "step": 8677
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.6570627689361572,
+      "learning_rate": 9.67874264385428e-06,
+      "loss": 0.5531,
+      "step": 8678
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.6458740234375,
+      "learning_rate": 9.678649653557317e-06,
+      "loss": 1.1259,
+      "step": 8679
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.9622087478637695,
+      "learning_rate": 9.678556650250784e-06,
+      "loss": 0.7965,
+      "step": 8680
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.9233338832855225,
+      "learning_rate": 9.678463633934936e-06,
+      "loss": 0.6133,
+      "step": 8681
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.8438867330551147,
+      "learning_rate": 9.678370604610034e-06,
+      "loss": 0.5094,
+      "step": 8682
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.1088695526123047,
+      "learning_rate": 9.678277562276334e-06,
+      "loss": 0.8009,
+      "step": 8683
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.3920812606811523,
+      "learning_rate": 9.678184506934099e-06,
+      "loss": 0.5835,
+      "step": 8684
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.4839417934417725,
+      "learning_rate": 9.678091438583586e-06,
+      "loss": 0.7768,
+      "step": 8685
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 6.555802345275879,
+      "learning_rate": 9.677998357225052e-06,
+      "loss": 0.5606,
+      "step": 8686
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 4.0452704429626465,
+      "learning_rate": 9.677905262858755e-06,
+      "loss": 0.8085,
+      "step": 8687
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 4.272977828979492,
+      "learning_rate": 9.677812155484958e-06,
+      "loss": 0.731,
+      "step": 8688
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.5181257724761963,
+      "learning_rate": 9.677719035103918e-06,
+      "loss": 0.5944,
+      "step": 8689
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.990808129310608,
+      "learning_rate": 9.677625901715892e-06,
+      "loss": 0.7009,
+      "step": 8690
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.0089926719665527,
+      "learning_rate": 9.677532755321142e-06,
+      "loss": 0.7643,
+      "step": 8691
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.0536069869995117,
+      "learning_rate": 9.677439595919925e-06,
+      "loss": 0.7566,
+      "step": 8692
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.9526302814483643,
+      "learning_rate": 9.677346423512502e-06,
+      "loss": 0.6623,
+      "step": 8693
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.607224464416504,
+      "learning_rate": 9.67725323809913e-06,
+      "loss": 0.7097,
+      "step": 8694
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.788073778152466,
+      "learning_rate": 9.677160039680067e-06,
+      "loss": 0.6104,
+      "step": 8695
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.043384075164795,
+      "learning_rate": 9.677066828255577e-06,
+      "loss": 0.446,
+      "step": 8696
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.773749589920044,
+      "learning_rate": 9.676973603825914e-06,
+      "loss": 0.595,
+      "step": 8697
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.4306883811950684,
+      "learning_rate": 9.67688036639134e-06,
+      "loss": 0.675,
+      "step": 8698
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.103383779525757,
+      "learning_rate": 9.676787115952114e-06,
+      "loss": 0.619,
+      "step": 8699
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.003915786743164,
+      "learning_rate": 9.676693852508496e-06,
+      "loss": 0.7121,
+      "step": 8700
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.445542335510254,
+      "learning_rate": 9.676600576060743e-06,
+      "loss": 0.9386,
+      "step": 8701
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.42360520362854,
+      "learning_rate": 9.676507286609116e-06,
+      "loss": 0.7038,
+      "step": 8702
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.6120866537094116,
+      "learning_rate": 9.676413984153875e-06,
+      "loss": 0.5687,
+      "step": 8703
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.2994561195373535,
+      "learning_rate": 9.676320668695275e-06,
+      "loss": 0.7207,
+      "step": 8704
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.960589647293091,
+      "learning_rate": 9.676227340233583e-06,
+      "loss": 0.5173,
+      "step": 8705
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.9838786125183105,
+      "learning_rate": 9.676133998769053e-06,
+      "loss": 0.759,
+      "step": 8706
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.6479034423828125,
+      "learning_rate": 9.676040644301947e-06,
+      "loss": 0.6887,
+      "step": 8707
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.725938320159912,
+      "learning_rate": 9.675947276832522e-06,
+      "loss": 0.6763,
+      "step": 8708
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.036618232727051,
+      "learning_rate": 9.67585389636104e-06,
+      "loss": 0.6804,
+      "step": 8709
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.2378828525543213,
+      "learning_rate": 9.67576050288776e-06,
+      "loss": 0.7628,
+      "step": 8710
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.9556877613067627,
+      "learning_rate": 9.675667096412942e-06,
+      "loss": 0.7129,
+      "step": 8711
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.8442976474761963,
+      "learning_rate": 9.675573676936846e-06,
+      "loss": 0.9477,
+      "step": 8712
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.958019733428955,
+      "learning_rate": 9.67548024445973e-06,
+      "loss": 0.5411,
+      "step": 8713
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.3187248706817627,
+      "learning_rate": 9.675386798981855e-06,
+      "loss": 0.8536,
+      "step": 8714
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.1851253509521484,
+      "learning_rate": 9.675293340503482e-06,
+      "loss": 0.7841,
+      "step": 8715
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.6243399381637573,
+      "learning_rate": 9.675199869024868e-06,
+      "loss": 0.4403,
+      "step": 8716
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.63442063331604,
+      "learning_rate": 9.675106384546275e-06,
+      "loss": 0.6689,
+      "step": 8717
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.1994502544403076,
+      "learning_rate": 9.675012887067962e-06,
+      "loss": 0.7237,
+      "step": 8718
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.5294344425201416,
+      "learning_rate": 9.67491937659019e-06,
+      "loss": 0.4852,
+      "step": 8719
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.18308162689209,
+      "learning_rate": 9.674825853113218e-06,
+      "loss": 0.9203,
+      "step": 8720
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.1974291801452637,
+      "learning_rate": 9.674732316637307e-06,
+      "loss": 0.5718,
+      "step": 8721
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.6926469802856445,
+      "learning_rate": 9.674638767162717e-06,
+      "loss": 0.7743,
+      "step": 8722
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.16975474357605,
+      "learning_rate": 9.674545204689706e-06,
+      "loss": 0.6526,
+      "step": 8723
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.9108003377914429,
+      "learning_rate": 9.674451629218538e-06,
+      "loss": 0.775,
+      "step": 8724
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.172274351119995,
+      "learning_rate": 9.674358040749471e-06,
+      "loss": 0.6242,
+      "step": 8725
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 8.825780868530273,
+      "learning_rate": 9.674264439282765e-06,
+      "loss": 1.031,
+      "step": 8726
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.798898458480835,
+      "learning_rate": 9.67417082481868e-06,
+      "loss": 0.6987,
+      "step": 8727
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.842627763748169,
+      "learning_rate": 9.674077197357477e-06,
+      "loss": 0.9338,
+      "step": 8728
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.106072425842285,
+      "learning_rate": 9.673983556899417e-06,
+      "loss": 0.7716,
+      "step": 8729
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.8894760608673096,
+      "learning_rate": 9.67388990344476e-06,
+      "loss": 0.6809,
+      "step": 8730
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.621091604232788,
+      "learning_rate": 9.673796236993764e-06,
+      "loss": 0.6714,
+      "step": 8731
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.202763795852661,
+      "learning_rate": 9.673702557546694e-06,
+      "loss": 0.3922,
+      "step": 8732
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.8042852878570557,
+      "learning_rate": 9.673608865103806e-06,
+      "loss": 0.5415,
+      "step": 8733
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.860476016998291,
+      "learning_rate": 9.673515159665362e-06,
+      "loss": 0.5734,
+      "step": 8734
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.512300491333008,
+      "learning_rate": 9.673421441231626e-06,
+      "loss": 0.5499,
+      "step": 8735
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.610044002532959,
+      "learning_rate": 9.673327709802854e-06,
+      "loss": 0.5256,
+      "step": 8736
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.9829243421554565,
+      "learning_rate": 9.673233965379308e-06,
+      "loss": 0.6554,
+      "step": 8737
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.1019744873046875,
+      "learning_rate": 9.673140207961249e-06,
+      "loss": 0.6257,
+      "step": 8738
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.1812944412231445,
+      "learning_rate": 9.673046437548938e-06,
+      "loss": 0.4996,
+      "step": 8739
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.554523229598999,
+      "learning_rate": 9.672952654142636e-06,
+      "loss": 0.5808,
+      "step": 8740
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.236131429672241,
+      "learning_rate": 9.672858857742602e-06,
+      "loss": 0.6207,
+      "step": 8741
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.114142417907715,
+      "learning_rate": 9.672765048349097e-06,
+      "loss": 0.7029,
+      "step": 8742
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.6232264041900635,
+      "learning_rate": 9.672671225962385e-06,
+      "loss": 0.7358,
+      "step": 8743
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.7193849086761475,
+      "learning_rate": 9.672577390582725e-06,
+      "loss": 0.7842,
+      "step": 8744
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.769548773765564,
+      "learning_rate": 9.672483542210375e-06,
+      "loss": 0.5124,
+      "step": 8745
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.360008478164673,
+      "learning_rate": 9.672389680845602e-06,
+      "loss": 0.7706,
+      "step": 8746
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 4.074578762054443,
+      "learning_rate": 9.672295806488663e-06,
+      "loss": 0.8405,
+      "step": 8747
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.8803458213806152,
+      "learning_rate": 9.672201919139818e-06,
+      "loss": 0.6307,
+      "step": 8748
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.645871639251709,
+      "learning_rate": 9.67210801879933e-06,
+      "loss": 0.6991,
+      "step": 8749
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.8200531005859375,
+      "learning_rate": 9.67201410546746e-06,
+      "loss": 0.7405,
+      "step": 8750
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.336754083633423,
+      "learning_rate": 9.67192017914447e-06,
+      "loss": 0.5838,
+      "step": 8751
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.444692611694336,
+      "learning_rate": 9.67182623983062e-06,
+      "loss": 0.5023,
+      "step": 8752
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 4.078394889831543,
+      "learning_rate": 9.67173228752617e-06,
+      "loss": 0.6084,
+      "step": 8753
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 4.279819011688232,
+      "learning_rate": 9.671638322231382e-06,
+      "loss": 0.8504,
+      "step": 8754
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.386571168899536,
+      "learning_rate": 9.67154434394652e-06,
+      "loss": 0.7297,
+      "step": 8755
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.312030076980591,
+      "learning_rate": 9.671450352671843e-06,
+      "loss": 0.625,
+      "step": 8756
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.8101210594177246,
+      "learning_rate": 9.67135634840761e-06,
+      "loss": 0.7874,
+      "step": 8757
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.5879853963851929,
+      "learning_rate": 9.671262331154088e-06,
+      "loss": 0.7504,
+      "step": 8758
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.4854512214660645,
+      "learning_rate": 9.671168300911535e-06,
+      "loss": 0.8383,
+      "step": 8759
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.1255242824554443,
+      "learning_rate": 9.671074257680211e-06,
+      "loss": 0.7583,
+      "step": 8760
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.8405210971832275,
+      "learning_rate": 9.670980201460381e-06,
+      "loss": 0.728,
+      "step": 8761
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.6694655418395996,
+      "learning_rate": 9.670886132252303e-06,
+      "loss": 0.7474,
+      "step": 8762
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.4494324922561646,
+      "learning_rate": 9.67079205005624e-06,
+      "loss": 0.6952,
+      "step": 8763
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.5141639709472656,
+      "learning_rate": 9.670697954872456e-06,
+      "loss": 0.5873,
+      "step": 8764
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.120415449142456,
+      "learning_rate": 9.670603846701212e-06,
+      "loss": 0.6405,
+      "step": 8765
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.363326072692871,
+      "learning_rate": 9.670509725542766e-06,
+      "loss": 0.6317,
+      "step": 8766
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.7950005531311035,
+      "learning_rate": 9.670415591397381e-06,
+      "loss": 0.6731,
+      "step": 8767
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.8251538276672363,
+      "learning_rate": 9.670321444265322e-06,
+      "loss": 0.4452,
+      "step": 8768
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.620973587036133,
+      "learning_rate": 9.67022728414685e-06,
+      "loss": 0.8464,
+      "step": 8769
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.326273202896118,
+      "learning_rate": 9.670133111042222e-06,
+      "loss": 0.7383,
+      "step": 8770
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 4.745613098144531,
+      "learning_rate": 9.670038924951706e-06,
+      "loss": 0.4285,
+      "step": 8771
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.3257217407226562,
+      "learning_rate": 9.66994472587556e-06,
+      "loss": 0.7004,
+      "step": 8772
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.219292640686035,
+      "learning_rate": 9.669850513814047e-06,
+      "loss": 0.5973,
+      "step": 8773
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.0437710285186768,
+      "learning_rate": 9.66975628876743e-06,
+      "loss": 0.7734,
+      "step": 8774
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.5671191215515137,
+      "learning_rate": 9.669662050735968e-06,
+      "loss": 0.4917,
+      "step": 8775
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.630298614501953,
+      "learning_rate": 9.669567799719926e-06,
+      "loss": 0.827,
+      "step": 8776
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.254654884338379,
+      "learning_rate": 9.669473535719566e-06,
+      "loss": 0.5125,
+      "step": 8777
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.95975399017334,
+      "learning_rate": 9.669379258735148e-06,
+      "loss": 0.7252,
+      "step": 8778
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.272780179977417,
+      "learning_rate": 9.669284968766937e-06,
+      "loss": 0.7051,
+      "step": 8779
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.609161376953125,
+      "learning_rate": 9.669190665815193e-06,
+      "loss": 0.7055,
+      "step": 8780
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.151419162750244,
+      "learning_rate": 9.669096349880179e-06,
+      "loss": 0.7363,
+      "step": 8781
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.4541139602661133,
+      "learning_rate": 9.669002020962157e-06,
+      "loss": 0.9359,
+      "step": 8782
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.6118983030319214,
+      "learning_rate": 9.66890767906139e-06,
+      "loss": 0.661,
+      "step": 8783
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 5.283218860626221,
+      "learning_rate": 9.668813324178139e-06,
+      "loss": 0.7376,
+      "step": 8784
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.9895248413085938,
+      "learning_rate": 9.668718956312668e-06,
+      "loss": 0.5521,
+      "step": 8785
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.2929766178131104,
+      "learning_rate": 9.668624575465238e-06,
+      "loss": 0.5026,
+      "step": 8786
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.6445133686065674,
+      "learning_rate": 9.668530181636112e-06,
+      "loss": 0.7237,
+      "step": 8787
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.2171130180358887,
+      "learning_rate": 9.668435774825551e-06,
+      "loss": 0.7621,
+      "step": 8788
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.794267416000366,
+      "learning_rate": 9.66834135503382e-06,
+      "loss": 0.6701,
+      "step": 8789
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 20.070720672607422,
+      "learning_rate": 9.668246922261182e-06,
+      "loss": 0.666,
+      "step": 8790
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.3533592224121094,
+      "learning_rate": 9.668152476507895e-06,
+      "loss": 0.672,
+      "step": 8791
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.037874937057495,
+      "learning_rate": 9.668058017774227e-06,
+      "loss": 0.5324,
+      "step": 8792
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.4371681213378906,
+      "learning_rate": 9.667963546060437e-06,
+      "loss": 0.6138,
+      "step": 8793
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.1373863220214844,
+      "learning_rate": 9.66786906136679e-06,
+      "loss": 0.6862,
+      "step": 8794
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.5227386951446533,
+      "learning_rate": 9.667774563693548e-06,
+      "loss": 0.7215,
+      "step": 8795
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.1058719158172607,
+      "learning_rate": 9.667680053040974e-06,
+      "loss": 0.9162,
+      "step": 8796
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.7121050357818604,
+      "learning_rate": 9.667585529409329e-06,
+      "loss": 0.8447,
+      "step": 8797
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.482879400253296,
+      "learning_rate": 9.667490992798879e-06,
+      "loss": 0.8503,
+      "step": 8798
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.0349955558776855,
+      "learning_rate": 9.667396443209882e-06,
+      "loss": 0.7528,
+      "step": 8799
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.142754316329956,
+      "learning_rate": 9.667301880642606e-06,
+      "loss": 0.8838,
+      "step": 8800
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.663055419921875,
+      "learning_rate": 9.66720730509731e-06,
+      "loss": 0.5897,
+      "step": 8801
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.6391189098358154,
+      "learning_rate": 9.667112716574261e-06,
+      "loss": 0.5458,
+      "step": 8802
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.4721394777297974,
+      "learning_rate": 9.66701811507372e-06,
+      "loss": 0.4281,
+      "step": 8803
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.110684394836426,
+      "learning_rate": 9.666923500595949e-06,
+      "loss": 0.8228,
+      "step": 8804
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.1202237606048584,
+      "learning_rate": 9.666828873141212e-06,
+      "loss": 0.4778,
+      "step": 8805
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.8150641918182373,
+      "learning_rate": 9.666734232709773e-06,
+      "loss": 0.8519,
+      "step": 8806
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.8232063055038452,
+      "learning_rate": 9.666639579301892e-06,
+      "loss": 0.6209,
+      "step": 8807
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.594722032546997,
+      "learning_rate": 9.666544912917836e-06,
+      "loss": 0.824,
+      "step": 8808
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.1112287044525146,
+      "learning_rate": 9.666450233557868e-06,
+      "loss": 0.489,
+      "step": 8809
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.3054707050323486,
+      "learning_rate": 9.666355541222249e-06,
+      "loss": 0.716,
+      "step": 8810
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.6277103424072266,
+      "learning_rate": 9.666260835911244e-06,
+      "loss": 0.6313,
+      "step": 8811
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.4738576412200928,
+      "learning_rate": 9.666166117625113e-06,
+      "loss": 0.5059,
+      "step": 8812
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.3930277824401855,
+      "learning_rate": 9.666071386364124e-06,
+      "loss": 0.7265,
+      "step": 8813
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.014721632003784,
+      "learning_rate": 9.665976642128538e-06,
+      "loss": 0.6576,
+      "step": 8814
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.2964611053466797,
+      "learning_rate": 9.66588188491862e-06,
+      "loss": 0.5306,
+      "step": 8815
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.9146775007247925,
+      "learning_rate": 9.66578711473463e-06,
+      "loss": 0.6966,
+      "step": 8816
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.7618193626403809,
+      "learning_rate": 9.665692331576837e-06,
+      "loss": 0.4954,
+      "step": 8817
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.1614928245544434,
+      "learning_rate": 9.665597535445499e-06,
+      "loss": 1.0168,
+      "step": 8818
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.171674966812134,
+      "learning_rate": 9.665502726340882e-06,
+      "loss": 0.669,
+      "step": 8819
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.2353391647338867,
+      "learning_rate": 9.665407904263252e-06,
+      "loss": 0.6883,
+      "step": 8820
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.0934557914733887,
+      "learning_rate": 9.665313069212867e-06,
+      "loss": 0.7894,
+      "step": 8821
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 5.728153705596924,
+      "learning_rate": 9.665218221189996e-06,
+      "loss": 0.5736,
+      "step": 8822
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.8794363737106323,
+      "learning_rate": 9.6651233601949e-06,
+      "loss": 0.8413,
+      "step": 8823
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.8286707401275635,
+      "learning_rate": 9.665028486227845e-06,
+      "loss": 0.4454,
+      "step": 8824
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.1394340991973877,
+      "learning_rate": 9.664933599289093e-06,
+      "loss": 0.6363,
+      "step": 8825
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.4835121631622314,
+      "learning_rate": 9.664838699378908e-06,
+      "loss": 0.5895,
+      "step": 8826
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.4755568504333496,
+      "learning_rate": 9.664743786497552e-06,
+      "loss": 0.7771,
+      "step": 8827
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.948798418045044,
+      "learning_rate": 9.664648860645294e-06,
+      "loss": 0.6807,
+      "step": 8828
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.0644304752349854,
+      "learning_rate": 9.664553921822393e-06,
+      "loss": 0.7878,
+      "step": 8829
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.2813198566436768,
+      "learning_rate": 9.664458970029116e-06,
+      "loss": 0.6915,
+      "step": 8830
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.2820544242858887,
+      "learning_rate": 9.664364005265724e-06,
+      "loss": 0.7125,
+      "step": 8831
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.2493960857391357,
+      "learning_rate": 9.664269027532484e-06,
+      "loss": 0.684,
+      "step": 8832
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.9182159900665283,
+      "learning_rate": 9.66417403682966e-06,
+      "loss": 0.7251,
+      "step": 8833
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.5450632572174072,
+      "learning_rate": 9.664079033157517e-06,
+      "loss": 0.732,
+      "step": 8834
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.830838918685913,
+      "learning_rate": 9.663984016516313e-06,
+      "loss": 0.7792,
+      "step": 8835
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.1844053268432617,
+      "learning_rate": 9.66388898690632e-06,
+      "loss": 0.4487,
+      "step": 8836
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.233731269836426,
+      "learning_rate": 9.663793944327798e-06,
+      "loss": 0.7106,
+      "step": 8837
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.302997350692749,
+      "learning_rate": 9.66369888878101e-06,
+      "loss": 0.8345,
+      "step": 8838
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.645581007003784,
+      "learning_rate": 9.663603820266225e-06,
+      "loss": 0.685,
+      "step": 8839
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 4.159610748291016,
+      "learning_rate": 9.663508738783704e-06,
+      "loss": 0.5525,
+      "step": 8840
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.271293878555298,
+      "learning_rate": 9.663413644333713e-06,
+      "loss": 0.5771,
+      "step": 8841
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.6176875829696655,
+      "learning_rate": 9.663318536916515e-06,
+      "loss": 0.5068,
+      "step": 8842
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.1451451778411865,
+      "learning_rate": 9.663223416532374e-06,
+      "loss": 0.6716,
+      "step": 8843
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 4.204276084899902,
+      "learning_rate": 9.663128283181558e-06,
+      "loss": 0.6075,
+      "step": 8844
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.34540057182312,
+      "learning_rate": 9.663033136864326e-06,
+      "loss": 0.6715,
+      "step": 8845
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.072725534439087,
+      "learning_rate": 9.662937977580948e-06,
+      "loss": 0.9667,
+      "step": 8846
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.7928375005722046,
+      "learning_rate": 9.662842805331684e-06,
+      "loss": 0.5742,
+      "step": 8847
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.5404961109161377,
+      "learning_rate": 9.662747620116804e-06,
+      "loss": 0.6501,
+      "step": 8848
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.3723618984222412,
+      "learning_rate": 9.662652421936568e-06,
+      "loss": 0.6017,
+      "step": 8849
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.242497205734253,
+      "learning_rate": 9.66255721079124e-06,
+      "loss": 0.6943,
+      "step": 8850
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.0281665325164795,
+      "learning_rate": 9.662461986681092e-06,
+      "loss": 0.6562,
+      "step": 8851
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.5044679641723633,
+      "learning_rate": 9.66236674960638e-06,
+      "loss": 0.6411,
+      "step": 8852
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.3353841304779053,
+      "learning_rate": 9.662271499567374e-06,
+      "loss": 0.5125,
+      "step": 8853
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.1067910194396973,
+      "learning_rate": 9.662176236564338e-06,
+      "loss": 0.7378,
+      "step": 8854
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.7649697065353394,
+      "learning_rate": 9.662080960597535e-06,
+      "loss": 0.4665,
+      "step": 8855
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.5407823324203491,
+      "learning_rate": 9.661985671667231e-06,
+      "loss": 0.6776,
+      "step": 8856
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.0574681758880615,
+      "learning_rate": 9.661890369773692e-06,
+      "loss": 0.738,
+      "step": 8857
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.693726062774658,
+      "learning_rate": 9.661795054917182e-06,
+      "loss": 0.6535,
+      "step": 8858
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 8.228614807128906,
+      "learning_rate": 9.66169972709797e-06,
+      "loss": 0.7087,
+      "step": 8859
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 4.104874610900879,
+      "learning_rate": 9.661604386316313e-06,
+      "loss": 1.0352,
+      "step": 8860
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 4.090569019317627,
+      "learning_rate": 9.661509032572481e-06,
+      "loss": 0.6404,
+      "step": 8861
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.8062289953231812,
+      "learning_rate": 9.66141366586674e-06,
+      "loss": 0.6024,
+      "step": 8862
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.039367914199829,
+      "learning_rate": 9.661318286199352e-06,
+      "loss": 0.568,
+      "step": 8863
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.1588101387023926,
+      "learning_rate": 9.661222893570586e-06,
+      "loss": 0.4784,
+      "step": 8864
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 6.575103759765625,
+      "learning_rate": 9.661127487980704e-06,
+      "loss": 0.4998,
+      "step": 8865
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.695946216583252,
+      "learning_rate": 9.661032069429971e-06,
+      "loss": 0.5613,
+      "step": 8866
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.8889232873916626,
+      "learning_rate": 9.660936637918657e-06,
+      "loss": 0.6259,
+      "step": 8867
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.625464916229248,
+      "learning_rate": 9.660841193447022e-06,
+      "loss": 0.4749,
+      "step": 8868
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.8360397815704346,
+      "learning_rate": 9.660745736015336e-06,
+      "loss": 0.5792,
+      "step": 8869
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.6611647605895996,
+      "learning_rate": 9.66065026562386e-06,
+      "loss": 0.7312,
+      "step": 8870
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.728693723678589,
+      "learning_rate": 9.660554782272861e-06,
+      "loss": 0.5175,
+      "step": 8871
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 5.5705413818359375,
+      "learning_rate": 9.660459285962607e-06,
+      "loss": 0.9568,
+      "step": 8872
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.693734645843506,
+      "learning_rate": 9.660363776693361e-06,
+      "loss": 0.5968,
+      "step": 8873
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.368353843688965,
+      "learning_rate": 9.66026825446539e-06,
+      "loss": 0.8224,
+      "step": 8874
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.0350871086120605,
+      "learning_rate": 9.660172719278957e-06,
+      "loss": 0.6193,
+      "step": 8875
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.991678237915039,
+      "learning_rate": 9.66007717113433e-06,
+      "loss": 0.5959,
+      "step": 8876
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.782475233078003,
+      "learning_rate": 9.659981610031774e-06,
+      "loss": 0.4936,
+      "step": 8877
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.6291141510009766,
+      "learning_rate": 9.659886035971555e-06,
+      "loss": 0.7538,
+      "step": 8878
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.8042261600494385,
+      "learning_rate": 9.659790448953938e-06,
+      "loss": 0.5873,
+      "step": 8879
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.2318739891052246,
+      "learning_rate": 9.65969484897919e-06,
+      "loss": 0.6534,
+      "step": 8880
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.5892741680145264,
+      "learning_rate": 9.659599236047576e-06,
+      "loss": 0.6296,
+      "step": 8881
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.341031789779663,
+      "learning_rate": 9.659503610159363e-06,
+      "loss": 0.8094,
+      "step": 8882
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.8002843856811523,
+      "learning_rate": 9.659407971314815e-06,
+      "loss": 0.7781,
+      "step": 8883
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.133571147918701,
+      "learning_rate": 9.6593123195142e-06,
+      "loss": 0.8648,
+      "step": 8884
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.8544716835021973,
+      "learning_rate": 9.659216654757782e-06,
+      "loss": 0.7824,
+      "step": 8885
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.5535001754760742,
+      "learning_rate": 9.659120977045827e-06,
+      "loss": 0.5544,
+      "step": 8886
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 4.587620258331299,
+      "learning_rate": 9.659025286378602e-06,
+      "loss": 0.5905,
+      "step": 8887
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.539450168609619,
+      "learning_rate": 9.658929582756373e-06,
+      "loss": 0.5651,
+      "step": 8888
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.8810675144195557,
+      "learning_rate": 9.658833866179407e-06,
+      "loss": 0.7174,
+      "step": 8889
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.804142951965332,
+      "learning_rate": 9.65873813664797e-06,
+      "loss": 0.7431,
+      "step": 8890
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.8588318824768066,
+      "learning_rate": 9.658642394162326e-06,
+      "loss": 0.6867,
+      "step": 8891
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.661200761795044,
+      "learning_rate": 9.658546638722743e-06,
+      "loss": 0.5258,
+      "step": 8892
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.8966474533081055,
+      "learning_rate": 9.658450870329488e-06,
+      "loss": 0.6731,
+      "step": 8893
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.306063652038574,
+      "learning_rate": 9.658355088982824e-06,
+      "loss": 0.4703,
+      "step": 8894
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.5537703037261963,
+      "learning_rate": 9.658259294683021e-06,
+      "loss": 0.8125,
+      "step": 8895
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.383596897125244,
+      "learning_rate": 9.658163487430345e-06,
+      "loss": 0.5966,
+      "step": 8896
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.5567710399627686,
+      "learning_rate": 9.658067667225059e-06,
+      "loss": 0.6376,
+      "step": 8897
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.2411725521087646,
+      "learning_rate": 9.657971834067433e-06,
+      "loss": 0.6228,
+      "step": 8898
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.154454469680786,
+      "learning_rate": 9.657875987957731e-06,
+      "loss": 0.9357,
+      "step": 8899
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.5275049209594727,
+      "learning_rate": 9.657780128896224e-06,
+      "loss": 0.7825,
+      "step": 8900
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.2394773960113525,
+      "learning_rate": 9.657684256883174e-06,
+      "loss": 0.3497,
+      "step": 8901
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.697767734527588,
+      "learning_rate": 9.657588371918846e-06,
+      "loss": 0.5526,
+      "step": 8902
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 4.703580379486084,
+      "learning_rate": 9.657492474003512e-06,
+      "loss": 0.8401,
+      "step": 8903
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.5667831897735596,
+      "learning_rate": 9.657396563137435e-06,
+      "loss": 0.6476,
+      "step": 8904
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.4764761924743652,
+      "learning_rate": 9.657300639320883e-06,
+      "loss": 1.0124,
+      "step": 8905
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.827434539794922,
+      "learning_rate": 9.657204702554123e-06,
+      "loss": 0.6317,
+      "step": 8906
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.7468079328536987,
+      "learning_rate": 9.657108752837423e-06,
+      "loss": 0.7015,
+      "step": 8907
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.583381175994873,
+      "learning_rate": 9.657012790171047e-06,
+      "loss": 0.7549,
+      "step": 8908
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.0743894577026367,
+      "learning_rate": 9.656916814555262e-06,
+      "loss": 0.4692,
+      "step": 8909
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.423967123031616,
+      "learning_rate": 9.656820825990337e-06,
+      "loss": 0.4273,
+      "step": 8910
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.391204595565796,
+      "learning_rate": 9.656724824476537e-06,
+      "loss": 0.7195,
+      "step": 8911
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.0121114253997803,
+      "learning_rate": 9.656628810014131e-06,
+      "loss": 0.5654,
+      "step": 8912
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.952876329421997,
+      "learning_rate": 9.656532782603383e-06,
+      "loss": 0.7683,
+      "step": 8913
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 5.346926689147949,
+      "learning_rate": 9.656436742244561e-06,
+      "loss": 0.6892,
+      "step": 8914
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.3486664295196533,
+      "learning_rate": 9.656340688937934e-06,
+      "loss": 0.8466,
+      "step": 8915
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.758354663848877,
+      "learning_rate": 9.656244622683768e-06,
+      "loss": 0.6829,
+      "step": 8916
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.115903377532959,
+      "learning_rate": 9.656148543482331e-06,
+      "loss": 0.8726,
+      "step": 8917
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.9196717739105225,
+      "learning_rate": 9.656052451333889e-06,
+      "loss": 0.7834,
+      "step": 8918
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 7.784257888793945,
+      "learning_rate": 9.655956346238708e-06,
+      "loss": 0.7081,
+      "step": 8919
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.6971824169158936,
+      "learning_rate": 9.655860228197057e-06,
+      "loss": 0.5941,
+      "step": 8920
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 4.694906711578369,
+      "learning_rate": 9.655764097209202e-06,
+      "loss": 0.8076,
+      "step": 8921
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.7469735145568848,
+      "learning_rate": 9.655667953275413e-06,
+      "loss": 0.6851,
+      "step": 8922
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.9527406692504883,
+      "learning_rate": 9.655571796395954e-06,
+      "loss": 0.5657,
+      "step": 8923
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.050011157989502,
+      "learning_rate": 9.655475626571095e-06,
+      "loss": 0.6994,
+      "step": 8924
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.5998365879058838,
+      "learning_rate": 9.6553794438011e-06,
+      "loss": 0.5388,
+      "step": 8925
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.785574197769165,
+      "learning_rate": 9.65528324808624e-06,
+      "loss": 0.8398,
+      "step": 8926
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.3392181396484375,
+      "learning_rate": 9.655187039426783e-06,
+      "loss": 0.5618,
+      "step": 8927
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.149937391281128,
+      "learning_rate": 9.655090817822993e-06,
+      "loss": 0.5143,
+      "step": 8928
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.146240711212158,
+      "learning_rate": 9.65499458327514e-06,
+      "loss": 0.7856,
+      "step": 8929
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.349750518798828,
+      "learning_rate": 9.654898335783489e-06,
+      "loss": 0.6849,
+      "step": 8930
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.3296492099761963,
+      "learning_rate": 9.65480207534831e-06,
+      "loss": 0.7566,
+      "step": 8931
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.6273622512817383,
+      "learning_rate": 9.654705801969872e-06,
+      "loss": 0.6479,
+      "step": 8932
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.5623583793640137,
+      "learning_rate": 9.65460951564844e-06,
+      "loss": 0.7244,
+      "step": 8933
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.22717022895813,
+      "learning_rate": 9.654513216384283e-06,
+      "loss": 0.8085,
+      "step": 8934
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.711453914642334,
+      "learning_rate": 9.654416904177668e-06,
+      "loss": 0.3434,
+      "step": 8935
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.7320539951324463,
+      "learning_rate": 9.654320579028861e-06,
+      "loss": 0.9547,
+      "step": 8936
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.883171319961548,
+      "learning_rate": 9.654224240938134e-06,
+      "loss": 0.6628,
+      "step": 8937
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.539255380630493,
+      "learning_rate": 9.654127889905754e-06,
+      "loss": 0.656,
+      "step": 8938
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.2381831407546997,
+      "learning_rate": 9.654031525931987e-06,
+      "loss": 0.5539,
+      "step": 8939
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.9671261310577393,
+      "learning_rate": 9.653935149017102e-06,
+      "loss": 0.445,
+      "step": 8940
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.7833564281463623,
+      "learning_rate": 9.653838759161366e-06,
+      "loss": 0.8383,
+      "step": 8941
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.7618911266326904,
+      "learning_rate": 9.653742356365049e-06,
+      "loss": 0.5691,
+      "step": 8942
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.1083908081054688,
+      "learning_rate": 9.653645940628416e-06,
+      "loss": 0.6462,
+      "step": 8943
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.0445311069488525,
+      "learning_rate": 9.653549511951737e-06,
+      "loss": 0.5127,
+      "step": 8944
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.018934726715088,
+      "learning_rate": 9.653453070335281e-06,
+      "loss": 0.7448,
+      "step": 8945
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.6759612560272217,
+      "learning_rate": 9.653356615779315e-06,
+      "loss": 0.5499,
+      "step": 8946
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.3914074897766113,
+      "learning_rate": 9.653260148284107e-06,
+      "loss": 0.5304,
+      "step": 8947
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.2965309619903564,
+      "learning_rate": 9.653163667849927e-06,
+      "loss": 0.6842,
+      "step": 8948
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.1838624477386475,
+      "learning_rate": 9.653067174477041e-06,
+      "loss": 0.6315,
+      "step": 8949
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.047696113586426,
+      "learning_rate": 9.652970668165718e-06,
+      "loss": 0.5682,
+      "step": 8950
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.8601951599121094,
+      "learning_rate": 9.652874148916226e-06,
+      "loss": 0.5846,
+      "step": 8951
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.0312421321868896,
+      "learning_rate": 9.652777616728837e-06,
+      "loss": 0.8194,
+      "step": 8952
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.379729747772217,
+      "learning_rate": 9.652681071603815e-06,
+      "loss": 0.6974,
+      "step": 8953
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.1391372680664062,
+      "learning_rate": 9.652584513541428e-06,
+      "loss": 0.8167,
+      "step": 8954
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.1637725830078125,
+      "learning_rate": 9.652487942541949e-06,
+      "loss": 0.4919,
+      "step": 8955
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.4900827407836914,
+      "learning_rate": 9.652391358605642e-06,
+      "loss": 0.4963,
+      "step": 8956
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.7169077396392822,
+      "learning_rate": 9.652294761732778e-06,
+      "loss": 0.6236,
+      "step": 8957
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.6700938940048218,
+      "learning_rate": 9.652198151923624e-06,
+      "loss": 0.684,
+      "step": 8958
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.423887252807617,
+      "learning_rate": 9.652101529178451e-06,
+      "loss": 0.8423,
+      "step": 8959
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.098507881164551,
+      "learning_rate": 9.652004893497527e-06,
+      "loss": 0.6854,
+      "step": 8960
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 4.698646545410156,
+      "learning_rate": 9.651908244881118e-06,
+      "loss": 0.7489,
+      "step": 8961
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.9189419746398926,
+      "learning_rate": 9.651811583329497e-06,
+      "loss": 0.5766,
+      "step": 8962
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.4463396072387695,
+      "learning_rate": 9.65171490884293e-06,
+      "loss": 0.869,
+      "step": 8963
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.591742992401123,
+      "learning_rate": 9.651618221421685e-06,
+      "loss": 0.7432,
+      "step": 8964
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.8019020557403564,
+      "learning_rate": 9.651521521066034e-06,
+      "loss": 0.7806,
+      "step": 8965
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.2668120861053467,
+      "learning_rate": 9.651424807776244e-06,
+      "loss": 0.7789,
+      "step": 8966
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.9751665592193604,
+      "learning_rate": 9.651328081552583e-06,
+      "loss": 0.5664,
+      "step": 8967
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.9656927585601807,
+      "learning_rate": 9.651231342395322e-06,
+      "loss": 0.8076,
+      "step": 8968
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.171710252761841,
+      "learning_rate": 9.651134590304728e-06,
+      "loss": 0.541,
+      "step": 8969
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.487367630004883,
+      "learning_rate": 9.65103782528107e-06,
+      "loss": 0.5635,
+      "step": 8970
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.9508566856384277,
+      "learning_rate": 9.650941047324622e-06,
+      "loss": 0.4345,
+      "step": 8971
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.3319437503814697,
+      "learning_rate": 9.650844256435646e-06,
+      "loss": 0.5837,
+      "step": 8972
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.348374366760254,
+      "learning_rate": 9.650747452614416e-06,
+      "loss": 0.6593,
+      "step": 8973
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.0323710441589355,
+      "learning_rate": 9.650650635861198e-06,
+      "loss": 0.7343,
+      "step": 8974
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.8283898830413818,
+      "learning_rate": 9.650553806176264e-06,
+      "loss": 0.5362,
+      "step": 8975
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.134399890899658,
+      "learning_rate": 9.650456963559883e-06,
+      "loss": 0.8996,
+      "step": 8976
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.7458109855651855,
+      "learning_rate": 9.65036010801232e-06,
+      "loss": 0.7936,
+      "step": 8977
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.2629895210266113,
+      "learning_rate": 9.65026323953385e-06,
+      "loss": 0.6835,
+      "step": 8978
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.5934970378875732,
+      "learning_rate": 9.650166358124739e-06,
+      "loss": 0.7953,
+      "step": 8979
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 4.518874168395996,
+      "learning_rate": 9.650069463785258e-06,
+      "loss": 0.6722,
+      "step": 8980
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 4.46768856048584,
+      "learning_rate": 9.649972556515677e-06,
+      "loss": 0.7139,
+      "step": 8981
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.908592939376831,
+      "learning_rate": 9.649875636316263e-06,
+      "loss": 0.7363,
+      "step": 8982
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.8313379287719727,
+      "learning_rate": 9.649778703187286e-06,
+      "loss": 0.7488,
+      "step": 8983
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.6592462062835693,
+      "learning_rate": 9.649681757129017e-06,
+      "loss": 0.7604,
+      "step": 8984
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 4.190092086791992,
+      "learning_rate": 9.649584798141724e-06,
+      "loss": 0.8095,
+      "step": 8985
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.0823800563812256,
+      "learning_rate": 9.649487826225679e-06,
+      "loss": 0.5607,
+      "step": 8986
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.1254770755767822,
+      "learning_rate": 9.64939084138115e-06,
+      "loss": 0.7019,
+      "step": 8987
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.1376538276672363,
+      "learning_rate": 9.649293843608405e-06,
+      "loss": 0.6382,
+      "step": 8988
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 4.274931907653809,
+      "learning_rate": 9.649196832907717e-06,
+      "loss": 0.8453,
+      "step": 8989
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.755249261856079,
+      "learning_rate": 9.649099809279355e-06,
+      "loss": 0.6318,
+      "step": 8990
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.0303497314453125,
+      "learning_rate": 9.649002772723586e-06,
+      "loss": 0.5734,
+      "step": 8991
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.185819149017334,
+      "learning_rate": 9.648905723240682e-06,
+      "loss": 0.6798,
+      "step": 8992
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 4.023479461669922,
+      "learning_rate": 9.648808660830914e-06,
+      "loss": 0.6541,
+      "step": 8993
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 4.576733112335205,
+      "learning_rate": 9.64871158549455e-06,
+      "loss": 0.758,
+      "step": 8994
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 5.896394729614258,
+      "learning_rate": 9.648614497231862e-06,
+      "loss": 0.7835,
+      "step": 8995
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.9486929178237915,
+      "learning_rate": 9.648517396043115e-06,
+      "loss": 0.4561,
+      "step": 8996
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.39420747756958,
+      "learning_rate": 9.648420281928584e-06,
+      "loss": 0.6105,
+      "step": 8997
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.130918025970459,
+      "learning_rate": 9.64832315488854e-06,
+      "loss": 0.7931,
+      "step": 8998
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.6747732162475586,
+      "learning_rate": 9.648226014923248e-06,
+      "loss": 0.7464,
+      "step": 8999
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.142186164855957,
+      "learning_rate": 9.64812886203298e-06,
+      "loss": 0.5072,
+      "step": 9000
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.772045135498047,
+      "learning_rate": 9.648031696218009e-06,
+      "loss": 0.8567,
+      "step": 9001
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.7058571577072144,
+      "learning_rate": 9.647934517478602e-06,
+      "loss": 0.5238,
+      "step": 9002
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 5.220035076141357,
+      "learning_rate": 9.64783732581503e-06,
+      "loss": 1.0313,
+      "step": 9003
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.692866802215576,
+      "learning_rate": 9.647740121227564e-06,
+      "loss": 0.5433,
+      "step": 9004
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 4.613753795623779,
+      "learning_rate": 9.647642903716474e-06,
+      "loss": 0.6717,
+      "step": 9005
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.173454999923706,
+      "learning_rate": 9.647545673282031e-06,
+      "loss": 0.3449,
+      "step": 9006
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.167635917663574,
+      "learning_rate": 9.647448429924503e-06,
+      "loss": 0.6645,
+      "step": 9007
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.9194788932800293,
+      "learning_rate": 9.647351173644161e-06,
+      "loss": 0.7412,
+      "step": 9008
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.1452698707580566,
+      "learning_rate": 9.647253904441278e-06,
+      "loss": 0.7674,
+      "step": 9009
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.457973003387451,
+      "learning_rate": 9.647156622316122e-06,
+      "loss": 0.8028,
+      "step": 9010
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.323556423187256,
+      "learning_rate": 9.647059327268966e-06,
+      "loss": 0.4981,
+      "step": 9011
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.2688498497009277,
+      "learning_rate": 9.646962019300076e-06,
+      "loss": 0.844,
+      "step": 9012
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.6100046634674072,
+      "learning_rate": 9.646864698409728e-06,
+      "loss": 0.6203,
+      "step": 9013
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.336200714111328,
+      "learning_rate": 9.64676736459819e-06,
+      "loss": 0.7441,
+      "step": 9014
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.4294016361236572,
+      "learning_rate": 9.64667001786573e-06,
+      "loss": 0.6621,
+      "step": 9015
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.8043638467788696,
+      "learning_rate": 9.646572658212625e-06,
+      "loss": 0.6535,
+      "step": 9016
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.693516492843628,
+      "learning_rate": 9.64647528563914e-06,
+      "loss": 0.7629,
+      "step": 9017
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.711160659790039,
+      "learning_rate": 9.646377900145547e-06,
+      "loss": 0.6382,
+      "step": 9018
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.2734789848327637,
+      "learning_rate": 9.646280501732119e-06,
+      "loss": 0.8713,
+      "step": 9019
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.5177861452102661,
+      "learning_rate": 9.646183090399127e-06,
+      "loss": 0.4039,
+      "step": 9020
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.9589673280715942,
+      "learning_rate": 9.646085666146837e-06,
+      "loss": 0.6613,
+      "step": 9021
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.2646594047546387,
+      "learning_rate": 9.645988228975524e-06,
+      "loss": 0.8408,
+      "step": 9022
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 5.3766655921936035,
+      "learning_rate": 9.64589077888546e-06,
+      "loss": 0.733,
+      "step": 9023
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.460942029953003,
+      "learning_rate": 9.645793315876912e-06,
+      "loss": 0.969,
+      "step": 9024
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.584542155265808,
+      "learning_rate": 9.645695839950154e-06,
+      "loss": 0.6206,
+      "step": 9025
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.2583229541778564,
+      "learning_rate": 9.645598351105456e-06,
+      "loss": 0.679,
+      "step": 9026
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.312166213989258,
+      "learning_rate": 9.64550084934309e-06,
+      "loss": 0.6745,
+      "step": 9027
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.3967554569244385,
+      "learning_rate": 9.645403334663326e-06,
+      "loss": 0.6123,
+      "step": 9028
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.359586000442505,
+      "learning_rate": 9.645305807066436e-06,
+      "loss": 0.7407,
+      "step": 9029
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.241316795349121,
+      "learning_rate": 9.64520826655269e-06,
+      "loss": 0.7623,
+      "step": 9030
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.3007590770721436,
+      "learning_rate": 9.64511071312236e-06,
+      "loss": 0.4657,
+      "step": 9031
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.0276224613189697,
+      "learning_rate": 9.645013146775716e-06,
+      "loss": 0.7497,
+      "step": 9032
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.6059722900390625,
+      "learning_rate": 9.644915567513034e-06,
+      "loss": 0.723,
+      "step": 9033
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.1668896675109863,
+      "learning_rate": 9.644817975334579e-06,
+      "loss": 0.6201,
+      "step": 9034
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 5.443350791931152,
+      "learning_rate": 9.644720370240626e-06,
+      "loss": 0.7955,
+      "step": 9035
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.1045851707458496,
+      "learning_rate": 9.644622752231445e-06,
+      "loss": 0.7465,
+      "step": 9036
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 4.216970443725586,
+      "learning_rate": 9.64452512130731e-06,
+      "loss": 0.8297,
+      "step": 9037
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.7657499313354492,
+      "learning_rate": 9.644427477468487e-06,
+      "loss": 0.6674,
+      "step": 9038
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 7.261105537414551,
+      "learning_rate": 9.644329820715254e-06,
+      "loss": 0.6149,
+      "step": 9039
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.300884246826172,
+      "learning_rate": 9.644232151047878e-06,
+      "loss": 0.8742,
+      "step": 9040
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.868029832839966,
+      "learning_rate": 9.644134468466633e-06,
+      "loss": 0.4648,
+      "step": 9041
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.0219764709472656,
+      "learning_rate": 9.64403677297179e-06,
+      "loss": 0.6561,
+      "step": 9042
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.158957004547119,
+      "learning_rate": 9.643939064563619e-06,
+      "loss": 0.72,
+      "step": 9043
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.5927720069885254,
+      "learning_rate": 9.643841343242395e-06,
+      "loss": 0.8305,
+      "step": 9044
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.7755725383758545,
+      "learning_rate": 9.643743609008386e-06,
+      "loss": 0.5309,
+      "step": 9045
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.42649507522583,
+      "learning_rate": 9.643645861861868e-06,
+      "loss": 0.9962,
+      "step": 9046
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.202354907989502,
+      "learning_rate": 9.64354810180311e-06,
+      "loss": 0.9766,
+      "step": 9047
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.097158908843994,
+      "learning_rate": 9.643450328832383e-06,
+      "loss": 0.7414,
+      "step": 9048
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.9944026470184326,
+      "learning_rate": 9.64335254294996e-06,
+      "loss": 0.6841,
+      "step": 9049
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.3741836547851562,
+      "learning_rate": 9.643254744156112e-06,
+      "loss": 0.5693,
+      "step": 9050
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.8548657894134521,
+      "learning_rate": 9.643156932451113e-06,
+      "loss": 0.5986,
+      "step": 9051
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.835768699645996,
+      "learning_rate": 9.643059107835235e-06,
+      "loss": 0.4433,
+      "step": 9052
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.7736570835113525,
+      "learning_rate": 9.642961270308748e-06,
+      "loss": 0.8176,
+      "step": 9053
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.6282951831817627,
+      "learning_rate": 9.642863419871926e-06,
+      "loss": 0.6931,
+      "step": 9054
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.485215663909912,
+      "learning_rate": 9.642765556525037e-06,
+      "loss": 0.7998,
+      "step": 9055
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.41902756690979,
+      "learning_rate": 9.64266768026836e-06,
+      "loss": 0.7224,
+      "step": 9056
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.8407992124557495,
+      "learning_rate": 9.64256979110216e-06,
+      "loss": 0.7487,
+      "step": 9057
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.568352222442627,
+      "learning_rate": 9.642471889026717e-06,
+      "loss": 0.5932,
+      "step": 9058
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.8993630409240723,
+      "learning_rate": 9.642373974042296e-06,
+      "loss": 0.6579,
+      "step": 9059
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.9005932807922363,
+      "learning_rate": 9.642276046149172e-06,
+      "loss": 0.87,
+      "step": 9060
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 4.623261451721191,
+      "learning_rate": 9.642178105347618e-06,
+      "loss": 0.6396,
+      "step": 9061
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.5869817733764648,
+      "learning_rate": 9.642080151637904e-06,
+      "loss": 0.5,
+      "step": 9062
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.449575185775757,
+      "learning_rate": 9.641982185020306e-06,
+      "loss": 0.598,
+      "step": 9063
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.3841345310211182,
+      "learning_rate": 9.641884205495093e-06,
+      "loss": 0.5741,
+      "step": 9064
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.6974503993988037,
+      "learning_rate": 9.64178621306254e-06,
+      "loss": 0.7839,
+      "step": 9065
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.528719663619995,
+      "learning_rate": 9.64168820772292e-06,
+      "loss": 0.6168,
+      "step": 9066
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.3297407627105713,
+      "learning_rate": 9.6415901894765e-06,
+      "loss": 0.7314,
+      "step": 9067
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.9798909425735474,
+      "learning_rate": 9.641492158323559e-06,
+      "loss": 0.3837,
+      "step": 9068
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.645381212234497,
+      "learning_rate": 9.641394114264366e-06,
+      "loss": 0.3833,
+      "step": 9069
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.0073609352111816,
+      "learning_rate": 9.641296057299197e-06,
+      "loss": 0.6723,
+      "step": 9070
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.075974941253662,
+      "learning_rate": 9.64119798742832e-06,
+      "loss": 0.833,
+      "step": 9071
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.18418025970459,
+      "learning_rate": 9.64109990465201e-06,
+      "loss": 0.7887,
+      "step": 9072
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.1711373329162598,
+      "learning_rate": 9.64100180897054e-06,
+      "loss": 0.4616,
+      "step": 9073
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.260305881500244,
+      "learning_rate": 9.640903700384183e-06,
+      "loss": 0.8447,
+      "step": 9074
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.160562038421631,
+      "learning_rate": 9.64080557889321e-06,
+      "loss": 0.8394,
+      "step": 9075
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.6719810962677,
+      "learning_rate": 9.640707444497898e-06,
+      "loss": 0.5998,
+      "step": 9076
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.8054897785186768,
+      "learning_rate": 9.640609297198515e-06,
+      "loss": 0.6952,
+      "step": 9077
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 4.297336101531982,
+      "learning_rate": 9.640511136995337e-06,
+      "loss": 0.7776,
+      "step": 9078
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.3039917945861816,
+      "learning_rate": 9.640412963888635e-06,
+      "loss": 1.0124,
+      "step": 9079
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 5.543468952178955,
+      "learning_rate": 9.640314777878684e-06,
+      "loss": 0.6944,
+      "step": 9080
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 5.240705966949463,
+      "learning_rate": 9.640216578965754e-06,
+      "loss": 0.8158,
+      "step": 9081
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.7893165349960327,
+      "learning_rate": 9.640118367150123e-06,
+      "loss": 0.6186,
+      "step": 9082
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.315391778945923,
+      "learning_rate": 9.640020142432057e-06,
+      "loss": 0.5394,
+      "step": 9083
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.3493874073028564,
+      "learning_rate": 9.639921904811836e-06,
+      "loss": 0.6537,
+      "step": 9084
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.2796595096588135,
+      "learning_rate": 9.639823654289728e-06,
+      "loss": 0.6702,
+      "step": 9085
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.259685754776001,
+      "learning_rate": 9.63972539086601e-06,
+      "loss": 0.8146,
+      "step": 9086
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.121725559234619,
+      "learning_rate": 9.639627114540954e-06,
+      "loss": 0.5109,
+      "step": 9087
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.4799175262451172,
+      "learning_rate": 9.639528825314834e-06,
+      "loss": 0.5734,
+      "step": 9088
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.1387975215911865,
+      "learning_rate": 9.639430523187921e-06,
+      "loss": 0.8593,
+      "step": 9089
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.409719467163086,
+      "learning_rate": 9.63933220816049e-06,
+      "loss": 0.5352,
+      "step": 9090
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.6094114780426025,
+      "learning_rate": 9.639233880232815e-06,
+      "loss": 0.7027,
+      "step": 9091
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.8280071020126343,
+      "learning_rate": 9.639135539405166e-06,
+      "loss": 0.714,
+      "step": 9092
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.1124026775360107,
+      "learning_rate": 9.639037185677821e-06,
+      "loss": 0.7375,
+      "step": 9093
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.316516399383545,
+      "learning_rate": 9.638938819051052e-06,
+      "loss": 0.901,
+      "step": 9094
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.913557529449463,
+      "learning_rate": 9.63884043952513e-06,
+      "loss": 0.6851,
+      "step": 9095
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.875072956085205,
+      "learning_rate": 9.63874204710033e-06,
+      "loss": 0.6829,
+      "step": 9096
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.9766740798950195,
+      "learning_rate": 9.638643641776929e-06,
+      "loss": 0.681,
+      "step": 9097
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.318445920944214,
+      "learning_rate": 9.638545223555196e-06,
+      "loss": 0.5805,
+      "step": 9098
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.5572521686553955,
+      "learning_rate": 9.638446792435406e-06,
+      "loss": 0.5053,
+      "step": 9099
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.3317313194274902,
+      "learning_rate": 9.638348348417835e-06,
+      "loss": 0.8296,
+      "step": 9100
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.4863994121551514,
+      "learning_rate": 9.63824989150275e-06,
+      "loss": 0.6778,
+      "step": 9101
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.604731321334839,
+      "learning_rate": 9.638151421690435e-06,
+      "loss": 0.8034,
+      "step": 9102
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.01792573928833,
+      "learning_rate": 9.638052938981156e-06,
+      "loss": 0.617,
+      "step": 9103
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.5949485301971436,
+      "learning_rate": 9.637954443375189e-06,
+      "loss": 0.7866,
+      "step": 9104
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.7707171440124512,
+      "learning_rate": 9.637855934872808e-06,
+      "loss": 0.5444,
+      "step": 9105
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.5201897621154785,
+      "learning_rate": 9.637757413474288e-06,
+      "loss": 0.7087,
+      "step": 9106
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.141629219055176,
+      "learning_rate": 9.637658879179902e-06,
+      "loss": 0.7733,
+      "step": 9107
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.2926437854766846,
+      "learning_rate": 9.637560331989923e-06,
+      "loss": 0.546,
+      "step": 9108
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.3731415271759033,
+      "learning_rate": 9.637461771904627e-06,
+      "loss": 0.5191,
+      "step": 9109
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.560654878616333,
+      "learning_rate": 9.637363198924287e-06,
+      "loss": 0.6292,
+      "step": 9110
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.0008137226104736,
+      "learning_rate": 9.637264613049176e-06,
+      "loss": 0.4797,
+      "step": 9111
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.8111706972122192,
+      "learning_rate": 9.637166014279571e-06,
+      "loss": 0.6549,
+      "step": 9112
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.339073896408081,
+      "learning_rate": 9.637067402615744e-06,
+      "loss": 0.6879,
+      "step": 9113
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.0972650051116943,
+      "learning_rate": 9.636968778057969e-06,
+      "loss": 0.5418,
+      "step": 9114
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.198732376098633,
+      "learning_rate": 9.63687014060652e-06,
+      "loss": 0.551,
+      "step": 9115
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.0318048000335693,
+      "learning_rate": 9.636771490261673e-06,
+      "loss": 0.725,
+      "step": 9116
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.5943262577056885,
+      "learning_rate": 9.636672827023702e-06,
+      "loss": 0.8407,
+      "step": 9117
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.2808542251586914,
+      "learning_rate": 9.63657415089288e-06,
+      "loss": 0.5445,
+      "step": 9118
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 8.846039772033691,
+      "learning_rate": 9.636475461869483e-06,
+      "loss": 0.713,
+      "step": 9119
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.5284671783447266,
+      "learning_rate": 9.636376759953784e-06,
+      "loss": 0.7164,
+      "step": 9120
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.155930995941162,
+      "learning_rate": 9.636278045146058e-06,
+      "loss": 0.7618,
+      "step": 9121
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.0200247764587402,
+      "learning_rate": 9.63617931744658e-06,
+      "loss": 0.7139,
+      "step": 9122
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.853576421737671,
+      "learning_rate": 9.636080576855625e-06,
+      "loss": 0.6596,
+      "step": 9123
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.597778558731079,
+      "learning_rate": 9.635981823373464e-06,
+      "loss": 0.7278,
+      "step": 9124
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.7141265869140625,
+      "learning_rate": 9.635883057000376e-06,
+      "loss": 0.6095,
+      "step": 9125
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.3520796298980713,
+      "learning_rate": 9.635784277736635e-06,
+      "loss": 0.5296,
+      "step": 9126
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.855483889579773,
+      "learning_rate": 9.635685485582512e-06,
+      "loss": 0.6745,
+      "step": 9127
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 4.729060173034668,
+      "learning_rate": 9.635586680538284e-06,
+      "loss": 0.8258,
+      "step": 9128
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 4.7564215660095215,
+      "learning_rate": 9.635487862604227e-06,
+      "loss": 0.57,
+      "step": 9129
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.2300148010253906,
+      "learning_rate": 9.635389031780616e-06,
+      "loss": 0.4778,
+      "step": 9130
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 4.612700462341309,
+      "learning_rate": 9.635290188067725e-06,
+      "loss": 0.8416,
+      "step": 9131
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.8707780838012695,
+      "learning_rate": 9.635191331465826e-06,
+      "loss": 0.6765,
+      "step": 9132
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 4.376836776733398,
+      "learning_rate": 9.635092461975197e-06,
+      "loss": 0.6276,
+      "step": 9133
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.957150459289551,
+      "learning_rate": 9.634993579596112e-06,
+      "loss": 0.8611,
+      "step": 9134
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.236206293106079,
+      "learning_rate": 9.634894684328848e-06,
+      "loss": 0.6305,
+      "step": 9135
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 4.60676908493042,
+      "learning_rate": 9.634795776173677e-06,
+      "loss": 0.6691,
+      "step": 9136
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 7.320321083068848,
+      "learning_rate": 9.634696855130874e-06,
+      "loss": 0.8322,
+      "step": 9137
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.3338258266448975,
+      "learning_rate": 9.634597921200717e-06,
+      "loss": 0.8156,
+      "step": 9138
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.9477417469024658,
+      "learning_rate": 9.634498974383479e-06,
+      "loss": 0.8232,
+      "step": 9139
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.4564549922943115,
+      "learning_rate": 9.634400014679435e-06,
+      "loss": 0.8318,
+      "step": 9140
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.222780227661133,
+      "learning_rate": 9.63430104208886e-06,
+      "loss": 0.797,
+      "step": 9141
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.4795610904693604,
+      "learning_rate": 9.63420205661203e-06,
+      "loss": 0.7032,
+      "step": 9142
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.8250125646591187,
+      "learning_rate": 9.63410305824922e-06,
+      "loss": 0.7268,
+      "step": 9143
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.033677816390991,
+      "learning_rate": 9.634004047000707e-06,
+      "loss": 0.9275,
+      "step": 9144
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.3569773435592651,
+      "learning_rate": 9.633905022866764e-06,
+      "loss": 0.4724,
+      "step": 9145
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.1866331100463867,
+      "learning_rate": 9.633805985847666e-06,
+      "loss": 0.8077,
+      "step": 9146
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.8902547359466553,
+      "learning_rate": 9.63370693594369e-06,
+      "loss": 0.9639,
+      "step": 9147
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 7.762159824371338,
+      "learning_rate": 9.633607873155113e-06,
+      "loss": 0.79,
+      "step": 9148
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 4.797220230102539,
+      "learning_rate": 9.633508797482205e-06,
+      "loss": 0.7951,
+      "step": 9149
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.298466920852661,
+      "learning_rate": 9.633409708925245e-06,
+      "loss": 0.4701,
+      "step": 9150
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.485834836959839,
+      "learning_rate": 9.63331060748451e-06,
+      "loss": 0.8679,
+      "step": 9151
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.85491943359375,
+      "learning_rate": 9.633211493160274e-06,
+      "loss": 0.737,
+      "step": 9152
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.4062693119049072,
+      "learning_rate": 9.633112365952811e-06,
+      "loss": 0.6135,
+      "step": 9153
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.4001564979553223,
+      "learning_rate": 9.6330132258624e-06,
+      "loss": 0.4888,
+      "step": 9154
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.6704118251800537,
+      "learning_rate": 9.632914072889312e-06,
+      "loss": 0.6255,
+      "step": 9155
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.4999866485595703,
+      "learning_rate": 9.632814907033829e-06,
+      "loss": 0.5725,
+      "step": 9156
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.7120914459228516,
+      "learning_rate": 9.63271572829622e-06,
+      "loss": 0.5503,
+      "step": 9157
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.4959442615509033,
+      "learning_rate": 9.632616536676767e-06,
+      "loss": 0.8288,
+      "step": 9158
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.3554654121398926,
+      "learning_rate": 9.63251733217574e-06,
+      "loss": 0.6214,
+      "step": 9159
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.885139226913452,
+      "learning_rate": 9.632418114793419e-06,
+      "loss": 0.6184,
+      "step": 9160
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.1213011741638184,
+      "learning_rate": 9.632318884530078e-06,
+      "loss": 0.5032,
+      "step": 9161
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.385004997253418,
+      "learning_rate": 9.632219641385995e-06,
+      "loss": 0.6696,
+      "step": 9162
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.6190775632858276,
+      "learning_rate": 9.632120385361443e-06,
+      "loss": 0.778,
+      "step": 9163
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.986325740814209,
+      "learning_rate": 9.632021116456701e-06,
+      "loss": 0.7634,
+      "step": 9164
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.875401735305786,
+      "learning_rate": 9.63192183467204e-06,
+      "loss": 0.5064,
+      "step": 9165
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.348681688308716,
+      "learning_rate": 9.631822540007743e-06,
+      "loss": 0.4316,
+      "step": 9166
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.4408364295959473,
+      "learning_rate": 9.63172323246408e-06,
+      "loss": 0.9466,
+      "step": 9167
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.369736671447754,
+      "learning_rate": 9.631623912041332e-06,
+      "loss": 0.8176,
+      "step": 9168
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.370767593383789,
+      "learning_rate": 9.631524578739772e-06,
+      "loss": 0.5104,
+      "step": 9169
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.958517551422119,
+      "learning_rate": 9.631425232559678e-06,
+      "loss": 0.613,
+      "step": 9170
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.005302667617798,
+      "learning_rate": 9.631325873501325e-06,
+      "loss": 0.7111,
+      "step": 9171
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.5800623893737793,
+      "learning_rate": 9.63122650156499e-06,
+      "loss": 0.3818,
+      "step": 9172
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 4.004000663757324,
+      "learning_rate": 9.631127116750948e-06,
+      "loss": 0.8392,
+      "step": 9173
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.633358955383301,
+      "learning_rate": 9.631027719059476e-06,
+      "loss": 0.6756,
+      "step": 9174
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.085622787475586,
+      "learning_rate": 9.630928308490852e-06,
+      "loss": 0.5804,
+      "step": 9175
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.144685983657837,
+      "learning_rate": 9.63082888504535e-06,
+      "loss": 0.7947,
+      "step": 9176
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.432814359664917,
+      "learning_rate": 9.63072944872325e-06,
+      "loss": 0.7977,
+      "step": 9177
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.572338819503784,
+      "learning_rate": 9.630629999524822e-06,
+      "loss": 0.6891,
+      "step": 9178
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.6647441387176514,
+      "learning_rate": 9.630530537450351e-06,
+      "loss": 0.7797,
+      "step": 9179
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.622718095779419,
+      "learning_rate": 9.630431062500108e-06,
+      "loss": 0.6257,
+      "step": 9180
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.284806251525879,
+      "learning_rate": 9.630331574674368e-06,
+      "loss": 0.7618,
+      "step": 9181
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.900129795074463,
+      "learning_rate": 9.630232073973414e-06,
+      "loss": 0.7556,
+      "step": 9182
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.1742324829101562,
+      "learning_rate": 9.630132560397517e-06,
+      "loss": 0.7013,
+      "step": 9183
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.721203327178955,
+      "learning_rate": 9.630033033946957e-06,
+      "loss": 0.7673,
+      "step": 9184
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.9659923315048218,
+      "learning_rate": 9.629933494622007e-06,
+      "loss": 0.6821,
+      "step": 9185
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.653113603591919,
+      "learning_rate": 9.629833942422951e-06,
+      "loss": 0.8148,
+      "step": 9186
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.4407970905303955,
+      "learning_rate": 9.629734377350058e-06,
+      "loss": 0.9539,
+      "step": 9187
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.288463592529297,
+      "learning_rate": 9.629634799403608e-06,
+      "loss": 0.6243,
+      "step": 9188
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.9192054271698,
+      "learning_rate": 9.629535208583878e-06,
+      "loss": 0.6335,
+      "step": 9189
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.5066535472869873,
+      "learning_rate": 9.629435604891146e-06,
+      "loss": 0.6962,
+      "step": 9190
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.018653392791748,
+      "learning_rate": 9.629335988325688e-06,
+      "loss": 0.361,
+      "step": 9191
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.6549975872039795,
+      "learning_rate": 9.62923635888778e-06,
+      "loss": 0.779,
+      "step": 9192
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.7975223064422607,
+      "learning_rate": 9.629136716577699e-06,
+      "loss": 0.8184,
+      "step": 9193
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.7613375186920166,
+      "learning_rate": 9.629037061395725e-06,
+      "loss": 0.8849,
+      "step": 9194
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.8259708881378174,
+      "learning_rate": 9.628937393342131e-06,
+      "loss": 0.5941,
+      "step": 9195
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.764789581298828,
+      "learning_rate": 9.628837712417197e-06,
+      "loss": 0.6232,
+      "step": 9196
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.5629210472106934,
+      "learning_rate": 9.6287380186212e-06,
+      "loss": 0.6593,
+      "step": 9197
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.269958972930908,
+      "learning_rate": 9.628638311954415e-06,
+      "loss": 0.7626,
+      "step": 9198
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.1478068828582764,
+      "learning_rate": 9.628538592417122e-06,
+      "loss": 0.6288,
+      "step": 9199
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.667252779006958,
+      "learning_rate": 9.628438860009597e-06,
+      "loss": 0.6362,
+      "step": 9200
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.8861584663391113,
+      "learning_rate": 9.628339114732118e-06,
+      "loss": 0.6334,
+      "step": 9201
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.279052257537842,
+      "learning_rate": 9.628239356584959e-06,
+      "loss": 0.4937,
+      "step": 9202
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.719318389892578,
+      "learning_rate": 9.628139585568402e-06,
+      "loss": 0.5423,
+      "step": 9203
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.7199742794036865,
+      "learning_rate": 9.628039801682722e-06,
+      "loss": 0.7488,
+      "step": 9204
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.440434217453003,
+      "learning_rate": 9.627940004928198e-06,
+      "loss": 0.7956,
+      "step": 9205
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.232495069503784,
+      "learning_rate": 9.627840195305105e-06,
+      "loss": 0.7785,
+      "step": 9206
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.9977788925170898,
+      "learning_rate": 9.627740372813724e-06,
+      "loss": 0.7506,
+      "step": 9207
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.9753663539886475,
+      "learning_rate": 9.62764053745433e-06,
+      "loss": 0.7276,
+      "step": 9208
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.277208685874939,
+      "learning_rate": 9.6275406892272e-06,
+      "loss": 0.5401,
+      "step": 9209
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.467670202255249,
+      "learning_rate": 9.627440828132615e-06,
+      "loss": 0.6225,
+      "step": 9210
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.7736117839813232,
+      "learning_rate": 9.627340954170848e-06,
+      "loss": 0.7381,
+      "step": 9211
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.0796358585357666,
+      "learning_rate": 9.627241067342181e-06,
+      "loss": 0.4595,
+      "step": 9212
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.7916176319122314,
+      "learning_rate": 9.627141167646889e-06,
+      "loss": 0.5613,
+      "step": 9213
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.7593108415603638,
+      "learning_rate": 9.627041255085253e-06,
+      "loss": 0.7934,
+      "step": 9214
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.5483808517456055,
+      "learning_rate": 9.626941329657547e-06,
+      "loss": 0.6074,
+      "step": 9215
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.8498321771621704,
+      "learning_rate": 9.62684139136405e-06,
+      "loss": 0.6773,
+      "step": 9216
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.201711654663086,
+      "learning_rate": 9.62674144020504e-06,
+      "loss": 0.7206,
+      "step": 9217
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.2689425945281982,
+      "learning_rate": 9.626641476180795e-06,
+      "loss": 0.4971,
+      "step": 9218
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.682175636291504,
+      "learning_rate": 9.626541499291594e-06,
+      "loss": 1.0005,
+      "step": 9219
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.8788950443267822,
+      "learning_rate": 9.626441509537716e-06,
+      "loss": 0.7541,
+      "step": 9220
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.0534775257110596,
+      "learning_rate": 9.626341506919434e-06,
+      "loss": 0.8298,
+      "step": 9221
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 4.600705623626709,
+      "learning_rate": 9.626241491437032e-06,
+      "loss": 0.9255,
+      "step": 9222
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.891151785850525,
+      "learning_rate": 9.626141463090784e-06,
+      "loss": 0.5744,
+      "step": 9223
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.181910276412964,
+      "learning_rate": 9.626041421880971e-06,
+      "loss": 0.8132,
+      "step": 9224
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 4.094030857086182,
+      "learning_rate": 9.625941367807868e-06,
+      "loss": 0.8545,
+      "step": 9225
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.4615516662597656,
+      "learning_rate": 9.625841300871757e-06,
+      "loss": 0.648,
+      "step": 9226
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.897463321685791,
+      "learning_rate": 9.625741221072913e-06,
+      "loss": 0.5092,
+      "step": 9227
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.4454262256622314,
+      "learning_rate": 9.625641128411615e-06,
+      "loss": 0.7483,
+      "step": 9228
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.3722240924835205,
+      "learning_rate": 9.625541022888142e-06,
+      "loss": 0.691,
+      "step": 9229
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.309617280960083,
+      "learning_rate": 9.625440904502771e-06,
+      "loss": 0.7064,
+      "step": 9230
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.1015586853027344,
+      "learning_rate": 9.625340773255785e-06,
+      "loss": 0.8227,
+      "step": 9231
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 4.188119888305664,
+      "learning_rate": 9.625240629147457e-06,
+      "loss": 0.749,
+      "step": 9232
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.20414662361145,
+      "learning_rate": 9.625140472178067e-06,
+      "loss": 0.9256,
+      "step": 9233
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.6929664611816406,
+      "learning_rate": 9.625040302347895e-06,
+      "loss": 0.7651,
+      "step": 9234
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.885590672492981,
+      "learning_rate": 9.624940119657218e-06,
+      "loss": 0.6229,
+      "step": 9235
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 5.0554986000061035,
+      "learning_rate": 9.624839924106315e-06,
+      "loss": 0.5943,
+      "step": 9236
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 4.004422187805176,
+      "learning_rate": 9.624739715695465e-06,
+      "loss": 0.7018,
+      "step": 9237
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.8546090126037598,
+      "learning_rate": 9.624639494424945e-06,
+      "loss": 0.6853,
+      "step": 9238
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.080018997192383,
+      "learning_rate": 9.624539260295036e-06,
+      "loss": 0.7974,
+      "step": 9239
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 4.97280216217041,
+      "learning_rate": 9.624439013306017e-06,
+      "loss": 0.8866,
+      "step": 9240
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.0639288425445557,
+      "learning_rate": 9.624338753458164e-06,
+      "loss": 0.8674,
+      "step": 9241
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.5961925983428955,
+      "learning_rate": 9.624238480751758e-06,
+      "loss": 0.693,
+      "step": 9242
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.596165418624878,
+      "learning_rate": 9.624138195187076e-06,
+      "loss": 0.5761,
+      "step": 9243
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.9534189701080322,
+      "learning_rate": 9.624037896764397e-06,
+      "loss": 0.5647,
+      "step": 9244
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.462174892425537,
+      "learning_rate": 9.623937585484004e-06,
+      "loss": 0.3241,
+      "step": 9245
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.2451133728027344,
+      "learning_rate": 9.62383726134617e-06,
+      "loss": 0.7625,
+      "step": 9246
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.8664538860321045,
+      "learning_rate": 9.62373692435118e-06,
+      "loss": 0.8933,
+      "step": 9247
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.0571110248565674,
+      "learning_rate": 9.623636574499306e-06,
+      "loss": 0.436,
+      "step": 9248
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.8766251802444458,
+      "learning_rate": 9.623536211790831e-06,
+      "loss": 0.7667,
+      "step": 9249
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.1329281330108643,
+      "learning_rate": 9.623435836226036e-06,
+      "loss": 0.6791,
+      "step": 9250
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.414661169052124,
+      "learning_rate": 9.623335447805198e-06,
+      "loss": 0.6898,
+      "step": 9251
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.8021678924560547,
+      "learning_rate": 9.623235046528594e-06,
+      "loss": 0.6636,
+      "step": 9252
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.473249912261963,
+      "learning_rate": 9.623134632396507e-06,
+      "loss": 0.7187,
+      "step": 9253
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 4.361594200134277,
+      "learning_rate": 9.623034205409214e-06,
+      "loss": 0.8701,
+      "step": 9254
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.9391567707061768,
+      "learning_rate": 9.622933765566993e-06,
+      "loss": 0.6755,
+      "step": 9255
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.1698925495147705,
+      "learning_rate": 9.622833312870127e-06,
+      "loss": 0.6859,
+      "step": 9256
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.639292001724243,
+      "learning_rate": 9.622732847318891e-06,
+      "loss": 0.7096,
+      "step": 9257
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.9368826150894165,
+      "learning_rate": 9.62263236891357e-06,
+      "loss": 0.5093,
+      "step": 9258
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.9387379884719849,
+      "learning_rate": 9.622531877654438e-06,
+      "loss": 0.4452,
+      "step": 9259
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.919856309890747,
+      "learning_rate": 9.622431373541776e-06,
+      "loss": 0.7812,
+      "step": 9260
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 4.504508972167969,
+      "learning_rate": 9.622330856575866e-06,
+      "loss": 0.5305,
+      "step": 9261
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.1697640419006348,
+      "learning_rate": 9.622230326756984e-06,
+      "loss": 0.5914,
+      "step": 9262
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.7600743770599365,
+      "learning_rate": 9.62212978408541e-06,
+      "loss": 0.6505,
+      "step": 9263
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.713416576385498,
+      "learning_rate": 9.622029228561427e-06,
+      "loss": 0.8582,
+      "step": 9264
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.7852973937988281,
+      "learning_rate": 9.62192866018531e-06,
+      "loss": 0.4168,
+      "step": 9265
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.808659553527832,
+      "learning_rate": 9.621828078957343e-06,
+      "loss": 0.4756,
+      "step": 9266
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.8755950927734375,
+      "learning_rate": 9.621727484877802e-06,
+      "loss": 0.7737,
+      "step": 9267
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.573814868927002,
+      "learning_rate": 9.621626877946969e-06,
+      "loss": 0.6238,
+      "step": 9268
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.8976504802703857,
+      "learning_rate": 9.621526258165121e-06,
+      "loss": 0.7149,
+      "step": 9269
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.735164165496826,
+      "learning_rate": 9.621425625532541e-06,
+      "loss": 0.7145,
+      "step": 9270
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.032008171081543,
+      "learning_rate": 9.621324980049508e-06,
+      "loss": 0.4682,
+      "step": 9271
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.7272837162017822,
+      "learning_rate": 9.621224321716303e-06,
+      "loss": 0.6266,
+      "step": 9272
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.983700156211853,
+      "learning_rate": 9.621123650533202e-06,
+      "loss": 0.5879,
+      "step": 9273
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 5.184019088745117,
+      "learning_rate": 9.621022966500487e-06,
+      "loss": 0.9024,
+      "step": 9274
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 4.431836128234863,
+      "learning_rate": 9.62092226961844e-06,
+      "loss": 0.7287,
+      "step": 9275
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.2462925910949707,
+      "learning_rate": 9.620821559887338e-06,
+      "loss": 0.5841,
+      "step": 9276
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.038663148880005,
+      "learning_rate": 9.620720837307463e-06,
+      "loss": 0.8382,
+      "step": 9277
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.5190069675445557,
+      "learning_rate": 9.620620101879092e-06,
+      "loss": 0.6364,
+      "step": 9278
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.1377665996551514,
+      "learning_rate": 9.62051935360251e-06,
+      "loss": 0.4451,
+      "step": 9279
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.3205103874206543,
+      "learning_rate": 9.620418592477994e-06,
+      "loss": 0.5646,
+      "step": 9280
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 4.247714042663574,
+      "learning_rate": 9.620317818505824e-06,
+      "loss": 0.9674,
+      "step": 9281
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.586759328842163,
+      "learning_rate": 9.62021703168628e-06,
+      "loss": 0.7059,
+      "step": 9282
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.8724048137664795,
+      "learning_rate": 9.620116232019644e-06,
+      "loss": 0.5761,
+      "step": 9283
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.6577773094177246,
+      "learning_rate": 9.620015419506198e-06,
+      "loss": 0.7007,
+      "step": 9284
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 4.378198146820068,
+      "learning_rate": 9.619914594146218e-06,
+      "loss": 0.8358,
+      "step": 9285
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.996290445327759,
+      "learning_rate": 9.619813755939986e-06,
+      "loss": 0.6922,
+      "step": 9286
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.99086856842041,
+      "learning_rate": 9.619712904887783e-06,
+      "loss": 0.7626,
+      "step": 9287
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.455630302429199,
+      "learning_rate": 9.619612040989887e-06,
+      "loss": 0.4912,
+      "step": 9288
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.4690463542938232,
+      "learning_rate": 9.619511164246583e-06,
+      "loss": 0.825,
+      "step": 9289
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.279906988143921,
+      "learning_rate": 9.619410274658147e-06,
+      "loss": 0.5518,
+      "step": 9290
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.866755247116089,
+      "learning_rate": 9.619309372224864e-06,
+      "loss": 0.5847,
+      "step": 9291
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 6.270918846130371,
+      "learning_rate": 9.619208456947012e-06,
+      "loss": 0.6369,
+      "step": 9292
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 4.857220649719238,
+      "learning_rate": 9.61910752882487e-06,
+      "loss": 0.7956,
+      "step": 9293
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.819601058959961,
+      "learning_rate": 9.61900658785872e-06,
+      "loss": 0.6444,
+      "step": 9294
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.4768617153167725,
+      "learning_rate": 9.618905634048843e-06,
+      "loss": 0.604,
+      "step": 9295
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.2077479362487793,
+      "learning_rate": 9.61880466739552e-06,
+      "loss": 0.4836,
+      "step": 9296
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.1878278255462646,
+      "learning_rate": 9.618703687899031e-06,
+      "loss": 0.819,
+      "step": 9297
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.754781723022461,
+      "learning_rate": 9.618602695559658e-06,
+      "loss": 0.7268,
+      "step": 9298
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.7633765935897827,
+      "learning_rate": 9.618501690377682e-06,
+      "loss": 0.5512,
+      "step": 9299
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.5548739433288574,
+      "learning_rate": 9.618400672353382e-06,
+      "loss": 0.7998,
+      "step": 9300
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.16007399559021,
+      "learning_rate": 9.61829964148704e-06,
+      "loss": 0.6301,
+      "step": 9301
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.6231623888015747,
+      "learning_rate": 9.618198597778935e-06,
+      "loss": 0.748,
+      "step": 9302
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.306713104248047,
+      "learning_rate": 9.618097541229353e-06,
+      "loss": 0.7678,
+      "step": 9303
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 5.1402435302734375,
+      "learning_rate": 9.617996471838569e-06,
+      "loss": 0.8958,
+      "step": 9304
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.274935007095337,
+      "learning_rate": 9.617895389606868e-06,
+      "loss": 0.5572,
+      "step": 9305
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.9416043758392334,
+      "learning_rate": 9.61779429453453e-06,
+      "loss": 0.5941,
+      "step": 9306
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.649076223373413,
+      "learning_rate": 9.617693186621836e-06,
+      "loss": 0.6928,
+      "step": 9307
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.715383291244507,
+      "learning_rate": 9.617592065869065e-06,
+      "loss": 0.5869,
+      "step": 9308
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.329418182373047,
+      "learning_rate": 9.617490932276503e-06,
+      "loss": 0.8137,
+      "step": 9309
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.893345355987549,
+      "learning_rate": 9.617389785844428e-06,
+      "loss": 0.5356,
+      "step": 9310
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.8980116844177246,
+      "learning_rate": 9.61728862657312e-06,
+      "loss": 0.7772,
+      "step": 9311
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.212078809738159,
+      "learning_rate": 9.617187454462864e-06,
+      "loss": 0.7005,
+      "step": 9312
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 4.25345516204834,
+      "learning_rate": 9.617086269513937e-06,
+      "loss": 0.702,
+      "step": 9313
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.399299383163452,
+      "learning_rate": 9.616985071726625e-06,
+      "loss": 0.7956,
+      "step": 9314
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.1939938068389893,
+      "learning_rate": 9.616883861101207e-06,
+      "loss": 0.7528,
+      "step": 9315
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.710381507873535,
+      "learning_rate": 9.616782637637963e-06,
+      "loss": 0.6789,
+      "step": 9316
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.818877696990967,
+      "learning_rate": 9.616681401337177e-06,
+      "loss": 0.5297,
+      "step": 9317
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.3020622730255127,
+      "learning_rate": 9.61658015219913e-06,
+      "loss": 0.8496,
+      "step": 9318
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 11.877994537353516,
+      "learning_rate": 9.6164788902241e-06,
+      "loss": 0.8303,
+      "step": 9319
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.7647833824157715,
+      "learning_rate": 9.616377615412376e-06,
+      "loss": 0.5512,
+      "step": 9320
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.9260588884353638,
+      "learning_rate": 9.616276327764231e-06,
+      "loss": 0.7173,
+      "step": 9321
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.9399852752685547,
+      "learning_rate": 9.616175027279954e-06,
+      "loss": 0.6303,
+      "step": 9322
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.9192118644714355,
+      "learning_rate": 9.616073713959821e-06,
+      "loss": 0.5373,
+      "step": 9323
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.854340076446533,
+      "learning_rate": 9.615972387804117e-06,
+      "loss": 0.8264,
+      "step": 9324
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 4.233853340148926,
+      "learning_rate": 9.615871048813124e-06,
+      "loss": 0.4698,
+      "step": 9325
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.039365530014038,
+      "learning_rate": 9.615769696987123e-06,
+      "loss": 0.5668,
+      "step": 9326
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.883851170539856,
+      "learning_rate": 9.615668332326395e-06,
+      "loss": 0.6093,
+      "step": 9327
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.598233938217163,
+      "learning_rate": 9.61556695483122e-06,
+      "loss": 0.5089,
+      "step": 9328
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.549323558807373,
+      "learning_rate": 9.615465564501886e-06,
+      "loss": 0.636,
+      "step": 9329
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.5549793243408203,
+      "learning_rate": 9.61536416133867e-06,
+      "loss": 0.7435,
+      "step": 9330
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.345106601715088,
+      "learning_rate": 9.615262745341857e-06,
+      "loss": 0.6385,
+      "step": 9331
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.1793737411499023,
+      "learning_rate": 9.615161316511724e-06,
+      "loss": 0.5466,
+      "step": 9332
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 4.023952484130859,
+      "learning_rate": 9.615059874848557e-06,
+      "loss": 0.6479,
+      "step": 9333
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 4.7019853591918945,
+      "learning_rate": 9.614958420352638e-06,
+      "loss": 0.7371,
+      "step": 9334
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.5624194145202637,
+      "learning_rate": 9.614856953024248e-06,
+      "loss": 0.5354,
+      "step": 9335
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.23028564453125,
+      "learning_rate": 9.614755472863671e-06,
+      "loss": 0.725,
+      "step": 9336
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.9788565635681152,
+      "learning_rate": 9.614653979871184e-06,
+      "loss": 0.7695,
+      "step": 9337
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.323652505874634,
+      "learning_rate": 9.614552474047076e-06,
+      "loss": 0.5962,
+      "step": 9338
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.223637342453003,
+      "learning_rate": 9.614450955391627e-06,
+      "loss": 0.7794,
+      "step": 9339
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.2367804050445557,
+      "learning_rate": 9.614349423905116e-06,
+      "loss": 0.6648,
+      "step": 9340
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.195786952972412,
+      "learning_rate": 9.61424787958783e-06,
+      "loss": 0.5608,
+      "step": 9341
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.6985347270965576,
+      "learning_rate": 9.614146322440048e-06,
+      "loss": 0.6347,
+      "step": 9342
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.456251382827759,
+      "learning_rate": 9.614044752462054e-06,
+      "loss": 0.6396,
+      "step": 9343
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.1449882984161377,
+      "learning_rate": 9.61394316965413e-06,
+      "loss": 0.8923,
+      "step": 9344
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.3514246940612793,
+      "learning_rate": 9.613841574016557e-06,
+      "loss": 0.6107,
+      "step": 9345
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.9102351665496826,
+      "learning_rate": 9.613739965549622e-06,
+      "loss": 0.4927,
+      "step": 9346
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.0848491191864014,
+      "learning_rate": 9.613638344253602e-06,
+      "loss": 0.8606,
+      "step": 9347
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.783212661743164,
+      "learning_rate": 9.613536710128783e-06,
+      "loss": 0.7614,
+      "step": 9348
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.4095206260681152,
+      "learning_rate": 9.613435063175447e-06,
+      "loss": 0.672,
+      "step": 9349
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.8068294525146484,
+      "learning_rate": 9.613333403393876e-06,
+      "loss": 0.6145,
+      "step": 9350
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.9675201177597046,
+      "learning_rate": 9.613231730784353e-06,
+      "loss": 0.6027,
+      "step": 9351
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.826292037963867,
+      "learning_rate": 9.61313004534716e-06,
+      "loss": 0.6269,
+      "step": 9352
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.4919509887695312,
+      "learning_rate": 9.613028347082581e-06,
+      "loss": 0.8729,
+      "step": 9353
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 6.881424903869629,
+      "learning_rate": 9.612926635990897e-06,
+      "loss": 0.555,
+      "step": 9354
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.854924201965332,
+      "learning_rate": 9.612824912072395e-06,
+      "loss": 0.5137,
+      "step": 9355
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.5309712886810303,
+      "learning_rate": 9.612723175327353e-06,
+      "loss": 0.5282,
+      "step": 9356
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.9477436542510986,
+      "learning_rate": 9.612621425756057e-06,
+      "loss": 0.9892,
+      "step": 9357
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.5035985708236694,
+      "learning_rate": 9.612519663358788e-06,
+      "loss": 0.7903,
+      "step": 9358
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.579436779022217,
+      "learning_rate": 9.612417888135829e-06,
+      "loss": 0.732,
+      "step": 9359
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.041377544403076,
+      "learning_rate": 9.612316100087464e-06,
+      "loss": 0.7426,
+      "step": 9360
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.5206170082092285,
+      "learning_rate": 9.612214299213976e-06,
+      "loss": 0.5824,
+      "step": 9361
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.0298895835876465,
+      "learning_rate": 9.612112485515648e-06,
+      "loss": 0.7387,
+      "step": 9362
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.787697434425354,
+      "learning_rate": 9.612010658992762e-06,
+      "loss": 0.4778,
+      "step": 9363
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.88810658454895,
+      "learning_rate": 9.611908819645603e-06,
+      "loss": 0.7508,
+      "step": 9364
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.6013882160186768,
+      "learning_rate": 9.611806967474452e-06,
+      "loss": 0.4007,
+      "step": 9365
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 4.77581262588501,
+      "learning_rate": 9.611705102479596e-06,
+      "loss": 0.6148,
+      "step": 9366
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.912545680999756,
+      "learning_rate": 9.611603224661314e-06,
+      "loss": 0.5075,
+      "step": 9367
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.23537278175354,
+      "learning_rate": 9.611501334019891e-06,
+      "loss": 0.7064,
+      "step": 9368
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.23241925239563,
+      "learning_rate": 9.61139943055561e-06,
+      "loss": 0.4447,
+      "step": 9369
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 4.893078327178955,
+      "learning_rate": 9.611297514268757e-06,
+      "loss": 0.4531,
+      "step": 9370
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.9692257642745972,
+      "learning_rate": 9.61119558515961e-06,
+      "loss": 0.6611,
+      "step": 9371
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.9442877769470215,
+      "learning_rate": 9.611093643228456e-06,
+      "loss": 0.9945,
+      "step": 9372
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.500776529312134,
+      "learning_rate": 9.61099168847558e-06,
+      "loss": 0.5402,
+      "step": 9373
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.728515148162842,
+      "learning_rate": 9.610889720901262e-06,
+      "loss": 0.7579,
+      "step": 9374
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.673114776611328,
+      "learning_rate": 9.610787740505787e-06,
+      "loss": 0.621,
+      "step": 9375
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.642455577850342,
+      "learning_rate": 9.610685747289437e-06,
+      "loss": 0.7733,
+      "step": 9376
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.1912455558776855,
+      "learning_rate": 9.610583741252499e-06,
+      "loss": 0.6594,
+      "step": 9377
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.838888168334961,
+      "learning_rate": 9.610481722395254e-06,
+      "loss": 0.597,
+      "step": 9378
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.6111741065979004,
+      "learning_rate": 9.610379690717988e-06,
+      "loss": 0.6168,
+      "step": 9379
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.9205989837646484,
+      "learning_rate": 9.61027764622098e-06,
+      "loss": 0.6105,
+      "step": 9380
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.182098150253296,
+      "learning_rate": 9.61017558890452e-06,
+      "loss": 0.4412,
+      "step": 9381
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.1399855613708496,
+      "learning_rate": 9.610073518768887e-06,
+      "loss": 0.6731,
+      "step": 9382
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.962782621383667,
+      "learning_rate": 9.609971435814367e-06,
+      "loss": 0.7524,
+      "step": 9383
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.9980413913726807,
+      "learning_rate": 9.609869340041244e-06,
+      "loss": 0.6508,
+      "step": 9384
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 5.282434940338135,
+      "learning_rate": 9.609767231449802e-06,
+      "loss": 0.8264,
+      "step": 9385
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.042769193649292,
+      "learning_rate": 9.609665110040323e-06,
+      "loss": 0.7318,
+      "step": 9386
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.0588531494140625,
+      "learning_rate": 9.609562975813092e-06,
+      "loss": 0.8531,
+      "step": 9387
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.318140745162964,
+      "learning_rate": 9.609460828768393e-06,
+      "loss": 0.7091,
+      "step": 9388
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.4876925945281982,
+      "learning_rate": 9.60935866890651e-06,
+      "loss": 0.6917,
+      "step": 9389
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.887075901031494,
+      "learning_rate": 9.609256496227728e-06,
+      "loss": 0.7652,
+      "step": 9390
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 4.507138252258301,
+      "learning_rate": 9.60915431073233e-06,
+      "loss": 0.6545,
+      "step": 9391
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.5160818099975586,
+      "learning_rate": 9.609052112420601e-06,
+      "loss": 0.9842,
+      "step": 9392
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.5163795948028564,
+      "learning_rate": 9.608949901292824e-06,
+      "loss": 0.4259,
+      "step": 9393
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.549483060836792,
+      "learning_rate": 9.608847677349284e-06,
+      "loss": 0.7109,
+      "step": 9394
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.597250461578369,
+      "learning_rate": 9.608745440590264e-06,
+      "loss": 0.2759,
+      "step": 9395
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.5434515476226807,
+      "learning_rate": 9.608643191016052e-06,
+      "loss": 0.6441,
+      "step": 9396
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.6520910263061523,
+      "learning_rate": 9.608540928626929e-06,
+      "loss": 0.5314,
+      "step": 9397
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.680987596511841,
+      "learning_rate": 9.608438653423177e-06,
+      "loss": 0.5762,
+      "step": 9398
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.893707752227783,
+      "learning_rate": 9.608336365405086e-06,
+      "loss": 0.8458,
+      "step": 9399
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.1939046382904053,
+      "learning_rate": 9.608234064572939e-06,
+      "loss": 0.6036,
+      "step": 9400
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 4.356733798980713,
+      "learning_rate": 9.608131750927017e-06,
+      "loss": 0.4957,
+      "step": 9401
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.3125839233398438,
+      "learning_rate": 9.608029424467608e-06,
+      "loss": 0.4905,
+      "step": 9402
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.5948922634124756,
+      "learning_rate": 9.607927085194996e-06,
+      "loss": 0.714,
+      "step": 9403
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.5257461071014404,
+      "learning_rate": 9.607824733109464e-06,
+      "loss": 0.5295,
+      "step": 9404
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.7516964673995972,
+      "learning_rate": 9.607722368211298e-06,
+      "loss": 0.8141,
+      "step": 9405
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 5.474801540374756,
+      "learning_rate": 9.607619990500782e-06,
+      "loss": 0.6797,
+      "step": 9406
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.3781425952911377,
+      "learning_rate": 9.607517599978201e-06,
+      "loss": 0.5833,
+      "step": 9407
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.2920992374420166,
+      "learning_rate": 9.607415196643838e-06,
+      "loss": 0.6339,
+      "step": 9408
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.4999213218688965,
+      "learning_rate": 9.607312780497982e-06,
+      "loss": 0.6012,
+      "step": 9409
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.4228787422180176,
+      "learning_rate": 9.607210351540914e-06,
+      "loss": 0.9017,
+      "step": 9410
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.8034703731536865,
+      "learning_rate": 9.607107909772919e-06,
+      "loss": 0.5631,
+      "step": 9411
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.31611704826355,
+      "learning_rate": 9.607005455194284e-06,
+      "loss": 0.5044,
+      "step": 9412
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.504732847213745,
+      "learning_rate": 9.606902987805293e-06,
+      "loss": 0.7018,
+      "step": 9413
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 4.579161643981934,
+      "learning_rate": 9.60680050760623e-06,
+      "loss": 0.8635,
+      "step": 9414
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 4.065528392791748,
+      "learning_rate": 9.60669801459738e-06,
+      "loss": 0.6599,
+      "step": 9415
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.6919770240783691,
+      "learning_rate": 9.606595508779029e-06,
+      "loss": 0.6141,
+      "step": 9416
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.092466115951538,
+      "learning_rate": 9.606492990151461e-06,
+      "loss": 0.6096,
+      "step": 9417
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.5526351928710938,
+      "learning_rate": 9.606390458714961e-06,
+      "loss": 0.6955,
+      "step": 9418
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.6000163555145264,
+      "learning_rate": 9.606287914469817e-06,
+      "loss": 0.6793,
+      "step": 9419
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.0461583137512207,
+      "learning_rate": 9.60618535741631e-06,
+      "loss": 0.6532,
+      "step": 9420
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.827998399734497,
+      "learning_rate": 9.606082787554728e-06,
+      "loss": 0.7938,
+      "step": 9421
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.670050024986267,
+      "learning_rate": 9.605980204885355e-06,
+      "loss": 0.6555,
+      "step": 9422
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.4127371311187744,
+      "learning_rate": 9.605877609408477e-06,
+      "loss": 0.6157,
+      "step": 9423
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.912238836288452,
+      "learning_rate": 9.605775001124377e-06,
+      "loss": 0.3482,
+      "step": 9424
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.7026402950286865,
+      "learning_rate": 9.605672380033343e-06,
+      "loss": 0.78,
+      "step": 9425
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.9359517097473145,
+      "learning_rate": 9.60556974613566e-06,
+      "loss": 0.6869,
+      "step": 9426
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.3750956058502197,
+      "learning_rate": 9.605467099431613e-06,
+      "loss": 0.618,
+      "step": 9427
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.7034037113189697,
+      "learning_rate": 9.605364439921487e-06,
+      "loss": 0.669,
+      "step": 9428
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.198690176010132,
+      "learning_rate": 9.605261767605568e-06,
+      "loss": 0.7395,
+      "step": 9429
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.1068291664123535,
+      "learning_rate": 9.605159082484138e-06,
+      "loss": 0.7291,
+      "step": 9430
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.708838701248169,
+      "learning_rate": 9.60505638455749e-06,
+      "loss": 0.6225,
+      "step": 9431
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.8721771240234375,
+      "learning_rate": 9.604953673825902e-06,
+      "loss": 0.7835,
+      "step": 9432
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.5209500789642334,
+      "learning_rate": 9.604850950289665e-06,
+      "loss": 0.6854,
+      "step": 9433
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.146364688873291,
+      "learning_rate": 9.604748213949061e-06,
+      "loss": 0.6368,
+      "step": 9434
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.735365867614746,
+      "learning_rate": 9.60464546480438e-06,
+      "loss": 0.7317,
+      "step": 9435
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.3995823860168457,
+      "learning_rate": 9.604542702855904e-06,
+      "loss": 0.5611,
+      "step": 9436
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.4938321113586426,
+      "learning_rate": 9.604439928103918e-06,
+      "loss": 0.9457,
+      "step": 9437
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.518712043762207,
+      "learning_rate": 9.604337140548709e-06,
+      "loss": 0.7561,
+      "step": 9438
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.7216696739196777,
+      "learning_rate": 9.604234340190566e-06,
+      "loss": 0.6691,
+      "step": 9439
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.4954135417938232,
+      "learning_rate": 9.60413152702977e-06,
+      "loss": 0.6552,
+      "step": 9440
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.8236114978790283,
+      "learning_rate": 9.60402870106661e-06,
+      "loss": 0.835,
+      "step": 9441
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.6898791790008545,
+      "learning_rate": 9.60392586230137e-06,
+      "loss": 0.8433,
+      "step": 9442
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.0503857135772705,
+      "learning_rate": 9.603823010734337e-06,
+      "loss": 0.5781,
+      "step": 9443
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.3325117826461792,
+      "learning_rate": 9.603720146365798e-06,
+      "loss": 0.4811,
+      "step": 9444
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.419990062713623,
+      "learning_rate": 9.603617269196036e-06,
+      "loss": 0.6668,
+      "step": 9445
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.2151882648468018,
+      "learning_rate": 9.60351437922534e-06,
+      "loss": 0.7609,
+      "step": 9446
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.3392271995544434,
+      "learning_rate": 9.603411476453994e-06,
+      "loss": 0.6821,
+      "step": 9447
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.6951136589050293,
+      "learning_rate": 9.603308560882287e-06,
+      "loss": 0.6704,
+      "step": 9448
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.336097478866577,
+      "learning_rate": 9.603205632510502e-06,
+      "loss": 0.6951,
+      "step": 9449
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.1231796741485596,
+      "learning_rate": 9.603102691338929e-06,
+      "loss": 0.4664,
+      "step": 9450
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.9334241151809692,
+      "learning_rate": 9.602999737367849e-06,
+      "loss": 0.4311,
+      "step": 9451
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.9788320064544678,
+      "learning_rate": 9.602896770597551e-06,
+      "loss": 0.9252,
+      "step": 9452
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.770284414291382,
+      "learning_rate": 9.602793791028321e-06,
+      "loss": 0.5185,
+      "step": 9453
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.871503472328186,
+      "learning_rate": 9.602690798660447e-06,
+      "loss": 0.5056,
+      "step": 9454
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.9777652025222778,
+      "learning_rate": 9.602587793494215e-06,
+      "loss": 0.7651,
+      "step": 9455
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.0555098056793213,
+      "learning_rate": 9.602484775529908e-06,
+      "loss": 0.6876,
+      "step": 9456
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.0513548851013184,
+      "learning_rate": 9.602381744767816e-06,
+      "loss": 0.5328,
+      "step": 9457
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.930782675743103,
+      "learning_rate": 9.602278701208227e-06,
+      "loss": 0.7812,
+      "step": 9458
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.300604820251465,
+      "learning_rate": 9.602175644851422e-06,
+      "loss": 0.7441,
+      "step": 9459
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.8348537683486938,
+      "learning_rate": 9.602072575697692e-06,
+      "loss": 0.7968,
+      "step": 9460
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.4969003200531006,
+      "learning_rate": 9.60196949374732e-06,
+      "loss": 0.6592,
+      "step": 9461
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.481107473373413,
+      "learning_rate": 9.601866399000598e-06,
+      "loss": 0.757,
+      "step": 9462
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.2859091758728027,
+      "learning_rate": 9.601763291457807e-06,
+      "loss": 0.6265,
+      "step": 9463
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.192394971847534,
+      "learning_rate": 9.601660171119238e-06,
+      "loss": 0.5943,
+      "step": 9464
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.418635606765747,
+      "learning_rate": 9.601557037985174e-06,
+      "loss": 0.6653,
+      "step": 9465
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.735344171524048,
+      "learning_rate": 9.601453892055905e-06,
+      "loss": 0.6593,
+      "step": 9466
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.9261598587036133,
+      "learning_rate": 9.601350733331716e-06,
+      "loss": 0.4957,
+      "step": 9467
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.9788124561309814,
+      "learning_rate": 9.601247561812894e-06,
+      "loss": 0.6074,
+      "step": 9468
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.8317711353302002,
+      "learning_rate": 9.601144377499727e-06,
+      "loss": 0.5775,
+      "step": 9469
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.1468007564544678,
+      "learning_rate": 9.601041180392499e-06,
+      "loss": 0.5434,
+      "step": 9470
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.318164587020874,
+      "learning_rate": 9.600937970491502e-06,
+      "loss": 0.775,
+      "step": 9471
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.7335093021392822,
+      "learning_rate": 9.60083474779702e-06,
+      "loss": 0.619,
+      "step": 9472
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.2109498977661133,
+      "learning_rate": 9.600731512309337e-06,
+      "loss": 0.7657,
+      "step": 9473
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.759622097015381,
+      "learning_rate": 9.600628264028744e-06,
+      "loss": 0.5644,
+      "step": 9474
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.0773842334747314,
+      "learning_rate": 9.600525002955529e-06,
+      "loss": 0.6646,
+      "step": 9475
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.1223297119140625,
+      "learning_rate": 9.600421729089977e-06,
+      "loss": 0.6359,
+      "step": 9476
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.1133484840393066,
+      "learning_rate": 9.600318442432376e-06,
+      "loss": 0.7022,
+      "step": 9477
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.030351400375366,
+      "learning_rate": 9.60021514298301e-06,
+      "loss": 0.781,
+      "step": 9478
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.524613857269287,
+      "learning_rate": 9.60011183074217e-06,
+      "loss": 0.5471,
+      "step": 9479
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 4.543669700622559,
+      "learning_rate": 9.600008505710143e-06,
+      "loss": 0.6323,
+      "step": 9480
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.9950777292251587,
+      "learning_rate": 9.599905167887215e-06,
+      "loss": 0.7271,
+      "step": 9481
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.181157350540161,
+      "learning_rate": 9.599801817273675e-06,
+      "loss": 0.2172,
+      "step": 9482
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.2736833095550537,
+      "learning_rate": 9.59969845386981e-06,
+      "loss": 0.7504,
+      "step": 9483
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.9293527603149414,
+      "learning_rate": 9.599595077675905e-06,
+      "loss": 0.452,
+      "step": 9484
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.530679702758789,
+      "learning_rate": 9.59949168869225e-06,
+      "loss": 0.8073,
+      "step": 9485
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.086515426635742,
+      "learning_rate": 9.59938828691913e-06,
+      "loss": 0.8008,
+      "step": 9486
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.4752376079559326,
+      "learning_rate": 9.599284872356836e-06,
+      "loss": 0.6214,
+      "step": 9487
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.141432285308838,
+      "learning_rate": 9.599181445005651e-06,
+      "loss": 0.725,
+      "step": 9488
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.0200626850128174,
+      "learning_rate": 9.59907800486587e-06,
+      "loss": 0.7675,
+      "step": 9489
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 4.7672438621521,
+      "learning_rate": 9.598974551937773e-06,
+      "loss": 0.5337,
+      "step": 9490
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 4.158195972442627,
+      "learning_rate": 9.59887108622165e-06,
+      "loss": 0.5408,
+      "step": 9491
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 4.759908199310303,
+      "learning_rate": 9.59876760771779e-06,
+      "loss": 0.7264,
+      "step": 9492
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.5824991464614868,
+      "learning_rate": 9.59866411642648e-06,
+      "loss": 0.5547,
+      "step": 9493
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.7052903175354004,
+      "learning_rate": 9.598560612348007e-06,
+      "loss": 0.8394,
+      "step": 9494
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.061932325363159,
+      "learning_rate": 9.59845709548266e-06,
+      "loss": 0.7147,
+      "step": 9495
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.6381187438964844,
+      "learning_rate": 9.598353565830729e-06,
+      "loss": 0.7191,
+      "step": 9496
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.7694512605667114,
+      "learning_rate": 9.598250023392496e-06,
+      "loss": 0.8021,
+      "step": 9497
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.700073003768921,
+      "learning_rate": 9.598146468168254e-06,
+      "loss": 0.7796,
+      "step": 9498
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.9474902153015137,
+      "learning_rate": 9.598042900158288e-06,
+      "loss": 0.6536,
+      "step": 9499
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.052617311477661,
+      "learning_rate": 9.597939319362888e-06,
+      "loss": 0.7018,
+      "step": 9500
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.1826775074005127,
+      "learning_rate": 9.597835725782341e-06,
+      "loss": 0.6665,
+      "step": 9501
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.568601369857788,
+      "learning_rate": 9.597732119416936e-06,
+      "loss": 0.8269,
+      "step": 9502
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 4.40510368347168,
+      "learning_rate": 9.597628500266958e-06,
+      "loss": 0.9358,
+      "step": 9503
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.7574055194854736,
+      "learning_rate": 9.5975248683327e-06,
+      "loss": 0.7007,
+      "step": 9504
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.105207920074463,
+      "learning_rate": 9.597421223614446e-06,
+      "loss": 0.6159,
+      "step": 9505
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.8496932983398438,
+      "learning_rate": 9.597317566112487e-06,
+      "loss": 0.6934,
+      "step": 9506
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.7635927200317383,
+      "learning_rate": 9.59721389582711e-06,
+      "loss": 0.6302,
+      "step": 9507
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.291236639022827,
+      "learning_rate": 9.597110212758602e-06,
+      "loss": 0.5066,
+      "step": 9508
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.176281690597534,
+      "learning_rate": 9.597006516907253e-06,
+      "loss": 0.5364,
+      "step": 9509
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.005006790161133,
+      "learning_rate": 9.596902808273351e-06,
+      "loss": 0.8216,
+      "step": 9510
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.9274601936340332,
+      "learning_rate": 9.596799086857184e-06,
+      "loss": 0.5879,
+      "step": 9511
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.598238945007324,
+      "learning_rate": 9.596695352659042e-06,
+      "loss": 0.639,
+      "step": 9512
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.4584171772003174,
+      "learning_rate": 9.596591605679213e-06,
+      "loss": 0.4336,
+      "step": 9513
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.280146598815918,
+      "learning_rate": 9.596487845917984e-06,
+      "loss": 0.7184,
+      "step": 9514
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.9547243118286133,
+      "learning_rate": 9.596384073375642e-06,
+      "loss": 0.9409,
+      "step": 9515
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 4.668095588684082,
+      "learning_rate": 9.59628028805248e-06,
+      "loss": 0.522,
+      "step": 9516
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 6.59360933303833,
+      "learning_rate": 9.596176489948782e-06,
+      "loss": 0.5504,
+      "step": 9517
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.8930540084838867,
+      "learning_rate": 9.596072679064841e-06,
+      "loss": 0.5855,
+      "step": 9518
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.5747604370117188,
+      "learning_rate": 9.595968855400943e-06,
+      "loss": 0.6587,
+      "step": 9519
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.290070056915283,
+      "learning_rate": 9.595865018957378e-06,
+      "loss": 0.5605,
+      "step": 9520
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.3601202964782715,
+      "learning_rate": 9.595761169734434e-06,
+      "loss": 0.6412,
+      "step": 9521
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.046886444091797,
+      "learning_rate": 9.595657307732399e-06,
+      "loss": 0.7567,
+      "step": 9522
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 6.977245330810547,
+      "learning_rate": 9.595553432951562e-06,
+      "loss": 0.8818,
+      "step": 9523
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.0487074851989746,
+      "learning_rate": 9.595449545392213e-06,
+      "loss": 0.7125,
+      "step": 9524
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.9202971458435059,
+      "learning_rate": 9.595345645054642e-06,
+      "loss": 0.8278,
+      "step": 9525
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.8270716667175293,
+      "learning_rate": 9.595241731939133e-06,
+      "loss": 0.8078,
+      "step": 9526
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 5.294354438781738,
+      "learning_rate": 9.59513780604598e-06,
+      "loss": 0.6542,
+      "step": 9527
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.8569673299789429,
+      "learning_rate": 9.595033867375469e-06,
+      "loss": 0.6396,
+      "step": 9528
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 6.708364963531494,
+      "learning_rate": 9.594929915927893e-06,
+      "loss": 0.5193,
+      "step": 9529
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.0373141765594482,
+      "learning_rate": 9.594825951703536e-06,
+      "loss": 0.7111,
+      "step": 9530
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.155618906021118,
+      "learning_rate": 9.59472197470269e-06,
+      "loss": 0.7563,
+      "step": 9531
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.563389301300049,
+      "learning_rate": 9.594617984925643e-06,
+      "loss": 0.7952,
+      "step": 9532
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.6556735038757324,
+      "learning_rate": 9.594513982372685e-06,
+      "loss": 0.541,
+      "step": 9533
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.373140573501587,
+      "learning_rate": 9.594409967044104e-06,
+      "loss": 0.5671,
+      "step": 9534
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.401442527770996,
+      "learning_rate": 9.59430593894019e-06,
+      "loss": 0.4551,
+      "step": 9535
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.7793748378753662,
+      "learning_rate": 9.594201898061232e-06,
+      "loss": 0.7142,
+      "step": 9536
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.3604705333709717,
+      "learning_rate": 9.594097844407522e-06,
+      "loss": 0.8815,
+      "step": 9537
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.42742919921875,
+      "learning_rate": 9.593993777979344e-06,
+      "loss": 0.7246,
+      "step": 9538
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.9855430126190186,
+      "learning_rate": 9.593889698776992e-06,
+      "loss": 0.661,
+      "step": 9539
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.6064469814300537,
+      "learning_rate": 9.593785606800751e-06,
+      "loss": 0.809,
+      "step": 9540
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.8741118907928467,
+      "learning_rate": 9.593681502050916e-06,
+      "loss": 0.4529,
+      "step": 9541
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.8695939779281616,
+      "learning_rate": 9.593577384527774e-06,
+      "loss": 0.7968,
+      "step": 9542
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.0311567783355713,
+      "learning_rate": 9.593473254231613e-06,
+      "loss": 0.8349,
+      "step": 9543
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.8234100341796875,
+      "learning_rate": 9.593369111162722e-06,
+      "loss": 0.6815,
+      "step": 9544
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.4705772399902344,
+      "learning_rate": 9.593264955321395e-06,
+      "loss": 0.7376,
+      "step": 9545
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 4.490925312042236,
+      "learning_rate": 9.593160786707917e-06,
+      "loss": 0.7501,
+      "step": 9546
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.1183555126190186,
+      "learning_rate": 9.59305660532258e-06,
+      "loss": 0.6786,
+      "step": 9547
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.1969258785247803,
+      "learning_rate": 9.592952411165673e-06,
+      "loss": 0.543,
+      "step": 9548
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.3537094593048096,
+      "learning_rate": 9.592848204237486e-06,
+      "loss": 0.6926,
+      "step": 9549
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.9952051639556885,
+      "learning_rate": 9.592743984538309e-06,
+      "loss": 0.7969,
+      "step": 9550
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.15545654296875,
+      "learning_rate": 9.592639752068431e-06,
+      "loss": 0.7,
+      "step": 9551
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.5274910926818848,
+      "learning_rate": 9.592535506828141e-06,
+      "loss": 0.8377,
+      "step": 9552
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.5584094524383545,
+      "learning_rate": 9.592431248817732e-06,
+      "loss": 0.9297,
+      "step": 9553
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.7697417736053467,
+      "learning_rate": 9.592326978037491e-06,
+      "loss": 0.6022,
+      "step": 9554
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.375220537185669,
+      "learning_rate": 9.592222694487712e-06,
+      "loss": 0.6987,
+      "step": 9555
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.6963284015655518,
+      "learning_rate": 9.592118398168679e-06,
+      "loss": 0.674,
+      "step": 9556
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.8780174255371094,
+      "learning_rate": 9.592014089080685e-06,
+      "loss": 0.9514,
+      "step": 9557
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.6337575912475586,
+      "learning_rate": 9.59190976722402e-06,
+      "loss": 0.4304,
+      "step": 9558
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.5554091930389404,
+      "learning_rate": 9.591805432598975e-06,
+      "loss": 0.7282,
+      "step": 9559
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.998399257659912,
+      "learning_rate": 9.59170108520584e-06,
+      "loss": 0.5364,
+      "step": 9560
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.4402291774749756,
+      "learning_rate": 9.591596725044903e-06,
+      "loss": 0.8046,
+      "step": 9561
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.5249561071395874,
+      "learning_rate": 9.591492352116456e-06,
+      "loss": 0.8376,
+      "step": 9562
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.019813060760498,
+      "learning_rate": 9.591387966420788e-06,
+      "loss": 0.9458,
+      "step": 9563
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.9564876556396484,
+      "learning_rate": 9.591283567958191e-06,
+      "loss": 0.4897,
+      "step": 9564
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 4.848633289337158,
+      "learning_rate": 9.591179156728954e-06,
+      "loss": 0.8629,
+      "step": 9565
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.9835829734802246,
+      "learning_rate": 9.591074732733367e-06,
+      "loss": 0.8151,
+      "step": 9566
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.082477569580078,
+      "learning_rate": 9.590970295971723e-06,
+      "loss": 0.6896,
+      "step": 9567
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.8209786415100098,
+      "learning_rate": 9.590865846444308e-06,
+      "loss": 0.8054,
+      "step": 9568
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 4.677137851715088,
+      "learning_rate": 9.590761384151417e-06,
+      "loss": 0.7536,
+      "step": 9569
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.876169443130493,
+      "learning_rate": 9.590656909093336e-06,
+      "loss": 0.6012,
+      "step": 9570
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.6798017024993896,
+      "learning_rate": 9.59055242127036e-06,
+      "loss": 0.6528,
+      "step": 9571
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.349917411804199,
+      "learning_rate": 9.590447920682778e-06,
+      "loss": 0.547,
+      "step": 9572
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.985037088394165,
+      "learning_rate": 9.590343407330878e-06,
+      "loss": 0.6667,
+      "step": 9573
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.709815263748169,
+      "learning_rate": 9.590238881214952e-06,
+      "loss": 0.5439,
+      "step": 9574
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.585406541824341,
+      "learning_rate": 9.590134342335294e-06,
+      "loss": 0.6331,
+      "step": 9575
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.7988953590393066,
+      "learning_rate": 9.590029790692189e-06,
+      "loss": 0.4709,
+      "step": 9576
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.175602912902832,
+      "learning_rate": 9.589925226285932e-06,
+      "loss": 0.4866,
+      "step": 9577
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.9703128337860107,
+      "learning_rate": 9.589820649116812e-06,
+      "loss": 0.6922,
+      "step": 9578
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.6335346698760986,
+      "learning_rate": 9.58971605918512e-06,
+      "loss": 0.4221,
+      "step": 9579
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.520473003387451,
+      "learning_rate": 9.589611456491148e-06,
+      "loss": 0.7354,
+      "step": 9580
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.2813875675201416,
+      "learning_rate": 9.589506841035185e-06,
+      "loss": 0.8901,
+      "step": 9581
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.1263926029205322,
+      "learning_rate": 9.589402212817523e-06,
+      "loss": 0.9525,
+      "step": 9582
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 4.629757881164551,
+      "learning_rate": 9.589297571838451e-06,
+      "loss": 0.5192,
+      "step": 9583
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.8351002931594849,
+      "learning_rate": 9.589192918098264e-06,
+      "loss": 0.7724,
+      "step": 9584
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.4773027896881104,
+      "learning_rate": 9.58908825159725e-06,
+      "loss": 0.6438,
+      "step": 9585
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.3997676372528076,
+      "learning_rate": 9.588983572335701e-06,
+      "loss": 0.5966,
+      "step": 9586
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.8673087358474731,
+      "learning_rate": 9.588878880313907e-06,
+      "loss": 0.5215,
+      "step": 9587
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.3438868522644043,
+      "learning_rate": 9.58877417553216e-06,
+      "loss": 0.8736,
+      "step": 9588
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 6.302083492279053,
+      "learning_rate": 9.58866945799075e-06,
+      "loss": 0.5751,
+      "step": 9589
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.2491700649261475,
+      "learning_rate": 9.588564727689972e-06,
+      "loss": 0.7522,
+      "step": 9590
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.9761745929718018,
+      "learning_rate": 9.588459984630113e-06,
+      "loss": 0.8789,
+      "step": 9591
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.28947114944458,
+      "learning_rate": 9.588355228811466e-06,
+      "loss": 0.7655,
+      "step": 9592
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 4.363162994384766,
+      "learning_rate": 9.588250460234321e-06,
+      "loss": 0.5758,
+      "step": 9593
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.1847779750823975,
+      "learning_rate": 9.588145678898971e-06,
+      "loss": 0.5971,
+      "step": 9594
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.31479811668396,
+      "learning_rate": 9.588040884805705e-06,
+      "loss": 0.7156,
+      "step": 9595
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.4511096477508545,
+      "learning_rate": 9.587936077954818e-06,
+      "loss": 0.8843,
+      "step": 9596
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.215172052383423,
+      "learning_rate": 9.587831258346599e-06,
+      "loss": 0.4955,
+      "step": 9597
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.0874249935150146,
+      "learning_rate": 9.58772642598134e-06,
+      "loss": 0.6596,
+      "step": 9598
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 4.131960391998291,
+      "learning_rate": 9.587621580859332e-06,
+      "loss": 0.8323,
+      "step": 9599
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.279832363128662,
+      "learning_rate": 9.587516722980867e-06,
+      "loss": 0.5668,
+      "step": 9600
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.729313611984253,
+      "learning_rate": 9.587411852346237e-06,
+      "loss": 0.7723,
+      "step": 9601
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.242436408996582,
+      "learning_rate": 9.587306968955732e-06,
+      "loss": 0.6007,
+      "step": 9602
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.202531099319458,
+      "learning_rate": 9.587202072809645e-06,
+      "loss": 0.8878,
+      "step": 9603
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.0886964797973633,
+      "learning_rate": 9.587097163908268e-06,
+      "loss": 0.6208,
+      "step": 9604
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.262948751449585,
+      "learning_rate": 9.586992242251892e-06,
+      "loss": 0.7298,
+      "step": 9605
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.9251415729522705,
+      "learning_rate": 9.586887307840808e-06,
+      "loss": 0.7081,
+      "step": 9606
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 4.180373668670654,
+      "learning_rate": 9.586782360675311e-06,
+      "loss": 0.7113,
+      "step": 9607
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.268801689147949,
+      "learning_rate": 9.586677400755688e-06,
+      "loss": 0.5288,
+      "step": 9608
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.4050862789154053,
+      "learning_rate": 9.586572428082236e-06,
+      "loss": 0.4549,
+      "step": 9609
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.8432615995407104,
+      "learning_rate": 9.586467442655243e-06,
+      "loss": 0.7666,
+      "step": 9610
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.2827093601226807,
+      "learning_rate": 9.586362444475e-06,
+      "loss": 0.7559,
+      "step": 9611
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.235532760620117,
+      "learning_rate": 9.586257433541804e-06,
+      "loss": 0.6221,
+      "step": 9612
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.3756866455078125,
+      "learning_rate": 9.586152409855943e-06,
+      "loss": 0.4962,
+      "step": 9613
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.496027946472168,
+      "learning_rate": 9.58604737341771e-06,
+      "loss": 0.7238,
+      "step": 9614
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.1965084075927734,
+      "learning_rate": 9.5859423242274e-06,
+      "loss": 0.6509,
+      "step": 9615
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.1834940910339355,
+      "learning_rate": 9.585837262285297e-06,
+      "loss": 0.5165,
+      "step": 9616
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.812258005142212,
+      "learning_rate": 9.585732187591704e-06,
+      "loss": 0.6629,
+      "step": 9617
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.334733247756958,
+      "learning_rate": 9.585627100146903e-06,
+      "loss": 0.6398,
+      "step": 9618
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.5247209072113037,
+      "learning_rate": 9.585521999951192e-06,
+      "loss": 0.6271,
+      "step": 9619
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.19236159324646,
+      "learning_rate": 9.585416887004864e-06,
+      "loss": 0.5685,
+      "step": 9620
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.417945146560669,
+      "learning_rate": 9.585311761308208e-06,
+      "loss": 0.511,
+      "step": 9621
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.1699790954589844,
+      "learning_rate": 9.585206622861518e-06,
+      "loss": 0.6612,
+      "step": 9622
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.050297737121582,
+      "learning_rate": 9.585101471665088e-06,
+      "loss": 0.8403,
+      "step": 9623
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.1701853275299072,
+      "learning_rate": 9.584996307719205e-06,
+      "loss": 0.5586,
+      "step": 9624
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.9779152870178223,
+      "learning_rate": 9.584891131024166e-06,
+      "loss": 0.6624,
+      "step": 9625
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.8500685691833496,
+      "learning_rate": 9.584785941580263e-06,
+      "loss": 0.4933,
+      "step": 9626
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.1180624961853027,
+      "learning_rate": 9.584680739387789e-06,
+      "loss": 0.5962,
+      "step": 9627
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 7.648984909057617,
+      "learning_rate": 9.584575524447032e-06,
+      "loss": 0.4956,
+      "step": 9628
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.194587469100952,
+      "learning_rate": 9.58447029675829e-06,
+      "loss": 0.8324,
+      "step": 9629
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.3266499042510986,
+      "learning_rate": 9.584365056321853e-06,
+      "loss": 0.5193,
+      "step": 9630
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.265183687210083,
+      "learning_rate": 9.584259803138015e-06,
+      "loss": 0.9145,
+      "step": 9631
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.350332260131836,
+      "learning_rate": 9.584154537207067e-06,
+      "loss": 0.9711,
+      "step": 9632
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.229005813598633,
+      "learning_rate": 9.584049258529301e-06,
+      "loss": 0.4079,
+      "step": 9633
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.8959556818008423,
+      "learning_rate": 9.583943967105013e-06,
+      "loss": 0.6872,
+      "step": 9634
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.1622588634490967,
+      "learning_rate": 9.583838662934495e-06,
+      "loss": 0.7134,
+      "step": 9635
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.698256015777588,
+      "learning_rate": 9.583733346018037e-06,
+      "loss": 0.4925,
+      "step": 9636
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.0449776649475098,
+      "learning_rate": 9.583628016355934e-06,
+      "loss": 0.6351,
+      "step": 9637
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.0785112380981445,
+      "learning_rate": 9.583522673948479e-06,
+      "loss": 0.7083,
+      "step": 9638
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.935166120529175,
+      "learning_rate": 9.583417318795964e-06,
+      "loss": 0.6021,
+      "step": 9639
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.9967041015625,
+      "learning_rate": 9.583311950898682e-06,
+      "loss": 1.008,
+      "step": 9640
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.708435535430908,
+      "learning_rate": 9.583206570256929e-06,
+      "loss": 0.612,
+      "step": 9641
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.458012104034424,
+      "learning_rate": 9.583101176870992e-06,
+      "loss": 0.6821,
+      "step": 9642
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.88608717918396,
+      "learning_rate": 9.58299577074117e-06,
+      "loss": 0.7319,
+      "step": 9643
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.4215126037597656,
+      "learning_rate": 9.582890351867753e-06,
+      "loss": 0.6997,
+      "step": 9644
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.773097038269043,
+      "learning_rate": 9.582784920251033e-06,
+      "loss": 0.6618,
+      "step": 9645
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 5.994002819061279,
+      "learning_rate": 9.582679475891307e-06,
+      "loss": 0.5711,
+      "step": 9646
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.9921594858169556,
+      "learning_rate": 9.582574018788865e-06,
+      "loss": 0.7431,
+      "step": 9647
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.8054683208465576,
+      "learning_rate": 9.582468548944002e-06,
+      "loss": 0.6593,
+      "step": 9648
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.4314754009246826,
+      "learning_rate": 9.58236306635701e-06,
+      "loss": 0.8049,
+      "step": 9649
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.997250556945801,
+      "learning_rate": 9.582257571028183e-06,
+      "loss": 0.5509,
+      "step": 9650
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.2827024459838867,
+      "learning_rate": 9.582152062957814e-06,
+      "loss": 0.669,
+      "step": 9651
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.5860257148742676,
+      "learning_rate": 9.582046542146197e-06,
+      "loss": 0.7768,
+      "step": 9652
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.241830348968506,
+      "learning_rate": 9.581941008593626e-06,
+      "loss": 0.6134,
+      "step": 9653
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.022212505340576,
+      "learning_rate": 9.581835462300392e-06,
+      "loss": 0.4647,
+      "step": 9654
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.5933048725128174,
+      "learning_rate": 9.581729903266791e-06,
+      "loss": 0.9135,
+      "step": 9655
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.9705746173858643,
+      "learning_rate": 9.581624331493116e-06,
+      "loss": 0.6569,
+      "step": 9656
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.4067039489746094,
+      "learning_rate": 9.58151874697966e-06,
+      "loss": 0.3707,
+      "step": 9657
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.4051768779754639,
+      "learning_rate": 9.581413149726714e-06,
+      "loss": 0.5468,
+      "step": 9658
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.3578126430511475,
+      "learning_rate": 9.581307539734576e-06,
+      "loss": 0.8907,
+      "step": 9659
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.8138067722320557,
+      "learning_rate": 9.581201917003537e-06,
+      "loss": 0.818,
+      "step": 9660
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.702528715133667,
+      "learning_rate": 9.581096281533893e-06,
+      "loss": 0.7202,
+      "step": 9661
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 4.379930019378662,
+      "learning_rate": 9.580990633325936e-06,
+      "loss": 0.6871,
+      "step": 9662
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.262882709503174,
+      "learning_rate": 9.58088497237996e-06,
+      "loss": 0.6208,
+      "step": 9663
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.5128889083862305,
+      "learning_rate": 9.58077929869626e-06,
+      "loss": 0.4118,
+      "step": 9664
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 4.814243793487549,
+      "learning_rate": 9.580673612275127e-06,
+      "loss": 0.6286,
+      "step": 9665
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.998807430267334,
+      "learning_rate": 9.580567913116858e-06,
+      "loss": 0.7744,
+      "step": 9666
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.4152703285217285,
+      "learning_rate": 9.580462201221743e-06,
+      "loss": 0.7579,
+      "step": 9667
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.969394326210022,
+      "learning_rate": 9.580356476590083e-06,
+      "loss": 0.8876,
+      "step": 9668
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.7667940855026245,
+      "learning_rate": 9.580250739222164e-06,
+      "loss": 0.8693,
+      "step": 9669
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.822360873222351,
+      "learning_rate": 9.580144989118284e-06,
+      "loss": 0.7551,
+      "step": 9670
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.3782849311828613,
+      "learning_rate": 9.580039226278736e-06,
+      "loss": 0.5877,
+      "step": 9671
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.3315417766571045,
+      "learning_rate": 9.579933450703817e-06,
+      "loss": 0.7082,
+      "step": 9672
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.571854829788208,
+      "learning_rate": 9.579827662393817e-06,
+      "loss": 0.7187,
+      "step": 9673
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.783111095428467,
+      "learning_rate": 9.579721861349032e-06,
+      "loss": 0.9493,
+      "step": 9674
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.8124969005584717,
+      "learning_rate": 9.579616047569756e-06,
+      "loss": 0.5978,
+      "step": 9675
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 5.118401527404785,
+      "learning_rate": 9.579510221056285e-06,
+      "loss": 0.881,
+      "step": 9676
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.466937780380249,
+      "learning_rate": 9.57940438180891e-06,
+      "loss": 0.7281,
+      "step": 9677
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.570122241973877,
+      "learning_rate": 9.579298529827927e-06,
+      "loss": 0.6491,
+      "step": 9678
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.0660202503204346,
+      "learning_rate": 9.57919266511363e-06,
+      "loss": 0.7571,
+      "step": 9679
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.331512928009033,
+      "learning_rate": 9.579086787666313e-06,
+      "loss": 0.666,
+      "step": 9680
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.3845529556274414,
+      "learning_rate": 9.578980897486273e-06,
+      "loss": 0.8385,
+      "step": 9681
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.553006172180176,
+      "learning_rate": 9.578874994573801e-06,
+      "loss": 0.6307,
+      "step": 9682
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.7666819095611572,
+      "learning_rate": 9.578769078929194e-06,
+      "loss": 0.93,
+      "step": 9683
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.3358588218688965,
+      "learning_rate": 9.578663150552745e-06,
+      "loss": 0.5822,
+      "step": 9684
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.1728146076202393,
+      "learning_rate": 9.578557209444748e-06,
+      "loss": 0.4739,
+      "step": 9685
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.768612861633301,
+      "learning_rate": 9.578451255605501e-06,
+      "loss": 0.5951,
+      "step": 9686
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.447622299194336,
+      "learning_rate": 9.578345289035293e-06,
+      "loss": 0.8967,
+      "step": 9687
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 5.0652384757995605,
+      "learning_rate": 9.578239309734425e-06,
+      "loss": 0.5721,
+      "step": 9688
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.9109249114990234,
+      "learning_rate": 9.578133317703186e-06,
+      "loss": 0.5911,
+      "step": 9689
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.2534193992614746,
+      "learning_rate": 9.578027312941875e-06,
+      "loss": 0.7341,
+      "step": 9690
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.698972702026367,
+      "learning_rate": 9.577921295450783e-06,
+      "loss": 0.6192,
+      "step": 9691
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.83575701713562,
+      "learning_rate": 9.577815265230208e-06,
+      "loss": 0.7334,
+      "step": 9692
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.6083409786224365,
+      "learning_rate": 9.577709222280445e-06,
+      "loss": 0.4928,
+      "step": 9693
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.8471341133117676,
+      "learning_rate": 9.577603166601785e-06,
+      "loss": 0.688,
+      "step": 9694
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.519015073776245,
+      "learning_rate": 9.577497098194527e-06,
+      "loss": 0.9088,
+      "step": 9695
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.1353447437286377,
+      "learning_rate": 9.577391017058964e-06,
+      "loss": 0.6116,
+      "step": 9696
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.5347161293029785,
+      "learning_rate": 9.577284923195392e-06,
+      "loss": 0.857,
+      "step": 9697
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.7601053714752197,
+      "learning_rate": 9.577178816604104e-06,
+      "loss": 0.8548,
+      "step": 9698
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.6335721015930176,
+      "learning_rate": 9.577072697285396e-06,
+      "loss": 0.6481,
+      "step": 9699
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.0393013954162598,
+      "learning_rate": 9.576966565239565e-06,
+      "loss": 0.6269,
+      "step": 9700
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 4.09230375289917,
+      "learning_rate": 9.576860420466903e-06,
+      "loss": 0.4768,
+      "step": 9701
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.672195315361023,
+      "learning_rate": 9.57675426296771e-06,
+      "loss": 0.6195,
+      "step": 9702
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.2016637325286865,
+      "learning_rate": 9.576648092742273e-06,
+      "loss": 0.4043,
+      "step": 9703
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 4.637990474700928,
+      "learning_rate": 9.576541909790895e-06,
+      "loss": 0.8386,
+      "step": 9704
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.4718586206436157,
+      "learning_rate": 9.576435714113867e-06,
+      "loss": 0.6239,
+      "step": 9705
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 4.5899977684021,
+      "learning_rate": 9.576329505711488e-06,
+      "loss": 0.5504,
+      "step": 9706
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.215564489364624,
+      "learning_rate": 9.57622328458405e-06,
+      "loss": 0.884,
+      "step": 9707
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.8571231365203857,
+      "learning_rate": 9.576117050731849e-06,
+      "loss": 0.5344,
+      "step": 9708
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.825881004333496,
+      "learning_rate": 9.57601080415518e-06,
+      "loss": 0.6972,
+      "step": 9709
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 5.654539108276367,
+      "learning_rate": 9.57590454485434e-06,
+      "loss": 0.6087,
+      "step": 9710
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.7283196449279785,
+      "learning_rate": 9.575798272829624e-06,
+      "loss": 0.8611,
+      "step": 9711
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 4.5550432205200195,
+      "learning_rate": 9.575691988081325e-06,
+      "loss": 0.8857,
+      "step": 9712
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.13666033744812,
+      "learning_rate": 9.575585690609744e-06,
+      "loss": 0.7741,
+      "step": 9713
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 4.807835102081299,
+      "learning_rate": 9.575479380415171e-06,
+      "loss": 0.5074,
+      "step": 9714
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.9859981536865234,
+      "learning_rate": 9.575373057497904e-06,
+      "loss": 0.79,
+      "step": 9715
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.3796420097351074,
+      "learning_rate": 9.575266721858239e-06,
+      "loss": 0.6249,
+      "step": 9716
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.8308537006378174,
+      "learning_rate": 9.575160373496471e-06,
+      "loss": 0.5822,
+      "step": 9717
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.6126034259796143,
+      "learning_rate": 9.575054012412897e-06,
+      "loss": 0.9266,
+      "step": 9718
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.8930916786193848,
+      "learning_rate": 9.57494763860781e-06,
+      "loss": 0.7526,
+      "step": 9719
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.6120526790618896,
+      "learning_rate": 9.574841252081509e-06,
+      "loss": 0.6645,
+      "step": 9720
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.2434568405151367,
+      "learning_rate": 9.574734852834286e-06,
+      "loss": 0.4461,
+      "step": 9721
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.5701935291290283,
+      "learning_rate": 9.57462844086644e-06,
+      "loss": 0.7232,
+      "step": 9722
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.9702439308166504,
+      "learning_rate": 9.574522016178266e-06,
+      "loss": 0.9316,
+      "step": 9723
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.2930991649627686,
+      "learning_rate": 9.574415578770061e-06,
+      "loss": 0.5852,
+      "step": 9724
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.8400425910949707,
+      "learning_rate": 9.574309128642118e-06,
+      "loss": 0.8273,
+      "step": 9725
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.736785650253296,
+      "learning_rate": 9.574202665794737e-06,
+      "loss": 0.8352,
+      "step": 9726
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.025648593902588,
+      "learning_rate": 9.57409619022821e-06,
+      "loss": 0.5992,
+      "step": 9727
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.6940979957580566,
+      "learning_rate": 9.573989701942836e-06,
+      "loss": 0.7201,
+      "step": 9728
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 4.910181522369385,
+      "learning_rate": 9.573883200938912e-06,
+      "loss": 1.0488,
+      "step": 9729
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.42167067527771,
+      "learning_rate": 9.57377668721673e-06,
+      "loss": 0.7585,
+      "step": 9730
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.777007579803467,
+      "learning_rate": 9.573670160776588e-06,
+      "loss": 0.606,
+      "step": 9731
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.1491870880126953,
+      "learning_rate": 9.573563621618784e-06,
+      "loss": 0.8017,
+      "step": 9732
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.141711711883545,
+      "learning_rate": 9.573457069743612e-06,
+      "loss": 0.8704,
+      "step": 9733
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.8954893350601196,
+      "learning_rate": 9.573350505151368e-06,
+      "loss": 0.6153,
+      "step": 9734
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.758685350418091,
+      "learning_rate": 9.573243927842351e-06,
+      "loss": 0.6781,
+      "step": 9735
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.829807758331299,
+      "learning_rate": 9.573137337816858e-06,
+      "loss": 0.6566,
+      "step": 9736
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.031712532043457,
+      "learning_rate": 9.57303073507518e-06,
+      "loss": 0.42,
+      "step": 9737
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.1098673343658447,
+      "learning_rate": 9.572924119617617e-06,
+      "loss": 0.7166,
+      "step": 9738
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.765876293182373,
+      "learning_rate": 9.572817491444466e-06,
+      "loss": 0.5236,
+      "step": 9739
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.3737587928771973,
+      "learning_rate": 9.572710850556021e-06,
+      "loss": 0.6124,
+      "step": 9740
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.6088435649871826,
+      "learning_rate": 9.572604196952582e-06,
+      "loss": 0.8609,
+      "step": 9741
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.806699275970459,
+      "learning_rate": 9.572497530634442e-06,
+      "loss": 0.8313,
+      "step": 9742
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.706173896789551,
+      "learning_rate": 9.5723908516019e-06,
+      "loss": 0.6844,
+      "step": 9743
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.266831159591675,
+      "learning_rate": 9.572284159855252e-06,
+      "loss": 0.6142,
+      "step": 9744
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.86826753616333,
+      "learning_rate": 9.572177455394796e-06,
+      "loss": 0.6802,
+      "step": 9745
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.8956406116485596,
+      "learning_rate": 9.572070738220825e-06,
+      "loss": 0.8119,
+      "step": 9746
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.9416329860687256,
+      "learning_rate": 9.571964008333638e-06,
+      "loss": 0.6631,
+      "step": 9747
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.0720479488372803,
+      "learning_rate": 9.571857265733534e-06,
+      "loss": 0.7337,
+      "step": 9748
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.2104592323303223,
+      "learning_rate": 9.571750510420806e-06,
+      "loss": 0.7078,
+      "step": 9749
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.8171300888061523,
+      "learning_rate": 9.571643742395753e-06,
+      "loss": 0.5375,
+      "step": 9750
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.9190375804901123,
+      "learning_rate": 9.571536961658672e-06,
+      "loss": 0.5571,
+      "step": 9751
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.8494765758514404,
+      "learning_rate": 9.571430168209857e-06,
+      "loss": 0.5592,
+      "step": 9752
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.3564789295196533,
+      "learning_rate": 9.571323362049609e-06,
+      "loss": 0.5148,
+      "step": 9753
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.381859302520752,
+      "learning_rate": 9.571216543178224e-06,
+      "loss": 0.9098,
+      "step": 9754
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.611635446548462,
+      "learning_rate": 9.571109711595996e-06,
+      "loss": 0.7993,
+      "step": 9755
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.991265892982483,
+      "learning_rate": 9.571002867303226e-06,
+      "loss": 0.5199,
+      "step": 9756
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.1992223262786865,
+      "learning_rate": 9.570896010300208e-06,
+      "loss": 0.9291,
+      "step": 9757
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.9513661861419678,
+      "learning_rate": 9.57078914058724e-06,
+      "loss": 0.6136,
+      "step": 9758
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.8191137313842773,
+      "learning_rate": 9.570682258164621e-06,
+      "loss": 0.886,
+      "step": 9759
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.8642668724060059,
+      "learning_rate": 9.570575363032645e-06,
+      "loss": 0.5281,
+      "step": 9760
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.4827346801757812,
+      "learning_rate": 9.570468455191615e-06,
+      "loss": 0.701,
+      "step": 9761
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 12.382190704345703,
+      "learning_rate": 9.570361534641822e-06,
+      "loss": 0.4555,
+      "step": 9762
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.6185832023620605,
+      "learning_rate": 9.570254601383563e-06,
+      "loss": 0.6337,
+      "step": 9763
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.8853323459625244,
+      "learning_rate": 9.570147655417142e-06,
+      "loss": 0.708,
+      "step": 9764
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.951383113861084,
+      "learning_rate": 9.570040696742852e-06,
+      "loss": 0.5382,
+      "step": 9765
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.1830332279205322,
+      "learning_rate": 9.56993372536099e-06,
+      "loss": 0.4558,
+      "step": 9766
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.05102801322937,
+      "learning_rate": 9.569826741271853e-06,
+      "loss": 0.8781,
+      "step": 9767
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.077399492263794,
+      "learning_rate": 9.56971974447574e-06,
+      "loss": 0.3185,
+      "step": 9768
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 4.025172233581543,
+      "learning_rate": 9.56961273497295e-06,
+      "loss": 0.7403,
+      "step": 9769
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.0785012245178223,
+      "learning_rate": 9.569505712763777e-06,
+      "loss": 0.7283,
+      "step": 9770
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.527066230773926,
+      "learning_rate": 9.569398677848522e-06,
+      "loss": 0.5632,
+      "step": 9771
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 4.230048179626465,
+      "learning_rate": 9.56929163022748e-06,
+      "loss": 0.5102,
+      "step": 9772
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.896314859390259,
+      "learning_rate": 9.569184569900951e-06,
+      "loss": 0.6912,
+      "step": 9773
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 4.342724800109863,
+      "learning_rate": 9.56907749686923e-06,
+      "loss": 0.7404,
+      "step": 9774
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.095921039581299,
+      "learning_rate": 9.568970411132617e-06,
+      "loss": 0.53,
+      "step": 9775
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.8229455947875977,
+      "learning_rate": 9.568863312691407e-06,
+      "loss": 0.4978,
+      "step": 9776
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.286160469055176,
+      "learning_rate": 9.568756201545901e-06,
+      "loss": 0.7596,
+      "step": 9777
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.0309555530548096,
+      "learning_rate": 9.568649077696397e-06,
+      "loss": 0.8789,
+      "step": 9778
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.6662838459014893,
+      "learning_rate": 9.568541941143189e-06,
+      "loss": 0.4694,
+      "step": 9779
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.9892568588256836,
+      "learning_rate": 9.568434791886579e-06,
+      "loss": 0.8294,
+      "step": 9780
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.0382883548736572,
+      "learning_rate": 9.568327629926861e-06,
+      "loss": 0.579,
+      "step": 9781
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.0111191272735596,
+      "learning_rate": 9.568220455264336e-06,
+      "loss": 0.9368,
+      "step": 9782
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.6350982189178467,
+      "learning_rate": 9.568113267899302e-06,
+      "loss": 0.7979,
+      "step": 9783
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.824620008468628,
+      "learning_rate": 9.568006067832056e-06,
+      "loss": 0.7233,
+      "step": 9784
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 4.6943278312683105,
+      "learning_rate": 9.567898855062897e-06,
+      "loss": 0.5846,
+      "step": 9785
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 4.684123992919922,
+      "learning_rate": 9.567791629592123e-06,
+      "loss": 0.5062,
+      "step": 9786
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 4.342680931091309,
+      "learning_rate": 9.56768439142003e-06,
+      "loss": 0.8452,
+      "step": 9787
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.480039358139038,
+      "learning_rate": 9.567577140546918e-06,
+      "loss": 0.8413,
+      "step": 9788
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.9750442504882812,
+      "learning_rate": 9.567469876973085e-06,
+      "loss": 0.7503,
+      "step": 9789
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 6.800575256347656,
+      "learning_rate": 9.56736260069883e-06,
+      "loss": 0.2477,
+      "step": 9790
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.9949727058410645,
+      "learning_rate": 9.56725531172445e-06,
+      "loss": 0.6693,
+      "step": 9791
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.9172663688659668,
+      "learning_rate": 9.567148010050244e-06,
+      "loss": 0.7902,
+      "step": 9792
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.0562384128570557,
+      "learning_rate": 9.567040695676509e-06,
+      "loss": 0.7145,
+      "step": 9793
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.448880195617676,
+      "learning_rate": 9.566933368603546e-06,
+      "loss": 0.7351,
+      "step": 9794
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.1156835556030273,
+      "learning_rate": 9.566826028831652e-06,
+      "loss": 0.8941,
+      "step": 9795
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.947418689727783,
+      "learning_rate": 9.566718676361126e-06,
+      "loss": 0.6058,
+      "step": 9796
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.3248252868652344,
+      "learning_rate": 9.566611311192267e-06,
+      "loss": 0.8795,
+      "step": 9797
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.1676955223083496,
+      "learning_rate": 9.566503933325371e-06,
+      "loss": 0.5977,
+      "step": 9798
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 5.520416259765625,
+      "learning_rate": 9.566396542760738e-06,
+      "loss": 0.6373,
+      "step": 9799
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.9033315181732178,
+      "learning_rate": 9.566289139498668e-06,
+      "loss": 0.4493,
+      "step": 9800
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 4.035064697265625,
+      "learning_rate": 9.566181723539457e-06,
+      "loss": 0.9407,
+      "step": 9801
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.6605267524719238,
+      "learning_rate": 9.566074294883406e-06,
+      "loss": 0.7529,
+      "step": 9802
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.316183090209961,
+      "learning_rate": 9.565966853530813e-06,
+      "loss": 0.9561,
+      "step": 9803
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.0230348110198975,
+      "learning_rate": 9.565859399481975e-06,
+      "loss": 0.7293,
+      "step": 9804
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 5.274302959442139,
+      "learning_rate": 9.565751932737194e-06,
+      "loss": 0.5649,
+      "step": 9805
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.03739857673645,
+      "learning_rate": 9.56564445329677e-06,
+      "loss": 0.7423,
+      "step": 9806
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.8657338619232178,
+      "learning_rate": 9.565536961160995e-06,
+      "loss": 0.5168,
+      "step": 9807
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.1153640747070312,
+      "learning_rate": 9.565429456330173e-06,
+      "loss": 0.8023,
+      "step": 9808
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.196272134780884,
+      "learning_rate": 9.565321938804602e-06,
+      "loss": 0.6948,
+      "step": 9809
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.726189374923706,
+      "learning_rate": 9.565214408584582e-06,
+      "loss": 0.74,
+      "step": 9810
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.5671980381011963,
+      "learning_rate": 9.56510686567041e-06,
+      "loss": 0.7631,
+      "step": 9811
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.6737732887268066,
+      "learning_rate": 9.564999310062387e-06,
+      "loss": 0.5265,
+      "step": 9812
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.205702543258667,
+      "learning_rate": 9.564891741760808e-06,
+      "loss": 0.6172,
+      "step": 9813
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.644850492477417,
+      "learning_rate": 9.564784160765979e-06,
+      "loss": 0.9262,
+      "step": 9814
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.6317226886749268,
+      "learning_rate": 9.564676567078193e-06,
+      "loss": 0.5413,
+      "step": 9815
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.8514792919158936,
+      "learning_rate": 9.564568960697751e-06,
+      "loss": 0.6069,
+      "step": 9816
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.5978550910949707,
+      "learning_rate": 9.564461341624954e-06,
+      "loss": 0.7752,
+      "step": 9817
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 5.129965305328369,
+      "learning_rate": 9.5643537098601e-06,
+      "loss": 0.9318,
+      "step": 9818
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.388442277908325,
+      "learning_rate": 9.564246065403486e-06,
+      "loss": 0.4063,
+      "step": 9819
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.7987406253814697,
+      "learning_rate": 9.564138408255416e-06,
+      "loss": 0.5079,
+      "step": 9820
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 4.639627933502197,
+      "learning_rate": 9.564030738416185e-06,
+      "loss": 0.6606,
+      "step": 9821
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.544037342071533,
+      "learning_rate": 9.563923055886095e-06,
+      "loss": 0.623,
+      "step": 9822
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.648394823074341,
+      "learning_rate": 9.563815360665446e-06,
+      "loss": 0.371,
+      "step": 9823
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.6223983764648438,
+      "learning_rate": 9.563707652754535e-06,
+      "loss": 0.5602,
+      "step": 9824
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.8561618328094482,
+      "learning_rate": 9.563599932153661e-06,
+      "loss": 0.8748,
+      "step": 9825
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.756612777709961,
+      "learning_rate": 9.563492198863126e-06,
+      "loss": 0.7841,
+      "step": 9826
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.13173770904541,
+      "learning_rate": 9.563384452883231e-06,
+      "loss": 0.5726,
+      "step": 9827
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.3774654865264893,
+      "learning_rate": 9.563276694214271e-06,
+      "loss": 0.892,
+      "step": 9828
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.899033546447754,
+      "learning_rate": 9.56316892285655e-06,
+      "loss": 0.5995,
+      "step": 9829
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.247166633605957,
+      "learning_rate": 9.563061138810363e-06,
+      "loss": 0.6597,
+      "step": 9830
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.7387428283691406,
+      "learning_rate": 9.562953342076014e-06,
+      "loss": 0.5809,
+      "step": 9831
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.3044328689575195,
+      "learning_rate": 9.5628455326538e-06,
+      "loss": 0.5988,
+      "step": 9832
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.9683332443237305,
+      "learning_rate": 9.562737710544024e-06,
+      "loss": 0.8357,
+      "step": 9833
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.453125476837158,
+      "learning_rate": 9.562629875746981e-06,
+      "loss": 0.4709,
+      "step": 9834
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.578566074371338,
+      "learning_rate": 9.562522028262974e-06,
+      "loss": 0.7325,
+      "step": 9835
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.5397377014160156,
+      "learning_rate": 9.562414168092304e-06,
+      "loss": 0.9132,
+      "step": 9836
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.7028377056121826,
+      "learning_rate": 9.562306295235268e-06,
+      "loss": 0.6561,
+      "step": 9837
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.524498701095581,
+      "learning_rate": 9.562198409692168e-06,
+      "loss": 0.6252,
+      "step": 9838
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.978860378265381,
+      "learning_rate": 9.562090511463304e-06,
+      "loss": 0.667,
+      "step": 9839
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.9929331541061401,
+      "learning_rate": 9.561982600548974e-06,
+      "loss": 0.4267,
+      "step": 9840
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.2718985080718994,
+      "learning_rate": 9.561874676949478e-06,
+      "loss": 0.29,
+      "step": 9841
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.009803533554077,
+      "learning_rate": 9.56176674066512e-06,
+      "loss": 0.737,
+      "step": 9842
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.3564441204071045,
+      "learning_rate": 9.561658791696197e-06,
+      "loss": 0.6968,
+      "step": 9843
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.930527687072754,
+      "learning_rate": 9.56155083004301e-06,
+      "loss": 0.6903,
+      "step": 9844
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.4689974784851074,
+      "learning_rate": 9.561442855705858e-06,
+      "loss": 0.7899,
+      "step": 9845
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.059934616088867,
+      "learning_rate": 9.561334868685041e-06,
+      "loss": 0.5732,
+      "step": 9846
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 5.251935958862305,
+      "learning_rate": 9.561226868980862e-06,
+      "loss": 0.464,
+      "step": 9847
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 5.17796516418457,
+      "learning_rate": 9.561118856593621e-06,
+      "loss": 0.8021,
+      "step": 9848
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.7618660926818848,
+      "learning_rate": 9.561010831523616e-06,
+      "loss": 0.3202,
+      "step": 9849
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.976901054382324,
+      "learning_rate": 9.560902793771147e-06,
+      "loss": 0.7681,
+      "step": 9850
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.9044525623321533,
+      "learning_rate": 9.560794743336518e-06,
+      "loss": 0.7085,
+      "step": 9851
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.9936485290527344,
+      "learning_rate": 9.560686680220027e-06,
+      "loss": 0.6585,
+      "step": 9852
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.876704454421997,
+      "learning_rate": 9.560578604421973e-06,
+      "loss": 0.3937,
+      "step": 9853
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.0862271785736084,
+      "learning_rate": 9.56047051594266e-06,
+      "loss": 0.8106,
+      "step": 9854
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.3598947525024414,
+      "learning_rate": 9.560362414782387e-06,
+      "loss": 0.8871,
+      "step": 9855
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.5167438983917236,
+      "learning_rate": 9.560254300941455e-06,
+      "loss": 0.3777,
+      "step": 9856
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 4.126410961151123,
+      "learning_rate": 9.560146174420162e-06,
+      "loss": 1.125,
+      "step": 9857
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.079984664916992,
+      "learning_rate": 9.560038035218812e-06,
+      "loss": 0.8338,
+      "step": 9858
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.838648796081543,
+      "learning_rate": 9.559929883337703e-06,
+      "loss": 0.9193,
+      "step": 9859
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.4276607036590576,
+      "learning_rate": 9.55982171877714e-06,
+      "loss": 0.402,
+      "step": 9860
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.0659661293029785,
+      "learning_rate": 9.559713541537418e-06,
+      "loss": 0.5907,
+      "step": 9861
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.942427396774292,
+      "learning_rate": 9.559605351618844e-06,
+      "loss": 0.6836,
+      "step": 9862
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.1294825077056885,
+      "learning_rate": 9.559497149021712e-06,
+      "loss": 0.7886,
+      "step": 9863
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.6447081565856934,
+      "learning_rate": 9.55938893374633e-06,
+      "loss": 0.5848,
+      "step": 9864
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 6.346417427062988,
+      "learning_rate": 9.559280705792992e-06,
+      "loss": 0.8238,
+      "step": 9865
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.754533052444458,
+      "learning_rate": 9.559172465162004e-06,
+      "loss": 0.6168,
+      "step": 9866
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.107421398162842,
+      "learning_rate": 9.559064211853666e-06,
+      "loss": 0.5977,
+      "step": 9867
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.548701047897339,
+      "learning_rate": 9.558955945868277e-06,
+      "loss": 0.5087,
+      "step": 9868
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.0277698040008545,
+      "learning_rate": 9.558847667206139e-06,
+      "loss": 0.6078,
+      "step": 9869
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.278437376022339,
+      "learning_rate": 9.558739375867556e-06,
+      "loss": 0.7983,
+      "step": 9870
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.7710394859313965,
+      "learning_rate": 9.558631071852824e-06,
+      "loss": 0.4202,
+      "step": 9871
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.4019312858581543,
+      "learning_rate": 9.558522755162248e-06,
+      "loss": 0.6639,
+      "step": 9872
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 4.018265247344971,
+      "learning_rate": 9.558414425796127e-06,
+      "loss": 0.9879,
+      "step": 9873
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.3754546642303467,
+      "learning_rate": 9.558306083754763e-06,
+      "loss": 0.8363,
+      "step": 9874
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.451524257659912,
+      "learning_rate": 9.558197729038457e-06,
+      "loss": 0.5541,
+      "step": 9875
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.2919185161590576,
+      "learning_rate": 9.55808936164751e-06,
+      "loss": 0.63,
+      "step": 9876
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.728198766708374,
+      "learning_rate": 9.557980981582228e-06,
+      "loss": 0.6491,
+      "step": 9877
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 4.910820007324219,
+      "learning_rate": 9.557872588842904e-06,
+      "loss": 0.641,
+      "step": 9878
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 4.423444747924805,
+      "learning_rate": 9.557764183429847e-06,
+      "loss": 0.4391,
+      "step": 9879
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 5.880538463592529,
+      "learning_rate": 9.557655765343354e-06,
+      "loss": 0.9481,
+      "step": 9880
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.584414482116699,
+      "learning_rate": 9.557547334583726e-06,
+      "loss": 0.665,
+      "step": 9881
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.159853219985962,
+      "learning_rate": 9.557438891151268e-06,
+      "loss": 0.5847,
+      "step": 9882
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.9297564029693604,
+      "learning_rate": 9.557330435046279e-06,
+      "loss": 0.6829,
+      "step": 9883
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.194007158279419,
+      "learning_rate": 9.557221966269062e-06,
+      "loss": 0.498,
+      "step": 9884
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.67551851272583,
+      "learning_rate": 9.557113484819918e-06,
+      "loss": 0.68,
+      "step": 9885
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.9704209566116333,
+      "learning_rate": 9.557004990699147e-06,
+      "loss": 0.6162,
+      "step": 9886
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.069786548614502,
+      "learning_rate": 9.556896483907054e-06,
+      "loss": 0.9077,
+      "step": 9887
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.500004529953003,
+      "learning_rate": 9.556787964443937e-06,
+      "loss": 0.5575,
+      "step": 9888
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.3947219848632812,
+      "learning_rate": 9.5566794323101e-06,
+      "loss": 0.6152,
+      "step": 9889
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 4.886354923248291,
+      "learning_rate": 9.556570887505846e-06,
+      "loss": 0.6482,
+      "step": 9890
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.8240156173706055,
+      "learning_rate": 9.556462330031475e-06,
+      "loss": 0.7923,
+      "step": 9891
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.35872483253479,
+      "learning_rate": 9.55635375988729e-06,
+      "loss": 0.5098,
+      "step": 9892
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.4055776596069336,
+      "learning_rate": 9.55624517707359e-06,
+      "loss": 0.4861,
+      "step": 9893
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.980097770690918,
+      "learning_rate": 9.556136581590679e-06,
+      "loss": 0.6171,
+      "step": 9894
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.895800828933716,
+      "learning_rate": 9.556027973438859e-06,
+      "loss": 0.5753,
+      "step": 9895
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.8086880445480347,
+      "learning_rate": 9.555919352618432e-06,
+      "loss": 0.5794,
+      "step": 9896
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.2958359718322754,
+      "learning_rate": 9.555810719129702e-06,
+      "loss": 0.606,
+      "step": 9897
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.1954243183135986,
+      "learning_rate": 9.555702072972967e-06,
+      "loss": 0.6408,
+      "step": 9898
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.9131903648376465,
+      "learning_rate": 9.555593414148531e-06,
+      "loss": 0.9228,
+      "step": 9899
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.391402244567871,
+      "learning_rate": 9.555484742656697e-06,
+      "loss": 0.7221,
+      "step": 9900
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.457822799682617,
+      "learning_rate": 9.555376058497767e-06,
+      "loss": 0.5677,
+      "step": 9901
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.5186946392059326,
+      "learning_rate": 9.555267361672043e-06,
+      "loss": 0.8309,
+      "step": 9902
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.9854886531829834,
+      "learning_rate": 9.555158652179824e-06,
+      "loss": 0.8122,
+      "step": 9903
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.8589978218078613,
+      "learning_rate": 9.555049930021419e-06,
+      "loss": 0.7589,
+      "step": 9904
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.7715420722961426,
+      "learning_rate": 9.554941195197124e-06,
+      "loss": 0.6076,
+      "step": 9905
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.502872943878174,
+      "learning_rate": 9.554832447707245e-06,
+      "loss": 0.5689,
+      "step": 9906
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.1934030055999756,
+      "learning_rate": 9.554723687552083e-06,
+      "loss": 0.6848,
+      "step": 9907
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.0013654232025146,
+      "learning_rate": 9.554614914731939e-06,
+      "loss": 0.9684,
+      "step": 9908
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.8530268669128418,
+      "learning_rate": 9.554506129247119e-06,
+      "loss": 0.6025,
+      "step": 9909
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.022693634033203,
+      "learning_rate": 9.554397331097923e-06,
+      "loss": 0.7189,
+      "step": 9910
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.632110595703125,
+      "learning_rate": 9.554288520284653e-06,
+      "loss": 0.5925,
+      "step": 9911
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.668820858001709,
+      "learning_rate": 9.554179696807615e-06,
+      "loss": 0.8019,
+      "step": 9912
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.1896796226501465,
+      "learning_rate": 9.554070860667108e-06,
+      "loss": 0.5455,
+      "step": 9913
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.4498629570007324,
+      "learning_rate": 9.553962011863436e-06,
+      "loss": 0.674,
+      "step": 9914
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.7542307376861572,
+      "learning_rate": 9.553853150396903e-06,
+      "loss": 0.6029,
+      "step": 9915
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.3131942749023438,
+      "learning_rate": 9.553744276267809e-06,
+      "loss": 0.682,
+      "step": 9916
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.9186440706253052,
+      "learning_rate": 9.553635389476458e-06,
+      "loss": 0.5576,
+      "step": 9917
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.081665277481079,
+      "learning_rate": 9.553526490023153e-06,
+      "loss": 0.6522,
+      "step": 9918
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.690891981124878,
+      "learning_rate": 9.553417577908197e-06,
+      "loss": 0.6442,
+      "step": 9919
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.5126595497131348,
+      "learning_rate": 9.553308653131892e-06,
+      "loss": 0.8149,
+      "step": 9920
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.9386684894561768,
+      "learning_rate": 9.553199715694542e-06,
+      "loss": 0.575,
+      "step": 9921
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.7694294452667236,
+      "learning_rate": 9.553090765596448e-06,
+      "loss": 0.5665,
+      "step": 9922
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.716355800628662,
+      "learning_rate": 9.552981802837915e-06,
+      "loss": 0.8512,
+      "step": 9923
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.5828135013580322,
+      "learning_rate": 9.552872827419243e-06,
+      "loss": 0.3913,
+      "step": 9924
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.3128199577331543,
+      "learning_rate": 9.552763839340741e-06,
+      "loss": 0.6432,
+      "step": 9925
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.2608816623687744,
+      "learning_rate": 9.552654838602706e-06,
+      "loss": 0.6463,
+      "step": 9926
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.5992491245269775,
+      "learning_rate": 9.552545825205444e-06,
+      "loss": 0.7227,
+      "step": 9927
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.9064269065856934,
+      "learning_rate": 9.552436799149259e-06,
+      "loss": 0.6258,
+      "step": 9928
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.6319127082824707,
+      "learning_rate": 9.55232776043445e-06,
+      "loss": 0.6501,
+      "step": 9929
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.606477975845337,
+      "learning_rate": 9.552218709061323e-06,
+      "loss": 0.5788,
+      "step": 9930
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.9874954223632812,
+      "learning_rate": 9.552109645030183e-06,
+      "loss": 0.9524,
+      "step": 9931
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.4094350337982178,
+      "learning_rate": 9.55200056834133e-06,
+      "loss": 0.6096,
+      "step": 9932
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.6496853828430176,
+      "learning_rate": 9.55189147899507e-06,
+      "loss": 0.9602,
+      "step": 9933
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.6862123012542725,
+      "learning_rate": 9.551782376991703e-06,
+      "loss": 0.5584,
+      "step": 9934
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.343508243560791,
+      "learning_rate": 9.551673262331535e-06,
+      "loss": 0.6715,
+      "step": 9935
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.1546027660369873,
+      "learning_rate": 9.551564135014868e-06,
+      "loss": 0.4495,
+      "step": 9936
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.9186586141586304,
+      "learning_rate": 9.551454995042008e-06,
+      "loss": 0.9833,
+      "step": 9937
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.5251119136810303,
+      "learning_rate": 9.551345842413256e-06,
+      "loss": 0.5601,
+      "step": 9938
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.250265598297119,
+      "learning_rate": 9.551236677128915e-06,
+      "loss": 0.8359,
+      "step": 9939
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.807934045791626,
+      "learning_rate": 9.551127499189292e-06,
+      "loss": 0.6624,
+      "step": 9940
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.4350197315216064,
+      "learning_rate": 9.551018308594685e-06,
+      "loss": 0.6059,
+      "step": 9941
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.8405252695083618,
+      "learning_rate": 9.550909105345405e-06,
+      "loss": 0.6378,
+      "step": 9942
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.2193307876586914,
+      "learning_rate": 9.55079988944175e-06,
+      "loss": 0.3498,
+      "step": 9943
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.1734211444854736,
+      "learning_rate": 9.550690660884024e-06,
+      "loss": 0.6552,
+      "step": 9944
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.6406359672546387,
+      "learning_rate": 9.550581419672532e-06,
+      "loss": 0.604,
+      "step": 9945
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.986671209335327,
+      "learning_rate": 9.55047216580758e-06,
+      "loss": 0.6165,
+      "step": 9946
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.6881613731384277,
+      "learning_rate": 9.550362899289468e-06,
+      "loss": 0.7255,
+      "step": 9947
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.87017560005188,
+      "learning_rate": 9.550253620118501e-06,
+      "loss": 0.7294,
+      "step": 9948
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.3522913455963135,
+      "learning_rate": 9.550144328294985e-06,
+      "loss": 0.5624,
+      "step": 9949
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.6912384033203125,
+      "learning_rate": 9.55003502381922e-06,
+      "loss": 0.7683,
+      "step": 9950
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.5462048053741455,
+      "learning_rate": 9.549925706691514e-06,
+      "loss": 0.7314,
+      "step": 9951
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.055619239807129,
+      "learning_rate": 9.549816376912167e-06,
+      "loss": 0.7412,
+      "step": 9952
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.3498294353485107,
+      "learning_rate": 9.549707034481487e-06,
+      "loss": 0.6814,
+      "step": 9953
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.5553696155548096,
+      "learning_rate": 9.549597679399775e-06,
+      "loss": 0.9117,
+      "step": 9954
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.1391398906707764,
+      "learning_rate": 9.549488311667336e-06,
+      "loss": 0.6553,
+      "step": 9955
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.9216763973236084,
+      "learning_rate": 9.549378931284476e-06,
+      "loss": 0.7934,
+      "step": 9956
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.57479989528656,
+      "learning_rate": 9.549269538251495e-06,
+      "loss": 0.5952,
+      "step": 9957
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.6018929481506348,
+      "learning_rate": 9.549160132568702e-06,
+      "loss": 0.579,
+      "step": 9958
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.0652949810028076,
+      "learning_rate": 9.549050714236397e-06,
+      "loss": 0.8974,
+      "step": 9959
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.1884493827819824,
+      "learning_rate": 9.548941283254887e-06,
+      "loss": 0.7147,
+      "step": 9960
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.465529203414917,
+      "learning_rate": 9.548831839624476e-06,
+      "loss": 0.7207,
+      "step": 9961
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.736881732940674,
+      "learning_rate": 9.548722383345466e-06,
+      "loss": 0.8596,
+      "step": 9962
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.6420090198516846,
+      "learning_rate": 9.548612914418162e-06,
+      "loss": 0.6652,
+      "step": 9963
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.5332701206207275,
+      "learning_rate": 9.548503432842872e-06,
+      "loss": 0.5447,
+      "step": 9964
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.8131459951400757,
+      "learning_rate": 9.548393938619898e-06,
+      "loss": 0.8036,
+      "step": 9965
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.865621566772461,
+      "learning_rate": 9.548284431749542e-06,
+      "loss": 0.6401,
+      "step": 9966
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.1528186798095703,
+      "learning_rate": 9.548174912232112e-06,
+      "loss": 0.6321,
+      "step": 9967
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.638716459274292,
+      "learning_rate": 9.548065380067912e-06,
+      "loss": 0.6268,
+      "step": 9968
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.2079577445983887,
+      "learning_rate": 9.547955835257245e-06,
+      "loss": 0.8853,
+      "step": 9969
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.9679503440856934,
+      "learning_rate": 9.547846277800416e-06,
+      "loss": 0.8408,
+      "step": 9970
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 4.2325215339660645,
+      "learning_rate": 9.547736707697732e-06,
+      "loss": 0.5755,
+      "step": 9971
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.132232666015625,
+      "learning_rate": 9.547627124949496e-06,
+      "loss": 0.5737,
+      "step": 9972
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.2992541790008545,
+      "learning_rate": 9.547517529556011e-06,
+      "loss": 0.782,
+      "step": 9973
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.7059829235076904,
+      "learning_rate": 9.547407921517584e-06,
+      "loss": 0.6559,
+      "step": 9974
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.7573598623275757,
+      "learning_rate": 9.547298300834519e-06,
+      "loss": 0.5483,
+      "step": 9975
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.5517141819000244,
+      "learning_rate": 9.54718866750712e-06,
+      "loss": 0.6507,
+      "step": 9976
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.7588428258895874,
+      "learning_rate": 9.547079021535694e-06,
+      "loss": 0.7179,
+      "step": 9977
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 6.870862007141113,
+      "learning_rate": 9.546969362920543e-06,
+      "loss": 0.526,
+      "step": 9978
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.011665105819702,
+      "learning_rate": 9.546859691661976e-06,
+      "loss": 0.7769,
+      "step": 9979
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.3949882984161377,
+      "learning_rate": 9.546750007760294e-06,
+      "loss": 0.5671,
+      "step": 9980
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 6.056945323944092,
+      "learning_rate": 9.546640311215804e-06,
+      "loss": 1.0011,
+      "step": 9981
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 4.1125688552856445,
+      "learning_rate": 9.546530602028809e-06,
+      "loss": 0.6475,
+      "step": 9982
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.2459545135498047,
+      "learning_rate": 9.546420880199617e-06,
+      "loss": 0.7356,
+      "step": 9983
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.8155505657196045,
+      "learning_rate": 9.546311145728531e-06,
+      "loss": 0.4327,
+      "step": 9984
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.6482863426208496,
+      "learning_rate": 9.546201398615858e-06,
+      "loss": 0.5403,
+      "step": 9985
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.674241304397583,
+      "learning_rate": 9.5460916388619e-06,
+      "loss": 0.768,
+      "step": 9986
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.570509672164917,
+      "learning_rate": 9.545981866466967e-06,
+      "loss": 0.7246,
+      "step": 9987
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.480600118637085,
+      "learning_rate": 9.545872081431359e-06,
+      "loss": 0.8933,
+      "step": 9988
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.3401479721069336,
+      "learning_rate": 9.545762283755384e-06,
+      "loss": 0.631,
+      "step": 9989
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.6985855102539062,
+      "learning_rate": 9.545652473439348e-06,
+      "loss": 0.5032,
+      "step": 9990
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.50523042678833,
+      "learning_rate": 9.545542650483555e-06,
+      "loss": 0.8725,
+      "step": 9991
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.108868360519409,
+      "learning_rate": 9.545432814888309e-06,
+      "loss": 0.5078,
+      "step": 9992
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.8238778114318848,
+      "learning_rate": 9.545322966653918e-06,
+      "loss": 0.4893,
+      "step": 9993
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.0586071014404297,
+      "learning_rate": 9.545213105780688e-06,
+      "loss": 0.4867,
+      "step": 9994
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 4.448045253753662,
+      "learning_rate": 9.545103232268922e-06,
+      "loss": 0.7073,
+      "step": 9995
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.3496851921081543,
+      "learning_rate": 9.544993346118926e-06,
+      "loss": 0.6568,
+      "step": 9996
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.7440450191497803,
+      "learning_rate": 9.544883447331007e-06,
+      "loss": 0.5705,
+      "step": 9997
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.056246519088745,
+      "learning_rate": 9.544773535905469e-06,
+      "loss": 0.55,
+      "step": 9998
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.117652654647827,
+      "learning_rate": 9.544663611842619e-06,
+      "loss": 0.3972,
+      "step": 9999
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.2699472904205322,
+      "learning_rate": 9.544553675142762e-06,
+      "loss": 0.5544,
+      "step": 10000
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.9115086793899536,
+      "learning_rate": 9.544443725806202e-06,
+      "loss": 0.8355,
+      "step": 10001
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.7674100399017334,
+      "learning_rate": 9.54433376383325e-06,
+      "loss": 0.34,
+      "step": 10002
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.307246446609497,
+      "learning_rate": 9.544223789224204e-06,
+      "loss": 0.8728,
+      "step": 10003
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.3160383701324463,
+      "learning_rate": 9.544113801979376e-06,
+      "loss": 0.7438,
+      "step": 10004
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.5999209880828857,
+      "learning_rate": 9.54400380209907e-06,
+      "loss": 0.709,
+      "step": 10005
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.559575080871582,
+      "learning_rate": 9.543893789583591e-06,
+      "loss": 0.6848,
+      "step": 10006
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 4.218487739562988,
+      "learning_rate": 9.543783764433245e-06,
+      "loss": 0.9402,
+      "step": 10007
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 4.687023162841797,
+      "learning_rate": 9.54367372664834e-06,
+      "loss": 0.749,
+      "step": 10008
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.6233476400375366,
+      "learning_rate": 9.543563676229179e-06,
+      "loss": 0.4305,
+      "step": 10009
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.0329267978668213,
+      "learning_rate": 9.54345361317607e-06,
+      "loss": 0.5574,
+      "step": 10010
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.192730665206909,
+      "learning_rate": 9.543343537489319e-06,
+      "loss": 0.6975,
+      "step": 10011
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.779508352279663,
+      "learning_rate": 9.543233449169231e-06,
+      "loss": 0.9087,
+      "step": 10012
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.2456302642822266,
+      "learning_rate": 9.543123348216112e-06,
+      "loss": 0.6058,
+      "step": 10013
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.3412792682647705,
+      "learning_rate": 9.54301323463027e-06,
+      "loss": 0.6984,
+      "step": 10014
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.742541790008545,
+      "learning_rate": 9.542903108412009e-06,
+      "loss": 0.4617,
+      "step": 10015
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.828937292098999,
+      "learning_rate": 9.542792969561638e-06,
+      "loss": 0.8353,
+      "step": 10016
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.03497052192688,
+      "learning_rate": 9.54268281807946e-06,
+      "loss": 0.4825,
+      "step": 10017
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.5806117057800293,
+      "learning_rate": 9.542572653965784e-06,
+      "loss": 0.6153,
+      "step": 10018
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.1637985706329346,
+      "learning_rate": 9.542462477220915e-06,
+      "loss": 0.663,
+      "step": 10019
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.158602237701416,
+      "learning_rate": 9.542352287845159e-06,
+      "loss": 0.8628,
+      "step": 10020
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.122044563293457,
+      "learning_rate": 9.542242085838823e-06,
+      "loss": 0.6504,
+      "step": 10021
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.4158270359039307,
+      "learning_rate": 9.542131871202214e-06,
+      "loss": 0.5473,
+      "step": 10022
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.0933456420898438,
+      "learning_rate": 9.542021643935638e-06,
+      "loss": 0.5585,
+      "step": 10023
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.5489213466644287,
+      "learning_rate": 9.5419114040394e-06,
+      "loss": 0.6068,
+      "step": 10024
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.2359464168548584,
+      "learning_rate": 9.541801151513808e-06,
+      "loss": 0.346,
+      "step": 10025
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.925541877746582,
+      "learning_rate": 9.541690886359167e-06,
+      "loss": 0.75,
+      "step": 10026
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.449756383895874,
+      "learning_rate": 9.541580608575787e-06,
+      "loss": 0.9724,
+      "step": 10027
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.438413143157959,
+      "learning_rate": 9.541470318163972e-06,
+      "loss": 0.7304,
+      "step": 10028
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.2872862815856934,
+      "learning_rate": 9.541360015124031e-06,
+      "loss": 0.724,
+      "step": 10029
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.966677188873291,
+      "learning_rate": 9.541249699456267e-06,
+      "loss": 0.7218,
+      "step": 10030
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.0858850479125977,
+      "learning_rate": 9.541139371160989e-06,
+      "loss": 0.6879,
+      "step": 10031
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.6765029430389404,
+      "learning_rate": 9.541029030238505e-06,
+      "loss": 0.4767,
+      "step": 10032
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.763856887817383,
+      "learning_rate": 9.540918676689119e-06,
+      "loss": 0.82,
+      "step": 10033
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.2701234817504883,
+      "learning_rate": 9.54080831051314e-06,
+      "loss": 0.3649,
+      "step": 10034
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.748669385910034,
+      "learning_rate": 9.540697931710872e-06,
+      "loss": 0.8863,
+      "step": 10035
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.7841534614562988,
+      "learning_rate": 9.540587540282626e-06,
+      "loss": 0.8402,
+      "step": 10036
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.561091661453247,
+      "learning_rate": 9.540477136228707e-06,
+      "loss": 0.6069,
+      "step": 10037
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.814286708831787,
+      "learning_rate": 9.540366719549421e-06,
+      "loss": 0.6668,
+      "step": 10038
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.6130847930908203,
+      "learning_rate": 9.540256290245078e-06,
+      "loss": 0.6188,
+      "step": 10039
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.8669018745422363,
+      "learning_rate": 9.54014584831598e-06,
+      "loss": 0.7981,
+      "step": 10040
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.484833240509033,
+      "learning_rate": 9.54003539376244e-06,
+      "loss": 0.6754,
+      "step": 10041
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.8146551847457886,
+      "learning_rate": 9.539924926584761e-06,
+      "loss": 0.7747,
+      "step": 10042
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.9629979133605957,
+      "learning_rate": 9.53981444678325e-06,
+      "loss": 0.3649,
+      "step": 10043
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.407999873161316,
+      "learning_rate": 9.53970395435822e-06,
+      "loss": 0.5632,
+      "step": 10044
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.8088442087173462,
+      "learning_rate": 9.53959344930997e-06,
+      "loss": 0.5342,
+      "step": 10045
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.825974464416504,
+      "learning_rate": 9.539482931638813e-06,
+      "loss": 0.6639,
+      "step": 10046
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.9467988014221191,
+      "learning_rate": 9.539372401345053e-06,
+      "loss": 0.6596,
+      "step": 10047
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.8473072052001953,
+      "learning_rate": 9.539261858429002e-06,
+      "loss": 0.602,
+      "step": 10048
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.074270248413086,
+      "learning_rate": 9.539151302890962e-06,
+      "loss": 0.5649,
+      "step": 10049
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.1628596782684326,
+      "learning_rate": 9.539040734731241e-06,
+      "loss": 0.5423,
+      "step": 10050
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.4590611457824707,
+      "learning_rate": 9.53893015395015e-06,
+      "loss": 0.7706,
+      "step": 10051
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 6.094876289367676,
+      "learning_rate": 9.538819560547995e-06,
+      "loss": 0.7649,
+      "step": 10052
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.8007171154022217,
+      "learning_rate": 9.538708954525082e-06,
+      "loss": 0.5188,
+      "step": 10053
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.7249879837036133,
+      "learning_rate": 9.53859833588172e-06,
+      "loss": 0.4318,
+      "step": 10054
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.165735960006714,
+      "learning_rate": 9.538487704618216e-06,
+      "loss": 0.3757,
+      "step": 10055
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.2154552936553955,
+      "learning_rate": 9.538377060734878e-06,
+      "loss": 0.4399,
+      "step": 10056
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.992588520050049,
+      "learning_rate": 9.538266404232014e-06,
+      "loss": 0.5765,
+      "step": 10057
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 4.8871002197265625,
+      "learning_rate": 9.53815573510993e-06,
+      "loss": 0.6916,
+      "step": 10058
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.047802686691284,
+      "learning_rate": 9.538045053368936e-06,
+      "loss": 0.7013,
+      "step": 10059
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.318777561187744,
+      "learning_rate": 9.537934359009338e-06,
+      "loss": 0.7465,
+      "step": 10060
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.8998163938522339,
+      "learning_rate": 9.537823652031446e-06,
+      "loss": 0.7061,
+      "step": 10061
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.8816070556640625,
+      "learning_rate": 9.537712932435566e-06,
+      "loss": 0.5822,
+      "step": 10062
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.0200588703155518,
+      "learning_rate": 9.537602200222004e-06,
+      "loss": 0.5056,
+      "step": 10063
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.8820343017578125,
+      "learning_rate": 9.537491455391072e-06,
+      "loss": 0.9503,
+      "step": 10064
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.270369291305542,
+      "learning_rate": 9.537380697943075e-06,
+      "loss": 0.8546,
+      "step": 10065
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.5528509616851807,
+      "learning_rate": 9.537269927878323e-06,
+      "loss": 0.7524,
+      "step": 10066
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.277357578277588,
+      "learning_rate": 9.537159145197124e-06,
+      "loss": 0.7713,
+      "step": 10067
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.9990612268447876,
+      "learning_rate": 9.537048349899783e-06,
+      "loss": 0.8965,
+      "step": 10068
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.5208425521850586,
+      "learning_rate": 9.536937541986611e-06,
+      "loss": 0.6342,
+      "step": 10069
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.1196470260620117,
+      "learning_rate": 9.536826721457915e-06,
+      "loss": 0.6845,
+      "step": 10070
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.0791919231414795,
+      "learning_rate": 9.536715888314003e-06,
+      "loss": 0.5794,
+      "step": 10071
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.8693456649780273,
+      "learning_rate": 9.536605042555184e-06,
+      "loss": 0.8099,
+      "step": 10072
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.026930570602417,
+      "learning_rate": 9.536494184181766e-06,
+      "loss": 0.9246,
+      "step": 10073
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.5695085525512695,
+      "learning_rate": 9.536383313194057e-06,
+      "loss": 0.5923,
+      "step": 10074
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.30311918258667,
+      "learning_rate": 9.536272429592365e-06,
+      "loss": 0.604,
+      "step": 10075
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.9869540929794312,
+      "learning_rate": 9.536161533377e-06,
+      "loss": 0.5942,
+      "step": 10076
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.1644387245178223,
+      "learning_rate": 9.536050624548268e-06,
+      "loss": 0.6693,
+      "step": 10077
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.0103530883789062,
+      "learning_rate": 9.535939703106479e-06,
+      "loss": 0.7704,
+      "step": 10078
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.7291452884674072,
+      "learning_rate": 9.53582876905194e-06,
+      "loss": 0.9114,
+      "step": 10079
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.6055426597595215,
+      "learning_rate": 9.535717822384962e-06,
+      "loss": 0.8085,
+      "step": 10080
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.4988999366760254,
+      "learning_rate": 9.53560686310585e-06,
+      "loss": 0.6317,
+      "step": 10081
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.8260304927825928,
+      "learning_rate": 9.535495891214916e-06,
+      "loss": 0.8874,
+      "step": 10082
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.6482203006744385,
+      "learning_rate": 9.535384906712466e-06,
+      "loss": 0.6229,
+      "step": 10083
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.4723005294799805,
+      "learning_rate": 9.53527390959881e-06,
+      "loss": 0.6954,
+      "step": 10084
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.0846943855285645,
+      "learning_rate": 9.535162899874255e-06,
+      "loss": 0.6061,
+      "step": 10085
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.2997376918792725,
+      "learning_rate": 9.535051877539112e-06,
+      "loss": 0.7093,
+      "step": 10086
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.876800060272217,
+      "learning_rate": 9.53494084259369e-06,
+      "loss": 0.8371,
+      "step": 10087
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.5868995189666748,
+      "learning_rate": 9.534829795038294e-06,
+      "loss": 0.6372,
+      "step": 10088
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.804968237876892,
+      "learning_rate": 9.534718734873237e-06,
+      "loss": 0.8642,
+      "step": 10089
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.378352403640747,
+      "learning_rate": 9.534607662098824e-06,
+      "loss": 0.6824,
+      "step": 10090
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.9846465587615967,
+      "learning_rate": 9.534496576715366e-06,
+      "loss": 0.7042,
+      "step": 10091
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.5396370887756348,
+      "learning_rate": 9.534385478723173e-06,
+      "loss": 0.4985,
+      "step": 10092
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.8865251541137695,
+      "learning_rate": 9.534274368122552e-06,
+      "loss": 0.969,
+      "step": 10093
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.2665908336639404,
+      "learning_rate": 9.534163244913812e-06,
+      "loss": 0.5827,
+      "step": 10094
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.1991803646087646,
+      "learning_rate": 9.534052109097262e-06,
+      "loss": 0.5816,
+      "step": 10095
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.4971230030059814,
+      "learning_rate": 9.533940960673213e-06,
+      "loss": 0.639,
+      "step": 10096
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.676687240600586,
+      "learning_rate": 9.533829799641973e-06,
+      "loss": 0.6837,
+      "step": 10097
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.1269099712371826,
+      "learning_rate": 9.53371862600385e-06,
+      "loss": 0.685,
+      "step": 10098
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.3789873123168945,
+      "learning_rate": 9.533607439759154e-06,
+      "loss": 0.7519,
+      "step": 10099
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.1922061443328857,
+      "learning_rate": 9.533496240908193e-06,
+      "loss": 0.9019,
+      "step": 10100
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.4761669635772705,
+      "learning_rate": 9.533385029451278e-06,
+      "loss": 0.6831,
+      "step": 10101
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.104382038116455,
+      "learning_rate": 9.533273805388718e-06,
+      "loss": 0.6608,
+      "step": 10102
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.758721113204956,
+      "learning_rate": 9.533162568720821e-06,
+      "loss": 0.5314,
+      "step": 10103
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.9752894639968872,
+      "learning_rate": 9.533051319447898e-06,
+      "loss": 0.7747,
+      "step": 10104
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.5392396450042725,
+      "learning_rate": 9.532940057570257e-06,
+      "loss": 0.6041,
+      "step": 10105
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.2779903411865234,
+      "learning_rate": 9.532828783088209e-06,
+      "loss": 0.7876,
+      "step": 10106
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.319488048553467,
+      "learning_rate": 9.53271749600206e-06,
+      "loss": 0.6611,
+      "step": 10107
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.4716593027114868,
+      "learning_rate": 9.532606196312122e-06,
+      "loss": 0.7439,
+      "step": 10108
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.0085999965667725,
+      "learning_rate": 9.532494884018704e-06,
+      "loss": 0.6925,
+      "step": 10109
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.098454236984253,
+      "learning_rate": 9.532383559122116e-06,
+      "loss": 0.7977,
+      "step": 10110
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.1064767837524414,
+      "learning_rate": 9.532272221622668e-06,
+      "loss": 0.7417,
+      "step": 10111
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.249027729034424,
+      "learning_rate": 9.532160871520669e-06,
+      "loss": 0.5671,
+      "step": 10112
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.4430227279663086,
+      "learning_rate": 9.532049508816427e-06,
+      "loss": 0.7132,
+      "step": 10113
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.615895986557007,
+      "learning_rate": 9.531938133510253e-06,
+      "loss": 0.6861,
+      "step": 10114
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.3997998237609863,
+      "learning_rate": 9.531826745602456e-06,
+      "loss": 0.6324,
+      "step": 10115
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.7182278633117676,
+      "learning_rate": 9.53171534509335e-06,
+      "loss": 0.3334,
+      "step": 10116
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.004659414291382,
+      "learning_rate": 9.531603931983239e-06,
+      "loss": 0.6202,
+      "step": 10117
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.327288866043091,
+      "learning_rate": 9.531492506272434e-06,
+      "loss": 0.7109,
+      "step": 10118
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.307752847671509,
+      "learning_rate": 9.531381067961248e-06,
+      "loss": 0.8513,
+      "step": 10119
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.14005446434021,
+      "learning_rate": 9.531269617049987e-06,
+      "loss": 0.6769,
+      "step": 10120
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.736250162124634,
+      "learning_rate": 9.531158153538963e-06,
+      "loss": 0.7289,
+      "step": 10121
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.1484038829803467,
+      "learning_rate": 9.531046677428487e-06,
+      "loss": 0.5662,
+      "step": 10122
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.3893500566482544,
+      "learning_rate": 9.530935188718866e-06,
+      "loss": 0.7129,
+      "step": 10123
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.550204873085022,
+      "learning_rate": 9.530823687410413e-06,
+      "loss": 0.589,
+      "step": 10124
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.3104305267333984,
+      "learning_rate": 9.530712173503435e-06,
+      "loss": 0.7244,
+      "step": 10125
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.1418402194976807,
+      "learning_rate": 9.530600646998244e-06,
+      "loss": 0.6968,
+      "step": 10126
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.4698660373687744,
+      "learning_rate": 9.530489107895151e-06,
+      "loss": 0.7532,
+      "step": 10127
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.430586338043213,
+      "learning_rate": 9.530377556194464e-06,
+      "loss": 0.9492,
+      "step": 10128
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.942626953125,
+      "learning_rate": 9.530265991896495e-06,
+      "loss": 0.729,
+      "step": 10129
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.7651076316833496,
+      "learning_rate": 9.530154415001553e-06,
+      "loss": 0.6967,
+      "step": 10130
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.300220012664795,
+      "learning_rate": 9.530042825509947e-06,
+      "loss": 0.7077,
+      "step": 10131
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.865198016166687,
+      "learning_rate": 9.52993122342199e-06,
+      "loss": 0.5942,
+      "step": 10132
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.1846165657043457,
+      "learning_rate": 9.529819608737992e-06,
+      "loss": 0.7172,
+      "step": 10133
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.2166061401367188,
+      "learning_rate": 9.529707981458262e-06,
+      "loss": 0.8094,
+      "step": 10134
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.933180570602417,
+      "learning_rate": 9.529596341583112e-06,
+      "loss": 0.5166,
+      "step": 10135
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 4.360199928283691,
+      "learning_rate": 9.529484689112848e-06,
+      "loss": 0.6927,
+      "step": 10136
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.8257620334625244,
+      "learning_rate": 9.529373024047787e-06,
+      "loss": 0.7433,
+      "step": 10137
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.5652315616607666,
+      "learning_rate": 9.529261346388237e-06,
+      "loss": 0.4937,
+      "step": 10138
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.1157147884368896,
+      "learning_rate": 9.529149656134505e-06,
+      "loss": 0.6559,
+      "step": 10139
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 5.113277435302734,
+      "learning_rate": 9.529037953286906e-06,
+      "loss": 0.6496,
+      "step": 10140
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.678004264831543,
+      "learning_rate": 9.52892623784575e-06,
+      "loss": 0.6121,
+      "step": 10141
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.2646260261535645,
+      "learning_rate": 9.528814509811344e-06,
+      "loss": 0.7328,
+      "step": 10142
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.334526300430298,
+      "learning_rate": 9.528702769184004e-06,
+      "loss": 0.5445,
+      "step": 10143
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.2351083755493164,
+      "learning_rate": 9.528591015964038e-06,
+      "loss": 0.511,
+      "step": 10144
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.4981930255889893,
+      "learning_rate": 9.528479250151755e-06,
+      "loss": 0.5467,
+      "step": 10145
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.369718074798584,
+      "learning_rate": 9.528367471747468e-06,
+      "loss": 0.6661,
+      "step": 10146
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.6800177097320557,
+      "learning_rate": 9.528255680751488e-06,
+      "loss": 0.5615,
+      "step": 10147
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 4.050826072692871,
+      "learning_rate": 9.528143877164127e-06,
+      "loss": 0.6068,
+      "step": 10148
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.3220155239105225,
+      "learning_rate": 9.528032060985693e-06,
+      "loss": 0.7816,
+      "step": 10149
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.2812654972076416,
+      "learning_rate": 9.527920232216497e-06,
+      "loss": 0.4987,
+      "step": 10150
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 4.180825233459473,
+      "learning_rate": 9.52780839085685e-06,
+      "loss": 0.6039,
+      "step": 10151
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.5067288875579834,
+      "learning_rate": 9.527696536907066e-06,
+      "loss": 0.6527,
+      "step": 10152
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 4.121750354766846,
+      "learning_rate": 9.527584670367453e-06,
+      "loss": 0.681,
+      "step": 10153
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.7186996936798096,
+      "learning_rate": 9.527472791238324e-06,
+      "loss": 0.764,
+      "step": 10154
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.2156317234039307,
+      "learning_rate": 9.527360899519991e-06,
+      "loss": 0.7256,
+      "step": 10155
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.7455782890319824,
+      "learning_rate": 9.52724899521276e-06,
+      "loss": 0.9619,
+      "step": 10156
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.739417791366577,
+      "learning_rate": 9.527137078316948e-06,
+      "loss": 0.6158,
+      "step": 10157
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.8203258514404297,
+      "learning_rate": 9.527025148832861e-06,
+      "loss": 0.8515,
+      "step": 10158
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.4197826385498047,
+      "learning_rate": 9.526913206760814e-06,
+      "loss": 0.5147,
+      "step": 10159
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 4.025177478790283,
+      "learning_rate": 9.52680125210112e-06,
+      "loss": 0.6982,
+      "step": 10160
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.963836669921875,
+      "learning_rate": 9.526689284854083e-06,
+      "loss": 0.945,
+      "step": 10161
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.7805184125900269,
+      "learning_rate": 9.526577305020022e-06,
+      "loss": 0.9027,
+      "step": 10162
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.846395969390869,
+      "learning_rate": 9.526465312599244e-06,
+      "loss": 0.8655,
+      "step": 10163
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.061797618865967,
+      "learning_rate": 9.526353307592062e-06,
+      "loss": 0.8226,
+      "step": 10164
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.633162260055542,
+      "learning_rate": 9.526241289998786e-06,
+      "loss": 0.6021,
+      "step": 10165
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.017263412475586,
+      "learning_rate": 9.52612925981973e-06,
+      "loss": 0.7264,
+      "step": 10166
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.7931668758392334,
+      "learning_rate": 9.526017217055203e-06,
+      "loss": 0.6302,
+      "step": 10167
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.0060858726501465,
+      "learning_rate": 9.525905161705518e-06,
+      "loss": 0.6822,
+      "step": 10168
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.7265491485595703,
+      "learning_rate": 9.525793093770986e-06,
+      "loss": 0.7675,
+      "step": 10169
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.303175449371338,
+      "learning_rate": 9.52568101325192e-06,
+      "loss": 0.6615,
+      "step": 10170
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 4.005038261413574,
+      "learning_rate": 9.52556892014863e-06,
+      "loss": 0.7533,
+      "step": 10171
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.6369974613189697,
+      "learning_rate": 9.525456814461427e-06,
+      "loss": 0.7166,
+      "step": 10172
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.402153730392456,
+      "learning_rate": 9.525344696190624e-06,
+      "loss": 0.6375,
+      "step": 10173
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 5.588066101074219,
+      "learning_rate": 9.525232565336532e-06,
+      "loss": 0.7699,
+      "step": 10174
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.5125622749328613,
+      "learning_rate": 9.525120421899465e-06,
+      "loss": 0.7184,
+      "step": 10175
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 4.73606538772583,
+      "learning_rate": 9.525008265879732e-06,
+      "loss": 0.6395,
+      "step": 10176
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.453115701675415,
+      "learning_rate": 9.524896097277648e-06,
+      "loss": 0.5548,
+      "step": 10177
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.839975118637085,
+      "learning_rate": 9.524783916093521e-06,
+      "loss": 0.7177,
+      "step": 10178
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.4408442974090576,
+      "learning_rate": 9.524671722327666e-06,
+      "loss": 0.6017,
+      "step": 10179
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.8730099201202393,
+      "learning_rate": 9.524559515980394e-06,
+      "loss": 0.8964,
+      "step": 10180
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.621114730834961,
+      "learning_rate": 9.524447297052017e-06,
+      "loss": 0.8011,
+      "step": 10181
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.9234715700149536,
+      "learning_rate": 9.524335065542847e-06,
+      "loss": 0.4919,
+      "step": 10182
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.803953170776367,
+      "learning_rate": 9.524222821453196e-06,
+      "loss": 0.7873,
+      "step": 10183
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 7.327319622039795,
+      "learning_rate": 9.524110564783375e-06,
+      "loss": 0.7552,
+      "step": 10184
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.1525821685791016,
+      "learning_rate": 9.523998295533699e-06,
+      "loss": 0.5389,
+      "step": 10185
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.4166173934936523,
+      "learning_rate": 9.523886013704477e-06,
+      "loss": 0.6571,
+      "step": 10186
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.58183217048645,
+      "learning_rate": 9.523773719296023e-06,
+      "loss": 0.7651,
+      "step": 10187
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.0046989917755127,
+      "learning_rate": 9.523661412308649e-06,
+      "loss": 0.5565,
+      "step": 10188
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 4.458701133728027,
+      "learning_rate": 9.523549092742668e-06,
+      "loss": 0.8574,
+      "step": 10189
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.6094489097595215,
+      "learning_rate": 9.523436760598391e-06,
+      "loss": 0.7428,
+      "step": 10190
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.91934871673584,
+      "learning_rate": 9.523324415876131e-06,
+      "loss": 0.6774,
+      "step": 10191
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.2343533039093018,
+      "learning_rate": 9.523212058576201e-06,
+      "loss": 0.6606,
+      "step": 10192
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.871098518371582,
+      "learning_rate": 9.52309968869891e-06,
+      "loss": 0.6179,
+      "step": 10193
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.373697280883789,
+      "learning_rate": 9.522987306244575e-06,
+      "loss": 0.9331,
+      "step": 10194
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.685900330543518,
+      "learning_rate": 9.522874911213508e-06,
+      "loss": 0.7403,
+      "step": 10195
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.0353267192840576,
+      "learning_rate": 9.52276250360602e-06,
+      "loss": 0.7141,
+      "step": 10196
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.2712647914886475,
+      "learning_rate": 9.522650083422421e-06,
+      "loss": 0.6606,
+      "step": 10197
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 4.28580379486084,
+      "learning_rate": 9.52253765066303e-06,
+      "loss": 0.8057,
+      "step": 10198
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.7205839157104492,
+      "learning_rate": 9.522425205328151e-06,
+      "loss": 0.7642,
+      "step": 10199
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.3312978744506836,
+      "learning_rate": 9.522312747418105e-06,
+      "loss": 0.8076,
+      "step": 10200
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.647197723388672,
+      "learning_rate": 9.522200276933202e-06,
+      "loss": 0.6915,
+      "step": 10201
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.5210509300231934,
+      "learning_rate": 9.522087793873753e-06,
+      "loss": 0.8372,
+      "step": 10202
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 4.283466339111328,
+      "learning_rate": 9.521975298240072e-06,
+      "loss": 0.754,
+      "step": 10203
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.5277230739593506,
+      "learning_rate": 9.52186279003247e-06,
+      "loss": 0.6517,
+      "step": 10204
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.0523579120635986,
+      "learning_rate": 9.521750269251265e-06,
+      "loss": 0.9453,
+      "step": 10205
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.6328976154327393,
+      "learning_rate": 9.521637735896763e-06,
+      "loss": 0.7699,
+      "step": 10206
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 5.005312442779541,
+      "learning_rate": 9.521525189969284e-06,
+      "loss": 0.7807,
+      "step": 10207
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.844050407409668,
+      "learning_rate": 9.521412631469133e-06,
+      "loss": 0.5205,
+      "step": 10208
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.6667779684066772,
+      "learning_rate": 9.52130006039663e-06,
+      "loss": 0.5637,
+      "step": 10209
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.103351593017578,
+      "learning_rate": 9.521187476752083e-06,
+      "loss": 0.9455,
+      "step": 10210
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.885805606842041,
+      "learning_rate": 9.52107488053581e-06,
+      "loss": 0.5138,
+      "step": 10211
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.3579564094543457,
+      "learning_rate": 9.520962271748119e-06,
+      "loss": 0.7017,
+      "step": 10212
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 4.061976909637451,
+      "learning_rate": 9.520849650389328e-06,
+      "loss": 0.8703,
+      "step": 10213
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.183725357055664,
+      "learning_rate": 9.520737016459744e-06,
+      "loss": 0.7417,
+      "step": 10214
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.28484845161438,
+      "learning_rate": 9.520624369959687e-06,
+      "loss": 0.8736,
+      "step": 10215
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.1663830280303955,
+      "learning_rate": 9.520511710889467e-06,
+      "loss": 0.6386,
+      "step": 10216
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.7545185089111328,
+      "learning_rate": 9.520399039249398e-06,
+      "loss": 0.7527,
+      "step": 10217
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.3548014163970947,
+      "learning_rate": 9.52028635503979e-06,
+      "loss": 0.5557,
+      "step": 10218
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.6167333126068115,
+      "learning_rate": 9.52017365826096e-06,
+      "loss": 0.5761,
+      "step": 10219
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.3634378910064697,
+      "learning_rate": 9.520060948913221e-06,
+      "loss": 0.8247,
+      "step": 10220
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.272085666656494,
+      "learning_rate": 9.519948226996884e-06,
+      "loss": 0.3416,
+      "step": 10221
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.879964828491211,
+      "learning_rate": 9.519835492512266e-06,
+      "loss": 0.783,
+      "step": 10222
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.2655043601989746,
+      "learning_rate": 9.519722745459678e-06,
+      "loss": 0.795,
+      "step": 10223
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.211954355239868,
+      "learning_rate": 9.519609985839435e-06,
+      "loss": 0.7946,
+      "step": 10224
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 4.118817329406738,
+      "learning_rate": 9.51949721365185e-06,
+      "loss": 1.034,
+      "step": 10225
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.055813789367676,
+      "learning_rate": 9.519384428897235e-06,
+      "loss": 0.79,
+      "step": 10226
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 4.543420314788818,
+      "learning_rate": 9.519271631575905e-06,
+      "loss": 0.6429,
+      "step": 10227
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.1513326168060303,
+      "learning_rate": 9.519158821688173e-06,
+      "loss": 0.6185,
+      "step": 10228
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 4.310660362243652,
+      "learning_rate": 9.519045999234356e-06,
+      "loss": 0.7881,
+      "step": 10229
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.1917898654937744,
+      "learning_rate": 9.518933164214762e-06,
+      "loss": 0.6605,
+      "step": 10230
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.1354589462280273,
+      "learning_rate": 9.518820316629709e-06,
+      "loss": 0.9641,
+      "step": 10231
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.7010698318481445,
+      "learning_rate": 9.51870745647951e-06,
+      "loss": 0.6513,
+      "step": 10232
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.0489232540130615,
+      "learning_rate": 9.518594583764476e-06,
+      "loss": 0.4811,
+      "step": 10233
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.69371497631073,
+      "learning_rate": 9.518481698484925e-06,
+      "loss": 0.7464,
+      "step": 10234
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.3133318424224854,
+      "learning_rate": 9.518368800641169e-06,
+      "loss": 0.5235,
+      "step": 10235
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.474661350250244,
+      "learning_rate": 9.518255890233522e-06,
+      "loss": 0.7157,
+      "step": 10236
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.7597529888153076,
+      "learning_rate": 9.518142967262297e-06,
+      "loss": 0.6633,
+      "step": 10237
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.5403823852539062,
+      "learning_rate": 9.51803003172781e-06,
+      "loss": 0.8053,
+      "step": 10238
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 5.553660869598389,
+      "learning_rate": 9.517917083630374e-06,
+      "loss": 0.4491,
+      "step": 10239
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 49.85354995727539,
+      "learning_rate": 9.517804122970302e-06,
+      "loss": 0.7997,
+      "step": 10240
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.5812958478927612,
+      "learning_rate": 9.51769114974791e-06,
+      "loss": 0.6306,
+      "step": 10241
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 4.403132438659668,
+      "learning_rate": 9.517578163963512e-06,
+      "loss": 0.7713,
+      "step": 10242
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.787087082862854,
+      "learning_rate": 9.517465165617419e-06,
+      "loss": 0.5314,
+      "step": 10243
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.515662431716919,
+      "learning_rate": 9.51735215470995e-06,
+      "loss": 0.6138,
+      "step": 10244
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.6485254764556885,
+      "learning_rate": 9.517239131241415e-06,
+      "loss": 0.7418,
+      "step": 10245
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.182870864868164,
+      "learning_rate": 9.517126095212133e-06,
+      "loss": 0.8352,
+      "step": 10246
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.0776526927948,
+      "learning_rate": 9.517013046622412e-06,
+      "loss": 0.5542,
+      "step": 10247
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.44179105758667,
+      "learning_rate": 9.516899985472572e-06,
+      "loss": 0.7845,
+      "step": 10248
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.1015686988830566,
+      "learning_rate": 9.516786911762926e-06,
+      "loss": 0.6631,
+      "step": 10249
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.501368761062622,
+      "learning_rate": 9.516673825493785e-06,
+      "loss": 0.8641,
+      "step": 10250
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.245772123336792,
+      "learning_rate": 9.516560726665468e-06,
+      "loss": 0.5625,
+      "step": 10251
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.735985279083252,
+      "learning_rate": 9.516447615278286e-06,
+      "loss": 0.5497,
+      "step": 10252
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.9705100059509277,
+      "learning_rate": 9.516334491332558e-06,
+      "loss": 0.5196,
+      "step": 10253
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.4611356258392334,
+      "learning_rate": 9.516221354828593e-06,
+      "loss": 0.6719,
+      "step": 10254
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.8032175302505493,
+      "learning_rate": 9.516108205766709e-06,
+      "loss": 0.9581,
+      "step": 10255
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.415888786315918,
+      "learning_rate": 9.51599504414722e-06,
+      "loss": 0.7177,
+      "step": 10256
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.7843017578125,
+      "learning_rate": 9.51588186997044e-06,
+      "loss": 0.7131,
+      "step": 10257
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.8552448749542236,
+      "learning_rate": 9.515768683236684e-06,
+      "loss": 0.6848,
+      "step": 10258
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.2281501293182373,
+      "learning_rate": 9.515655483946269e-06,
+      "loss": 0.4816,
+      "step": 10259
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.8828301429748535,
+      "learning_rate": 9.515542272099504e-06,
+      "loss": 0.768,
+      "step": 10260
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.931317090988159,
+      "learning_rate": 9.515429047696711e-06,
+      "loss": 0.5209,
+      "step": 10261
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.128697395324707,
+      "learning_rate": 9.5153158107382e-06,
+      "loss": 0.6406,
+      "step": 10262
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.806093692779541,
+      "learning_rate": 9.515202561224287e-06,
+      "loss": 0.7825,
+      "step": 10263
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.931646466255188,
+      "learning_rate": 9.515089299155287e-06,
+      "loss": 0.5854,
+      "step": 10264
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.1700761318206787,
+      "learning_rate": 9.514976024531516e-06,
+      "loss": 0.4378,
+      "step": 10265
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.8790807723999023,
+      "learning_rate": 9.514862737353287e-06,
+      "loss": 0.8985,
+      "step": 10266
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.1623659133911133,
+      "learning_rate": 9.514749437620917e-06,
+      "loss": 0.7314,
+      "step": 10267
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.8806407451629639,
+      "learning_rate": 9.51463612533472e-06,
+      "loss": 0.611,
+      "step": 10268
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.435960531234741,
+      "learning_rate": 9.51452280049501e-06,
+      "loss": 0.5659,
+      "step": 10269
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.426346778869629,
+      "learning_rate": 9.514409463102104e-06,
+      "loss": 0.8791,
+      "step": 10270
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 5.642628192901611,
+      "learning_rate": 9.514296113156316e-06,
+      "loss": 0.4091,
+      "step": 10271
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.1208841800689697,
+      "learning_rate": 9.514182750657964e-06,
+      "loss": 0.5832,
+      "step": 10272
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.362509250640869,
+      "learning_rate": 9.514069375607358e-06,
+      "loss": 0.7112,
+      "step": 10273
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.154269218444824,
+      "learning_rate": 9.513955988004816e-06,
+      "loss": 0.7869,
+      "step": 10274
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.3268849849700928,
+      "learning_rate": 9.513842587850654e-06,
+      "loss": 0.8865,
+      "step": 10275
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.8052048683166504,
+      "learning_rate": 9.513729175145188e-06,
+      "loss": 0.76,
+      "step": 10276
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.0169272422790527,
+      "learning_rate": 9.51361574988873e-06,
+      "loss": 0.6555,
+      "step": 10277
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.9675045013427734,
+      "learning_rate": 9.513502312081598e-06,
+      "loss": 0.432,
+      "step": 10278
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.073096752166748,
+      "learning_rate": 9.513388861724107e-06,
+      "loss": 0.7488,
+      "step": 10279
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.2706825733184814,
+      "learning_rate": 9.513275398816572e-06,
+      "loss": 0.6486,
+      "step": 10280
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.27449631690979,
+      "learning_rate": 9.51316192335931e-06,
+      "loss": 0.3867,
+      "step": 10281
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.3465182781219482,
+      "learning_rate": 9.513048435352636e-06,
+      "loss": 0.6995,
+      "step": 10282
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.8659071922302246,
+      "learning_rate": 9.512934934796863e-06,
+      "loss": 0.7986,
+      "step": 10283
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.0711684226989746,
+      "learning_rate": 9.512821421692309e-06,
+      "loss": 0.8983,
+      "step": 10284
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.7802085876464844,
+      "learning_rate": 9.51270789603929e-06,
+      "loss": 0.5867,
+      "step": 10285
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.1581106185913086,
+      "learning_rate": 9.512594357838121e-06,
+      "loss": 0.7016,
+      "step": 10286
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 4.94696044921875,
+      "learning_rate": 9.512480807089116e-06,
+      "loss": 0.9501,
+      "step": 10287
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.8669962882995605,
+      "learning_rate": 9.512367243792593e-06,
+      "loss": 0.4791,
+      "step": 10288
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.965036153793335,
+      "learning_rate": 9.512253667948868e-06,
+      "loss": 0.6281,
+      "step": 10289
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.4726932048797607,
+      "learning_rate": 9.512140079558253e-06,
+      "loss": 0.6184,
+      "step": 10290
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.106729745864868,
+      "learning_rate": 9.51202647862107e-06,
+      "loss": 0.4883,
+      "step": 10291
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.9862715005874634,
+      "learning_rate": 9.511912865137632e-06,
+      "loss": 0.6484,
+      "step": 10292
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.6336441040039062,
+      "learning_rate": 9.511799239108251e-06,
+      "loss": 0.8719,
+      "step": 10293
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.700315475463867,
+      "learning_rate": 9.511685600533249e-06,
+      "loss": 0.8465,
+      "step": 10294
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.1529736518859863,
+      "learning_rate": 9.511571949412938e-06,
+      "loss": 0.5666,
+      "step": 10295
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.8053480386734009,
+      "learning_rate": 9.511458285747638e-06,
+      "loss": 0.7337,
+      "step": 10296
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 6.374292850494385,
+      "learning_rate": 9.51134460953766e-06,
+      "loss": 0.5859,
+      "step": 10297
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.772294282913208,
+      "learning_rate": 9.511230920783323e-06,
+      "loss": 0.6289,
+      "step": 10298
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.021082639694214,
+      "learning_rate": 9.511117219484945e-06,
+      "loss": 0.8303,
+      "step": 10299
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.944436550140381,
+      "learning_rate": 9.511003505642837e-06,
+      "loss": 0.706,
+      "step": 10300
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.270374059677124,
+      "learning_rate": 9.51088977925732e-06,
+      "loss": 0.6898,
+      "step": 10301
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.7226914167404175,
+      "learning_rate": 9.510776040328707e-06,
+      "loss": 0.4535,
+      "step": 10302
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 7.392149925231934,
+      "learning_rate": 9.510662288857315e-06,
+      "loss": 0.5957,
+      "step": 10303
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.420355796813965,
+      "learning_rate": 9.510548524843464e-06,
+      "loss": 0.8197,
+      "step": 10304
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.5383379459381104,
+      "learning_rate": 9.510434748287465e-06,
+      "loss": 0.8026,
+      "step": 10305
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.119243860244751,
+      "learning_rate": 9.510320959189636e-06,
+      "loss": 0.7881,
+      "step": 10306
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.1628525257110596,
+      "learning_rate": 9.510207157550295e-06,
+      "loss": 0.5648,
+      "step": 10307
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.6038259267807007,
+      "learning_rate": 9.510093343369757e-06,
+      "loss": 0.7055,
+      "step": 10308
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.8321571350097656,
+      "learning_rate": 9.50997951664834e-06,
+      "loss": 0.6203,
+      "step": 10309
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.668952226638794,
+      "learning_rate": 9.509865677386359e-06,
+      "loss": 0.8564,
+      "step": 10310
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.3299078941345215,
+      "learning_rate": 9.509751825584131e-06,
+      "loss": 0.6646,
+      "step": 10311
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.9908604621887207,
+      "learning_rate": 9.509637961241971e-06,
+      "loss": 0.7608,
+      "step": 10312
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.2300233840942383,
+      "learning_rate": 9.5095240843602e-06,
+      "loss": 0.7362,
+      "step": 10313
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.420527935028076,
+      "learning_rate": 9.50941019493913e-06,
+      "loss": 0.8043,
+      "step": 10314
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.5351274013519287,
+      "learning_rate": 9.50929629297908e-06,
+      "loss": 0.6023,
+      "step": 10315
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.848787784576416,
+      "learning_rate": 9.509182378480367e-06,
+      "loss": 0.7018,
+      "step": 10316
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.160438060760498,
+      "learning_rate": 9.509068451443305e-06,
+      "loss": 0.8707,
+      "step": 10317
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 6.201223373413086,
+      "learning_rate": 9.508954511868213e-06,
+      "loss": 0.6385,
+      "step": 10318
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.9545360803604126,
+      "learning_rate": 9.50884055975541e-06,
+      "loss": 0.6846,
+      "step": 10319
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.961760997772217,
+      "learning_rate": 9.508726595105208e-06,
+      "loss": 0.6046,
+      "step": 10320
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.0439562797546387,
+      "learning_rate": 9.508612617917926e-06,
+      "loss": 0.6738,
+      "step": 10321
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.143430471420288,
+      "learning_rate": 9.508498628193884e-06,
+      "loss": 0.6551,
+      "step": 10322
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 4.21409797668457,
+      "learning_rate": 9.508384625933395e-06,
+      "loss": 0.4598,
+      "step": 10323
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.1265804767608643,
+      "learning_rate": 9.508270611136777e-06,
+      "loss": 0.6587,
+      "step": 10324
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.8297059535980225,
+      "learning_rate": 9.508156583804348e-06,
+      "loss": 0.9868,
+      "step": 10325
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.2555105686187744,
+      "learning_rate": 9.508042543936423e-06,
+      "loss": 0.8019,
+      "step": 10326
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.9843708276748657,
+      "learning_rate": 9.507928491533322e-06,
+      "loss": 0.4399,
+      "step": 10327
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.6290197372436523,
+      "learning_rate": 9.50781442659536e-06,
+      "loss": 0.8199,
+      "step": 10328
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 5.200539588928223,
+      "learning_rate": 9.507700349122853e-06,
+      "loss": 0.7214,
+      "step": 10329
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.888765573501587,
+      "learning_rate": 9.507586259116122e-06,
+      "loss": 0.5778,
+      "step": 10330
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.176344394683838,
+      "learning_rate": 9.507472156575482e-06,
+      "loss": 0.7183,
+      "step": 10331
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.3764026165008545,
+      "learning_rate": 9.507358041501248e-06,
+      "loss": 0.4857,
+      "step": 10332
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.867663860321045,
+      "learning_rate": 9.507243913893742e-06,
+      "loss": 0.4332,
+      "step": 10333
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.928310513496399,
+      "learning_rate": 9.50712977375328e-06,
+      "loss": 0.6716,
+      "step": 10334
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.9104784727096558,
+      "learning_rate": 9.507015621080178e-06,
+      "loss": 0.5729,
+      "step": 10335
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.0910027027130127,
+      "learning_rate": 9.506901455874753e-06,
+      "loss": 0.7586,
+      "step": 10336
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.863759994506836,
+      "learning_rate": 9.506787278137324e-06,
+      "loss": 0.5477,
+      "step": 10337
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.3780248165130615,
+      "learning_rate": 9.506673087868207e-06,
+      "loss": 0.8567,
+      "step": 10338
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.9899587631225586,
+      "learning_rate": 9.50655888506772e-06,
+      "loss": 0.7846,
+      "step": 10339
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.6999050378799438,
+      "learning_rate": 9.506444669736183e-06,
+      "loss": 0.7408,
+      "step": 10340
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.3911757469177246,
+      "learning_rate": 9.50633044187391e-06,
+      "loss": 0.7514,
+      "step": 10341
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.1921043395996094,
+      "learning_rate": 9.506216201481221e-06,
+      "loss": 0.7261,
+      "step": 10342
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.4939348697662354,
+      "learning_rate": 9.506101948558433e-06,
+      "loss": 0.5089,
+      "step": 10343
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.0382766723632812,
+      "learning_rate": 9.505987683105861e-06,
+      "loss": 0.78,
+      "step": 10344
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.5794427394866943,
+      "learning_rate": 9.505873405123828e-06,
+      "loss": 0.6852,
+      "step": 10345
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.7733997106552124,
+      "learning_rate": 9.505759114612649e-06,
+      "loss": 0.4798,
+      "step": 10346
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.064375877380371,
+      "learning_rate": 9.505644811572641e-06,
+      "loss": 0.7769,
+      "step": 10347
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.9131062030792236,
+      "learning_rate": 9.505530496004121e-06,
+      "loss": 0.7006,
+      "step": 10348
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.6857311725616455,
+      "learning_rate": 9.505416167907412e-06,
+      "loss": 0.6192,
+      "step": 10349
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 4.562222003936768,
+      "learning_rate": 9.505301827282825e-06,
+      "loss": 0.7058,
+      "step": 10350
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.0739879608154297,
+      "learning_rate": 9.505187474130683e-06,
+      "loss": 0.8744,
+      "step": 10351
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.6175358295440674,
+      "learning_rate": 9.505073108451301e-06,
+      "loss": 0.5715,
+      "step": 10352
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.5604963302612305,
+      "learning_rate": 9.504958730245001e-06,
+      "loss": 0.8323,
+      "step": 10353
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.2892323732376099,
+      "learning_rate": 9.504844339512096e-06,
+      "loss": 0.5534,
+      "step": 10354
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.2611584663391113,
+      "learning_rate": 9.504729936252908e-06,
+      "loss": 0.5948,
+      "step": 10355
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.148409366607666,
+      "learning_rate": 9.504615520467753e-06,
+      "loss": 0.9746,
+      "step": 10356
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.804096221923828,
+      "learning_rate": 9.504501092156949e-06,
+      "loss": 0.9079,
+      "step": 10357
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.234635829925537,
+      "learning_rate": 9.504386651320816e-06,
+      "loss": 0.7388,
+      "step": 10358
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.048332929611206,
+      "learning_rate": 9.50427219795967e-06,
+      "loss": 0.599,
+      "step": 10359
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.5090014934539795,
+      "learning_rate": 9.50415773207383e-06,
+      "loss": 0.7143,
+      "step": 10360
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.300133228302002,
+      "learning_rate": 9.504043253663617e-06,
+      "loss": 0.6395,
+      "step": 10361
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.483245372772217,
+      "learning_rate": 9.503928762729344e-06,
+      "loss": 0.7421,
+      "step": 10362
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.503690481185913,
+      "learning_rate": 9.503814259271334e-06,
+      "loss": 0.8884,
+      "step": 10363
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.19749116897583,
+      "learning_rate": 9.503699743289904e-06,
+      "loss": 0.7744,
+      "step": 10364
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 5.015937328338623,
+      "learning_rate": 9.50358521478537e-06,
+      "loss": 0.6535,
+      "step": 10365
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.33366060256958,
+      "learning_rate": 9.503470673758056e-06,
+      "loss": 0.7493,
+      "step": 10366
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.4582059383392334,
+      "learning_rate": 9.503356120208276e-06,
+      "loss": 0.6411,
+      "step": 10367
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.9939601421356201,
+      "learning_rate": 9.503241554136348e-06,
+      "loss": 0.6768,
+      "step": 10368
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.8593411445617676,
+      "learning_rate": 9.503126975542594e-06,
+      "loss": 0.5342,
+      "step": 10369
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.671372413635254,
+      "learning_rate": 9.50301238442733e-06,
+      "loss": 0.8521,
+      "step": 10370
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.6840384006500244,
+      "learning_rate": 9.502897780790875e-06,
+      "loss": 0.6474,
+      "step": 10371
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.2515957355499268,
+      "learning_rate": 9.502783164633549e-06,
+      "loss": 0.5749,
+      "step": 10372
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.4579503536224365,
+      "learning_rate": 9.502668535955668e-06,
+      "loss": 0.6537,
+      "step": 10373
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.398876428604126,
+      "learning_rate": 9.502553894757554e-06,
+      "loss": 0.7186,
+      "step": 10374
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.7991513013839722,
+      "learning_rate": 9.502439241039524e-06,
+      "loss": 0.6408,
+      "step": 10375
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.408522367477417,
+      "learning_rate": 9.502324574801897e-06,
+      "loss": 0.465,
+      "step": 10376
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 4.181957721710205,
+      "learning_rate": 9.502209896044992e-06,
+      "loss": 0.7553,
+      "step": 10377
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.617824077606201,
+      "learning_rate": 9.502095204769129e-06,
+      "loss": 0.648,
+      "step": 10378
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.6615511178970337,
+      "learning_rate": 9.501980500974625e-06,
+      "loss": 0.7852,
+      "step": 10379
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.5429993867874146,
+      "learning_rate": 9.501865784661799e-06,
+      "loss": 0.8459,
+      "step": 10380
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.7779524326324463,
+      "learning_rate": 9.501751055830973e-06,
+      "loss": 0.6222,
+      "step": 10381
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.334580898284912,
+      "learning_rate": 9.501636314482462e-06,
+      "loss": 0.7448,
+      "step": 10382
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 4.302098751068115,
+      "learning_rate": 9.501521560616584e-06,
+      "loss": 1.1006,
+      "step": 10383
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.4145541191101074,
+      "learning_rate": 9.501406794233665e-06,
+      "loss": 0.8395,
+      "step": 10384
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.768483281135559,
+      "learning_rate": 9.501292015334018e-06,
+      "loss": 0.6646,
+      "step": 10385
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.6435892581939697,
+      "learning_rate": 9.501177223917964e-06,
+      "loss": 0.7252,
+      "step": 10386
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.5014662742614746,
+      "learning_rate": 9.501062419985824e-06,
+      "loss": 0.874,
+      "step": 10387
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.8542028665542603,
+      "learning_rate": 9.500947603537915e-06,
+      "loss": 0.6246,
+      "step": 10388
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 5.303249359130859,
+      "learning_rate": 9.500832774574555e-06,
+      "loss": 0.671,
+      "step": 10389
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 4.120805740356445,
+      "learning_rate": 9.500717933096067e-06,
+      "loss": 0.9322,
+      "step": 10390
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.079190254211426,
+      "learning_rate": 9.500603079102769e-06,
+      "loss": 0.6078,
+      "step": 10391
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.800370693206787,
+      "learning_rate": 9.500488212594977e-06,
+      "loss": 0.6887,
+      "step": 10392
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.594526529312134,
+      "learning_rate": 9.500373333573016e-06,
+      "loss": 0.7235,
+      "step": 10393
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.4160168170928955,
+      "learning_rate": 9.500258442037201e-06,
+      "loss": 0.6429,
+      "step": 10394
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.9811166524887085,
+      "learning_rate": 9.500143537987854e-06,
+      "loss": 0.5709,
+      "step": 10395
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.243696451187134,
+      "learning_rate": 9.500028621425292e-06,
+      "loss": 0.6006,
+      "step": 10396
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.0417072772979736,
+      "learning_rate": 9.499913692349838e-06,
+      "loss": 0.4513,
+      "step": 10397
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.086174964904785,
+      "learning_rate": 9.499798750761808e-06,
+      "loss": 0.6266,
+      "step": 10398
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.101405620574951,
+      "learning_rate": 9.499683796661526e-06,
+      "loss": 0.7105,
+      "step": 10399
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.3123583793640137,
+      "learning_rate": 9.499568830049305e-06,
+      "loss": 0.6928,
+      "step": 10400
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.0929958820343018,
+      "learning_rate": 9.499453850925473e-06,
+      "loss": 0.8067,
+      "step": 10401
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.5037612915039062,
+      "learning_rate": 9.499338859290344e-06,
+      "loss": 0.5718,
+      "step": 10402
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.6613402366638184,
+      "learning_rate": 9.499223855144236e-06,
+      "loss": 0.6362,
+      "step": 10403
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.2597360610961914,
+      "learning_rate": 9.499108838487476e-06,
+      "loss": 0.6404,
+      "step": 10404
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.490504264831543,
+      "learning_rate": 9.498993809320378e-06,
+      "loss": 0.7198,
+      "step": 10405
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.002176523208618,
+      "learning_rate": 9.498878767643264e-06,
+      "loss": 0.4391,
+      "step": 10406
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.5406653881073,
+      "learning_rate": 9.498763713456453e-06,
+      "loss": 0.5226,
+      "step": 10407
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.188713788986206,
+      "learning_rate": 9.498648646760268e-06,
+      "loss": 0.685,
+      "step": 10408
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.3441269397735596,
+      "learning_rate": 9.498533567555022e-06,
+      "loss": 0.5824,
+      "step": 10409
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.5617988109588623,
+      "learning_rate": 9.49841847584104e-06,
+      "loss": 0.6158,
+      "step": 10410
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.21574068069458,
+      "learning_rate": 9.498303371618645e-06,
+      "loss": 0.4933,
+      "step": 10411
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.615159034729004,
+      "learning_rate": 9.49818825488815e-06,
+      "loss": 0.6376,
+      "step": 10412
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.4264140129089355,
+      "learning_rate": 9.49807312564988e-06,
+      "loss": 0.7135,
+      "step": 10413
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.5353585481643677,
+      "learning_rate": 9.497957983904154e-06,
+      "loss": 0.6046,
+      "step": 10414
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.824878454208374,
+      "learning_rate": 9.49784282965129e-06,
+      "loss": 0.6881,
+      "step": 10415
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.282379150390625,
+      "learning_rate": 9.497727662891611e-06,
+      "loss": 0.8288,
+      "step": 10416
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.2112462520599365,
+      "learning_rate": 9.497612483625436e-06,
+      "loss": 0.8155,
+      "step": 10417
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.376997470855713,
+      "learning_rate": 9.497497291853086e-06,
+      "loss": 0.822,
+      "step": 10418
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.933297872543335,
+      "learning_rate": 9.497382087574881e-06,
+      "loss": 0.6388,
+      "step": 10419
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.3402457237243652,
+      "learning_rate": 9.497266870791142e-06,
+      "loss": 0.5295,
+      "step": 10420
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.111476182937622,
+      "learning_rate": 9.497151641502187e-06,
+      "loss": 0.6608,
+      "step": 10421
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.257914662361145,
+      "learning_rate": 9.497036399708338e-06,
+      "loss": 0.5166,
+      "step": 10422
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 6.7440409660339355,
+      "learning_rate": 9.496921145409917e-06,
+      "loss": 0.3699,
+      "step": 10423
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.193127155303955,
+      "learning_rate": 9.496805878607241e-06,
+      "loss": 0.8367,
+      "step": 10424
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.5150694847106934,
+      "learning_rate": 9.496690599300633e-06,
+      "loss": 0.8314,
+      "step": 10425
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.6196986436843872,
+      "learning_rate": 9.496575307490414e-06,
+      "loss": 0.6873,
+      "step": 10426
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.4718201160430908,
+      "learning_rate": 9.496460003176902e-06,
+      "loss": 0.7953,
+      "step": 10427
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.8933509588241577,
+      "learning_rate": 9.496344686360419e-06,
+      "loss": 0.6367,
+      "step": 10428
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.1450400352478027,
+      "learning_rate": 9.496229357041287e-06,
+      "loss": 0.5912,
+      "step": 10429
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.621860980987549,
+      "learning_rate": 9.496114015219826e-06,
+      "loss": 0.9477,
+      "step": 10430
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.8403124809265137,
+      "learning_rate": 9.495998660896354e-06,
+      "loss": 0.4967,
+      "step": 10431
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.95809268951416,
+      "learning_rate": 9.495883294071195e-06,
+      "loss": 0.8351,
+      "step": 10432
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.505676746368408,
+      "learning_rate": 9.495767914744669e-06,
+      "loss": 0.5886,
+      "step": 10433
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.592207670211792,
+      "learning_rate": 9.495652522917096e-06,
+      "loss": 0.7267,
+      "step": 10434
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.2845561504364014,
+      "learning_rate": 9.495537118588797e-06,
+      "loss": 0.6413,
+      "step": 10435
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.2147603034973145,
+      "learning_rate": 9.495421701760096e-06,
+      "loss": 0.7306,
+      "step": 10436
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.7021124362945557,
+      "learning_rate": 9.495306272431309e-06,
+      "loss": 0.3803,
+      "step": 10437
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.1840126514434814,
+      "learning_rate": 9.49519083060276e-06,
+      "loss": 0.9254,
+      "step": 10438
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.9184470176696777,
+      "learning_rate": 9.49507537627477e-06,
+      "loss": 0.9336,
+      "step": 10439
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.8395369052886963,
+      "learning_rate": 9.494959909447657e-06,
+      "loss": 0.615,
+      "step": 10440
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.4750802516937256,
+      "learning_rate": 9.494844430121746e-06,
+      "loss": 0.7804,
+      "step": 10441
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.634310722351074,
+      "learning_rate": 9.494728938297355e-06,
+      "loss": 0.7112,
+      "step": 10442
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.695814609527588,
+      "learning_rate": 9.49461343397481e-06,
+      "loss": 0.8218,
+      "step": 10443
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.9067130088806152,
+      "learning_rate": 9.494497917154425e-06,
+      "loss": 0.8006,
+      "step": 10444
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.595428705215454,
+      "learning_rate": 9.494382387836528e-06,
+      "loss": 0.8245,
+      "step": 10445
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.6662044525146484,
+      "learning_rate": 9.494266846021434e-06,
+      "loss": 0.7429,
+      "step": 10446
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.2339529991149902,
+      "learning_rate": 9.494151291709469e-06,
+      "loss": 1.0599,
+      "step": 10447
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.0537731647491455,
+      "learning_rate": 9.494035724900952e-06,
+      "loss": 0.9421,
+      "step": 10448
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.6656476259231567,
+      "learning_rate": 9.493920145596207e-06,
+      "loss": 0.3389,
+      "step": 10449
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.625241994857788,
+      "learning_rate": 9.493804553795553e-06,
+      "loss": 0.7624,
+      "step": 10450
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.1076619625091553,
+      "learning_rate": 9.493688949499311e-06,
+      "loss": 0.6868,
+      "step": 10451
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.2036454677581787,
+      "learning_rate": 9.493573332707804e-06,
+      "loss": 0.7279,
+      "step": 10452
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 10.36037540435791,
+      "learning_rate": 9.493457703421354e-06,
+      "loss": 0.7283,
+      "step": 10453
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.5440095663070679,
+      "learning_rate": 9.49334206164028e-06,
+      "loss": 0.401,
+      "step": 10454
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.805882692337036,
+      "learning_rate": 9.493226407364904e-06,
+      "loss": 0.8654,
+      "step": 10455
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.186142683029175,
+      "learning_rate": 9.49311074059555e-06,
+      "loss": 0.7581,
+      "step": 10456
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 4.451480388641357,
+      "learning_rate": 9.492995061332538e-06,
+      "loss": 0.5594,
+      "step": 10457
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.8687106370925903,
+      "learning_rate": 9.49287936957619e-06,
+      "loss": 0.7004,
+      "step": 10458
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.432910919189453,
+      "learning_rate": 9.492763665326827e-06,
+      "loss": 0.5917,
+      "step": 10459
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.528226375579834,
+      "learning_rate": 9.492647948584772e-06,
+      "loss": 0.6466,
+      "step": 10460
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.74302077293396,
+      "learning_rate": 9.492532219350346e-06,
+      "loss": 0.7246,
+      "step": 10461
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.810833215713501,
+      "learning_rate": 9.49241647762387e-06,
+      "loss": 0.6003,
+      "step": 10462
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.9010322093963623,
+      "learning_rate": 9.492300723405667e-06,
+      "loss": 0.7779,
+      "step": 10463
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.7741286754608154,
+      "learning_rate": 9.49218495669606e-06,
+      "loss": 0.7336,
+      "step": 10464
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.482110619544983,
+      "learning_rate": 9.492069177495368e-06,
+      "loss": 0.6465,
+      "step": 10465
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.1729047298431396,
+      "learning_rate": 9.491953385803914e-06,
+      "loss": 0.646,
+      "step": 10466
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.6947133541107178,
+      "learning_rate": 9.491837581622021e-06,
+      "loss": 0.8336,
+      "step": 10467
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.2784225940704346,
+      "learning_rate": 9.49172176495001e-06,
+      "loss": 0.8651,
+      "step": 10468
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.88940691947937,
+      "learning_rate": 9.491605935788205e-06,
+      "loss": 0.9838,
+      "step": 10469
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 4.155715465545654,
+      "learning_rate": 9.491490094136924e-06,
+      "loss": 0.7592,
+      "step": 10470
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.193814992904663,
+      "learning_rate": 9.491374239996493e-06,
+      "loss": 0.958,
+      "step": 10471
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.3698832988739014,
+      "learning_rate": 9.491258373367234e-06,
+      "loss": 0.6225,
+      "step": 10472
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 4.237252235412598,
+      "learning_rate": 9.491142494249465e-06,
+      "loss": 0.8915,
+      "step": 10473
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.8968802690505981,
+      "learning_rate": 9.491026602643512e-06,
+      "loss": 0.5072,
+      "step": 10474
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.1828198432922363,
+      "learning_rate": 9.490910698549698e-06,
+      "loss": 0.4926,
+      "step": 10475
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.5994880199432373,
+      "learning_rate": 9.490794781968343e-06,
+      "loss": 1.005,
+      "step": 10476
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.412193536758423,
+      "learning_rate": 9.490678852899769e-06,
+      "loss": 0.7237,
+      "step": 10477
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.8580392599105835,
+      "learning_rate": 9.490562911344301e-06,
+      "loss": 0.5076,
+      "step": 10478
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 7.076424598693848,
+      "learning_rate": 9.490446957302257e-06,
+      "loss": 0.6465,
+      "step": 10479
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.935369849205017,
+      "learning_rate": 9.490330990773966e-06,
+      "loss": 0.5548,
+      "step": 10480
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.0231149196624756,
+      "learning_rate": 9.490215011759743e-06,
+      "loss": 0.7482,
+      "step": 10481
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.504882335662842,
+      "learning_rate": 9.490099020259917e-06,
+      "loss": 0.4971,
+      "step": 10482
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.697723627090454,
+      "learning_rate": 9.489983016274807e-06,
+      "loss": 0.7432,
+      "step": 10483
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.9430259466171265,
+      "learning_rate": 9.489866999804734e-06,
+      "loss": 0.7236,
+      "step": 10484
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.2045929431915283,
+      "learning_rate": 9.489750970850025e-06,
+      "loss": 0.746,
+      "step": 10485
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.476693630218506,
+      "learning_rate": 9.489634929411e-06,
+      "loss": 0.8136,
+      "step": 10486
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.81790030002594,
+      "learning_rate": 9.489518875487981e-06,
+      "loss": 0.7588,
+      "step": 10487
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.253382682800293,
+      "learning_rate": 9.489402809081294e-06,
+      "loss": 0.7125,
+      "step": 10488
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.384922742843628,
+      "learning_rate": 9.489286730191258e-06,
+      "loss": 0.7058,
+      "step": 10489
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.1513445377349854,
+      "learning_rate": 9.489170638818197e-06,
+      "loss": 0.7642,
+      "step": 10490
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.460270404815674,
+      "learning_rate": 9.489054534962436e-06,
+      "loss": 0.879,
+      "step": 10491
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.929008960723877,
+      "learning_rate": 9.488938418624293e-06,
+      "loss": 0.5694,
+      "step": 10492
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.274059295654297,
+      "learning_rate": 9.488822289804096e-06,
+      "loss": 0.6359,
+      "step": 10493
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.543063163757324,
+      "learning_rate": 9.488706148502166e-06,
+      "loss": 0.4981,
+      "step": 10494
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.915736436843872,
+      "learning_rate": 9.488589994718826e-06,
+      "loss": 0.6043,
+      "step": 10495
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.706517457962036,
+      "learning_rate": 9.488473828454396e-06,
+      "loss": 0.8195,
+      "step": 10496
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.6324901580810547,
+      "learning_rate": 9.488357649709205e-06,
+      "loss": 0.5911,
+      "step": 10497
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.227588415145874,
+      "learning_rate": 9.488241458483571e-06,
+      "loss": 0.6912,
+      "step": 10498
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.3750642538070679,
+      "learning_rate": 9.488125254777819e-06,
+      "loss": 0.5422,
+      "step": 10499
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.6910104751586914,
+      "learning_rate": 9.488009038592272e-06,
+      "loss": 0.6802,
+      "step": 10500
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.7488882541656494,
+      "learning_rate": 9.487892809927252e-06,
+      "loss": 0.5742,
+      "step": 10501
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 4.097071170806885,
+      "learning_rate": 9.487776568783085e-06,
+      "loss": 0.8579,
+      "step": 10502
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.474846839904785,
+      "learning_rate": 9.487660315160092e-06,
+      "loss": 0.5953,
+      "step": 10503
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.5823605060577393,
+      "learning_rate": 9.487544049058597e-06,
+      "loss": 0.6868,
+      "step": 10504
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.570481061935425,
+      "learning_rate": 9.487427770478923e-06,
+      "loss": 0.6097,
+      "step": 10505
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.3169522285461426,
+      "learning_rate": 9.487311479421393e-06,
+      "loss": 0.5517,
+      "step": 10506
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.8925031423568726,
+      "learning_rate": 9.487195175886333e-06,
+      "loss": 0.8891,
+      "step": 10507
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.226954936981201,
+      "learning_rate": 9.48707885987406e-06,
+      "loss": 0.5143,
+      "step": 10508
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.6002063751220703,
+      "learning_rate": 9.486962531384905e-06,
+      "loss": 0.547,
+      "step": 10509
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.3541698455810547,
+      "learning_rate": 9.486846190419186e-06,
+      "loss": 0.6789,
+      "step": 10510
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.158748149871826,
+      "learning_rate": 9.486729836977229e-06,
+      "loss": 0.7091,
+      "step": 10511
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.5441279411315918,
+      "learning_rate": 9.486613471059358e-06,
+      "loss": 0.6346,
+      "step": 10512
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.4577157497406006,
+      "learning_rate": 9.486497092665896e-06,
+      "loss": 0.5559,
+      "step": 10513
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.1816794872283936,
+      "learning_rate": 9.486380701797164e-06,
+      "loss": 0.6685,
+      "step": 10514
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.3222496509552,
+      "learning_rate": 9.48626429845349e-06,
+      "loss": 0.819,
+      "step": 10515
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 8.13864517211914,
+      "learning_rate": 9.486147882635195e-06,
+      "loss": 0.8749,
+      "step": 10516
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.0724995136260986,
+      "learning_rate": 9.486031454342603e-06,
+      "loss": 0.699,
+      "step": 10517
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.3345272541046143,
+      "learning_rate": 9.485915013576039e-06,
+      "loss": 0.6885,
+      "step": 10518
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.204432964324951,
+      "learning_rate": 9.485798560335825e-06,
+      "loss": 0.4868,
+      "step": 10519
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.6585631370544434,
+      "learning_rate": 9.485682094622283e-06,
+      "loss": 0.6421,
+      "step": 10520
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.4183330535888672,
+      "learning_rate": 9.485565616435744e-06,
+      "loss": 0.5621,
+      "step": 10521
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.9073290824890137,
+      "learning_rate": 9.485449125776525e-06,
+      "loss": 0.7191,
+      "step": 10522
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.1241607666015625,
+      "learning_rate": 9.485332622644953e-06,
+      "loss": 0.6539,
+      "step": 10523
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.5909345149993896,
+      "learning_rate": 9.48521610704135e-06,
+      "loss": 0.7397,
+      "step": 10524
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.4115231037139893,
+      "learning_rate": 9.485099578966043e-06,
+      "loss": 0.6895,
+      "step": 10525
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 4.10917329788208,
+      "learning_rate": 9.484983038419353e-06,
+      "loss": 0.7587,
+      "step": 10526
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.818233609199524,
+      "learning_rate": 9.484866485401605e-06,
+      "loss": 0.4316,
+      "step": 10527
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.9067022800445557,
+      "learning_rate": 9.484749919913126e-06,
+      "loss": 0.7811,
+      "step": 10528
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.83923077583313,
+      "learning_rate": 9.484633341954235e-06,
+      "loss": 0.8048,
+      "step": 10529
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.7465736865997314,
+      "learning_rate": 9.484516751525259e-06,
+      "loss": 0.7978,
+      "step": 10530
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.629798412322998,
+      "learning_rate": 9.484400148626522e-06,
+      "loss": 0.745,
+      "step": 10531
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.0552148818969727,
+      "learning_rate": 9.484283533258347e-06,
+      "loss": 0.6378,
+      "step": 10532
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.3016107082366943,
+      "learning_rate": 9.484166905421061e-06,
+      "loss": 0.6205,
+      "step": 10533
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 4.502875804901123,
+      "learning_rate": 9.484050265114986e-06,
+      "loss": 0.5547,
+      "step": 10534
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.292834520339966,
+      "learning_rate": 9.483933612340447e-06,
+      "loss": 0.4869,
+      "step": 10535
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.6882078647613525,
+      "learning_rate": 9.483816947097768e-06,
+      "loss": 0.7612,
+      "step": 10536
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.6806468963623047,
+      "learning_rate": 9.483700269387274e-06,
+      "loss": 0.6964,
+      "step": 10537
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.4076740741729736,
+      "learning_rate": 9.48358357920929e-06,
+      "loss": 0.8176,
+      "step": 10538
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.49035382270813,
+      "learning_rate": 9.48346687656414e-06,
+      "loss": 0.7933,
+      "step": 10539
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.7301900386810303,
+      "learning_rate": 9.483350161452147e-06,
+      "loss": 0.7926,
+      "step": 10540
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.2806546688079834,
+      "learning_rate": 9.483233433873636e-06,
+      "loss": 0.5357,
+      "step": 10541
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.2213289737701416,
+      "learning_rate": 9.483116693828934e-06,
+      "loss": 0.6361,
+      "step": 10542
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.4364349842071533,
+      "learning_rate": 9.482999941318363e-06,
+      "loss": 0.669,
+      "step": 10543
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.8092615604400635,
+      "learning_rate": 9.482883176342248e-06,
+      "loss": 0.6901,
+      "step": 10544
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 6.30463981628418,
+      "learning_rate": 9.482766398900916e-06,
+      "loss": 0.8011,
+      "step": 10545
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.8606741428375244,
+      "learning_rate": 9.482649608994687e-06,
+      "loss": 0.725,
+      "step": 10546
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.8304851055145264,
+      "learning_rate": 9.48253280662389e-06,
+      "loss": 0.5535,
+      "step": 10547
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 4.486219882965088,
+      "learning_rate": 9.482415991788849e-06,
+      "loss": 0.8196,
+      "step": 10548
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.1686971187591553,
+      "learning_rate": 9.482299164489888e-06,
+      "loss": 0.7912,
+      "step": 10549
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.3265955448150635,
+      "learning_rate": 9.482182324727334e-06,
+      "loss": 0.5748,
+      "step": 10550
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.249286651611328,
+      "learning_rate": 9.482065472501507e-06,
+      "loss": 0.6168,
+      "step": 10551
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.506998300552368,
+      "learning_rate": 9.481948607812736e-06,
+      "loss": 0.6093,
+      "step": 10552
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.735158681869507,
+      "learning_rate": 9.481831730661347e-06,
+      "loss": 0.5538,
+      "step": 10553
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.8518626689910889,
+      "learning_rate": 9.48171484104766e-06,
+      "loss": 0.4567,
+      "step": 10554
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.6638877391815186,
+      "learning_rate": 9.481597938972003e-06,
+      "loss": 0.51,
+      "step": 10555
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.6486951112747192,
+      "learning_rate": 9.481481024434702e-06,
+      "loss": 0.6935,
+      "step": 10556
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.420889139175415,
+      "learning_rate": 9.48136409743608e-06,
+      "loss": 0.6434,
+      "step": 10557
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.9606144428253174,
+      "learning_rate": 9.481247157976466e-06,
+      "loss": 0.6222,
+      "step": 10558
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.891723155975342,
+      "learning_rate": 9.481130206056179e-06,
+      "loss": 0.5941,
+      "step": 10559
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.8966915607452393,
+      "learning_rate": 9.481013241675549e-06,
+      "loss": 0.7579,
+      "step": 10560
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.986267566680908,
+      "learning_rate": 9.480896264834902e-06,
+      "loss": 0.7018,
+      "step": 10561
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.1986083984375,
+      "learning_rate": 9.480779275534558e-06,
+      "loss": 0.7503,
+      "step": 10562
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.808972120285034,
+      "learning_rate": 9.480662273774847e-06,
+      "loss": 0.5198,
+      "step": 10563
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.321810245513916,
+      "learning_rate": 9.48054525955609e-06,
+      "loss": 0.6686,
+      "step": 10564
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.280142307281494,
+      "learning_rate": 9.480428232878619e-06,
+      "loss": 0.5829,
+      "step": 10565
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.247267246246338,
+      "learning_rate": 9.480311193742752e-06,
+      "loss": 0.4983,
+      "step": 10566
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.376124382019043,
+      "learning_rate": 9.480194142148822e-06,
+      "loss": 0.5825,
+      "step": 10567
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 5.58064079284668,
+      "learning_rate": 9.480077078097147e-06,
+      "loss": 0.666,
+      "step": 10568
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.1733834743499756,
+      "learning_rate": 9.479960001588056e-06,
+      "loss": 0.5261,
+      "step": 10569
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.072023630142212,
+      "learning_rate": 9.479842912621876e-06,
+      "loss": 0.776,
+      "step": 10570
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.9361355304718018,
+      "learning_rate": 9.47972581119893e-06,
+      "loss": 0.4803,
+      "step": 10571
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.2251259088516235,
+      "learning_rate": 9.479608697319545e-06,
+      "loss": 0.4045,
+      "step": 10572
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.362170696258545,
+      "learning_rate": 9.479491570984047e-06,
+      "loss": 0.6945,
+      "step": 10573
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.613978624343872,
+      "learning_rate": 9.47937443219276e-06,
+      "loss": 0.4395,
+      "step": 10574
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.5516645908355713,
+      "learning_rate": 9.479257280946012e-06,
+      "loss": 0.6384,
+      "step": 10575
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 8.010658264160156,
+      "learning_rate": 9.479140117244125e-06,
+      "loss": 0.4525,
+      "step": 10576
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.427335500717163,
+      "learning_rate": 9.47902294108743e-06,
+      "loss": 0.6083,
+      "step": 10577
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 4.257616996765137,
+      "learning_rate": 9.478905752476249e-06,
+      "loss": 0.784,
+      "step": 10578
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.8000601530075073,
+      "learning_rate": 9.478788551410909e-06,
+      "loss": 0.6186,
+      "step": 10579
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.6824159622192383,
+      "learning_rate": 9.478671337891734e-06,
+      "loss": 0.5333,
+      "step": 10580
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.6375234127044678,
+      "learning_rate": 9.478554111919053e-06,
+      "loss": 0.6951,
+      "step": 10581
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.847238302230835,
+      "learning_rate": 9.478436873493193e-06,
+      "loss": 0.4386,
+      "step": 10582
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.156665563583374,
+      "learning_rate": 9.478319622614476e-06,
+      "loss": 0.6831,
+      "step": 10583
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.3432726860046387,
+      "learning_rate": 9.478202359283228e-06,
+      "loss": 0.698,
+      "step": 10584
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.038853406906128,
+      "learning_rate": 9.47808508349978e-06,
+      "loss": 0.6983,
+      "step": 10585
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.2927896976470947,
+      "learning_rate": 9.477967795264453e-06,
+      "loss": 0.5537,
+      "step": 10586
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.9252958297729492,
+      "learning_rate": 9.477850494577577e-06,
+      "loss": 0.7918,
+      "step": 10587
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.0408592224121094,
+      "learning_rate": 9.477733181439473e-06,
+      "loss": 0.5028,
+      "step": 10588
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.558305263519287,
+      "learning_rate": 9.477615855850472e-06,
+      "loss": 0.8052,
+      "step": 10589
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.5939443111419678,
+      "learning_rate": 9.477498517810898e-06,
+      "loss": 0.7087,
+      "step": 10590
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.64555823802948,
+      "learning_rate": 9.477381167321079e-06,
+      "loss": 0.8415,
+      "step": 10591
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.283090353012085,
+      "learning_rate": 9.47726380438134e-06,
+      "loss": 0.7052,
+      "step": 10592
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.804222822189331,
+      "learning_rate": 9.477146428992008e-06,
+      "loss": 0.6281,
+      "step": 10593
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.833308219909668,
+      "learning_rate": 9.477029041153408e-06,
+      "loss": 0.7567,
+      "step": 10594
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.4484379291534424,
+      "learning_rate": 9.476911640865867e-06,
+      "loss": 0.6534,
+      "step": 10595
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.8758794069290161,
+      "learning_rate": 9.476794228129712e-06,
+      "loss": 0.7162,
+      "step": 10596
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.8968820571899414,
+      "learning_rate": 9.47667680294527e-06,
+      "loss": 0.8954,
+      "step": 10597
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.4656617641448975,
+      "learning_rate": 9.476559365312867e-06,
+      "loss": 0.5716,
+      "step": 10598
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.4578897953033447,
+      "learning_rate": 9.47644191523283e-06,
+      "loss": 0.7851,
+      "step": 10599
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.156003952026367,
+      "learning_rate": 9.476324452705483e-06,
+      "loss": 0.6401,
+      "step": 10600
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.3754547834396362,
+      "learning_rate": 9.476206977731157e-06,
+      "loss": 0.3109,
+      "step": 10601
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.1903343200683594,
+      "learning_rate": 9.476089490310174e-06,
+      "loss": 0.8339,
+      "step": 10602
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.9593040943145752,
+      "learning_rate": 9.475971990442865e-06,
+      "loss": 0.8441,
+      "step": 10603
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.614734172821045,
+      "learning_rate": 9.475854478129552e-06,
+      "loss": 0.6535,
+      "step": 10604
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.805114984512329,
+      "learning_rate": 9.475736953370566e-06,
+      "loss": 0.6973,
+      "step": 10605
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.253253221511841,
+      "learning_rate": 9.475619416166232e-06,
+      "loss": 0.7418,
+      "step": 10606
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.024226427078247,
+      "learning_rate": 9.475501866516876e-06,
+      "loss": 0.6378,
+      "step": 10607
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.6736257076263428,
+      "learning_rate": 9.475384304422828e-06,
+      "loss": 0.6439,
+      "step": 10608
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.075942039489746,
+      "learning_rate": 9.475266729884412e-06,
+      "loss": 0.6699,
+      "step": 10609
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.391895294189453,
+      "learning_rate": 9.475149142901954e-06,
+      "loss": 0.8658,
+      "step": 10610
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.5941162109375,
+      "learning_rate": 9.475031543475783e-06,
+      "loss": 0.9937,
+      "step": 10611
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.4991254806518555,
+      "learning_rate": 9.474913931606228e-06,
+      "loss": 0.5868,
+      "step": 10612
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.7235450744628906,
+      "learning_rate": 9.474796307293613e-06,
+      "loss": 0.751,
+      "step": 10613
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.502690076828003,
+      "learning_rate": 9.474678670538265e-06,
+      "loss": 0.9297,
+      "step": 10614
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.117034912109375,
+      "learning_rate": 9.474561021340511e-06,
+      "loss": 0.7278,
+      "step": 10615
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.0165743827819824,
+      "learning_rate": 9.47444335970068e-06,
+      "loss": 0.595,
+      "step": 10616
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.2153115272521973,
+      "learning_rate": 9.474325685619098e-06,
+      "loss": 0.5396,
+      "step": 10617
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 4.912493705749512,
+      "learning_rate": 9.474207999096092e-06,
+      "loss": 0.8788,
+      "step": 10618
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.3082833290100098,
+      "learning_rate": 9.47409030013199e-06,
+      "loss": 0.8721,
+      "step": 10619
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.5185935497283936,
+      "learning_rate": 9.473972588727119e-06,
+      "loss": 0.5702,
+      "step": 10620
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.5127086639404297,
+      "learning_rate": 9.473854864881807e-06,
+      "loss": 0.7025,
+      "step": 10621
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.956839084625244,
+      "learning_rate": 9.47373712859638e-06,
+      "loss": 0.7561,
+      "step": 10622
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.2030770778656006,
+      "learning_rate": 9.473619379871165e-06,
+      "loss": 0.7327,
+      "step": 10623
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.5370371341705322,
+      "learning_rate": 9.473501618706491e-06,
+      "loss": 0.6038,
+      "step": 10624
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.8276759386062622,
+      "learning_rate": 9.473383845102686e-06,
+      "loss": 0.3491,
+      "step": 10625
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.0350236892700195,
+      "learning_rate": 9.473266059060075e-06,
+      "loss": 0.6491,
+      "step": 10626
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.315904378890991,
+      "learning_rate": 9.473148260578985e-06,
+      "loss": 0.5204,
+      "step": 10627
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.904785633087158,
+      "learning_rate": 9.473030449659748e-06,
+      "loss": 1.0385,
+      "step": 10628
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.370518445968628,
+      "learning_rate": 9.47291262630269e-06,
+      "loss": 0.5909,
+      "step": 10629
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.937161922454834,
+      "learning_rate": 9.472794790508135e-06,
+      "loss": 0.6367,
+      "step": 10630
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.5351500511169434,
+      "learning_rate": 9.472676942276414e-06,
+      "loss": 0.6626,
+      "step": 10631
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.7795634269714355,
+      "learning_rate": 9.472559081607854e-06,
+      "loss": 0.6111,
+      "step": 10632
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.8793718814849854,
+      "learning_rate": 9.472441208502782e-06,
+      "loss": 0.7272,
+      "step": 10633
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.801804780960083,
+      "learning_rate": 9.472323322961529e-06,
+      "loss": 0.8387,
+      "step": 10634
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.262141466140747,
+      "learning_rate": 9.472205424984418e-06,
+      "loss": 0.6688,
+      "step": 10635
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.085623264312744,
+      "learning_rate": 9.47208751457178e-06,
+      "loss": 0.7472,
+      "step": 10636
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 4.778904438018799,
+      "learning_rate": 9.471969591723941e-06,
+      "loss": 0.5973,
+      "step": 10637
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 5.78367280960083,
+      "learning_rate": 9.47185165644123e-06,
+      "loss": 0.6712,
+      "step": 10638
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.3375236988067627,
+      "learning_rate": 9.471733708723974e-06,
+      "loss": 0.6463,
+      "step": 10639
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.392666816711426,
+      "learning_rate": 9.471615748572503e-06,
+      "loss": 0.8561,
+      "step": 10640
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.9072295427322388,
+      "learning_rate": 9.471497775987143e-06,
+      "loss": 0.7081,
+      "step": 10641
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 4.207918643951416,
+      "learning_rate": 9.471379790968222e-06,
+      "loss": 0.7693,
+      "step": 10642
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.4026310443878174,
+      "learning_rate": 9.471261793516071e-06,
+      "loss": 0.6052,
+      "step": 10643
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.1570310592651367,
+      "learning_rate": 9.471143783631016e-06,
+      "loss": 0.4315,
+      "step": 10644
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.210540294647217,
+      "learning_rate": 9.471025761313382e-06,
+      "loss": 0.7823,
+      "step": 10645
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.9383751153945923,
+      "learning_rate": 9.470907726563504e-06,
+      "loss": 0.4808,
+      "step": 10646
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.9483840465545654,
+      "learning_rate": 9.470789679381704e-06,
+      "loss": 0.7675,
+      "step": 10647
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.5032174587249756,
+      "learning_rate": 9.470671619768312e-06,
+      "loss": 0.7494,
+      "step": 10648
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 4.21278190612793,
+      "learning_rate": 9.470553547723659e-06,
+      "loss": 0.7648,
+      "step": 10649
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.8184905052185059,
+      "learning_rate": 9.47043546324807e-06,
+      "loss": 0.7466,
+      "step": 10650
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.203277349472046,
+      "learning_rate": 9.470317366341875e-06,
+      "loss": 0.8857,
+      "step": 10651
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.295179605484009,
+      "learning_rate": 9.470199257005402e-06,
+      "loss": 0.7414,
+      "step": 10652
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.7792396545410156,
+      "learning_rate": 9.47008113523898e-06,
+      "loss": 0.6746,
+      "step": 10653
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.3513123989105225,
+      "learning_rate": 9.469963001042936e-06,
+      "loss": 0.835,
+      "step": 10654
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 4.185072898864746,
+      "learning_rate": 9.4698448544176e-06,
+      "loss": 0.7325,
+      "step": 10655
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.9845689535140991,
+      "learning_rate": 9.4697266953633e-06,
+      "loss": 0.7334,
+      "step": 10656
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.6052888631820679,
+      "learning_rate": 9.469608523880364e-06,
+      "loss": 0.5507,
+      "step": 10657
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.221104621887207,
+      "learning_rate": 9.46949033996912e-06,
+      "loss": 0.5356,
+      "step": 10658
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.444990396499634,
+      "learning_rate": 9.469372143629901e-06,
+      "loss": 0.6962,
+      "step": 10659
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.8339579105377197,
+      "learning_rate": 9.46925393486303e-06,
+      "loss": 0.5978,
+      "step": 10660
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.721719980239868,
+      "learning_rate": 9.469135713668837e-06,
+      "loss": 0.7151,
+      "step": 10661
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.5940771102905273,
+      "learning_rate": 9.469017480047654e-06,
+      "loss": 0.7309,
+      "step": 10662
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.9468930959701538,
+      "learning_rate": 9.468899233999807e-06,
+      "loss": 0.6238,
+      "step": 10663
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.756399631500244,
+      "learning_rate": 9.468780975525625e-06,
+      "loss": 0.7264,
+      "step": 10664
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 4.417372226715088,
+      "learning_rate": 9.468662704625437e-06,
+      "loss": 0.7588,
+      "step": 10665
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.5723319053649902,
+      "learning_rate": 9.468544421299571e-06,
+      "loss": 0.5689,
+      "step": 10666
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.280410051345825,
+      "learning_rate": 9.468426125548359e-06,
+      "loss": 0.8115,
+      "step": 10667
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.927154302597046,
+      "learning_rate": 9.468307817372126e-06,
+      "loss": 0.5623,
+      "step": 10668
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.7397496700286865,
+      "learning_rate": 9.468189496771203e-06,
+      "loss": 0.6273,
+      "step": 10669
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.8308863639831543,
+      "learning_rate": 9.46807116374592e-06,
+      "loss": 0.5759,
+      "step": 10670
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 4.466609954833984,
+      "learning_rate": 9.467952818296606e-06,
+      "loss": 0.8123,
+      "step": 10671
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.8583920001983643,
+      "learning_rate": 9.467834460423585e-06,
+      "loss": 0.5462,
+      "step": 10672
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 4.207177639007568,
+      "learning_rate": 9.467716090127194e-06,
+      "loss": 0.98,
+      "step": 10673
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.2886695861816406,
+      "learning_rate": 9.467597707407756e-06,
+      "loss": 0.6599,
+      "step": 10674
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.123368978500366,
+      "learning_rate": 9.467479312265604e-06,
+      "loss": 0.6037,
+      "step": 10675
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.4392318725585938,
+      "learning_rate": 9.467360904701064e-06,
+      "loss": 0.9256,
+      "step": 10676
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.1341896057128906,
+      "learning_rate": 9.467242484714466e-06,
+      "loss": 0.7135,
+      "step": 10677
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 6.029886722564697,
+      "learning_rate": 9.467124052306143e-06,
+      "loss": 0.7405,
+      "step": 10678
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.9384899139404297,
+      "learning_rate": 9.467005607476419e-06,
+      "loss": 0.9115,
+      "step": 10679
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.5434274673461914,
+      "learning_rate": 9.466887150225627e-06,
+      "loss": 0.5614,
+      "step": 10680
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.9237064123153687,
+      "learning_rate": 9.466768680554095e-06,
+      "loss": 0.7289,
+      "step": 10681
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.7422375679016113,
+      "learning_rate": 9.466650198462153e-06,
+      "loss": 0.5957,
+      "step": 10682
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 4.249253749847412,
+      "learning_rate": 9.466531703950129e-06,
+      "loss": 0.634,
+      "step": 10683
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.219738006591797,
+      "learning_rate": 9.466413197018354e-06,
+      "loss": 0.4323,
+      "step": 10684
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.6698689460754395,
+      "learning_rate": 9.466294677667157e-06,
+      "loss": 0.8492,
+      "step": 10685
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.999068021774292,
+      "learning_rate": 9.466176145896868e-06,
+      "loss": 0.4856,
+      "step": 10686
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.1698074340820312,
+      "learning_rate": 9.466057601707816e-06,
+      "loss": 0.6178,
+      "step": 10687
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.247678756713867,
+      "learning_rate": 9.46593904510033e-06,
+      "loss": 0.4556,
+      "step": 10688
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.7483644485473633,
+      "learning_rate": 9.465820476074742e-06,
+      "loss": 0.6456,
+      "step": 10689
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.591893434524536,
+      "learning_rate": 9.46570189463138e-06,
+      "loss": 0.8205,
+      "step": 10690
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.172982692718506,
+      "learning_rate": 9.465583300770573e-06,
+      "loss": 0.8466,
+      "step": 10691
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.189426898956299,
+      "learning_rate": 9.465464694492652e-06,
+      "loss": 0.5903,
+      "step": 10692
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.0947060585021973,
+      "learning_rate": 9.465346075797946e-06,
+      "loss": 0.5376,
+      "step": 10693
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.1681792736053467,
+      "learning_rate": 9.465227444686786e-06,
+      "loss": 0.6922,
+      "step": 10694
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.7460262775421143,
+      "learning_rate": 9.465108801159502e-06,
+      "loss": 0.8315,
+      "step": 10695
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.9647703170776367,
+      "learning_rate": 9.464990145216422e-06,
+      "loss": 0.6212,
+      "step": 10696
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.2755417823791504,
+      "learning_rate": 9.464871476857876e-06,
+      "loss": 0.7241,
+      "step": 10697
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 4.345125675201416,
+      "learning_rate": 9.464752796084197e-06,
+      "loss": 1.241,
+      "step": 10698
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.5738613605499268,
+      "learning_rate": 9.464634102895711e-06,
+      "loss": 0.7883,
+      "step": 10699
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.3104299306869507,
+      "learning_rate": 9.464515397292752e-06,
+      "loss": 0.8492,
+      "step": 10700
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 4.024679660797119,
+      "learning_rate": 9.464396679275647e-06,
+      "loss": 0.8014,
+      "step": 10701
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.363654375076294,
+      "learning_rate": 9.464277948844728e-06,
+      "loss": 0.6587,
+      "step": 10702
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.6236326694488525,
+      "learning_rate": 9.464159206000325e-06,
+      "loss": 0.8295,
+      "step": 10703
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.6451902389526367,
+      "learning_rate": 9.464040450742764e-06,
+      "loss": 0.9713,
+      "step": 10704
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 4.3640923500061035,
+      "learning_rate": 9.463921683072383e-06,
+      "loss": 0.6372,
+      "step": 10705
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.3409489393234253,
+      "learning_rate": 9.463802902989505e-06,
+      "loss": 0.8204,
+      "step": 10706
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.3247175216674805,
+      "learning_rate": 9.463684110494466e-06,
+      "loss": 0.7914,
+      "step": 10707
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.8910285234451294,
+      "learning_rate": 9.463565305587593e-06,
+      "loss": 0.6841,
+      "step": 10708
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 7.850323677062988,
+      "learning_rate": 9.463446488269214e-06,
+      "loss": 0.4944,
+      "step": 10709
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.8099967241287231,
+      "learning_rate": 9.463327658539665e-06,
+      "loss": 0.7494,
+      "step": 10710
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.183399200439453,
+      "learning_rate": 9.463208816399274e-06,
+      "loss": 0.7646,
+      "step": 10711
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.849496841430664,
+      "learning_rate": 9.46308996184837e-06,
+      "loss": 0.6372,
+      "step": 10712
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.195448637008667,
+      "learning_rate": 9.462971094887285e-06,
+      "loss": 0.9749,
+      "step": 10713
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.0502355098724365,
+      "learning_rate": 9.46285221551635e-06,
+      "loss": 0.5443,
+      "step": 10714
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.195547580718994,
+      "learning_rate": 9.462733323735896e-06,
+      "loss": 0.7006,
+      "step": 10715
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.4923148155212402,
+      "learning_rate": 9.46261441954625e-06,
+      "loss": 0.8278,
+      "step": 10716
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.9256699085235596,
+      "learning_rate": 9.462495502947746e-06,
+      "loss": 0.6107,
+      "step": 10717
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.5365333557128906,
+      "learning_rate": 9.462376573940713e-06,
+      "loss": 0.6444,
+      "step": 10718
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.328930616378784,
+      "learning_rate": 9.462257632525482e-06,
+      "loss": 0.7463,
+      "step": 10719
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.195544958114624,
+      "learning_rate": 9.462138678702385e-06,
+      "loss": 0.7315,
+      "step": 10720
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.94962739944458,
+      "learning_rate": 9.462019712471752e-06,
+      "loss": 0.6573,
+      "step": 10721
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.412020206451416,
+      "learning_rate": 9.461900733833916e-06,
+      "loss": 0.6778,
+      "step": 10722
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.9522459506988525,
+      "learning_rate": 9.461781742789202e-06,
+      "loss": 0.7868,
+      "step": 10723
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.0279855728149414,
+      "learning_rate": 9.461662739337945e-06,
+      "loss": 0.5565,
+      "step": 10724
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.338468074798584,
+      "learning_rate": 9.461543723480476e-06,
+      "loss": 0.7419,
+      "step": 10725
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.5203135013580322,
+      "learning_rate": 9.461424695217126e-06,
+      "loss": 0.5438,
+      "step": 10726
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.161494255065918,
+      "learning_rate": 9.461305654548223e-06,
+      "loss": 0.4394,
+      "step": 10727
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.201827049255371,
+      "learning_rate": 9.461186601474101e-06,
+      "loss": 0.8262,
+      "step": 10728
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.3167407512664795,
+      "learning_rate": 9.461067535995091e-06,
+      "loss": 0.7367,
+      "step": 10729
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.7664002180099487,
+      "learning_rate": 9.460948458111524e-06,
+      "loss": 0.6872,
+      "step": 10730
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.0983426570892334,
+      "learning_rate": 9.460829367823729e-06,
+      "loss": 0.9759,
+      "step": 10731
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.840078353881836,
+      "learning_rate": 9.46071026513204e-06,
+      "loss": 0.7228,
+      "step": 10732
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.765059232711792,
+      "learning_rate": 9.460591150036785e-06,
+      "loss": 0.6901,
+      "step": 10733
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 4.475542068481445,
+      "learning_rate": 9.460472022538299e-06,
+      "loss": 0.7247,
+      "step": 10734
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.733582019805908,
+      "learning_rate": 9.460352882636909e-06,
+      "loss": 0.6235,
+      "step": 10735
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 5.639922618865967,
+      "learning_rate": 9.46023373033295e-06,
+      "loss": 0.7367,
+      "step": 10736
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.5625661611557007,
+      "learning_rate": 9.460114565626752e-06,
+      "loss": 0.7755,
+      "step": 10737
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.3997714519500732,
+      "learning_rate": 9.459995388518645e-06,
+      "loss": 0.7645,
+      "step": 10738
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.0463643074035645,
+      "learning_rate": 9.459876199008963e-06,
+      "loss": 0.7938,
+      "step": 10739
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.4105167388916016,
+      "learning_rate": 9.459756997098036e-06,
+      "loss": 0.8191,
+      "step": 10740
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.5874297618865967,
+      "learning_rate": 9.459637782786194e-06,
+      "loss": 0.5196,
+      "step": 10741
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.967484474182129,
+      "learning_rate": 9.45951855607377e-06,
+      "loss": 0.636,
+      "step": 10742
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.609935998916626,
+      "learning_rate": 9.459399316961096e-06,
+      "loss": 0.8863,
+      "step": 10743
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.639068841934204,
+      "learning_rate": 9.459280065448504e-06,
+      "loss": 0.6483,
+      "step": 10744
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.930962562561035,
+      "learning_rate": 9.459160801536324e-06,
+      "loss": 0.6891,
+      "step": 10745
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.359109401702881,
+      "learning_rate": 9.459041525224887e-06,
+      "loss": 0.5623,
+      "step": 10746
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.9155690670013428,
+      "learning_rate": 9.458922236514527e-06,
+      "loss": 0.6427,
+      "step": 10747
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.747655153274536,
+      "learning_rate": 9.458802935405576e-06,
+      "loss": 0.7281,
+      "step": 10748
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.127436637878418,
+      "learning_rate": 9.458683621898361e-06,
+      "loss": 0.6669,
+      "step": 10749
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.5615854263305664,
+      "learning_rate": 9.45856429599322e-06,
+      "loss": 0.8044,
+      "step": 10750
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.411273956298828,
+      "learning_rate": 9.45844495769048e-06,
+      "loss": 0.5286,
+      "step": 10751
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.8037726879119873,
+      "learning_rate": 9.458325606990477e-06,
+      "loss": 0.5699,
+      "step": 10752
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.644188642501831,
+      "learning_rate": 9.458206243893539e-06,
+      "loss": 0.8429,
+      "step": 10753
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.1681931018829346,
+      "learning_rate": 9.458086868399999e-06,
+      "loss": 0.8116,
+      "step": 10754
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.7696008682250977,
+      "learning_rate": 9.457967480510191e-06,
+      "loss": 0.5848,
+      "step": 10755
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.320373296737671,
+      "learning_rate": 9.457848080224446e-06,
+      "loss": 0.8519,
+      "step": 10756
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.407341241836548,
+      "learning_rate": 9.457728667543092e-06,
+      "loss": 0.6641,
+      "step": 10757
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.8487441539764404,
+      "learning_rate": 9.457609242466468e-06,
+      "loss": 0.7145,
+      "step": 10758
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 4.021374702453613,
+      "learning_rate": 9.4574898049949e-06,
+      "loss": 0.8268,
+      "step": 10759
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.3669660091400146,
+      "learning_rate": 9.457370355128724e-06,
+      "loss": 0.7185,
+      "step": 10760
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.610190749168396,
+      "learning_rate": 9.45725089286827e-06,
+      "loss": 0.6554,
+      "step": 10761
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 4.234034538269043,
+      "learning_rate": 9.457131418213872e-06,
+      "loss": 0.5961,
+      "step": 10762
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 4.526187896728516,
+      "learning_rate": 9.457011931165861e-06,
+      "loss": 0.6118,
+      "step": 10763
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.05942964553833,
+      "learning_rate": 9.456892431724569e-06,
+      "loss": 0.4601,
+      "step": 10764
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.271235227584839,
+      "learning_rate": 9.45677291989033e-06,
+      "loss": 0.6914,
+      "step": 10765
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.2148115634918213,
+      "learning_rate": 9.456653395663472e-06,
+      "loss": 0.4734,
+      "step": 10766
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.056448221206665,
+      "learning_rate": 9.456533859044334e-06,
+      "loss": 0.674,
+      "step": 10767
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.785046339035034,
+      "learning_rate": 9.456414310033244e-06,
+      "loss": 0.6982,
+      "step": 10768
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.2926337718963623,
+      "learning_rate": 9.456294748630534e-06,
+      "loss": 0.7425,
+      "step": 10769
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.295605421066284,
+      "learning_rate": 9.456175174836538e-06,
+      "loss": 0.4441,
+      "step": 10770
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.6186575889587402,
+      "learning_rate": 9.456055588651589e-06,
+      "loss": 0.7899,
+      "step": 10771
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.5259044170379639,
+      "learning_rate": 9.455935990076017e-06,
+      "loss": 0.4678,
+      "step": 10772
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.866652727127075,
+      "learning_rate": 9.455816379110157e-06,
+      "loss": 0.641,
+      "step": 10773
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.2640397548675537,
+      "learning_rate": 9.455696755754341e-06,
+      "loss": 0.6833,
+      "step": 10774
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.7247512340545654,
+      "learning_rate": 9.455577120008904e-06,
+      "loss": 0.479,
+      "step": 10775
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.4343714714050293,
+      "learning_rate": 9.455457471874173e-06,
+      "loss": 0.6915,
+      "step": 10776
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.299705743789673,
+      "learning_rate": 9.455337811350484e-06,
+      "loss": 0.6876,
+      "step": 10777
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 5.307137489318848,
+      "learning_rate": 9.455218138438169e-06,
+      "loss": 0.8657,
+      "step": 10778
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.406400442123413,
+      "learning_rate": 9.455098453137563e-06,
+      "loss": 0.9496,
+      "step": 10779
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.458646059036255,
+      "learning_rate": 9.454978755448995e-06,
+      "loss": 0.4588,
+      "step": 10780
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 5.042985916137695,
+      "learning_rate": 9.454859045372803e-06,
+      "loss": 0.7692,
+      "step": 10781
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.279721260070801,
+      "learning_rate": 9.454739322909315e-06,
+      "loss": 0.6705,
+      "step": 10782
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 4.307494163513184,
+      "learning_rate": 9.454619588058866e-06,
+      "loss": 0.8204,
+      "step": 10783
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.2554612159729004,
+      "learning_rate": 9.45449984082179e-06,
+      "loss": 0.7413,
+      "step": 10784
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.327019214630127,
+      "learning_rate": 9.454380081198415e-06,
+      "loss": 0.6673,
+      "step": 10785
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.0631420612335205,
+      "learning_rate": 9.45426030918908e-06,
+      "loss": 0.8915,
+      "step": 10786
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.1597702503204346,
+      "learning_rate": 9.454140524794117e-06,
+      "loss": 0.3763,
+      "step": 10787
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.3001632690429688,
+      "learning_rate": 9.454020728013857e-06,
+      "loss": 0.7082,
+      "step": 10788
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.6574902534484863,
+      "learning_rate": 9.453900918848633e-06,
+      "loss": 0.699,
+      "step": 10789
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.5819625854492188,
+      "learning_rate": 9.453781097298779e-06,
+      "loss": 0.6958,
+      "step": 10790
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 5.354830741882324,
+      "learning_rate": 9.453661263364628e-06,
+      "loss": 0.8264,
+      "step": 10791
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.608548164367676,
+      "learning_rate": 9.453541417046515e-06,
+      "loss": 0.6289,
+      "step": 10792
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.86495041847229,
+      "learning_rate": 9.45342155834477e-06,
+      "loss": 0.5245,
+      "step": 10793
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.3078722953796387,
+      "learning_rate": 9.453301687259729e-06,
+      "loss": 1.0061,
+      "step": 10794
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.5761091709136963,
+      "learning_rate": 9.453181803791721e-06,
+      "loss": 0.6585,
+      "step": 10795
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 5.415067672729492,
+      "learning_rate": 9.453061907941087e-06,
+      "loss": 0.7043,
+      "step": 10796
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.823923110961914,
+      "learning_rate": 9.452941999708154e-06,
+      "loss": 0.7353,
+      "step": 10797
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.335245370864868,
+      "learning_rate": 9.452822079093257e-06,
+      "loss": 0.7061,
+      "step": 10798
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.6575195789337158,
+      "learning_rate": 9.452702146096731e-06,
+      "loss": 0.5244,
+      "step": 10799
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 5.198136806488037,
+      "learning_rate": 9.452582200718908e-06,
+      "loss": 0.813,
+      "step": 10800
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.058291435241699,
+      "learning_rate": 9.452462242960123e-06,
+      "loss": 0.3956,
+      "step": 10801
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.728337287902832,
+      "learning_rate": 9.452342272820707e-06,
+      "loss": 0.7795,
+      "step": 10802
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.09710693359375,
+      "learning_rate": 9.452222290300994e-06,
+      "loss": 0.7488,
+      "step": 10803
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.6108858585357666,
+      "learning_rate": 9.45210229540132e-06,
+      "loss": 0.6454,
+      "step": 10804
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.7709121704101562,
+      "learning_rate": 9.451982288122017e-06,
+      "loss": 0.468,
+      "step": 10805
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.8526860475540161,
+      "learning_rate": 9.451862268463421e-06,
+      "loss": 0.4854,
+      "step": 10806
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.2947092056274414,
+      "learning_rate": 9.45174223642586e-06,
+      "loss": 0.3207,
+      "step": 10807
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.5121591091156006,
+      "learning_rate": 9.451622192009675e-06,
+      "loss": 0.5903,
+      "step": 10808
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.825718641281128,
+      "learning_rate": 9.451502135215195e-06,
+      "loss": 0.7226,
+      "step": 10809
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.7764503955841064,
+      "learning_rate": 9.451382066042754e-06,
+      "loss": 0.7133,
+      "step": 10810
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.9941954612731934,
+      "learning_rate": 9.451261984492688e-06,
+      "loss": 0.643,
+      "step": 10811
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.3469189405441284,
+      "learning_rate": 9.45114189056533e-06,
+      "loss": 0.5076,
+      "step": 10812
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.6300206184387207,
+      "learning_rate": 9.451021784261015e-06,
+      "loss": 0.8664,
+      "step": 10813
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.3989944458007812,
+      "learning_rate": 9.450901665580073e-06,
+      "loss": 0.3529,
+      "step": 10814
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.033111095428467,
+      "learning_rate": 9.450781534522842e-06,
+      "loss": 0.6705,
+      "step": 10815
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.639739990234375,
+      "learning_rate": 9.450661391089654e-06,
+      "loss": 0.8308,
+      "step": 10816
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.4191677570343018,
+      "learning_rate": 9.450541235280847e-06,
+      "loss": 0.5318,
+      "step": 10817
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.9724866151809692,
+      "learning_rate": 9.450421067096749e-06,
+      "loss": 0.5409,
+      "step": 10818
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 4.562854290008545,
+      "learning_rate": 9.450300886537698e-06,
+      "loss": 0.5954,
+      "step": 10819
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.2501072883605957,
+      "learning_rate": 9.450180693604028e-06,
+      "loss": 0.6846,
+      "step": 10820
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.3175389766693115,
+      "learning_rate": 9.450060488296073e-06,
+      "loss": 0.6802,
+      "step": 10821
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.6622276306152344,
+      "learning_rate": 9.449940270614165e-06,
+      "loss": 0.7399,
+      "step": 10822
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.3728678226470947,
+      "learning_rate": 9.449820040558642e-06,
+      "loss": 0.708,
+      "step": 10823
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.1973819732666016,
+      "learning_rate": 9.449699798129835e-06,
+      "loss": 0.6776,
+      "step": 10824
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.0642406940460205,
+      "learning_rate": 9.44957954332808e-06,
+      "loss": 0.4944,
+      "step": 10825
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.6468260288238525,
+      "learning_rate": 9.449459276153711e-06,
+      "loss": 0.6704,
+      "step": 10826
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.0978267192840576,
+      "learning_rate": 9.449338996607065e-06,
+      "loss": 0.9038,
+      "step": 10827
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.9506523609161377,
+      "learning_rate": 9.449218704688471e-06,
+      "loss": 0.8122,
+      "step": 10828
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.1333682537078857,
+      "learning_rate": 9.449098400398266e-06,
+      "loss": 0.7296,
+      "step": 10829
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.855626106262207,
+      "learning_rate": 9.448978083736788e-06,
+      "loss": 0.9225,
+      "step": 10830
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.9008073806762695,
+      "learning_rate": 9.448857754704366e-06,
+      "loss": 0.4147,
+      "step": 10831
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.161646604537964,
+      "learning_rate": 9.448737413301339e-06,
+      "loss": 0.6776,
+      "step": 10832
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.595844268798828,
+      "learning_rate": 9.448617059528038e-06,
+      "loss": 0.5868,
+      "step": 10833
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.2704598903656006,
+      "learning_rate": 9.448496693384801e-06,
+      "loss": 0.514,
+      "step": 10834
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.6578919887542725,
+      "learning_rate": 9.44837631487196e-06,
+      "loss": 0.6145,
+      "step": 10835
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.816132068634033,
+      "learning_rate": 9.448255923989854e-06,
+      "loss": 0.914,
+      "step": 10836
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.8067936897277832,
+      "learning_rate": 9.44813552073881e-06,
+      "loss": 0.8143,
+      "step": 10837
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.8599244356155396,
+      "learning_rate": 9.44801510511917e-06,
+      "loss": 0.5009,
+      "step": 10838
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.856908917427063,
+      "learning_rate": 9.447894677131267e-06,
+      "loss": 0.8557,
+      "step": 10839
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.3017656803131104,
+      "learning_rate": 9.447774236775433e-06,
+      "loss": 0.8738,
+      "step": 10840
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.607539415359497,
+      "learning_rate": 9.447653784052006e-06,
+      "loss": 0.6542,
+      "step": 10841
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.7786004543304443,
+      "learning_rate": 9.44753331896132e-06,
+      "loss": 0.639,
+      "step": 10842
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.5590696334838867,
+      "learning_rate": 9.44741284150371e-06,
+      "loss": 0.7302,
+      "step": 10843
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.8998444080352783,
+      "learning_rate": 9.44729235167951e-06,
+      "loss": 0.3961,
+      "step": 10844
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.9670826196670532,
+      "learning_rate": 9.447171849489057e-06,
+      "loss": 0.3744,
+      "step": 10845
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.4691996574401855,
+      "learning_rate": 9.447051334932685e-06,
+      "loss": 0.7538,
+      "step": 10846
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.9804043769836426,
+      "learning_rate": 9.446930808010729e-06,
+      "loss": 0.4354,
+      "step": 10847
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.2944865226745605,
+      "learning_rate": 9.446810268723525e-06,
+      "loss": 0.6159,
+      "step": 10848
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.568437337875366,
+      "learning_rate": 9.446689717071405e-06,
+      "loss": 0.655,
+      "step": 10849
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.8644258975982666,
+      "learning_rate": 9.446569153054709e-06,
+      "loss": 0.6179,
+      "step": 10850
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.530625343322754,
+      "learning_rate": 9.446448576673769e-06,
+      "loss": 0.8235,
+      "step": 10851
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.8302133083343506,
+      "learning_rate": 9.446327987928921e-06,
+      "loss": 0.7117,
+      "step": 10852
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.4429070949554443,
+      "learning_rate": 9.446207386820501e-06,
+      "loss": 0.7764,
+      "step": 10853
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 4.436441898345947,
+      "learning_rate": 9.446086773348844e-06,
+      "loss": 0.8528,
+      "step": 10854
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.3259117603302,
+      "learning_rate": 9.445966147514285e-06,
+      "loss": 0.9872,
+      "step": 10855
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.4429690837860107,
+      "learning_rate": 9.44584550931716e-06,
+      "loss": 0.5638,
+      "step": 10856
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.8449393510818481,
+      "learning_rate": 9.445724858757803e-06,
+      "loss": 0.6619,
+      "step": 10857
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.342743396759033,
+      "learning_rate": 9.44560419583655e-06,
+      "loss": 0.6501,
+      "step": 10858
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.5048906803131104,
+      "learning_rate": 9.44548352055374e-06,
+      "loss": 0.8111,
+      "step": 10859
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.4585325717926025,
+      "learning_rate": 9.445362832909703e-06,
+      "loss": 0.7282,
+      "step": 10860
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.0680785179138184,
+      "learning_rate": 9.44524213290478e-06,
+      "loss": 0.6999,
+      "step": 10861
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.8380342721939087,
+      "learning_rate": 9.4451214205393e-06,
+      "loss": 0.7386,
+      "step": 10862
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.4585282802581787,
+      "learning_rate": 9.445000695813606e-06,
+      "loss": 0.6241,
+      "step": 10863
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.6914100646972656,
+      "learning_rate": 9.444879958728029e-06,
+      "loss": 0.6872,
+      "step": 10864
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.1503093242645264,
+      "learning_rate": 9.444759209282906e-06,
+      "loss": 0.9172,
+      "step": 10865
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.171081304550171,
+      "learning_rate": 9.444638447478571e-06,
+      "loss": 0.7518,
+      "step": 10866
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 4.165119647979736,
+      "learning_rate": 9.444517673315364e-06,
+      "loss": 0.8896,
+      "step": 10867
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.6286392211914062,
+      "learning_rate": 9.444396886793617e-06,
+      "loss": 0.5783,
+      "step": 10868
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.088283061981201,
+      "learning_rate": 9.444276087913666e-06,
+      "loss": 0.3427,
+      "step": 10869
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.0636870861053467,
+      "learning_rate": 9.44415527667585e-06,
+      "loss": 0.9554,
+      "step": 10870
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.193281412124634,
+      "learning_rate": 9.444034453080501e-06,
+      "loss": 0.6556,
+      "step": 10871
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.8212504386901855,
+      "learning_rate": 9.44391361712796e-06,
+      "loss": 0.5253,
+      "step": 10872
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.985412120819092,
+      "learning_rate": 9.443792768818558e-06,
+      "loss": 0.7088,
+      "step": 10873
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.6397731304168701,
+      "learning_rate": 9.443671908152634e-06,
+      "loss": 0.5437,
+      "step": 10874
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 5.726080894470215,
+      "learning_rate": 9.443551035130522e-06,
+      "loss": 0.5543,
+      "step": 10875
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.64676833152771,
+      "learning_rate": 9.44343014975256e-06,
+      "loss": 0.5994,
+      "step": 10876
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.201387882232666,
+      "learning_rate": 9.443309252019081e-06,
+      "loss": 0.6475,
+      "step": 10877
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.8255109786987305,
+      "learning_rate": 9.443188341930426e-06,
+      "loss": 0.5586,
+      "step": 10878
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.8089162111282349,
+      "learning_rate": 9.443067419486929e-06,
+      "loss": 0.6123,
+      "step": 10879
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.3066890239715576,
+      "learning_rate": 9.442946484688924e-06,
+      "loss": 0.5199,
+      "step": 10880
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.010655403137207,
+      "learning_rate": 9.44282553753675e-06,
+      "loss": 0.8443,
+      "step": 10881
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.3428754806518555,
+      "learning_rate": 9.442704578030743e-06,
+      "loss": 0.6786,
+      "step": 10882
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.3209245204925537,
+      "learning_rate": 9.442583606171238e-06,
+      "loss": 0.7342,
+      "step": 10883
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.501556634902954,
+      "learning_rate": 9.442462621958572e-06,
+      "loss": 0.3911,
+      "step": 10884
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.8243606090545654,
+      "learning_rate": 9.442341625393083e-06,
+      "loss": 0.6955,
+      "step": 10885
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.744112730026245,
+      "learning_rate": 9.442220616475106e-06,
+      "loss": 0.4546,
+      "step": 10886
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.922027587890625,
+      "learning_rate": 9.442099595204976e-06,
+      "loss": 0.5194,
+      "step": 10887
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.9996984004974365,
+      "learning_rate": 9.441978561583032e-06,
+      "loss": 0.6344,
+      "step": 10888
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.246840238571167,
+      "learning_rate": 9.44185751560961e-06,
+      "loss": 0.6365,
+      "step": 10889
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.8968093395233154,
+      "learning_rate": 9.441736457285046e-06,
+      "loss": 0.8719,
+      "step": 10890
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.566741943359375,
+      "learning_rate": 9.441615386609677e-06,
+      "loss": 0.8025,
+      "step": 10891
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.823715090751648,
+      "learning_rate": 9.44149430358384e-06,
+      "loss": 0.856,
+      "step": 10892
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.8248825073242188,
+      "learning_rate": 9.44137320820787e-06,
+      "loss": 0.6337,
+      "step": 10893
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.573488235473633,
+      "learning_rate": 9.441252100482105e-06,
+      "loss": 0.6381,
+      "step": 10894
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.772270679473877,
+      "learning_rate": 9.441130980406881e-06,
+      "loss": 0.7206,
+      "step": 10895
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.235203266143799,
+      "learning_rate": 9.441009847982537e-06,
+      "loss": 0.7306,
+      "step": 10896
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.0670528411865234,
+      "learning_rate": 9.440888703209408e-06,
+      "loss": 0.8304,
+      "step": 10897
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.562513589859009,
+      "learning_rate": 9.44076754608783e-06,
+      "loss": 0.902,
+      "step": 10898
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.0605130195617676,
+      "learning_rate": 9.440646376618141e-06,
+      "loss": 0.6122,
+      "step": 10899
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.0386691093444824,
+      "learning_rate": 9.440525194800678e-06,
+      "loss": 0.7542,
+      "step": 10900
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.221141815185547,
+      "learning_rate": 9.440404000635779e-06,
+      "loss": 1.0683,
+      "step": 10901
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.97432017326355,
+      "learning_rate": 9.440282794123779e-06,
+      "loss": 0.49,
+      "step": 10902
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.028312921524048,
+      "learning_rate": 9.440161575265016e-06,
+      "loss": 0.6929,
+      "step": 10903
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.532275915145874,
+      "learning_rate": 9.440040344059829e-06,
+      "loss": 0.8667,
+      "step": 10904
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.1880943775177,
+      "learning_rate": 9.43991910050855e-06,
+      "loss": 0.8023,
+      "step": 10905
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.4329127073287964,
+      "learning_rate": 9.43979784461152e-06,
+      "loss": 0.6093,
+      "step": 10906
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.975536584854126,
+      "learning_rate": 9.439676576369076e-06,
+      "loss": 0.6042,
+      "step": 10907
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.5961716175079346,
+      "learning_rate": 9.439555295781554e-06,
+      "loss": 0.6133,
+      "step": 10908
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.311441421508789,
+      "learning_rate": 9.439434002849293e-06,
+      "loss": 0.6103,
+      "step": 10909
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.336646556854248,
+      "learning_rate": 9.43931269757263e-06,
+      "loss": 0.6586,
+      "step": 10910
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.295391321182251,
+      "learning_rate": 9.439191379951899e-06,
+      "loss": 0.9439,
+      "step": 10911
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.913789749145508,
+      "learning_rate": 9.43907004998744e-06,
+      "loss": 0.6349,
+      "step": 10912
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.324530601501465,
+      "learning_rate": 9.438948707679592e-06,
+      "loss": 0.5122,
+      "step": 10913
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.0552773475646973,
+      "learning_rate": 9.438827353028688e-06,
+      "loss": 0.8875,
+      "step": 10914
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.9044663906097412,
+      "learning_rate": 9.438705986035067e-06,
+      "loss": 0.5725,
+      "step": 10915
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.4793386459350586,
+      "learning_rate": 9.438584606699072e-06,
+      "loss": 0.6509,
+      "step": 10916
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.378636598587036,
+      "learning_rate": 9.438463215021033e-06,
+      "loss": 0.5442,
+      "step": 10917
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.9512776136398315,
+      "learning_rate": 9.43834181100129e-06,
+      "loss": 0.8138,
+      "step": 10918
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.3950705528259277,
+      "learning_rate": 9.438220394640182e-06,
+      "loss": 0.6665,
+      "step": 10919
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.5465123653411865,
+      "learning_rate": 9.438098965938046e-06,
+      "loss": 0.6031,
+      "step": 10920
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.5791966915130615,
+      "learning_rate": 9.437977524895218e-06,
+      "loss": 0.7603,
+      "step": 10921
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.861023187637329,
+      "learning_rate": 9.437856071512038e-06,
+      "loss": 0.731,
+      "step": 10922
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.8034024238586426,
+      "learning_rate": 9.437734605788844e-06,
+      "loss": 1.0289,
+      "step": 10923
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.245155096054077,
+      "learning_rate": 9.437613127725972e-06,
+      "loss": 0.4973,
+      "step": 10924
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.075096607208252,
+      "learning_rate": 9.437491637323759e-06,
+      "loss": 0.5597,
+      "step": 10925
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.993335485458374,
+      "learning_rate": 9.437370134582544e-06,
+      "loss": 0.7393,
+      "step": 10926
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.3205268383026123,
+      "learning_rate": 9.437248619502665e-06,
+      "loss": 0.5663,
+      "step": 10927
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.2943906784057617,
+      "learning_rate": 9.437127092084462e-06,
+      "loss": 0.7626,
+      "step": 10928
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.961060047149658,
+      "learning_rate": 9.437005552328269e-06,
+      "loss": 0.5531,
+      "step": 10929
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.340744733810425,
+      "learning_rate": 9.436884000234425e-06,
+      "loss": 0.6489,
+      "step": 10930
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.8403501510620117,
+      "learning_rate": 9.43676243580327e-06,
+      "loss": 0.7495,
+      "step": 10931
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.496065139770508,
+      "learning_rate": 9.43664085903514e-06,
+      "loss": 0.8233,
+      "step": 10932
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.6739344596862793,
+      "learning_rate": 9.436519269930373e-06,
+      "loss": 0.7949,
+      "step": 10933
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.9608519077301025,
+      "learning_rate": 9.43639766848931e-06,
+      "loss": 0.4178,
+      "step": 10934
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.350726366043091,
+      "learning_rate": 9.436276054712285e-06,
+      "loss": 0.5072,
+      "step": 10935
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.9608473777770996,
+      "learning_rate": 9.436154428599639e-06,
+      "loss": 0.5879,
+      "step": 10936
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.9644527435302734,
+      "learning_rate": 9.43603279015171e-06,
+      "loss": 0.515,
+      "step": 10937
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.6984763145446777,
+      "learning_rate": 9.435911139368833e-06,
+      "loss": 0.8103,
+      "step": 10938
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.6870312690734863,
+      "learning_rate": 9.435789476251353e-06,
+      "loss": 0.49,
+      "step": 10939
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 4.798371315002441,
+      "learning_rate": 9.435667800799602e-06,
+      "loss": 0.501,
+      "step": 10940
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.373554229736328,
+      "learning_rate": 9.43554611301392e-06,
+      "loss": 0.7714,
+      "step": 10941
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.1171724796295166,
+      "learning_rate": 9.435424412894647e-06,
+      "loss": 0.744,
+      "step": 10942
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.6238868236541748,
+      "learning_rate": 9.435302700442118e-06,
+      "loss": 0.7397,
+      "step": 10943
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.7884669303894043,
+      "learning_rate": 9.435180975656675e-06,
+      "loss": 0.5722,
+      "step": 10944
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 4.13140869140625,
+      "learning_rate": 9.435059238538656e-06,
+      "loss": 1.077,
+      "step": 10945
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.4701778888702393,
+      "learning_rate": 9.434937489088399e-06,
+      "loss": 0.9344,
+      "step": 10946
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.1342194080352783,
+      "learning_rate": 9.43481572730624e-06,
+      "loss": 0.7174,
+      "step": 10947
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.8220584392547607,
+      "learning_rate": 9.434693953192522e-06,
+      "loss": 0.5685,
+      "step": 10948
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.976811170578003,
+      "learning_rate": 9.43457216674758e-06,
+      "loss": 0.5738,
+      "step": 10949
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.3820009231567383,
+      "learning_rate": 9.434450367971754e-06,
+      "loss": 0.6875,
+      "step": 10950
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.421724796295166,
+      "learning_rate": 9.434328556865383e-06,
+      "loss": 0.816,
+      "step": 10951
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.923435926437378,
+      "learning_rate": 9.434206733428805e-06,
+      "loss": 0.88,
+      "step": 10952
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.236030340194702,
+      "learning_rate": 9.434084897662361e-06,
+      "loss": 0.8914,
+      "step": 10953
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.8820817470550537,
+      "learning_rate": 9.433963049566386e-06,
+      "loss": 0.6976,
+      "step": 10954
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.4505558013916016,
+      "learning_rate": 9.433841189141222e-06,
+      "loss": 0.7863,
+      "step": 10955
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.181964874267578,
+      "learning_rate": 9.433719316387205e-06,
+      "loss": 0.912,
+      "step": 10956
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.2366487979888916,
+      "learning_rate": 9.433597431304678e-06,
+      "loss": 0.6708,
+      "step": 10957
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.5709619522094727,
+      "learning_rate": 9.433475533893974e-06,
+      "loss": 0.7738,
+      "step": 10958
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.5908639430999756,
+      "learning_rate": 9.43335362415544e-06,
+      "loss": 0.433,
+      "step": 10959
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.8615360260009766,
+      "learning_rate": 9.433231702089406e-06,
+      "loss": 0.8389,
+      "step": 10960
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.4980762004852295,
+      "learning_rate": 9.433109767696217e-06,
+      "loss": 0.8842,
+      "step": 10961
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.7385271787643433,
+      "learning_rate": 9.43298782097621e-06,
+      "loss": 0.661,
+      "step": 10962
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.1583328247070312,
+      "learning_rate": 9.432865861929726e-06,
+      "loss": 0.7675,
+      "step": 10963
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.1372382640838623,
+      "learning_rate": 9.432743890557101e-06,
+      "loss": 0.7801,
+      "step": 10964
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 4.5598320960998535,
+      "learning_rate": 9.432621906858677e-06,
+      "loss": 0.7674,
+      "step": 10965
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.1355655193328857,
+      "learning_rate": 9.43249991083479e-06,
+      "loss": 0.8426,
+      "step": 10966
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.8903189897537231,
+      "learning_rate": 9.432377902485784e-06,
+      "loss": 0.3842,
+      "step": 10967
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.1913414001464844,
+      "learning_rate": 9.432255881811994e-06,
+      "loss": 0.7869,
+      "step": 10968
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.3204078674316406,
+      "learning_rate": 9.43213384881376e-06,
+      "loss": 0.5978,
+      "step": 10969
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.6554737091064453,
+      "learning_rate": 9.432011803491422e-06,
+      "loss": 0.501,
+      "step": 10970
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.5545499324798584,
+      "learning_rate": 9.431889745845321e-06,
+      "loss": 0.6034,
+      "step": 10971
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 4.317105770111084,
+      "learning_rate": 9.431767675875793e-06,
+      "loss": 0.9035,
+      "step": 10972
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.7048065662384033,
+      "learning_rate": 9.43164559358318e-06,
+      "loss": 0.9112,
+      "step": 10973
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.6504125595092773,
+      "learning_rate": 9.431523498967823e-06,
+      "loss": 0.583,
+      "step": 10974
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.5053799152374268,
+      "learning_rate": 9.431401392030057e-06,
+      "loss": 0.822,
+      "step": 10975
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.4654463529586792,
+      "learning_rate": 9.431279272770225e-06,
+      "loss": 0.5405,
+      "step": 10976
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.5140738487243652,
+      "learning_rate": 9.431157141188662e-06,
+      "loss": 0.7019,
+      "step": 10977
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.6833451986312866,
+      "learning_rate": 9.431034997285716e-06,
+      "loss": 0.5206,
+      "step": 10978
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.1572258472442627,
+      "learning_rate": 9.430912841061718e-06,
+      "loss": 0.7537,
+      "step": 10979
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.4591935873031616,
+      "learning_rate": 9.430790672517012e-06,
+      "loss": 0.594,
+      "step": 10980
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.2204134464263916,
+      "learning_rate": 9.430668491651938e-06,
+      "loss": 0.6023,
+      "step": 10981
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.1851134300231934,
+      "learning_rate": 9.430546298466833e-06,
+      "loss": 1.0341,
+      "step": 10982
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.920462131500244,
+      "learning_rate": 9.430424092962038e-06,
+      "loss": 0.6785,
+      "step": 10983
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.8107032775878906,
+      "learning_rate": 9.430301875137895e-06,
+      "loss": 0.6569,
+      "step": 10984
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 4.1000518798828125,
+      "learning_rate": 9.430179644994743e-06,
+      "loss": 0.8729,
+      "step": 10985
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.156960964202881,
+      "learning_rate": 9.43005740253292e-06,
+      "loss": 0.803,
+      "step": 10986
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.8314822912216187,
+      "learning_rate": 9.429935147752766e-06,
+      "loss": 0.7636,
+      "step": 10987
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 4.244826316833496,
+      "learning_rate": 9.429812880654623e-06,
+      "loss": 0.5753,
+      "step": 10988
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.5718626976013184,
+      "learning_rate": 9.429690601238828e-06,
+      "loss": 0.5355,
+      "step": 10989
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.3482754230499268,
+      "learning_rate": 9.429568309505725e-06,
+      "loss": 0.872,
+      "step": 10990
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.1284890174865723,
+      "learning_rate": 9.42944600545565e-06,
+      "loss": 0.8313,
+      "step": 10991
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.621087908744812,
+      "learning_rate": 9.429323689088946e-06,
+      "loss": 0.6084,
+      "step": 10992
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.0946385860443115,
+      "learning_rate": 9.429201360405953e-06,
+      "loss": 0.3076,
+      "step": 10993
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.232724666595459,
+      "learning_rate": 9.429079019407007e-06,
+      "loss": 0.7909,
+      "step": 10994
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.304067611694336,
+      "learning_rate": 9.428956666092454e-06,
+      "loss": 0.5526,
+      "step": 10995
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.9172791242599487,
+      "learning_rate": 9.428834300462632e-06,
+      "loss": 0.4279,
+      "step": 10996
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.2131686210632324,
+      "learning_rate": 9.428711922517881e-06,
+      "loss": 0.9573,
+      "step": 10997
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.512552499771118,
+      "learning_rate": 9.428589532258541e-06,
+      "loss": 0.5648,
+      "step": 10998
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.407752275466919,
+      "learning_rate": 9.42846712968495e-06,
+      "loss": 0.7167,
+      "step": 10999
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.991593599319458,
+      "learning_rate": 9.428344714797455e-06,
+      "loss": 0.5497,
+      "step": 11000
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.1713247299194336,
+      "learning_rate": 9.428222287596389e-06,
+      "loss": 0.4252,
+      "step": 11001
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.7983202934265137,
+      "learning_rate": 9.428099848082097e-06,
+      "loss": 0.7502,
+      "step": 11002
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.2739429473876953,
+      "learning_rate": 9.427977396254917e-06,
+      "loss": 0.6993,
+      "step": 11003
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.749498128890991,
+      "learning_rate": 9.427854932115192e-06,
+      "loss": 0.528,
+      "step": 11004
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 4.015427112579346,
+      "learning_rate": 9.427732455663262e-06,
+      "loss": 0.6971,
+      "step": 11005
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.048252582550049,
+      "learning_rate": 9.427609966899464e-06,
+      "loss": 0.8056,
+      "step": 11006
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.225562810897827,
+      "learning_rate": 9.427487465824142e-06,
+      "loss": 0.5144,
+      "step": 11007
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.5293009281158447,
+      "learning_rate": 9.427364952437636e-06,
+      "loss": 0.8671,
+      "step": 11008
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.745347261428833,
+      "learning_rate": 9.427242426740288e-06,
+      "loss": 0.6938,
+      "step": 11009
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.497211456298828,
+      "learning_rate": 9.427119888732437e-06,
+      "loss": 0.7647,
+      "step": 11010
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.7968924045562744,
+      "learning_rate": 9.426997338414423e-06,
+      "loss": 0.9031,
+      "step": 11011
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.175305128097534,
+      "learning_rate": 9.426874775786587e-06,
+      "loss": 0.8094,
+      "step": 11012
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.4359066486358643,
+      "learning_rate": 9.426752200849272e-06,
+      "loss": 0.7168,
+      "step": 11013
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.8383498191833496,
+      "learning_rate": 9.426629613602818e-06,
+      "loss": 0.745,
+      "step": 11014
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.8916677236557007,
+      "learning_rate": 9.426507014047565e-06,
+      "loss": 0.4861,
+      "step": 11015
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.9377143383026123,
+      "learning_rate": 9.426384402183852e-06,
+      "loss": 0.5977,
+      "step": 11016
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.0380589962005615,
+      "learning_rate": 9.426261778012025e-06,
+      "loss": 0.5275,
+      "step": 11017
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.1606576442718506,
+      "learning_rate": 9.426139141532422e-06,
+      "loss": 0.6538,
+      "step": 11018
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.9726486206054688,
+      "learning_rate": 9.426016492745382e-06,
+      "loss": 0.4529,
+      "step": 11019
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.1725404262542725,
+      "learning_rate": 9.425893831651249e-06,
+      "loss": 0.8137,
+      "step": 11020
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.442869186401367,
+      "learning_rate": 9.425771158250364e-06,
+      "loss": 0.457,
+      "step": 11021
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.6060051918029785,
+      "learning_rate": 9.425648472543066e-06,
+      "loss": 0.4719,
+      "step": 11022
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.597039222717285,
+      "learning_rate": 9.425525774529698e-06,
+      "loss": 0.7234,
+      "step": 11023
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 4.21103048324585,
+      "learning_rate": 9.425403064210603e-06,
+      "loss": 0.6084,
+      "step": 11024
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.6579840183258057,
+      "learning_rate": 9.425280341586115e-06,
+      "loss": 0.5277,
+      "step": 11025
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.3728373050689697,
+      "learning_rate": 9.425157606656586e-06,
+      "loss": 0.621,
+      "step": 11026
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.3294758796691895,
+      "learning_rate": 9.425034859422346e-06,
+      "loss": 0.8579,
+      "step": 11027
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 5.482442378997803,
+      "learning_rate": 9.424912099883744e-06,
+      "loss": 0.7816,
+      "step": 11028
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.188915252685547,
+      "learning_rate": 9.42478932804112e-06,
+      "loss": 0.6566,
+      "step": 11029
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.1711907386779785,
+      "learning_rate": 9.424666543894814e-06,
+      "loss": 0.8729,
+      "step": 11030
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.032270908355713,
+      "learning_rate": 9.424543747445168e-06,
+      "loss": 0.4823,
+      "step": 11031
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.3745250701904297,
+      "learning_rate": 9.424420938692523e-06,
+      "loss": 0.4878,
+      "step": 11032
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.343252420425415,
+      "learning_rate": 9.42429811763722e-06,
+      "loss": 0.6626,
+      "step": 11033
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.7308783531188965,
+      "learning_rate": 9.424175284279602e-06,
+      "loss": 0.6672,
+      "step": 11034
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.7125766277313232,
+      "learning_rate": 9.424052438620008e-06,
+      "loss": 0.8462,
+      "step": 11035
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.32975435256958,
+      "learning_rate": 9.423929580658784e-06,
+      "loss": 0.4797,
+      "step": 11036
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.4924453496932983,
+      "learning_rate": 9.423806710396268e-06,
+      "loss": 0.6615,
+      "step": 11037
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.016094446182251,
+      "learning_rate": 9.423683827832804e-06,
+      "loss": 0.698,
+      "step": 11038
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.0401012897491455,
+      "learning_rate": 9.423560932968732e-06,
+      "loss": 0.6712,
+      "step": 11039
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.764334201812744,
+      "learning_rate": 9.423438025804392e-06,
+      "loss": 0.4853,
+      "step": 11040
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.633094072341919,
+      "learning_rate": 9.423315106340129e-06,
+      "loss": 0.6427,
+      "step": 11041
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.424090623855591,
+      "learning_rate": 9.423192174576285e-06,
+      "loss": 0.5607,
+      "step": 11042
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.0249154567718506,
+      "learning_rate": 9.4230692305132e-06,
+      "loss": 0.8016,
+      "step": 11043
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.0175249576568604,
+      "learning_rate": 9.422946274151216e-06,
+      "loss": 0.6556,
+      "step": 11044
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.0966949462890625,
+      "learning_rate": 9.422823305490676e-06,
+      "loss": 0.816,
+      "step": 11045
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.121436834335327,
+      "learning_rate": 9.42270032453192e-06,
+      "loss": 0.5446,
+      "step": 11046
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.7546112537384033,
+      "learning_rate": 9.422577331275291e-06,
+      "loss": 0.9868,
+      "step": 11047
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.167708396911621,
+      "learning_rate": 9.422454325721134e-06,
+      "loss": 0.6081,
+      "step": 11048
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.944630742073059,
+      "learning_rate": 9.422331307869786e-06,
+      "loss": 0.665,
+      "step": 11049
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.0330758094787598,
+      "learning_rate": 9.42220827772159e-06,
+      "loss": 0.7971,
+      "step": 11050
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.357003688812256,
+      "learning_rate": 9.42208523527689e-06,
+      "loss": 0.6338,
+      "step": 11051
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.8175278902053833,
+      "learning_rate": 9.421962180536028e-06,
+      "loss": 1.0088,
+      "step": 11052
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.7253072261810303,
+      "learning_rate": 9.421839113499347e-06,
+      "loss": 0.6716,
+      "step": 11053
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.4881982803344727,
+      "learning_rate": 9.421716034167187e-06,
+      "loss": 0.5516,
+      "step": 11054
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.2827346324920654,
+      "learning_rate": 9.421592942539891e-06,
+      "loss": 0.8025,
+      "step": 11055
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.8374119997024536,
+      "learning_rate": 9.421469838617802e-06,
+      "loss": 0.6914,
+      "step": 11056
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.455444097518921,
+      "learning_rate": 9.421346722401261e-06,
+      "loss": 0.6385,
+      "step": 11057
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.851374626159668,
+      "learning_rate": 9.42122359389061e-06,
+      "loss": 0.7445,
+      "step": 11058
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.3006885051727295,
+      "learning_rate": 9.421100453086194e-06,
+      "loss": 0.6773,
+      "step": 11059
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.7394611835479736,
+      "learning_rate": 9.420977299988354e-06,
+      "loss": 0.7866,
+      "step": 11060
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.8801047801971436,
+      "learning_rate": 9.42085413459743e-06,
+      "loss": 0.8331,
+      "step": 11061
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.3067054748535156,
+      "learning_rate": 9.42073095691377e-06,
+      "loss": 0.6046,
+      "step": 11062
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.915045738220215,
+      "learning_rate": 9.42060776693771e-06,
+      "loss": 0.9455,
+      "step": 11063
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.876382350921631,
+      "learning_rate": 9.420484564669598e-06,
+      "loss": 0.6857,
+      "step": 11064
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.2780725955963135,
+      "learning_rate": 9.420361350109775e-06,
+      "loss": 0.4881,
+      "step": 11065
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.9946141242980957,
+      "learning_rate": 9.420238123258582e-06,
+      "loss": 0.5673,
+      "step": 11066
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.9566220045089722,
+      "learning_rate": 9.420114884116361e-06,
+      "loss": 0.5424,
+      "step": 11067
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.3450751304626465,
+      "learning_rate": 9.419991632683459e-06,
+      "loss": 0.8519,
+      "step": 11068
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.9768311977386475,
+      "learning_rate": 9.419868368960215e-06,
+      "loss": 0.6974,
+      "step": 11069
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.151665449142456,
+      "learning_rate": 9.419745092946972e-06,
+      "loss": 0.8041,
+      "step": 11070
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.453331708908081,
+      "learning_rate": 9.419621804644074e-06,
+      "loss": 0.7475,
+      "step": 11071
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.9714953899383545,
+      "learning_rate": 9.419498504051863e-06,
+      "loss": 0.7837,
+      "step": 11072
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.4314756393432617,
+      "learning_rate": 9.419375191170683e-06,
+      "loss": 0.8274,
+      "step": 11073
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.1410107612609863,
+      "learning_rate": 9.419251866000875e-06,
+      "loss": 0.6205,
+      "step": 11074
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.36271595954895,
+      "learning_rate": 9.419128528542785e-06,
+      "loss": 0.6033,
+      "step": 11075
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.619112968444824,
+      "learning_rate": 9.419005178796752e-06,
+      "loss": 0.6458,
+      "step": 11076
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.7114661931991577,
+      "learning_rate": 9.418881816763124e-06,
+      "loss": 0.8861,
+      "step": 11077
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.2672295570373535,
+      "learning_rate": 9.418758442442238e-06,
+      "loss": 0.8573,
+      "step": 11078
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.375216484069824,
+      "learning_rate": 9.41863505583444e-06,
+      "loss": 0.7849,
+      "step": 11079
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.0928022861480713,
+      "learning_rate": 9.418511656940075e-06,
+      "loss": 0.6462,
+      "step": 11080
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.424175977706909,
+      "learning_rate": 9.418388245759484e-06,
+      "loss": 0.5728,
+      "step": 11081
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.8775157928466797,
+      "learning_rate": 9.418264822293011e-06,
+      "loss": 0.4943,
+      "step": 11082
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.7612040042877197,
+      "learning_rate": 9.418141386540996e-06,
+      "loss": 0.7555,
+      "step": 11083
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.238825559616089,
+      "learning_rate": 9.418017938503788e-06,
+      "loss": 0.8519,
+      "step": 11084
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.5956473350524902,
+      "learning_rate": 9.417894478181724e-06,
+      "loss": 0.7377,
+      "step": 11085
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.69336199760437,
+      "learning_rate": 9.417771005575153e-06,
+      "loss": 0.7537,
+      "step": 11086
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.5093514919281006,
+      "learning_rate": 9.417647520684413e-06,
+      "loss": 0.8007,
+      "step": 11087
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.386387825012207,
+      "learning_rate": 9.417524023509855e-06,
+      "loss": 0.8437,
+      "step": 11088
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.947991371154785,
+      "learning_rate": 9.417400514051813e-06,
+      "loss": 0.7178,
+      "step": 11089
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.330814242362976,
+      "learning_rate": 9.417276992310637e-06,
+      "loss": 0.6631,
+      "step": 11090
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.3141963481903076,
+      "learning_rate": 9.417153458286668e-06,
+      "loss": 0.5469,
+      "step": 11091
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.1112141609191895,
+      "learning_rate": 9.41702991198025e-06,
+      "loss": 0.722,
+      "step": 11092
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.448038101196289,
+      "learning_rate": 9.416906353391727e-06,
+      "loss": 0.7569,
+      "step": 11093
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.397905111312866,
+      "learning_rate": 9.41678278252144e-06,
+      "loss": 0.3606,
+      "step": 11094
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.3870458602905273,
+      "learning_rate": 9.416659199369735e-06,
+      "loss": 0.8465,
+      "step": 11095
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.4744913578033447,
+      "learning_rate": 9.416535603936957e-06,
+      "loss": 0.8012,
+      "step": 11096
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.9811311960220337,
+      "learning_rate": 9.416411996223447e-06,
+      "loss": 0.5637,
+      "step": 11097
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.0054688453674316,
+      "learning_rate": 9.41628837622955e-06,
+      "loss": 0.477,
+      "step": 11098
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 11.763630867004395,
+      "learning_rate": 9.416164743955608e-06,
+      "loss": 0.5699,
+      "step": 11099
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.6277835369110107,
+      "learning_rate": 9.416041099401969e-06,
+      "loss": 0.7213,
+      "step": 11100
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.0440261363983154,
+      "learning_rate": 9.415917442568973e-06,
+      "loss": 0.6557,
+      "step": 11101
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.4174859523773193,
+      "learning_rate": 9.415793773456965e-06,
+      "loss": 0.8359,
+      "step": 11102
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.7322797775268555,
+      "learning_rate": 9.415670092066286e-06,
+      "loss": 0.4123,
+      "step": 11103
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.4796531200408936,
+      "learning_rate": 9.415546398397285e-06,
+      "loss": 0.6509,
+      "step": 11104
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.4151830673217773,
+      "learning_rate": 9.415422692450305e-06,
+      "loss": 1.0449,
+      "step": 11105
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.8579427003860474,
+      "learning_rate": 9.415298974225685e-06,
+      "loss": 0.7364,
+      "step": 11106
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.113833427429199,
+      "learning_rate": 9.415175243723775e-06,
+      "loss": 0.7175,
+      "step": 11107
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.658200979232788,
+      "learning_rate": 9.415051500944916e-06,
+      "loss": 0.8624,
+      "step": 11108
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.2603490352630615,
+      "learning_rate": 9.414927745889454e-06,
+      "loss": 0.8293,
+      "step": 11109
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 4.142967700958252,
+      "learning_rate": 9.41480397855773e-06,
+      "loss": 0.7723,
+      "step": 11110
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.2330329418182373,
+      "learning_rate": 9.41468019895009e-06,
+      "loss": 0.6319,
+      "step": 11111
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.298095226287842,
+      "learning_rate": 9.414556407066878e-06,
+      "loss": 0.4612,
+      "step": 11112
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.319847583770752,
+      "learning_rate": 9.414432602908438e-06,
+      "loss": 0.4391,
+      "step": 11113
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.242358684539795,
+      "learning_rate": 9.414308786475117e-06,
+      "loss": 0.9359,
+      "step": 11114
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.6694462299346924,
+      "learning_rate": 9.414184957767255e-06,
+      "loss": 0.6282,
+      "step": 11115
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.007664680480957,
+      "learning_rate": 9.414061116785198e-06,
+      "loss": 0.7531,
+      "step": 11116
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.0671653747558594,
+      "learning_rate": 9.413937263529291e-06,
+      "loss": 0.7016,
+      "step": 11117
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 4.022116184234619,
+      "learning_rate": 9.413813397999877e-06,
+      "loss": 0.7019,
+      "step": 11118
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.228813648223877,
+      "learning_rate": 9.413689520197305e-06,
+      "loss": 0.5283,
+      "step": 11119
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.6171677112579346,
+      "learning_rate": 9.413565630121913e-06,
+      "loss": 0.6869,
+      "step": 11120
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.3968358039855957,
+      "learning_rate": 9.413441727774048e-06,
+      "loss": 0.9477,
+      "step": 11121
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.613001823425293,
+      "learning_rate": 9.413317813154057e-06,
+      "loss": 0.5261,
+      "step": 11122
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.5948456525802612,
+      "learning_rate": 9.41319388626228e-06,
+      "loss": 0.7552,
+      "step": 11123
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.9666553735733032,
+      "learning_rate": 9.413069947099065e-06,
+      "loss": 0.8133,
+      "step": 11124
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.565964937210083,
+      "learning_rate": 9.412945995664757e-06,
+      "loss": 0.5017,
+      "step": 11125
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.831080913543701,
+      "learning_rate": 9.412822031959698e-06,
+      "loss": 0.4296,
+      "step": 11126
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.9732463359832764,
+      "learning_rate": 9.412698055984233e-06,
+      "loss": 0.9127,
+      "step": 11127
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.055112361907959,
+      "learning_rate": 9.41257406773871e-06,
+      "loss": 0.6325,
+      "step": 11128
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.4501662254333496,
+      "learning_rate": 9.412450067223469e-06,
+      "loss": 0.5668,
+      "step": 11129
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.236626148223877,
+      "learning_rate": 9.412326054438859e-06,
+      "loss": 0.5422,
+      "step": 11130
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.2569174766540527,
+      "learning_rate": 9.412202029385223e-06,
+      "loss": 0.6712,
+      "step": 11131
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.7156431674957275,
+      "learning_rate": 9.412077992062906e-06,
+      "loss": 0.5033,
+      "step": 11132
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.3590545654296875,
+      "learning_rate": 9.411953942472251e-06,
+      "loss": 0.5633,
+      "step": 11133
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 6.000082492828369,
+      "learning_rate": 9.411829880613607e-06,
+      "loss": 1.0097,
+      "step": 11134
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.7335305213928223,
+      "learning_rate": 9.411705806487317e-06,
+      "loss": 0.7156,
+      "step": 11135
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.153830051422119,
+      "learning_rate": 9.411581720093725e-06,
+      "loss": 0.4931,
+      "step": 11136
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.182105541229248,
+      "learning_rate": 9.411457621433178e-06,
+      "loss": 0.4537,
+      "step": 11137
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.9737200736999512,
+      "learning_rate": 9.411333510506018e-06,
+      "loss": 0.5624,
+      "step": 11138
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.9611378908157349,
+      "learning_rate": 9.411209387312593e-06,
+      "loss": 0.766,
+      "step": 11139
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.6560767889022827,
+      "learning_rate": 9.411085251853247e-06,
+      "loss": 0.8188,
+      "step": 11140
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.401043653488159,
+      "learning_rate": 9.410961104128325e-06,
+      "loss": 0.4435,
+      "step": 11141
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.551811695098877,
+      "learning_rate": 9.410836944138173e-06,
+      "loss": 0.4769,
+      "step": 11142
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.898859977722168,
+      "learning_rate": 9.410712771883136e-06,
+      "loss": 0.9284,
+      "step": 11143
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.876610279083252,
+      "learning_rate": 9.410588587363559e-06,
+      "loss": 0.5697,
+      "step": 11144
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.5334763526916504,
+      "learning_rate": 9.410464390579788e-06,
+      "loss": 0.6819,
+      "step": 11145
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.996887445449829,
+      "learning_rate": 9.410340181532167e-06,
+      "loss": 0.6966,
+      "step": 11146
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.4973337650299072,
+      "learning_rate": 9.410215960221041e-06,
+      "loss": 0.6168,
+      "step": 11147
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.4107439517974854,
+      "learning_rate": 9.410091726646759e-06,
+      "loss": 0.6025,
+      "step": 11148
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.040337085723877,
+      "learning_rate": 9.409967480809661e-06,
+      "loss": 0.5733,
+      "step": 11149
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.5658671855926514,
+      "learning_rate": 9.409843222710097e-06,
+      "loss": 0.4894,
+      "step": 11150
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.3207807540893555,
+      "learning_rate": 9.409718952348413e-06,
+      "loss": 0.7735,
+      "step": 11151
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 4.739053249359131,
+      "learning_rate": 9.409594669724949e-06,
+      "loss": 0.6185,
+      "step": 11152
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.8298118114471436,
+      "learning_rate": 9.409470374840057e-06,
+      "loss": 0.6479,
+      "step": 11153
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.32320237159729,
+      "learning_rate": 9.409346067694078e-06,
+      "loss": 0.6649,
+      "step": 11154
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.095400094985962,
+      "learning_rate": 9.40922174828736e-06,
+      "loss": 0.7889,
+      "step": 11155
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.4658334255218506,
+      "learning_rate": 9.409097416620248e-06,
+      "loss": 0.7465,
+      "step": 11156
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.7076678276062012,
+      "learning_rate": 9.408973072693086e-06,
+      "loss": 0.6115,
+      "step": 11157
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.4408211708068848,
+      "learning_rate": 9.408848716506224e-06,
+      "loss": 0.7122,
+      "step": 11158
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.107429027557373,
+      "learning_rate": 9.408724348060003e-06,
+      "loss": 0.5596,
+      "step": 11159
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.055828094482422,
+      "learning_rate": 9.408599967354774e-06,
+      "loss": 0.8504,
+      "step": 11160
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.8948416709899902,
+      "learning_rate": 9.408475574390878e-06,
+      "loss": 0.6266,
+      "step": 11161
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.6004562377929688,
+      "learning_rate": 9.408351169168664e-06,
+      "loss": 0.543,
+      "step": 11162
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.131312131881714,
+      "learning_rate": 9.408226751688476e-06,
+      "loss": 0.4482,
+      "step": 11163
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.7192538976669312,
+      "learning_rate": 9.40810232195066e-06,
+      "loss": 0.5158,
+      "step": 11164
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.076453447341919,
+      "learning_rate": 9.407977879955564e-06,
+      "loss": 0.7543,
+      "step": 11165
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.5264780521392822,
+      "learning_rate": 9.407853425703533e-06,
+      "loss": 0.6508,
+      "step": 11166
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 4.308371067047119,
+      "learning_rate": 9.407728959194912e-06,
+      "loss": 0.7529,
+      "step": 11167
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.1738569736480713,
+      "learning_rate": 9.407604480430048e-06,
+      "loss": 0.6319,
+      "step": 11168
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.1257200241088867,
+      "learning_rate": 9.407479989409287e-06,
+      "loss": 0.6302,
+      "step": 11169
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.1168923377990723,
+      "learning_rate": 9.407355486132977e-06,
+      "loss": 0.7092,
+      "step": 11170
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.005230665206909,
+      "learning_rate": 9.40723097060146e-06,
+      "loss": 0.5512,
+      "step": 11171
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.4612860679626465,
+      "learning_rate": 9.407106442815086e-06,
+      "loss": 0.5426,
+      "step": 11172
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.8886089324951172,
+      "learning_rate": 9.4069819027742e-06,
+      "loss": 0.8042,
+      "step": 11173
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.740675687789917,
+      "learning_rate": 9.406857350479146e-06,
+      "loss": 0.8502,
+      "step": 11174
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.2449469566345215,
+      "learning_rate": 9.406732785930274e-06,
+      "loss": 0.3424,
+      "step": 11175
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.8774759769439697,
+      "learning_rate": 9.40660820912793e-06,
+      "loss": 0.8546,
+      "step": 11176
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.8612428903579712,
+      "learning_rate": 9.406483620072459e-06,
+      "loss": 0.6119,
+      "step": 11177
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.626042127609253,
+      "learning_rate": 9.406359018764207e-06,
+      "loss": 0.8422,
+      "step": 11178
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.9693557024002075,
+      "learning_rate": 9.406234405203521e-06,
+      "loss": 0.8111,
+      "step": 11179
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.172605514526367,
+      "learning_rate": 9.406109779390749e-06,
+      "loss": 0.6805,
+      "step": 11180
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.490159034729004,
+      "learning_rate": 9.405985141326234e-06,
+      "loss": 0.6429,
+      "step": 11181
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.6485438346862793,
+      "learning_rate": 9.405860491010329e-06,
+      "loss": 0.4132,
+      "step": 11182
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.7445950508117676,
+      "learning_rate": 9.405735828443372e-06,
+      "loss": 0.4574,
+      "step": 11183
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.469003915786743,
+      "learning_rate": 9.405611153625717e-06,
+      "loss": 0.4804,
+      "step": 11184
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.516918420791626,
+      "learning_rate": 9.405486466557706e-06,
+      "loss": 0.6452,
+      "step": 11185
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.0520646572113037,
+      "learning_rate": 9.405361767239689e-06,
+      "loss": 0.7518,
+      "step": 11186
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.7567296028137207,
+      "learning_rate": 9.40523705567201e-06,
+      "loss": 0.5963,
+      "step": 11187
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.8441386222839355,
+      "learning_rate": 9.405112331855017e-06,
+      "loss": 0.7193,
+      "step": 11188
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.5392107963562012,
+      "learning_rate": 9.404987595789057e-06,
+      "loss": 0.8717,
+      "step": 11189
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.97408390045166,
+      "learning_rate": 9.404862847474476e-06,
+      "loss": 0.6042,
+      "step": 11190
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.8824498653411865,
+      "learning_rate": 9.404738086911624e-06,
+      "loss": 0.8871,
+      "step": 11191
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.9629024267196655,
+      "learning_rate": 9.404613314100844e-06,
+      "loss": 0.5317,
+      "step": 11192
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 4.160228729248047,
+      "learning_rate": 9.404488529042484e-06,
+      "loss": 0.8078,
+      "step": 11193
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.7512404918670654,
+      "learning_rate": 9.40436373173689e-06,
+      "loss": 0.4488,
+      "step": 11194
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.098231792449951,
+      "learning_rate": 9.404238922184412e-06,
+      "loss": 0.669,
+      "step": 11195
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.6322101354599,
+      "learning_rate": 9.404114100385395e-06,
+      "loss": 0.8148,
+      "step": 11196
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.1009292602539062,
+      "learning_rate": 9.403989266340188e-06,
+      "loss": 0.7433,
+      "step": 11197
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.561352014541626,
+      "learning_rate": 9.403864420049136e-06,
+      "loss": 0.6866,
+      "step": 11198
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.542954206466675,
+      "learning_rate": 9.403739561512585e-06,
+      "loss": 0.66,
+      "step": 11199
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.5258021354675293,
+      "learning_rate": 9.403614690730885e-06,
+      "loss": 0.5986,
+      "step": 11200
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.312811851501465,
+      "learning_rate": 9.403489807704383e-06,
+      "loss": 0.6137,
+      "step": 11201
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.87465500831604,
+      "learning_rate": 9.403364912433424e-06,
+      "loss": 0.8073,
+      "step": 11202
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.932677745819092,
+      "learning_rate": 9.403240004918359e-06,
+      "loss": 0.8302,
+      "step": 11203
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.4912407398223877,
+      "learning_rate": 9.403115085159531e-06,
+      "loss": 0.5161,
+      "step": 11204
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.7661426067352295,
+      "learning_rate": 9.40299015315729e-06,
+      "loss": 0.623,
+      "step": 11205
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.605994462966919,
+      "learning_rate": 9.402865208911983e-06,
+      "loss": 0.532,
+      "step": 11206
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.3150928020477295,
+      "learning_rate": 9.402740252423957e-06,
+      "loss": 0.805,
+      "step": 11207
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.2000696659088135,
+      "learning_rate": 9.40261528369356e-06,
+      "loss": 0.6633,
+      "step": 11208
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.7196896076202393,
+      "learning_rate": 9.402490302721137e-06,
+      "loss": 0.7302,
+      "step": 11209
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 12.979425430297852,
+      "learning_rate": 9.40236530950704e-06,
+      "loss": 0.8799,
+      "step": 11210
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.8809219598770142,
+      "learning_rate": 9.402240304051613e-06,
+      "loss": 0.6537,
+      "step": 11211
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.0790209770202637,
+      "learning_rate": 9.402115286355206e-06,
+      "loss": 0.8985,
+      "step": 11212
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.3181092739105225,
+      "learning_rate": 9.401990256418166e-06,
+      "loss": 0.5402,
+      "step": 11213
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 5.804421901702881,
+      "learning_rate": 9.401865214240838e-06,
+      "loss": 0.7924,
+      "step": 11214
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.1997134685516357,
+      "learning_rate": 9.401740159823572e-06,
+      "loss": 0.6513,
+      "step": 11215
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 4.806216239929199,
+      "learning_rate": 9.401615093166717e-06,
+      "loss": 0.7014,
+      "step": 11216
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.171142578125,
+      "learning_rate": 9.401490014270617e-06,
+      "loss": 0.36,
+      "step": 11217
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.4833577871322632,
+      "learning_rate": 9.401364923135625e-06,
+      "loss": 0.5941,
+      "step": 11218
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.6852095127105713,
+      "learning_rate": 9.401239819762085e-06,
+      "loss": 0.7576,
+      "step": 11219
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.4485586881637573,
+      "learning_rate": 9.401114704150345e-06,
+      "loss": 0.6645,
+      "step": 11220
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.648336410522461,
+      "learning_rate": 9.400989576300755e-06,
+      "loss": 0.7679,
+      "step": 11221
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.2908072471618652,
+      "learning_rate": 9.400864436213662e-06,
+      "loss": 0.5658,
+      "step": 11222
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.9278783798217773,
+      "learning_rate": 9.40073928388941e-06,
+      "loss": 0.4836,
+      "step": 11223
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.8345468044281006,
+      "learning_rate": 9.400614119328354e-06,
+      "loss": 0.7507,
+      "step": 11224
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.8691742420196533,
+      "learning_rate": 9.400488942530838e-06,
+      "loss": 0.634,
+      "step": 11225
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.265517234802246,
+      "learning_rate": 9.40036375349721e-06,
+      "loss": 0.6803,
+      "step": 11226
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.8305935859680176,
+      "learning_rate": 9.40023855222782e-06,
+      "loss": 0.8826,
+      "step": 11227
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.9428205490112305,
+      "learning_rate": 9.400113338723015e-06,
+      "loss": 0.755,
+      "step": 11228
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.1507341861724854,
+      "learning_rate": 9.39998811298314e-06,
+      "loss": 0.6132,
+      "step": 11229
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 10.997653007507324,
+      "learning_rate": 9.39986287500855e-06,
+      "loss": 0.8984,
+      "step": 11230
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.4245736598968506,
+      "learning_rate": 9.399737624799588e-06,
+      "loss": 0.6185,
+      "step": 11231
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.6753737926483154,
+      "learning_rate": 9.399612362356605e-06,
+      "loss": 0.863,
+      "step": 11232
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.9941108226776123,
+      "learning_rate": 9.399487087679946e-06,
+      "loss": 0.7522,
+      "step": 11233
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.5819182395935059,
+      "learning_rate": 9.399361800769964e-06,
+      "loss": 0.4775,
+      "step": 11234
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.3053781986236572,
+      "learning_rate": 9.399236501627002e-06,
+      "loss": 0.5652,
+      "step": 11235
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.5894811153411865,
+      "learning_rate": 9.399111190251413e-06,
+      "loss": 0.5907,
+      "step": 11236
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.5463030338287354,
+      "learning_rate": 9.398985866643544e-06,
+      "loss": 0.6329,
+      "step": 11237
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.3372151851654053,
+      "learning_rate": 9.398860530803743e-06,
+      "loss": 0.8055,
+      "step": 11238
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.4998435974121094,
+      "learning_rate": 9.398735182732358e-06,
+      "loss": 0.6445,
+      "step": 11239
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.0666141510009766,
+      "learning_rate": 9.39860982242974e-06,
+      "loss": 0.4779,
+      "step": 11240
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.606938600540161,
+      "learning_rate": 9.398484449896235e-06,
+      "loss": 0.913,
+      "step": 11241
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.2076895236968994,
+      "learning_rate": 9.398359065132191e-06,
+      "loss": 0.5531,
+      "step": 11242
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.093533515930176,
+      "learning_rate": 9.398233668137959e-06,
+      "loss": 0.5817,
+      "step": 11243
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.5174801349639893,
+      "learning_rate": 9.398108258913887e-06,
+      "loss": 0.7828,
+      "step": 11244
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.940358877182007,
+      "learning_rate": 9.397982837460324e-06,
+      "loss": 0.7066,
+      "step": 11245
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.9191480875015259,
+      "learning_rate": 9.397857403777617e-06,
+      "loss": 0.8235,
+      "step": 11246
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.7097365856170654,
+      "learning_rate": 9.397731957866117e-06,
+      "loss": 0.5962,
+      "step": 11247
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.822585105895996,
+      "learning_rate": 9.397606499726172e-06,
+      "loss": 0.5725,
+      "step": 11248
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.377361297607422,
+      "learning_rate": 9.39748102935813e-06,
+      "loss": 0.7389,
+      "step": 11249
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.0707027912139893,
+      "learning_rate": 9.397355546762342e-06,
+      "loss": 0.8677,
+      "step": 11250
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.2011778354644775,
+      "learning_rate": 9.397230051939153e-06,
+      "loss": 0.8954,
+      "step": 11251
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.1602234840393066,
+      "learning_rate": 9.397104544888917e-06,
+      "loss": 0.6858,
+      "step": 11252
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.5437216758728027,
+      "learning_rate": 9.396979025611978e-06,
+      "loss": 0.7418,
+      "step": 11253
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.03196382522583,
+      "learning_rate": 9.39685349410869e-06,
+      "loss": 0.8202,
+      "step": 11254
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.9534695148468018,
+      "learning_rate": 9.396727950379398e-06,
+      "loss": 0.7605,
+      "step": 11255
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.052554130554199,
+      "learning_rate": 9.396602394424452e-06,
+      "loss": 0.6476,
+      "step": 11256
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.9370455741882324,
+      "learning_rate": 9.396476826244202e-06,
+      "loss": 0.7648,
+      "step": 11257
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.401928663253784,
+      "learning_rate": 9.396351245839e-06,
+      "loss": 0.6367,
+      "step": 11258
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.178969144821167,
+      "learning_rate": 9.39622565320919e-06,
+      "loss": 0.7198,
+      "step": 11259
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.397505044937134,
+      "learning_rate": 9.396100048355122e-06,
+      "loss": 0.5366,
+      "step": 11260
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.57072377204895,
+      "learning_rate": 9.395974431277147e-06,
+      "loss": 0.5165,
+      "step": 11261
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.184065103530884,
+      "learning_rate": 9.395848801975616e-06,
+      "loss": 0.6975,
+      "step": 11262
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.5823118686676025,
+      "learning_rate": 9.395723160450873e-06,
+      "loss": 0.5393,
+      "step": 11263
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 4.403531551361084,
+      "learning_rate": 9.395597506703273e-06,
+      "loss": 0.7901,
+      "step": 11264
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.873396396636963,
+      "learning_rate": 9.395471840733162e-06,
+      "loss": 0.8407,
+      "step": 11265
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 4.041815757751465,
+      "learning_rate": 9.39534616254089e-06,
+      "loss": 0.5336,
+      "step": 11266
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.342679262161255,
+      "learning_rate": 9.395220472126808e-06,
+      "loss": 0.4739,
+      "step": 11267
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 20.376047134399414,
+      "learning_rate": 9.395094769491265e-06,
+      "loss": 0.6211,
+      "step": 11268
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.1398472785949707,
+      "learning_rate": 9.394969054634609e-06,
+      "loss": 0.7685,
+      "step": 11269
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.914816379547119,
+      "learning_rate": 9.39484332755719e-06,
+      "loss": 0.7369,
+      "step": 11270
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.7310680150985718,
+      "learning_rate": 9.394717588259357e-06,
+      "loss": 0.6979,
+      "step": 11271
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.5918843746185303,
+      "learning_rate": 9.394591836741462e-06,
+      "loss": 0.7869,
+      "step": 11272
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.074451208114624,
+      "learning_rate": 9.394466073003854e-06,
+      "loss": 0.4967,
+      "step": 11273
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.956739902496338,
+      "learning_rate": 9.39434029704688e-06,
+      "loss": 0.6714,
+      "step": 11274
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 4.233849048614502,
+      "learning_rate": 9.394214508870893e-06,
+      "loss": 0.8207,
+      "step": 11275
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.4965031147003174,
+      "learning_rate": 9.394088708476244e-06,
+      "loss": 0.6178,
+      "step": 11276
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.2165353298187256,
+      "learning_rate": 9.393962895863276e-06,
+      "loss": 0.8656,
+      "step": 11277
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.550102710723877,
+      "learning_rate": 9.393837071032345e-06,
+      "loss": 0.6866,
+      "step": 11278
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 5.421934127807617,
+      "learning_rate": 9.3937112339838e-06,
+      "loss": 0.7415,
+      "step": 11279
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.5841939449310303,
+      "learning_rate": 9.393585384717989e-06,
+      "loss": 0.8976,
+      "step": 11280
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.9381890296936035,
+      "learning_rate": 9.393459523235264e-06,
+      "loss": 0.8975,
+      "step": 11281
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.1762495040893555,
+      "learning_rate": 9.393333649535973e-06,
+      "loss": 0.611,
+      "step": 11282
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.27601957321167,
+      "learning_rate": 9.393207763620467e-06,
+      "loss": 0.6512,
+      "step": 11283
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.0915520191192627,
+      "learning_rate": 9.393081865489095e-06,
+      "loss": 0.82,
+      "step": 11284
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.415823221206665,
+      "learning_rate": 9.39295595514221e-06,
+      "loss": 0.8917,
+      "step": 11285
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.7444286346435547,
+      "learning_rate": 9.392830032580158e-06,
+      "loss": 0.739,
+      "step": 11286
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 4.249607086181641,
+      "learning_rate": 9.392704097803293e-06,
+      "loss": 0.6945,
+      "step": 11287
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.1589787006378174,
+      "learning_rate": 9.392578150811963e-06,
+      "loss": 0.5356,
+      "step": 11288
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.4214026927948,
+      "learning_rate": 9.392452191606517e-06,
+      "loss": 0.4633,
+      "step": 11289
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.6538684368133545,
+      "learning_rate": 9.39232622018731e-06,
+      "loss": 0.6832,
+      "step": 11290
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.8749886751174927,
+      "learning_rate": 9.392200236554687e-06,
+      "loss": 0.7823,
+      "step": 11291
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.0555074214935303,
+      "learning_rate": 9.392074240709001e-06,
+      "loss": 0.5859,
+      "step": 11292
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.5292608737945557,
+      "learning_rate": 9.391948232650603e-06,
+      "loss": 0.6434,
+      "step": 11293
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 4.632567882537842,
+      "learning_rate": 9.39182221237984e-06,
+      "loss": 0.9408,
+      "step": 11294
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.799156665802002,
+      "learning_rate": 9.391696179897066e-06,
+      "loss": 0.6391,
+      "step": 11295
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.7179605960845947,
+      "learning_rate": 9.39157013520263e-06,
+      "loss": 0.8436,
+      "step": 11296
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.3822596073150635,
+      "learning_rate": 9.391444078296883e-06,
+      "loss": 0.5,
+      "step": 11297
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.4124948978424072,
+      "learning_rate": 9.391318009180176e-06,
+      "loss": 0.748,
+      "step": 11298
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.5386269092559814,
+      "learning_rate": 9.391191927852855e-06,
+      "loss": 0.8236,
+      "step": 11299
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.7453930377960205,
+      "learning_rate": 9.391065834315276e-06,
+      "loss": 0.7735,
+      "step": 11300
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.1830735206604004,
+      "learning_rate": 9.39093972856779e-06,
+      "loss": 0.6581,
+      "step": 11301
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.0879862308502197,
+      "learning_rate": 9.390813610610743e-06,
+      "loss": 0.7828,
+      "step": 11302
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.7577403783798218,
+      "learning_rate": 9.39068748044449e-06,
+      "loss": 0.3881,
+      "step": 11303
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.712989091873169,
+      "learning_rate": 9.390561338069377e-06,
+      "loss": 0.7082,
+      "step": 11304
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.5614469051361084,
+      "learning_rate": 9.39043518348576e-06,
+      "loss": 0.5739,
+      "step": 11305
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.7445223331451416,
+      "learning_rate": 9.390309016693986e-06,
+      "loss": 0.5542,
+      "step": 11306
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.479652166366577,
+      "learning_rate": 9.39018283769441e-06,
+      "loss": 0.685,
+      "step": 11307
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.4869425296783447,
+      "learning_rate": 9.390056646487378e-06,
+      "loss": 0.6097,
+      "step": 11308
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.5369672775268555,
+      "learning_rate": 9.389930443073243e-06,
+      "loss": 0.7361,
+      "step": 11309
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.1781389713287354,
+      "learning_rate": 9.389804227452357e-06,
+      "loss": 0.706,
+      "step": 11310
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.234013795852661,
+      "learning_rate": 9.389677999625068e-06,
+      "loss": 0.8582,
+      "step": 11311
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.943272352218628,
+      "learning_rate": 9.38955175959173e-06,
+      "loss": 0.8785,
+      "step": 11312
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.4561405181884766,
+      "learning_rate": 9.389425507352695e-06,
+      "loss": 0.6892,
+      "step": 11313
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 5.988938808441162,
+      "learning_rate": 9.389299242908308e-06,
+      "loss": 0.8758,
+      "step": 11314
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.848145008087158,
+      "learning_rate": 9.389172966258926e-06,
+      "loss": 0.6186,
+      "step": 11315
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 4.0560150146484375,
+      "learning_rate": 9.389046677404898e-06,
+      "loss": 0.7078,
+      "step": 11316
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 4.035661697387695,
+      "learning_rate": 9.388920376346576e-06,
+      "loss": 0.7489,
+      "step": 11317
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.132171392440796,
+      "learning_rate": 9.38879406308431e-06,
+      "loss": 0.8354,
+      "step": 11318
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.980463981628418,
+      "learning_rate": 9.388667737618452e-06,
+      "loss": 0.7735,
+      "step": 11319
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.0837323665618896,
+      "learning_rate": 9.388541399949353e-06,
+      "loss": 0.779,
+      "step": 11320
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.5262463092803955,
+      "learning_rate": 9.388415050077365e-06,
+      "loss": 0.8712,
+      "step": 11321
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.2455387115478516,
+      "learning_rate": 9.388288688002838e-06,
+      "loss": 0.5711,
+      "step": 11322
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.8673534393310547,
+      "learning_rate": 9.388162313726125e-06,
+      "loss": 0.8249,
+      "step": 11323
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.737626314163208,
+      "learning_rate": 9.388035927247575e-06,
+      "loss": 0.6351,
+      "step": 11324
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.3754714727401733,
+      "learning_rate": 9.387909528567541e-06,
+      "loss": 0.6515,
+      "step": 11325
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.5752370357513428,
+      "learning_rate": 9.387783117686375e-06,
+      "loss": 0.7296,
+      "step": 11326
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.337623357772827,
+      "learning_rate": 9.387656694604428e-06,
+      "loss": 0.8178,
+      "step": 11327
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.6388978958129883,
+      "learning_rate": 9.38753025932205e-06,
+      "loss": 0.9897,
+      "step": 11328
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.9540765285491943,
+      "learning_rate": 9.387403811839596e-06,
+      "loss": 1.0191,
+      "step": 11329
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.5243163108825684,
+      "learning_rate": 9.387277352157415e-06,
+      "loss": 0.6075,
+      "step": 11330
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.280639410018921,
+      "learning_rate": 9.387150880275859e-06,
+      "loss": 0.787,
+      "step": 11331
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 4.175056457519531,
+      "learning_rate": 9.38702439619528e-06,
+      "loss": 0.7841,
+      "step": 11332
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.652233839035034,
+      "learning_rate": 9.386897899916028e-06,
+      "loss": 0.508,
+      "step": 11333
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.6416363716125488,
+      "learning_rate": 9.386771391438459e-06,
+      "loss": 0.4878,
+      "step": 11334
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.4562768936157227,
+      "learning_rate": 9.38664487076292e-06,
+      "loss": 0.518,
+      "step": 11335
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.2027676105499268,
+      "learning_rate": 9.386518337889766e-06,
+      "loss": 0.3951,
+      "step": 11336
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.5495567321777344,
+      "learning_rate": 9.386391792819346e-06,
+      "loss": 0.5891,
+      "step": 11337
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.9405221939086914,
+      "learning_rate": 9.386265235552015e-06,
+      "loss": 0.7147,
+      "step": 11338
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.355886459350586,
+      "learning_rate": 9.386138666088123e-06,
+      "loss": 0.4978,
+      "step": 11339
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.6558327674865723,
+      "learning_rate": 9.386012084428023e-06,
+      "loss": 0.6784,
+      "step": 11340
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 4.484047889709473,
+      "learning_rate": 9.385885490572066e-06,
+      "loss": 0.9966,
+      "step": 11341
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.381248712539673,
+      "learning_rate": 9.385758884520605e-06,
+      "loss": 0.6305,
+      "step": 11342
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.3333966732025146,
+      "learning_rate": 9.385632266273992e-06,
+      "loss": 0.6439,
+      "step": 11343
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.8697363138198853,
+      "learning_rate": 9.385505635832577e-06,
+      "loss": 0.5665,
+      "step": 11344
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.7494041919708252,
+      "learning_rate": 9.385378993196714e-06,
+      "loss": 0.6775,
+      "step": 11345
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.1408820152282715,
+      "learning_rate": 9.385252338366755e-06,
+      "loss": 0.5794,
+      "step": 11346
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.9981499910354614,
+      "learning_rate": 9.385125671343053e-06,
+      "loss": 0.7515,
+      "step": 11347
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.1286778450012207,
+      "learning_rate": 9.384998992125958e-06,
+      "loss": 0.5505,
+      "step": 11348
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 6.62282133102417,
+      "learning_rate": 9.384872300715825e-06,
+      "loss": 0.8626,
+      "step": 11349
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.2382216453552246,
+      "learning_rate": 9.384745597113003e-06,
+      "loss": 0.5728,
+      "step": 11350
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.32322096824646,
+      "learning_rate": 9.384618881317847e-06,
+      "loss": 0.7907,
+      "step": 11351
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.4516143798828125,
+      "learning_rate": 9.384492153330707e-06,
+      "loss": 0.7241,
+      "step": 11352
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.4793860912323,
+      "learning_rate": 9.384365413151938e-06,
+      "loss": 0.7406,
+      "step": 11353
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.9484293460845947,
+      "learning_rate": 9.384238660781891e-06,
+      "loss": 0.5249,
+      "step": 11354
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.7828483581542969,
+      "learning_rate": 9.384111896220919e-06,
+      "loss": 0.4918,
+      "step": 11355
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.2063260078430176,
+      "learning_rate": 9.383985119469375e-06,
+      "loss": 0.4376,
+      "step": 11356
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.6633455753326416,
+      "learning_rate": 9.38385833052761e-06,
+      "loss": 0.6044,
+      "step": 11357
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.1934006214141846,
+      "learning_rate": 9.383731529395974e-06,
+      "loss": 0.9046,
+      "step": 11358
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.3821699619293213,
+      "learning_rate": 9.383604716074827e-06,
+      "loss": 0.5954,
+      "step": 11359
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.3659400939941406,
+      "learning_rate": 9.383477890564516e-06,
+      "loss": 0.8118,
+      "step": 11360
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.9317363500595093,
+      "learning_rate": 9.383351052865395e-06,
+      "loss": 0.64,
+      "step": 11361
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.7320095300674438,
+      "learning_rate": 9.383224202977816e-06,
+      "loss": 0.6529,
+      "step": 11362
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.5365842580795288,
+      "learning_rate": 9.383097340902133e-06,
+      "loss": 0.5869,
+      "step": 11363
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.2888622283935547,
+      "learning_rate": 9.382970466638698e-06,
+      "loss": 0.7115,
+      "step": 11364
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 4.253566265106201,
+      "learning_rate": 9.382843580187865e-06,
+      "loss": 0.6972,
+      "step": 11365
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.1805875301361084,
+      "learning_rate": 9.382716681549987e-06,
+      "loss": 0.6024,
+      "step": 11366
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.1735880374908447,
+      "learning_rate": 9.382589770725413e-06,
+      "loss": 0.6546,
+      "step": 11367
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.7031123638153076,
+      "learning_rate": 9.3824628477145e-06,
+      "loss": 0.7471,
+      "step": 11368
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.5084728002548218,
+      "learning_rate": 9.382335912517599e-06,
+      "loss": 0.6406,
+      "step": 11369
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.881387233734131,
+      "learning_rate": 9.382208965135064e-06,
+      "loss": 0.7078,
+      "step": 11370
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.994325637817383,
+      "learning_rate": 9.382082005567248e-06,
+      "loss": 0.6213,
+      "step": 11371
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.808034896850586,
+      "learning_rate": 9.381955033814503e-06,
+      "loss": 0.8811,
+      "step": 11372
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.5134811401367188,
+      "learning_rate": 9.381828049877182e-06,
+      "loss": 0.7051,
+      "step": 11373
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.1947991847991943,
+      "learning_rate": 9.381701053755638e-06,
+      "loss": 0.7156,
+      "step": 11374
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.3464925289154053,
+      "learning_rate": 9.381574045450225e-06,
+      "loss": 0.5955,
+      "step": 11375
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.2489070892333984,
+      "learning_rate": 9.381447024961299e-06,
+      "loss": 0.4359,
+      "step": 11376
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.5202629566192627,
+      "learning_rate": 9.381319992289206e-06,
+      "loss": 0.9825,
+      "step": 11377
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.2364864349365234,
+      "learning_rate": 9.381192947434306e-06,
+      "loss": 0.7367,
+      "step": 11378
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.881882905960083,
+      "learning_rate": 9.38106589039695e-06,
+      "loss": 0.5838,
+      "step": 11379
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.9226677417755127,
+      "learning_rate": 9.38093882117749e-06,
+      "loss": 0.8651,
+      "step": 11380
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.491645574569702,
+      "learning_rate": 9.380811739776279e-06,
+      "loss": 0.865,
+      "step": 11381
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 4.489715099334717,
+      "learning_rate": 9.380684646193673e-06,
+      "loss": 0.5892,
+      "step": 11382
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.9162148237228394,
+      "learning_rate": 9.380557540430023e-06,
+      "loss": 0.8868,
+      "step": 11383
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.217621088027954,
+      "learning_rate": 9.380430422485685e-06,
+      "loss": 0.4012,
+      "step": 11384
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.4957773685455322,
+      "learning_rate": 9.38030329236101e-06,
+      "loss": 0.7792,
+      "step": 11385
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.012058973312378,
+      "learning_rate": 9.380176150056352e-06,
+      "loss": 0.4243,
+      "step": 11386
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.0815658569335938,
+      "learning_rate": 9.380048995572067e-06,
+      "loss": 1.0508,
+      "step": 11387
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.4013218879699707,
+      "learning_rate": 9.379921828908504e-06,
+      "loss": 0.8752,
+      "step": 11388
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.482520580291748,
+      "learning_rate": 9.379794650066021e-06,
+      "loss": 0.8236,
+      "step": 11389
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.7826476097106934,
+      "learning_rate": 9.379667459044968e-06,
+      "loss": 0.5771,
+      "step": 11390
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.5571117401123047,
+      "learning_rate": 9.379540255845702e-06,
+      "loss": 0.9169,
+      "step": 11391
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.394099473953247,
+      "learning_rate": 9.379413040468574e-06,
+      "loss": 0.6587,
+      "step": 11392
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.0282692909240723,
+      "learning_rate": 9.37928581291394e-06,
+      "loss": 0.5223,
+      "step": 11393
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.0968620777130127,
+      "learning_rate": 9.379158573182154e-06,
+      "loss": 0.6357,
+      "step": 11394
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.451953649520874,
+      "learning_rate": 9.379031321273567e-06,
+      "loss": 0.6042,
+      "step": 11395
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.033036231994629,
+      "learning_rate": 9.378904057188534e-06,
+      "loss": 0.6199,
+      "step": 11396
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.5770809650421143,
+      "learning_rate": 9.378776780927408e-06,
+      "loss": 0.6341,
+      "step": 11397
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.5486972332000732,
+      "learning_rate": 9.378649492490545e-06,
+      "loss": 0.5405,
+      "step": 11398
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.8021129369735718,
+      "learning_rate": 9.3785221918783e-06,
+      "loss": 0.6431,
+      "step": 11399
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.3796629905700684,
+      "learning_rate": 9.378394879091023e-06,
+      "loss": 0.5512,
+      "step": 11400
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.511936902999878,
+      "learning_rate": 9.37826755412907e-06,
+      "loss": 0.5786,
+      "step": 11401
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 5.867277145385742,
+      "learning_rate": 9.378140216992796e-06,
+      "loss": 0.866,
+      "step": 11402
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 4.70305061340332,
+      "learning_rate": 9.378012867682554e-06,
+      "loss": 0.6956,
+      "step": 11403
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.2387566566467285,
+      "learning_rate": 9.377885506198698e-06,
+      "loss": 0.6942,
+      "step": 11404
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.245187282562256,
+      "learning_rate": 9.377758132541581e-06,
+      "loss": 0.4462,
+      "step": 11405
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.182913303375244,
+      "learning_rate": 9.377630746711561e-06,
+      "loss": 0.5944,
+      "step": 11406
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.3173322677612305,
+      "learning_rate": 9.377503348708988e-06,
+      "loss": 0.7035,
+      "step": 11407
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.3424129486083984,
+      "learning_rate": 9.37737593853422e-06,
+      "loss": 0.4575,
+      "step": 11408
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.060633420944214,
+      "learning_rate": 9.377248516187608e-06,
+      "loss": 0.4489,
+      "step": 11409
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.4399945735931396,
+      "learning_rate": 9.377121081669507e-06,
+      "loss": 0.7073,
+      "step": 11410
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.840198278427124,
+      "learning_rate": 9.376993634980273e-06,
+      "loss": 0.6594,
+      "step": 11411
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.882866144180298,
+      "learning_rate": 9.376866176120258e-06,
+      "loss": 0.8798,
+      "step": 11412
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.6709911823272705,
+      "learning_rate": 9.376738705089819e-06,
+      "loss": 0.6269,
+      "step": 11413
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.7989749908447266,
+      "learning_rate": 9.37661122188931e-06,
+      "loss": 0.5891,
+      "step": 11414
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.176198720932007,
+      "learning_rate": 9.376483726519083e-06,
+      "loss": 0.7257,
+      "step": 11415
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.154407501220703,
+      "learning_rate": 9.376356218979496e-06,
+      "loss": 0.6578,
+      "step": 11416
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 9.767399787902832,
+      "learning_rate": 9.376228699270899e-06,
+      "loss": 0.7375,
+      "step": 11417
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.406205415725708,
+      "learning_rate": 9.376101167393652e-06,
+      "loss": 0.5001,
+      "step": 11418
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.9059438705444336,
+      "learning_rate": 9.375973623348106e-06,
+      "loss": 0.6937,
+      "step": 11419
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.369929313659668,
+      "learning_rate": 9.375846067134617e-06,
+      "loss": 0.8753,
+      "step": 11420
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.5811753273010254,
+      "learning_rate": 9.375718498753538e-06,
+      "loss": 0.5964,
+      "step": 11421
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.540254831314087,
+      "learning_rate": 9.375590918205228e-06,
+      "loss": 0.5197,
+      "step": 11422
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.1309092044830322,
+      "learning_rate": 9.375463325490036e-06,
+      "loss": 0.7509,
+      "step": 11423
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 4.494478225708008,
+      "learning_rate": 9.37533572060832e-06,
+      "loss": 0.8876,
+      "step": 11424
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.6379735469818115,
+      "learning_rate": 9.375208103560437e-06,
+      "loss": 1.0392,
+      "step": 11425
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.0561342239379883,
+      "learning_rate": 9.375080474346736e-06,
+      "loss": 0.8565,
+      "step": 11426
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.9014644622802734,
+      "learning_rate": 9.374952832967577e-06,
+      "loss": 0.7632,
+      "step": 11427
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 5.048196792602539,
+      "learning_rate": 9.37482517942331e-06,
+      "loss": 0.6111,
+      "step": 11428
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.2013559341430664,
+      "learning_rate": 9.374697513714297e-06,
+      "loss": 0.6602,
+      "step": 11429
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.2219552993774414,
+      "learning_rate": 9.374569835840886e-06,
+      "loss": 0.9295,
+      "step": 11430
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.7139501571655273,
+      "learning_rate": 9.374442145803438e-06,
+      "loss": 0.7615,
+      "step": 11431
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.003939628601074,
+      "learning_rate": 9.374314443602302e-06,
+      "loss": 0.4354,
+      "step": 11432
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.9602051973342896,
+      "learning_rate": 9.374186729237838e-06,
+      "loss": 0.731,
+      "step": 11433
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.9285081624984741,
+      "learning_rate": 9.374059002710399e-06,
+      "loss": 0.8566,
+      "step": 11434
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.15179181098938,
+      "learning_rate": 9.37393126402034e-06,
+      "loss": 0.8069,
+      "step": 11435
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.8017570972442627,
+      "learning_rate": 9.373803513168018e-06,
+      "loss": 0.6661,
+      "step": 11436
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.1124117374420166,
+      "learning_rate": 9.373675750153786e-06,
+      "loss": 0.4623,
+      "step": 11437
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.290571928024292,
+      "learning_rate": 9.373547974978e-06,
+      "loss": 0.6069,
+      "step": 11438
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.8318822383880615,
+      "learning_rate": 9.373420187641015e-06,
+      "loss": 0.5973,
+      "step": 11439
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.312574863433838,
+      "learning_rate": 9.373292388143189e-06,
+      "loss": 0.8357,
+      "step": 11440
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.81648588180542,
+      "learning_rate": 9.373164576484873e-06,
+      "loss": 0.5147,
+      "step": 11441
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.5350947380065918,
+      "learning_rate": 9.373036752666425e-06,
+      "loss": 0.9121,
+      "step": 11442
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.4698214530944824,
+      "learning_rate": 9.372908916688198e-06,
+      "loss": 0.709,
+      "step": 11443
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.642137050628662,
+      "learning_rate": 9.372781068550553e-06,
+      "loss": 0.6058,
+      "step": 11444
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.4381072521209717,
+      "learning_rate": 9.372653208253842e-06,
+      "loss": 0.4264,
+      "step": 11445
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.860438108444214,
+      "learning_rate": 9.372525335798416e-06,
+      "loss": 0.6781,
+      "step": 11446
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 5.928210735321045,
+      "learning_rate": 9.37239745118464e-06,
+      "loss": 0.7771,
+      "step": 11447
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.542754650115967,
+      "learning_rate": 9.372269554412863e-06,
+      "loss": 0.6912,
+      "step": 11448
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.5648043155670166,
+      "learning_rate": 9.372141645483441e-06,
+      "loss": 0.688,
+      "step": 11449
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.611751079559326,
+      "learning_rate": 9.372013724396732e-06,
+      "loss": 0.6648,
+      "step": 11450
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.1541428565979004,
+      "learning_rate": 9.371885791153091e-06,
+      "loss": 0.6675,
+      "step": 11451
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.3672752380371094,
+      "learning_rate": 9.371757845752872e-06,
+      "loss": 0.7853,
+      "step": 11452
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.256044864654541,
+      "learning_rate": 9.371629888196432e-06,
+      "loss": 0.5786,
+      "step": 11453
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.157630681991577,
+      "learning_rate": 9.371501918484128e-06,
+      "loss": 0.6029,
+      "step": 11454
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.114455461502075,
+      "learning_rate": 9.371373936616315e-06,
+      "loss": 0.8681,
+      "step": 11455
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.7438806295394897,
+      "learning_rate": 9.371245942593348e-06,
+      "loss": 0.7522,
+      "step": 11456
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.4911868572235107,
+      "learning_rate": 9.371117936415583e-06,
+      "loss": 0.7218,
+      "step": 11457
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.110321283340454,
+      "learning_rate": 9.370989918083377e-06,
+      "loss": 0.6833,
+      "step": 11458
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.9347962141036987,
+      "learning_rate": 9.370861887597088e-06,
+      "loss": 0.7765,
+      "step": 11459
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.378023624420166,
+      "learning_rate": 9.370733844957065e-06,
+      "loss": 0.4053,
+      "step": 11460
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.3722333908081055,
+      "learning_rate": 9.370605790163672e-06,
+      "loss": 0.7636,
+      "step": 11461
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.3372535705566406,
+      "learning_rate": 9.370477723217259e-06,
+      "loss": 0.6571,
+      "step": 11462
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.0724856853485107,
+      "learning_rate": 9.370349644118186e-06,
+      "loss": 0.648,
+      "step": 11463
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.555164337158203,
+      "learning_rate": 9.370221552866807e-06,
+      "loss": 0.7279,
+      "step": 11464
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.403376340866089,
+      "learning_rate": 9.37009344946348e-06,
+      "loss": 0.4692,
+      "step": 11465
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.6450438499450684,
+      "learning_rate": 9.36996533390856e-06,
+      "loss": 0.669,
+      "step": 11466
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.2129271030426025,
+      "learning_rate": 9.369837206202403e-06,
+      "loss": 0.6503,
+      "step": 11467
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.057771921157837,
+      "learning_rate": 9.369709066345367e-06,
+      "loss": 0.8226,
+      "step": 11468
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.530261993408203,
+      "learning_rate": 9.369580914337805e-06,
+      "loss": 0.6251,
+      "step": 11469
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 4.7804436683654785,
+      "learning_rate": 9.369452750180077e-06,
+      "loss": 0.9164,
+      "step": 11470
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.4889286756515503,
+      "learning_rate": 9.369324573872536e-06,
+      "loss": 0.3495,
+      "step": 11471
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.0325183868408203,
+      "learning_rate": 9.369196385415542e-06,
+      "loss": 0.5989,
+      "step": 11472
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.3640084266662598,
+      "learning_rate": 9.369068184809449e-06,
+      "loss": 0.8881,
+      "step": 11473
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.636804223060608,
+      "learning_rate": 9.368939972054613e-06,
+      "loss": 0.3683,
+      "step": 11474
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.0721397399902344,
+      "learning_rate": 9.368811747151392e-06,
+      "loss": 0.7968,
+      "step": 11475
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.9091076850891113,
+      "learning_rate": 9.368683510100143e-06,
+      "loss": 0.8541,
+      "step": 11476
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.5974600315093994,
+      "learning_rate": 9.368555260901222e-06,
+      "loss": 0.4465,
+      "step": 11477
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.242008924484253,
+      "learning_rate": 9.368426999554985e-06,
+      "loss": 0.3825,
+      "step": 11478
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.8190598487854004,
+      "learning_rate": 9.368298726061788e-06,
+      "loss": 0.6728,
+      "step": 11479
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.6509288549423218,
+      "learning_rate": 9.368170440421988e-06,
+      "loss": 0.4768,
+      "step": 11480
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.3694489002227783,
+      "learning_rate": 9.368042142635944e-06,
+      "loss": 0.6276,
+      "step": 11481
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.337794303894043,
+      "learning_rate": 9.36791383270401e-06,
+      "loss": 0.611,
+      "step": 11482
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.8326539993286133,
+      "learning_rate": 9.367785510626546e-06,
+      "loss": 0.8443,
+      "step": 11483
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.313600778579712,
+      "learning_rate": 9.367657176403904e-06,
+      "loss": 0.5503,
+      "step": 11484
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.6648250818252563,
+      "learning_rate": 9.367528830036445e-06,
+      "loss": 0.4153,
+      "step": 11485
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.4749722480773926,
+      "learning_rate": 9.367400471524525e-06,
+      "loss": 0.6559,
+      "step": 11486
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 4.7582807540893555,
+      "learning_rate": 9.3672721008685e-06,
+      "loss": 0.9018,
+      "step": 11487
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.7568222284317017,
+      "learning_rate": 9.367143718068727e-06,
+      "loss": 0.5773,
+      "step": 11488
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.7017732858657837,
+      "learning_rate": 9.367015323125563e-06,
+      "loss": 0.4741,
+      "step": 11489
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.6289901733398438,
+      "learning_rate": 9.366886916039365e-06,
+      "loss": 0.7583,
+      "step": 11490
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.8408421277999878,
+      "learning_rate": 9.36675849681049e-06,
+      "loss": 0.3783,
+      "step": 11491
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.0178275108337402,
+      "learning_rate": 9.366630065439297e-06,
+      "loss": 0.7043,
+      "step": 11492
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.073237657546997,
+      "learning_rate": 9.36650162192614e-06,
+      "loss": 0.7553,
+      "step": 11493
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.640230655670166,
+      "learning_rate": 9.36637316627138e-06,
+      "loss": 0.7621,
+      "step": 11494
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.847668409347534,
+      "learning_rate": 9.366244698475371e-06,
+      "loss": 0.8448,
+      "step": 11495
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.068347930908203,
+      "learning_rate": 9.36611621853847e-06,
+      "loss": 0.745,
+      "step": 11496
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.5971901416778564,
+      "learning_rate": 9.365987726461037e-06,
+      "loss": 0.4809,
+      "step": 11497
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.3498793840408325,
+      "learning_rate": 9.365859222243425e-06,
+      "loss": 0.5403,
+      "step": 11498
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.5607253313064575,
+      "learning_rate": 9.365730705885995e-06,
+      "loss": 0.3255,
+      "step": 11499
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.2357072830200195,
+      "learning_rate": 9.365602177389105e-06,
+      "loss": 0.6411,
+      "step": 11500
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.6265007257461548,
+      "learning_rate": 9.365473636753111e-06,
+      "loss": 0.5562,
+      "step": 11501
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 5.1494035720825195,
+      "learning_rate": 9.365345083978368e-06,
+      "loss": 0.802,
+      "step": 11502
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.2411277294158936,
+      "learning_rate": 9.365216519065236e-06,
+      "loss": 0.6923,
+      "step": 11503
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.0836541652679443,
+      "learning_rate": 9.365087942014073e-06,
+      "loss": 0.7278,
+      "step": 11504
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.2699785232543945,
+      "learning_rate": 9.364959352825235e-06,
+      "loss": 0.7305,
+      "step": 11505
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.516326427459717,
+      "learning_rate": 9.36483075149908e-06,
+      "loss": 0.4404,
+      "step": 11506
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.8001492023468018,
+      "learning_rate": 9.364702138035968e-06,
+      "loss": 0.7574,
+      "step": 11507
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 4.436690330505371,
+      "learning_rate": 9.364573512436252e-06,
+      "loss": 0.6377,
+      "step": 11508
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.061342477798462,
+      "learning_rate": 9.364444874700293e-06,
+      "loss": 1.0188,
+      "step": 11509
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 4.786138534545898,
+      "learning_rate": 9.364316224828446e-06,
+      "loss": 0.5699,
+      "step": 11510
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.978698492050171,
+      "learning_rate": 9.364187562821073e-06,
+      "loss": 0.4345,
+      "step": 11511
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.9910151958465576,
+      "learning_rate": 9.364058888678527e-06,
+      "loss": 0.8749,
+      "step": 11512
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.3207759857177734,
+      "learning_rate": 9.36393020240117e-06,
+      "loss": 0.6361,
+      "step": 11513
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.187889575958252,
+      "learning_rate": 9.363801503989358e-06,
+      "loss": 0.5434,
+      "step": 11514
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 4.921619892120361,
+      "learning_rate": 9.363672793443446e-06,
+      "loss": 0.4913,
+      "step": 11515
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 9.176040649414062,
+      "learning_rate": 9.363544070763797e-06,
+      "loss": 0.6614,
+      "step": 11516
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.9560686349868774,
+      "learning_rate": 9.363415335950767e-06,
+      "loss": 0.5489,
+      "step": 11517
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.173588275909424,
+      "learning_rate": 9.363286589004711e-06,
+      "loss": 0.7315,
+      "step": 11518
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.231006145477295,
+      "learning_rate": 9.363157829925993e-06,
+      "loss": 0.8896,
+      "step": 11519
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.012934684753418,
+      "learning_rate": 9.363029058714965e-06,
+      "loss": 0.6406,
+      "step": 11520
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.876489520072937,
+      "learning_rate": 9.362900275371988e-06,
+      "loss": 0.7753,
+      "step": 11521
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.27555775642395,
+      "learning_rate": 9.362771479897421e-06,
+      "loss": 0.578,
+      "step": 11522
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.8202145099639893,
+      "learning_rate": 9.36264267229162e-06,
+      "loss": 0.6051,
+      "step": 11523
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 5.5173115730285645,
+      "learning_rate": 9.362513852554943e-06,
+      "loss": 0.7002,
+      "step": 11524
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.8427557945251465,
+      "learning_rate": 9.362385020687749e-06,
+      "loss": 0.5505,
+      "step": 11525
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.9896926879882812,
+      "learning_rate": 9.362256176690398e-06,
+      "loss": 0.8829,
+      "step": 11526
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.96372389793396,
+      "learning_rate": 9.362127320563246e-06,
+      "loss": 0.4855,
+      "step": 11527
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.9720659255981445,
+      "learning_rate": 9.361998452306653e-06,
+      "loss": 0.7283,
+      "step": 11528
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.9526526927947998,
+      "learning_rate": 9.361869571920975e-06,
+      "loss": 0.7869,
+      "step": 11529
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.7107036113739014,
+      "learning_rate": 9.361740679406572e-06,
+      "loss": 0.4623,
+      "step": 11530
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.8468860387802124,
+      "learning_rate": 9.361611774763804e-06,
+      "loss": 0.7135,
+      "step": 11531
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.4041560888290405,
+      "learning_rate": 9.361482857993024e-06,
+      "loss": 0.8453,
+      "step": 11532
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.543905258178711,
+      "learning_rate": 9.361353929094597e-06,
+      "loss": 0.8474,
+      "step": 11533
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.3008201122283936,
+      "learning_rate": 9.361224988068878e-06,
+      "loss": 0.5429,
+      "step": 11534
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.6368255615234375,
+      "learning_rate": 9.361096034916224e-06,
+      "loss": 0.7047,
+      "step": 11535
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.6332478523254395,
+      "learning_rate": 9.360967069637e-06,
+      "loss": 0.7428,
+      "step": 11536
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.796046018600464,
+      "learning_rate": 9.360838092231557e-06,
+      "loss": 0.3712,
+      "step": 11537
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.8762998580932617,
+      "learning_rate": 9.360709102700256e-06,
+      "loss": 1.0141,
+      "step": 11538
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.418567657470703,
+      "learning_rate": 9.360580101043457e-06,
+      "loss": 0.8567,
+      "step": 11539
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.192120313644409,
+      "learning_rate": 9.36045108726152e-06,
+      "loss": 0.8402,
+      "step": 11540
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.986350417137146,
+      "learning_rate": 9.3603220613548e-06,
+      "loss": 0.8104,
+      "step": 11541
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.5143144130706787,
+      "learning_rate": 9.36019302332366e-06,
+      "loss": 0.7234,
+      "step": 11542
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.8577046394348145,
+      "learning_rate": 9.360063973168455e-06,
+      "loss": 0.764,
+      "step": 11543
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.8967267274856567,
+      "learning_rate": 9.359934910889546e-06,
+      "loss": 0.551,
+      "step": 11544
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.667419672012329,
+      "learning_rate": 9.35980583648729e-06,
+      "loss": 0.6307,
+      "step": 11545
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.0254313945770264,
+      "learning_rate": 9.35967674996205e-06,
+      "loss": 0.3938,
+      "step": 11546
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.839027166366577,
+      "learning_rate": 9.35954765131418e-06,
+      "loss": 0.6923,
+      "step": 11547
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.6440560817718506,
+      "learning_rate": 9.359418540544041e-06,
+      "loss": 0.6291,
+      "step": 11548
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.9146807193756104,
+      "learning_rate": 9.359289417651994e-06,
+      "loss": 0.9165,
+      "step": 11549
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.763216018676758,
+      "learning_rate": 9.359160282638393e-06,
+      "loss": 0.7149,
+      "step": 11550
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 169.72564697265625,
+      "learning_rate": 9.359031135503602e-06,
+      "loss": 0.678,
+      "step": 11551
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.213226556777954,
+      "learning_rate": 9.35890197624798e-06,
+      "loss": 0.6044,
+      "step": 11552
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.2740273475646973,
+      "learning_rate": 9.358772804871883e-06,
+      "loss": 0.8817,
+      "step": 11553
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.4841153621673584,
+      "learning_rate": 9.35864362137567e-06,
+      "loss": 0.6663,
+      "step": 11554
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.146599054336548,
+      "learning_rate": 9.358514425759703e-06,
+      "loss": 0.7322,
+      "step": 11555
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.7284823656082153,
+      "learning_rate": 9.358385218024342e-06,
+      "loss": 0.74,
+      "step": 11556
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.1923491954803467,
+      "learning_rate": 9.358255998169942e-06,
+      "loss": 0.8383,
+      "step": 11557
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.0981638431549072,
+      "learning_rate": 9.358126766196867e-06,
+      "loss": 0.6204,
+      "step": 11558
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.5668678283691406,
+      "learning_rate": 9.357997522105472e-06,
+      "loss": 0.5363,
+      "step": 11559
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.9222867488861084,
+      "learning_rate": 9.357868265896119e-06,
+      "loss": 0.5306,
+      "step": 11560
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.6488419771194458,
+      "learning_rate": 9.357738997569166e-06,
+      "loss": 0.9181,
+      "step": 11561
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.4568681716918945,
+      "learning_rate": 9.357609717124976e-06,
+      "loss": 0.798,
+      "step": 11562
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.662583351135254,
+      "learning_rate": 9.357480424563903e-06,
+      "loss": 0.9744,
+      "step": 11563
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.5926730632781982,
+      "learning_rate": 9.357351119886311e-06,
+      "loss": 0.4981,
+      "step": 11564
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.348235845565796,
+      "learning_rate": 9.357221803092557e-06,
+      "loss": 0.5666,
+      "step": 11565
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.9464435577392578,
+      "learning_rate": 9.357092474183e-06,
+      "loss": 0.7586,
+      "step": 11566
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.271867275238037,
+      "learning_rate": 9.356963133158003e-06,
+      "loss": 0.858,
+      "step": 11567
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.8925464153289795,
+      "learning_rate": 9.356833780017925e-06,
+      "loss": 0.7255,
+      "step": 11568
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.2207179069519043,
+      "learning_rate": 9.35670441476312e-06,
+      "loss": 0.5675,
+      "step": 11569
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.520810842514038,
+      "learning_rate": 9.356575037393956e-06,
+      "loss": 0.6443,
+      "step": 11570
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.612301826477051,
+      "learning_rate": 9.356445647910787e-06,
+      "loss": 0.6268,
+      "step": 11571
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.5186707973480225,
+      "learning_rate": 9.356316246313975e-06,
+      "loss": 0.6016,
+      "step": 11572
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.923858165740967,
+      "learning_rate": 9.35618683260388e-06,
+      "loss": 0.7515,
+      "step": 11573
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.440225601196289,
+      "learning_rate": 9.35605740678086e-06,
+      "loss": 0.6204,
+      "step": 11574
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.1005940437316895,
+      "learning_rate": 9.355927968845279e-06,
+      "loss": 0.4886,
+      "step": 11575
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.8481335639953613,
+      "learning_rate": 9.355798518797493e-06,
+      "loss": 0.5883,
+      "step": 11576
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.818028211593628,
+      "learning_rate": 9.35566905663786e-06,
+      "loss": 0.794,
+      "step": 11577
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.4002180099487305,
+      "learning_rate": 9.355539582366746e-06,
+      "loss": 0.7091,
+      "step": 11578
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.7654304504394531,
+      "learning_rate": 9.355410095984507e-06,
+      "loss": 0.7214,
+      "step": 11579
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 8.004246711730957,
+      "learning_rate": 9.355280597491505e-06,
+      "loss": 0.8364,
+      "step": 11580
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.0542385578155518,
+      "learning_rate": 9.355151086888098e-06,
+      "loss": 0.9213,
+      "step": 11581
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.9080415964126587,
+      "learning_rate": 9.355021564174648e-06,
+      "loss": 0.6935,
+      "step": 11582
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.5442166328430176,
+      "learning_rate": 9.354892029351515e-06,
+      "loss": 0.4896,
+      "step": 11583
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.240750551223755,
+      "learning_rate": 9.354762482419057e-06,
+      "loss": 0.9334,
+      "step": 11584
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.6961746215820312,
+      "learning_rate": 9.354632923377637e-06,
+      "loss": 0.7856,
+      "step": 11585
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.949275493621826,
+      "learning_rate": 9.354503352227614e-06,
+      "loss": 0.6904,
+      "step": 11586
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.536425828933716,
+      "learning_rate": 9.354373768969348e-06,
+      "loss": 0.6453,
+      "step": 11587
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.0650672912597656,
+      "learning_rate": 9.3542441736032e-06,
+      "loss": 0.7875,
+      "step": 11588
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.6569292545318604,
+      "learning_rate": 9.354114566129531e-06,
+      "loss": 0.5498,
+      "step": 11589
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.8203248977661133,
+      "learning_rate": 9.3539849465487e-06,
+      "loss": 0.5274,
+      "step": 11590
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.1783287525177,
+      "learning_rate": 9.353855314861066e-06,
+      "loss": 0.5099,
+      "step": 11591
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.8621255159378052,
+      "learning_rate": 9.353725671066994e-06,
+      "loss": 0.5057,
+      "step": 11592
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.556403398513794,
+      "learning_rate": 9.35359601516684e-06,
+      "loss": 0.753,
+      "step": 11593
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.9191430807113647,
+      "learning_rate": 9.353466347160968e-06,
+      "loss": 0.4966,
+      "step": 11594
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 4.293036937713623,
+      "learning_rate": 9.353336667049734e-06,
+      "loss": 0.8114,
+      "step": 11595
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.5871614217758179,
+      "learning_rate": 9.353206974833506e-06,
+      "loss": 0.6683,
+      "step": 11596
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.842824101448059,
+      "learning_rate": 9.353077270512637e-06,
+      "loss": 0.6583,
+      "step": 11597
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.123307228088379,
+      "learning_rate": 9.352947554087492e-06,
+      "loss": 0.8467,
+      "step": 11598
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.652726173400879,
+      "learning_rate": 9.352817825558429e-06,
+      "loss": 0.7396,
+      "step": 11599
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.227266788482666,
+      "learning_rate": 9.352688084925812e-06,
+      "loss": 0.8648,
+      "step": 11600
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.2804203033447266,
+      "learning_rate": 9.352558332189999e-06,
+      "loss": 0.5473,
+      "step": 11601
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 4.35103178024292,
+      "learning_rate": 9.352428567351352e-06,
+      "loss": 0.7794,
+      "step": 11602
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.324859380722046,
+      "learning_rate": 9.352298790410233e-06,
+      "loss": 0.9029,
+      "step": 11603
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.80016827583313,
+      "learning_rate": 9.352169001367e-06,
+      "loss": 0.3631,
+      "step": 11604
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 4.301228046417236,
+      "learning_rate": 9.352039200222014e-06,
+      "loss": 0.6287,
+      "step": 11605
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.823641061782837,
+      "learning_rate": 9.351909386975639e-06,
+      "loss": 0.4949,
+      "step": 11606
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 4.0539727210998535,
+      "learning_rate": 9.351779561628233e-06,
+      "loss": 0.6342,
+      "step": 11607
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.472867965698242,
+      "learning_rate": 9.35164972418016e-06,
+      "loss": 0.5793,
+      "step": 11608
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.562450885772705,
+      "learning_rate": 9.35151987463178e-06,
+      "loss": 0.7539,
+      "step": 11609
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.3858091831207275,
+      "learning_rate": 9.35139001298345e-06,
+      "loss": 0.5726,
+      "step": 11610
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.393691062927246,
+      "learning_rate": 9.351260139235537e-06,
+      "loss": 0.8364,
+      "step": 11611
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.937692403793335,
+      "learning_rate": 9.3511302533884e-06,
+      "loss": 0.5557,
+      "step": 11612
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.8815698623657227,
+      "learning_rate": 9.351000355442398e-06,
+      "loss": 0.756,
+      "step": 11613
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.042219400405884,
+      "learning_rate": 9.350870445397895e-06,
+      "loss": 0.6909,
+      "step": 11614
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.183722734451294,
+      "learning_rate": 9.35074052325525e-06,
+      "loss": 0.5814,
+      "step": 11615
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.09928035736084,
+      "learning_rate": 9.350610589014827e-06,
+      "loss": 0.7892,
+      "step": 11616
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.0357580184936523,
+      "learning_rate": 9.350480642676984e-06,
+      "loss": 0.7712,
+      "step": 11617
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.4424759149551392,
+      "learning_rate": 9.350350684242086e-06,
+      "loss": 0.4736,
+      "step": 11618
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.1659865379333496,
+      "learning_rate": 9.350220713710492e-06,
+      "loss": 0.7023,
+      "step": 11619
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.2648191452026367,
+      "learning_rate": 9.350090731082563e-06,
+      "loss": 0.7433,
+      "step": 11620
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.2228436470031738,
+      "learning_rate": 9.349960736358661e-06,
+      "loss": 0.9829,
+      "step": 11621
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.610452890396118,
+      "learning_rate": 9.349830729539149e-06,
+      "loss": 0.5037,
+      "step": 11622
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.929396629333496,
+      "learning_rate": 9.349700710624386e-06,
+      "loss": 0.7482,
+      "step": 11623
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.07749605178833,
+      "learning_rate": 9.349570679614734e-06,
+      "loss": 0.645,
+      "step": 11624
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.6635410785675049,
+      "learning_rate": 9.349440636510559e-06,
+      "loss": 0.7238,
+      "step": 11625
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.109414577484131,
+      "learning_rate": 9.349310581312216e-06,
+      "loss": 0.6474,
+      "step": 11626
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 4.83278751373291,
+      "learning_rate": 9.34918051402007e-06,
+      "loss": 0.5174,
+      "step": 11627
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.715319871902466,
+      "learning_rate": 9.34905043463448e-06,
+      "loss": 0.6864,
+      "step": 11628
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.4772415161132812,
+      "learning_rate": 9.348920343155814e-06,
+      "loss": 0.4762,
+      "step": 11629
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.006890058517456,
+      "learning_rate": 9.348790239584428e-06,
+      "loss": 0.5576,
+      "step": 11630
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.9950459003448486,
+      "learning_rate": 9.348660123920686e-06,
+      "loss": 0.7442,
+      "step": 11631
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.105970859527588,
+      "learning_rate": 9.348529996164947e-06,
+      "loss": 0.8812,
+      "step": 11632
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.4253616333007812,
+      "learning_rate": 9.348399856317578e-06,
+      "loss": 0.9266,
+      "step": 11633
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.8504674434661865,
+      "learning_rate": 9.348269704378936e-06,
+      "loss": 0.5193,
+      "step": 11634
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.3662233352661133,
+      "learning_rate": 9.348139540349386e-06,
+      "loss": 0.5532,
+      "step": 11635
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.386770009994507,
+      "learning_rate": 9.348009364229287e-06,
+      "loss": 0.5118,
+      "step": 11636
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.7069579362869263,
+      "learning_rate": 9.347879176019003e-06,
+      "loss": 0.6482,
+      "step": 11637
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.249622106552124,
+      "learning_rate": 9.347748975718897e-06,
+      "loss": 0.5657,
+      "step": 11638
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.970962405204773,
+      "learning_rate": 9.347618763329332e-06,
+      "loss": 0.6648,
+      "step": 11639
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 4.78306770324707,
+      "learning_rate": 9.347488538850664e-06,
+      "loss": 0.6152,
+      "step": 11640
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 4.473427772521973,
+      "learning_rate": 9.34735830228326e-06,
+      "loss": 0.7189,
+      "step": 11641
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.9442013502120972,
+      "learning_rate": 9.347228053627481e-06,
+      "loss": 0.8302,
+      "step": 11642
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.3914883136749268,
+      "learning_rate": 9.34709779288369e-06,
+      "loss": 0.7502,
+      "step": 11643
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.9464284181594849,
+      "learning_rate": 9.34696752005225e-06,
+      "loss": 0.459,
+      "step": 11644
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.2350120544433594,
+      "learning_rate": 9.346837235133518e-06,
+      "loss": 0.4261,
+      "step": 11645
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.0138471126556396,
+      "learning_rate": 9.346706938127862e-06,
+      "loss": 0.6627,
+      "step": 11646
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.5356762409210205,
+      "learning_rate": 9.346576629035644e-06,
+      "loss": 0.578,
+      "step": 11647
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.6438237428665161,
+      "learning_rate": 9.346446307857221e-06,
+      "loss": 0.8223,
+      "step": 11648
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.184431791305542,
+      "learning_rate": 9.346315974592963e-06,
+      "loss": 0.5849,
+      "step": 11649
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.485664129257202,
+      "learning_rate": 9.346185629243227e-06,
+      "loss": 0.7325,
+      "step": 11650
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.9859700202941895,
+      "learning_rate": 9.346055271808377e-06,
+      "loss": 0.6384,
+      "step": 11651
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.1582982540130615,
+      "learning_rate": 9.345924902288773e-06,
+      "loss": 0.7546,
+      "step": 11652
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.6166346073150635,
+      "learning_rate": 9.345794520684783e-06,
+      "loss": 0.701,
+      "step": 11653
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.8820767402648926,
+      "learning_rate": 9.345664126996764e-06,
+      "loss": 0.8632,
+      "step": 11654
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.627044916152954,
+      "learning_rate": 9.345533721225082e-06,
+      "loss": 0.7329,
+      "step": 11655
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 4.039275646209717,
+      "learning_rate": 9.3454033033701e-06,
+      "loss": 0.6452,
+      "step": 11656
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.271939516067505,
+      "learning_rate": 9.345272873432176e-06,
+      "loss": 0.6801,
+      "step": 11657
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 4.924674034118652,
+      "learning_rate": 9.345142431411677e-06,
+      "loss": 0.8389,
+      "step": 11658
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.5350351333618164,
+      "learning_rate": 9.345011977308967e-06,
+      "loss": 0.6997,
+      "step": 11659
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.7092154026031494,
+      "learning_rate": 9.344881511124404e-06,
+      "loss": 0.7065,
+      "step": 11660
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.9051575660705566,
+      "learning_rate": 9.344751032858355e-06,
+      "loss": 0.5399,
+      "step": 11661
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.5944981575012207,
+      "learning_rate": 9.34462054251118e-06,
+      "loss": 0.6319,
+      "step": 11662
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.488710403442383,
+      "learning_rate": 9.344490040083243e-06,
+      "loss": 0.8519,
+      "step": 11663
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.726936101913452,
+      "learning_rate": 9.344359525574905e-06,
+      "loss": 0.7584,
+      "step": 11664
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.273735284805298,
+      "learning_rate": 9.344228998986534e-06,
+      "loss": 0.4423,
+      "step": 11665
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 4.30676794052124,
+      "learning_rate": 9.344098460318485e-06,
+      "loss": 0.9938,
+      "step": 11666
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.6699331998825073,
+      "learning_rate": 9.343967909571128e-06,
+      "loss": 0.8898,
+      "step": 11667
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.1059277057647705,
+      "learning_rate": 9.343837346744824e-06,
+      "loss": 0.8289,
+      "step": 11668
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.6958677768707275,
+      "learning_rate": 9.343706771839935e-06,
+      "loss": 0.5741,
+      "step": 11669
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.503915548324585,
+      "learning_rate": 9.343576184856825e-06,
+      "loss": 0.8079,
+      "step": 11670
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.336745500564575,
+      "learning_rate": 9.343445585795856e-06,
+      "loss": 0.5684,
+      "step": 11671
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.9846529960632324,
+      "learning_rate": 9.343314974657394e-06,
+      "loss": 0.8329,
+      "step": 11672
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.1681511402130127,
+      "learning_rate": 9.343184351441798e-06,
+      "loss": 0.7814,
+      "step": 11673
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.7259182929992676,
+      "learning_rate": 9.343053716149433e-06,
+      "loss": 0.692,
+      "step": 11674
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.973252058029175,
+      "learning_rate": 9.342923068780666e-06,
+      "loss": 0.8838,
+      "step": 11675
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.2232465744018555,
+      "learning_rate": 9.342792409335852e-06,
+      "loss": 0.8616,
+      "step": 11676
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.4863991737365723,
+      "learning_rate": 9.342661737815361e-06,
+      "loss": 0.5686,
+      "step": 11677
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.8564995527267456,
+      "learning_rate": 9.342531054219555e-06,
+      "loss": 0.899,
+      "step": 11678
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.837302327156067,
+      "learning_rate": 9.342400358548798e-06,
+      "loss": 0.7684,
+      "step": 11679
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 4.397166728973389,
+      "learning_rate": 9.34226965080345e-06,
+      "loss": 0.7123,
+      "step": 11680
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 4.29719877243042,
+      "learning_rate": 9.342138930983879e-06,
+      "loss": 0.8758,
+      "step": 11681
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.416897773742676,
+      "learning_rate": 9.342008199090444e-06,
+      "loss": 0.465,
+      "step": 11682
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.161792755126953,
+      "learning_rate": 9.341877455123513e-06,
+      "loss": 0.7995,
+      "step": 11683
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.9696296453475952,
+      "learning_rate": 9.341746699083445e-06,
+      "loss": 0.8097,
+      "step": 11684
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.337395668029785,
+      "learning_rate": 9.341615930970607e-06,
+      "loss": 0.5283,
+      "step": 11685
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.411005735397339,
+      "learning_rate": 9.341485150785364e-06,
+      "loss": 0.5025,
+      "step": 11686
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.4363536834716797,
+      "learning_rate": 9.341354358528073e-06,
+      "loss": 0.655,
+      "step": 11687
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.100152015686035,
+      "learning_rate": 9.341223554199103e-06,
+      "loss": 0.7332,
+      "step": 11688
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.9121849536895752,
+      "learning_rate": 9.341092737798819e-06,
+      "loss": 0.6942,
+      "step": 11689
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.5907230377197266,
+      "learning_rate": 9.34096190932758e-06,
+      "loss": 0.5742,
+      "step": 11690
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.224238157272339,
+      "learning_rate": 9.340831068785752e-06,
+      "loss": 0.6056,
+      "step": 11691
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.260510802268982,
+      "learning_rate": 9.3407002161737e-06,
+      "loss": 0.9981,
+      "step": 11692
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.68904709815979,
+      "learning_rate": 9.340569351491787e-06,
+      "loss": 0.5286,
+      "step": 11693
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.4442479610443115,
+      "learning_rate": 9.340438474740375e-06,
+      "loss": 0.8422,
+      "step": 11694
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.6331398487091064,
+      "learning_rate": 9.34030758591983e-06,
+      "loss": 0.642,
+      "step": 11695
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.290520191192627,
+      "learning_rate": 9.340176685030515e-06,
+      "loss": 0.5085,
+      "step": 11696
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.72182035446167,
+      "learning_rate": 9.340045772072797e-06,
+      "loss": 0.6836,
+      "step": 11697
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.6908719539642334,
+      "learning_rate": 9.339914847047035e-06,
+      "loss": 0.7698,
+      "step": 11698
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.4605159759521484,
+      "learning_rate": 9.339783909953597e-06,
+      "loss": 0.6945,
+      "step": 11699
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 4.027771949768066,
+      "learning_rate": 9.339652960792845e-06,
+      "loss": 0.6497,
+      "step": 11700
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.8144252300262451,
+      "learning_rate": 9.339521999565145e-06,
+      "loss": 0.6749,
+      "step": 11701
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.634614944458008,
+      "learning_rate": 9.339391026270858e-06,
+      "loss": 0.6752,
+      "step": 11702
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.967916965484619,
+      "learning_rate": 9.33926004091035e-06,
+      "loss": 0.7003,
+      "step": 11703
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.1822683811187744,
+      "learning_rate": 9.339129043483988e-06,
+      "loss": 0.7486,
+      "step": 11704
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 4.487982273101807,
+      "learning_rate": 9.338998033992131e-06,
+      "loss": 0.7361,
+      "step": 11705
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.0636746883392334,
+      "learning_rate": 9.338867012435146e-06,
+      "loss": 0.7336,
+      "step": 11706
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.588618516921997,
+      "learning_rate": 9.3387359788134e-06,
+      "loss": 0.7457,
+      "step": 11707
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.379237174987793,
+      "learning_rate": 9.338604933127251e-06,
+      "loss": 1.0363,
+      "step": 11708
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.531512498855591,
+      "learning_rate": 9.33847387537707e-06,
+      "loss": 0.8199,
+      "step": 11709
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.172679901123047,
+      "learning_rate": 9.338342805563217e-06,
+      "loss": 0.8254,
+      "step": 11710
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.6660043001174927,
+      "learning_rate": 9.338211723686058e-06,
+      "loss": 0.5552,
+      "step": 11711
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.899930238723755,
+      "learning_rate": 9.338080629745955e-06,
+      "loss": 0.673,
+      "step": 11712
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.2984113693237305,
+      "learning_rate": 9.337949523743278e-06,
+      "loss": 0.6994,
+      "step": 11713
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.2125489711761475,
+      "learning_rate": 9.337818405678387e-06,
+      "loss": 0.6961,
+      "step": 11714
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.613878011703491,
+      "learning_rate": 9.33768727555165e-06,
+      "loss": 0.6859,
+      "step": 11715
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 6.80631685256958,
+      "learning_rate": 9.337556133363428e-06,
+      "loss": 0.729,
+      "step": 11716
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.7916969060897827,
+      "learning_rate": 9.337424979114087e-06,
+      "loss": 0.5568,
+      "step": 11717
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.9296796321868896,
+      "learning_rate": 9.337293812803992e-06,
+      "loss": 0.6965,
+      "step": 11718
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.3792104721069336,
+      "learning_rate": 9.337162634433507e-06,
+      "loss": 0.7292,
+      "step": 11719
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.558090925216675,
+      "learning_rate": 9.337031444003e-06,
+      "loss": 0.8211,
+      "step": 11720
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.430354118347168,
+      "learning_rate": 9.336900241512833e-06,
+      "loss": 0.7709,
+      "step": 11721
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.2126200199127197,
+      "learning_rate": 9.336769026963368e-06,
+      "loss": 0.4921,
+      "step": 11722
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.2047231197357178,
+      "learning_rate": 9.336637800354975e-06,
+      "loss": 0.6453,
+      "step": 11723
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.7307610511779785,
+      "learning_rate": 9.336506561688018e-06,
+      "loss": 0.674,
+      "step": 11724
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.4226629734039307,
+      "learning_rate": 9.336375310962857e-06,
+      "loss": 0.4985,
+      "step": 11725
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.460892915725708,
+      "learning_rate": 9.336244048179863e-06,
+      "loss": 0.6887,
+      "step": 11726
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.6217966079711914,
+      "learning_rate": 9.336112773339398e-06,
+      "loss": 0.7008,
+      "step": 11727
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.025752305984497,
+      "learning_rate": 9.33598148644183e-06,
+      "loss": 0.863,
+      "step": 11728
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.44295072555542,
+      "learning_rate": 9.335850187487519e-06,
+      "loss": 0.668,
+      "step": 11729
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.345877170562744,
+      "learning_rate": 9.335718876476834e-06,
+      "loss": 0.6211,
+      "step": 11730
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.4144492149353027,
+      "learning_rate": 9.335587553410139e-06,
+      "loss": 0.5284,
+      "step": 11731
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.7447233200073242,
+      "learning_rate": 9.335456218287799e-06,
+      "loss": 0.5971,
+      "step": 11732
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.74330735206604,
+      "learning_rate": 9.335324871110182e-06,
+      "loss": 0.5488,
+      "step": 11733
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.9631505012512207,
+      "learning_rate": 9.335193511877647e-06,
+      "loss": 0.5577,
+      "step": 11734
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.896094799041748,
+      "learning_rate": 9.335062140590563e-06,
+      "loss": 0.5269,
+      "step": 11735
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.6887965202331543,
+      "learning_rate": 9.334930757249296e-06,
+      "loss": 0.4662,
+      "step": 11736
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.0736141204833984,
+      "learning_rate": 9.334799361854213e-06,
+      "loss": 0.9885,
+      "step": 11737
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.2305569648742676,
+      "learning_rate": 9.334667954405673e-06,
+      "loss": 0.6676,
+      "step": 11738
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.506659984588623,
+      "learning_rate": 9.334536534904046e-06,
+      "loss": 0.8186,
+      "step": 11739
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.665491819381714,
+      "learning_rate": 9.334405103349697e-06,
+      "loss": 0.641,
+      "step": 11740
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.664147138595581,
+      "learning_rate": 9.334273659742991e-06,
+      "loss": 0.5634,
+      "step": 11741
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.258439302444458,
+      "learning_rate": 9.334142204084295e-06,
+      "loss": 0.6463,
+      "step": 11742
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.5413429737091064,
+      "learning_rate": 9.334010736373973e-06,
+      "loss": 0.5875,
+      "step": 11743
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.291658401489258,
+      "learning_rate": 9.333879256612389e-06,
+      "loss": 0.6105,
+      "step": 11744
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.642223596572876,
+      "learning_rate": 9.333747764799913e-06,
+      "loss": 0.5886,
+      "step": 11745
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.877098798751831,
+      "learning_rate": 9.333616260936904e-06,
+      "loss": 0.6083,
+      "step": 11746
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.9197009801864624,
+      "learning_rate": 9.333484745023734e-06,
+      "loss": 0.9981,
+      "step": 11747
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.279494524002075,
+      "learning_rate": 9.333353217060766e-06,
+      "loss": 0.5271,
+      "step": 11748
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.8454289436340332,
+      "learning_rate": 9.333221677048366e-06,
+      "loss": 0.5373,
+      "step": 11749
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.1449966430664062,
+      "learning_rate": 9.3330901249869e-06,
+      "loss": 0.6915,
+      "step": 11750
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.672494888305664,
+      "learning_rate": 9.332958560876732e-06,
+      "loss": 0.7796,
+      "step": 11751
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.6354206800460815,
+      "learning_rate": 9.332826984718233e-06,
+      "loss": 0.5046,
+      "step": 11752
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.126079559326172,
+      "learning_rate": 9.332695396511761e-06,
+      "loss": 0.7369,
+      "step": 11753
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.912869691848755,
+      "learning_rate": 9.33256379625769e-06,
+      "loss": 0.5489,
+      "step": 11754
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.401517391204834,
+      "learning_rate": 9.332432183956379e-06,
+      "loss": 0.6245,
+      "step": 11755
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.2945494651794434,
+      "learning_rate": 9.332300559608199e-06,
+      "loss": 0.629,
+      "step": 11756
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.174345016479492,
+      "learning_rate": 9.332168923213514e-06,
+      "loss": 0.6057,
+      "step": 11757
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.093583583831787,
+      "learning_rate": 9.332037274772688e-06,
+      "loss": 0.8772,
+      "step": 11758
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.047440767288208,
+      "learning_rate": 9.331905614286093e-06,
+      "loss": 0.5563,
+      "step": 11759
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.3980729579925537,
+      "learning_rate": 9.331773941754088e-06,
+      "loss": 0.8695,
+      "step": 11760
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.8789879083633423,
+      "learning_rate": 9.331642257177043e-06,
+      "loss": 0.5369,
+      "step": 11761
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.945452928543091,
+      "learning_rate": 9.331510560555323e-06,
+      "loss": 0.8385,
+      "step": 11762
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.8309476375579834,
+      "learning_rate": 9.331378851889295e-06,
+      "loss": 0.9478,
+      "step": 11763
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.984384775161743,
+      "learning_rate": 9.331247131179327e-06,
+      "loss": 0.5698,
+      "step": 11764
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.955205202102661,
+      "learning_rate": 9.331115398425781e-06,
+      "loss": 0.6189,
+      "step": 11765
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.063715934753418,
+      "learning_rate": 9.330983653629026e-06,
+      "loss": 0.7047,
+      "step": 11766
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.620577335357666,
+      "learning_rate": 9.330851896789428e-06,
+      "loss": 0.6114,
+      "step": 11767
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 4.627536296844482,
+      "learning_rate": 9.330720127907354e-06,
+      "loss": 0.5302,
+      "step": 11768
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.9588682651519775,
+      "learning_rate": 9.33058834698317e-06,
+      "loss": 0.7031,
+      "step": 11769
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.12095046043396,
+      "learning_rate": 9.33045655401724e-06,
+      "loss": 0.5964,
+      "step": 11770
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 4.5406718254089355,
+      "learning_rate": 9.330324749009934e-06,
+      "loss": 0.521,
+      "step": 11771
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.3320538997650146,
+      "learning_rate": 9.330192931961616e-06,
+      "loss": 1.0294,
+      "step": 11772
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.9115841388702393,
+      "learning_rate": 9.330061102872657e-06,
+      "loss": 0.5604,
+      "step": 11773
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.281156539916992,
+      "learning_rate": 9.329929261743416e-06,
+      "loss": 0.7569,
+      "step": 11774
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.7612615823745728,
+      "learning_rate": 9.329797408574265e-06,
+      "loss": 0.4702,
+      "step": 11775
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.661391019821167,
+      "learning_rate": 9.32966554336557e-06,
+      "loss": 0.8246,
+      "step": 11776
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.325244426727295,
+      "learning_rate": 9.329533666117697e-06,
+      "loss": 0.6149,
+      "step": 11777
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 4.327699661254883,
+      "learning_rate": 9.329401776831013e-06,
+      "loss": 0.6768,
+      "step": 11778
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.716422438621521,
+      "learning_rate": 9.329269875505884e-06,
+      "loss": 0.425,
+      "step": 11779
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.5573924779891968,
+      "learning_rate": 9.329137962142676e-06,
+      "loss": 0.7015,
+      "step": 11780
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.2901034355163574,
+      "learning_rate": 9.32900603674176e-06,
+      "loss": 0.6255,
+      "step": 11781
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.8192939758300781,
+      "learning_rate": 9.328874099303497e-06,
+      "loss": 0.534,
+      "step": 11782
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.987565517425537,
+      "learning_rate": 9.328742149828259e-06,
+      "loss": 0.8191,
+      "step": 11783
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 4.056896209716797,
+      "learning_rate": 9.32861018831641e-06,
+      "loss": 0.9226,
+      "step": 11784
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.8149008750915527,
+      "learning_rate": 9.328478214768316e-06,
+      "loss": 0.3527,
+      "step": 11785
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.004992961883545,
+      "learning_rate": 9.328346229184347e-06,
+      "loss": 0.6723,
+      "step": 11786
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.579662799835205,
+      "learning_rate": 9.328214231564869e-06,
+      "loss": 0.4217,
+      "step": 11787
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.846816301345825,
+      "learning_rate": 9.328082221910247e-06,
+      "loss": 0.5232,
+      "step": 11788
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.1108412742614746,
+      "learning_rate": 9.327950200220851e-06,
+      "loss": 0.6672,
+      "step": 11789
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.406278610229492,
+      "learning_rate": 9.327818166497046e-06,
+      "loss": 0.7544,
+      "step": 11790
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.9432435035705566,
+      "learning_rate": 9.3276861207392e-06,
+      "loss": 0.6273,
+      "step": 11791
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.1250940561294556,
+      "learning_rate": 9.327554062947682e-06,
+      "loss": 0.6436,
+      "step": 11792
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 5.4818949699401855,
+      "learning_rate": 9.327421993122855e-06,
+      "loss": 0.5562,
+      "step": 11793
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 4.62794828414917,
+      "learning_rate": 9.32728991126509e-06,
+      "loss": 0.4533,
+      "step": 11794
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.1997904777526855,
+      "learning_rate": 9.327157817374751e-06,
+      "loss": 0.5476,
+      "step": 11795
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.0086333751678467,
+      "learning_rate": 9.32702571145221e-06,
+      "loss": 0.7252,
+      "step": 11796
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.6794034242630005,
+      "learning_rate": 9.326893593497827e-06,
+      "loss": 0.4817,
+      "step": 11797
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.5250167846679688,
+      "learning_rate": 9.326761463511977e-06,
+      "loss": 0.9584,
+      "step": 11798
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.877162218093872,
+      "learning_rate": 9.326629321495024e-06,
+      "loss": 0.9159,
+      "step": 11799
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.07261061668396,
+      "learning_rate": 9.326497167447334e-06,
+      "loss": 0.6073,
+      "step": 11800
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.1223514080047607,
+      "learning_rate": 9.326365001369278e-06,
+      "loss": 0.6038,
+      "step": 11801
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.6748740673065186,
+      "learning_rate": 9.326232823261221e-06,
+      "loss": 0.5101,
+      "step": 11802
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.410881519317627,
+      "learning_rate": 9.326100633123532e-06,
+      "loss": 0.7209,
+      "step": 11803
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.4362757205963135,
+      "learning_rate": 9.325968430956577e-06,
+      "loss": 0.7813,
+      "step": 11804
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 5.048228740692139,
+      "learning_rate": 9.325836216760723e-06,
+      "loss": 0.6668,
+      "step": 11805
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.0876543521881104,
+      "learning_rate": 9.325703990536342e-06,
+      "loss": 0.6374,
+      "step": 11806
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.9472393989562988,
+      "learning_rate": 9.325571752283796e-06,
+      "loss": 0.7139,
+      "step": 11807
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.8731160163879395,
+      "learning_rate": 9.325439502003457e-06,
+      "loss": 0.6805,
+      "step": 11808
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.9112095832824707,
+      "learning_rate": 9.325307239695688e-06,
+      "loss": 0.6782,
+      "step": 11809
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.0106403827667236,
+      "learning_rate": 9.325174965360864e-06,
+      "loss": 0.7368,
+      "step": 11810
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.0311734676361084,
+      "learning_rate": 9.325042678999347e-06,
+      "loss": 0.8654,
+      "step": 11811
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.848191499710083,
+      "learning_rate": 9.324910380611506e-06,
+      "loss": 0.4545,
+      "step": 11812
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.5481531620025635,
+      "learning_rate": 9.32477807019771e-06,
+      "loss": 0.8029,
+      "step": 11813
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 4.439029216766357,
+      "learning_rate": 9.324645747758326e-06,
+      "loss": 0.6661,
+      "step": 11814
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.9828609228134155,
+      "learning_rate": 9.324513413293722e-06,
+      "loss": 0.4828,
+      "step": 11815
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.9961422681808472,
+      "learning_rate": 9.324381066804267e-06,
+      "loss": 0.7948,
+      "step": 11816
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 4.2561936378479,
+      "learning_rate": 9.324248708290327e-06,
+      "loss": 0.8774,
+      "step": 11817
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.362780809402466,
+      "learning_rate": 9.324116337752272e-06,
+      "loss": 0.4614,
+      "step": 11818
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.9104914665222168,
+      "learning_rate": 9.32398395519047e-06,
+      "loss": 0.5322,
+      "step": 11819
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.4034743309020996,
+      "learning_rate": 9.323851560605287e-06,
+      "loss": 0.5193,
+      "step": 11820
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.01938796043396,
+      "learning_rate": 9.323719153997092e-06,
+      "loss": 0.5094,
+      "step": 11821
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.411440134048462,
+      "learning_rate": 9.323586735366256e-06,
+      "loss": 0.706,
+      "step": 11822
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.5928313732147217,
+      "learning_rate": 9.323454304713144e-06,
+      "loss": 0.6599,
+      "step": 11823
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.8476972579956055,
+      "learning_rate": 9.323321862038124e-06,
+      "loss": 0.8057,
+      "step": 11824
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.450335741043091,
+      "learning_rate": 9.323189407341567e-06,
+      "loss": 0.5719,
+      "step": 11825
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.8777459859848022,
+      "learning_rate": 9.32305694062384e-06,
+      "loss": 0.7727,
+      "step": 11826
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.9543933868408203,
+      "learning_rate": 9.32292446188531e-06,
+      "loss": 0.8218,
+      "step": 11827
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.1263625621795654,
+      "learning_rate": 9.322791971126347e-06,
+      "loss": 0.9211,
+      "step": 11828
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.005045175552368,
+      "learning_rate": 9.32265946834732e-06,
+      "loss": 0.5978,
+      "step": 11829
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 4.203628063201904,
+      "learning_rate": 9.322526953548594e-06,
+      "loss": 0.5183,
+      "step": 11830
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.268705368041992,
+      "learning_rate": 9.322394426730541e-06,
+      "loss": 0.7083,
+      "step": 11831
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.2227132320404053,
+      "learning_rate": 9.322261887893527e-06,
+      "loss": 0.7856,
+      "step": 11832
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.5554585456848145,
+      "learning_rate": 9.322129337037924e-06,
+      "loss": 0.6541,
+      "step": 11833
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.9967398643493652,
+      "learning_rate": 9.321996774164098e-06,
+      "loss": 0.4646,
+      "step": 11834
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.2331840991973877,
+      "learning_rate": 9.321864199272417e-06,
+      "loss": 0.45,
+      "step": 11835
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.0744028091430664,
+      "learning_rate": 9.321731612363251e-06,
+      "loss": 0.5348,
+      "step": 11836
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.2758102416992188,
+      "learning_rate": 9.321599013436968e-06,
+      "loss": 0.6741,
+      "step": 11837
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.920168399810791,
+      "learning_rate": 9.321466402493937e-06,
+      "loss": 0.5957,
+      "step": 11838
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.58979868888855,
+      "learning_rate": 9.321333779534528e-06,
+      "loss": 0.8538,
+      "step": 11839
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.570474147796631,
+      "learning_rate": 9.321201144559108e-06,
+      "loss": 0.5471,
+      "step": 11840
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.672578811645508,
+      "learning_rate": 9.321068497568047e-06,
+      "loss": 0.7001,
+      "step": 11841
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.0753767490386963,
+      "learning_rate": 9.320935838561711e-06,
+      "loss": 0.8125,
+      "step": 11842
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.2144668102264404,
+      "learning_rate": 9.320803167540473e-06,
+      "loss": 0.516,
+      "step": 11843
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.3157191276550293,
+      "learning_rate": 9.3206704845047e-06,
+      "loss": 0.6103,
+      "step": 11844
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 5.4449543952941895,
+      "learning_rate": 9.32053778945476e-06,
+      "loss": 0.5865,
+      "step": 11845
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.464570164680481,
+      "learning_rate": 9.320405082391024e-06,
+      "loss": 0.6435,
+      "step": 11846
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.0591275691986084,
+      "learning_rate": 9.320272363313859e-06,
+      "loss": 0.6863,
+      "step": 11847
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.0234735012054443,
+      "learning_rate": 9.320139632223634e-06,
+      "loss": 0.5942,
+      "step": 11848
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.9855382442474365,
+      "learning_rate": 9.320006889120722e-06,
+      "loss": 0.7746,
+      "step": 11849
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.225980520248413,
+      "learning_rate": 9.319874134005487e-06,
+      "loss": 0.5079,
+      "step": 11850
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.7776286602020264,
+      "learning_rate": 9.319741366878301e-06,
+      "loss": 0.6047,
+      "step": 11851
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.7064229249954224,
+      "learning_rate": 9.319608587739533e-06,
+      "loss": 0.5088,
+      "step": 11852
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.498041868209839,
+      "learning_rate": 9.31947579658955e-06,
+      "loss": 0.7254,
+      "step": 11853
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.3360519409179688,
+      "learning_rate": 9.319342993428723e-06,
+      "loss": 0.5937,
+      "step": 11854
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.40384840965271,
+      "learning_rate": 9.319210178257423e-06,
+      "loss": 0.4685,
+      "step": 11855
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.822266101837158,
+      "learning_rate": 9.319077351076016e-06,
+      "loss": 0.7697,
+      "step": 11856
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.103971481323242,
+      "learning_rate": 9.318944511884875e-06,
+      "loss": 0.4604,
+      "step": 11857
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.024325370788574,
+      "learning_rate": 9.318811660684365e-06,
+      "loss": 0.4173,
+      "step": 11858
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.9204814434051514,
+      "learning_rate": 9.318678797474858e-06,
+      "loss": 0.6186,
+      "step": 11859
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.6499335765838623,
+      "learning_rate": 9.318545922256724e-06,
+      "loss": 0.5962,
+      "step": 11860
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.3636138439178467,
+      "learning_rate": 9.318413035030331e-06,
+      "loss": 0.637,
+      "step": 11861
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.765114188194275,
+      "learning_rate": 9.318280135796048e-06,
+      "loss": 0.6439,
+      "step": 11862
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.3721303939819336,
+      "learning_rate": 9.318147224554247e-06,
+      "loss": 0.6226,
+      "step": 11863
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 4.045658588409424,
+      "learning_rate": 9.318014301305296e-06,
+      "loss": 0.9311,
+      "step": 11864
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 6.897544860839844,
+      "learning_rate": 9.317881366049563e-06,
+      "loss": 0.8346,
+      "step": 11865
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.8708146810531616,
+      "learning_rate": 9.317748418787423e-06,
+      "loss": 0.8567,
+      "step": 11866
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.4691627025604248,
+      "learning_rate": 9.31761545951924e-06,
+      "loss": 0.3156,
+      "step": 11867
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.2177135944366455,
+      "learning_rate": 9.317482488245386e-06,
+      "loss": 0.7562,
+      "step": 11868
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 4.185089111328125,
+      "learning_rate": 9.31734950496623e-06,
+      "loss": 0.8228,
+      "step": 11869
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.6321990489959717,
+      "learning_rate": 9.317216509682144e-06,
+      "loss": 0.8974,
+      "step": 11870
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.6643645763397217,
+      "learning_rate": 9.317083502393494e-06,
+      "loss": 0.7897,
+      "step": 11871
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.803323745727539,
+      "learning_rate": 9.316950483100653e-06,
+      "loss": 0.6298,
+      "step": 11872
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.8963760137557983,
+      "learning_rate": 9.31681745180399e-06,
+      "loss": 0.6801,
+      "step": 11873
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.3091750144958496,
+      "learning_rate": 9.316684408503875e-06,
+      "loss": 0.64,
+      "step": 11874
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.9226685762405396,
+      "learning_rate": 9.316551353200678e-06,
+      "loss": 0.6459,
+      "step": 11875
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.9266464710235596,
+      "learning_rate": 9.316418285894767e-06,
+      "loss": 0.7097,
+      "step": 11876
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.6687099933624268,
+      "learning_rate": 9.316285206586514e-06,
+      "loss": 0.7474,
+      "step": 11877
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.3118200302124023,
+      "learning_rate": 9.316152115276291e-06,
+      "loss": 0.5214,
+      "step": 11878
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.9396238327026367,
+      "learning_rate": 9.316019011964465e-06,
+      "loss": 0.5622,
+      "step": 11879
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.849409341812134,
+      "learning_rate": 9.315885896651406e-06,
+      "loss": 0.7849,
+      "step": 11880
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.6655497550964355,
+      "learning_rate": 9.315752769337485e-06,
+      "loss": 0.7003,
+      "step": 11881
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.4813778400421143,
+      "learning_rate": 9.315619630023072e-06,
+      "loss": 0.6709,
+      "step": 11882
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.307135581970215,
+      "learning_rate": 9.315486478708538e-06,
+      "loss": 0.4832,
+      "step": 11883
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 5.805148601531982,
+      "learning_rate": 9.315353315394253e-06,
+      "loss": 0.7626,
+      "step": 11884
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.5301264524459839,
+      "learning_rate": 9.315220140080588e-06,
+      "loss": 0.6477,
+      "step": 11885
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.4515414237976074,
+      "learning_rate": 9.31508695276791e-06,
+      "loss": 0.6096,
+      "step": 11886
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.197783946990967,
+      "learning_rate": 9.314953753456594e-06,
+      "loss": 0.5416,
+      "step": 11887
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.502031087875366,
+      "learning_rate": 9.314820542147008e-06,
+      "loss": 0.6331,
+      "step": 11888
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.5603346824645996,
+      "learning_rate": 9.314687318839519e-06,
+      "loss": 0.9642,
+      "step": 11889
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 6.7842698097229,
+      "learning_rate": 9.314554083534504e-06,
+      "loss": 0.819,
+      "step": 11890
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 4.285874843597412,
+      "learning_rate": 9.31442083623233e-06,
+      "loss": 0.6257,
+      "step": 11891
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.482701063156128,
+      "learning_rate": 9.314287576933367e-06,
+      "loss": 0.5332,
+      "step": 11892
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.962354302406311,
+      "learning_rate": 9.314154305637987e-06,
+      "loss": 0.5781,
+      "step": 11893
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.9308419227600098,
+      "learning_rate": 9.31402102234656e-06,
+      "loss": 0.5392,
+      "step": 11894
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.6854898929595947,
+      "learning_rate": 9.313887727059457e-06,
+      "loss": 0.7499,
+      "step": 11895
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 4.157299518585205,
+      "learning_rate": 9.313754419777046e-06,
+      "loss": 0.6883,
+      "step": 11896
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.691312074661255,
+      "learning_rate": 9.313621100499703e-06,
+      "loss": 0.6743,
+      "step": 11897
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.524036407470703,
+      "learning_rate": 9.313487769227793e-06,
+      "loss": 0.8657,
+      "step": 11898
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.1713852882385254,
+      "learning_rate": 9.313354425961692e-06,
+      "loss": 0.5608,
+      "step": 11899
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.638895034790039,
+      "learning_rate": 9.313221070701767e-06,
+      "loss": 0.8001,
+      "step": 11900
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.1524276733398438,
+      "learning_rate": 9.31308770344839e-06,
+      "loss": 0.8129,
+      "step": 11901
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.24385142326355,
+      "learning_rate": 9.31295432420193e-06,
+      "loss": 0.7317,
+      "step": 11902
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.787583827972412,
+      "learning_rate": 9.312820932962762e-06,
+      "loss": 0.9248,
+      "step": 11903
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.7997983694076538,
+      "learning_rate": 9.312687529731255e-06,
+      "loss": 0.8517,
+      "step": 11904
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.7732837200164795,
+      "learning_rate": 9.312554114507777e-06,
+      "loss": 0.7514,
+      "step": 11905
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.9331928491592407,
+      "learning_rate": 9.312420687292703e-06,
+      "loss": 0.3315,
+      "step": 11906
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 4.023275852203369,
+      "learning_rate": 9.312287248086404e-06,
+      "loss": 0.8379,
+      "step": 11907
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.598823070526123,
+      "learning_rate": 9.312153796889247e-06,
+      "loss": 0.8346,
+      "step": 11908
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.330841302871704,
+      "learning_rate": 9.312020333701607e-06,
+      "loss": 0.7796,
+      "step": 11909
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.6558196544647217,
+      "learning_rate": 9.311886858523854e-06,
+      "loss": 0.7192,
+      "step": 11910
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.7979302406311035,
+      "learning_rate": 9.311753371356358e-06,
+      "loss": 0.784,
+      "step": 11911
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 4.914438247680664,
+      "learning_rate": 9.311619872199492e-06,
+      "loss": 0.7932,
+      "step": 11912
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.1464338302612305,
+      "learning_rate": 9.311486361053625e-06,
+      "loss": 0.8852,
+      "step": 11913
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.504174828529358,
+      "learning_rate": 9.311352837919131e-06,
+      "loss": 0.612,
+      "step": 11914
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.5560283660888672,
+      "learning_rate": 9.31121930279638e-06,
+      "loss": 0.6193,
+      "step": 11915
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.921565294265747,
+      "learning_rate": 9.311085755685743e-06,
+      "loss": 0.7599,
+      "step": 11916
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.143993377685547,
+      "learning_rate": 9.310952196587589e-06,
+      "loss": 0.815,
+      "step": 11917
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.813258171081543,
+      "learning_rate": 9.310818625502293e-06,
+      "loss": 0.8401,
+      "step": 11918
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.1420655250549316,
+      "learning_rate": 9.310685042430228e-06,
+      "loss": 1.034,
+      "step": 11919
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.9427874088287354,
+      "learning_rate": 9.31055144737176e-06,
+      "loss": 0.62,
+      "step": 11920
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.8097316026687622,
+      "learning_rate": 9.310417840327263e-06,
+      "loss": 0.6868,
+      "step": 11921
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.227397918701172,
+      "learning_rate": 9.31028422129711e-06,
+      "loss": 0.6497,
+      "step": 11922
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.100022315979004,
+      "learning_rate": 9.310150590281671e-06,
+      "loss": 0.7889,
+      "step": 11923
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.637787938117981,
+      "learning_rate": 9.310016947281319e-06,
+      "loss": 0.4668,
+      "step": 11924
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.532217502593994,
+      "learning_rate": 9.309883292296422e-06,
+      "loss": 0.8319,
+      "step": 11925
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.7392373085021973,
+      "learning_rate": 9.309749625327356e-06,
+      "loss": 0.5227,
+      "step": 11926
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.454636335372925,
+      "learning_rate": 9.30961594637449e-06,
+      "loss": 0.8841,
+      "step": 11927
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.9732491970062256,
+      "learning_rate": 9.309482255438196e-06,
+      "loss": 0.5281,
+      "step": 11928
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.347783088684082,
+      "learning_rate": 9.309348552518847e-06,
+      "loss": 1.0658,
+      "step": 11929
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.4120702743530273,
+      "learning_rate": 9.309214837616813e-06,
+      "loss": 0.4565,
+      "step": 11930
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.8554550409317017,
+      "learning_rate": 9.309081110732468e-06,
+      "loss": 0.5841,
+      "step": 11931
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.6503067016601562,
+      "learning_rate": 9.308947371866182e-06,
+      "loss": 0.7867,
+      "step": 11932
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.7745940685272217,
+      "learning_rate": 9.308813621018328e-06,
+      "loss": 0.9453,
+      "step": 11933
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.8288657665252686,
+      "learning_rate": 9.308679858189276e-06,
+      "loss": 0.6475,
+      "step": 11934
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.5385866165161133,
+      "learning_rate": 9.308546083379402e-06,
+      "loss": 0.7058,
+      "step": 11935
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.9923008680343628,
+      "learning_rate": 9.308412296589073e-06,
+      "loss": 0.798,
+      "step": 11936
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.168644905090332,
+      "learning_rate": 9.308278497818663e-06,
+      "loss": 0.5935,
+      "step": 11937
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.724824905395508,
+      "learning_rate": 9.308144687068546e-06,
+      "loss": 0.7052,
+      "step": 11938
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.8413002490997314,
+      "learning_rate": 9.308010864339093e-06,
+      "loss": 0.4526,
+      "step": 11939
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.991044521331787,
+      "learning_rate": 9.307877029630674e-06,
+      "loss": 0.7489,
+      "step": 11940
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.8803513050079346,
+      "learning_rate": 9.307743182943665e-06,
+      "loss": 0.657,
+      "step": 11941
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 4.213972091674805,
+      "learning_rate": 9.307609324278432e-06,
+      "loss": 0.7483,
+      "step": 11942
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.417346239089966,
+      "learning_rate": 9.307475453635354e-06,
+      "loss": 0.5392,
+      "step": 11943
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.7147706747055054,
+      "learning_rate": 9.3073415710148e-06,
+      "loss": 0.3539,
+      "step": 11944
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.8560320138931274,
+      "learning_rate": 9.307207676417142e-06,
+      "loss": 0.6113,
+      "step": 11945
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.6907222270965576,
+      "learning_rate": 9.307073769842756e-06,
+      "loss": 0.9862,
+      "step": 11946
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.859821319580078,
+      "learning_rate": 9.306939851292007e-06,
+      "loss": 0.7275,
+      "step": 11947
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.5354349613189697,
+      "learning_rate": 9.306805920765274e-06,
+      "loss": 0.6713,
+      "step": 11948
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.8553515672683716,
+      "learning_rate": 9.306671978262927e-06,
+      "loss": 0.8295,
+      "step": 11949
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.6726938486099243,
+      "learning_rate": 9.30653802378534e-06,
+      "loss": 0.5934,
+      "step": 11950
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 6.44894552230835,
+      "learning_rate": 9.30640405733288e-06,
+      "loss": 0.7948,
+      "step": 11951
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.326677083969116,
+      "learning_rate": 9.306270078905926e-06,
+      "loss": 0.6429,
+      "step": 11952
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.5599181652069092,
+      "learning_rate": 9.306136088504849e-06,
+      "loss": 0.7222,
+      "step": 11953
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 6.779192924499512,
+      "learning_rate": 9.306002086130018e-06,
+      "loss": 0.974,
+      "step": 11954
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.43927264213562,
+      "learning_rate": 9.30586807178181e-06,
+      "loss": 0.7069,
+      "step": 11955
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.0247178077697754,
+      "learning_rate": 9.305734045460596e-06,
+      "loss": 0.7843,
+      "step": 11956
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.1627578735351562,
+      "learning_rate": 9.305600007166749e-06,
+      "loss": 0.7559,
+      "step": 11957
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.184746742248535,
+      "learning_rate": 9.305465956900639e-06,
+      "loss": 0.4657,
+      "step": 11958
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.9712252616882324,
+      "learning_rate": 9.305331894662644e-06,
+      "loss": 0.6369,
+      "step": 11959
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.0295939445495605,
+      "learning_rate": 9.305197820453132e-06,
+      "loss": 0.8363,
+      "step": 11960
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.4338839054107666,
+      "learning_rate": 9.305063734272478e-06,
+      "loss": 0.6851,
+      "step": 11961
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.3347352743148804,
+      "learning_rate": 9.304929636121055e-06,
+      "loss": 0.8636,
+      "step": 11962
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.1143715381622314,
+      "learning_rate": 9.304795525999233e-06,
+      "loss": 0.558,
+      "step": 11963
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.2943484783172607,
+      "learning_rate": 9.304661403907389e-06,
+      "loss": 0.5295,
+      "step": 11964
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.433009147644043,
+      "learning_rate": 9.304527269845896e-06,
+      "loss": 0.727,
+      "step": 11965
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.798341751098633,
+      "learning_rate": 9.304393123815123e-06,
+      "loss": 0.7128,
+      "step": 11966
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.2093982696533203,
+      "learning_rate": 9.304258965815446e-06,
+      "loss": 0.5611,
+      "step": 11967
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.8437498807907104,
+      "learning_rate": 9.304124795847236e-06,
+      "loss": 0.666,
+      "step": 11968
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.7721259593963623,
+      "learning_rate": 9.303990613910868e-06,
+      "loss": 0.672,
+      "step": 11969
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.5425164699554443,
+      "learning_rate": 9.303856420006715e-06,
+      "loss": 0.6315,
+      "step": 11970
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.8217363357543945,
+      "learning_rate": 9.303722214135149e-06,
+      "loss": 0.7659,
+      "step": 11971
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.497346878051758,
+      "learning_rate": 9.303587996296544e-06,
+      "loss": 0.7248,
+      "step": 11972
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.9302175045013428,
+      "learning_rate": 9.303453766491272e-06,
+      "loss": 0.5038,
+      "step": 11973
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.8945019245147705,
+      "learning_rate": 9.303319524719709e-06,
+      "loss": 0.8139,
+      "step": 11974
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.2848165035247803,
+      "learning_rate": 9.303185270982225e-06,
+      "loss": 0.807,
+      "step": 11975
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.1481876373291016,
+      "learning_rate": 9.303051005279196e-06,
+      "loss": 0.8265,
+      "step": 11976
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.3170926570892334,
+      "learning_rate": 9.302916727610993e-06,
+      "loss": 0.3003,
+      "step": 11977
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.170884609222412,
+      "learning_rate": 9.30278243797799e-06,
+      "loss": 0.8863,
+      "step": 11978
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.7176928520202637,
+      "learning_rate": 9.302648136380562e-06,
+      "loss": 0.6265,
+      "step": 11979
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.0742809772491455,
+      "learning_rate": 9.30251382281908e-06,
+      "loss": 0.8546,
+      "step": 11980
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.1822617053985596,
+      "learning_rate": 9.302379497293921e-06,
+      "loss": 0.6879,
+      "step": 11981
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.06612229347229,
+      "learning_rate": 9.302245159805454e-06,
+      "loss": 0.6246,
+      "step": 11982
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.2846181392669678,
+      "learning_rate": 9.302110810354057e-06,
+      "loss": 0.4286,
+      "step": 11983
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.4026131629943848,
+      "learning_rate": 9.3019764489401e-06,
+      "loss": 0.759,
+      "step": 11984
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.7847230434417725,
+      "learning_rate": 9.30184207556396e-06,
+      "loss": 0.6763,
+      "step": 11985
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.713772177696228,
+      "learning_rate": 9.301707690226007e-06,
+      "loss": 0.7413,
+      "step": 11986
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.9729183912277222,
+      "learning_rate": 9.301573292926616e-06,
+      "loss": 0.6033,
+      "step": 11987
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.18306565284729,
+      "learning_rate": 9.301438883666161e-06,
+      "loss": 0.3797,
+      "step": 11988
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.334535837173462,
+      "learning_rate": 9.301304462445018e-06,
+      "loss": 0.8027,
+      "step": 11989
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.2260684967041016,
+      "learning_rate": 9.301170029263556e-06,
+      "loss": 0.5274,
+      "step": 11990
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.0677502155303955,
+      "learning_rate": 9.301035584122154e-06,
+      "loss": 0.6437,
+      "step": 11991
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.6544976234436035,
+      "learning_rate": 9.30090112702118e-06,
+      "loss": 0.7969,
+      "step": 11992
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.3170583248138428,
+      "learning_rate": 9.300766657961014e-06,
+      "loss": 0.5029,
+      "step": 11993
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.9249403476715088,
+      "learning_rate": 9.300632176942026e-06,
+      "loss": 0.8272,
+      "step": 11994
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.181474208831787,
+      "learning_rate": 9.30049768396459e-06,
+      "loss": 0.6295,
+      "step": 11995
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.6666793823242188,
+      "learning_rate": 9.300363179029082e-06,
+      "loss": 0.7067,
+      "step": 11996
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.504439353942871,
+      "learning_rate": 9.300228662135875e-06,
+      "loss": 0.8093,
+      "step": 11997
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.7674779891967773,
+      "learning_rate": 9.300094133285343e-06,
+      "loss": 0.4956,
+      "step": 11998
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.8213471174240112,
+      "learning_rate": 9.299959592477861e-06,
+      "loss": 0.69,
+      "step": 11999
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.3258543014526367,
+      "learning_rate": 9.2998250397138e-06,
+      "loss": 0.6269,
+      "step": 12000
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.5191736221313477,
+      "learning_rate": 9.299690474993537e-06,
+      "loss": 0.6838,
+      "step": 12001
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.46687912940979,
+      "learning_rate": 9.299555898317447e-06,
+      "loss": 0.6643,
+      "step": 12002
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.755819320678711,
+      "learning_rate": 9.299421309685901e-06,
+      "loss": 0.6002,
+      "step": 12003
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.5928969383239746,
+      "learning_rate": 9.299286709099276e-06,
+      "loss": 0.8052,
+      "step": 12004
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.49100399017334,
+      "learning_rate": 9.299152096557944e-06,
+      "loss": 0.8623,
+      "step": 12005
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.090205669403076,
+      "learning_rate": 9.299017472062282e-06,
+      "loss": 0.7651,
+      "step": 12006
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.4384846687316895,
+      "learning_rate": 9.298882835612662e-06,
+      "loss": 0.6149,
+      "step": 12007
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.0264246463775635,
+      "learning_rate": 9.29874818720946e-06,
+      "loss": 0.6741,
+      "step": 12008
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.875427484512329,
+      "learning_rate": 9.298613526853048e-06,
+      "loss": 0.4472,
+      "step": 12009
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.3409297466278076,
+      "learning_rate": 9.298478854543803e-06,
+      "loss": 0.6487,
+      "step": 12010
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.3313939571380615,
+      "learning_rate": 9.298344170282099e-06,
+      "loss": 0.7068,
+      "step": 12011
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 4.8358564376831055,
+      "learning_rate": 9.29820947406831e-06,
+      "loss": 0.7143,
+      "step": 12012
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.1737794876098633,
+      "learning_rate": 9.298074765902809e-06,
+      "loss": 0.6133,
+      "step": 12013
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 4.099835395812988,
+      "learning_rate": 9.297940045785974e-06,
+      "loss": 0.6056,
+      "step": 12014
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.7470550537109375,
+      "learning_rate": 9.297805313718176e-06,
+      "loss": 0.406,
+      "step": 12015
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.5049097537994385,
+      "learning_rate": 9.297670569699793e-06,
+      "loss": 0.6154,
+      "step": 12016
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.0464019775390625,
+      "learning_rate": 9.297535813731198e-06,
+      "loss": 0.4392,
+      "step": 12017
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.9091603755950928,
+      "learning_rate": 9.297401045812766e-06,
+      "loss": 0.4065,
+      "step": 12018
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.9257094860076904,
+      "learning_rate": 9.297266265944871e-06,
+      "loss": 0.538,
+      "step": 12019
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.385694742202759,
+      "learning_rate": 9.297131474127888e-06,
+      "loss": 0.559,
+      "step": 12020
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.0631296634674072,
+      "learning_rate": 9.296996670362193e-06,
+      "loss": 0.4634,
+      "step": 12021
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.605146646499634,
+      "learning_rate": 9.296861854648159e-06,
+      "loss": 0.4463,
+      "step": 12022
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.975609064102173,
+      "learning_rate": 9.29672702698616e-06,
+      "loss": 0.665,
+      "step": 12023
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 4.2875189781188965,
+      "learning_rate": 9.296592187376576e-06,
+      "loss": 0.7244,
+      "step": 12024
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.3667893409729004,
+      "learning_rate": 9.296457335819777e-06,
+      "loss": 0.5041,
+      "step": 12025
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.4082884788513184,
+      "learning_rate": 9.296322472316142e-06,
+      "loss": 0.4884,
+      "step": 12026
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 4.430501937866211,
+      "learning_rate": 9.296187596866042e-06,
+      "loss": 0.4304,
+      "step": 12027
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.7528188228607178,
+      "learning_rate": 9.296052709469853e-06,
+      "loss": 0.5391,
+      "step": 12028
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.4570257663726807,
+      "learning_rate": 9.29591781012795e-06,
+      "loss": 0.7283,
+      "step": 12029
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.4248180389404297,
+      "learning_rate": 9.295782898840712e-06,
+      "loss": 0.953,
+      "step": 12030
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.440190315246582,
+      "learning_rate": 9.295647975608508e-06,
+      "loss": 0.801,
+      "step": 12031
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.1256771087646484,
+      "learning_rate": 9.295513040431718e-06,
+      "loss": 0.8,
+      "step": 12032
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 4.129363536834717,
+      "learning_rate": 9.295378093310714e-06,
+      "loss": 0.8805,
+      "step": 12033
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.037241220474243,
+      "learning_rate": 9.295243134245874e-06,
+      "loss": 0.7787,
+      "step": 12034
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.756953477859497,
+      "learning_rate": 9.295108163237571e-06,
+      "loss": 0.5495,
+      "step": 12035
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.995936393737793,
+      "learning_rate": 9.294973180286182e-06,
+      "loss": 0.5504,
+      "step": 12036
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.5572938919067383,
+      "learning_rate": 9.294838185392082e-06,
+      "loss": 0.5611,
+      "step": 12037
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.2773494720458984,
+      "learning_rate": 9.294703178555642e-06,
+      "loss": 0.5662,
+      "step": 12038
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.2920138835906982,
+      "learning_rate": 9.294568159777246e-06,
+      "loss": 0.8288,
+      "step": 12039
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.8477094173431396,
+      "learning_rate": 9.294433129057263e-06,
+      "loss": 0.8731,
+      "step": 12040
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.301201820373535,
+      "learning_rate": 9.29429808639607e-06,
+      "loss": 0.8976,
+      "step": 12041
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.9837453365325928,
+      "learning_rate": 9.294163031794041e-06,
+      "loss": 0.6371,
+      "step": 12042
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 4.56164026260376,
+      "learning_rate": 9.294027965251555e-06,
+      "loss": 0.7027,
+      "step": 12043
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.2615904808044434,
+      "learning_rate": 9.293892886768986e-06,
+      "loss": 0.8868,
+      "step": 12044
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.173339366912842,
+      "learning_rate": 9.293757796346709e-06,
+      "loss": 0.7234,
+      "step": 12045
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.1474270820617676,
+      "learning_rate": 9.2936226939851e-06,
+      "loss": 0.7751,
+      "step": 12046
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.2654311656951904,
+      "learning_rate": 9.293487579684536e-06,
+      "loss": 1.0251,
+      "step": 12047
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 4.799258232116699,
+      "learning_rate": 9.29335245344539e-06,
+      "loss": 0.5677,
+      "step": 12048
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.101395606994629,
+      "learning_rate": 9.293217315268039e-06,
+      "loss": 0.8348,
+      "step": 12049
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.9704760313034058,
+      "learning_rate": 9.29308216515286e-06,
+      "loss": 0.5971,
+      "step": 12050
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.5296459197998047,
+      "learning_rate": 9.292947003100228e-06,
+      "loss": 0.707,
+      "step": 12051
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 5.962486743927002,
+      "learning_rate": 9.292811829110518e-06,
+      "loss": 0.5927,
+      "step": 12052
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.212265968322754,
+      "learning_rate": 9.292676643184104e-06,
+      "loss": 0.4369,
+      "step": 12053
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.595828056335449,
+      "learning_rate": 9.292541445321368e-06,
+      "loss": 0.6424,
+      "step": 12054
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.7502083778381348,
+      "learning_rate": 9.29240623552268e-06,
+      "loss": 0.6669,
+      "step": 12055
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.3194470405578613,
+      "learning_rate": 9.292271013788419e-06,
+      "loss": 0.6124,
+      "step": 12056
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.9007463455200195,
+      "learning_rate": 9.29213578011896e-06,
+      "loss": 0.6353,
+      "step": 12057
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.1592299938201904,
+      "learning_rate": 9.29200053451468e-06,
+      "loss": 0.5598,
+      "step": 12058
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.1868765354156494,
+      "learning_rate": 9.291865276975953e-06,
+      "loss": 0.7363,
+      "step": 12059
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.2174532413482666,
+      "learning_rate": 9.291730007503158e-06,
+      "loss": 0.6767,
+      "step": 12060
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 9.213817596435547,
+      "learning_rate": 9.29159472609667e-06,
+      "loss": 0.6407,
+      "step": 12061
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.1281182765960693,
+      "learning_rate": 9.291459432756861e-06,
+      "loss": 0.5313,
+      "step": 12062
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.0264389514923096,
+      "learning_rate": 9.291324127484114e-06,
+      "loss": 0.5815,
+      "step": 12063
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.0285873413085938,
+      "learning_rate": 9.291188810278803e-06,
+      "loss": 0.7428,
+      "step": 12064
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 15.493531227111816,
+      "learning_rate": 9.2910534811413e-06,
+      "loss": 0.7277,
+      "step": 12065
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.77799916267395,
+      "learning_rate": 9.290918140071989e-06,
+      "loss": 0.96,
+      "step": 12066
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.31184720993042,
+      "learning_rate": 9.29078278707124e-06,
+      "loss": 0.6695,
+      "step": 12067
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.383301258087158,
+      "learning_rate": 9.29064742213943e-06,
+      "loss": 0.6729,
+      "step": 12068
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.5310521125793457,
+      "learning_rate": 9.290512045276938e-06,
+      "loss": 0.9139,
+      "step": 12069
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.073470115661621,
+      "learning_rate": 9.29037665648414e-06,
+      "loss": 0.5092,
+      "step": 12070
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.0254342555999756,
+      "learning_rate": 9.290241255761412e-06,
+      "loss": 0.592,
+      "step": 12071
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.8664183616638184,
+      "learning_rate": 9.290105843109128e-06,
+      "loss": 0.7034,
+      "step": 12072
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.7092063426971436,
+      "learning_rate": 9.28997041852767e-06,
+      "loss": 0.7459,
+      "step": 12073
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.8400976657867432,
+      "learning_rate": 9.289834982017407e-06,
+      "loss": 0.672,
+      "step": 12074
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.920262336730957,
+      "learning_rate": 9.289699533578723e-06,
+      "loss": 0.4019,
+      "step": 12075
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.2875421047210693,
+      "learning_rate": 9.289564073211992e-06,
+      "loss": 0.6595,
+      "step": 12076
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 5.672563552856445,
+      "learning_rate": 9.28942860091759e-06,
+      "loss": 0.769,
+      "step": 12077
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.8804631233215332,
+      "learning_rate": 9.289293116695892e-06,
+      "loss": 0.5614,
+      "step": 12078
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.7588672637939453,
+      "learning_rate": 9.289157620547277e-06,
+      "loss": 0.4786,
+      "step": 12079
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 4.602984428405762,
+      "learning_rate": 9.289022112472122e-06,
+      "loss": 0.6856,
+      "step": 12080
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.8965415954589844,
+      "learning_rate": 9.288886592470804e-06,
+      "loss": 0.9347,
+      "step": 12081
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.207160711288452,
+      "learning_rate": 9.288751060543698e-06,
+      "loss": 0.8424,
+      "step": 12082
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.228245735168457,
+      "learning_rate": 9.288615516691183e-06,
+      "loss": 0.6835,
+      "step": 12083
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.9126663208007812,
+      "learning_rate": 9.288479960913634e-06,
+      "loss": 0.5826,
+      "step": 12084
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.7321637868881226,
+      "learning_rate": 9.288344393211428e-06,
+      "loss": 0.783,
+      "step": 12085
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 9.20428466796875,
+      "learning_rate": 9.288208813584945e-06,
+      "loss": 0.8598,
+      "step": 12086
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.0483031272888184,
+      "learning_rate": 9.288073222034557e-06,
+      "loss": 0.791,
+      "step": 12087
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.8254120349884033,
+      "learning_rate": 9.287937618560646e-06,
+      "loss": 0.7655,
+      "step": 12088
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.187262773513794,
+      "learning_rate": 9.287802003163583e-06,
+      "loss": 0.4449,
+      "step": 12089
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.144707202911377,
+      "learning_rate": 9.287666375843753e-06,
+      "loss": 0.7986,
+      "step": 12090
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.754892110824585,
+      "learning_rate": 9.287530736601527e-06,
+      "loss": 0.5585,
+      "step": 12091
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.9358842372894287,
+      "learning_rate": 9.287395085437286e-06,
+      "loss": 0.5619,
+      "step": 12092
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.884176015853882,
+      "learning_rate": 9.287259422351405e-06,
+      "loss": 0.4915,
+      "step": 12093
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.79397714138031,
+      "learning_rate": 9.28712374734426e-06,
+      "loss": 0.656,
+      "step": 12094
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.8585867881774902,
+      "learning_rate": 9.286988060416229e-06,
+      "loss": 0.5572,
+      "step": 12095
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 4.225172996520996,
+      "learning_rate": 9.286852361567694e-06,
+      "loss": 0.7017,
+      "step": 12096
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.615339756011963,
+      "learning_rate": 9.286716650799024e-06,
+      "loss": 0.7567,
+      "step": 12097
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.1614291667938232,
+      "learning_rate": 9.286580928110604e-06,
+      "loss": 0.705,
+      "step": 12098
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.788529872894287,
+      "learning_rate": 9.286445193502806e-06,
+      "loss": 0.6589,
+      "step": 12099
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.863767623901367,
+      "learning_rate": 9.286309446976012e-06,
+      "loss": 0.5906,
+      "step": 12100
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.4936094284057617,
+      "learning_rate": 9.286173688530595e-06,
+      "loss": 0.6705,
+      "step": 12101
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.3686683177948,
+      "learning_rate": 9.286037918166937e-06,
+      "loss": 0.6559,
+      "step": 12102
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 6.065394878387451,
+      "learning_rate": 9.285902135885412e-06,
+      "loss": 0.8435,
+      "step": 12103
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.8127775192260742,
+      "learning_rate": 9.285766341686398e-06,
+      "loss": 0.5743,
+      "step": 12104
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.218320608139038,
+      "learning_rate": 9.285630535570274e-06,
+      "loss": 0.6478,
+      "step": 12105
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 4.3270745277404785,
+      "learning_rate": 9.285494717537415e-06,
+      "loss": 0.5408,
+      "step": 12106
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.4112869501113892,
+      "learning_rate": 9.285358887588203e-06,
+      "loss": 0.7264,
+      "step": 12107
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.1753954887390137,
+      "learning_rate": 9.285223045723013e-06,
+      "loss": 0.8682,
+      "step": 12108
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.275164842605591,
+      "learning_rate": 9.285087191942221e-06,
+      "loss": 0.7802,
+      "step": 12109
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.0081841945648193,
+      "learning_rate": 9.28495132624621e-06,
+      "loss": 0.4958,
+      "step": 12110
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.054046630859375,
+      "learning_rate": 9.284815448635352e-06,
+      "loss": 0.7888,
+      "step": 12111
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.8796558380126953,
+      "learning_rate": 9.284679559110028e-06,
+      "loss": 0.6849,
+      "step": 12112
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.4354958534240723,
+      "learning_rate": 9.284543657670616e-06,
+      "loss": 0.5666,
+      "step": 12113
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 5.860469341278076,
+      "learning_rate": 9.284407744317491e-06,
+      "loss": 0.6119,
+      "step": 12114
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 4.146963596343994,
+      "learning_rate": 9.284271819051037e-06,
+      "loss": 0.4698,
+      "step": 12115
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.7557471990585327,
+      "learning_rate": 9.284135881871625e-06,
+      "loss": 0.4952,
+      "step": 12116
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.2331247329711914,
+      "learning_rate": 9.283999932779636e-06,
+      "loss": 0.7511,
+      "step": 12117
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.7247133255004883,
+      "learning_rate": 9.283863971775449e-06,
+      "loss": 0.8526,
+      "step": 12118
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.2609047889709473,
+      "learning_rate": 9.28372799885944e-06,
+      "loss": 0.6942,
+      "step": 12119
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.7295596599578857,
+      "learning_rate": 9.283592014031989e-06,
+      "loss": 0.5354,
+      "step": 12120
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.360602378845215,
+      "learning_rate": 9.283456017293473e-06,
+      "loss": 0.3706,
+      "step": 12121
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 4.564781188964844,
+      "learning_rate": 9.28332000864427e-06,
+      "loss": 0.6661,
+      "step": 12122
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.2163937091827393,
+      "learning_rate": 9.283183988084758e-06,
+      "loss": 0.7556,
+      "step": 12123
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.767616033554077,
+      "learning_rate": 9.283047955615318e-06,
+      "loss": 0.566,
+      "step": 12124
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.7796480655670166,
+      "learning_rate": 9.282911911236326e-06,
+      "loss": 0.727,
+      "step": 12125
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.116560459136963,
+      "learning_rate": 9.282775854948159e-06,
+      "loss": 0.6087,
+      "step": 12126
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 4.550478458404541,
+      "learning_rate": 9.282639786751196e-06,
+      "loss": 0.4589,
+      "step": 12127
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.2065091133117676,
+      "learning_rate": 9.282503706645818e-06,
+      "loss": 0.5435,
+      "step": 12128
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.649355411529541,
+      "learning_rate": 9.2823676146324e-06,
+      "loss": 0.6446,
+      "step": 12129
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.579777956008911,
+      "learning_rate": 9.282231510711323e-06,
+      "loss": 0.753,
+      "step": 12130
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 4.00351619720459,
+      "learning_rate": 9.282095394882964e-06,
+      "loss": 0.7525,
+      "step": 12131
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.0384068489074707,
+      "learning_rate": 9.281959267147701e-06,
+      "loss": 0.8645,
+      "step": 12132
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.989235281944275,
+      "learning_rate": 9.281823127505915e-06,
+      "loss": 0.4444,
+      "step": 12133
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.2943830490112305,
+      "learning_rate": 9.281686975957982e-06,
+      "loss": 0.8511,
+      "step": 12134
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.7091069221496582,
+      "learning_rate": 9.28155081250428e-06,
+      "loss": 0.5786,
+      "step": 12135
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.849430799484253,
+      "learning_rate": 9.28141463714519e-06,
+      "loss": 0.7569,
+      "step": 12136
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.212040662765503,
+      "learning_rate": 9.281278449881091e-06,
+      "loss": 0.8457,
+      "step": 12137
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.229520797729492,
+      "learning_rate": 9.28114225071236e-06,
+      "loss": 0.6977,
+      "step": 12138
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 4.182148456573486,
+      "learning_rate": 9.281006039639375e-06,
+      "loss": 0.5483,
+      "step": 12139
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.989235758781433,
+      "learning_rate": 9.280869816662516e-06,
+      "loss": 0.6696,
+      "step": 12140
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.0718941688537598,
+      "learning_rate": 9.280733581782163e-06,
+      "loss": 0.69,
+      "step": 12141
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 5.376554489135742,
+      "learning_rate": 9.280597334998694e-06,
+      "loss": 0.4196,
+      "step": 12142
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.7000176906585693,
+      "learning_rate": 9.280461076312485e-06,
+      "loss": 0.7929,
+      "step": 12143
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.8971103429794312,
+      "learning_rate": 9.28032480572392e-06,
+      "loss": 0.6149,
+      "step": 12144
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.07548451423645,
+      "learning_rate": 9.280188523233373e-06,
+      "loss": 0.5169,
+      "step": 12145
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.219827651977539,
+      "learning_rate": 9.280052228841226e-06,
+      "loss": 0.5806,
+      "step": 12146
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.2987754344940186,
+      "learning_rate": 9.279915922547857e-06,
+      "loss": 0.6794,
+      "step": 12147
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.8834047317504883,
+      "learning_rate": 9.279779604353645e-06,
+      "loss": 0.8143,
+      "step": 12148
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.4241559505462646,
+      "learning_rate": 9.27964327425897e-06,
+      "loss": 0.7414,
+      "step": 12149
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.881864309310913,
+      "learning_rate": 9.279506932264209e-06,
+      "loss": 0.6152,
+      "step": 12150
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.1102840900421143,
+      "learning_rate": 9.279370578369742e-06,
+      "loss": 0.6552,
+      "step": 12151
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.809814691543579,
+      "learning_rate": 9.27923421257595e-06,
+      "loss": 0.8706,
+      "step": 12152
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.9434757232666016,
+      "learning_rate": 9.279097834883209e-06,
+      "loss": 0.5499,
+      "step": 12153
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.4348573684692383,
+      "learning_rate": 9.2789614452919e-06,
+      "loss": 0.685,
+      "step": 12154
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.714503049850464,
+      "learning_rate": 9.278825043802405e-06,
+      "loss": 0.6468,
+      "step": 12155
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.7991563081741333,
+      "learning_rate": 9.278688630415098e-06,
+      "loss": 0.9239,
+      "step": 12156
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.8034579753875732,
+      "learning_rate": 9.278552205130361e-06,
+      "loss": 0.4394,
+      "step": 12157
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.2659389972686768,
+      "learning_rate": 9.278415767948573e-06,
+      "loss": 0.7895,
+      "step": 12158
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.272387742996216,
+      "learning_rate": 9.278279318870113e-06,
+      "loss": 0.7611,
+      "step": 12159
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.6463102102279663,
+      "learning_rate": 9.278142857895362e-06,
+      "loss": 0.7659,
+      "step": 12160
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 6.689394474029541,
+      "learning_rate": 9.278006385024697e-06,
+      "loss": 0.5479,
+      "step": 12161
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.6225590705871582,
+      "learning_rate": 9.277869900258502e-06,
+      "loss": 0.7819,
+      "step": 12162
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.8865857124328613,
+      "learning_rate": 9.277733403597152e-06,
+      "loss": 0.7215,
+      "step": 12163
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.8363921642303467,
+      "learning_rate": 9.277596895041026e-06,
+      "loss": 0.5533,
+      "step": 12164
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 5.3143086433410645,
+      "learning_rate": 9.277460374590505e-06,
+      "loss": 0.654,
+      "step": 12165
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.093940019607544,
+      "learning_rate": 9.27732384224597e-06,
+      "loss": 0.6819,
+      "step": 12166
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.6713335514068604,
+      "learning_rate": 9.277187298007801e-06,
+      "loss": 0.5703,
+      "step": 12167
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.2567343711853027,
+      "learning_rate": 9.277050741876376e-06,
+      "loss": 0.7738,
+      "step": 12168
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.414114236831665,
+      "learning_rate": 9.276914173852077e-06,
+      "loss": 0.5332,
+      "step": 12169
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.161900520324707,
+      "learning_rate": 9.276777593935277e-06,
+      "loss": 0.8837,
+      "step": 12170
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.0231895446777344,
+      "learning_rate": 9.276641002126365e-06,
+      "loss": 0.792,
+      "step": 12171
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.816367506980896,
+      "learning_rate": 9.276504398425714e-06,
+      "loss": 0.4615,
+      "step": 12172
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 4.420854091644287,
+      "learning_rate": 9.276367782833707e-06,
+      "loss": 0.8104,
+      "step": 12173
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.8424859046936035,
+      "learning_rate": 9.276231155350724e-06,
+      "loss": 0.9154,
+      "step": 12174
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.779146671295166,
+      "learning_rate": 9.276094515977143e-06,
+      "loss": 0.8226,
+      "step": 12175
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.7663798332214355,
+      "learning_rate": 9.275957864713345e-06,
+      "loss": 0.4934,
+      "step": 12176
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.569833517074585,
+      "learning_rate": 9.27582120155971e-06,
+      "loss": 0.5555,
+      "step": 12177
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.6557462215423584,
+      "learning_rate": 9.275684526516618e-06,
+      "loss": 0.7748,
+      "step": 12178
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.068943738937378,
+      "learning_rate": 9.27554783958445e-06,
+      "loss": 0.839,
+      "step": 12179
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.0547609329223633,
+      "learning_rate": 9.275411140763584e-06,
+      "loss": 0.8194,
+      "step": 12180
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.708627223968506,
+      "learning_rate": 9.275274430054402e-06,
+      "loss": 0.7393,
+      "step": 12181
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.626004934310913,
+      "learning_rate": 9.275137707457283e-06,
+      "loss": 0.9174,
+      "step": 12182
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.023421049118042,
+      "learning_rate": 9.275000972972608e-06,
+      "loss": 0.7405,
+      "step": 12183
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.4048807621002197,
+      "learning_rate": 9.274864226600756e-06,
+      "loss": 0.6215,
+      "step": 12184
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.8397011756896973,
+      "learning_rate": 9.274727468342108e-06,
+      "loss": 0.8526,
+      "step": 12185
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.7404427528381348,
+      "learning_rate": 9.274590698197044e-06,
+      "loss": 0.8887,
+      "step": 12186
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.00002384185791,
+      "learning_rate": 9.274453916165946e-06,
+      "loss": 0.9603,
+      "step": 12187
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.1428449153900146,
+      "learning_rate": 9.274317122249192e-06,
+      "loss": 0.7918,
+      "step": 12188
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.3802433013916016,
+      "learning_rate": 9.274180316447162e-06,
+      "loss": 0.4968,
+      "step": 12189
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.212554693222046,
+      "learning_rate": 9.27404349876024e-06,
+      "loss": 0.5713,
+      "step": 12190
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.8817356824874878,
+      "learning_rate": 9.273906669188804e-06,
+      "loss": 0.6306,
+      "step": 12191
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.497724771499634,
+      "learning_rate": 9.273769827733231e-06,
+      "loss": 0.7916,
+      "step": 12192
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.773533582687378,
+      "learning_rate": 9.273632974393908e-06,
+      "loss": 0.5268,
+      "step": 12193
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.66001296043396,
+      "learning_rate": 9.273496109171215e-06,
+      "loss": 0.7461,
+      "step": 12194
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.696167469024658,
+      "learning_rate": 9.273359232065526e-06,
+      "loss": 0.7219,
+      "step": 12195
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.140195846557617,
+      "learning_rate": 9.273222343077229e-06,
+      "loss": 0.5975,
+      "step": 12196
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.9323149919509888,
+      "learning_rate": 9.273085442206698e-06,
+      "loss": 0.5546,
+      "step": 12197
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.6448168754577637,
+      "learning_rate": 9.27294852945432e-06,
+      "loss": 0.4447,
+      "step": 12198
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.331914186477661,
+      "learning_rate": 9.272811604820471e-06,
+      "loss": 0.5577,
+      "step": 12199
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.04386830329895,
+      "learning_rate": 9.272674668305535e-06,
+      "loss": 0.8186,
+      "step": 12200
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.8123021125793457,
+      "learning_rate": 9.272537719909891e-06,
+      "loss": 0.4837,
+      "step": 12201
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.3674213886260986,
+      "learning_rate": 9.272400759633919e-06,
+      "loss": 0.7251,
+      "step": 12202
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.561168670654297,
+      "learning_rate": 9.272263787478001e-06,
+      "loss": 0.5079,
+      "step": 12203
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.5547306537628174,
+      "learning_rate": 9.272126803442521e-06,
+      "loss": 0.7665,
+      "step": 12204
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.0491621494293213,
+      "learning_rate": 9.271989807527853e-06,
+      "loss": 0.7603,
+      "step": 12205
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.5485191345214844,
+      "learning_rate": 9.271852799734384e-06,
+      "loss": 0.428,
+      "step": 12206
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.5885114669799805,
+      "learning_rate": 9.271715780062491e-06,
+      "loss": 0.4859,
+      "step": 12207
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.1127512454986572,
+      "learning_rate": 9.271578748512557e-06,
+      "loss": 0.7455,
+      "step": 12208
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.4628303050994873,
+      "learning_rate": 9.271441705084963e-06,
+      "loss": 0.9094,
+      "step": 12209
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.0854973793029785,
+      "learning_rate": 9.27130464978009e-06,
+      "loss": 0.6195,
+      "step": 12210
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.0034148693084717,
+      "learning_rate": 9.271167582598318e-06,
+      "loss": 0.5191,
+      "step": 12211
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.883847713470459,
+      "learning_rate": 9.27103050354003e-06,
+      "loss": 0.6187,
+      "step": 12212
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.7142407894134521,
+      "learning_rate": 9.270893412605606e-06,
+      "loss": 0.7839,
+      "step": 12213
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.827287435531616,
+      "learning_rate": 9.270756309795427e-06,
+      "loss": 0.5432,
+      "step": 12214
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.2303466796875,
+      "learning_rate": 9.270619195109875e-06,
+      "loss": 0.6363,
+      "step": 12215
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.2328624725341797,
+      "learning_rate": 9.270482068549329e-06,
+      "loss": 0.4399,
+      "step": 12216
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.311493158340454,
+      "learning_rate": 9.270344930114174e-06,
+      "loss": 0.8008,
+      "step": 12217
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.8061370849609375,
+      "learning_rate": 9.27020777980479e-06,
+      "loss": 0.592,
+      "step": 12218
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.303274154663086,
+      "learning_rate": 9.270070617621555e-06,
+      "loss": 0.6639,
+      "step": 12219
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.2921979427337646,
+      "learning_rate": 9.269933443564855e-06,
+      "loss": 0.5549,
+      "step": 12220
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.321951150894165,
+      "learning_rate": 9.26979625763507e-06,
+      "loss": 0.4731,
+      "step": 12221
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.858480453491211,
+      "learning_rate": 9.269659059832581e-06,
+      "loss": 0.9855,
+      "step": 12222
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.264187812805176,
+      "learning_rate": 9.26952185015777e-06,
+      "loss": 0.6933,
+      "step": 12223
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.9304301738739014,
+      "learning_rate": 9.269384628611019e-06,
+      "loss": 0.7751,
+      "step": 12224
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.8690788745880127,
+      "learning_rate": 9.269247395192706e-06,
+      "loss": 0.6299,
+      "step": 12225
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.538602113723755,
+      "learning_rate": 9.269110149903215e-06,
+      "loss": 0.798,
+      "step": 12226
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.670628309249878,
+      "learning_rate": 9.268972892742932e-06,
+      "loss": 0.3382,
+      "step": 12227
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.733410120010376,
+      "learning_rate": 9.26883562371223e-06,
+      "loss": 0.5297,
+      "step": 12228
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.9162665605545044,
+      "learning_rate": 9.268698342811497e-06,
+      "loss": 0.517,
+      "step": 12229
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.885911464691162,
+      "learning_rate": 9.268561050041114e-06,
+      "loss": 0.5605,
+      "step": 12230
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 4.163054943084717,
+      "learning_rate": 9.26842374540146e-06,
+      "loss": 0.9515,
+      "step": 12231
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.974374771118164,
+      "learning_rate": 9.268286428892921e-06,
+      "loss": 0.5262,
+      "step": 12232
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.072082757949829,
+      "learning_rate": 9.268149100515874e-06,
+      "loss": 0.5587,
+      "step": 12233
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.3108322620391846,
+      "learning_rate": 9.268011760270705e-06,
+      "loss": 0.6532,
+      "step": 12234
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.8345584869384766,
+      "learning_rate": 9.267874408157793e-06,
+      "loss": 1.0501,
+      "step": 12235
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.1573383808135986,
+      "learning_rate": 9.26773704417752e-06,
+      "loss": 0.8179,
+      "step": 12236
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 13.351454734802246,
+      "learning_rate": 9.267599668330271e-06,
+      "loss": 0.806,
+      "step": 12237
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.2950937747955322,
+      "learning_rate": 9.267462280616427e-06,
+      "loss": 0.4733,
+      "step": 12238
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.791813611984253,
+      "learning_rate": 9.267324881036366e-06,
+      "loss": 0.6374,
+      "step": 12239
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.713646411895752,
+      "learning_rate": 9.267187469590474e-06,
+      "loss": 0.6694,
+      "step": 12240
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.2264857292175293,
+      "learning_rate": 9.267050046279134e-06,
+      "loss": 0.6971,
+      "step": 12241
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 4.381303310394287,
+      "learning_rate": 9.266912611102725e-06,
+      "loss": 0.7941,
+      "step": 12242
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.6096723079681396,
+      "learning_rate": 9.26677516406163e-06,
+      "loss": 0.5232,
+      "step": 12243
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.544187545776367,
+      "learning_rate": 9.266637705156231e-06,
+      "loss": 0.8049,
+      "step": 12244
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.375502109527588,
+      "learning_rate": 9.266500234386912e-06,
+      "loss": 0.6685,
+      "step": 12245
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.550549030303955,
+      "learning_rate": 9.266362751754054e-06,
+      "loss": 0.5922,
+      "step": 12246
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.9592071771621704,
+      "learning_rate": 9.266225257258041e-06,
+      "loss": 0.7059,
+      "step": 12247
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.276900053024292,
+      "learning_rate": 9.266087750899252e-06,
+      "loss": 0.4452,
+      "step": 12248
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.671545386314392,
+      "learning_rate": 9.26595023267807e-06,
+      "loss": 0.5078,
+      "step": 12249
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.81536602973938,
+      "learning_rate": 9.265812702594881e-06,
+      "loss": 0.7651,
+      "step": 12250
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.292879104614258,
+      "learning_rate": 9.265675160650063e-06,
+      "loss": 0.4092,
+      "step": 12251
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.588656425476074,
+      "learning_rate": 9.265537606844001e-06,
+      "loss": 0.9028,
+      "step": 12252
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.4347848892211914,
+      "learning_rate": 9.265400041177078e-06,
+      "loss": 0.5516,
+      "step": 12253
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.789199113845825,
+      "learning_rate": 9.265262463649674e-06,
+      "loss": 0.7233,
+      "step": 12254
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.7066004276275635,
+      "learning_rate": 9.265124874262173e-06,
+      "loss": 0.736,
+      "step": 12255
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.295508623123169,
+      "learning_rate": 9.264987273014956e-06,
+      "loss": 0.6066,
+      "step": 12256
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.002897262573242,
+      "learning_rate": 9.264849659908408e-06,
+      "loss": 0.7585,
+      "step": 12257
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.444289207458496,
+      "learning_rate": 9.264712034942912e-06,
+      "loss": 0.7054,
+      "step": 12258
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.4845988750457764,
+      "learning_rate": 9.264574398118848e-06,
+      "loss": 0.71,
+      "step": 12259
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.9915757179260254,
+      "learning_rate": 9.264436749436602e-06,
+      "loss": 0.5928,
+      "step": 12260
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.517395257949829,
+      "learning_rate": 9.264299088896553e-06,
+      "loss": 0.7223,
+      "step": 12261
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.4973106384277344,
+      "learning_rate": 9.264161416499087e-06,
+      "loss": 0.5821,
+      "step": 12262
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.0638606548309326,
+      "learning_rate": 9.264023732244585e-06,
+      "loss": 0.7422,
+      "step": 12263
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 4.054327964782715,
+      "learning_rate": 9.263886036133429e-06,
+      "loss": 0.6932,
+      "step": 12264
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.991775393486023,
+      "learning_rate": 9.263748328166006e-06,
+      "loss": 0.4933,
+      "step": 12265
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.3345460891723633,
+      "learning_rate": 9.263610608342694e-06,
+      "loss": 0.8,
+      "step": 12266
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.6596043109893799,
+      "learning_rate": 9.26347287666388e-06,
+      "loss": 0.7187,
+      "step": 12267
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.9248459339141846,
+      "learning_rate": 9.263335133129943e-06,
+      "loss": 0.6623,
+      "step": 12268
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.36875319480896,
+      "learning_rate": 9.263197377741268e-06,
+      "loss": 0.5587,
+      "step": 12269
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.3263914585113525,
+      "learning_rate": 9.26305961049824e-06,
+      "loss": 0.8684,
+      "step": 12270
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.2325544357299805,
+      "learning_rate": 9.262921831401239e-06,
+      "loss": 0.6294,
+      "step": 12271
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.4579250812530518,
+      "learning_rate": 9.26278404045065e-06,
+      "loss": 0.5962,
+      "step": 12272
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.6611994504928589,
+      "learning_rate": 9.262646237646854e-06,
+      "loss": 0.6198,
+      "step": 12273
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.3274478912353516,
+      "learning_rate": 9.262508422990238e-06,
+      "loss": 0.5964,
+      "step": 12274
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 4.7791266441345215,
+      "learning_rate": 9.26237059648118e-06,
+      "loss": 0.7912,
+      "step": 12275
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.578946590423584,
+      "learning_rate": 9.262232758120068e-06,
+      "loss": 0.734,
+      "step": 12276
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.2220406532287598,
+      "learning_rate": 9.262094907907282e-06,
+      "loss": 0.7472,
+      "step": 12277
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 4.2390456199646,
+      "learning_rate": 9.261957045843208e-06,
+      "loss": 0.7116,
+      "step": 12278
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.6763391494750977,
+      "learning_rate": 9.261819171928227e-06,
+      "loss": 0.5525,
+      "step": 12279
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.4907093048095703,
+      "learning_rate": 9.261681286162725e-06,
+      "loss": 0.7611,
+      "step": 12280
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.2218198776245117,
+      "learning_rate": 9.261543388547083e-06,
+      "loss": 0.5028,
+      "step": 12281
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 5.139666557312012,
+      "learning_rate": 9.261405479081684e-06,
+      "loss": 0.9327,
+      "step": 12282
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.280456066131592,
+      "learning_rate": 9.261267557766913e-06,
+      "loss": 0.6616,
+      "step": 12283
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.0420963764190674,
+      "learning_rate": 9.261129624603155e-06,
+      "loss": 0.6642,
+      "step": 12284
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.0399506092071533,
+      "learning_rate": 9.260991679590792e-06,
+      "loss": 0.5924,
+      "step": 12285
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.81125545501709,
+      "learning_rate": 9.260853722730206e-06,
+      "loss": 0.7181,
+      "step": 12286
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.0773065090179443,
+      "learning_rate": 9.260715754021781e-06,
+      "loss": 0.4912,
+      "step": 12287
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.4172422885894775,
+      "learning_rate": 9.260577773465902e-06,
+      "loss": 0.649,
+      "step": 12288
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.8971006870269775,
+      "learning_rate": 9.260439781062954e-06,
+      "loss": 0.7309,
+      "step": 12289
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.7464499473571777,
+      "learning_rate": 9.260301776813317e-06,
+      "loss": 0.625,
+      "step": 12290
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 5.543396949768066,
+      "learning_rate": 9.260163760717378e-06,
+      "loss": 0.6428,
+      "step": 12291
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.8027719259262085,
+      "learning_rate": 9.260025732775519e-06,
+      "loss": 0.6495,
+      "step": 12292
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.309952974319458,
+      "learning_rate": 9.259887692988124e-06,
+      "loss": 0.756,
+      "step": 12293
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.120455741882324,
+      "learning_rate": 9.259749641355578e-06,
+      "loss": 0.5927,
+      "step": 12294
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.9063079357147217,
+      "learning_rate": 9.259611577878264e-06,
+      "loss": 0.6962,
+      "step": 12295
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.429896354675293,
+      "learning_rate": 9.259473502556565e-06,
+      "loss": 0.6528,
+      "step": 12296
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.9915623664855957,
+      "learning_rate": 9.259335415390865e-06,
+      "loss": 0.8609,
+      "step": 12297
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.9591680765151978,
+      "learning_rate": 9.25919731638155e-06,
+      "loss": 0.8123,
+      "step": 12298
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.933361530303955,
+      "learning_rate": 9.259059205529004e-06,
+      "loss": 0.4899,
+      "step": 12299
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.8586313724517822,
+      "learning_rate": 9.258921082833608e-06,
+      "loss": 0.8213,
+      "step": 12300
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.189824104309082,
+      "learning_rate": 9.258782948295749e-06,
+      "loss": 0.7658,
+      "step": 12301
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.2822535037994385,
+      "learning_rate": 9.258644801915809e-06,
+      "loss": 0.6731,
+      "step": 12302
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.9249932765960693,
+      "learning_rate": 9.258506643694173e-06,
+      "loss": 0.579,
+      "step": 12303
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 5.41797399520874,
+      "learning_rate": 9.258368473631226e-06,
+      "loss": 0.463,
+      "step": 12304
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.8232877254486084,
+      "learning_rate": 9.258230291727352e-06,
+      "loss": 0.8485,
+      "step": 12305
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.6146433353424072,
+      "learning_rate": 9.258092097982933e-06,
+      "loss": 0.7724,
+      "step": 12306
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.47886061668396,
+      "learning_rate": 9.257953892398356e-06,
+      "loss": 0.8237,
+      "step": 12307
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.2275075912475586,
+      "learning_rate": 9.257815674974005e-06,
+      "loss": 0.5178,
+      "step": 12308
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.3822412490844727,
+      "learning_rate": 9.257677445710263e-06,
+      "loss": 0.8164,
+      "step": 12309
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.546931028366089,
+      "learning_rate": 9.257539204607516e-06,
+      "loss": 0.5712,
+      "step": 12310
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.040219306945801,
+      "learning_rate": 9.257400951666147e-06,
+      "loss": 0.6851,
+      "step": 12311
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.049591541290283,
+      "learning_rate": 9.257262686886538e-06,
+      "loss": 0.6817,
+      "step": 12312
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.054051160812378,
+      "learning_rate": 9.257124410269079e-06,
+      "loss": 0.5191,
+      "step": 12313
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.0143656730651855,
+      "learning_rate": 9.25698612181415e-06,
+      "loss": 0.672,
+      "step": 12314
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.2785511016845703,
+      "learning_rate": 9.25684782152214e-06,
+      "loss": 0.7557,
+      "step": 12315
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 5.426827430725098,
+      "learning_rate": 9.25670950939343e-06,
+      "loss": 0.5732,
+      "step": 12316
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.607241153717041,
+      "learning_rate": 9.256571185428405e-06,
+      "loss": 0.6853,
+      "step": 12317
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.814347743988037,
+      "learning_rate": 9.25643284962745e-06,
+      "loss": 0.7113,
+      "step": 12318
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.2715537548065186,
+      "learning_rate": 9.25629450199095e-06,
+      "loss": 0.6947,
+      "step": 12319
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.2358317375183105,
+      "learning_rate": 9.256156142519288e-06,
+      "loss": 0.5543,
+      "step": 12320
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.9548975229263306,
+      "learning_rate": 9.25601777121285e-06,
+      "loss": 0.6772,
+      "step": 12321
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 13.53371810913086,
+      "learning_rate": 9.255879388072024e-06,
+      "loss": 0.6772,
+      "step": 12322
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.6783063411712646,
+      "learning_rate": 9.25574099309719e-06,
+      "loss": 0.7396,
+      "step": 12323
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.8216254711151123,
+      "learning_rate": 9.255602586288734e-06,
+      "loss": 0.8559,
+      "step": 12324
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.1521501541137695,
+      "learning_rate": 9.25546416764704e-06,
+      "loss": 0.7518,
+      "step": 12325
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 5.341104030609131,
+      "learning_rate": 9.255325737172497e-06,
+      "loss": 0.6214,
+      "step": 12326
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.9201459884643555,
+      "learning_rate": 9.255187294865488e-06,
+      "loss": 0.6841,
+      "step": 12327
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.4647046327590942,
+      "learning_rate": 9.255048840726396e-06,
+      "loss": 0.5654,
+      "step": 12328
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 4.888077259063721,
+      "learning_rate": 9.254910374755605e-06,
+      "loss": 0.6365,
+      "step": 12329
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.9004082679748535,
+      "learning_rate": 9.254771896953505e-06,
+      "loss": 0.8872,
+      "step": 12330
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.208829641342163,
+      "learning_rate": 9.254633407320477e-06,
+      "loss": 0.572,
+      "step": 12331
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.1557459831237793,
+      "learning_rate": 9.254494905856907e-06,
+      "loss": 0.9735,
+      "step": 12332
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.907535195350647,
+      "learning_rate": 9.25435639256318e-06,
+      "loss": 0.5825,
+      "step": 12333
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.8691487312316895,
+      "learning_rate": 9.254217867439684e-06,
+      "loss": 0.7298,
+      "step": 12334
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.7736005783081055,
+      "learning_rate": 9.254079330486799e-06,
+      "loss": 0.6894,
+      "step": 12335
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.0160892009735107,
+      "learning_rate": 9.253940781704914e-06,
+      "loss": 0.44,
+      "step": 12336
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.039957046508789,
+      "learning_rate": 9.253802221094413e-06,
+      "loss": 0.5571,
+      "step": 12337
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.867669105529785,
+      "learning_rate": 9.253663648655683e-06,
+      "loss": 0.7446,
+      "step": 12338
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.7538492679595947,
+      "learning_rate": 9.253525064389106e-06,
+      "loss": 0.7748,
+      "step": 12339
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.164307117462158,
+      "learning_rate": 9.25338646829507e-06,
+      "loss": 0.706,
+      "step": 12340
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.7782866954803467,
+      "learning_rate": 9.25324786037396e-06,
+      "loss": 0.6512,
+      "step": 12341
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.9809660911560059,
+      "learning_rate": 9.25310924062616e-06,
+      "loss": 0.514,
+      "step": 12342
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.0764846801757812,
+      "learning_rate": 9.252970609052057e-06,
+      "loss": 0.5113,
+      "step": 12343
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.855653762817383,
+      "learning_rate": 9.252831965652036e-06,
+      "loss": 0.7457,
+      "step": 12344
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.206857204437256,
+      "learning_rate": 9.252693310426482e-06,
+      "loss": 0.8274,
+      "step": 12345
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.575619697570801,
+      "learning_rate": 9.252554643375782e-06,
+      "loss": 0.4344,
+      "step": 12346
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.3060545921325684,
+      "learning_rate": 9.25241596450032e-06,
+      "loss": 0.5692,
+      "step": 12347
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.9556819200515747,
+      "learning_rate": 9.252277273800483e-06,
+      "loss": 0.7276,
+      "step": 12348
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.7145277261734009,
+      "learning_rate": 9.252138571276655e-06,
+      "loss": 0.7283,
+      "step": 12349
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.6865360736846924,
+      "learning_rate": 9.25199985692922e-06,
+      "loss": 0.5747,
+      "step": 12350
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.3498175144195557,
+      "learning_rate": 9.25186113075857e-06,
+      "loss": 0.6525,
+      "step": 12351
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.9981071949005127,
+      "learning_rate": 9.251722392765085e-06,
+      "loss": 0.6903,
+      "step": 12352
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.3813388347625732,
+      "learning_rate": 9.251583642949156e-06,
+      "loss": 0.7586,
+      "step": 12353
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.558032274246216,
+      "learning_rate": 9.251444881311162e-06,
+      "loss": 0.7145,
+      "step": 12354
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.039466142654419,
+      "learning_rate": 9.251306107851494e-06,
+      "loss": 0.6834,
+      "step": 12355
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.0357935428619385,
+      "learning_rate": 9.251167322570536e-06,
+      "loss": 0.7222,
+      "step": 12356
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.4582135677337646,
+      "learning_rate": 9.251028525468673e-06,
+      "loss": 0.7718,
+      "step": 12357
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.3585351705551147,
+      "learning_rate": 9.250889716546293e-06,
+      "loss": 0.5096,
+      "step": 12358
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.491936445236206,
+      "learning_rate": 9.250750895803782e-06,
+      "loss": 0.7825,
+      "step": 12359
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.331367254257202,
+      "learning_rate": 9.250612063241525e-06,
+      "loss": 0.4221,
+      "step": 12360
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.878396987915039,
+      "learning_rate": 9.250473218859907e-06,
+      "loss": 0.4948,
+      "step": 12361
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.8209755420684814,
+      "learning_rate": 9.250334362659316e-06,
+      "loss": 0.6969,
+      "step": 12362
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.992044448852539,
+      "learning_rate": 9.250195494640137e-06,
+      "loss": 0.5036,
+      "step": 12363
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.4056942462921143,
+      "learning_rate": 9.250056614802755e-06,
+      "loss": 0.6991,
+      "step": 12364
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.2059731483459473,
+      "learning_rate": 9.249917723147561e-06,
+      "loss": 0.7063,
+      "step": 12365
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.9221125841140747,
+      "learning_rate": 9.249778819674936e-06,
+      "loss": 0.4626,
+      "step": 12366
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.664033889770508,
+      "learning_rate": 9.249639904385267e-06,
+      "loss": 0.745,
+      "step": 12367
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.022918224334717,
+      "learning_rate": 9.249500977278944e-06,
+      "loss": 0.6331,
+      "step": 12368
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 4.170511245727539,
+      "learning_rate": 9.249362038356348e-06,
+      "loss": 0.7111,
+      "step": 12369
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.0551750659942627,
+      "learning_rate": 9.24922308761787e-06,
+      "loss": 0.4069,
+      "step": 12370
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 8.434218406677246,
+      "learning_rate": 9.249084125063893e-06,
+      "loss": 0.732,
+      "step": 12371
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 4.468972206115723,
+      "learning_rate": 9.248945150694807e-06,
+      "loss": 0.6401,
+      "step": 12372
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.8383349180221558,
+      "learning_rate": 9.248806164510995e-06,
+      "loss": 0.5169,
+      "step": 12373
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 5.020917892456055,
+      "learning_rate": 9.248667166512843e-06,
+      "loss": 0.495,
+      "step": 12374
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.804445266723633,
+      "learning_rate": 9.24852815670074e-06,
+      "loss": 0.6876,
+      "step": 12375
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.830322027206421,
+      "learning_rate": 9.248389135075073e-06,
+      "loss": 0.4549,
+      "step": 12376
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 5.043947219848633,
+      "learning_rate": 9.248250101636226e-06,
+      "loss": 0.969,
+      "step": 12377
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.503241777420044,
+      "learning_rate": 9.24811105638459e-06,
+      "loss": 0.9119,
+      "step": 12378
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.5476701259613037,
+      "learning_rate": 9.247971999320546e-06,
+      "loss": 0.6558,
+      "step": 12379
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.902095079421997,
+      "learning_rate": 9.247832930444482e-06,
+      "loss": 0.5576,
+      "step": 12380
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.2265024185180664,
+      "learning_rate": 9.247693849756787e-06,
+      "loss": 0.6369,
+      "step": 12381
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.7180111408233643,
+      "learning_rate": 9.247554757257846e-06,
+      "loss": 0.5315,
+      "step": 12382
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.0590975284576416,
+      "learning_rate": 9.247415652948048e-06,
+      "loss": 0.7169,
+      "step": 12383
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.1919562816619873,
+      "learning_rate": 9.247276536827779e-06,
+      "loss": 0.9145,
+      "step": 12384
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.326895236968994,
+      "learning_rate": 9.247137408897421e-06,
+      "loss": 0.576,
+      "step": 12385
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.6211618185043335,
+      "learning_rate": 9.246998269157368e-06,
+      "loss": 0.4941,
+      "step": 12386
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.6400129795074463,
+      "learning_rate": 9.246859117608003e-06,
+      "loss": 0.6223,
+      "step": 12387
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.980117082595825,
+      "learning_rate": 9.246719954249715e-06,
+      "loss": 0.6195,
+      "step": 12388
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.519190788269043,
+      "learning_rate": 9.246580779082888e-06,
+      "loss": 0.7605,
+      "step": 12389
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.1944990158081055,
+      "learning_rate": 9.246441592107911e-06,
+      "loss": 0.6591,
+      "step": 12390
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.0882928371429443,
+      "learning_rate": 9.24630239332517e-06,
+      "loss": 0.8111,
+      "step": 12391
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.7350564002990723,
+      "learning_rate": 9.246163182735054e-06,
+      "loss": 0.4128,
+      "step": 12392
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.6730711460113525,
+      "learning_rate": 9.246023960337948e-06,
+      "loss": 0.5774,
+      "step": 12393
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.4786179065704346,
+      "learning_rate": 9.245884726134241e-06,
+      "loss": 0.5234,
+      "step": 12394
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.9031963348388672,
+      "learning_rate": 9.245745480124319e-06,
+      "loss": 0.8977,
+      "step": 12395
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.5657188892364502,
+      "learning_rate": 9.245606222308569e-06,
+      "loss": 0.7685,
+      "step": 12396
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.585869789123535,
+      "learning_rate": 9.245466952687377e-06,
+      "loss": 0.6662,
+      "step": 12397
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.293689012527466,
+      "learning_rate": 9.245327671261134e-06,
+      "loss": 0.8593,
+      "step": 12398
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.211344003677368,
+      "learning_rate": 9.245188378030223e-06,
+      "loss": 0.5811,
+      "step": 12399
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.630772352218628,
+      "learning_rate": 9.245049072995036e-06,
+      "loss": 0.5833,
+      "step": 12400
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.7430909872055054,
+      "learning_rate": 9.244909756155956e-06,
+      "loss": 0.563,
+      "step": 12401
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.011892318725586,
+      "learning_rate": 9.244770427513373e-06,
+      "loss": 0.5618,
+      "step": 12402
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.386631727218628,
+      "learning_rate": 9.244631087067673e-06,
+      "loss": 0.7725,
+      "step": 12403
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.0834274291992188,
+      "learning_rate": 9.244491734819244e-06,
+      "loss": 0.8633,
+      "step": 12404
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.474576950073242,
+      "learning_rate": 9.244352370768472e-06,
+      "loss": 0.6002,
+      "step": 12405
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.209557056427002,
+      "learning_rate": 9.244212994915748e-06,
+      "loss": 0.6865,
+      "step": 12406
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.9666974544525146,
+      "learning_rate": 9.244073607261459e-06,
+      "loss": 0.8369,
+      "step": 12407
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.842097759246826,
+      "learning_rate": 9.243934207805989e-06,
+      "loss": 0.51,
+      "step": 12408
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.879873514175415,
+      "learning_rate": 9.243794796549727e-06,
+      "loss": 0.6211,
+      "step": 12409
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.544738531112671,
+      "learning_rate": 9.243655373493064e-06,
+      "loss": 0.704,
+      "step": 12410
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 4.92232608795166,
+      "learning_rate": 9.243515938636382e-06,
+      "loss": 0.6758,
+      "step": 12411
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 4.277561664581299,
+      "learning_rate": 9.243376491980075e-06,
+      "loss": 0.6555,
+      "step": 12412
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.859069347381592,
+      "learning_rate": 9.243237033524525e-06,
+      "loss": 0.6623,
+      "step": 12413
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.3261852264404297,
+      "learning_rate": 9.243097563270123e-06,
+      "loss": 0.6947,
+      "step": 12414
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.59100341796875,
+      "learning_rate": 9.242958081217257e-06,
+      "loss": 0.9213,
+      "step": 12415
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.805638074874878,
+      "learning_rate": 9.242818587366313e-06,
+      "loss": 0.8251,
+      "step": 12416
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.3984410762786865,
+      "learning_rate": 9.242679081717681e-06,
+      "loss": 0.7288,
+      "step": 12417
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.0137457847595215,
+      "learning_rate": 9.242539564271748e-06,
+      "loss": 0.8483,
+      "step": 12418
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.215233564376831,
+      "learning_rate": 9.242400035028902e-06,
+      "loss": 0.4675,
+      "step": 12419
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.643981456756592,
+      "learning_rate": 9.242260493989529e-06,
+      "loss": 0.7433,
+      "step": 12420
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.8635258674621582,
+      "learning_rate": 9.24212094115402e-06,
+      "loss": 0.6283,
+      "step": 12421
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 4.3089280128479,
+      "learning_rate": 9.241981376522762e-06,
+      "loss": 1.0029,
+      "step": 12422
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.6704065799713135,
+      "learning_rate": 9.24184180009614e-06,
+      "loss": 0.8506,
+      "step": 12423
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.122138023376465,
+      "learning_rate": 9.241702211874548e-06,
+      "loss": 0.8737,
+      "step": 12424
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.1634910106658936,
+      "learning_rate": 9.241562611858372e-06,
+      "loss": 0.7766,
+      "step": 12425
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.238161563873291,
+      "learning_rate": 9.241423000047997e-06,
+      "loss": 0.7436,
+      "step": 12426
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.0288643836975098,
+      "learning_rate": 9.241283376443815e-06,
+      "loss": 0.9307,
+      "step": 12427
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.048780918121338,
+      "learning_rate": 9.241143741046211e-06,
+      "loss": 0.7314,
+      "step": 12428
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.435723066329956,
+      "learning_rate": 9.241004093855577e-06,
+      "loss": 0.8423,
+      "step": 12429
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.071012496948242,
+      "learning_rate": 9.240864434872297e-06,
+      "loss": 0.6265,
+      "step": 12430
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.6060549020767212,
+      "learning_rate": 9.240724764096763e-06,
+      "loss": 0.4291,
+      "step": 12431
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.1700785160064697,
+      "learning_rate": 9.240585081529362e-06,
+      "loss": 0.4806,
+      "step": 12432
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.281038761138916,
+      "learning_rate": 9.240445387170483e-06,
+      "loss": 0.8247,
+      "step": 12433
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.813568592071533,
+      "learning_rate": 9.240305681020514e-06,
+      "loss": 0.7545,
+      "step": 12434
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.4586069583892822,
+      "learning_rate": 9.240165963079842e-06,
+      "loss": 0.6096,
+      "step": 12435
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.302046537399292,
+      "learning_rate": 9.240026233348858e-06,
+      "loss": 0.8933,
+      "step": 12436
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.876852512359619,
+      "learning_rate": 9.239886491827949e-06,
+      "loss": 0.8371,
+      "step": 12437
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.6633803844451904,
+      "learning_rate": 9.239746738517503e-06,
+      "loss": 0.6496,
+      "step": 12438
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.4729278087615967,
+      "learning_rate": 9.239606973417912e-06,
+      "loss": 0.4597,
+      "step": 12439
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.30427622795105,
+      "learning_rate": 9.23946719652956e-06,
+      "loss": 0.6057,
+      "step": 12440
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.2565901279449463,
+      "learning_rate": 9.239327407852839e-06,
+      "loss": 0.5696,
+      "step": 12441
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.967308282852173,
+      "learning_rate": 9.239187607388135e-06,
+      "loss": 0.5756,
+      "step": 12442
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.359919786453247,
+      "learning_rate": 9.239047795135841e-06,
+      "loss": 0.6979,
+      "step": 12443
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 4.01053524017334,
+      "learning_rate": 9.23890797109634e-06,
+      "loss": 1.0037,
+      "step": 12444
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.724396228790283,
+      "learning_rate": 9.238768135270026e-06,
+      "loss": 0.7904,
+      "step": 12445
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.1084437370300293,
+      "learning_rate": 9.238628287657285e-06,
+      "loss": 0.7046,
+      "step": 12446
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.109597682952881,
+      "learning_rate": 9.238488428258508e-06,
+      "loss": 0.5403,
+      "step": 12447
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.4228028059005737,
+      "learning_rate": 9.23834855707408e-06,
+      "loss": 0.6764,
+      "step": 12448
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 5.281374454498291,
+      "learning_rate": 9.238208674104394e-06,
+      "loss": 0.5786,
+      "step": 12449
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.2669928073883057,
+      "learning_rate": 9.238068779349836e-06,
+      "loss": 0.5127,
+      "step": 12450
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.6516168117523193,
+      "learning_rate": 9.237928872810798e-06,
+      "loss": 0.7033,
+      "step": 12451
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 6.03978967666626,
+      "learning_rate": 9.237788954487666e-06,
+      "loss": 0.8733,
+      "step": 12452
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.7156982421875,
+      "learning_rate": 9.237649024380832e-06,
+      "loss": 0.5215,
+      "step": 12453
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.3148934841156006,
+      "learning_rate": 9.237509082490681e-06,
+      "loss": 0.8641,
+      "step": 12454
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.881591796875,
+      "learning_rate": 9.237369128817607e-06,
+      "loss": 0.6102,
+      "step": 12455
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.0742197036743164,
+      "learning_rate": 9.237229163361996e-06,
+      "loss": 0.5781,
+      "step": 12456
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.291358470916748,
+      "learning_rate": 9.237089186124238e-06,
+      "loss": 0.835,
+      "step": 12457
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 4.045928001403809,
+      "learning_rate": 9.236949197104721e-06,
+      "loss": 0.8545,
+      "step": 12458
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.581808090209961,
+      "learning_rate": 9.236809196303837e-06,
+      "loss": 0.8708,
+      "step": 12459
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.9234122037887573,
+      "learning_rate": 9.236669183721974e-06,
+      "loss": 0.6836,
+      "step": 12460
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.4448890686035156,
+      "learning_rate": 9.236529159359521e-06,
+      "loss": 0.6371,
+      "step": 12461
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.863875389099121,
+      "learning_rate": 9.236389123216867e-06,
+      "loss": 0.6691,
+      "step": 12462
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.9132838249206543,
+      "learning_rate": 9.2362490752944e-06,
+      "loss": 0.647,
+      "step": 12463
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.3933703899383545,
+      "learning_rate": 9.236109015592515e-06,
+      "loss": 0.9037,
+      "step": 12464
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.325589418411255,
+      "learning_rate": 9.235968944111595e-06,
+      "loss": 0.5687,
+      "step": 12465
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.087864398956299,
+      "learning_rate": 9.235828860852033e-06,
+      "loss": 1.0014,
+      "step": 12466
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.076434373855591,
+      "learning_rate": 9.235688765814218e-06,
+      "loss": 0.7199,
+      "step": 12467
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.2209551334381104,
+      "learning_rate": 9.23554865899854e-06,
+      "loss": 0.9066,
+      "step": 12468
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.6512937545776367,
+      "learning_rate": 9.235408540405385e-06,
+      "loss": 0.766,
+      "step": 12469
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.3855135440826416,
+      "learning_rate": 9.235268410035148e-06,
+      "loss": 0.6288,
+      "step": 12470
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.0940654277801514,
+      "learning_rate": 9.235128267888216e-06,
+      "loss": 0.8137,
+      "step": 12471
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.2502896785736084,
+      "learning_rate": 9.234988113964977e-06,
+      "loss": 0.7414,
+      "step": 12472
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.0234971046447754,
+      "learning_rate": 9.234847948265824e-06,
+      "loss": 0.8674,
+      "step": 12473
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.479001045227051,
+      "learning_rate": 9.234707770791145e-06,
+      "loss": 0.9456,
+      "step": 12474
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.808678150177002,
+      "learning_rate": 9.23456758154133e-06,
+      "loss": 0.6292,
+      "step": 12475
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.3697168827056885,
+      "learning_rate": 9.23442738051677e-06,
+      "loss": 0.7254,
+      "step": 12476
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.574604868888855,
+      "learning_rate": 9.234287167717853e-06,
+      "loss": 0.6452,
+      "step": 12477
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.322032928466797,
+      "learning_rate": 9.234146943144969e-06,
+      "loss": 0.2703,
+      "step": 12478
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.909510374069214,
+      "learning_rate": 9.234006706798508e-06,
+      "loss": 0.5587,
+      "step": 12479
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.6981528997421265,
+      "learning_rate": 9.233866458678861e-06,
+      "loss": 0.5305,
+      "step": 12480
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.253063440322876,
+      "learning_rate": 9.233726198786417e-06,
+      "loss": 0.3931,
+      "step": 12481
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.4420692920684814,
+      "learning_rate": 9.233585927121567e-06,
+      "loss": 0.6568,
+      "step": 12482
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.8937408924102783,
+      "learning_rate": 9.2334456436847e-06,
+      "loss": 0.7447,
+      "step": 12483
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.9541633129119873,
+      "learning_rate": 9.233305348476208e-06,
+      "loss": 0.7215,
+      "step": 12484
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.2144720554351807,
+      "learning_rate": 9.233165041496479e-06,
+      "loss": 0.6631,
+      "step": 12485
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.6477553844451904,
+      "learning_rate": 9.233024722745902e-06,
+      "loss": 0.8608,
+      "step": 12486
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.064216136932373,
+      "learning_rate": 9.232884392224871e-06,
+      "loss": 0.8107,
+      "step": 12487
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.1332051753997803,
+      "learning_rate": 9.232744049933774e-06,
+      "loss": 0.7165,
+      "step": 12488
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.735093116760254,
+      "learning_rate": 9.232603695873001e-06,
+      "loss": 0.7781,
+      "step": 12489
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.4364612102508545,
+      "learning_rate": 9.232463330042943e-06,
+      "loss": 0.445,
+      "step": 12490
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.8044382333755493,
+      "learning_rate": 9.23232295244399e-06,
+      "loss": 0.7659,
+      "step": 12491
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.1426913738250732,
+      "learning_rate": 9.232182563076531e-06,
+      "loss": 0.7782,
+      "step": 12492
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.785299301147461,
+      "learning_rate": 9.23204216194096e-06,
+      "loss": 0.7135,
+      "step": 12493
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.638582229614258,
+      "learning_rate": 9.231901749037664e-06,
+      "loss": 0.4476,
+      "step": 12494
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.9703367948532104,
+      "learning_rate": 9.231761324367034e-06,
+      "loss": 0.6125,
+      "step": 12495
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.720838785171509,
+      "learning_rate": 9.231620887929461e-06,
+      "loss": 0.9602,
+      "step": 12496
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.1608097553253174,
+      "learning_rate": 9.231480439725335e-06,
+      "loss": 0.5879,
+      "step": 12497
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.268277168273926,
+      "learning_rate": 9.231339979755049e-06,
+      "loss": 0.601,
+      "step": 12498
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.242241621017456,
+      "learning_rate": 9.23119950801899e-06,
+      "loss": 0.7147,
+      "step": 12499
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.2783896923065186,
+      "learning_rate": 9.231059024517552e-06,
+      "loss": 0.7693,
+      "step": 12500
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.1027722358703613,
+      "learning_rate": 9.230918529251123e-06,
+      "loss": 0.7662,
+      "step": 12501
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.897822856903076,
+      "learning_rate": 9.230778022220093e-06,
+      "loss": 0.4264,
+      "step": 12502
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.8145487308502197,
+      "learning_rate": 9.230637503424856e-06,
+      "loss": 0.5983,
+      "step": 12503
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.9370108842849731,
+      "learning_rate": 9.2304969728658e-06,
+      "loss": 0.624,
+      "step": 12504
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.699851632118225,
+      "learning_rate": 9.230356430543317e-06,
+      "loss": 0.6332,
+      "step": 12505
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 5.152247905731201,
+      "learning_rate": 9.230215876457797e-06,
+      "loss": 0.3086,
+      "step": 12506
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.9627952575683594,
+      "learning_rate": 9.230075310609631e-06,
+      "loss": 0.8532,
+      "step": 12507
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.762157917022705,
+      "learning_rate": 9.229934732999213e-06,
+      "loss": 0.7342,
+      "step": 12508
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.804769515991211,
+      "learning_rate": 9.229794143626928e-06,
+      "loss": 0.6458,
+      "step": 12509
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.2904515266418457,
+      "learning_rate": 9.229653542493172e-06,
+      "loss": 0.6358,
+      "step": 12510
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.6092631816864014,
+      "learning_rate": 9.229512929598332e-06,
+      "loss": 0.6451,
+      "step": 12511
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.5188310146331787,
+      "learning_rate": 9.229372304942801e-06,
+      "loss": 0.7769,
+      "step": 12512
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.007556438446045,
+      "learning_rate": 9.229231668526972e-06,
+      "loss": 0.7235,
+      "step": 12513
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.084427833557129,
+      "learning_rate": 9.229091020351234e-06,
+      "loss": 0.5032,
+      "step": 12514
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.3150734901428223,
+      "learning_rate": 9.228950360415976e-06,
+      "loss": 0.5384,
+      "step": 12515
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.2797852754592896,
+      "learning_rate": 9.228809688721593e-06,
+      "loss": 0.7076,
+      "step": 12516
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.9447929859161377,
+      "learning_rate": 9.228669005268474e-06,
+      "loss": 0.7852,
+      "step": 12517
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.459329843521118,
+      "learning_rate": 9.22852831005701e-06,
+      "loss": 0.6077,
+      "step": 12518
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.643657684326172,
+      "learning_rate": 9.228387603087593e-06,
+      "loss": 0.7635,
+      "step": 12519
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.9708025455474854,
+      "learning_rate": 9.228246884360614e-06,
+      "loss": 0.72,
+      "step": 12520
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.1364240646362305,
+      "learning_rate": 9.228106153876464e-06,
+      "loss": 0.6899,
+      "step": 12521
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.952735662460327,
+      "learning_rate": 9.227965411635535e-06,
+      "loss": 0.8827,
+      "step": 12522
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.586808204650879,
+      "learning_rate": 9.22782465763822e-06,
+      "loss": 0.7706,
+      "step": 12523
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 4.543497085571289,
+      "learning_rate": 9.227683891884907e-06,
+      "loss": 0.7824,
+      "step": 12524
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.129917621612549,
+      "learning_rate": 9.227543114375987e-06,
+      "loss": 0.7574,
+      "step": 12525
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.620213270187378,
+      "learning_rate": 9.227402325111856e-06,
+      "loss": 0.8016,
+      "step": 12526
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.279648542404175,
+      "learning_rate": 9.2272615240929e-06,
+      "loss": 0.5397,
+      "step": 12527
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.5375142097473145,
+      "learning_rate": 9.227120711319516e-06,
+      "loss": 0.7912,
+      "step": 12528
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.055577039718628,
+      "learning_rate": 9.226979886792092e-06,
+      "loss": 0.613,
+      "step": 12529
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.446528911590576,
+      "learning_rate": 9.22683905051102e-06,
+      "loss": 0.7073,
+      "step": 12530
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.926478862762451,
+      "learning_rate": 9.22669820247669e-06,
+      "loss": 0.6108,
+      "step": 12531
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.2567265033721924,
+      "learning_rate": 9.2265573426895e-06,
+      "loss": 0.7477,
+      "step": 12532
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.471010446548462,
+      "learning_rate": 9.226416471149835e-06,
+      "loss": 0.5236,
+      "step": 12533
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.956939697265625,
+      "learning_rate": 9.226275587858089e-06,
+      "loss": 0.6829,
+      "step": 12534
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.8125638961791992,
+      "learning_rate": 9.226134692814654e-06,
+      "loss": 0.8938,
+      "step": 12535
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 6.390133857727051,
+      "learning_rate": 9.225993786019921e-06,
+      "loss": 0.5942,
+      "step": 12536
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.9591516256332397,
+      "learning_rate": 9.225852867474283e-06,
+      "loss": 0.6381,
+      "step": 12537
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.977221965789795,
+      "learning_rate": 9.225711937178131e-06,
+      "loss": 0.6236,
+      "step": 12538
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.6235506534576416,
+      "learning_rate": 9.225570995131857e-06,
+      "loss": 0.4161,
+      "step": 12539
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.221256971359253,
+      "learning_rate": 9.225430041335853e-06,
+      "loss": 0.5803,
+      "step": 12540
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.608591914176941,
+      "learning_rate": 9.22528907579051e-06,
+      "loss": 0.8344,
+      "step": 12541
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.69348406791687,
+      "learning_rate": 9.225148098496224e-06,
+      "loss": 0.483,
+      "step": 12542
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.2196147441864014,
+      "learning_rate": 9.22500710945338e-06,
+      "loss": 0.6178,
+      "step": 12543
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.6764872074127197,
+      "learning_rate": 9.224866108662376e-06,
+      "loss": 0.5282,
+      "step": 12544
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.2808003425598145,
+      "learning_rate": 9.224725096123603e-06,
+      "loss": 0.9476,
+      "step": 12545
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.4179062843322754,
+      "learning_rate": 9.22458407183745e-06,
+      "loss": 0.3923,
+      "step": 12546
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.078428030014038,
+      "learning_rate": 9.224443035804313e-06,
+      "loss": 0.4649,
+      "step": 12547
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.062676429748535,
+      "learning_rate": 9.224301988024582e-06,
+      "loss": 0.8748,
+      "step": 12548
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.0881283283233643,
+      "learning_rate": 9.22416092849865e-06,
+      "loss": 0.8716,
+      "step": 12549
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 4.049964427947998,
+      "learning_rate": 9.224019857226906e-06,
+      "loss": 0.3764,
+      "step": 12550
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.715399742126465,
+      "learning_rate": 9.223878774209748e-06,
+      "loss": 0.4357,
+      "step": 12551
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.3814523220062256,
+      "learning_rate": 9.223737679447566e-06,
+      "loss": 0.5985,
+      "step": 12552
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.4296467304229736,
+      "learning_rate": 9.22359657294075e-06,
+      "loss": 0.5602,
+      "step": 12553
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.033048391342163,
+      "learning_rate": 9.223455454689694e-06,
+      "loss": 0.8188,
+      "step": 12554
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.731144428253174,
+      "learning_rate": 9.223314324694792e-06,
+      "loss": 0.7041,
+      "step": 12555
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.8055109977722168,
+      "learning_rate": 9.223173182956434e-06,
+      "loss": 0.6903,
+      "step": 12556
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.7528791427612305,
+      "learning_rate": 9.223032029475012e-06,
+      "loss": 0.6798,
+      "step": 12557
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.9766985177993774,
+      "learning_rate": 9.222890864250922e-06,
+      "loss": 0.709,
+      "step": 12558
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.3137835264205933,
+      "learning_rate": 9.222749687284554e-06,
+      "loss": 0.752,
+      "step": 12559
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 4.264050006866455,
+      "learning_rate": 9.222608498576302e-06,
+      "loss": 0.5423,
+      "step": 12560
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.2166452407836914,
+      "learning_rate": 9.222467298126556e-06,
+      "loss": 0.9152,
+      "step": 12561
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.615168809890747,
+      "learning_rate": 9.222326085935712e-06,
+      "loss": 0.5639,
+      "step": 12562
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.6135427951812744,
+      "learning_rate": 9.222184862004157e-06,
+      "loss": 0.734,
+      "step": 12563
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.8100800514221191,
+      "learning_rate": 9.22204362633229e-06,
+      "loss": 0.8492,
+      "step": 12564
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.590782880783081,
+      "learning_rate": 9.221902378920502e-06,
+      "loss": 0.7305,
+      "step": 12565
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.257709503173828,
+      "learning_rate": 9.221761119769185e-06,
+      "loss": 0.6155,
+      "step": 12566
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.25681209564209,
+      "learning_rate": 9.22161984887873e-06,
+      "loss": 0.7573,
+      "step": 12567
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.9260931015014648,
+      "learning_rate": 9.221478566249534e-06,
+      "loss": 0.7067,
+      "step": 12568
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.547590732574463,
+      "learning_rate": 9.221337271881988e-06,
+      "loss": 0.6496,
+      "step": 12569
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.8202400207519531,
+      "learning_rate": 9.221195965776482e-06,
+      "loss": 0.5161,
+      "step": 12570
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.8929543495178223,
+      "learning_rate": 9.221054647933413e-06,
+      "loss": 0.803,
+      "step": 12571
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.0759429931640625,
+      "learning_rate": 9.22091331835317e-06,
+      "loss": 0.7429,
+      "step": 12572
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.1511709690093994,
+      "learning_rate": 9.220771977036149e-06,
+      "loss": 0.7928,
+      "step": 12573
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.5796995162963867,
+      "learning_rate": 9.220630623982743e-06,
+      "loss": 0.7415,
+      "step": 12574
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.1615805625915527,
+      "learning_rate": 9.220489259193345e-06,
+      "loss": 0.5633,
+      "step": 12575
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.8268343210220337,
+      "learning_rate": 9.220347882668347e-06,
+      "loss": 0.4924,
+      "step": 12576
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.8333961963653564,
+      "learning_rate": 9.220206494408142e-06,
+      "loss": 0.8398,
+      "step": 12577
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.088759183883667,
+      "learning_rate": 9.220065094413124e-06,
+      "loss": 0.6975,
+      "step": 12578
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.906778335571289,
+      "learning_rate": 9.219923682683684e-06,
+      "loss": 0.7703,
+      "step": 12579
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.6805884838104248,
+      "learning_rate": 9.21978225922022e-06,
+      "loss": 0.723,
+      "step": 12580
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.8527075052261353,
+      "learning_rate": 9.219640824023121e-06,
+      "loss": 0.5964,
+      "step": 12581
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 6.131479263305664,
+      "learning_rate": 9.219499377092782e-06,
+      "loss": 0.7719,
+      "step": 12582
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.7294673919677734,
+      "learning_rate": 9.219357918429594e-06,
+      "loss": 0.3677,
+      "step": 12583
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.4809324741363525,
+      "learning_rate": 9.219216448033956e-06,
+      "loss": 0.5976,
+      "step": 12584
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.8943865299224854,
+      "learning_rate": 9.219074965906254e-06,
+      "loss": 0.694,
+      "step": 12585
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.918992280960083,
+      "learning_rate": 9.218933472046887e-06,
+      "loss": 0.5266,
+      "step": 12586
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.876474142074585,
+      "learning_rate": 9.218791966456247e-06,
+      "loss": 0.4856,
+      "step": 12587
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.7890996932983398,
+      "learning_rate": 9.218650449134724e-06,
+      "loss": 0.4417,
+      "step": 12588
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.505664587020874,
+      "learning_rate": 9.218508920082717e-06,
+      "loss": 0.6376,
+      "step": 12589
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.7800554037094116,
+      "learning_rate": 9.218367379300616e-06,
+      "loss": 0.6689,
+      "step": 12590
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.9497182369232178,
+      "learning_rate": 9.218225826788815e-06,
+      "loss": 0.5716,
+      "step": 12591
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.748012900352478,
+      "learning_rate": 9.21808426254771e-06,
+      "loss": 0.7084,
+      "step": 12592
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.574145793914795,
+      "learning_rate": 9.217942686577692e-06,
+      "loss": 0.6124,
+      "step": 12593
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.0071821212768555,
+      "learning_rate": 9.217801098879156e-06,
+      "loss": 0.7149,
+      "step": 12594
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 3.4257211685180664,
+      "learning_rate": 9.217659499452494e-06,
+      "loss": 0.7111,
+      "step": 12595
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.8422884941101074,
+      "learning_rate": 9.2175178882981e-06,
+      "loss": 0.734,
+      "step": 12596
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.2319886684417725,
+      "learning_rate": 9.21737626541637e-06,
+      "loss": 0.5423,
+      "step": 12597
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.1245274543762207,
+      "learning_rate": 9.217234630807696e-06,
+      "loss": 0.7171,
+      "step": 12598
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.233170509338379,
+      "learning_rate": 9.217092984472473e-06,
+      "loss": 0.6359,
+      "step": 12599
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 3.375264883041382,
+      "learning_rate": 9.216951326411095e-06,
+      "loss": 0.643,
+      "step": 12600
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.685068130493164,
+      "learning_rate": 9.216809656623953e-06,
+      "loss": 0.3908,
+      "step": 12601
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.9926544427871704,
+      "learning_rate": 9.216667975111445e-06,
+      "loss": 0.6344,
+      "step": 12602
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.7121317386627197,
+      "learning_rate": 9.21652628187396e-06,
+      "loss": 0.8278,
+      "step": 12603
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 3.5454790592193604,
+      "learning_rate": 9.216384576911899e-06,
+      "loss": 0.7205,
+      "step": 12604
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.024989604949951,
+      "learning_rate": 9.216242860225648e-06,
+      "loss": 0.7439,
+      "step": 12605
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 3.1078155040740967,
+      "learning_rate": 9.216101131815609e-06,
+      "loss": 0.8635,
+      "step": 12606
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.398090362548828,
+      "learning_rate": 9.215959391682169e-06,
+      "loss": 0.6909,
+      "step": 12607
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.9122759103775024,
+      "learning_rate": 9.215817639825726e-06,
+      "loss": 0.5876,
+      "step": 12608
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 4.146350383758545,
+      "learning_rate": 9.215675876246674e-06,
+      "loss": 0.7384,
+      "step": 12609
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 3.018558979034424,
+      "learning_rate": 9.215534100945407e-06,
+      "loss": 0.8349,
+      "step": 12610
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.3783390522003174,
+      "learning_rate": 9.215392313922318e-06,
+      "loss": 0.6823,
+      "step": 12611
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.9925901889801025,
+      "learning_rate": 9.215250515177803e-06,
+      "loss": 0.4625,
+      "step": 12612
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.1424574851989746,
+      "learning_rate": 9.215108704712253e-06,
+      "loss": 0.8085,
+      "step": 12613
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.980588674545288,
+      "learning_rate": 9.214966882526067e-06,
+      "loss": 0.7151,
+      "step": 12614
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.5825607776641846,
+      "learning_rate": 9.214825048619636e-06,
+      "loss": 0.6977,
+      "step": 12615
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.248112201690674,
+      "learning_rate": 9.214683202993357e-06,
+      "loss": 0.6917,
+      "step": 12616
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 5.481602668762207,
+      "learning_rate": 9.21454134564762e-06,
+      "loss": 0.5653,
+      "step": 12617
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.054347515106201,
+      "learning_rate": 9.214399476582825e-06,
+      "loss": 0.9518,
+      "step": 12618
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.8319151401519775,
+      "learning_rate": 9.214257595799363e-06,
+      "loss": 0.6083,
+      "step": 12619
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 3.124462604522705,
+      "learning_rate": 9.214115703297629e-06,
+      "loss": 0.5679,
+      "step": 12620
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.7661571502685547,
+      "learning_rate": 9.213973799078017e-06,
+      "loss": 0.6292,
+      "step": 12621
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.3150992393493652,
+      "learning_rate": 9.213831883140925e-06,
+      "loss": 0.5724,
+      "step": 12622
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.546724557876587,
+      "learning_rate": 9.213689955486744e-06,
+      "loss": 0.796,
+      "step": 12623
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 3.799065113067627,
+      "learning_rate": 9.213548016115868e-06,
+      "loss": 0.7461,
+      "step": 12624
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.5219719409942627,
+      "learning_rate": 9.213406065028697e-06,
+      "loss": 0.455,
+      "step": 12625
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.7492650747299194,
+      "learning_rate": 9.213264102225618e-06,
+      "loss": 0.5935,
+      "step": 12626
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 4.046304225921631,
+      "learning_rate": 9.213122127707032e-06,
+      "loss": 0.7462,
+      "step": 12627
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.2150392532348633,
+      "learning_rate": 9.212980141473332e-06,
+      "loss": 0.7031,
+      "step": 12628
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.9930732250213623,
+      "learning_rate": 9.212838143524912e-06,
+      "loss": 0.7565,
+      "step": 12629
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.8134121894836426,
+      "learning_rate": 9.212696133862166e-06,
+      "loss": 0.5665,
+      "step": 12630
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.205732822418213,
+      "learning_rate": 9.212554112485492e-06,
+      "loss": 0.5748,
+      "step": 12631
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.8457581996917725,
+      "learning_rate": 9.212412079395281e-06,
+      "loss": 0.6607,
+      "step": 12632
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.4777705669403076,
+      "learning_rate": 9.212270034591932e-06,
+      "loss": 0.8275,
+      "step": 12633
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.8297882080078125,
+      "learning_rate": 9.212127978075837e-06,
+      "loss": 0.7939,
+      "step": 12634
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.38720965385437,
+      "learning_rate": 9.211985909847393e-06,
+      "loss": 0.4675,
+      "step": 12635
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.179983139038086,
+      "learning_rate": 9.211843829906993e-06,
+      "loss": 0.6956,
+      "step": 12636
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.6611883640289307,
+      "learning_rate": 9.211701738255033e-06,
+      "loss": 0.6704,
+      "step": 12637
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.5594241619110107,
+      "learning_rate": 9.211559634891909e-06,
+      "loss": 0.5334,
+      "step": 12638
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 3.228337287902832,
+      "learning_rate": 9.211417519818015e-06,
+      "loss": 0.6451,
+      "step": 12639
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.751563310623169,
+      "learning_rate": 9.211275393033746e-06,
+      "loss": 0.5715,
+      "step": 12640
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.757843255996704,
+      "learning_rate": 9.2111332545395e-06,
+      "loss": 0.6489,
+      "step": 12641
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.628127336502075,
+      "learning_rate": 9.210991104335668e-06,
+      "loss": 0.7416,
+      "step": 12642
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.7582337856292725,
+      "learning_rate": 9.210848942422645e-06,
+      "loss": 0.7554,
+      "step": 12643
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.131995916366577,
+      "learning_rate": 9.210706768800833e-06,
+      "loss": 0.7147,
+      "step": 12644
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 3.61264705657959,
+      "learning_rate": 9.21056458347062e-06,
+      "loss": 0.6996,
+      "step": 12645
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.8482601642608643,
+      "learning_rate": 9.210422386432405e-06,
+      "loss": 0.8228,
+      "step": 12646
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.8238847255706787,
+      "learning_rate": 9.210280177686583e-06,
+      "loss": 0.7592,
+      "step": 12647
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 4.001441955566406,
+      "learning_rate": 9.210137957233548e-06,
+      "loss": 0.5306,
+      "step": 12648
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.1801748275756836,
+      "learning_rate": 9.209995725073697e-06,
+      "loss": 0.675,
+      "step": 12649
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.9809799194335938,
+      "learning_rate": 9.209853481207425e-06,
+      "loss": 0.6838,
+      "step": 12650
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 3.0674872398376465,
+      "learning_rate": 9.209711225635128e-06,
+      "loss": 0.6266,
+      "step": 12651
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.5058728456497192,
+      "learning_rate": 9.2095689583572e-06,
+      "loss": 0.6722,
+      "step": 12652
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 4.673649787902832,
+      "learning_rate": 9.209426679374039e-06,
+      "loss": 0.7391,
+      "step": 12653
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 3.181621789932251,
+      "learning_rate": 9.209284388686038e-06,
+      "loss": 0.6251,
+      "step": 12654
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 3.7506842613220215,
+      "learning_rate": 9.209142086293594e-06,
+      "loss": 0.6815,
+      "step": 12655
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.0671169757843018,
+      "learning_rate": 9.208999772197104e-06,
+      "loss": 0.8535,
+      "step": 12656
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.0332300662994385,
+      "learning_rate": 9.208857446396961e-06,
+      "loss": 0.7433,
+      "step": 12657
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 3.144956350326538,
+      "learning_rate": 9.208715108893563e-06,
+      "loss": 0.6381,
+      "step": 12658
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.61617374420166,
+      "learning_rate": 9.208572759687305e-06,
+      "loss": 0.6368,
+      "step": 12659
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.391141653060913,
+      "learning_rate": 9.208430398778582e-06,
+      "loss": 0.6346,
+      "step": 12660
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.293999195098877,
+      "learning_rate": 9.20828802616779e-06,
+      "loss": 0.8591,
+      "step": 12661
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 3.136676549911499,
+      "learning_rate": 9.208145641855327e-06,
+      "loss": 0.6463,
+      "step": 12662
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.7572304010391235,
+      "learning_rate": 9.208003245841586e-06,
+      "loss": 0.7903,
+      "step": 12663
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.760211944580078,
+      "learning_rate": 9.207860838126966e-06,
+      "loss": 0.6071,
+      "step": 12664
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 3.7785756587982178,
+      "learning_rate": 9.20771841871186e-06,
+      "loss": 0.5994,
+      "step": 12665
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.351851463317871,
+      "learning_rate": 9.207575987596665e-06,
+      "loss": 0.7198,
+      "step": 12666
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.7117629051208496,
+      "learning_rate": 9.207433544781779e-06,
+      "loss": 0.4475,
+      "step": 12667
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.589327573776245,
+      "learning_rate": 9.207291090267594e-06,
+      "loss": 0.6323,
+      "step": 12668
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.6379921436309814,
+      "learning_rate": 9.20714862405451e-06,
+      "loss": 0.7784,
+      "step": 12669
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.825004816055298,
+      "learning_rate": 9.207006146142923e-06,
+      "loss": 0.4035,
+      "step": 12670
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 7.059628486633301,
+      "learning_rate": 9.206863656533227e-06,
+      "loss": 0.6865,
+      "step": 12671
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.6932289600372314,
+      "learning_rate": 9.206721155225817e-06,
+      "loss": 0.54,
+      "step": 12672
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.606825590133667,
+      "learning_rate": 9.206578642221093e-06,
+      "loss": 0.5599,
+      "step": 12673
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.0118987560272217,
+      "learning_rate": 9.20643611751945e-06,
+      "loss": 0.6689,
+      "step": 12674
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.6534324884414673,
+      "learning_rate": 9.206293581121285e-06,
+      "loss": 0.7083,
+      "step": 12675
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.5395588874816895,
+      "learning_rate": 9.206151033026991e-06,
+      "loss": 0.5263,
+      "step": 12676
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.6911940574645996,
+      "learning_rate": 9.206008473236969e-06,
+      "loss": 0.7484,
+      "step": 12677
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 3.6189167499542236,
+      "learning_rate": 9.20586590175161e-06,
+      "loss": 0.8192,
+      "step": 12678
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.550219178199768,
+      "learning_rate": 9.205723318571316e-06,
+      "loss": 0.8008,
+      "step": 12679
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.599794387817383,
+      "learning_rate": 9.205580723696479e-06,
+      "loss": 0.6818,
+      "step": 12680
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.7410829067230225,
+      "learning_rate": 9.205438117127498e-06,
+      "loss": 0.6051,
+      "step": 12681
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.360584020614624,
+      "learning_rate": 9.20529549886477e-06,
+      "loss": 0.632,
+      "step": 12682
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.845848321914673,
+      "learning_rate": 9.205152868908691e-06,
+      "loss": 0.8516,
+      "step": 12683
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.29023814201355,
+      "learning_rate": 9.205010227259656e-06,
+      "loss": 0.6498,
+      "step": 12684
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.3893303871154785,
+      "learning_rate": 9.204867573918063e-06,
+      "loss": 0.5384,
+      "step": 12685
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.8861345052719116,
+      "learning_rate": 9.204724908884309e-06,
+      "loss": 0.7989,
+      "step": 12686
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.643306255340576,
+      "learning_rate": 9.204582232158789e-06,
+      "loss": 0.6292,
+      "step": 12687
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.6376177072525024,
+      "learning_rate": 9.2044395437419e-06,
+      "loss": 0.7754,
+      "step": 12688
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.2409374713897705,
+      "learning_rate": 9.204296843634041e-06,
+      "loss": 0.618,
+      "step": 12689
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 3.799917697906494,
+      "learning_rate": 9.20415413183561e-06,
+      "loss": 0.9201,
+      "step": 12690
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.7751599550247192,
+      "learning_rate": 9.204011408346998e-06,
+      "loss": 0.6531,
+      "step": 12691
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.28428053855896,
+      "learning_rate": 9.203868673168606e-06,
+      "loss": 0.5233,
+      "step": 12692
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.7973806858062744,
+      "learning_rate": 9.20372592630083e-06,
+      "loss": 0.7476,
+      "step": 12693
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.189854621887207,
+      "learning_rate": 9.203583167744065e-06,
+      "loss": 0.6673,
+      "step": 12694
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.863605260848999,
+      "learning_rate": 9.203440397498713e-06,
+      "loss": 0.3249,
+      "step": 12695
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.5698869228363037,
+      "learning_rate": 9.203297615565166e-06,
+      "loss": 0.5854,
+      "step": 12696
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.613109827041626,
+      "learning_rate": 9.203154821943826e-06,
+      "loss": 0.5119,
+      "step": 12697
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.489539861679077,
+      "learning_rate": 9.203012016635083e-06,
+      "loss": 0.5965,
+      "step": 12698
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.4820283651351929,
+      "learning_rate": 9.20286919963934e-06,
+      "loss": 0.6813,
+      "step": 12699
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.995190143585205,
+      "learning_rate": 9.202726370956992e-06,
+      "loss": 0.6857,
+      "step": 12700
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.7726150751113892,
+      "learning_rate": 9.202583530588435e-06,
+      "loss": 0.6636,
+      "step": 12701
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.4142119884490967,
+      "learning_rate": 9.202440678534068e-06,
+      "loss": 0.5521,
+      "step": 12702
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.3650816679000854,
+      "learning_rate": 9.20229781479429e-06,
+      "loss": 0.4561,
+      "step": 12703
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.5448367595672607,
+      "learning_rate": 9.202154939369493e-06,
+      "loss": 0.6731,
+      "step": 12704
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 3.6119823455810547,
+      "learning_rate": 9.20201205226008e-06,
+      "loss": 0.5535,
+      "step": 12705
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.028679609298706,
+      "learning_rate": 9.201869153466445e-06,
+      "loss": 0.4422,
+      "step": 12706
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.2809014320373535,
+      "learning_rate": 9.201726242988983e-06,
+      "loss": 0.7368,
+      "step": 12707
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.8887741565704346,
+      "learning_rate": 9.201583320828098e-06,
+      "loss": 0.7342,
+      "step": 12708
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 3.8069376945495605,
+      "learning_rate": 9.201440386984182e-06,
+      "loss": 0.8225,
+      "step": 12709
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.0203561782836914,
+      "learning_rate": 9.201297441457636e-06,
+      "loss": 0.6572,
+      "step": 12710
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.707066774368286,
+      "learning_rate": 9.201154484248853e-06,
+      "loss": 0.6969,
+      "step": 12711
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.7047717571258545,
+      "learning_rate": 9.201011515358234e-06,
+      "loss": 0.6641,
+      "step": 12712
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.040815591812134,
+      "learning_rate": 9.200868534786177e-06,
+      "loss": 0.593,
+      "step": 12713
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.8343148231506348,
+      "learning_rate": 9.200725542533078e-06,
+      "loss": 0.5375,
+      "step": 12714
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.579420804977417,
+      "learning_rate": 9.200582538599334e-06,
+      "loss": 0.5938,
+      "step": 12715
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.365905523300171,
+      "learning_rate": 9.200439522985342e-06,
+      "loss": 0.6175,
+      "step": 12716
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.454341173171997,
+      "learning_rate": 9.200296495691504e-06,
+      "loss": 0.8475,
+      "step": 12717
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.6356818675994873,
+      "learning_rate": 9.200153456718215e-06,
+      "loss": 0.5284,
+      "step": 12718
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.166733741760254,
+      "learning_rate": 9.200010406065871e-06,
+      "loss": 0.7575,
+      "step": 12719
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.711897373199463,
+      "learning_rate": 9.199867343734872e-06,
+      "loss": 0.5774,
+      "step": 12720
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.3475077152252197,
+      "learning_rate": 9.199724269725614e-06,
+      "loss": 0.679,
+      "step": 12721
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.7408289909362793,
+      "learning_rate": 9.199581184038498e-06,
+      "loss": 0.7399,
+      "step": 12722
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.708890676498413,
+      "learning_rate": 9.19943808667392e-06,
+      "loss": 0.9552,
+      "step": 12723
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 3.5177083015441895,
+      "learning_rate": 9.199294977632275e-06,
+      "loss": 0.8493,
+      "step": 12724
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.3991572856903076,
+      "learning_rate": 9.199151856913967e-06,
+      "loss": 0.8236,
+      "step": 12725
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 3.360039472579956,
+      "learning_rate": 9.199008724519389e-06,
+      "loss": 0.7324,
+      "step": 12726
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 3.7429494857788086,
+      "learning_rate": 9.198865580448941e-06,
+      "loss": 0.8043,
+      "step": 12727
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.2391161918640137,
+      "learning_rate": 9.198722424703022e-06,
+      "loss": 0.6048,
+      "step": 12728
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.665215015411377,
+      "learning_rate": 9.198579257282027e-06,
+      "loss": 0.5865,
+      "step": 12729
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 5.488815784454346,
+      "learning_rate": 9.198436078186357e-06,
+      "loss": 0.679,
+      "step": 12730
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 3.8323891162872314,
+      "learning_rate": 9.19829288741641e-06,
+      "loss": 0.7711,
+      "step": 12731
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.371929883956909,
+      "learning_rate": 9.19814968497258e-06,
+      "loss": 0.7623,
+      "step": 12732
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.2381551265716553,
+      "learning_rate": 9.19800647085527e-06,
+      "loss": 0.6472,
+      "step": 12733
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.7624828815460205,
+      "learning_rate": 9.197863245064878e-06,
+      "loss": 0.613,
+      "step": 12734
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 3.3852877616882324,
+      "learning_rate": 9.197720007601799e-06,
+      "loss": 0.6538,
+      "step": 12735
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 4.1268534660339355,
+      "learning_rate": 9.197576758466434e-06,
+      "loss": 0.735,
+      "step": 12736
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.4783387184143066,
+      "learning_rate": 9.19743349765918e-06,
+      "loss": 0.6087,
+      "step": 12737
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.978574275970459,
+      "learning_rate": 9.197290225180438e-06,
+      "loss": 0.674,
+      "step": 12738
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 4.072201251983643,
+      "learning_rate": 9.197146941030602e-06,
+      "loss": 1.0673,
+      "step": 12739
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.340719699859619,
+      "learning_rate": 9.197003645210072e-06,
+      "loss": 1.0408,
+      "step": 12740
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.23081111907959,
+      "learning_rate": 9.196860337719248e-06,
+      "loss": 0.7313,
+      "step": 12741
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.8630762100219727,
+      "learning_rate": 9.196717018558529e-06,
+      "loss": 0.4602,
+      "step": 12742
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.4723198413848877,
+      "learning_rate": 9.19657368772831e-06,
+      "loss": 0.5924,
+      "step": 12743
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.7307453155517578,
+      "learning_rate": 9.196430345228992e-06,
+      "loss": 0.6427,
+      "step": 12744
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 3.4955809116363525,
+      "learning_rate": 9.196286991060975e-06,
+      "loss": 0.6814,
+      "step": 12745
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 3.3079874515533447,
+      "learning_rate": 9.196143625224654e-06,
+      "loss": 0.6597,
+      "step": 12746
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.5228843688964844,
+      "learning_rate": 9.196000247720431e-06,
+      "loss": 0.6695,
+      "step": 12747
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.7121808528900146,
+      "learning_rate": 9.195856858548701e-06,
+      "loss": 0.7762,
+      "step": 12748
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.5702043771743774,
+      "learning_rate": 9.195713457709867e-06,
+      "loss": 0.7552,
+      "step": 12749
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.1601037979125977,
+      "learning_rate": 9.195570045204326e-06,
+      "loss": 0.7609,
+      "step": 12750
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.833166241645813,
+      "learning_rate": 9.195426621032474e-06,
+      "loss": 0.5305,
+      "step": 12751
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.4798858165740967,
+      "learning_rate": 9.195283185194714e-06,
+      "loss": 0.5551,
+      "step": 12752
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.3369839191436768,
+      "learning_rate": 9.195139737691442e-06,
+      "loss": 0.6291,
+      "step": 12753
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.8015663623809814,
+      "learning_rate": 9.19499627852306e-06,
+      "loss": 0.6877,
+      "step": 12754
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.6999808549880981,
+      "learning_rate": 9.194852807689962e-06,
+      "loss": 0.468,
+      "step": 12755
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 3.081904888153076,
+      "learning_rate": 9.19470932519255e-06,
+      "loss": 0.8328,
+      "step": 12756
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.9879435300827026,
+      "learning_rate": 9.194565831031225e-06,
+      "loss": 0.6314,
+      "step": 12757
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.4725098609924316,
+      "learning_rate": 9.194422325206382e-06,
+      "loss": 0.644,
+      "step": 12758
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.291466474533081,
+      "learning_rate": 9.194278807718422e-06,
+      "loss": 0.9589,
+      "step": 12759
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.177659034729004,
+      "learning_rate": 9.194135278567743e-06,
+      "loss": 0.5502,
+      "step": 12760
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.2806971073150635,
+      "learning_rate": 9.193991737754747e-06,
+      "loss": 0.5715,
+      "step": 12761
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.8029550313949585,
+      "learning_rate": 9.19384818527983e-06,
+      "loss": 0.4294,
+      "step": 12762
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.6376923322677612,
+      "learning_rate": 9.193704621143393e-06,
+      "loss": 0.8018,
+      "step": 12763
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.555217981338501,
+      "learning_rate": 9.193561045345834e-06,
+      "loss": 0.7629,
+      "step": 12764
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.131535530090332,
+      "learning_rate": 9.193417457887552e-06,
+      "loss": 0.6666,
+      "step": 12765
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.4468986988067627,
+      "learning_rate": 9.193273858768948e-06,
+      "loss": 0.6711,
+      "step": 12766
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.8136781454086304,
+      "learning_rate": 9.193130247990419e-06,
+      "loss": 0.6104,
+      "step": 12767
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.492905855178833,
+      "learning_rate": 9.192986625552366e-06,
+      "loss": 0.5921,
+      "step": 12768
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 3.4952218532562256,
+      "learning_rate": 9.192842991455188e-06,
+      "loss": 0.5402,
+      "step": 12769
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.1692771911621094,
+      "learning_rate": 9.192699345699286e-06,
+      "loss": 0.6715,
+      "step": 12770
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.6757129430770874,
+      "learning_rate": 9.192555688285054e-06,
+      "loss": 0.7853,
+      "step": 12771
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 3.9565088748931885,
+      "learning_rate": 9.192412019212899e-06,
+      "loss": 0.2733,
+      "step": 12772
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 3.5393102169036865,
+      "learning_rate": 9.192268338483215e-06,
+      "loss": 0.658,
+      "step": 12773
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 3.2802369594573975,
+      "learning_rate": 9.192124646096404e-06,
+      "loss": 0.5831,
+      "step": 12774
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.0495173931121826,
+      "learning_rate": 9.191980942052862e-06,
+      "loss": 0.5566,
+      "step": 12775
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.9895907640457153,
+      "learning_rate": 9.191837226352994e-06,
+      "loss": 0.6963,
+      "step": 12776
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.000457763671875,
+      "learning_rate": 9.191693498997197e-06,
+      "loss": 0.6694,
+      "step": 12777
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.9996378421783447,
+      "learning_rate": 9.191549759985869e-06,
+      "loss": 0.5024,
+      "step": 12778
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 3.247129440307617,
+      "learning_rate": 9.191406009319413e-06,
+      "loss": 0.5234,
+      "step": 12779
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.473780632019043,
+      "learning_rate": 9.191262246998227e-06,
+      "loss": 0.6452,
+      "step": 12780
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.441820740699768,
+      "learning_rate": 9.191118473022708e-06,
+      "loss": 0.7683,
+      "step": 12781
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 3.1586368083953857,
+      "learning_rate": 9.190974687393262e-06,
+      "loss": 0.5599,
+      "step": 12782
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.041145086288452,
+      "learning_rate": 9.190830890110286e-06,
+      "loss": 0.6381,
+      "step": 12783
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.8613951206207275,
+      "learning_rate": 9.190687081174175e-06,
+      "loss": 0.5856,
+      "step": 12784
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.0327978134155273,
+      "learning_rate": 9.190543260585336e-06,
+      "loss": 0.6799,
+      "step": 12785
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.7980778217315674,
+      "learning_rate": 9.190399428344165e-06,
+      "loss": 0.4622,
+      "step": 12786
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.8249951601028442,
+      "learning_rate": 9.190255584451064e-06,
+      "loss": 0.3945,
+      "step": 12787
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.5132510662078857,
+      "learning_rate": 9.19011172890643e-06,
+      "loss": 0.6678,
+      "step": 12788
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.751981496810913,
+      "learning_rate": 9.189967861710668e-06,
+      "loss": 0.7127,
+      "step": 12789
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.0935404300689697,
+      "learning_rate": 9.189823982864171e-06,
+      "loss": 0.485,
+      "step": 12790
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 3.209939479827881,
+      "learning_rate": 9.189680092367345e-06,
+      "loss": 0.6695,
+      "step": 12791
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 3.3368492126464844,
+      "learning_rate": 9.189536190220589e-06,
+      "loss": 0.7037,
+      "step": 12792
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.7951425313949585,
+      "learning_rate": 9.189392276424303e-06,
+      "loss": 0.78,
+      "step": 12793
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.3649890422821045,
+      "learning_rate": 9.189248350978884e-06,
+      "loss": 0.7423,
+      "step": 12794
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 3.145507335662842,
+      "learning_rate": 9.189104413884736e-06,
+      "loss": 0.4884,
+      "step": 12795
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.8091524839401245,
+      "learning_rate": 9.188960465142258e-06,
+      "loss": 0.6035,
+      "step": 12796
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.6658921241760254,
+      "learning_rate": 9.188816504751849e-06,
+      "loss": 0.7194,
+      "step": 12797
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.052260637283325,
+      "learning_rate": 9.188672532713912e-06,
+      "loss": 0.6193,
+      "step": 12798
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.8153417110443115,
+      "learning_rate": 9.188528549028845e-06,
+      "loss": 0.4575,
+      "step": 12799
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 3.3302199840545654,
+      "learning_rate": 9.18838455369705e-06,
+      "loss": 1.0611,
+      "step": 12800
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.1532938480377197,
+      "learning_rate": 9.188240546718926e-06,
+      "loss": 0.6121,
+      "step": 12801
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.1826300621032715,
+      "learning_rate": 9.188096528094873e-06,
+      "loss": 0.6631,
+      "step": 12802
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.2624679803848267,
+      "learning_rate": 9.187952497825293e-06,
+      "loss": 0.5793,
+      "step": 12803
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 3.040531873703003,
+      "learning_rate": 9.187808455910587e-06,
+      "loss": 0.4784,
+      "step": 12804
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.461051106452942,
+      "learning_rate": 9.187664402351153e-06,
+      "loss": 0.6661,
+      "step": 12805
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.3612325191497803,
+      "learning_rate": 9.187520337147393e-06,
+      "loss": 0.539,
+      "step": 12806
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.4328701496124268,
+      "learning_rate": 9.18737626029971e-06,
+      "loss": 0.813,
+      "step": 12807
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.7566299438476562,
+      "learning_rate": 9.1872321718085e-06,
+      "loss": 0.8732,
+      "step": 12808
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 4.327056407928467,
+      "learning_rate": 9.187088071674166e-06,
+      "loss": 0.5942,
+      "step": 12809
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.8756656646728516,
+      "learning_rate": 9.18694395989711e-06,
+      "loss": 0.6475,
+      "step": 12810
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.211698055267334,
+      "learning_rate": 9.18679983647773e-06,
+      "loss": 0.5988,
+      "step": 12811
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.9935016632080078,
+      "learning_rate": 9.186655701416428e-06,
+      "loss": 0.5676,
+      "step": 12812
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.067072868347168,
+      "learning_rate": 9.186511554713606e-06,
+      "loss": 0.548,
+      "step": 12813
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.2721312046051025,
+      "learning_rate": 9.186367396369663e-06,
+      "loss": 0.643,
+      "step": 12814
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.8395259380340576,
+      "learning_rate": 9.186223226385e-06,
+      "loss": 0.7716,
+      "step": 12815
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 4.524062156677246,
+      "learning_rate": 9.186079044760019e-06,
+      "loss": 0.747,
+      "step": 12816
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 3.2226192951202393,
+      "learning_rate": 9.18593485149512e-06,
+      "loss": 0.4396,
+      "step": 12817
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 3.448179244995117,
+      "learning_rate": 9.185790646590705e-06,
+      "loss": 0.6195,
+      "step": 12818
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.8231674432754517,
+      "learning_rate": 9.185646430047174e-06,
+      "loss": 0.7022,
+      "step": 12819
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.054290533065796,
+      "learning_rate": 9.185502201864927e-06,
+      "loss": 0.5796,
+      "step": 12820
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 3.1962685585021973,
+      "learning_rate": 9.185357962044368e-06,
+      "loss": 0.7459,
+      "step": 12821
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 3.163135528564453,
+      "learning_rate": 9.185213710585896e-06,
+      "loss": 0.6801,
+      "step": 12822
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.5613434314727783,
+      "learning_rate": 9.185069447489911e-06,
+      "loss": 0.6298,
+      "step": 12823
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.5815303325653076,
+      "learning_rate": 9.184925172756817e-06,
+      "loss": 0.8789,
+      "step": 12824
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.1794865131378174,
+      "learning_rate": 9.184780886387014e-06,
+      "loss": 0.8024,
+      "step": 12825
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.5365419387817383,
+      "learning_rate": 9.184636588380902e-06,
+      "loss": 0.534,
+      "step": 12826
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 3.7436938285827637,
+      "learning_rate": 9.184492278738883e-06,
+      "loss": 0.6256,
+      "step": 12827
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.8495410680770874,
+      "learning_rate": 9.18434795746136e-06,
+      "loss": 0.6144,
+      "step": 12828
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.864957332611084,
+      "learning_rate": 9.184203624548733e-06,
+      "loss": 0.5043,
+      "step": 12829
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.9034457206726074,
+      "learning_rate": 9.184059280001403e-06,
+      "loss": 0.6445,
+      "step": 12830
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.9189304113388062,
+      "learning_rate": 9.18391492381977e-06,
+      "loss": 0.6797,
+      "step": 12831
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 4.525580883026123,
+      "learning_rate": 9.183770556004237e-06,
+      "loss": 0.6694,
+      "step": 12832
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.381004810333252,
+      "learning_rate": 9.183626176555208e-06,
+      "loss": 0.3722,
+      "step": 12833
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.8888839483261108,
+      "learning_rate": 9.183481785473079e-06,
+      "loss": 0.8602,
+      "step": 12834
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 3.4682037830352783,
+      "learning_rate": 9.183337382758255e-06,
+      "loss": 0.8534,
+      "step": 12835
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.765556812286377,
+      "learning_rate": 9.183192968411137e-06,
+      "loss": 0.7468,
+      "step": 12836
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.5041468143463135,
+      "learning_rate": 9.183048542432127e-06,
+      "loss": 0.4219,
+      "step": 12837
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 3.171942710876465,
+      "learning_rate": 9.182904104821626e-06,
+      "loss": 0.7115,
+      "step": 12838
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.6499056816101074,
+      "learning_rate": 9.182759655580034e-06,
+      "loss": 0.671,
+      "step": 12839
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 3.277135133743286,
+      "learning_rate": 9.182615194707755e-06,
+      "loss": 0.5758,
+      "step": 12840
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.9766954183578491,
+      "learning_rate": 9.182470722205192e-06,
+      "loss": 0.6128,
+      "step": 12841
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.8605804443359375,
+      "learning_rate": 9.182326238072743e-06,
+      "loss": 0.8233,
+      "step": 12842
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.1431705951690674,
+      "learning_rate": 9.182181742310811e-06,
+      "loss": 0.5141,
+      "step": 12843
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.2250022888183594,
+      "learning_rate": 9.182037234919801e-06,
+      "loss": 0.5722,
+      "step": 12844
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 4.4847025871276855,
+      "learning_rate": 9.18189271590011e-06,
+      "loss": 0.87,
+      "step": 12845
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.5674314498901367,
+      "learning_rate": 9.18174818525214e-06,
+      "loss": 0.8465,
+      "step": 12846
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 4.280275821685791,
+      "learning_rate": 9.181603642976298e-06,
+      "loss": 0.7708,
+      "step": 12847
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 4.111727237701416,
+      "learning_rate": 9.181459089072981e-06,
+      "loss": 0.7361,
+      "step": 12848
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.5278284549713135,
+      "learning_rate": 9.181314523542592e-06,
+      "loss": 0.4214,
+      "step": 12849
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 3.895998001098633,
+      "learning_rate": 9.181169946385536e-06,
+      "loss": 0.5644,
+      "step": 12850
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.9177452325820923,
+      "learning_rate": 9.181025357602212e-06,
+      "loss": 0.8122,
+      "step": 12851
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.8515524864196777,
+      "learning_rate": 9.180880757193022e-06,
+      "loss": 0.6763,
+      "step": 12852
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.1734418869018555,
+      "learning_rate": 9.180736145158369e-06,
+      "loss": 0.7842,
+      "step": 12853
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.4229981899261475,
+      "learning_rate": 9.180591521498653e-06,
+      "loss": 0.7836,
+      "step": 12854
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.373715877532959,
+      "learning_rate": 9.180446886214283e-06,
+      "loss": 0.5201,
+      "step": 12855
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.5177505016326904,
+      "learning_rate": 9.180302239305652e-06,
+      "loss": 0.5045,
+      "step": 12856
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.920541763305664,
+      "learning_rate": 9.180157580773168e-06,
+      "loss": 0.6339,
+      "step": 12857
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.9386043548583984,
+      "learning_rate": 9.18001291061723e-06,
+      "loss": 0.5998,
+      "step": 12858
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.2031004428863525,
+      "learning_rate": 9.179868228838245e-06,
+      "loss": 0.8395,
+      "step": 12859
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.624161958694458,
+      "learning_rate": 9.179723535436611e-06,
+      "loss": 0.7885,
+      "step": 12860
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.642074465751648,
+      "learning_rate": 9.17957883041273e-06,
+      "loss": 0.6553,
+      "step": 12861
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.9019657373428345,
+      "learning_rate": 9.179434113767008e-06,
+      "loss": 0.7191,
+      "step": 12862
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.031895160675049,
+      "learning_rate": 9.179289385499845e-06,
+      "loss": 0.3977,
+      "step": 12863
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.3548483848571777,
+      "learning_rate": 9.179144645611643e-06,
+      "loss": 0.9144,
+      "step": 12864
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 4.677771091461182,
+      "learning_rate": 9.178999894102806e-06,
+      "loss": 0.7143,
+      "step": 12865
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 3.438237190246582,
+      "learning_rate": 9.178855130973736e-06,
+      "loss": 0.6147,
+      "step": 12866
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.8055171966552734,
+      "learning_rate": 9.178710356224834e-06,
+      "loss": 0.6864,
+      "step": 12867
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.3141679763793945,
+      "learning_rate": 9.178565569856504e-06,
+      "loss": 0.6062,
+      "step": 12868
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.8571078777313232,
+      "learning_rate": 9.178420771869149e-06,
+      "loss": 0.5772,
+      "step": 12869
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.8560945987701416,
+      "learning_rate": 9.178275962263172e-06,
+      "loss": 0.66,
+      "step": 12870
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.945866346359253,
+      "learning_rate": 9.178131141038976e-06,
+      "loss": 0.6665,
+      "step": 12871
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.692792534828186,
+      "learning_rate": 9.177986308196959e-06,
+      "loss": 0.6249,
+      "step": 12872
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.6351590156555176,
+      "learning_rate": 9.177841463737529e-06,
+      "loss": 0.8054,
+      "step": 12873
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.9022804498672485,
+      "learning_rate": 9.177696607661087e-06,
+      "loss": 0.7041,
+      "step": 12874
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.940618634223938,
+      "learning_rate": 9.177551739968035e-06,
+      "loss": 0.6592,
+      "step": 12875
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.3933796882629395,
+      "learning_rate": 9.177406860658778e-06,
+      "loss": 0.5409,
+      "step": 12876
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.239866256713867,
+      "learning_rate": 9.177261969733716e-06,
+      "loss": 0.7269,
+      "step": 12877
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 3.247511863708496,
+      "learning_rate": 9.177117067193252e-06,
+      "loss": 0.7915,
+      "step": 12878
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 5.321187496185303,
+      "learning_rate": 9.176972153037793e-06,
+      "loss": 0.643,
+      "step": 12879
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.700685977935791,
+      "learning_rate": 9.176827227267738e-06,
+      "loss": 0.2964,
+      "step": 12880
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.792163372039795,
+      "learning_rate": 9.176682289883491e-06,
+      "loss": 0.4807,
+      "step": 12881
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 3.389634132385254,
+      "learning_rate": 9.176537340885456e-06,
+      "loss": 0.7767,
+      "step": 12882
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.196011543273926,
+      "learning_rate": 9.176392380274033e-06,
+      "loss": 0.7324,
+      "step": 12883
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.904395580291748,
+      "learning_rate": 9.17624740804963e-06,
+      "loss": 0.5271,
+      "step": 12884
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.9042983055114746,
+      "learning_rate": 9.176102424212646e-06,
+      "loss": 0.4691,
+      "step": 12885
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 5.543762683868408,
+      "learning_rate": 9.175957428763486e-06,
+      "loss": 0.8895,
+      "step": 12886
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.207737445831299,
+      "learning_rate": 9.175812421702553e-06,
+      "loss": 0.6165,
+      "step": 12887
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 5.407472610473633,
+      "learning_rate": 9.175667403030249e-06,
+      "loss": 0.7338,
+      "step": 12888
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.948871374130249,
+      "learning_rate": 9.17552237274698e-06,
+      "loss": 0.6005,
+      "step": 12889
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.8108572959899902,
+      "learning_rate": 9.175377330853145e-06,
+      "loss": 0.8366,
+      "step": 12890
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 4.786581993103027,
+      "learning_rate": 9.17523227734915e-06,
+      "loss": 1.1485,
+      "step": 12891
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.5026962757110596,
+      "learning_rate": 9.1750872122354e-06,
+      "loss": 0.7403,
+      "step": 12892
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.728378415107727,
+      "learning_rate": 9.174942135512294e-06,
+      "loss": 0.6879,
+      "step": 12893
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.421949625015259,
+      "learning_rate": 9.174797047180242e-06,
+      "loss": 0.7509,
+      "step": 12894
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.3063912391662598,
+      "learning_rate": 9.17465194723964e-06,
+      "loss": 0.54,
+      "step": 12895
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.189406633377075,
+      "learning_rate": 9.174506835690895e-06,
+      "loss": 0.841,
+      "step": 12896
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.6941406726837158,
+      "learning_rate": 9.174361712534412e-06,
+      "loss": 0.4823,
+      "step": 12897
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.392876386642456,
+      "learning_rate": 9.174216577770592e-06,
+      "loss": 0.6973,
+      "step": 12898
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.1412441730499268,
+      "learning_rate": 9.174071431399838e-06,
+      "loss": 0.5205,
+      "step": 12899
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.8489298820495605,
+      "learning_rate": 9.173926273422557e-06,
+      "loss": 0.4211,
+      "step": 12900
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 5.17540168762207,
+      "learning_rate": 9.17378110383915e-06,
+      "loss": 0.7152,
+      "step": 12901
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.6750539541244507,
+      "learning_rate": 9.17363592265002e-06,
+      "loss": 0.6122,
+      "step": 12902
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.2886970043182373,
+      "learning_rate": 9.173490729855574e-06,
+      "loss": 0.8518,
+      "step": 12903
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 3.5985546112060547,
+      "learning_rate": 9.173345525456212e-06,
+      "loss": 0.6154,
+      "step": 12904
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.571071982383728,
+      "learning_rate": 9.173200309452341e-06,
+      "loss": 0.6651,
+      "step": 12905
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.7650220394134521,
+      "learning_rate": 9.173055081844363e-06,
+      "loss": 0.9537,
+      "step": 12906
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.1103572845458984,
+      "learning_rate": 9.17290984263268e-06,
+      "loss": 0.5554,
+      "step": 12907
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.6705331802368164,
+      "learning_rate": 9.1727645918177e-06,
+      "loss": 0.4686,
+      "step": 12908
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.6467320919036865,
+      "learning_rate": 9.172619329399823e-06,
+      "loss": 0.7297,
+      "step": 12909
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.9709092378616333,
+      "learning_rate": 9.172474055379457e-06,
+      "loss": 0.5206,
+      "step": 12910
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.0431952476501465,
+      "learning_rate": 9.172328769757003e-06,
+      "loss": 0.5038,
+      "step": 12911
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.0678372383117676,
+      "learning_rate": 9.172183472532864e-06,
+      "loss": 0.6042,
+      "step": 12912
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.641141653060913,
+      "learning_rate": 9.172038163707449e-06,
+      "loss": 0.53,
+      "step": 12913
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 3.8890607357025146,
+      "learning_rate": 9.171892843281156e-06,
+      "loss": 0.7772,
+      "step": 12914
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.9697895050048828,
+      "learning_rate": 9.171747511254394e-06,
+      "loss": 0.508,
+      "step": 12915
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 5.1777849197387695,
+      "learning_rate": 9.171602167627562e-06,
+      "loss": 0.6198,
+      "step": 12916
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.7191095352172852,
+      "learning_rate": 9.17145681240107e-06,
+      "loss": 0.4726,
+      "step": 12917
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 5.55944299697876,
+      "learning_rate": 9.171311445575317e-06,
+      "loss": 0.9755,
+      "step": 12918
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.2215332984924316,
+      "learning_rate": 9.17116606715071e-06,
+      "loss": 0.4087,
+      "step": 12919
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.2204346656799316,
+      "learning_rate": 9.171020677127654e-06,
+      "loss": 0.5067,
+      "step": 12920
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.616426467895508,
+      "learning_rate": 9.170875275506552e-06,
+      "loss": 0.669,
+      "step": 12921
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 3.5495853424072266,
+      "learning_rate": 9.170729862287807e-06,
+      "loss": 0.6625,
+      "step": 12922
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.446331024169922,
+      "learning_rate": 9.170584437471824e-06,
+      "loss": 0.8092,
+      "step": 12923
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.3648338317871094,
+      "learning_rate": 9.17043900105901e-06,
+      "loss": 0.4897,
+      "step": 12924
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.0606257915496826,
+      "learning_rate": 9.170293553049766e-06,
+      "loss": 0.5581,
+      "step": 12925
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.982696294784546,
+      "learning_rate": 9.170148093444499e-06,
+      "loss": 0.6545,
+      "step": 12926
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.7998712062835693,
+      "learning_rate": 9.170002622243612e-06,
+      "loss": 0.6508,
+      "step": 12927
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.632218599319458,
+      "learning_rate": 9.16985713944751e-06,
+      "loss": 0.8712,
+      "step": 12928
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.0953627824783325,
+      "learning_rate": 9.169711645056596e-06,
+      "loss": 0.6383,
+      "step": 12929
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 3.4415173530578613,
+      "learning_rate": 9.169566139071276e-06,
+      "loss": 0.8322,
+      "step": 12930
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.736254334449768,
+      "learning_rate": 9.169420621491955e-06,
+      "loss": 0.5221,
+      "step": 12931
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.150672435760498,
+      "learning_rate": 9.16927509231904e-06,
+      "loss": 0.6575,
+      "step": 12932
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.1008293628692627,
+      "learning_rate": 9.169129551552928e-06,
+      "loss": 0.7369,
+      "step": 12933
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.7155669927597046,
+      "learning_rate": 9.168983999194032e-06,
+      "loss": 0.6773,
+      "step": 12934
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.8257884979248047,
+      "learning_rate": 9.168838435242752e-06,
+      "loss": 0.4918,
+      "step": 12935
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.569242238998413,
+      "learning_rate": 9.168692859699494e-06,
+      "loss": 0.7209,
+      "step": 12936
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.7916110754013062,
+      "learning_rate": 9.168547272564663e-06,
+      "loss": 0.9191,
+      "step": 12937
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 3.5409531593322754,
+      "learning_rate": 9.168401673838663e-06,
+      "loss": 0.677,
+      "step": 12938
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.5618419647216797,
+      "learning_rate": 9.1682560635219e-06,
+      "loss": 0.4441,
+      "step": 12939
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 3.780400037765503,
+      "learning_rate": 9.16811044161478e-06,
+      "loss": 0.734,
+      "step": 12940
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.8700486421585083,
+      "learning_rate": 9.167964808117703e-06,
+      "loss": 0.6482,
+      "step": 12941
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.588095188140869,
+      "learning_rate": 9.16781916303108e-06,
+      "loss": 0.4331,
+      "step": 12942
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.063659191131592,
+      "learning_rate": 9.16767350635531e-06,
+      "loss": 0.528,
+      "step": 12943
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.3790526390075684,
+      "learning_rate": 9.167527838090804e-06,
+      "loss": 0.5054,
+      "step": 12944
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.496967315673828,
+      "learning_rate": 9.167382158237962e-06,
+      "loss": 0.7302,
+      "step": 12945
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.836822509765625,
+      "learning_rate": 9.167236466797194e-06,
+      "loss": 0.6651,
+      "step": 12946
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.8738694190979004,
+      "learning_rate": 9.167090763768902e-06,
+      "loss": 0.652,
+      "step": 12947
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.0228981971740723,
+      "learning_rate": 9.166945049153491e-06,
+      "loss": 0.7076,
+      "step": 12948
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.7024261951446533,
+      "learning_rate": 9.166799322951366e-06,
+      "loss": 0.6816,
+      "step": 12949
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.294809579849243,
+      "learning_rate": 9.166653585162935e-06,
+      "loss": 0.9254,
+      "step": 12950
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.4977805614471436,
+      "learning_rate": 9.166507835788599e-06,
+      "loss": 0.5538,
+      "step": 12951
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.5585746765136719,
+      "learning_rate": 9.166362074828768e-06,
+      "loss": 0.7389,
+      "step": 12952
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.914142608642578,
+      "learning_rate": 9.166216302283842e-06,
+      "loss": 0.9098,
+      "step": 12953
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.573789358139038,
+      "learning_rate": 9.166070518154231e-06,
+      "loss": 0.5793,
+      "step": 12954
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.7553069591522217,
+      "learning_rate": 9.165924722440339e-06,
+      "loss": 0.7278,
+      "step": 12955
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.6450035572052,
+      "learning_rate": 9.16577891514257e-06,
+      "loss": 0.7788,
+      "step": 12956
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.5955560207366943,
+      "learning_rate": 9.165633096261328e-06,
+      "loss": 0.5678,
+      "step": 12957
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.4622621536254883,
+      "learning_rate": 9.165487265797025e-06,
+      "loss": 0.7642,
+      "step": 12958
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.4330060482025146,
+      "learning_rate": 9.165341423750059e-06,
+      "loss": 0.7161,
+      "step": 12959
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.0196261405944824,
+      "learning_rate": 9.165195570120841e-06,
+      "loss": 0.5976,
+      "step": 12960
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.8629318475723267,
+      "learning_rate": 9.165049704909774e-06,
+      "loss": 0.6634,
+      "step": 12961
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.8485214710235596,
+      "learning_rate": 9.164903828117264e-06,
+      "loss": 0.6466,
+      "step": 12962
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.305119514465332,
+      "learning_rate": 9.164757939743715e-06,
+      "loss": 0.6945,
+      "step": 12963
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.7022414207458496,
+      "learning_rate": 9.164612039789537e-06,
+      "loss": 0.5167,
+      "step": 12964
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.284802198410034,
+      "learning_rate": 9.164466128255132e-06,
+      "loss": 0.6362,
+      "step": 12965
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.8074376583099365,
+      "learning_rate": 9.164320205140906e-06,
+      "loss": 0.7585,
+      "step": 12966
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.355074405670166,
+      "learning_rate": 9.164174270447266e-06,
+      "loss": 0.6247,
+      "step": 12967
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.9954054355621338,
+      "learning_rate": 9.164028324174617e-06,
+      "loss": 0.3891,
+      "step": 12968
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.677276849746704,
+      "learning_rate": 9.163882366323367e-06,
+      "loss": 0.6911,
+      "step": 12969
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.5305027961730957,
+      "learning_rate": 9.163736396893918e-06,
+      "loss": 0.6414,
+      "step": 12970
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.2412970066070557,
+      "learning_rate": 9.163590415886678e-06,
+      "loss": 0.6607,
+      "step": 12971
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.041090250015259,
+      "learning_rate": 9.163444423302052e-06,
+      "loss": 0.6737,
+      "step": 12972
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.7214024066925049,
+      "learning_rate": 9.16329841914045e-06,
+      "loss": 0.6417,
+      "step": 12973
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.787792205810547,
+      "learning_rate": 9.163152403402271e-06,
+      "loss": 0.6104,
+      "step": 12974
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 3.167271375656128,
+      "learning_rate": 9.163006376087926e-06,
+      "loss": 0.7158,
+      "step": 12975
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.609437346458435,
+      "learning_rate": 9.162860337197822e-06,
+      "loss": 0.6099,
+      "step": 12976
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.75380539894104,
+      "learning_rate": 9.16271428673236e-06,
+      "loss": 0.5377,
+      "step": 12977
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.1947109699249268,
+      "learning_rate": 9.162568224691949e-06,
+      "loss": 0.6436,
+      "step": 12978
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.8603813648223877,
+      "learning_rate": 9.162422151076995e-06,
+      "loss": 0.617,
+      "step": 12979
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 3.2543065547943115,
+      "learning_rate": 9.162276065887906e-06,
+      "loss": 0.7274,
+      "step": 12980
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 3.8545353412628174,
+      "learning_rate": 9.162129969125087e-06,
+      "loss": 0.6595,
+      "step": 12981
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.8849714994430542,
+      "learning_rate": 9.161983860788942e-06,
+      "loss": 0.5937,
+      "step": 12982
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.9724304676055908,
+      "learning_rate": 9.161837740879879e-06,
+      "loss": 0.7848,
+      "step": 12983
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.3849077224731445,
+      "learning_rate": 9.161691609398304e-06,
+      "loss": 0.5221,
+      "step": 12984
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 3.3637564182281494,
+      "learning_rate": 9.161545466344625e-06,
+      "loss": 0.6093,
+      "step": 12985
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.7858467102050781,
+      "learning_rate": 9.161399311719245e-06,
+      "loss": 0.6056,
+      "step": 12986
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 3.2142040729522705,
+      "learning_rate": 9.161253145522576e-06,
+      "loss": 0.7663,
+      "step": 12987
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.2539548873901367,
+      "learning_rate": 9.161106967755018e-06,
+      "loss": 0.7991,
+      "step": 12988
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 4.686570167541504,
+      "learning_rate": 9.160960778416981e-06,
+      "loss": 1.1548,
+      "step": 12989
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.73789644241333,
+      "learning_rate": 9.16081457750887e-06,
+      "loss": 0.5523,
+      "step": 12990
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.1611688137054443,
+      "learning_rate": 9.160668365031094e-06,
+      "loss": 0.8267,
+      "step": 12991
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 3.1466052532196045,
+      "learning_rate": 9.160522140984057e-06,
+      "loss": 0.6774,
+      "step": 12992
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.1778275966644287,
+      "learning_rate": 9.160375905368167e-06,
+      "loss": 0.4559,
+      "step": 12993
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.8984692096710205,
+      "learning_rate": 9.160229658183828e-06,
+      "loss": 0.5551,
+      "step": 12994
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.5531246662139893,
+      "learning_rate": 9.16008339943145e-06,
+      "loss": 0.5418,
+      "step": 12995
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.5816376209259033,
+      "learning_rate": 9.15993712911144e-06,
+      "loss": 0.7267,
+      "step": 12996
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 3.787977695465088,
+      "learning_rate": 9.159790847224201e-06,
+      "loss": 0.7032,
+      "step": 12997
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 3.9905178546905518,
+      "learning_rate": 9.159644553770144e-06,
+      "loss": 0.6975,
+      "step": 12998
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 3.1747796535491943,
+      "learning_rate": 9.159498248749672e-06,
+      "loss": 0.5893,
+      "step": 12999
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 3.34295392036438,
+      "learning_rate": 9.159351932163194e-06,
+      "loss": 0.7225,
+      "step": 13000
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.60935378074646,
+      "learning_rate": 9.159205604011116e-06,
+      "loss": 0.5391,
+      "step": 13001
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 3.2090647220611572,
+      "learning_rate": 9.159059264293846e-06,
+      "loss": 0.9752,
+      "step": 13002
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.1667919158935547,
+      "learning_rate": 9.15891291301179e-06,
+      "loss": 0.4915,
+      "step": 13003
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 3.4092562198638916,
+      "learning_rate": 9.158766550165356e-06,
+      "loss": 0.4495,
+      "step": 13004
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.7842961549758911,
+      "learning_rate": 9.15862017575495e-06,
+      "loss": 0.8174,
+      "step": 13005
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.8180489540100098,
+      "learning_rate": 9.158473789780977e-06,
+      "loss": 0.6908,
+      "step": 13006
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.3462374210357666,
+      "learning_rate": 9.158327392243848e-06,
+      "loss": 0.5148,
+      "step": 13007
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 3.2761800289154053,
+      "learning_rate": 9.158180983143968e-06,
+      "loss": 0.6574,
+      "step": 13008
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.6556084156036377,
+      "learning_rate": 9.158034562481745e-06,
+      "loss": 0.5473,
+      "step": 13009
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.5855369567871094,
+      "learning_rate": 9.157888130257585e-06,
+      "loss": 0.9603,
+      "step": 13010
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.268481492996216,
+      "learning_rate": 9.157741686471895e-06,
+      "loss": 0.7113,
+      "step": 13011
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.5809552669525146,
+      "learning_rate": 9.157595231125084e-06,
+      "loss": 0.6976,
+      "step": 13012
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 4.239136695861816,
+      "learning_rate": 9.157448764217558e-06,
+      "loss": 0.7272,
+      "step": 13013
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.9842618703842163,
+      "learning_rate": 9.157302285749723e-06,
+      "loss": 0.4838,
+      "step": 13014
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.740856647491455,
+      "learning_rate": 9.15715579572199e-06,
+      "loss": 0.6703,
+      "step": 13015
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 3.4907476902008057,
+      "learning_rate": 9.157009294134763e-06,
+      "loss": 0.5127,
+      "step": 13016
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 4.782325744628906,
+      "learning_rate": 9.15686278098845e-06,
+      "loss": 0.842,
+      "step": 13017
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.010399103164673,
+      "learning_rate": 9.156716256283457e-06,
+      "loss": 0.5899,
+      "step": 13018
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.091212272644043,
+      "learning_rate": 9.156569720020196e-06,
+      "loss": 0.6881,
+      "step": 13019
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.9761590957641602,
+      "learning_rate": 9.156423172199071e-06,
+      "loss": 0.7135,
+      "step": 13020
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.8584929704666138,
+      "learning_rate": 9.15627661282049e-06,
+      "loss": 0.8692,
+      "step": 13021
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.588998317718506,
+      "learning_rate": 9.15613004188486e-06,
+      "loss": 0.6443,
+      "step": 13022
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 3.8955936431884766,
+      "learning_rate": 9.155983459392592e-06,
+      "loss": 0.996,
+      "step": 13023
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.9833558797836304,
+      "learning_rate": 9.155836865344088e-06,
+      "loss": 0.8308,
+      "step": 13024
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.0317275524139404,
+      "learning_rate": 9.15569025973976e-06,
+      "loss": 0.6154,
+      "step": 13025
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.4682029485702515,
+      "learning_rate": 9.155543642580014e-06,
+      "loss": 0.5008,
+      "step": 13026
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.6824676990509033,
+      "learning_rate": 9.155397013865257e-06,
+      "loss": 0.8146,
+      "step": 13027
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.433620572090149,
+      "learning_rate": 9.1552503735959e-06,
+      "loss": 0.6242,
+      "step": 13028
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.927640199661255,
+      "learning_rate": 9.155103721772344e-06,
+      "loss": 0.9162,
+      "step": 13029
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.640148878097534,
+      "learning_rate": 9.154957058395004e-06,
+      "loss": 0.6236,
+      "step": 13030
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 3.1762192249298096,
+      "learning_rate": 9.154810383464283e-06,
+      "loss": 0.6785,
+      "step": 13031
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.156294822692871,
+      "learning_rate": 9.154663696980592e-06,
+      "loss": 0.7318,
+      "step": 13032
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 6.876437187194824,
+      "learning_rate": 9.154516998944338e-06,
+      "loss": 0.7599,
+      "step": 13033
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.5460543632507324,
+      "learning_rate": 9.15437028935593e-06,
+      "loss": 0.6478,
+      "step": 13034
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.6543217897415161,
+      "learning_rate": 9.154223568215772e-06,
+      "loss": 0.96,
+      "step": 13035
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.2935945987701416,
+      "learning_rate": 9.154076835524275e-06,
+      "loss": 0.7677,
+      "step": 13036
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.666074275970459,
+      "learning_rate": 9.153930091281847e-06,
+      "loss": 0.5398,
+      "step": 13037
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 3.028285264968872,
+      "learning_rate": 9.153783335488895e-06,
+      "loss": 0.8333,
+      "step": 13038
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 3.4485037326812744,
+      "learning_rate": 9.153636568145826e-06,
+      "loss": 0.7066,
+      "step": 13039
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.8774844408035278,
+      "learning_rate": 9.153489789253052e-06,
+      "loss": 0.4022,
+      "step": 13040
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 3.903658390045166,
+      "learning_rate": 9.15334299881098e-06,
+      "loss": 0.6526,
+      "step": 13041
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.132977247238159,
+      "learning_rate": 9.153196196820013e-06,
+      "loss": 0.4775,
+      "step": 13042
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.3984036445617676,
+      "learning_rate": 9.153049383280565e-06,
+      "loss": 0.5316,
+      "step": 13043
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 4.5248870849609375,
+      "learning_rate": 9.152902558193044e-06,
+      "loss": 0.6956,
+      "step": 13044
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.916656732559204,
+      "learning_rate": 9.152755721557856e-06,
+      "loss": 0.6978,
+      "step": 13045
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.3951001167297363,
+      "learning_rate": 9.152608873375409e-06,
+      "loss": 0.6426,
+      "step": 13046
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 3.2585718631744385,
+      "learning_rate": 9.152462013646113e-06,
+      "loss": 0.5109,
+      "step": 13047
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 3.7724406719207764,
+      "learning_rate": 9.152315142370377e-06,
+      "loss": 0.922,
+      "step": 13048
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.9215984344482422,
+      "learning_rate": 9.152168259548606e-06,
+      "loss": 0.7676,
+      "step": 13049
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.7141611576080322,
+      "learning_rate": 9.15202136518121e-06,
+      "loss": 0.3863,
+      "step": 13050
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.5959551334381104,
+      "learning_rate": 9.1518744592686e-06,
+      "loss": 0.3156,
+      "step": 13051
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 3.2547271251678467,
+      "learning_rate": 9.151727541811182e-06,
+      "loss": 0.5805,
+      "step": 13052
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.676185131072998,
+      "learning_rate": 9.151580612809363e-06,
+      "loss": 0.6838,
+      "step": 13053
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 4.301938533782959,
+      "learning_rate": 9.151433672263557e-06,
+      "loss": 0.9386,
+      "step": 13054
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 4.326162815093994,
+      "learning_rate": 9.151286720174166e-06,
+      "loss": 0.9134,
+      "step": 13055
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.939131021499634,
+      "learning_rate": 9.151139756541603e-06,
+      "loss": 0.9073,
+      "step": 13056
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 3.361299514770508,
+      "learning_rate": 9.150992781366275e-06,
+      "loss": 0.8042,
+      "step": 13057
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.7616865634918213,
+      "learning_rate": 9.150845794648591e-06,
+      "loss": 0.6773,
+      "step": 13058
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.948369264602661,
+      "learning_rate": 9.15069879638896e-06,
+      "loss": 0.527,
+      "step": 13059
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.6075525283813477,
+      "learning_rate": 9.150551786587791e-06,
+      "loss": 0.612,
+      "step": 13060
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.272026300430298,
+      "learning_rate": 9.150404765245493e-06,
+      "loss": 0.6703,
+      "step": 13061
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.6506001949310303,
+      "learning_rate": 9.150257732362472e-06,
+      "loss": 0.5636,
+      "step": 13062
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 5.025625228881836,
+      "learning_rate": 9.150110687939141e-06,
+      "loss": 0.7261,
+      "step": 13063
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.9075963497161865,
+      "learning_rate": 9.149963631975906e-06,
+      "loss": 0.5081,
+      "step": 13064
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 3.3113367557525635,
+      "learning_rate": 9.149816564473177e-06,
+      "loss": 0.7807,
+      "step": 13065
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.8013274669647217,
+      "learning_rate": 9.149669485431363e-06,
+      "loss": 0.4998,
+      "step": 13066
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.306272506713867,
+      "learning_rate": 9.14952239485087e-06,
+      "loss": 0.7091,
+      "step": 13067
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.888054847717285,
+      "learning_rate": 9.149375292732114e-06,
+      "loss": 0.8556,
+      "step": 13068
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 5.035773277282715,
+      "learning_rate": 9.149228179075498e-06,
+      "loss": 0.5866,
+      "step": 13069
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 4.983370304107666,
+      "learning_rate": 9.149081053881431e-06,
+      "loss": 0.573,
+      "step": 13070
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 3.1444249153137207,
+      "learning_rate": 9.148933917150325e-06,
+      "loss": 0.5363,
+      "step": 13071
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.7436299324035645,
+      "learning_rate": 9.148786768882589e-06,
+      "loss": 0.7269,
+      "step": 13072
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 3.1048696041107178,
+      "learning_rate": 9.148639609078628e-06,
+      "loss": 0.8729,
+      "step": 13073
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 3.0030806064605713,
+      "learning_rate": 9.148492437738859e-06,
+      "loss": 0.6907,
+      "step": 13074
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 4.625672340393066,
+      "learning_rate": 9.148345254863684e-06,
+      "loss": 0.8927,
+      "step": 13075
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 3.720355987548828,
+      "learning_rate": 9.148198060453515e-06,
+      "loss": 0.9769,
+      "step": 13076
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.4953079223632812,
+      "learning_rate": 9.148050854508761e-06,
+      "loss": 0.7136,
+      "step": 13077
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.759425640106201,
+      "learning_rate": 9.147903637029831e-06,
+      "loss": 0.6987,
+      "step": 13078
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.9089722633361816,
+      "learning_rate": 9.147756408017136e-06,
+      "loss": 0.8172,
+      "step": 13079
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 3.497185707092285,
+      "learning_rate": 9.147609167471084e-06,
+      "loss": 0.7582,
+      "step": 13080
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.361236333847046,
+      "learning_rate": 9.147461915392084e-06,
+      "loss": 0.6454,
+      "step": 13081
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.964324474334717,
+      "learning_rate": 9.147314651780546e-06,
+      "loss": 0.8727,
+      "step": 13082
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.359084367752075,
+      "learning_rate": 9.14716737663688e-06,
+      "loss": 0.7422,
+      "step": 13083
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.647111177444458,
+      "learning_rate": 9.147020089961496e-06,
+      "loss": 0.7238,
+      "step": 13084
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.9816322326660156,
+      "learning_rate": 9.146872791754802e-06,
+      "loss": 0.8036,
+      "step": 13085
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 3.3852550983428955,
+      "learning_rate": 9.146725482017207e-06,
+      "loss": 0.7683,
+      "step": 13086
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 3.526150941848755,
+      "learning_rate": 9.146578160749123e-06,
+      "loss": 0.8091,
+      "step": 13087
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.773756980895996,
+      "learning_rate": 9.146430827950958e-06,
+      "loss": 0.4946,
+      "step": 13088
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.85326886177063,
+      "learning_rate": 9.146283483623124e-06,
+      "loss": 0.7646,
+      "step": 13089
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.511093020439148,
+      "learning_rate": 9.146136127766025e-06,
+      "loss": 0.5897,
+      "step": 13090
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.6534969806671143,
+      "learning_rate": 9.145988760380078e-06,
+      "loss": 0.7674,
+      "step": 13091
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.5202937126159668,
+      "learning_rate": 9.14584138146569e-06,
+      "loss": 1.0536,
+      "step": 13092
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.4084676504135132,
+      "learning_rate": 9.145693991023268e-06,
+      "loss": 0.6031,
+      "step": 13093
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.065840244293213,
+      "learning_rate": 9.145546589053224e-06,
+      "loss": 0.7151,
+      "step": 13094
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 3.3951659202575684,
+      "learning_rate": 9.145399175555969e-06,
+      "loss": 0.6961,
+      "step": 13095
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.0696732997894287,
+      "learning_rate": 9.14525175053191e-06,
+      "loss": 0.5363,
+      "step": 13096
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.9328157901763916,
+      "learning_rate": 9.14510431398146e-06,
+      "loss": 0.4512,
+      "step": 13097
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 3.7819736003875732,
+      "learning_rate": 9.144956865905029e-06,
+      "loss": 1.0753,
+      "step": 13098
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.482684850692749,
+      "learning_rate": 9.144809406303023e-06,
+      "loss": 0.4595,
+      "step": 13099
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.9789987802505493,
+      "learning_rate": 9.144661935175855e-06,
+      "loss": 0.5158,
+      "step": 13100
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.6619112491607666,
+      "learning_rate": 9.144514452523937e-06,
+      "loss": 0.8267,
+      "step": 13101
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.6385722160339355,
+      "learning_rate": 9.144366958347676e-06,
+      "loss": 0.5984,
+      "step": 13102
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.171208143234253,
+      "learning_rate": 9.144219452647483e-06,
+      "loss": 0.782,
+      "step": 13103
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 4.574158191680908,
+      "learning_rate": 9.144071935423768e-06,
+      "loss": 0.4817,
+      "step": 13104
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.3935710191726685,
+      "learning_rate": 9.14392440667694e-06,
+      "loss": 0.4665,
+      "step": 13105
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.6891980171203613,
+      "learning_rate": 9.143776866407413e-06,
+      "loss": 0.6579,
+      "step": 13106
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.799371361732483,
+      "learning_rate": 9.143629314615593e-06,
+      "loss": 0.6614,
+      "step": 13107
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 3.8068437576293945,
+      "learning_rate": 9.143481751301893e-06,
+      "loss": 0.6561,
+      "step": 13108
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 3.7652950286865234,
+      "learning_rate": 9.143334176466722e-06,
+      "loss": 0.5852,
+      "step": 13109
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.9162747859954834,
+      "learning_rate": 9.143186590110491e-06,
+      "loss": 0.4733,
+      "step": 13110
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.781670331954956,
+      "learning_rate": 9.143038992233611e-06,
+      "loss": 0.6968,
+      "step": 13111
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 3.6957502365112305,
+      "learning_rate": 9.14289138283649e-06,
+      "loss": 0.8488,
+      "step": 13112
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.9646986722946167,
+      "learning_rate": 9.14274376191954e-06,
+      "loss": 0.7689,
+      "step": 13113
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.069166660308838,
+      "learning_rate": 9.142596129483172e-06,
+      "loss": 0.3251,
+      "step": 13114
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.1812989711761475,
+      "learning_rate": 9.142448485527797e-06,
+      "loss": 0.8585,
+      "step": 13115
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 3.088437557220459,
+      "learning_rate": 9.142300830053823e-06,
+      "loss": 0.7233,
+      "step": 13116
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 3.402740240097046,
+      "learning_rate": 9.142153163061664e-06,
+      "loss": 0.6595,
+      "step": 13117
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.881204843521118,
+      "learning_rate": 9.142005484551728e-06,
+      "loss": 0.525,
+      "step": 13118
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.6845204830169678,
+      "learning_rate": 9.141857794524425e-06,
+      "loss": 0.6863,
+      "step": 13119
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.7729618549346924,
+      "learning_rate": 9.14171009298017e-06,
+      "loss": 0.6822,
+      "step": 13120
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.812323808670044,
+      "learning_rate": 9.141562379919366e-06,
+      "loss": 0.6162,
+      "step": 13121
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 3.350635290145874,
+      "learning_rate": 9.141414655342434e-06,
+      "loss": 0.8326,
+      "step": 13122
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.4563158750534058,
+      "learning_rate": 9.141266919249776e-06,
+      "loss": 0.6592,
+      "step": 13123
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.149223566055298,
+      "learning_rate": 9.141119171641806e-06,
+      "loss": 0.5522,
+      "step": 13124
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.1119024753570557,
+      "learning_rate": 9.140971412518935e-06,
+      "loss": 0.5935,
+      "step": 13125
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.1839990615844727,
+      "learning_rate": 9.140823641881576e-06,
+      "loss": 0.5777,
+      "step": 13126
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.7880282402038574,
+      "learning_rate": 9.140675859730134e-06,
+      "loss": 0.9283,
+      "step": 13127
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.723858594894409,
+      "learning_rate": 9.140528066065026e-06,
+      "loss": 0.7887,
+      "step": 13128
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.557792901992798,
+      "learning_rate": 9.140380260886661e-06,
+      "loss": 0.7544,
+      "step": 13129
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.6493263244628906,
+      "learning_rate": 9.140232444195448e-06,
+      "loss": 0.6573,
+      "step": 13130
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.9374138116836548,
+      "learning_rate": 9.1400846159918e-06,
+      "loss": 0.785,
+      "step": 13131
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.925751805305481,
+      "learning_rate": 9.139936776276128e-06,
+      "loss": 0.7931,
+      "step": 13132
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.973182201385498,
+      "learning_rate": 9.139788925048843e-06,
+      "loss": 0.9075,
+      "step": 13133
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.2086005210876465,
+      "learning_rate": 9.139641062310354e-06,
+      "loss": 0.5989,
+      "step": 13134
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.883362293243408,
+      "learning_rate": 9.139493188061074e-06,
+      "loss": 0.576,
+      "step": 13135
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 4.213723182678223,
+      "learning_rate": 9.139345302301416e-06,
+      "loss": 0.8716,
+      "step": 13136
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.513972759246826,
+      "learning_rate": 9.13919740503179e-06,
+      "loss": 0.7109,
+      "step": 13137
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 3.888495445251465,
+      "learning_rate": 9.139049496252603e-06,
+      "loss": 0.6799,
+      "step": 13138
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.8132169246673584,
+      "learning_rate": 9.138901575964273e-06,
+      "loss": 0.5948,
+      "step": 13139
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.300004720687866,
+      "learning_rate": 9.138753644167207e-06,
+      "loss": 0.6724,
+      "step": 13140
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 4.876774787902832,
+      "learning_rate": 9.138605700861818e-06,
+      "loss": 0.5832,
+      "step": 13141
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.620971441268921,
+      "learning_rate": 9.138457746048517e-06,
+      "loss": 0.549,
+      "step": 13142
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.3919371366500854,
+      "learning_rate": 9.138309779727714e-06,
+      "loss": 0.3443,
+      "step": 13143
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.0458409786224365,
+      "learning_rate": 9.138161801899823e-06,
+      "loss": 0.4227,
+      "step": 13144
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 3.010042667388916,
+      "learning_rate": 9.138013812565254e-06,
+      "loss": 0.5694,
+      "step": 13145
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.9419398307800293,
+      "learning_rate": 9.13786581172442e-06,
+      "loss": 0.8914,
+      "step": 13146
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 3.1540591716766357,
+      "learning_rate": 9.137717799377729e-06,
+      "loss": 0.759,
+      "step": 13147
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.914467692375183,
+      "learning_rate": 9.137569775525595e-06,
+      "loss": 0.5622,
+      "step": 13148
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.8524353504180908,
+      "learning_rate": 9.13742174016843e-06,
+      "loss": 0.8043,
+      "step": 13149
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 3.154383897781372,
+      "learning_rate": 9.137273693306646e-06,
+      "loss": 0.9496,
+      "step": 13150
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.7525708675384521,
+      "learning_rate": 9.137125634940654e-06,
+      "loss": 0.7901,
+      "step": 13151
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.475266933441162,
+      "learning_rate": 9.136977565070864e-06,
+      "loss": 0.4268,
+      "step": 13152
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.8249621391296387,
+      "learning_rate": 9.136829483697688e-06,
+      "loss": 0.8152,
+      "step": 13153
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.3008675575256348,
+      "learning_rate": 9.136681390821542e-06,
+      "loss": 0.4065,
+      "step": 13154
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.8722832202911377,
+      "learning_rate": 9.136533286442833e-06,
+      "loss": 0.5849,
+      "step": 13155
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.862840175628662,
+      "learning_rate": 9.136385170561975e-06,
+      "loss": 0.6926,
+      "step": 13156
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 3.777641773223877,
+      "learning_rate": 9.136237043179378e-06,
+      "loss": 0.4965,
+      "step": 13157
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.5626513957977295,
+      "learning_rate": 9.136088904295457e-06,
+      "loss": 0.5763,
+      "step": 13158
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.8261091709136963,
+      "learning_rate": 9.13594075391062e-06,
+      "loss": 0.9445,
+      "step": 13159
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.831390142440796,
+      "learning_rate": 9.135792592025284e-06,
+      "loss": 0.9172,
+      "step": 13160
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 3.343987226486206,
+      "learning_rate": 9.135644418639856e-06,
+      "loss": 0.706,
+      "step": 13161
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.137028455734253,
+      "learning_rate": 9.13549623375475e-06,
+      "loss": 0.5675,
+      "step": 13162
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.5072920322418213,
+      "learning_rate": 9.135348037370379e-06,
+      "loss": 0.6854,
+      "step": 13163
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.9263098239898682,
+      "learning_rate": 9.135199829487155e-06,
+      "loss": 0.822,
+      "step": 13164
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.5851292610168457,
+      "learning_rate": 9.135051610105488e-06,
+      "loss": 0.5111,
+      "step": 13165
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.9780535697937012,
+      "learning_rate": 9.134903379225793e-06,
+      "loss": 0.5701,
+      "step": 13166
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.2248694896698,
+      "learning_rate": 9.134755136848479e-06,
+      "loss": 0.663,
+      "step": 13167
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.881791114807129,
+      "learning_rate": 9.13460688297396e-06,
+      "loss": 0.8128,
+      "step": 13168
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.6993775367736816,
+      "learning_rate": 9.13445861760265e-06,
+      "loss": 0.7508,
+      "step": 13169
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.862908124923706,
+      "learning_rate": 9.134310340734958e-06,
+      "loss": 0.6876,
+      "step": 13170
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.6107686758041382,
+      "learning_rate": 9.134162052371296e-06,
+      "loss": 0.6895,
+      "step": 13171
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.6540539264678955,
+      "learning_rate": 9.134013752512081e-06,
+      "loss": 0.5433,
+      "step": 13172
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.8569653034210205,
+      "learning_rate": 9.13386544115772e-06,
+      "loss": 0.7455,
+      "step": 13173
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.7437626123428345,
+      "learning_rate": 9.133717118308629e-06,
+      "loss": 0.9429,
+      "step": 13174
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 3.068009853363037,
+      "learning_rate": 9.13356878396522e-06,
+      "loss": 0.8685,
+      "step": 13175
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.8772921562194824,
+      "learning_rate": 9.133420438127903e-06,
+      "loss": 0.4165,
+      "step": 13176
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.8650404214859009,
+      "learning_rate": 9.133272080797093e-06,
+      "loss": 0.7937,
+      "step": 13177
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.9505897760391235,
+      "learning_rate": 9.133123711973202e-06,
+      "loss": 0.5212,
+      "step": 13178
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.9360053539276123,
+      "learning_rate": 9.132975331656642e-06,
+      "loss": 0.8288,
+      "step": 13179
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.4432425498962402,
+      "learning_rate": 9.132826939847825e-06,
+      "loss": 0.7264,
+      "step": 13180
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.4790546894073486,
+      "learning_rate": 9.132678536547165e-06,
+      "loss": 0.4803,
+      "step": 13181
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.955668568611145,
+      "learning_rate": 9.132530121755075e-06,
+      "loss": 0.572,
+      "step": 13182
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.492943525314331,
+      "learning_rate": 9.132381695471967e-06,
+      "loss": 0.5352,
+      "step": 13183
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.894403338432312,
+      "learning_rate": 9.132233257698254e-06,
+      "loss": 0.6078,
+      "step": 13184
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 3.1467437744140625,
+      "learning_rate": 9.132084808434347e-06,
+      "loss": 0.8186,
+      "step": 13185
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 3.0705959796905518,
+      "learning_rate": 9.131936347680658e-06,
+      "loss": 0.7208,
+      "step": 13186
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.9769928455352783,
+      "learning_rate": 9.131787875437605e-06,
+      "loss": 0.6604,
+      "step": 13187
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.542754888534546,
+      "learning_rate": 9.131639391705596e-06,
+      "loss": 0.5166,
+      "step": 13188
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.6800525188446045,
+      "learning_rate": 9.131490896485046e-06,
+      "loss": 0.3922,
+      "step": 13189
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.3096258640289307,
+      "learning_rate": 9.13134238977637e-06,
+      "loss": 0.5071,
+      "step": 13190
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.807614803314209,
+      "learning_rate": 9.131193871579975e-06,
+      "loss": 0.6091,
+      "step": 13191
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.6406261920928955,
+      "learning_rate": 9.131045341896278e-06,
+      "loss": 0.4927,
+      "step": 13192
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.8356214761734009,
+      "learning_rate": 9.130896800725693e-06,
+      "loss": 0.5608,
+      "step": 13193
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.9227862358093262,
+      "learning_rate": 9.13074824806863e-06,
+      "loss": 0.685,
+      "step": 13194
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.6585140228271484,
+      "learning_rate": 9.130599683925506e-06,
+      "loss": 0.4916,
+      "step": 13195
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 3.57647442817688,
+      "learning_rate": 9.130451108296728e-06,
+      "loss": 0.4763,
+      "step": 13196
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.79819655418396,
+      "learning_rate": 9.130302521182715e-06,
+      "loss": 0.5916,
+      "step": 13197
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.5659046173095703,
+      "learning_rate": 9.130153922583877e-06,
+      "loss": 0.5905,
+      "step": 13198
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.4371862411499023,
+      "learning_rate": 9.13000531250063e-06,
+      "loss": 0.7762,
+      "step": 13199
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.4085843563079834,
+      "learning_rate": 9.129856690933383e-06,
+      "loss": 0.5302,
+      "step": 13200
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.2225749492645264,
+      "learning_rate": 9.129708057882552e-06,
+      "loss": 0.718,
+      "step": 13201
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.8599785566329956,
+      "learning_rate": 9.12955941334855e-06,
+      "loss": 0.53,
+      "step": 13202
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.3976306915283203,
+      "learning_rate": 9.12941075733179e-06,
+      "loss": 0.7614,
+      "step": 13203
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 3.226186990737915,
+      "learning_rate": 9.12926208983269e-06,
+      "loss": 0.6836,
+      "step": 13204
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.9030818939208984,
+      "learning_rate": 9.129113410851653e-06,
+      "loss": 0.3753,
+      "step": 13205
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.6299935579299927,
+      "learning_rate": 9.128964720389102e-06,
+      "loss": 0.4991,
+      "step": 13206
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.746349811553955,
+      "learning_rate": 9.128816018445445e-06,
+      "loss": 0.4701,
+      "step": 13207
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.8993568420410156,
+      "learning_rate": 9.128667305021097e-06,
+      "loss": 0.5385,
+      "step": 13208
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.113050699234009,
+      "learning_rate": 9.128518580116474e-06,
+      "loss": 0.7714,
+      "step": 13209
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 4.011818885803223,
+      "learning_rate": 9.128369843731986e-06,
+      "loss": 0.444,
+      "step": 13210
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.379549503326416,
+      "learning_rate": 9.128221095868049e-06,
+      "loss": 0.5887,
+      "step": 13211
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 4.072395324707031,
+      "learning_rate": 9.128072336525074e-06,
+      "loss": 0.6153,
+      "step": 13212
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.1904895305633545,
+      "learning_rate": 9.127923565703477e-06,
+      "loss": 0.4132,
+      "step": 13213
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.7099182605743408,
+      "learning_rate": 9.127774783403671e-06,
+      "loss": 0.3507,
+      "step": 13214
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.071309804916382,
+      "learning_rate": 9.12762598962607e-06,
+      "loss": 0.6505,
+      "step": 13215
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.314701557159424,
+      "learning_rate": 9.127477184371086e-06,
+      "loss": 0.6415,
+      "step": 13216
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.1692357063293457,
+      "learning_rate": 9.127328367639136e-06,
+      "loss": 0.8819,
+      "step": 13217
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.499143123626709,
+      "learning_rate": 9.12717953943063e-06,
+      "loss": 0.5405,
+      "step": 13218
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.670830249786377,
+      "learning_rate": 9.127030699745984e-06,
+      "loss": 0.7231,
+      "step": 13219
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.2728073596954346,
+      "learning_rate": 9.126881848585613e-06,
+      "loss": 0.487,
+      "step": 13220
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 9.074621200561523,
+      "learning_rate": 9.12673298594993e-06,
+      "loss": 0.5701,
+      "step": 13221
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.9697229862213135,
+      "learning_rate": 9.126584111839347e-06,
+      "loss": 0.8374,
+      "step": 13222
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.460791826248169,
+      "learning_rate": 9.12643522625428e-06,
+      "loss": 0.7402,
+      "step": 13223
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.0372512340545654,
+      "learning_rate": 9.126286329195141e-06,
+      "loss": 0.6691,
+      "step": 13224
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.4868403673171997,
+      "learning_rate": 9.126137420662349e-06,
+      "loss": 0.6485,
+      "step": 13225
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 4.370301246643066,
+      "learning_rate": 9.125988500656312e-06,
+      "loss": 0.6645,
+      "step": 13226
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.0700552463531494,
+      "learning_rate": 9.125839569177446e-06,
+      "loss": 0.8918,
+      "step": 13227
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.8705554008483887,
+      "learning_rate": 9.125690626226166e-06,
+      "loss": 0.5041,
+      "step": 13228
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.5595011711120605,
+      "learning_rate": 9.125541671802887e-06,
+      "loss": 0.7299,
+      "step": 13229
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.2206408977508545,
+      "learning_rate": 9.125392705908023e-06,
+      "loss": 0.6515,
+      "step": 13230
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.89353609085083,
+      "learning_rate": 9.125243728541986e-06,
+      "loss": 0.5782,
+      "step": 13231
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.29400372505188,
+      "learning_rate": 9.125094739705191e-06,
+      "loss": 0.7488,
+      "step": 13232
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.1908984184265137,
+      "learning_rate": 9.124945739398053e-06,
+      "loss": 0.7687,
+      "step": 13233
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.870398998260498,
+      "learning_rate": 9.124796727620985e-06,
+      "loss": 0.6235,
+      "step": 13234
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.454364538192749,
+      "learning_rate": 9.124647704374406e-06,
+      "loss": 0.8362,
+      "step": 13235
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.199533700942993,
+      "learning_rate": 9.124498669658724e-06,
+      "loss": 0.9309,
+      "step": 13236
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.1637277603149414,
+      "learning_rate": 9.124349623474358e-06,
+      "loss": 0.8678,
+      "step": 13237
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.7956600189208984,
+      "learning_rate": 9.124200565821719e-06,
+      "loss": 0.8502,
+      "step": 13238
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.066178560256958,
+      "learning_rate": 9.124051496701224e-06,
+      "loss": 0.7625,
+      "step": 13239
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.8855226039886475,
+      "learning_rate": 9.123902416113287e-06,
+      "loss": 0.804,
+      "step": 13240
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 5.113556385040283,
+      "learning_rate": 9.123753324058323e-06,
+      "loss": 0.7292,
+      "step": 13241
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.052920341491699,
+      "learning_rate": 9.123604220536743e-06,
+      "loss": 0.7237,
+      "step": 13242
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 5.146976947784424,
+      "learning_rate": 9.123455105548966e-06,
+      "loss": 0.7144,
+      "step": 13243
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.2272069454193115,
+      "learning_rate": 9.123305979095404e-06,
+      "loss": 0.4356,
+      "step": 13244
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.476914644241333,
+      "learning_rate": 9.123156841176474e-06,
+      "loss": 0.7436,
+      "step": 13245
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.7269859313964844,
+      "learning_rate": 9.12300769179259e-06,
+      "loss": 0.751,
+      "step": 13246
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.181791305541992,
+      "learning_rate": 9.122858530944165e-06,
+      "loss": 0.8262,
+      "step": 13247
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.4700608253479004,
+      "learning_rate": 9.122709358631617e-06,
+      "loss": 0.543,
+      "step": 13248
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.819058895111084,
+      "learning_rate": 9.122560174855354e-06,
+      "loss": 0.8357,
+      "step": 13249
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.7793415784835815,
+      "learning_rate": 9.1224109796158e-06,
+      "loss": 0.6034,
+      "step": 13250
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.86387300491333,
+      "learning_rate": 9.122261772913364e-06,
+      "loss": 0.7692,
+      "step": 13251
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.6185752153396606,
+      "learning_rate": 9.122112554748462e-06,
+      "loss": 0.7122,
+      "step": 13252
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.6785014867782593,
+      "learning_rate": 9.121963325121508e-06,
+      "loss": 0.5763,
+      "step": 13253
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 8.854635238647461,
+      "learning_rate": 9.12181408403292e-06,
+      "loss": 1.0735,
+      "step": 13254
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.6597793102264404,
+      "learning_rate": 9.12166483148311e-06,
+      "loss": 0.7458,
+      "step": 13255
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.227308750152588,
+      "learning_rate": 9.121515567472495e-06,
+      "loss": 0.5354,
+      "step": 13256
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.124159812927246,
+      "learning_rate": 9.121366292001489e-06,
+      "loss": 0.7286,
+      "step": 13257
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.0284409523010254,
+      "learning_rate": 9.121217005070507e-06,
+      "loss": 0.6646,
+      "step": 13258
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.0134551525115967,
+      "learning_rate": 9.121067706679963e-06,
+      "loss": 0.6885,
+      "step": 13259
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.2137582302093506,
+      "learning_rate": 9.120918396830276e-06,
+      "loss": 0.7457,
+      "step": 13260
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.3807642459869385,
+      "learning_rate": 9.120769075521855e-06,
+      "loss": 0.4448,
+      "step": 13261
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.7410537004470825,
+      "learning_rate": 9.120619742755123e-06,
+      "loss": 0.4841,
+      "step": 13262
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.6732490062713623,
+      "learning_rate": 9.12047039853049e-06,
+      "loss": 0.932,
+      "step": 13263
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.1580801010131836,
+      "learning_rate": 9.12032104284837e-06,
+      "loss": 0.6038,
+      "step": 13264
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.9493234157562256,
+      "learning_rate": 9.120171675709184e-06,
+      "loss": 0.6764,
+      "step": 13265
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.8902987241744995,
+      "learning_rate": 9.120022297113343e-06,
+      "loss": 0.8372,
+      "step": 13266
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.9961017370224,
+      "learning_rate": 9.119872907061263e-06,
+      "loss": 0.6438,
+      "step": 13267
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.733557105064392,
+      "learning_rate": 9.119723505553359e-06,
+      "loss": 0.6336,
+      "step": 13268
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.8936210870742798,
+      "learning_rate": 9.119574092590046e-06,
+      "loss": 0.8429,
+      "step": 13269
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.303553581237793,
+      "learning_rate": 9.119424668171744e-06,
+      "loss": 0.6364,
+      "step": 13270
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.596984386444092,
+      "learning_rate": 9.119275232298863e-06,
+      "loss": 0.7908,
+      "step": 13271
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.2570531368255615,
+      "learning_rate": 9.119125784971823e-06,
+      "loss": 0.5752,
+      "step": 13272
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.5006769895553589,
+      "learning_rate": 9.118976326191034e-06,
+      "loss": 0.8125,
+      "step": 13273
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.33674693107605,
+      "learning_rate": 9.118826855956916e-06,
+      "loss": 0.5229,
+      "step": 13274
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.880873680114746,
+      "learning_rate": 9.118677374269885e-06,
+      "loss": 0.8967,
+      "step": 13275
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.3087059259414673,
+      "learning_rate": 9.118527881130355e-06,
+      "loss": 0.6997,
+      "step": 13276
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.6988914012908936,
+      "learning_rate": 9.11837837653874e-06,
+      "loss": 0.8806,
+      "step": 13277
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.2021076679229736,
+      "learning_rate": 9.118228860495459e-06,
+      "loss": 0.4944,
+      "step": 13278
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.179086923599243,
+      "learning_rate": 9.118079333000926e-06,
+      "loss": 0.8309,
+      "step": 13279
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.232820987701416,
+      "learning_rate": 9.117929794055555e-06,
+      "loss": 0.3484,
+      "step": 13280
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.416084051132202,
+      "learning_rate": 9.117780243659766e-06,
+      "loss": 1.1853,
+      "step": 13281
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.4824321269989014,
+      "learning_rate": 9.117630681813973e-06,
+      "loss": 0.8202,
+      "step": 13282
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.9930721521377563,
+      "learning_rate": 9.117481108518592e-06,
+      "loss": 0.7973,
+      "step": 13283
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.7658207416534424,
+      "learning_rate": 9.117331523774036e-06,
+      "loss": 0.5971,
+      "step": 13284
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.098478078842163,
+      "learning_rate": 9.117181927580725e-06,
+      "loss": 0.5861,
+      "step": 13285
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.7959580421447754,
+      "learning_rate": 9.117032319939073e-06,
+      "loss": 0.65,
+      "step": 13286
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.9162042140960693,
+      "learning_rate": 9.116882700849497e-06,
+      "loss": 0.6528,
+      "step": 13287
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.0438966751098633,
+      "learning_rate": 9.11673307031241e-06,
+      "loss": 0.7526,
+      "step": 13288
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.754636764526367,
+      "learning_rate": 9.116583428328234e-06,
+      "loss": 0.3666,
+      "step": 13289
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.7006549835205078,
+      "learning_rate": 9.11643377489738e-06,
+      "loss": 0.5862,
+      "step": 13290
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.6027098894119263,
+      "learning_rate": 9.116284110020267e-06,
+      "loss": 0.7865,
+      "step": 13291
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.23740553855896,
+      "learning_rate": 9.116134433697307e-06,
+      "loss": 0.689,
+      "step": 13292
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.7181692123413086,
+      "learning_rate": 9.115984745928921e-06,
+      "loss": 0.7149,
+      "step": 13293
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.8857595920562744,
+      "learning_rate": 9.115835046715523e-06,
+      "loss": 0.832,
+      "step": 13294
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.1707839965820312,
+      "learning_rate": 9.115685336057529e-06,
+      "loss": 0.731,
+      "step": 13295
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.0967652797698975,
+      "learning_rate": 9.115535613955357e-06,
+      "loss": 0.6172,
+      "step": 13296
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.849414825439453,
+      "learning_rate": 9.115385880409421e-06,
+      "loss": 0.6001,
+      "step": 13297
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.497669219970703,
+      "learning_rate": 9.115236135420138e-06,
+      "loss": 0.6994,
+      "step": 13298
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.2281177043914795,
+      "learning_rate": 9.115086378987928e-06,
+      "loss": 0.7215,
+      "step": 13299
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.6573830842971802,
+      "learning_rate": 9.1149366111132e-06,
+      "loss": 0.8068,
+      "step": 13300
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.1466500759124756,
+      "learning_rate": 9.114786831796378e-06,
+      "loss": 0.6862,
+      "step": 13301
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.7659660577774048,
+      "learning_rate": 9.114637041037874e-06,
+      "loss": 0.7056,
+      "step": 13302
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.6964123249053955,
+      "learning_rate": 9.114487238838107e-06,
+      "loss": 0.6641,
+      "step": 13303
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.4526305198669434,
+      "learning_rate": 9.114337425197491e-06,
+      "loss": 0.9231,
+      "step": 13304
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.86350679397583,
+      "learning_rate": 9.114187600116446e-06,
+      "loss": 0.8739,
+      "step": 13305
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.2991409301757812,
+      "learning_rate": 9.114037763595385e-06,
+      "loss": 1.0518,
+      "step": 13306
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.7764389514923096,
+      "learning_rate": 9.113887915634725e-06,
+      "loss": 0.7882,
+      "step": 13307
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.5964685678482056,
+      "learning_rate": 9.113738056234884e-06,
+      "loss": 0.7134,
+      "step": 13308
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.7832175493240356,
+      "learning_rate": 9.11358818539628e-06,
+      "loss": 0.5097,
+      "step": 13309
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.1036696434020996,
+      "learning_rate": 9.113438303119328e-06,
+      "loss": 0.5381,
+      "step": 13310
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.527525782585144,
+      "learning_rate": 9.113288409404444e-06,
+      "loss": 0.5514,
+      "step": 13311
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.5818490982055664,
+      "learning_rate": 9.113138504252046e-06,
+      "loss": 0.7244,
+      "step": 13312
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.7112247943878174,
+      "learning_rate": 9.112988587662553e-06,
+      "loss": 0.924,
+      "step": 13313
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.4677116870880127,
+      "learning_rate": 9.112838659636377e-06,
+      "loss": 0.5641,
+      "step": 13314
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 4.512304306030273,
+      "learning_rate": 9.112688720173936e-06,
+      "loss": 0.6609,
+      "step": 13315
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.7997713088989258,
+      "learning_rate": 9.112538769275652e-06,
+      "loss": 0.6504,
+      "step": 13316
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.9349143505096436,
+      "learning_rate": 9.112388806941936e-06,
+      "loss": 0.8144,
+      "step": 13317
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.1447737216949463,
+      "learning_rate": 9.112238833173208e-06,
+      "loss": 0.6141,
+      "step": 13318
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.4541640281677246,
+      "learning_rate": 9.112088847969883e-06,
+      "loss": 0.5084,
+      "step": 13319
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.024303913116455,
+      "learning_rate": 9.11193885133238e-06,
+      "loss": 0.7248,
+      "step": 13320
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.6294939517974854,
+      "learning_rate": 9.111788843261117e-06,
+      "loss": 0.6093,
+      "step": 13321
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.764355182647705,
+      "learning_rate": 9.111638823756507e-06,
+      "loss": 0.9022,
+      "step": 13322
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 4.659295082092285,
+      "learning_rate": 9.111488792818972e-06,
+      "loss": 0.5382,
+      "step": 13323
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.7284106016159058,
+      "learning_rate": 9.111338750448925e-06,
+      "loss": 0.7897,
+      "step": 13324
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.1496126651763916,
+      "learning_rate": 9.111188696646786e-06,
+      "loss": 0.5986,
+      "step": 13325
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.5614523887634277,
+      "learning_rate": 9.111038631412972e-06,
+      "loss": 0.5126,
+      "step": 13326
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.0910840034484863,
+      "learning_rate": 9.110888554747898e-06,
+      "loss": 0.8271,
+      "step": 13327
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.0590784549713135,
+      "learning_rate": 9.110738466651985e-06,
+      "loss": 0.759,
+      "step": 13328
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 4.6343464851379395,
+      "learning_rate": 9.110588367125646e-06,
+      "loss": 0.6188,
+      "step": 13329
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.0406687259674072,
+      "learning_rate": 9.110438256169301e-06,
+      "loss": 0.5628,
+      "step": 13330
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.852622389793396,
+      "learning_rate": 9.110288133783369e-06,
+      "loss": 0.5613,
+      "step": 13331
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.2480859756469727,
+      "learning_rate": 9.110137999968263e-06,
+      "loss": 0.5893,
+      "step": 13332
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.267303228378296,
+      "learning_rate": 9.109987854724405e-06,
+      "loss": 0.6993,
+      "step": 13333
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.3680663108825684,
+      "learning_rate": 9.109837698052208e-06,
+      "loss": 0.6413,
+      "step": 13334
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.415590524673462,
+      "learning_rate": 9.109687529952094e-06,
+      "loss": 0.7645,
+      "step": 13335
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.4935269355773926,
+      "learning_rate": 9.109537350424477e-06,
+      "loss": 0.7036,
+      "step": 13336
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.840585708618164,
+      "learning_rate": 9.109387159469777e-06,
+      "loss": 0.8163,
+      "step": 13337
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.0103836059570312,
+      "learning_rate": 9.109236957088412e-06,
+      "loss": 0.552,
+      "step": 13338
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.608635187149048,
+      "learning_rate": 9.109086743280797e-06,
+      "loss": 0.976,
+      "step": 13339
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.7172178030014038,
+      "learning_rate": 9.108936518047352e-06,
+      "loss": 0.3368,
+      "step": 13340
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 4.194417476654053,
+      "learning_rate": 9.108786281388492e-06,
+      "loss": 0.6097,
+      "step": 13341
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.7221972942352295,
+      "learning_rate": 9.108636033304637e-06,
+      "loss": 0.8594,
+      "step": 13342
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.2174949645996094,
+      "learning_rate": 9.108485773796205e-06,
+      "loss": 0.4904,
+      "step": 13343
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.7820932865142822,
+      "learning_rate": 9.108335502863613e-06,
+      "loss": 0.4611,
+      "step": 13344
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.053478717803955,
+      "learning_rate": 9.10818522050728e-06,
+      "loss": 0.5622,
+      "step": 13345
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.7484631538391113,
+      "learning_rate": 9.10803492672762e-06,
+      "loss": 0.7338,
+      "step": 13346
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 4.153586387634277,
+      "learning_rate": 9.107884621525056e-06,
+      "loss": 0.7772,
+      "step": 13347
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.5175552368164062,
+      "learning_rate": 9.107734304900005e-06,
+      "loss": 0.5159,
+      "step": 13348
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.8400477170944214,
+      "learning_rate": 9.10758397685288e-06,
+      "loss": 0.4755,
+      "step": 13349
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.6261013746261597,
+      "learning_rate": 9.107433637384106e-06,
+      "loss": 0.7993,
+      "step": 13350
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.7388015985488892,
+      "learning_rate": 9.107283286494095e-06,
+      "loss": 0.6632,
+      "step": 13351
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.778552770614624,
+      "learning_rate": 9.10713292418327e-06,
+      "loss": 0.3792,
+      "step": 13352
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.0692975521087646,
+      "learning_rate": 9.106982550452045e-06,
+      "loss": 0.4546,
+      "step": 13353
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.2091150283813477,
+      "learning_rate": 9.106832165300841e-06,
+      "loss": 0.6471,
+      "step": 13354
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.4303042888641357,
+      "learning_rate": 9.106681768730078e-06,
+      "loss": 0.839,
+      "step": 13355
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.8122847080230713,
+      "learning_rate": 9.106531360740168e-06,
+      "loss": 0.6133,
+      "step": 13356
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 6.141816139221191,
+      "learning_rate": 9.106380941331531e-06,
+      "loss": 0.7181,
+      "step": 13357
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.680264949798584,
+      "learning_rate": 9.10623051050459e-06,
+      "loss": 0.7067,
+      "step": 13358
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.547744035720825,
+      "learning_rate": 9.106080068259759e-06,
+      "loss": 0.6584,
+      "step": 13359
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.296625852584839,
+      "learning_rate": 9.105929614597458e-06,
+      "loss": 0.5243,
+      "step": 13360
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.2513022422790527,
+      "learning_rate": 9.105779149518106e-06,
+      "loss": 0.5047,
+      "step": 13361
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.8520350456237793,
+      "learning_rate": 9.105628673022118e-06,
+      "loss": 0.7202,
+      "step": 13362
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.475987434387207,
+      "learning_rate": 9.105478185109916e-06,
+      "loss": 0.7145,
+      "step": 13363
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.0840375423431396,
+      "learning_rate": 9.105327685781918e-06,
+      "loss": 0.3918,
+      "step": 13364
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.470615863800049,
+      "learning_rate": 9.10517717503854e-06,
+      "loss": 0.4275,
+      "step": 13365
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.181130886077881,
+      "learning_rate": 9.105026652880202e-06,
+      "loss": 0.5123,
+      "step": 13366
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.1638166904449463,
+      "learning_rate": 9.104876119307322e-06,
+      "loss": 0.7157,
+      "step": 13367
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.6880764961242676,
+      "learning_rate": 9.104725574320322e-06,
+      "loss": 0.644,
+      "step": 13368
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.02245831489563,
+      "learning_rate": 9.104575017919616e-06,
+      "loss": 0.5814,
+      "step": 13369
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.0107269287109375,
+      "learning_rate": 9.104424450105625e-06,
+      "loss": 0.8654,
+      "step": 13370
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.277120351791382,
+      "learning_rate": 9.104273870878766e-06,
+      "loss": 0.6074,
+      "step": 13371
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.5259318351745605,
+      "learning_rate": 9.10412328023946e-06,
+      "loss": 0.901,
+      "step": 13372
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.6528992652893066,
+      "learning_rate": 9.103972678188125e-06,
+      "loss": 0.5795,
+      "step": 13373
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.6143877506256104,
+      "learning_rate": 9.103822064725178e-06,
+      "loss": 0.5434,
+      "step": 13374
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.667884111404419,
+      "learning_rate": 9.103671439851041e-06,
+      "loss": 0.7855,
+      "step": 13375
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.657715082168579,
+      "learning_rate": 9.103520803566128e-06,
+      "loss": 0.411,
+      "step": 13376
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.4914209842681885,
+      "learning_rate": 9.103370155870864e-06,
+      "loss": 0.7094,
+      "step": 13377
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.906656265258789,
+      "learning_rate": 9.103219496765664e-06,
+      "loss": 0.6138,
+      "step": 13378
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.126070976257324,
+      "learning_rate": 9.103068826250947e-06,
+      "loss": 0.8013,
+      "step": 13379
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.0158064365386963,
+      "learning_rate": 9.102918144327131e-06,
+      "loss": 0.5734,
+      "step": 13380
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.8618059158325195,
+      "learning_rate": 9.10276745099464e-06,
+      "loss": 0.6402,
+      "step": 13381
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.6751455068588257,
+      "learning_rate": 9.102616746253887e-06,
+      "loss": 0.5126,
+      "step": 13382
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.1949470043182373,
+      "learning_rate": 9.102466030105294e-06,
+      "loss": 0.5656,
+      "step": 13383
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.898087978363037,
+      "learning_rate": 9.102315302549281e-06,
+      "loss": 0.4441,
+      "step": 13384
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.0866386890411377,
+      "learning_rate": 9.102164563586267e-06,
+      "loss": 0.5883,
+      "step": 13385
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.444359302520752,
+      "learning_rate": 9.102013813216669e-06,
+      "loss": 0.7997,
+      "step": 13386
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.9181774854660034,
+      "learning_rate": 9.101863051440904e-06,
+      "loss": 0.7185,
+      "step": 13387
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 4.141602993011475,
+      "learning_rate": 9.101712278259397e-06,
+      "loss": 0.4874,
+      "step": 13388
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.9370530843734741,
+      "learning_rate": 9.101561493672566e-06,
+      "loss": 0.6473,
+      "step": 13389
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.1593780517578125,
+      "learning_rate": 9.101410697680828e-06,
+      "loss": 0.6652,
+      "step": 13390
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.4462977647781372,
+      "learning_rate": 9.101259890284604e-06,
+      "loss": 0.8344,
+      "step": 13391
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.463425397872925,
+      "learning_rate": 9.10110907148431e-06,
+      "loss": 0.9423,
+      "step": 13392
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.5020198822021484,
+      "learning_rate": 9.10095824128037e-06,
+      "loss": 0.8209,
+      "step": 13393
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.6778435707092285,
+      "learning_rate": 9.100807399673201e-06,
+      "loss": 0.9259,
+      "step": 13394
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.8558851480484009,
+      "learning_rate": 9.100656546663225e-06,
+      "loss": 0.6485,
+      "step": 13395
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.026949882507324,
+      "learning_rate": 9.100505682250857e-06,
+      "loss": 0.619,
+      "step": 13396
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.617025375366211,
+      "learning_rate": 9.100354806436517e-06,
+      "loss": 0.637,
+      "step": 13397
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.3117473125457764,
+      "learning_rate": 9.10020391922063e-06,
+      "loss": 0.8835,
+      "step": 13398
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.76330304145813,
+      "learning_rate": 9.10005302060361e-06,
+      "loss": 0.7749,
+      "step": 13399
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.085738182067871,
+      "learning_rate": 9.099902110585878e-06,
+      "loss": 0.7893,
+      "step": 13400
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.663769245147705,
+      "learning_rate": 9.099751189167856e-06,
+      "loss": 0.9102,
+      "step": 13401
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.1780471801757812,
+      "learning_rate": 9.09960025634996e-06,
+      "loss": 0.6976,
+      "step": 13402
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.010070323944092,
+      "learning_rate": 9.09944931213261e-06,
+      "loss": 0.8028,
+      "step": 13403
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.2315924167633057,
+      "learning_rate": 9.09929835651623e-06,
+      "loss": 0.4725,
+      "step": 13404
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.0908384323120117,
+      "learning_rate": 9.099147389501237e-06,
+      "loss": 0.4488,
+      "step": 13405
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.374779462814331,
+      "learning_rate": 9.098996411088048e-06,
+      "loss": 0.73,
+      "step": 13406
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 12.621086120605469,
+      "learning_rate": 9.098845421277088e-06,
+      "loss": 0.6195,
+      "step": 13407
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.066389322280884,
+      "learning_rate": 9.098694420068775e-06,
+      "loss": 0.6001,
+      "step": 13408
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.774188756942749,
+      "learning_rate": 9.098543407463526e-06,
+      "loss": 0.6337,
+      "step": 13409
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.333631992340088,
+      "learning_rate": 9.098392383461763e-06,
+      "loss": 0.7499,
+      "step": 13410
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.2454845905303955,
+      "learning_rate": 9.098241348063908e-06,
+      "loss": 0.7514,
+      "step": 13411
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.2358853816986084,
+      "learning_rate": 9.098090301270378e-06,
+      "loss": 0.7916,
+      "step": 13412
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.8204317092895508,
+      "learning_rate": 9.097939243081595e-06,
+      "loss": 0.8929,
+      "step": 13413
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.8815157413482666,
+      "learning_rate": 9.097788173497975e-06,
+      "loss": 0.6416,
+      "step": 13414
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.773761510848999,
+      "learning_rate": 9.097637092519944e-06,
+      "loss": 0.7847,
+      "step": 13415
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.736686944961548,
+      "learning_rate": 9.09748600014792e-06,
+      "loss": 0.7917,
+      "step": 13416
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.116056442260742,
+      "learning_rate": 9.09733489638232e-06,
+      "loss": 0.6536,
+      "step": 13417
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.586057662963867,
+      "learning_rate": 9.097183781223566e-06,
+      "loss": 0.9263,
+      "step": 13418
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.1817212104797363,
+      "learning_rate": 9.09703265467208e-06,
+      "loss": 0.7844,
+      "step": 13419
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.3390493392944336,
+      "learning_rate": 9.096881516728283e-06,
+      "loss": 0.4304,
+      "step": 13420
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.5631825923919678,
+      "learning_rate": 9.09673036739259e-06,
+      "loss": 0.8552,
+      "step": 13421
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.0441410541534424,
+      "learning_rate": 9.096579206665425e-06,
+      "loss": 0.7511,
+      "step": 13422
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.7441903352737427,
+      "learning_rate": 9.09642803454721e-06,
+      "loss": 0.6048,
+      "step": 13423
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.8282585144042969,
+      "learning_rate": 9.096276851038362e-06,
+      "loss": 0.5997,
+      "step": 13424
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.8489322662353516,
+      "learning_rate": 9.096125656139303e-06,
+      "loss": 0.7347,
+      "step": 13425
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.4278669357299805,
+      "learning_rate": 9.095974449850451e-06,
+      "loss": 0.6909,
+      "step": 13426
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.8594081401824951,
+      "learning_rate": 9.095823232172232e-06,
+      "loss": 0.6426,
+      "step": 13427
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.030575752258301,
+      "learning_rate": 9.09567200310506e-06,
+      "loss": 0.5345,
+      "step": 13428
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.6940174102783203,
+      "learning_rate": 9.09552076264936e-06,
+      "loss": 0.7485,
+      "step": 13429
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.762345314025879,
+      "learning_rate": 9.09536951080555e-06,
+      "loss": 0.7951,
+      "step": 13430
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.9324936866760254,
+      "learning_rate": 9.09521824757405e-06,
+      "loss": 0.5642,
+      "step": 13431
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.0988078117370605,
+      "learning_rate": 9.095066972955287e-06,
+      "loss": 0.519,
+      "step": 13432
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.1511478424072266,
+      "learning_rate": 9.094915686949672e-06,
+      "loss": 0.7793,
+      "step": 13433
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.949933648109436,
+      "learning_rate": 9.094764389557633e-06,
+      "loss": 0.7312,
+      "step": 13434
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.8877010345458984,
+      "learning_rate": 9.094613080779587e-06,
+      "loss": 0.862,
+      "step": 13435
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.4911015033721924,
+      "learning_rate": 9.094461760615957e-06,
+      "loss": 0.6749,
+      "step": 13436
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.8538212776184082,
+      "learning_rate": 9.094310429067161e-06,
+      "loss": 0.5823,
+      "step": 13437
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 4.615832328796387,
+      "learning_rate": 9.094159086133623e-06,
+      "loss": 0.8618,
+      "step": 13438
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.0591320991516113,
+      "learning_rate": 9.094007731815762e-06,
+      "loss": 0.8259,
+      "step": 13439
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 5.895419597625732,
+      "learning_rate": 9.093856366113999e-06,
+      "loss": 0.7357,
+      "step": 13440
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.561591863632202,
+      "learning_rate": 9.093704989028756e-06,
+      "loss": 0.5827,
+      "step": 13441
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.7894811630249023,
+      "learning_rate": 9.093553600560451e-06,
+      "loss": 0.8406,
+      "step": 13442
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.21750545501709,
+      "learning_rate": 9.093402200709507e-06,
+      "loss": 0.3487,
+      "step": 13443
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.5491268634796143,
+      "learning_rate": 9.093250789476347e-06,
+      "loss": 0.6774,
+      "step": 13444
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.244197130203247,
+      "learning_rate": 9.093099366861389e-06,
+      "loss": 0.7262,
+      "step": 13445
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.1564393043518066,
+      "learning_rate": 9.092947932865053e-06,
+      "loss": 0.4909,
+      "step": 13446
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.5938810110092163,
+      "learning_rate": 9.092796487487766e-06,
+      "loss": 0.4509,
+      "step": 13447
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.5702085494995117,
+      "learning_rate": 9.09264503072994e-06,
+      "loss": 0.5091,
+      "step": 13448
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.1968204975128174,
+      "learning_rate": 9.092493562592005e-06,
+      "loss": 0.723,
+      "step": 13449
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.3655591011047363,
+      "learning_rate": 9.092342083074378e-06,
+      "loss": 0.5697,
+      "step": 13450
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.5011515617370605,
+      "learning_rate": 9.09219059217748e-06,
+      "loss": 0.8409,
+      "step": 13451
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.9890614748001099,
+      "learning_rate": 9.092039089901733e-06,
+      "loss": 0.5172,
+      "step": 13452
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.4790306091308594,
+      "learning_rate": 9.091887576247559e-06,
+      "loss": 0.6544,
+      "step": 13453
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.630140542984009,
+      "learning_rate": 9.091736051215378e-06,
+      "loss": 0.4012,
+      "step": 13454
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.391695022583008,
+      "learning_rate": 9.091584514805611e-06,
+      "loss": 0.5539,
+      "step": 13455
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.202531099319458,
+      "learning_rate": 9.091432967018681e-06,
+      "loss": 0.6695,
+      "step": 13456
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 4.0574188232421875,
+      "learning_rate": 9.091281407855009e-06,
+      "loss": 0.9032,
+      "step": 13457
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.5806257724761963,
+      "learning_rate": 9.091129837315016e-06,
+      "loss": 0.5846,
+      "step": 13458
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.6563315391540527,
+      "learning_rate": 9.090978255399124e-06,
+      "loss": 0.725,
+      "step": 13459
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.669569730758667,
+      "learning_rate": 9.090826662107753e-06,
+      "loss": 0.8166,
+      "step": 13460
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.3139851093292236,
+      "learning_rate": 9.090675057441325e-06,
+      "loss": 0.4614,
+      "step": 13461
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.6326589584350586,
+      "learning_rate": 9.090523441400263e-06,
+      "loss": 0.4517,
+      "step": 13462
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.3409016132354736,
+      "learning_rate": 9.090371813984987e-06,
+      "loss": 0.5927,
+      "step": 13463
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.077990770339966,
+      "learning_rate": 9.09022017519592e-06,
+      "loss": 0.7818,
+      "step": 13464
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.6322970390319824,
+      "learning_rate": 9.090068525033481e-06,
+      "loss": 0.715,
+      "step": 13465
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.7286577224731445,
+      "learning_rate": 9.089916863498097e-06,
+      "loss": 0.8241,
+      "step": 13466
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.8267276287078857,
+      "learning_rate": 9.089765190590184e-06,
+      "loss": 0.7447,
+      "step": 13467
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.935219645500183,
+      "learning_rate": 9.089613506310167e-06,
+      "loss": 0.6393,
+      "step": 13468
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.9323416948318481,
+      "learning_rate": 9.089461810658466e-06,
+      "loss": 0.6889,
+      "step": 13469
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.931649923324585,
+      "learning_rate": 9.089310103635505e-06,
+      "loss": 0.4522,
+      "step": 13470
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.04424786567688,
+      "learning_rate": 9.089158385241704e-06,
+      "loss": 0.756,
+      "step": 13471
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.144164562225342,
+      "learning_rate": 9.089006655477483e-06,
+      "loss": 0.7258,
+      "step": 13472
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 4.199191093444824,
+      "learning_rate": 9.088854914343269e-06,
+      "loss": 0.5253,
+      "step": 13473
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.080522060394287,
+      "learning_rate": 9.08870316183948e-06,
+      "loss": 0.7115,
+      "step": 13474
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.648909568786621,
+      "learning_rate": 9.08855139796654e-06,
+      "loss": 0.3931,
+      "step": 13475
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.6123244762420654,
+      "learning_rate": 9.08839962272487e-06,
+      "loss": 0.7786,
+      "step": 13476
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.5362167358398438,
+      "learning_rate": 9.088247836114892e-06,
+      "loss": 0.6438,
+      "step": 13477
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.3636317253112793,
+      "learning_rate": 9.088096038137028e-06,
+      "loss": 0.8283,
+      "step": 13478
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.403486490249634,
+      "learning_rate": 9.0879442287917e-06,
+      "loss": 0.5824,
+      "step": 13479
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.7505943775177002,
+      "learning_rate": 9.08779240807933e-06,
+      "loss": 0.6708,
+      "step": 13480
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.8360118865966797,
+      "learning_rate": 9.087640576000343e-06,
+      "loss": 0.6333,
+      "step": 13481
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.089433431625366,
+      "learning_rate": 9.087488732555156e-06,
+      "loss": 0.4873,
+      "step": 13482
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.200270891189575,
+      "learning_rate": 9.087336877744196e-06,
+      "loss": 0.788,
+      "step": 13483
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.687997341156006,
+      "learning_rate": 9.087185011567882e-06,
+      "loss": 0.6689,
+      "step": 13484
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.5728986263275146,
+      "learning_rate": 9.08703313402664e-06,
+      "loss": 0.6329,
+      "step": 13485
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.8009579181671143,
+      "learning_rate": 9.086881245120888e-06,
+      "loss": 0.7905,
+      "step": 13486
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.6362189054489136,
+      "learning_rate": 9.08672934485105e-06,
+      "loss": 0.5749,
+      "step": 13487
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.664557695388794,
+      "learning_rate": 9.08657743321755e-06,
+      "loss": 0.4922,
+      "step": 13488
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.5197622776031494,
+      "learning_rate": 9.086425510220807e-06,
+      "loss": 0.7275,
+      "step": 13489
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 4.0703887939453125,
+      "learning_rate": 9.086273575861247e-06,
+      "loss": 0.6658,
+      "step": 13490
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.38702130317688,
+      "learning_rate": 9.08612163013929e-06,
+      "loss": 0.6553,
+      "step": 13491
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.1513102054595947,
+      "learning_rate": 9.08596967305536e-06,
+      "loss": 0.783,
+      "step": 13492
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.2335093021392822,
+      "learning_rate": 9.08581770460988e-06,
+      "loss": 0.7644,
+      "step": 13493
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.780374526977539,
+      "learning_rate": 9.08566572480327e-06,
+      "loss": 0.5974,
+      "step": 13494
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.933862328529358,
+      "learning_rate": 9.085513733635956e-06,
+      "loss": 0.6564,
+      "step": 13495
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.299888849258423,
+      "learning_rate": 9.085361731108358e-06,
+      "loss": 0.6447,
+      "step": 13496
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.957972288131714,
+      "learning_rate": 9.085209717220898e-06,
+      "loss": 0.7337,
+      "step": 13497
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.566352367401123,
+      "learning_rate": 9.085057691974001e-06,
+      "loss": 0.6305,
+      "step": 13498
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.4112794399261475,
+      "learning_rate": 9.084905655368087e-06,
+      "loss": 0.7493,
+      "step": 13499
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.8751726150512695,
+      "learning_rate": 9.084753607403584e-06,
+      "loss": 0.93,
+      "step": 13500
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.2281737327575684,
+      "learning_rate": 9.08460154808091e-06,
+      "loss": 0.669,
+      "step": 13501
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.8868359327316284,
+      "learning_rate": 9.08444947740049e-06,
+      "loss": 0.847,
+      "step": 13502
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.5356953144073486,
+      "learning_rate": 9.084297395362744e-06,
+      "loss": 0.5505,
+      "step": 13503
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.0897281169891357,
+      "learning_rate": 9.0841453019681e-06,
+      "loss": 0.6908,
+      "step": 13504
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.2997565269470215,
+      "learning_rate": 9.083993197216974e-06,
+      "loss": 0.6618,
+      "step": 13505
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.9824628829956055,
+      "learning_rate": 9.083841081109796e-06,
+      "loss": 0.4833,
+      "step": 13506
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.4404754638671875,
+      "learning_rate": 9.083688953646983e-06,
+      "loss": 0.6138,
+      "step": 13507
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.884061574935913,
+      "learning_rate": 9.083536814828963e-06,
+      "loss": 0.8741,
+      "step": 13508
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 4.184769630432129,
+      "learning_rate": 9.083384664656157e-06,
+      "loss": 0.4338,
+      "step": 13509
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.356879472732544,
+      "learning_rate": 9.083232503128987e-06,
+      "loss": 0.149,
+      "step": 13510
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.2699124813079834,
+      "learning_rate": 9.083080330247878e-06,
+      "loss": 0.875,
+      "step": 13511
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.6882816553115845,
+      "learning_rate": 9.08292814601325e-06,
+      "loss": 0.7043,
+      "step": 13512
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.1475653648376465,
+      "learning_rate": 9.08277595042553e-06,
+      "loss": 0.6015,
+      "step": 13513
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.9470796585083008,
+      "learning_rate": 9.08262374348514e-06,
+      "loss": 0.518,
+      "step": 13514
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.3494322299957275,
+      "learning_rate": 9.082471525192502e-06,
+      "loss": 0.8007,
+      "step": 13515
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.102414131164551,
+      "learning_rate": 9.082319295548038e-06,
+      "loss": 0.2414,
+      "step": 13516
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.7699503898620605,
+      "learning_rate": 9.082167054552175e-06,
+      "loss": 0.5117,
+      "step": 13517
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.352673053741455,
+      "learning_rate": 9.082014802205337e-06,
+      "loss": 0.7978,
+      "step": 13518
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.192307472229004,
+      "learning_rate": 9.081862538507942e-06,
+      "loss": 0.7144,
+      "step": 13519
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.7161515951156616,
+      "learning_rate": 9.081710263460417e-06,
+      "loss": 0.5242,
+      "step": 13520
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.0267529487609863,
+      "learning_rate": 9.081557977063186e-06,
+      "loss": 0.4944,
+      "step": 13521
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.6835594177246094,
+      "learning_rate": 9.08140567931667e-06,
+      "loss": 0.6976,
+      "step": 13522
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.0321011543273926,
+      "learning_rate": 9.081253370221295e-06,
+      "loss": 0.6988,
+      "step": 13523
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.2316699028015137,
+      "learning_rate": 9.081101049777481e-06,
+      "loss": 0.7873,
+      "step": 13524
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.330624580383301,
+      "learning_rate": 9.080948717985656e-06,
+      "loss": 0.6786,
+      "step": 13525
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.9944789409637451,
+      "learning_rate": 9.080796374846242e-06,
+      "loss": 0.4332,
+      "step": 13526
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.8336408138275146,
+      "learning_rate": 9.08064402035966e-06,
+      "loss": 0.8192,
+      "step": 13527
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.8574299812316895,
+      "learning_rate": 9.080491654526335e-06,
+      "loss": 0.3022,
+      "step": 13528
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.3397040367126465,
+      "learning_rate": 9.080339277346693e-06,
+      "loss": 0.8148,
+      "step": 13529
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.878993511199951,
+      "learning_rate": 9.080186888821155e-06,
+      "loss": 0.8469,
+      "step": 13530
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.4327380657196045,
+      "learning_rate": 9.080034488950147e-06,
+      "loss": 0.3536,
+      "step": 13531
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 68.00572967529297,
+      "learning_rate": 9.07988207773409e-06,
+      "loss": 0.47,
+      "step": 13532
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.9922919273376465,
+      "learning_rate": 9.07972965517341e-06,
+      "loss": 0.5612,
+      "step": 13533
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.6827266216278076,
+      "learning_rate": 9.07957722126853e-06,
+      "loss": 0.9936,
+      "step": 13534
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.8851189613342285,
+      "learning_rate": 9.079424776019874e-06,
+      "loss": 0.7762,
+      "step": 13535
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.029430866241455,
+      "learning_rate": 9.079272319427866e-06,
+      "loss": 0.8726,
+      "step": 13536
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.267934799194336,
+      "learning_rate": 9.079119851492931e-06,
+      "loss": 0.8363,
+      "step": 13537
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.9048361778259277,
+      "learning_rate": 9.07896737221549e-06,
+      "loss": 0.5901,
+      "step": 13538
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.426300048828125,
+      "learning_rate": 9.07881488159597e-06,
+      "loss": 0.9829,
+      "step": 13539
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.6072936058044434,
+      "learning_rate": 9.078662379634791e-06,
+      "loss": 0.8578,
+      "step": 13540
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.077542543411255,
+      "learning_rate": 9.078509866332382e-06,
+      "loss": 0.4553,
+      "step": 13541
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.756479024887085,
+      "learning_rate": 9.078357341689166e-06,
+      "loss": 0.5905,
+      "step": 13542
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.0807151794433594,
+      "learning_rate": 9.078204805705564e-06,
+      "loss": 0.6784,
+      "step": 13543
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.000887632369995,
+      "learning_rate": 9.078052258382002e-06,
+      "loss": 0.6468,
+      "step": 13544
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.2514166831970215,
+      "learning_rate": 9.077899699718905e-06,
+      "loss": 0.5664,
+      "step": 13545
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 6.914203643798828,
+      "learning_rate": 9.077747129716699e-06,
+      "loss": 0.335,
+      "step": 13546
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.399772882461548,
+      "learning_rate": 9.077594548375803e-06,
+      "loss": 0.6668,
+      "step": 13547
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.8374102115631104,
+      "learning_rate": 9.077441955696643e-06,
+      "loss": 0.7446,
+      "step": 13548
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.4687010049819946,
+      "learning_rate": 9.077289351679649e-06,
+      "loss": 0.5989,
+      "step": 13549
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.0824434757232666,
+      "learning_rate": 9.077136736325236e-06,
+      "loss": 0.6473,
+      "step": 13550
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.4847865104675293,
+      "learning_rate": 9.076984109633835e-06,
+      "loss": 0.5585,
+      "step": 13551
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.357797145843506,
+      "learning_rate": 9.076831471605869e-06,
+      "loss": 0.8722,
+      "step": 13552
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.6934895515441895,
+      "learning_rate": 9.07667882224176e-06,
+      "loss": 0.6904,
+      "step": 13553
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.1925883293151855,
+      "learning_rate": 9.076526161541936e-06,
+      "loss": 0.768,
+      "step": 13554
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.7173608541488647,
+      "learning_rate": 9.07637348950682e-06,
+      "loss": 0.7107,
+      "step": 13555
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.2303783893585205,
+      "learning_rate": 9.076220806136836e-06,
+      "loss": 0.2189,
+      "step": 13556
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.4626426696777344,
+      "learning_rate": 9.076068111432409e-06,
+      "loss": 0.4759,
+      "step": 13557
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.3073973655700684,
+      "learning_rate": 9.075915405393963e-06,
+      "loss": 0.6056,
+      "step": 13558
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.6637661457061768,
+      "learning_rate": 9.075762688021924e-06,
+      "loss": 0.7189,
+      "step": 13559
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.169015884399414,
+      "learning_rate": 9.075609959316715e-06,
+      "loss": 0.6181,
+      "step": 13560
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.1387081146240234,
+      "learning_rate": 9.075457219278762e-06,
+      "loss": 0.5371,
+      "step": 13561
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.5564824342727661,
+      "learning_rate": 9.075304467908488e-06,
+      "loss": 0.5948,
+      "step": 13562
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.891160488128662,
+      "learning_rate": 9.07515170520632e-06,
+      "loss": 1.0072,
+      "step": 13563
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.846433639526367,
+      "learning_rate": 9.07499893117268e-06,
+      "loss": 0.8555,
+      "step": 13564
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.0325114727020264,
+      "learning_rate": 9.074846145807997e-06,
+      "loss": 0.5466,
+      "step": 13565
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.828020453453064,
+      "learning_rate": 9.074693349112694e-06,
+      "loss": 0.6759,
+      "step": 13566
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.8602445125579834,
+      "learning_rate": 9.074540541087192e-06,
+      "loss": 0.9546,
+      "step": 13567
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.9552929401397705,
+      "learning_rate": 9.074387721731921e-06,
+      "loss": 0.5387,
+      "step": 13568
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.5630979537963867,
+      "learning_rate": 9.074234891047302e-06,
+      "loss": 0.8131,
+      "step": 13569
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.4582746028900146,
+      "learning_rate": 9.074082049033764e-06,
+      "loss": 0.7169,
+      "step": 13570
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.1175785064697266,
+      "learning_rate": 9.07392919569173e-06,
+      "loss": 0.5111,
+      "step": 13571
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.069384813308716,
+      "learning_rate": 9.073776331021625e-06,
+      "loss": 0.916,
+      "step": 13572
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.3973236083984375,
+      "learning_rate": 9.073623455023875e-06,
+      "loss": 0.651,
+      "step": 13573
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.420290231704712,
+      "learning_rate": 9.073470567698901e-06,
+      "loss": 0.9532,
+      "step": 13574
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 5.020055294036865,
+      "learning_rate": 9.073317669047134e-06,
+      "loss": 0.8046,
+      "step": 13575
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.154090404510498,
+      "learning_rate": 9.073164759068995e-06,
+      "loss": 0.7206,
+      "step": 13576
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.2843542098999023,
+      "learning_rate": 9.07301183776491e-06,
+      "loss": 0.6157,
+      "step": 13577
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.919222116470337,
+      "learning_rate": 9.072858905135307e-06,
+      "loss": 0.603,
+      "step": 13578
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 4.22925329208374,
+      "learning_rate": 9.07270596118061e-06,
+      "loss": 0.5278,
+      "step": 13579
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.320491075515747,
+      "learning_rate": 9.072553005901241e-06,
+      "loss": 0.5965,
+      "step": 13580
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.1220390796661377,
+      "learning_rate": 9.072400039297629e-06,
+      "loss": 0.7536,
+      "step": 13581
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.4096370935440063,
+      "learning_rate": 9.072247061370198e-06,
+      "loss": 0.4213,
+      "step": 13582
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.3784178495407104,
+      "learning_rate": 9.072094072119373e-06,
+      "loss": 0.5853,
+      "step": 13583
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.7312450408935547,
+      "learning_rate": 9.071941071545577e-06,
+      "loss": 0.5043,
+      "step": 13584
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.3986828327178955,
+      "learning_rate": 9.071788059649243e-06,
+      "loss": 0.6448,
+      "step": 13585
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.272085189819336,
+      "learning_rate": 9.071635036430791e-06,
+      "loss": 0.8056,
+      "step": 13586
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.95550537109375,
+      "learning_rate": 9.071482001890644e-06,
+      "loss": 0.4159,
+      "step": 13587
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.9209731817245483,
+      "learning_rate": 9.071328956029232e-06,
+      "loss": 0.6884,
+      "step": 13588
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.347412347793579,
+      "learning_rate": 9.071175898846981e-06,
+      "loss": 0.418,
+      "step": 13589
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.955533981323242,
+      "learning_rate": 9.071022830344313e-06,
+      "loss": 0.5805,
+      "step": 13590
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.823232650756836,
+      "learning_rate": 9.070869750521658e-06,
+      "loss": 0.6673,
+      "step": 13591
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 4.888542175292969,
+      "learning_rate": 9.070716659379437e-06,
+      "loss": 0.9341,
+      "step": 13592
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.155158281326294,
+      "learning_rate": 9.070563556918079e-06,
+      "loss": 0.6643,
+      "step": 13593
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.621525287628174,
+      "learning_rate": 9.07041044313801e-06,
+      "loss": 0.6311,
+      "step": 13594
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.054607629776001,
+      "learning_rate": 9.070257318039651e-06,
+      "loss": 0.7855,
+      "step": 13595
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.0573782920837402,
+      "learning_rate": 9.070104181623434e-06,
+      "loss": 0.6266,
+      "step": 13596
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.2557461261749268,
+      "learning_rate": 9.06995103388978e-06,
+      "loss": 0.7051,
+      "step": 13597
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.2956557273864746,
+      "learning_rate": 9.069797874839117e-06,
+      "loss": 0.5533,
+      "step": 13598
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.407284736633301,
+      "learning_rate": 9.069644704471872e-06,
+      "loss": 0.6966,
+      "step": 13599
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.0948991775512695,
+      "learning_rate": 9.069491522788468e-06,
+      "loss": 0.6489,
+      "step": 13600
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.248645305633545,
+      "learning_rate": 9.069338329789334e-06,
+      "loss": 0.6492,
+      "step": 13601
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.9176509380340576,
+      "learning_rate": 9.069185125474894e-06,
+      "loss": 0.7618,
+      "step": 13602
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.144188404083252,
+      "learning_rate": 9.069031909845574e-06,
+      "loss": 0.6553,
+      "step": 13603
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.4930496215820312,
+      "learning_rate": 9.068878682901801e-06,
+      "loss": 0.7249,
+      "step": 13604
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.7444247007369995,
+      "learning_rate": 9.068725444644e-06,
+      "loss": 0.6183,
+      "step": 13605
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.0129897594451904,
+      "learning_rate": 9.068572195072597e-06,
+      "loss": 0.7004,
+      "step": 13606
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.454028606414795,
+      "learning_rate": 9.068418934188022e-06,
+      "loss": 0.9126,
+      "step": 13607
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.7638814449310303,
+      "learning_rate": 9.068265661990696e-06,
+      "loss": 0.6288,
+      "step": 13608
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.9928274154663086,
+      "learning_rate": 9.068112378481046e-06,
+      "loss": 0.4874,
+      "step": 13609
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.8129377365112305,
+      "learning_rate": 9.067959083659503e-06,
+      "loss": 0.5736,
+      "step": 13610
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.5888917446136475,
+      "learning_rate": 9.067805777526486e-06,
+      "loss": 0.3614,
+      "step": 13611
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.103649377822876,
+      "learning_rate": 9.067652460082429e-06,
+      "loss": 0.4782,
+      "step": 13612
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.5611937046051025,
+      "learning_rate": 9.06749913132775e-06,
+      "loss": 0.6136,
+      "step": 13613
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.4354817867279053,
+      "learning_rate": 9.06734579126288e-06,
+      "loss": 0.7757,
+      "step": 13614
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.249786138534546,
+      "learning_rate": 9.067192439888248e-06,
+      "loss": 0.6671,
+      "step": 13615
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.6489267349243164,
+      "learning_rate": 9.067039077204276e-06,
+      "loss": 0.7408,
+      "step": 13616
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.247156858444214,
+      "learning_rate": 9.066885703211391e-06,
+      "loss": 0.4189,
+      "step": 13617
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.816559314727783,
+      "learning_rate": 9.06673231791002e-06,
+      "loss": 0.6895,
+      "step": 13618
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.8764283657073975,
+      "learning_rate": 9.066578921300591e-06,
+      "loss": 0.6028,
+      "step": 13619
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.9207963943481445,
+      "learning_rate": 9.06642551338353e-06,
+      "loss": 0.6503,
+      "step": 13620
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.2203822135925293,
+      "learning_rate": 9.066272094159262e-06,
+      "loss": 0.5844,
+      "step": 13621
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.304717779159546,
+      "learning_rate": 9.066118663628215e-06,
+      "loss": 0.6773,
+      "step": 13622
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.0440239906311035,
+      "learning_rate": 9.065965221790815e-06,
+      "loss": 0.6456,
+      "step": 13623
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.001593828201294,
+      "learning_rate": 9.06581176864749e-06,
+      "loss": 0.502,
+      "step": 13624
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.9538497924804688,
+      "learning_rate": 9.065658304198663e-06,
+      "loss": 0.5492,
+      "step": 13625
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.043736696243286,
+      "learning_rate": 9.065504828444766e-06,
+      "loss": 0.9075,
+      "step": 13626
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.939732074737549,
+      "learning_rate": 9.065351341386222e-06,
+      "loss": 0.3311,
+      "step": 13627
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.3908812999725342,
+      "learning_rate": 9.065197843023459e-06,
+      "loss": 0.6416,
+      "step": 13628
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.6743606328964233,
+      "learning_rate": 9.065044333356903e-06,
+      "loss": 0.5635,
+      "step": 13629
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.1474790573120117,
+      "learning_rate": 9.064890812386982e-06,
+      "loss": 0.7986,
+      "step": 13630
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 4.316041946411133,
+      "learning_rate": 9.06473728011412e-06,
+      "loss": 0.7283,
+      "step": 13631
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.1894690990448,
+      "learning_rate": 9.064583736538751e-06,
+      "loss": 0.478,
+      "step": 13632
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.8006844520568848,
+      "learning_rate": 9.064430181661294e-06,
+      "loss": 0.5311,
+      "step": 13633
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.7495739459991455,
+      "learning_rate": 9.06427661548218e-06,
+      "loss": 0.8088,
+      "step": 13634
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.97133469581604,
+      "learning_rate": 9.064123038001835e-06,
+      "loss": 0.7671,
+      "step": 13635
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.9145348072052,
+      "learning_rate": 9.063969449220686e-06,
+      "loss": 0.5223,
+      "step": 13636
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.3819358348846436,
+      "learning_rate": 9.063815849139161e-06,
+      "loss": 0.9794,
+      "step": 13637
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.2240567207336426,
+      "learning_rate": 9.063662237757686e-06,
+      "loss": 0.908,
+      "step": 13638
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 4.47125768661499,
+      "learning_rate": 9.063508615076689e-06,
+      "loss": 0.6756,
+      "step": 13639
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.8932151794433594,
+      "learning_rate": 9.063354981096597e-06,
+      "loss": 0.7596,
+      "step": 13640
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 4.119964122772217,
+      "learning_rate": 9.063201335817836e-06,
+      "loss": 0.6478,
+      "step": 13641
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.8714368343353271,
+      "learning_rate": 9.063047679240834e-06,
+      "loss": 0.8888,
+      "step": 13642
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.263183355331421,
+      "learning_rate": 9.062894011366021e-06,
+      "loss": 0.689,
+      "step": 13643
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.9132182598114014,
+      "learning_rate": 9.06274033219382e-06,
+      "loss": 0.5224,
+      "step": 13644
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.3169660568237305,
+      "learning_rate": 9.06258664172466e-06,
+      "loss": 0.5591,
+      "step": 13645
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.8006457090377808,
+      "learning_rate": 9.062432939958967e-06,
+      "loss": 0.6983,
+      "step": 13646
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.650048017501831,
+      "learning_rate": 9.062279226897172e-06,
+      "loss": 0.6653,
+      "step": 13647
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.69269061088562,
+      "learning_rate": 9.0621255025397e-06,
+      "loss": 0.8365,
+      "step": 13648
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.9880597591400146,
+      "learning_rate": 9.061971766886979e-06,
+      "loss": 0.6658,
+      "step": 13649
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.0036139488220215,
+      "learning_rate": 9.061818019939435e-06,
+      "loss": 0.7189,
+      "step": 13650
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.7926589250564575,
+      "learning_rate": 9.061664261697497e-06,
+      "loss": 0.6587,
+      "step": 13651
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.6893128156661987,
+      "learning_rate": 9.061510492161593e-06,
+      "loss": 0.6619,
+      "step": 13652
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.6186187267303467,
+      "learning_rate": 9.061356711332149e-06,
+      "loss": 0.7294,
+      "step": 13653
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.4773359298706055,
+      "learning_rate": 9.061202919209593e-06,
+      "loss": 0.8952,
+      "step": 13654
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.0315608978271484,
+      "learning_rate": 9.061049115794353e-06,
+      "loss": 0.6943,
+      "step": 13655
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.1212873458862305,
+      "learning_rate": 9.060895301086858e-06,
+      "loss": 0.957,
+      "step": 13656
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.287308692932129,
+      "learning_rate": 9.060741475087533e-06,
+      "loss": 0.6862,
+      "step": 13657
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.3270723819732666,
+      "learning_rate": 9.060587637796809e-06,
+      "loss": 1.0179,
+      "step": 13658
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.3883309364318848,
+      "learning_rate": 9.060433789215111e-06,
+      "loss": 0.7862,
+      "step": 13659
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.0332794189453125,
+      "learning_rate": 9.060279929342867e-06,
+      "loss": 0.5773,
+      "step": 13660
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.526174783706665,
+      "learning_rate": 9.060126058180505e-06,
+      "loss": 0.5272,
+      "step": 13661
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.838343858718872,
+      "learning_rate": 9.059972175728455e-06,
+      "loss": 0.9426,
+      "step": 13662
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 4.654810905456543,
+      "learning_rate": 9.059818281987143e-06,
+      "loss": 0.7778,
+      "step": 13663
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.2979836463928223,
+      "learning_rate": 9.059664376956998e-06,
+      "loss": 0.5027,
+      "step": 13664
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.5589399337768555,
+      "learning_rate": 9.059510460638446e-06,
+      "loss": 0.564,
+      "step": 13665
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.1498520374298096,
+      "learning_rate": 9.059356533031916e-06,
+      "loss": 0.4797,
+      "step": 13666
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.09755539894104,
+      "learning_rate": 9.059202594137835e-06,
+      "loss": 0.4703,
+      "step": 13667
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.7775731086730957,
+      "learning_rate": 9.059048643956634e-06,
+      "loss": 0.6851,
+      "step": 13668
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.756840467453003,
+      "learning_rate": 9.05889468248874e-06,
+      "loss": 0.5087,
+      "step": 13669
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.5405781269073486,
+      "learning_rate": 9.058740709734578e-06,
+      "loss": 0.5705,
+      "step": 13670
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.961988925933838,
+      "learning_rate": 9.05858672569458e-06,
+      "loss": 0.6325,
+      "step": 13671
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.032553195953369,
+      "learning_rate": 9.058432730369172e-06,
+      "loss": 0.7829,
+      "step": 13672
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.7534724473953247,
+      "learning_rate": 9.058278723758783e-06,
+      "loss": 0.6783,
+      "step": 13673
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.628647804260254,
+      "learning_rate": 9.058124705863842e-06,
+      "loss": 0.6134,
+      "step": 13674
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.8160629272460938,
+      "learning_rate": 9.057970676684776e-06,
+      "loss": 0.5505,
+      "step": 13675
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.486321210861206,
+      "learning_rate": 9.057816636222014e-06,
+      "loss": 0.8933,
+      "step": 13676
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.8592801094055176,
+      "learning_rate": 9.057662584475984e-06,
+      "loss": 0.6085,
+      "step": 13677
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.3631579875946045,
+      "learning_rate": 9.057508521447114e-06,
+      "loss": 0.8827,
+      "step": 13678
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.090442657470703,
+      "learning_rate": 9.057354447135831e-06,
+      "loss": 0.374,
+      "step": 13679
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.3266351222991943,
+      "learning_rate": 9.057200361542568e-06,
+      "loss": 0.833,
+      "step": 13680
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.234630823135376,
+      "learning_rate": 9.057046264667749e-06,
+      "loss": 0.5355,
+      "step": 13681
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.8592190742492676,
+      "learning_rate": 9.056892156511805e-06,
+      "loss": 0.6221,
+      "step": 13682
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.237055540084839,
+      "learning_rate": 9.056738037075164e-06,
+      "loss": 0.4814,
+      "step": 13683
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.1394193172454834,
+      "learning_rate": 9.056583906358253e-06,
+      "loss": 0.3977,
+      "step": 13684
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.760971784591675,
+      "learning_rate": 9.056429764361503e-06,
+      "loss": 0.6449,
+      "step": 13685
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.536176919937134,
+      "learning_rate": 9.05627561108534e-06,
+      "loss": 0.3794,
+      "step": 13686
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.2560486793518066,
+      "learning_rate": 9.056121446530195e-06,
+      "loss": 0.5107,
+      "step": 13687
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.37880802154541,
+      "learning_rate": 9.055967270696496e-06,
+      "loss": 0.6722,
+      "step": 13688
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.006776809692383,
+      "learning_rate": 9.05581308358467e-06,
+      "loss": 0.8419,
+      "step": 13689
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.7494263648986816,
+      "learning_rate": 9.055658885195148e-06,
+      "loss": 0.7808,
+      "step": 13690
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.6533586978912354,
+      "learning_rate": 9.05550467552836e-06,
+      "loss": 0.6691,
+      "step": 13691
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.5135014057159424,
+      "learning_rate": 9.055350454584729e-06,
+      "loss": 0.6048,
+      "step": 13692
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.177312135696411,
+      "learning_rate": 9.05519622236469e-06,
+      "loss": 0.7432,
+      "step": 13693
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.5700974464416504,
+      "learning_rate": 9.055041978868668e-06,
+      "loss": 0.5504,
+      "step": 13694
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.8548738956451416,
+      "learning_rate": 9.054887724097094e-06,
+      "loss": 0.7985,
+      "step": 13695
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 4.1736297607421875,
+      "learning_rate": 9.054733458050398e-06,
+      "loss": 0.9204,
+      "step": 13696
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.830343008041382,
+      "learning_rate": 9.054579180729005e-06,
+      "loss": 0.7097,
+      "step": 13697
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.5303982496261597,
+      "learning_rate": 9.054424892133347e-06,
+      "loss": 0.6792,
+      "step": 13698
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.560546398162842,
+      "learning_rate": 9.054270592263852e-06,
+      "loss": 0.49,
+      "step": 13699
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.0979082584381104,
+      "learning_rate": 9.054116281120949e-06,
+      "loss": 0.7239,
+      "step": 13700
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.4206817150115967,
+      "learning_rate": 9.053961958705066e-06,
+      "loss": 0.74,
+      "step": 13701
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.492784023284912,
+      "learning_rate": 9.053807625016635e-06,
+      "loss": 0.4475,
+      "step": 13702
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.8295197486877441,
+      "learning_rate": 9.053653280056083e-06,
+      "loss": 0.6325,
+      "step": 13703
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.9794046878814697,
+      "learning_rate": 9.053498923823841e-06,
+      "loss": 0.6872,
+      "step": 13704
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.604586362838745,
+      "learning_rate": 9.053344556320336e-06,
+      "loss": 0.5665,
+      "step": 13705
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.9498491287231445,
+      "learning_rate": 9.053190177545998e-06,
+      "loss": 0.3852,
+      "step": 13706
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.991424322128296,
+      "learning_rate": 9.053035787501258e-06,
+      "loss": 0.6525,
+      "step": 13707
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.1589748859405518,
+      "learning_rate": 9.052881386186543e-06,
+      "loss": 0.6195,
+      "step": 13708
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.849381685256958,
+      "learning_rate": 9.052726973602282e-06,
+      "loss": 0.9051,
+      "step": 13709
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.534457206726074,
+      "learning_rate": 9.052572549748907e-06,
+      "loss": 0.6988,
+      "step": 13710
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.136777639389038,
+      "learning_rate": 9.052418114626844e-06,
+      "loss": 0.6323,
+      "step": 13711
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.7139946222305298,
+      "learning_rate": 9.052263668236525e-06,
+      "loss": 0.6536,
+      "step": 13712
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.087850570678711,
+      "learning_rate": 9.05210921057838e-06,
+      "loss": 0.631,
+      "step": 13713
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.7987405061721802,
+      "learning_rate": 9.051954741652836e-06,
+      "loss": 0.6174,
+      "step": 13714
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.5699784755706787,
+      "learning_rate": 9.051800261460325e-06,
+      "loss": 0.7776,
+      "step": 13715
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.4515135288238525,
+      "learning_rate": 9.051645770001274e-06,
+      "loss": 0.6381,
+      "step": 13716
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.9632132053375244,
+      "learning_rate": 9.051491267276114e-06,
+      "loss": 0.6739,
+      "step": 13717
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.6155483722686768,
+      "learning_rate": 9.051336753285276e-06,
+      "loss": 0.6487,
+      "step": 13718
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.0344510078430176,
+      "learning_rate": 9.051182228029186e-06,
+      "loss": 0.8983,
+      "step": 13719
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.212958574295044,
+      "learning_rate": 9.051027691508277e-06,
+      "loss": 0.6258,
+      "step": 13720
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.6715129613876343,
+      "learning_rate": 9.050873143722978e-06,
+      "loss": 0.7027,
+      "step": 13721
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.196727991104126,
+      "learning_rate": 9.050718584673717e-06,
+      "loss": 0.5604,
+      "step": 13722
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.1791257858276367,
+      "learning_rate": 9.050564014360924e-06,
+      "loss": 0.9941,
+      "step": 13723
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.3730108737945557,
+      "learning_rate": 9.050409432785034e-06,
+      "loss": 0.7689,
+      "step": 13724
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 4.493328094482422,
+      "learning_rate": 9.050254839946471e-06,
+      "loss": 0.5387,
+      "step": 13725
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.56187105178833,
+      "learning_rate": 9.050100235845665e-06,
+      "loss": 0.519,
+      "step": 13726
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.688053846359253,
+      "learning_rate": 9.049945620483048e-06,
+      "loss": 0.8354,
+      "step": 13727
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.673760175704956,
+      "learning_rate": 9.049790993859048e-06,
+      "loss": 0.8048,
+      "step": 13728
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.717373013496399,
+      "learning_rate": 9.0496363559741e-06,
+      "loss": 0.7456,
+      "step": 13729
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.514664888381958,
+      "learning_rate": 9.049481706828628e-06,
+      "loss": 0.7541,
+      "step": 13730
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.2245819568634033,
+      "learning_rate": 9.049327046423064e-06,
+      "loss": 0.7367,
+      "step": 13731
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.5562567710876465,
+      "learning_rate": 9.04917237475784e-06,
+      "loss": 0.6335,
+      "step": 13732
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.6442129611968994,
+      "learning_rate": 9.049017691833383e-06,
+      "loss": 0.6666,
+      "step": 13733
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.785679817199707,
+      "learning_rate": 9.048862997650125e-06,
+      "loss": 0.7827,
+      "step": 13734
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 5.276504039764404,
+      "learning_rate": 9.048708292208495e-06,
+      "loss": 0.571,
+      "step": 13735
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.6014482975006104,
+      "learning_rate": 9.048553575508924e-06,
+      "loss": 0.63,
+      "step": 13736
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.112997531890869,
+      "learning_rate": 9.048398847551846e-06,
+      "loss": 0.6805,
+      "step": 13737
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.1457579135894775,
+      "learning_rate": 9.048244108337682e-06,
+      "loss": 0.7012,
+      "step": 13738
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.0514135360717773,
+      "learning_rate": 9.04808935786687e-06,
+      "loss": 0.6194,
+      "step": 13739
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.302279472351074,
+      "learning_rate": 9.047934596139838e-06,
+      "loss": 0.6624,
+      "step": 13740
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.2682716846466064,
+      "learning_rate": 9.047779823157017e-06,
+      "loss": 0.711,
+      "step": 13741
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.6537089347839355,
+      "learning_rate": 9.047625038918834e-06,
+      "loss": 0.8283,
+      "step": 13742
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.2991881370544434,
+      "learning_rate": 9.047470243425725e-06,
+      "loss": 0.8722,
+      "step": 13743
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.194399118423462,
+      "learning_rate": 9.047315436678117e-06,
+      "loss": 0.7517,
+      "step": 13744
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.2789621353149414,
+      "learning_rate": 9.04716061867644e-06,
+      "loss": 0.8483,
+      "step": 13745
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.6680790185928345,
+      "learning_rate": 9.047005789421124e-06,
+      "loss": 0.6784,
+      "step": 13746
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.538088321685791,
+      "learning_rate": 9.046850948912604e-06,
+      "loss": 0.8575,
+      "step": 13747
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.8985605239868164,
+      "learning_rate": 9.046696097151306e-06,
+      "loss": 0.7596,
+      "step": 13748
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.040586471557617,
+      "learning_rate": 9.046541234137663e-06,
+      "loss": 0.6373,
+      "step": 13749
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.4277617931365967,
+      "learning_rate": 9.046386359872104e-06,
+      "loss": 0.7085,
+      "step": 13750
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.850901484489441,
+      "learning_rate": 9.046231474355059e-06,
+      "loss": 0.7058,
+      "step": 13751
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.875288963317871,
+      "learning_rate": 9.046076577586961e-06,
+      "loss": 0.8235,
+      "step": 13752
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.373767137527466,
+      "learning_rate": 9.04592166956824e-06,
+      "loss": 0.5481,
+      "step": 13753
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.6719861030578613,
+      "learning_rate": 9.045766750299328e-06,
+      "loss": 0.6375,
+      "step": 13754
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.5190539360046387,
+      "learning_rate": 9.04561181978065e-06,
+      "loss": 0.9269,
+      "step": 13755
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.825394868850708,
+      "learning_rate": 9.045456878012645e-06,
+      "loss": 0.4598,
+      "step": 13756
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.8021585941314697,
+      "learning_rate": 9.045301924995739e-06,
+      "loss": 0.3554,
+      "step": 13757
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.2473175525665283,
+      "learning_rate": 9.045146960730363e-06,
+      "loss": 0.6086,
+      "step": 13758
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.9876456260681152,
+      "learning_rate": 9.04499198521695e-06,
+      "loss": 0.6994,
+      "step": 13759
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.5088460445404053,
+      "learning_rate": 9.044836998455928e-06,
+      "loss": 0.4567,
+      "step": 13760
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.816037654876709,
+      "learning_rate": 9.04468200044773e-06,
+      "loss": 0.7131,
+      "step": 13761
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.2252306938171387,
+      "learning_rate": 9.044526991192785e-06,
+      "loss": 0.6943,
+      "step": 13762
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.454848527908325,
+      "learning_rate": 9.044371970691528e-06,
+      "loss": 0.4737,
+      "step": 13763
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.8161442279815674,
+      "learning_rate": 9.044216938944388e-06,
+      "loss": 0.766,
+      "step": 13764
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.6238653659820557,
+      "learning_rate": 9.044061895951794e-06,
+      "loss": 0.7504,
+      "step": 13765
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.9996391534805298,
+      "learning_rate": 9.04390684171418e-06,
+      "loss": 0.7661,
+      "step": 13766
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.3647470474243164,
+      "learning_rate": 9.043751776231974e-06,
+      "loss": 0.6397,
+      "step": 13767
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.9252853393554688,
+      "learning_rate": 9.04359669950561e-06,
+      "loss": 0.6348,
+      "step": 13768
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.874582290649414,
+      "learning_rate": 9.043441611535519e-06,
+      "loss": 0.7579,
+      "step": 13769
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.2808940410614014,
+      "learning_rate": 9.043286512322132e-06,
+      "loss": 0.5935,
+      "step": 13770
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.1149682998657227,
+      "learning_rate": 9.04313140186588e-06,
+      "loss": 0.592,
+      "step": 13771
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.9357961416244507,
+      "learning_rate": 9.042976280167193e-06,
+      "loss": 0.7405,
+      "step": 13772
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.801463007926941,
+      "learning_rate": 9.042821147226503e-06,
+      "loss": 0.5652,
+      "step": 13773
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.4656596183776855,
+      "learning_rate": 9.042666003044243e-06,
+      "loss": 0.5149,
+      "step": 13774
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.249938726425171,
+      "learning_rate": 9.042510847620843e-06,
+      "loss": 0.8825,
+      "step": 13775
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.1036980152130127,
+      "learning_rate": 9.042355680956734e-06,
+      "loss": 0.8427,
+      "step": 13776
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.2570292949676514,
+      "learning_rate": 9.04220050305235e-06,
+      "loss": 0.4359,
+      "step": 13777
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.4265012741088867,
+      "learning_rate": 9.042045313908119e-06,
+      "loss": 0.6542,
+      "step": 13778
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.6305232048034668,
+      "learning_rate": 9.041890113524473e-06,
+      "loss": 0.5456,
+      "step": 13779
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.4002492427825928,
+      "learning_rate": 9.04173490190185e-06,
+      "loss": 0.8396,
+      "step": 13780
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.5964921712875366,
+      "learning_rate": 9.04157967904067e-06,
+      "loss": 0.6845,
+      "step": 13781
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.6101927757263184,
+      "learning_rate": 9.041424444941374e-06,
+      "loss": 0.6317,
+      "step": 13782
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.4191536903381348,
+      "learning_rate": 9.04126919960439e-06,
+      "loss": 0.7003,
+      "step": 13783
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.6437675952911377,
+      "learning_rate": 9.04111394303015e-06,
+      "loss": 0.4762,
+      "step": 13784
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 4.5255231857299805,
+      "learning_rate": 9.040958675219087e-06,
+      "loss": 0.8859,
+      "step": 13785
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.899646282196045,
+      "learning_rate": 9.04080339617163e-06,
+      "loss": 0.6037,
+      "step": 13786
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.1222660541534424,
+      "learning_rate": 9.040648105888215e-06,
+      "loss": 0.7345,
+      "step": 13787
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.7156190872192383,
+      "learning_rate": 9.040492804369269e-06,
+      "loss": 0.625,
+      "step": 13788
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.5101006031036377,
+      "learning_rate": 9.040337491615227e-06,
+      "loss": 0.8608,
+      "step": 13789
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.5178563594818115,
+      "learning_rate": 9.04018216762652e-06,
+      "loss": 0.6572,
+      "step": 13790
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.180541515350342,
+      "learning_rate": 9.04002683240358e-06,
+      "loss": 0.6621,
+      "step": 13791
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.6646485328674316,
+      "learning_rate": 9.039871485946839e-06,
+      "loss": 0.9901,
+      "step": 13792
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.3168137073516846,
+      "learning_rate": 9.039716128256727e-06,
+      "loss": 0.5192,
+      "step": 13793
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.3062381744384766,
+      "learning_rate": 9.039560759333681e-06,
+      "loss": 0.7679,
+      "step": 13794
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.2043378353118896,
+      "learning_rate": 9.039405379178128e-06,
+      "loss": 0.6205,
+      "step": 13795
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.671459436416626,
+      "learning_rate": 9.039249987790501e-06,
+      "loss": 0.452,
+      "step": 13796
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.705578327178955,
+      "learning_rate": 9.039094585171233e-06,
+      "loss": 0.6192,
+      "step": 13797
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.9783036708831787,
+      "learning_rate": 9.038939171320758e-06,
+      "loss": 0.6779,
+      "step": 13798
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.429363250732422,
+      "learning_rate": 9.038783746239505e-06,
+      "loss": 0.62,
+      "step": 13799
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.4925975799560547,
+      "learning_rate": 9.038628309927905e-06,
+      "loss": 0.4932,
+      "step": 13800
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.9054009914398193,
+      "learning_rate": 9.038472862386395e-06,
+      "loss": 0.6899,
+      "step": 13801
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.3641533851623535,
+      "learning_rate": 9.038317403615407e-06,
+      "loss": 0.6888,
+      "step": 13802
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.804718255996704,
+      "learning_rate": 9.038161933615369e-06,
+      "loss": 0.423,
+      "step": 13803
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 4.2273406982421875,
+      "learning_rate": 9.038006452386714e-06,
+      "loss": 0.8231,
+      "step": 13804
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.970745325088501,
+      "learning_rate": 9.037850959929876e-06,
+      "loss": 0.7139,
+      "step": 13805
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.9408729076385498,
+      "learning_rate": 9.03769545624529e-06,
+      "loss": 0.6112,
+      "step": 13806
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.274662494659424,
+      "learning_rate": 9.037539941333382e-06,
+      "loss": 0.7422,
+      "step": 13807
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.8322948217391968,
+      "learning_rate": 9.03738441519459e-06,
+      "loss": 0.7345,
+      "step": 13808
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.759078025817871,
+      "learning_rate": 9.037228877829343e-06,
+      "loss": 0.6741,
+      "step": 13809
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.9904122352600098,
+      "learning_rate": 9.037073329238076e-06,
+      "loss": 0.8779,
+      "step": 13810
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.8646774291992188,
+      "learning_rate": 9.03691776942122e-06,
+      "loss": 0.5607,
+      "step": 13811
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 5.0012006759643555,
+      "learning_rate": 9.036762198379208e-06,
+      "loss": 0.7308,
+      "step": 13812
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.9582897424697876,
+      "learning_rate": 9.036606616112473e-06,
+      "loss": 0.5055,
+      "step": 13813
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.4718849658966064,
+      "learning_rate": 9.036451022621446e-06,
+      "loss": 0.7043,
+      "step": 13814
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.1511645317077637,
+      "learning_rate": 9.036295417906562e-06,
+      "loss": 0.5337,
+      "step": 13815
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.2933990955352783,
+      "learning_rate": 9.036139801968251e-06,
+      "loss": 0.6305,
+      "step": 13816
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.483157157897949,
+      "learning_rate": 9.035984174806949e-06,
+      "loss": 0.4687,
+      "step": 13817
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.8487614393234253,
+      "learning_rate": 9.035828536423086e-06,
+      "loss": 0.4636,
+      "step": 13818
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.909360885620117,
+      "learning_rate": 9.035672886817094e-06,
+      "loss": 0.7157,
+      "step": 13819
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.5914852619171143,
+      "learning_rate": 9.03551722598941e-06,
+      "loss": 0.6961,
+      "step": 13820
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.1742703914642334,
+      "learning_rate": 9.035361553940465e-06,
+      "loss": 0.485,
+      "step": 13821
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.392742395401001,
+      "learning_rate": 9.035205870670689e-06,
+      "loss": 0.5945,
+      "step": 13822
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.879354238510132,
+      "learning_rate": 9.035050176180518e-06,
+      "loss": 0.6537,
+      "step": 13823
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.716035842895508,
+      "learning_rate": 9.034894470470384e-06,
+      "loss": 0.7061,
+      "step": 13824
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.4149012565612793,
+      "learning_rate": 9.034738753540719e-06,
+      "loss": 0.9059,
+      "step": 13825
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.705479383468628,
+      "learning_rate": 9.034583025391959e-06,
+      "loss": 0.8073,
+      "step": 13826
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.2589833736419678,
+      "learning_rate": 9.034427286024534e-06,
+      "loss": 0.904,
+      "step": 13827
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.739731550216675,
+      "learning_rate": 9.034271535438877e-06,
+      "loss": 0.627,
+      "step": 13828
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.8652892112731934,
+      "learning_rate": 9.034115773635423e-06,
+      "loss": 0.6359,
+      "step": 13829
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.708111524581909,
+      "learning_rate": 9.033960000614603e-06,
+      "loss": 0.7,
+      "step": 13830
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.3945603370666504,
+      "learning_rate": 9.033804216376854e-06,
+      "loss": 0.8771,
+      "step": 13831
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.8960227966308594,
+      "learning_rate": 9.033648420922605e-06,
+      "loss": 0.8692,
+      "step": 13832
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.1482620239257812,
+      "learning_rate": 9.03349261425229e-06,
+      "loss": 0.5994,
+      "step": 13833
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.9800450801849365,
+      "learning_rate": 9.033336796366346e-06,
+      "loss": 0.6467,
+      "step": 13834
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 4.258978366851807,
+      "learning_rate": 9.0331809672652e-06,
+      "loss": 0.5957,
+      "step": 13835
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 8.416656494140625,
+      "learning_rate": 9.033025126949291e-06,
+      "loss": 0.9508,
+      "step": 13836
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.6262885332107544,
+      "learning_rate": 9.032869275419048e-06,
+      "loss": 0.4178,
+      "step": 13837
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.4930362701416016,
+      "learning_rate": 9.032713412674907e-06,
+      "loss": 0.7218,
+      "step": 13838
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.4628422260284424,
+      "learning_rate": 9.032557538717301e-06,
+      "loss": 0.6381,
+      "step": 13839
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.9315388202667236,
+      "learning_rate": 9.032401653546662e-06,
+      "loss": 0.7246,
+      "step": 13840
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.108999013900757,
+      "learning_rate": 9.032245757163427e-06,
+      "loss": 0.6958,
+      "step": 13841
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.7463459968566895,
+      "learning_rate": 9.032089849568025e-06,
+      "loss": 0.7944,
+      "step": 13842
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 5.303126811981201,
+      "learning_rate": 9.031933930760893e-06,
+      "loss": 0.7064,
+      "step": 13843
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.9244740009307861,
+      "learning_rate": 9.031778000742462e-06,
+      "loss": 0.4172,
+      "step": 13844
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.0923852920532227,
+      "learning_rate": 9.031622059513167e-06,
+      "loss": 0.6117,
+      "step": 13845
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.3868651390075684,
+      "learning_rate": 9.031466107073442e-06,
+      "loss": 0.6236,
+      "step": 13846
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.4240612983703613,
+      "learning_rate": 9.031310143423718e-06,
+      "loss": 0.618,
+      "step": 13847
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.3336498737335205,
+      "learning_rate": 9.03115416856443e-06,
+      "loss": 0.6674,
+      "step": 13848
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.5523370504379272,
+      "learning_rate": 9.030998182496017e-06,
+      "loss": 0.7747,
+      "step": 13849
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 4.62847375869751,
+      "learning_rate": 9.030842185218904e-06,
+      "loss": 0.6427,
+      "step": 13850
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.8178210258483887,
+      "learning_rate": 9.030686176733533e-06,
+      "loss": 0.7216,
+      "step": 13851
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.0199809074401855,
+      "learning_rate": 9.03053015704033e-06,
+      "loss": 0.6997,
+      "step": 13852
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.790701389312744,
+      "learning_rate": 9.030374126139731e-06,
+      "loss": 0.7427,
+      "step": 13853
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.8690261840820312,
+      "learning_rate": 9.030218084032175e-06,
+      "loss": 0.4335,
+      "step": 13854
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.5908493995666504,
+      "learning_rate": 9.030062030718092e-06,
+      "loss": 0.5733,
+      "step": 13855
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.7220702171325684,
+      "learning_rate": 9.029905966197913e-06,
+      "loss": 0.6455,
+      "step": 13856
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.6406795978546143,
+      "learning_rate": 9.02974989047208e-06,
+      "loss": 0.4609,
+      "step": 13857
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.6652615070343018,
+      "learning_rate": 9.029593803541019e-06,
+      "loss": 0.7291,
+      "step": 13858
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.5861337184906006,
+      "learning_rate": 9.029437705405166e-06,
+      "loss": 0.789,
+      "step": 13859
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.1343884468078613,
+      "learning_rate": 9.029281596064959e-06,
+      "loss": 0.6809,
+      "step": 13860
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.6421077251434326,
+      "learning_rate": 9.029125475520829e-06,
+      "loss": 0.7159,
+      "step": 13861
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 4.50379753112793,
+      "learning_rate": 9.028969343773209e-06,
+      "loss": 0.7759,
+      "step": 13862
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.1666100025177,
+      "learning_rate": 9.028813200822535e-06,
+      "loss": 0.6048,
+      "step": 13863
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.83952260017395,
+      "learning_rate": 9.02865704666924e-06,
+      "loss": 0.8311,
+      "step": 13864
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.9007115364074707,
+      "learning_rate": 9.028500881313761e-06,
+      "loss": 0.5979,
+      "step": 13865
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.653686046600342,
+      "learning_rate": 9.028344704756528e-06,
+      "loss": 0.7026,
+      "step": 13866
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.7670931816101074,
+      "learning_rate": 9.02818851699798e-06,
+      "loss": 0.7998,
+      "step": 13867
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 4.479480743408203,
+      "learning_rate": 9.028032318038547e-06,
+      "loss": 0.7115,
+      "step": 13868
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.0221309661865234,
+      "learning_rate": 9.027876107878663e-06,
+      "loss": 0.6224,
+      "step": 13869
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.931692123413086,
+      "learning_rate": 9.027719886518769e-06,
+      "loss": 0.8018,
+      "step": 13870
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.3698484897613525,
+      "learning_rate": 9.027563653959292e-06,
+      "loss": 0.6475,
+      "step": 13871
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.737946033477783,
+      "learning_rate": 9.027407410200669e-06,
+      "loss": 0.6355,
+      "step": 13872
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.566089630126953,
+      "learning_rate": 9.027251155243335e-06,
+      "loss": 0.7409,
+      "step": 13873
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.115764856338501,
+      "learning_rate": 9.027094889087724e-06,
+      "loss": 0.4967,
+      "step": 13874
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.3704915046691895,
+      "learning_rate": 9.026938611734273e-06,
+      "loss": 0.6025,
+      "step": 13875
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.1387221813201904,
+      "learning_rate": 9.026782323183412e-06,
+      "loss": 0.6682,
+      "step": 13876
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.018892288208008,
+      "learning_rate": 9.02662602343558e-06,
+      "loss": 0.9051,
+      "step": 13877
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.747286319732666,
+      "learning_rate": 9.026469712491206e-06,
+      "loss": 0.6503,
+      "step": 13878
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.736264705657959,
+      "learning_rate": 9.02631339035073e-06,
+      "loss": 0.8507,
+      "step": 13879
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.251472234725952,
+      "learning_rate": 9.026157057014585e-06,
+      "loss": 0.5742,
+      "step": 13880
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 4.514065742492676,
+      "learning_rate": 9.026000712483203e-06,
+      "loss": 0.7384,
+      "step": 13881
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.2353551387786865,
+      "learning_rate": 9.025844356757024e-06,
+      "loss": 0.5347,
+      "step": 13882
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.7179635763168335,
+      "learning_rate": 9.025687989836478e-06,
+      "loss": 0.5834,
+      "step": 13883
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.9082448482513428,
+      "learning_rate": 9.025531611722006e-06,
+      "loss": 0.7099,
+      "step": 13884
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.559715509414673,
+      "learning_rate": 9.025375222414035e-06,
+      "loss": 0.7871,
+      "step": 13885
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.848430871963501,
+      "learning_rate": 9.025218821913002e-06,
+      "loss": 0.7579,
+      "step": 13886
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.3391435146331787,
+      "learning_rate": 9.025062410219346e-06,
+      "loss": 0.7572,
+      "step": 13887
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 4.26291036605835,
+      "learning_rate": 9.024905987333498e-06,
+      "loss": 1.0438,
+      "step": 13888
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.8455026149749756,
+      "learning_rate": 9.024749553255894e-06,
+      "loss": 0.508,
+      "step": 13889
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.8969457149505615,
+      "learning_rate": 9.024593107986971e-06,
+      "loss": 0.711,
+      "step": 13890
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.7670338153839111,
+      "learning_rate": 9.024436651527162e-06,
+      "loss": 0.6373,
+      "step": 13891
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.680068254470825,
+      "learning_rate": 9.024280183876902e-06,
+      "loss": 0.4733,
+      "step": 13892
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.4822840690612793,
+      "learning_rate": 9.024123705036625e-06,
+      "loss": 0.5674,
+      "step": 13893
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.068934202194214,
+      "learning_rate": 9.023967215006767e-06,
+      "loss": 0.6803,
+      "step": 13894
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.516932725906372,
+      "learning_rate": 9.023810713787767e-06,
+      "loss": 0.3391,
+      "step": 13895
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 4.637246608734131,
+      "learning_rate": 9.023654201380053e-06,
+      "loss": 0.4089,
+      "step": 13896
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.5599631071090698,
+      "learning_rate": 9.023497677784066e-06,
+      "loss": 0.4187,
+      "step": 13897
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.7053542137145996,
+      "learning_rate": 9.02334114300024e-06,
+      "loss": 0.7171,
+      "step": 13898
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.6527899503707886,
+      "learning_rate": 9.023184597029006e-06,
+      "loss": 0.5887,
+      "step": 13899
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 5.583285808563232,
+      "learning_rate": 9.023028039870805e-06,
+      "loss": 0.8843,
+      "step": 13900
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.024082899093628,
+      "learning_rate": 9.022871471526071e-06,
+      "loss": 0.5402,
+      "step": 13901
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.742111325263977,
+      "learning_rate": 9.022714891995237e-06,
+      "loss": 0.6941,
+      "step": 13902
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.6914026737213135,
+      "learning_rate": 9.02255830127874e-06,
+      "loss": 0.9732,
+      "step": 13903
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.2486422061920166,
+      "learning_rate": 9.022401699377017e-06,
+      "loss": 0.6881,
+      "step": 13904
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.7231922149658203,
+      "learning_rate": 9.0222450862905e-06,
+      "loss": 0.7566,
+      "step": 13905
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 4.618488311767578,
+      "learning_rate": 9.022088462019625e-06,
+      "loss": 0.7123,
+      "step": 13906
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.5367939472198486,
+      "learning_rate": 9.021931826564832e-06,
+      "loss": 0.4522,
+      "step": 13907
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.4551854133605957,
+      "learning_rate": 9.021775179926551e-06,
+      "loss": 0.5627,
+      "step": 13908
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.8343448638916016,
+      "learning_rate": 9.02161852210522e-06,
+      "loss": 0.4796,
+      "step": 13909
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.214679718017578,
+      "learning_rate": 9.021461853101274e-06,
+      "loss": 0.9751,
+      "step": 13910
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.4323734045028687,
+      "learning_rate": 9.021305172915149e-06,
+      "loss": 0.6185,
+      "step": 13911
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.41225266456604,
+      "learning_rate": 9.021148481547282e-06,
+      "loss": 0.5886,
+      "step": 13912
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.7209110260009766,
+      "learning_rate": 9.020991778998106e-06,
+      "loss": 0.6786,
+      "step": 13913
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.8386682271957397,
+      "learning_rate": 9.02083506526806e-06,
+      "loss": 0.5936,
+      "step": 13914
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.6399924755096436,
+      "learning_rate": 9.020678340357575e-06,
+      "loss": 0.6108,
+      "step": 13915
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.193732738494873,
+      "learning_rate": 9.020521604267094e-06,
+      "loss": 0.4968,
+      "step": 13916
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.2237486839294434,
+      "learning_rate": 9.020364856997046e-06,
+      "loss": 0.6494,
+      "step": 13917
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.085707902908325,
+      "learning_rate": 9.02020809854787e-06,
+      "loss": 0.798,
+      "step": 13918
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.9193321466445923,
+      "learning_rate": 9.02005132892e-06,
+      "loss": 0.7476,
+      "step": 13919
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.6169471740722656,
+      "learning_rate": 9.019894548113873e-06,
+      "loss": 0.6214,
+      "step": 13920
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.5088424682617188,
+      "learning_rate": 9.019737756129928e-06,
+      "loss": 0.7404,
+      "step": 13921
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 4.770052909851074,
+      "learning_rate": 9.019580952968594e-06,
+      "loss": 0.6361,
+      "step": 13922
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.571316719055176,
+      "learning_rate": 9.019424138630314e-06,
+      "loss": 0.657,
+      "step": 13923
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.452519178390503,
+      "learning_rate": 9.01926731311552e-06,
+      "loss": 0.5753,
+      "step": 13924
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.4950647354125977,
+      "learning_rate": 9.01911047642465e-06,
+      "loss": 0.5486,
+      "step": 13925
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.1367275714874268,
+      "learning_rate": 9.01895362855814e-06,
+      "loss": 0.8902,
+      "step": 13926
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.5855441093444824,
+      "learning_rate": 9.018796769516425e-06,
+      "loss": 0.635,
+      "step": 13927
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.2744944095611572,
+      "learning_rate": 9.01863989929994e-06,
+      "loss": 0.6634,
+      "step": 13928
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.4053630828857422,
+      "learning_rate": 9.018483017909126e-06,
+      "loss": 0.5551,
+      "step": 13929
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.256829261779785,
+      "learning_rate": 9.018326125344414e-06,
+      "loss": 0.6676,
+      "step": 13930
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.8774040937423706,
+      "learning_rate": 9.018169221606243e-06,
+      "loss": 0.6906,
+      "step": 13931
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.859837532043457,
+      "learning_rate": 9.018012306695048e-06,
+      "loss": 0.6068,
+      "step": 13932
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.3244271278381348,
+      "learning_rate": 9.017855380611267e-06,
+      "loss": 0.557,
+      "step": 13933
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.8652987480163574,
+      "learning_rate": 9.017698443355333e-06,
+      "loss": 0.5075,
+      "step": 13934
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.7755135297775269,
+      "learning_rate": 9.017541494927687e-06,
+      "loss": 0.5659,
+      "step": 13935
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.2958896160125732,
+      "learning_rate": 9.017384535328762e-06,
+      "loss": 0.9173,
+      "step": 13936
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.9583890438079834,
+      "learning_rate": 9.017227564558997e-06,
+      "loss": 0.6937,
+      "step": 13937
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.976123332977295,
+      "learning_rate": 9.017070582618826e-06,
+      "loss": 0.6238,
+      "step": 13938
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 5.057216167449951,
+      "learning_rate": 9.016913589508687e-06,
+      "loss": 0.5232,
+      "step": 13939
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.4735636711120605,
+      "learning_rate": 9.016756585229013e-06,
+      "loss": 0.7234,
+      "step": 13940
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.4331798553466797,
+      "learning_rate": 9.01659956978025e-06,
+      "loss": 0.9101,
+      "step": 13941
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.0491738319396973,
+      "learning_rate": 9.016442543162822e-06,
+      "loss": 0.3978,
+      "step": 13942
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.0652809143066406,
+      "learning_rate": 9.016285505377176e-06,
+      "loss": 0.4672,
+      "step": 13943
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.7820687294006348,
+      "learning_rate": 9.016128456423742e-06,
+      "loss": 0.6034,
+      "step": 13944
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.7488279342651367,
+      "learning_rate": 9.01597139630296e-06,
+      "loss": 0.7723,
+      "step": 13945
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.9045063257217407,
+      "learning_rate": 9.015814325015266e-06,
+      "loss": 0.7869,
+      "step": 13946
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.9428632259368896,
+      "learning_rate": 9.015657242561098e-06,
+      "loss": 0.5058,
+      "step": 13947
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.4812915325164795,
+      "learning_rate": 9.01550014894089e-06,
+      "loss": 0.6199,
+      "step": 13948
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.9339845180511475,
+      "learning_rate": 9.01534304415508e-06,
+      "loss": 0.7853,
+      "step": 13949
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 4.335233211517334,
+      "learning_rate": 9.015185928204106e-06,
+      "loss": 0.5567,
+      "step": 13950
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.0806634426116943,
+      "learning_rate": 9.015028801088404e-06,
+      "loss": 0.7169,
+      "step": 13951
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.88250732421875,
+      "learning_rate": 9.01487166280841e-06,
+      "loss": 0.5772,
+      "step": 13952
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.4476687908172607,
+      "learning_rate": 9.014714513364564e-06,
+      "loss": 0.6138,
+      "step": 13953
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.60146164894104,
+      "learning_rate": 9.014557352757299e-06,
+      "loss": 0.6101,
+      "step": 13954
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.7700178623199463,
+      "learning_rate": 9.014400180987056e-06,
+      "loss": 0.638,
+      "step": 13955
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.5694525241851807,
+      "learning_rate": 9.014242998054267e-06,
+      "loss": 0.6527,
+      "step": 13956
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.7521231174468994,
+      "learning_rate": 9.014085803959373e-06,
+      "loss": 0.4103,
+      "step": 13957
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.2067677974700928,
+      "learning_rate": 9.013928598702811e-06,
+      "loss": 0.4416,
+      "step": 13958
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.7155132293701172,
+      "learning_rate": 9.013771382285018e-06,
+      "loss": 0.8569,
+      "step": 13959
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.514894962310791,
+      "learning_rate": 9.013614154706427e-06,
+      "loss": 0.8279,
+      "step": 13960
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 7.122725486755371,
+      "learning_rate": 9.013456915967483e-06,
+      "loss": 0.9701,
+      "step": 13961
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.578472375869751,
+      "learning_rate": 9.013299666068616e-06,
+      "loss": 0.7243,
+      "step": 13962
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.9523894786834717,
+      "learning_rate": 9.013142405010267e-06,
+      "loss": 0.722,
+      "step": 13963
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.4202704429626465,
+      "learning_rate": 9.012985132792871e-06,
+      "loss": 0.6508,
+      "step": 13964
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.914508581161499,
+      "learning_rate": 9.012827849416867e-06,
+      "loss": 0.6217,
+      "step": 13965
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.6350955963134766,
+      "learning_rate": 9.012670554882692e-06,
+      "loss": 0.62,
+      "step": 13966
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.8298826217651367,
+      "learning_rate": 9.012513249190784e-06,
+      "loss": 0.5808,
+      "step": 13967
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.1208043098449707,
+      "learning_rate": 9.012355932341581e-06,
+      "loss": 0.7879,
+      "step": 13968
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.080404758453369,
+      "learning_rate": 9.012198604335519e-06,
+      "loss": 0.4476,
+      "step": 13969
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.3826167583465576,
+      "learning_rate": 9.012041265173032e-06,
+      "loss": 0.7944,
+      "step": 13970
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.0236098766326904,
+      "learning_rate": 9.011883914854565e-06,
+      "loss": 0.6794,
+      "step": 13971
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.507648468017578,
+      "learning_rate": 9.011726553380551e-06,
+      "loss": 0.9427,
+      "step": 13972
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.0598323345184326,
+      "learning_rate": 9.011569180751427e-06,
+      "loss": 0.9316,
+      "step": 13973
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.099688768386841,
+      "learning_rate": 9.011411796967633e-06,
+      "loss": 0.7131,
+      "step": 13974
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.2596561908721924,
+      "learning_rate": 9.011254402029605e-06,
+      "loss": 0.8383,
+      "step": 13975
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.7497291564941406,
+      "learning_rate": 9.01109699593778e-06,
+      "loss": 0.6626,
+      "step": 13976
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.8783137798309326,
+      "learning_rate": 9.010939578692598e-06,
+      "loss": 0.7115,
+      "step": 13977
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.6507575511932373,
+      "learning_rate": 9.010782150294496e-06,
+      "loss": 0.7042,
+      "step": 13978
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.6183481216430664,
+      "learning_rate": 9.010624710743912e-06,
+      "loss": 0.6755,
+      "step": 13979
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.8333730697631836,
+      "learning_rate": 9.010467260041282e-06,
+      "loss": 0.8414,
+      "step": 13980
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.303769111633301,
+      "learning_rate": 9.010309798187044e-06,
+      "loss": 0.5181,
+      "step": 13981
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.6846632957458496,
+      "learning_rate": 9.010152325181638e-06,
+      "loss": 0.7528,
+      "step": 13982
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.833099842071533,
+      "learning_rate": 9.009994841025502e-06,
+      "loss": 0.602,
+      "step": 13983
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.94589900970459,
+      "learning_rate": 9.00983734571907e-06,
+      "loss": 0.7219,
+      "step": 13984
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.1146469116210938,
+      "learning_rate": 9.009679839262784e-06,
+      "loss": 0.8426,
+      "step": 13985
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.7571403980255127,
+      "learning_rate": 9.00952232165708e-06,
+      "loss": 0.4978,
+      "step": 13986
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.720874547958374,
+      "learning_rate": 9.009364792902396e-06,
+      "loss": 0.5011,
+      "step": 13987
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.471693277359009,
+      "learning_rate": 9.00920725299917e-06,
+      "loss": 0.7657,
+      "step": 13988
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.1352598667144775,
+      "learning_rate": 9.00904970194784e-06,
+      "loss": 0.6571,
+      "step": 13989
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.361945390701294,
+      "learning_rate": 9.008892139748846e-06,
+      "loss": 0.807,
+      "step": 13990
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.241363286972046,
+      "learning_rate": 9.008734566402626e-06,
+      "loss": 0.476,
+      "step": 13991
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.6141750812530518,
+      "learning_rate": 9.008576981909615e-06,
+      "loss": 0.6785,
+      "step": 13992
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 5.987361431121826,
+      "learning_rate": 9.008419386270253e-06,
+      "loss": 0.5551,
+      "step": 13993
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.927051305770874,
+      "learning_rate": 9.00826177948498e-06,
+      "loss": 0.7186,
+      "step": 13994
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.149219036102295,
+      "learning_rate": 9.00810416155423e-06,
+      "loss": 0.5983,
+      "step": 13995
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.1301631927490234,
+      "learning_rate": 9.007946532478446e-06,
+      "loss": 0.6855,
+      "step": 13996
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.2106194496154785,
+      "learning_rate": 9.007788892258063e-06,
+      "loss": 0.7444,
+      "step": 13997
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.506685256958008,
+      "learning_rate": 9.00763124089352e-06,
+      "loss": 0.7133,
+      "step": 13998
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.538114070892334,
+      "learning_rate": 9.007473578385258e-06,
+      "loss": 0.6625,
+      "step": 13999
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.3563313484191895,
+      "learning_rate": 9.007315904733711e-06,
+      "loss": 0.4449,
+      "step": 14000
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.851022720336914,
+      "learning_rate": 9.00715821993932e-06,
+      "loss": 0.5786,
+      "step": 14001
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.9221205711364746,
+      "learning_rate": 9.007000524002523e-06,
+      "loss": 0.6114,
+      "step": 14002
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.3475840091705322,
+      "learning_rate": 9.00684281692376e-06,
+      "loss": 1.0702,
+      "step": 14003
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.2841455936431885,
+      "learning_rate": 9.006685098703467e-06,
+      "loss": 0.6009,
+      "step": 14004
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.8078806400299072,
+      "learning_rate": 9.006527369342082e-06,
+      "loss": 0.5029,
+      "step": 14005
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.036618947982788,
+      "learning_rate": 9.006369628840048e-06,
+      "loss": 0.87,
+      "step": 14006
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 22.80233383178711,
+      "learning_rate": 9.0062118771978e-06,
+      "loss": 0.5923,
+      "step": 14007
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.627448320388794,
+      "learning_rate": 9.006054114415778e-06,
+      "loss": 0.4651,
+      "step": 14008
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.7376842498779297,
+      "learning_rate": 9.00589634049442e-06,
+      "loss": 0.6104,
+      "step": 14009
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.68856680393219,
+      "learning_rate": 9.005738555434164e-06,
+      "loss": 0.6567,
+      "step": 14010
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.1822662353515625,
+      "learning_rate": 9.00558075923545e-06,
+      "loss": 0.6425,
+      "step": 14011
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.7753052711486816,
+      "learning_rate": 9.005422951898716e-06,
+      "loss": 0.4824,
+      "step": 14012
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.837749719619751,
+      "learning_rate": 9.005265133424402e-06,
+      "loss": 0.636,
+      "step": 14013
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.174983024597168,
+      "learning_rate": 9.005107303812946e-06,
+      "loss": 0.7446,
+      "step": 14014
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.340059757232666,
+      "learning_rate": 9.004949463064787e-06,
+      "loss": 0.482,
+      "step": 14015
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.6928250789642334,
+      "learning_rate": 9.004791611180363e-06,
+      "loss": 0.5827,
+      "step": 14016
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.7065125703811646,
+      "learning_rate": 9.004633748160115e-06,
+      "loss": 0.5938,
+      "step": 14017
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.8621628284454346,
+      "learning_rate": 9.004475874004482e-06,
+      "loss": 0.9902,
+      "step": 14018
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 4.608409881591797,
+      "learning_rate": 9.004317988713898e-06,
+      "loss": 0.6062,
+      "step": 14019
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.8161122798919678,
+      "learning_rate": 9.00416009228881e-06,
+      "loss": 0.7574,
+      "step": 14020
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.865177869796753,
+      "learning_rate": 9.004002184729649e-06,
+      "loss": 0.6714,
+      "step": 14021
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.4139842987060547,
+      "learning_rate": 9.003844266036859e-06,
+      "loss": 0.8334,
+      "step": 14022
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.1106581687927246,
+      "learning_rate": 9.003686336210878e-06,
+      "loss": 0.6917,
+      "step": 14023
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.2048494815826416,
+      "learning_rate": 9.003528395252146e-06,
+      "loss": 0.5087,
+      "step": 14024
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.9163308143615723,
+      "learning_rate": 9.003370443161101e-06,
+      "loss": 0.592,
+      "step": 14025
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.8333673477172852,
+      "learning_rate": 9.003212479938182e-06,
+      "loss": 0.8213,
+      "step": 14026
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 4.806340217590332,
+      "learning_rate": 9.003054505583828e-06,
+      "loss": 0.7501,
+      "step": 14027
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.934818744659424,
+      "learning_rate": 9.002896520098481e-06,
+      "loss": 0.3778,
+      "step": 14028
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.0275650024414062,
+      "learning_rate": 9.002738523482577e-06,
+      "loss": 0.8064,
+      "step": 14029
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 4.785223484039307,
+      "learning_rate": 9.002580515736557e-06,
+      "loss": 0.4023,
+      "step": 14030
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.394174575805664,
+      "learning_rate": 9.002422496860859e-06,
+      "loss": 0.7748,
+      "step": 14031
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.3994877338409424,
+      "learning_rate": 9.002264466855924e-06,
+      "loss": 0.7683,
+      "step": 14032
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.3910696506500244,
+      "learning_rate": 9.002106425722193e-06,
+      "loss": 0.6508,
+      "step": 14033
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.64066219329834,
+      "learning_rate": 9.0019483734601e-06,
+      "loss": 0.6109,
+      "step": 14034
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.139627695083618,
+      "learning_rate": 9.001790310070088e-06,
+      "loss": 0.8392,
+      "step": 14035
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.4740610122680664,
+      "learning_rate": 9.001632235552598e-06,
+      "loss": 0.7242,
+      "step": 14036
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.3717037439346313,
+      "learning_rate": 9.001474149908068e-06,
+      "loss": 0.6756,
+      "step": 14037
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.999849796295166,
+      "learning_rate": 9.001316053136935e-06,
+      "loss": 0.8303,
+      "step": 14038
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.2084131240844727,
+      "learning_rate": 9.001157945239643e-06,
+      "loss": 0.8136,
+      "step": 14039
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.068147659301758,
+      "learning_rate": 9.00099982621663e-06,
+      "loss": 0.7123,
+      "step": 14040
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.3124704360961914,
+      "learning_rate": 9.000841696068333e-06,
+      "loss": 0.8301,
+      "step": 14041
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.385411262512207,
+      "learning_rate": 9.000683554795195e-06,
+      "loss": 0.6508,
+      "step": 14042
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.3318471908569336,
+      "learning_rate": 9.000525402397655e-06,
+      "loss": 0.7576,
+      "step": 14043
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.533951759338379,
+      "learning_rate": 9.000367238876153e-06,
+      "loss": 0.5711,
+      "step": 14044
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 4.258266448974609,
+      "learning_rate": 9.000209064231129e-06,
+      "loss": 0.5973,
+      "step": 14045
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.8464906215667725,
+      "learning_rate": 9.00005087846302e-06,
+      "loss": 0.5869,
+      "step": 14046
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 7.1532087326049805,
+      "learning_rate": 8.999892681572269e-06,
+      "loss": 0.4864,
+      "step": 14047
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.734816074371338,
+      "learning_rate": 8.999734473559313e-06,
+      "loss": 0.757,
+      "step": 14048
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.4399569034576416,
+      "learning_rate": 8.999576254424596e-06,
+      "loss": 0.6045,
+      "step": 14049
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.2054052352905273,
+      "learning_rate": 8.999418024168555e-06,
+      "loss": 0.7525,
+      "step": 14050
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.359745502471924,
+      "learning_rate": 8.999259782791632e-06,
+      "loss": 0.698,
+      "step": 14051
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.0246098041534424,
+      "learning_rate": 8.999101530294265e-06,
+      "loss": 0.7936,
+      "step": 14052
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.9289275407791138,
+      "learning_rate": 8.998943266676894e-06,
+      "loss": 0.5547,
+      "step": 14053
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.888702392578125,
+      "learning_rate": 8.99878499193996e-06,
+      "loss": 0.6933,
+      "step": 14054
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 6.670708179473877,
+      "learning_rate": 8.998626706083903e-06,
+      "loss": 0.9131,
+      "step": 14055
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.4873507022857666,
+      "learning_rate": 8.998468409109163e-06,
+      "loss": 0.5134,
+      "step": 14056
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.299588441848755,
+      "learning_rate": 8.99831010101618e-06,
+      "loss": 0.7281,
+      "step": 14057
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.7490992546081543,
+      "learning_rate": 8.998151781805395e-06,
+      "loss": 0.4842,
+      "step": 14058
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.5466270446777344,
+      "learning_rate": 8.997993451477246e-06,
+      "loss": 0.6971,
+      "step": 14059
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.4894068241119385,
+      "learning_rate": 8.997835110032178e-06,
+      "loss": 0.607,
+      "step": 14060
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.2033369541168213,
+      "learning_rate": 8.997676757470626e-06,
+      "loss": 0.4642,
+      "step": 14061
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 4.3630805015563965,
+      "learning_rate": 8.997518393793033e-06,
+      "loss": 0.8108,
+      "step": 14062
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.8940802812576294,
+      "learning_rate": 8.997360018999839e-06,
+      "loss": 0.613,
+      "step": 14063
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.830249547958374,
+      "learning_rate": 8.997201633091483e-06,
+      "loss": 0.9169,
+      "step": 14064
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.1873772144317627,
+      "learning_rate": 8.997043236068407e-06,
+      "loss": 0.6325,
+      "step": 14065
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.1842567920684814,
+      "learning_rate": 8.996884827931053e-06,
+      "loss": 0.7794,
+      "step": 14066
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.7233171463012695,
+      "learning_rate": 8.996726408679859e-06,
+      "loss": 0.6125,
+      "step": 14067
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.165748357772827,
+      "learning_rate": 8.996567978315264e-06,
+      "loss": 0.9096,
+      "step": 14068
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.6454066038131714,
+      "learning_rate": 8.996409536837714e-06,
+      "loss": 0.568,
+      "step": 14069
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.149909734725952,
+      "learning_rate": 8.996251084247642e-06,
+      "loss": 0.5449,
+      "step": 14070
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.4639315605163574,
+      "learning_rate": 8.996092620545495e-06,
+      "loss": 0.6289,
+      "step": 14071
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.798107862472534,
+      "learning_rate": 8.995934145731711e-06,
+      "loss": 0.5823,
+      "step": 14072
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 4.547476768493652,
+      "learning_rate": 8.995775659806732e-06,
+      "loss": 0.7453,
+      "step": 14073
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 4.262748718261719,
+      "learning_rate": 8.995617162770998e-06,
+      "loss": 0.8206,
+      "step": 14074
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.9920756816864014,
+      "learning_rate": 8.995458654624948e-06,
+      "loss": 0.6123,
+      "step": 14075
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.301757574081421,
+      "learning_rate": 8.995300135369025e-06,
+      "loss": 0.7993,
+      "step": 14076
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.8219423294067383,
+      "learning_rate": 8.995141605003667e-06,
+      "loss": 0.802,
+      "step": 14077
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.811217784881592,
+      "learning_rate": 8.99498306352932e-06,
+      "loss": 0.7796,
+      "step": 14078
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.7025058269500732,
+      "learning_rate": 8.994824510946421e-06,
+      "loss": 0.4369,
+      "step": 14079
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.9099528789520264,
+      "learning_rate": 8.99466594725541e-06,
+      "loss": 0.7271,
+      "step": 14080
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.271284341812134,
+      "learning_rate": 8.99450737245673e-06,
+      "loss": 0.7615,
+      "step": 14081
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.1101279258728027,
+      "learning_rate": 8.994348786550821e-06,
+      "loss": 0.8119,
+      "step": 14082
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.181403398513794,
+      "learning_rate": 8.994190189538125e-06,
+      "loss": 0.8094,
+      "step": 14083
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.7411577701568604,
+      "learning_rate": 8.994031581419082e-06,
+      "loss": 0.8703,
+      "step": 14084
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.9447011947631836,
+      "learning_rate": 8.993872962194134e-06,
+      "loss": 0.6301,
+      "step": 14085
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.7250921726226807,
+      "learning_rate": 8.99371433186372e-06,
+      "loss": 0.7587,
+      "step": 14086
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.26629376411438,
+      "learning_rate": 8.993555690428285e-06,
+      "loss": 0.6885,
+      "step": 14087
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.2450830936431885,
+      "learning_rate": 8.993397037888266e-06,
+      "loss": 0.7078,
+      "step": 14088
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.4962069988250732,
+      "learning_rate": 8.993238374244106e-06,
+      "loss": 0.6213,
+      "step": 14089
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.3300387859344482,
+      "learning_rate": 8.993079699496247e-06,
+      "loss": 0.801,
+      "step": 14090
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 4.2601094245910645,
+      "learning_rate": 8.992921013645129e-06,
+      "loss": 0.28,
+      "step": 14091
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.5574233531951904,
+      "learning_rate": 8.99276231669119e-06,
+      "loss": 0.7023,
+      "step": 14092
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.120530605316162,
+      "learning_rate": 8.992603608634878e-06,
+      "loss": 0.5587,
+      "step": 14093
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.3723011016845703,
+      "learning_rate": 8.99244488947663e-06,
+      "loss": 0.8963,
+      "step": 14094
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.9721159934997559,
+      "learning_rate": 8.99228615921689e-06,
+      "loss": 0.7138,
+      "step": 14095
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.5351009368896484,
+      "learning_rate": 8.992127417856095e-06,
+      "loss": 0.7536,
+      "step": 14096
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.8239054679870605,
+      "learning_rate": 8.991968665394693e-06,
+      "loss": 0.8867,
+      "step": 14097
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.1087377071380615,
+      "learning_rate": 8.991809901833116e-06,
+      "loss": 0.9144,
+      "step": 14098
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.6651177406311035,
+      "learning_rate": 8.991651127171817e-06,
+      "loss": 0.5932,
+      "step": 14099
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.9749364852905273,
+      "learning_rate": 8.991492341411227e-06,
+      "loss": 0.6475,
+      "step": 14100
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 5.584768295288086,
+      "learning_rate": 8.991333544551795e-06,
+      "loss": 0.5177,
+      "step": 14101
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.8842593431472778,
+      "learning_rate": 8.991174736593958e-06,
+      "loss": 0.6692,
+      "step": 14102
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.802801251411438,
+      "learning_rate": 8.991015917538159e-06,
+      "loss": 0.8816,
+      "step": 14103
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.706106662750244,
+      "learning_rate": 8.990857087384841e-06,
+      "loss": 0.8962,
+      "step": 14104
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.435938835144043,
+      "learning_rate": 8.990698246134444e-06,
+      "loss": 0.7617,
+      "step": 14105
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.580308198928833,
+      "learning_rate": 8.99053939378741e-06,
+      "loss": 0.6485,
+      "step": 14106
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.199968099594116,
+      "learning_rate": 8.99038053034418e-06,
+      "loss": 0.6523,
+      "step": 14107
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 5.555122375488281,
+      "learning_rate": 8.990221655805198e-06,
+      "loss": 0.8022,
+      "step": 14108
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.1333017349243164,
+      "learning_rate": 8.990062770170904e-06,
+      "loss": 0.8086,
+      "step": 14109
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.365666151046753,
+      "learning_rate": 8.98990387344174e-06,
+      "loss": 0.5942,
+      "step": 14110
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.5012106895446777,
+      "learning_rate": 8.989744965618148e-06,
+      "loss": 0.6643,
+      "step": 14111
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.409857749938965,
+      "learning_rate": 8.98958604670057e-06,
+      "loss": 0.6904,
+      "step": 14112
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.05469012260437,
+      "learning_rate": 8.989427116689447e-06,
+      "loss": 0.5141,
+      "step": 14113
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.750746726989746,
+      "learning_rate": 8.989268175585222e-06,
+      "loss": 0.7531,
+      "step": 14114
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.3261842727661133,
+      "learning_rate": 8.989109223388338e-06,
+      "loss": 0.6014,
+      "step": 14115
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.2257704734802246,
+      "learning_rate": 8.988950260099234e-06,
+      "loss": 0.5021,
+      "step": 14116
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.6524235010147095,
+      "learning_rate": 8.988791285718352e-06,
+      "loss": 0.8382,
+      "step": 14117
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.9585530757904053,
+      "learning_rate": 8.988632300246139e-06,
+      "loss": 0.8594,
+      "step": 14118
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.8018885850906372,
+      "learning_rate": 8.988473303683031e-06,
+      "loss": 0.7046,
+      "step": 14119
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.0853562355041504,
+      "learning_rate": 8.988314296029475e-06,
+      "loss": 0.6785,
+      "step": 14120
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.8015578985214233,
+      "learning_rate": 8.98815527728591e-06,
+      "loss": 0.7351,
+      "step": 14121
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.60001277923584,
+      "learning_rate": 8.98799624745278e-06,
+      "loss": 0.7769,
+      "step": 14122
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.473285436630249,
+      "learning_rate": 8.987837206530526e-06,
+      "loss": 0.7171,
+      "step": 14123
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.996070146560669,
+      "learning_rate": 8.987678154519591e-06,
+      "loss": 0.6519,
+      "step": 14124
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.246234893798828,
+      "learning_rate": 8.987519091420415e-06,
+      "loss": 0.7391,
+      "step": 14125
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.9889473915100098,
+      "learning_rate": 8.987360017233445e-06,
+      "loss": 0.6962,
+      "step": 14126
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.8207823038101196,
+      "learning_rate": 8.987200931959118e-06,
+      "loss": 0.5668,
+      "step": 14127
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.2210934162139893,
+      "learning_rate": 8.987041835597878e-06,
+      "loss": 0.6778,
+      "step": 14128
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.9660661220550537,
+      "learning_rate": 8.98688272815017e-06,
+      "loss": 0.9655,
+      "step": 14129
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.5979490280151367,
+      "learning_rate": 8.986723609616435e-06,
+      "loss": 0.5918,
+      "step": 14130
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.8119338750839233,
+      "learning_rate": 8.986564479997115e-06,
+      "loss": 0.5799,
+      "step": 14131
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 4.371266841888428,
+      "learning_rate": 8.986405339292652e-06,
+      "loss": 0.8759,
+      "step": 14132
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.9525043964385986,
+      "learning_rate": 8.98624618750349e-06,
+      "loss": 0.7911,
+      "step": 14133
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.51025652885437,
+      "learning_rate": 8.98608702463007e-06,
+      "loss": 0.8624,
+      "step": 14134
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.619649648666382,
+      "learning_rate": 8.985927850672835e-06,
+      "loss": 0.771,
+      "step": 14135
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.1338508129119873,
+      "learning_rate": 8.985768665632228e-06,
+      "loss": 0.6348,
+      "step": 14136
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.395875930786133,
+      "learning_rate": 8.985609469508691e-06,
+      "loss": 0.5643,
+      "step": 14137
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.190402030944824,
+      "learning_rate": 8.985450262302669e-06,
+      "loss": 0.5349,
+      "step": 14138
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.5772457122802734,
+      "learning_rate": 8.9852910440146e-06,
+      "loss": 0.6431,
+      "step": 14139
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.7247529029846191,
+      "learning_rate": 8.985131814644931e-06,
+      "loss": 0.7144,
+      "step": 14140
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.920250177383423,
+      "learning_rate": 8.984972574194104e-06,
+      "loss": 0.4596,
+      "step": 14141
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.586371421813965,
+      "learning_rate": 8.98481332266256e-06,
+      "loss": 1.0172,
+      "step": 14142
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 6.992191314697266,
+      "learning_rate": 8.984654060050744e-06,
+      "loss": 0.7404,
+      "step": 14143
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.568960428237915,
+      "learning_rate": 8.984494786359097e-06,
+      "loss": 0.527,
+      "step": 14144
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.110107421875,
+      "learning_rate": 8.984335501588062e-06,
+      "loss": 0.4744,
+      "step": 14145
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.5757858753204346,
+      "learning_rate": 8.984176205738083e-06,
+      "loss": 0.6826,
+      "step": 14146
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.4094254970550537,
+      "learning_rate": 8.984016898809603e-06,
+      "loss": 0.9114,
+      "step": 14147
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.374891757965088,
+      "learning_rate": 8.983857580803063e-06,
+      "loss": 0.7236,
+      "step": 14148
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.0524134635925293,
+      "learning_rate": 8.983698251718908e-06,
+      "loss": 0.739,
+      "step": 14149
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.9328441619873047,
+      "learning_rate": 8.983538911557583e-06,
+      "loss": 0.5164,
+      "step": 14150
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.7015540599823,
+      "learning_rate": 8.983379560319525e-06,
+      "loss": 0.7066,
+      "step": 14151
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.7731945514678955,
+      "learning_rate": 8.983220198005182e-06,
+      "loss": 0.713,
+      "step": 14152
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.763954520225525,
+      "learning_rate": 8.983060824614998e-06,
+      "loss": 0.7592,
+      "step": 14153
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.5195250511169434,
+      "learning_rate": 8.98290144014941e-06,
+      "loss": 0.8925,
+      "step": 14154
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.26714825630188,
+      "learning_rate": 8.982742044608868e-06,
+      "loss": 0.7193,
+      "step": 14155
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.8284430503845215,
+      "learning_rate": 8.98258263799381e-06,
+      "loss": 0.8318,
+      "step": 14156
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.948428988456726,
+      "learning_rate": 8.982423220304684e-06,
+      "loss": 0.8332,
+      "step": 14157
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.0420517921447754,
+      "learning_rate": 8.982263791541929e-06,
+      "loss": 0.7679,
+      "step": 14158
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.599918007850647,
+      "learning_rate": 8.982104351705992e-06,
+      "loss": 0.5332,
+      "step": 14159
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 5.105205535888672,
+      "learning_rate": 8.981944900797313e-06,
+      "loss": 0.6995,
+      "step": 14160
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.432518720626831,
+      "learning_rate": 8.981785438816338e-06,
+      "loss": 0.6231,
+      "step": 14161
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.3928961753845215,
+      "learning_rate": 8.98162596576351e-06,
+      "loss": 0.3501,
+      "step": 14162
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.4677144289016724,
+      "learning_rate": 8.981466481639269e-06,
+      "loss": 0.7394,
+      "step": 14163
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.6921958923339844,
+      "learning_rate": 8.981306986444063e-06,
+      "loss": 0.659,
+      "step": 14164
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.263571262359619,
+      "learning_rate": 8.981147480178334e-06,
+      "loss": 0.595,
+      "step": 14165
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.197028398513794,
+      "learning_rate": 8.980987962842524e-06,
+      "loss": 0.6083,
+      "step": 14166
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.981449842453003,
+      "learning_rate": 8.980828434437078e-06,
+      "loss": 0.7206,
+      "step": 14167
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.0357885360717773,
+      "learning_rate": 8.98066889496244e-06,
+      "loss": 0.6116,
+      "step": 14168
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.9470248222351074,
+      "learning_rate": 8.980509344419052e-06,
+      "loss": 0.5319,
+      "step": 14169
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.13334321975708,
+      "learning_rate": 8.98034978280736e-06,
+      "loss": 0.7358,
+      "step": 14170
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.8641984462738037,
+      "learning_rate": 8.980190210127805e-06,
+      "loss": 0.7677,
+      "step": 14171
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.6030185222625732,
+      "learning_rate": 8.980030626380833e-06,
+      "loss": 0.5668,
+      "step": 14172
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.7929136753082275,
+      "learning_rate": 8.979871031566888e-06,
+      "loss": 0.6237,
+      "step": 14173
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.3529653549194336,
+      "learning_rate": 8.979711425686411e-06,
+      "loss": 0.8037,
+      "step": 14174
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.4153947830200195,
+      "learning_rate": 8.979551808739847e-06,
+      "loss": 0.422,
+      "step": 14175
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.6220521926879883,
+      "learning_rate": 8.979392180727641e-06,
+      "loss": 0.6031,
+      "step": 14176
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.3230481147766113,
+      "learning_rate": 8.979232541650236e-06,
+      "loss": 0.5779,
+      "step": 14177
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.5195157527923584,
+      "learning_rate": 8.979072891508074e-06,
+      "loss": 0.5367,
+      "step": 14178
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.456580638885498,
+      "learning_rate": 8.978913230301604e-06,
+      "loss": 0.7902,
+      "step": 14179
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.8771286010742188,
+      "learning_rate": 8.978753558031266e-06,
+      "loss": 0.6741,
+      "step": 14180
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 4.007027626037598,
+      "learning_rate": 8.978593874697504e-06,
+      "loss": 0.6948,
+      "step": 14181
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.2278687953948975,
+      "learning_rate": 8.978434180300763e-06,
+      "loss": 0.4768,
+      "step": 14182
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.5713634490966797,
+      "learning_rate": 8.978274474841487e-06,
+      "loss": 0.8694,
+      "step": 14183
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.9373788833618164,
+      "learning_rate": 8.97811475832012e-06,
+      "loss": 0.604,
+      "step": 14184
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.9362053871154785,
+      "learning_rate": 8.977955030737106e-06,
+      "loss": 0.6046,
+      "step": 14185
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.0650877952575684,
+      "learning_rate": 8.977795292092888e-06,
+      "loss": 0.6344,
+      "step": 14186
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.9677928686141968,
+      "learning_rate": 8.977635542387915e-06,
+      "loss": 0.8907,
+      "step": 14187
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.6015241146087646,
+      "learning_rate": 8.977475781622624e-06,
+      "loss": 0.7001,
+      "step": 14188
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.8228061199188232,
+      "learning_rate": 8.977316009797464e-06,
+      "loss": 0.8877,
+      "step": 14189
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.2558999061584473,
+      "learning_rate": 8.977156226912878e-06,
+      "loss": 0.5666,
+      "step": 14190
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.9002315998077393,
+      "learning_rate": 8.976996432969312e-06,
+      "loss": 0.7255,
+      "step": 14191
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 4.158173084259033,
+      "learning_rate": 8.976836627967207e-06,
+      "loss": 0.693,
+      "step": 14192
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.4095990657806396,
+      "learning_rate": 8.97667681190701e-06,
+      "loss": 0.989,
+      "step": 14193
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.8804779052734375,
+      "learning_rate": 8.976516984789163e-06,
+      "loss": 0.4617,
+      "step": 14194
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.8744823932647705,
+      "learning_rate": 8.976357146614114e-06,
+      "loss": 0.6141,
+      "step": 14195
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 4.721320629119873,
+      "learning_rate": 8.976197297382304e-06,
+      "loss": 0.4724,
+      "step": 14196
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.6142313480377197,
+      "learning_rate": 8.976037437094178e-06,
+      "loss": 0.5864,
+      "step": 14197
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.644639730453491,
+      "learning_rate": 8.975877565750184e-06,
+      "loss": 0.7826,
+      "step": 14198
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.8273811340332031,
+      "learning_rate": 8.975717683350762e-06,
+      "loss": 0.6927,
+      "step": 14199
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.7887721061706543,
+      "learning_rate": 8.975557789896358e-06,
+      "loss": 0.788,
+      "step": 14200
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.3645381927490234,
+      "learning_rate": 8.975397885387418e-06,
+      "loss": 0.5039,
+      "step": 14201
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.955149531364441,
+      "learning_rate": 8.975237969824386e-06,
+      "loss": 0.624,
+      "step": 14202
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.524900197982788,
+      "learning_rate": 8.975078043207706e-06,
+      "loss": 0.8181,
+      "step": 14203
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.6683340072631836,
+      "learning_rate": 8.974918105537822e-06,
+      "loss": 0.7393,
+      "step": 14204
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.1129491329193115,
+      "learning_rate": 8.974758156815181e-06,
+      "loss": 0.5483,
+      "step": 14205
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.4343459606170654,
+      "learning_rate": 8.974598197040228e-06,
+      "loss": 0.8178,
+      "step": 14206
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.3837509155273438,
+      "learning_rate": 8.974438226213405e-06,
+      "loss": 0.5349,
+      "step": 14207
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.2506089210510254,
+      "learning_rate": 8.974278244335157e-06,
+      "loss": 0.5026,
+      "step": 14208
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.599953532218933,
+      "learning_rate": 8.97411825140593e-06,
+      "loss": 0.6628,
+      "step": 14209
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 4.65084981918335,
+      "learning_rate": 8.97395824742617e-06,
+      "loss": 0.8286,
+      "step": 14210
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.262119770050049,
+      "learning_rate": 8.97379823239632e-06,
+      "loss": 0.6971,
+      "step": 14211
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.6957449913024902,
+      "learning_rate": 8.973638206316827e-06,
+      "loss": 0.7489,
+      "step": 14212
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.7439578771591187,
+      "learning_rate": 8.973478169188134e-06,
+      "loss": 0.5472,
+      "step": 14213
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.240994453430176,
+      "learning_rate": 8.973318121010686e-06,
+      "loss": 0.581,
+      "step": 14214
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.6228712797164917,
+      "learning_rate": 8.97315806178493e-06,
+      "loss": 0.6295,
+      "step": 14215
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.143735408782959,
+      "learning_rate": 8.97299799151131e-06,
+      "loss": 0.575,
+      "step": 14216
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.6957290172576904,
+      "learning_rate": 8.972837910190268e-06,
+      "loss": 0.6267,
+      "step": 14217
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.883577346801758,
+      "learning_rate": 8.972677817822256e-06,
+      "loss": 0.6379,
+      "step": 14218
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.0622363090515137,
+      "learning_rate": 8.972517714407712e-06,
+      "loss": 0.7348,
+      "step": 14219
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.3501155376434326,
+      "learning_rate": 8.972357599947085e-06,
+      "loss": 0.6623,
+      "step": 14220
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 4.758203983306885,
+      "learning_rate": 8.972197474440822e-06,
+      "loss": 0.7529,
+      "step": 14221
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.4882030487060547,
+      "learning_rate": 8.972037337889363e-06,
+      "loss": 0.61,
+      "step": 14222
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.35064959526062,
+      "learning_rate": 8.971877190293158e-06,
+      "loss": 0.8021,
+      "step": 14223
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.140509605407715,
+      "learning_rate": 8.97171703165265e-06,
+      "loss": 0.7783,
+      "step": 14224
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.242546558380127,
+      "learning_rate": 8.971556861968285e-06,
+      "loss": 0.5426,
+      "step": 14225
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.5411869287490845,
+      "learning_rate": 8.971396681240508e-06,
+      "loss": 0.7108,
+      "step": 14226
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.523071050643921,
+      "learning_rate": 8.971236489469764e-06,
+      "loss": 0.8426,
+      "step": 14227
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.9816805124282837,
+      "learning_rate": 8.971076286656499e-06,
+      "loss": 0.7014,
+      "step": 14228
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.638171434402466,
+      "learning_rate": 8.970916072801157e-06,
+      "loss": 0.702,
+      "step": 14229
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.8251773118972778,
+      "learning_rate": 8.970755847904188e-06,
+      "loss": 0.6021,
+      "step": 14230
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.112180233001709,
+      "learning_rate": 8.970595611966032e-06,
+      "loss": 0.7545,
+      "step": 14231
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.358727216720581,
+      "learning_rate": 8.970435364987138e-06,
+      "loss": 0.6331,
+      "step": 14232
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.000877857208252,
+      "learning_rate": 8.97027510696795e-06,
+      "loss": 0.8254,
+      "step": 14233
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.9357244968414307,
+      "learning_rate": 8.970114837908916e-06,
+      "loss": 0.5915,
+      "step": 14234
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.682370185852051,
+      "learning_rate": 8.969954557810479e-06,
+      "loss": 0.7307,
+      "step": 14235
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.654844880104065,
+      "learning_rate": 8.969794266673084e-06,
+      "loss": 0.6922,
+      "step": 14236
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.902665138244629,
+      "learning_rate": 8.969633964497179e-06,
+      "loss": 0.8348,
+      "step": 14237
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.053826332092285,
+      "learning_rate": 8.969473651283212e-06,
+      "loss": 0.8167,
+      "step": 14238
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.642552137374878,
+      "learning_rate": 8.969313327031621e-06,
+      "loss": 0.6798,
+      "step": 14239
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.5451393127441406,
+      "learning_rate": 8.96915299174286e-06,
+      "loss": 1.036,
+      "step": 14240
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.4008376598358154,
+      "learning_rate": 8.96899264541737e-06,
+      "loss": 0.8112,
+      "step": 14241
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.0325963497161865,
+      "learning_rate": 8.968832288055599e-06,
+      "loss": 0.7122,
+      "step": 14242
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.652190923690796,
+      "learning_rate": 8.968671919657993e-06,
+      "loss": 0.6521,
+      "step": 14243
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.9101479053497314,
+      "learning_rate": 8.968511540224994e-06,
+      "loss": 0.7769,
+      "step": 14244
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 4.481254577636719,
+      "learning_rate": 8.968351149757053e-06,
+      "loss": 0.7639,
+      "step": 14245
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.65354323387146,
+      "learning_rate": 8.968190748254615e-06,
+      "loss": 0.4778,
+      "step": 14246
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.41526198387146,
+      "learning_rate": 8.968030335718124e-06,
+      "loss": 0.6358,
+      "step": 14247
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.087111711502075,
+      "learning_rate": 8.967869912148027e-06,
+      "loss": 0.7641,
+      "step": 14248
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.9968147277832031,
+      "learning_rate": 8.967709477544768e-06,
+      "loss": 0.7347,
+      "step": 14249
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.7905056476593018,
+      "learning_rate": 8.9675490319088e-06,
+      "loss": 0.8269,
+      "step": 14250
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.6732330322265625,
+      "learning_rate": 8.967388575240561e-06,
+      "loss": 0.6501,
+      "step": 14251
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.141315221786499,
+      "learning_rate": 8.967228107540502e-06,
+      "loss": 0.6606,
+      "step": 14252
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.108125686645508,
+      "learning_rate": 8.967067628809068e-06,
+      "loss": 0.4977,
+      "step": 14253
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.328659772872925,
+      "learning_rate": 8.966907139046704e-06,
+      "loss": 0.884,
+      "step": 14254
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.7632312774658203,
+      "learning_rate": 8.966746638253857e-06,
+      "loss": 0.5864,
+      "step": 14255
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.1831815242767334,
+      "learning_rate": 8.966586126430973e-06,
+      "loss": 0.5422,
+      "step": 14256
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.5084259510040283,
+      "learning_rate": 8.966425603578502e-06,
+      "loss": 0.7411,
+      "step": 14257
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.6637825965881348,
+      "learning_rate": 8.966265069696885e-06,
+      "loss": 0.5637,
+      "step": 14258
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.9663424491882324,
+      "learning_rate": 8.96610452478657e-06,
+      "loss": 0.5988,
+      "step": 14259
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.5718772411346436,
+      "learning_rate": 8.965943968848006e-06,
+      "loss": 0.6598,
+      "step": 14260
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.856729030609131,
+      "learning_rate": 8.965783401881636e-06,
+      "loss": 0.7291,
+      "step": 14261
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.7216169834136963,
+      "learning_rate": 8.965622823887909e-06,
+      "loss": 0.7445,
+      "step": 14262
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.9249515533447266,
+      "learning_rate": 8.96546223486727e-06,
+      "loss": 0.5024,
+      "step": 14263
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.2487778663635254,
+      "learning_rate": 8.965301634820165e-06,
+      "loss": 0.7813,
+      "step": 14264
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.9182548522949219,
+      "learning_rate": 8.965141023747043e-06,
+      "loss": 0.6365,
+      "step": 14265
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.666832685470581,
+      "learning_rate": 8.96498040164835e-06,
+      "loss": 0.8208,
+      "step": 14266
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.7758796215057373,
+      "learning_rate": 8.96481976852453e-06,
+      "loss": 0.7038,
+      "step": 14267
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 4.293801784515381,
+      "learning_rate": 8.964659124376031e-06,
+      "loss": 0.8461,
+      "step": 14268
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.099334955215454,
+      "learning_rate": 8.964498469203303e-06,
+      "loss": 0.6986,
+      "step": 14269
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.3936431407928467,
+      "learning_rate": 8.964337803006786e-06,
+      "loss": 0.6887,
+      "step": 14270
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.027647018432617,
+      "learning_rate": 8.964177125786934e-06,
+      "loss": 0.6696,
+      "step": 14271
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.3964297771453857,
+      "learning_rate": 8.96401643754419e-06,
+      "loss": 0.7048,
+      "step": 14272
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.2051029205322266,
+      "learning_rate": 8.963855738279e-06,
+      "loss": 0.7106,
+      "step": 14273
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.318110942840576,
+      "learning_rate": 8.963695027991813e-06,
+      "loss": 0.9104,
+      "step": 14274
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.003270387649536,
+      "learning_rate": 8.963534306683072e-06,
+      "loss": 0.5478,
+      "step": 14275
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.7203911542892456,
+      "learning_rate": 8.963373574353232e-06,
+      "loss": 0.7602,
+      "step": 14276
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.4820809364318848,
+      "learning_rate": 8.96321283100273e-06,
+      "loss": 0.3793,
+      "step": 14277
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.1940128803253174,
+      "learning_rate": 8.963052076632022e-06,
+      "loss": 0.6574,
+      "step": 14278
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 4.30666446685791,
+      "learning_rate": 8.962891311241549e-06,
+      "loss": 0.6072,
+      "step": 14279
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.8589552640914917,
+      "learning_rate": 8.962730534831758e-06,
+      "loss": 0.5626,
+      "step": 14280
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.6152122020721436,
+      "learning_rate": 8.962569747403099e-06,
+      "loss": 0.7988,
+      "step": 14281
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.858006238937378,
+      "learning_rate": 8.962408948956018e-06,
+      "loss": 0.4982,
+      "step": 14282
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.9131903648376465,
+      "learning_rate": 8.962248139490963e-06,
+      "loss": 0.8197,
+      "step": 14283
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.781275987625122,
+      "learning_rate": 8.962087319008381e-06,
+      "loss": 0.7431,
+      "step": 14284
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.664442777633667,
+      "learning_rate": 8.961926487508716e-06,
+      "loss": 0.7195,
+      "step": 14285
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.230529308319092,
+      "learning_rate": 8.961765644992418e-06,
+      "loss": 0.768,
+      "step": 14286
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.926225185394287,
+      "learning_rate": 8.961604791459936e-06,
+      "loss": 0.843,
+      "step": 14287
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.254863977432251,
+      "learning_rate": 8.961443926911713e-06,
+      "loss": 0.5937,
+      "step": 14288
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.082991123199463,
+      "learning_rate": 8.961283051348198e-06,
+      "loss": 0.5288,
+      "step": 14289
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.1997880935668945,
+      "learning_rate": 8.96112216476984e-06,
+      "loss": 0.7726,
+      "step": 14290
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.8853877782821655,
+      "learning_rate": 8.960961267177085e-06,
+      "loss": 0.628,
+      "step": 14291
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.243738889694214,
+      "learning_rate": 8.96080035857038e-06,
+      "loss": 0.6285,
+      "step": 14292
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.610089659690857,
+      "learning_rate": 8.960639438950173e-06,
+      "loss": 0.6488,
+      "step": 14293
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.1118662357330322,
+      "learning_rate": 8.960478508316913e-06,
+      "loss": 0.5402,
+      "step": 14294
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.105125665664673,
+      "learning_rate": 8.960317566671043e-06,
+      "loss": 0.6001,
+      "step": 14295
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.3530936241149902,
+      "learning_rate": 8.960156614013015e-06,
+      "loss": 0.6748,
+      "step": 14296
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.8439853191375732,
+      "learning_rate": 8.959995650343276e-06,
+      "loss": 0.7243,
+      "step": 14297
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.217376470565796,
+      "learning_rate": 8.959834675662273e-06,
+      "loss": 0.6579,
+      "step": 14298
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.514571189880371,
+      "learning_rate": 8.959673689970451e-06,
+      "loss": 0.6408,
+      "step": 14299
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.2820470333099365,
+      "learning_rate": 8.95951269326826e-06,
+      "loss": 1.0447,
+      "step": 14300
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 8.037911415100098,
+      "learning_rate": 8.959351685556148e-06,
+      "loss": 0.6253,
+      "step": 14301
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.70106840133667,
+      "learning_rate": 8.959190666834563e-06,
+      "loss": 0.569,
+      "step": 14302
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.070352077484131,
+      "learning_rate": 8.95902963710395e-06,
+      "loss": 0.5118,
+      "step": 14303
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.0514397621154785,
+      "learning_rate": 8.95886859636476e-06,
+      "loss": 0.5338,
+      "step": 14304
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.757784366607666,
+      "learning_rate": 8.95870754461744e-06,
+      "loss": 0.882,
+      "step": 14305
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.5668137073516846,
+      "learning_rate": 8.958546481862436e-06,
+      "loss": 0.6018,
+      "step": 14306
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.4264309406280518,
+      "learning_rate": 8.958385408100199e-06,
+      "loss": 0.514,
+      "step": 14307
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.3073606491088867,
+      "learning_rate": 8.958224323331172e-06,
+      "loss": 0.711,
+      "step": 14308
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.264042615890503,
+      "learning_rate": 8.958063227555808e-06,
+      "loss": 0.8299,
+      "step": 14309
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.5314799547195435,
+      "learning_rate": 8.957902120774553e-06,
+      "loss": 0.6669,
+      "step": 14310
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.8835787773132324,
+      "learning_rate": 8.957741002987854e-06,
+      "loss": 0.9387,
+      "step": 14311
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.374701976776123,
+      "learning_rate": 8.95757987419616e-06,
+      "loss": 0.494,
+      "step": 14312
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.5538721084594727,
+      "learning_rate": 8.95741873439992e-06,
+      "loss": 0.8027,
+      "step": 14313
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.5285987854003906,
+      "learning_rate": 8.95725758359958e-06,
+      "loss": 0.686,
+      "step": 14314
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.0058510303497314,
+      "learning_rate": 8.957096421795589e-06,
+      "loss": 0.6868,
+      "step": 14315
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.2523677349090576,
+      "learning_rate": 8.956935248988396e-06,
+      "loss": 0.5513,
+      "step": 14316
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.2274978160858154,
+      "learning_rate": 8.956774065178448e-06,
+      "loss": 0.7832,
+      "step": 14317
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.245573043823242,
+      "learning_rate": 8.956612870366192e-06,
+      "loss": 0.5714,
+      "step": 14318
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.0990025997161865,
+      "learning_rate": 8.95645166455208e-06,
+      "loss": 0.8185,
+      "step": 14319
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.9125128984451294,
+      "learning_rate": 8.956290447736558e-06,
+      "loss": 0.6119,
+      "step": 14320
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.566632032394409,
+      "learning_rate": 8.956129219920073e-06,
+      "loss": 0.6858,
+      "step": 14321
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 4.819601535797119,
+      "learning_rate": 8.955967981103076e-06,
+      "loss": 0.6876,
+      "step": 14322
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.1635921001434326,
+      "learning_rate": 8.955806731286012e-06,
+      "loss": 0.5384,
+      "step": 14323
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.0093109607696533,
+      "learning_rate": 8.955645470469332e-06,
+      "loss": 0.6393,
+      "step": 14324
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.9374823570251465,
+      "learning_rate": 8.955484198653486e-06,
+      "loss": 0.653,
+      "step": 14325
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.80234956741333,
+      "learning_rate": 8.955322915838918e-06,
+      "loss": 0.4582,
+      "step": 14326
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.2432005405426025,
+      "learning_rate": 8.955161622026079e-06,
+      "loss": 0.6969,
+      "step": 14327
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.437772512435913,
+      "learning_rate": 8.955000317215416e-06,
+      "loss": 0.7438,
+      "step": 14328
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.2869784832000732,
+      "learning_rate": 8.95483900140738e-06,
+      "loss": 0.687,
+      "step": 14329
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.2425878047943115,
+      "learning_rate": 8.95467767460242e-06,
+      "loss": 0.8264,
+      "step": 14330
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.920339345932007,
+      "learning_rate": 8.95451633680098e-06,
+      "loss": 0.5083,
+      "step": 14331
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.0330989360809326,
+      "learning_rate": 8.954354988003513e-06,
+      "loss": 0.7795,
+      "step": 14332
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.456796407699585,
+      "learning_rate": 8.954193628210466e-06,
+      "loss": 0.6361,
+      "step": 14333
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.0850703716278076,
+      "learning_rate": 8.954032257422286e-06,
+      "loss": 0.7253,
+      "step": 14334
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.273650646209717,
+      "learning_rate": 8.953870875639425e-06,
+      "loss": 1.0321,
+      "step": 14335
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.1231436729431152,
+      "learning_rate": 8.953709482862331e-06,
+      "loss": 0.4603,
+      "step": 14336
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.6008200645446777,
+      "learning_rate": 8.953548079091452e-06,
+      "loss": 0.7973,
+      "step": 14337
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.8885116577148438,
+      "learning_rate": 8.953386664327236e-06,
+      "loss": 0.9017,
+      "step": 14338
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.827885627746582,
+      "learning_rate": 8.953225238570132e-06,
+      "loss": 0.686,
+      "step": 14339
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.4861574172973633,
+      "learning_rate": 8.953063801820591e-06,
+      "loss": 0.6546,
+      "step": 14340
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.8573318719863892,
+      "learning_rate": 8.95290235407906e-06,
+      "loss": 0.6852,
+      "step": 14341
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.515637159347534,
+      "learning_rate": 8.952740895345988e-06,
+      "loss": 0.7699,
+      "step": 14342
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.2090678215026855,
+      "learning_rate": 8.952579425621824e-06,
+      "loss": 0.7296,
+      "step": 14343
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.077346086502075,
+      "learning_rate": 8.952417944907017e-06,
+      "loss": 0.8726,
+      "step": 14344
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.9609330892562866,
+      "learning_rate": 8.952256453202018e-06,
+      "loss": 0.8952,
+      "step": 14345
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.351238250732422,
+      "learning_rate": 8.952094950507274e-06,
+      "loss": 0.5112,
+      "step": 14346
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.306823253631592,
+      "learning_rate": 8.951933436823233e-06,
+      "loss": 0.4745,
+      "step": 14347
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.5879359245300293,
+      "learning_rate": 8.951771912150347e-06,
+      "loss": 0.6616,
+      "step": 14348
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.5873281955718994,
+      "learning_rate": 8.951610376489062e-06,
+      "loss": 0.8687,
+      "step": 14349
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.476007342338562,
+      "learning_rate": 8.95144882983983e-06,
+      "loss": 0.5331,
+      "step": 14350
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.4384117126464844,
+      "learning_rate": 8.951287272203097e-06,
+      "loss": 0.6905,
+      "step": 14351
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.053046703338623,
+      "learning_rate": 8.951125703579318e-06,
+      "loss": 0.6893,
+      "step": 14352
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.950870990753174,
+      "learning_rate": 8.950964123968936e-06,
+      "loss": 0.6318,
+      "step": 14353
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.9755698442459106,
+      "learning_rate": 8.950802533372403e-06,
+      "loss": 0.5884,
+      "step": 14354
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.0623555183410645,
+      "learning_rate": 8.950640931790169e-06,
+      "loss": 0.809,
+      "step": 14355
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.476991891860962,
+      "learning_rate": 8.950479319222682e-06,
+      "loss": 0.6004,
+      "step": 14356
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.584930419921875,
+      "learning_rate": 8.950317695670392e-06,
+      "loss": 0.8554,
+      "step": 14357
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.6499345302581787,
+      "learning_rate": 8.950156061133748e-06,
+      "loss": 0.7024,
+      "step": 14358
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.05745530128479,
+      "learning_rate": 8.949994415613201e-06,
+      "loss": 0.6928,
+      "step": 14359
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.1961491107940674,
+      "learning_rate": 8.949832759109199e-06,
+      "loss": 0.6896,
+      "step": 14360
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.8013575077056885,
+      "learning_rate": 8.94967109162219e-06,
+      "loss": 0.5886,
+      "step": 14361
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.9227211475372314,
+      "learning_rate": 8.949509413152627e-06,
+      "loss": 0.6076,
+      "step": 14362
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.8942837715148926,
+      "learning_rate": 8.949347723700957e-06,
+      "loss": 0.7138,
+      "step": 14363
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.77663254737854,
+      "learning_rate": 8.949186023267632e-06,
+      "loss": 0.5935,
+      "step": 14364
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.2486047744750977,
+      "learning_rate": 8.949024311853097e-06,
+      "loss": 0.781,
+      "step": 14365
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.8428661823272705,
+      "learning_rate": 8.948862589457807e-06,
+      "loss": 0.7074,
+      "step": 14366
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.6225168704986572,
+      "learning_rate": 8.948700856082208e-06,
+      "loss": 0.7371,
+      "step": 14367
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.3931455612182617,
+      "learning_rate": 8.948539111726753e-06,
+      "loss": 0.6029,
+      "step": 14368
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.3803834915161133,
+      "learning_rate": 8.948377356391888e-06,
+      "loss": 0.5068,
+      "step": 14369
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.6940884590148926,
+      "learning_rate": 8.948215590078067e-06,
+      "loss": 0.8076,
+      "step": 14370
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.800396203994751,
+      "learning_rate": 8.948053812785735e-06,
+      "loss": 0.8552,
+      "step": 14371
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.358819007873535,
+      "learning_rate": 8.947892024515345e-06,
+      "loss": 0.5143,
+      "step": 14372
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.171773910522461,
+      "learning_rate": 8.947730225267346e-06,
+      "loss": 0.8111,
+      "step": 14373
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.9527435302734375,
+      "learning_rate": 8.94756841504219e-06,
+      "loss": 0.6457,
+      "step": 14374
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.9404687881469727,
+      "learning_rate": 8.947406593840325e-06,
+      "loss": 0.5588,
+      "step": 14375
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.337045192718506,
+      "learning_rate": 8.947244761662199e-06,
+      "loss": 0.8978,
+      "step": 14376
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.013429880142212,
+      "learning_rate": 8.947082918508264e-06,
+      "loss": 0.5648,
+      "step": 14377
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.3851637840270996,
+      "learning_rate": 8.946921064378971e-06,
+      "loss": 0.6171,
+      "step": 14378
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.9747720956802368,
+      "learning_rate": 8.946759199274769e-06,
+      "loss": 0.5241,
+      "step": 14379
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.2143452167510986,
+      "learning_rate": 8.946597323196107e-06,
+      "loss": 0.8855,
+      "step": 14380
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.062742233276367,
+      "learning_rate": 8.946435436143438e-06,
+      "loss": 0.4488,
+      "step": 14381
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.930065155029297,
+      "learning_rate": 8.94627353811721e-06,
+      "loss": 0.629,
+      "step": 14382
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.7217488288879395,
+      "learning_rate": 8.946111629117873e-06,
+      "loss": 0.582,
+      "step": 14383
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 5.150395393371582,
+      "learning_rate": 8.945949709145878e-06,
+      "loss": 0.8773,
+      "step": 14384
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 4.1285600662231445,
+      "learning_rate": 8.945787778201674e-06,
+      "loss": 0.897,
+      "step": 14385
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.1986727714538574,
+      "learning_rate": 8.945625836285713e-06,
+      "loss": 0.9441,
+      "step": 14386
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.4096835851669312,
+      "learning_rate": 8.945463883398447e-06,
+      "loss": 0.3682,
+      "step": 14387
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.960206389427185,
+      "learning_rate": 8.945301919540321e-06,
+      "loss": 0.8343,
+      "step": 14388
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.3085384368896484,
+      "learning_rate": 8.945139944711789e-06,
+      "loss": 0.7574,
+      "step": 14389
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.137769937515259,
+      "learning_rate": 8.944977958913301e-06,
+      "loss": 0.675,
+      "step": 14390
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.8298306465148926,
+      "learning_rate": 8.944815962145306e-06,
+      "loss": 0.6975,
+      "step": 14391
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.630571126937866,
+      "learning_rate": 8.944653954408259e-06,
+      "loss": 0.5827,
+      "step": 14392
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.5458309650421143,
+      "learning_rate": 8.944491935702603e-06,
+      "loss": 0.5643,
+      "step": 14393
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.8186454772949219,
+      "learning_rate": 8.944329906028795e-06,
+      "loss": 0.6299,
+      "step": 14394
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.6517055034637451,
+      "learning_rate": 8.944167865387282e-06,
+      "loss": 0.6439,
+      "step": 14395
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.8895293474197388,
+      "learning_rate": 8.944005813778517e-06,
+      "loss": 0.6527,
+      "step": 14396
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.6311919689178467,
+      "learning_rate": 8.943843751202948e-06,
+      "loss": 0.515,
+      "step": 14397
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.3278627395629883,
+      "learning_rate": 8.943681677661027e-06,
+      "loss": 0.8765,
+      "step": 14398
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.6849926710128784,
+      "learning_rate": 8.943519593153206e-06,
+      "loss": 0.7041,
+      "step": 14399
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.7079603672027588,
+      "learning_rate": 8.943357497679931e-06,
+      "loss": 0.5379,
+      "step": 14400
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.428368330001831,
+      "learning_rate": 8.94319539124166e-06,
+      "loss": 0.6989,
+      "step": 14401
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.7438912391662598,
+      "learning_rate": 8.943033273838838e-06,
+      "loss": 0.4262,
+      "step": 14402
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.3832080364227295,
+      "learning_rate": 8.942871145471916e-06,
+      "loss": 0.5258,
+      "step": 14403
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.352604866027832,
+      "learning_rate": 8.942709006141348e-06,
+      "loss": 0.6744,
+      "step": 14404
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.582913637161255,
+      "learning_rate": 8.942546855847582e-06,
+      "loss": 0.631,
+      "step": 14405
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.1970114707946777,
+      "learning_rate": 8.942384694591073e-06,
+      "loss": 0.8702,
+      "step": 14406
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.5936055183410645,
+      "learning_rate": 8.942222522372267e-06,
+      "loss": 0.7305,
+      "step": 14407
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.642026424407959,
+      "learning_rate": 8.942060339191616e-06,
+      "loss": 0.6276,
+      "step": 14408
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.228767156600952,
+      "learning_rate": 8.941898145049574e-06,
+      "loss": 0.5776,
+      "step": 14409
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.3853816986083984,
+      "learning_rate": 8.941735939946588e-06,
+      "loss": 0.6934,
+      "step": 14410
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.3057703971862793,
+      "learning_rate": 8.941573723883112e-06,
+      "loss": 0.7347,
+      "step": 14411
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.429176092147827,
+      "learning_rate": 8.941411496859597e-06,
+      "loss": 1.0603,
+      "step": 14412
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.2699153423309326,
+      "learning_rate": 8.941249258876492e-06,
+      "loss": 0.6951,
+      "step": 14413
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.0684964656829834,
+      "learning_rate": 8.941087009934249e-06,
+      "loss": 0.5743,
+      "step": 14414
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.164402484893799,
+      "learning_rate": 8.94092475003332e-06,
+      "loss": 0.6323,
+      "step": 14415
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.4933481216430664,
+      "learning_rate": 8.940762479174155e-06,
+      "loss": 0.4779,
+      "step": 14416
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.5646989345550537,
+      "learning_rate": 8.940600197357207e-06,
+      "loss": 0.758,
+      "step": 14417
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.3328545093536377,
+      "learning_rate": 8.940437904582925e-06,
+      "loss": 0.4834,
+      "step": 14418
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.787893533706665,
+      "learning_rate": 8.94027560085176e-06,
+      "loss": 0.6154,
+      "step": 14419
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.116570472717285,
+      "learning_rate": 8.940113286164167e-06,
+      "loss": 0.8083,
+      "step": 14420
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.2342472076416016,
+      "learning_rate": 8.939950960520594e-06,
+      "loss": 0.7243,
+      "step": 14421
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.8593688011169434,
+      "learning_rate": 8.939788623921493e-06,
+      "loss": 0.7151,
+      "step": 14422
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.7971079349517822,
+      "learning_rate": 8.939626276367317e-06,
+      "loss": 0.5748,
+      "step": 14423
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.1648547649383545,
+      "learning_rate": 8.939463917858515e-06,
+      "loss": 0.5362,
+      "step": 14424
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 5.332996845245361,
+      "learning_rate": 8.93930154839554e-06,
+      "loss": 0.7787,
+      "step": 14425
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.4750993251800537,
+      "learning_rate": 8.939139167978843e-06,
+      "loss": 0.7775,
+      "step": 14426
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 4.592913627624512,
+      "learning_rate": 8.938976776608874e-06,
+      "loss": 0.9629,
+      "step": 14427
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.9010376930236816,
+      "learning_rate": 8.938814374286088e-06,
+      "loss": 0.6487,
+      "step": 14428
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.3894926309585571,
+      "learning_rate": 8.938651961010935e-06,
+      "loss": 0.7372,
+      "step": 14429
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.346325397491455,
+      "learning_rate": 8.938489536783865e-06,
+      "loss": 0.8523,
+      "step": 14430
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.05436110496521,
+      "learning_rate": 8.938327101605333e-06,
+      "loss": 0.7892,
+      "step": 14431
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.6469225883483887,
+      "learning_rate": 8.938164655475787e-06,
+      "loss": 0.6729,
+      "step": 14432
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.498138189315796,
+      "learning_rate": 8.93800219839568e-06,
+      "loss": 0.9013,
+      "step": 14433
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.586411237716675,
+      "learning_rate": 8.937839730365464e-06,
+      "loss": 0.4915,
+      "step": 14434
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.4075798988342285,
+      "learning_rate": 8.937677251385592e-06,
+      "loss": 0.8909,
+      "step": 14435
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.238492250442505,
+      "learning_rate": 8.937514761456516e-06,
+      "loss": 0.7005,
+      "step": 14436
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.429093837738037,
+      "learning_rate": 8.937352260578684e-06,
+      "loss": 0.5741,
+      "step": 14437
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.019381046295166,
+      "learning_rate": 8.937189748752551e-06,
+      "loss": 0.5176,
+      "step": 14438
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.669611692428589,
+      "learning_rate": 8.937027225978568e-06,
+      "loss": 0.5776,
+      "step": 14439
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 4.019074440002441,
+      "learning_rate": 8.936864692257187e-06,
+      "loss": 0.7461,
+      "step": 14440
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.953755259513855,
+      "learning_rate": 8.936702147588863e-06,
+      "loss": 0.6516,
+      "step": 14441
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 3.852137327194214,
+      "learning_rate": 8.936539591974042e-06,
+      "loss": 1.0995,
+      "step": 14442
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.920172929763794,
+      "learning_rate": 8.936377025413179e-06,
+      "loss": 0.5856,
+      "step": 14443
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.5137200355529785,
+      "learning_rate": 8.936214447906727e-06,
+      "loss": 0.4294,
+      "step": 14444
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 3.4185609817504883,
+      "learning_rate": 8.936051859455138e-06,
+      "loss": 0.6682,
+      "step": 14445
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.0528483390808105,
+      "learning_rate": 8.935889260058861e-06,
+      "loss": 0.6298,
+      "step": 14446
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.9235224723815918,
+      "learning_rate": 8.935726649718353e-06,
+      "loss": 0.5891,
+      "step": 14447
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 4.166275501251221,
+      "learning_rate": 8.93556402843406e-06,
+      "loss": 0.4692,
+      "step": 14448
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.1364150047302246,
+      "learning_rate": 8.93540139620644e-06,
+      "loss": 0.4012,
+      "step": 14449
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.5767865180969238,
+      "learning_rate": 8.935238753035944e-06,
+      "loss": 0.6463,
+      "step": 14450
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 3.3181567192077637,
+      "learning_rate": 8.93507609892302e-06,
+      "loss": 0.425,
+      "step": 14451
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.9237003326416016,
+      "learning_rate": 8.934913433868126e-06,
+      "loss": 0.7365,
+      "step": 14452
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.216398239135742,
+      "learning_rate": 8.934750757871713e-06,
+      "loss": 0.5458,
+      "step": 14453
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.189789056777954,
+      "learning_rate": 8.93458807093423e-06,
+      "loss": 0.4775,
+      "step": 14454
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.058476448059082,
+      "learning_rate": 8.934425373056131e-06,
+      "loss": 0.6565,
+      "step": 14455
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.4761918783187866,
+      "learning_rate": 8.93426266423787e-06,
+      "loss": 0.7124,
+      "step": 14456
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 4.318071365356445,
+      "learning_rate": 8.934099944479899e-06,
+      "loss": 0.6342,
+      "step": 14457
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.890814781188965,
+      "learning_rate": 8.933937213782668e-06,
+      "loss": 0.4787,
+      "step": 14458
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 3.914724111557007,
+      "learning_rate": 8.933774472146632e-06,
+      "loss": 0.9496,
+      "step": 14459
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.4196629524230957,
+      "learning_rate": 8.933611719572245e-06,
+      "loss": 0.7131,
+      "step": 14460
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.1698615550994873,
+      "learning_rate": 8.933448956059953e-06,
+      "loss": 0.8443,
+      "step": 14461
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 4.984518051147461,
+      "learning_rate": 8.933286181610216e-06,
+      "loss": 0.7061,
+      "step": 14462
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.6938655376434326,
+      "learning_rate": 8.933123396223484e-06,
+      "loss": 0.6128,
+      "step": 14463
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 4.113766670227051,
+      "learning_rate": 8.932960599900209e-06,
+      "loss": 0.7417,
+      "step": 14464
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.0142266750335693,
+      "learning_rate": 8.932797792640842e-06,
+      "loss": 0.49,
+      "step": 14465
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.523009777069092,
+      "learning_rate": 8.93263497444584e-06,
+      "loss": 0.6613,
+      "step": 14466
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.1406052112579346,
+      "learning_rate": 8.93247214531565e-06,
+      "loss": 0.7295,
+      "step": 14467
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 3.087284564971924,
+      "learning_rate": 8.93230930525073e-06,
+      "loss": 0.7786,
+      "step": 14468
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.4759912490844727,
+      "learning_rate": 8.932146454251531e-06,
+      "loss": 0.6518,
+      "step": 14469
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.5177927017211914,
+      "learning_rate": 8.931983592318507e-06,
+      "loss": 0.8118,
+      "step": 14470
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.7156693935394287,
+      "learning_rate": 8.931820719452109e-06,
+      "loss": 0.7569,
+      "step": 14471
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 3.5037691593170166,
+      "learning_rate": 8.931657835652789e-06,
+      "loss": 0.935,
+      "step": 14472
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.3068361282348633,
+      "learning_rate": 8.931494940921002e-06,
+      "loss": 0.473,
+      "step": 14473
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.743076205253601,
+      "learning_rate": 8.9313320352572e-06,
+      "loss": 0.5846,
+      "step": 14474
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.2796146869659424,
+      "learning_rate": 8.931169118661837e-06,
+      "loss": 0.7598,
+      "step": 14475
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 3.9718713760375977,
+      "learning_rate": 8.931006191135366e-06,
+      "loss": 0.6678,
+      "step": 14476
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 3.7078304290771484,
+      "learning_rate": 8.930843252678238e-06,
+      "loss": 0.6996,
+      "step": 14477
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 3.231478214263916,
+      "learning_rate": 8.930680303290909e-06,
+      "loss": 0.9358,
+      "step": 14478
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.1766550540924072,
+      "learning_rate": 8.930517342973829e-06,
+      "loss": 0.769,
+      "step": 14479
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.2340950965881348,
+      "learning_rate": 8.930354371727454e-06,
+      "loss": 0.9321,
+      "step": 14480
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.7273612022399902,
+      "learning_rate": 8.930191389552235e-06,
+      "loss": 0.7003,
+      "step": 14481
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.0640053749084473,
+      "learning_rate": 8.930028396448625e-06,
+      "loss": 0.9378,
+      "step": 14482
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.3189029693603516,
+      "learning_rate": 8.92986539241708e-06,
+      "loss": 0.5517,
+      "step": 14483
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.7500569820404053,
+      "learning_rate": 8.92970237745805e-06,
+      "loss": 0.6417,
+      "step": 14484
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 3.9571332931518555,
+      "learning_rate": 8.92953935157199e-06,
+      "loss": 0.5214,
+      "step": 14485
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.8665435314178467,
+      "learning_rate": 8.929376314759355e-06,
+      "loss": 0.6481,
+      "step": 14486
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 3.4866974353790283,
+      "learning_rate": 8.929213267020596e-06,
+      "loss": 0.8838,
+      "step": 14487
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.3730130195617676,
+      "learning_rate": 8.929050208356165e-06,
+      "loss": 0.7265,
+      "step": 14488
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.914732575416565,
+      "learning_rate": 8.928887138766517e-06,
+      "loss": 0.575,
+      "step": 14489
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.8682862520217896,
+      "learning_rate": 8.928724058252108e-06,
+      "loss": 0.7742,
+      "step": 14490
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.5902124643325806,
+      "learning_rate": 8.92856096681339e-06,
+      "loss": 0.8004,
+      "step": 14491
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.934884548187256,
+      "learning_rate": 8.928397864450812e-06,
+      "loss": 0.6895,
+      "step": 14492
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 6.259133338928223,
+      "learning_rate": 8.928234751164833e-06,
+      "loss": 0.5741,
+      "step": 14493
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.5976333618164062,
+      "learning_rate": 8.928071626955904e-06,
+      "loss": 0.7194,
+      "step": 14494
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.22851824760437,
+      "learning_rate": 8.92790849182448e-06,
+      "loss": 0.7175,
+      "step": 14495
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 3.109893321990967,
+      "learning_rate": 8.927745345771013e-06,
+      "loss": 0.4957,
+      "step": 14496
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.401366710662842,
+      "learning_rate": 8.92758218879596e-06,
+      "loss": 0.5799,
+      "step": 14497
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.7652602195739746,
+      "learning_rate": 8.92741902089977e-06,
+      "loss": 0.5595,
+      "step": 14498
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 4.316258430480957,
+      "learning_rate": 8.927255842082899e-06,
+      "loss": 0.5928,
+      "step": 14499
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.158543109893799,
+      "learning_rate": 8.927092652345802e-06,
+      "loss": 0.7009,
+      "step": 14500
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 4.48809289932251,
+      "learning_rate": 8.92692945168893e-06,
+      "loss": 0.7176,
+      "step": 14501
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.4719455242156982,
+      "learning_rate": 8.926766240112739e-06,
+      "loss": 0.6839,
+      "step": 14502
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.68232798576355,
+      "learning_rate": 8.92660301761768e-06,
+      "loss": 0.8647,
+      "step": 14503
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 4.1450958251953125,
+      "learning_rate": 8.926439784204212e-06,
+      "loss": 0.6921,
+      "step": 14504
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.456737995147705,
+      "learning_rate": 8.926276539872785e-06,
+      "loss": 0.6817,
+      "step": 14505
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.7736434936523438,
+      "learning_rate": 8.926113284623854e-06,
+      "loss": 0.4748,
+      "step": 14506
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.20481014251709,
+      "learning_rate": 8.925950018457872e-06,
+      "loss": 0.6487,
+      "step": 14507
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.0337939262390137,
+      "learning_rate": 8.925786741375294e-06,
+      "loss": 0.4153,
+      "step": 14508
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 3.152097225189209,
+      "learning_rate": 8.925623453376573e-06,
+      "loss": 0.7395,
+      "step": 14509
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 4.163193225860596,
+      "learning_rate": 8.925460154462164e-06,
+      "loss": 0.4822,
+      "step": 14510
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 3.038900136947632,
+      "learning_rate": 8.925296844632523e-06,
+      "loss": 0.4931,
+      "step": 14511
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.885190010070801,
+      "learning_rate": 8.925133523888098e-06,
+      "loss": 0.5531,
+      "step": 14512
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.58337664604187,
+      "learning_rate": 8.92497019222935e-06,
+      "loss": 0.7897,
+      "step": 14513
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 3.105998992919922,
+      "learning_rate": 8.92480684965673e-06,
+      "loss": 0.819,
+      "step": 14514
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.404290199279785,
+      "learning_rate": 8.924643496170691e-06,
+      "loss": 0.3475,
+      "step": 14515
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.1592040061950684,
+      "learning_rate": 8.924480131771688e-06,
+      "loss": 0.5213,
+      "step": 14516
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 4.5431694984436035,
+      "learning_rate": 8.924316756460179e-06,
+      "loss": 1.1004,
+      "step": 14517
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.5631920099258423,
+      "learning_rate": 8.924153370236614e-06,
+      "loss": 0.6217,
+      "step": 14518
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 3.66949200630188,
+      "learning_rate": 8.923989973101448e-06,
+      "loss": 0.8034,
+      "step": 14519
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.2884843349456787,
+      "learning_rate": 8.923826565055136e-06,
+      "loss": 0.4977,
+      "step": 14520
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.5274288654327393,
+      "learning_rate": 8.923663146098131e-06,
+      "loss": 0.4162,
+      "step": 14521
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.435567617416382,
+      "learning_rate": 8.92349971623089e-06,
+      "loss": 0.9652,
+      "step": 14522
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.7944960594177246,
+      "learning_rate": 8.923336275453866e-06,
+      "loss": 0.4377,
+      "step": 14523
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.850914478302002,
+      "learning_rate": 8.923172823767513e-06,
+      "loss": 0.7248,
+      "step": 14524
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 3.9799466133117676,
+      "learning_rate": 8.923009361172286e-06,
+      "loss": 0.6005,
+      "step": 14525
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.8322279453277588,
+      "learning_rate": 8.922845887668642e-06,
+      "loss": 0.7404,
+      "step": 14526
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 4.390231609344482,
+      "learning_rate": 8.92268240325703e-06,
+      "loss": 0.7184,
+      "step": 14527
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.1814169883728027,
+      "learning_rate": 8.92251890793791e-06,
+      "loss": 0.3332,
+      "step": 14528
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.56604266166687,
+      "learning_rate": 8.922355401711731e-06,
+      "loss": 0.7802,
+      "step": 14529
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.3906941413879395,
+      "learning_rate": 8.922191884578953e-06,
+      "loss": 0.6452,
+      "step": 14530
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.7610241174697876,
+      "learning_rate": 8.922028356540029e-06,
+      "loss": 0.6356,
+      "step": 14531
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.27958083152771,
+      "learning_rate": 8.921864817595413e-06,
+      "loss": 0.7359,
+      "step": 14532
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 3.089603900909424,
+      "learning_rate": 8.92170126774556e-06,
+      "loss": 0.3857,
+      "step": 14533
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.9388020038604736,
+      "learning_rate": 8.921537706990925e-06,
+      "loss": 0.7339,
+      "step": 14534
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.4464914798736572,
+      "learning_rate": 8.921374135331965e-06,
+      "loss": 0.7211,
+      "step": 14535
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.0899243354797363,
+      "learning_rate": 8.921210552769128e-06,
+      "loss": 0.7008,
+      "step": 14536
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 5.603985786437988,
+      "learning_rate": 8.921046959302876e-06,
+      "loss": 0.5502,
+      "step": 14537
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.465203285217285,
+      "learning_rate": 8.92088335493366e-06,
+      "loss": 0.5538,
+      "step": 14538
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.6999502182006836,
+      "learning_rate": 8.920719739661939e-06,
+      "loss": 0.7116,
+      "step": 14539
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.1502304077148438,
+      "learning_rate": 8.920556113488163e-06,
+      "loss": 0.668,
+      "step": 14540
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.3943042755126953,
+      "learning_rate": 8.920392476412788e-06,
+      "loss": 0.6798,
+      "step": 14541
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.715284824371338,
+      "learning_rate": 8.920228828436272e-06,
+      "loss": 0.5128,
+      "step": 14542
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.942193627357483,
+      "learning_rate": 8.920065169559066e-06,
+      "loss": 0.5426,
+      "step": 14543
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 4.181576251983643,
+      "learning_rate": 8.91990149978163e-06,
+      "loss": 0.9705,
+      "step": 14544
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.751162052154541,
+      "learning_rate": 8.919737819104416e-06,
+      "loss": 0.5527,
+      "step": 14545
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.842041254043579,
+      "learning_rate": 8.919574127527879e-06,
+      "loss": 0.7521,
+      "step": 14546
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 3.6231045722961426,
+      "learning_rate": 8.919410425052473e-06,
+      "loss": 0.7594,
+      "step": 14547
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.418739676475525,
+      "learning_rate": 8.919246711678658e-06,
+      "loss": 0.5486,
+      "step": 14548
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 3.6457650661468506,
+      "learning_rate": 8.919082987406883e-06,
+      "loss": 0.5945,
+      "step": 14549
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.6461119651794434,
+      "learning_rate": 8.918919252237608e-06,
+      "loss": 0.7479,
+      "step": 14550
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.055943012237549,
+      "learning_rate": 8.918755506171286e-06,
+      "loss": 0.5867,
+      "step": 14551
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 3.51406192779541,
+      "learning_rate": 8.918591749208374e-06,
+      "loss": 0.7129,
+      "step": 14552
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.1755728721618652,
+      "learning_rate": 8.918427981349325e-06,
+      "loss": 0.5426,
+      "step": 14553
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.9303176403045654,
+      "learning_rate": 8.918264202594596e-06,
+      "loss": 0.7406,
+      "step": 14554
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 3.925659418106079,
+      "learning_rate": 8.918100412944642e-06,
+      "loss": 0.4989,
+      "step": 14555
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.378406286239624,
+      "learning_rate": 8.917936612399918e-06,
+      "loss": 0.5027,
+      "step": 14556
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.3616271018981934,
+      "learning_rate": 8.917772800960882e-06,
+      "loss": 0.4023,
+      "step": 14557
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.720771312713623,
+      "learning_rate": 8.917608978627985e-06,
+      "loss": 0.7065,
+      "step": 14558
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 3.281874179840088,
+      "learning_rate": 8.917445145401686e-06,
+      "loss": 0.3093,
+      "step": 14559
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 3.850057363510132,
+      "learning_rate": 8.917281301282441e-06,
+      "loss": 0.6135,
+      "step": 14560
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.6839576959609985,
+      "learning_rate": 8.917117446270702e-06,
+      "loss": 0.624,
+      "step": 14561
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.9171496629714966,
+      "learning_rate": 8.916953580366929e-06,
+      "loss": 0.8029,
+      "step": 14562
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 3.089653968811035,
+      "learning_rate": 8.916789703571574e-06,
+      "loss": 0.5111,
+      "step": 14563
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.9404594898223877,
+      "learning_rate": 8.916625815885093e-06,
+      "loss": 0.6344,
+      "step": 14564
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.31306529045105,
+      "learning_rate": 8.916461917307944e-06,
+      "loss": 0.3389,
+      "step": 14565
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.8165847063064575,
+      "learning_rate": 8.916298007840583e-06,
+      "loss": 0.8038,
+      "step": 14566
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.0351743698120117,
+      "learning_rate": 8.91613408748346e-06,
+      "loss": 0.5266,
+      "step": 14567
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 3.3775341510772705,
+      "learning_rate": 8.91597015623704e-06,
+      "loss": 0.7151,
+      "step": 14568
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 4.628840923309326,
+      "learning_rate": 8.915806214101772e-06,
+      "loss": 0.6987,
+      "step": 14569
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.9881250858306885,
+      "learning_rate": 8.915642261078114e-06,
+      "loss": 0.6495,
+      "step": 14570
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.5036587715148926,
+      "learning_rate": 8.915478297166522e-06,
+      "loss": 0.5386,
+      "step": 14571
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.744144916534424,
+      "learning_rate": 8.91531432236745e-06,
+      "loss": 0.7548,
+      "step": 14572
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.6409826278686523,
+      "learning_rate": 8.915150336681357e-06,
+      "loss": 0.629,
+      "step": 14573
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.8239426612854004,
+      "learning_rate": 8.914986340108697e-06,
+      "loss": 0.3553,
+      "step": 14574
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.0033631324768066,
+      "learning_rate": 8.914822332649927e-06,
+      "loss": 0.6195,
+      "step": 14575
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 5.135595798492432,
+      "learning_rate": 8.914658314305502e-06,
+      "loss": 0.6906,
+      "step": 14576
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.9349308013916016,
+      "learning_rate": 8.914494285075878e-06,
+      "loss": 0.4221,
+      "step": 14577
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.4766552448272705,
+      "learning_rate": 8.914330244961514e-06,
+      "loss": 0.6291,
+      "step": 14578
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 6.8307719230651855,
+      "learning_rate": 8.914166193962863e-06,
+      "loss": 0.9304,
+      "step": 14579
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.1894421577453613,
+      "learning_rate": 8.914002132080382e-06,
+      "loss": 0.6719,
+      "step": 14580
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 3.157907724380493,
+      "learning_rate": 8.913838059314527e-06,
+      "loss": 0.9295,
+      "step": 14581
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.02008318901062,
+      "learning_rate": 8.913673975665754e-06,
+      "loss": 0.7497,
+      "step": 14582
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.6821366548538208,
+      "learning_rate": 8.91350988113452e-06,
+      "loss": 0.5286,
+      "step": 14583
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.8912876844406128,
+      "learning_rate": 8.913345775721282e-06,
+      "loss": 0.5156,
+      "step": 14584
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.4137020111083984,
+      "learning_rate": 8.913181659426496e-06,
+      "loss": 0.588,
+      "step": 14585
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.9191985130310059,
+      "learning_rate": 8.913017532250616e-06,
+      "loss": 0.6647,
+      "step": 14586
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.678738594055176,
+      "learning_rate": 8.912853394194101e-06,
+      "loss": 0.7079,
+      "step": 14587
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 3.15991473197937,
+      "learning_rate": 8.912689245257406e-06,
+      "loss": 0.5437,
+      "step": 14588
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 8.304712295532227,
+      "learning_rate": 8.912525085440989e-06,
+      "loss": 0.7772,
+      "step": 14589
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.5890886783599854,
+      "learning_rate": 8.912360914745303e-06,
+      "loss": 0.6588,
+      "step": 14590
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 3.0682311058044434,
+      "learning_rate": 8.91219673317081e-06,
+      "loss": 0.671,
+      "step": 14591
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.694242238998413,
+      "learning_rate": 8.91203254071796e-06,
+      "loss": 0.7762,
+      "step": 14592
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.9695130586624146,
+      "learning_rate": 8.911868337387217e-06,
+      "loss": 0.5242,
+      "step": 14593
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 3.0273213386535645,
+      "learning_rate": 8.91170412317903e-06,
+      "loss": 0.7072,
+      "step": 14594
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.4529948234558105,
+      "learning_rate": 8.911539898093861e-06,
+      "loss": 0.828,
+      "step": 14595
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.5659806728363037,
+      "learning_rate": 8.911375662132164e-06,
+      "loss": 0.8469,
+      "step": 14596
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.5455892086029053,
+      "learning_rate": 8.911211415294398e-06,
+      "loss": 0.6961,
+      "step": 14597
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.7596980333328247,
+      "learning_rate": 8.911047157581017e-06,
+      "loss": 0.6073,
+      "step": 14598
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 3.0889852046966553,
+      "learning_rate": 8.91088288899248e-06,
+      "loss": 1.031,
+      "step": 14599
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 4.14651346206665,
+      "learning_rate": 8.91071860952924e-06,
+      "loss": 0.7931,
+      "step": 14600
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.7236875295639038,
+      "learning_rate": 8.91055431919176e-06,
+      "loss": 0.6688,
+      "step": 14601
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.4506677389144897,
+      "learning_rate": 8.91039001798049e-06,
+      "loss": 0.507,
+      "step": 14602
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.2860875129699707,
+      "learning_rate": 8.910225705895893e-06,
+      "loss": 0.6984,
+      "step": 14603
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 3.7815775871276855,
+      "learning_rate": 8.910061382938421e-06,
+      "loss": 0.7048,
+      "step": 14604
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.869844675064087,
+      "learning_rate": 8.909897049108535e-06,
+      "loss": 0.5799,
+      "step": 14605
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.3568899631500244,
+      "learning_rate": 8.90973270440669e-06,
+      "loss": 0.7086,
+      "step": 14606
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.6062183380126953,
+      "learning_rate": 8.909568348833343e-06,
+      "loss": 0.5486,
+      "step": 14607
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.5016467571258545,
+      "learning_rate": 8.909403982388949e-06,
+      "loss": 0.7793,
+      "step": 14608
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.554844379425049,
+      "learning_rate": 8.909239605073967e-06,
+      "loss": 0.872,
+      "step": 14609
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 4.378716468811035,
+      "learning_rate": 8.909075216888857e-06,
+      "loss": 0.9169,
+      "step": 14610
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 4.433804035186768,
+      "learning_rate": 8.908910817834071e-06,
+      "loss": 0.6172,
+      "step": 14611
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.9309419393539429,
+      "learning_rate": 8.908746407910069e-06,
+      "loss": 0.681,
+      "step": 14612
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.2658472061157227,
+      "learning_rate": 8.908581987117307e-06,
+      "loss": 0.6765,
+      "step": 14613
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.118885040283203,
+      "learning_rate": 8.908417555456242e-06,
+      "loss": 0.475,
+      "step": 14614
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.0078775882720947,
+      "learning_rate": 8.908253112927333e-06,
+      "loss": 0.7291,
+      "step": 14615
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.418593168258667,
+      "learning_rate": 8.908088659531036e-06,
+      "loss": 0.5562,
+      "step": 14616
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.7477610111236572,
+      "learning_rate": 8.907924195267809e-06,
+      "loss": 0.6152,
+      "step": 14617
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.946272611618042,
+      "learning_rate": 8.907759720138109e-06,
+      "loss": 0.7022,
+      "step": 14618
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.3745546340942383,
+      "learning_rate": 8.907595234142393e-06,
+      "loss": 0.396,
+      "step": 14619
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.8648028373718262,
+      "learning_rate": 8.907430737281117e-06,
+      "loss": 0.6595,
+      "step": 14620
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 3.2253472805023193,
+      "learning_rate": 8.90726622955474e-06,
+      "loss": 0.7678,
+      "step": 14621
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.7926255464553833,
+      "learning_rate": 8.907101710963719e-06,
+      "loss": 0.5951,
+      "step": 14622
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.314791202545166,
+      "learning_rate": 8.906937181508514e-06,
+      "loss": 0.5099,
+      "step": 14623
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.363909959793091,
+      "learning_rate": 8.906772641189578e-06,
+      "loss": 0.6685,
+      "step": 14624
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 3.171295404434204,
+      "learning_rate": 8.906608090007373e-06,
+      "loss": 0.8573,
+      "step": 14625
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.9935598373413086,
+      "learning_rate": 8.906443527962352e-06,
+      "loss": 0.5928,
+      "step": 14626
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.469619035720825,
+      "learning_rate": 8.906278955054977e-06,
+      "loss": 0.7313,
+      "step": 14627
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.78761887550354,
+      "learning_rate": 8.906114371285702e-06,
+      "loss": 0.7263,
+      "step": 14628
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.633301258087158,
+      "learning_rate": 8.905949776654987e-06,
+      "loss": 0.5529,
+      "step": 14629
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.112016201019287,
+      "learning_rate": 8.905785171163288e-06,
+      "loss": 0.8599,
+      "step": 14630
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 3.0503878593444824,
+      "learning_rate": 8.905620554811066e-06,
+      "loss": 0.5743,
+      "step": 14631
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 3.6144325733184814,
+      "learning_rate": 8.905455927598774e-06,
+      "loss": 0.7913,
+      "step": 14632
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 4.1377644538879395,
+      "learning_rate": 8.905291289526872e-06,
+      "loss": 0.8913,
+      "step": 14633
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.8314006328582764,
+      "learning_rate": 8.905126640595819e-06,
+      "loss": 0.5603,
+      "step": 14634
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.89814031124115,
+      "learning_rate": 8.90496198080607e-06,
+      "loss": 0.7561,
+      "step": 14635
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.342664957046509,
+      "learning_rate": 8.904797310158086e-06,
+      "loss": 0.7652,
+      "step": 14636
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.9336891174316406,
+      "learning_rate": 8.904632628652324e-06,
+      "loss": 0.4036,
+      "step": 14637
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.4767141342163086,
+      "learning_rate": 8.904467936289241e-06,
+      "loss": 0.7088,
+      "step": 14638
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.845311164855957,
+      "learning_rate": 8.904303233069295e-06,
+      "loss": 0.729,
+      "step": 14639
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.4190924167633057,
+      "learning_rate": 8.904138518992944e-06,
+      "loss": 0.8456,
+      "step": 14640
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.947226047515869,
+      "learning_rate": 8.903973794060648e-06,
+      "loss": 0.6122,
+      "step": 14641
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 4.454245090484619,
+      "learning_rate": 8.90380905827286e-06,
+      "loss": 1.0714,
+      "step": 14642
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.6741487979888916,
+      "learning_rate": 8.903644311630044e-06,
+      "loss": 0.4344,
+      "step": 14643
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 3.378380060195923,
+      "learning_rate": 8.903479554132655e-06,
+      "loss": 0.5814,
+      "step": 14644
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.6887283325195312,
+      "learning_rate": 8.903314785781152e-06,
+      "loss": 0.5445,
+      "step": 14645
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.4523509740829468,
+      "learning_rate": 8.90315000657599e-06,
+      "loss": 0.557,
+      "step": 14646
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 3.3634982109069824,
+      "learning_rate": 8.902985216517634e-06,
+      "loss": 0.8521,
+      "step": 14647
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 4.795854568481445,
+      "learning_rate": 8.902820415606536e-06,
+      "loss": 0.6451,
+      "step": 14648
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.3997511863708496,
+      "learning_rate": 8.902655603843158e-06,
+      "loss": 0.7629,
+      "step": 14649
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.3081488609313965,
+      "learning_rate": 8.902490781227955e-06,
+      "loss": 0.5359,
+      "step": 14650
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.3285915851593018,
+      "learning_rate": 8.902325947761388e-06,
+      "loss": 0.7228,
+      "step": 14651
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.458008289337158,
+      "learning_rate": 8.902161103443913e-06,
+      "loss": 0.7741,
+      "step": 14652
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 4.744142532348633,
+      "learning_rate": 8.901996248275992e-06,
+      "loss": 0.8332,
+      "step": 14653
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 3.782881259918213,
+      "learning_rate": 8.90183138225808e-06,
+      "loss": 0.8616,
+      "step": 14654
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 3.880605936050415,
+      "learning_rate": 8.901666505390636e-06,
+      "loss": 0.9077,
+      "step": 14655
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.3961868286132812,
+      "learning_rate": 8.90150161767412e-06,
+      "loss": 0.8076,
+      "step": 14656
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.796459913253784,
+      "learning_rate": 8.901336719108989e-06,
+      "loss": 0.8507,
+      "step": 14657
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 3.0899910926818848,
+      "learning_rate": 8.9011718096957e-06,
+      "loss": 0.6845,
+      "step": 14658
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 3.9265198707580566,
+      "learning_rate": 8.901006889434718e-06,
+      "loss": 0.5182,
+      "step": 14659
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.911782145500183,
+      "learning_rate": 8.900841958326495e-06,
+      "loss": 0.5105,
+      "step": 14660
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.377974510192871,
+      "learning_rate": 8.90067701637149e-06,
+      "loss": 0.7935,
+      "step": 14661
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.9396177530288696,
+      "learning_rate": 8.900512063570167e-06,
+      "loss": 0.7666,
+      "step": 14662
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 3.052367925643921,
+      "learning_rate": 8.90034709992298e-06,
+      "loss": 0.7219,
+      "step": 14663
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.0906994342803955,
+      "learning_rate": 8.900182125430388e-06,
+      "loss": 0.6879,
+      "step": 14664
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 3.6687216758728027,
+      "learning_rate": 8.900017140092849e-06,
+      "loss": 0.5741,
+      "step": 14665
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 6.6663079261779785,
+      "learning_rate": 8.899852143910826e-06,
+      "loss": 0.6297,
+      "step": 14666
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.9864916801452637,
+      "learning_rate": 8.899687136884773e-06,
+      "loss": 0.6608,
+      "step": 14667
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 3.612638473510742,
+      "learning_rate": 8.899522119015153e-06,
+      "loss": 0.4123,
+      "step": 14668
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.6373708248138428,
+      "learning_rate": 8.899357090302424e-06,
+      "loss": 0.7266,
+      "step": 14669
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.9703768491744995,
+      "learning_rate": 8.899192050747039e-06,
+      "loss": 0.4759,
+      "step": 14670
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.5529091358184814,
+      "learning_rate": 8.899027000349465e-06,
+      "loss": 0.9945,
+      "step": 14671
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.0332891941070557,
+      "learning_rate": 8.898861939110156e-06,
+      "loss": 0.7317,
+      "step": 14672
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 3.8039820194244385,
+      "learning_rate": 8.898696867029573e-06,
+      "loss": 0.6818,
+      "step": 14673
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.879807710647583,
+      "learning_rate": 8.898531784108175e-06,
+      "loss": 0.6348,
+      "step": 14674
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.7582014799118042,
+      "learning_rate": 8.89836669034642e-06,
+      "loss": 0.4331,
+      "step": 14675
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.0362253189086914,
+      "learning_rate": 8.898201585744767e-06,
+      "loss": 0.4215,
+      "step": 14676
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.1614861488342285,
+      "learning_rate": 8.898036470303677e-06,
+      "loss": 0.7786,
+      "step": 14677
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.3952643871307373,
+      "learning_rate": 8.897871344023608e-06,
+      "loss": 0.6719,
+      "step": 14678
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 3.2852184772491455,
+      "learning_rate": 8.897706206905019e-06,
+      "loss": 0.5401,
+      "step": 14679
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.4907286167144775,
+      "learning_rate": 8.897541058948366e-06,
+      "loss": 0.6615,
+      "step": 14680
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.489190697669983,
+      "learning_rate": 8.897375900154114e-06,
+      "loss": 0.7569,
+      "step": 14681
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.7263152599334717,
+      "learning_rate": 8.897210730522721e-06,
+      "loss": 0.5911,
+      "step": 14682
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.8601489067077637,
+      "learning_rate": 8.897045550054642e-06,
+      "loss": 0.7039,
+      "step": 14683
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.603079319000244,
+      "learning_rate": 8.896880358750342e-06,
+      "loss": 0.7052,
+      "step": 14684
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.03562331199646,
+      "learning_rate": 8.896715156610275e-06,
+      "loss": 0.5956,
+      "step": 14685
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 3.7618162631988525,
+      "learning_rate": 8.896549943634904e-06,
+      "loss": 0.7515,
+      "step": 14686
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.1474521160125732,
+      "learning_rate": 8.896384719824687e-06,
+      "loss": 0.6439,
+      "step": 14687
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.6478687524795532,
+      "learning_rate": 8.896219485180084e-06,
+      "loss": 0.5399,
+      "step": 14688
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 6.272049427032471,
+      "learning_rate": 8.896054239701554e-06,
+      "loss": 0.6911,
+      "step": 14689
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.9689693450927734,
+      "learning_rate": 8.895888983389556e-06,
+      "loss": 0.7449,
+      "step": 14690
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.559646725654602,
+      "learning_rate": 8.895723716244551e-06,
+      "loss": 0.3973,
+      "step": 14691
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.3520112037658691,
+      "learning_rate": 8.895558438266999e-06,
+      "loss": 0.8523,
+      "step": 14692
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.7242319583892822,
+      "learning_rate": 8.895393149457356e-06,
+      "loss": 0.72,
+      "step": 14693
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.768648624420166,
+      "learning_rate": 8.895227849816084e-06,
+      "loss": 0.7346,
+      "step": 14694
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.7662962675094604,
+      "learning_rate": 8.895062539343643e-06,
+      "loss": 0.4878,
+      "step": 14695
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 3.543156147003174,
+      "learning_rate": 8.894897218040494e-06,
+      "loss": 0.8144,
+      "step": 14696
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.114945650100708,
+      "learning_rate": 8.894731885907094e-06,
+      "loss": 0.6615,
+      "step": 14697
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.8469996452331543,
+      "learning_rate": 8.8945665429439e-06,
+      "loss": 0.6388,
+      "step": 14698
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.9192769527435303,
+      "learning_rate": 8.89440118915138e-06,
+      "loss": 0.5726,
+      "step": 14699
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 3.2630364894866943,
+      "learning_rate": 8.894235824529988e-06,
+      "loss": 0.6804,
+      "step": 14700
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 3.0430374145507812,
+      "learning_rate": 8.894070449080183e-06,
+      "loss": 0.6058,
+      "step": 14701
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.7332754135131836,
+      "learning_rate": 8.89390506280243e-06,
+      "loss": 0.9371,
+      "step": 14702
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 3.0588693618774414,
+      "learning_rate": 8.893739665697184e-06,
+      "loss": 0.8509,
+      "step": 14703
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 3.6608378887176514,
+      "learning_rate": 8.893574257764905e-06,
+      "loss": 0.4611,
+      "step": 14704
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 4.694374084472656,
+      "learning_rate": 8.893408839006057e-06,
+      "loss": 0.7822,
+      "step": 14705
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 5.895643711090088,
+      "learning_rate": 8.893243409421097e-06,
+      "loss": 0.68,
+      "step": 14706
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 7.091893672943115,
+      "learning_rate": 8.893077969010486e-06,
+      "loss": 0.7445,
+      "step": 14707
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 5.0292181968688965,
+      "learning_rate": 8.892912517774682e-06,
+      "loss": 0.672,
+      "step": 14708
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.031559467315674,
+      "learning_rate": 8.892747055714148e-06,
+      "loss": 0.7246,
+      "step": 14709
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.88669753074646,
+      "learning_rate": 8.892581582829342e-06,
+      "loss": 0.7317,
+      "step": 14710
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.3710455894470215,
+      "learning_rate": 8.892416099120726e-06,
+      "loss": 0.7925,
+      "step": 14711
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.0182089805603027,
+      "learning_rate": 8.892250604588758e-06,
+      "loss": 1.0087,
+      "step": 14712
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.5228148698806763,
+      "learning_rate": 8.892085099233897e-06,
+      "loss": 0.5802,
+      "step": 14713
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.2195446491241455,
+      "learning_rate": 8.891919583056609e-06,
+      "loss": 0.4724,
+      "step": 14714
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.407137632369995,
+      "learning_rate": 8.89175405605735e-06,
+      "loss": 0.5188,
+      "step": 14715
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.233837366104126,
+      "learning_rate": 8.891588518236579e-06,
+      "loss": 0.9296,
+      "step": 14716
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.796748638153076,
+      "learning_rate": 8.891422969594758e-06,
+      "loss": 0.4922,
+      "step": 14717
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.031630516052246,
+      "learning_rate": 8.891257410132348e-06,
+      "loss": 0.6062,
+      "step": 14718
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.76517915725708,
+      "learning_rate": 8.89109183984981e-06,
+      "loss": 0.935,
+      "step": 14719
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.8394808769226074,
+      "learning_rate": 8.890926258747602e-06,
+      "loss": 0.831,
+      "step": 14720
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 3.9149928092956543,
+      "learning_rate": 8.890760666826187e-06,
+      "loss": 0.7433,
+      "step": 14721
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.10585355758667,
+      "learning_rate": 8.890595064086022e-06,
+      "loss": 0.7185,
+      "step": 14722
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.0650243759155273,
+      "learning_rate": 8.890429450527573e-06,
+      "loss": 0.6468,
+      "step": 14723
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.4397146701812744,
+      "learning_rate": 8.890263826151294e-06,
+      "loss": 0.7876,
+      "step": 14724
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 3.03847599029541,
+      "learning_rate": 8.89009819095765e-06,
+      "loss": 0.5235,
+      "step": 14725
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.033329486846924,
+      "learning_rate": 8.8899325449471e-06,
+      "loss": 0.8068,
+      "step": 14726
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.7156126499176025,
+      "learning_rate": 8.889766888120104e-06,
+      "loss": 0.762,
+      "step": 14727
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.306096076965332,
+      "learning_rate": 8.889601220477125e-06,
+      "loss": 0.5272,
+      "step": 14728
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 3.9320969581604004,
+      "learning_rate": 8.889435542018622e-06,
+      "loss": 0.7949,
+      "step": 14729
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 3.307161808013916,
+      "learning_rate": 8.889269852745053e-06,
+      "loss": 0.4277,
+      "step": 14730
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 3.184101104736328,
+      "learning_rate": 8.889104152656882e-06,
+      "loss": 0.4698,
+      "step": 14731
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 3.8073341846466064,
+      "learning_rate": 8.888938441754571e-06,
+      "loss": 0.7696,
+      "step": 14732
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.5713694095611572,
+      "learning_rate": 8.88877272003858e-06,
+      "loss": 0.7037,
+      "step": 14733
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 3.1657557487487793,
+      "learning_rate": 8.888606987509367e-06,
+      "loss": 0.7751,
+      "step": 14734
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.2863333225250244,
+      "learning_rate": 8.888441244167394e-06,
+      "loss": 0.8959,
+      "step": 14735
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.302424430847168,
+      "learning_rate": 8.888275490013125e-06,
+      "loss": 0.8592,
+      "step": 14736
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.065830707550049,
+      "learning_rate": 8.888109725047016e-06,
+      "loss": 0.6567,
+      "step": 14737
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.857113003730774,
+      "learning_rate": 8.887943949269534e-06,
+      "loss": 0.594,
+      "step": 14738
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.7917064428329468,
+      "learning_rate": 8.887778162681132e-06,
+      "loss": 0.7576,
+      "step": 14739
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 3.97560453414917,
+      "learning_rate": 8.887612365282278e-06,
+      "loss": 0.7541,
+      "step": 14740
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 7.018538951873779,
+      "learning_rate": 8.887446557073428e-06,
+      "loss": 0.7351,
+      "step": 14741
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.1690313816070557,
+      "learning_rate": 8.88728073805505e-06,
+      "loss": 0.7786,
+      "step": 14742
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.560084819793701,
+      "learning_rate": 8.887114908227595e-06,
+      "loss": 0.5601,
+      "step": 14743
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.8640947341918945,
+      "learning_rate": 8.886949067591532e-06,
+      "loss": 0.8471,
+      "step": 14744
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.298403024673462,
+      "learning_rate": 8.88678321614732e-06,
+      "loss": 0.5026,
+      "step": 14745
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.6779316663742065,
+      "learning_rate": 8.88661735389542e-06,
+      "loss": 0.6533,
+      "step": 14746
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 5.608635902404785,
+      "learning_rate": 8.886451480836292e-06,
+      "loss": 0.7401,
+      "step": 14747
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 3.3052756786346436,
+      "learning_rate": 8.886285596970399e-06,
+      "loss": 0.8387,
+      "step": 14748
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.2829411029815674,
+      "learning_rate": 8.886119702298201e-06,
+      "loss": 0.5703,
+      "step": 14749
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.139695882797241,
+      "learning_rate": 8.88595379682016e-06,
+      "loss": 0.5822,
+      "step": 14750
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.8910938501358032,
+      "learning_rate": 8.885787880536739e-06,
+      "loss": 0.6617,
+      "step": 14751
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.205042839050293,
+      "learning_rate": 8.885621953448396e-06,
+      "loss": 0.5236,
+      "step": 14752
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.4046523571014404,
+      "learning_rate": 8.885456015555595e-06,
+      "loss": 0.6637,
+      "step": 14753
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.4769389629364014,
+      "learning_rate": 8.885290066858794e-06,
+      "loss": 0.5419,
+      "step": 14754
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.5708528757095337,
+      "learning_rate": 8.88512410735846e-06,
+      "loss": 0.5083,
+      "step": 14755
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.8120152950286865,
+      "learning_rate": 8.88495813705505e-06,
+      "loss": 0.8131,
+      "step": 14756
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.036569356918335,
+      "learning_rate": 8.884792155949026e-06,
+      "loss": 0.495,
+      "step": 14757
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.883015513420105,
+      "learning_rate": 8.88462616404085e-06,
+      "loss": 0.6562,
+      "step": 14758
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.9742932319641113,
+      "learning_rate": 8.884460161330984e-06,
+      "loss": 0.4602,
+      "step": 14759
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.4994698762893677,
+      "learning_rate": 8.88429414781989e-06,
+      "loss": 0.3456,
+      "step": 14760
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.627194881439209,
+      "learning_rate": 8.884128123508029e-06,
+      "loss": 0.7059,
+      "step": 14761
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.875486135482788,
+      "learning_rate": 8.883962088395863e-06,
+      "loss": 1.1478,
+      "step": 14762
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.2161054611206055,
+      "learning_rate": 8.883796042483853e-06,
+      "loss": 0.6688,
+      "step": 14763
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.6695632934570312,
+      "learning_rate": 8.883629985772462e-06,
+      "loss": 0.3785,
+      "step": 14764
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 4.596360683441162,
+      "learning_rate": 8.88346391826215e-06,
+      "loss": 0.7164,
+      "step": 14765
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.779330015182495,
+      "learning_rate": 8.883297839953378e-06,
+      "loss": 0.6454,
+      "step": 14766
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.6119112968444824,
+      "learning_rate": 8.883131750846613e-06,
+      "loss": 0.472,
+      "step": 14767
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.6699745655059814,
+      "learning_rate": 8.88296565094231e-06,
+      "loss": 0.6997,
+      "step": 14768
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.627103805541992,
+      "learning_rate": 8.882799540240935e-06,
+      "loss": 0.6204,
+      "step": 14769
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.716583251953125,
+      "learning_rate": 8.882633418742949e-06,
+      "loss": 0.5734,
+      "step": 14770
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 3.4486100673675537,
+      "learning_rate": 8.882467286448816e-06,
+      "loss": 0.6492,
+      "step": 14771
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 3.460941791534424,
+      "learning_rate": 8.882301143358994e-06,
+      "loss": 0.4874,
+      "step": 14772
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.16532564163208,
+      "learning_rate": 8.882134989473947e-06,
+      "loss": 0.9249,
+      "step": 14773
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.9352478981018066,
+      "learning_rate": 8.881968824794138e-06,
+      "loss": 0.6596,
+      "step": 14774
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 4.006577491760254,
+      "learning_rate": 8.881802649320027e-06,
+      "loss": 0.8122,
+      "step": 14775
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.08974289894104,
+      "learning_rate": 8.881636463052077e-06,
+      "loss": 0.874,
+      "step": 14776
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.6659934520721436,
+      "learning_rate": 8.881470265990749e-06,
+      "loss": 0.727,
+      "step": 14777
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 6.661565780639648,
+      "learning_rate": 8.881304058136508e-06,
+      "loss": 0.4058,
+      "step": 14778
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.1451704502105713,
+      "learning_rate": 8.881137839489812e-06,
+      "loss": 0.3897,
+      "step": 14779
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 5.32816743850708,
+      "learning_rate": 8.880971610051127e-06,
+      "loss": 0.734,
+      "step": 14780
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.7350962162017822,
+      "learning_rate": 8.880805369820916e-06,
+      "loss": 0.9759,
+      "step": 14781
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.096256971359253,
+      "learning_rate": 8.880639118799636e-06,
+      "loss": 0.525,
+      "step": 14782
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.8519885540008545,
+      "learning_rate": 8.880472856987754e-06,
+      "loss": 0.5936,
+      "step": 14783
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.2754921913146973,
+      "learning_rate": 8.88030658438573e-06,
+      "loss": 0.5264,
+      "step": 14784
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.5855188369750977,
+      "learning_rate": 8.880140300994026e-06,
+      "loss": 0.4463,
+      "step": 14785
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.483654260635376,
+      "learning_rate": 8.879974006813108e-06,
+      "loss": 0.5771,
+      "step": 14786
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 3.3670175075531006,
+      "learning_rate": 8.879807701843433e-06,
+      "loss": 0.7067,
+      "step": 14787
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.426248550415039,
+      "learning_rate": 8.879641386085467e-06,
+      "loss": 0.7686,
+      "step": 14788
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 3.0158677101135254,
+      "learning_rate": 8.879475059539672e-06,
+      "loss": 0.8321,
+      "step": 14789
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 3.3889756202697754,
+      "learning_rate": 8.879308722206512e-06,
+      "loss": 0.6389,
+      "step": 14790
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.3755886554718018,
+      "learning_rate": 8.879142374086444e-06,
+      "loss": 0.7635,
+      "step": 14791
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.8799922466278076,
+      "learning_rate": 8.878976015179937e-06,
+      "loss": 0.676,
+      "step": 14792
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.3079333305358887,
+      "learning_rate": 8.87880964548745e-06,
+      "loss": 0.6594,
+      "step": 14793
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.731851816177368,
+      "learning_rate": 8.878643265009446e-06,
+      "loss": 0.8761,
+      "step": 14794
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.3199143409729004,
+      "learning_rate": 8.878476873746389e-06,
+      "loss": 0.6614,
+      "step": 14795
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.3761537075042725,
+      "learning_rate": 8.878310471698738e-06,
+      "loss": 0.6868,
+      "step": 14796
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 5.947066307067871,
+      "learning_rate": 8.878144058866961e-06,
+      "loss": 0.8904,
+      "step": 14797
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.2870856523513794,
+      "learning_rate": 8.877977635251518e-06,
+      "loss": 0.718,
+      "step": 14798
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.1927437782287598,
+      "learning_rate": 8.87781120085287e-06,
+      "loss": 0.8258,
+      "step": 14799
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.1339452266693115,
+      "learning_rate": 8.877644755671484e-06,
+      "loss": 0.6776,
+      "step": 14800
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.55787992477417,
+      "learning_rate": 8.87747829970782e-06,
+      "loss": 0.8461,
+      "step": 14801
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.3615760803222656,
+      "learning_rate": 8.877311832962341e-06,
+      "loss": 0.5908,
+      "step": 14802
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 3.398716926574707,
+      "learning_rate": 8.877145355435511e-06,
+      "loss": 0.7405,
+      "step": 14803
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 3.9847795963287354,
+      "learning_rate": 8.87697886712779e-06,
+      "loss": 0.8152,
+      "step": 14804
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 4.4743828773498535,
+      "learning_rate": 8.876812368039645e-06,
+      "loss": 0.4837,
+      "step": 14805
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 3.7411653995513916,
+      "learning_rate": 8.876645858171538e-06,
+      "loss": 0.5544,
+      "step": 14806
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.7714424133300781,
+      "learning_rate": 8.87647933752393e-06,
+      "loss": 0.3664,
+      "step": 14807
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 3.5642495155334473,
+      "learning_rate": 8.876312806097284e-06,
+      "loss": 0.7735,
+      "step": 14808
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.2378499507904053,
+      "learning_rate": 8.876146263892067e-06,
+      "loss": 0.5513,
+      "step": 14809
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.579960823059082,
+      "learning_rate": 8.875979710908737e-06,
+      "loss": 0.622,
+      "step": 14810
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.7307195663452148,
+      "learning_rate": 8.875813147147759e-06,
+      "loss": 0.7073,
+      "step": 14811
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 3.1382644176483154,
+      "learning_rate": 8.875646572609598e-06,
+      "loss": 0.7572,
+      "step": 14812
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.6796947717666626,
+      "learning_rate": 8.875479987294714e-06,
+      "loss": 0.5557,
+      "step": 14813
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 4.582531452178955,
+      "learning_rate": 8.875313391203575e-06,
+      "loss": 0.5005,
+      "step": 14814
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.744290828704834,
+      "learning_rate": 8.875146784336637e-06,
+      "loss": 0.7739,
+      "step": 14815
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 3.825268030166626,
+      "learning_rate": 8.874980166694372e-06,
+      "loss": 0.789,
+      "step": 14816
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.5566611289978027,
+      "learning_rate": 8.874813538277236e-06,
+      "loss": 0.7171,
+      "step": 14817
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.8682506084442139,
+      "learning_rate": 8.874646899085696e-06,
+      "loss": 0.4953,
+      "step": 14818
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.113485336303711,
+      "learning_rate": 8.874480249120213e-06,
+      "loss": 0.5287,
+      "step": 14819
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.2100934982299805,
+      "learning_rate": 8.874313588381253e-06,
+      "loss": 0.4998,
+      "step": 14820
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.304198980331421,
+      "learning_rate": 8.874146916869277e-06,
+      "loss": 0.6541,
+      "step": 14821
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.5288567543029785,
+      "learning_rate": 8.873980234584752e-06,
+      "loss": 0.628,
+      "step": 14822
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 4.040139198303223,
+      "learning_rate": 8.873813541528138e-06,
+      "loss": 0.6079,
+      "step": 14823
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 3.685091733932495,
+      "learning_rate": 8.8736468376999e-06,
+      "loss": 0.7036,
+      "step": 14824
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.0057222843170166,
+      "learning_rate": 8.8734801231005e-06,
+      "loss": 0.4946,
+      "step": 14825
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.3419885635375977,
+      "learning_rate": 8.873313397730405e-06,
+      "loss": 0.7217,
+      "step": 14826
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.214982748031616,
+      "learning_rate": 8.873146661590075e-06,
+      "loss": 0.8154,
+      "step": 14827
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.1867470741271973,
+      "learning_rate": 8.872979914679975e-06,
+      "loss": 0.6647,
+      "step": 14828
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.5110269784927368,
+      "learning_rate": 8.872813157000569e-06,
+      "loss": 0.8753,
+      "step": 14829
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.855336904525757,
+      "learning_rate": 8.87264638855232e-06,
+      "loss": 1.0458,
+      "step": 14830
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.9618980884552002,
+      "learning_rate": 8.872479609335693e-06,
+      "loss": 0.8165,
+      "step": 14831
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.8486324548721313,
+      "learning_rate": 8.872312819351149e-06,
+      "loss": 0.7409,
+      "step": 14832
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 4.328358173370361,
+      "learning_rate": 8.872146018599156e-06,
+      "loss": 0.7843,
+      "step": 14833
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.8340836763381958,
+      "learning_rate": 8.871979207080175e-06,
+      "loss": 0.568,
+      "step": 14834
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.247570276260376,
+      "learning_rate": 8.871812384794669e-06,
+      "loss": 0.5766,
+      "step": 14835
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.941856622695923,
+      "learning_rate": 8.871645551743104e-06,
+      "loss": 0.6133,
+      "step": 14836
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.4992363452911377,
+      "learning_rate": 8.871478707925944e-06,
+      "loss": 0.6247,
+      "step": 14837
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 3.3965554237365723,
+      "learning_rate": 8.87131185334365e-06,
+      "loss": 0.8476,
+      "step": 14838
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 4.235958576202393,
+      "learning_rate": 8.87114498799669e-06,
+      "loss": 0.8249,
+      "step": 14839
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.7920923233032227,
+      "learning_rate": 8.870978111885524e-06,
+      "loss": 0.5046,
+      "step": 14840
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.0613040924072266,
+      "learning_rate": 8.87081122501062e-06,
+      "loss": 0.7884,
+      "step": 14841
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.669590711593628,
+      "learning_rate": 8.870644327372437e-06,
+      "loss": 0.5406,
+      "step": 14842
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.7473275661468506,
+      "learning_rate": 8.870477418971445e-06,
+      "loss": 0.7235,
+      "step": 14843
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.9278817176818848,
+      "learning_rate": 8.870310499808104e-06,
+      "loss": 0.3977,
+      "step": 14844
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.486067771911621,
+      "learning_rate": 8.87014356988288e-06,
+      "loss": 0.439,
+      "step": 14845
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.9625180959701538,
+      "learning_rate": 8.869976629196237e-06,
+      "loss": 0.895,
+      "step": 14846
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.3206512928009033,
+      "learning_rate": 8.869809677748638e-06,
+      "loss": 0.5254,
+      "step": 14847
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 3.8555121421813965,
+      "learning_rate": 8.869642715540547e-06,
+      "loss": 0.4917,
+      "step": 14848
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.3328087329864502,
+      "learning_rate": 8.86947574257243e-06,
+      "loss": 0.6612,
+      "step": 14849
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 4.258424282073975,
+      "learning_rate": 8.86930875884475e-06,
+      "loss": 0.453,
+      "step": 14850
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.8665666580200195,
+      "learning_rate": 8.869141764357973e-06,
+      "loss": 0.6704,
+      "step": 14851
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.526128053665161,
+      "learning_rate": 8.86897475911256e-06,
+      "loss": 0.5985,
+      "step": 14852
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.0245256423950195,
+      "learning_rate": 8.86880774310898e-06,
+      "loss": 0.7771,
+      "step": 14853
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.5778322219848633,
+      "learning_rate": 8.868640716347694e-06,
+      "loss": 0.5551,
+      "step": 14854
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.765263795852661,
+      "learning_rate": 8.868473678829168e-06,
+      "loss": 0.6662,
+      "step": 14855
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.711007833480835,
+      "learning_rate": 8.868306630553865e-06,
+      "loss": 0.7066,
+      "step": 14856
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 4.2410888671875,
+      "learning_rate": 8.868139571522251e-06,
+      "loss": 0.649,
+      "step": 14857
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.529919385910034,
+      "learning_rate": 8.86797250173479e-06,
+      "loss": 0.958,
+      "step": 14858
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 3.1934475898742676,
+      "learning_rate": 8.867805421191945e-06,
+      "loss": 0.8146,
+      "step": 14859
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.8794870376586914,
+      "learning_rate": 8.867638329894183e-06,
+      "loss": 0.538,
+      "step": 14860
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.028712272644043,
+      "learning_rate": 8.867471227841968e-06,
+      "loss": 0.7268,
+      "step": 14861
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 3.1380984783172607,
+      "learning_rate": 8.867304115035763e-06,
+      "loss": 0.6369,
+      "step": 14862
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.887648820877075,
+      "learning_rate": 8.867136991476034e-06,
+      "loss": 0.677,
+      "step": 14863
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.6286098957061768,
+      "learning_rate": 8.866969857163248e-06,
+      "loss": 1.0005,
+      "step": 14864
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.0885238647460938,
+      "learning_rate": 8.866802712097866e-06,
+      "loss": 0.7706,
+      "step": 14865
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 4.578159809112549,
+      "learning_rate": 8.866635556280353e-06,
+      "loss": 0.8053,
+      "step": 14866
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.5158069133758545,
+      "learning_rate": 8.866468389711178e-06,
+      "loss": 0.923,
+      "step": 14867
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.302717685699463,
+      "learning_rate": 8.8663012123908e-06,
+      "loss": 0.7723,
+      "step": 14868
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.8401930332183838,
+      "learning_rate": 8.866134024319687e-06,
+      "loss": 0.8156,
+      "step": 14869
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 4.115475654602051,
+      "learning_rate": 8.865966825498305e-06,
+      "loss": 0.6621,
+      "step": 14870
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.4004133939743042,
+      "learning_rate": 8.865799615927117e-06,
+      "loss": 0.5174,
+      "step": 14871
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.5415849685668945,
+      "learning_rate": 8.865632395606588e-06,
+      "loss": 0.7928,
+      "step": 14872
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.1570260524749756,
+      "learning_rate": 8.865465164537184e-06,
+      "loss": 0.5942,
+      "step": 14873
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.1800332069396973,
+      "learning_rate": 8.86529792271937e-06,
+      "loss": 0.6371,
+      "step": 14874
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 3.3665034770965576,
+      "learning_rate": 8.865130670153607e-06,
+      "loss": 0.7975,
+      "step": 14875
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.838015556335449,
+      "learning_rate": 8.864963406840368e-06,
+      "loss": 0.4131,
+      "step": 14876
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.0188992023468018,
+      "learning_rate": 8.864796132780112e-06,
+      "loss": 0.7422,
+      "step": 14877
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.9523375034332275,
+      "learning_rate": 8.864628847973304e-06,
+      "loss": 0.6846,
+      "step": 14878
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 3.3137166500091553,
+      "learning_rate": 8.864461552420414e-06,
+      "loss": 0.7317,
+      "step": 14879
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.831536054611206,
+      "learning_rate": 8.8642942461219e-06,
+      "loss": 0.6712,
+      "step": 14880
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 5.509222507476807,
+      "learning_rate": 8.864126929078234e-06,
+      "loss": 0.7605,
+      "step": 14881
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.9176435470581055,
+      "learning_rate": 8.863959601289878e-06,
+      "loss": 0.6655,
+      "step": 14882
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.370124340057373,
+      "learning_rate": 8.863792262757297e-06,
+      "loss": 0.5385,
+      "step": 14883
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.6216739416122437,
+      "learning_rate": 8.863624913480959e-06,
+      "loss": 0.5987,
+      "step": 14884
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.0390639305114746,
+      "learning_rate": 8.863457553461328e-06,
+      "loss": 0.4916,
+      "step": 14885
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.265838623046875,
+      "learning_rate": 8.863290182698867e-06,
+      "loss": 0.5769,
+      "step": 14886
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.8624258041381836,
+      "learning_rate": 8.863122801194043e-06,
+      "loss": 0.8575,
+      "step": 14887
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 3.034605026245117,
+      "learning_rate": 8.862955408947322e-06,
+      "loss": 0.3646,
+      "step": 14888
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.0800578594207764,
+      "learning_rate": 8.862788005959171e-06,
+      "loss": 0.457,
+      "step": 14889
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 4.514899730682373,
+      "learning_rate": 8.862620592230051e-06,
+      "loss": 0.7493,
+      "step": 14890
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.642984628677368,
+      "learning_rate": 8.86245316776043e-06,
+      "loss": 0.5296,
+      "step": 14891
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 3.8389761447906494,
+      "learning_rate": 8.862285732550775e-06,
+      "loss": 0.375,
+      "step": 14892
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.1013119220733643,
+      "learning_rate": 8.862118286601552e-06,
+      "loss": 0.7162,
+      "step": 14893
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.812848448753357,
+      "learning_rate": 8.861950829913221e-06,
+      "loss": 0.5734,
+      "step": 14894
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.3671412467956543,
+      "learning_rate": 8.861783362486255e-06,
+      "loss": 0.8018,
+      "step": 14895
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 3.0618984699249268,
+      "learning_rate": 8.861615884321112e-06,
+      "loss": 0.7575,
+      "step": 14896
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 3.203798294067383,
+      "learning_rate": 8.861448395418266e-06,
+      "loss": 0.7064,
+      "step": 14897
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.9520084857940674,
+      "learning_rate": 8.861280895778175e-06,
+      "loss": 0.4386,
+      "step": 14898
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.166419744491577,
+      "learning_rate": 8.86111338540131e-06,
+      "loss": 0.8109,
+      "step": 14899
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.574491262435913,
+      "learning_rate": 8.860945864288135e-06,
+      "loss": 0.4527,
+      "step": 14900
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.30380916595459,
+      "learning_rate": 8.860778332439117e-06,
+      "loss": 0.691,
+      "step": 14901
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 3.378401756286621,
+      "learning_rate": 8.86061078985472e-06,
+      "loss": 0.4648,
+      "step": 14902
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.7989107370376587,
+      "learning_rate": 8.860443236535409e-06,
+      "loss": 0.9935,
+      "step": 14903
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 3.6971867084503174,
+      "learning_rate": 8.860275672481655e-06,
+      "loss": 0.8459,
+      "step": 14904
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.6039140224456787,
+      "learning_rate": 8.860108097693917e-06,
+      "loss": 0.6872,
+      "step": 14905
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.4907565116882324,
+      "learning_rate": 8.859940512172667e-06,
+      "loss": 0.7933,
+      "step": 14906
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.5621964931488037,
+      "learning_rate": 8.859772915918367e-06,
+      "loss": 0.6754,
+      "step": 14907
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.9662259817123413,
+      "learning_rate": 8.859605308931484e-06,
+      "loss": 0.6998,
+      "step": 14908
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 3.3692173957824707,
+      "learning_rate": 8.859437691212485e-06,
+      "loss": 0.5583,
+      "step": 14909
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.374075174331665,
+      "learning_rate": 8.859270062761836e-06,
+      "loss": 0.317,
+      "step": 14910
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.4423904418945312,
+      "learning_rate": 8.859102423580001e-06,
+      "loss": 0.7654,
+      "step": 14911
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.0570099353790283,
+      "learning_rate": 8.85893477366745e-06,
+      "loss": 0.5616,
+      "step": 14912
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 3.71008563041687,
+      "learning_rate": 8.858767113024646e-06,
+      "loss": 0.7352,
+      "step": 14913
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.037681818008423,
+      "learning_rate": 8.858599441652056e-06,
+      "loss": 0.5966,
+      "step": 14914
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.044952154159546,
+      "learning_rate": 8.858431759550147e-06,
+      "loss": 0.5118,
+      "step": 14915
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.0533995628356934,
+      "learning_rate": 8.858264066719384e-06,
+      "loss": 0.7854,
+      "step": 14916
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.147036552429199,
+      "learning_rate": 8.858096363160234e-06,
+      "loss": 0.6098,
+      "step": 14917
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 3.476679563522339,
+      "learning_rate": 8.857928648873164e-06,
+      "loss": 0.6967,
+      "step": 14918
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.1216325759887695,
+      "learning_rate": 8.85776092385864e-06,
+      "loss": 0.6068,
+      "step": 14919
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.2107276916503906,
+      "learning_rate": 8.857593188117126e-06,
+      "loss": 0.7093,
+      "step": 14920
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.1064391136169434,
+      "learning_rate": 8.857425441649092e-06,
+      "loss": 0.6565,
+      "step": 14921
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.469454765319824,
+      "learning_rate": 8.857257684455003e-06,
+      "loss": 0.6008,
+      "step": 14922
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.135632038116455,
+      "learning_rate": 8.857089916535325e-06,
+      "loss": 0.7152,
+      "step": 14923
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.916234254837036,
+      "learning_rate": 8.856922137890525e-06,
+      "loss": 0.6009,
+      "step": 14924
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.259821653366089,
+      "learning_rate": 8.856754348521068e-06,
+      "loss": 0.5156,
+      "step": 14925
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.814917802810669,
+      "learning_rate": 8.856586548427424e-06,
+      "loss": 0.552,
+      "step": 14926
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.6283924579620361,
+      "learning_rate": 8.856418737610054e-06,
+      "loss": 0.6,
+      "step": 14927
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.724608898162842,
+      "learning_rate": 8.856250916069433e-06,
+      "loss": 0.7469,
+      "step": 14928
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 3.1548800468444824,
+      "learning_rate": 8.856083083806019e-06,
+      "loss": 0.6909,
+      "step": 14929
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 3.029566526412964,
+      "learning_rate": 8.855915240820285e-06,
+      "loss": 0.6702,
+      "step": 14930
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.205764055252075,
+      "learning_rate": 8.855747387112694e-06,
+      "loss": 0.7651,
+      "step": 14931
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.1988916397094727,
+      "learning_rate": 8.855579522683714e-06,
+      "loss": 0.5902,
+      "step": 14932
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 3.0429768562316895,
+      "learning_rate": 8.855411647533812e-06,
+      "loss": 0.6212,
+      "step": 14933
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.7730673551559448,
+      "learning_rate": 8.855243761663455e-06,
+      "loss": 0.9568,
+      "step": 14934
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.2984097003936768,
+      "learning_rate": 8.855075865073108e-06,
+      "loss": 0.4096,
+      "step": 14935
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 3.559638738632202,
+      "learning_rate": 8.85490795776324e-06,
+      "loss": 0.933,
+      "step": 14936
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.9322822093963623,
+      "learning_rate": 8.854740039734317e-06,
+      "loss": 0.8814,
+      "step": 14937
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 3.0445218086242676,
+      "learning_rate": 8.854572110986805e-06,
+      "loss": 0.664,
+      "step": 14938
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.9802892208099365,
+      "learning_rate": 8.854404171521172e-06,
+      "loss": 0.6086,
+      "step": 14939
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 3.4565269947052,
+      "learning_rate": 8.854236221337886e-06,
+      "loss": 0.76,
+      "step": 14940
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 3.0307211875915527,
+      "learning_rate": 8.854068260437411e-06,
+      "loss": 1.0965,
+      "step": 14941
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.7785959243774414,
+      "learning_rate": 8.853900288820219e-06,
+      "loss": 0.6008,
+      "step": 14942
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 3.3411052227020264,
+      "learning_rate": 8.853732306486772e-06,
+      "loss": 0.658,
+      "step": 14943
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.9254822731018066,
+      "learning_rate": 8.85356431343754e-06,
+      "loss": 0.6462,
+      "step": 14944
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.0616955757141113,
+      "learning_rate": 8.853396309672989e-06,
+      "loss": 0.6103,
+      "step": 14945
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 3.3625311851501465,
+      "learning_rate": 8.853228295193586e-06,
+      "loss": 0.6109,
+      "step": 14946
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.4550976753234863,
+      "learning_rate": 8.853060269999798e-06,
+      "loss": 0.5654,
+      "step": 14947
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 3.171447515487671,
+      "learning_rate": 8.852892234092093e-06,
+      "loss": 0.7609,
+      "step": 14948
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.5746042728424072,
+      "learning_rate": 8.852724187470937e-06,
+      "loss": 0.5594,
+      "step": 14949
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 3.5135204792022705,
+      "learning_rate": 8.8525561301368e-06,
+      "loss": 0.6069,
+      "step": 14950
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.6040492057800293,
+      "learning_rate": 8.85238806209015e-06,
+      "loss": 0.6963,
+      "step": 14951
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.7087886333465576,
+      "learning_rate": 8.852219983331446e-06,
+      "loss": 0.6984,
+      "step": 14952
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.1389920711517334,
+      "learning_rate": 8.852051893861166e-06,
+      "loss": 0.4447,
+      "step": 14953
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 3.2990171909332275,
+      "learning_rate": 8.851883793679773e-06,
+      "loss": 0.8749,
+      "step": 14954
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 3.8501598834991455,
+      "learning_rate": 8.85171568278773e-06,
+      "loss": 0.3871,
+      "step": 14955
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.933195948600769,
+      "learning_rate": 8.851547561185511e-06,
+      "loss": 0.7052,
+      "step": 14956
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.600435495376587,
+      "learning_rate": 8.851379428873581e-06,
+      "loss": 0.6349,
+      "step": 14957
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 3.4602527618408203,
+      "learning_rate": 8.851211285852408e-06,
+      "loss": 0.6215,
+      "step": 14958
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.5474414825439453,
+      "learning_rate": 8.851043132122459e-06,
+      "loss": 0.6411,
+      "step": 14959
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.768178939819336,
+      "learning_rate": 8.8508749676842e-06,
+      "loss": 0.7561,
+      "step": 14960
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.860438585281372,
+      "learning_rate": 8.850706792538101e-06,
+      "loss": 0.6514,
+      "step": 14961
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.6668667793273926,
+      "learning_rate": 8.850538606684631e-06,
+      "loss": 0.741,
+      "step": 14962
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.6380579471588135,
+      "learning_rate": 8.850370410124253e-06,
+      "loss": 0.5905,
+      "step": 14963
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.2557830810546875,
+      "learning_rate": 8.85020220285744e-06,
+      "loss": 0.5335,
+      "step": 14964
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.3947181701660156,
+      "learning_rate": 8.850033984884654e-06,
+      "loss": 0.5968,
+      "step": 14965
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 3.1507492065429688,
+      "learning_rate": 8.849865756206369e-06,
+      "loss": 0.7233,
+      "step": 14966
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 3.461409091949463,
+      "learning_rate": 8.849697516823048e-06,
+      "loss": 0.7297,
+      "step": 14967
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.640930414199829,
+      "learning_rate": 8.84952926673516e-06,
+      "loss": 0.5793,
+      "step": 14968
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.9945178031921387,
+      "learning_rate": 8.849361005943171e-06,
+      "loss": 0.464,
+      "step": 14969
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.043689727783203,
+      "learning_rate": 8.849192734447554e-06,
+      "loss": 0.6983,
+      "step": 14970
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.5616331100463867,
+      "learning_rate": 8.849024452248773e-06,
+      "loss": 0.5511,
+      "step": 14971
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.8946971893310547,
+      "learning_rate": 8.848856159347296e-06,
+      "loss": 0.6974,
+      "step": 14972
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.058032751083374,
+      "learning_rate": 8.848687855743593e-06,
+      "loss": 0.8031,
+      "step": 14973
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.323859453201294,
+      "learning_rate": 8.848519541438133e-06,
+      "loss": 0.6531,
+      "step": 14974
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.5749244689941406,
+      "learning_rate": 8.84835121643138e-06,
+      "loss": 0.6273,
+      "step": 14975
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.6764425039291382,
+      "learning_rate": 8.848182880723802e-06,
+      "loss": 0.5264,
+      "step": 14976
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.8875963687896729,
+      "learning_rate": 8.84801453431587e-06,
+      "loss": 0.8059,
+      "step": 14977
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.9458746910095215,
+      "learning_rate": 8.847846177208052e-06,
+      "loss": 0.7134,
+      "step": 14978
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.834235668182373,
+      "learning_rate": 8.847677809400814e-06,
+      "loss": 0.8378,
+      "step": 14979
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.345665216445923,
+      "learning_rate": 8.847509430894628e-06,
+      "loss": 0.7387,
+      "step": 14980
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.436922550201416,
+      "learning_rate": 8.847341041689957e-06,
+      "loss": 0.6133,
+      "step": 14981
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.3783867359161377,
+      "learning_rate": 8.847172641787271e-06,
+      "loss": 0.6747,
+      "step": 14982
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.698400855064392,
+      "learning_rate": 8.847004231187043e-06,
+      "loss": 0.6578,
+      "step": 14983
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.7455998659133911,
+      "learning_rate": 8.846835809889733e-06,
+      "loss": 0.6198,
+      "step": 14984
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.988809108734131,
+      "learning_rate": 8.846667377895815e-06,
+      "loss": 0.6674,
+      "step": 14985
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 3.3882689476013184,
+      "learning_rate": 8.846498935205757e-06,
+      "loss": 0.5177,
+      "step": 14986
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 3.264364719390869,
+      "learning_rate": 8.846330481820026e-06,
+      "loss": 0.8059,
+      "step": 14987
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.1670033931732178,
+      "learning_rate": 8.84616201773909e-06,
+      "loss": 0.8087,
+      "step": 14988
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.3188838958740234,
+      "learning_rate": 8.84599354296342e-06,
+      "loss": 0.5265,
+      "step": 14989
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.2610912322998047,
+      "learning_rate": 8.845825057493482e-06,
+      "loss": 0.7143,
+      "step": 14990
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.890357494354248,
+      "learning_rate": 8.845656561329746e-06,
+      "loss": 0.8117,
+      "step": 14991
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.743622303009033,
+      "learning_rate": 8.845488054472678e-06,
+      "loss": 0.5975,
+      "step": 14992
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.8093671798706055,
+      "learning_rate": 8.845319536922747e-06,
+      "loss": 0.8595,
+      "step": 14993
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 3.2399227619171143,
+      "learning_rate": 8.845151008680426e-06,
+      "loss": 0.6207,
+      "step": 14994
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.082432508468628,
+      "learning_rate": 8.84498246974618e-06,
+      "loss": 0.586,
+      "step": 14995
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.735148310661316,
+      "learning_rate": 8.844813920120477e-06,
+      "loss": 0.6508,
+      "step": 14996
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.027773380279541,
+      "learning_rate": 8.844645359803787e-06,
+      "loss": 0.8667,
+      "step": 14997
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.4000258445739746,
+      "learning_rate": 8.844476788796578e-06,
+      "loss": 0.8461,
+      "step": 14998
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 6.0648698806762695,
+      "learning_rate": 8.84430820709932e-06,
+      "loss": 0.6603,
+      "step": 14999
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 4.65696382522583,
+      "learning_rate": 8.844139614712481e-06,
+      "loss": 0.4801,
+      "step": 15000
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.8921704292297363,
+      "learning_rate": 8.84397101163653e-06,
+      "loss": 0.7595,
+      "step": 15001
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.200784206390381,
+      "learning_rate": 8.843802397871935e-06,
+      "loss": 0.4582,
+      "step": 15002
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.7137724161148071,
+      "learning_rate": 8.843633773419165e-06,
+      "loss": 0.3008,
+      "step": 15003
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 4.058442115783691,
+      "learning_rate": 8.843465138278691e-06,
+      "loss": 0.7827,
+      "step": 15004
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 3.7274649143218994,
+      "learning_rate": 8.843296492450979e-06,
+      "loss": 0.4588,
+      "step": 15005
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.6551620960235596,
+      "learning_rate": 8.8431278359365e-06,
+      "loss": 0.4286,
+      "step": 15006
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.46942138671875,
+      "learning_rate": 8.84295916873572e-06,
+      "loss": 0.5536,
+      "step": 15007
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.9614231586456299,
+      "learning_rate": 8.842790490849112e-06,
+      "loss": 0.7185,
+      "step": 15008
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.9370994567871094,
+      "learning_rate": 8.842621802277142e-06,
+      "loss": 0.5044,
+      "step": 15009
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 3.0046491622924805,
+      "learning_rate": 8.842453103020282e-06,
+      "loss": 0.5748,
+      "step": 15010
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 3.0251870155334473,
+      "learning_rate": 8.842284393079e-06,
+      "loss": 0.7817,
+      "step": 15011
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 3.4150121212005615,
+      "learning_rate": 8.842115672453762e-06,
+      "loss": 0.7525,
+      "step": 15012
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.8405338525772095,
+      "learning_rate": 8.841946941145042e-06,
+      "loss": 0.7431,
+      "step": 15013
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 3.681748628616333,
+      "learning_rate": 8.841778199153303e-06,
+      "loss": 0.7685,
+      "step": 15014
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.724160075187683,
+      "learning_rate": 8.84160944647902e-06,
+      "loss": 0.6794,
+      "step": 15015
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.1323554515838623,
+      "learning_rate": 8.841440683122662e-06,
+      "loss": 0.69,
+      "step": 15016
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.5867624282836914,
+      "learning_rate": 8.841271909084694e-06,
+      "loss": 0.8188,
+      "step": 15017
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.3219223022460938,
+      "learning_rate": 8.84110312436559e-06,
+      "loss": 0.4321,
+      "step": 15018
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.677811861038208,
+      "learning_rate": 8.840934328965816e-06,
+      "loss": 0.7403,
+      "step": 15019
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.8454535007476807,
+      "learning_rate": 8.840765522885842e-06,
+      "loss": 0.6319,
+      "step": 15020
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 4.21033239364624,
+      "learning_rate": 8.840596706126139e-06,
+      "loss": 0.7642,
+      "step": 15021
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 3.76234769821167,
+      "learning_rate": 8.840427878687173e-06,
+      "loss": 0.5548,
+      "step": 15022
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 3.1402175426483154,
+      "learning_rate": 8.840259040569417e-06,
+      "loss": 0.7065,
+      "step": 15023
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.9605915546417236,
+      "learning_rate": 8.840090191773341e-06,
+      "loss": 0.6079,
+      "step": 15024
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.7747457027435303,
+      "learning_rate": 8.83992133229941e-06,
+      "loss": 0.5437,
+      "step": 15025
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.7014654874801636,
+      "learning_rate": 8.8397524621481e-06,
+      "loss": 0.6882,
+      "step": 15026
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 4.059200763702393,
+      "learning_rate": 8.839583581319873e-06,
+      "loss": 0.6194,
+      "step": 15027
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.7045447826385498,
+      "learning_rate": 8.839414689815203e-06,
+      "loss": 0.4383,
+      "step": 15028
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.1743290424346924,
+      "learning_rate": 8.83924578763456e-06,
+      "loss": 0.7154,
+      "step": 15029
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.2885024547576904,
+      "learning_rate": 8.839076874778412e-06,
+      "loss": 0.8422,
+      "step": 15030
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.9703856706619263,
+      "learning_rate": 8.838907951247231e-06,
+      "loss": 0.7793,
+      "step": 15031
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.716556429862976,
+      "learning_rate": 8.838739017041482e-06,
+      "loss": 0.5888,
+      "step": 15032
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.8520221710205078,
+      "learning_rate": 8.83857007216164e-06,
+      "loss": 0.6999,
+      "step": 15033
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.108553171157837,
+      "learning_rate": 8.838401116608172e-06,
+      "loss": 0.8918,
+      "step": 15034
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.0821704864501953,
+      "learning_rate": 8.838232150381549e-06,
+      "loss": 0.6833,
+      "step": 15035
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.9754698276519775,
+      "learning_rate": 8.83806317348224e-06,
+      "loss": 0.6751,
+      "step": 15036
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.914702296257019,
+      "learning_rate": 8.837894185910716e-06,
+      "loss": 0.7231,
+      "step": 15037
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.7492917776107788,
+      "learning_rate": 8.837725187667444e-06,
+      "loss": 0.6448,
+      "step": 15038
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.3717206716537476,
+      "learning_rate": 8.837556178752898e-06,
+      "loss": 0.6013,
+      "step": 15039
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.800853967666626,
+      "learning_rate": 8.837387159167542e-06,
+      "loss": 0.4422,
+      "step": 15040
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.4933385848999023,
+      "learning_rate": 8.837218128911855e-06,
+      "loss": 0.6569,
+      "step": 15041
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.98374605178833,
+      "learning_rate": 8.837049087986299e-06,
+      "loss": 0.6418,
+      "step": 15042
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.590869665145874,
+      "learning_rate": 8.836880036391346e-06,
+      "loss": 0.4525,
+      "step": 15043
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.9302741289138794,
+      "learning_rate": 8.83671097412747e-06,
+      "loss": 0.5826,
+      "step": 15044
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.2593830823898315,
+      "learning_rate": 8.836541901195136e-06,
+      "loss": 0.7167,
+      "step": 15045
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.8339338302612305,
+      "learning_rate": 8.836372817594816e-06,
+      "loss": 0.6997,
+      "step": 15046
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.1438050270080566,
+      "learning_rate": 8.83620372332698e-06,
+      "loss": 0.7461,
+      "step": 15047
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.7466965913772583,
+      "learning_rate": 8.8360346183921e-06,
+      "loss": 0.4654,
+      "step": 15048
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.561016321182251,
+      "learning_rate": 8.835865502790643e-06,
+      "loss": 0.6669,
+      "step": 15049
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.9394631385803223,
+      "learning_rate": 8.835696376523082e-06,
+      "loss": 0.6704,
+      "step": 15050
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.929321050643921,
+      "learning_rate": 8.835527239589887e-06,
+      "loss": 0.8316,
+      "step": 15051
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.198214530944824,
+      "learning_rate": 8.835358091991526e-06,
+      "loss": 0.8935,
+      "step": 15052
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.4759888648986816,
+      "learning_rate": 8.83518893372847e-06,
+      "loss": 0.4876,
+      "step": 15053
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.4616544246673584,
+      "learning_rate": 8.835019764801194e-06,
+      "loss": 0.5425,
+      "step": 15054
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.6886157989501953,
+      "learning_rate": 8.834850585210161e-06,
+      "loss": 0.8863,
+      "step": 15055
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.7457600831985474,
+      "learning_rate": 8.834681394955849e-06,
+      "loss": 0.6791,
+      "step": 15056
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.7381823062896729,
+      "learning_rate": 8.834512194038722e-06,
+      "loss": 0.3237,
+      "step": 15057
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.8712003231048584,
+      "learning_rate": 8.834342982459252e-06,
+      "loss": 0.9476,
+      "step": 15058
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.429656982421875,
+      "learning_rate": 8.834173760217912e-06,
+      "loss": 0.6391,
+      "step": 15059
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.4885077476501465,
+      "learning_rate": 8.834004527315172e-06,
+      "loss": 0.729,
+      "step": 15060
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.1152381896972656,
+      "learning_rate": 8.8338352837515e-06,
+      "loss": 0.6289,
+      "step": 15061
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.1606690883636475,
+      "learning_rate": 8.833666029527369e-06,
+      "loss": 0.3788,
+      "step": 15062
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.2564988136291504,
+      "learning_rate": 8.833496764643249e-06,
+      "loss": 0.7327,
+      "step": 15063
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.126124858856201,
+      "learning_rate": 8.83332748909961e-06,
+      "loss": 0.3609,
+      "step": 15064
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.6759345531463623,
+      "learning_rate": 8.833158202896923e-06,
+      "loss": 0.5869,
+      "step": 15065
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.2438395023345947,
+      "learning_rate": 8.83298890603566e-06,
+      "loss": 0.6335,
+      "step": 15066
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.874683141708374,
+      "learning_rate": 8.83281959851629e-06,
+      "loss": 0.6697,
+      "step": 15067
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 4.479242324829102,
+      "learning_rate": 8.832650280339285e-06,
+      "loss": 0.9532,
+      "step": 15068
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.3353490829467773,
+      "learning_rate": 8.832480951505116e-06,
+      "loss": 0.5211,
+      "step": 15069
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.041666269302368,
+      "learning_rate": 8.83231161201425e-06,
+      "loss": 0.6968,
+      "step": 15070
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.7236149311065674,
+      "learning_rate": 8.832142261867166e-06,
+      "loss": 0.6615,
+      "step": 15071
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.8963731527328491,
+      "learning_rate": 8.831972901064326e-06,
+      "loss": 0.5288,
+      "step": 15072
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.21686053276062,
+      "learning_rate": 8.831803529606207e-06,
+      "loss": 0.3516,
+      "step": 15073
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 6.369561672210693,
+      "learning_rate": 8.831634147493276e-06,
+      "loss": 0.8813,
+      "step": 15074
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.587337017059326,
+      "learning_rate": 8.831464754726007e-06,
+      "loss": 0.785,
+      "step": 15075
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.6762983798980713,
+      "learning_rate": 8.83129535130487e-06,
+      "loss": 0.671,
+      "step": 15076
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.139195203781128,
+      "learning_rate": 8.831125937230334e-06,
+      "loss": 0.4515,
+      "step": 15077
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.515549659729004,
+      "learning_rate": 8.830956512502872e-06,
+      "loss": 0.6844,
+      "step": 15078
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.9563424587249756,
+      "learning_rate": 8.830787077122958e-06,
+      "loss": 0.4644,
+      "step": 15079
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.4820556640625,
+      "learning_rate": 8.830617631091058e-06,
+      "loss": 0.5112,
+      "step": 15080
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.361555814743042,
+      "learning_rate": 8.830448174407647e-06,
+      "loss": 0.5132,
+      "step": 15081
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.994290351867676,
+      "learning_rate": 8.830278707073193e-06,
+      "loss": 0.4223,
+      "step": 15082
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.9019999504089355,
+      "learning_rate": 8.830109229088168e-06,
+      "loss": 0.4939,
+      "step": 15083
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.8182296752929688,
+      "learning_rate": 8.829939740453045e-06,
+      "loss": 0.8768,
+      "step": 15084
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.8278839588165283,
+      "learning_rate": 8.829770241168296e-06,
+      "loss": 0.6705,
+      "step": 15085
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.152972936630249,
+      "learning_rate": 8.829600731234388e-06,
+      "loss": 0.7209,
+      "step": 15086
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.654715895652771,
+      "learning_rate": 8.829431210651795e-06,
+      "loss": 0.7809,
+      "step": 15087
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.3446462154388428,
+      "learning_rate": 8.82926167942099e-06,
+      "loss": 0.6452,
+      "step": 15088
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.123652935028076,
+      "learning_rate": 8.829092137542443e-06,
+      "loss": 0.7902,
+      "step": 15089
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.702911615371704,
+      "learning_rate": 8.828922585016623e-06,
+      "loss": 0.5081,
+      "step": 15090
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.3010151386260986,
+      "learning_rate": 8.828753021844005e-06,
+      "loss": 0.7132,
+      "step": 15091
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.832901954650879,
+      "learning_rate": 8.828583448025061e-06,
+      "loss": 0.7492,
+      "step": 15092
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 4.401388168334961,
+      "learning_rate": 8.828413863560258e-06,
+      "loss": 0.6573,
+      "step": 15093
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.829822063446045,
+      "learning_rate": 8.82824426845007e-06,
+      "loss": 0.7077,
+      "step": 15094
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 4.5740203857421875,
+      "learning_rate": 8.82807466269497e-06,
+      "loss": 0.7362,
+      "step": 15095
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.8042850494384766,
+      "learning_rate": 8.827905046295429e-06,
+      "loss": 0.4769,
+      "step": 15096
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.313086986541748,
+      "learning_rate": 8.827735419251918e-06,
+      "loss": 0.398,
+      "step": 15097
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.6409714221954346,
+      "learning_rate": 8.827565781564906e-06,
+      "loss": 0.7186,
+      "step": 15098
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.589881181716919,
+      "learning_rate": 8.82739613323487e-06,
+      "loss": 0.5695,
+      "step": 15099
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.50464129447937,
+      "learning_rate": 8.827226474262279e-06,
+      "loss": 0.743,
+      "step": 15100
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.8726907968521118,
+      "learning_rate": 8.827056804647603e-06,
+      "loss": 0.5972,
+      "step": 15101
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.485840320587158,
+      "learning_rate": 8.826887124391318e-06,
+      "loss": 0.7197,
+      "step": 15102
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.0005993843078613,
+      "learning_rate": 8.826717433493893e-06,
+      "loss": 0.4793,
+      "step": 15103
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.9437551498413086,
+      "learning_rate": 8.8265477319558e-06,
+      "loss": 0.6679,
+      "step": 15104
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.7948129177093506,
+      "learning_rate": 8.82637801977751e-06,
+      "loss": 0.3813,
+      "step": 15105
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.679680824279785,
+      "learning_rate": 8.826208296959498e-06,
+      "loss": 0.8582,
+      "step": 15106
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.4621262550354004,
+      "learning_rate": 8.826038563502233e-06,
+      "loss": 0.6959,
+      "step": 15107
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.977443218231201,
+      "learning_rate": 8.825868819406187e-06,
+      "loss": 0.6653,
+      "step": 15108
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.2612063884735107,
+      "learning_rate": 8.825699064671834e-06,
+      "loss": 0.719,
+      "step": 15109
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.768094778060913,
+      "learning_rate": 8.825529299299646e-06,
+      "loss": 0.6931,
+      "step": 15110
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.104602813720703,
+      "learning_rate": 8.825359523290094e-06,
+      "loss": 0.8091,
+      "step": 15111
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.350202798843384,
+      "learning_rate": 8.82518973664365e-06,
+      "loss": 0.4374,
+      "step": 15112
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.5580275058746338,
+      "learning_rate": 8.825019939360783e-06,
+      "loss": 0.4479,
+      "step": 15113
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.998939275741577,
+      "learning_rate": 8.824850131441974e-06,
+      "loss": 0.6352,
+      "step": 15114
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.6851420402526855,
+      "learning_rate": 8.824680312887686e-06,
+      "loss": 0.8349,
+      "step": 15115
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.3566079139709473,
+      "learning_rate": 8.824510483698396e-06,
+      "loss": 0.8263,
+      "step": 15116
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.454533815383911,
+      "learning_rate": 8.824340643874574e-06,
+      "loss": 0.8807,
+      "step": 15117
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 4.304869651794434,
+      "learning_rate": 8.824170793416692e-06,
+      "loss": 0.7193,
+      "step": 15118
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.3074283599853516,
+      "learning_rate": 8.824000932325227e-06,
+      "loss": 0.6893,
+      "step": 15119
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.0457944869995117,
+      "learning_rate": 8.823831060600647e-06,
+      "loss": 0.7636,
+      "step": 15120
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.035996437072754,
+      "learning_rate": 8.823661178243425e-06,
+      "loss": 0.7542,
+      "step": 15121
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.239319086074829,
+      "learning_rate": 8.823491285254034e-06,
+      "loss": 0.7614,
+      "step": 15122
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.7523307800292969,
+      "learning_rate": 8.823321381632946e-06,
+      "loss": 0.5626,
+      "step": 15123
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.1999249458312988,
+      "learning_rate": 8.82315146738063e-06,
+      "loss": 0.7318,
+      "step": 15124
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.801569700241089,
+      "learning_rate": 8.822981542497567e-06,
+      "loss": 0.7058,
+      "step": 15125
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.504194736480713,
+      "learning_rate": 8.822811606984221e-06,
+      "loss": 0.6658,
+      "step": 15126
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.2671172618865967,
+      "learning_rate": 8.82264166084107e-06,
+      "loss": 0.618,
+      "step": 15127
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.8695721626281738,
+      "learning_rate": 8.822471704068585e-06,
+      "loss": 0.7919,
+      "step": 15128
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 4.227978229522705,
+      "learning_rate": 8.822301736667236e-06,
+      "loss": 0.5637,
+      "step": 15129
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.2848258018493652,
+      "learning_rate": 8.822131758637498e-06,
+      "loss": 0.9203,
+      "step": 15130
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.8878788948059082,
+      "learning_rate": 8.821961769979845e-06,
+      "loss": 0.6414,
+      "step": 15131
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.0191965103149414,
+      "learning_rate": 8.821791770694748e-06,
+      "loss": 0.7419,
+      "step": 15132
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.838928461074829,
+      "learning_rate": 8.821621760782678e-06,
+      "loss": 0.5523,
+      "step": 15133
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.9643876552581787,
+      "learning_rate": 8.82145174024411e-06,
+      "loss": 0.6136,
+      "step": 15134
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.190397024154663,
+      "learning_rate": 8.821281709079516e-06,
+      "loss": 0.5127,
+      "step": 15135
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.021367073059082,
+      "learning_rate": 8.82111166728937e-06,
+      "loss": 0.6352,
+      "step": 15136
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.766768455505371,
+      "learning_rate": 8.820941614874145e-06,
+      "loss": 0.6978,
+      "step": 15137
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.6793172359466553,
+      "learning_rate": 8.82077155183431e-06,
+      "loss": 0.6764,
+      "step": 15138
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.8322434425354004,
+      "learning_rate": 8.820601478170344e-06,
+      "loss": 0.5524,
+      "step": 15139
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.6044881343841553,
+      "learning_rate": 8.820431393882713e-06,
+      "loss": 0.7311,
+      "step": 15140
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.5544620752334595,
+      "learning_rate": 8.820261298971894e-06,
+      "loss": 0.9064,
+      "step": 15141
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.672281265258789,
+      "learning_rate": 8.82009119343836e-06,
+      "loss": 0.6084,
+      "step": 15142
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.2437586784362793,
+      "learning_rate": 8.819921077282586e-06,
+      "loss": 0.7153,
+      "step": 15143
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.694753885269165,
+      "learning_rate": 8.81975095050504e-06,
+      "loss": 0.7508,
+      "step": 15144
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.0570785999298096,
+      "learning_rate": 8.819580813106196e-06,
+      "loss": 0.7072,
+      "step": 15145
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.273130178451538,
+      "learning_rate": 8.819410665086533e-06,
+      "loss": 0.3625,
+      "step": 15146
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.450270414352417,
+      "learning_rate": 8.819240506446516e-06,
+      "loss": 0.569,
+      "step": 15147
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.4650678634643555,
+      "learning_rate": 8.819070337186623e-06,
+      "loss": 0.9484,
+      "step": 15148
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.7115721702575684,
+      "learning_rate": 8.818900157307325e-06,
+      "loss": 0.7677,
+      "step": 15149
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.042426586151123,
+      "learning_rate": 8.818729966809098e-06,
+      "loss": 0.4054,
+      "step": 15150
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.3319156169891357,
+      "learning_rate": 8.818559765692412e-06,
+      "loss": 0.7194,
+      "step": 15151
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.1223416328430176,
+      "learning_rate": 8.818389553957742e-06,
+      "loss": 0.7875,
+      "step": 15152
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.1349592208862305,
+      "learning_rate": 8.818219331605562e-06,
+      "loss": 0.7749,
+      "step": 15153
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.547708749771118,
+      "learning_rate": 8.818049098636345e-06,
+      "loss": 0.7685,
+      "step": 15154
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.454085111618042,
+      "learning_rate": 8.817878855050562e-06,
+      "loss": 0.5317,
+      "step": 15155
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.9414446353912354,
+      "learning_rate": 8.817708600848688e-06,
+      "loss": 0.679,
+      "step": 15156
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.1074163913726807,
+      "learning_rate": 8.817538336031198e-06,
+      "loss": 0.7834,
+      "step": 15157
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.6032723188400269,
+      "learning_rate": 8.817368060598564e-06,
+      "loss": 0.4469,
+      "step": 15158
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.9720981121063232,
+      "learning_rate": 8.817197774551258e-06,
+      "loss": 0.5659,
+      "step": 15159
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.1896636486053467,
+      "learning_rate": 8.817027477889756e-06,
+      "loss": 0.5507,
+      "step": 15160
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.9232282638549805,
+      "learning_rate": 8.81685717061453e-06,
+      "loss": 0.634,
+      "step": 15161
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.89471435546875,
+      "learning_rate": 8.816686852726054e-06,
+      "loss": 0.3867,
+      "step": 15162
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.1708574295043945,
+      "learning_rate": 8.816516524224803e-06,
+      "loss": 0.7119,
+      "step": 15163
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.112438201904297,
+      "learning_rate": 8.816346185111246e-06,
+      "loss": 0.5784,
+      "step": 15164
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.8785300254821777,
+      "learning_rate": 8.816175835385863e-06,
+      "loss": 0.5414,
+      "step": 15165
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 4.4908037185668945,
+      "learning_rate": 8.816005475049123e-06,
+      "loss": 0.4691,
+      "step": 15166
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.0949513912200928,
+      "learning_rate": 8.815835104101501e-06,
+      "loss": 0.4231,
+      "step": 15167
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 4.831174850463867,
+      "learning_rate": 8.815664722543474e-06,
+      "loss": 1.1062,
+      "step": 15168
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.3385796546936035,
+      "learning_rate": 8.815494330375508e-06,
+      "loss": 0.6478,
+      "step": 15169
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.367403984069824,
+      "learning_rate": 8.815323927598086e-06,
+      "loss": 0.681,
+      "step": 15170
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.9099828004837036,
+      "learning_rate": 8.815153514211676e-06,
+      "loss": 0.5871,
+      "step": 15171
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.454869031906128,
+      "learning_rate": 8.814983090216752e-06,
+      "loss": 0.4395,
+      "step": 15172
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.0873329639434814,
+      "learning_rate": 8.814812655613791e-06,
+      "loss": 0.6752,
+      "step": 15173
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.496812343597412,
+      "learning_rate": 8.814642210403264e-06,
+      "loss": 0.8127,
+      "step": 15174
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.196943998336792,
+      "learning_rate": 8.814471754585644e-06,
+      "loss": 0.5813,
+      "step": 15175
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.5802721977233887,
+      "learning_rate": 8.81430128816141e-06,
+      "loss": 0.9021,
+      "step": 15176
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.8419110774993896,
+      "learning_rate": 8.814130811131032e-06,
+      "loss": 0.6585,
+      "step": 15177
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.4068331718444824,
+      "learning_rate": 8.813960323494986e-06,
+      "loss": 0.775,
+      "step": 15178
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.1556246280670166,
+      "learning_rate": 8.813789825253745e-06,
+      "loss": 0.7141,
+      "step": 15179
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.7479521036148071,
+      "learning_rate": 8.81361931640778e-06,
+      "loss": 0.6267,
+      "step": 15180
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.134704113006592,
+      "learning_rate": 8.813448796957571e-06,
+      "loss": 0.3893,
+      "step": 15181
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.0760281085968018,
+      "learning_rate": 8.81327826690359e-06,
+      "loss": 0.822,
+      "step": 15182
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.8151439428329468,
+      "learning_rate": 8.813107726246311e-06,
+      "loss": 0.7062,
+      "step": 15183
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.171292304992676,
+      "learning_rate": 8.812937174986205e-06,
+      "loss": 0.6289,
+      "step": 15184
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.899653911590576,
+      "learning_rate": 8.81276661312375e-06,
+      "loss": 0.787,
+      "step": 15185
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.7453458309173584,
+      "learning_rate": 8.81259604065942e-06,
+      "loss": 1.0239,
+      "step": 15186
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.235079288482666,
+      "learning_rate": 8.812425457593689e-06,
+      "loss": 0.9511,
+      "step": 15187
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 8.131799697875977,
+      "learning_rate": 8.81225486392703e-06,
+      "loss": 0.6407,
+      "step": 15188
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.6591315269470215,
+      "learning_rate": 8.812084259659921e-06,
+      "loss": 0.5171,
+      "step": 15189
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.3527491092681885,
+      "learning_rate": 8.811913644792832e-06,
+      "loss": 0.9166,
+      "step": 15190
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.1518843173980713,
+      "learning_rate": 8.811743019326237e-06,
+      "loss": 0.7835,
+      "step": 15191
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.5560121536254883,
+      "learning_rate": 8.811572383260615e-06,
+      "loss": 0.6405,
+      "step": 15192
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.085639715194702,
+      "learning_rate": 8.811401736596437e-06,
+      "loss": 0.4285,
+      "step": 15193
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.2167445421218872,
+      "learning_rate": 8.81123107933418e-06,
+      "loss": 0.5874,
+      "step": 15194
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.1079769134521484,
+      "learning_rate": 8.811060411474315e-06,
+      "loss": 0.4146,
+      "step": 15195
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.9779186248779297,
+      "learning_rate": 8.81088973301732e-06,
+      "loss": 0.6318,
+      "step": 15196
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 5.037085056304932,
+      "learning_rate": 8.810719043963667e-06,
+      "loss": 0.8126,
+      "step": 15197
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.6744266748428345,
+      "learning_rate": 8.810548344313834e-06,
+      "loss": 0.5715,
+      "step": 15198
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.6346973180770874,
+      "learning_rate": 8.810377634068292e-06,
+      "loss": 0.5554,
+      "step": 15199
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.575023889541626,
+      "learning_rate": 8.810206913227517e-06,
+      "loss": 0.5672,
+      "step": 15200
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.007183313369751,
+      "learning_rate": 8.810036181791984e-06,
+      "loss": 0.4802,
+      "step": 15201
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.890669107437134,
+      "learning_rate": 8.809865439762169e-06,
+      "loss": 0.4385,
+      "step": 15202
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.016403913497925,
+      "learning_rate": 8.809694687138543e-06,
+      "loss": 0.6795,
+      "step": 15203
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.2665722370147705,
+      "learning_rate": 8.809523923921586e-06,
+      "loss": 0.7916,
+      "step": 15204
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.6135334968566895,
+      "learning_rate": 8.809353150111767e-06,
+      "loss": 0.6175,
+      "step": 15205
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.7898110151290894,
+      "learning_rate": 8.809182365709568e-06,
+      "loss": 0.6692,
+      "step": 15206
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.344693422317505,
+      "learning_rate": 8.809011570715457e-06,
+      "loss": 0.4171,
+      "step": 15207
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.6450682878494263,
+      "learning_rate": 8.808840765129912e-06,
+      "loss": 0.6496,
+      "step": 15208
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.9642555713653564,
+      "learning_rate": 8.808669948953407e-06,
+      "loss": 0.8477,
+      "step": 15209
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.1381876468658447,
+      "learning_rate": 8.808499122186418e-06,
+      "loss": 0.9424,
+      "step": 15210
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.059382677078247,
+      "learning_rate": 8.808328284829421e-06,
+      "loss": 0.6726,
+      "step": 15211
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 5.023487091064453,
+      "learning_rate": 8.80815743688289e-06,
+      "loss": 0.7408,
+      "step": 15212
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 6.143955707550049,
+      "learning_rate": 8.807986578347297e-06,
+      "loss": 0.7442,
+      "step": 15213
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.735749363899231,
+      "learning_rate": 8.807815709223123e-06,
+      "loss": 0.8782,
+      "step": 15214
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.2541110515594482,
+      "learning_rate": 8.807644829510838e-06,
+      "loss": 0.6041,
+      "step": 15215
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.3296892642974854,
+      "learning_rate": 8.807473939210918e-06,
+      "loss": 0.7456,
+      "step": 15216
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.823709487915039,
+      "learning_rate": 8.80730303832384e-06,
+      "loss": 0.6091,
+      "step": 15217
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 5.0928168296813965,
+      "learning_rate": 8.80713212685008e-06,
+      "loss": 0.6085,
+      "step": 15218
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.115267276763916,
+      "learning_rate": 8.80696120479011e-06,
+      "loss": 0.6233,
+      "step": 15219
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.277977466583252,
+      "learning_rate": 8.80679027214441e-06,
+      "loss": 0.8254,
+      "step": 15220
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.308701753616333,
+      "learning_rate": 8.806619328913449e-06,
+      "loss": 0.743,
+      "step": 15221
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.850633382797241,
+      "learning_rate": 8.806448375097708e-06,
+      "loss": 0.7295,
+      "step": 15222
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.674997091293335,
+      "learning_rate": 8.806277410697658e-06,
+      "loss": 0.7914,
+      "step": 15223
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.5040266513824463,
+      "learning_rate": 8.806106435713777e-06,
+      "loss": 0.3724,
+      "step": 15224
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.0516207218170166,
+      "learning_rate": 8.80593545014654e-06,
+      "loss": 0.637,
+      "step": 15225
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.3159701824188232,
+      "learning_rate": 8.805764453996423e-06,
+      "loss": 0.8409,
+      "step": 15226
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.218853712081909,
+      "learning_rate": 8.805593447263901e-06,
+      "loss": 0.7033,
+      "step": 15227
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.7136015892028809,
+      "learning_rate": 8.805422429949446e-06,
+      "loss": 0.8012,
+      "step": 15228
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.226449966430664,
+      "learning_rate": 8.80525140205354e-06,
+      "loss": 0.5734,
+      "step": 15229
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.8104915618896484,
+      "learning_rate": 8.805080363576655e-06,
+      "loss": 0.6196,
+      "step": 15230
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.427130103111267,
+      "learning_rate": 8.804909314519265e-06,
+      "loss": 0.2459,
+      "step": 15231
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.6537649631500244,
+      "learning_rate": 8.80473825488185e-06,
+      "loss": 0.7406,
+      "step": 15232
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.0938124656677246,
+      "learning_rate": 8.804567184664882e-06,
+      "loss": 0.6214,
+      "step": 15233
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.564835548400879,
+      "learning_rate": 8.804396103868837e-06,
+      "loss": 0.8428,
+      "step": 15234
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.7950356006622314,
+      "learning_rate": 8.804225012494191e-06,
+      "loss": 0.7076,
+      "step": 15235
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.499965190887451,
+      "learning_rate": 8.804053910541423e-06,
+      "loss": 0.5687,
+      "step": 15236
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.9952714443206787,
+      "learning_rate": 8.803882798011005e-06,
+      "loss": 0.5373,
+      "step": 15237
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 4.42133903503418,
+      "learning_rate": 8.803711674903415e-06,
+      "loss": 0.7649,
+      "step": 15238
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.3278791904449463,
+      "learning_rate": 8.803540541219126e-06,
+      "loss": 0.7288,
+      "step": 15239
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.845406413078308,
+      "learning_rate": 8.803369396958616e-06,
+      "loss": 0.6218,
+      "step": 15240
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.1720449924468994,
+      "learning_rate": 8.80319824212236e-06,
+      "loss": 0.8588,
+      "step": 15241
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 4.91230583190918,
+      "learning_rate": 8.803027076710835e-06,
+      "loss": 0.7962,
+      "step": 15242
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.779510498046875,
+      "learning_rate": 8.802855900724518e-06,
+      "loss": 0.5203,
+      "step": 15243
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.3223414421081543,
+      "learning_rate": 8.80268471416388e-06,
+      "loss": 0.7732,
+      "step": 15244
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.5228604078292847,
+      "learning_rate": 8.802513517029404e-06,
+      "loss": 0.393,
+      "step": 15245
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.7811940908432007,
+      "learning_rate": 8.802342309321559e-06,
+      "loss": 0.6651,
+      "step": 15246
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 4.994194984436035,
+      "learning_rate": 8.802171091040827e-06,
+      "loss": 0.7644,
+      "step": 15247
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.4516232013702393,
+      "learning_rate": 8.80199986218768e-06,
+      "loss": 0.7854,
+      "step": 15248
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.0103096961975098,
+      "learning_rate": 8.801828622762596e-06,
+      "loss": 0.9821,
+      "step": 15249
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.1631157398223877,
+      "learning_rate": 8.80165737276605e-06,
+      "loss": 0.4787,
+      "step": 15250
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.7182061672210693,
+      "learning_rate": 8.80148611219852e-06,
+      "loss": 0.6414,
+      "step": 15251
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.557051181793213,
+      "learning_rate": 8.80131484106048e-06,
+      "loss": 0.6646,
+      "step": 15252
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.608220338821411,
+      "learning_rate": 8.80114355935241e-06,
+      "loss": 0.6699,
+      "step": 15253
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.6801514625549316,
+      "learning_rate": 8.800972267074782e-06,
+      "loss": 0.5465,
+      "step": 15254
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.508357048034668,
+      "learning_rate": 8.800800964228074e-06,
+      "loss": 0.5232,
+      "step": 15255
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 4.3600897789001465,
+      "learning_rate": 8.800629650812762e-06,
+      "loss": 0.5425,
+      "step": 15256
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.2844882011413574,
+      "learning_rate": 8.800458326829325e-06,
+      "loss": 0.7935,
+      "step": 15257
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.8883824348449707,
+      "learning_rate": 8.800286992278233e-06,
+      "loss": 0.7378,
+      "step": 15258
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.869217872619629,
+      "learning_rate": 8.80011564715997e-06,
+      "loss": 0.6445,
+      "step": 15259
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.227918863296509,
+      "learning_rate": 8.799944291475007e-06,
+      "loss": 0.7101,
+      "step": 15260
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.5159311294555664,
+      "learning_rate": 8.799772925223822e-06,
+      "loss": 0.6874,
+      "step": 15261
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.7059036493301392,
+      "learning_rate": 8.799601548406895e-06,
+      "loss": 0.8231,
+      "step": 15262
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.9943349361419678,
+      "learning_rate": 8.799430161024697e-06,
+      "loss": 0.5681,
+      "step": 15263
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.3947582244873047,
+      "learning_rate": 8.799258763077707e-06,
+      "loss": 0.5944,
+      "step": 15264
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.7218008041381836,
+      "learning_rate": 8.799087354566402e-06,
+      "loss": 0.5557,
+      "step": 15265
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.711632251739502,
+      "learning_rate": 8.798915935491258e-06,
+      "loss": 0.4975,
+      "step": 15266
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.4076144695281982,
+      "learning_rate": 8.798744505852751e-06,
+      "loss": 0.6448,
+      "step": 15267
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.7633490562438965,
+      "learning_rate": 8.79857306565136e-06,
+      "loss": 0.8072,
+      "step": 15268
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.261082172393799,
+      "learning_rate": 8.798401614887562e-06,
+      "loss": 0.6678,
+      "step": 15269
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.797690987586975,
+      "learning_rate": 8.79823015356183e-06,
+      "loss": 0.4662,
+      "step": 15270
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.1376090049743652,
+      "learning_rate": 8.798058681674641e-06,
+      "loss": 0.8472,
+      "step": 15271
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.407203435897827,
+      "learning_rate": 8.797887199226476e-06,
+      "loss": 0.9302,
+      "step": 15272
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.8967790603637695,
+      "learning_rate": 8.797715706217809e-06,
+      "loss": 0.8371,
+      "step": 15273
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.386385440826416,
+      "learning_rate": 8.797544202649118e-06,
+      "loss": 0.5092,
+      "step": 15274
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.0077075958251953,
+      "learning_rate": 8.797372688520877e-06,
+      "loss": 0.5169,
+      "step": 15275
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.1913816928863525,
+      "learning_rate": 8.797201163833567e-06,
+      "loss": 0.6677,
+      "step": 15276
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.3584649562835693,
+      "learning_rate": 8.797029628587663e-06,
+      "loss": 0.7622,
+      "step": 15277
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.354034900665283,
+      "learning_rate": 8.79685808278364e-06,
+      "loss": 0.571,
+      "step": 15278
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.965027332305908,
+      "learning_rate": 8.79668652642198e-06,
+      "loss": 0.8964,
+      "step": 15279
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.542548418045044,
+      "learning_rate": 8.796514959503155e-06,
+      "loss": 0.6448,
+      "step": 15280
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.216426134109497,
+      "learning_rate": 8.796343382027646e-06,
+      "loss": 0.694,
+      "step": 15281
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.3239333629608154,
+      "learning_rate": 8.796171793995926e-06,
+      "loss": 0.4068,
+      "step": 15282
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.0821404457092285,
+      "learning_rate": 8.796000195408475e-06,
+      "loss": 0.6989,
+      "step": 15283
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.2669272422790527,
+      "learning_rate": 8.79582858626577e-06,
+      "loss": 0.8069,
+      "step": 15284
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.4810492992401123,
+      "learning_rate": 8.795656966568287e-06,
+      "loss": 0.3178,
+      "step": 15285
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.8359874486923218,
+      "learning_rate": 8.795485336316503e-06,
+      "loss": 0.6898,
+      "step": 15286
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.3896827697753906,
+      "learning_rate": 8.795313695510899e-06,
+      "loss": 0.3924,
+      "step": 15287
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.76662540435791,
+      "learning_rate": 8.795142044151948e-06,
+      "loss": 0.6547,
+      "step": 15288
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.954662799835205,
+      "learning_rate": 8.794970382240127e-06,
+      "loss": 0.4334,
+      "step": 15289
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.5204570293426514,
+      "learning_rate": 8.794798709775916e-06,
+      "loss": 0.7731,
+      "step": 15290
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.6554665565490723,
+      "learning_rate": 8.794627026759793e-06,
+      "loss": 0.651,
+      "step": 15291
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.675212860107422,
+      "learning_rate": 8.794455333192231e-06,
+      "loss": 0.5805,
+      "step": 15292
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.1767947673797607,
+      "learning_rate": 8.794283629073713e-06,
+      "loss": 0.7092,
+      "step": 15293
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.6646132469177246,
+      "learning_rate": 8.79411191440471e-06,
+      "loss": 0.6731,
+      "step": 15294
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.798643708229065,
+      "learning_rate": 8.793940189185705e-06,
+      "loss": 0.5834,
+      "step": 15295
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.0226690769195557,
+      "learning_rate": 8.793768453417175e-06,
+      "loss": 0.8773,
+      "step": 15296
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.4264161586761475,
+      "learning_rate": 8.793596707099595e-06,
+      "loss": 0.7661,
+      "step": 15297
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.1470820903778076,
+      "learning_rate": 8.793424950233443e-06,
+      "loss": 0.7937,
+      "step": 15298
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.000488758087158,
+      "learning_rate": 8.793253182819198e-06,
+      "loss": 0.3619,
+      "step": 15299
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.886918544769287,
+      "learning_rate": 8.793081404857336e-06,
+      "loss": 0.9079,
+      "step": 15300
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.516167402267456,
+      "learning_rate": 8.792909616348337e-06,
+      "loss": 0.5051,
+      "step": 15301
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.9268836975097656,
+      "learning_rate": 8.792737817292676e-06,
+      "loss": 0.7981,
+      "step": 15302
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 5.67045259475708,
+      "learning_rate": 8.792566007690832e-06,
+      "loss": 0.7417,
+      "step": 15303
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.7163407802581787,
+      "learning_rate": 8.792394187543284e-06,
+      "loss": 0.4638,
+      "step": 15304
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.400192975997925,
+      "learning_rate": 8.792222356850505e-06,
+      "loss": 0.6279,
+      "step": 15305
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.5189623832702637,
+      "learning_rate": 8.79205051561298e-06,
+      "loss": 0.7518,
+      "step": 15306
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.925901412963867,
+      "learning_rate": 8.791878663831182e-06,
+      "loss": 0.8052,
+      "step": 15307
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.2321202754974365,
+      "learning_rate": 8.79170680150559e-06,
+      "loss": 0.7986,
+      "step": 15308
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.310384511947632,
+      "learning_rate": 8.791534928636681e-06,
+      "loss": 0.8397,
+      "step": 15309
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.0265376567840576,
+      "learning_rate": 8.791363045224935e-06,
+      "loss": 0.3975,
+      "step": 15310
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.8116745948791504,
+      "learning_rate": 8.791191151270826e-06,
+      "loss": 0.4183,
+      "step": 15311
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.1216647624969482,
+      "learning_rate": 8.791019246774836e-06,
+      "loss": 0.4509,
+      "step": 15312
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.0917282104492188,
+      "learning_rate": 8.790847331737443e-06,
+      "loss": 0.6439,
+      "step": 15313
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.345553398132324,
+      "learning_rate": 8.790675406159122e-06,
+      "loss": 0.8202,
+      "step": 15314
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.901768207550049,
+      "learning_rate": 8.790503470040354e-06,
+      "loss": 0.8664,
+      "step": 15315
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 4.6550068855285645,
+      "learning_rate": 8.790331523381615e-06,
+      "loss": 0.6177,
+      "step": 15316
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.166015863418579,
+      "learning_rate": 8.790159566183385e-06,
+      "loss": 0.6484,
+      "step": 15317
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 5.738861083984375,
+      "learning_rate": 8.78998759844614e-06,
+      "loss": 0.6172,
+      "step": 15318
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.3226451873779297,
+      "learning_rate": 8.789815620170359e-06,
+      "loss": 0.4471,
+      "step": 15319
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 4.023512363433838,
+      "learning_rate": 8.789643631356521e-06,
+      "loss": 0.7363,
+      "step": 15320
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.6049160957336426,
+      "learning_rate": 8.789471632005104e-06,
+      "loss": 0.797,
+      "step": 15321
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.073336362838745,
+      "learning_rate": 8.789299622116585e-06,
+      "loss": 0.5207,
+      "step": 15322
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.0187244415283203,
+      "learning_rate": 8.789127601691444e-06,
+      "loss": 0.5982,
+      "step": 15323
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.0836009979248047,
+      "learning_rate": 8.788955570730158e-06,
+      "loss": 0.9467,
+      "step": 15324
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.922530174255371,
+      "learning_rate": 8.788783529233207e-06,
+      "loss": 0.6715,
+      "step": 15325
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.5306084156036377,
+      "learning_rate": 8.788611477201067e-06,
+      "loss": 0.7409,
+      "step": 15326
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.1818509101867676,
+      "learning_rate": 8.788439414634217e-06,
+      "loss": 0.8989,
+      "step": 15327
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.4819278717041016,
+      "learning_rate": 8.788267341533139e-06,
+      "loss": 0.3301,
+      "step": 15328
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.2642018795013428,
+      "learning_rate": 8.788095257898306e-06,
+      "loss": 0.7426,
+      "step": 15329
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.135075092315674,
+      "learning_rate": 8.7879231637302e-06,
+      "loss": 0.5859,
+      "step": 15330
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.389571189880371,
+      "learning_rate": 8.787751059029301e-06,
+      "loss": 0.6734,
+      "step": 15331
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.3572959899902344,
+      "learning_rate": 8.787578943796082e-06,
+      "loss": 0.6367,
+      "step": 15332
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.9736437797546387,
+      "learning_rate": 8.787406818031025e-06,
+      "loss": 0.5886,
+      "step": 15333
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.8040101528167725,
+      "learning_rate": 8.78723468173461e-06,
+      "loss": 0.7469,
+      "step": 15334
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.7532455921173096,
+      "learning_rate": 8.787062534907314e-06,
+      "loss": 0.7236,
+      "step": 15335
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.547738552093506,
+      "learning_rate": 8.786890377549615e-06,
+      "loss": 0.6834,
+      "step": 15336
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.9651200771331787,
+      "learning_rate": 8.786718209661992e-06,
+      "loss": 0.555,
+      "step": 15337
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.2408995628356934,
+      "learning_rate": 8.786546031244925e-06,
+      "loss": 0.8042,
+      "step": 15338
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.6208419799804688,
+      "learning_rate": 8.786373842298893e-06,
+      "loss": 0.5369,
+      "step": 15339
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.8167953491210938,
+      "learning_rate": 8.786201642824373e-06,
+      "loss": 0.6847,
+      "step": 15340
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.644601821899414,
+      "learning_rate": 8.786029432821843e-06,
+      "loss": 0.4123,
+      "step": 15341
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.092092514038086,
+      "learning_rate": 8.785857212291785e-06,
+      "loss": 0.862,
+      "step": 15342
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.9424052238464355,
+      "learning_rate": 8.785684981234676e-06,
+      "loss": 0.7197,
+      "step": 15343
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.116212844848633,
+      "learning_rate": 8.785512739650996e-06,
+      "loss": 0.6468,
+      "step": 15344
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.8700571060180664,
+      "learning_rate": 8.785340487541222e-06,
+      "loss": 0.5805,
+      "step": 15345
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.998879313468933,
+      "learning_rate": 8.785168224905834e-06,
+      "loss": 0.576,
+      "step": 15346
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.833160400390625,
+      "learning_rate": 8.784995951745313e-06,
+      "loss": 0.6129,
+      "step": 15347
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.4726977348327637,
+      "learning_rate": 8.784823668060134e-06,
+      "loss": 0.6919,
+      "step": 15348
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.6170730590820312,
+      "learning_rate": 8.784651373850778e-06,
+      "loss": 0.7836,
+      "step": 15349
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.8158586025238037,
+      "learning_rate": 8.784479069117725e-06,
+      "loss": 0.493,
+      "step": 15350
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.8994807004928589,
+      "learning_rate": 8.784306753861454e-06,
+      "loss": 0.7465,
+      "step": 15351
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.5223681926727295,
+      "learning_rate": 8.784134428082443e-06,
+      "loss": 0.727,
+      "step": 15352
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.8325181007385254,
+      "learning_rate": 8.783962091781173e-06,
+      "loss": 0.7953,
+      "step": 15353
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.260100841522217,
+      "learning_rate": 8.78378974495812e-06,
+      "loss": 0.7445,
+      "step": 15354
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.1253933906555176,
+      "learning_rate": 8.783617387613766e-06,
+      "loss": 0.8086,
+      "step": 15355
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.124077558517456,
+      "learning_rate": 8.78344501974859e-06,
+      "loss": 0.3187,
+      "step": 15356
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.5342164039611816,
+      "learning_rate": 8.78327264136307e-06,
+      "loss": 0.7374,
+      "step": 15357
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.004565954208374,
+      "learning_rate": 8.783100252457686e-06,
+      "loss": 0.7097,
+      "step": 15358
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.6852312088012695,
+      "learning_rate": 8.782927853032918e-06,
+      "loss": 0.634,
+      "step": 15359
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.1269595623016357,
+      "learning_rate": 8.782755443089244e-06,
+      "loss": 0.8463,
+      "step": 15360
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.9126695394515991,
+      "learning_rate": 8.782583022627145e-06,
+      "loss": 0.53,
+      "step": 15361
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.8411744832992554,
+      "learning_rate": 8.782410591647098e-06,
+      "loss": 0.6831,
+      "step": 15362
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.0961997509002686,
+      "learning_rate": 8.782238150149584e-06,
+      "loss": 0.4683,
+      "step": 15363
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.9028990268707275,
+      "learning_rate": 8.782065698135085e-06,
+      "loss": 0.6407,
+      "step": 15364
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.8627054691314697,
+      "learning_rate": 8.781893235604078e-06,
+      "loss": 0.4668,
+      "step": 15365
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.0745646953582764,
+      "learning_rate": 8.78172076255704e-06,
+      "loss": 0.4995,
+      "step": 15366
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.2975411415100098,
+      "learning_rate": 8.781548278994455e-06,
+      "loss": 0.7141,
+      "step": 15367
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.7820358276367188,
+      "learning_rate": 8.7813757849168e-06,
+      "loss": 0.8078,
+      "step": 15368
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.697274923324585,
+      "learning_rate": 8.781203280324556e-06,
+      "loss": 0.7084,
+      "step": 15369
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.22747540473938,
+      "learning_rate": 8.781030765218202e-06,
+      "loss": 0.5567,
+      "step": 15370
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.3438363075256348,
+      "learning_rate": 8.780858239598219e-06,
+      "loss": 0.7457,
+      "step": 15371
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.418449878692627,
+      "learning_rate": 8.780685703465084e-06,
+      "loss": 0.9053,
+      "step": 15372
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.0428762435913086,
+      "learning_rate": 8.780513156819279e-06,
+      "loss": 0.6831,
+      "step": 15373
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 5.0718607902526855,
+      "learning_rate": 8.780340599661284e-06,
+      "loss": 0.6863,
+      "step": 15374
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.7991318702697754,
+      "learning_rate": 8.780168031991576e-06,
+      "loss": 0.7628,
+      "step": 15375
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.210050582885742,
+      "learning_rate": 8.779995453810636e-06,
+      "loss": 0.7788,
+      "step": 15376
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.174600601196289,
+      "learning_rate": 8.779822865118948e-06,
+      "loss": 0.5592,
+      "step": 15377
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.919928789138794,
+      "learning_rate": 8.779650265916986e-06,
+      "loss": 0.8719,
+      "step": 15378
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.0721843242645264,
+      "learning_rate": 8.779477656205233e-06,
+      "loss": 0.5146,
+      "step": 15379
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.642812490463257,
+      "learning_rate": 8.77930503598417e-06,
+      "loss": 0.5183,
+      "step": 15380
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.7707730531692505,
+      "learning_rate": 8.779132405254275e-06,
+      "loss": 0.6801,
+      "step": 15381
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.0620994567871094,
+      "learning_rate": 8.778959764016027e-06,
+      "loss": 0.6139,
+      "step": 15382
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.273651361465454,
+      "learning_rate": 8.778787112269908e-06,
+      "loss": 0.644,
+      "step": 15383
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.403229236602783,
+      "learning_rate": 8.778614450016398e-06,
+      "loss": 0.6184,
+      "step": 15384
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.5963642597198486,
+      "learning_rate": 8.778441777255977e-06,
+      "loss": 0.7002,
+      "step": 15385
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.8896515369415283,
+      "learning_rate": 8.778269093989124e-06,
+      "loss": 0.7813,
+      "step": 15386
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.9095168113708496,
+      "learning_rate": 8.778096400216322e-06,
+      "loss": 0.9936,
+      "step": 15387
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.0980801582336426,
+      "learning_rate": 8.777923695938047e-06,
+      "loss": 0.7468,
+      "step": 15388
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.1195273399353027,
+      "learning_rate": 8.777750981154783e-06,
+      "loss": 0.7541,
+      "step": 15389
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.0437023639678955,
+      "learning_rate": 8.777578255867007e-06,
+      "loss": 0.6783,
+      "step": 15390
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 10.056321144104004,
+      "learning_rate": 8.777405520075201e-06,
+      "loss": 0.7576,
+      "step": 15391
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.506836175918579,
+      "learning_rate": 8.777232773779846e-06,
+      "loss": 0.7379,
+      "step": 15392
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.8335665464401245,
+      "learning_rate": 8.777060016981423e-06,
+      "loss": 0.5231,
+      "step": 15393
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.1632540225982666,
+      "learning_rate": 8.77688724968041e-06,
+      "loss": 0.5417,
+      "step": 15394
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.1345179080963135,
+      "learning_rate": 8.776714471877288e-06,
+      "loss": 0.498,
+      "step": 15395
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.4778921604156494,
+      "learning_rate": 8.776541683572538e-06,
+      "loss": 0.6841,
+      "step": 15396
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.8976821899414062,
+      "learning_rate": 8.77636888476664e-06,
+      "loss": 0.6827,
+      "step": 15397
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.532658338546753,
+      "learning_rate": 8.776196075460075e-06,
+      "loss": 0.4807,
+      "step": 15398
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.437111258506775,
+      "learning_rate": 8.776023255653324e-06,
+      "loss": 0.2781,
+      "step": 15399
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.2983829975128174,
+      "learning_rate": 8.775850425346866e-06,
+      "loss": 0.7015,
+      "step": 15400
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.8575916290283203,
+      "learning_rate": 8.775677584541184e-06,
+      "loss": 0.5232,
+      "step": 15401
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.1565868854522705,
+      "learning_rate": 8.775504733236754e-06,
+      "loss": 0.8562,
+      "step": 15402
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.0804805755615234,
+      "learning_rate": 8.775331871434063e-06,
+      "loss": 0.6664,
+      "step": 15403
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.9448837041854858,
+      "learning_rate": 8.775158999133587e-06,
+      "loss": 0.5226,
+      "step": 15404
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.0085062980651855,
+      "learning_rate": 8.774986116335808e-06,
+      "loss": 0.672,
+      "step": 15405
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.0258374214172363,
+      "learning_rate": 8.774813223041207e-06,
+      "loss": 0.4332,
+      "step": 15406
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.8462457656860352,
+      "learning_rate": 8.774640319250264e-06,
+      "loss": 0.7844,
+      "step": 15407
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.4649605751037598,
+      "learning_rate": 8.77446740496346e-06,
+      "loss": 0.7659,
+      "step": 15408
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.278062343597412,
+      "learning_rate": 8.774294480181276e-06,
+      "loss": 0.5897,
+      "step": 15409
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.9867944717407227,
+      "learning_rate": 8.774121544904193e-06,
+      "loss": 0.5243,
+      "step": 15410
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.0965983867645264,
+      "learning_rate": 8.773948599132693e-06,
+      "loss": 0.5438,
+      "step": 15411
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.2155227661132812,
+      "learning_rate": 8.773775642867255e-06,
+      "loss": 0.6344,
+      "step": 15412
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.895857810974121,
+      "learning_rate": 8.773602676108362e-06,
+      "loss": 0.5743,
+      "step": 15413
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.536768674850464,
+      "learning_rate": 8.77342969885649e-06,
+      "loss": 0.6728,
+      "step": 15414
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.903625965118408,
+      "learning_rate": 8.773256711112127e-06,
+      "loss": 0.6664,
+      "step": 15415
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.8568921089172363,
+      "learning_rate": 8.77308371287575e-06,
+      "loss": 0.4101,
+      "step": 15416
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.382286548614502,
+      "learning_rate": 8.77291070414784e-06,
+      "loss": 0.7937,
+      "step": 15417
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.1768710613250732,
+      "learning_rate": 8.772737684928877e-06,
+      "loss": 0.6026,
+      "step": 15418
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.0131137371063232,
+      "learning_rate": 8.772564655219346e-06,
+      "loss": 0.7123,
+      "step": 15419
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.4275530576705933,
+      "learning_rate": 8.772391615019725e-06,
+      "loss": 0.6992,
+      "step": 15420
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.3718326091766357,
+      "learning_rate": 8.772218564330497e-06,
+      "loss": 0.5346,
+      "step": 15421
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.9736214876174927,
+      "learning_rate": 8.772045503152141e-06,
+      "loss": 0.6169,
+      "step": 15422
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.483171820640564,
+      "learning_rate": 8.77187243148514e-06,
+      "loss": 0.5756,
+      "step": 15423
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.2649600505828857,
+      "learning_rate": 8.771699349329976e-06,
+      "loss": 0.6503,
+      "step": 15424
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.142192840576172,
+      "learning_rate": 8.771526256687128e-06,
+      "loss": 0.6168,
+      "step": 15425
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.5396103858947754,
+      "learning_rate": 8.771353153557079e-06,
+      "loss": 0.3929,
+      "step": 15426
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.853708028793335,
+      "learning_rate": 8.771180039940307e-06,
+      "loss": 0.6923,
+      "step": 15427
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.3637330532073975,
+      "learning_rate": 8.771006915837299e-06,
+      "loss": 0.5638,
+      "step": 15428
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.548161029815674,
+      "learning_rate": 8.770833781248533e-06,
+      "loss": 0.8534,
+      "step": 15429
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.8363564014434814,
+      "learning_rate": 8.77066063617449e-06,
+      "loss": 0.6511,
+      "step": 15430
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.1413521766662598,
+      "learning_rate": 8.770487480615652e-06,
+      "loss": 0.8806,
+      "step": 15431
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.3492443561553955,
+      "learning_rate": 8.770314314572502e-06,
+      "loss": 0.7582,
+      "step": 15432
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.9891738891601562,
+      "learning_rate": 8.77014113804552e-06,
+      "loss": 0.5947,
+      "step": 15433
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.948405146598816,
+      "learning_rate": 8.769967951035187e-06,
+      "loss": 0.6137,
+      "step": 15434
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 4.6944355964660645,
+      "learning_rate": 8.769794753541984e-06,
+      "loss": 0.7178,
+      "step": 15435
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.1038453578948975,
+      "learning_rate": 8.769621545566397e-06,
+      "loss": 0.6759,
+      "step": 15436
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.884098529815674,
+      "learning_rate": 8.769448327108904e-06,
+      "loss": 0.4832,
+      "step": 15437
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.939294695854187,
+      "learning_rate": 8.769275098169988e-06,
+      "loss": 0.5918,
+      "step": 15438
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.9808694124221802,
+      "learning_rate": 8.769101858750127e-06,
+      "loss": 0.759,
+      "step": 15439
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.8322439193725586,
+      "learning_rate": 8.768928608849807e-06,
+      "loss": 0.6383,
+      "step": 15440
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.7811195850372314,
+      "learning_rate": 8.768755348469509e-06,
+      "loss": 0.5696,
+      "step": 15441
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.6618897914886475,
+      "learning_rate": 8.768582077609714e-06,
+      "loss": 0.8577,
+      "step": 15442
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 4.439010143280029,
+      "learning_rate": 8.768408796270904e-06,
+      "loss": 0.4748,
+      "step": 15443
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.3967232704162598,
+      "learning_rate": 8.768235504453561e-06,
+      "loss": 0.5842,
+      "step": 15444
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.6806881427764893,
+      "learning_rate": 8.768062202158165e-06,
+      "loss": 0.5081,
+      "step": 15445
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.384373188018799,
+      "learning_rate": 8.767888889385203e-06,
+      "loss": 0.8143,
+      "step": 15446
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.2018872499465942,
+      "learning_rate": 8.767715566135152e-06,
+      "loss": 0.6583,
+      "step": 15447
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.8862180709838867,
+      "learning_rate": 8.767542232408494e-06,
+      "loss": 0.5573,
+      "step": 15448
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.8305423259735107,
+      "learning_rate": 8.767368888205712e-06,
+      "loss": 0.939,
+      "step": 15449
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 4.527393817901611,
+      "learning_rate": 8.767195533527291e-06,
+      "loss": 0.7963,
+      "step": 15450
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.980792999267578,
+      "learning_rate": 8.767022168373708e-06,
+      "loss": 0.6984,
+      "step": 15451
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.013094186782837,
+      "learning_rate": 8.766848792745449e-06,
+      "loss": 0.686,
+      "step": 15452
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.248836040496826,
+      "learning_rate": 8.766675406642993e-06,
+      "loss": 0.7078,
+      "step": 15453
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.5737950801849365,
+      "learning_rate": 8.766502010066826e-06,
+      "loss": 0.6343,
+      "step": 15454
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.9720666408538818,
+      "learning_rate": 8.766328603017427e-06,
+      "loss": 0.5496,
+      "step": 15455
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.230471134185791,
+      "learning_rate": 8.766155185495278e-06,
+      "loss": 0.7296,
+      "step": 15456
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.3446853160858154,
+      "learning_rate": 8.765981757500864e-06,
+      "loss": 0.8189,
+      "step": 15457
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.644892454147339,
+      "learning_rate": 8.765808319034665e-06,
+      "loss": 0.7165,
+      "step": 15458
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.6207175254821777,
+      "learning_rate": 8.765634870097162e-06,
+      "loss": 0.5692,
+      "step": 15459
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.256122350692749,
+      "learning_rate": 8.765461410688842e-06,
+      "loss": 0.4869,
+      "step": 15460
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.3652422428131104,
+      "learning_rate": 8.765287940810182e-06,
+      "loss": 0.6769,
+      "step": 15461
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 7.716115951538086,
+      "learning_rate": 8.765114460461667e-06,
+      "loss": 0.6567,
+      "step": 15462
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.5517756938934326,
+      "learning_rate": 8.764940969643781e-06,
+      "loss": 0.6186,
+      "step": 15463
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.0158209800720215,
+      "learning_rate": 8.764767468357003e-06,
+      "loss": 0.7777,
+      "step": 15464
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.6484705209732056,
+      "learning_rate": 8.764593956601817e-06,
+      "loss": 0.6637,
+      "step": 15465
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.6887753009796143,
+      "learning_rate": 8.764420434378707e-06,
+      "loss": 0.6118,
+      "step": 15466
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 7.238039493560791,
+      "learning_rate": 8.764246901688152e-06,
+      "loss": 0.7573,
+      "step": 15467
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.544409155845642,
+      "learning_rate": 8.764073358530636e-06,
+      "loss": 0.4941,
+      "step": 15468
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.9656175374984741,
+      "learning_rate": 8.763899804906644e-06,
+      "loss": 0.4643,
+      "step": 15469
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.2080726623535156,
+      "learning_rate": 8.763726240816656e-06,
+      "loss": 0.4744,
+      "step": 15470
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.5309736728668213,
+      "learning_rate": 8.763552666261155e-06,
+      "loss": 0.4699,
+      "step": 15471
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.5242743492126465,
+      "learning_rate": 8.763379081240623e-06,
+      "loss": 0.7739,
+      "step": 15472
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.751532793045044,
+      "learning_rate": 8.763205485755546e-06,
+      "loss": 0.6431,
+      "step": 15473
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.8264925479888916,
+      "learning_rate": 8.763031879806402e-06,
+      "loss": 0.5443,
+      "step": 15474
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.3805418014526367,
+      "learning_rate": 8.762858263393676e-06,
+      "loss": 0.7787,
+      "step": 15475
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.351478338241577,
+      "learning_rate": 8.762684636517853e-06,
+      "loss": 0.5374,
+      "step": 15476
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.5794029235839844,
+      "learning_rate": 8.762510999179413e-06,
+      "loss": 0.6114,
+      "step": 15477
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.5339603424072266,
+      "learning_rate": 8.762337351378838e-06,
+      "loss": 0.5898,
+      "step": 15478
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.7181291580200195,
+      "learning_rate": 8.762163693116613e-06,
+      "loss": 0.3313,
+      "step": 15479
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.46600604057312,
+      "learning_rate": 8.761990024393218e-06,
+      "loss": 0.7052,
+      "step": 15480
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.1199259757995605,
+      "learning_rate": 8.761816345209142e-06,
+      "loss": 0.7999,
+      "step": 15481
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.5659332275390625,
+      "learning_rate": 8.761642655564862e-06,
+      "loss": 0.5237,
+      "step": 15482
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.8361544609069824,
+      "learning_rate": 8.76146895546086e-06,
+      "loss": 0.5369,
+      "step": 15483
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.2519826889038086,
+      "learning_rate": 8.761295244897627e-06,
+      "loss": 0.5879,
+      "step": 15484
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.9500393867492676,
+      "learning_rate": 8.761121523875638e-06,
+      "loss": 0.6765,
+      "step": 15485
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 4.023960590362549,
+      "learning_rate": 8.76094779239538e-06,
+      "loss": 0.9703,
+      "step": 15486
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.378276824951172,
+      "learning_rate": 8.760774050457334e-06,
+      "loss": 0.542,
+      "step": 15487
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.2590713500976562,
+      "learning_rate": 8.760600298061985e-06,
+      "loss": 0.5405,
+      "step": 15488
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.72572660446167,
+      "learning_rate": 8.760426535209815e-06,
+      "loss": 0.4719,
+      "step": 15489
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.0912413597106934,
+      "learning_rate": 8.760252761901308e-06,
+      "loss": 0.78,
+      "step": 15490
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.8462717533111572,
+      "learning_rate": 8.760078978136945e-06,
+      "loss": 0.7786,
+      "step": 15491
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.3310329914093018,
+      "learning_rate": 8.759905183917213e-06,
+      "loss": 0.7681,
+      "step": 15492
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.2347464561462402,
+      "learning_rate": 8.759731379242593e-06,
+      "loss": 0.6795,
+      "step": 15493
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.962085485458374,
+      "learning_rate": 8.759557564113567e-06,
+      "loss": 0.7326,
+      "step": 15494
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 4.531918525695801,
+      "learning_rate": 8.759383738530622e-06,
+      "loss": 0.7753,
+      "step": 15495
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.023301124572754,
+      "learning_rate": 8.759209902494237e-06,
+      "loss": 0.7831,
+      "step": 15496
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.1707992553710938,
+      "learning_rate": 8.759036056004899e-06,
+      "loss": 0.5755,
+      "step": 15497
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.7339069843292236,
+      "learning_rate": 8.75886219906309e-06,
+      "loss": 0.5052,
+      "step": 15498
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.936795949935913,
+      "learning_rate": 8.758688331669291e-06,
+      "loss": 0.6215,
+      "step": 15499
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.52059006690979,
+      "learning_rate": 8.75851445382399e-06,
+      "loss": 0.5588,
+      "step": 15500
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.531548023223877,
+      "learning_rate": 8.758340565527668e-06,
+      "loss": 0.7826,
+      "step": 15501
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.1324262619018555,
+      "learning_rate": 8.758166666780808e-06,
+      "loss": 0.4023,
+      "step": 15502
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.657060146331787,
+      "learning_rate": 8.757992757583896e-06,
+      "loss": 0.8069,
+      "step": 15503
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.2167763710021973,
+      "learning_rate": 8.757818837937411e-06,
+      "loss": 0.8641,
+      "step": 15504
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.8806718587875366,
+      "learning_rate": 8.757644907841842e-06,
+      "loss": 0.9999,
+      "step": 15505
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.2655959129333496,
+      "learning_rate": 8.757470967297668e-06,
+      "loss": 0.6747,
+      "step": 15506
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.64424467086792,
+      "learning_rate": 8.757297016305378e-06,
+      "loss": 0.6445,
+      "step": 15507
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 4.267336845397949,
+      "learning_rate": 8.75712305486545e-06,
+      "loss": 0.6593,
+      "step": 15508
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.7648786306381226,
+      "learning_rate": 8.756949082978371e-06,
+      "loss": 0.4635,
+      "step": 15509
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.3853952884674072,
+      "learning_rate": 8.756775100644623e-06,
+      "loss": 0.4752,
+      "step": 15510
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.68769907951355,
+      "learning_rate": 8.756601107864694e-06,
+      "loss": 0.6825,
+      "step": 15511
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.2077221870422363,
+      "learning_rate": 8.75642710463906e-06,
+      "loss": 0.7293,
+      "step": 15512
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.0260486602783203,
+      "learning_rate": 8.756253090968212e-06,
+      "loss": 0.6252,
+      "step": 15513
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.904855251312256,
+      "learning_rate": 8.756079066852631e-06,
+      "loss": 0.4942,
+      "step": 15514
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.4529643058776855,
+      "learning_rate": 8.755905032292801e-06,
+      "loss": 0.5189,
+      "step": 15515
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.0759682655334473,
+      "learning_rate": 8.755730987289206e-06,
+      "loss": 0.5617,
+      "step": 15516
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.311128854751587,
+      "learning_rate": 8.75555693184233e-06,
+      "loss": 0.8717,
+      "step": 15517
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.0077924728393555,
+      "learning_rate": 8.755382865952658e-06,
+      "loss": 0.793,
+      "step": 15518
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.865872621536255,
+      "learning_rate": 8.755208789620674e-06,
+      "loss": 0.81,
+      "step": 15519
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.5151946544647217,
+      "learning_rate": 8.755034702846857e-06,
+      "loss": 0.6436,
+      "step": 15520
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.9300867319107056,
+      "learning_rate": 8.754860605631698e-06,
+      "loss": 0.5723,
+      "step": 15521
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.4267098903656006,
+      "learning_rate": 8.754686497975677e-06,
+      "loss": 0.7291,
+      "step": 15522
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.0978708267211914,
+      "learning_rate": 8.754512379879281e-06,
+      "loss": 0.5352,
+      "step": 15523
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.401353359222412,
+      "learning_rate": 8.754338251342993e-06,
+      "loss": 0.6602,
+      "step": 15524
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.1787047386169434,
+      "learning_rate": 8.754164112367294e-06,
+      "loss": 0.8031,
+      "step": 15525
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.0376813411712646,
+      "learning_rate": 8.753989962952673e-06,
+      "loss": 0.7109,
+      "step": 15526
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.5408666133880615,
+      "learning_rate": 8.753815803099612e-06,
+      "loss": 0.6062,
+      "step": 15527
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.1305339336395264,
+      "learning_rate": 8.753641632808595e-06,
+      "loss": 0.5794,
+      "step": 15528
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.678849458694458,
+      "learning_rate": 8.753467452080107e-06,
+      "loss": 0.5501,
+      "step": 15529
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.5332436561584473,
+      "learning_rate": 8.753293260914632e-06,
+      "loss": 0.5579,
+      "step": 15530
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.3738512992858887,
+      "learning_rate": 8.753119059312653e-06,
+      "loss": 0.7319,
+      "step": 15531
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.8533538579940796,
+      "learning_rate": 8.752944847274657e-06,
+      "loss": 0.3126,
+      "step": 15532
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.1190786361694336,
+      "learning_rate": 8.75277062480113e-06,
+      "loss": 0.9742,
+      "step": 15533
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.9028573036193848,
+      "learning_rate": 8.75259639189255e-06,
+      "loss": 0.5123,
+      "step": 15534
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.683304786682129,
+      "learning_rate": 8.752422148549407e-06,
+      "loss": 0.8855,
+      "step": 15535
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.5384278297424316,
+      "learning_rate": 8.752247894772184e-06,
+      "loss": 0.5969,
+      "step": 15536
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.5166802406311035,
+      "learning_rate": 8.752073630561363e-06,
+      "loss": 0.9114,
+      "step": 15537
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.2835235595703125,
+      "learning_rate": 8.751899355917433e-06,
+      "loss": 0.8001,
+      "step": 15538
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.8433657884597778,
+      "learning_rate": 8.751725070840875e-06,
+      "loss": 0.836,
+      "step": 15539
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.03629994392395,
+      "learning_rate": 8.751550775332177e-06,
+      "loss": 0.7723,
+      "step": 15540
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.6508445739746094,
+      "learning_rate": 8.751376469391819e-06,
+      "loss": 0.7315,
+      "step": 15541
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.8149229288101196,
+      "learning_rate": 8.751202153020289e-06,
+      "loss": 0.5519,
+      "step": 15542
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.6341603994369507,
+      "learning_rate": 8.751027826218073e-06,
+      "loss": 0.7245,
+      "step": 15543
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.0356624126434326,
+      "learning_rate": 8.750853488985652e-06,
+      "loss": 0.6451,
+      "step": 15544
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.6979928016662598,
+      "learning_rate": 8.750679141323512e-06,
+      "loss": 0.8629,
+      "step": 15545
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.471190929412842,
+      "learning_rate": 8.750504783232138e-06,
+      "loss": 0.7877,
+      "step": 15546
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.7326887845993042,
+      "learning_rate": 8.750330414712019e-06,
+      "loss": 0.5466,
+      "step": 15547
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.486953854560852,
+      "learning_rate": 8.750156035763633e-06,
+      "loss": 0.6928,
+      "step": 15548
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.423706293106079,
+      "learning_rate": 8.749981646387468e-06,
+      "loss": 0.9225,
+      "step": 15549
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.280846118927002,
+      "learning_rate": 8.74980724658401e-06,
+      "loss": 0.6159,
+      "step": 15550
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.1345126628875732,
+      "learning_rate": 8.749632836353741e-06,
+      "loss": 0.7755,
+      "step": 15551
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.5441434383392334,
+      "learning_rate": 8.749458415697148e-06,
+      "loss": 0.5734,
+      "step": 15552
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.377096176147461,
+      "learning_rate": 8.749283984614716e-06,
+      "loss": 0.7057,
+      "step": 15553
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.4251461029052734,
+      "learning_rate": 8.749109543106931e-06,
+      "loss": 1.0691,
+      "step": 15554
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.436321973800659,
+      "learning_rate": 8.748935091174275e-06,
+      "loss": 0.7033,
+      "step": 15555
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.6709253787994385,
+      "learning_rate": 8.748760628817237e-06,
+      "loss": 0.84,
+      "step": 15556
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.899780511856079,
+      "learning_rate": 8.748586156036298e-06,
+      "loss": 0.5428,
+      "step": 15557
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.953487753868103,
+      "learning_rate": 8.748411672831947e-06,
+      "loss": 0.8166,
+      "step": 15558
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.6209970712661743,
+      "learning_rate": 8.748237179204667e-06,
+      "loss": 0.7365,
+      "step": 15559
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.035352945327759,
+      "learning_rate": 8.748062675154942e-06,
+      "loss": 0.7065,
+      "step": 15560
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.568601131439209,
+      "learning_rate": 8.747888160683258e-06,
+      "loss": 0.6496,
+      "step": 15561
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.835151195526123,
+      "learning_rate": 8.747713635790104e-06,
+      "loss": 0.7261,
+      "step": 15562
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.9248369932174683,
+      "learning_rate": 8.747539100475962e-06,
+      "loss": 0.6889,
+      "step": 15563
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 4.243217945098877,
+      "learning_rate": 8.747364554741315e-06,
+      "loss": 0.4761,
+      "step": 15564
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.214958906173706,
+      "learning_rate": 8.747189998586654e-06,
+      "loss": 0.4889,
+      "step": 15565
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.6866440773010254,
+      "learning_rate": 8.747015432012459e-06,
+      "loss": 0.6532,
+      "step": 15566
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.298368453979492,
+      "learning_rate": 8.746840855019218e-06,
+      "loss": 0.6366,
+      "step": 15567
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.9625673294067383,
+      "learning_rate": 8.746666267607418e-06,
+      "loss": 0.5379,
+      "step": 15568
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 4.329989910125732,
+      "learning_rate": 8.746491669777541e-06,
+      "loss": 0.8708,
+      "step": 15569
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.3477585315704346,
+      "learning_rate": 8.746317061530074e-06,
+      "loss": 0.5858,
+      "step": 15570
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.359931230545044,
+      "learning_rate": 8.746142442865504e-06,
+      "loss": 0.566,
+      "step": 15571
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.8060882091522217,
+      "learning_rate": 8.745967813784314e-06,
+      "loss": 0.7672,
+      "step": 15572
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.1753063201904297,
+      "learning_rate": 8.745793174286991e-06,
+      "loss": 0.6013,
+      "step": 15573
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.874086380004883,
+      "learning_rate": 8.74561852437402e-06,
+      "loss": 0.9178,
+      "step": 15574
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.3133926391601562,
+      "learning_rate": 8.745443864045889e-06,
+      "loss": 0.4518,
+      "step": 15575
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.6516320705413818,
+      "learning_rate": 8.74526919330308e-06,
+      "loss": 0.6405,
+      "step": 15576
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.712509870529175,
+      "learning_rate": 8.74509451214608e-06,
+      "loss": 0.8414,
+      "step": 15577
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.8804280757904053,
+      "learning_rate": 8.744919820575375e-06,
+      "loss": 0.941,
+      "step": 15578
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.8223004341125488,
+      "learning_rate": 8.744745118591454e-06,
+      "loss": 0.6954,
+      "step": 15579
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.7454217672348022,
+      "learning_rate": 8.744570406194796e-06,
+      "loss": 0.4275,
+      "step": 15580
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.9983749389648438,
+      "learning_rate": 8.74439568338589e-06,
+      "loss": 0.8972,
+      "step": 15581
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.3779494762420654,
+      "learning_rate": 8.744220950165226e-06,
+      "loss": 0.8151,
+      "step": 15582
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.8573317527770996,
+      "learning_rate": 8.744046206533282e-06,
+      "loss": 0.8274,
+      "step": 15583
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 8.60457706451416,
+      "learning_rate": 8.743871452490552e-06,
+      "loss": 0.6824,
+      "step": 15584
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.9742916822433472,
+      "learning_rate": 8.743696688037516e-06,
+      "loss": 0.5144,
+      "step": 15585
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.1088945865631104,
+      "learning_rate": 8.743521913174661e-06,
+      "loss": 0.488,
+      "step": 15586
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.4861724376678467,
+      "learning_rate": 8.743347127902477e-06,
+      "loss": 0.8478,
+      "step": 15587
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.7997000217437744,
+      "learning_rate": 8.743172332221444e-06,
+      "loss": 0.5809,
+      "step": 15588
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.3458175659179688,
+      "learning_rate": 8.742997526132053e-06,
+      "loss": 0.771,
+      "step": 15589
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.3314599990844727,
+      "learning_rate": 8.742822709634786e-06,
+      "loss": 0.8658,
+      "step": 15590
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.8079543113708496,
+      "learning_rate": 8.742647882730133e-06,
+      "loss": 0.7385,
+      "step": 15591
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.4047205448150635,
+      "learning_rate": 8.742473045418577e-06,
+      "loss": 0.7519,
+      "step": 15592
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.292250633239746,
+      "learning_rate": 8.742298197700606e-06,
+      "loss": 0.6659,
+      "step": 15593
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.5895925760269165,
+      "learning_rate": 8.742123339576705e-06,
+      "loss": 0.7883,
+      "step": 15594
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.920491337776184,
+      "learning_rate": 8.741948471047361e-06,
+      "loss": 0.6426,
+      "step": 15595
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.3003146648406982,
+      "learning_rate": 8.74177359211306e-06,
+      "loss": 0.472,
+      "step": 15596
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.5330886840820312,
+      "learning_rate": 8.74159870277429e-06,
+      "loss": 0.6127,
+      "step": 15597
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.5788981914520264,
+      "learning_rate": 8.741423803031534e-06,
+      "loss": 0.6963,
+      "step": 15598
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.4518089294433594,
+      "learning_rate": 8.74124889288528e-06,
+      "loss": 0.7052,
+      "step": 15599
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.42555570602417,
+      "learning_rate": 8.741073972336015e-06,
+      "loss": 0.6623,
+      "step": 15600
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.095412492752075,
+      "learning_rate": 8.740899041384224e-06,
+      "loss": 0.5767,
+      "step": 15601
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.3571276664733887,
+      "learning_rate": 8.740724100030395e-06,
+      "loss": 0.5628,
+      "step": 15602
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 4.238399028778076,
+      "learning_rate": 8.740549148275013e-06,
+      "loss": 0.712,
+      "step": 15603
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.1251659393310547,
+      "learning_rate": 8.740374186118566e-06,
+      "loss": 0.7092,
+      "step": 15604
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.130521774291992,
+      "learning_rate": 8.740199213561538e-06,
+      "loss": 0.947,
+      "step": 15605
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.5697219371795654,
+      "learning_rate": 8.740024230604418e-06,
+      "loss": 0.5561,
+      "step": 15606
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.708345651626587,
+      "learning_rate": 8.73984923724769e-06,
+      "loss": 0.7638,
+      "step": 15607
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.426617383956909,
+      "learning_rate": 8.739674233491844e-06,
+      "loss": 0.8157,
+      "step": 15608
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.584563970565796,
+      "learning_rate": 8.739499219337364e-06,
+      "loss": 0.9537,
+      "step": 15609
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.4029994010925293,
+      "learning_rate": 8.73932419478474e-06,
+      "loss": 0.6731,
+      "step": 15610
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.661219835281372,
+      "learning_rate": 8.739149159834453e-06,
+      "loss": 0.5519,
+      "step": 15611
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.0955052375793457,
+      "learning_rate": 8.738974114486994e-06,
+      "loss": 0.7012,
+      "step": 15612
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.839899778366089,
+      "learning_rate": 8.738799058742849e-06,
+      "loss": 0.8292,
+      "step": 15613
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.5799474716186523,
+      "learning_rate": 8.738623992602504e-06,
+      "loss": 0.6506,
+      "step": 15614
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.139949083328247,
+      "learning_rate": 8.738448916066446e-06,
+      "loss": 0.3822,
+      "step": 15615
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.2951653003692627,
+      "learning_rate": 8.738273829135165e-06,
+      "loss": 0.7163,
+      "step": 15616
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.4017438888549805,
+      "learning_rate": 8.738098731809141e-06,
+      "loss": 0.6434,
+      "step": 15617
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 4.984151363372803,
+      "learning_rate": 8.737923624088866e-06,
+      "loss": 0.5728,
+      "step": 15618
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 4.02589750289917,
+      "learning_rate": 8.737748505974825e-06,
+      "loss": 0.6017,
+      "step": 15619
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.1384689807891846,
+      "learning_rate": 8.737573377467507e-06,
+      "loss": 0.5063,
+      "step": 15620
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.2173845767974854,
+      "learning_rate": 8.737398238567395e-06,
+      "loss": 0.8872,
+      "step": 15621
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.6605148315429688,
+      "learning_rate": 8.737223089274982e-06,
+      "loss": 0.7037,
+      "step": 15622
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.855370283126831,
+      "learning_rate": 8.73704792959075e-06,
+      "loss": 0.6138,
+      "step": 15623
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.6332556009292603,
+      "learning_rate": 8.736872759515188e-06,
+      "loss": 0.6182,
+      "step": 15624
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.6927520036697388,
+      "learning_rate": 8.736697579048783e-06,
+      "loss": 0.5137,
+      "step": 15625
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.134908676147461,
+      "learning_rate": 8.736522388192022e-06,
+      "loss": 0.669,
+      "step": 15626
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.7639505863189697,
+      "learning_rate": 8.736347186945389e-06,
+      "loss": 0.5419,
+      "step": 15627
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.776189088821411,
+      "learning_rate": 8.736171975309377e-06,
+      "loss": 0.846,
+      "step": 15628
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.0464680194854736,
+      "learning_rate": 8.73599675328447e-06,
+      "loss": 0.856,
+      "step": 15629
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.507776141166687,
+      "learning_rate": 8.735821520871155e-06,
+      "loss": 0.4493,
+      "step": 15630
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.5226101875305176,
+      "learning_rate": 8.735646278069921e-06,
+      "loss": 0.4679,
+      "step": 15631
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.5270164012908936,
+      "learning_rate": 8.735471024881253e-06,
+      "loss": 0.4264,
+      "step": 15632
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.0374488830566406,
+      "learning_rate": 8.73529576130564e-06,
+      "loss": 0.3882,
+      "step": 15633
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.002671241760254,
+      "learning_rate": 8.735120487343567e-06,
+      "loss": 0.877,
+      "step": 15634
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.8369977474212646,
+      "learning_rate": 8.734945202995523e-06,
+      "loss": 0.6549,
+      "step": 15635
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.7490413188934326,
+      "learning_rate": 8.734769908262e-06,
+      "loss": 0.7313,
+      "step": 15636
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.2288498878479004,
+      "learning_rate": 8.734594603143476e-06,
+      "loss": 0.7714,
+      "step": 15637
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 4.16254997253418,
+      "learning_rate": 8.734419287640445e-06,
+      "loss": 0.6754,
+      "step": 15638
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.5518991947174072,
+      "learning_rate": 8.734243961753395e-06,
+      "loss": 0.5657,
+      "step": 15639
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.0507965087890625,
+      "learning_rate": 8.734068625482809e-06,
+      "loss": 0.8085,
+      "step": 15640
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.9145082235336304,
+      "learning_rate": 8.733893278829177e-06,
+      "loss": 1.0149,
+      "step": 15641
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.8144054412841797,
+      "learning_rate": 8.733717921792987e-06,
+      "loss": 0.7989,
+      "step": 15642
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.63980770111084,
+      "learning_rate": 8.733542554374728e-06,
+      "loss": 0.7732,
+      "step": 15643
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.896831512451172,
+      "learning_rate": 8.733367176574883e-06,
+      "loss": 0.597,
+      "step": 15644
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.47194504737854,
+      "learning_rate": 8.733191788393944e-06,
+      "loss": 0.6883,
+      "step": 15645
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.371889591217041,
+      "learning_rate": 8.733016389832395e-06,
+      "loss": 0.5092,
+      "step": 15646
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.722299575805664,
+      "learning_rate": 8.73284098089073e-06,
+      "loss": 0.8466,
+      "step": 15647
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.9911298751831055,
+      "learning_rate": 8.732665561569428e-06,
+      "loss": 0.7923,
+      "step": 15648
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.832833766937256,
+      "learning_rate": 8.732490131868985e-06,
+      "loss": 0.575,
+      "step": 15649
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.6450706720352173,
+      "learning_rate": 8.732314691789883e-06,
+      "loss": 0.6388,
+      "step": 15650
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.29876971244812,
+      "learning_rate": 8.732139241332613e-06,
+      "loss": 0.566,
+      "step": 15651
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.2280352115631104,
+      "learning_rate": 8.731963780497661e-06,
+      "loss": 0.7153,
+      "step": 15652
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.415557622909546,
+      "learning_rate": 8.731788309285516e-06,
+      "loss": 0.7621,
+      "step": 15653
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.0320558547973633,
+      "learning_rate": 8.731612827696666e-06,
+      "loss": 0.8997,
+      "step": 15654
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.406327962875366,
+      "learning_rate": 8.731437335731599e-06,
+      "loss": 0.6344,
+      "step": 15655
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.016355276107788,
+      "learning_rate": 8.731261833390803e-06,
+      "loss": 0.6486,
+      "step": 15656
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.8827924728393555,
+      "learning_rate": 8.731086320674765e-06,
+      "loss": 0.8018,
+      "step": 15657
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 5.169730186462402,
+      "learning_rate": 8.730910797583972e-06,
+      "loss": 1.0432,
+      "step": 15658
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.7752691507339478,
+      "learning_rate": 8.730735264118915e-06,
+      "loss": 0.6932,
+      "step": 15659
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.1752681732177734,
+      "learning_rate": 8.730559720280081e-06,
+      "loss": 0.8127,
+      "step": 15660
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.2185349464416504,
+      "learning_rate": 8.730384166067958e-06,
+      "loss": 0.4326,
+      "step": 15661
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.24210262298584,
+      "learning_rate": 8.730208601483034e-06,
+      "loss": 0.7247,
+      "step": 15662
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.992440104484558,
+      "learning_rate": 8.730033026525797e-06,
+      "loss": 0.4385,
+      "step": 15663
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.4058289527893066,
+      "learning_rate": 8.729857441196736e-06,
+      "loss": 0.5985,
+      "step": 15664
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.8109318017959595,
+      "learning_rate": 8.729681845496338e-06,
+      "loss": 0.7198,
+      "step": 15665
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.9202929735183716,
+      "learning_rate": 8.729506239425092e-06,
+      "loss": 0.7643,
+      "step": 15666
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.449291944503784,
+      "learning_rate": 8.729330622983486e-06,
+      "loss": 0.6862,
+      "step": 15667
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.974698543548584,
+      "learning_rate": 8.72915499617201e-06,
+      "loss": 0.6419,
+      "step": 15668
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.0604021549224854,
+      "learning_rate": 8.72897935899115e-06,
+      "loss": 0.6521,
+      "step": 15669
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.498155117034912,
+      "learning_rate": 8.728803711441395e-06,
+      "loss": 0.6472,
+      "step": 15670
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.6647493839263916,
+      "learning_rate": 8.728628053523234e-06,
+      "loss": 0.7021,
+      "step": 15671
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 4.397945880889893,
+      "learning_rate": 8.728452385237156e-06,
+      "loss": 0.9712,
+      "step": 15672
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.8276335000991821,
+      "learning_rate": 8.728276706583646e-06,
+      "loss": 0.6696,
+      "step": 15673
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 4.476650238037109,
+      "learning_rate": 8.728101017563197e-06,
+      "loss": 0.6716,
+      "step": 15674
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.8436646461486816,
+      "learning_rate": 8.727925318176297e-06,
+      "loss": 0.6841,
+      "step": 15675
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.024338483810425,
+      "learning_rate": 8.727749608423432e-06,
+      "loss": 0.5951,
+      "step": 15676
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.7941515445709229,
+      "learning_rate": 8.727573888305092e-06,
+      "loss": 0.725,
+      "step": 15677
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.556439161300659,
+      "learning_rate": 8.727398157821766e-06,
+      "loss": 0.7197,
+      "step": 15678
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.4449143409729004,
+      "learning_rate": 8.727222416973939e-06,
+      "loss": 0.4347,
+      "step": 15679
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.0043859481811523,
+      "learning_rate": 8.727046665762105e-06,
+      "loss": 0.8017,
+      "step": 15680
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.1159517765045166,
+      "learning_rate": 8.72687090418675e-06,
+      "loss": 0.5317,
+      "step": 15681
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.1089560985565186,
+      "learning_rate": 8.726695132248364e-06,
+      "loss": 0.8779,
+      "step": 15682
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.1638522148132324,
+      "learning_rate": 8.726519349947433e-06,
+      "loss": 0.5625,
+      "step": 15683
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.9998682737350464,
+      "learning_rate": 8.726343557284449e-06,
+      "loss": 0.7572,
+      "step": 15684
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.035170555114746,
+      "learning_rate": 8.7261677542599e-06,
+      "loss": 0.6018,
+      "step": 15685
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.8344086408615112,
+      "learning_rate": 8.725991940874275e-06,
+      "loss": 0.4617,
+      "step": 15686
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.0070571899414062,
+      "learning_rate": 8.725816117128059e-06,
+      "loss": 0.548,
+      "step": 15687
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.8474780321121216,
+      "learning_rate": 8.725640283021746e-06,
+      "loss": 0.7835,
+      "step": 15688
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.8774068355560303,
+      "learning_rate": 8.725464438555823e-06,
+      "loss": 0.7746,
+      "step": 15689
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.583733558654785,
+      "learning_rate": 8.72528858373078e-06,
+      "loss": 0.8706,
+      "step": 15690
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.3970584869384766,
+      "learning_rate": 8.725112718547104e-06,
+      "loss": 0.6473,
+      "step": 15691
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.4336812496185303,
+      "learning_rate": 8.724936843005286e-06,
+      "loss": 0.6534,
+      "step": 15692
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.7081384658813477,
+      "learning_rate": 8.724760957105811e-06,
+      "loss": 0.6711,
+      "step": 15693
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.1985561847686768,
+      "learning_rate": 8.724585060849173e-06,
+      "loss": 0.6542,
+      "step": 15694
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.123666286468506,
+      "learning_rate": 8.72440915423586e-06,
+      "loss": 0.8044,
+      "step": 15695
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.0889036655426025,
+      "learning_rate": 8.72423323726636e-06,
+      "loss": 0.4232,
+      "step": 15696
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.3680636882781982,
+      "learning_rate": 8.72405730994116e-06,
+      "loss": 0.7649,
+      "step": 15697
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.2537834644317627,
+      "learning_rate": 8.723881372260754e-06,
+      "loss": 0.5587,
+      "step": 15698
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.5792407989501953,
+      "learning_rate": 8.72370542422563e-06,
+      "loss": 0.8882,
+      "step": 15699
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.9109344482421875,
+      "learning_rate": 8.723529465836274e-06,
+      "loss": 0.9774,
+      "step": 15700
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.0258610248565674,
+      "learning_rate": 8.723353497093179e-06,
+      "loss": 0.5674,
+      "step": 15701
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.5931003093719482,
+      "learning_rate": 8.723177517996831e-06,
+      "loss": 0.9342,
+      "step": 15702
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.7713570594787598,
+      "learning_rate": 8.723001528547723e-06,
+      "loss": 0.7078,
+      "step": 15703
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.2318336963653564,
+      "learning_rate": 8.722825528746342e-06,
+      "loss": 0.6249,
+      "step": 15704
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.4036173820495605,
+      "learning_rate": 8.722649518593175e-06,
+      "loss": 0.9168,
+      "step": 15705
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 9.983814239501953,
+      "learning_rate": 8.722473498088717e-06,
+      "loss": 0.8214,
+      "step": 15706
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 4.571307182312012,
+      "learning_rate": 8.722297467233455e-06,
+      "loss": 0.9074,
+      "step": 15707
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 4.097952842712402,
+      "learning_rate": 8.722121426027877e-06,
+      "loss": 0.5957,
+      "step": 15708
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 4.375496864318848,
+      "learning_rate": 8.721945374472473e-06,
+      "loss": 0.7162,
+      "step": 15709
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.9999783039093018,
+      "learning_rate": 8.721769312567736e-06,
+      "loss": 0.6463,
+      "step": 15710
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.01584792137146,
+      "learning_rate": 8.72159324031415e-06,
+      "loss": 0.672,
+      "step": 15711
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.1569788455963135,
+      "learning_rate": 8.72141715771221e-06,
+      "loss": 0.5034,
+      "step": 15712
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.2755978107452393,
+      "learning_rate": 8.7212410647624e-06,
+      "loss": 0.5379,
+      "step": 15713
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.154445171356201,
+      "learning_rate": 8.721064961465213e-06,
+      "loss": 0.3384,
+      "step": 15714
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.556673288345337,
+      "learning_rate": 8.720888847821139e-06,
+      "loss": 0.7305,
+      "step": 15715
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.445580005645752,
+      "learning_rate": 8.720712723830667e-06,
+      "loss": 0.5942,
+      "step": 15716
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.38529634475708,
+      "learning_rate": 8.720536589494288e-06,
+      "loss": 0.7835,
+      "step": 15717
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.7870640754699707,
+      "learning_rate": 8.720360444812489e-06,
+      "loss": 0.5735,
+      "step": 15718
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.349682092666626,
+      "learning_rate": 8.720184289785761e-06,
+      "loss": 0.3612,
+      "step": 15719
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.1072163581848145,
+      "learning_rate": 8.720008124414594e-06,
+      "loss": 0.7409,
+      "step": 15720
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.200770139694214,
+      "learning_rate": 8.71983194869948e-06,
+      "loss": 0.5366,
+      "step": 15721
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.3344404697418213,
+      "learning_rate": 8.719655762640906e-06,
+      "loss": 0.6689,
+      "step": 15722
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.2996108531951904,
+      "learning_rate": 8.719479566239361e-06,
+      "loss": 0.6043,
+      "step": 15723
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.460742950439453,
+      "learning_rate": 8.719303359495338e-06,
+      "loss": 0.5583,
+      "step": 15724
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 4.213554859161377,
+      "learning_rate": 8.719127142409325e-06,
+      "loss": 0.6222,
+      "step": 15725
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.2231929302215576,
+      "learning_rate": 8.718950914981814e-06,
+      "loss": 0.9082,
+      "step": 15726
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.856072425842285,
+      "learning_rate": 8.718774677213293e-06,
+      "loss": 0.7437,
+      "step": 15727
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.508716106414795,
+      "learning_rate": 8.718598429104253e-06,
+      "loss": 0.5823,
+      "step": 15728
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.2000045776367188,
+      "learning_rate": 8.718422170655183e-06,
+      "loss": 0.5865,
+      "step": 15729
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.284491539001465,
+      "learning_rate": 8.718245901866572e-06,
+      "loss": 0.4195,
+      "step": 15730
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.6859523057937622,
+      "learning_rate": 8.718069622738917e-06,
+      "loss": 0.7681,
+      "step": 15731
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.861064910888672,
+      "learning_rate": 8.7178933332727e-06,
+      "loss": 0.3543,
+      "step": 15732
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.542440891265869,
+      "learning_rate": 8.717717033468414e-06,
+      "loss": 0.7489,
+      "step": 15733
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.9293582439422607,
+      "learning_rate": 8.71754072332655e-06,
+      "loss": 0.5688,
+      "step": 15734
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.462003469467163,
+      "learning_rate": 8.7173644028476e-06,
+      "loss": 0.9326,
+      "step": 15735
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.2215652465820312,
+      "learning_rate": 8.717188072032048e-06,
+      "loss": 0.5523,
+      "step": 15736
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.41418719291687,
+      "learning_rate": 8.717011730880391e-06,
+      "loss": 0.7763,
+      "step": 15737
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 7.949345111846924,
+      "learning_rate": 8.716835379393118e-06,
+      "loss": 0.7835,
+      "step": 15738
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.2945120334625244,
+      "learning_rate": 8.716659017570716e-06,
+      "loss": 0.6189,
+      "step": 15739
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.106717824935913,
+      "learning_rate": 8.716482645413679e-06,
+      "loss": 0.7689,
+      "step": 15740
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.4181339740753174,
+      "learning_rate": 8.716306262922496e-06,
+      "loss": 0.6325,
+      "step": 15741
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.6252551078796387,
+      "learning_rate": 8.716129870097657e-06,
+      "loss": 0.5864,
+      "step": 15742
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.0266506671905518,
+      "learning_rate": 8.715953466939652e-06,
+      "loss": 0.7678,
+      "step": 15743
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.6234688758850098,
+      "learning_rate": 8.715777053448974e-06,
+      "loss": 1.1756,
+      "step": 15744
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.4003703594207764,
+      "learning_rate": 8.71560062962611e-06,
+      "loss": 0.6787,
+      "step": 15745
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.6280972957611084,
+      "learning_rate": 8.715424195471555e-06,
+      "loss": 0.5893,
+      "step": 15746
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.0215065479278564,
+      "learning_rate": 8.715247750985795e-06,
+      "loss": 0.459,
+      "step": 15747
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.6542348861694336,
+      "learning_rate": 8.715071296169323e-06,
+      "loss": 0.6773,
+      "step": 15748
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.271430253982544,
+      "learning_rate": 8.71489483102263e-06,
+      "loss": 0.8761,
+      "step": 15749
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.9893982410430908,
+      "learning_rate": 8.714718355546207e-06,
+      "loss": 0.6736,
+      "step": 15750
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.0156986713409424,
+      "learning_rate": 8.714541869740543e-06,
+      "loss": 0.4984,
+      "step": 15751
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.337688684463501,
+      "learning_rate": 8.714365373606131e-06,
+      "loss": 0.6953,
+      "step": 15752
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.0373764038085938,
+      "learning_rate": 8.714188867143458e-06,
+      "loss": 0.6232,
+      "step": 15753
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.4044127464294434,
+      "learning_rate": 8.714012350353018e-06,
+      "loss": 0.9373,
+      "step": 15754
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.6475520133972168,
+      "learning_rate": 8.713835823235301e-06,
+      "loss": 0.5747,
+      "step": 15755
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 5.175723552703857,
+      "learning_rate": 8.713659285790799e-06,
+      "loss": 0.5028,
+      "step": 15756
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.246858596801758,
+      "learning_rate": 8.71348273802e-06,
+      "loss": 0.5682,
+      "step": 15757
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.459498405456543,
+      "learning_rate": 8.713306179923398e-06,
+      "loss": 0.5633,
+      "step": 15758
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.3602614402770996,
+      "learning_rate": 8.713129611501481e-06,
+      "loss": 0.5476,
+      "step": 15759
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.0736780166625977,
+      "learning_rate": 8.712953032754741e-06,
+      "loss": 0.5145,
+      "step": 15760
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.9545844793319702,
+      "learning_rate": 8.71277644368367e-06,
+      "loss": 0.565,
+      "step": 15761
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.606074810028076,
+      "learning_rate": 8.712599844288761e-06,
+      "loss": 0.7359,
+      "step": 15762
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.962860345840454,
+      "learning_rate": 8.7124232345705e-06,
+      "loss": 0.9311,
+      "step": 15763
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.1353352069854736,
+      "learning_rate": 8.712246614529383e-06,
+      "loss": 0.6363,
+      "step": 15764
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.679168224334717,
+      "learning_rate": 8.712069984165897e-06,
+      "loss": 0.6976,
+      "step": 15765
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.9770468473434448,
+      "learning_rate": 8.711893343480535e-06,
+      "loss": 0.6088,
+      "step": 15766
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.4397168159484863,
+      "learning_rate": 8.711716692473789e-06,
+      "loss": 0.504,
+      "step": 15767
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.5208489894866943,
+      "learning_rate": 8.71154003114615e-06,
+      "loss": 0.6102,
+      "step": 15768
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.655043125152588,
+      "learning_rate": 8.711363359498104e-06,
+      "loss": 0.868,
+      "step": 15769
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.0692906379699707,
+      "learning_rate": 8.71118667753015e-06,
+      "loss": 0.7205,
+      "step": 15770
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 5.566799640655518,
+      "learning_rate": 8.711009985242778e-06,
+      "loss": 0.5997,
+      "step": 15771
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.8631442785263062,
+      "learning_rate": 8.710833282636475e-06,
+      "loss": 0.5987,
+      "step": 15772
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.8023170232772827,
+      "learning_rate": 8.710656569711735e-06,
+      "loss": 0.7314,
+      "step": 15773
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.6344916820526123,
+      "learning_rate": 8.71047984646905e-06,
+      "loss": 0.7415,
+      "step": 15774
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.6923813819885254,
+      "learning_rate": 8.71030311290891e-06,
+      "loss": 0.6241,
+      "step": 15775
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.0119636058807373,
+      "learning_rate": 8.710126369031805e-06,
+      "loss": 0.4592,
+      "step": 15776
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.6486754417419434,
+      "learning_rate": 8.709949614838233e-06,
+      "loss": 0.6362,
+      "step": 15777
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.5577585697174072,
+      "learning_rate": 8.709772850328677e-06,
+      "loss": 0.5657,
+      "step": 15778
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.8634727001190186,
+      "learning_rate": 8.709596075503634e-06,
+      "loss": 0.7621,
+      "step": 15779
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.9078052043914795,
+      "learning_rate": 8.709419290363594e-06,
+      "loss": 0.5182,
+      "step": 15780
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.1232426166534424,
+      "learning_rate": 8.709242494909048e-06,
+      "loss": 0.6888,
+      "step": 15781
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.766550064086914,
+      "learning_rate": 8.709065689140489e-06,
+      "loss": 0.6616,
+      "step": 15782
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.1103971004486084,
+      "learning_rate": 8.708888873058407e-06,
+      "loss": 0.6983,
+      "step": 15783
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.414841651916504,
+      "learning_rate": 8.708712046663295e-06,
+      "loss": 0.7421,
+      "step": 15784
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.286506175994873,
+      "learning_rate": 8.708535209955643e-06,
+      "loss": 0.7903,
+      "step": 15785
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.211082935333252,
+      "learning_rate": 8.708358362935945e-06,
+      "loss": 0.8239,
+      "step": 15786
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.905756950378418,
+      "learning_rate": 8.708181505604688e-06,
+      "loss": 0.795,
+      "step": 15787
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.4241302013397217,
+      "learning_rate": 8.708004637962373e-06,
+      "loss": 0.8297,
+      "step": 15788
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.683580160140991,
+      "learning_rate": 8.707827760009482e-06,
+      "loss": 0.6512,
+      "step": 15789
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.5355379581451416,
+      "learning_rate": 8.707650871746514e-06,
+      "loss": 0.431,
+      "step": 15790
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.655531644821167,
+      "learning_rate": 8.707473973173955e-06,
+      "loss": 0.7188,
+      "step": 15791
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.256976842880249,
+      "learning_rate": 8.707297064292302e-06,
+      "loss": 0.5516,
+      "step": 15792
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.325279712677002,
+      "learning_rate": 8.707120145102044e-06,
+      "loss": 0.8384,
+      "step": 15793
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.1613733768463135,
+      "learning_rate": 8.706943215603674e-06,
+      "loss": 0.9232,
+      "step": 15794
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.4052629470825195,
+      "learning_rate": 8.706766275797683e-06,
+      "loss": 0.5709,
+      "step": 15795
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.650980234146118,
+      "learning_rate": 8.706589325684562e-06,
+      "loss": 0.5201,
+      "step": 15796
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.9722059965133667,
+      "learning_rate": 8.706412365264806e-06,
+      "loss": 0.531,
+      "step": 15797
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 4.462671279907227,
+      "learning_rate": 8.706235394538906e-06,
+      "loss": 0.6951,
+      "step": 15798
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.8745996952056885,
+      "learning_rate": 8.706058413507353e-06,
+      "loss": 0.761,
+      "step": 15799
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.9584836959838867,
+      "learning_rate": 8.70588142217064e-06,
+      "loss": 0.6064,
+      "step": 15800
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.6818692684173584,
+      "learning_rate": 8.705704420529259e-06,
+      "loss": 0.5781,
+      "step": 15801
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.4171953201293945,
+      "learning_rate": 8.705527408583703e-06,
+      "loss": 0.6158,
+      "step": 15802
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.0833375453948975,
+      "learning_rate": 8.705350386334463e-06,
+      "loss": 0.6021,
+      "step": 15803
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.4255802631378174,
+      "learning_rate": 8.705173353782032e-06,
+      "loss": 0.2942,
+      "step": 15804
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.0847692489624023,
+      "learning_rate": 8.7049963109269e-06,
+      "loss": 0.5909,
+      "step": 15805
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 4.90632438659668,
+      "learning_rate": 8.704819257769565e-06,
+      "loss": 0.8455,
+      "step": 15806
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.6683619022369385,
+      "learning_rate": 8.704642194310513e-06,
+      "loss": 0.4765,
+      "step": 15807
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.5036325454711914,
+      "learning_rate": 8.704465120550238e-06,
+      "loss": 0.8208,
+      "step": 15808
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.435676336288452,
+      "learning_rate": 8.704288036489235e-06,
+      "loss": 0.7191,
+      "step": 15809
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.900191307067871,
+      "learning_rate": 8.704110942127994e-06,
+      "loss": 0.7789,
+      "step": 15810
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.069955348968506,
+      "learning_rate": 8.703933837467008e-06,
+      "loss": 0.3616,
+      "step": 15811
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.5845284461975098,
+      "learning_rate": 8.703756722506769e-06,
+      "loss": 0.8408,
+      "step": 15812
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.292160987854004,
+      "learning_rate": 8.703579597247771e-06,
+      "loss": 0.5544,
+      "step": 15813
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.8245488405227661,
+      "learning_rate": 8.703402461690505e-06,
+      "loss": 0.3693,
+      "step": 15814
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.016075611114502,
+      "learning_rate": 8.703225315835464e-06,
+      "loss": 0.8724,
+      "step": 15815
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.4418015480041504,
+      "learning_rate": 8.70304815968314e-06,
+      "loss": 0.5982,
+      "step": 15816
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.5937368869781494,
+      "learning_rate": 8.702870993234028e-06,
+      "loss": 0.5937,
+      "step": 15817
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.8013193607330322,
+      "learning_rate": 8.702693816488617e-06,
+      "loss": 0.9986,
+      "step": 15818
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.33454966545105,
+      "learning_rate": 8.702516629447404e-06,
+      "loss": 0.6043,
+      "step": 15819
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.931457042694092,
+      "learning_rate": 8.702339432110876e-06,
+      "loss": 0.7532,
+      "step": 15820
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.935817003250122,
+      "learning_rate": 8.70216222447953e-06,
+      "loss": 0.6761,
+      "step": 15821
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.8924282789230347,
+      "learning_rate": 8.70198500655386e-06,
+      "loss": 0.6242,
+      "step": 15822
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.7820658683776855,
+      "learning_rate": 8.701807778334355e-06,
+      "loss": 0.8409,
+      "step": 15823
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.671173095703125,
+      "learning_rate": 8.701630539821509e-06,
+      "loss": 0.8168,
+      "step": 15824
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.697225332260132,
+      "learning_rate": 8.701453291015815e-06,
+      "loss": 0.7738,
+      "step": 15825
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.6505849361419678,
+      "learning_rate": 8.701276031917767e-06,
+      "loss": 0.7446,
+      "step": 15826
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.8660690784454346,
+      "learning_rate": 8.701098762527855e-06,
+      "loss": 0.3956,
+      "step": 15827
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 4.091864585876465,
+      "learning_rate": 8.700921482846575e-06,
+      "loss": 0.4806,
+      "step": 15828
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.8931864500045776,
+      "learning_rate": 8.70074419287442e-06,
+      "loss": 0.7685,
+      "step": 15829
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.8621412515640259,
+      "learning_rate": 8.70056689261188e-06,
+      "loss": 0.5075,
+      "step": 15830
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.719016194343567,
+      "learning_rate": 8.70038958205945e-06,
+      "loss": 0.5589,
+      "step": 15831
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.554830551147461,
+      "learning_rate": 8.700212261217624e-06,
+      "loss": 0.623,
+      "step": 15832
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.5336320400238037,
+      "learning_rate": 8.700034930086892e-06,
+      "loss": 0.6723,
+      "step": 15833
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.8616421222686768,
+      "learning_rate": 8.699857588667751e-06,
+      "loss": 0.6392,
+      "step": 15834
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.9436466693878174,
+      "learning_rate": 8.69968023696069e-06,
+      "loss": 0.8658,
+      "step": 15835
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.787562608718872,
+      "learning_rate": 8.699502874966204e-06,
+      "loss": 0.6357,
+      "step": 15836
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.762606143951416,
+      "learning_rate": 8.699325502684787e-06,
+      "loss": 0.7301,
+      "step": 15837
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 4.136106967926025,
+      "learning_rate": 8.699148120116933e-06,
+      "loss": 1.0774,
+      "step": 15838
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.3051209449768066,
+      "learning_rate": 8.698970727263133e-06,
+      "loss": 0.6309,
+      "step": 15839
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.040281295776367,
+      "learning_rate": 8.69879332412388e-06,
+      "loss": 0.5832,
+      "step": 15840
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.630781888961792,
+      "learning_rate": 8.69861591069967e-06,
+      "loss": 0.6852,
+      "step": 15841
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.9123064279556274,
+      "learning_rate": 8.698438486990995e-06,
+      "loss": 0.9077,
+      "step": 15842
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.0535693168640137,
+      "learning_rate": 8.698261052998346e-06,
+      "loss": 0.592,
+      "step": 15843
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.3451638221740723,
+      "learning_rate": 8.698083608722219e-06,
+      "loss": 0.7612,
+      "step": 15844
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 5.4002885818481445,
+      "learning_rate": 8.697906154163107e-06,
+      "loss": 0.6733,
+      "step": 15845
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.578873872756958,
+      "learning_rate": 8.697728689321502e-06,
+      "loss": 0.5519,
+      "step": 15846
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.826797366142273,
+      "learning_rate": 8.6975512141979e-06,
+      "loss": 0.6645,
+      "step": 15847
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.71150279045105,
+      "learning_rate": 8.697373728792794e-06,
+      "loss": 0.8361,
+      "step": 15848
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.9303486347198486,
+      "learning_rate": 8.697196233106677e-06,
+      "loss": 0.7502,
+      "step": 15849
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 5.815851211547852,
+      "learning_rate": 8.69701872714004e-06,
+      "loss": 0.8631,
+      "step": 15850
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.0751497745513916,
+      "learning_rate": 8.69684121089338e-06,
+      "loss": 0.7411,
+      "step": 15851
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.3511102199554443,
+      "learning_rate": 8.69666368436719e-06,
+      "loss": 0.6042,
+      "step": 15852
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.178271532058716,
+      "learning_rate": 8.696486147561964e-06,
+      "loss": 0.5986,
+      "step": 15853
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 26.644268035888672,
+      "learning_rate": 8.696308600478194e-06,
+      "loss": 0.9653,
+      "step": 15854
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.0834245681762695,
+      "learning_rate": 8.696131043116374e-06,
+      "loss": 0.9077,
+      "step": 15855
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.0805304050445557,
+      "learning_rate": 8.695953475476997e-06,
+      "loss": 0.7641,
+      "step": 15856
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.753981590270996,
+      "learning_rate": 8.695775897560559e-06,
+      "loss": 0.5888,
+      "step": 15857
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.2865138053894043,
+      "learning_rate": 8.695598309367554e-06,
+      "loss": 0.7289,
+      "step": 15858
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.38891339302063,
+      "learning_rate": 8.695420710898472e-06,
+      "loss": 0.5718,
+      "step": 15859
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.0441551208496094,
+      "learning_rate": 8.69524310215381e-06,
+      "loss": 0.6812,
+      "step": 15860
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 4.142175674438477,
+      "learning_rate": 8.695065483134063e-06,
+      "loss": 0.8762,
+      "step": 15861
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.6749491691589355,
+      "learning_rate": 8.694887853839722e-06,
+      "loss": 0.7452,
+      "step": 15862
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.4764528274536133,
+      "learning_rate": 8.694710214271281e-06,
+      "loss": 0.8378,
+      "step": 15863
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.769836902618408,
+      "learning_rate": 8.694532564429236e-06,
+      "loss": 0.6081,
+      "step": 15864
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.69600510597229,
+      "learning_rate": 8.694354904314079e-06,
+      "loss": 0.5885,
+      "step": 15865
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.1373491287231445,
+      "learning_rate": 8.694177233926306e-06,
+      "loss": 0.6311,
+      "step": 15866
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.476287603378296,
+      "learning_rate": 8.69399955326641e-06,
+      "loss": 0.7773,
+      "step": 15867
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.2967255115509033,
+      "learning_rate": 8.693821862334886e-06,
+      "loss": 0.4406,
+      "step": 15868
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.9648959636688232,
+      "learning_rate": 8.693644161132226e-06,
+      "loss": 0.8787,
+      "step": 15869
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.590052604675293,
+      "learning_rate": 8.693466449658925e-06,
+      "loss": 0.6865,
+      "step": 15870
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.8100569248199463,
+      "learning_rate": 8.693288727915477e-06,
+      "loss": 0.5779,
+      "step": 15871
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.7536110877990723,
+      "learning_rate": 8.693110995902376e-06,
+      "loss": 0.6717,
+      "step": 15872
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.6908414363861084,
+      "learning_rate": 8.69293325362012e-06,
+      "loss": 0.7304,
+      "step": 15873
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 6.6785454750061035,
+      "learning_rate": 8.692755501069197e-06,
+      "loss": 0.7151,
+      "step": 15874
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.1354730129241943,
+      "learning_rate": 8.692577738250106e-06,
+      "loss": 0.5683,
+      "step": 15875
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.2098605632781982,
+      "learning_rate": 8.692399965163339e-06,
+      "loss": 0.4435,
+      "step": 15876
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.5412265062332153,
+      "learning_rate": 8.69222218180939e-06,
+      "loss": 0.411,
+      "step": 15877
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.2787747383117676,
+      "learning_rate": 8.692044388188755e-06,
+      "loss": 0.842,
+      "step": 15878
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.0728678703308105,
+      "learning_rate": 8.691866584301927e-06,
+      "loss": 0.5672,
+      "step": 15879
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.1431076526641846,
+      "learning_rate": 8.691688770149402e-06,
+      "loss": 0.6921,
+      "step": 15880
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.404447078704834,
+      "learning_rate": 8.691510945731673e-06,
+      "loss": 0.5573,
+      "step": 15881
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.9450161457061768,
+      "learning_rate": 8.691333111049236e-06,
+      "loss": 0.6571,
+      "step": 15882
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.9374836683273315,
+      "learning_rate": 8.691155266102583e-06,
+      "loss": 0.3812,
+      "step": 15883
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.292196035385132,
+      "learning_rate": 8.69097741089221e-06,
+      "loss": 0.4405,
+      "step": 15884
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.1836600303649902,
+      "learning_rate": 8.690799545418612e-06,
+      "loss": 0.5851,
+      "step": 15885
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.755723476409912,
+      "learning_rate": 8.690621669682283e-06,
+      "loss": 0.5376,
+      "step": 15886
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.50445294380188,
+      "learning_rate": 8.690443783683717e-06,
+      "loss": 0.7274,
+      "step": 15887
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.9864674806594849,
+      "learning_rate": 8.69026588742341e-06,
+      "loss": 0.5546,
+      "step": 15888
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.2983226776123047,
+      "learning_rate": 8.690087980901857e-06,
+      "loss": 0.8075,
+      "step": 15889
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.144474744796753,
+      "learning_rate": 8.68991006411955e-06,
+      "loss": 0.7982,
+      "step": 15890
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.6857974529266357,
+      "learning_rate": 8.689732137076987e-06,
+      "loss": 0.8557,
+      "step": 15891
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.7505943775177,
+      "learning_rate": 8.689554199774658e-06,
+      "loss": 0.6777,
+      "step": 15892
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.558285713195801,
+      "learning_rate": 8.689376252213064e-06,
+      "loss": 0.5935,
+      "step": 15893
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.2419276237487793,
+      "learning_rate": 8.689198294392695e-06,
+      "loss": 0.6191,
+      "step": 15894
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 4.070488452911377,
+      "learning_rate": 8.68902032631405e-06,
+      "loss": 0.6457,
+      "step": 15895
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.673685073852539,
+      "learning_rate": 8.68884234797762e-06,
+      "loss": 0.5305,
+      "step": 15896
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.628786563873291,
+      "learning_rate": 8.688664359383901e-06,
+      "loss": 0.6139,
+      "step": 15897
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.6775000095367432,
+      "learning_rate": 8.688486360533387e-06,
+      "loss": 0.6148,
+      "step": 15898
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.8117061853408813,
+      "learning_rate": 8.688308351426575e-06,
+      "loss": 0.6078,
+      "step": 15899
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.5501585006713867,
+      "learning_rate": 8.68813033206396e-06,
+      "loss": 0.8571,
+      "step": 15900
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.689176082611084,
+      "learning_rate": 8.687952302446037e-06,
+      "loss": 0.639,
+      "step": 15901
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.611698627471924,
+      "learning_rate": 8.687774262573299e-06,
+      "loss": 0.6824,
+      "step": 15902
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.969156265258789,
+      "learning_rate": 8.687596212446241e-06,
+      "loss": 0.6158,
+      "step": 15903
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.869912624359131,
+      "learning_rate": 8.68741815206536e-06,
+      "loss": 0.6466,
+      "step": 15904
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.3641769886016846,
+      "learning_rate": 8.687240081431152e-06,
+      "loss": 0.4752,
+      "step": 15905
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.1978747844696045,
+      "learning_rate": 8.687062000544108e-06,
+      "loss": 0.5602,
+      "step": 15906
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.37245774269104,
+      "learning_rate": 8.686883909404728e-06,
+      "loss": 0.5035,
+      "step": 15907
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.9389114379882812,
+      "learning_rate": 8.686705808013505e-06,
+      "loss": 0.5722,
+      "step": 15908
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.863309621810913,
+      "learning_rate": 8.686527696370934e-06,
+      "loss": 0.7939,
+      "step": 15909
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.140807867050171,
+      "learning_rate": 8.686349574477509e-06,
+      "loss": 0.3609,
+      "step": 15910
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.052124500274658,
+      "learning_rate": 8.686171442333728e-06,
+      "loss": 0.7587,
+      "step": 15911
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.9351706504821777,
+      "learning_rate": 8.685993299940086e-06,
+      "loss": 0.6847,
+      "step": 15912
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.9966944456100464,
+      "learning_rate": 8.685815147297075e-06,
+      "loss": 0.6596,
+      "step": 15913
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.491349220275879,
+      "learning_rate": 8.685636984405195e-06,
+      "loss": 0.6048,
+      "step": 15914
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.475357174873352,
+      "learning_rate": 8.685458811264938e-06,
+      "loss": 0.7363,
+      "step": 15915
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.8421125411987305,
+      "learning_rate": 8.685280627876801e-06,
+      "loss": 0.5478,
+      "step": 15916
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.2678630352020264,
+      "learning_rate": 8.68510243424128e-06,
+      "loss": 0.6581,
+      "step": 15917
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.991370439529419,
+      "learning_rate": 8.684924230358868e-06,
+      "loss": 0.5834,
+      "step": 15918
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.9602134227752686,
+      "learning_rate": 8.684746016230062e-06,
+      "loss": 0.5523,
+      "step": 15919
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.476189136505127,
+      "learning_rate": 8.684567791855359e-06,
+      "loss": 0.6192,
+      "step": 15920
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.114161491394043,
+      "learning_rate": 8.684389557235254e-06,
+      "loss": 0.7247,
+      "step": 15921
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.269197940826416,
+      "learning_rate": 8.684211312370239e-06,
+      "loss": 0.5357,
+      "step": 15922
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.873943328857422,
+      "learning_rate": 8.684033057260815e-06,
+      "loss": 0.6089,
+      "step": 15923
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.474144697189331,
+      "learning_rate": 8.683854791907473e-06,
+      "loss": 0.832,
+      "step": 15924
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.9997645616531372,
+      "learning_rate": 8.683676516310712e-06,
+      "loss": 0.9153,
+      "step": 15925
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.5917619466781616,
+      "learning_rate": 8.683498230471026e-06,
+      "loss": 0.5579,
+      "step": 15926
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.4706571102142334,
+      "learning_rate": 8.683319934388913e-06,
+      "loss": 0.7275,
+      "step": 15927
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.7177464962005615,
+      "learning_rate": 8.683141628064866e-06,
+      "loss": 0.6272,
+      "step": 15928
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.0094122886657715,
+      "learning_rate": 8.682963311499382e-06,
+      "loss": 0.6932,
+      "step": 15929
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.1237597465515137,
+      "learning_rate": 8.682784984692956e-06,
+      "loss": 0.8092,
+      "step": 15930
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.3220746517181396,
+      "learning_rate": 8.682606647646086e-06,
+      "loss": 0.7226,
+      "step": 15931
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.1054646968841553,
+      "learning_rate": 8.682428300359265e-06,
+      "loss": 0.8505,
+      "step": 15932
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.746140956878662,
+      "learning_rate": 8.68224994283299e-06,
+      "loss": 0.9312,
+      "step": 15933
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.6988232135772705,
+      "learning_rate": 8.68207157506776e-06,
+      "loss": 0.7405,
+      "step": 15934
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.891392707824707,
+      "learning_rate": 8.681893197064065e-06,
+      "loss": 0.675,
+      "step": 15935
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.4324474334716797,
+      "learning_rate": 8.681714808822405e-06,
+      "loss": 0.4399,
+      "step": 15936
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.9146947860717773,
+      "learning_rate": 8.681536410343277e-06,
+      "loss": 0.7615,
+      "step": 15937
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.18855357170105,
+      "learning_rate": 8.681358001627174e-06,
+      "loss": 0.6905,
+      "step": 15938
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.044262409210205,
+      "learning_rate": 8.681179582674595e-06,
+      "loss": 0.6358,
+      "step": 15939
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.4760584831237793,
+      "learning_rate": 8.681001153486032e-06,
+      "loss": 0.6611,
+      "step": 15940
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.3950276374816895,
+      "learning_rate": 8.680822714061987e-06,
+      "loss": 0.8075,
+      "step": 15941
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.089102268218994,
+      "learning_rate": 8.68064426440295e-06,
+      "loss": 0.5553,
+      "step": 15942
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.003216028213501,
+      "learning_rate": 8.680465804509422e-06,
+      "loss": 0.7509,
+      "step": 15943
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.018139362335205,
+      "learning_rate": 8.680287334381896e-06,
+      "loss": 0.7286,
+      "step": 15944
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.500089645385742,
+      "learning_rate": 8.68010885402087e-06,
+      "loss": 0.5831,
+      "step": 15945
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.0334479808807373,
+      "learning_rate": 8.67993036342684e-06,
+      "loss": 0.6919,
+      "step": 15946
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.6051784753799438,
+      "learning_rate": 8.679751862600302e-06,
+      "loss": 0.8701,
+      "step": 15947
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.6295955181121826,
+      "learning_rate": 8.679573351541752e-06,
+      "loss": 0.7952,
+      "step": 15948
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.084831476211548,
+      "learning_rate": 8.679394830251689e-06,
+      "loss": 0.6631,
+      "step": 15949
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.614586353302002,
+      "learning_rate": 8.679216298730606e-06,
+      "loss": 0.5731,
+      "step": 15950
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.7149580717086792,
+      "learning_rate": 8.679037756979e-06,
+      "loss": 0.6188,
+      "step": 15951
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.3220272064208984,
+      "learning_rate": 8.67885920499737e-06,
+      "loss": 0.5432,
+      "step": 15952
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.7247092723846436,
+      "learning_rate": 8.678680642786208e-06,
+      "loss": 0.9715,
+      "step": 15953
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.012619733810425,
+      "learning_rate": 8.678502070346016e-06,
+      "loss": 0.7424,
+      "step": 15954
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.6354565620422363,
+      "learning_rate": 8.678323487677286e-06,
+      "loss": 0.4927,
+      "step": 15955
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.700540781021118,
+      "learning_rate": 8.678144894780518e-06,
+      "loss": 0.7887,
+      "step": 15956
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.5835988521575928,
+      "learning_rate": 8.677966291656204e-06,
+      "loss": 0.5124,
+      "step": 15957
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.7581675052642822,
+      "learning_rate": 8.677787678304846e-06,
+      "loss": 0.5698,
+      "step": 15958
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.9742509126663208,
+      "learning_rate": 8.677609054726937e-06,
+      "loss": 0.4976,
+      "step": 15959
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.3208274841308594,
+      "learning_rate": 8.677430420922975e-06,
+      "loss": 0.5307,
+      "step": 15960
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.0417540073394775,
+      "learning_rate": 8.677251776893459e-06,
+      "loss": 0.8793,
+      "step": 15961
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.130368232727051,
+      "learning_rate": 8.67707312263888e-06,
+      "loss": 0.5537,
+      "step": 15962
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.64068603515625,
+      "learning_rate": 8.676894458159738e-06,
+      "loss": 0.6677,
+      "step": 15963
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.1701085567474365,
+      "learning_rate": 8.676715783456533e-06,
+      "loss": 0.7437,
+      "step": 15964
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.8250277042388916,
+      "learning_rate": 8.676537098529756e-06,
+      "loss": 0.9734,
+      "step": 15965
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.081289529800415,
+      "learning_rate": 8.676358403379907e-06,
+      "loss": 0.7144,
+      "step": 15966
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.0173110961914062,
+      "learning_rate": 8.676179698007486e-06,
+      "loss": 0.3938,
+      "step": 15967
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.5374605655670166,
+      "learning_rate": 8.676000982412983e-06,
+      "loss": 0.3922,
+      "step": 15968
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.4252467155456543,
+      "learning_rate": 8.675822256596898e-06,
+      "loss": 0.6373,
+      "step": 15969
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.336106300354004,
+      "learning_rate": 8.675643520559729e-06,
+      "loss": 0.5604,
+      "step": 15970
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.9920744895935059,
+      "learning_rate": 8.675464774301972e-06,
+      "loss": 0.4245,
+      "step": 15971
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.4216995239257812,
+      "learning_rate": 8.675286017824124e-06,
+      "loss": 0.6727,
+      "step": 15972
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.4032328128814697,
+      "learning_rate": 8.675107251126685e-06,
+      "loss": 0.62,
+      "step": 15973
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.9876923561096191,
+      "learning_rate": 8.674928474210147e-06,
+      "loss": 0.5344,
+      "step": 15974
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.580425500869751,
+      "learning_rate": 8.67474968707501e-06,
+      "loss": 0.824,
+      "step": 15975
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.6269235610961914,
+      "learning_rate": 8.674570889721773e-06,
+      "loss": 0.6105,
+      "step": 15976
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.1342833042144775,
+      "learning_rate": 8.674392082150926e-06,
+      "loss": 0.7073,
+      "step": 15977
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.462567090988159,
+      "learning_rate": 8.674213264362974e-06,
+      "loss": 0.8553,
+      "step": 15978
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.0522987842559814,
+      "learning_rate": 8.674034436358411e-06,
+      "loss": 0.638,
+      "step": 15979
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.831113815307617,
+      "learning_rate": 8.673855598137735e-06,
+      "loss": 0.852,
+      "step": 15980
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.021123170852661,
+      "learning_rate": 8.673676749701442e-06,
+      "loss": 0.4551,
+      "step": 15981
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.6263670921325684,
+      "learning_rate": 8.673497891050032e-06,
+      "loss": 0.5333,
+      "step": 15982
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.9013203382492065,
+      "learning_rate": 8.673319022183999e-06,
+      "loss": 0.8155,
+      "step": 15983
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.6216375827789307,
+      "learning_rate": 8.673140143103842e-06,
+      "loss": 0.836,
+      "step": 15984
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.661494493484497,
+      "learning_rate": 8.672961253810058e-06,
+      "loss": 0.8356,
+      "step": 15985
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.9668141603469849,
+      "learning_rate": 8.672782354303145e-06,
+      "loss": 0.5688,
+      "step": 15986
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.6267073154449463,
+      "learning_rate": 8.6726034445836e-06,
+      "loss": 0.5541,
+      "step": 15987
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.6085078716278076,
+      "learning_rate": 8.67242452465192e-06,
+      "loss": 0.506,
+      "step": 15988
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.9132418632507324,
+      "learning_rate": 8.672245594508603e-06,
+      "loss": 0.7255,
+      "step": 15989
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.7479482889175415,
+      "learning_rate": 8.672066654154148e-06,
+      "loss": 0.5329,
+      "step": 15990
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 5.072632789611816,
+      "learning_rate": 8.671887703589053e-06,
+      "loss": 0.5454,
+      "step": 15991
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.2517595291137695,
+      "learning_rate": 8.67170874281381e-06,
+      "loss": 0.7325,
+      "step": 15992
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.3554494380950928,
+      "learning_rate": 8.671529771828921e-06,
+      "loss": 0.832,
+      "step": 15993
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.463074803352356,
+      "learning_rate": 8.671350790634884e-06,
+      "loss": 0.7594,
+      "step": 15994
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.5385351181030273,
+      "learning_rate": 8.671171799232196e-06,
+      "loss": 0.855,
+      "step": 15995
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.48750638961792,
+      "learning_rate": 8.670992797621355e-06,
+      "loss": 0.8491,
+      "step": 15996
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.6437153816223145,
+      "learning_rate": 8.670813785802859e-06,
+      "loss": 0.9383,
+      "step": 15997
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.7436859607696533,
+      "learning_rate": 8.670634763777202e-06,
+      "loss": 0.8628,
+      "step": 15998
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.8335108757019043,
+      "learning_rate": 8.670455731544887e-06,
+      "loss": 0.6944,
+      "step": 15999
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.758056640625,
+      "learning_rate": 8.670276689106409e-06,
+      "loss": 1.0431,
+      "step": 16000
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.4720518589019775,
+      "learning_rate": 8.670097636462266e-06,
+      "loss": 0.7907,
+      "step": 16001
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.051922559738159,
+      "learning_rate": 8.669918573612957e-06,
+      "loss": 0.6632,
+      "step": 16002
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.9935153722763062,
+      "learning_rate": 8.669739500558978e-06,
+      "loss": 0.6633,
+      "step": 16003
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.6885604858398438,
+      "learning_rate": 8.669560417300829e-06,
+      "loss": 0.5929,
+      "step": 16004
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.1727707386016846,
+      "learning_rate": 8.669381323839009e-06,
+      "loss": 0.6375,
+      "step": 16005
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.670106291770935,
+      "learning_rate": 8.669202220174011e-06,
+      "loss": 0.4729,
+      "step": 16006
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.1883232593536377,
+      "learning_rate": 8.669023106306337e-06,
+      "loss": 0.5772,
+      "step": 16007
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.438218116760254,
+      "learning_rate": 8.668843982236485e-06,
+      "loss": 0.6705,
+      "step": 16008
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.5237293243408203,
+      "learning_rate": 8.668664847964952e-06,
+      "loss": 0.7295,
+      "step": 16009
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.559630870819092,
+      "learning_rate": 8.668485703492236e-06,
+      "loss": 0.7001,
+      "step": 16010
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 6.040968418121338,
+      "learning_rate": 8.668306548818836e-06,
+      "loss": 0.6515,
+      "step": 16011
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.3364884853363037,
+      "learning_rate": 8.668127383945249e-06,
+      "loss": 0.3639,
+      "step": 16012
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.7379381656646729,
+      "learning_rate": 8.667948208871974e-06,
+      "loss": 0.5277,
+      "step": 16013
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.2507097721099854,
+      "learning_rate": 8.66776902359951e-06,
+      "loss": 0.4087,
+      "step": 16014
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.5829209089279175,
+      "learning_rate": 8.667589828128354e-06,
+      "loss": 0.5022,
+      "step": 16015
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 6.1941633224487305,
+      "learning_rate": 8.667410622459004e-06,
+      "loss": 0.6196,
+      "step": 16016
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.0289268493652344,
+      "learning_rate": 8.667231406591958e-06,
+      "loss": 0.5622,
+      "step": 16017
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.270655393600464,
+      "learning_rate": 8.667052180527718e-06,
+      "loss": 0.7002,
+      "step": 16018
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.608736991882324,
+      "learning_rate": 8.666872944266778e-06,
+      "loss": 0.7862,
+      "step": 16019
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.389498710632324,
+      "learning_rate": 8.666693697809637e-06,
+      "loss": 0.463,
+      "step": 16020
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 5.62608003616333,
+      "learning_rate": 8.666514441156794e-06,
+      "loss": 0.5912,
+      "step": 16021
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.6915571689605713,
+      "learning_rate": 8.666335174308751e-06,
+      "loss": 0.6651,
+      "step": 16022
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.167309522628784,
+      "learning_rate": 8.666155897266003e-06,
+      "loss": 0.8366,
+      "step": 16023
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.983669400215149,
+      "learning_rate": 8.665976610029045e-06,
+      "loss": 0.5512,
+      "step": 16024
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.657074213027954,
+      "learning_rate": 8.665797312598383e-06,
+      "loss": 0.7837,
+      "step": 16025
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.851615071296692,
+      "learning_rate": 8.665618004974509e-06,
+      "loss": 0.8053,
+      "step": 16026
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.230041265487671,
+      "learning_rate": 8.665438687157927e-06,
+      "loss": 0.693,
+      "step": 16027
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.29138445854187,
+      "learning_rate": 8.665259359149132e-06,
+      "loss": 0.734,
+      "step": 16028
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.2059552669525146,
+      "learning_rate": 8.665080020948624e-06,
+      "loss": 0.4494,
+      "step": 16029
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.8310656547546387,
+      "learning_rate": 8.664900672556902e-06,
+      "loss": 0.6176,
+      "step": 16030
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 4.42653226852417,
+      "learning_rate": 8.664721313974464e-06,
+      "loss": 0.9714,
+      "step": 16031
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.7383716106414795,
+      "learning_rate": 8.664541945201808e-06,
+      "loss": 0.641,
+      "step": 16032
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.5596694946289062,
+      "learning_rate": 8.664362566239435e-06,
+      "loss": 0.5267,
+      "step": 16033
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.398186445236206,
+      "learning_rate": 8.664183177087841e-06,
+      "loss": 0.9594,
+      "step": 16034
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.7787013053894043,
+      "learning_rate": 8.664003777747526e-06,
+      "loss": 0.7997,
+      "step": 16035
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.701035261154175,
+      "learning_rate": 8.663824368218989e-06,
+      "loss": 0.3717,
+      "step": 16036
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.9613107442855835,
+      "learning_rate": 8.66364494850273e-06,
+      "loss": 0.8474,
+      "step": 16037
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.2718067169189453,
+      "learning_rate": 8.663465518599248e-06,
+      "loss": 0.7585,
+      "step": 16038
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 27.13405418395996,
+      "learning_rate": 8.663286078509038e-06,
+      "loss": 0.8144,
+      "step": 16039
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.7067675590515137,
+      "learning_rate": 8.663106628232602e-06,
+      "loss": 0.7885,
+      "step": 16040
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.340451955795288,
+      "learning_rate": 8.66292716777044e-06,
+      "loss": 0.5008,
+      "step": 16041
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.640317440032959,
+      "learning_rate": 8.66274769712305e-06,
+      "loss": 0.635,
+      "step": 16042
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.2943265438079834,
+      "learning_rate": 8.66256821629093e-06,
+      "loss": 0.593,
+      "step": 16043
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.05657696723938,
+      "learning_rate": 8.66238872527458e-06,
+      "loss": 0.7972,
+      "step": 16044
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.5566362142562866,
+      "learning_rate": 8.662209224074499e-06,
+      "loss": 0.6179,
+      "step": 16045
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.000849485397339,
+      "learning_rate": 8.662029712691186e-06,
+      "loss": 0.5608,
+      "step": 16046
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.943760633468628,
+      "learning_rate": 8.66185019112514e-06,
+      "loss": 0.4874,
+      "step": 16047
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.621215581893921,
+      "learning_rate": 8.66167065937686e-06,
+      "loss": 0.4744,
+      "step": 16048
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.9256235361099243,
+      "learning_rate": 8.661491117446848e-06,
+      "loss": 0.6243,
+      "step": 16049
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.1542327404022217,
+      "learning_rate": 8.661311565335598e-06,
+      "loss": 0.4123,
+      "step": 16050
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.341804027557373,
+      "learning_rate": 8.661132003043613e-06,
+      "loss": 0.6386,
+      "step": 16051
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.9858430624008179,
+      "learning_rate": 8.660952430571392e-06,
+      "loss": 0.8628,
+      "step": 16052
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 4.193539142608643,
+      "learning_rate": 8.660772847919432e-06,
+      "loss": 0.4022,
+      "step": 16053
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 4.3524675369262695,
+      "learning_rate": 8.660593255088235e-06,
+      "loss": 0.7254,
+      "step": 16054
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.4623425006866455,
+      "learning_rate": 8.6604136520783e-06,
+      "loss": 0.5806,
+      "step": 16055
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.928560256958008,
+      "learning_rate": 8.660234038890125e-06,
+      "loss": 0.5166,
+      "step": 16056
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.584731340408325,
+      "learning_rate": 8.660054415524212e-06,
+      "loss": 0.453,
+      "step": 16057
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.4622575044631958,
+      "learning_rate": 8.659874781981056e-06,
+      "loss": 0.658,
+      "step": 16058
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.026829957962036,
+      "learning_rate": 8.659695138261162e-06,
+      "loss": 0.7108,
+      "step": 16059
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.432588577270508,
+      "learning_rate": 8.659515484365025e-06,
+      "loss": 0.8382,
+      "step": 16060
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.866896629333496,
+      "learning_rate": 8.659335820293146e-06,
+      "loss": 0.3306,
+      "step": 16061
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.4142870903015137,
+      "learning_rate": 8.659156146046027e-06,
+      "loss": 0.4701,
+      "step": 16062
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.592250347137451,
+      "learning_rate": 8.658976461624165e-06,
+      "loss": 0.7848,
+      "step": 16063
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.6859219074249268,
+      "learning_rate": 8.65879676702806e-06,
+      "loss": 0.4761,
+      "step": 16064
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.6544911861419678,
+      "learning_rate": 8.65861706225821e-06,
+      "loss": 0.7945,
+      "step": 16065
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.1589362621307373,
+      "learning_rate": 8.65843734731512e-06,
+      "loss": 0.662,
+      "step": 16066
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.873211622238159,
+      "learning_rate": 8.658257622199282e-06,
+      "loss": 0.8408,
+      "step": 16067
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.9141753911972046,
+      "learning_rate": 8.658077886911202e-06,
+      "loss": 0.5592,
+      "step": 16068
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.8100594282150269,
+      "learning_rate": 8.657898141451377e-06,
+      "loss": 0.8602,
+      "step": 16069
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.911212682723999,
+      "learning_rate": 8.657718385820308e-06,
+      "loss": 0.6104,
+      "step": 16070
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.1070265769958496,
+      "learning_rate": 8.657538620018495e-06,
+      "loss": 0.548,
+      "step": 16071
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.581697463989258,
+      "learning_rate": 8.657358844046437e-06,
+      "loss": 0.5404,
+      "step": 16072
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.0082287788391113,
+      "learning_rate": 8.657179057904634e-06,
+      "loss": 0.5367,
+      "step": 16073
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.7542321681976318,
+      "learning_rate": 8.656999261593585e-06,
+      "loss": 0.7212,
+      "step": 16074
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.607466220855713,
+      "learning_rate": 8.656819455113793e-06,
+      "loss": 0.5454,
+      "step": 16075
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.126058578491211,
+      "learning_rate": 8.656639638465756e-06,
+      "loss": 0.7852,
+      "step": 16076
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.8826537132263184,
+      "learning_rate": 8.656459811649972e-06,
+      "loss": 0.7333,
+      "step": 16077
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 4.707444190979004,
+      "learning_rate": 8.656279974666943e-06,
+      "loss": 0.511,
+      "step": 16078
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.9054274559020996,
+      "learning_rate": 8.656100127517171e-06,
+      "loss": 0.7841,
+      "step": 16079
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.4374544620513916,
+      "learning_rate": 8.655920270201153e-06,
+      "loss": 0.4235,
+      "step": 16080
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.0477218627929688,
+      "learning_rate": 8.65574040271939e-06,
+      "loss": 0.7117,
+      "step": 16081
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.720285654067993,
+      "learning_rate": 8.655560525072385e-06,
+      "loss": 0.4573,
+      "step": 16082
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.520585536956787,
+      "learning_rate": 8.655380637260632e-06,
+      "loss": 0.5182,
+      "step": 16083
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.510925054550171,
+      "learning_rate": 8.655200739284637e-06,
+      "loss": 0.6568,
+      "step": 16084
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.753816604614258,
+      "learning_rate": 8.655020831144895e-06,
+      "loss": 0.5505,
+      "step": 16085
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.0492613315582275,
+      "learning_rate": 8.654840912841913e-06,
+      "loss": 0.7286,
+      "step": 16086
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.45154070854187,
+      "learning_rate": 8.654660984376186e-06,
+      "loss": 0.7815,
+      "step": 16087
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 6.58304500579834,
+      "learning_rate": 8.654481045748215e-06,
+      "loss": 0.9246,
+      "step": 16088
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.0462520122528076,
+      "learning_rate": 8.654301096958503e-06,
+      "loss": 0.9213,
+      "step": 16089
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.054982900619507,
+      "learning_rate": 8.654121138007547e-06,
+      "loss": 0.6828,
+      "step": 16090
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.7649595737457275,
+      "learning_rate": 8.653941168895851e-06,
+      "loss": 0.3311,
+      "step": 16091
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 7.8888444900512695,
+      "learning_rate": 8.653761189623912e-06,
+      "loss": 0.9212,
+      "step": 16092
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.3375205993652344,
+      "learning_rate": 8.653581200192232e-06,
+      "loss": 0.6231,
+      "step": 16093
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.152266502380371,
+      "learning_rate": 8.65340120060131e-06,
+      "loss": 0.4474,
+      "step": 16094
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.2617733478546143,
+      "learning_rate": 8.65322119085165e-06,
+      "loss": 0.7688,
+      "step": 16095
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.3776228427886963,
+      "learning_rate": 8.65304117094375e-06,
+      "loss": 0.6821,
+      "step": 16096
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.3980283737182617,
+      "learning_rate": 8.652861140878111e-06,
+      "loss": 0.4371,
+      "step": 16097
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.368135929107666,
+      "learning_rate": 8.652681100655233e-06,
+      "loss": 0.6393,
+      "step": 16098
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.871290922164917,
+      "learning_rate": 8.652501050275617e-06,
+      "loss": 0.7072,
+      "step": 16099
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.0809450149536133,
+      "learning_rate": 8.652320989739764e-06,
+      "loss": 0.3801,
+      "step": 16100
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.794127345085144,
+      "learning_rate": 8.652140919048175e-06,
+      "loss": 0.5889,
+      "step": 16101
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.2520534992218018,
+      "learning_rate": 8.65196083820135e-06,
+      "loss": 0.7073,
+      "step": 16102
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.8791284561157227,
+      "learning_rate": 8.65178074719979e-06,
+      "loss": 0.7115,
+      "step": 16103
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.9077190160751343,
+      "learning_rate": 8.651600646043996e-06,
+      "loss": 0.5619,
+      "step": 16104
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.9205642938613892,
+      "learning_rate": 8.651420534734467e-06,
+      "loss": 0.6547,
+      "step": 16105
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.4960471391677856,
+      "learning_rate": 8.651240413271707e-06,
+      "loss": 0.8779,
+      "step": 16106
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.1291096210479736,
+      "learning_rate": 8.651060281656214e-06,
+      "loss": 0.7046,
+      "step": 16107
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.391134262084961,
+      "learning_rate": 8.650880139888492e-06,
+      "loss": 0.6533,
+      "step": 16108
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.8730828762054443,
+      "learning_rate": 8.650699987969037e-06,
+      "loss": 0.6722,
+      "step": 16109
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.8764044046401978,
+      "learning_rate": 8.650519825898355e-06,
+      "loss": 0.4774,
+      "step": 16110
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 4.779319763183594,
+      "learning_rate": 8.650339653676944e-06,
+      "loss": 0.5066,
+      "step": 16111
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.67699933052063,
+      "learning_rate": 8.650159471305307e-06,
+      "loss": 0.5508,
+      "step": 16112
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.972639560699463,
+      "learning_rate": 8.649979278783945e-06,
+      "loss": 0.7257,
+      "step": 16113
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.4443275928497314,
+      "learning_rate": 8.649799076113354e-06,
+      "loss": 0.3794,
+      "step": 16114
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.386648178100586,
+      "learning_rate": 8.649618863294042e-06,
+      "loss": 0.8201,
+      "step": 16115
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.7263739109039307,
+      "learning_rate": 8.649438640326506e-06,
+      "loss": 0.7796,
+      "step": 16116
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.9917134046554565,
+      "learning_rate": 8.649258407211249e-06,
+      "loss": 0.7301,
+      "step": 16117
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.1173346042633057,
+      "learning_rate": 8.649078163948771e-06,
+      "loss": 0.5033,
+      "step": 16118
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.4699958562850952,
+      "learning_rate": 8.648897910539572e-06,
+      "loss": 0.6724,
+      "step": 16119
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.1036376953125,
+      "learning_rate": 8.648717646984157e-06,
+      "loss": 0.6811,
+      "step": 16120
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.9757905006408691,
+      "learning_rate": 8.648537373283025e-06,
+      "loss": 0.6483,
+      "step": 16121
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.917187213897705,
+      "learning_rate": 8.648357089436677e-06,
+      "loss": 0.591,
+      "step": 16122
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.6629198789596558,
+      "learning_rate": 8.648176795445616e-06,
+      "loss": 0.7871,
+      "step": 16123
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.5273756980895996,
+      "learning_rate": 8.64799649131034e-06,
+      "loss": 0.8032,
+      "step": 16124
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.7772071361541748,
+      "learning_rate": 8.647816177031352e-06,
+      "loss": 0.5008,
+      "step": 16125
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.192282199859619,
+      "learning_rate": 8.647635852609154e-06,
+      "loss": 0.4584,
+      "step": 16126
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.5351364612579346,
+      "learning_rate": 8.647455518044248e-06,
+      "loss": 0.555,
+      "step": 16127
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.9507204294204712,
+      "learning_rate": 8.647275173337133e-06,
+      "loss": 0.6427,
+      "step": 16128
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.282804012298584,
+      "learning_rate": 8.647094818488316e-06,
+      "loss": 0.9266,
+      "step": 16129
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.2358226776123047,
+      "learning_rate": 8.646914453498289e-06,
+      "loss": 0.6487,
+      "step": 16130
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.2692224979400635,
+      "learning_rate": 8.646734078367562e-06,
+      "loss": 0.8157,
+      "step": 16131
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.9224351644515991,
+      "learning_rate": 8.646553693096635e-06,
+      "loss": 0.7555,
+      "step": 16132
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.694247007369995,
+      "learning_rate": 8.646373297686006e-06,
+      "loss": 0.7426,
+      "step": 16133
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 5.185919284820557,
+      "learning_rate": 8.64619289213618e-06,
+      "loss": 0.4917,
+      "step": 16134
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.291470766067505,
+      "learning_rate": 8.646012476447657e-06,
+      "loss": 0.801,
+      "step": 16135
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.1371521949768066,
+      "learning_rate": 8.645832050620939e-06,
+      "loss": 0.6112,
+      "step": 16136
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.4113810062408447,
+      "learning_rate": 8.645651614656528e-06,
+      "loss": 0.464,
+      "step": 16137
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.3057861328125,
+      "learning_rate": 8.645471168554926e-06,
+      "loss": 0.74,
+      "step": 16138
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.6162867546081543,
+      "learning_rate": 8.645290712316632e-06,
+      "loss": 0.4667,
+      "step": 16139
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.0750362873077393,
+      "learning_rate": 8.645110245942152e-06,
+      "loss": 0.4792,
+      "step": 16140
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 8.443188667297363,
+      "learning_rate": 8.644929769431987e-06,
+      "loss": 0.7733,
+      "step": 16141
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 5.299589157104492,
+      "learning_rate": 8.644749282786636e-06,
+      "loss": 0.5602,
+      "step": 16142
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 5.725451469421387,
+      "learning_rate": 8.644568786006603e-06,
+      "loss": 0.7225,
+      "step": 16143
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.540771722793579,
+      "learning_rate": 8.64438827909239e-06,
+      "loss": 0.74,
+      "step": 16144
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.6983505487442017,
+      "learning_rate": 8.644207762044497e-06,
+      "loss": 0.9353,
+      "step": 16145
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.5497095584869385,
+      "learning_rate": 8.64402723486343e-06,
+      "loss": 0.7284,
+      "step": 16146
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.6653721332550049,
+      "learning_rate": 8.643846697549685e-06,
+      "loss": 0.7084,
+      "step": 16147
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.7231285572052,
+      "learning_rate": 8.64366615010377e-06,
+      "loss": 0.6692,
+      "step": 16148
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.5376827716827393,
+      "learning_rate": 8.643485592526184e-06,
+      "loss": 0.4866,
+      "step": 16149
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.138261079788208,
+      "learning_rate": 8.643305024817428e-06,
+      "loss": 0.7003,
+      "step": 16150
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.757441997528076,
+      "learning_rate": 8.643124446978007e-06,
+      "loss": 0.4685,
+      "step": 16151
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.1213178634643555,
+      "learning_rate": 8.642943859008422e-06,
+      "loss": 0.7688,
+      "step": 16152
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.8731191158294678,
+      "learning_rate": 8.642763260909172e-06,
+      "loss": 0.8775,
+      "step": 16153
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.1283819675445557,
+      "learning_rate": 8.642582652680765e-06,
+      "loss": 0.6878,
+      "step": 16154
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.9579006433486938,
+      "learning_rate": 8.642402034323699e-06,
+      "loss": 0.6427,
+      "step": 16155
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.314542055130005,
+      "learning_rate": 8.642221405838477e-06,
+      "loss": 0.6659,
+      "step": 16156
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.283475160598755,
+      "learning_rate": 8.642040767225602e-06,
+      "loss": 0.3379,
+      "step": 16157
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.152761459350586,
+      "learning_rate": 8.641860118485576e-06,
+      "loss": 0.7329,
+      "step": 16158
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.8749070167541504,
+      "learning_rate": 8.641679459618901e-06,
+      "loss": 0.7299,
+      "step": 16159
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.9217849969863892,
+      "learning_rate": 8.641498790626082e-06,
+      "loss": 0.7107,
+      "step": 16160
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.623284101486206,
+      "learning_rate": 8.641318111507616e-06,
+      "loss": 0.5753,
+      "step": 16161
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.164506673812866,
+      "learning_rate": 8.64113742226401e-06,
+      "loss": 0.7858,
+      "step": 16162
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.6185641288757324,
+      "learning_rate": 8.640956722895763e-06,
+      "loss": 0.6629,
+      "step": 16163
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.191575288772583,
+      "learning_rate": 8.640776013403381e-06,
+      "loss": 0.7804,
+      "step": 16164
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.4784483909606934,
+      "learning_rate": 8.640595293787363e-06,
+      "loss": 0.7848,
+      "step": 16165
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.056894063949585,
+      "learning_rate": 8.640414564048216e-06,
+      "loss": 0.6166,
+      "step": 16166
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.0185599327087402,
+      "learning_rate": 8.640233824186438e-06,
+      "loss": 0.5942,
+      "step": 16167
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.6096854209899902,
+      "learning_rate": 8.640053074202534e-06,
+      "loss": 0.8773,
+      "step": 16168
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.2795913219451904,
+      "learning_rate": 8.639872314097007e-06,
+      "loss": 0.9756,
+      "step": 16169
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.9362525939941406,
+      "learning_rate": 8.639691543870357e-06,
+      "loss": 0.7363,
+      "step": 16170
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.7799805402755737,
+      "learning_rate": 8.63951076352309e-06,
+      "loss": 0.8061,
+      "step": 16171
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.958794355392456,
+      "learning_rate": 8.639329973055703e-06,
+      "loss": 0.751,
+      "step": 16172
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.6903085708618164,
+      "learning_rate": 8.639149172468706e-06,
+      "loss": 0.6263,
+      "step": 16173
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.813350796699524,
+      "learning_rate": 8.638968361762599e-06,
+      "loss": 0.6848,
+      "step": 16174
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.028606414794922,
+      "learning_rate": 8.638787540937882e-06,
+      "loss": 0.7849,
+      "step": 16175
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.1431784629821777,
+      "learning_rate": 8.638606709995062e-06,
+      "loss": 0.8152,
+      "step": 16176
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.756885290145874,
+      "learning_rate": 8.638425868934636e-06,
+      "loss": 0.7904,
+      "step": 16177
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.89639949798584,
+      "learning_rate": 8.638245017757113e-06,
+      "loss": 0.5925,
+      "step": 16178
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.5109877586364746,
+      "learning_rate": 8.638064156462995e-06,
+      "loss": 0.7485,
+      "step": 16179
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.082754135131836,
+      "learning_rate": 8.637883285052783e-06,
+      "loss": 0.5619,
+      "step": 16180
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.6094043254852295,
+      "learning_rate": 8.63770240352698e-06,
+      "loss": 0.9542,
+      "step": 16181
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.897690534591675,
+      "learning_rate": 8.637521511886088e-06,
+      "loss": 0.7812,
+      "step": 16182
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.212480306625366,
+      "learning_rate": 8.637340610130614e-06,
+      "loss": 0.7257,
+      "step": 16183
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.52931809425354,
+      "learning_rate": 8.637159698261055e-06,
+      "loss": 0.5204,
+      "step": 16184
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.7428529262542725,
+      "learning_rate": 8.63697877627792e-06,
+      "loss": 0.5094,
+      "step": 16185
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 4.563608169555664,
+      "learning_rate": 8.636797844181707e-06,
+      "loss": 0.6488,
+      "step": 16186
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.2529478073120117,
+      "learning_rate": 8.636616901972924e-06,
+      "loss": 0.4047,
+      "step": 16187
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.3068442344665527,
+      "learning_rate": 8.636435949652071e-06,
+      "loss": 0.5988,
+      "step": 16188
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.3280386924743652,
+      "learning_rate": 8.636254987219653e-06,
+      "loss": 0.5739,
+      "step": 16189
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.6830782890319824,
+      "learning_rate": 8.63607401467617e-06,
+      "loss": 0.5722,
+      "step": 16190
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.0789883136749268,
+      "learning_rate": 8.635893032022127e-06,
+      "loss": 0.988,
+      "step": 16191
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.157683849334717,
+      "learning_rate": 8.635712039258028e-06,
+      "loss": 0.7043,
+      "step": 16192
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.539621114730835,
+      "learning_rate": 8.635531036384377e-06,
+      "loss": 0.9925,
+      "step": 16193
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 5.437475681304932,
+      "learning_rate": 8.635350023401676e-06,
+      "loss": 0.8343,
+      "step": 16194
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.820988178253174,
+      "learning_rate": 8.635169000310428e-06,
+      "loss": 0.5298,
+      "step": 16195
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.77592134475708,
+      "learning_rate": 8.634987967111135e-06,
+      "loss": 0.8921,
+      "step": 16196
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.5897557735443115,
+      "learning_rate": 8.634806923804304e-06,
+      "loss": 0.9682,
+      "step": 16197
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.600017786026001,
+      "learning_rate": 8.634625870390437e-06,
+      "loss": 0.8081,
+      "step": 16198
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.2120673656463623,
+      "learning_rate": 8.634444806870037e-06,
+      "loss": 0.526,
+      "step": 16199
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.068164110183716,
+      "learning_rate": 8.634263733243607e-06,
+      "loss": 0.5967,
+      "step": 16200
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.0810728073120117,
+      "learning_rate": 8.63408264951165e-06,
+      "loss": 0.5804,
+      "step": 16201
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.215986728668213,
+      "learning_rate": 8.63390155567467e-06,
+      "loss": 0.6169,
+      "step": 16202
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.9868603944778442,
+      "learning_rate": 8.633720451733174e-06,
+      "loss": 0.8486,
+      "step": 16203
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.383002758026123,
+      "learning_rate": 8.633539337687661e-06,
+      "loss": 0.8098,
+      "step": 16204
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.6132279634475708,
+      "learning_rate": 8.633358213538637e-06,
+      "loss": 0.3608,
+      "step": 16205
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.0163168907165527,
+      "learning_rate": 8.633177079286603e-06,
+      "loss": 0.5295,
+      "step": 16206
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.084540605545044,
+      "learning_rate": 8.632995934932065e-06,
+      "loss": 0.8361,
+      "step": 16207
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.795492649078369,
+      "learning_rate": 8.632814780475527e-06,
+      "loss": 0.7306,
+      "step": 16208
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.7082176208496094,
+      "learning_rate": 8.632633615917493e-06,
+      "loss": 0.3849,
+      "step": 16209
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.900362968444824,
+      "learning_rate": 8.632452441258464e-06,
+      "loss": 0.8488,
+      "step": 16210
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.2343904972076416,
+      "learning_rate": 8.632271256498947e-06,
+      "loss": 0.734,
+      "step": 16211
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.6925524473190308,
+      "learning_rate": 8.632090061639442e-06,
+      "loss": 0.8296,
+      "step": 16212
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.1801793575286865,
+      "learning_rate": 8.631908856680457e-06,
+      "loss": 0.5834,
+      "step": 16213
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.221768856048584,
+      "learning_rate": 8.631727641622493e-06,
+      "loss": 0.6209,
+      "step": 16214
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.7706899642944336,
+      "learning_rate": 8.631546416466057e-06,
+      "loss": 0.5143,
+      "step": 16215
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.2002599239349365,
+      "learning_rate": 8.63136518121165e-06,
+      "loss": 0.8953,
+      "step": 16216
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.783549427986145,
+      "learning_rate": 8.631183935859774e-06,
+      "loss": 0.6941,
+      "step": 16217
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.7671151161193848,
+      "learning_rate": 8.63100268041094e-06,
+      "loss": 0.8802,
+      "step": 16218
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.3149070739746094,
+      "learning_rate": 8.630821414865644e-06,
+      "loss": 0.596,
+      "step": 16219
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.8141546249389648,
+      "learning_rate": 8.630640139224396e-06,
+      "loss": 0.5499,
+      "step": 16220
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.1446919441223145,
+      "learning_rate": 8.630458853487697e-06,
+      "loss": 0.5402,
+      "step": 16221
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.945732831954956,
+      "learning_rate": 8.630277557656053e-06,
+      "loss": 0.6724,
+      "step": 16222
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.2951505184173584,
+      "learning_rate": 8.630096251729967e-06,
+      "loss": 0.7587,
+      "step": 16223
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.9662423133850098,
+      "learning_rate": 8.629914935709941e-06,
+      "loss": 0.7936,
+      "step": 16224
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.9391695261001587,
+      "learning_rate": 8.629733609596485e-06,
+      "loss": 0.775,
+      "step": 16225
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.0197479724884033,
+      "learning_rate": 8.629552273390095e-06,
+      "loss": 0.9011,
+      "step": 16226
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.7447855472564697,
+      "learning_rate": 8.629370927091283e-06,
+      "loss": 0.4402,
+      "step": 16227
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.303901195526123,
+      "learning_rate": 8.62918957070055e-06,
+      "loss": 0.563,
+      "step": 16228
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.4054934978485107,
+      "learning_rate": 8.629008204218399e-06,
+      "loss": 0.6328,
+      "step": 16229
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.587818145751953,
+      "learning_rate": 8.628826827645336e-06,
+      "loss": 1.0159,
+      "step": 16230
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.6064608097076416,
+      "learning_rate": 8.628645440981867e-06,
+      "loss": 0.7144,
+      "step": 16231
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.2275640964508057,
+      "learning_rate": 8.628464044228491e-06,
+      "loss": 0.7799,
+      "step": 16232
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.569833993911743,
+      "learning_rate": 8.628282637385717e-06,
+      "loss": 0.7554,
+      "step": 16233
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.196225881576538,
+      "learning_rate": 8.628101220454049e-06,
+      "loss": 0.6873,
+      "step": 16234
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.9128835201263428,
+      "learning_rate": 8.62791979343399e-06,
+      "loss": 0.7383,
+      "step": 16235
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.9857065677642822,
+      "learning_rate": 8.627738356326045e-06,
+      "loss": 0.4763,
+      "step": 16236
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.813171625137329,
+      "learning_rate": 8.627556909130718e-06,
+      "loss": 0.6686,
+      "step": 16237
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.8390932083129883,
+      "learning_rate": 8.627375451848516e-06,
+      "loss": 0.6961,
+      "step": 16238
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.03847599029541,
+      "learning_rate": 8.62719398447994e-06,
+      "loss": 0.8417,
+      "step": 16239
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.3381383419036865,
+      "learning_rate": 8.627012507025498e-06,
+      "loss": 0.5204,
+      "step": 16240
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.928095817565918,
+      "learning_rate": 8.626831019485691e-06,
+      "loss": 0.8396,
+      "step": 16241
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.409494161605835,
+      "learning_rate": 8.626649521861026e-06,
+      "loss": 0.5046,
+      "step": 16242
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.5572330951690674,
+      "learning_rate": 8.626468014152008e-06,
+      "loss": 0.8066,
+      "step": 16243
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.2228121757507324,
+      "learning_rate": 8.626286496359141e-06,
+      "loss": 0.4392,
+      "step": 16244
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.6801044940948486,
+      "learning_rate": 8.62610496848293e-06,
+      "loss": 0.5363,
+      "step": 16245
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.602524995803833,
+      "learning_rate": 8.625923430523878e-06,
+      "loss": 0.6062,
+      "step": 16246
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.147021532058716,
+      "learning_rate": 8.625741882482492e-06,
+      "loss": 0.501,
+      "step": 16247
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.6798551082611084,
+      "learning_rate": 8.625560324359276e-06,
+      "loss": 0.8842,
+      "step": 16248
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.8869527578353882,
+      "learning_rate": 8.625378756154735e-06,
+      "loss": 0.6312,
+      "step": 16249
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.644260883331299,
+      "learning_rate": 8.625197177869374e-06,
+      "loss": 0.8645,
+      "step": 16250
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.34549617767334,
+      "learning_rate": 8.625015589503698e-06,
+      "loss": 0.6946,
+      "step": 16251
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.6055419445037842,
+      "learning_rate": 8.62483399105821e-06,
+      "loss": 0.7906,
+      "step": 16252
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.606945753097534,
+      "learning_rate": 8.624652382533419e-06,
+      "loss": 0.7216,
+      "step": 16253
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.1373727321624756,
+      "learning_rate": 8.624470763929825e-06,
+      "loss": 0.6637,
+      "step": 16254
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.31376576423645,
+      "learning_rate": 8.624289135247937e-06,
+      "loss": 0.6038,
+      "step": 16255
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.146221876144409,
+      "learning_rate": 8.624107496488259e-06,
+      "loss": 0.9184,
+      "step": 16256
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.1066699028015137,
+      "learning_rate": 8.623925847651294e-06,
+      "loss": 0.3236,
+      "step": 16257
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.182645320892334,
+      "learning_rate": 8.62374418873755e-06,
+      "loss": 0.3651,
+      "step": 16258
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.879166841506958,
+      "learning_rate": 8.62356251974753e-06,
+      "loss": 0.9277,
+      "step": 16259
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.9639713764190674,
+      "learning_rate": 8.623380840681742e-06,
+      "loss": 0.5334,
+      "step": 16260
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.3301000595092773,
+      "learning_rate": 8.623199151540689e-06,
+      "loss": 0.615,
+      "step": 16261
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 4.655129432678223,
+      "learning_rate": 8.623017452324874e-06,
+      "loss": 0.7327,
+      "step": 16262
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.1237452030181885,
+      "learning_rate": 8.622835743034807e-06,
+      "loss": 0.726,
+      "step": 16263
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.5999786853790283,
+      "learning_rate": 8.622654023670988e-06,
+      "loss": 0.7737,
+      "step": 16264
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.521838903427124,
+      "learning_rate": 8.622472294233929e-06,
+      "loss": 0.7803,
+      "step": 16265
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.5570404529571533,
+      "learning_rate": 8.622290554724128e-06,
+      "loss": 0.7563,
+      "step": 16266
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.60198712348938,
+      "learning_rate": 8.622108805142098e-06,
+      "loss": 0.615,
+      "step": 16267
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.2220587730407715,
+      "learning_rate": 8.621927045488338e-06,
+      "loss": 0.4559,
+      "step": 16268
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.7459356784820557,
+      "learning_rate": 8.621745275763355e-06,
+      "loss": 0.6427,
+      "step": 16269
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 4.590092658996582,
+      "learning_rate": 8.621563495967655e-06,
+      "loss": 0.6349,
+      "step": 16270
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.4631872177124023,
+      "learning_rate": 8.621381706101745e-06,
+      "loss": 0.4605,
+      "step": 16271
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.375753164291382,
+      "learning_rate": 8.621199906166129e-06,
+      "loss": 0.6342,
+      "step": 16272
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.8033223152160645,
+      "learning_rate": 8.621018096161311e-06,
+      "loss": 0.622,
+      "step": 16273
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.4679272174835205,
+      "learning_rate": 8.620836276087801e-06,
+      "loss": 0.7355,
+      "step": 16274
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.60115647315979,
+      "learning_rate": 8.620654445946099e-06,
+      "loss": 0.7568,
+      "step": 16275
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.4539682865142822,
+      "learning_rate": 8.620472605736713e-06,
+      "loss": 0.6976,
+      "step": 16276
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.7057788372039795,
+      "learning_rate": 8.620290755460153e-06,
+      "loss": 0.7324,
+      "step": 16277
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.10756516456604,
+      "learning_rate": 8.620108895116916e-06,
+      "loss": 0.8539,
+      "step": 16278
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 3.906029224395752,
+      "learning_rate": 8.619927024707515e-06,
+      "loss": 0.6201,
+      "step": 16279
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.383094072341919,
+      "learning_rate": 8.619745144232451e-06,
+      "loss": 1.0022,
+      "step": 16280
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.0562691688537598,
+      "learning_rate": 8.619563253692234e-06,
+      "loss": 0.5343,
+      "step": 16281
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.7966426610946655,
+      "learning_rate": 8.619381353087367e-06,
+      "loss": 0.8713,
+      "step": 16282
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.7131222486495972,
+      "learning_rate": 8.619199442418354e-06,
+      "loss": 0.6845,
+      "step": 16283
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.5067557096481323,
+      "learning_rate": 8.619017521685705e-06,
+      "loss": 0.4599,
+      "step": 16284
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.6787197589874268,
+      "learning_rate": 8.618835590889925e-06,
+      "loss": 0.7162,
+      "step": 16285
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.840275764465332,
+      "learning_rate": 8.618653650031516e-06,
+      "loss": 0.5383,
+      "step": 16286
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.4401090145111084,
+      "learning_rate": 8.618471699110989e-06,
+      "loss": 0.8227,
+      "step": 16287
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.810023307800293,
+      "learning_rate": 8.618289738128847e-06,
+      "loss": 0.9274,
+      "step": 16288
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 5.7020649909973145,
+      "learning_rate": 8.618107767085597e-06,
+      "loss": 0.8923,
+      "step": 16289
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.1826188564300537,
+      "learning_rate": 8.617925785981744e-06,
+      "loss": 0.6878,
+      "step": 16290
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.807215452194214,
+      "learning_rate": 8.617743794817796e-06,
+      "loss": 0.7555,
+      "step": 16291
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.061197280883789,
+      "learning_rate": 8.617561793594257e-06,
+      "loss": 0.6648,
+      "step": 16292
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.69848895072937,
+      "learning_rate": 8.617379782311635e-06,
+      "loss": 0.8232,
+      "step": 16293
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.3995985984802246,
+      "learning_rate": 8.617197760970432e-06,
+      "loss": 0.6944,
+      "step": 16294
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 3.01918625831604,
+      "learning_rate": 8.617015729571158e-06,
+      "loss": 0.7728,
+      "step": 16295
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.0843186378479004,
+      "learning_rate": 8.61683368811432e-06,
+      "loss": 0.6178,
+      "step": 16296
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.2855594158172607,
+      "learning_rate": 8.616651636600421e-06,
+      "loss": 0.6411,
+      "step": 16297
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.7043826580047607,
+      "learning_rate": 8.61646957502997e-06,
+      "loss": 0.6685,
+      "step": 16298
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 3.6447253227233887,
+      "learning_rate": 8.616287503403472e-06,
+      "loss": 0.7338,
+      "step": 16299
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 3.5367178916931152,
+      "learning_rate": 8.61610542172143e-06,
+      "loss": 0.8661,
+      "step": 16300
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.161999225616455,
+      "learning_rate": 8.615923329984356e-06,
+      "loss": 0.4309,
+      "step": 16301
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.8243000507354736,
+      "learning_rate": 8.615741228192753e-06,
+      "loss": 0.5338,
+      "step": 16302
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.114001512527466,
+      "learning_rate": 8.615559116347128e-06,
+      "loss": 0.5259,
+      "step": 16303
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 4.408656120300293,
+      "learning_rate": 8.615376994447988e-06,
+      "loss": 0.6205,
+      "step": 16304
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.414918899536133,
+      "learning_rate": 8.615194862495839e-06,
+      "loss": 0.8059,
+      "step": 16305
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 4.110039234161377,
+      "learning_rate": 8.615012720491188e-06,
+      "loss": 0.7836,
+      "step": 16306
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.3448338508605957,
+      "learning_rate": 8.61483056843454e-06,
+      "loss": 0.5839,
+      "step": 16307
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.716017723083496,
+      "learning_rate": 8.614648406326403e-06,
+      "loss": 0.6349,
+      "step": 16308
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 3.593010187149048,
+      "learning_rate": 8.61446623416728e-06,
+      "loss": 0.6166,
+      "step": 16309
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 3.1768455505371094,
+      "learning_rate": 8.614284051957685e-06,
+      "loss": 0.671,
+      "step": 16310
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.041308879852295,
+      "learning_rate": 8.614101859698116e-06,
+      "loss": 0.5571,
+      "step": 16311
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.037773609161377,
+      "learning_rate": 8.613919657389087e-06,
+      "loss": 0.6589,
+      "step": 16312
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.643993377685547,
+      "learning_rate": 8.613737445031098e-06,
+      "loss": 0.8317,
+      "step": 16313
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.477283477783203,
+      "learning_rate": 8.613555222624662e-06,
+      "loss": 0.6466,
+      "step": 16314
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.738189935684204,
+      "learning_rate": 8.613372990170278e-06,
+      "loss": 0.6393,
+      "step": 16315
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.1000874042510986,
+      "learning_rate": 8.613190747668461e-06,
+      "loss": 0.8696,
+      "step": 16316
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.0105173587799072,
+      "learning_rate": 8.613008495119715e-06,
+      "loss": 0.4771,
+      "step": 16317
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.1626346111297607,
+      "learning_rate": 8.612826232524543e-06,
+      "loss": 0.3782,
+      "step": 16318
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.9671305418014526,
+      "learning_rate": 8.612643959883456e-06,
+      "loss": 0.7525,
+      "step": 16319
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.9560573101043701,
+      "learning_rate": 8.612461677196959e-06,
+      "loss": 0.4951,
+      "step": 16320
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.8981729745864868,
+      "learning_rate": 8.612279384465561e-06,
+      "loss": 0.4085,
+      "step": 16321
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.9707400798797607,
+      "learning_rate": 8.612097081689765e-06,
+      "loss": 0.7807,
+      "step": 16322
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 3.115736722946167,
+      "learning_rate": 8.611914768870081e-06,
+      "loss": 0.7026,
+      "step": 16323
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.3658792972564697,
+      "learning_rate": 8.611732446007015e-06,
+      "loss": 0.915,
+      "step": 16324
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.441504716873169,
+      "learning_rate": 8.611550113101074e-06,
+      "loss": 0.6354,
+      "step": 16325
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.191929578781128,
+      "learning_rate": 8.611367770152766e-06,
+      "loss": 0.7965,
+      "step": 16326
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 3.829136371612549,
+      "learning_rate": 8.611185417162595e-06,
+      "loss": 0.7254,
+      "step": 16327
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.045290470123291,
+      "learning_rate": 8.611003054131073e-06,
+      "loss": 0.9314,
+      "step": 16328
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.7431014776229858,
+      "learning_rate": 8.610820681058702e-06,
+      "loss": 0.4981,
+      "step": 16329
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.9803500175476074,
+      "learning_rate": 8.610638297945993e-06,
+      "loss": 0.5127,
+      "step": 16330
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.9550538063049316,
+      "learning_rate": 8.610455904793449e-06,
+      "loss": 0.635,
+      "step": 16331
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.694301962852478,
+      "learning_rate": 8.610273501601582e-06,
+      "loss": 0.8255,
+      "step": 16332
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.0121755599975586,
+      "learning_rate": 8.610091088370896e-06,
+      "loss": 0.6115,
+      "step": 16333
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.0509777069091797,
+      "learning_rate": 8.609908665101899e-06,
+      "loss": 0.4499,
+      "step": 16334
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.34462571144104,
+      "learning_rate": 8.609726231795097e-06,
+      "loss": 0.7946,
+      "step": 16335
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 3.340757369995117,
+      "learning_rate": 8.609543788451e-06,
+      "loss": 0.4936,
+      "step": 16336
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 27.196779251098633,
+      "learning_rate": 8.609361335070114e-06,
+      "loss": 0.5839,
+      "step": 16337
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 4.246116638183594,
+      "learning_rate": 8.609178871652946e-06,
+      "loss": 0.6769,
+      "step": 16338
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 4.906302452087402,
+      "learning_rate": 8.608996398200003e-06,
+      "loss": 0.8554,
+      "step": 16339
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.6764333248138428,
+      "learning_rate": 8.608813914711793e-06,
+      "loss": 0.7499,
+      "step": 16340
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.5371372699737549,
+      "learning_rate": 8.608631421188824e-06,
+      "loss": 0.6898,
+      "step": 16341
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.034489631652832,
+      "learning_rate": 8.608448917631604e-06,
+      "loss": 0.6687,
+      "step": 16342
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.822180986404419,
+      "learning_rate": 8.608266404040637e-06,
+      "loss": 0.4683,
+      "step": 16343
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 4.237945079803467,
+      "learning_rate": 8.608083880416435e-06,
+      "loss": 0.8417,
+      "step": 16344
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.9615129232406616,
+      "learning_rate": 8.607901346759502e-06,
+      "loss": 0.7303,
+      "step": 16345
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.4672930240631104,
+      "learning_rate": 8.607718803070348e-06,
+      "loss": 0.5714,
+      "step": 16346
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.8333014249801636,
+      "learning_rate": 8.607536249349478e-06,
+      "loss": 0.7392,
+      "step": 16347
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.8366667032241821,
+      "learning_rate": 8.607353685597402e-06,
+      "loss": 0.605,
+      "step": 16348
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.02097487449646,
+      "learning_rate": 8.607171111814626e-06,
+      "loss": 0.772,
+      "step": 16349
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.1111345291137695,
+      "learning_rate": 8.606988528001658e-06,
+      "loss": 0.3529,
+      "step": 16350
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.4592015743255615,
+      "learning_rate": 8.606805934159008e-06,
+      "loss": 0.71,
+      "step": 16351
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.040454864501953,
+      "learning_rate": 8.60662333028718e-06,
+      "loss": 0.625,
+      "step": 16352
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.9323418140411377,
+      "learning_rate": 8.606440716386686e-06,
+      "loss": 0.8039,
+      "step": 16353
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 3.7834725379943848,
+      "learning_rate": 8.606258092458028e-06,
+      "loss": 1.0565,
+      "step": 16354
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.0440781116485596,
+      "learning_rate": 8.60607545850172e-06,
+      "loss": 0.7669,
+      "step": 16355
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.4184207916259766,
+      "learning_rate": 8.605892814518266e-06,
+      "loss": 0.8163,
+      "step": 16356
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.6857938766479492,
+      "learning_rate": 8.605710160508175e-06,
+      "loss": 0.721,
+      "step": 16357
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.6683669090270996,
+      "learning_rate": 8.605527496471955e-06,
+      "loss": 0.7369,
+      "step": 16358
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.009953498840332,
+      "learning_rate": 8.605344822410114e-06,
+      "loss": 0.7978,
+      "step": 16359
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.7909362316131592,
+      "learning_rate": 8.605162138323158e-06,
+      "loss": 0.6128,
+      "step": 16360
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.1415305137634277,
+      "learning_rate": 8.604979444211598e-06,
+      "loss": 0.5804,
+      "step": 16361
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 3.0640833377838135,
+      "learning_rate": 8.60479674007594e-06,
+      "loss": 0.6692,
+      "step": 16362
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.949345588684082,
+      "learning_rate": 8.604614025916693e-06,
+      "loss": 0.7437,
+      "step": 16363
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.750899314880371,
+      "learning_rate": 8.604431301734364e-06,
+      "loss": 0.4942,
+      "step": 16364
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.758450746536255,
+      "learning_rate": 8.604248567529464e-06,
+      "loss": 0.8515,
+      "step": 16365
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 3.5095396041870117,
+      "learning_rate": 8.604065823302496e-06,
+      "loss": 0.8285,
+      "step": 16366
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.542848825454712,
+      "learning_rate": 8.603883069053974e-06,
+      "loss": 0.6052,
+      "step": 16367
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.36803936958313,
+      "learning_rate": 8.6037003047844e-06,
+      "loss": 0.78,
+      "step": 16368
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 3.018768072128296,
+      "learning_rate": 8.603517530494288e-06,
+      "loss": 0.9658,
+      "step": 16369
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.9603132009506226,
+      "learning_rate": 8.603334746184141e-06,
+      "loss": 0.6798,
+      "step": 16370
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.743401288986206,
+      "learning_rate": 8.603151951854472e-06,
+      "loss": 0.6583,
+      "step": 16371
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.5832347869873047,
+      "learning_rate": 8.602969147505788e-06,
+      "loss": 0.4654,
+      "step": 16372
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.6333733797073364,
+      "learning_rate": 8.602786333138595e-06,
+      "loss": 0.6964,
+      "step": 16373
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.719516634941101,
+      "learning_rate": 8.602603508753402e-06,
+      "loss": 0.7217,
+      "step": 16374
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.5604366064071655,
+      "learning_rate": 8.60242067435072e-06,
+      "loss": 0.7129,
+      "step": 16375
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.5607898235321045,
+      "learning_rate": 8.602237829931056e-06,
+      "loss": 0.7437,
+      "step": 16376
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 4.430799961090088,
+      "learning_rate": 8.602054975494915e-06,
+      "loss": 0.7885,
+      "step": 16377
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.647379994392395,
+      "learning_rate": 8.601872111042812e-06,
+      "loss": 0.5989,
+      "step": 16378
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.901464819908142,
+      "learning_rate": 8.601689236575251e-06,
+      "loss": 0.7831,
+      "step": 16379
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 6.809023857116699,
+      "learning_rate": 8.60150635209274e-06,
+      "loss": 0.7166,
+      "step": 16380
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.2854076623916626,
+      "learning_rate": 8.601323457595789e-06,
+      "loss": 0.5581,
+      "step": 16381
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.3762292861938477,
+      "learning_rate": 8.601140553084907e-06,
+      "loss": 0.5248,
+      "step": 16382
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.0924465656280518,
+      "learning_rate": 8.600957638560604e-06,
+      "loss": 0.5611,
+      "step": 16383
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.29647159576416,
+      "learning_rate": 8.600774714023385e-06,
+      "loss": 0.534,
+      "step": 16384
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.9042255878448486,
+      "learning_rate": 8.600591779473761e-06,
+      "loss": 0.7175,
+      "step": 16385
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.1760332584381104,
+      "learning_rate": 8.60040883491224e-06,
+      "loss": 0.6042,
+      "step": 16386
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.32859468460083,
+      "learning_rate": 8.60022588033933e-06,
+      "loss": 0.6104,
+      "step": 16387
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 3.725825548171997,
+      "learning_rate": 8.600042915755541e-06,
+      "loss": 0.7946,
+      "step": 16388
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.9546575546264648,
+      "learning_rate": 8.59985994116138e-06,
+      "loss": 0.546,
+      "step": 16389
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.818493127822876,
+      "learning_rate": 8.59967695655736e-06,
+      "loss": 0.634,
+      "step": 16390
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.783322811126709,
+      "learning_rate": 8.599493961943985e-06,
+      "loss": 0.8681,
+      "step": 16391
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 5.436459064483643,
+      "learning_rate": 8.599310957321766e-06,
+      "loss": 0.8488,
+      "step": 16392
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.1836886405944824,
+      "learning_rate": 8.599127942691211e-06,
+      "loss": 0.7221,
+      "step": 16393
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 3.11017107963562,
+      "learning_rate": 8.59894491805283e-06,
+      "loss": 0.751,
+      "step": 16394
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.1655592918395996,
+      "learning_rate": 8.598761883407132e-06,
+      "loss": 0.4251,
+      "step": 16395
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.146458148956299,
+      "learning_rate": 8.598578838754624e-06,
+      "loss": 0.8011,
+      "step": 16396
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.534138560295105,
+      "learning_rate": 8.598395784095816e-06,
+      "loss": 0.6675,
+      "step": 16397
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.3948824405670166,
+      "learning_rate": 8.59821271943122e-06,
+      "loss": 0.7034,
+      "step": 16398
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.534989595413208,
+      "learning_rate": 8.59802964476134e-06,
+      "loss": 0.413,
+      "step": 16399
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 3.577253818511963,
+      "learning_rate": 8.597846560086687e-06,
+      "loss": 0.7165,
+      "step": 16400
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 3.6822187900543213,
+      "learning_rate": 8.597663465407773e-06,
+      "loss": 0.7411,
+      "step": 16401
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 3.9353952407836914,
+      "learning_rate": 8.597480360725104e-06,
+      "loss": 1.0179,
+      "step": 16402
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.0003767013549805,
+      "learning_rate": 8.597297246039187e-06,
+      "loss": 0.4988,
+      "step": 16403
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.2993221282958984,
+      "learning_rate": 8.597114121350535e-06,
+      "loss": 0.8059,
+      "step": 16404
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 3.152008533477783,
+      "learning_rate": 8.596930986659658e-06,
+      "loss": 1.0294,
+      "step": 16405
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 3.0176169872283936,
+      "learning_rate": 8.596747841967062e-06,
+      "loss": 0.605,
+      "step": 16406
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.6657873392105103,
+      "learning_rate": 8.596564687273258e-06,
+      "loss": 0.6131,
+      "step": 16407
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.96964168548584,
+      "learning_rate": 8.596381522578757e-06,
+      "loss": 0.6822,
+      "step": 16408
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 3.223442792892456,
+      "learning_rate": 8.596198347884064e-06,
+      "loss": 0.6363,
+      "step": 16409
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.274996519088745,
+      "learning_rate": 8.59601516318969e-06,
+      "loss": 0.8085,
+      "step": 16410
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.524555206298828,
+      "learning_rate": 8.595831968496145e-06,
+      "loss": 0.8412,
+      "step": 16411
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 3.0598626136779785,
+      "learning_rate": 8.59564876380394e-06,
+      "loss": 0.4625,
+      "step": 16412
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.47770357131958,
+      "learning_rate": 8.59546554911358e-06,
+      "loss": 0.8218,
+      "step": 16413
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.201582193374634,
+      "learning_rate": 8.59528232442558e-06,
+      "loss": 0.6863,
+      "step": 16414
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.157863140106201,
+      "learning_rate": 8.595099089740445e-06,
+      "loss": 0.762,
+      "step": 16415
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.5554864406585693,
+      "learning_rate": 8.594915845058687e-06,
+      "loss": 0.6065,
+      "step": 16416
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 3.296396017074585,
+      "learning_rate": 8.594732590380814e-06,
+      "loss": 0.8378,
+      "step": 16417
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 3.9245007038116455,
+      "learning_rate": 8.594549325707337e-06,
+      "loss": 0.601,
+      "step": 16418
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.1272411346435547,
+      "learning_rate": 8.594366051038765e-06,
+      "loss": 0.5162,
+      "step": 16419
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 3.081456422805786,
+      "learning_rate": 8.594182766375606e-06,
+      "loss": 0.864,
+      "step": 16420
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.6279103755950928,
+      "learning_rate": 8.593999471718373e-06,
+      "loss": 0.6345,
+      "step": 16421
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.93896746635437,
+      "learning_rate": 8.593816167067572e-06,
+      "loss": 0.5929,
+      "step": 16422
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.5925402641296387,
+      "learning_rate": 8.593632852423717e-06,
+      "loss": 0.5099,
+      "step": 16423
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.9438250064849854,
+      "learning_rate": 8.593449527787312e-06,
+      "loss": 0.7069,
+      "step": 16424
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.6456197500228882,
+      "learning_rate": 8.593266193158874e-06,
+      "loss": 0.8868,
+      "step": 16425
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.8761926889419556,
+      "learning_rate": 8.593082848538906e-06,
+      "loss": 0.6424,
+      "step": 16426
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.3482147455215454,
+      "learning_rate": 8.59289949392792e-06,
+      "loss": 0.4757,
+      "step": 16427
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.320812702178955,
+      "learning_rate": 8.592716129326429e-06,
+      "loss": 0.7074,
+      "step": 16428
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.8837673664093018,
+      "learning_rate": 8.592532754734939e-06,
+      "loss": 0.6693,
+      "step": 16429
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.618740439414978,
+      "learning_rate": 8.59234937015396e-06,
+      "loss": 0.4645,
+      "step": 16430
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.4378533363342285,
+      "learning_rate": 8.592165975584003e-06,
+      "loss": 0.6932,
+      "step": 16431
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.9986237287521362,
+      "learning_rate": 8.591982571025582e-06,
+      "loss": 0.5729,
+      "step": 16432
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.773853063583374,
+      "learning_rate": 8.5917991564792e-06,
+      "loss": 0.885,
+      "step": 16433
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.1794931888580322,
+      "learning_rate": 8.59161573194537e-06,
+      "loss": 0.6528,
+      "step": 16434
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.4754538536071777,
+      "learning_rate": 8.591432297424603e-06,
+      "loss": 0.6033,
+      "step": 16435
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.9555375576019287,
+      "learning_rate": 8.591248852917407e-06,
+      "loss": 0.813,
+      "step": 16436
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.170856475830078,
+      "learning_rate": 8.591065398424294e-06,
+      "loss": 0.5135,
+      "step": 16437
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.180701971054077,
+      "learning_rate": 8.590881933945774e-06,
+      "loss": 0.9593,
+      "step": 16438
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 3.4577033519744873,
+      "learning_rate": 8.590698459482355e-06,
+      "loss": 0.8815,
+      "step": 16439
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.8554773330688477,
+      "learning_rate": 8.590514975034549e-06,
+      "loss": 0.7059,
+      "step": 16440
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.2350339889526367,
+      "learning_rate": 8.590331480602867e-06,
+      "loss": 0.7424,
+      "step": 16441
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.8539011478424072,
+      "learning_rate": 8.590147976187817e-06,
+      "loss": 0.5313,
+      "step": 16442
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.766577124595642,
+      "learning_rate": 8.589964461789912e-06,
+      "loss": 0.5755,
+      "step": 16443
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.570884108543396,
+      "learning_rate": 8.589780937409659e-06,
+      "loss": 0.6751,
+      "step": 16444
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.959791898727417,
+      "learning_rate": 8.589597403047572e-06,
+      "loss": 0.779,
+      "step": 16445
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.9788771867752075,
+      "learning_rate": 8.589413858704157e-06,
+      "loss": 0.3052,
+      "step": 16446
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.058555841445923,
+      "learning_rate": 8.589230304379927e-06,
+      "loss": 0.7785,
+      "step": 16447
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.85872483253479,
+      "learning_rate": 8.589046740075393e-06,
+      "loss": 0.9577,
+      "step": 16448
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.907261610031128,
+      "learning_rate": 8.588863165791064e-06,
+      "loss": 0.4629,
+      "step": 16449
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 3.126812219619751,
+      "learning_rate": 8.588679581527452e-06,
+      "loss": 0.7316,
+      "step": 16450
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.0935871601104736,
+      "learning_rate": 8.588495987285065e-06,
+      "loss": 0.5812,
+      "step": 16451
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 3.929908037185669,
+      "learning_rate": 8.588312383064416e-06,
+      "loss": 0.7664,
+      "step": 16452
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 3.5904977321624756,
+      "learning_rate": 8.588128768866013e-06,
+      "loss": 0.608,
+      "step": 16453
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.3602311611175537,
+      "learning_rate": 8.587945144690369e-06,
+      "loss": 0.6332,
+      "step": 16454
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.8164554834365845,
+      "learning_rate": 8.587761510537994e-06,
+      "loss": 0.5655,
+      "step": 16455
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 3.348860740661621,
+      "learning_rate": 8.587577866409398e-06,
+      "loss": 0.7631,
+      "step": 16456
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.1491448879241943,
+      "learning_rate": 8.587394212305092e-06,
+      "loss": 0.6254,
+      "step": 16457
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.8297841548919678,
+      "learning_rate": 8.587210548225585e-06,
+      "loss": 0.6912,
+      "step": 16458
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 3.0254437923431396,
+      "learning_rate": 8.587026874171392e-06,
+      "loss": 0.4578,
+      "step": 16459
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 3.0687708854675293,
+      "learning_rate": 8.58684319014302e-06,
+      "loss": 0.8004,
+      "step": 16460
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.8243050575256348,
+      "learning_rate": 8.586659496140978e-06,
+      "loss": 0.7479,
+      "step": 16461
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.8233321905136108,
+      "learning_rate": 8.586475792165783e-06,
+      "loss": 0.3374,
+      "step": 16462
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.9459701776504517,
+      "learning_rate": 8.586292078217942e-06,
+      "loss": 0.6714,
+      "step": 16463
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.979881763458252,
+      "learning_rate": 8.586108354297965e-06,
+      "loss": 0.6311,
+      "step": 16464
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.7000417709350586,
+      "learning_rate": 8.585924620406366e-06,
+      "loss": 0.5288,
+      "step": 16465
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.4037020206451416,
+      "learning_rate": 8.58574087654365e-06,
+      "loss": 0.9097,
+      "step": 16466
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.949535608291626,
+      "learning_rate": 8.585557122710336e-06,
+      "loss": 0.5378,
+      "step": 16467
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.9199094772338867,
+      "learning_rate": 8.585373358906928e-06,
+      "loss": 0.5676,
+      "step": 16468
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.8946398496627808,
+      "learning_rate": 8.585189585133942e-06,
+      "loss": 0.6191,
+      "step": 16469
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.097684383392334,
+      "learning_rate": 8.585005801391885e-06,
+      "loss": 0.525,
+      "step": 16470
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.1427528858184814,
+      "learning_rate": 8.58482200768127e-06,
+      "loss": 0.4384,
+      "step": 16471
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.546670436859131,
+      "learning_rate": 8.584638204002608e-06,
+      "loss": 0.477,
+      "step": 16472
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.3213162422180176,
+      "learning_rate": 8.58445439035641e-06,
+      "loss": 0.7129,
+      "step": 16473
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.7677737474441528,
+      "learning_rate": 8.584270566743189e-06,
+      "loss": 0.4615,
+      "step": 16474
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.3193774223327637,
+      "learning_rate": 8.584086733163453e-06,
+      "loss": 0.7675,
+      "step": 16475
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.096604585647583,
+      "learning_rate": 8.583902889617715e-06,
+      "loss": 0.8436,
+      "step": 16476
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.8863550424575806,
+      "learning_rate": 8.583719036106484e-06,
+      "loss": 0.7478,
+      "step": 16477
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.9215422868728638,
+      "learning_rate": 8.583535172630272e-06,
+      "loss": 0.8628,
+      "step": 16478
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.976989984512329,
+      "learning_rate": 8.583351299189595e-06,
+      "loss": 0.832,
+      "step": 16479
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.245206356048584,
+      "learning_rate": 8.583167415784958e-06,
+      "loss": 0.4168,
+      "step": 16480
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.6696441173553467,
+      "learning_rate": 8.582983522416875e-06,
+      "loss": 0.8854,
+      "step": 16481
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 3.535954713821411,
+      "learning_rate": 8.582799619085857e-06,
+      "loss": 0.5422,
+      "step": 16482
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.2557315826416016,
+      "learning_rate": 8.582615705792415e-06,
+      "loss": 0.6612,
+      "step": 16483
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.28110671043396,
+      "learning_rate": 8.582431782537063e-06,
+      "loss": 0.7924,
+      "step": 16484
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.8208634853363037,
+      "learning_rate": 8.582247849320306e-06,
+      "loss": 0.442,
+      "step": 16485
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.325718402862549,
+      "learning_rate": 8.582063906142663e-06,
+      "loss": 0.8043,
+      "step": 16486
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.2020623683929443,
+      "learning_rate": 8.581879953004642e-06,
+      "loss": 0.6152,
+      "step": 16487
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.5628513097763062,
+      "learning_rate": 8.581695989906752e-06,
+      "loss": 0.3835,
+      "step": 16488
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.8151196241378784,
+      "learning_rate": 8.58151201684951e-06,
+      "loss": 0.7514,
+      "step": 16489
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.9397484064102173,
+      "learning_rate": 8.581328033833422e-06,
+      "loss": 0.3821,
+      "step": 16490
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.9293766021728516,
+      "learning_rate": 8.581144040859004e-06,
+      "loss": 0.5401,
+      "step": 16491
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.6145830154418945,
+      "learning_rate": 8.580960037926766e-06,
+      "loss": 0.7761,
+      "step": 16492
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.8129874467849731,
+      "learning_rate": 8.58077602503722e-06,
+      "loss": 0.5012,
+      "step": 16493
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.6593809127807617,
+      "learning_rate": 8.580592002190878e-06,
+      "loss": 0.7236,
+      "step": 16494
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.7693912982940674,
+      "learning_rate": 8.58040796938825e-06,
+      "loss": 0.8037,
+      "step": 16495
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 3.255697727203369,
+      "learning_rate": 8.580223926629849e-06,
+      "loss": 0.689,
+      "step": 16496
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.835381507873535,
+      "learning_rate": 8.580039873916187e-06,
+      "loss": 0.3424,
+      "step": 16497
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.4100701808929443,
+      "learning_rate": 8.579855811247773e-06,
+      "loss": 0.5135,
+      "step": 16498
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 3.2098636627197266,
+      "learning_rate": 8.579671738625121e-06,
+      "loss": 0.5098,
+      "step": 16499
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.4231395721435547,
+      "learning_rate": 8.579487656048747e-06,
+      "loss": 0.7706,
+      "step": 16500
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.81526780128479,
+      "learning_rate": 8.579303563519154e-06,
+      "loss": 0.5258,
+      "step": 16501
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 5.13504695892334,
+      "learning_rate": 8.57911946103686e-06,
+      "loss": 0.6231,
+      "step": 16502
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.7683680057525635,
+      "learning_rate": 8.578935348602377e-06,
+      "loss": 0.5112,
+      "step": 16503
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.339789867401123,
+      "learning_rate": 8.578751226216214e-06,
+      "loss": 0.7071,
+      "step": 16504
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.6596872806549072,
+      "learning_rate": 8.578567093878885e-06,
+      "loss": 0.5902,
+      "step": 16505
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.6403310298919678,
+      "learning_rate": 8.578382951590902e-06,
+      "loss": 0.6375,
+      "step": 16506
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.289485216140747,
+      "learning_rate": 8.578198799352775e-06,
+      "loss": 0.6427,
+      "step": 16507
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 4.066699028015137,
+      "learning_rate": 8.57801463716502e-06,
+      "loss": 0.6891,
+      "step": 16508
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.7577359676361084,
+      "learning_rate": 8.577830465028145e-06,
+      "loss": 0.61,
+      "step": 16509
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.8520969152450562,
+      "learning_rate": 8.577646282942664e-06,
+      "loss": 0.7895,
+      "step": 16510
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.4831387996673584,
+      "learning_rate": 8.577462090909088e-06,
+      "loss": 0.6328,
+      "step": 16511
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.1032168865203857,
+      "learning_rate": 8.57727788892793e-06,
+      "loss": 0.7805,
+      "step": 16512
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.465385913848877,
+      "learning_rate": 8.577093676999705e-06,
+      "loss": 0.4791,
+      "step": 16513
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.9755440950393677,
+      "learning_rate": 8.57690945512492e-06,
+      "loss": 0.8198,
+      "step": 16514
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.2591991424560547,
+      "learning_rate": 8.576725223304091e-06,
+      "loss": 0.6973,
+      "step": 16515
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.276684522628784,
+      "learning_rate": 8.576540981537728e-06,
+      "loss": 0.7389,
+      "step": 16516
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.8503625392913818,
+      "learning_rate": 8.576356729826345e-06,
+      "loss": 0.6842,
+      "step": 16517
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 3.3219430446624756,
+      "learning_rate": 8.576172468170454e-06,
+      "loss": 0.5842,
+      "step": 16518
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.065788507461548,
+      "learning_rate": 8.575988196570566e-06,
+      "loss": 0.7868,
+      "step": 16519
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.0566444396972656,
+      "learning_rate": 8.575803915027194e-06,
+      "loss": 0.3719,
+      "step": 16520
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.4418981075286865,
+      "learning_rate": 8.575619623540853e-06,
+      "loss": 0.527,
+      "step": 16521
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.9075779914855957,
+      "learning_rate": 8.57543532211205e-06,
+      "loss": 0.3852,
+      "step": 16522
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.6387137174606323,
+      "learning_rate": 8.575251010741304e-06,
+      "loss": 0.7948,
+      "step": 16523
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 4.296513557434082,
+      "learning_rate": 8.575066689429124e-06,
+      "loss": 0.9011,
+      "step": 16524
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.9769636392593384,
+      "learning_rate": 8.574882358176022e-06,
+      "loss": 0.7117,
+      "step": 16525
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.074901819229126,
+      "learning_rate": 8.574698016982511e-06,
+      "loss": 0.5743,
+      "step": 16526
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.7425403594970703,
+      "learning_rate": 8.574513665849105e-06,
+      "loss": 0.5724,
+      "step": 16527
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.498497247695923,
+      "learning_rate": 8.574329304776315e-06,
+      "loss": 0.7386,
+      "step": 16528
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.9638029336929321,
+      "learning_rate": 8.574144933764652e-06,
+      "loss": 0.6402,
+      "step": 16529
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.705757737159729,
+      "learning_rate": 8.573960552814634e-06,
+      "loss": 0.8311,
+      "step": 16530
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 3.102020263671875,
+      "learning_rate": 8.573776161926769e-06,
+      "loss": 0.7892,
+      "step": 16531
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.9289062023162842,
+      "learning_rate": 8.57359176110157e-06,
+      "loss": 0.7804,
+      "step": 16532
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.551181674003601,
+      "learning_rate": 8.573407350339553e-06,
+      "loss": 0.5344,
+      "step": 16533
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 3.9647889137268066,
+      "learning_rate": 8.57322292964123e-06,
+      "loss": 0.9083,
+      "step": 16534
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.612851858139038,
+      "learning_rate": 8.573038499007109e-06,
+      "loss": 0.8601,
+      "step": 16535
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.7635581493377686,
+      "learning_rate": 8.572854058437709e-06,
+      "loss": 0.7219,
+      "step": 16536
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.4636231660842896,
+      "learning_rate": 8.57266960793354e-06,
+      "loss": 0.5176,
+      "step": 16537
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.0348658561706543,
+      "learning_rate": 8.572485147495115e-06,
+      "loss": 0.6125,
+      "step": 16538
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.9188934564590454,
+      "learning_rate": 8.572300677122947e-06,
+      "loss": 0.549,
+      "step": 16539
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.891679286956787,
+      "learning_rate": 8.572116196817549e-06,
+      "loss": 0.7616,
+      "step": 16540
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.9914414882659912,
+      "learning_rate": 8.571931706579433e-06,
+      "loss": 0.76,
+      "step": 16541
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.2728655338287354,
+      "learning_rate": 8.571747206409113e-06,
+      "loss": 0.6055,
+      "step": 16542
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.5561301708221436,
+      "learning_rate": 8.571562696307104e-06,
+      "loss": 0.6101,
+      "step": 16543
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.901017427444458,
+      "learning_rate": 8.571378176273916e-06,
+      "loss": 0.3514,
+      "step": 16544
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.6303409337997437,
+      "learning_rate": 8.571193646310063e-06,
+      "loss": 0.6426,
+      "step": 16545
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.5189898014068604,
+      "learning_rate": 8.571009106416059e-06,
+      "loss": 0.6376,
+      "step": 16546
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.1501338481903076,
+      "learning_rate": 8.570824556592416e-06,
+      "loss": 0.697,
+      "step": 16547
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.246521234512329,
+      "learning_rate": 8.570639996839647e-06,
+      "loss": 0.6386,
+      "step": 16548
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.4852960109710693,
+      "learning_rate": 8.570455427158267e-06,
+      "loss": 0.6369,
+      "step": 16549
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.1313226222991943,
+      "learning_rate": 8.570270847548786e-06,
+      "loss": 0.741,
+      "step": 16550
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.472884178161621,
+      "learning_rate": 8.57008625801172e-06,
+      "loss": 0.5346,
+      "step": 16551
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.6554315090179443,
+      "learning_rate": 8.569901658547583e-06,
+      "loss": 0.5633,
+      "step": 16552
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.9985687732696533,
+      "learning_rate": 8.569717049156885e-06,
+      "loss": 0.7611,
+      "step": 16553
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 5.192220211029053,
+      "learning_rate": 8.569532429840141e-06,
+      "loss": 0.6127,
+      "step": 16554
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.3618617057800293,
+      "learning_rate": 8.569347800597867e-06,
+      "loss": 0.7313,
+      "step": 16555
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 3.3336639404296875,
+      "learning_rate": 8.569163161430571e-06,
+      "loss": 0.7779,
+      "step": 16556
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.0052244663238525,
+      "learning_rate": 8.56897851233877e-06,
+      "loss": 0.5074,
+      "step": 16557
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.7018043994903564,
+      "learning_rate": 8.56879385332298e-06,
+      "loss": 0.7266,
+      "step": 16558
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.5665740966796875,
+      "learning_rate": 8.568609184383707e-06,
+      "loss": 0.5841,
+      "step": 16559
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.0691611766815186,
+      "learning_rate": 8.56842450552147e-06,
+      "loss": 0.7531,
+      "step": 16560
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.719186305999756,
+      "learning_rate": 8.568239816736781e-06,
+      "loss": 0.6394,
+      "step": 16561
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.7026084661483765,
+      "learning_rate": 8.568055118030154e-06,
+      "loss": 0.2853,
+      "step": 16562
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.9162688255310059,
+      "learning_rate": 8.567870409402102e-06,
+      "loss": 0.8922,
+      "step": 16563
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.3429372310638428,
+      "learning_rate": 8.56768569085314e-06,
+      "loss": 0.8275,
+      "step": 16564
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.2196781635284424,
+      "learning_rate": 8.567500962383778e-06,
+      "loss": 0.4776,
+      "step": 16565
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 4.177454948425293,
+      "learning_rate": 8.567316223994536e-06,
+      "loss": 0.5556,
+      "step": 16566
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.7005324363708496,
+      "learning_rate": 8.567131475685922e-06,
+      "loss": 0.855,
+      "step": 16567
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.176844358444214,
+      "learning_rate": 8.566946717458452e-06,
+      "loss": 0.6249,
+      "step": 16568
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.9179294109344482,
+      "learning_rate": 8.566761949312639e-06,
+      "loss": 0.7586,
+      "step": 16569
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.365311861038208,
+      "learning_rate": 8.566577171248997e-06,
+      "loss": 0.6084,
+      "step": 16570
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.140167236328125,
+      "learning_rate": 8.566392383268041e-06,
+      "loss": 0.5661,
+      "step": 16571
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.8645148277282715,
+      "learning_rate": 8.566207585370282e-06,
+      "loss": 0.5858,
+      "step": 16572
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 4.277562141418457,
+      "learning_rate": 8.566022777556238e-06,
+      "loss": 1.0989,
+      "step": 16573
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.287275552749634,
+      "learning_rate": 8.565837959826417e-06,
+      "loss": 0.6782,
+      "step": 16574
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.050363063812256,
+      "learning_rate": 8.56565313218134e-06,
+      "loss": 0.6978,
+      "step": 16575
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 3.173196792602539,
+      "learning_rate": 8.565468294621515e-06,
+      "loss": 0.6818,
+      "step": 16576
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.7805402278900146,
+      "learning_rate": 8.56528344714746e-06,
+      "loss": 0.6297,
+      "step": 16577
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.283576011657715,
+      "learning_rate": 8.565098589759686e-06,
+      "loss": 0.6547,
+      "step": 16578
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.6750924587249756,
+      "learning_rate": 8.56491372245871e-06,
+      "loss": 0.7395,
+      "step": 16579
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.9401544332504272,
+      "learning_rate": 8.564728845245043e-06,
+      "loss": 0.7332,
+      "step": 16580
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.9857457876205444,
+      "learning_rate": 8.5645439581192e-06,
+      "loss": 0.5896,
+      "step": 16581
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.104139804840088,
+      "learning_rate": 8.564359061081697e-06,
+      "loss": 0.5648,
+      "step": 16582
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.1965746879577637,
+      "learning_rate": 8.564174154133046e-06,
+      "loss": 0.7612,
+      "step": 16583
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.185274839401245,
+      "learning_rate": 8.563989237273762e-06,
+      "loss": 0.7405,
+      "step": 16584
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.7594493627548218,
+      "learning_rate": 8.563804310504359e-06,
+      "loss": 0.4582,
+      "step": 16585
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 3.284085750579834,
+      "learning_rate": 8.563619373825351e-06,
+      "loss": 0.7584,
+      "step": 16586
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.8848354816436768,
+      "learning_rate": 8.563434427237252e-06,
+      "loss": 0.5955,
+      "step": 16587
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.0719504356384277,
+      "learning_rate": 8.563249470740578e-06,
+      "loss": 0.5829,
+      "step": 16588
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 3.6222918033599854,
+      "learning_rate": 8.563064504335842e-06,
+      "loss": 0.9104,
+      "step": 16589
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.8208272457122803,
+      "learning_rate": 8.562879528023558e-06,
+      "loss": 0.7044,
+      "step": 16590
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 4.393877029418945,
+      "learning_rate": 8.562694541804239e-06,
+      "loss": 0.754,
+      "step": 16591
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.4100289344787598,
+      "learning_rate": 8.562509545678404e-06,
+      "loss": 0.5955,
+      "step": 16592
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.9988503456115723,
+      "learning_rate": 8.562324539646563e-06,
+      "loss": 0.874,
+      "step": 16593
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 3.616305112838745,
+      "learning_rate": 8.562139523709232e-06,
+      "loss": 0.788,
+      "step": 16594
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.0291507244110107,
+      "learning_rate": 8.561954497866926e-06,
+      "loss": 0.6149,
+      "step": 16595
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.8346078395843506,
+      "learning_rate": 8.561769462120158e-06,
+      "loss": 0.5013,
+      "step": 16596
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 4.276188850402832,
+      "learning_rate": 8.561584416469443e-06,
+      "loss": 0.7099,
+      "step": 16597
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 3.071013927459717,
+      "learning_rate": 8.561399360915297e-06,
+      "loss": 0.6841,
+      "step": 16598
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.3847596645355225,
+      "learning_rate": 8.561214295458233e-06,
+      "loss": 0.7948,
+      "step": 16599
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.868643045425415,
+      "learning_rate": 8.561029220098766e-06,
+      "loss": 0.5582,
+      "step": 16600
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.9484732151031494,
+      "learning_rate": 8.560844134837411e-06,
+      "loss": 0.4907,
+      "step": 16601
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.8889633417129517,
+      "learning_rate": 8.560659039674683e-06,
+      "loss": 0.6863,
+      "step": 16602
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.2690694332122803,
+      "learning_rate": 8.560473934611095e-06,
+      "loss": 0.8025,
+      "step": 16603
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.11590838432312,
+      "learning_rate": 8.560288819647165e-06,
+      "loss": 0.9103,
+      "step": 16604
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 3.5675270557403564,
+      "learning_rate": 8.560103694783403e-06,
+      "loss": 0.7781,
+      "step": 16605
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.213136672973633,
+      "learning_rate": 8.559918560020326e-06,
+      "loss": 0.8831,
+      "step": 16606
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.4932749271392822,
+      "learning_rate": 8.559733415358451e-06,
+      "loss": 0.7557,
+      "step": 16607
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.56840980052948,
+      "learning_rate": 8.55954826079829e-06,
+      "loss": 0.7495,
+      "step": 16608
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.805678606033325,
+      "learning_rate": 8.559363096340359e-06,
+      "loss": 0.6811,
+      "step": 16609
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.450502634048462,
+      "learning_rate": 8.559177921985171e-06,
+      "loss": 0.5298,
+      "step": 16610
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.156477928161621,
+      "learning_rate": 8.558992737733244e-06,
+      "loss": 0.6564,
+      "step": 16611
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.8737926483154297,
+      "learning_rate": 8.558807543585092e-06,
+      "loss": 0.5872,
+      "step": 16612
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.559100389480591,
+      "learning_rate": 8.558622339541229e-06,
+      "loss": 0.7493,
+      "step": 16613
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.1007847785949707,
+      "learning_rate": 8.558437125602168e-06,
+      "loss": 0.3863,
+      "step": 16614
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.304079294204712,
+      "learning_rate": 8.558251901768428e-06,
+      "loss": 0.4301,
+      "step": 16615
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.716315507888794,
+      "learning_rate": 8.558066668040524e-06,
+      "loss": 0.7318,
+      "step": 16616
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.4819467067718506,
+      "learning_rate": 8.557881424418968e-06,
+      "loss": 0.6852,
+      "step": 16617
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 3.118483304977417,
+      "learning_rate": 8.557696170904278e-06,
+      "loss": 0.705,
+      "step": 16618
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 4.573771953582764,
+      "learning_rate": 8.557510907496966e-06,
+      "loss": 0.7366,
+      "step": 16619
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.8902900218963623,
+      "learning_rate": 8.557325634197549e-06,
+      "loss": 0.5996,
+      "step": 16620
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.6973981857299805,
+      "learning_rate": 8.557140351006543e-06,
+      "loss": 0.9309,
+      "step": 16621
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.393763542175293,
+      "learning_rate": 8.556955057924461e-06,
+      "loss": 0.5391,
+      "step": 16622
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.722206950187683,
+      "learning_rate": 8.556769754951819e-06,
+      "loss": 0.6527,
+      "step": 16623
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.948734998703003,
+      "learning_rate": 8.556584442089135e-06,
+      "loss": 0.6729,
+      "step": 16624
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.8758933544158936,
+      "learning_rate": 8.55639911933692e-06,
+      "loss": 0.7379,
+      "step": 16625
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 4.542972087860107,
+      "learning_rate": 8.556213786695694e-06,
+      "loss": 0.6797,
+      "step": 16626
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.5640469789505005,
+      "learning_rate": 8.556028444165965e-06,
+      "loss": 0.8356,
+      "step": 16627
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 3.3345301151275635,
+      "learning_rate": 8.555843091748255e-06,
+      "loss": 0.7702,
+      "step": 16628
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.9833014011383057,
+      "learning_rate": 8.555657729443079e-06,
+      "loss": 0.6752,
+      "step": 16629
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 7.405829429626465,
+      "learning_rate": 8.55547235725095e-06,
+      "loss": 0.6395,
+      "step": 16630
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.2083842754364014,
+      "learning_rate": 8.555286975172386e-06,
+      "loss": 0.5095,
+      "step": 16631
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 3.0357565879821777,
+      "learning_rate": 8.555101583207898e-06,
+      "loss": 0.6847,
+      "step": 16632
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.0992884635925293,
+      "learning_rate": 8.554916181358005e-06,
+      "loss": 0.8586,
+      "step": 16633
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.6081247329711914,
+      "learning_rate": 8.554730769623221e-06,
+      "loss": 0.7475,
+      "step": 16634
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 3.989614725112915,
+      "learning_rate": 8.554545348004065e-06,
+      "loss": 0.6025,
+      "step": 16635
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.1539053916931152,
+      "learning_rate": 8.554359916501047e-06,
+      "loss": 0.6124,
+      "step": 16636
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.0177245140075684,
+      "learning_rate": 8.554174475114687e-06,
+      "loss": 0.804,
+      "step": 16637
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.7172685861587524,
+      "learning_rate": 8.553989023845498e-06,
+      "loss": 0.5893,
+      "step": 16638
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.8583872318267822,
+      "learning_rate": 8.553803562694e-06,
+      "loss": 0.7697,
+      "step": 16639
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 3.6574134826660156,
+      "learning_rate": 8.553618091660703e-06,
+      "loss": 0.8961,
+      "step": 16640
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 3.7682886123657227,
+      "learning_rate": 8.553432610746126e-06,
+      "loss": 0.5138,
+      "step": 16641
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.7795974016189575,
+      "learning_rate": 8.553247119950784e-06,
+      "loss": 0.6612,
+      "step": 16642
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.3207144737243652,
+      "learning_rate": 8.553061619275193e-06,
+      "loss": 0.5628,
+      "step": 16643
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.8046993017196655,
+      "learning_rate": 8.552876108719867e-06,
+      "loss": 0.8662,
+      "step": 16644
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 3.636126756668091,
+      "learning_rate": 8.552690588285327e-06,
+      "loss": 0.5334,
+      "step": 16645
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.8756203651428223,
+      "learning_rate": 8.552505057972082e-06,
+      "loss": 0.7048,
+      "step": 16646
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.360980749130249,
+      "learning_rate": 8.552319517780652e-06,
+      "loss": 0.6069,
+      "step": 16647
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.724146604537964,
+      "learning_rate": 8.552133967711553e-06,
+      "loss": 0.416,
+      "step": 16648
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 3.185936689376831,
+      "learning_rate": 8.5519484077653e-06,
+      "loss": 0.3835,
+      "step": 16649
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.716154098510742,
+      "learning_rate": 8.55176283794241e-06,
+      "loss": 0.7568,
+      "step": 16650
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 3.657383918762207,
+      "learning_rate": 8.551577258243395e-06,
+      "loss": 0.6619,
+      "step": 16651
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.872816324234009,
+      "learning_rate": 8.551391668668777e-06,
+      "loss": 0.5366,
+      "step": 16652
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.739867091178894,
+      "learning_rate": 8.551206069219068e-06,
+      "loss": 0.6755,
+      "step": 16653
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 3.922112464904785,
+      "learning_rate": 8.551020459894785e-06,
+      "loss": 0.7928,
+      "step": 16654
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.9298959970474243,
+      "learning_rate": 8.550834840696445e-06,
+      "loss": 0.5768,
+      "step": 16655
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.799572706222534,
+      "learning_rate": 8.550649211624563e-06,
+      "loss": 0.8729,
+      "step": 16656
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.5840564966201782,
+      "learning_rate": 8.550463572679656e-06,
+      "loss": 0.7671,
+      "step": 16657
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.297940492630005,
+      "learning_rate": 8.55027792386224e-06,
+      "loss": 0.69,
+      "step": 16658
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.2451441287994385,
+      "learning_rate": 8.55009226517283e-06,
+      "loss": 0.652,
+      "step": 16659
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.7142928838729858,
+      "learning_rate": 8.549906596611945e-06,
+      "loss": 0.749,
+      "step": 16660
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 3.140944242477417,
+      "learning_rate": 8.549720918180098e-06,
+      "loss": 0.7241,
+      "step": 16661
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.709672689437866,
+      "learning_rate": 8.549535229877807e-06,
+      "loss": 0.7367,
+      "step": 16662
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.643648147583008,
+      "learning_rate": 8.549349531705588e-06,
+      "loss": 0.5444,
+      "step": 16663
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.282733917236328,
+      "learning_rate": 8.54916382366396e-06,
+      "loss": 0.4841,
+      "step": 16664
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.823474645614624,
+      "learning_rate": 8.548978105753434e-06,
+      "loss": 0.56,
+      "step": 16665
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.919288992881775,
+      "learning_rate": 8.548792377974531e-06,
+      "loss": 0.7652,
+      "step": 16666
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.1677732467651367,
+      "learning_rate": 8.548606640327766e-06,
+      "loss": 0.6141,
+      "step": 16667
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.178474187850952,
+      "learning_rate": 8.548420892813654e-06,
+      "loss": 0.4222,
+      "step": 16668
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.318328380584717,
+      "learning_rate": 8.548235135432714e-06,
+      "loss": 0.6639,
+      "step": 16669
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 3.670264720916748,
+      "learning_rate": 8.54804936818546e-06,
+      "loss": 0.8126,
+      "step": 16670
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 3.320307970046997,
+      "learning_rate": 8.54786359107241e-06,
+      "loss": 0.7489,
+      "step": 16671
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 3.748032331466675,
+      "learning_rate": 8.54767780409408e-06,
+      "loss": 0.8207,
+      "step": 16672
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.867978096008301,
+      "learning_rate": 8.547492007250989e-06,
+      "loss": 0.6579,
+      "step": 16673
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.109497308731079,
+      "learning_rate": 8.547306200543651e-06,
+      "loss": 0.5519,
+      "step": 16674
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.9284441471099854,
+      "learning_rate": 8.547120383972583e-06,
+      "loss": 0.621,
+      "step": 16675
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.59249210357666,
+      "learning_rate": 8.546934557538301e-06,
+      "loss": 0.6298,
+      "step": 16676
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.3635523319244385,
+      "learning_rate": 8.546748721241325e-06,
+      "loss": 0.8085,
+      "step": 16677
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.8563427925109863,
+      "learning_rate": 8.546562875082167e-06,
+      "loss": 0.4561,
+      "step": 16678
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 3.346337080001831,
+      "learning_rate": 8.546377019061347e-06,
+      "loss": 0.6318,
+      "step": 16679
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.104520797729492,
+      "learning_rate": 8.546191153179381e-06,
+      "loss": 0.7789,
+      "step": 16680
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.067831516265869,
+      "learning_rate": 8.546005277436788e-06,
+      "loss": 0.7329,
+      "step": 16681
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.574883222579956,
+      "learning_rate": 8.545819391834081e-06,
+      "loss": 0.7009,
+      "step": 16682
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.349381923675537,
+      "learning_rate": 8.545633496371778e-06,
+      "loss": 0.6176,
+      "step": 16683
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.9217281341552734,
+      "learning_rate": 8.545447591050397e-06,
+      "loss": 0.8811,
+      "step": 16684
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.7973992824554443,
+      "learning_rate": 8.545261675870453e-06,
+      "loss": 0.3938,
+      "step": 16685
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.215993642807007,
+      "learning_rate": 8.545075750832467e-06,
+      "loss": 0.844,
+      "step": 16686
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.203923225402832,
+      "learning_rate": 8.544889815936952e-06,
+      "loss": 0.5583,
+      "step": 16687
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.812164545059204,
+      "learning_rate": 8.544703871184426e-06,
+      "loss": 0.7096,
+      "step": 16688
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 3.860679864883423,
+      "learning_rate": 8.544517916575407e-06,
+      "loss": 0.662,
+      "step": 16689
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.944460868835449,
+      "learning_rate": 8.54433195211041e-06,
+      "loss": 0.6058,
+      "step": 16690
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.8090357780456543,
+      "learning_rate": 8.544145977789955e-06,
+      "loss": 0.6567,
+      "step": 16691
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.4020142555236816,
+      "learning_rate": 8.54395999361456e-06,
+      "loss": 0.586,
+      "step": 16692
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.966535210609436,
+      "learning_rate": 8.543773999584735e-06,
+      "loss": 0.7707,
+      "step": 16693
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.8838571310043335,
+      "learning_rate": 8.543587995701006e-06,
+      "loss": 0.4155,
+      "step": 16694
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.7691802978515625,
+      "learning_rate": 8.543401981963884e-06,
+      "loss": 0.5328,
+      "step": 16695
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.8788559436798096,
+      "learning_rate": 8.543215958373889e-06,
+      "loss": 0.6334,
+      "step": 16696
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.914335012435913,
+      "learning_rate": 8.543029924931539e-06,
+      "loss": 0.6274,
+      "step": 16697
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 3.6338956356048584,
+      "learning_rate": 8.542843881637348e-06,
+      "loss": 0.525,
+      "step": 16698
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.311513900756836,
+      "learning_rate": 8.542657828491836e-06,
+      "loss": 0.5252,
+      "step": 16699
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 3.1388022899627686,
+      "learning_rate": 8.542471765495518e-06,
+      "loss": 0.6853,
+      "step": 16700
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 3.2513487339019775,
+      "learning_rate": 8.542285692648918e-06,
+      "loss": 0.6049,
+      "step": 16701
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.9302029609680176,
+      "learning_rate": 8.542099609952545e-06,
+      "loss": 0.6452,
+      "step": 16702
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.397420883178711,
+      "learning_rate": 8.54191351740692e-06,
+      "loss": 0.5167,
+      "step": 16703
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.6280277967453003,
+      "learning_rate": 8.54172741501256e-06,
+      "loss": 0.6751,
+      "step": 16704
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.7813316583633423,
+      "learning_rate": 8.541541302769984e-06,
+      "loss": 0.7257,
+      "step": 16705
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.7716169357299805,
+      "learning_rate": 8.541355180679708e-06,
+      "loss": 0.5745,
+      "step": 16706
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.651724100112915,
+      "learning_rate": 8.541169048742249e-06,
+      "loss": 0.7979,
+      "step": 16707
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.9814257621765137,
+      "learning_rate": 8.540982906958127e-06,
+      "loss": 0.9272,
+      "step": 16708
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.6857224702835083,
+      "learning_rate": 8.540796755327857e-06,
+      "loss": 0.5013,
+      "step": 16709
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 3.335050106048584,
+      "learning_rate": 8.540610593851957e-06,
+      "loss": 0.3332,
+      "step": 16710
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.6842172145843506,
+      "learning_rate": 8.540424422530947e-06,
+      "loss": 0.7165,
+      "step": 16711
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 3.061185359954834,
+      "learning_rate": 8.540238241365342e-06,
+      "loss": 0.7925,
+      "step": 16712
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 3.456179618835449,
+      "learning_rate": 8.54005205035566e-06,
+      "loss": 0.7413,
+      "step": 16713
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.2641788721084595,
+      "learning_rate": 8.53986584950242e-06,
+      "loss": 0.5444,
+      "step": 16714
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.6661782264709473,
+      "learning_rate": 8.53967963880614e-06,
+      "loss": 0.5367,
+      "step": 16715
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.554264545440674,
+      "learning_rate": 8.539493418267335e-06,
+      "loss": 0.5705,
+      "step": 16716
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 3.096935987472534,
+      "learning_rate": 8.539307187886526e-06,
+      "loss": 0.6749,
+      "step": 16717
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.510450601577759,
+      "learning_rate": 8.53912094766423e-06,
+      "loss": 0.5014,
+      "step": 16718
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.2426226139068604,
+      "learning_rate": 8.538934697600964e-06,
+      "loss": 0.5933,
+      "step": 16719
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.8998661041259766,
+      "learning_rate": 8.538748437697247e-06,
+      "loss": 0.7814,
+      "step": 16720
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.9848382472991943,
+      "learning_rate": 8.538562167953595e-06,
+      "loss": 0.6248,
+      "step": 16721
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 3.567009210586548,
+      "learning_rate": 8.538375888370528e-06,
+      "loss": 0.7938,
+      "step": 16722
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 3.020467758178711,
+      "learning_rate": 8.538189598948561e-06,
+      "loss": 0.5482,
+      "step": 16723
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.6928741931915283,
+      "learning_rate": 8.538003299688217e-06,
+      "loss": 0.5592,
+      "step": 16724
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 3.0733559131622314,
+      "learning_rate": 8.537816990590009e-06,
+      "loss": 0.6206,
+      "step": 16725
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.6509517431259155,
+      "learning_rate": 8.537630671654459e-06,
+      "loss": 0.7483,
+      "step": 16726
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.7748692035675049,
+      "learning_rate": 8.537444342882082e-06,
+      "loss": 0.5675,
+      "step": 16727
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.7879716157913208,
+      "learning_rate": 8.537258004273399e-06,
+      "loss": 0.5816,
+      "step": 16728
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.8179638385772705,
+      "learning_rate": 8.537071655828926e-06,
+      "loss": 0.6965,
+      "step": 16729
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.7916898727416992,
+      "learning_rate": 8.53688529754918e-06,
+      "loss": 0.6287,
+      "step": 16730
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.4014017581939697,
+      "learning_rate": 8.536698929434684e-06,
+      "loss": 0.8698,
+      "step": 16731
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.915626883506775,
+      "learning_rate": 8.536512551485952e-06,
+      "loss": 0.4925,
+      "step": 16732
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.7807703018188477,
+      "learning_rate": 8.5363261637035e-06,
+      "loss": 0.4378,
+      "step": 16733
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.1150248050689697,
+      "learning_rate": 8.536139766087854e-06,
+      "loss": 0.4323,
+      "step": 16734
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.2653322219848633,
+      "learning_rate": 8.535953358639526e-06,
+      "loss": 0.5653,
+      "step": 16735
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.0964138507843018,
+      "learning_rate": 8.535766941359039e-06,
+      "loss": 0.4721,
+      "step": 16736
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.0818848609924316,
+      "learning_rate": 8.535580514246904e-06,
+      "loss": 0.5685,
+      "step": 16737
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.2838525772094727,
+      "learning_rate": 8.535394077303647e-06,
+      "loss": 0.901,
+      "step": 16738
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 3.942633628845215,
+      "learning_rate": 8.535207630529784e-06,
+      "loss": 0.7719,
+      "step": 16739
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.800061821937561,
+      "learning_rate": 8.535021173925832e-06,
+      "loss": 0.7051,
+      "step": 16740
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.4651308059692383,
+      "learning_rate": 8.53483470749231e-06,
+      "loss": 0.6713,
+      "step": 16741
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.4583096504211426,
+      "learning_rate": 8.534648231229738e-06,
+      "loss": 0.6479,
+      "step": 16742
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.7146925926208496,
+      "learning_rate": 8.534461745138632e-06,
+      "loss": 0.59,
+      "step": 16743
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.7421330213546753,
+      "learning_rate": 8.534275249219513e-06,
+      "loss": 0.7205,
+      "step": 16744
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.000980854034424,
+      "learning_rate": 8.534088743472897e-06,
+      "loss": 0.5687,
+      "step": 16745
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 4.067720890045166,
+      "learning_rate": 8.533902227899304e-06,
+      "loss": 0.8188,
+      "step": 16746
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 3.844616413116455,
+      "learning_rate": 8.533715702499255e-06,
+      "loss": 0.7153,
+      "step": 16747
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 3.9344091415405273,
+      "learning_rate": 8.533529167273264e-06,
+      "loss": 0.9799,
+      "step": 16748
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.6204557418823242,
+      "learning_rate": 8.533342622221855e-06,
+      "loss": 0.678,
+      "step": 16749
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.3195927143096924,
+      "learning_rate": 8.533156067345541e-06,
+      "loss": 0.5451,
+      "step": 16750
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.094846725463867,
+      "learning_rate": 8.532969502644846e-06,
+      "loss": 0.7229,
+      "step": 16751
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 3.3463497161865234,
+      "learning_rate": 8.532782928120285e-06,
+      "loss": 0.5855,
+      "step": 16752
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.917575478553772,
+      "learning_rate": 8.532596343772377e-06,
+      "loss": 0.895,
+      "step": 16753
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.9515589475631714,
+      "learning_rate": 8.532409749601642e-06,
+      "loss": 0.807,
+      "step": 16754
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.6501107215881348,
+      "learning_rate": 8.5322231456086e-06,
+      "loss": 0.5773,
+      "step": 16755
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 4.0478925704956055,
+      "learning_rate": 8.532036531793768e-06,
+      "loss": 0.7737,
+      "step": 16756
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 3.054338216781616,
+      "learning_rate": 8.531849908157666e-06,
+      "loss": 0.6129,
+      "step": 16757
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.193748712539673,
+      "learning_rate": 8.531663274700813e-06,
+      "loss": 0.7397,
+      "step": 16758
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.641972780227661,
+      "learning_rate": 8.531476631423728e-06,
+      "loss": 0.6206,
+      "step": 16759
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.574728488922119,
+      "learning_rate": 8.531289978326926e-06,
+      "loss": 0.5503,
+      "step": 16760
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.7398419380187988,
+      "learning_rate": 8.531103315410933e-06,
+      "loss": 0.669,
+      "step": 16761
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.857334852218628,
+      "learning_rate": 8.530916642676264e-06,
+      "loss": 0.7238,
+      "step": 16762
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.4099843502044678,
+      "learning_rate": 8.530729960123435e-06,
+      "loss": 0.7689,
+      "step": 16763
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.929164171218872,
+      "learning_rate": 8.530543267752973e-06,
+      "loss": 0.7891,
+      "step": 16764
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.3651010990142822,
+      "learning_rate": 8.53035656556539e-06,
+      "loss": 0.5257,
+      "step": 16765
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 3.3922119140625,
+      "learning_rate": 8.530169853561208e-06,
+      "loss": 0.82,
+      "step": 16766
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.1232030391693115,
+      "learning_rate": 8.529983131740948e-06,
+      "loss": 0.3827,
+      "step": 16767
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.421264886856079,
+      "learning_rate": 8.529796400105127e-06,
+      "loss": 0.8309,
+      "step": 16768
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 3.503063678741455,
+      "learning_rate": 8.529609658654263e-06,
+      "loss": 0.9205,
+      "step": 16769
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 3.1181533336639404,
+      "learning_rate": 8.529422907388876e-06,
+      "loss": 0.4487,
+      "step": 16770
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.3672077655792236,
+      "learning_rate": 8.529236146309488e-06,
+      "loss": 0.5106,
+      "step": 16771
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.9864532947540283,
+      "learning_rate": 8.529049375416617e-06,
+      "loss": 0.7251,
+      "step": 16772
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 3.9026544094085693,
+      "learning_rate": 8.52886259471078e-06,
+      "loss": 0.6568,
+      "step": 16773
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.5126092433929443,
+      "learning_rate": 8.528675804192498e-06,
+      "loss": 0.4796,
+      "step": 16774
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 4.269010543823242,
+      "learning_rate": 8.52848900386229e-06,
+      "loss": 0.8809,
+      "step": 16775
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.599142074584961,
+      "learning_rate": 8.528302193720678e-06,
+      "loss": 0.7985,
+      "step": 16776
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.1788551807403564,
+      "learning_rate": 8.528115373768177e-06,
+      "loss": 0.6614,
+      "step": 16777
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.008010149002075,
+      "learning_rate": 8.527928544005311e-06,
+      "loss": 0.8484,
+      "step": 16778
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 3.8270301818847656,
+      "learning_rate": 8.527741704432598e-06,
+      "loss": 0.7218,
+      "step": 16779
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.0516297817230225,
+      "learning_rate": 8.527554855050555e-06,
+      "loss": 0.8319,
+      "step": 16780
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.951968193054199,
+      "learning_rate": 8.527367995859703e-06,
+      "loss": 0.4973,
+      "step": 16781
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.1524477005004883,
+      "learning_rate": 8.527181126860562e-06,
+      "loss": 0.731,
+      "step": 16782
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.9372719526290894,
+      "learning_rate": 8.526994248053653e-06,
+      "loss": 0.604,
+      "step": 16783
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.4977694749832153,
+      "learning_rate": 8.526807359439494e-06,
+      "loss": 0.7598,
+      "step": 16784
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.215209484100342,
+      "learning_rate": 8.526620461018604e-06,
+      "loss": 0.8626,
+      "step": 16785
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.9351646900177,
+      "learning_rate": 8.526433552791503e-06,
+      "loss": 0.7934,
+      "step": 16786
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 4.203113555908203,
+      "learning_rate": 8.526246634758712e-06,
+      "loss": 0.5737,
+      "step": 16787
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.5498149394989014,
+      "learning_rate": 8.52605970692075e-06,
+      "loss": 0.6377,
+      "step": 16788
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.8154367208480835,
+      "learning_rate": 8.525872769278139e-06,
+      "loss": 0.6779,
+      "step": 16789
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.5364630222320557,
+      "learning_rate": 8.525685821831394e-06,
+      "loss": 0.5587,
+      "step": 16790
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.655857801437378,
+      "learning_rate": 8.525498864581038e-06,
+      "loss": 0.6775,
+      "step": 16791
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 3.9139554500579834,
+      "learning_rate": 8.525311897527591e-06,
+      "loss": 0.9963,
+      "step": 16792
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.952401876449585,
+      "learning_rate": 8.525124920671572e-06,
+      "loss": 0.7164,
+      "step": 16793
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.4473960399627686,
+      "learning_rate": 8.5249379340135e-06,
+      "loss": 0.62,
+      "step": 16794
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.1117310523986816,
+      "learning_rate": 8.524750937553898e-06,
+      "loss": 0.5484,
+      "step": 16795
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.3596601486206055,
+      "learning_rate": 8.524563931293283e-06,
+      "loss": 0.5981,
+      "step": 16796
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.38424551486969,
+      "learning_rate": 8.524376915232175e-06,
+      "loss": 0.5448,
+      "step": 16797
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.5905672311782837,
+      "learning_rate": 8.524189889371097e-06,
+      "loss": 0.7671,
+      "step": 16798
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.729215621948242,
+      "learning_rate": 8.524002853710568e-06,
+      "loss": 0.4007,
+      "step": 16799
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.857457160949707,
+      "learning_rate": 8.523815808251104e-06,
+      "loss": 0.6554,
+      "step": 16800
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.170778512954712,
+      "learning_rate": 8.52362875299323e-06,
+      "loss": 0.7636,
+      "step": 16801
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 3.373133897781372,
+      "learning_rate": 8.523441687937464e-06,
+      "loss": 0.7821,
+      "step": 16802
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.077881336212158,
+      "learning_rate": 8.523254613084327e-06,
+      "loss": 0.6581,
+      "step": 16803
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.927515745162964,
+      "learning_rate": 8.52306752843434e-06,
+      "loss": 0.8169,
+      "step": 16804
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.5068819522857666,
+      "learning_rate": 8.52288043398802e-06,
+      "loss": 0.8204,
+      "step": 16805
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 3.7904582023620605,
+      "learning_rate": 8.522693329745887e-06,
+      "loss": 0.6289,
+      "step": 16806
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.082444190979004,
+      "learning_rate": 8.522506215708468e-06,
+      "loss": 0.6539,
+      "step": 16807
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.102644681930542,
+      "learning_rate": 8.522319091876276e-06,
+      "loss": 0.665,
+      "step": 16808
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.5485615730285645,
+      "learning_rate": 8.522131958249836e-06,
+      "loss": 0.7456,
+      "step": 16809
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.6980135440826416,
+      "learning_rate": 8.521944814829665e-06,
+      "loss": 0.4152,
+      "step": 16810
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.242232084274292,
+      "learning_rate": 8.521757661616286e-06,
+      "loss": 0.7032,
+      "step": 16811
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.210055112838745,
+      "learning_rate": 8.521570498610218e-06,
+      "loss": 0.6618,
+      "step": 16812
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.580124855041504,
+      "learning_rate": 8.521383325811982e-06,
+      "loss": 0.7658,
+      "step": 16813
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 3.5969924926757812,
+      "learning_rate": 8.521196143222095e-06,
+      "loss": 0.9077,
+      "step": 16814
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.749905824661255,
+      "learning_rate": 8.521008950841084e-06,
+      "loss": 0.7829,
+      "step": 16815
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.214698553085327,
+      "learning_rate": 8.520821748669466e-06,
+      "loss": 0.6682,
+      "step": 16816
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.291292428970337,
+      "learning_rate": 8.520634536707759e-06,
+      "loss": 0.7946,
+      "step": 16817
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.3541245460510254,
+      "learning_rate": 8.52044731495649e-06,
+      "loss": 0.6931,
+      "step": 16818
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.6833128929138184,
+      "learning_rate": 8.520260083416173e-06,
+      "loss": 0.6263,
+      "step": 16819
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 3.0125699043273926,
+      "learning_rate": 8.520072842087332e-06,
+      "loss": 0.6424,
+      "step": 16820
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.941474199295044,
+      "learning_rate": 8.519885590970486e-06,
+      "loss": 0.8761,
+      "step": 16821
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.1243064403533936,
+      "learning_rate": 8.51969833006616e-06,
+      "loss": 0.4881,
+      "step": 16822
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 4.263283729553223,
+      "learning_rate": 8.51951105937487e-06,
+      "loss": 0.9064,
+      "step": 16823
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.6604971885681152,
+      "learning_rate": 8.519323778897135e-06,
+      "loss": 0.6253,
+      "step": 16824
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.4796736240386963,
+      "learning_rate": 8.519136488633483e-06,
+      "loss": 0.6267,
+      "step": 16825
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.0909481048583984,
+      "learning_rate": 8.518949188584428e-06,
+      "loss": 0.4502,
+      "step": 16826
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.31540584564209,
+      "learning_rate": 8.518761878750496e-06,
+      "loss": 0.5223,
+      "step": 16827
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.5876203775405884,
+      "learning_rate": 8.518574559132204e-06,
+      "loss": 0.933,
+      "step": 16828
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.2567906379699707,
+      "learning_rate": 8.518387229730074e-06,
+      "loss": 0.8124,
+      "step": 16829
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.2932698726654053,
+      "learning_rate": 8.518199890544628e-06,
+      "loss": 0.5837,
+      "step": 16830
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.5928025245666504,
+      "learning_rate": 8.518012541576387e-06,
+      "loss": 0.5372,
+      "step": 16831
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.1356842517852783,
+      "learning_rate": 8.517825182825868e-06,
+      "loss": 0.5264,
+      "step": 16832
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.026599168777466,
+      "learning_rate": 8.517637814293598e-06,
+      "loss": 0.5372,
+      "step": 16833
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.368215799331665,
+      "learning_rate": 8.517450435980092e-06,
+      "loss": 0.6859,
+      "step": 16834
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.7396080493927,
+      "learning_rate": 8.517263047885875e-06,
+      "loss": 0.7522,
+      "step": 16835
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.6784398555755615,
+      "learning_rate": 8.517075650011468e-06,
+      "loss": 0.8046,
+      "step": 16836
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.5764216184616089,
+      "learning_rate": 8.51688824235739e-06,
+      "loss": 0.7065,
+      "step": 16837
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.87136709690094,
+      "learning_rate": 8.516700824924165e-06,
+      "loss": 0.7443,
+      "step": 16838
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.2331604957580566,
+      "learning_rate": 8.516513397712312e-06,
+      "loss": 0.607,
+      "step": 16839
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 4.160686016082764,
+      "learning_rate": 8.516325960722352e-06,
+      "loss": 0.6735,
+      "step": 16840
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.8780629634857178,
+      "learning_rate": 8.516138513954807e-06,
+      "loss": 0.7646,
+      "step": 16841
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.2393102645874023,
+      "learning_rate": 8.515951057410198e-06,
+      "loss": 0.6708,
+      "step": 16842
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.0494747161865234,
+      "learning_rate": 8.515763591089046e-06,
+      "loss": 0.8402,
+      "step": 16843
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 3.928464651107788,
+      "learning_rate": 8.515576114991873e-06,
+      "loss": 0.7002,
+      "step": 16844
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.000701904296875,
+      "learning_rate": 8.515388629119199e-06,
+      "loss": 0.8748,
+      "step": 16845
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.5981402397155762,
+      "learning_rate": 8.515201133471544e-06,
+      "loss": 0.5211,
+      "step": 16846
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.367159843444824,
+      "learning_rate": 8.515013628049436e-06,
+      "loss": 0.5813,
+      "step": 16847
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.7384142875671387,
+      "learning_rate": 8.514826112853389e-06,
+      "loss": 0.7207,
+      "step": 16848
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.6348016262054443,
+      "learning_rate": 8.514638587883927e-06,
+      "loss": 0.8345,
+      "step": 16849
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 4.808608055114746,
+      "learning_rate": 8.514451053141574e-06,
+      "loss": 0.8191,
+      "step": 16850
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.1507389545440674,
+      "learning_rate": 8.514263508626847e-06,
+      "loss": 0.8067,
+      "step": 16851
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.5628304481506348,
+      "learning_rate": 8.51407595434027e-06,
+      "loss": 0.5588,
+      "step": 16852
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 6.222600936889648,
+      "learning_rate": 8.513888390282366e-06,
+      "loss": 0.9581,
+      "step": 16853
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.992852807044983,
+      "learning_rate": 8.513700816453653e-06,
+      "loss": 0.9007,
+      "step": 16854
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.9727003574371338,
+      "learning_rate": 8.513513232854652e-06,
+      "loss": 0.3413,
+      "step": 16855
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 3.0991597175598145,
+      "learning_rate": 8.513325639485893e-06,
+      "loss": 0.6452,
+      "step": 16856
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.5673744678497314,
+      "learning_rate": 8.513138036347887e-06,
+      "loss": 0.4072,
+      "step": 16857
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 3.084136724472046,
+      "learning_rate": 8.51295042344116e-06,
+      "loss": 0.6174,
+      "step": 16858
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.3386497497558594,
+      "learning_rate": 8.512762800766236e-06,
+      "loss": 0.7572,
+      "step": 16859
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 3.916290521621704,
+      "learning_rate": 8.512575168323633e-06,
+      "loss": 0.3984,
+      "step": 16860
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.7722249031066895,
+      "learning_rate": 8.512387526113876e-06,
+      "loss": 0.7449,
+      "step": 16861
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.7528327703475952,
+      "learning_rate": 8.512199874137483e-06,
+      "loss": 0.7057,
+      "step": 16862
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.3379108905792236,
+      "learning_rate": 8.51201221239498e-06,
+      "loss": 0.4217,
+      "step": 16863
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.9628832340240479,
+      "learning_rate": 8.511824540886884e-06,
+      "loss": 0.6879,
+      "step": 16864
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.3992807865142822,
+      "learning_rate": 8.511636859613721e-06,
+      "loss": 0.682,
+      "step": 16865
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.2185394763946533,
+      "learning_rate": 8.511449168576011e-06,
+      "loss": 0.6304,
+      "step": 16866
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.1063859462738037,
+      "learning_rate": 8.511261467774276e-06,
+      "loss": 0.5772,
+      "step": 16867
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.9948205947875977,
+      "learning_rate": 8.51107375720904e-06,
+      "loss": 0.6318,
+      "step": 16868
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.9651660919189453,
+      "learning_rate": 8.510886036880823e-06,
+      "loss": 0.7095,
+      "step": 16869
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 3.6279876232147217,
+      "learning_rate": 8.510698306790145e-06,
+      "loss": 0.7253,
+      "step": 16870
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 3.2827630043029785,
+      "learning_rate": 8.510510566937532e-06,
+      "loss": 0.9233,
+      "step": 16871
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.4933669567108154,
+      "learning_rate": 8.510322817323504e-06,
+      "loss": 0.4696,
+      "step": 16872
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.191153049468994,
+      "learning_rate": 8.510135057948583e-06,
+      "loss": 0.5367,
+      "step": 16873
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.49778413772583,
+      "learning_rate": 8.50994728881329e-06,
+      "loss": 0.6075,
+      "step": 16874
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.103922128677368,
+      "learning_rate": 8.50975950991815e-06,
+      "loss": 0.7834,
+      "step": 16875
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 4.248005390167236,
+      "learning_rate": 8.509571721263682e-06,
+      "loss": 0.7921,
+      "step": 16876
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 3.22892165184021,
+      "learning_rate": 8.509383922850411e-06,
+      "loss": 1.0133,
+      "step": 16877
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.6456172466278076,
+      "learning_rate": 8.50919611467886e-06,
+      "loss": 0.6284,
+      "step": 16878
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.5153467655181885,
+      "learning_rate": 8.509008296749546e-06,
+      "loss": 0.831,
+      "step": 16879
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.6337382793426514,
+      "learning_rate": 8.508820469062996e-06,
+      "loss": 0.5834,
+      "step": 16880
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 3.87880277633667,
+      "learning_rate": 8.508632631619729e-06,
+      "loss": 0.7597,
+      "step": 16881
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 3.026801824569702,
+      "learning_rate": 8.508444784420272e-06,
+      "loss": 0.8668,
+      "step": 16882
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 3.4970219135284424,
+      "learning_rate": 8.508256927465142e-06,
+      "loss": 0.6634,
+      "step": 16883
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.429652214050293,
+      "learning_rate": 8.508069060754864e-06,
+      "loss": 0.6354,
+      "step": 16884
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.9863663911819458,
+      "learning_rate": 8.50788118428996e-06,
+      "loss": 0.659,
+      "step": 16885
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.9513068199157715,
+      "learning_rate": 8.50769329807095e-06,
+      "loss": 0.5964,
+      "step": 16886
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 3.7776618003845215,
+      "learning_rate": 8.507505402098364e-06,
+      "loss": 0.7798,
+      "step": 16887
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.5972936153411865,
+      "learning_rate": 8.507317496372717e-06,
+      "loss": 0.5617,
+      "step": 16888
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 3.2838735580444336,
+      "learning_rate": 8.507129580894534e-06,
+      "loss": 0.8245,
+      "step": 16889
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.460966110229492,
+      "learning_rate": 8.506941655664339e-06,
+      "loss": 0.4651,
+      "step": 16890
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 3.758253574371338,
+      "learning_rate": 8.50675372068265e-06,
+      "loss": 0.697,
+      "step": 16891
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.0542874336242676,
+      "learning_rate": 8.506565775949995e-06,
+      "loss": 0.718,
+      "step": 16892
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.366931438446045,
+      "learning_rate": 8.506377821466893e-06,
+      "loss": 0.4602,
+      "step": 16893
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 3.4403984546661377,
+      "learning_rate": 8.506189857233868e-06,
+      "loss": 0.6254,
+      "step": 16894
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.941646099090576,
+      "learning_rate": 8.506001883251443e-06,
+      "loss": 0.9077,
+      "step": 16895
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.3175792694091797,
+      "learning_rate": 8.50581389952014e-06,
+      "loss": 0.7179,
+      "step": 16896
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.2010741233825684,
+      "learning_rate": 8.50562590604048e-06,
+      "loss": 0.5838,
+      "step": 16897
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.2123093605041504,
+      "learning_rate": 8.50543790281299e-06,
+      "loss": 0.7606,
+      "step": 16898
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.6281046867370605,
+      "learning_rate": 8.505249889838188e-06,
+      "loss": 0.6581,
+      "step": 16899
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.2124245166778564,
+      "learning_rate": 8.505061867116603e-06,
+      "loss": 0.5122,
+      "step": 16900
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.3619186878204346,
+      "learning_rate": 8.50487383464875e-06,
+      "loss": 0.7579,
+      "step": 16901
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.7728848457336426,
+      "learning_rate": 8.504685792435157e-06,
+      "loss": 0.9083,
+      "step": 16902
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.072498083114624,
+      "learning_rate": 8.504497740476348e-06,
+      "loss": 0.5193,
+      "step": 16903
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.3128793239593506,
+      "learning_rate": 8.50430967877284e-06,
+      "loss": 1.0373,
+      "step": 16904
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.9503068923950195,
+      "learning_rate": 8.504121607325162e-06,
+      "loss": 0.5866,
+      "step": 16905
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.103045701980591,
+      "learning_rate": 8.503933526133833e-06,
+      "loss": 0.5979,
+      "step": 16906
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.8766053915023804,
+      "learning_rate": 8.503745435199378e-06,
+      "loss": 0.6289,
+      "step": 16907
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 3.737642526626587,
+      "learning_rate": 8.50355733452232e-06,
+      "loss": 0.8517,
+      "step": 16908
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.690523147583008,
+      "learning_rate": 8.503369224103182e-06,
+      "loss": 0.5227,
+      "step": 16909
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.1461753845214844,
+      "learning_rate": 8.503181103942484e-06,
+      "loss": 0.6238,
+      "step": 16910
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.4719350337982178,
+      "learning_rate": 8.502992974040755e-06,
+      "loss": 0.4306,
+      "step": 16911
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.765979051589966,
+      "learning_rate": 8.502804834398513e-06,
+      "loss": 0.7321,
+      "step": 16912
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.5725041627883911,
+      "learning_rate": 8.502616685016282e-06,
+      "loss": 0.6056,
+      "step": 16913
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.4886815547943115,
+      "learning_rate": 8.502428525894588e-06,
+      "loss": 0.6312,
+      "step": 16914
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.3909502029418945,
+      "learning_rate": 8.502240357033951e-06,
+      "loss": 0.6781,
+      "step": 16915
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 3.129612445831299,
+      "learning_rate": 8.502052178434897e-06,
+      "loss": 0.7717,
+      "step": 16916
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 3.348396062850952,
+      "learning_rate": 8.501863990097947e-06,
+      "loss": 0.9741,
+      "step": 16917
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.9526398181915283,
+      "learning_rate": 8.501675792023624e-06,
+      "loss": 0.6071,
+      "step": 16918
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.191347360610962,
+      "learning_rate": 8.501487584212453e-06,
+      "loss": 0.7578,
+      "step": 16919
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.01104474067688,
+      "learning_rate": 8.501299366664957e-06,
+      "loss": 0.6391,
+      "step": 16920
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.8009289503097534,
+      "learning_rate": 8.50111113938166e-06,
+      "loss": 0.5655,
+      "step": 16921
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.81997013092041,
+      "learning_rate": 8.500922902363085e-06,
+      "loss": 0.6973,
+      "step": 16922
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.575821876525879,
+      "learning_rate": 8.500734655609752e-06,
+      "loss": 0.6198,
+      "step": 16923
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.957775592803955,
+      "learning_rate": 8.50054639912219e-06,
+      "loss": 0.8428,
+      "step": 16924
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 3.739967107772827,
+      "learning_rate": 8.500358132900917e-06,
+      "loss": 0.5134,
+      "step": 16925
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 5.471301555633545,
+      "learning_rate": 8.50016985694646e-06,
+      "loss": 0.6897,
+      "step": 16926
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.295712947845459,
+      "learning_rate": 8.499981571259343e-06,
+      "loss": 0.5479,
+      "step": 16927
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 3.071298837661743,
+      "learning_rate": 8.49979327584009e-06,
+      "loss": 0.6541,
+      "step": 16928
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.185814380645752,
+      "learning_rate": 8.499604970689219e-06,
+      "loss": 0.5466,
+      "step": 16929
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.5414886474609375,
+      "learning_rate": 8.49941665580726e-06,
+      "loss": 0.7726,
+      "step": 16930
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.0374042987823486,
+      "learning_rate": 8.499228331194734e-06,
+      "loss": 0.8041,
+      "step": 16931
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.4913299083709717,
+      "learning_rate": 8.499039996852165e-06,
+      "loss": 0.5603,
+      "step": 16932
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.258263349533081,
+      "learning_rate": 8.498851652780076e-06,
+      "loss": 0.709,
+      "step": 16933
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 3.3689308166503906,
+      "learning_rate": 8.498663298978992e-06,
+      "loss": 0.8239,
+      "step": 16934
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.600977897644043,
+      "learning_rate": 8.498474935449435e-06,
+      "loss": 0.4083,
+      "step": 16935
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.1043994426727295,
+      "learning_rate": 8.498286562191932e-06,
+      "loss": 0.4774,
+      "step": 16936
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.5132265090942383,
+      "learning_rate": 8.498098179207002e-06,
+      "loss": 0.6122,
+      "step": 16937
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 4.080072402954102,
+      "learning_rate": 8.49790978649517e-06,
+      "loss": 0.6144,
+      "step": 16938
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.2896299362182617,
+      "learning_rate": 8.497721384056965e-06,
+      "loss": 0.5818,
+      "step": 16939
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.5790631771087646,
+      "learning_rate": 8.497532971892906e-06,
+      "loss": 0.7174,
+      "step": 16940
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.522887706756592,
+      "learning_rate": 8.497344550003517e-06,
+      "loss": 0.6161,
+      "step": 16941
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.249777317047119,
+      "learning_rate": 8.497156118389324e-06,
+      "loss": 0.7114,
+      "step": 16942
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.7918305397033691,
+      "learning_rate": 8.49696767705085e-06,
+      "loss": 0.5352,
+      "step": 16943
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.7880604267120361,
+      "learning_rate": 8.496779225988618e-06,
+      "loss": 0.7862,
+      "step": 16944
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.8244874477386475,
+      "learning_rate": 8.496590765203153e-06,
+      "loss": 0.6094,
+      "step": 16945
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 7.054228782653809,
+      "learning_rate": 8.496402294694978e-06,
+      "loss": 0.4931,
+      "step": 16946
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.585098385810852,
+      "learning_rate": 8.49621381446462e-06,
+      "loss": 0.6248,
+      "step": 16947
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.5289249420166016,
+      "learning_rate": 8.4960253245126e-06,
+      "loss": 0.6126,
+      "step": 16948
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.6801918745040894,
+      "learning_rate": 8.495836824839443e-06,
+      "loss": 0.9116,
+      "step": 16949
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.8926726579666138,
+      "learning_rate": 8.495648315445673e-06,
+      "loss": 0.7828,
+      "step": 16950
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.267299175262451,
+      "learning_rate": 8.495459796331815e-06,
+      "loss": 1.0469,
+      "step": 16951
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.7272686958312988,
+      "learning_rate": 8.495271267498394e-06,
+      "loss": 0.7551,
+      "step": 16952
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.427577018737793,
+      "learning_rate": 8.495082728945931e-06,
+      "loss": 0.7548,
+      "step": 16953
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 3.319336175918579,
+      "learning_rate": 8.494894180674953e-06,
+      "loss": 0.8776,
+      "step": 16954
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 4.774655342102051,
+      "learning_rate": 8.494705622685984e-06,
+      "loss": 0.7399,
+      "step": 16955
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.73354172706604,
+      "learning_rate": 8.494517054979548e-06,
+      "loss": 0.7712,
+      "step": 16956
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.3926572799682617,
+      "learning_rate": 8.494328477556169e-06,
+      "loss": 0.7852,
+      "step": 16957
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.4718780517578125,
+      "learning_rate": 8.494139890416371e-06,
+      "loss": 0.568,
+      "step": 16958
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.5225844383239746,
+      "learning_rate": 8.493951293560679e-06,
+      "loss": 0.547,
+      "step": 16959
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.2478362321853638,
+      "learning_rate": 8.493762686989617e-06,
+      "loss": 0.5505,
+      "step": 16960
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.547466278076172,
+      "learning_rate": 8.49357407070371e-06,
+      "loss": 0.4697,
+      "step": 16961
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.8501485586166382,
+      "learning_rate": 8.493385444703483e-06,
+      "loss": 0.8151,
+      "step": 16962
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.5983060598373413,
+      "learning_rate": 8.49319680898946e-06,
+      "loss": 0.5935,
+      "step": 16963
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.4800355434417725,
+      "learning_rate": 8.493008163562164e-06,
+      "loss": 0.4277,
+      "step": 16964
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.2838551998138428,
+      "learning_rate": 8.49281950842212e-06,
+      "loss": 0.2768,
+      "step": 16965
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.048600912094116,
+      "learning_rate": 8.492630843569856e-06,
+      "loss": 0.5815,
+      "step": 16966
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.8084352016448975,
+      "learning_rate": 8.49244216900589e-06,
+      "loss": 0.5407,
+      "step": 16967
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.9920365810394287,
+      "learning_rate": 8.492253484730755e-06,
+      "loss": 0.4994,
+      "step": 16968
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 7.801927089691162,
+      "learning_rate": 8.492064790744968e-06,
+      "loss": 0.6817,
+      "step": 16969
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 3.038200616836548,
+      "learning_rate": 8.49187608704906e-06,
+      "loss": 0.7206,
+      "step": 16970
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 3.8813602924346924,
+      "learning_rate": 8.49168737364355e-06,
+      "loss": 0.5849,
+      "step": 16971
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 3.1553142070770264,
+      "learning_rate": 8.491498650528967e-06,
+      "loss": 0.7682,
+      "step": 16972
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.3256173133850098,
+      "learning_rate": 8.491309917705833e-06,
+      "loss": 0.8358,
+      "step": 16973
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.6730027198791504,
+      "learning_rate": 8.491121175174675e-06,
+      "loss": 0.497,
+      "step": 16974
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.969093918800354,
+      "learning_rate": 8.490932422936017e-06,
+      "loss": 0.7944,
+      "step": 16975
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 5.069087028503418,
+      "learning_rate": 8.490743660990384e-06,
+      "loss": 0.6978,
+      "step": 16976
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.053581714630127,
+      "learning_rate": 8.490554889338298e-06,
+      "loss": 0.7656,
+      "step": 16977
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.3973560333251953,
+      "learning_rate": 8.490366107980288e-06,
+      "loss": 0.5754,
+      "step": 16978
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.218534231185913,
+      "learning_rate": 8.490177316916878e-06,
+      "loss": 0.9181,
+      "step": 16979
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.8999342918395996,
+      "learning_rate": 8.489988516148594e-06,
+      "loss": 0.6697,
+      "step": 16980
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 3.047391414642334,
+      "learning_rate": 8.489799705675956e-06,
+      "loss": 0.4529,
+      "step": 16981
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.285101890563965,
+      "learning_rate": 8.489610885499493e-06,
+      "loss": 0.6789,
+      "step": 16982
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.640990972518921,
+      "learning_rate": 8.48942205561973e-06,
+      "loss": 0.3305,
+      "step": 16983
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.8383655548095703,
+      "learning_rate": 8.489233216037192e-06,
+      "loss": 0.6572,
+      "step": 16984
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.5292093753814697,
+      "learning_rate": 8.489044366752401e-06,
+      "loss": 0.66,
+      "step": 16985
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.188697338104248,
+      "learning_rate": 8.488855507765887e-06,
+      "loss": 0.9223,
+      "step": 16986
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 3.5638844966888428,
+      "learning_rate": 8.488666639078174e-06,
+      "loss": 0.674,
+      "step": 16987
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.846158742904663,
+      "learning_rate": 8.488477760689784e-06,
+      "loss": 0.3,
+      "step": 16988
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.2774300575256348,
+      "learning_rate": 8.488288872601243e-06,
+      "loss": 0.5164,
+      "step": 16989
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 3.6842868328094482,
+      "learning_rate": 8.48809997481308e-06,
+      "loss": 0.7747,
+      "step": 16990
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 3.414802312850952,
+      "learning_rate": 8.487911067325816e-06,
+      "loss": 0.8228,
+      "step": 16991
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.0204787254333496,
+      "learning_rate": 8.487722150139978e-06,
+      "loss": 0.6428,
+      "step": 16992
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 3.0860865116119385,
+      "learning_rate": 8.487533223256092e-06,
+      "loss": 0.7091,
+      "step": 16993
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.0865395069122314,
+      "learning_rate": 8.487344286674682e-06,
+      "loss": 0.7719,
+      "step": 16994
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.042360305786133,
+      "learning_rate": 8.487155340396274e-06,
+      "loss": 0.8404,
+      "step": 16995
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.7880991697311401,
+      "learning_rate": 8.486966384421394e-06,
+      "loss": 0.7448,
+      "step": 16996
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.401728630065918,
+      "learning_rate": 8.486777418750565e-06,
+      "loss": 0.8108,
+      "step": 16997
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.0230393409729004,
+      "learning_rate": 8.486588443384316e-06,
+      "loss": 0.5021,
+      "step": 16998
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.6084370613098145,
+      "learning_rate": 8.48639945832317e-06,
+      "loss": 0.6321,
+      "step": 16999
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.3078429698944092,
+      "learning_rate": 8.486210463567652e-06,
+      "loss": 0.5509,
+      "step": 17000
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.7750980854034424,
+      "learning_rate": 8.48602145911829e-06,
+      "loss": 0.4422,
+      "step": 17001
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.7262778282165527,
+      "learning_rate": 8.485832444975608e-06,
+      "loss": 0.7043,
+      "step": 17002
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.4277102947235107,
+      "learning_rate": 8.485643421140133e-06,
+      "loss": 0.6493,
+      "step": 17003
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.8326252698898315,
+      "learning_rate": 8.485454387612388e-06,
+      "loss": 0.7181,
+      "step": 17004
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.5318139791488647,
+      "learning_rate": 8.485265344392901e-06,
+      "loss": 0.5374,
+      "step": 17005
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 3.511049747467041,
+      "learning_rate": 8.485076291482195e-06,
+      "loss": 0.7615,
+      "step": 17006
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.8021661043167114,
+      "learning_rate": 8.484887228880798e-06,
+      "loss": 0.3936,
+      "step": 17007
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.8373771905899048,
+      "learning_rate": 8.484698156589236e-06,
+      "loss": 0.5746,
+      "step": 17008
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 4.179495334625244,
+      "learning_rate": 8.484509074608032e-06,
+      "loss": 0.6579,
+      "step": 17009
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.4746854305267334,
+      "learning_rate": 8.484319982937715e-06,
+      "loss": 0.6986,
+      "step": 17010
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 3.3466033935546875,
+      "learning_rate": 8.48413088157881e-06,
+      "loss": 0.8312,
+      "step": 17011
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 3.483724594116211,
+      "learning_rate": 8.483941770531841e-06,
+      "loss": 0.4706,
+      "step": 17012
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.4295217990875244,
+      "learning_rate": 8.483752649797336e-06,
+      "loss": 0.9153,
+      "step": 17013
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.4281749725341797,
+      "learning_rate": 8.483563519375819e-06,
+      "loss": 0.6115,
+      "step": 17014
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.966282844543457,
+      "learning_rate": 8.483374379267818e-06,
+      "loss": 0.5262,
+      "step": 17015
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.6218483448028564,
+      "learning_rate": 8.483185229473856e-06,
+      "loss": 0.5306,
+      "step": 17016
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.1199493408203125,
+      "learning_rate": 8.48299606999446e-06,
+      "loss": 0.5296,
+      "step": 17017
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 3.5256507396698,
+      "learning_rate": 8.482806900830159e-06,
+      "loss": 0.8973,
+      "step": 17018
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.406142473220825,
+      "learning_rate": 8.482617721981476e-06,
+      "loss": 0.7726,
+      "step": 17019
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.5648021697998047,
+      "learning_rate": 8.482428533448938e-06,
+      "loss": 0.6145,
+      "step": 17020
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 3.025073766708374,
+      "learning_rate": 8.48223933523307e-06,
+      "loss": 0.5012,
+      "step": 17021
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 3.351627826690674,
+      "learning_rate": 8.4820501273344e-06,
+      "loss": 0.7784,
+      "step": 17022
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 4.947874069213867,
+      "learning_rate": 8.481860909753452e-06,
+      "loss": 0.7877,
+      "step": 17023
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 3.5052337646484375,
+      "learning_rate": 8.481671682490754e-06,
+      "loss": 0.7431,
+      "step": 17024
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.9387729167938232,
+      "learning_rate": 8.48148244554683e-06,
+      "loss": 0.4191,
+      "step": 17025
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.9925129413604736,
+      "learning_rate": 8.481293198922209e-06,
+      "loss": 0.7041,
+      "step": 17026
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.9002432823181152,
+      "learning_rate": 8.481103942617416e-06,
+      "loss": 0.5551,
+      "step": 17027
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 3.093071222305298,
+      "learning_rate": 8.480914676632975e-06,
+      "loss": 0.6764,
+      "step": 17028
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.2388482093811035,
+      "learning_rate": 8.480725400969415e-06,
+      "loss": 0.5683,
+      "step": 17029
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.9393694400787354,
+      "learning_rate": 8.480536115627263e-06,
+      "loss": 0.748,
+      "step": 17030
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 3.0875771045684814,
+      "learning_rate": 8.480346820607042e-06,
+      "loss": 0.9365,
+      "step": 17031
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.1233794689178467,
+      "learning_rate": 8.480157515909282e-06,
+      "loss": 0.7366,
+      "step": 17032
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.793073892593384,
+      "learning_rate": 8.479968201534506e-06,
+      "loss": 0.6601,
+      "step": 17033
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 3.363330125808716,
+      "learning_rate": 8.479778877483244e-06,
+      "loss": 0.6211,
+      "step": 17034
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.837438702583313,
+      "learning_rate": 8.47958954375602e-06,
+      "loss": 0.8008,
+      "step": 17035
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.5935425758361816,
+      "learning_rate": 8.479400200353359e-06,
+      "loss": 0.6789,
+      "step": 17036
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 3.4770452976226807,
+      "learning_rate": 8.479210847275792e-06,
+      "loss": 0.6861,
+      "step": 17037
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 3.245565891265869,
+      "learning_rate": 8.479021484523839e-06,
+      "loss": 0.8687,
+      "step": 17038
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 3.0023136138916016,
+      "learning_rate": 8.478832112098034e-06,
+      "loss": 0.606,
+      "step": 17039
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.731450080871582,
+      "learning_rate": 8.4786427299989e-06,
+      "loss": 0.8284,
+      "step": 17040
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.445084810256958,
+      "learning_rate": 8.478453338226963e-06,
+      "loss": 0.7403,
+      "step": 17041
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.096036434173584,
+      "learning_rate": 8.47826393678275e-06,
+      "loss": 0.4506,
+      "step": 17042
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.5807057619094849,
+      "learning_rate": 8.478074525666789e-06,
+      "loss": 0.4156,
+      "step": 17043
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 3.089120388031006,
+      "learning_rate": 8.477885104879604e-06,
+      "loss": 0.6342,
+      "step": 17044
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 3.1521291732788086,
+      "learning_rate": 8.477695674421725e-06,
+      "loss": 0.8335,
+      "step": 17045
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.609527826309204,
+      "learning_rate": 8.477506234293677e-06,
+      "loss": 0.7929,
+      "step": 17046
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 3.1620845794677734,
+      "learning_rate": 8.477316784495986e-06,
+      "loss": 0.888,
+      "step": 17047
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 3.1351397037506104,
+      "learning_rate": 8.477127325029179e-06,
+      "loss": 0.8079,
+      "step": 17048
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.441779851913452,
+      "learning_rate": 8.476937855893786e-06,
+      "loss": 0.5721,
+      "step": 17049
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.371525526046753,
+      "learning_rate": 8.47674837709033e-06,
+      "loss": 0.962,
+      "step": 17050
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.4270682334899902,
+      "learning_rate": 8.476558888619337e-06,
+      "loss": 0.5672,
+      "step": 17051
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.1062474250793457,
+      "learning_rate": 8.476369390481337e-06,
+      "loss": 0.7157,
+      "step": 17052
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.2783563137054443,
+      "learning_rate": 8.476179882676859e-06,
+      "loss": 0.6235,
+      "step": 17053
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.34340238571167,
+      "learning_rate": 8.475990365206425e-06,
+      "loss": 0.6023,
+      "step": 17054
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.357510805130005,
+      "learning_rate": 8.475800838070564e-06,
+      "loss": 0.6276,
+      "step": 17055
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.4470921754837036,
+      "learning_rate": 8.475611301269803e-06,
+      "loss": 0.4657,
+      "step": 17056
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.592069387435913,
+      "learning_rate": 8.475421754804669e-06,
+      "loss": 0.6496,
+      "step": 17057
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 3.2749764919281006,
+      "learning_rate": 8.475232198675687e-06,
+      "loss": 0.7795,
+      "step": 17058
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.9863932132720947,
+      "learning_rate": 8.475042632883388e-06,
+      "loss": 0.797,
+      "step": 17059
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.7536721229553223,
+      "learning_rate": 8.474853057428299e-06,
+      "loss": 0.6061,
+      "step": 17060
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.8117762804031372,
+      "learning_rate": 8.474663472310942e-06,
+      "loss": 0.803,
+      "step": 17061
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.8670767545700073,
+      "learning_rate": 8.47447387753185e-06,
+      "loss": 0.5464,
+      "step": 17062
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.9464757442474365,
+      "learning_rate": 8.474284273091546e-06,
+      "loss": 0.7112,
+      "step": 17063
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.2634074687957764,
+      "learning_rate": 8.47409465899056e-06,
+      "loss": 0.6542,
+      "step": 17064
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 3.1708908081054688,
+      "learning_rate": 8.473905035229419e-06,
+      "loss": 0.6421,
+      "step": 17065
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.1105833053588867,
+      "learning_rate": 8.473715401808648e-06,
+      "loss": 0.625,
+      "step": 17066
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.9688186645507812,
+      "learning_rate": 8.473525758728776e-06,
+      "loss": 0.6328,
+      "step": 17067
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.2957000732421875,
+      "learning_rate": 8.473336105990328e-06,
+      "loss": 0.7072,
+      "step": 17068
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.2056331634521484,
+      "learning_rate": 8.473146443593837e-06,
+      "loss": 0.6406,
+      "step": 17069
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.9802544116973877,
+      "learning_rate": 8.472956771539826e-06,
+      "loss": 0.6162,
+      "step": 17070
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 3.3693249225616455,
+      "learning_rate": 8.47276708982882e-06,
+      "loss": 0.5498,
+      "step": 17071
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.9941298961639404,
+      "learning_rate": 8.472577398461354e-06,
+      "loss": 0.652,
+      "step": 17072
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.2279698848724365,
+      "learning_rate": 8.47238769743795e-06,
+      "loss": 0.8076,
+      "step": 17073
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.131319046020508,
+      "learning_rate": 8.472197986759136e-06,
+      "loss": 0.6448,
+      "step": 17074
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.7853889465332031,
+      "learning_rate": 8.47200826642544e-06,
+      "loss": 0.6051,
+      "step": 17075
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.7746357917785645,
+      "learning_rate": 8.47181853643739e-06,
+      "loss": 0.5333,
+      "step": 17076
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 3.879847764968872,
+      "learning_rate": 8.471628796795513e-06,
+      "loss": 0.8362,
+      "step": 17077
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 3.2785544395446777,
+      "learning_rate": 8.471439047500335e-06,
+      "loss": 0.7136,
+      "step": 17078
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 3.2150046825408936,
+      "learning_rate": 8.471249288552387e-06,
+      "loss": 0.6394,
+      "step": 17079
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 3.2869811058044434,
+      "learning_rate": 8.471059519952197e-06,
+      "loss": 0.7249,
+      "step": 17080
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.60945987701416,
+      "learning_rate": 8.47086974170029e-06,
+      "loss": 0.6206,
+      "step": 17081
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.013153553009033,
+      "learning_rate": 8.470679953797192e-06,
+      "loss": 0.8021,
+      "step": 17082
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 4.030055522918701,
+      "learning_rate": 8.470490156243435e-06,
+      "loss": 0.7016,
+      "step": 17083
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.8175179958343506,
+      "learning_rate": 8.470300349039544e-06,
+      "loss": 0.749,
+      "step": 17084
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.2441837787628174,
+      "learning_rate": 8.470110532186048e-06,
+      "loss": 0.6235,
+      "step": 17085
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.3207008838653564,
+      "learning_rate": 8.469920705683477e-06,
+      "loss": 0.5132,
+      "step": 17086
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.147439956665039,
+      "learning_rate": 8.469730869532353e-06,
+      "loss": 0.5444,
+      "step": 17087
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.8423025608062744,
+      "learning_rate": 8.46954102373321e-06,
+      "loss": 0.4448,
+      "step": 17088
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.0310418605804443,
+      "learning_rate": 8.46935116828657e-06,
+      "loss": 0.7122,
+      "step": 17089
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 5.554298400878906,
+      "learning_rate": 8.469161303192965e-06,
+      "loss": 0.615,
+      "step": 17090
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 3.923065185546875,
+      "learning_rate": 8.468971428452923e-06,
+      "loss": 0.9606,
+      "step": 17091
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.7390761375427246,
+      "learning_rate": 8.468781544066972e-06,
+      "loss": 0.4993,
+      "step": 17092
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.6625194549560547,
+      "learning_rate": 8.468591650035638e-06,
+      "loss": 0.3698,
+      "step": 17093
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 3.323249578475952,
+      "learning_rate": 8.468401746359449e-06,
+      "loss": 0.8532,
+      "step": 17094
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.7256064414978027,
+      "learning_rate": 8.468211833038934e-06,
+      "loss": 0.567,
+      "step": 17095
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.9709769487380981,
+      "learning_rate": 8.468021910074623e-06,
+      "loss": 0.6131,
+      "step": 17096
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.7193338871002197,
+      "learning_rate": 8.467831977467041e-06,
+      "loss": 0.6635,
+      "step": 17097
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.9224568605422974,
+      "learning_rate": 8.467642035216718e-06,
+      "loss": 0.7599,
+      "step": 17098
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.3483242988586426,
+      "learning_rate": 8.46745208332418e-06,
+      "loss": 0.9571,
+      "step": 17099
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.9180240631103516,
+      "learning_rate": 8.467262121789957e-06,
+      "loss": 0.3848,
+      "step": 17100
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.6046931743621826,
+      "learning_rate": 8.467072150614578e-06,
+      "loss": 0.9063,
+      "step": 17101
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 7.092075347900391,
+      "learning_rate": 8.466882169798568e-06,
+      "loss": 0.7029,
+      "step": 17102
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 3.1197869777679443,
+      "learning_rate": 8.466692179342462e-06,
+      "loss": 0.5385,
+      "step": 17103
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.0971667766571045,
+      "learning_rate": 8.466502179246781e-06,
+      "loss": 0.664,
+      "step": 17104
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.5605391263961792,
+      "learning_rate": 8.466312169512054e-06,
+      "loss": 0.4959,
+      "step": 17105
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.1229519844055176,
+      "learning_rate": 8.466122150138813e-06,
+      "loss": 0.4454,
+      "step": 17106
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.668799877166748,
+      "learning_rate": 8.465932121127586e-06,
+      "loss": 0.4956,
+      "step": 17107
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 3.6790337562561035,
+      "learning_rate": 8.465742082478899e-06,
+      "loss": 0.8818,
+      "step": 17108
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.650625228881836,
+      "learning_rate": 8.465552034193281e-06,
+      "loss": 0.4002,
+      "step": 17109
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.5807366371154785,
+      "learning_rate": 8.465361976271261e-06,
+      "loss": 0.8385,
+      "step": 17110
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.924566388130188,
+      "learning_rate": 8.465171908713369e-06,
+      "loss": 0.7734,
+      "step": 17111
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.6719595193862915,
+      "learning_rate": 8.464981831520131e-06,
+      "loss": 0.5464,
+      "step": 17112
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.0691983699798584,
+      "learning_rate": 8.464791744692076e-06,
+      "loss": 0.6482,
+      "step": 17113
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.9722682237625122,
+      "learning_rate": 8.464601648229733e-06,
+      "loss": 0.8236,
+      "step": 17114
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.3450632095336914,
+      "learning_rate": 8.464411542133633e-06,
+      "loss": 0.7493,
+      "step": 17115
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.927708625793457,
+      "learning_rate": 8.4642214264043e-06,
+      "loss": 0.6531,
+      "step": 17116
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.48753821849823,
+      "learning_rate": 8.464031301042266e-06,
+      "loss": 0.7108,
+      "step": 17117
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.399303436279297,
+      "learning_rate": 8.463841166048059e-06,
+      "loss": 0.6148,
+      "step": 17118
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 3.0465049743652344,
+      "learning_rate": 8.463651021422206e-06,
+      "loss": 0.4048,
+      "step": 17119
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 4.2588911056518555,
+      "learning_rate": 8.463460867165237e-06,
+      "loss": 0.8267,
+      "step": 17120
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.6823111772537231,
+      "learning_rate": 8.463270703277682e-06,
+      "loss": 0.5803,
+      "step": 17121
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.4731053113937378,
+      "learning_rate": 8.463080529760068e-06,
+      "loss": 0.6832,
+      "step": 17122
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.8685060739517212,
+      "learning_rate": 8.462890346612923e-06,
+      "loss": 0.8744,
+      "step": 17123
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.6341578960418701,
+      "learning_rate": 8.46270015383678e-06,
+      "loss": 0.7422,
+      "step": 17124
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.224142074584961,
+      "learning_rate": 8.462509951432162e-06,
+      "loss": 0.7633,
+      "step": 17125
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.598867177963257,
+      "learning_rate": 8.462319739399602e-06,
+      "loss": 0.8637,
+      "step": 17126
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.8857016563415527,
+      "learning_rate": 8.46212951773963e-06,
+      "loss": 0.8389,
+      "step": 17127
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.8997138738632202,
+      "learning_rate": 8.461939286452771e-06,
+      "loss": 0.57,
+      "step": 17128
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.334765911102295,
+      "learning_rate": 8.461749045539554e-06,
+      "loss": 0.7553,
+      "step": 17129
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.1999619007110596,
+      "learning_rate": 8.461558795000512e-06,
+      "loss": 0.8098,
+      "step": 17130
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.890311360359192,
+      "learning_rate": 8.461368534836172e-06,
+      "loss": 0.8403,
+      "step": 17131
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.4588537216186523,
+      "learning_rate": 8.461178265047061e-06,
+      "loss": 0.8241,
+      "step": 17132
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.6439824104309082,
+      "learning_rate": 8.460987985633711e-06,
+      "loss": 0.678,
+      "step": 17133
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.412522554397583,
+      "learning_rate": 8.46079769659665e-06,
+      "loss": 0.6445,
+      "step": 17134
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.783541679382324,
+      "learning_rate": 8.460607397936406e-06,
+      "loss": 0.7753,
+      "step": 17135
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 4.083739757537842,
+      "learning_rate": 8.46041708965351e-06,
+      "loss": 0.6701,
+      "step": 17136
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.0446343421936035,
+      "learning_rate": 8.46022677174849e-06,
+      "loss": 0.5213,
+      "step": 17137
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.1847519874572754,
+      "learning_rate": 8.460036444221876e-06,
+      "loss": 0.7177,
+      "step": 17138
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.6494905948638916,
+      "learning_rate": 8.459846107074196e-06,
+      "loss": 0.6979,
+      "step": 17139
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.520543098449707,
+      "learning_rate": 8.45965576030598e-06,
+      "loss": 0.7773,
+      "step": 17140
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.6568140983581543,
+      "learning_rate": 8.45946540391776e-06,
+      "loss": 0.5116,
+      "step": 17141
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 3.061094045639038,
+      "learning_rate": 8.45927503791006e-06,
+      "loss": 0.4745,
+      "step": 17142
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.110340118408203,
+      "learning_rate": 8.459084662283413e-06,
+      "loss": 0.6035,
+      "step": 17143
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.0522615909576416,
+      "learning_rate": 8.458894277038348e-06,
+      "loss": 0.8368,
+      "step": 17144
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 87.2000961303711,
+      "learning_rate": 8.458703882175391e-06,
+      "loss": 0.6739,
+      "step": 17145
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 5.887093544006348,
+      "learning_rate": 8.458513477695075e-06,
+      "loss": 0.5993,
+      "step": 17146
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.736555576324463,
+      "learning_rate": 8.45832306359793e-06,
+      "loss": 0.5883,
+      "step": 17147
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 3.501648187637329,
+      "learning_rate": 8.458132639884484e-06,
+      "loss": 0.7329,
+      "step": 17148
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.5649640560150146,
+      "learning_rate": 8.457942206555268e-06,
+      "loss": 0.7101,
+      "step": 17149
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.7921903133392334,
+      "learning_rate": 8.457751763610807e-06,
+      "loss": 0.5394,
+      "step": 17150
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.131446361541748,
+      "learning_rate": 8.457561311051634e-06,
+      "loss": 0.6035,
+      "step": 17151
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.753605842590332,
+      "learning_rate": 8.45737084887828e-06,
+      "loss": 0.8222,
+      "step": 17152
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.976393222808838,
+      "learning_rate": 8.457180377091272e-06,
+      "loss": 0.7011,
+      "step": 17153
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 5.790304183959961,
+      "learning_rate": 8.456989895691142e-06,
+      "loss": 0.7418,
+      "step": 17154
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.4164392948150635,
+      "learning_rate": 8.456799404678416e-06,
+      "loss": 0.8904,
+      "step": 17155
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.7496023178100586,
+      "learning_rate": 8.456608904053626e-06,
+      "loss": 0.7023,
+      "step": 17156
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.736140489578247,
+      "learning_rate": 8.456418393817301e-06,
+      "loss": 0.5781,
+      "step": 17157
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.538756847381592,
+      "learning_rate": 8.456227873969973e-06,
+      "loss": 0.5755,
+      "step": 17158
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.5559203624725342,
+      "learning_rate": 8.456037344512169e-06,
+      "loss": 0.5916,
+      "step": 17159
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.795315980911255,
+      "learning_rate": 8.45584680544442e-06,
+      "loss": 0.7163,
+      "step": 17160
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 3.283536911010742,
+      "learning_rate": 8.455656256767256e-06,
+      "loss": 0.6211,
+      "step": 17161
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.1584296226501465,
+      "learning_rate": 8.455465698481205e-06,
+      "loss": 0.6733,
+      "step": 17162
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.5990190505981445,
+      "learning_rate": 8.4552751305868e-06,
+      "loss": 0.7597,
+      "step": 17163
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.4101879596710205,
+      "learning_rate": 8.455084553084569e-06,
+      "loss": 0.4407,
+      "step": 17164
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.6188347339630127,
+      "learning_rate": 8.454893965975042e-06,
+      "loss": 0.6312,
+      "step": 17165
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.6376497745513916,
+      "learning_rate": 8.454703369258749e-06,
+      "loss": 0.9391,
+      "step": 17166
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.4710724353790283,
+      "learning_rate": 8.45451276293622e-06,
+      "loss": 0.6745,
+      "step": 17167
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.5626368522644043,
+      "learning_rate": 8.454322147007985e-06,
+      "loss": 0.5697,
+      "step": 17168
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.216777801513672,
+      "learning_rate": 8.454131521474574e-06,
+      "loss": 0.4354,
+      "step": 17169
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.1641221046447754,
+      "learning_rate": 8.453940886336518e-06,
+      "loss": 0.5866,
+      "step": 17170
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 3.2158334255218506,
+      "learning_rate": 8.453750241594346e-06,
+      "loss": 0.5267,
+      "step": 17171
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.0937306880950928,
+      "learning_rate": 8.453559587248587e-06,
+      "loss": 0.5724,
+      "step": 17172
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 3.1845788955688477,
+      "learning_rate": 8.453368923299774e-06,
+      "loss": 0.8584,
+      "step": 17173
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 4.69972562789917,
+      "learning_rate": 8.453178249748436e-06,
+      "loss": 0.5864,
+      "step": 17174
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.9581491947174072,
+      "learning_rate": 8.452987566595102e-06,
+      "loss": 0.766,
+      "step": 17175
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.4157793521881104,
+      "learning_rate": 8.452796873840302e-06,
+      "loss": 0.7048,
+      "step": 17176
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.020709276199341,
+      "learning_rate": 8.452606171484569e-06,
+      "loss": 0.5286,
+      "step": 17177
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.0523645877838135,
+      "learning_rate": 8.452415459528432e-06,
+      "loss": 0.748,
+      "step": 17178
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.7806593179702759,
+      "learning_rate": 8.452224737972418e-06,
+      "loss": 0.5084,
+      "step": 17179
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 3.1080915927886963,
+      "learning_rate": 8.452034006817063e-06,
+      "loss": 0.6059,
+      "step": 17180
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.912288188934326,
+      "learning_rate": 8.451843266062893e-06,
+      "loss": 0.8076,
+      "step": 17181
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.0248594284057617,
+      "learning_rate": 8.451652515710443e-06,
+      "loss": 0.7747,
+      "step": 17182
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.355682373046875,
+      "learning_rate": 8.451461755760238e-06,
+      "loss": 0.6869,
+      "step": 17183
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 13.836934089660645,
+      "learning_rate": 8.451270986212811e-06,
+      "loss": 0.5814,
+      "step": 17184
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.652021884918213,
+      "learning_rate": 8.451080207068691e-06,
+      "loss": 0.6513,
+      "step": 17185
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.8460018634796143,
+      "learning_rate": 8.450889418328411e-06,
+      "loss": 0.7709,
+      "step": 17186
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.8691303730010986,
+      "learning_rate": 8.450698619992501e-06,
+      "loss": 0.5997,
+      "step": 17187
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.7626397609710693,
+      "learning_rate": 8.45050781206149e-06,
+      "loss": 0.744,
+      "step": 17188
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.9781118631362915,
+      "learning_rate": 8.450316994535912e-06,
+      "loss": 0.5575,
+      "step": 17189
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.9662269353866577,
+      "learning_rate": 8.45012616741629e-06,
+      "loss": 0.5215,
+      "step": 17190
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.1262056827545166,
+      "learning_rate": 8.449935330703163e-06,
+      "loss": 0.6867,
+      "step": 17191
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.5113983154296875,
+      "learning_rate": 8.449744484397058e-06,
+      "loss": 0.6777,
+      "step": 17192
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.1240410804748535,
+      "learning_rate": 8.449553628498504e-06,
+      "loss": 0.8232,
+      "step": 17193
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.2463722229003906,
+      "learning_rate": 8.449362763008036e-06,
+      "loss": 0.6491,
+      "step": 17194
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.7246570587158203,
+      "learning_rate": 8.44917188792618e-06,
+      "loss": 0.6982,
+      "step": 17195
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.0120275020599365,
+      "learning_rate": 8.44898100325347e-06,
+      "loss": 0.7158,
+      "step": 17196
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 3.865025043487549,
+      "learning_rate": 8.448790108990437e-06,
+      "loss": 0.6811,
+      "step": 17197
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 3.364863157272339,
+      "learning_rate": 8.448599205137608e-06,
+      "loss": 0.8999,
+      "step": 17198
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 3.3796303272247314,
+      "learning_rate": 8.44840829169552e-06,
+      "loss": 0.9013,
+      "step": 17199
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.825914144515991,
+      "learning_rate": 8.448217368664697e-06,
+      "loss": 0.9871,
+      "step": 17200
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.694237470626831,
+      "learning_rate": 8.448026436045675e-06,
+      "loss": 0.7167,
+      "step": 17201
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.9333622455596924,
+      "learning_rate": 8.447835493838983e-06,
+      "loss": 0.6273,
+      "step": 17202
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.2715439796447754,
+      "learning_rate": 8.447644542045152e-06,
+      "loss": 0.5665,
+      "step": 17203
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.965853214263916,
+      "learning_rate": 8.447453580664712e-06,
+      "loss": 0.783,
+      "step": 17204
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.4996185302734375,
+      "learning_rate": 8.447262609698197e-06,
+      "loss": 0.695,
+      "step": 17205
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.7127840518951416,
+      "learning_rate": 8.447071629146136e-06,
+      "loss": 0.8068,
+      "step": 17206
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.8246731758117676,
+      "learning_rate": 8.44688063900906e-06,
+      "loss": 0.7187,
+      "step": 17207
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.2676522731781006,
+      "learning_rate": 8.446689639287497e-06,
+      "loss": 0.8116,
+      "step": 17208
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.9753334522247314,
+      "learning_rate": 8.446498629981985e-06,
+      "loss": 0.4457,
+      "step": 17209
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 3.384723424911499,
+      "learning_rate": 8.44630761109305e-06,
+      "loss": 0.4434,
+      "step": 17210
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.864967703819275,
+      "learning_rate": 8.446116582621224e-06,
+      "loss": 0.5796,
+      "step": 17211
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 3.23050856590271,
+      "learning_rate": 8.44592554456704e-06,
+      "loss": 0.7114,
+      "step": 17212
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.6777690649032593,
+      "learning_rate": 8.445734496931026e-06,
+      "loss": 0.6563,
+      "step": 17213
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.6286094188690186,
+      "learning_rate": 8.445543439713717e-06,
+      "loss": 0.5383,
+      "step": 17214
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.4000637531280518,
+      "learning_rate": 8.445352372915641e-06,
+      "loss": 0.572,
+      "step": 17215
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.7037174701690674,
+      "learning_rate": 8.445161296537333e-06,
+      "loss": 0.6796,
+      "step": 17216
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.8885093927383423,
+      "learning_rate": 8.44497021057932e-06,
+      "loss": 0.558,
+      "step": 17217
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.283621311187744,
+      "learning_rate": 8.444779115042136e-06,
+      "loss": 0.7715,
+      "step": 17218
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.4240329265594482,
+      "learning_rate": 8.44458800992631e-06,
+      "loss": 0.6888,
+      "step": 17219
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.6963423490524292,
+      "learning_rate": 8.444396895232379e-06,
+      "loss": 0.6095,
+      "step": 17220
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.0825130939483643,
+      "learning_rate": 8.444205770960867e-06,
+      "loss": 0.6632,
+      "step": 17221
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 5.034461498260498,
+      "learning_rate": 8.44401463711231e-06,
+      "loss": 0.9302,
+      "step": 17222
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.4703409671783447,
+      "learning_rate": 8.44382349368724e-06,
+      "loss": 0.6942,
+      "step": 17223
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.198418378829956,
+      "learning_rate": 8.443632340686185e-06,
+      "loss": 0.4562,
+      "step": 17224
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.0761423110961914,
+      "learning_rate": 8.443441178109678e-06,
+      "loss": 0.5902,
+      "step": 17225
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.358999490737915,
+      "learning_rate": 8.443250005958254e-06,
+      "loss": 0.609,
+      "step": 17226
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.4150282144546509,
+      "learning_rate": 8.443058824232439e-06,
+      "loss": 0.3988,
+      "step": 17227
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.801849126815796,
+      "learning_rate": 8.44286763293277e-06,
+      "loss": 0.5113,
+      "step": 17228
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.7883551120758057,
+      "learning_rate": 8.442676432059772e-06,
+      "loss": 0.7655,
+      "step": 17229
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.479346990585327,
+      "learning_rate": 8.442485221613983e-06,
+      "loss": 0.6889,
+      "step": 17230
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 4.617532730102539,
+      "learning_rate": 8.44229400159593e-06,
+      "loss": 1.0727,
+      "step": 17231
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.3997883796691895,
+      "learning_rate": 8.442102772006149e-06,
+      "loss": 0.7336,
+      "step": 17232
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.6321384906768799,
+      "learning_rate": 8.441911532845169e-06,
+      "loss": 0.7357,
+      "step": 17233
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.9180017709732056,
+      "learning_rate": 8.441720284113523e-06,
+      "loss": 0.5663,
+      "step": 17234
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.320215940475464,
+      "learning_rate": 8.441529025811741e-06,
+      "loss": 0.6978,
+      "step": 17235
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 4.224651336669922,
+      "learning_rate": 8.441337757940358e-06,
+      "loss": 1.0499,
+      "step": 17236
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 3.257831335067749,
+      "learning_rate": 8.4411464804999e-06,
+      "loss": 0.6776,
+      "step": 17237
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.7162771224975586,
+      "learning_rate": 8.440955193490907e-06,
+      "loss": 0.4673,
+      "step": 17238
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.3137612342834473,
+      "learning_rate": 8.440763896913904e-06,
+      "loss": 0.6188,
+      "step": 17239
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 3.355430841445923,
+      "learning_rate": 8.440572590769426e-06,
+      "loss": 0.5621,
+      "step": 17240
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.4025858640670776,
+      "learning_rate": 8.440381275058004e-06,
+      "loss": 0.57,
+      "step": 17241
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.664482831954956,
+      "learning_rate": 8.440189949780172e-06,
+      "loss": 0.6918,
+      "step": 17242
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 4.093806743621826,
+      "learning_rate": 8.43999861493646e-06,
+      "loss": 1.0557,
+      "step": 17243
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.121344566345215,
+      "learning_rate": 8.4398072705274e-06,
+      "loss": 0.6103,
+      "step": 17244
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.7968820333480835,
+      "learning_rate": 8.439615916553524e-06,
+      "loss": 0.8865,
+      "step": 17245
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.374375581741333,
+      "learning_rate": 8.439424553015367e-06,
+      "loss": 0.7453,
+      "step": 17246
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.85390305519104,
+      "learning_rate": 8.439233179913458e-06,
+      "loss": 0.643,
+      "step": 17247
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.711501121520996,
+      "learning_rate": 8.439041797248329e-06,
+      "loss": 0.7665,
+      "step": 17248
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.9024829864501953,
+      "learning_rate": 8.438850405020513e-06,
+      "loss": 0.6191,
+      "step": 17249
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 4.592313766479492,
+      "learning_rate": 8.438659003230542e-06,
+      "loss": 0.7016,
+      "step": 17250
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 5.03046178817749,
+      "learning_rate": 8.43846759187895e-06,
+      "loss": 0.8116,
+      "step": 17251
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.1427230834960938,
+      "learning_rate": 8.438276170966266e-06,
+      "loss": 0.4405,
+      "step": 17252
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.093038558959961,
+      "learning_rate": 8.438084740493024e-06,
+      "loss": 0.7558,
+      "step": 17253
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 3.0807530879974365,
+      "learning_rate": 8.437893300459758e-06,
+      "loss": 0.5255,
+      "step": 17254
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.393878698348999,
+      "learning_rate": 8.437701850866998e-06,
+      "loss": 0.6342,
+      "step": 17255
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.781175971031189,
+      "learning_rate": 8.437510391715276e-06,
+      "loss": 0.8963,
+      "step": 17256
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.8826531171798706,
+      "learning_rate": 8.437318923005125e-06,
+      "loss": 0.6415,
+      "step": 17257
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.282498836517334,
+      "learning_rate": 8.437127444737078e-06,
+      "loss": 0.5845,
+      "step": 17258
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.0079338550567627,
+      "learning_rate": 8.436935956911668e-06,
+      "loss": 0.5854,
+      "step": 17259
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.1758384704589844,
+      "learning_rate": 8.436744459529427e-06,
+      "loss": 0.6544,
+      "step": 17260
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.6967599391937256,
+      "learning_rate": 8.436552952590885e-06,
+      "loss": 0.6563,
+      "step": 17261
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 3.021723508834839,
+      "learning_rate": 8.43636143609658e-06,
+      "loss": 0.5919,
+      "step": 17262
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.9717845916748047,
+      "learning_rate": 8.436169910047038e-06,
+      "loss": 0.6067,
+      "step": 17263
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 3.5519676208496094,
+      "learning_rate": 8.435978374442795e-06,
+      "loss": 0.8102,
+      "step": 17264
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 3.6575002670288086,
+      "learning_rate": 8.435786829284385e-06,
+      "loss": 0.695,
+      "step": 17265
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.0492382049560547,
+      "learning_rate": 8.435595274572337e-06,
+      "loss": 0.4786,
+      "step": 17266
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.5050904750823975,
+      "learning_rate": 8.435403710307187e-06,
+      "loss": 0.5435,
+      "step": 17267
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.9286816120147705,
+      "learning_rate": 8.435212136489466e-06,
+      "loss": 0.7104,
+      "step": 17268
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 3.6727211475372314,
+      "learning_rate": 8.435020553119705e-06,
+      "loss": 1.0804,
+      "step": 17269
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 3.2666029930114746,
+      "learning_rate": 8.43482896019844e-06,
+      "loss": 0.8922,
+      "step": 17270
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.477320671081543,
+      "learning_rate": 8.434637357726205e-06,
+      "loss": 0.5276,
+      "step": 17271
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.74285888671875,
+      "learning_rate": 8.434445745703527e-06,
+      "loss": 0.6621,
+      "step": 17272
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.953056812286377,
+      "learning_rate": 8.434254124130943e-06,
+      "loss": 0.8112,
+      "step": 17273
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.9538516998291016,
+      "learning_rate": 8.434062493008983e-06,
+      "loss": 0.88,
+      "step": 17274
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.0478005409240723,
+      "learning_rate": 8.433870852338184e-06,
+      "loss": 0.6782,
+      "step": 17275
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.643920421600342,
+      "learning_rate": 8.433679202119076e-06,
+      "loss": 0.6642,
+      "step": 17276
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.8803839683532715,
+      "learning_rate": 8.433487542352192e-06,
+      "loss": 0.5301,
+      "step": 17277
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 3.808924913406372,
+      "learning_rate": 8.433295873038066e-06,
+      "loss": 0.5517,
+      "step": 17278
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.4426074028015137,
+      "learning_rate": 8.43310419417723e-06,
+      "loss": 0.6603,
+      "step": 17279
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 3.0322258472442627,
+      "learning_rate": 8.432912505770217e-06,
+      "loss": 0.8024,
+      "step": 17280
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 3.6482975482940674,
+      "learning_rate": 8.43272080781756e-06,
+      "loss": 0.5855,
+      "step": 17281
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.860285997390747,
+      "learning_rate": 8.432529100319792e-06,
+      "loss": 0.7731,
+      "step": 17282
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.8480116128921509,
+      "learning_rate": 8.432337383277447e-06,
+      "loss": 0.7016,
+      "step": 17283
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 3.7031943798065186,
+      "learning_rate": 8.432145656691057e-06,
+      "loss": 0.5418,
+      "step": 17284
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.301494836807251,
+      "learning_rate": 8.431953920561158e-06,
+      "loss": 0.6148,
+      "step": 17285
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.1750779151916504,
+      "learning_rate": 8.431762174888279e-06,
+      "loss": 0.7,
+      "step": 17286
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.857041597366333,
+      "learning_rate": 8.431570419672954e-06,
+      "loss": 0.5579,
+      "step": 17287
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.3410937786102295,
+      "learning_rate": 8.431378654915719e-06,
+      "loss": 0.6874,
+      "step": 17288
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 3.5006046295166016,
+      "learning_rate": 8.431186880617106e-06,
+      "loss": 0.8362,
+      "step": 17289
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.192734956741333,
+      "learning_rate": 8.430995096777645e-06,
+      "loss": 0.7975,
+      "step": 17290
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.6528102159500122,
+      "learning_rate": 8.430803303397875e-06,
+      "loss": 0.5851,
+      "step": 17291
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 3.280651092529297,
+      "learning_rate": 8.430611500478323e-06,
+      "loss": 0.8091,
+      "step": 17292
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 3.4380645751953125,
+      "learning_rate": 8.430419688019527e-06,
+      "loss": 0.4976,
+      "step": 17293
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.5246028900146484,
+      "learning_rate": 8.430227866022018e-06,
+      "loss": 0.501,
+      "step": 17294
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.8259801864624023,
+      "learning_rate": 8.43003603448633e-06,
+      "loss": 0.871,
+      "step": 17295
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 3.230198383331299,
+      "learning_rate": 8.429844193413e-06,
+      "loss": 0.6918,
+      "step": 17296
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 3.2562899589538574,
+      "learning_rate": 8.429652342802555e-06,
+      "loss": 0.3584,
+      "step": 17297
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.809164524078369,
+      "learning_rate": 8.429460482655533e-06,
+      "loss": 0.6903,
+      "step": 17298
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.24086332321167,
+      "learning_rate": 8.429268612972465e-06,
+      "loss": 0.6107,
+      "step": 17299
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.76491117477417,
+      "learning_rate": 8.429076733753888e-06,
+      "loss": 0.5943,
+      "step": 17300
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.080284595489502,
+      "learning_rate": 8.428884845000332e-06,
+      "loss": 0.473,
+      "step": 17301
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.3166351318359375,
+      "learning_rate": 8.42869294671233e-06,
+      "loss": 0.7548,
+      "step": 17302
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.0449671745300293,
+      "learning_rate": 8.428501038890418e-06,
+      "loss": 0.606,
+      "step": 17303
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.5235018730163574,
+      "learning_rate": 8.42830912153513e-06,
+      "loss": 0.7312,
+      "step": 17304
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.6224185228347778,
+      "learning_rate": 8.428117194646997e-06,
+      "loss": 0.5561,
+      "step": 17305
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.7789766788482666,
+      "learning_rate": 8.427925258226557e-06,
+      "loss": 0.5295,
+      "step": 17306
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.5579750537872314,
+      "learning_rate": 8.427733312274338e-06,
+      "loss": 0.8723,
+      "step": 17307
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.383948802947998,
+      "learning_rate": 8.42754135679088e-06,
+      "loss": 0.4913,
+      "step": 17308
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.531540870666504,
+      "learning_rate": 8.427349391776711e-06,
+      "loss": 0.3777,
+      "step": 17309
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.7181944847106934,
+      "learning_rate": 8.427157417232369e-06,
+      "loss": 0.6761,
+      "step": 17310
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.312415361404419,
+      "learning_rate": 8.426965433158384e-06,
+      "loss": 0.7041,
+      "step": 17311
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.8227864503860474,
+      "learning_rate": 8.426773439555291e-06,
+      "loss": 0.5421,
+      "step": 17312
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.7816197872161865,
+      "learning_rate": 8.426581436423627e-06,
+      "loss": 0.6222,
+      "step": 17313
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.3936665058135986,
+      "learning_rate": 8.426389423763924e-06,
+      "loss": 0.6973,
+      "step": 17314
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.324500560760498,
+      "learning_rate": 8.426197401576715e-06,
+      "loss": 0.4962,
+      "step": 17315
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.1308584213256836,
+      "learning_rate": 8.426005369862534e-06,
+      "loss": 0.3758,
+      "step": 17316
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 3.7837255001068115,
+      "learning_rate": 8.425813328621916e-06,
+      "loss": 0.6312,
+      "step": 17317
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.6528840065002441,
+      "learning_rate": 8.425621277855394e-06,
+      "loss": 0.5362,
+      "step": 17318
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.132535457611084,
+      "learning_rate": 8.425429217563502e-06,
+      "loss": 0.5445,
+      "step": 17319
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.3060989379882812,
+      "learning_rate": 8.425237147746776e-06,
+      "loss": 0.5492,
+      "step": 17320
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 3.633115530014038,
+      "learning_rate": 8.425045068405748e-06,
+      "loss": 0.7232,
+      "step": 17321
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.7584421634674072,
+      "learning_rate": 8.424852979540954e-06,
+      "loss": 0.9755,
+      "step": 17322
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.480022668838501,
+      "learning_rate": 8.424660881152923e-06,
+      "loss": 0.7532,
+      "step": 17323
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.239825487136841,
+      "learning_rate": 8.424468773242198e-06,
+      "loss": 0.9009,
+      "step": 17324
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.1456639766693115,
+      "learning_rate": 8.424276655809304e-06,
+      "loss": 0.6529,
+      "step": 17325
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.107231616973877,
+      "learning_rate": 8.424084528854781e-06,
+      "loss": 0.6109,
+      "step": 17326
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.5962789058685303,
+      "learning_rate": 8.423892392379162e-06,
+      "loss": 0.8808,
+      "step": 17327
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.7715502977371216,
+      "learning_rate": 8.42370024638298e-06,
+      "loss": 0.7719,
+      "step": 17328
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.268282890319824,
+      "learning_rate": 8.423508090866772e-06,
+      "loss": 0.6038,
+      "step": 17329
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.9052369594573975,
+      "learning_rate": 8.423315925831068e-06,
+      "loss": 0.6902,
+      "step": 17330
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 3.3032243251800537,
+      "learning_rate": 8.423123751276408e-06,
+      "loss": 0.5468,
+      "step": 17331
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.9137632846832275,
+      "learning_rate": 8.422931567203321e-06,
+      "loss": 0.7795,
+      "step": 17332
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.0955557823181152,
+      "learning_rate": 8.422739373612343e-06,
+      "loss": 0.6049,
+      "step": 17333
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 3.4665157794952393,
+      "learning_rate": 8.42254717050401e-06,
+      "loss": 0.8429,
+      "step": 17334
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.267709255218506,
+      "learning_rate": 8.422354957878855e-06,
+      "loss": 0.5087,
+      "step": 17335
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 4.077171325683594,
+      "learning_rate": 8.422162735737412e-06,
+      "loss": 0.5193,
+      "step": 17336
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.129981756210327,
+      "learning_rate": 8.421970504080218e-06,
+      "loss": 0.5951,
+      "step": 17337
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.7797508239746094,
+      "learning_rate": 8.421778262907805e-06,
+      "loss": 0.8338,
+      "step": 17338
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.8056528568267822,
+      "learning_rate": 8.42158601222071e-06,
+      "loss": 0.5473,
+      "step": 17339
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.1030538082122803,
+      "learning_rate": 8.421393752019464e-06,
+      "loss": 0.5559,
+      "step": 17340
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.8817803859710693,
+      "learning_rate": 8.421201482304605e-06,
+      "loss": 0.5944,
+      "step": 17341
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.8684552907943726,
+      "learning_rate": 8.421009203076667e-06,
+      "loss": 0.6112,
+      "step": 17342
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.077202558517456,
+      "learning_rate": 8.42081691433618e-06,
+      "loss": 0.679,
+      "step": 17343
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.8767383098602295,
+      "learning_rate": 8.420624616083686e-06,
+      "loss": 0.6219,
+      "step": 17344
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.3456318378448486,
+      "learning_rate": 8.420432308319717e-06,
+      "loss": 0.5747,
+      "step": 17345
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.060058832168579,
+      "learning_rate": 8.420239991044806e-06,
+      "loss": 0.6997,
+      "step": 17346
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.111267566680908,
+      "learning_rate": 8.420047664259488e-06,
+      "loss": 0.5786,
+      "step": 17347
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.305302858352661,
+      "learning_rate": 8.419855327964299e-06,
+      "loss": 0.5767,
+      "step": 17348
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.9605557918548584,
+      "learning_rate": 8.419662982159776e-06,
+      "loss": 0.6674,
+      "step": 17349
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 3.088639259338379,
+      "learning_rate": 8.419470626846447e-06,
+      "loss": 0.3701,
+      "step": 17350
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.4461843967437744,
+      "learning_rate": 8.419278262024854e-06,
+      "loss": 0.456,
+      "step": 17351
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.364109992980957,
+      "learning_rate": 8.419085887695529e-06,
+      "loss": 0.7196,
+      "step": 17352
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.581265449523926,
+      "learning_rate": 8.418893503859005e-06,
+      "loss": 0.86,
+      "step": 17353
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.784485101699829,
+      "learning_rate": 8.41870111051582e-06,
+      "loss": 0.5399,
+      "step": 17354
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.6347177028656006,
+      "learning_rate": 8.418508707666508e-06,
+      "loss": 0.7756,
+      "step": 17355
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 3.1122539043426514,
+      "learning_rate": 8.418316295311604e-06,
+      "loss": 0.7011,
+      "step": 17356
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.011415719985962,
+      "learning_rate": 8.418123873451642e-06,
+      "loss": 0.7467,
+      "step": 17357
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.8804702758789062,
+      "learning_rate": 8.417931442087159e-06,
+      "loss": 0.7998,
+      "step": 17358
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.9816621541976929,
+      "learning_rate": 8.41773900121869e-06,
+      "loss": 0.7049,
+      "step": 17359
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.6772007942199707,
+      "learning_rate": 8.417546550846769e-06,
+      "loss": 0.3622,
+      "step": 17360
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.9194326400756836,
+      "learning_rate": 8.417354090971929e-06,
+      "loss": 0.5671,
+      "step": 17361
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.5871707201004028,
+      "learning_rate": 8.417161621594708e-06,
+      "loss": 0.4888,
+      "step": 17362
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.6799848079681396,
+      "learning_rate": 8.41696914271564e-06,
+      "loss": 0.7685,
+      "step": 17363
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 4.807890892028809,
+      "learning_rate": 8.416776654335264e-06,
+      "loss": 0.5426,
+      "step": 17364
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.3395960330963135,
+      "learning_rate": 8.41658415645411e-06,
+      "loss": 0.5923,
+      "step": 17365
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 4.390196323394775,
+      "learning_rate": 8.416391649072714e-06,
+      "loss": 0.6108,
+      "step": 17366
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 3.9022579193115234,
+      "learning_rate": 8.416199132191616e-06,
+      "loss": 0.8601,
+      "step": 17367
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 3.314694404602051,
+      "learning_rate": 8.416006605811347e-06,
+      "loss": 0.6858,
+      "step": 17368
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.1899983882904053,
+      "learning_rate": 8.415814069932443e-06,
+      "loss": 0.6449,
+      "step": 17369
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.4549418687820435,
+      "learning_rate": 8.415621524555438e-06,
+      "loss": 0.7995,
+      "step": 17370
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.3761056661605835,
+      "learning_rate": 8.41542896968087e-06,
+      "loss": 0.3869,
+      "step": 17371
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 3.757512331008911,
+      "learning_rate": 8.415236405309274e-06,
+      "loss": 0.8934,
+      "step": 17372
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 3.695547103881836,
+      "learning_rate": 8.415043831441182e-06,
+      "loss": 0.7443,
+      "step": 17373
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.7420995235443115,
+      "learning_rate": 8.414851248077137e-06,
+      "loss": 0.6372,
+      "step": 17374
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.35042405128479,
+      "learning_rate": 8.414658655217669e-06,
+      "loss": 0.6426,
+      "step": 17375
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.7568247318267822,
+      "learning_rate": 8.414466052863312e-06,
+      "loss": 0.7205,
+      "step": 17376
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 4.575131416320801,
+      "learning_rate": 8.414273441014605e-06,
+      "loss": 0.6204,
+      "step": 17377
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.769752860069275,
+      "learning_rate": 8.414080819672084e-06,
+      "loss": 0.7225,
+      "step": 17378
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 4.734886646270752,
+      "learning_rate": 8.413888188836282e-06,
+      "loss": 0.6122,
+      "step": 17379
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.9944648742675781,
+      "learning_rate": 8.413695548507737e-06,
+      "loss": 0.7043,
+      "step": 17380
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.6384470462799072,
+      "learning_rate": 8.413502898686982e-06,
+      "loss": 0.7887,
+      "step": 17381
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.580262541770935,
+      "learning_rate": 8.413310239374555e-06,
+      "loss": 0.6754,
+      "step": 17382
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.967835783958435,
+      "learning_rate": 8.413117570570989e-06,
+      "loss": 0.7184,
+      "step": 17383
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.1456260681152344,
+      "learning_rate": 8.412924892276824e-06,
+      "loss": 0.586,
+      "step": 17384
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.736502766609192,
+      "learning_rate": 8.412732204492592e-06,
+      "loss": 0.6494,
+      "step": 17385
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.519322395324707,
+      "learning_rate": 8.41253950721883e-06,
+      "loss": 0.6314,
+      "step": 17386
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.8780653476715088,
+      "learning_rate": 8.412346800456074e-06,
+      "loss": 0.5581,
+      "step": 17387
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.8583595752716064,
+      "learning_rate": 8.412154084204863e-06,
+      "loss": 0.6003,
+      "step": 17388
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.0528695583343506,
+      "learning_rate": 8.411961358465726e-06,
+      "loss": 0.3934,
+      "step": 17389
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.0504136085510254,
+      "learning_rate": 8.411768623239205e-06,
+      "loss": 0.6816,
+      "step": 17390
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.861537218093872,
+      "learning_rate": 8.411575878525831e-06,
+      "loss": 0.6259,
+      "step": 17391
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 3.556753158569336,
+      "learning_rate": 8.411383124326145e-06,
+      "loss": 0.6824,
+      "step": 17392
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.041581869125366,
+      "learning_rate": 8.411190360640677e-06,
+      "loss": 0.6389,
+      "step": 17393
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 3.200030565261841,
+      "learning_rate": 8.41099758746997e-06,
+      "loss": 0.6882,
+      "step": 17394
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.8807510137557983,
+      "learning_rate": 8.410804804814555e-06,
+      "loss": 0.6012,
+      "step": 17395
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.29416823387146,
+      "learning_rate": 8.41061201267497e-06,
+      "loss": 0.6011,
+      "step": 17396
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.6628270149230957,
+      "learning_rate": 8.41041921105175e-06,
+      "loss": 0.724,
+      "step": 17397
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.615621328353882,
+      "learning_rate": 8.410226399945433e-06,
+      "loss": 0.7301,
+      "step": 17398
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.484891653060913,
+      "learning_rate": 8.410033579356554e-06,
+      "loss": 0.4949,
+      "step": 17399
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.219512701034546,
+      "learning_rate": 8.409840749285646e-06,
+      "loss": 0.7606,
+      "step": 17400
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.8562021255493164,
+      "learning_rate": 8.409647909733252e-06,
+      "loss": 0.5531,
+      "step": 17401
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.1001315116882324,
+      "learning_rate": 8.409455060699902e-06,
+      "loss": 0.798,
+      "step": 17402
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 4.662204742431641,
+      "learning_rate": 8.409262202186136e-06,
+      "loss": 0.8205,
+      "step": 17403
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 3.997144937515259,
+      "learning_rate": 8.409069334192488e-06,
+      "loss": 0.6587,
+      "step": 17404
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.0239951610565186,
+      "learning_rate": 8.408876456719495e-06,
+      "loss": 0.7497,
+      "step": 17405
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.4164791107177734,
+      "learning_rate": 8.408683569767694e-06,
+      "loss": 0.6153,
+      "step": 17406
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.653024196624756,
+      "learning_rate": 8.408490673337622e-06,
+      "loss": 0.817,
+      "step": 17407
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.6812238693237305,
+      "learning_rate": 8.408297767429815e-06,
+      "loss": 0.7633,
+      "step": 17408
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.64798903465271,
+      "learning_rate": 8.408104852044807e-06,
+      "loss": 0.5335,
+      "step": 17409
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 3.1440343856811523,
+      "learning_rate": 8.407911927183135e-06,
+      "loss": 0.6284,
+      "step": 17410
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.9808931350708008,
+      "learning_rate": 8.407718992845339e-06,
+      "loss": 0.6335,
+      "step": 17411
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.881413698196411,
+      "learning_rate": 8.40752604903195e-06,
+      "loss": 0.6116,
+      "step": 17412
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.9116542339324951,
+      "learning_rate": 8.40733309574351e-06,
+      "loss": 0.7121,
+      "step": 17413
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.7042529582977295,
+      "learning_rate": 8.407140132980553e-06,
+      "loss": 0.8464,
+      "step": 17414
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.5661871433258057,
+      "learning_rate": 8.406947160743615e-06,
+      "loss": 0.6229,
+      "step": 17415
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 3.179978132247925,
+      "learning_rate": 8.406754179033234e-06,
+      "loss": 0.6011,
+      "step": 17416
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.6368186473846436,
+      "learning_rate": 8.406561187849944e-06,
+      "loss": 0.7074,
+      "step": 17417
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.301950454711914,
+      "learning_rate": 8.406368187194285e-06,
+      "loss": 0.4423,
+      "step": 17418
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.414358377456665,
+      "learning_rate": 8.406175177066792e-06,
+      "loss": 0.4798,
+      "step": 17419
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 3.1349706649780273,
+      "learning_rate": 8.405982157468002e-06,
+      "loss": 0.6599,
+      "step": 17420
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.3741469383239746,
+      "learning_rate": 8.405789128398452e-06,
+      "loss": 0.7914,
+      "step": 17421
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 3.0010931491851807,
+      "learning_rate": 8.405596089858676e-06,
+      "loss": 0.6936,
+      "step": 17422
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.68210506439209,
+      "learning_rate": 8.405403041849214e-06,
+      "loss": 0.6998,
+      "step": 17423
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.1981990337371826,
+      "learning_rate": 8.405209984370602e-06,
+      "loss": 0.6214,
+      "step": 17424
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 4.6466217041015625,
+      "learning_rate": 8.405016917423378e-06,
+      "loss": 0.6683,
+      "step": 17425
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.6650334596633911,
+      "learning_rate": 8.404823841008076e-06,
+      "loss": 0.7189,
+      "step": 17426
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.0481929779052734,
+      "learning_rate": 8.404630755125235e-06,
+      "loss": 0.6958,
+      "step": 17427
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.177844285964966,
+      "learning_rate": 8.40443765977539e-06,
+      "loss": 0.4479,
+      "step": 17428
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 3.18593430519104,
+      "learning_rate": 8.40424455495908e-06,
+      "loss": 0.5778,
+      "step": 17429
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.9409606456756592,
+      "learning_rate": 8.404051440676842e-06,
+      "loss": 0.7515,
+      "step": 17430
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.0333235263824463,
+      "learning_rate": 8.403858316929211e-06,
+      "loss": 0.7341,
+      "step": 17431
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.77077054977417,
+      "learning_rate": 8.403665183716726e-06,
+      "loss": 0.9326,
+      "step": 17432
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.6415133476257324,
+      "learning_rate": 8.403472041039924e-06,
+      "loss": 0.509,
+      "step": 17433
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.6789666414260864,
+      "learning_rate": 8.403278888899339e-06,
+      "loss": 0.7807,
+      "step": 17434
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.8090050220489502,
+      "learning_rate": 8.40308572729551e-06,
+      "loss": 0.5861,
+      "step": 17435
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.5844351053237915,
+      "learning_rate": 8.402892556228978e-06,
+      "loss": 0.494,
+      "step": 17436
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.327204704284668,
+      "learning_rate": 8.402699375700273e-06,
+      "loss": 0.5635,
+      "step": 17437
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.8840444087982178,
+      "learning_rate": 8.402506185709938e-06,
+      "loss": 0.9115,
+      "step": 17438
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.417179584503174,
+      "learning_rate": 8.402312986258508e-06,
+      "loss": 0.873,
+      "step": 17439
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 3.474553108215332,
+      "learning_rate": 8.402119777346517e-06,
+      "loss": 0.6612,
+      "step": 17440
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 3.3788115978240967,
+      "learning_rate": 8.401926558974509e-06,
+      "loss": 0.5999,
+      "step": 17441
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.8815972805023193,
+      "learning_rate": 8.401733331143016e-06,
+      "loss": 0.5784,
+      "step": 17442
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.7110562324523926,
+      "learning_rate": 8.401540093852577e-06,
+      "loss": 0.5043,
+      "step": 17443
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.469383716583252,
+      "learning_rate": 8.401346847103728e-06,
+      "loss": 0.5936,
+      "step": 17444
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.8247785568237305,
+      "learning_rate": 8.40115359089701e-06,
+      "loss": 0.528,
+      "step": 17445
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 3.0983994007110596,
+      "learning_rate": 8.400960325232956e-06,
+      "loss": 0.4096,
+      "step": 17446
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.1874542236328125,
+      "learning_rate": 8.400767050112106e-06,
+      "loss": 0.7991,
+      "step": 17447
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.8772648572921753,
+      "learning_rate": 8.400573765534998e-06,
+      "loss": 0.7616,
+      "step": 17448
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.2468667030334473,
+      "learning_rate": 8.400380471502168e-06,
+      "loss": 0.5395,
+      "step": 17449
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 3.0220627784729004,
+      "learning_rate": 8.400187168014153e-06,
+      "loss": 0.6237,
+      "step": 17450
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 4.2204790115356445,
+      "learning_rate": 8.399993855071492e-06,
+      "loss": 0.9847,
+      "step": 17451
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.5482332706451416,
+      "learning_rate": 8.39980053267472e-06,
+      "loss": 0.5853,
+      "step": 17452
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 4.123030662536621,
+      "learning_rate": 8.399607200824376e-06,
+      "loss": 0.6383,
+      "step": 17453
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 3.651320219039917,
+      "learning_rate": 8.399413859521e-06,
+      "loss": 0.6315,
+      "step": 17454
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.309403657913208,
+      "learning_rate": 8.399220508765126e-06,
+      "loss": 0.6969,
+      "step": 17455
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.7199307680130005,
+      "learning_rate": 8.399027148557295e-06,
+      "loss": 0.6142,
+      "step": 17456
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.3045003414154053,
+      "learning_rate": 8.39883377889804e-06,
+      "loss": 0.7766,
+      "step": 17457
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.8562071323394775,
+      "learning_rate": 8.398640399787903e-06,
+      "loss": 0.7715,
+      "step": 17458
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.6085995435714722,
+      "learning_rate": 8.39844701122742e-06,
+      "loss": 0.4505,
+      "step": 17459
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.7465972900390625,
+      "learning_rate": 8.39825361321713e-06,
+      "loss": 0.7116,
+      "step": 17460
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 3.7737877368927,
+      "learning_rate": 8.398060205757568e-06,
+      "loss": 0.8052,
+      "step": 17461
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.9465551376342773,
+      "learning_rate": 8.397866788849277e-06,
+      "loss": 0.7625,
+      "step": 17462
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.006542921066284,
+      "learning_rate": 8.397673362492787e-06,
+      "loss": 0.6125,
+      "step": 17463
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.1539177894592285,
+      "learning_rate": 8.397479926688642e-06,
+      "loss": 0.7408,
+      "step": 17464
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.841557502746582,
+      "learning_rate": 8.397286481437378e-06,
+      "loss": 0.3671,
+      "step": 17465
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 3.2826476097106934,
+      "learning_rate": 8.397093026739533e-06,
+      "loss": 0.7948,
+      "step": 17466
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.53912353515625,
+      "learning_rate": 8.396899562595644e-06,
+      "loss": 0.7458,
+      "step": 17467
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.334045886993408,
+      "learning_rate": 8.396706089006251e-06,
+      "loss": 0.8082,
+      "step": 17468
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.4502429962158203,
+      "learning_rate": 8.396512605971891e-06,
+      "loss": 0.7419,
+      "step": 17469
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.439774990081787,
+      "learning_rate": 8.3963191134931e-06,
+      "loss": 0.4978,
+      "step": 17470
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.630305528640747,
+      "learning_rate": 8.39612561157042e-06,
+      "loss": 0.7249,
+      "step": 17471
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.675084352493286,
+      "learning_rate": 8.395932100204387e-06,
+      "loss": 0.8916,
+      "step": 17472
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.8062455654144287,
+      "learning_rate": 8.395738579395538e-06,
+      "loss": 0.3895,
+      "step": 17473
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.3454039096832275,
+      "learning_rate": 8.395545049144412e-06,
+      "loss": 0.6479,
+      "step": 17474
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.637253761291504,
+      "learning_rate": 8.395351509451546e-06,
+      "loss": 0.9303,
+      "step": 17475
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.4633047580718994,
+      "learning_rate": 8.395157960317481e-06,
+      "loss": 0.6942,
+      "step": 17476
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 3.723712205886841,
+      "learning_rate": 8.394964401742753e-06,
+      "loss": 0.7669,
+      "step": 17477
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.317012071609497,
+      "learning_rate": 8.3947708337279e-06,
+      "loss": 0.5674,
+      "step": 17478
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.066866159439087,
+      "learning_rate": 8.394577256273463e-06,
+      "loss": 0.6515,
+      "step": 17479
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.099393606185913,
+      "learning_rate": 8.394383669379978e-06,
+      "loss": 0.7992,
+      "step": 17480
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 3.1602864265441895,
+      "learning_rate": 8.394190073047983e-06,
+      "loss": 0.7635,
+      "step": 17481
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.8917396068572998,
+      "learning_rate": 8.393996467278016e-06,
+      "loss": 0.8499,
+      "step": 17482
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.131629705429077,
+      "learning_rate": 8.393802852070618e-06,
+      "loss": 0.9162,
+      "step": 17483
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.144460439682007,
+      "learning_rate": 8.393609227426325e-06,
+      "loss": 0.7284,
+      "step": 17484
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 4.071224689483643,
+      "learning_rate": 8.393415593345676e-06,
+      "loss": 0.9192,
+      "step": 17485
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.7843637466430664,
+      "learning_rate": 8.393221949829208e-06,
+      "loss": 0.7069,
+      "step": 17486
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.705094814300537,
+      "learning_rate": 8.393028296877461e-06,
+      "loss": 0.4407,
+      "step": 17487
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 3.094456672668457,
+      "learning_rate": 8.392834634490975e-06,
+      "loss": 1.0401,
+      "step": 17488
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 3.531993865966797,
+      "learning_rate": 8.392640962670286e-06,
+      "loss": 0.9325,
+      "step": 17489
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.2394301891326904,
+      "learning_rate": 8.392447281415934e-06,
+      "loss": 0.7058,
+      "step": 17490
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 3.5137226581573486,
+      "learning_rate": 8.392253590728456e-06,
+      "loss": 0.7915,
+      "step": 17491
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.5606160163879395,
+      "learning_rate": 8.392059890608393e-06,
+      "loss": 0.696,
+      "step": 17492
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 4.917855262756348,
+      "learning_rate": 8.391866181056281e-06,
+      "loss": 0.6271,
+      "step": 17493
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.9736953973770142,
+      "learning_rate": 8.39167246207266e-06,
+      "loss": 0.6726,
+      "step": 17494
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 3.6358282566070557,
+      "learning_rate": 8.391478733658067e-06,
+      "loss": 0.6218,
+      "step": 17495
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.962812900543213,
+      "learning_rate": 8.391284995813044e-06,
+      "loss": 0.6646,
+      "step": 17496
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.161921262741089,
+      "learning_rate": 8.391091248538127e-06,
+      "loss": 0.5145,
+      "step": 17497
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.2648138999938965,
+      "learning_rate": 8.390897491833856e-06,
+      "loss": 0.6271,
+      "step": 17498
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.860910415649414,
+      "learning_rate": 8.39070372570077e-06,
+      "loss": 0.7908,
+      "step": 17499
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.9767229557037354,
+      "learning_rate": 8.390509950139407e-06,
+      "loss": 0.5871,
+      "step": 17500
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 3.3038854598999023,
+      "learning_rate": 8.390316165150305e-06,
+      "loss": 0.8554,
+      "step": 17501
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 6.605527877807617,
+      "learning_rate": 8.390122370734005e-06,
+      "loss": 0.6045,
+      "step": 17502
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 3.3283345699310303,
+      "learning_rate": 8.389928566891044e-06,
+      "loss": 0.846,
+      "step": 17503
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.4596962928771973,
+      "learning_rate": 8.38973475362196e-06,
+      "loss": 0.6939,
+      "step": 17504
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.5775970220565796,
+      "learning_rate": 8.389540930927294e-06,
+      "loss": 0.5987,
+      "step": 17505
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 5.2659592628479,
+      "learning_rate": 8.389347098807586e-06,
+      "loss": 0.6914,
+      "step": 17506
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.4802800416946411,
+      "learning_rate": 8.389153257263373e-06,
+      "loss": 0.7369,
+      "step": 17507
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.5540952682495117,
+      "learning_rate": 8.388959406295194e-06,
+      "loss": 0.7079,
+      "step": 17508
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.3444349765777588,
+      "learning_rate": 8.388765545903588e-06,
+      "loss": 0.4702,
+      "step": 17509
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.5283546447753906,
+      "learning_rate": 8.388571676089095e-06,
+      "loss": 0.9889,
+      "step": 17510
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.677574634552002,
+      "learning_rate": 8.388377796852254e-06,
+      "loss": 0.4303,
+      "step": 17511
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 3.0888864994049072,
+      "learning_rate": 8.388183908193604e-06,
+      "loss": 0.8984,
+      "step": 17512
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 12.863924026489258,
+      "learning_rate": 8.387990010113683e-06,
+      "loss": 0.5918,
+      "step": 17513
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.6660059690475464,
+      "learning_rate": 8.38779610261303e-06,
+      "loss": 0.7195,
+      "step": 17514
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 9.699686050415039,
+      "learning_rate": 8.387602185692187e-06,
+      "loss": 0.8653,
+      "step": 17515
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.8050696849823,
+      "learning_rate": 8.387408259351689e-06,
+      "loss": 0.7809,
+      "step": 17516
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.113863945007324,
+      "learning_rate": 8.387214323592078e-06,
+      "loss": 0.5697,
+      "step": 17517
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.0965685844421387,
+      "learning_rate": 8.387020378413895e-06,
+      "loss": 0.6506,
+      "step": 17518
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.3082261085510254,
+      "learning_rate": 8.386826423817675e-06,
+      "loss": 0.5663,
+      "step": 17519
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.6803066730499268,
+      "learning_rate": 8.386632459803961e-06,
+      "loss": 0.5503,
+      "step": 17520
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.7164866924285889,
+      "learning_rate": 8.38643848637329e-06,
+      "loss": 0.4301,
+      "step": 17521
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.6055500507354736,
+      "learning_rate": 8.386244503526203e-06,
+      "loss": 0.4589,
+      "step": 17522
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.823113441467285,
+      "learning_rate": 8.386050511263238e-06,
+      "loss": 0.7628,
+      "step": 17523
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 3.9562394618988037,
+      "learning_rate": 8.385856509584935e-06,
+      "loss": 0.6877,
+      "step": 17524
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.472161293029785,
+      "learning_rate": 8.385662498491833e-06,
+      "loss": 0.81,
+      "step": 17525
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 3.703198194503784,
+      "learning_rate": 8.385468477984472e-06,
+      "loss": 0.6668,
+      "step": 17526
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 3.301215887069702,
+      "learning_rate": 8.385274448063392e-06,
+      "loss": 0.9454,
+      "step": 17527
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 3.3559820652008057,
+      "learning_rate": 8.385080408729132e-06,
+      "loss": 0.9543,
+      "step": 17528
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 4.220170974731445,
+      "learning_rate": 8.38488635998223e-06,
+      "loss": 0.6394,
+      "step": 17529
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 3.2408864498138428,
+      "learning_rate": 8.384692301823227e-06,
+      "loss": 0.8395,
+      "step": 17530
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 4.082276344299316,
+      "learning_rate": 8.384498234252664e-06,
+      "loss": 0.3649,
+      "step": 17531
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.7145711183547974,
+      "learning_rate": 8.384304157271079e-06,
+      "loss": 0.5024,
+      "step": 17532
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.932374119758606,
+      "learning_rate": 8.384110070879013e-06,
+      "loss": 0.4603,
+      "step": 17533
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.3500761985778809,
+      "learning_rate": 8.383915975077004e-06,
+      "loss": 0.8014,
+      "step": 17534
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 3.0362532138824463,
+      "learning_rate": 8.383721869865591e-06,
+      "loss": 0.6961,
+      "step": 17535
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.7245309352874756,
+      "learning_rate": 8.383527755245315e-06,
+      "loss": 0.7072,
+      "step": 17536
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.0470364093780518,
+      "learning_rate": 8.383333631216716e-06,
+      "loss": 0.7226,
+      "step": 17537
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.215527057647705,
+      "learning_rate": 8.383139497780334e-06,
+      "loss": 0.4693,
+      "step": 17538
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 3.9288008213043213,
+      "learning_rate": 8.38294535493671e-06,
+      "loss": 0.528,
+      "step": 17539
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.7838455438613892,
+      "learning_rate": 8.382751202686382e-06,
+      "loss": 0.759,
+      "step": 17540
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.5976946353912354,
+      "learning_rate": 8.382557041029889e-06,
+      "loss": 0.9249,
+      "step": 17541
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 3.274867296218872,
+      "learning_rate": 8.382362869967771e-06,
+      "loss": 0.6766,
+      "step": 17542
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.1796083450317383,
+      "learning_rate": 8.382168689500569e-06,
+      "loss": 0.8598,
+      "step": 17543
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.9009021520614624,
+      "learning_rate": 8.381974499628824e-06,
+      "loss": 0.9858,
+      "step": 17544
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.3635449409484863,
+      "learning_rate": 8.381780300353076e-06,
+      "loss": 0.6982,
+      "step": 17545
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.0094921588897705,
+      "learning_rate": 8.38158609167386e-06,
+      "loss": 0.7334,
+      "step": 17546
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.500506639480591,
+      "learning_rate": 8.381391873591724e-06,
+      "loss": 0.7793,
+      "step": 17547
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.7621941566467285,
+      "learning_rate": 8.381197646107202e-06,
+      "loss": 0.7349,
+      "step": 17548
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.8842275142669678,
+      "learning_rate": 8.381003409220836e-06,
+      "loss": 0.5477,
+      "step": 17549
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 5.508273124694824,
+      "learning_rate": 8.380809162933166e-06,
+      "loss": 0.7321,
+      "step": 17550
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.9106037616729736,
+      "learning_rate": 8.380614907244732e-06,
+      "loss": 0.5947,
+      "step": 17551
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.9552781581878662,
+      "learning_rate": 8.380420642156076e-06,
+      "loss": 0.6065,
+      "step": 17552
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 3.1436409950256348,
+      "learning_rate": 8.380226367667735e-06,
+      "loss": 0.8569,
+      "step": 17553
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 3.2870900630950928,
+      "learning_rate": 8.380032083780251e-06,
+      "loss": 0.826,
+      "step": 17554
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.0422677993774414,
+      "learning_rate": 8.379837790494164e-06,
+      "loss": 0.5701,
+      "step": 17555
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.287379503250122,
+      "learning_rate": 8.379643487810014e-06,
+      "loss": 0.7113,
+      "step": 17556
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.7626020908355713,
+      "learning_rate": 8.37944917572834e-06,
+      "loss": 0.596,
+      "step": 17557
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.7230308055877686,
+      "learning_rate": 8.379254854249687e-06,
+      "loss": 0.6976,
+      "step": 17558
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 3.182924747467041,
+      "learning_rate": 8.379060523374592e-06,
+      "loss": 0.5059,
+      "step": 17559
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.368217706680298,
+      "learning_rate": 8.378866183103594e-06,
+      "loss": 0.6806,
+      "step": 17560
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.8934694528579712,
+      "learning_rate": 8.378671833437234e-06,
+      "loss": 0.7174,
+      "step": 17561
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.5977911949157715,
+      "learning_rate": 8.378477474376055e-06,
+      "loss": 0.7641,
+      "step": 17562
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.3576651811599731,
+      "learning_rate": 8.378283105920595e-06,
+      "loss": 0.4699,
+      "step": 17563
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 3.3040108680725098,
+      "learning_rate": 8.378088728071396e-06,
+      "loss": 0.7623,
+      "step": 17564
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.92657470703125,
+      "learning_rate": 8.377894340828995e-06,
+      "loss": 0.4549,
+      "step": 17565
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 6.045033931732178,
+      "learning_rate": 8.377699944193938e-06,
+      "loss": 0.8516,
+      "step": 17566
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.7285666465759277,
+      "learning_rate": 8.377505538166763e-06,
+      "loss": 0.6555,
+      "step": 17567
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.6626110076904297,
+      "learning_rate": 8.377311122748009e-06,
+      "loss": 0.7428,
+      "step": 17568
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 3.4505741596221924,
+      "learning_rate": 8.377116697938218e-06,
+      "loss": 0.6105,
+      "step": 17569
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.641472816467285,
+      "learning_rate": 8.376922263737933e-06,
+      "loss": 0.5608,
+      "step": 17570
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.6810389757156372,
+      "learning_rate": 8.376727820147689e-06,
+      "loss": 0.6975,
+      "step": 17571
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.0531551837921143,
+      "learning_rate": 8.37653336716803e-06,
+      "loss": 0.6308,
+      "step": 17572
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.2762131690979004,
+      "learning_rate": 8.376338904799499e-06,
+      "loss": 0.4899,
+      "step": 17573
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.9394371509552,
+      "learning_rate": 8.376144433042633e-06,
+      "loss": 0.5939,
+      "step": 17574
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.8550293445587158,
+      "learning_rate": 8.375949951897974e-06,
+      "loss": 0.2932,
+      "step": 17575
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.1132287979125977,
+      "learning_rate": 8.375755461366062e-06,
+      "loss": 0.387,
+      "step": 17576
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.9134443998336792,
+      "learning_rate": 8.37556096144744e-06,
+      "loss": 0.7373,
+      "step": 17577
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 3.8796849250793457,
+      "learning_rate": 8.375366452142646e-06,
+      "loss": 0.8634,
+      "step": 17578
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 4.269119739532471,
+      "learning_rate": 8.375171933452224e-06,
+      "loss": 0.996,
+      "step": 17579
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.8573808670043945,
+      "learning_rate": 8.374977405376712e-06,
+      "loss": 0.5671,
+      "step": 17580
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.0454957485198975,
+      "learning_rate": 8.374782867916654e-06,
+      "loss": 0.5606,
+      "step": 17581
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.090325355529785,
+      "learning_rate": 8.374588321072585e-06,
+      "loss": 0.7369,
+      "step": 17582
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.1618340015411377,
+      "learning_rate": 8.374393764845054e-06,
+      "loss": 0.8741,
+      "step": 17583
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 3.1730916500091553,
+      "learning_rate": 8.374199199234596e-06,
+      "loss": 0.7269,
+      "step": 17584
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.7430055141448975,
+      "learning_rate": 8.374004624241756e-06,
+      "loss": 0.8193,
+      "step": 17585
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.210421323776245,
+      "learning_rate": 8.373810039867071e-06,
+      "loss": 0.638,
+      "step": 17586
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.474440574645996,
+      "learning_rate": 8.373615446111086e-06,
+      "loss": 0.7069,
+      "step": 17587
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 3.7259891033172607,
+      "learning_rate": 8.373420842974338e-06,
+      "loss": 0.8489,
+      "step": 17588
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.9619220495224,
+      "learning_rate": 8.373226230457373e-06,
+      "loss": 0.6815,
+      "step": 17589
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.882288932800293,
+      "learning_rate": 8.373031608560727e-06,
+      "loss": 0.7478,
+      "step": 17590
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.9391672611236572,
+      "learning_rate": 8.372836977284945e-06,
+      "loss": 0.4568,
+      "step": 17591
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.638256788253784,
+      "learning_rate": 8.372642336630566e-06,
+      "loss": 0.5883,
+      "step": 17592
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.5008450746536255,
+      "learning_rate": 8.372447686598132e-06,
+      "loss": 0.7229,
+      "step": 17593
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.6294586658477783,
+      "learning_rate": 8.372253027188186e-06,
+      "loss": 0.818,
+      "step": 17594
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.752957582473755,
+      "learning_rate": 8.372058358401267e-06,
+      "loss": 0.7505,
+      "step": 17595
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.8324224948883057,
+      "learning_rate": 8.371863680237917e-06,
+      "loss": 0.6007,
+      "step": 17596
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.816926956176758,
+      "learning_rate": 8.371668992698676e-06,
+      "loss": 0.6546,
+      "step": 17597
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 3.7511942386627197,
+      "learning_rate": 8.371474295784086e-06,
+      "loss": 0.9435,
+      "step": 17598
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.6705883741378784,
+      "learning_rate": 8.371279589494692e-06,
+      "loss": 0.5899,
+      "step": 17599
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 3.450885057449341,
+      "learning_rate": 8.37108487383103e-06,
+      "loss": 0.4098,
+      "step": 17600
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.9021482467651367,
+      "learning_rate": 8.370890148793644e-06,
+      "loss": 0.5545,
+      "step": 17601
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.464097738265991,
+      "learning_rate": 8.370695414383077e-06,
+      "loss": 0.5459,
+      "step": 17602
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.3746156692504883,
+      "learning_rate": 8.370500670599868e-06,
+      "loss": 0.7814,
+      "step": 17603
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.0794596672058105,
+      "learning_rate": 8.370305917444557e-06,
+      "loss": 0.7353,
+      "step": 17604
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.6465704441070557,
+      "learning_rate": 8.370111154917691e-06,
+      "loss": 0.8018,
+      "step": 17605
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.1669466495513916,
+      "learning_rate": 8.369916383019808e-06,
+      "loss": 0.7135,
+      "step": 17606
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.9681676626205444,
+      "learning_rate": 8.369721601751447e-06,
+      "loss": 0.5397,
+      "step": 17607
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 3.336510181427002,
+      "learning_rate": 8.369526811113153e-06,
+      "loss": 0.5439,
+      "step": 17608
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.745442271232605,
+      "learning_rate": 8.369332011105468e-06,
+      "loss": 0.4907,
+      "step": 17609
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.0137124061584473,
+      "learning_rate": 8.369137201728935e-06,
+      "loss": 0.9124,
+      "step": 17610
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.379112958908081,
+      "learning_rate": 8.368942382984092e-06,
+      "loss": 0.4681,
+      "step": 17611
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.2177083492279053,
+      "learning_rate": 8.368747554871482e-06,
+      "loss": 0.6141,
+      "step": 17612
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.137359857559204,
+      "learning_rate": 8.368552717391648e-06,
+      "loss": 0.4476,
+      "step": 17613
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.010369062423706,
+      "learning_rate": 8.36835787054513e-06,
+      "loss": 0.5363,
+      "step": 17614
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.7531373500823975,
+      "learning_rate": 8.36816301433247e-06,
+      "loss": 0.473,
+      "step": 17615
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.1216728687286377,
+      "learning_rate": 8.367968148754211e-06,
+      "loss": 0.5796,
+      "step": 17616
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.5359554290771484,
+      "learning_rate": 8.367773273810894e-06,
+      "loss": 0.7188,
+      "step": 17617
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.5275802612304688,
+      "learning_rate": 8.367578389503062e-06,
+      "loss": 0.5006,
+      "step": 17618
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.1476855278015137,
+      "learning_rate": 8.367383495831253e-06,
+      "loss": 0.6719,
+      "step": 17619
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.5636703968048096,
+      "learning_rate": 8.367188592796016e-06,
+      "loss": 0.6677,
+      "step": 17620
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.846057653427124,
+      "learning_rate": 8.366993680397887e-06,
+      "loss": 0.7681,
+      "step": 17621
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.9266852140426636,
+      "learning_rate": 8.366798758637411e-06,
+      "loss": 0.7219,
+      "step": 17622
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 3.128204345703125,
+      "learning_rate": 8.366603827515127e-06,
+      "loss": 0.4865,
+      "step": 17623
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.8663923740386963,
+      "learning_rate": 8.366408887031581e-06,
+      "loss": 0.9315,
+      "step": 17624
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.1062426567077637,
+      "learning_rate": 8.36621393718731e-06,
+      "loss": 0.6089,
+      "step": 17625
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 3.528560161590576,
+      "learning_rate": 8.366018977982862e-06,
+      "loss": 0.6676,
+      "step": 17626
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 3.5607192516326904,
+      "learning_rate": 8.365824009418775e-06,
+      "loss": 1.0092,
+      "step": 17627
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.7381551265716553,
+      "learning_rate": 8.365629031495593e-06,
+      "loss": 0.8235,
+      "step": 17628
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.6760826110839844,
+      "learning_rate": 8.365434044213855e-06,
+      "loss": 0.851,
+      "step": 17629
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.6019178628921509,
+      "learning_rate": 8.36523904757411e-06,
+      "loss": 0.6506,
+      "step": 17630
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.7602885961532593,
+      "learning_rate": 8.365044041576892e-06,
+      "loss": 0.6985,
+      "step": 17631
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.0855464935302734,
+      "learning_rate": 8.364849026222748e-06,
+      "loss": 0.811,
+      "step": 17632
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.1494044065475464,
+      "learning_rate": 8.36465400151222e-06,
+      "loss": 0.4325,
+      "step": 17633
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.7661399841308594,
+      "learning_rate": 8.36445896744585e-06,
+      "loss": 0.7356,
+      "step": 17634
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.1450271606445312,
+      "learning_rate": 8.36426392402418e-06,
+      "loss": 0.588,
+      "step": 17635
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 3.448368787765503,
+      "learning_rate": 8.364068871247753e-06,
+      "loss": 1.1298,
+      "step": 17636
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.3023440837860107,
+      "learning_rate": 8.363873809117109e-06,
+      "loss": 0.596,
+      "step": 17637
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.4756929874420166,
+      "learning_rate": 8.363678737632792e-06,
+      "loss": 0.7224,
+      "step": 17638
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.6235935688018799,
+      "learning_rate": 8.363483656795343e-06,
+      "loss": 0.7633,
+      "step": 17639
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.229569673538208,
+      "learning_rate": 8.36328856660531e-06,
+      "loss": 0.9086,
+      "step": 17640
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.4486947059631348,
+      "learning_rate": 8.363093467063227e-06,
+      "loss": 0.7445,
+      "step": 17641
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.053844451904297,
+      "learning_rate": 8.362898358169644e-06,
+      "loss": 0.9809,
+      "step": 17642
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.7188129425048828,
+      "learning_rate": 8.362703239925101e-06,
+      "loss": 0.5714,
+      "step": 17643
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.8913483619689941,
+      "learning_rate": 8.362508112330138e-06,
+      "loss": 0.664,
+      "step": 17644
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.0054593086242676,
+      "learning_rate": 8.3623129753853e-06,
+      "loss": 0.5027,
+      "step": 17645
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.779098629951477,
+      "learning_rate": 8.36211782909113e-06,
+      "loss": 0.6862,
+      "step": 17646
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 3.936445713043213,
+      "learning_rate": 8.361922673448168e-06,
+      "loss": 0.6901,
+      "step": 17647
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.4161717891693115,
+      "learning_rate": 8.36172750845696e-06,
+      "loss": 0.4605,
+      "step": 17648
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.7755292654037476,
+      "learning_rate": 8.361532334118047e-06,
+      "loss": 0.827,
+      "step": 17649
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.9369680881500244,
+      "learning_rate": 8.361337150431973e-06,
+      "loss": 0.6522,
+      "step": 17650
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.880060076713562,
+      "learning_rate": 8.361141957399278e-06,
+      "loss": 0.5377,
+      "step": 17651
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 3.465144634246826,
+      "learning_rate": 8.360946755020504e-06,
+      "loss": 0.7716,
+      "step": 17652
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.6295568943023682,
+      "learning_rate": 8.3607515432962e-06,
+      "loss": 0.7991,
+      "step": 17653
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.7001243829727173,
+      "learning_rate": 8.360556322226904e-06,
+      "loss": 0.7388,
+      "step": 17654
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.608327627182007,
+      "learning_rate": 8.360361091813158e-06,
+      "loss": 0.3886,
+      "step": 17655
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 3.4775705337524414,
+      "learning_rate": 8.360165852055508e-06,
+      "loss": 0.6654,
+      "step": 17656
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.3635472059249878,
+      "learning_rate": 8.359970602954497e-06,
+      "loss": 0.693,
+      "step": 17657
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 4.354608058929443,
+      "learning_rate": 8.359775344510664e-06,
+      "loss": 0.7145,
+      "step": 17658
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.981109619140625,
+      "learning_rate": 8.359580076724555e-06,
+      "loss": 0.7087,
+      "step": 17659
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.7548943758010864,
+      "learning_rate": 8.359384799596713e-06,
+      "loss": 0.5822,
+      "step": 17660
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.9847512245178223,
+      "learning_rate": 8.359189513127678e-06,
+      "loss": 0.4746,
+      "step": 17661
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 4.215170860290527,
+      "learning_rate": 8.358994217318e-06,
+      "loss": 0.5299,
+      "step": 17662
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.770280361175537,
+      "learning_rate": 8.358798912168214e-06,
+      "loss": 0.7049,
+      "step": 17663
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.237293004989624,
+      "learning_rate": 8.358603597678867e-06,
+      "loss": 0.8027,
+      "step": 17664
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 3.4375078678131104,
+      "learning_rate": 8.358408273850502e-06,
+      "loss": 0.7615,
+      "step": 17665
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.6242141723632812,
+      "learning_rate": 8.358212940683661e-06,
+      "loss": 0.6286,
+      "step": 17666
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 3.034151315689087,
+      "learning_rate": 8.358017598178889e-06,
+      "loss": 0.8522,
+      "step": 17667
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 4.07211971282959,
+      "learning_rate": 8.357822246336727e-06,
+      "loss": 0.6581,
+      "step": 17668
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.6394155025482178,
+      "learning_rate": 8.35762688515772e-06,
+      "loss": 0.4891,
+      "step": 17669
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.1299948692321777,
+      "learning_rate": 8.35743151464241e-06,
+      "loss": 0.7354,
+      "step": 17670
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.7681589126586914,
+      "learning_rate": 8.357236134791338e-06,
+      "loss": 0.58,
+      "step": 17671
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 4.810450553894043,
+      "learning_rate": 8.357040745605054e-06,
+      "loss": 0.8079,
+      "step": 17672
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.4689733982086182,
+      "learning_rate": 8.356845347084096e-06,
+      "loss": 0.5343,
+      "step": 17673
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.7519859075546265,
+      "learning_rate": 8.356649939229007e-06,
+      "loss": 0.7259,
+      "step": 17674
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.730175018310547,
+      "learning_rate": 8.356454522040334e-06,
+      "loss": 0.6642,
+      "step": 17675
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.354825735092163,
+      "learning_rate": 8.356259095518617e-06,
+      "loss": 0.4638,
+      "step": 17676
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.457932710647583,
+      "learning_rate": 8.356063659664402e-06,
+      "loss": 0.4527,
+      "step": 17677
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.1829111576080322,
+      "learning_rate": 8.355868214478229e-06,
+      "loss": 0.7199,
+      "step": 17678
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.4978296756744385,
+      "learning_rate": 8.355672759960646e-06,
+      "loss": 0.8907,
+      "step": 17679
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.0817742347717285,
+      "learning_rate": 8.355477296112193e-06,
+      "loss": 0.6224,
+      "step": 17680
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 3.394174575805664,
+      "learning_rate": 8.355281822933414e-06,
+      "loss": 0.7507,
+      "step": 17681
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.4968764781951904,
+      "learning_rate": 8.355086340424855e-06,
+      "loss": 0.8478,
+      "step": 17682
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 5.480932235717773,
+      "learning_rate": 8.354890848587056e-06,
+      "loss": 0.754,
+      "step": 17683
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.2805376052856445,
+      "learning_rate": 8.354695347420565e-06,
+      "loss": 0.5796,
+      "step": 17684
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 3.0068767070770264,
+      "learning_rate": 8.35449983692592e-06,
+      "loss": 0.5658,
+      "step": 17685
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 4.418600082397461,
+      "learning_rate": 8.354304317103668e-06,
+      "loss": 0.7069,
+      "step": 17686
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.7277922630310059,
+      "learning_rate": 8.354108787954353e-06,
+      "loss": 0.8675,
+      "step": 17687
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.518437623977661,
+      "learning_rate": 8.353913249478518e-06,
+      "loss": 0.6602,
+      "step": 17688
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.963657855987549,
+      "learning_rate": 8.353717701676706e-06,
+      "loss": 0.6954,
+      "step": 17689
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.0415351390838623,
+      "learning_rate": 8.353522144549461e-06,
+      "loss": 0.6916,
+      "step": 17690
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.4496475458145142,
+      "learning_rate": 8.35332657809733e-06,
+      "loss": 0.573,
+      "step": 17691
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.6655807495117188,
+      "learning_rate": 8.353131002320851e-06,
+      "loss": 0.7308,
+      "step": 17692
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 3.2632222175598145,
+      "learning_rate": 8.352935417220571e-06,
+      "loss": 0.7531,
+      "step": 17693
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.2352771759033203,
+      "learning_rate": 8.352739822797036e-06,
+      "loss": 0.4604,
+      "step": 17694
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.0265510082244873,
+      "learning_rate": 8.352544219050786e-06,
+      "loss": 0.732,
+      "step": 17695
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.725457191467285,
+      "learning_rate": 8.352348605982366e-06,
+      "loss": 0.7947,
+      "step": 17696
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.5501770973205566,
+      "learning_rate": 8.35215298359232e-06,
+      "loss": 0.8045,
+      "step": 17697
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.7034811973571777,
+      "learning_rate": 8.351957351881194e-06,
+      "loss": 0.729,
+      "step": 17698
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 3.4474174976348877,
+      "learning_rate": 8.351761710849528e-06,
+      "loss": 0.6519,
+      "step": 17699
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 3.1487958431243896,
+      "learning_rate": 8.35156606049787e-06,
+      "loss": 0.8753,
+      "step": 17700
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 6.059223175048828,
+      "learning_rate": 8.351370400826762e-06,
+      "loss": 0.6208,
+      "step": 17701
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.0373501777648926,
+      "learning_rate": 8.351174731836749e-06,
+      "loss": 0.6462,
+      "step": 17702
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.7429373264312744,
+      "learning_rate": 8.350979053528374e-06,
+      "loss": 0.6723,
+      "step": 17703
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.9900109767913818,
+      "learning_rate": 8.35078336590218e-06,
+      "loss": 0.659,
+      "step": 17704
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.6849873065948486,
+      "learning_rate": 8.350587668958714e-06,
+      "loss": 0.5734,
+      "step": 17705
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.1100172996520996,
+      "learning_rate": 8.35039196269852e-06,
+      "loss": 0.5726,
+      "step": 17706
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.4966042041778564,
+      "learning_rate": 8.35019624712214e-06,
+      "loss": 0.7454,
+      "step": 17707
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.4096944332122803,
+      "learning_rate": 8.350000522230116e-06,
+      "loss": 0.6598,
+      "step": 17708
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.430281400680542,
+      "learning_rate": 8.349804788023e-06,
+      "loss": 0.5459,
+      "step": 17709
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 3.180790662765503,
+      "learning_rate": 8.34960904450133e-06,
+      "loss": 0.6159,
+      "step": 17710
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 3.4096102714538574,
+      "learning_rate": 8.349413291665654e-06,
+      "loss": 0.6438,
+      "step": 17711
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 3.5805106163024902,
+      "learning_rate": 8.34921752951651e-06,
+      "loss": 0.4385,
+      "step": 17712
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.6336066722869873,
+      "learning_rate": 8.34902175805445e-06,
+      "loss": 0.572,
+      "step": 17713
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.532815456390381,
+      "learning_rate": 8.348825977280015e-06,
+      "loss": 0.7352,
+      "step": 17714
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 3.3297672271728516,
+      "learning_rate": 8.34863018719375e-06,
+      "loss": 0.8135,
+      "step": 17715
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.6429978609085083,
+      "learning_rate": 8.348434387796197e-06,
+      "loss": 0.5286,
+      "step": 17716
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 3.1213090419769287,
+      "learning_rate": 8.348238579087904e-06,
+      "loss": 0.4323,
+      "step": 17717
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.0143613815307617,
+      "learning_rate": 8.348042761069412e-06,
+      "loss": 0.8016,
+      "step": 17718
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.0551774501800537,
+      "learning_rate": 8.347846933741268e-06,
+      "loss": 0.6397,
+      "step": 17719
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 3.326425313949585,
+      "learning_rate": 8.347651097104015e-06,
+      "loss": 0.9668,
+      "step": 17720
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 3.0984086990356445,
+      "learning_rate": 8.347455251158201e-06,
+      "loss": 0.7582,
+      "step": 17721
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.9345684051513672,
+      "learning_rate": 8.347259395904365e-06,
+      "loss": 0.8366,
+      "step": 17722
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.9499740600585938,
+      "learning_rate": 8.347063531343055e-06,
+      "loss": 0.6815,
+      "step": 17723
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 3.1109185218811035,
+      "learning_rate": 8.346867657474818e-06,
+      "loss": 0.9295,
+      "step": 17724
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 3.1199519634246826,
+      "learning_rate": 8.346671774300193e-06,
+      "loss": 0.6083,
+      "step": 17725
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.533628225326538,
+      "learning_rate": 8.346475881819729e-06,
+      "loss": 0.7027,
+      "step": 17726
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.2792904376983643,
+      "learning_rate": 8.346279980033968e-06,
+      "loss": 0.6396,
+      "step": 17727
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.1187753677368164,
+      "learning_rate": 8.346084068943456e-06,
+      "loss": 0.8744,
+      "step": 17728
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.9713125228881836,
+      "learning_rate": 8.345888148548738e-06,
+      "loss": 0.4264,
+      "step": 17729
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.5830212831497192,
+      "learning_rate": 8.34569221885036e-06,
+      "loss": 0.7099,
+      "step": 17730
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 3.862689256668091,
+      "learning_rate": 8.345496279848863e-06,
+      "loss": 0.591,
+      "step": 17731
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.417726993560791,
+      "learning_rate": 8.345300331544794e-06,
+      "loss": 0.5329,
+      "step": 17732
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.8745808601379395,
+      "learning_rate": 8.3451043739387e-06,
+      "loss": 0.9685,
+      "step": 17733
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.4113879203796387,
+      "learning_rate": 8.344908407031123e-06,
+      "loss": 0.6528,
+      "step": 17734
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.661957025527954,
+      "learning_rate": 8.344712430822608e-06,
+      "loss": 0.6189,
+      "step": 17735
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.499997138977051,
+      "learning_rate": 8.344516445313701e-06,
+      "loss": 0.5834,
+      "step": 17736
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.9463014602661133,
+      "learning_rate": 8.344320450504946e-06,
+      "loss": 0.546,
+      "step": 17737
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 5.115029335021973,
+      "learning_rate": 8.344124446396891e-06,
+      "loss": 0.9191,
+      "step": 17738
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.7503355741500854,
+      "learning_rate": 8.343928432990076e-06,
+      "loss": 0.649,
+      "step": 17739
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.91428804397583,
+      "learning_rate": 8.34373241028505e-06,
+      "loss": 0.7138,
+      "step": 17740
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.0028107166290283,
+      "learning_rate": 8.343536378282357e-06,
+      "loss": 0.7426,
+      "step": 17741
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.6038603782653809,
+      "learning_rate": 8.34334033698254e-06,
+      "loss": 0.4228,
+      "step": 17742
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 3.086641788482666,
+      "learning_rate": 8.343144286386148e-06,
+      "loss": 0.6863,
+      "step": 17743
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.2865724563598633,
+      "learning_rate": 8.342948226493724e-06,
+      "loss": 0.5614,
+      "step": 17744
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 6.2041754722595215,
+      "learning_rate": 8.342752157305812e-06,
+      "loss": 0.5305,
+      "step": 17745
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.55411958694458,
+      "learning_rate": 8.342556078822958e-06,
+      "loss": 0.552,
+      "step": 17746
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 4.128890514373779,
+      "learning_rate": 8.34235999104571e-06,
+      "loss": 0.6949,
+      "step": 17747
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.3661835193634033,
+      "learning_rate": 8.34216389397461e-06,
+      "loss": 0.5938,
+      "step": 17748
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.5325547456741333,
+      "learning_rate": 8.341967787610203e-06,
+      "loss": 0.3189,
+      "step": 17749
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 3.53054141998291,
+      "learning_rate": 8.341771671953037e-06,
+      "loss": 0.6396,
+      "step": 17750
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.1484053134918213,
+      "learning_rate": 8.341575547003655e-06,
+      "loss": 0.7102,
+      "step": 17751
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.9458935260772705,
+      "learning_rate": 8.341379412762605e-06,
+      "loss": 0.7002,
+      "step": 17752
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.469888687133789,
+      "learning_rate": 8.341183269230428e-06,
+      "loss": 0.6245,
+      "step": 17753
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 3.4671428203582764,
+      "learning_rate": 8.340987116407673e-06,
+      "loss": 0.9012,
+      "step": 17754
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.4088025093078613,
+      "learning_rate": 8.340790954294887e-06,
+      "loss": 0.735,
+      "step": 17755
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.510098457336426,
+      "learning_rate": 8.340594782892609e-06,
+      "loss": 0.6678,
+      "step": 17756
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.4475948810577393,
+      "learning_rate": 8.34039860220139e-06,
+      "loss": 0.6382,
+      "step": 17757
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.6302490234375,
+      "learning_rate": 8.340202412221774e-06,
+      "loss": 0.6791,
+      "step": 17758
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.8497397899627686,
+      "learning_rate": 8.340006212954306e-06,
+      "loss": 0.9321,
+      "step": 17759
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.5606075525283813,
+      "learning_rate": 8.33981000439953e-06,
+      "loss": 0.5132,
+      "step": 17760
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.7964327335357666,
+      "learning_rate": 8.339613786557996e-06,
+      "loss": 0.5728,
+      "step": 17761
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.1738858222961426,
+      "learning_rate": 8.339417559430247e-06,
+      "loss": 0.5397,
+      "step": 17762
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.2584800720214844,
+      "learning_rate": 8.339221323016828e-06,
+      "loss": 0.5263,
+      "step": 17763
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 3.491586685180664,
+      "learning_rate": 8.339025077318285e-06,
+      "loss": 0.7137,
+      "step": 17764
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.1273417472839355,
+      "learning_rate": 8.338828822335164e-06,
+      "loss": 0.8171,
+      "step": 17765
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.6712898015975952,
+      "learning_rate": 8.338632558068012e-06,
+      "loss": 0.8462,
+      "step": 17766
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.77953839302063,
+      "learning_rate": 8.338436284517372e-06,
+      "loss": 0.5948,
+      "step": 17767
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.8778976202011108,
+      "learning_rate": 8.338240001683793e-06,
+      "loss": 0.6412,
+      "step": 17768
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.2099509239196777,
+      "learning_rate": 8.338043709567818e-06,
+      "loss": 0.7138,
+      "step": 17769
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.9963228702545166,
+      "learning_rate": 8.337847408169995e-06,
+      "loss": 0.6579,
+      "step": 17770
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.08872652053833,
+      "learning_rate": 8.337651097490867e-06,
+      "loss": 0.517,
+      "step": 17771
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.4464006423950195,
+      "learning_rate": 8.337454777530982e-06,
+      "loss": 0.6203,
+      "step": 17772
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.413609266281128,
+      "learning_rate": 8.337258448290887e-06,
+      "loss": 0.7097,
+      "step": 17773
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.305311679840088,
+      "learning_rate": 8.337062109771126e-06,
+      "loss": 0.6522,
+      "step": 17774
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.7339251041412354,
+      "learning_rate": 8.336865761972244e-06,
+      "loss": 0.8448,
+      "step": 17775
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.171144723892212,
+      "learning_rate": 8.33666940489479e-06,
+      "loss": 0.6274,
+      "step": 17776
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 3.3733737468719482,
+      "learning_rate": 8.336473038539308e-06,
+      "loss": 0.6298,
+      "step": 17777
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.711130976676941,
+      "learning_rate": 8.336276662906343e-06,
+      "loss": 0.5914,
+      "step": 17778
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 3.270907402038574,
+      "learning_rate": 8.336080277996444e-06,
+      "loss": 0.6466,
+      "step": 17779
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.037614107131958,
+      "learning_rate": 8.335883883810153e-06,
+      "loss": 0.6299,
+      "step": 17780
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.7519071102142334,
+      "learning_rate": 8.33568748034802e-06,
+      "loss": 0.571,
+      "step": 17781
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.655468463897705,
+      "learning_rate": 8.33549106761059e-06,
+      "loss": 0.6884,
+      "step": 17782
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 3.9960296154022217,
+      "learning_rate": 8.33529464559841e-06,
+      "loss": 0.8809,
+      "step": 17783
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 3.126445770263672,
+      "learning_rate": 8.335098214312025e-06,
+      "loss": 0.5367,
+      "step": 17784
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.600172281265259,
+      "learning_rate": 8.33490177375198e-06,
+      "loss": 0.8408,
+      "step": 17785
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 4.225794792175293,
+      "learning_rate": 8.334705323918821e-06,
+      "loss": 0.6538,
+      "step": 17786
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 3.643951416015625,
+      "learning_rate": 8.334508864813099e-06,
+      "loss": 0.9304,
+      "step": 17787
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 3.086089611053467,
+      "learning_rate": 8.334312396435355e-06,
+      "loss": 0.4239,
+      "step": 17788
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.9744750261306763,
+      "learning_rate": 8.334115918786136e-06,
+      "loss": 0.5108,
+      "step": 17789
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.478771209716797,
+      "learning_rate": 8.333919431865991e-06,
+      "loss": 0.5139,
+      "step": 17790
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 4.147990703582764,
+      "learning_rate": 8.333722935675465e-06,
+      "loss": 0.7174,
+      "step": 17791
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.5038883686065674,
+      "learning_rate": 8.333526430215105e-06,
+      "loss": 0.6909,
+      "step": 17792
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.1483771800994873,
+      "learning_rate": 8.333329915485457e-06,
+      "loss": 0.67,
+      "step": 17793
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.1365268230438232,
+      "learning_rate": 8.333133391487066e-06,
+      "loss": 0.8253,
+      "step": 17794
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.0321671962738037,
+      "learning_rate": 8.33293685822048e-06,
+      "loss": 0.499,
+      "step": 17795
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 3.7810418605804443,
+      "learning_rate": 8.332740315686245e-06,
+      "loss": 0.6748,
+      "step": 17796
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 3.0512096881866455,
+      "learning_rate": 8.332543763884908e-06,
+      "loss": 0.7742,
+      "step": 17797
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.0605928897857666,
+      "learning_rate": 8.332347202817015e-06,
+      "loss": 0.638,
+      "step": 17798
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.2239983081817627,
+      "learning_rate": 8.332150632483113e-06,
+      "loss": 0.6669,
+      "step": 17799
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.3308606147766113,
+      "learning_rate": 8.331954052883748e-06,
+      "loss": 0.5065,
+      "step": 17800
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 4.173402786254883,
+      "learning_rate": 8.331757464019468e-06,
+      "loss": 0.5887,
+      "step": 17801
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.2865889072418213,
+      "learning_rate": 8.331560865890818e-06,
+      "loss": 0.6118,
+      "step": 17802
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.094006061553955,
+      "learning_rate": 8.331364258498344e-06,
+      "loss": 0.4629,
+      "step": 17803
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.5289711952209473,
+      "learning_rate": 8.331167641842594e-06,
+      "loss": 0.8978,
+      "step": 17804
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.189444065093994,
+      "learning_rate": 8.330971015924117e-06,
+      "loss": 0.5646,
+      "step": 17805
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 4.146525859832764,
+      "learning_rate": 8.330774380743457e-06,
+      "loss": 0.5742,
+      "step": 17806
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 4.001610279083252,
+      "learning_rate": 8.33057773630116e-06,
+      "loss": 0.5701,
+      "step": 17807
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.2702934741973877,
+      "learning_rate": 8.330381082597775e-06,
+      "loss": 0.6547,
+      "step": 17808
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.232515335083008,
+      "learning_rate": 8.330184419633847e-06,
+      "loss": 0.8646,
+      "step": 17809
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.6136374473571777,
+      "learning_rate": 8.329987747409922e-06,
+      "loss": 0.4535,
+      "step": 17810
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 3.1614022254943848,
+      "learning_rate": 8.32979106592655e-06,
+      "loss": 0.5776,
+      "step": 17811
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 4.110596656799316,
+      "learning_rate": 8.329594375184277e-06,
+      "loss": 0.7568,
+      "step": 17812
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.69608473777771,
+      "learning_rate": 8.329397675183647e-06,
+      "loss": 0.6761,
+      "step": 17813
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 3.367396593093872,
+      "learning_rate": 8.329200965925212e-06,
+      "loss": 0.6325,
+      "step": 17814
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.204092264175415,
+      "learning_rate": 8.329004247409517e-06,
+      "loss": 0.5776,
+      "step": 17815
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.802573323249817,
+      "learning_rate": 8.328807519637106e-06,
+      "loss": 0.5547,
+      "step": 17816
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.2243402004241943,
+      "learning_rate": 8.32861078260853e-06,
+      "loss": 0.6644,
+      "step": 17817
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 3.253085136413574,
+      "learning_rate": 8.328414036324333e-06,
+      "loss": 0.9335,
+      "step": 17818
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.7816389799118042,
+      "learning_rate": 8.328217280785065e-06,
+      "loss": 0.5726,
+      "step": 17819
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.3741483688354492,
+      "learning_rate": 8.328020515991269e-06,
+      "loss": 0.6066,
+      "step": 17820
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.3499177694320679,
+      "learning_rate": 8.327823741943495e-06,
+      "loss": 0.3368,
+      "step": 17821
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.9078502655029297,
+      "learning_rate": 8.327626958642292e-06,
+      "loss": 0.8393,
+      "step": 17822
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.509786605834961,
+      "learning_rate": 8.327430166088204e-06,
+      "loss": 0.7037,
+      "step": 17823
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.2095537185668945,
+      "learning_rate": 8.327233364281778e-06,
+      "loss": 0.5662,
+      "step": 17824
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 3.3533666133880615,
+      "learning_rate": 8.327036553223564e-06,
+      "loss": 0.6489,
+      "step": 17825
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.9601086378097534,
+      "learning_rate": 8.326839732914107e-06,
+      "loss": 0.7763,
+      "step": 17826
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.6180872917175293,
+      "learning_rate": 8.326642903353954e-06,
+      "loss": 0.794,
+      "step": 17827
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 3.1983535289764404,
+      "learning_rate": 8.326446064543654e-06,
+      "loss": 0.5521,
+      "step": 17828
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.8526277542114258,
+      "learning_rate": 8.326249216483755e-06,
+      "loss": 0.8312,
+      "step": 17829
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.079559087753296,
+      "learning_rate": 8.3260523591748e-06,
+      "loss": 0.7227,
+      "step": 17830
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.5995378494262695,
+      "learning_rate": 8.325855492617343e-06,
+      "loss": 0.8613,
+      "step": 17831
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.216712474822998,
+      "learning_rate": 8.325658616811926e-06,
+      "loss": 0.8226,
+      "step": 17832
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 3.0803701877593994,
+      "learning_rate": 8.325461731759099e-06,
+      "loss": 0.797,
+      "step": 17833
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.6251702308654785,
+      "learning_rate": 8.325264837459408e-06,
+      "loss": 0.7958,
+      "step": 17834
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.531911611557007,
+      "learning_rate": 8.3250679339134e-06,
+      "loss": 0.8504,
+      "step": 17835
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 3.5326120853424072,
+      "learning_rate": 8.324871021121625e-06,
+      "loss": 0.5749,
+      "step": 17836
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.7098114490509033,
+      "learning_rate": 8.324674099084629e-06,
+      "loss": 0.7144,
+      "step": 17837
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.176626205444336,
+      "learning_rate": 8.32447716780296e-06,
+      "loss": 0.5937,
+      "step": 17838
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.8586026430130005,
+      "learning_rate": 8.324280227277164e-06,
+      "loss": 0.7766,
+      "step": 17839
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 5.304523944854736,
+      "learning_rate": 8.324083277507791e-06,
+      "loss": 0.9229,
+      "step": 17840
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.134490489959717,
+      "learning_rate": 8.323886318495388e-06,
+      "loss": 0.4347,
+      "step": 17841
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.690821886062622,
+      "learning_rate": 8.323689350240503e-06,
+      "loss": 0.8518,
+      "step": 17842
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.2803964614868164,
+      "learning_rate": 8.32349237274368e-06,
+      "loss": 0.4962,
+      "step": 17843
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.6231095790863037,
+      "learning_rate": 8.323295386005472e-06,
+      "loss": 0.6954,
+      "step": 17844
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.7994544506072998,
+      "learning_rate": 8.323098390026425e-06,
+      "loss": 0.5596,
+      "step": 17845
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.6917986869812012,
+      "learning_rate": 8.322901384807086e-06,
+      "loss": 0.7232,
+      "step": 17846
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.5118393898010254,
+      "learning_rate": 8.322704370348002e-06,
+      "loss": 0.689,
+      "step": 17847
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.3650355339050293,
+      "learning_rate": 8.322507346649722e-06,
+      "loss": 0.4294,
+      "step": 17848
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 3.2000844478607178,
+      "learning_rate": 8.322310313712795e-06,
+      "loss": 0.8143,
+      "step": 17849
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.3423819541931152,
+      "learning_rate": 8.322113271537767e-06,
+      "loss": 0.4636,
+      "step": 17850
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.0954936742782593,
+      "learning_rate": 8.321916220125185e-06,
+      "loss": 0.5912,
+      "step": 17851
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 3.5825579166412354,
+      "learning_rate": 8.321719159475598e-06,
+      "loss": 0.729,
+      "step": 17852
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.574758291244507,
+      "learning_rate": 8.321522089589558e-06,
+      "loss": 0.6893,
+      "step": 17853
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.03100848197937,
+      "learning_rate": 8.321325010467607e-06,
+      "loss": 0.5985,
+      "step": 17854
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 3.015316963195801,
+      "learning_rate": 8.321127922110295e-06,
+      "loss": 0.8953,
+      "step": 17855
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 4.170287609100342,
+      "learning_rate": 8.320930824518172e-06,
+      "loss": 0.6299,
+      "step": 17856
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.872563362121582,
+      "learning_rate": 8.320733717691782e-06,
+      "loss": 0.5352,
+      "step": 17857
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.1040422916412354,
+      "learning_rate": 8.320536601631678e-06,
+      "loss": 0.721,
+      "step": 17858
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.269491195678711,
+      "learning_rate": 8.320339476338404e-06,
+      "loss": 0.8508,
+      "step": 17859
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 3.5985708236694336,
+      "learning_rate": 8.320142341812509e-06,
+      "loss": 0.75,
+      "step": 17860
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.9902470111846924,
+      "learning_rate": 8.319945198054544e-06,
+      "loss": 0.697,
+      "step": 17861
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.2490270137786865,
+      "learning_rate": 8.319748045065055e-06,
+      "loss": 0.7022,
+      "step": 17862
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.197226047515869,
+      "learning_rate": 8.31955088284459e-06,
+      "loss": 0.4895,
+      "step": 17863
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 3.584158420562744,
+      "learning_rate": 8.319353711393696e-06,
+      "loss": 0.4665,
+      "step": 17864
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.5101723670959473,
+      "learning_rate": 8.319156530712924e-06,
+      "loss": 0.4798,
+      "step": 17865
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.780745506286621,
+      "learning_rate": 8.318959340802821e-06,
+      "loss": 0.6433,
+      "step": 17866
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 5.1451568603515625,
+      "learning_rate": 8.318762141663935e-06,
+      "loss": 0.5453,
+      "step": 17867
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.8175623416900635,
+      "learning_rate": 8.318564933296814e-06,
+      "loss": 0.8092,
+      "step": 17868
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 4.200446128845215,
+      "learning_rate": 8.31836771570201e-06,
+      "loss": 0.9207,
+      "step": 17869
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 4.395917892456055,
+      "learning_rate": 8.318170488880065e-06,
+      "loss": 0.7895,
+      "step": 17870
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.4713187217712402,
+      "learning_rate": 8.317973252831532e-06,
+      "loss": 0.4848,
+      "step": 17871
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.975202798843384,
+      "learning_rate": 8.31777600755696e-06,
+      "loss": 0.6997,
+      "step": 17872
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.704564094543457,
+      "learning_rate": 8.317578753056894e-06,
+      "loss": 0.7119,
+      "step": 17873
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.5723657608032227,
+      "learning_rate": 8.317381489331883e-06,
+      "loss": 0.3137,
+      "step": 17874
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.6836633682250977,
+      "learning_rate": 8.31718421638248e-06,
+      "loss": 0.8751,
+      "step": 17875
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.2319021224975586,
+      "learning_rate": 8.31698693420923e-06,
+      "loss": 0.4809,
+      "step": 17876
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.3057117462158203,
+      "learning_rate": 8.31678964281268e-06,
+      "loss": 0.5601,
+      "step": 17877
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 4.249724388122559,
+      "learning_rate": 8.316592342193382e-06,
+      "loss": 0.8598,
+      "step": 17878
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 3.8338534832000732,
+      "learning_rate": 8.316395032351883e-06,
+      "loss": 0.8594,
+      "step": 17879
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 5.744667053222656,
+      "learning_rate": 8.31619771328873e-06,
+      "loss": 0.77,
+      "step": 17880
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.914585828781128,
+      "learning_rate": 8.316000385004475e-06,
+      "loss": 0.6902,
+      "step": 17881
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.4975974559783936,
+      "learning_rate": 8.315803047499664e-06,
+      "loss": 0.7971,
+      "step": 17882
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 3.2594454288482666,
+      "learning_rate": 8.315605700774848e-06,
+      "loss": 0.7373,
+      "step": 17883
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.2044599056243896,
+      "learning_rate": 8.315408344830574e-06,
+      "loss": 0.7107,
+      "step": 17884
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.8767626285552979,
+      "learning_rate": 8.315210979667392e-06,
+      "loss": 0.5928,
+      "step": 17885
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.8456246852874756,
+      "learning_rate": 8.315013605285851e-06,
+      "loss": 0.8791,
+      "step": 17886
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.378675699234009,
+      "learning_rate": 8.314816221686497e-06,
+      "loss": 0.6995,
+      "step": 17887
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.041440725326538,
+      "learning_rate": 8.314618828869882e-06,
+      "loss": 0.6984,
+      "step": 17888
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.8766733407974243,
+      "learning_rate": 8.314421426836553e-06,
+      "loss": 0.6819,
+      "step": 17889
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.425654888153076,
+      "learning_rate": 8.31422401558706e-06,
+      "loss": 0.6075,
+      "step": 17890
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.033489942550659,
+      "learning_rate": 8.314026595121952e-06,
+      "loss": 0.6874,
+      "step": 17891
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.934165358543396,
+      "learning_rate": 8.313829165441777e-06,
+      "loss": 0.6485,
+      "step": 17892
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.790883183479309,
+      "learning_rate": 8.313631726547084e-06,
+      "loss": 0.5426,
+      "step": 17893
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.8854788541793823,
+      "learning_rate": 8.313434278438424e-06,
+      "loss": 0.7004,
+      "step": 17894
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 3.124347448348999,
+      "learning_rate": 8.313236821116343e-06,
+      "loss": 0.7134,
+      "step": 17895
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.786250591278076,
+      "learning_rate": 8.313039354581392e-06,
+      "loss": 0.6312,
+      "step": 17896
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.5325539112091064,
+      "learning_rate": 8.31284187883412e-06,
+      "loss": 0.8183,
+      "step": 17897
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.584591031074524,
+      "learning_rate": 8.312644393875075e-06,
+      "loss": 0.7005,
+      "step": 17898
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 4.8747735023498535,
+      "learning_rate": 8.312446899704806e-06,
+      "loss": 0.9556,
+      "step": 17899
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.799175500869751,
+      "learning_rate": 8.312249396323865e-06,
+      "loss": 0.667,
+      "step": 17900
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.838894248008728,
+      "learning_rate": 8.312051883732798e-06,
+      "loss": 0.5635,
+      "step": 17901
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.5951223373413086,
+      "learning_rate": 8.311854361932157e-06,
+      "loss": 0.4979,
+      "step": 17902
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.62669038772583,
+      "learning_rate": 8.311656830922489e-06,
+      "loss": 0.5848,
+      "step": 17903
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.078458070755005,
+      "learning_rate": 8.311459290704343e-06,
+      "loss": 0.805,
+      "step": 17904
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 3.919536828994751,
+      "learning_rate": 8.311261741278271e-06,
+      "loss": 0.742,
+      "step": 17905
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 3.3211727142333984,
+      "learning_rate": 8.311064182644819e-06,
+      "loss": 0.5874,
+      "step": 17906
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.5722341537475586,
+      "learning_rate": 8.310866614804539e-06,
+      "loss": 0.5289,
+      "step": 17907
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.204052209854126,
+      "learning_rate": 8.310669037757978e-06,
+      "loss": 0.6493,
+      "step": 17908
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 3.740997791290283,
+      "learning_rate": 8.310471451505686e-06,
+      "loss": 0.9939,
+      "step": 17909
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 3.341900587081909,
+      "learning_rate": 8.310273856048216e-06,
+      "loss": 0.928,
+      "step": 17910
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.4987850189208984,
+      "learning_rate": 8.310076251386112e-06,
+      "loss": 0.4896,
+      "step": 17911
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.5593204498291016,
+      "learning_rate": 8.309878637519925e-06,
+      "loss": 0.6113,
+      "step": 17912
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.0879034996032715,
+      "learning_rate": 8.309681014450208e-06,
+      "loss": 0.7075,
+      "step": 17913
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.5650157928466797,
+      "learning_rate": 8.309483382177507e-06,
+      "loss": 0.5932,
+      "step": 17914
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 3.647318124771118,
+      "learning_rate": 8.309285740702373e-06,
+      "loss": 0.5316,
+      "step": 17915
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 4.438194751739502,
+      "learning_rate": 8.309088090025354e-06,
+      "loss": 0.7891,
+      "step": 17916
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 6.288387298583984,
+      "learning_rate": 8.308890430147e-06,
+      "loss": 0.6324,
+      "step": 17917
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.648809552192688,
+      "learning_rate": 8.308692761067863e-06,
+      "loss": 0.7814,
+      "step": 17918
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 4.4890313148498535,
+      "learning_rate": 8.30849508278849e-06,
+      "loss": 0.9167,
+      "step": 17919
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.2966458797454834,
+      "learning_rate": 8.308297395309433e-06,
+      "loss": 0.6417,
+      "step": 17920
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 3.9433863162994385,
+      "learning_rate": 8.308099698631238e-06,
+      "loss": 0.7332,
+      "step": 17921
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.0985355377197266,
+      "learning_rate": 8.307901992754457e-06,
+      "loss": 0.6239,
+      "step": 17922
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.0685923099517822,
+      "learning_rate": 8.307704277679643e-06,
+      "loss": 0.5505,
+      "step": 17923
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.922354817390442,
+      "learning_rate": 8.307506553407339e-06,
+      "loss": 0.9951,
+      "step": 17924
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.775862455368042,
+      "learning_rate": 8.3073088199381e-06,
+      "loss": 0.2609,
+      "step": 17925
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.4423677921295166,
+      "learning_rate": 8.307111077272473e-06,
+      "loss": 0.8438,
+      "step": 17926
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.8227076530456543,
+      "learning_rate": 8.306913325411011e-06,
+      "loss": 0.7757,
+      "step": 17927
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 3.033315420150757,
+      "learning_rate": 8.30671556435426e-06,
+      "loss": 0.6494,
+      "step": 17928
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.074451208114624,
+      "learning_rate": 8.30651779410277e-06,
+      "loss": 0.6025,
+      "step": 17929
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.853184700012207,
+      "learning_rate": 8.306320014657097e-06,
+      "loss": 0.7377,
+      "step": 17930
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 3.113776206970215,
+      "learning_rate": 8.306122226017783e-06,
+      "loss": 0.7205,
+      "step": 17931
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.1350250244140625,
+      "learning_rate": 8.305924428185384e-06,
+      "loss": 0.8197,
+      "step": 17932
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.039994478225708,
+      "learning_rate": 8.305726621160447e-06,
+      "loss": 0.5807,
+      "step": 17933
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.3508176803588867,
+      "learning_rate": 8.305528804943521e-06,
+      "loss": 0.632,
+      "step": 17934
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.5095443725585938,
+      "learning_rate": 8.30533097953516e-06,
+      "loss": 0.7508,
+      "step": 17935
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.520181655883789,
+      "learning_rate": 8.30513314493591e-06,
+      "loss": 0.8473,
+      "step": 17936
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 3.176652431488037,
+      "learning_rate": 8.304935301146325e-06,
+      "loss": 0.8066,
+      "step": 17937
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 3.30696177482605,
+      "learning_rate": 8.30473744816695e-06,
+      "loss": 0.8067,
+      "step": 17938
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.0170421600341797,
+      "learning_rate": 8.30453958599834e-06,
+      "loss": 0.4465,
+      "step": 17939
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.3350929021835327,
+      "learning_rate": 8.304341714641044e-06,
+      "loss": 0.7171,
+      "step": 17940
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.9051228761672974,
+      "learning_rate": 8.30414383409561e-06,
+      "loss": 0.5366,
+      "step": 17941
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.7291063070297241,
+      "learning_rate": 8.303945944362591e-06,
+      "loss": 0.6209,
+      "step": 17942
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.170790433883667,
+      "learning_rate": 8.303748045442534e-06,
+      "loss": 0.628,
+      "step": 17943
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.34727144241333,
+      "learning_rate": 8.303550137335992e-06,
+      "loss": 0.6852,
+      "step": 17944
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 3.0396127700805664,
+      "learning_rate": 8.303352220043516e-06,
+      "loss": 0.7501,
+      "step": 17945
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 4.473313808441162,
+      "learning_rate": 8.303154293565654e-06,
+      "loss": 0.8604,
+      "step": 17946
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.9597893953323364,
+      "learning_rate": 8.302956357902958e-06,
+      "loss": 0.7074,
+      "step": 17947
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.303053855895996,
+      "learning_rate": 8.302758413055977e-06,
+      "loss": 0.8676,
+      "step": 17948
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.2328145503997803,
+      "learning_rate": 8.302560459025262e-06,
+      "loss": 0.5799,
+      "step": 17949
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.0086073875427246,
+      "learning_rate": 8.302362495811364e-06,
+      "loss": 0.5341,
+      "step": 17950
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.3270068168640137,
+      "learning_rate": 8.302164523414833e-06,
+      "loss": 0.6819,
+      "step": 17951
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.1949219703674316,
+      "learning_rate": 8.301966541836219e-06,
+      "loss": 0.6031,
+      "step": 17952
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.7420114278793335,
+      "learning_rate": 8.301768551076073e-06,
+      "loss": 0.7125,
+      "step": 17953
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.8136513233184814,
+      "learning_rate": 8.301570551134945e-06,
+      "loss": 0.5622,
+      "step": 17954
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.0677082538604736,
+      "learning_rate": 8.301372542013388e-06,
+      "loss": 0.4548,
+      "step": 17955
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.5547610521316528,
+      "learning_rate": 8.301174523711948e-06,
+      "loss": 0.427,
+      "step": 17956
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.0444164276123047,
+      "learning_rate": 8.30097649623118e-06,
+      "loss": 0.3712,
+      "step": 17957
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.771446466445923,
+      "learning_rate": 8.300778459571632e-06,
+      "loss": 0.6401,
+      "step": 17958
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.1378977298736572,
+      "learning_rate": 8.300580413733858e-06,
+      "loss": 0.7047,
+      "step": 17959
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.7334831953048706,
+      "learning_rate": 8.300382358718404e-06,
+      "loss": 0.7103,
+      "step": 17960
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.293384552001953,
+      "learning_rate": 8.300184294525824e-06,
+      "loss": 0.5093,
+      "step": 17961
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.992340326309204,
+      "learning_rate": 8.299986221156667e-06,
+      "loss": 0.8029,
+      "step": 17962
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 3.6361007690429688,
+      "learning_rate": 8.299788138611484e-06,
+      "loss": 0.6661,
+      "step": 17963
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.2908315658569336,
+      "learning_rate": 8.29959004689083e-06,
+      "loss": 0.6228,
+      "step": 17964
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 3.4667868614196777,
+      "learning_rate": 8.299391945995248e-06,
+      "loss": 0.6295,
+      "step": 17965
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 6.7256574630737305,
+      "learning_rate": 8.299193835925294e-06,
+      "loss": 0.7098,
+      "step": 17966
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 8.604479789733887,
+      "learning_rate": 8.298995716681518e-06,
+      "loss": 0.6546,
+      "step": 17967
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.0307228565216064,
+      "learning_rate": 8.298797588264471e-06,
+      "loss": 0.4454,
+      "step": 17968
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.637675166130066,
+      "learning_rate": 8.298599450674704e-06,
+      "loss": 0.7073,
+      "step": 17969
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.7426645755767822,
+      "learning_rate": 8.298401303912766e-06,
+      "loss": 0.6844,
+      "step": 17970
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.3669216632843018,
+      "learning_rate": 8.298203147979212e-06,
+      "loss": 0.7302,
+      "step": 17971
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 3.656496286392212,
+      "learning_rate": 8.29800498287459e-06,
+      "loss": 0.5737,
+      "step": 17972
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.0623767375946045,
+      "learning_rate": 8.29780680859945e-06,
+      "loss": 0.7987,
+      "step": 17973
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 3.0872678756713867,
+      "learning_rate": 8.297608625154346e-06,
+      "loss": 0.6485,
+      "step": 17974
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.8742780685424805,
+      "learning_rate": 8.297410432539828e-06,
+      "loss": 0.6475,
+      "step": 17975
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.7055039405822754,
+      "learning_rate": 8.297212230756446e-06,
+      "loss": 0.769,
+      "step": 17976
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.207451581954956,
+      "learning_rate": 8.297014019804752e-06,
+      "loss": 0.5996,
+      "step": 17977
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 3.7937896251678467,
+      "learning_rate": 8.296815799685297e-06,
+      "loss": 0.9053,
+      "step": 17978
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.9385886192321777,
+      "learning_rate": 8.296617570398634e-06,
+      "loss": 0.401,
+      "step": 17979
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.9820808172225952,
+      "learning_rate": 8.29641933194531e-06,
+      "loss": 0.8734,
+      "step": 17980
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.642601490020752,
+      "learning_rate": 8.29622108432588e-06,
+      "loss": 0.482,
+      "step": 17981
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.950246810913086,
+      "learning_rate": 8.296022827540893e-06,
+      "loss": 0.5287,
+      "step": 17982
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 3.9981584548950195,
+      "learning_rate": 8.295824561590901e-06,
+      "loss": 0.5457,
+      "step": 17983
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.855818510055542,
+      "learning_rate": 8.295626286476457e-06,
+      "loss": 0.5907,
+      "step": 17984
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.9429088830947876,
+      "learning_rate": 8.29542800219811e-06,
+      "loss": 0.6084,
+      "step": 17985
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.0692532062530518,
+      "learning_rate": 8.295229708756412e-06,
+      "loss": 0.7264,
+      "step": 17986
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 3.301313638687134,
+      "learning_rate": 8.295031406151916e-06,
+      "loss": 0.6708,
+      "step": 17987
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.753626823425293,
+      "learning_rate": 8.29483309438517e-06,
+      "loss": 0.7366,
+      "step": 17988
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 3.6936628818511963,
+      "learning_rate": 8.29463477345673e-06,
+      "loss": 0.9476,
+      "step": 17989
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 3.313527822494507,
+      "learning_rate": 8.294436443367143e-06,
+      "loss": 0.8845,
+      "step": 17990
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 3.0900988578796387,
+      "learning_rate": 8.294238104116963e-06,
+      "loss": 0.5614,
+      "step": 17991
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.0189754962921143,
+      "learning_rate": 8.294039755706741e-06,
+      "loss": 0.5408,
+      "step": 17992
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.8400142192840576,
+      "learning_rate": 8.293841398137028e-06,
+      "loss": 0.4334,
+      "step": 17993
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.6428420543670654,
+      "learning_rate": 8.293643031408375e-06,
+      "loss": 0.5397,
+      "step": 17994
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 4.310129642486572,
+      "learning_rate": 8.293444655521337e-06,
+      "loss": 0.8879,
+      "step": 17995
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.054126739501953,
+      "learning_rate": 8.293246270476461e-06,
+      "loss": 0.7277,
+      "step": 17996
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.5597254037857056,
+      "learning_rate": 8.293047876274302e-06,
+      "loss": 0.7153,
+      "step": 17997
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 3.2459466457366943,
+      "learning_rate": 8.29284947291541e-06,
+      "loss": 0.7092,
+      "step": 17998
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.6304566860198975,
+      "learning_rate": 8.292651060400338e-06,
+      "loss": 0.8831,
+      "step": 17999
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.0460410118103027,
+      "learning_rate": 8.292452638729635e-06,
+      "loss": 0.6831,
+      "step": 18000
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.0718626976013184,
+      "learning_rate": 8.292254207903855e-06,
+      "loss": 0.7256,
+      "step": 18001
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.5480998754501343,
+      "learning_rate": 8.292055767923551e-06,
+      "loss": 0.9252,
+      "step": 18002
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.7517167329788208,
+      "learning_rate": 8.291857318789273e-06,
+      "loss": 0.7606,
+      "step": 18003
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 3.218679904937744,
+      "learning_rate": 8.291658860501571e-06,
+      "loss": 0.9559,
+      "step": 18004
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 3.4380414485931396,
+      "learning_rate": 8.291460393061e-06,
+      "loss": 0.41,
+      "step": 18005
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 3.72171688079834,
+      "learning_rate": 8.29126191646811e-06,
+      "loss": 1.035,
+      "step": 18006
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.447138786315918,
+      "learning_rate": 8.291063430723455e-06,
+      "loss": 0.6062,
+      "step": 18007
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.918837070465088,
+      "learning_rate": 8.290864935827584e-06,
+      "loss": 0.8377,
+      "step": 18008
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.2770450115203857,
+      "learning_rate": 8.29066643178105e-06,
+      "loss": 0.6291,
+      "step": 18009
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.8379266262054443,
+      "learning_rate": 8.290467918584408e-06,
+      "loss": 0.5238,
+      "step": 18010
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.7889572381973267,
+      "learning_rate": 8.290269396238206e-06,
+      "loss": 0.4421,
+      "step": 18011
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 3.400250196456909,
+      "learning_rate": 8.290070864742997e-06,
+      "loss": 0.4304,
+      "step": 18012
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.7688651084899902,
+      "learning_rate": 8.289872324099334e-06,
+      "loss": 0.387,
+      "step": 18013
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.361586570739746,
+      "learning_rate": 8.289673774307766e-06,
+      "loss": 0.5602,
+      "step": 18014
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.0354535579681396,
+      "learning_rate": 8.289475215368851e-06,
+      "loss": 0.5899,
+      "step": 18015
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.346637010574341,
+      "learning_rate": 8.289276647283135e-06,
+      "loss": 0.8522,
+      "step": 18016
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.3821325302124023,
+      "learning_rate": 8.289078070051175e-06,
+      "loss": 0.7281,
+      "step": 18017
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.1123850345611572,
+      "learning_rate": 8.28887948367352e-06,
+      "loss": 0.7571,
+      "step": 18018
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.0116164684295654,
+      "learning_rate": 8.288680888150724e-06,
+      "loss": 0.8102,
+      "step": 18019
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.220031261444092,
+      "learning_rate": 8.288482283483336e-06,
+      "loss": 0.5459,
+      "step": 18020
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.7096173763275146,
+      "learning_rate": 8.288283669671914e-06,
+      "loss": 0.7517,
+      "step": 18021
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.8872418403625488,
+      "learning_rate": 8.288085046717005e-06,
+      "loss": 0.7744,
+      "step": 18022
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.2515294551849365,
+      "learning_rate": 8.287886414619164e-06,
+      "loss": 0.7726,
+      "step": 18023
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.8919742107391357,
+      "learning_rate": 8.287687773378941e-06,
+      "loss": 0.6613,
+      "step": 18024
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.0021560192108154,
+      "learning_rate": 8.287489122996892e-06,
+      "loss": 0.6954,
+      "step": 18025
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.5983778238296509,
+      "learning_rate": 8.287290463473564e-06,
+      "loss": 0.4103,
+      "step": 18026
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.015584945678711,
+      "learning_rate": 8.287091794809516e-06,
+      "loss": 0.5693,
+      "step": 18027
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.9434146881103516,
+      "learning_rate": 8.286893117005297e-06,
+      "loss": 0.6849,
+      "step": 18028
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 3.777682304382324,
+      "learning_rate": 8.286694430061458e-06,
+      "loss": 0.7573,
+      "step": 18029
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.9907925128936768,
+      "learning_rate": 8.286495733978554e-06,
+      "loss": 0.7665,
+      "step": 18030
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.4384961128234863,
+      "learning_rate": 8.286297028757135e-06,
+      "loss": 0.6856,
+      "step": 18031
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.6493483781814575,
+      "learning_rate": 8.286098314397757e-06,
+      "loss": 0.649,
+      "step": 18032
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 4.240105152130127,
+      "learning_rate": 8.28589959090097e-06,
+      "loss": 0.5791,
+      "step": 18033
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 5.2573723793029785,
+      "learning_rate": 8.285700858267326e-06,
+      "loss": 0.4875,
+      "step": 18034
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.6758568286895752,
+      "learning_rate": 8.285502116497379e-06,
+      "loss": 0.4781,
+      "step": 18035
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.6982429027557373,
+      "learning_rate": 8.285303365591683e-06,
+      "loss": 0.8398,
+      "step": 18036
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.9503655433654785,
+      "learning_rate": 8.285104605550788e-06,
+      "loss": 0.6267,
+      "step": 18037
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.5388174057006836,
+      "learning_rate": 8.284905836375248e-06,
+      "loss": 0.4518,
+      "step": 18038
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 3.0316646099090576,
+      "learning_rate": 8.284707058065616e-06,
+      "loss": 0.7545,
+      "step": 18039
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 3.575676441192627,
+      "learning_rate": 8.284508270622443e-06,
+      "loss": 0.9089,
+      "step": 18040
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.6251332759857178,
+      "learning_rate": 8.284309474046284e-06,
+      "loss": 0.5305,
+      "step": 18041
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 3.5296566486358643,
+      "learning_rate": 8.28411066833769e-06,
+      "loss": 0.517,
+      "step": 18042
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.9206520318984985,
+      "learning_rate": 8.283911853497214e-06,
+      "loss": 0.7633,
+      "step": 18043
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.1512866020202637,
+      "learning_rate": 8.28371302952541e-06,
+      "loss": 0.5978,
+      "step": 18044
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.3898417949676514,
+      "learning_rate": 8.283514196422828e-06,
+      "loss": 0.5531,
+      "step": 18045
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 3.2154383659362793,
+      "learning_rate": 8.283315354190025e-06,
+      "loss": 0.4715,
+      "step": 18046
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.9093021154403687,
+      "learning_rate": 8.283116502827552e-06,
+      "loss": 0.6361,
+      "step": 18047
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.8980554342269897,
+      "learning_rate": 8.282917642335963e-06,
+      "loss": 0.6657,
+      "step": 18048
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.156230926513672,
+      "learning_rate": 8.282718772715808e-06,
+      "loss": 0.5392,
+      "step": 18049
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.9738037586212158,
+      "learning_rate": 8.282519893967644e-06,
+      "loss": 0.6277,
+      "step": 18050
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.4837193489074707,
+      "learning_rate": 8.28232100609202e-06,
+      "loss": 0.5322,
+      "step": 18051
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.8983877897262573,
+      "learning_rate": 8.28212210908949e-06,
+      "loss": 0.7972,
+      "step": 18052
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.358144521713257,
+      "learning_rate": 8.28192320296061e-06,
+      "loss": 0.7064,
+      "step": 18053
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.776547908782959,
+      "learning_rate": 8.28172428770593e-06,
+      "loss": 0.6973,
+      "step": 18054
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.0765585899353027,
+      "learning_rate": 8.281525363326004e-06,
+      "loss": 0.7692,
+      "step": 18055
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.251215696334839,
+      "learning_rate": 8.281326429821386e-06,
+      "loss": 0.5297,
+      "step": 18056
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.8445913791656494,
+      "learning_rate": 8.281127487192626e-06,
+      "loss": 0.7147,
+      "step": 18057
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.306098461151123,
+      "learning_rate": 8.280928535440282e-06,
+      "loss": 0.7585,
+      "step": 18058
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.734018325805664,
+      "learning_rate": 8.280729574564906e-06,
+      "loss": 0.8209,
+      "step": 18059
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 3.171114206314087,
+      "learning_rate": 8.280530604567047e-06,
+      "loss": 0.563,
+      "step": 18060
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.3091788291931152,
+      "learning_rate": 8.280331625447264e-06,
+      "loss": 0.6105,
+      "step": 18061
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 3.58095645904541,
+      "learning_rate": 8.280132637206105e-06,
+      "loss": 0.7258,
+      "step": 18062
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 4.648130893707275,
+      "learning_rate": 8.279933639844128e-06,
+      "loss": 0.6717,
+      "step": 18063
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.3373425006866455,
+      "learning_rate": 8.279734633361884e-06,
+      "loss": 0.7209,
+      "step": 18064
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.3956644535064697,
+      "learning_rate": 8.279535617759924e-06,
+      "loss": 0.5438,
+      "step": 18065
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.781351923942566,
+      "learning_rate": 8.279336593038807e-06,
+      "loss": 0.7868,
+      "step": 18066
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 3.4998154640197754,
+      "learning_rate": 8.279137559199081e-06,
+      "loss": 0.831,
+      "step": 18067
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 6.014233112335205,
+      "learning_rate": 8.278938516241304e-06,
+      "loss": 1.148,
+      "step": 18068
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.4974617958068848,
+      "learning_rate": 8.278739464166026e-06,
+      "loss": 0.6803,
+      "step": 18069
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.7405441999435425,
+      "learning_rate": 8.278540402973802e-06,
+      "loss": 0.7856,
+      "step": 18070
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.9695093631744385,
+      "learning_rate": 8.278341332665187e-06,
+      "loss": 0.687,
+      "step": 18071
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.0245699882507324,
+      "learning_rate": 8.278142253240731e-06,
+      "loss": 0.9014,
+      "step": 18072
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.9726035594940186,
+      "learning_rate": 8.27794316470099e-06,
+      "loss": 0.5502,
+      "step": 18073
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.4619073867797852,
+      "learning_rate": 8.277744067046518e-06,
+      "loss": 0.5437,
+      "step": 18074
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.89030385017395,
+      "learning_rate": 8.277544960277866e-06,
+      "loss": 0.6269,
+      "step": 18075
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.924359917640686,
+      "learning_rate": 8.277345844395591e-06,
+      "loss": 0.5623,
+      "step": 18076
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.0603299140930176,
+      "learning_rate": 8.277146719400242e-06,
+      "loss": 0.5648,
+      "step": 18077
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.9683327674865723,
+      "learning_rate": 8.276947585292378e-06,
+      "loss": 0.6772,
+      "step": 18078
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.4982093572616577,
+      "learning_rate": 8.27674844207255e-06,
+      "loss": 0.9204,
+      "step": 18079
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.2193543910980225,
+      "learning_rate": 8.276549289741311e-06,
+      "loss": 0.6349,
+      "step": 18080
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.450153350830078,
+      "learning_rate": 8.276350128299217e-06,
+      "loss": 0.4623,
+      "step": 18081
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.8978564739227295,
+      "learning_rate": 8.27615095774682e-06,
+      "loss": 0.4424,
+      "step": 18082
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.5290887355804443,
+      "learning_rate": 8.275951778084675e-06,
+      "loss": 0.6066,
+      "step": 18083
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.61643385887146,
+      "learning_rate": 8.275752589313336e-06,
+      "loss": 0.6249,
+      "step": 18084
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.6200135946273804,
+      "learning_rate": 8.275553391433353e-06,
+      "loss": 0.8784,
+      "step": 18085
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.0330541133880615,
+      "learning_rate": 8.275354184445286e-06,
+      "loss": 0.7758,
+      "step": 18086
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 3.2787952423095703,
+      "learning_rate": 8.275154968349686e-06,
+      "loss": 0.6488,
+      "step": 18087
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 4.578986167907715,
+      "learning_rate": 8.274955743147106e-06,
+      "loss": 1.19,
+      "step": 18088
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.2234721183776855,
+      "learning_rate": 8.2747565088381e-06,
+      "loss": 0.6038,
+      "step": 18089
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.116927146911621,
+      "learning_rate": 8.274557265423225e-06,
+      "loss": 0.5906,
+      "step": 18090
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.9589111804962158,
+      "learning_rate": 8.27435801290303e-06,
+      "loss": 0.5486,
+      "step": 18091
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 4.172830104827881,
+      "learning_rate": 8.274158751278074e-06,
+      "loss": 0.7428,
+      "step": 18092
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.5294588804244995,
+      "learning_rate": 8.27395948054891e-06,
+      "loss": 0.6479,
+      "step": 18093
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.7865654230117798,
+      "learning_rate": 8.27376020071609e-06,
+      "loss": 0.8923,
+      "step": 18094
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 3.218362331390381,
+      "learning_rate": 8.27356091178017e-06,
+      "loss": 0.6286,
+      "step": 18095
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 5.544905185699463,
+      "learning_rate": 8.2733616137417e-06,
+      "loss": 0.9804,
+      "step": 18096
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.6212146282196045,
+      "learning_rate": 8.273162306601243e-06,
+      "loss": 0.3984,
+      "step": 18097
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 3.073223829269409,
+      "learning_rate": 8.272962990359344e-06,
+      "loss": 0.5749,
+      "step": 18098
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.279310464859009,
+      "learning_rate": 8.272763665016563e-06,
+      "loss": 0.5392,
+      "step": 18099
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.5022454261779785,
+      "learning_rate": 8.272564330573452e-06,
+      "loss": 0.6389,
+      "step": 18100
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.8404412269592285,
+      "learning_rate": 8.272364987030563e-06,
+      "loss": 0.6174,
+      "step": 18101
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.7103415727615356,
+      "learning_rate": 8.272165634388456e-06,
+      "loss": 0.644,
+      "step": 18102
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.4642369747161865,
+      "learning_rate": 8.27196627264768e-06,
+      "loss": 0.6384,
+      "step": 18103
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 3.011296510696411,
+      "learning_rate": 8.271766901808794e-06,
+      "loss": 0.7762,
+      "step": 18104
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.9659512042999268,
+      "learning_rate": 8.271567521872346e-06,
+      "loss": 0.6407,
+      "step": 18105
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.0340988636016846,
+      "learning_rate": 8.271368132838897e-06,
+      "loss": 0.8157,
+      "step": 18106
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.7264413833618164,
+      "learning_rate": 8.271168734708998e-06,
+      "loss": 0.7099,
+      "step": 18107
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.4990031719207764,
+      "learning_rate": 8.270969327483205e-06,
+      "loss": 0.5085,
+      "step": 18108
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.2421975135803223,
+      "learning_rate": 8.270769911162072e-06,
+      "loss": 0.6913,
+      "step": 18109
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.5974671840667725,
+      "learning_rate": 8.270570485746151e-06,
+      "loss": 0.6573,
+      "step": 18110
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.1246368885040283,
+      "learning_rate": 8.270371051236e-06,
+      "loss": 0.7776,
+      "step": 18111
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.7769737243652344,
+      "learning_rate": 8.270171607632171e-06,
+      "loss": 0.638,
+      "step": 18112
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.008155345916748,
+      "learning_rate": 8.269972154935222e-06,
+      "loss": 0.7992,
+      "step": 18113
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.923638105392456,
+      "learning_rate": 8.269772693145704e-06,
+      "loss": 0.6891,
+      "step": 18114
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 3.051093101501465,
+      "learning_rate": 8.269573222264174e-06,
+      "loss": 0.684,
+      "step": 18115
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.6038848161697388,
+      "learning_rate": 8.269373742291184e-06,
+      "loss": 0.6244,
+      "step": 18116
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.294140577316284,
+      "learning_rate": 8.26917425322729e-06,
+      "loss": 0.805,
+      "step": 18117
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.0531129837036133,
+      "learning_rate": 8.268974755073049e-06,
+      "loss": 0.6465,
+      "step": 18118
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.4289021492004395,
+      "learning_rate": 8.268775247829012e-06,
+      "loss": 0.4973,
+      "step": 18119
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 3.702667474746704,
+      "learning_rate": 8.268575731495738e-06,
+      "loss": 0.9068,
+      "step": 18120
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.529095411300659,
+      "learning_rate": 8.268376206073778e-06,
+      "loss": 0.5889,
+      "step": 18121
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.363957166671753,
+      "learning_rate": 8.268176671563688e-06,
+      "loss": 0.6431,
+      "step": 18122
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.6003780364990234,
+      "learning_rate": 8.267977127966024e-06,
+      "loss": 0.4796,
+      "step": 18123
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.380270481109619,
+      "learning_rate": 8.26777757528134e-06,
+      "loss": 0.6439,
+      "step": 18124
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 3.089031934738159,
+      "learning_rate": 8.267578013510189e-06,
+      "loss": 0.7164,
+      "step": 18125
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.730623722076416,
+      "learning_rate": 8.267378442653126e-06,
+      "loss": 0.6796,
+      "step": 18126
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.8695534467697144,
+      "learning_rate": 8.267178862710711e-06,
+      "loss": 0.7503,
+      "step": 18127
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.225205183029175,
+      "learning_rate": 8.266979273683494e-06,
+      "loss": 0.7542,
+      "step": 18128
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.4505465030670166,
+      "learning_rate": 8.266779675572031e-06,
+      "loss": 0.4325,
+      "step": 18129
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.3527228832244873,
+      "learning_rate": 8.266580068376879e-06,
+      "loss": 0.5611,
+      "step": 18130
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.8653546571731567,
+      "learning_rate": 8.26638045209859e-06,
+      "loss": 0.6091,
+      "step": 18131
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.4533400535583496,
+      "learning_rate": 8.266180826737722e-06,
+      "loss": 0.5923,
+      "step": 18132
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.988525390625,
+      "learning_rate": 8.265981192294829e-06,
+      "loss": 0.7247,
+      "step": 18133
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 3.0278053283691406,
+      "learning_rate": 8.265781548770464e-06,
+      "loss": 0.6698,
+      "step": 18134
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.405398368835449,
+      "learning_rate": 8.265581896165184e-06,
+      "loss": 0.7008,
+      "step": 18135
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.4590208530426025,
+      "learning_rate": 8.265382234479545e-06,
+      "loss": 0.6563,
+      "step": 18136
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.3257498741149902,
+      "learning_rate": 8.2651825637141e-06,
+      "loss": 0.8172,
+      "step": 18137
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.9983599185943604,
+      "learning_rate": 8.264982883869408e-06,
+      "loss": 0.4488,
+      "step": 18138
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.5623602867126465,
+      "learning_rate": 8.26478319494602e-06,
+      "loss": 0.5614,
+      "step": 18139
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.754017949104309,
+      "learning_rate": 8.264583496944493e-06,
+      "loss": 0.7348,
+      "step": 18140
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.655975341796875,
+      "learning_rate": 8.264383789865383e-06,
+      "loss": 0.8115,
+      "step": 18141
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 3.8321454524993896,
+      "learning_rate": 8.264184073709243e-06,
+      "loss": 0.7669,
+      "step": 18142
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.12579083442688,
+      "learning_rate": 8.263984348476631e-06,
+      "loss": 0.3889,
+      "step": 18143
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.9950898885726929,
+      "learning_rate": 8.2637846141681e-06,
+      "loss": 0.6379,
+      "step": 18144
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.8239176273345947,
+      "learning_rate": 8.263584870784208e-06,
+      "loss": 0.6595,
+      "step": 18145
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.7482918500900269,
+      "learning_rate": 8.263385118325509e-06,
+      "loss": 0.6434,
+      "step": 18146
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.5972959995269775,
+      "learning_rate": 8.263185356792557e-06,
+      "loss": 0.6149,
+      "step": 18147
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.0409843921661377,
+      "learning_rate": 8.262985586185911e-06,
+      "loss": 0.6912,
+      "step": 18148
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.5526554584503174,
+      "learning_rate": 8.262785806506123e-06,
+      "loss": 0.5699,
+      "step": 18149
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.7340047359466553,
+      "learning_rate": 8.26258601775375e-06,
+      "loss": 0.4631,
+      "step": 18150
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 4.41455078125,
+      "learning_rate": 8.262386219929348e-06,
+      "loss": 0.5027,
+      "step": 18151
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.131490468978882,
+      "learning_rate": 8.262186413033472e-06,
+      "loss": 0.4073,
+      "step": 18152
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 3.6438381671905518,
+      "learning_rate": 8.261986597066679e-06,
+      "loss": 0.6484,
+      "step": 18153
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.7144078016281128,
+      "learning_rate": 8.261786772029521e-06,
+      "loss": 0.5725,
+      "step": 18154
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.8550400733947754,
+      "learning_rate": 8.261586937922556e-06,
+      "loss": 0.583,
+      "step": 18155
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 3.007668972015381,
+      "learning_rate": 8.26138709474634e-06,
+      "loss": 0.527,
+      "step": 18156
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 3.5983998775482178,
+      "learning_rate": 8.261187242501429e-06,
+      "loss": 0.7605,
+      "step": 18157
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.1697187423706055,
+      "learning_rate": 8.260987381188376e-06,
+      "loss": 0.6384,
+      "step": 18158
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.8542959690093994,
+      "learning_rate": 8.260787510807742e-06,
+      "loss": 0.7266,
+      "step": 18159
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 3.027859687805176,
+      "learning_rate": 8.260587631360078e-06,
+      "loss": 0.8467,
+      "step": 18160
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.3984885215759277,
+      "learning_rate": 8.26038774284594e-06,
+      "loss": 0.4133,
+      "step": 18161
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.749097228050232,
+      "learning_rate": 8.260187845265886e-06,
+      "loss": 0.4407,
+      "step": 18162
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.0452470779418945,
+      "learning_rate": 8.25998793862047e-06,
+      "loss": 0.9186,
+      "step": 18163
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.8420076370239258,
+      "learning_rate": 8.25978802291025e-06,
+      "loss": 0.6257,
+      "step": 18164
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.156850576400757,
+      "learning_rate": 8.25958809813578e-06,
+      "loss": 0.4945,
+      "step": 18165
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.838424563407898,
+      "learning_rate": 8.259388164297617e-06,
+      "loss": 0.5619,
+      "step": 18166
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.9205894470214844,
+      "learning_rate": 8.259188221396316e-06,
+      "loss": 0.5809,
+      "step": 18167
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.7896275520324707,
+      "learning_rate": 8.258988269432433e-06,
+      "loss": 0.9558,
+      "step": 18168
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.5479120016098022,
+      "learning_rate": 8.258788308406526e-06,
+      "loss": 0.9127,
+      "step": 18169
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.7270047664642334,
+      "learning_rate": 8.258588338319148e-06,
+      "loss": 0.6268,
+      "step": 18170
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.892215609550476,
+      "learning_rate": 8.258388359170858e-06,
+      "loss": 0.6307,
+      "step": 18171
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.9577479362487793,
+      "learning_rate": 8.25818837096221e-06,
+      "loss": 0.5841,
+      "step": 18172
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.9471979141235352,
+      "learning_rate": 8.25798837369376e-06,
+      "loss": 0.3838,
+      "step": 18173
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.045332670211792,
+      "learning_rate": 8.257788367366065e-06,
+      "loss": 0.9074,
+      "step": 18174
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.3829479217529297,
+      "learning_rate": 8.257588351979682e-06,
+      "loss": 0.8225,
+      "step": 18175
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.578248143196106,
+      "learning_rate": 8.257388327535164e-06,
+      "loss": 0.5391,
+      "step": 18176
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 3.6141746044158936,
+      "learning_rate": 8.257188294033071e-06,
+      "loss": 0.6344,
+      "step": 18177
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.8304771184921265,
+      "learning_rate": 8.256988251473958e-06,
+      "loss": 0.483,
+      "step": 18178
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.71726655960083,
+      "learning_rate": 8.256788199858378e-06,
+      "loss": 0.653,
+      "step": 18179
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.6559759378433228,
+      "learning_rate": 8.256588139186894e-06,
+      "loss": 0.6301,
+      "step": 18180
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.6483638286590576,
+      "learning_rate": 8.256388069460056e-06,
+      "loss": 0.5682,
+      "step": 18181
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.4631497859954834,
+      "learning_rate": 8.256187990678422e-06,
+      "loss": 0.8171,
+      "step": 18182
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.2088513374328613,
+      "learning_rate": 8.25598790284255e-06,
+      "loss": 0.6307,
+      "step": 18183
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.5639281272888184,
+      "learning_rate": 8.255787805952995e-06,
+      "loss": 0.5383,
+      "step": 18184
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.238260269165039,
+      "learning_rate": 8.255587700010315e-06,
+      "loss": 0.8623,
+      "step": 18185
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.2995051145553589,
+      "learning_rate": 8.255387585015065e-06,
+      "loss": 0.5222,
+      "step": 18186
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.6013762950897217,
+      "learning_rate": 8.2551874609678e-06,
+      "loss": 0.655,
+      "step": 18187
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 3.929165840148926,
+      "learning_rate": 8.254987327869079e-06,
+      "loss": 0.6055,
+      "step": 18188
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.6514060497283936,
+      "learning_rate": 8.254787185719457e-06,
+      "loss": 0.4918,
+      "step": 18189
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.5605392456054688,
+      "learning_rate": 8.254587034519493e-06,
+      "loss": 0.7908,
+      "step": 18190
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 3.3460288047790527,
+      "learning_rate": 8.254386874269739e-06,
+      "loss": 1.0139,
+      "step": 18191
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.794027090072632,
+      "learning_rate": 8.254186704970756e-06,
+      "loss": 0.7586,
+      "step": 18192
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.655637264251709,
+      "learning_rate": 8.253986526623098e-06,
+      "loss": 0.7848,
+      "step": 18193
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.7485300302505493,
+      "learning_rate": 8.253786339227323e-06,
+      "loss": 0.8426,
+      "step": 18194
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.06042218208313,
+      "learning_rate": 8.253586142783987e-06,
+      "loss": 0.5557,
+      "step": 18195
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.3835158348083496,
+      "learning_rate": 8.253385937293647e-06,
+      "loss": 0.4169,
+      "step": 18196
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.3354175090789795,
+      "learning_rate": 8.253185722756858e-06,
+      "loss": 0.8255,
+      "step": 18197
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.225714683532715,
+      "learning_rate": 8.25298549917418e-06,
+      "loss": 0.7435,
+      "step": 18198
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.53080677986145,
+      "learning_rate": 8.252785266546167e-06,
+      "loss": 0.6857,
+      "step": 18199
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.15669584274292,
+      "learning_rate": 8.252585024873377e-06,
+      "loss": 0.8537,
+      "step": 18200
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 3.5262327194213867,
+      "learning_rate": 8.252384774156366e-06,
+      "loss": 0.6565,
+      "step": 18201
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.7734594345092773,
+      "learning_rate": 8.252184514395692e-06,
+      "loss": 0.5806,
+      "step": 18202
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 3.134990930557251,
+      "learning_rate": 8.25198424559191e-06,
+      "loss": 0.818,
+      "step": 18203
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.6134461164474487,
+      "learning_rate": 8.25178396774558e-06,
+      "loss": 0.5595,
+      "step": 18204
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.8862130641937256,
+      "learning_rate": 8.251583680857256e-06,
+      "loss": 0.732,
+      "step": 18205
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.8379143476486206,
+      "learning_rate": 8.251383384927496e-06,
+      "loss": 0.4695,
+      "step": 18206
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.4950761795043945,
+      "learning_rate": 8.251183079956854e-06,
+      "loss": 0.4387,
+      "step": 18207
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.570168972015381,
+      "learning_rate": 8.250982765945894e-06,
+      "loss": 0.6028,
+      "step": 18208
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.173633337020874,
+      "learning_rate": 8.250782442895166e-06,
+      "loss": 0.681,
+      "step": 18209
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.017124652862549,
+      "learning_rate": 8.250582110805233e-06,
+      "loss": 0.5444,
+      "step": 18210
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.7244343757629395,
+      "learning_rate": 8.250381769676645e-06,
+      "loss": 0.4599,
+      "step": 18211
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.003528118133545,
+      "learning_rate": 8.250181419509964e-06,
+      "loss": 0.6104,
+      "step": 18212
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.5711116790771484,
+      "learning_rate": 8.249981060305748e-06,
+      "loss": 0.7557,
+      "step": 18213
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.305493116378784,
+      "learning_rate": 8.24978069206455e-06,
+      "loss": 0.4899,
+      "step": 18214
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.8396188020706177,
+      "learning_rate": 8.24958031478693e-06,
+      "loss": 0.4424,
+      "step": 18215
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.862744688987732,
+      "learning_rate": 8.249379928473444e-06,
+      "loss": 0.594,
+      "step": 18216
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.1511433124542236,
+      "learning_rate": 8.249179533124651e-06,
+      "loss": 0.7086,
+      "step": 18217
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.7662568092346191,
+      "learning_rate": 8.248979128741106e-06,
+      "loss": 0.405,
+      "step": 18218
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.3258578777313232,
+      "learning_rate": 8.248778715323367e-06,
+      "loss": 0.5702,
+      "step": 18219
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.015418291091919,
+      "learning_rate": 8.248578292871991e-06,
+      "loss": 0.5654,
+      "step": 18220
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.6274306774139404,
+      "learning_rate": 8.248377861387538e-06,
+      "loss": 0.6552,
+      "step": 18221
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.6705960035324097,
+      "learning_rate": 8.24817742087056e-06,
+      "loss": 0.5682,
+      "step": 18222
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.0913302898406982,
+      "learning_rate": 8.247976971321618e-06,
+      "loss": 0.7129,
+      "step": 18223
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.8384337425231934,
+      "learning_rate": 8.24777651274127e-06,
+      "loss": 0.5903,
+      "step": 18224
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.8350067138671875,
+      "learning_rate": 8.247576045130073e-06,
+      "loss": 0.8241,
+      "step": 18225
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.7087615728378296,
+      "learning_rate": 8.247375568488583e-06,
+      "loss": 0.6348,
+      "step": 18226
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.567237138748169,
+      "learning_rate": 8.247175082817355e-06,
+      "loss": 0.7763,
+      "step": 18227
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.871591567993164,
+      "learning_rate": 8.246974588116952e-06,
+      "loss": 0.5449,
+      "step": 18228
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.927015781402588,
+      "learning_rate": 8.24677408438793e-06,
+      "loss": 0.7254,
+      "step": 18229
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.8816299438476562,
+      "learning_rate": 8.246573571630843e-06,
+      "loss": 0.6476,
+      "step": 18230
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.2970399856567383,
+      "learning_rate": 8.246373049846253e-06,
+      "loss": 0.667,
+      "step": 18231
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 8.51439094543457,
+      "learning_rate": 8.246172519034715e-06,
+      "loss": 0.6089,
+      "step": 18232
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.6181024312973022,
+      "learning_rate": 8.245971979196788e-06,
+      "loss": 0.4843,
+      "step": 18233
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.724413514137268,
+      "learning_rate": 8.245771430333028e-06,
+      "loss": 0.6459,
+      "step": 18234
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.026886224746704,
+      "learning_rate": 8.245570872443994e-06,
+      "loss": 0.5929,
+      "step": 18235
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.160358190536499,
+      "learning_rate": 8.245370305530243e-06,
+      "loss": 0.6178,
+      "step": 18236
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 4.700367450714111,
+      "learning_rate": 8.245169729592333e-06,
+      "loss": 0.4103,
+      "step": 18237
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 3.017650604248047,
+      "learning_rate": 8.244969144630822e-06,
+      "loss": 0.6877,
+      "step": 18238
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.278928756713867,
+      "learning_rate": 8.244768550646267e-06,
+      "loss": 0.453,
+      "step": 18239
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.448886513710022,
+      "learning_rate": 8.244567947639227e-06,
+      "loss": 0.7358,
+      "step": 18240
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 3.1175649166107178,
+      "learning_rate": 8.244367335610258e-06,
+      "loss": 0.6754,
+      "step": 18241
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.7324609756469727,
+      "learning_rate": 8.244166714559918e-06,
+      "loss": 0.7808,
+      "step": 18242
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.435015916824341,
+      "learning_rate": 8.243966084488766e-06,
+      "loss": 0.8762,
+      "step": 18243
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 4.2507710456848145,
+      "learning_rate": 8.24376544539736e-06,
+      "loss": 0.6513,
+      "step": 18244
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.008216381072998,
+      "learning_rate": 8.243564797286258e-06,
+      "loss": 0.5647,
+      "step": 18245
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 3.3513283729553223,
+      "learning_rate": 8.243364140156019e-06,
+      "loss": 0.8301,
+      "step": 18246
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 3.3815925121307373,
+      "learning_rate": 8.243163474007196e-06,
+      "loss": 0.6761,
+      "step": 18247
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 3.56679368019104,
+      "learning_rate": 8.24296279884035e-06,
+      "loss": 0.6709,
+      "step": 18248
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 3.271226167678833,
+      "learning_rate": 8.242762114656043e-06,
+      "loss": 0.6793,
+      "step": 18249
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.212839126586914,
+      "learning_rate": 8.242561421454827e-06,
+      "loss": 0.5873,
+      "step": 18250
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.615409016609192,
+      "learning_rate": 8.242360719237261e-06,
+      "loss": 0.5979,
+      "step": 18251
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.7280333042144775,
+      "learning_rate": 8.242160008003908e-06,
+      "loss": 0.6777,
+      "step": 18252
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.9781129360198975,
+      "learning_rate": 8.241959287755318e-06,
+      "loss": 0.3477,
+      "step": 18253
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.8529428243637085,
+      "learning_rate": 8.241758558492058e-06,
+      "loss": 0.8791,
+      "step": 18254
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 3.963320016860962,
+      "learning_rate": 8.241557820214681e-06,
+      "loss": 0.8209,
+      "step": 18255
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 6.218343257904053,
+      "learning_rate": 8.241357072923744e-06,
+      "loss": 0.7442,
+      "step": 18256
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 3.214298725128174,
+      "learning_rate": 8.241156316619809e-06,
+      "loss": 0.839,
+      "step": 18257
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.9535837173461914,
+      "learning_rate": 8.240955551303432e-06,
+      "loss": 0.8912,
+      "step": 18258
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.4173614978790283,
+      "learning_rate": 8.24075477697517e-06,
+      "loss": 0.5618,
+      "step": 18259
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 4.1958909034729,
+      "learning_rate": 8.240553993635585e-06,
+      "loss": 0.6334,
+      "step": 18260
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.018291711807251,
+      "learning_rate": 8.240353201285233e-06,
+      "loss": 0.7402,
+      "step": 18261
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.395876407623291,
+      "learning_rate": 8.240152399924673e-06,
+      "loss": 0.4975,
+      "step": 18262
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.0447041988372803,
+      "learning_rate": 8.239951589554462e-06,
+      "loss": 0.6896,
+      "step": 18263
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 3.4042584896087646,
+      "learning_rate": 8.23975077017516e-06,
+      "loss": 0.9931,
+      "step": 18264
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.1176564693450928,
+      "learning_rate": 8.239549941787325e-06,
+      "loss": 0.8262,
+      "step": 18265
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.423574209213257,
+      "learning_rate": 8.239349104391513e-06,
+      "loss": 0.6735,
+      "step": 18266
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.2162373065948486,
+      "learning_rate": 8.239148257988287e-06,
+      "loss": 0.6325,
+      "step": 18267
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.2332344055175781,
+      "learning_rate": 8.238947402578202e-06,
+      "loss": 0.6332,
+      "step": 18268
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.7234437465667725,
+      "learning_rate": 8.23874653816182e-06,
+      "loss": 0.7841,
+      "step": 18269
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.0633466243743896,
+      "learning_rate": 8.238545664739694e-06,
+      "loss": 0.699,
+      "step": 18270
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.943751573562622,
+      "learning_rate": 8.238344782312387e-06,
+      "loss": 0.8024,
+      "step": 18271
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.6208250522613525,
+      "learning_rate": 8.238143890880456e-06,
+      "loss": 0.7588,
+      "step": 18272
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.5968092679977417,
+      "learning_rate": 8.23794299044446e-06,
+      "loss": 0.9596,
+      "step": 18273
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.7180323600769043,
+      "learning_rate": 8.237742081004956e-06,
+      "loss": 0.5474,
+      "step": 18274
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.752162218093872,
+      "learning_rate": 8.237541162562506e-06,
+      "loss": 0.5528,
+      "step": 18275
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.895289421081543,
+      "learning_rate": 8.237340235117668e-06,
+      "loss": 0.6918,
+      "step": 18276
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.879692792892456,
+      "learning_rate": 8.237139298670998e-06,
+      "loss": 0.5637,
+      "step": 18277
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.8799853324890137,
+      "learning_rate": 8.236938353223056e-06,
+      "loss": 0.65,
+      "step": 18278
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.266819715499878,
+      "learning_rate": 8.2367373987744e-06,
+      "loss": 0.8114,
+      "step": 18279
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.569132089614868,
+      "learning_rate": 8.236536435325592e-06,
+      "loss": 0.6275,
+      "step": 18280
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.0589542388916016,
+      "learning_rate": 8.236335462877187e-06,
+      "loss": 0.8019,
+      "step": 18281
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.042386770248413,
+      "learning_rate": 8.236134481429748e-06,
+      "loss": 0.5384,
+      "step": 18282
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.924830436706543,
+      "learning_rate": 8.235933490983827e-06,
+      "loss": 0.5693,
+      "step": 18283
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.5566234588623047,
+      "learning_rate": 8.235732491539991e-06,
+      "loss": 0.6143,
+      "step": 18284
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.277956962585449,
+      "learning_rate": 8.235531483098793e-06,
+      "loss": 0.601,
+      "step": 18285
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 3.584902286529541,
+      "learning_rate": 8.235330465660795e-06,
+      "loss": 0.6779,
+      "step": 18286
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.6162705421447754,
+      "learning_rate": 8.235129439226554e-06,
+      "loss": 0.7425,
+      "step": 18287
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 3.242074489593506,
+      "learning_rate": 8.234928403796631e-06,
+      "loss": 0.8532,
+      "step": 18288
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 3.172218084335327,
+      "learning_rate": 8.234727359371583e-06,
+      "loss": 0.6241,
+      "step": 18289
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.02001690864563,
+      "learning_rate": 8.234526305951972e-06,
+      "loss": 0.7252,
+      "step": 18290
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.2983052730560303,
+      "learning_rate": 8.234325243538352e-06,
+      "loss": 0.636,
+      "step": 18291
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.0524580478668213,
+      "learning_rate": 8.234124172131288e-06,
+      "loss": 0.5695,
+      "step": 18292
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.6885433197021484,
+      "learning_rate": 8.233923091731334e-06,
+      "loss": 0.5326,
+      "step": 18293
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.5141005516052246,
+      "learning_rate": 8.233722002339053e-06,
+      "loss": 0.6284,
+      "step": 18294
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.3482534885406494,
+      "learning_rate": 8.233520903955001e-06,
+      "loss": 0.779,
+      "step": 18295
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 3.592442274093628,
+      "learning_rate": 8.23331979657974e-06,
+      "loss": 0.954,
+      "step": 18296
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.872419834136963,
+      "learning_rate": 8.233118680213827e-06,
+      "loss": 0.6603,
+      "step": 18297
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.4104464054107666,
+      "learning_rate": 8.232917554857823e-06,
+      "loss": 0.6507,
+      "step": 18298
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.8663651943206787,
+      "learning_rate": 8.232716420512284e-06,
+      "loss": 0.4004,
+      "step": 18299
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.5813961029052734,
+      "learning_rate": 8.232515277177773e-06,
+      "loss": 0.6307,
+      "step": 18300
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.3747427463531494,
+      "learning_rate": 8.23231412485485e-06,
+      "loss": 0.591,
+      "step": 18301
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 6.356713771820068,
+      "learning_rate": 8.23211296354407e-06,
+      "loss": 0.7931,
+      "step": 18302
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.1307249069213867,
+      "learning_rate": 8.231911793245994e-06,
+      "loss": 0.7242,
+      "step": 18303
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.799553632736206,
+      "learning_rate": 8.231710613961183e-06,
+      "loss": 0.4347,
+      "step": 18304
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.32073974609375,
+      "learning_rate": 8.231509425690196e-06,
+      "loss": 0.5704,
+      "step": 18305
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.3276150226593018,
+      "learning_rate": 8.231308228433592e-06,
+      "loss": 0.9039,
+      "step": 18306
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.401951789855957,
+      "learning_rate": 8.23110702219193e-06,
+      "loss": 0.8142,
+      "step": 18307
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.8930498361587524,
+      "learning_rate": 8.23090580696577e-06,
+      "loss": 0.6089,
+      "step": 18308
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 3.769929885864258,
+      "learning_rate": 8.230704582755669e-06,
+      "loss": 0.7457,
+      "step": 18309
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.9098289012908936,
+      "learning_rate": 8.230503349562191e-06,
+      "loss": 0.8532,
+      "step": 18310
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.1216700077056885,
+      "learning_rate": 8.230302107385892e-06,
+      "loss": 0.7558,
+      "step": 18311
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 3.594024181365967,
+      "learning_rate": 8.230100856227335e-06,
+      "loss": 0.8039,
+      "step": 18312
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.0990304946899414,
+      "learning_rate": 8.229899596087076e-06,
+      "loss": 0.6324,
+      "step": 18313
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.3544039726257324,
+      "learning_rate": 8.229698326965678e-06,
+      "loss": 0.5554,
+      "step": 18314
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 3.304072856903076,
+      "learning_rate": 8.229497048863698e-06,
+      "loss": 0.8035,
+      "step": 18315
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.6481900215148926,
+      "learning_rate": 8.229295761781696e-06,
+      "loss": 0.5479,
+      "step": 18316
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.4204599857330322,
+      "learning_rate": 8.229094465720232e-06,
+      "loss": 0.7071,
+      "step": 18317
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.508414387702942,
+      "learning_rate": 8.228893160679866e-06,
+      "loss": 0.846,
+      "step": 18318
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.2863211631774902,
+      "learning_rate": 8.228691846661156e-06,
+      "loss": 0.6161,
+      "step": 18319
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.5135130882263184,
+      "learning_rate": 8.228490523664666e-06,
+      "loss": 0.614,
+      "step": 18320
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.225518226623535,
+      "learning_rate": 8.228289191690953e-06,
+      "loss": 0.6356,
+      "step": 18321
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.8779864311218262,
+      "learning_rate": 8.228087850740576e-06,
+      "loss": 0.7531,
+      "step": 18322
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.0408992767333984,
+      "learning_rate": 8.227886500814096e-06,
+      "loss": 0.559,
+      "step": 18323
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 3.1477818489074707,
+      "learning_rate": 8.227685141912074e-06,
+      "loss": 0.7629,
+      "step": 18324
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.3482253551483154,
+      "learning_rate": 8.227483774035067e-06,
+      "loss": 0.3801,
+      "step": 18325
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.7527709007263184,
+      "learning_rate": 8.227282397183638e-06,
+      "loss": 0.4221,
+      "step": 18326
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.5362740755081177,
+      "learning_rate": 8.227081011358347e-06,
+      "loss": 0.4436,
+      "step": 18327
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.92512845993042,
+      "learning_rate": 8.226879616559751e-06,
+      "loss": 0.6004,
+      "step": 18328
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.871050238609314,
+      "learning_rate": 8.22667821278841e-06,
+      "loss": 0.4877,
+      "step": 18329
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.924262762069702,
+      "learning_rate": 8.226476800044886e-06,
+      "loss": 0.5533,
+      "step": 18330
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.1755118370056152,
+      "learning_rate": 8.22627537832974e-06,
+      "loss": 0.5916,
+      "step": 18331
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.7925490140914917,
+      "learning_rate": 8.226073947643531e-06,
+      "loss": 0.5298,
+      "step": 18332
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.600537061691284,
+      "learning_rate": 8.22587250798682e-06,
+      "loss": 0.4466,
+      "step": 18333
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.0611329078674316,
+      "learning_rate": 8.225671059360164e-06,
+      "loss": 0.5266,
+      "step": 18334
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 3.141383409500122,
+      "learning_rate": 8.225469601764125e-06,
+      "loss": 0.7099,
+      "step": 18335
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.5171194076538086,
+      "learning_rate": 8.225268135199265e-06,
+      "loss": 0.7585,
+      "step": 18336
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 3.0456511974334717,
+      "learning_rate": 8.225066659666141e-06,
+      "loss": 0.5293,
+      "step": 18337
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.4856810569763184,
+      "learning_rate": 8.224865175165315e-06,
+      "loss": 0.7584,
+      "step": 18338
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.7235054969787598,
+      "learning_rate": 8.224663681697347e-06,
+      "loss": 0.607,
+      "step": 18339
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.684066653251648,
+      "learning_rate": 8.224462179262797e-06,
+      "loss": 0.6726,
+      "step": 18340
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.890605926513672,
+      "learning_rate": 8.224260667862227e-06,
+      "loss": 0.6781,
+      "step": 18341
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 3.1481668949127197,
+      "learning_rate": 8.224059147496194e-06,
+      "loss": 0.7197,
+      "step": 18342
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.1188509464263916,
+      "learning_rate": 8.223857618165263e-06,
+      "loss": 0.7751,
+      "step": 18343
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.147364854812622,
+      "learning_rate": 8.223656079869991e-06,
+      "loss": 0.5788,
+      "step": 18344
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.173020601272583,
+      "learning_rate": 8.223454532610937e-06,
+      "loss": 0.7637,
+      "step": 18345
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.511274814605713,
+      "learning_rate": 8.223252976388665e-06,
+      "loss": 0.8706,
+      "step": 18346
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 3.7734789848327637,
+      "learning_rate": 8.223051411203734e-06,
+      "loss": 0.7866,
+      "step": 18347
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.8801894187927246,
+      "learning_rate": 8.222849837056703e-06,
+      "loss": 0.8122,
+      "step": 18348
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 3.9930663108825684,
+      "learning_rate": 8.222648253948134e-06,
+      "loss": 0.7226,
+      "step": 18349
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.3558645248413086,
+      "learning_rate": 8.22244666187859e-06,
+      "loss": 0.4327,
+      "step": 18350
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.5658440589904785,
+      "learning_rate": 8.222245060848626e-06,
+      "loss": 0.5967,
+      "step": 18351
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 3.2431747913360596,
+      "learning_rate": 8.222043450858807e-06,
+      "loss": 0.7596,
+      "step": 18352
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 4.478899955749512,
+      "learning_rate": 8.221841831909691e-06,
+      "loss": 0.406,
+      "step": 18353
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.8400824069976807,
+      "learning_rate": 8.221640204001841e-06,
+      "loss": 0.6973,
+      "step": 18354
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 3.544057846069336,
+      "learning_rate": 8.221438567135817e-06,
+      "loss": 0.6965,
+      "step": 18355
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 3.220451831817627,
+      "learning_rate": 8.221236921312179e-06,
+      "loss": 0.9185,
+      "step": 18356
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.2235560417175293,
+      "learning_rate": 8.221035266531487e-06,
+      "loss": 0.6416,
+      "step": 18357
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 4.39286470413208,
+      "learning_rate": 8.220833602794303e-06,
+      "loss": 0.7247,
+      "step": 18358
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.678171157836914,
+      "learning_rate": 8.220631930101186e-06,
+      "loss": 0.6666,
+      "step": 18359
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.9260815382003784,
+      "learning_rate": 8.220430248452699e-06,
+      "loss": 0.8718,
+      "step": 18360
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.3461177349090576,
+      "learning_rate": 8.220228557849402e-06,
+      "loss": 0.8237,
+      "step": 18361
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.7020206451416016,
+      "learning_rate": 8.220026858291857e-06,
+      "loss": 0.7128,
+      "step": 18362
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.0549368858337402,
+      "learning_rate": 8.21982514978062e-06,
+      "loss": 0.6545,
+      "step": 18363
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.5290348529815674,
+      "learning_rate": 8.21962343231626e-06,
+      "loss": 0.6206,
+      "step": 18364
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.911355495452881,
+      "learning_rate": 8.21942170589933e-06,
+      "loss": 0.6132,
+      "step": 18365
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.2178797721862793,
+      "learning_rate": 8.219219970530394e-06,
+      "loss": 0.6511,
+      "step": 18366
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.5925289392471313,
+      "learning_rate": 8.219018226210016e-06,
+      "loss": 0.5961,
+      "step": 18367
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.807129979133606,
+      "learning_rate": 8.218816472938753e-06,
+      "loss": 0.5558,
+      "step": 18368
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 3.496474504470825,
+      "learning_rate": 8.218614710717165e-06,
+      "loss": 0.4724,
+      "step": 18369
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.018871307373047,
+      "learning_rate": 8.218412939545818e-06,
+      "loss": 0.5454,
+      "step": 18370
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 3.157931089401245,
+      "learning_rate": 8.218211159425269e-06,
+      "loss": 0.7549,
+      "step": 18371
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.646048069000244,
+      "learning_rate": 8.21800937035608e-06,
+      "loss": 0.7916,
+      "step": 18372
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.3398306369781494,
+      "learning_rate": 8.217807572338814e-06,
+      "loss": 0.6603,
+      "step": 18373
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 3.0379090309143066,
+      "learning_rate": 8.217605765374029e-06,
+      "loss": 0.7745,
+      "step": 18374
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.221001386642456,
+      "learning_rate": 8.217403949462289e-06,
+      "loss": 0.7234,
+      "step": 18375
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 4.29793643951416,
+      "learning_rate": 8.217202124604153e-06,
+      "loss": 0.5803,
+      "step": 18376
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.3617289066314697,
+      "learning_rate": 8.217000290800183e-06,
+      "loss": 0.6624,
+      "step": 18377
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.078894853591919,
+      "learning_rate": 8.21679844805094e-06,
+      "loss": 1.0052,
+      "step": 18378
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.310671091079712,
+      "learning_rate": 8.216596596356986e-06,
+      "loss": 0.7713,
+      "step": 18379
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.832528829574585,
+      "learning_rate": 8.216394735718883e-06,
+      "loss": 0.4647,
+      "step": 18380
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.9142709970474243,
+      "learning_rate": 8.21619286613719e-06,
+      "loss": 0.6429,
+      "step": 18381
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.1850740909576416,
+      "learning_rate": 8.21599098761247e-06,
+      "loss": 0.6648,
+      "step": 18382
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.020275115966797,
+      "learning_rate": 8.215789100145283e-06,
+      "loss": 0.8253,
+      "step": 18383
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 4.094560146331787,
+      "learning_rate": 8.215587203736192e-06,
+      "loss": 0.7003,
+      "step": 18384
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.8416253328323364,
+      "learning_rate": 8.215385298385757e-06,
+      "loss": 0.5036,
+      "step": 18385
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.9782003164291382,
+      "learning_rate": 8.215183384094541e-06,
+      "loss": 0.5041,
+      "step": 18386
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.2395668029785156,
+      "learning_rate": 8.214981460863103e-06,
+      "loss": 0.6318,
+      "step": 18387
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.8126940727233887,
+      "learning_rate": 8.214779528692006e-06,
+      "loss": 0.6806,
+      "step": 18388
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.3914096355438232,
+      "learning_rate": 8.214577587581812e-06,
+      "loss": 0.7833,
+      "step": 18389
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 3.0238962173461914,
+      "learning_rate": 8.21437563753308e-06,
+      "loss": 0.8027,
+      "step": 18390
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.988429307937622,
+      "learning_rate": 8.214173678546377e-06,
+      "loss": 0.7912,
+      "step": 18391
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.176206350326538,
+      "learning_rate": 8.213971710622259e-06,
+      "loss": 0.6361,
+      "step": 18392
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.0345675945281982,
+      "learning_rate": 8.21376973376129e-06,
+      "loss": 0.7515,
+      "step": 18393
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.6086623668670654,
+      "learning_rate": 8.213567747964032e-06,
+      "loss": 0.6996,
+      "step": 18394
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.202591896057129,
+      "learning_rate": 8.213365753231046e-06,
+      "loss": 0.6006,
+      "step": 18395
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.185018539428711,
+      "learning_rate": 8.21316374956289e-06,
+      "loss": 0.587,
+      "step": 18396
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.334350824356079,
+      "learning_rate": 8.212961736960134e-06,
+      "loss": 0.6702,
+      "step": 18397
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.551978588104248,
+      "learning_rate": 8.212759715423333e-06,
+      "loss": 0.5332,
+      "step": 18398
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.278597593307495,
+      "learning_rate": 8.212557684953051e-06,
+      "loss": 0.6798,
+      "step": 18399
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.481501579284668,
+      "learning_rate": 8.21235564554985e-06,
+      "loss": 0.8243,
+      "step": 18400
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.7729263305664062,
+      "learning_rate": 8.21215359721429e-06,
+      "loss": 0.6098,
+      "step": 18401
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.2327892780303955,
+      "learning_rate": 8.211951539946934e-06,
+      "loss": 0.6186,
+      "step": 18402
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 3.164842367172241,
+      "learning_rate": 8.211749473748345e-06,
+      "loss": 0.7115,
+      "step": 18403
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.1023027896881104,
+      "learning_rate": 8.211547398619085e-06,
+      "loss": 0.6123,
+      "step": 18404
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.115919589996338,
+      "learning_rate": 8.211345314559714e-06,
+      "loss": 0.8875,
+      "step": 18405
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 4.283769130706787,
+      "learning_rate": 8.211143221570793e-06,
+      "loss": 0.8071,
+      "step": 18406
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 5.442386627197266,
+      "learning_rate": 8.210941119652887e-06,
+      "loss": 0.6659,
+      "step": 18407
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.4814250469207764,
+      "learning_rate": 8.210739008806557e-06,
+      "loss": 0.4831,
+      "step": 18408
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 3.6339879035949707,
+      "learning_rate": 8.210536889032363e-06,
+      "loss": 0.5577,
+      "step": 18409
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 7.880307197570801,
+      "learning_rate": 8.21033476033087e-06,
+      "loss": 0.8497,
+      "step": 18410
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.797489643096924,
+      "learning_rate": 8.210132622702638e-06,
+      "loss": 0.7662,
+      "step": 18411
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.3756392002105713,
+      "learning_rate": 8.20993047614823e-06,
+      "loss": 0.9229,
+      "step": 18412
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.8493199348449707,
+      "learning_rate": 8.209728320668207e-06,
+      "loss": 0.5864,
+      "step": 18413
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.4482333660125732,
+      "learning_rate": 8.209526156263133e-06,
+      "loss": 0.7873,
+      "step": 18414
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.857262134552002,
+      "learning_rate": 8.20932398293357e-06,
+      "loss": 0.4729,
+      "step": 18415
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.6500393152236938,
+      "learning_rate": 8.209121800680077e-06,
+      "loss": 0.4915,
+      "step": 18416
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.6313915252685547,
+      "learning_rate": 8.20891960950322e-06,
+      "loss": 0.7349,
+      "step": 18417
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.5089151859283447,
+      "learning_rate": 8.20871740940356e-06,
+      "loss": 0.7058,
+      "step": 18418
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.8853389024734497,
+      "learning_rate": 8.208515200381656e-06,
+      "loss": 0.4668,
+      "step": 18419
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.949216604232788,
+      "learning_rate": 8.208312982438076e-06,
+      "loss": 0.7261,
+      "step": 18420
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.88038969039917,
+      "learning_rate": 8.208110755573378e-06,
+      "loss": 0.6787,
+      "step": 18421
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.68015456199646,
+      "learning_rate": 8.207908519788128e-06,
+      "loss": 0.6997,
+      "step": 18422
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.804913878440857,
+      "learning_rate": 8.207706275082883e-06,
+      "loss": 0.4657,
+      "step": 18423
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.134260416030884,
+      "learning_rate": 8.20750402145821e-06,
+      "loss": 0.6466,
+      "step": 18424
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 3.8190810680389404,
+      "learning_rate": 8.207301758914671e-06,
+      "loss": 0.6305,
+      "step": 18425
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.826855182647705,
+      "learning_rate": 8.207099487452825e-06,
+      "loss": 0.6073,
+      "step": 18426
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.987534284591675,
+      "learning_rate": 8.20689720707324e-06,
+      "loss": 0.5155,
+      "step": 18427
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.0171058177948,
+      "learning_rate": 8.206694917776472e-06,
+      "loss": 0.6921,
+      "step": 18428
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.2450199127197266,
+      "learning_rate": 8.206492619563088e-06,
+      "loss": 0.5996,
+      "step": 18429
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 4.514499664306641,
+      "learning_rate": 8.20629031243365e-06,
+      "loss": 0.867,
+      "step": 18430
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.9210888147354126,
+      "learning_rate": 8.206087996388717e-06,
+      "loss": 0.3996,
+      "step": 18431
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.9496238231658936,
+      "learning_rate": 8.205885671428858e-06,
+      "loss": 0.625,
+      "step": 18432
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.492403268814087,
+      "learning_rate": 8.205683337554629e-06,
+      "loss": 0.5884,
+      "step": 18433
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.470468759536743,
+      "learning_rate": 8.205480994766594e-06,
+      "loss": 0.8088,
+      "step": 18434
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.9757611751556396,
+      "learning_rate": 8.205278643065322e-06,
+      "loss": 0.7561,
+      "step": 18435
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.4457536935806274,
+      "learning_rate": 8.205076282451368e-06,
+      "loss": 0.6641,
+      "step": 18436
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.327634572982788,
+      "learning_rate": 8.204873912925298e-06,
+      "loss": 0.7578,
+      "step": 18437
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 3.1568782329559326,
+      "learning_rate": 8.204671534487673e-06,
+      "loss": 0.7878,
+      "step": 18438
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.600313663482666,
+      "learning_rate": 8.20446914713906e-06,
+      "loss": 0.4065,
+      "step": 18439
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.3280463218688965,
+      "learning_rate": 8.204266750880018e-06,
+      "loss": 0.8784,
+      "step": 18440
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.2785439491271973,
+      "learning_rate": 8.204064345711107e-06,
+      "loss": 0.5481,
+      "step": 18441
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 3.491201400756836,
+      "learning_rate": 8.203861931632895e-06,
+      "loss": 0.4876,
+      "step": 18442
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.7851405143737793,
+      "learning_rate": 8.203659508645945e-06,
+      "loss": 0.6315,
+      "step": 18443
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 3.1538591384887695,
+      "learning_rate": 8.203457076750815e-06,
+      "loss": 0.5891,
+      "step": 18444
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.3408923149108887,
+      "learning_rate": 8.203254635948075e-06,
+      "loss": 0.6901,
+      "step": 18445
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.907212018966675,
+      "learning_rate": 8.20305218623828e-06,
+      "loss": 0.636,
+      "step": 18446
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 3.9665017127990723,
+      "learning_rate": 8.202849727622e-06,
+      "loss": 0.5145,
+      "step": 18447
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.2238504886627197,
+      "learning_rate": 8.202647260099793e-06,
+      "loss": 0.4859,
+      "step": 18448
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.9614918231964111,
+      "learning_rate": 8.202444783672223e-06,
+      "loss": 0.4317,
+      "step": 18449
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.7887424230575562,
+      "learning_rate": 8.202242298339855e-06,
+      "loss": 0.6311,
+      "step": 18450
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 3.0846283435821533,
+      "learning_rate": 8.20203980410325e-06,
+      "loss": 0.5503,
+      "step": 18451
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.3420681953430176,
+      "learning_rate": 8.20183730096297e-06,
+      "loss": 0.5894,
+      "step": 18452
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.4324545860290527,
+      "learning_rate": 8.201634788919583e-06,
+      "loss": 0.8668,
+      "step": 18453
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.143314838409424,
+      "learning_rate": 8.201432267973648e-06,
+      "loss": 0.5218,
+      "step": 18454
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.4019129276275635,
+      "learning_rate": 8.201229738125731e-06,
+      "loss": 0.5135,
+      "step": 18455
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.7736177444458008,
+      "learning_rate": 8.201027199376391e-06,
+      "loss": 0.6195,
+      "step": 18456
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 3.6729016304016113,
+      "learning_rate": 8.200824651726193e-06,
+      "loss": 0.707,
+      "step": 18457
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.2832727432250977,
+      "learning_rate": 8.200622095175703e-06,
+      "loss": 0.8421,
+      "step": 18458
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 3.060138702392578,
+      "learning_rate": 8.200419529725481e-06,
+      "loss": 0.6795,
+      "step": 18459
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.950195550918579,
+      "learning_rate": 8.20021695537609e-06,
+      "loss": 0.6956,
+      "step": 18460
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 3.030012607574463,
+      "learning_rate": 8.200014372128095e-06,
+      "loss": 0.6602,
+      "step": 18461
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.4890053272247314,
+      "learning_rate": 8.199811779982059e-06,
+      "loss": 0.6486,
+      "step": 18462
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.054145097732544,
+      "learning_rate": 8.199609178938546e-06,
+      "loss": 0.5229,
+      "step": 18463
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.5194213390350342,
+      "learning_rate": 8.199406568998116e-06,
+      "loss": 0.4255,
+      "step": 18464
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.872126817703247,
+      "learning_rate": 8.199203950161337e-06,
+      "loss": 0.5076,
+      "step": 18465
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.93718421459198,
+      "learning_rate": 8.19900132242877e-06,
+      "loss": 0.7975,
+      "step": 18466
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.9918196201324463,
+      "learning_rate": 8.19879868580098e-06,
+      "loss": 0.6844,
+      "step": 18467
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.245298385620117,
+      "learning_rate": 8.198596040278526e-06,
+      "loss": 0.6151,
+      "step": 18468
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 3.7562978267669678,
+      "learning_rate": 8.198393385861976e-06,
+      "loss": 0.5956,
+      "step": 18469
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.8163714408874512,
+      "learning_rate": 8.198190722551893e-06,
+      "loss": 0.8064,
+      "step": 18470
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.1234545707702637,
+      "learning_rate": 8.197988050348838e-06,
+      "loss": 0.8192,
+      "step": 18471
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.5280215740203857,
+      "learning_rate": 8.197785369253377e-06,
+      "loss": 0.4864,
+      "step": 18472
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.930911064147949,
+      "learning_rate": 8.197582679266075e-06,
+      "loss": 0.6937,
+      "step": 18473
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.007611036300659,
+      "learning_rate": 8.19737998038749e-06,
+      "loss": 0.5428,
+      "step": 18474
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 3.462149143218994,
+      "learning_rate": 8.197177272618192e-06,
+      "loss": 0.786,
+      "step": 18475
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.918261766433716,
+      "learning_rate": 8.196974555958739e-06,
+      "loss": 0.6708,
+      "step": 18476
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.5195183753967285,
+      "learning_rate": 8.196771830409699e-06,
+      "loss": 0.8052,
+      "step": 18477
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 3.573291063308716,
+      "learning_rate": 8.196569095971634e-06,
+      "loss": 0.7598,
+      "step": 18478
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.3321239948272705,
+      "learning_rate": 8.196366352645107e-06,
+      "loss": 0.4871,
+      "step": 18479
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.76165771484375,
+      "learning_rate": 8.196163600430684e-06,
+      "loss": 0.5589,
+      "step": 18480
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.6634035110473633,
+      "learning_rate": 8.195960839328925e-06,
+      "loss": 0.686,
+      "step": 18481
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.9643670320510864,
+      "learning_rate": 8.195758069340398e-06,
+      "loss": 0.6002,
+      "step": 18482
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.5533266067504883,
+      "learning_rate": 8.195555290465665e-06,
+      "loss": 0.489,
+      "step": 18483
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.244325876235962,
+      "learning_rate": 8.195352502705289e-06,
+      "loss": 0.6919,
+      "step": 18484
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.3915278911590576,
+      "learning_rate": 8.195149706059835e-06,
+      "loss": 0.4545,
+      "step": 18485
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.9138524532318115,
+      "learning_rate": 8.194946900529867e-06,
+      "loss": 0.6677,
+      "step": 18486
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.4327056407928467,
+      "learning_rate": 8.194744086115947e-06,
+      "loss": 0.6786,
+      "step": 18487
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.1743690967559814,
+      "learning_rate": 8.194541262818643e-06,
+      "loss": 0.8202,
+      "step": 18488
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.7290098667144775,
+      "learning_rate": 8.194338430638514e-06,
+      "loss": 0.5212,
+      "step": 18489
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.722132444381714,
+      "learning_rate": 8.194135589576128e-06,
+      "loss": 0.5994,
+      "step": 18490
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.5874674320220947,
+      "learning_rate": 8.193932739632047e-06,
+      "loss": 0.6385,
+      "step": 18491
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.290510654449463,
+      "learning_rate": 8.193729880806836e-06,
+      "loss": 0.5454,
+      "step": 18492
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.3168351650238037,
+      "learning_rate": 8.193527013101056e-06,
+      "loss": 0.7657,
+      "step": 18493
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.6538174152374268,
+      "learning_rate": 8.193324136515278e-06,
+      "loss": 0.6748,
+      "step": 18494
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.232485771179199,
+      "learning_rate": 8.193121251050058e-06,
+      "loss": 0.4926,
+      "step": 18495
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.9425363540649414,
+      "learning_rate": 8.192918356705966e-06,
+      "loss": 0.6026,
+      "step": 18496
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.7984081506729126,
+      "learning_rate": 8.192715453483564e-06,
+      "loss": 0.4993,
+      "step": 18497
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.491029977798462,
+      "learning_rate": 8.192512541383417e-06,
+      "loss": 0.3315,
+      "step": 18498
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 5.776723384857178,
+      "learning_rate": 8.192309620406086e-06,
+      "loss": 0.7405,
+      "step": 18499
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.8652732372283936,
+      "learning_rate": 8.19210669055214e-06,
+      "loss": 0.7184,
+      "step": 18500
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.2038381099700928,
+      "learning_rate": 8.191903751822141e-06,
+      "loss": 0.4445,
+      "step": 18501
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.2485857009887695,
+      "learning_rate": 8.191700804216653e-06,
+      "loss": 0.8774,
+      "step": 18502
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.5623078346252441,
+      "learning_rate": 8.19149784773624e-06,
+      "loss": 0.6294,
+      "step": 18503
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.7377288341522217,
+      "learning_rate": 8.191294882381468e-06,
+      "loss": 0.3886,
+      "step": 18504
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.1545867919921875,
+      "learning_rate": 8.191091908152899e-06,
+      "loss": 0.5413,
+      "step": 18505
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.98030948638916,
+      "learning_rate": 8.1908889250511e-06,
+      "loss": 0.9532,
+      "step": 18506
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 3.5675814151763916,
+      "learning_rate": 8.190685933076636e-06,
+      "loss": 0.9488,
+      "step": 18507
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.7016751766204834,
+      "learning_rate": 8.190482932230067e-06,
+      "loss": 0.4389,
+      "step": 18508
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.1954007148742676,
+      "learning_rate": 8.190279922511961e-06,
+      "loss": 0.7624,
+      "step": 18509
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.9572124481201172,
+      "learning_rate": 8.19007690392288e-06,
+      "loss": 0.6461,
+      "step": 18510
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.6726126670837402,
+      "learning_rate": 8.189873876463394e-06,
+      "loss": 0.8374,
+      "step": 18511
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 3.4092535972595215,
+      "learning_rate": 8.18967084013406e-06,
+      "loss": 0.5766,
+      "step": 18512
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.4050068855285645,
+      "learning_rate": 8.189467794935449e-06,
+      "loss": 0.8742,
+      "step": 18513
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.7208558320999146,
+      "learning_rate": 8.18926474086812e-06,
+      "loss": 0.8403,
+      "step": 18514
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.6568323373794556,
+      "learning_rate": 8.189061677932644e-06,
+      "loss": 0.5214,
+      "step": 18515
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.1675219535827637,
+      "learning_rate": 8.18885860612958e-06,
+      "loss": 0.708,
+      "step": 18516
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.4814741611480713,
+      "learning_rate": 8.188655525459495e-06,
+      "loss": 0.6694,
+      "step": 18517
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.5503766536712646,
+      "learning_rate": 8.188452435922953e-06,
+      "loss": 0.6476,
+      "step": 18518
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.790346622467041,
+      "learning_rate": 8.18824933752052e-06,
+      "loss": 0.5097,
+      "step": 18519
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.6022720336914062,
+      "learning_rate": 8.18804623025276e-06,
+      "loss": 0.6263,
+      "step": 18520
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.1615421772003174,
+      "learning_rate": 8.187843114120236e-06,
+      "loss": 0.6265,
+      "step": 18521
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.074697256088257,
+      "learning_rate": 8.187639989123516e-06,
+      "loss": 0.7043,
+      "step": 18522
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.9815385341644287,
+      "learning_rate": 8.187436855263164e-06,
+      "loss": 0.7821,
+      "step": 18523
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.738116502761841,
+      "learning_rate": 8.187233712539743e-06,
+      "loss": 0.5455,
+      "step": 18524
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.010871171951294,
+      "learning_rate": 8.187030560953819e-06,
+      "loss": 0.5942,
+      "step": 18525
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.441495180130005,
+      "learning_rate": 8.186827400505956e-06,
+      "loss": 0.531,
+      "step": 18526
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.037285566329956,
+      "learning_rate": 8.186624231196723e-06,
+      "loss": 0.6542,
+      "step": 18527
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.0409300327301025,
+      "learning_rate": 8.186421053026679e-06,
+      "loss": 0.7322,
+      "step": 18528
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.8951711654663086,
+      "learning_rate": 8.186217865996392e-06,
+      "loss": 0.6433,
+      "step": 18529
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.9559582471847534,
+      "learning_rate": 8.186014670106427e-06,
+      "loss": 0.7428,
+      "step": 18530
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 5.332690238952637,
+      "learning_rate": 8.185811465357349e-06,
+      "loss": 1.0048,
+      "step": 18531
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.723840355873108,
+      "learning_rate": 8.185608251749722e-06,
+      "loss": 0.4183,
+      "step": 18532
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.8968520164489746,
+      "learning_rate": 8.185405029284113e-06,
+      "loss": 0.7584,
+      "step": 18533
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.5343129634857178,
+      "learning_rate": 8.185201797961085e-06,
+      "loss": 0.6788,
+      "step": 18534
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.6082569360733032,
+      "learning_rate": 8.184998557781206e-06,
+      "loss": 0.7232,
+      "step": 18535
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.0362489223480225,
+      "learning_rate": 8.184795308745035e-06,
+      "loss": 0.4435,
+      "step": 18536
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.1844606399536133,
+      "learning_rate": 8.184592050853145e-06,
+      "loss": 0.5404,
+      "step": 18537
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.8844523429870605,
+      "learning_rate": 8.184388784106097e-06,
+      "loss": 0.5035,
+      "step": 18538
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.971634864807129,
+      "learning_rate": 8.184185508504455e-06,
+      "loss": 0.6794,
+      "step": 18539
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.6157500743865967,
+      "learning_rate": 8.183982224048787e-06,
+      "loss": 0.6787,
+      "step": 18540
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.1158101558685303,
+      "learning_rate": 8.183778930739658e-06,
+      "loss": 0.7091,
+      "step": 18541
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.6502511501312256,
+      "learning_rate": 8.18357562857763e-06,
+      "loss": 0.7433,
+      "step": 18542
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.2444956302642822,
+      "learning_rate": 8.183372317563272e-06,
+      "loss": 0.5741,
+      "step": 18543
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.7961277961730957,
+      "learning_rate": 8.183168997697148e-06,
+      "loss": 0.8447,
+      "step": 18544
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.6843888759613037,
+      "learning_rate": 8.182965668979825e-06,
+      "loss": 0.6488,
+      "step": 18545
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 4.159528732299805,
+      "learning_rate": 8.182762331411866e-06,
+      "loss": 0.4935,
+      "step": 18546
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.725709080696106,
+      "learning_rate": 8.182558984993835e-06,
+      "loss": 0.6773,
+      "step": 18547
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.208270311355591,
+      "learning_rate": 8.182355629726304e-06,
+      "loss": 0.5349,
+      "step": 18548
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.9180606603622437,
+      "learning_rate": 8.18215226560983e-06,
+      "loss": 0.5692,
+      "step": 18549
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.09674072265625,
+      "learning_rate": 8.181948892644985e-06,
+      "loss": 0.5517,
+      "step": 18550
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.0322320461273193,
+      "learning_rate": 8.181745510832331e-06,
+      "loss": 0.6984,
+      "step": 18551
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 4.339781284332275,
+      "learning_rate": 8.181542120172436e-06,
+      "loss": 0.8338,
+      "step": 18552
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.8950204849243164,
+      "learning_rate": 8.181338720665863e-06,
+      "loss": 0.6742,
+      "step": 18553
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.4841220378875732,
+      "learning_rate": 8.181135312313178e-06,
+      "loss": 0.8523,
+      "step": 18554
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.8996989727020264,
+      "learning_rate": 8.18093189511495e-06,
+      "loss": 0.6452,
+      "step": 18555
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.7017062902450562,
+      "learning_rate": 8.18072846907174e-06,
+      "loss": 0.6469,
+      "step": 18556
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 4.3074140548706055,
+      "learning_rate": 8.180525034184116e-06,
+      "loss": 0.5909,
+      "step": 18557
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.6865242719650269,
+      "learning_rate": 8.180321590452643e-06,
+      "loss": 0.4758,
+      "step": 18558
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.7874479293823242,
+      "learning_rate": 8.180118137877888e-06,
+      "loss": 0.7149,
+      "step": 18559
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.3362019062042236,
+      "learning_rate": 8.179914676460415e-06,
+      "loss": 0.4568,
+      "step": 18560
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.5856412649154663,
+      "learning_rate": 8.179711206200792e-06,
+      "loss": 0.6425,
+      "step": 18561
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.45819354057312,
+      "learning_rate": 8.17950772709958e-06,
+      "loss": 0.4378,
+      "step": 18562
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.76138973236084,
+      "learning_rate": 8.179304239157351e-06,
+      "loss": 0.5929,
+      "step": 18563
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.776341199874878,
+      "learning_rate": 8.179100742374665e-06,
+      "loss": 0.6745,
+      "step": 18564
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.8134384155273438,
+      "learning_rate": 8.178897236752094e-06,
+      "loss": 0.7445,
+      "step": 18565
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.8153181076049805,
+      "learning_rate": 8.1786937222902e-06,
+      "loss": 0.6631,
+      "step": 18566
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.2479262351989746,
+      "learning_rate": 8.178490198989548e-06,
+      "loss": 0.1947,
+      "step": 18567
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.929681420326233,
+      "learning_rate": 8.178286666850707e-06,
+      "loss": 0.6118,
+      "step": 18568
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.5629682540893555,
+      "learning_rate": 8.178083125874242e-06,
+      "loss": 0.5158,
+      "step": 18569
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 3.1305179595947266,
+      "learning_rate": 8.177879576060717e-06,
+      "loss": 0.6437,
+      "step": 18570
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.1371536254882812,
+      "learning_rate": 8.177676017410699e-06,
+      "loss": 0.5672,
+      "step": 18571
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.6003336906433105,
+      "learning_rate": 8.177472449924755e-06,
+      "loss": 0.7143,
+      "step": 18572
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.3657374382019043,
+      "learning_rate": 8.17726887360345e-06,
+      "loss": 0.7622,
+      "step": 18573
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.6334525346755981,
+      "learning_rate": 8.17706528844735e-06,
+      "loss": 0.506,
+      "step": 18574
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.4720486402511597,
+      "learning_rate": 8.176861694457024e-06,
+      "loss": 0.4566,
+      "step": 18575
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.9565176963806152,
+      "learning_rate": 8.176658091633033e-06,
+      "loss": 0.6103,
+      "step": 18576
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.9597389698028564,
+      "learning_rate": 8.176454479975946e-06,
+      "loss": 0.7346,
+      "step": 18577
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.255878448486328,
+      "learning_rate": 8.17625085948633e-06,
+      "loss": 0.8186,
+      "step": 18578
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.390946865081787,
+      "learning_rate": 8.17604723016475e-06,
+      "loss": 0.4788,
+      "step": 18579
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.360258102416992,
+      "learning_rate": 8.175843592011774e-06,
+      "loss": 0.6729,
+      "step": 18580
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.8635352849960327,
+      "learning_rate": 8.175639945027964e-06,
+      "loss": 0.6791,
+      "step": 18581
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.2853822708129883,
+      "learning_rate": 8.175436289213892e-06,
+      "loss": 0.5171,
+      "step": 18582
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.86007022857666,
+      "learning_rate": 8.175232624570118e-06,
+      "loss": 0.6315,
+      "step": 18583
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.6155245304107666,
+      "learning_rate": 8.175028951097213e-06,
+      "loss": 0.8571,
+      "step": 18584
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.910306453704834,
+      "learning_rate": 8.174825268795742e-06,
+      "loss": 0.7098,
+      "step": 18585
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.78853178024292,
+      "learning_rate": 8.17462157766627e-06,
+      "loss": 0.6749,
+      "step": 18586
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.5383996963500977,
+      "learning_rate": 8.174417877709366e-06,
+      "loss": 0.4567,
+      "step": 18587
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 3.6272151470184326,
+      "learning_rate": 8.174214168925595e-06,
+      "loss": 0.8201,
+      "step": 18588
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.301647186279297,
+      "learning_rate": 8.174010451315523e-06,
+      "loss": 0.8063,
+      "step": 18589
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.909149408340454,
+      "learning_rate": 8.173806724879718e-06,
+      "loss": 0.6565,
+      "step": 18590
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.296949625015259,
+      "learning_rate": 8.173602989618745e-06,
+      "loss": 0.7562,
+      "step": 18591
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.5448070764541626,
+      "learning_rate": 8.17339924553317e-06,
+      "loss": 0.6613,
+      "step": 18592
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.1482863426208496,
+      "learning_rate": 8.17319549262356e-06,
+      "loss": 0.5169,
+      "step": 18593
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 3.128370761871338,
+      "learning_rate": 8.172991730890483e-06,
+      "loss": 0.4995,
+      "step": 18594
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.1853413581848145,
+      "learning_rate": 8.172787960334505e-06,
+      "loss": 0.5819,
+      "step": 18595
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.3896734714508057,
+      "learning_rate": 8.172584180956192e-06,
+      "loss": 0.8157,
+      "step": 18596
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.6958377361297607,
+      "learning_rate": 8.172380392756111e-06,
+      "loss": 0.5237,
+      "step": 18597
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.849095344543457,
+      "learning_rate": 8.172176595734828e-06,
+      "loss": 0.4315,
+      "step": 18598
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.5457631349563599,
+      "learning_rate": 8.171972789892912e-06,
+      "loss": 0.7009,
+      "step": 18599
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.786911964416504,
+      "learning_rate": 8.171768975230926e-06,
+      "loss": 0.5552,
+      "step": 18600
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.9137721061706543,
+      "learning_rate": 8.171565151749438e-06,
+      "loss": 0.7277,
+      "step": 18601
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.962329626083374,
+      "learning_rate": 8.171361319449017e-06,
+      "loss": 0.5359,
+      "step": 18602
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.3711795806884766,
+      "learning_rate": 8.171157478330227e-06,
+      "loss": 0.7583,
+      "step": 18603
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 3.115387201309204,
+      "learning_rate": 8.170953628393637e-06,
+      "loss": 0.7497,
+      "step": 18604
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 3.658860445022583,
+      "learning_rate": 8.170749769639813e-06,
+      "loss": 0.756,
+      "step": 18605
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.9930193424224854,
+      "learning_rate": 8.170545902069322e-06,
+      "loss": 0.5881,
+      "step": 18606
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.0029516220092773,
+      "learning_rate": 8.170342025682729e-06,
+      "loss": 0.8077,
+      "step": 18607
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.5369040966033936,
+      "learning_rate": 8.170138140480603e-06,
+      "loss": 0.7948,
+      "step": 18608
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 3.5128026008605957,
+      "learning_rate": 8.169934246463512e-06,
+      "loss": 0.3991,
+      "step": 18609
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.2597591876983643,
+      "learning_rate": 8.16973034363202e-06,
+      "loss": 0.6584,
+      "step": 18610
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.2778356075286865,
+      "learning_rate": 8.169526431986694e-06,
+      "loss": 0.5476,
+      "step": 18611
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.1214213371276855,
+      "learning_rate": 8.169322511528105e-06,
+      "loss": 0.6958,
+      "step": 18612
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 3.0737361907958984,
+      "learning_rate": 8.169118582256815e-06,
+      "loss": 0.6515,
+      "step": 18613
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.934880018234253,
+      "learning_rate": 8.168914644173395e-06,
+      "loss": 0.6123,
+      "step": 18614
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.425764322280884,
+      "learning_rate": 8.168710697278409e-06,
+      "loss": 0.5034,
+      "step": 18615
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.0624778270721436,
+      "learning_rate": 8.168506741572427e-06,
+      "loss": 0.653,
+      "step": 18616
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.677151679992676,
+      "learning_rate": 8.168302777056015e-06,
+      "loss": 0.569,
+      "step": 18617
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.429814577102661,
+      "learning_rate": 8.168098803729738e-06,
+      "loss": 0.5897,
+      "step": 18618
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.048182964324951,
+      "learning_rate": 8.167894821594166e-06,
+      "loss": 0.4755,
+      "step": 18619
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.3188693523406982,
+      "learning_rate": 8.167690830649865e-06,
+      "loss": 0.849,
+      "step": 18620
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.9872912168502808,
+      "learning_rate": 8.167486830897403e-06,
+      "loss": 0.4658,
+      "step": 18621
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.022448778152466,
+      "learning_rate": 8.167282822337346e-06,
+      "loss": 0.57,
+      "step": 18622
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 3.251873731613159,
+      "learning_rate": 8.167078804970263e-06,
+      "loss": 0.6583,
+      "step": 18623
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.938518762588501,
+      "learning_rate": 8.166874778796719e-06,
+      "loss": 0.7212,
+      "step": 18624
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.9988999366760254,
+      "learning_rate": 8.166670743817284e-06,
+      "loss": 0.8124,
+      "step": 18625
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.856158971786499,
+      "learning_rate": 8.166466700032523e-06,
+      "loss": 0.593,
+      "step": 18626
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.688180685043335,
+      "learning_rate": 8.166262647443004e-06,
+      "loss": 0.5449,
+      "step": 18627
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.171285390853882,
+      "learning_rate": 8.166058586049294e-06,
+      "loss": 0.4995,
+      "step": 18628
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 5.253480911254883,
+      "learning_rate": 8.165854515851964e-06,
+      "loss": 0.9994,
+      "step": 18629
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.6154669523239136,
+      "learning_rate": 8.165650436851576e-06,
+      "loss": 0.6904,
+      "step": 18630
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.984834909439087,
+      "learning_rate": 8.1654463490487e-06,
+      "loss": 0.93,
+      "step": 18631
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.173781394958496,
+      "learning_rate": 8.165242252443905e-06,
+      "loss": 0.8519,
+      "step": 18632
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.142831802368164,
+      "learning_rate": 8.165038147037757e-06,
+      "loss": 0.5194,
+      "step": 18633
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.2673850059509277,
+      "learning_rate": 8.164834032830822e-06,
+      "loss": 0.7329,
+      "step": 18634
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.5664854049682617,
+      "learning_rate": 8.16462990982367e-06,
+      "loss": 0.8485,
+      "step": 18635
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.129577398300171,
+      "learning_rate": 8.164425778016866e-06,
+      "loss": 0.7569,
+      "step": 18636
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.0166828632354736,
+      "learning_rate": 8.164221637410983e-06,
+      "loss": 0.6816,
+      "step": 18637
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.906726837158203,
+      "learning_rate": 8.164017488006582e-06,
+      "loss": 0.5464,
+      "step": 18638
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.8107411861419678,
+      "learning_rate": 8.163813329804235e-06,
+      "loss": 0.6296,
+      "step": 18639
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.8341915607452393,
+      "learning_rate": 8.163609162804509e-06,
+      "loss": 0.662,
+      "step": 18640
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 3.6766698360443115,
+      "learning_rate": 8.163404987007969e-06,
+      "loss": 0.7124,
+      "step": 18641
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.3606812953948975,
+      "learning_rate": 8.163200802415185e-06,
+      "loss": 0.5176,
+      "step": 18642
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 3.5018630027770996,
+      "learning_rate": 8.162996609026725e-06,
+      "loss": 0.4718,
+      "step": 18643
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.9747873544692993,
+      "learning_rate": 8.162792406843158e-06,
+      "loss": 0.83,
+      "step": 18644
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.005774974822998,
+      "learning_rate": 8.162588195865049e-06,
+      "loss": 0.6259,
+      "step": 18645
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.789478302001953,
+      "learning_rate": 8.162383976092966e-06,
+      "loss": 0.523,
+      "step": 18646
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.170379877090454,
+      "learning_rate": 8.16217974752748e-06,
+      "loss": 0.3472,
+      "step": 18647
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.140213966369629,
+      "learning_rate": 8.161975510169153e-06,
+      "loss": 0.5172,
+      "step": 18648
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.6429388523101807,
+      "learning_rate": 8.161771264018558e-06,
+      "loss": 0.4701,
+      "step": 18649
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 3.3929269313812256,
+      "learning_rate": 8.161567009076263e-06,
+      "loss": 0.707,
+      "step": 18650
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.4102081060409546,
+      "learning_rate": 8.161362745342834e-06,
+      "loss": 0.7344,
+      "step": 18651
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.157820224761963,
+      "learning_rate": 8.16115847281884e-06,
+      "loss": 0.8382,
+      "step": 18652
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.05946683883667,
+      "learning_rate": 8.160954191504848e-06,
+      "loss": 0.5269,
+      "step": 18653
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.7109971046447754,
+      "learning_rate": 8.160749901401426e-06,
+      "loss": 0.4706,
+      "step": 18654
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.5102286338806152,
+      "learning_rate": 8.160545602509142e-06,
+      "loss": 0.4439,
+      "step": 18655
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.9339182376861572,
+      "learning_rate": 8.160341294828566e-06,
+      "loss": 0.8922,
+      "step": 18656
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.474480628967285,
+      "learning_rate": 8.160136978360264e-06,
+      "loss": 0.6988,
+      "step": 18657
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.9427495002746582,
+      "learning_rate": 8.159932653104805e-06,
+      "loss": 0.5948,
+      "step": 18658
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 3.1427078247070312,
+      "learning_rate": 8.159728319062758e-06,
+      "loss": 0.6753,
+      "step": 18659
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.3002851009368896,
+      "learning_rate": 8.15952397623469e-06,
+      "loss": 0.5986,
+      "step": 18660
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.8389577865600586,
+      "learning_rate": 8.159319624621167e-06,
+      "loss": 0.7723,
+      "step": 18661
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.5316622257232666,
+      "learning_rate": 8.159115264222764e-06,
+      "loss": 0.7059,
+      "step": 18662
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.5537049770355225,
+      "learning_rate": 8.158910895040042e-06,
+      "loss": 0.3789,
+      "step": 18663
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 3.463625907897949,
+      "learning_rate": 8.158706517073572e-06,
+      "loss": 0.7594,
+      "step": 18664
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.3950326442718506,
+      "learning_rate": 8.158502130323922e-06,
+      "loss": 0.6969,
+      "step": 18665
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.478259563446045,
+      "learning_rate": 8.15829773479166e-06,
+      "loss": 0.6029,
+      "step": 18666
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.6909658908843994,
+      "learning_rate": 8.15809333047736e-06,
+      "loss": 0.4854,
+      "step": 18667
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 5.508438587188721,
+      "learning_rate": 8.15788891738158e-06,
+      "loss": 0.6161,
+      "step": 18668
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 3.1194264888763428,
+      "learning_rate": 8.157684495504897e-06,
+      "loss": 0.6992,
+      "step": 18669
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.3863449096679688,
+      "learning_rate": 8.157480064847877e-06,
+      "loss": 0.5487,
+      "step": 18670
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.6423145532608032,
+      "learning_rate": 8.157275625411085e-06,
+      "loss": 0.449,
+      "step": 18671
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.9707013368606567,
+      "learning_rate": 8.157071177195094e-06,
+      "loss": 0.5508,
+      "step": 18672
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 3.3043315410614014,
+      "learning_rate": 8.15686672020047e-06,
+      "loss": 0.6118,
+      "step": 18673
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.3487257957458496,
+      "learning_rate": 8.156662254427783e-06,
+      "loss": 0.9202,
+      "step": 18674
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.2826812267303467,
+      "learning_rate": 8.1564577798776e-06,
+      "loss": 0.2424,
+      "step": 18675
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.6319835186004639,
+      "learning_rate": 8.15625329655049e-06,
+      "loss": 0.4762,
+      "step": 18676
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 3.277235269546509,
+      "learning_rate": 8.156048804447023e-06,
+      "loss": 0.8654,
+      "step": 18677
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 3.031482458114624,
+      "learning_rate": 8.155844303567766e-06,
+      "loss": 0.7651,
+      "step": 18678
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.864699125289917,
+      "learning_rate": 8.155639793913289e-06,
+      "loss": 0.6162,
+      "step": 18679
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 3.1786606311798096,
+      "learning_rate": 8.155435275484158e-06,
+      "loss": 0.7609,
+      "step": 18680
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.9735565185546875,
+      "learning_rate": 8.155230748280945e-06,
+      "loss": 0.7657,
+      "step": 18681
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 3.3093621730804443,
+      "learning_rate": 8.155026212304217e-06,
+      "loss": 0.8055,
+      "step": 18682
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.722111701965332,
+      "learning_rate": 8.154821667554542e-06,
+      "loss": 0.6452,
+      "step": 18683
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.655597448348999,
+      "learning_rate": 8.154617114032493e-06,
+      "loss": 0.4848,
+      "step": 18684
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 3.103485584259033,
+      "learning_rate": 8.154412551738631e-06,
+      "loss": 0.7416,
+      "step": 18685
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 3.42792010307312,
+      "learning_rate": 8.154207980673531e-06,
+      "loss": 0.7179,
+      "step": 18686
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.365831136703491,
+      "learning_rate": 8.154003400837763e-06,
+      "loss": 0.3658,
+      "step": 18687
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 5.296168804168701,
+      "learning_rate": 8.15379881223189e-06,
+      "loss": 0.7187,
+      "step": 18688
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.0428366661071777,
+      "learning_rate": 8.153594214856485e-06,
+      "loss": 0.7073,
+      "step": 18689
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.771835207939148,
+      "learning_rate": 8.153389608712117e-06,
+      "loss": 0.5358,
+      "step": 18690
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.5682919025421143,
+      "learning_rate": 8.153184993799352e-06,
+      "loss": 0.5788,
+      "step": 18691
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.696936845779419,
+      "learning_rate": 8.152980370118761e-06,
+      "loss": 0.6701,
+      "step": 18692
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 3.171931028366089,
+      "learning_rate": 8.152775737670913e-06,
+      "loss": 0.7287,
+      "step": 18693
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.059934616088867,
+      "learning_rate": 8.152571096456376e-06,
+      "loss": 0.8414,
+      "step": 18694
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.1832168102264404,
+      "learning_rate": 8.15236644647572e-06,
+      "loss": 0.5934,
+      "step": 18695
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.9102550745010376,
+      "learning_rate": 8.152161787729516e-06,
+      "loss": 0.7429,
+      "step": 18696
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.327609658241272,
+      "learning_rate": 8.151957120218327e-06,
+      "loss": 0.57,
+      "step": 18697
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.6072096824645996,
+      "learning_rate": 8.15175244394273e-06,
+      "loss": 0.7059,
+      "step": 18698
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.555320978164673,
+      "learning_rate": 8.151547758903289e-06,
+      "loss": 0.7007,
+      "step": 18699
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.3909481763839722,
+      "learning_rate": 8.151343065100572e-06,
+      "loss": 0.6648,
+      "step": 18700
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.8759496212005615,
+      "learning_rate": 8.151138362535154e-06,
+      "loss": 0.7442,
+      "step": 18701
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.514894962310791,
+      "learning_rate": 8.150933651207597e-06,
+      "loss": 0.6415,
+      "step": 18702
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.1400306224823,
+      "learning_rate": 8.150728931118477e-06,
+      "loss": 0.8148,
+      "step": 18703
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.321605920791626,
+      "learning_rate": 8.15052420226836e-06,
+      "loss": 0.6765,
+      "step": 18704
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.874581813812256,
+      "learning_rate": 8.150319464657812e-06,
+      "loss": 0.7756,
+      "step": 18705
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.109668731689453,
+      "learning_rate": 8.150114718287407e-06,
+      "loss": 0.5433,
+      "step": 18706
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 3.037264108657837,
+      "learning_rate": 8.149909963157714e-06,
+      "loss": 0.4694,
+      "step": 18707
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.9032479524612427,
+      "learning_rate": 8.149705199269301e-06,
+      "loss": 0.524,
+      "step": 18708
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.5104130506515503,
+      "learning_rate": 8.149500426622739e-06,
+      "loss": 0.6756,
+      "step": 18709
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.7448177337646484,
+      "learning_rate": 8.149295645218596e-06,
+      "loss": 0.599,
+      "step": 18710
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.4383013248443604,
+      "learning_rate": 8.14909085505744e-06,
+      "loss": 0.6202,
+      "step": 18711
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.9488821029663086,
+      "learning_rate": 8.148886056139842e-06,
+      "loss": 0.8289,
+      "step": 18712
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 4.226846218109131,
+      "learning_rate": 8.148681248466373e-06,
+      "loss": 0.7357,
+      "step": 18713
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.593698024749756,
+      "learning_rate": 8.148476432037599e-06,
+      "loss": 0.5535,
+      "step": 18714
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.5799670219421387,
+      "learning_rate": 8.148271606854092e-06,
+      "loss": 0.6244,
+      "step": 18715
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.3566203117370605,
+      "learning_rate": 8.148066772916424e-06,
+      "loss": 0.731,
+      "step": 18716
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 3.084383487701416,
+      "learning_rate": 8.147861930225158e-06,
+      "loss": 0.8006,
+      "step": 18717
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.0013010501861572,
+      "learning_rate": 8.147657078780868e-06,
+      "loss": 0.4921,
+      "step": 18718
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.1599655151367188,
+      "learning_rate": 8.147452218584125e-06,
+      "loss": 0.5848,
+      "step": 18719
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.706080436706543,
+      "learning_rate": 8.147247349635493e-06,
+      "loss": 0.6711,
+      "step": 18720
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.0820915699005127,
+      "learning_rate": 8.147042471935547e-06,
+      "loss": 0.4826,
+      "step": 18721
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.3743013143539429,
+      "learning_rate": 8.146837585484854e-06,
+      "loss": 0.5676,
+      "step": 18722
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.189265251159668,
+      "learning_rate": 8.146632690283986e-06,
+      "loss": 0.8478,
+      "step": 18723
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.438491106033325,
+      "learning_rate": 8.14642778633351e-06,
+      "loss": 0.6219,
+      "step": 18724
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.7311049699783325,
+      "learning_rate": 8.146222873633997e-06,
+      "loss": 0.526,
+      "step": 18725
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 3.0425920486450195,
+      "learning_rate": 8.146017952186017e-06,
+      "loss": 0.8521,
+      "step": 18726
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 4.34625768661499,
+      "learning_rate": 8.14581302199014e-06,
+      "loss": 0.5983,
+      "step": 18727
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.7417383193969727,
+      "learning_rate": 8.145608083046936e-06,
+      "loss": 0.8182,
+      "step": 18728
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 3.2034196853637695,
+      "learning_rate": 8.145403135356973e-06,
+      "loss": 0.8085,
+      "step": 18729
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.9574528932571411,
+      "learning_rate": 8.145198178920823e-06,
+      "loss": 0.9159,
+      "step": 18730
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 5.657988548278809,
+      "learning_rate": 8.144993213739054e-06,
+      "loss": 0.6749,
+      "step": 18731
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.5555205345153809,
+      "learning_rate": 8.144788239812238e-06,
+      "loss": 0.5824,
+      "step": 18732
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 3.0846986770629883,
+      "learning_rate": 8.144583257140944e-06,
+      "loss": 0.4805,
+      "step": 18733
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.7630419731140137,
+      "learning_rate": 8.144378265725743e-06,
+      "loss": 0.5634,
+      "step": 18734
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.6863598823547363,
+      "learning_rate": 8.144173265567201e-06,
+      "loss": 0.4228,
+      "step": 18735
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.157107353210449,
+      "learning_rate": 8.143968256665894e-06,
+      "loss": 0.5687,
+      "step": 18736
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 3.8442485332489014,
+      "learning_rate": 8.143763239022386e-06,
+      "loss": 0.6048,
+      "step": 18737
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.1913321018218994,
+      "learning_rate": 8.143558212637253e-06,
+      "loss": 0.7168,
+      "step": 18738
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.2963693141937256,
+      "learning_rate": 8.14335317751106e-06,
+      "loss": 0.5375,
+      "step": 18739
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.777655601501465,
+      "learning_rate": 8.14314813364438e-06,
+      "loss": 0.7392,
+      "step": 18740
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.5887577533721924,
+      "learning_rate": 8.142943081037782e-06,
+      "loss": 0.5547,
+      "step": 18741
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 3.1480493545532227,
+      "learning_rate": 8.142738019691839e-06,
+      "loss": 0.7438,
+      "step": 18742
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 4.07824182510376,
+      "learning_rate": 8.142532949607117e-06,
+      "loss": 0.5236,
+      "step": 18743
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.4611663818359375,
+      "learning_rate": 8.142327870784189e-06,
+      "loss": 0.5623,
+      "step": 18744
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 4.1259074211120605,
+      "learning_rate": 8.142122783223622e-06,
+      "loss": 0.7563,
+      "step": 18745
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 7.826419353485107,
+      "learning_rate": 8.14191768692599e-06,
+      "loss": 0.8665,
+      "step": 18746
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.8872663974761963,
+      "learning_rate": 8.141712581891863e-06,
+      "loss": 0.748,
+      "step": 18747
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.2373850345611572,
+      "learning_rate": 8.14150746812181e-06,
+      "loss": 0.7006,
+      "step": 18748
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.0132267475128174,
+      "learning_rate": 8.1413023456164e-06,
+      "loss": 0.6627,
+      "step": 18749
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.9597715139389038,
+      "learning_rate": 8.141097214376208e-06,
+      "loss": 0.7803,
+      "step": 18750
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.7922714948654175,
+      "learning_rate": 8.140892074401799e-06,
+      "loss": 0.7928,
+      "step": 18751
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 3.567265272140503,
+      "learning_rate": 8.140686925693746e-06,
+      "loss": 0.8553,
+      "step": 18752
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 3.0181312561035156,
+      "learning_rate": 8.14048176825262e-06,
+      "loss": 0.7801,
+      "step": 18753
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.4496960639953613,
+      "learning_rate": 8.14027660207899e-06,
+      "loss": 0.7331,
+      "step": 18754
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 3.13832688331604,
+      "learning_rate": 8.140071427173426e-06,
+      "loss": 0.8948,
+      "step": 18755
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 3.410756826400757,
+      "learning_rate": 8.1398662435365e-06,
+      "loss": 0.5688,
+      "step": 18756
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.837763786315918,
+      "learning_rate": 8.139661051168784e-06,
+      "loss": 0.711,
+      "step": 18757
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.7273786067962646,
+      "learning_rate": 8.139455850070845e-06,
+      "loss": 0.6964,
+      "step": 18758
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 3.2478671073913574,
+      "learning_rate": 8.139250640243257e-06,
+      "loss": 0.7288,
+      "step": 18759
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.6251909732818604,
+      "learning_rate": 8.139045421686587e-06,
+      "loss": 0.5426,
+      "step": 18760
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.18825364112854,
+      "learning_rate": 8.138840194401409e-06,
+      "loss": 0.6372,
+      "step": 18761
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.7580565214157104,
+      "learning_rate": 8.138634958388291e-06,
+      "loss": 0.6969,
+      "step": 18762
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.3009531497955322,
+      "learning_rate": 8.138429713647807e-06,
+      "loss": 0.5027,
+      "step": 18763
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 3.125419855117798,
+      "learning_rate": 8.138224460180524e-06,
+      "loss": 0.9073,
+      "step": 18764
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.084949016571045,
+      "learning_rate": 8.138019197987015e-06,
+      "loss": 0.4588,
+      "step": 18765
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.9208056926727295,
+      "learning_rate": 8.13781392706785e-06,
+      "loss": 0.4958,
+      "step": 18766
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.8871872425079346,
+      "learning_rate": 8.137608647423599e-06,
+      "loss": 0.7973,
+      "step": 18767
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.5431363582611084,
+      "learning_rate": 8.137403359054835e-06,
+      "loss": 0.705,
+      "step": 18768
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.7944433689117432,
+      "learning_rate": 8.137198061962126e-06,
+      "loss": 0.5676,
+      "step": 18769
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.882875919342041,
+      "learning_rate": 8.136992756146048e-06,
+      "loss": 0.9729,
+      "step": 18770
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 5.899862289428711,
+      "learning_rate": 8.136787441607164e-06,
+      "loss": 0.6925,
+      "step": 18771
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.109570026397705,
+      "learning_rate": 8.136582118346051e-06,
+      "loss": 0.7196,
+      "step": 18772
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.292112112045288,
+      "learning_rate": 8.136376786363279e-06,
+      "loss": 0.7863,
+      "step": 18773
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.471127986907959,
+      "learning_rate": 8.136171445659417e-06,
+      "loss": 0.6998,
+      "step": 18774
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.6859962940216064,
+      "learning_rate": 8.135966096235037e-06,
+      "loss": 0.6188,
+      "step": 18775
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 3.1190850734710693,
+      "learning_rate": 8.13576073809071e-06,
+      "loss": 0.8382,
+      "step": 18776
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 3.633256196975708,
+      "learning_rate": 8.135555371227008e-06,
+      "loss": 0.6134,
+      "step": 18777
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.898348093032837,
+      "learning_rate": 8.135349995644499e-06,
+      "loss": 0.6583,
+      "step": 18778
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.112720012664795,
+      "learning_rate": 8.135144611343758e-06,
+      "loss": 0.3139,
+      "step": 18779
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.512620210647583,
+      "learning_rate": 8.134939218325355e-06,
+      "loss": 0.3635,
+      "step": 18780
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 3.8382980823516846,
+      "learning_rate": 8.134733816589859e-06,
+      "loss": 0.827,
+      "step": 18781
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.7752156257629395,
+      "learning_rate": 8.134528406137842e-06,
+      "loss": 0.6737,
+      "step": 18782
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.069798231124878,
+      "learning_rate": 8.134322986969878e-06,
+      "loss": 0.6566,
+      "step": 18783
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.8627005815505981,
+      "learning_rate": 8.134117559086533e-06,
+      "loss": 0.6047,
+      "step": 18784
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.9563820362091064,
+      "learning_rate": 8.133912122488384e-06,
+      "loss": 0.7501,
+      "step": 18785
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.669468641281128,
+      "learning_rate": 8.133706677175997e-06,
+      "loss": 0.4367,
+      "step": 18786
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.234854221343994,
+      "learning_rate": 8.133501223149946e-06,
+      "loss": 0.5196,
+      "step": 18787
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.8407548666000366,
+      "learning_rate": 8.133295760410805e-06,
+      "loss": 0.8661,
+      "step": 18788
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 4.144472599029541,
+      "learning_rate": 8.133090288959138e-06,
+      "loss": 0.5432,
+      "step": 18789
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 3.1781134605407715,
+      "learning_rate": 8.132884808795524e-06,
+      "loss": 0.835,
+      "step": 18790
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.9338126182556152,
+      "learning_rate": 8.13267931992053e-06,
+      "loss": 0.7131,
+      "step": 18791
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 4.957011699676514,
+      "learning_rate": 8.132473822334729e-06,
+      "loss": 0.7769,
+      "step": 18792
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.126889944076538,
+      "learning_rate": 8.132268316038688e-06,
+      "loss": 0.7363,
+      "step": 18793
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.9152024984359741,
+      "learning_rate": 8.132062801032986e-06,
+      "loss": 0.6938,
+      "step": 18794
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.812319278717041,
+      "learning_rate": 8.131857277318189e-06,
+      "loss": 0.5238,
+      "step": 18795
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.4405481815338135,
+      "learning_rate": 8.13165174489487e-06,
+      "loss": 0.7528,
+      "step": 18796
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.9266400337219238,
+      "learning_rate": 8.131446203763602e-06,
+      "loss": 0.6082,
+      "step": 18797
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.0334725379943848,
+      "learning_rate": 8.131240653924953e-06,
+      "loss": 0.8151,
+      "step": 18798
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.3088600635528564,
+      "learning_rate": 8.131035095379498e-06,
+      "loss": 0.7907,
+      "step": 18799
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.9522467851638794,
+      "learning_rate": 8.130829528127808e-06,
+      "loss": 0.7017,
+      "step": 18800
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.1555233001708984,
+      "learning_rate": 8.130623952170454e-06,
+      "loss": 0.5956,
+      "step": 18801
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.879499673843384,
+      "learning_rate": 8.130418367508006e-06,
+      "loss": 0.3674,
+      "step": 18802
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 3.0884907245635986,
+      "learning_rate": 8.130212774141037e-06,
+      "loss": 0.509,
+      "step": 18803
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.920229911804199,
+      "learning_rate": 8.130007172070121e-06,
+      "loss": 0.6625,
+      "step": 18804
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.66447377204895,
+      "learning_rate": 8.129801561295828e-06,
+      "loss": 0.5207,
+      "step": 18805
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.8352644443511963,
+      "learning_rate": 8.129595941818726e-06,
+      "loss": 0.4019,
+      "step": 18806
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 3.239293098449707,
+      "learning_rate": 8.129390313639392e-06,
+      "loss": 0.7896,
+      "step": 18807
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 3.426734447479248,
+      "learning_rate": 8.129184676758396e-06,
+      "loss": 0.6468,
+      "step": 18808
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.0897703170776367,
+      "learning_rate": 8.12897903117631e-06,
+      "loss": 0.5157,
+      "step": 18809
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.8050973415374756,
+      "learning_rate": 8.128773376893704e-06,
+      "loss": 0.5162,
+      "step": 18810
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.505394458770752,
+      "learning_rate": 8.128567713911152e-06,
+      "loss": 0.5554,
+      "step": 18811
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 3.4182095527648926,
+      "learning_rate": 8.128362042229225e-06,
+      "loss": 0.4137,
+      "step": 18812
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.458127975463867,
+      "learning_rate": 8.128156361848497e-06,
+      "loss": 0.4717,
+      "step": 18813
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 3.1189303398132324,
+      "learning_rate": 8.127950672769536e-06,
+      "loss": 0.7414,
+      "step": 18814
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.8773839473724365,
+      "learning_rate": 8.127744974992916e-06,
+      "loss": 0.8011,
+      "step": 18815
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.4928078651428223,
+      "learning_rate": 8.127539268519209e-06,
+      "loss": 0.5562,
+      "step": 18816
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.1389191150665283,
+      "learning_rate": 8.127333553348985e-06,
+      "loss": 0.5841,
+      "step": 18817
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 3.7045648097991943,
+      "learning_rate": 8.127127829482822e-06,
+      "loss": 0.5519,
+      "step": 18818
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.7618275880813599,
+      "learning_rate": 8.126922096921286e-06,
+      "loss": 0.5785,
+      "step": 18819
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.3851587772369385,
+      "learning_rate": 8.12671635566495e-06,
+      "loss": 0.7934,
+      "step": 18820
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.786844253540039,
+      "learning_rate": 8.126510605714389e-06,
+      "loss": 0.9951,
+      "step": 18821
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.8277180194854736,
+      "learning_rate": 8.126304847070172e-06,
+      "loss": 0.6832,
+      "step": 18822
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.0794219970703125,
+      "learning_rate": 8.126099079732872e-06,
+      "loss": 0.6582,
+      "step": 18823
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 4.061596393585205,
+      "learning_rate": 8.125893303703062e-06,
+      "loss": 0.592,
+      "step": 18824
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.0958077907562256,
+      "learning_rate": 8.125687518981315e-06,
+      "loss": 0.9584,
+      "step": 18825
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.789188265800476,
+      "learning_rate": 8.1254817255682e-06,
+      "loss": 0.4386,
+      "step": 18826
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.762359142303467,
+      "learning_rate": 8.125275923464293e-06,
+      "loss": 0.7935,
+      "step": 18827
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.9118661880493164,
+      "learning_rate": 8.125070112670162e-06,
+      "loss": 0.6624,
+      "step": 18828
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 3.5130889415740967,
+      "learning_rate": 8.124864293186383e-06,
+      "loss": 0.7302,
+      "step": 18829
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.684070110321045,
+      "learning_rate": 8.124658465013528e-06,
+      "loss": 0.4634,
+      "step": 18830
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.439152479171753,
+      "learning_rate": 8.124452628152168e-06,
+      "loss": 0.5634,
+      "step": 18831
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 3.2400050163269043,
+      "learning_rate": 8.124246782602875e-06,
+      "loss": 0.576,
+      "step": 18832
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.6734809875488281,
+      "learning_rate": 8.124040928366221e-06,
+      "loss": 0.6291,
+      "step": 18833
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.6998629570007324,
+      "learning_rate": 8.123835065442781e-06,
+      "loss": 0.6389,
+      "step": 18834
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 3.535979986190796,
+      "learning_rate": 8.123629193833126e-06,
+      "loss": 0.6291,
+      "step": 18835
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.170205593109131,
+      "learning_rate": 8.123423313537828e-06,
+      "loss": 0.9695,
+      "step": 18836
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 3.55322527885437,
+      "learning_rate": 8.123217424557459e-06,
+      "loss": 0.6427,
+      "step": 18837
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.6043128967285156,
+      "learning_rate": 8.123011526892593e-06,
+      "loss": 0.3886,
+      "step": 18838
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.9131267070770264,
+      "learning_rate": 8.122805620543803e-06,
+      "loss": 0.7468,
+      "step": 18839
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 3.54512882232666,
+      "learning_rate": 8.122599705511658e-06,
+      "loss": 0.6712,
+      "step": 18840
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 3.266235828399658,
+      "learning_rate": 8.122393781796734e-06,
+      "loss": 0.6973,
+      "step": 18841
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.733840823173523,
+      "learning_rate": 8.122187849399602e-06,
+      "loss": 0.6244,
+      "step": 18842
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.9357213973999023,
+      "learning_rate": 8.121981908320837e-06,
+      "loss": 0.7829,
+      "step": 18843
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.8618688583374023,
+      "learning_rate": 8.121775958561009e-06,
+      "loss": 0.5194,
+      "step": 18844
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.855398416519165,
+      "learning_rate": 8.121570000120692e-06,
+      "loss": 0.5236,
+      "step": 18845
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.7335999011993408,
+      "learning_rate": 8.121364033000458e-06,
+      "loss": 0.4733,
+      "step": 18846
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.2258877754211426,
+      "learning_rate": 8.121158057200878e-06,
+      "loss": 0.9359,
+      "step": 18847
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.1591789722442627,
+      "learning_rate": 8.120952072722528e-06,
+      "loss": 0.506,
+      "step": 18848
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.7639377117156982,
+      "learning_rate": 8.12074607956598e-06,
+      "loss": 0.5164,
+      "step": 18849
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.5789754390716553,
+      "learning_rate": 8.120540077731806e-06,
+      "loss": 0.7287,
+      "step": 18850
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.6194205284118652,
+      "learning_rate": 8.12033406722058e-06,
+      "loss": 0.6746,
+      "step": 18851
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 3.1942949295043945,
+      "learning_rate": 8.120128048032872e-06,
+      "loss": 0.7896,
+      "step": 18852
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 3.150028944015503,
+      "learning_rate": 8.119922020169257e-06,
+      "loss": 0.7029,
+      "step": 18853
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 5.294883728027344,
+      "learning_rate": 8.119715983630307e-06,
+      "loss": 0.9251,
+      "step": 18854
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 3.386524200439453,
+      "learning_rate": 8.119509938416598e-06,
+      "loss": 0.5112,
+      "step": 18855
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.4247665405273438,
+      "learning_rate": 8.119303884528699e-06,
+      "loss": 0.5969,
+      "step": 18856
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.7402758598327637,
+      "learning_rate": 8.119097821967184e-06,
+      "loss": 0.9945,
+      "step": 18857
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.9641920328140259,
+      "learning_rate": 8.118891750732626e-06,
+      "loss": 0.4598,
+      "step": 18858
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.8135175704956055,
+      "learning_rate": 8.1186856708256e-06,
+      "loss": 0.8529,
+      "step": 18859
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.9381214380264282,
+      "learning_rate": 8.118479582246676e-06,
+      "loss": 0.6932,
+      "step": 18860
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.5562729835510254,
+      "learning_rate": 8.11827348499643e-06,
+      "loss": 0.672,
+      "step": 18861
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.8590749502182007,
+      "learning_rate": 8.118067379075432e-06,
+      "loss": 0.5473,
+      "step": 18862
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.8416097164154053,
+      "learning_rate": 8.117861264484256e-06,
+      "loss": 0.69,
+      "step": 18863
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.6748898029327393,
+      "learning_rate": 8.11765514122348e-06,
+      "loss": 0.6202,
+      "step": 18864
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.3006503582000732,
+      "learning_rate": 8.117449009293668e-06,
+      "loss": 0.6232,
+      "step": 18865
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.227027654647827,
+      "learning_rate": 8.1172428686954e-06,
+      "loss": 0.7087,
+      "step": 18866
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 3.0798492431640625,
+      "learning_rate": 8.117036719429247e-06,
+      "loss": 0.5133,
+      "step": 18867
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.541938543319702,
+      "learning_rate": 8.116830561495785e-06,
+      "loss": 0.5697,
+      "step": 18868
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.8434416055679321,
+      "learning_rate": 8.116624394895583e-06,
+      "loss": 0.7026,
+      "step": 18869
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.437849283218384,
+      "learning_rate": 8.116418219629216e-06,
+      "loss": 0.6055,
+      "step": 18870
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.890751838684082,
+      "learning_rate": 8.116212035697256e-06,
+      "loss": 0.8323,
+      "step": 18871
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.923515558242798,
+      "learning_rate": 8.116005843100279e-06,
+      "loss": 0.7478,
+      "step": 18872
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.717599868774414,
+      "learning_rate": 8.115799641838857e-06,
+      "loss": 0.6549,
+      "step": 18873
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.5065507888793945,
+      "learning_rate": 8.115593431913565e-06,
+      "loss": 0.489,
+      "step": 18874
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 3.30808687210083,
+      "learning_rate": 8.115387213324973e-06,
+      "loss": 0.6187,
+      "step": 18875
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.6834938526153564,
+      "learning_rate": 8.115180986073656e-06,
+      "loss": 0.4987,
+      "step": 18876
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 3.2018911838531494,
+      "learning_rate": 8.114974750160188e-06,
+      "loss": 0.617,
+      "step": 18877
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.647115468978882,
+      "learning_rate": 8.114768505585142e-06,
+      "loss": 0.5564,
+      "step": 18878
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.49495792388916,
+      "learning_rate": 8.114562252349092e-06,
+      "loss": 0.4616,
+      "step": 18879
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 3.6270360946655273,
+      "learning_rate": 8.114355990452612e-06,
+      "loss": 0.7846,
+      "step": 18880
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.323835849761963,
+      "learning_rate": 8.114149719896271e-06,
+      "loss": 0.7347,
+      "step": 18881
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.113085985183716,
+      "learning_rate": 8.11394344068065e-06,
+      "loss": 0.6537,
+      "step": 18882
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.9062869548797607,
+      "learning_rate": 8.113737152806319e-06,
+      "loss": 0.502,
+      "step": 18883
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.0442495346069336,
+      "learning_rate": 8.11353085627385e-06,
+      "loss": 0.829,
+      "step": 18884
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.5436255931854248,
+      "learning_rate": 8.113324551083816e-06,
+      "loss": 0.3814,
+      "step": 18885
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.986112117767334,
+      "learning_rate": 8.113118237236796e-06,
+      "loss": 0.725,
+      "step": 18886
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.2509565353393555,
+      "learning_rate": 8.112911914733358e-06,
+      "loss": 0.9918,
+      "step": 18887
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.538266181945801,
+      "learning_rate": 8.112705583574079e-06,
+      "loss": 0.4196,
+      "step": 18888
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.3240621089935303,
+      "learning_rate": 8.112499243759532e-06,
+      "loss": 0.6541,
+      "step": 18889
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.9225919246673584,
+      "learning_rate": 8.112292895290293e-06,
+      "loss": 0.8073,
+      "step": 18890
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.0427353382110596,
+      "learning_rate": 8.11208653816693e-06,
+      "loss": 0.7893,
+      "step": 18891
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.5974667072296143,
+      "learning_rate": 8.11188017239002e-06,
+      "loss": 0.7617,
+      "step": 18892
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.9465620517730713,
+      "learning_rate": 8.11167379796014e-06,
+      "loss": 0.7967,
+      "step": 18893
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.5618085861206055,
+      "learning_rate": 8.111467414877859e-06,
+      "loss": 0.6611,
+      "step": 18894
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.9930460453033447,
+      "learning_rate": 8.111261023143752e-06,
+      "loss": 0.4147,
+      "step": 18895
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.1914896965026855,
+      "learning_rate": 8.111054622758394e-06,
+      "loss": 0.7355,
+      "step": 18896
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.4778802394866943,
+      "learning_rate": 8.110848213722359e-06,
+      "loss": 0.8436,
+      "step": 18897
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.5394905805587769,
+      "learning_rate": 8.110641796036221e-06,
+      "loss": 0.6795,
+      "step": 18898
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 3.606955051422119,
+      "learning_rate": 8.11043536970055e-06,
+      "loss": 0.2482,
+      "step": 18899
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.3927781581878662,
+      "learning_rate": 8.110228934715928e-06,
+      "loss": 0.4812,
+      "step": 18900
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.7356109619140625,
+      "learning_rate": 8.110022491082923e-06,
+      "loss": 0.6224,
+      "step": 18901
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.9687496423721313,
+      "learning_rate": 8.10981603880211e-06,
+      "loss": 0.911,
+      "step": 18902
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.574871301651001,
+      "learning_rate": 8.109609577874063e-06,
+      "loss": 0.5942,
+      "step": 18903
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.814791202545166,
+      "learning_rate": 8.109403108299358e-06,
+      "loss": 0.8405,
+      "step": 18904
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 3.0843803882598877,
+      "learning_rate": 8.109196630078566e-06,
+      "loss": 0.3888,
+      "step": 18905
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.354325532913208,
+      "learning_rate": 8.108990143212266e-06,
+      "loss": 0.86,
+      "step": 18906
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.9931530952453613,
+      "learning_rate": 8.108783647701025e-06,
+      "loss": 0.7548,
+      "step": 18907
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.9224348068237305,
+      "learning_rate": 8.108577143545423e-06,
+      "loss": 0.5301,
+      "step": 18908
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.879083275794983,
+      "learning_rate": 8.108370630746033e-06,
+      "loss": 0.7616,
+      "step": 18909
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.0296149253845215,
+      "learning_rate": 8.10816410930343e-06,
+      "loss": 0.6602,
+      "step": 18910
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.1040356159210205,
+      "learning_rate": 8.107957579218183e-06,
+      "loss": 0.6179,
+      "step": 18911
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.1660122871398926,
+      "learning_rate": 8.107751040490875e-06,
+      "loss": 0.5313,
+      "step": 18912
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.407327890396118,
+      "learning_rate": 8.107544493122072e-06,
+      "loss": 0.9696,
+      "step": 18913
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.52497935295105,
+      "learning_rate": 8.107337937112352e-06,
+      "loss": 0.5221,
+      "step": 18914
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.9512453079223633,
+      "learning_rate": 8.107131372462292e-06,
+      "loss": 0.5064,
+      "step": 18915
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.5518343448638916,
+      "learning_rate": 8.106924799172463e-06,
+      "loss": 0.6508,
+      "step": 18916
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.3458139896392822,
+      "learning_rate": 8.106718217243437e-06,
+      "loss": 1.0272,
+      "step": 18917
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.5388224124908447,
+      "learning_rate": 8.106511626675794e-06,
+      "loss": 0.7439,
+      "step": 18918
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.3971898555755615,
+      "learning_rate": 8.106305027470104e-06,
+      "loss": 0.5402,
+      "step": 18919
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.7577507495880127,
+      "learning_rate": 8.106098419626944e-06,
+      "loss": 0.5244,
+      "step": 18920
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.1408541202545166,
+      "learning_rate": 8.10589180314689e-06,
+      "loss": 0.8066,
+      "step": 18921
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 4.1522016525268555,
+      "learning_rate": 8.105685178030513e-06,
+      "loss": 0.6624,
+      "step": 18922
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.7832282781600952,
+      "learning_rate": 8.105478544278388e-06,
+      "loss": 0.7632,
+      "step": 18923
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.8859847784042358,
+      "learning_rate": 8.105271901891091e-06,
+      "loss": 0.7787,
+      "step": 18924
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 3.943591594696045,
+      "learning_rate": 8.105065250869198e-06,
+      "loss": 0.8881,
+      "step": 18925
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.0688538551330566,
+      "learning_rate": 8.104858591213279e-06,
+      "loss": 0.609,
+      "step": 18926
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.5087013244628906,
+      "learning_rate": 8.104651922923913e-06,
+      "loss": 0.721,
+      "step": 18927
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.9078642129898071,
+      "learning_rate": 8.104445246001673e-06,
+      "loss": 0.624,
+      "step": 18928
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.935868501663208,
+      "learning_rate": 8.104238560447132e-06,
+      "loss": 0.7314,
+      "step": 18929
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.589109182357788,
+      "learning_rate": 8.104031866260868e-06,
+      "loss": 0.5261,
+      "step": 18930
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.267256259918213,
+      "learning_rate": 8.103825163443454e-06,
+      "loss": 0.7718,
+      "step": 18931
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.890561580657959,
+      "learning_rate": 8.103618451995464e-06,
+      "loss": 0.6423,
+      "step": 18932
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 3.768502712249756,
+      "learning_rate": 8.103411731917474e-06,
+      "loss": 0.6976,
+      "step": 18933
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.5417237281799316,
+      "learning_rate": 8.10320500321006e-06,
+      "loss": 0.507,
+      "step": 18934
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 4.975924491882324,
+      "learning_rate": 8.102998265873794e-06,
+      "loss": 0.6634,
+      "step": 18935
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.5117197036743164,
+      "learning_rate": 8.102791519909253e-06,
+      "loss": 0.7473,
+      "step": 18936
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.9031765460968018,
+      "learning_rate": 8.102584765317011e-06,
+      "loss": 0.834,
+      "step": 18937
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.4043943881988525,
+      "learning_rate": 8.102378002097644e-06,
+      "loss": 0.9687,
+      "step": 18938
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.6953047513961792,
+      "learning_rate": 8.102171230251723e-06,
+      "loss": 0.5048,
+      "step": 18939
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 3.2105729579925537,
+      "learning_rate": 8.101964449779827e-06,
+      "loss": 0.5166,
+      "step": 18940
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.9518463611602783,
+      "learning_rate": 8.101757660682533e-06,
+      "loss": 0.4425,
+      "step": 18941
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.7917765378952026,
+      "learning_rate": 8.10155086296041e-06,
+      "loss": 0.5879,
+      "step": 18942
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.6059074401855469,
+      "learning_rate": 8.101344056614036e-06,
+      "loss": 0.4403,
+      "step": 18943
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.6480183601379395,
+      "learning_rate": 8.101137241643986e-06,
+      "loss": 0.6687,
+      "step": 18944
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.544365406036377,
+      "learning_rate": 8.100930418050835e-06,
+      "loss": 0.5833,
+      "step": 18945
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.4104201793670654,
+      "learning_rate": 8.100723585835159e-06,
+      "loss": 0.5394,
+      "step": 18946
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.4445853233337402,
+      "learning_rate": 8.100516744997532e-06,
+      "loss": 0.6057,
+      "step": 18947
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.999895691871643,
+      "learning_rate": 8.10030989553853e-06,
+      "loss": 0.8459,
+      "step": 18948
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.450666666030884,
+      "learning_rate": 8.100103037458725e-06,
+      "loss": 0.5117,
+      "step": 18949
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 4.1521525382995605,
+      "learning_rate": 8.099896170758697e-06,
+      "loss": 0.734,
+      "step": 18950
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.4074792861938477,
+      "learning_rate": 8.099689295439018e-06,
+      "loss": 0.8005,
+      "step": 18951
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.911147117614746,
+      "learning_rate": 8.099482411500266e-06,
+      "loss": 0.6972,
+      "step": 18952
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.873631477355957,
+      "learning_rate": 8.099275518943013e-06,
+      "loss": 0.9731,
+      "step": 18953
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.181020498275757,
+      "learning_rate": 8.099068617767836e-06,
+      "loss": 0.8672,
+      "step": 18954
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.2816126346588135,
+      "learning_rate": 8.098861707975309e-06,
+      "loss": 0.8323,
+      "step": 18955
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.8597065210342407,
+      "learning_rate": 8.098654789566012e-06,
+      "loss": 0.6831,
+      "step": 18956
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.3480427265167236,
+      "learning_rate": 8.098447862540513e-06,
+      "loss": 0.6791,
+      "step": 18957
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.4479410648345947,
+      "learning_rate": 8.098240926899392e-06,
+      "loss": 0.6401,
+      "step": 18958
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.486510992050171,
+      "learning_rate": 8.098033982643225e-06,
+      "loss": 0.5738,
+      "step": 18959
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.8813273906707764,
+      "learning_rate": 8.097827029772584e-06,
+      "loss": 0.5749,
+      "step": 18960
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.523207187652588,
+      "learning_rate": 8.097620068288049e-06,
+      "loss": 0.8506,
+      "step": 18961
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 4.276193141937256,
+      "learning_rate": 8.097413098190192e-06,
+      "loss": 0.8241,
+      "step": 18962
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.6337014436721802,
+      "learning_rate": 8.09720611947959e-06,
+      "loss": 0.7228,
+      "step": 18963
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.3013017177581787,
+      "learning_rate": 8.096999132156816e-06,
+      "loss": 0.5577,
+      "step": 18964
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 4.413869380950928,
+      "learning_rate": 8.096792136222449e-06,
+      "loss": 0.5165,
+      "step": 18965
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.9185765981674194,
+      "learning_rate": 8.096585131677064e-06,
+      "loss": 0.6063,
+      "step": 18966
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.817190170288086,
+      "learning_rate": 8.096378118521235e-06,
+      "loss": 0.6585,
+      "step": 18967
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.838176965713501,
+      "learning_rate": 8.096171096755538e-06,
+      "loss": 0.7647,
+      "step": 18968
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.938560128211975,
+      "learning_rate": 8.095964066380549e-06,
+      "loss": 0.7827,
+      "step": 18969
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.774372935295105,
+      "learning_rate": 8.095757027396844e-06,
+      "loss": 0.5826,
+      "step": 18970
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.816514253616333,
+      "learning_rate": 8.095549979804997e-06,
+      "loss": 0.8178,
+      "step": 18971
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.0442349910736084,
+      "learning_rate": 8.095342923605586e-06,
+      "loss": 0.816,
+      "step": 18972
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.233165979385376,
+      "learning_rate": 8.095135858799188e-06,
+      "loss": 0.7522,
+      "step": 18973
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.7646428346633911,
+      "learning_rate": 8.094928785386375e-06,
+      "loss": 0.6091,
+      "step": 18974
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.029829502105713,
+      "learning_rate": 8.094721703367724e-06,
+      "loss": 0.6475,
+      "step": 18975
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.7286605834960938,
+      "learning_rate": 8.094514612743812e-06,
+      "loss": 0.6698,
+      "step": 18976
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.2513651847839355,
+      "learning_rate": 8.094307513515214e-06,
+      "loss": 0.6928,
+      "step": 18977
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 5.854820251464844,
+      "learning_rate": 8.094100405682506e-06,
+      "loss": 0.505,
+      "step": 18978
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.181333303451538,
+      "learning_rate": 8.093893289246264e-06,
+      "loss": 0.4685,
+      "step": 18979
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.4693386554718018,
+      "learning_rate": 8.093686164207063e-06,
+      "loss": 0.4859,
+      "step": 18980
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.3227550983428955,
+      "learning_rate": 8.09347903056548e-06,
+      "loss": 0.8162,
+      "step": 18981
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.750171422958374,
+      "learning_rate": 8.09327188832209e-06,
+      "loss": 0.5082,
+      "step": 18982
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.693917989730835,
+      "learning_rate": 8.093064737477469e-06,
+      "loss": 0.623,
+      "step": 18983
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.6946887969970703,
+      "learning_rate": 8.092857578032196e-06,
+      "loss": 0.7559,
+      "step": 18984
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.9638922214508057,
+      "learning_rate": 8.092650409986845e-06,
+      "loss": 0.7688,
+      "step": 18985
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.7670609951019287,
+      "learning_rate": 8.092443233341988e-06,
+      "loss": 0.7028,
+      "step": 18986
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 3.9949848651885986,
+      "learning_rate": 8.092236048098209e-06,
+      "loss": 0.7131,
+      "step": 18987
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.5416955947875977,
+      "learning_rate": 8.092028854256078e-06,
+      "loss": 0.8176,
+      "step": 18988
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.445730447769165,
+      "learning_rate": 8.091821651816173e-06,
+      "loss": 0.659,
+      "step": 18989
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.431215763092041,
+      "learning_rate": 8.09161444077907e-06,
+      "loss": 0.5705,
+      "step": 18990
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.6649930477142334,
+      "learning_rate": 8.091407221145347e-06,
+      "loss": 0.5691,
+      "step": 18991
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 3.1882119178771973,
+      "learning_rate": 8.091199992915577e-06,
+      "loss": 0.5741,
+      "step": 18992
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.794355630874634,
+      "learning_rate": 8.090992756090337e-06,
+      "loss": 0.3544,
+      "step": 18993
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.885802984237671,
+      "learning_rate": 8.090785510670205e-06,
+      "loss": 0.6795,
+      "step": 18994
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.6661789417266846,
+      "learning_rate": 8.090578256655757e-06,
+      "loss": 0.7581,
+      "step": 18995
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.2865822315216064,
+      "learning_rate": 8.09037099404757e-06,
+      "loss": 0.4101,
+      "step": 18996
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.1450717449188232,
+      "learning_rate": 8.090163722846216e-06,
+      "loss": 0.6819,
+      "step": 18997
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 3.530853033065796,
+      "learning_rate": 8.089956443052276e-06,
+      "loss": 0.7695,
+      "step": 18998
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 3.537766456604004,
+      "learning_rate": 8.089749154666324e-06,
+      "loss": 0.5002,
+      "step": 18999
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.4768331050872803,
+      "learning_rate": 8.089541857688937e-06,
+      "loss": 0.7845,
+      "step": 19000
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.0781707763671875,
+      "learning_rate": 8.089334552120692e-06,
+      "loss": 0.8003,
+      "step": 19001
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.61218523979187,
+      "learning_rate": 8.089127237962164e-06,
+      "loss": 0.7407,
+      "step": 19002
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 4.035547256469727,
+      "learning_rate": 8.088919915213931e-06,
+      "loss": 0.6153,
+      "step": 19003
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.7088840007781982,
+      "learning_rate": 8.088712583876567e-06,
+      "loss": 0.6465,
+      "step": 19004
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.9344456195831299,
+      "learning_rate": 8.088505243950653e-06,
+      "loss": 0.778,
+      "step": 19005
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 3.493664026260376,
+      "learning_rate": 8.088297895436764e-06,
+      "loss": 0.7562,
+      "step": 19006
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.114863634109497,
+      "learning_rate": 8.088090538335472e-06,
+      "loss": 0.6602,
+      "step": 19007
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.9567432403564453,
+      "learning_rate": 8.087883172647359e-06,
+      "loss": 0.5707,
+      "step": 19008
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.097954511642456,
+      "learning_rate": 8.087675798372999e-06,
+      "loss": 0.576,
+      "step": 19009
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.443875312805176,
+      "learning_rate": 8.087468415512969e-06,
+      "loss": 0.6432,
+      "step": 19010
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.4440832138061523,
+      "learning_rate": 8.087261024067846e-06,
+      "loss": 0.5484,
+      "step": 19011
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.33716082572937,
+      "learning_rate": 8.087053624038207e-06,
+      "loss": 0.6051,
+      "step": 19012
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.2264232635498047,
+      "learning_rate": 8.086846215424628e-06,
+      "loss": 0.6039,
+      "step": 19013
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 3.20996356010437,
+      "learning_rate": 8.086638798227687e-06,
+      "loss": 0.8702,
+      "step": 19014
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 3.372615337371826,
+      "learning_rate": 8.086431372447959e-06,
+      "loss": 0.5669,
+      "step": 19015
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 3.114586591720581,
+      "learning_rate": 8.086223938086021e-06,
+      "loss": 0.6176,
+      "step": 19016
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.132453203201294,
+      "learning_rate": 8.086016495142453e-06,
+      "loss": 0.5425,
+      "step": 19017
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.46482515335083,
+      "learning_rate": 8.085809043617825e-06,
+      "loss": 0.7069,
+      "step": 19018
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 3.823225498199463,
+      "learning_rate": 8.08560158351272e-06,
+      "loss": 0.7063,
+      "step": 19019
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.8688596487045288,
+      "learning_rate": 8.085394114827714e-06,
+      "loss": 0.5651,
+      "step": 19020
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.88924503326416,
+      "learning_rate": 8.085186637563382e-06,
+      "loss": 0.6686,
+      "step": 19021
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.880164384841919,
+      "learning_rate": 8.084979151720302e-06,
+      "loss": 0.4616,
+      "step": 19022
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 3.63370943069458,
+      "learning_rate": 8.08477165729905e-06,
+      "loss": 0.7721,
+      "step": 19023
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 3.097550868988037,
+      "learning_rate": 8.084564154300203e-06,
+      "loss": 0.6612,
+      "step": 19024
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.9361765384674072,
+      "learning_rate": 8.08435664272434e-06,
+      "loss": 0.7713,
+      "step": 19025
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 3.147272825241089,
+      "learning_rate": 8.084149122572036e-06,
+      "loss": 0.762,
+      "step": 19026
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 3.078650951385498,
+      "learning_rate": 8.083941593843869e-06,
+      "loss": 0.8569,
+      "step": 19027
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.164777994155884,
+      "learning_rate": 8.083734056540415e-06,
+      "loss": 0.556,
+      "step": 19028
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 3.151414155960083,
+      "learning_rate": 8.083526510662251e-06,
+      "loss": 0.2547,
+      "step": 19029
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.0225610733032227,
+      "learning_rate": 8.083318956209955e-06,
+      "loss": 0.554,
+      "step": 19030
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 3.0582518577575684,
+      "learning_rate": 8.083111393184106e-06,
+      "loss": 0.6783,
+      "step": 19031
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.9140543937683105,
+      "learning_rate": 8.082903821585279e-06,
+      "loss": 0.7322,
+      "step": 19032
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 3.524813175201416,
+      "learning_rate": 8.08269624141405e-06,
+      "loss": 0.4768,
+      "step": 19033
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.7984611988067627,
+      "learning_rate": 8.082488652670998e-06,
+      "loss": 0.6823,
+      "step": 19034
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.6612436771392822,
+      "learning_rate": 8.0822810553567e-06,
+      "loss": 0.5878,
+      "step": 19035
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.9118722677230835,
+      "learning_rate": 8.082073449471733e-06,
+      "loss": 0.6277,
+      "step": 19036
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.0874876976013184,
+      "learning_rate": 8.081865835016674e-06,
+      "loss": 0.5215,
+      "step": 19037
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.2520217895507812,
+      "learning_rate": 8.081658211992102e-06,
+      "loss": 0.72,
+      "step": 19038
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.8659396171569824,
+      "learning_rate": 8.08145058039859e-06,
+      "loss": 0.7409,
+      "step": 19039
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.1639764308929443,
+      "learning_rate": 8.081242940236722e-06,
+      "loss": 0.6206,
+      "step": 19040
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.4764115810394287,
+      "learning_rate": 8.081035291507068e-06,
+      "loss": 0.5469,
+      "step": 19041
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 3.1375362873077393,
+      "learning_rate": 8.080827634210211e-06,
+      "loss": 0.7572,
+      "step": 19042
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.820326805114746,
+      "learning_rate": 8.080619968346726e-06,
+      "loss": 0.5709,
+      "step": 19043
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 3.2991204261779785,
+      "learning_rate": 8.080412293917191e-06,
+      "loss": 0.6719,
+      "step": 19044
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.879201650619507,
+      "learning_rate": 8.080204610922185e-06,
+      "loss": 0.6753,
+      "step": 19045
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.1816635131835938,
+      "learning_rate": 8.079996919362281e-06,
+      "loss": 0.8361,
+      "step": 19046
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.910968542098999,
+      "learning_rate": 8.079789219238061e-06,
+      "loss": 0.5646,
+      "step": 19047
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.8311017751693726,
+      "learning_rate": 8.079581510550101e-06,
+      "loss": 0.8099,
+      "step": 19048
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 3.455871343612671,
+      "learning_rate": 8.079373793298978e-06,
+      "loss": 0.8957,
+      "step": 19049
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.668999671936035,
+      "learning_rate": 8.079166067485269e-06,
+      "loss": 0.8217,
+      "step": 19050
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.887486219406128,
+      "learning_rate": 8.078958333109554e-06,
+      "loss": 0.8374,
+      "step": 19051
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 7.6266069412231445,
+      "learning_rate": 8.078750590172408e-06,
+      "loss": 0.7872,
+      "step": 19052
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.226341724395752,
+      "learning_rate": 8.078542838674413e-06,
+      "loss": 0.8596,
+      "step": 19053
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.6842975616455078,
+      "learning_rate": 8.07833507861614e-06,
+      "loss": 0.7036,
+      "step": 19054
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 3.2029778957366943,
+      "learning_rate": 8.078127309998172e-06,
+      "loss": 0.7733,
+      "step": 19055
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.3858118057250977,
+      "learning_rate": 8.077919532821084e-06,
+      "loss": 0.6387,
+      "step": 19056
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.9298441410064697,
+      "learning_rate": 8.077711747085457e-06,
+      "loss": 0.667,
+      "step": 19057
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.867825984954834,
+      "learning_rate": 8.077503952791865e-06,
+      "loss": 0.738,
+      "step": 19058
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.383756160736084,
+      "learning_rate": 8.077296149940887e-06,
+      "loss": 0.6592,
+      "step": 19059
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.297614812850952,
+      "learning_rate": 8.077088338533102e-06,
+      "loss": 0.7572,
+      "step": 19060
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.7799489498138428,
+      "learning_rate": 8.076880518569087e-06,
+      "loss": 0.6439,
+      "step": 19061
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.901233196258545,
+      "learning_rate": 8.07667269004942e-06,
+      "loss": 0.4682,
+      "step": 19062
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.1228244304656982,
+      "learning_rate": 8.076464852974678e-06,
+      "loss": 0.4094,
+      "step": 19063
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.6504839658737183,
+      "learning_rate": 8.076257007345441e-06,
+      "loss": 0.8836,
+      "step": 19064
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 4.195005893707275,
+      "learning_rate": 8.076049153162284e-06,
+      "loss": 0.6347,
+      "step": 19065
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.6950515508651733,
+      "learning_rate": 8.075841290425787e-06,
+      "loss": 0.7641,
+      "step": 19066
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 3.0246176719665527,
+      "learning_rate": 8.075633419136529e-06,
+      "loss": 0.5118,
+      "step": 19067
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 3.6025450229644775,
+      "learning_rate": 8.075425539295085e-06,
+      "loss": 0.8098,
+      "step": 19068
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.3676183223724365,
+      "learning_rate": 8.075217650902036e-06,
+      "loss": 0.6342,
+      "step": 19069
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.9410333633422852,
+      "learning_rate": 8.07500975395796e-06,
+      "loss": 0.7636,
+      "step": 19070
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.237901210784912,
+      "learning_rate": 8.074801848463429e-06,
+      "loss": 0.6915,
+      "step": 19071
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 3.3903095722198486,
+      "learning_rate": 8.074593934419029e-06,
+      "loss": 0.6136,
+      "step": 19072
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.453035354614258,
+      "learning_rate": 8.074386011825334e-06,
+      "loss": 0.5181,
+      "step": 19073
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.886291980743408,
+      "learning_rate": 8.074178080682927e-06,
+      "loss": 0.8057,
+      "step": 19074
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 3.4421048164367676,
+      "learning_rate": 8.073970140992377e-06,
+      "loss": 0.573,
+      "step": 19075
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.7060531377792358,
+      "learning_rate": 8.073762192754272e-06,
+      "loss": 0.8004,
+      "step": 19076
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 3.376936912536621,
+      "learning_rate": 8.073554235969184e-06,
+      "loss": 0.5982,
+      "step": 19077
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.2137720584869385,
+      "learning_rate": 8.073346270637693e-06,
+      "loss": 0.7885,
+      "step": 19078
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 3.7179768085479736,
+      "learning_rate": 8.073138296760376e-06,
+      "loss": 0.7981,
+      "step": 19079
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 3.015495777130127,
+      "learning_rate": 8.072930314337815e-06,
+      "loss": 0.5989,
+      "step": 19080
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.2079875469207764,
+      "learning_rate": 8.072722323370582e-06,
+      "loss": 0.6603,
+      "step": 19081
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.6126281023025513,
+      "learning_rate": 8.072514323859262e-06,
+      "loss": 0.4576,
+      "step": 19082
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.6868911981582642,
+      "learning_rate": 8.072306315804432e-06,
+      "loss": 0.5857,
+      "step": 19083
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 4.033559322357178,
+      "learning_rate": 8.072098299206668e-06,
+      "loss": 0.6431,
+      "step": 19084
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.7650835514068604,
+      "learning_rate": 8.071890274066549e-06,
+      "loss": 0.4069,
+      "step": 19085
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.4465893507003784,
+      "learning_rate": 8.071682240384654e-06,
+      "loss": 0.6774,
+      "step": 19086
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 3.074399709701538,
+      "learning_rate": 8.071474198161562e-06,
+      "loss": 0.4882,
+      "step": 19087
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.9446680545806885,
+      "learning_rate": 8.07126614739785e-06,
+      "loss": 0.8711,
+      "step": 19088
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.9169758558273315,
+      "learning_rate": 8.071058088094096e-06,
+      "loss": 0.6833,
+      "step": 19089
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.2439796924591064,
+      "learning_rate": 8.070850020250883e-06,
+      "loss": 0.8497,
+      "step": 19090
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 4.423413276672363,
+      "learning_rate": 8.070641943868785e-06,
+      "loss": 0.6957,
+      "step": 19091
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.8156890869140625,
+      "learning_rate": 8.070433858948382e-06,
+      "loss": 0.5343,
+      "step": 19092
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.6907901763916016,
+      "learning_rate": 8.070225765490252e-06,
+      "loss": 0.4818,
+      "step": 19093
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.854182243347168,
+      "learning_rate": 8.070017663494976e-06,
+      "loss": 0.694,
+      "step": 19094
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.9178218841552734,
+      "learning_rate": 8.06980955296313e-06,
+      "loss": 0.5082,
+      "step": 19095
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 4.569687366485596,
+      "learning_rate": 8.069601433895293e-06,
+      "loss": 0.5117,
+      "step": 19096
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.1878451108932495,
+      "learning_rate": 8.069393306292045e-06,
+      "loss": 0.4335,
+      "step": 19097
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.8219022750854492,
+      "learning_rate": 8.069185170153963e-06,
+      "loss": 0.6123,
+      "step": 19098
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 4.0085768699646,
+      "learning_rate": 8.068977025481628e-06,
+      "loss": 0.6701,
+      "step": 19099
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.0019028186798096,
+      "learning_rate": 8.068768872275619e-06,
+      "loss": 0.6092,
+      "step": 19100
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.571950674057007,
+      "learning_rate": 8.06856071053651e-06,
+      "loss": 0.5508,
+      "step": 19101
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.8714444637298584,
+      "learning_rate": 8.068352540264886e-06,
+      "loss": 0.5834,
+      "step": 19102
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 3.4709951877593994,
+      "learning_rate": 8.06814436146132e-06,
+      "loss": 0.7505,
+      "step": 19103
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.7443344593048096,
+      "learning_rate": 8.067936174126396e-06,
+      "loss": 0.7783,
+      "step": 19104
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.1539695262908936,
+      "learning_rate": 8.067727978260692e-06,
+      "loss": 0.6127,
+      "step": 19105
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.7446160316467285,
+      "learning_rate": 8.067519773864783e-06,
+      "loss": 0.9234,
+      "step": 19106
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.8860018253326416,
+      "learning_rate": 8.067311560939253e-06,
+      "loss": 0.6821,
+      "step": 19107
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.982081413269043,
+      "learning_rate": 8.067103339484677e-06,
+      "loss": 0.8892,
+      "step": 19108
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.4927607774734497,
+      "learning_rate": 8.066895109501636e-06,
+      "loss": 0.4603,
+      "step": 19109
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 3.4896578788757324,
+      "learning_rate": 8.066686870990707e-06,
+      "loss": 0.6023,
+      "step": 19110
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.5471692085266113,
+      "learning_rate": 8.066478623952473e-06,
+      "loss": 0.7097,
+      "step": 19111
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.8495326042175293,
+      "learning_rate": 8.06627036838751e-06,
+      "loss": 0.6282,
+      "step": 19112
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.7337369918823242,
+      "learning_rate": 8.066062104296397e-06,
+      "loss": 0.8165,
+      "step": 19113
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.790393590927124,
+      "learning_rate": 8.065853831679714e-06,
+      "loss": 0.6029,
+      "step": 19114
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 3.0677027702331543,
+      "learning_rate": 8.065645550538042e-06,
+      "loss": 0.6528,
+      "step": 19115
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.8473563194274902,
+      "learning_rate": 8.065437260871955e-06,
+      "loss": 0.5132,
+      "step": 19116
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.7150007486343384,
+      "learning_rate": 8.065228962682037e-06,
+      "loss": 0.8099,
+      "step": 19117
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.745522975921631,
+      "learning_rate": 8.065020655968865e-06,
+      "loss": 0.7402,
+      "step": 19118
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.2518537044525146,
+      "learning_rate": 8.064812340733019e-06,
+      "loss": 0.7405,
+      "step": 19119
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 3.1309173107147217,
+      "learning_rate": 8.064604016975079e-06,
+      "loss": 0.6109,
+      "step": 19120
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.206972360610962,
+      "learning_rate": 8.064395684695624e-06,
+      "loss": 0.6738,
+      "step": 19121
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.8841770887374878,
+      "learning_rate": 8.064187343895228e-06,
+      "loss": 0.6375,
+      "step": 19122
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 3.2921881675720215,
+      "learning_rate": 8.063978994574478e-06,
+      "loss": 0.8044,
+      "step": 19123
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.8249660730361938,
+      "learning_rate": 8.06377063673395e-06,
+      "loss": 0.6886,
+      "step": 19124
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.8882675170898438,
+      "learning_rate": 8.063562270374224e-06,
+      "loss": 0.5601,
+      "step": 19125
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.40921688079834,
+      "learning_rate": 8.06335389549588e-06,
+      "loss": 0.6959,
+      "step": 19126
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.8530231714248657,
+      "learning_rate": 8.063145512099493e-06,
+      "loss": 0.5101,
+      "step": 19127
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.0070366859436035,
+      "learning_rate": 8.062937120185649e-06,
+      "loss": 0.5175,
+      "step": 19128
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.6574883460998535,
+      "learning_rate": 8.062728719754922e-06,
+      "loss": 0.6749,
+      "step": 19129
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.899416446685791,
+      "learning_rate": 8.062520310807895e-06,
+      "loss": 0.6637,
+      "step": 19130
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.252291202545166,
+      "learning_rate": 8.062311893345147e-06,
+      "loss": 0.94,
+      "step": 19131
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.381749391555786,
+      "learning_rate": 8.062103467367254e-06,
+      "loss": 0.5252,
+      "step": 19132
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 3.4987125396728516,
+      "learning_rate": 8.0618950328748e-06,
+      "loss": 0.8214,
+      "step": 19133
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.9540290832519531,
+      "learning_rate": 8.061686589868364e-06,
+      "loss": 0.8164,
+      "step": 19134
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.624226689338684,
+      "learning_rate": 8.061478138348524e-06,
+      "loss": 0.6942,
+      "step": 19135
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.706721544265747,
+      "learning_rate": 8.06126967831586e-06,
+      "loss": 0.7619,
+      "step": 19136
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.8710963726043701,
+      "learning_rate": 8.06106120977095e-06,
+      "loss": 0.662,
+      "step": 19137
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.0114688873291016,
+      "learning_rate": 8.060852732714379e-06,
+      "loss": 0.8257,
+      "step": 19138
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.847597360610962,
+      "learning_rate": 8.06064424714672e-06,
+      "loss": 0.6657,
+      "step": 19139
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.9046928882598877,
+      "learning_rate": 8.06043575306856e-06,
+      "loss": 0.5834,
+      "step": 19140
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.958677053451538,
+      "learning_rate": 8.06022725048047e-06,
+      "loss": 0.6699,
+      "step": 19141
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 3.3648602962493896,
+      "learning_rate": 8.060018739383035e-06,
+      "loss": 0.528,
+      "step": 19142
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 3.320553779602051,
+      "learning_rate": 8.059810219776836e-06,
+      "loss": 0.6631,
+      "step": 19143
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.7484561204910278,
+      "learning_rate": 8.05960169166245e-06,
+      "loss": 0.5488,
+      "step": 19144
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.186344623565674,
+      "learning_rate": 8.059393155040459e-06,
+      "loss": 0.8576,
+      "step": 19145
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.6294695138931274,
+      "learning_rate": 8.05918460991144e-06,
+      "loss": 0.3689,
+      "step": 19146
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 3.4542524814605713,
+      "learning_rate": 8.058976056275976e-06,
+      "loss": 0.4758,
+      "step": 19147
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.5493342876434326,
+      "learning_rate": 8.058767494134643e-06,
+      "loss": 0.6407,
+      "step": 19148
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.5665509700775146,
+      "learning_rate": 8.058558923488026e-06,
+      "loss": 0.5326,
+      "step": 19149
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.601701259613037,
+      "learning_rate": 8.058350344336703e-06,
+      "loss": 0.6775,
+      "step": 19150
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 38.87457275390625,
+      "learning_rate": 8.058141756681253e-06,
+      "loss": 0.6886,
+      "step": 19151
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.776155471801758,
+      "learning_rate": 8.057933160522254e-06,
+      "loss": 0.9902,
+      "step": 19152
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.6792106628417969,
+      "learning_rate": 8.057724555860291e-06,
+      "loss": 0.5566,
+      "step": 19153
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.607623338699341,
+      "learning_rate": 8.05751594269594e-06,
+      "loss": 0.5876,
+      "step": 19154
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.5972644090652466,
+      "learning_rate": 8.057307321029782e-06,
+      "loss": 0.607,
+      "step": 19155
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.588935375213623,
+      "learning_rate": 8.057098690862399e-06,
+      "loss": 0.7952,
+      "step": 19156
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.2613751888275146,
+      "learning_rate": 8.05689005219437e-06,
+      "loss": 0.5437,
+      "step": 19157
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.083054304122925,
+      "learning_rate": 8.056681405026275e-06,
+      "loss": 0.6951,
+      "step": 19158
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.1921796798706055,
+      "learning_rate": 8.056472749358693e-06,
+      "loss": 0.8468,
+      "step": 19159
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 6.851712703704834,
+      "learning_rate": 8.056264085192205e-06,
+      "loss": 0.5356,
+      "step": 19160
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.0120952129364014,
+      "learning_rate": 8.056055412527393e-06,
+      "loss": 0.6552,
+      "step": 19161
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.6995224952697754,
+      "learning_rate": 8.055846731364834e-06,
+      "loss": 0.8695,
+      "step": 19162
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.760718584060669,
+      "learning_rate": 8.055638041705109e-06,
+      "loss": 0.4395,
+      "step": 19163
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.7465592622756958,
+      "learning_rate": 8.055429343548801e-06,
+      "loss": 0.7074,
+      "step": 19164
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.6647413969039917,
+      "learning_rate": 8.05522063689649e-06,
+      "loss": 0.6621,
+      "step": 19165
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.3282582759857178,
+      "learning_rate": 8.055011921748755e-06,
+      "loss": 0.6021,
+      "step": 19166
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.3783246278762817,
+      "learning_rate": 8.054803198106174e-06,
+      "loss": 0.7416,
+      "step": 19167
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.7760847806930542,
+      "learning_rate": 8.05459446596933e-06,
+      "loss": 0.7443,
+      "step": 19168
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 3.5746304988861084,
+      "learning_rate": 8.054385725338805e-06,
+      "loss": 0.7393,
+      "step": 19169
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.8637735843658447,
+      "learning_rate": 8.054176976215176e-06,
+      "loss": 0.6501,
+      "step": 19170
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 3.2555296421051025,
+      "learning_rate": 8.053968218599024e-06,
+      "loss": 0.5542,
+      "step": 19171
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 4.946590900421143,
+      "learning_rate": 8.053759452490932e-06,
+      "loss": 0.6599,
+      "step": 19172
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.714657187461853,
+      "learning_rate": 8.053550677891478e-06,
+      "loss": 0.625,
+      "step": 19173
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.5931003093719482,
+      "learning_rate": 8.053341894801243e-06,
+      "loss": 0.6055,
+      "step": 19174
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.157959222793579,
+      "learning_rate": 8.05313310322081e-06,
+      "loss": 0.6968,
+      "step": 19175
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.818476438522339,
+      "learning_rate": 8.052924303150755e-06,
+      "loss": 0.8043,
+      "step": 19176
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 3.228236436843872,
+      "learning_rate": 8.052715494591663e-06,
+      "loss": 0.8619,
+      "step": 19177
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.9463870525360107,
+      "learning_rate": 8.052506677544114e-06,
+      "loss": 0.5979,
+      "step": 19178
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 3.494149923324585,
+      "learning_rate": 8.052297852008684e-06,
+      "loss": 0.9054,
+      "step": 19179
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 4.06119966506958,
+      "learning_rate": 8.05208901798596e-06,
+      "loss": 0.6074,
+      "step": 19180
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 3.020399332046509,
+      "learning_rate": 8.05188017547652e-06,
+      "loss": 0.862,
+      "step": 19181
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.7270199060440063,
+      "learning_rate": 8.051671324480941e-06,
+      "loss": 0.6005,
+      "step": 19182
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.80865478515625,
+      "learning_rate": 8.051462464999809e-06,
+      "loss": 0.5819,
+      "step": 19183
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.292232036590576,
+      "learning_rate": 8.051253597033704e-06,
+      "loss": 0.6497,
+      "step": 19184
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.400214195251465,
+      "learning_rate": 8.051044720583205e-06,
+      "loss": 0.6237,
+      "step": 19185
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 4.16713285446167,
+      "learning_rate": 8.050835835648894e-06,
+      "loss": 0.5977,
+      "step": 19186
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.4876190423965454,
+      "learning_rate": 8.050626942231351e-06,
+      "loss": 0.6722,
+      "step": 19187
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.7535688877105713,
+      "learning_rate": 8.050418040331158e-06,
+      "loss": 0.9338,
+      "step": 19188
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.5173611640930176,
+      "learning_rate": 8.050209129948894e-06,
+      "loss": 0.5902,
+      "step": 19189
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.7520256042480469,
+      "learning_rate": 8.050000211085141e-06,
+      "loss": 0.4455,
+      "step": 19190
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.1079206466674805,
+      "learning_rate": 8.049791283740482e-06,
+      "loss": 0.5536,
+      "step": 19191
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.7298394441604614,
+      "learning_rate": 8.049582347915494e-06,
+      "loss": 0.8831,
+      "step": 19192
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 4.289029121398926,
+      "learning_rate": 8.04937340361076e-06,
+      "loss": 0.6571,
+      "step": 19193
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.4574708938598633,
+      "learning_rate": 8.049164450826861e-06,
+      "loss": 0.6463,
+      "step": 19194
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.9285742044448853,
+      "learning_rate": 8.048955489564378e-06,
+      "loss": 0.6192,
+      "step": 19195
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.4446847438812256,
+      "learning_rate": 8.048746519823892e-06,
+      "loss": 0.5597,
+      "step": 19196
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.8364615440368652,
+      "learning_rate": 8.048537541605986e-06,
+      "loss": 0.8474,
+      "step": 19197
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 4.035257816314697,
+      "learning_rate": 8.048328554911235e-06,
+      "loss": 0.9776,
+      "step": 19198
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 3.4030044078826904,
+      "learning_rate": 8.048119559740228e-06,
+      "loss": 0.5802,
+      "step": 19199
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.6136727333068848,
+      "learning_rate": 8.04791055609354e-06,
+      "loss": 0.7873,
+      "step": 19200
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.499821424484253,
+      "learning_rate": 8.047701543971755e-06,
+      "loss": 0.5623,
+      "step": 19201
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.0211598873138428,
+      "learning_rate": 8.047492523375455e-06,
+      "loss": 0.5555,
+      "step": 19202
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.887163519859314,
+      "learning_rate": 8.047283494305218e-06,
+      "loss": 0.7659,
+      "step": 19203
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.704350471496582,
+      "learning_rate": 8.047074456761628e-06,
+      "loss": 0.7276,
+      "step": 19204
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.097968578338623,
+      "learning_rate": 8.046865410745267e-06,
+      "loss": 0.5908,
+      "step": 19205
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.630446195602417,
+      "learning_rate": 8.046656356256713e-06,
+      "loss": 0.7336,
+      "step": 19206
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.3807947635650635,
+      "learning_rate": 8.046447293296549e-06,
+      "loss": 0.6603,
+      "step": 19207
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.7431976795196533,
+      "learning_rate": 8.046238221865356e-06,
+      "loss": 0.4376,
+      "step": 19208
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.6611852645874023,
+      "learning_rate": 8.046029141963716e-06,
+      "loss": 0.5736,
+      "step": 19209
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 3.565232515335083,
+      "learning_rate": 8.04582005359221e-06,
+      "loss": 0.7304,
+      "step": 19210
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 3.8449833393096924,
+      "learning_rate": 8.045610956751419e-06,
+      "loss": 0.5275,
+      "step": 19211
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.5479722023010254,
+      "learning_rate": 8.045401851441925e-06,
+      "loss": 0.7682,
+      "step": 19212
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.9212186336517334,
+      "learning_rate": 8.04519273766431e-06,
+      "loss": 0.5381,
+      "step": 19213
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.157848596572876,
+      "learning_rate": 8.044983615419154e-06,
+      "loss": 0.8304,
+      "step": 19214
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.1277785301208496,
+      "learning_rate": 8.04477448470704e-06,
+      "loss": 0.5791,
+      "step": 19215
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.397285223007202,
+      "learning_rate": 8.044565345528548e-06,
+      "loss": 0.7081,
+      "step": 19216
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 3.086232900619507,
+      "learning_rate": 8.04435619788426e-06,
+      "loss": 0.651,
+      "step": 19217
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.7733874320983887,
+      "learning_rate": 8.044147041774759e-06,
+      "loss": 0.5612,
+      "step": 19218
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.4274232387542725,
+      "learning_rate": 8.043937877200624e-06,
+      "loss": 0.5706,
+      "step": 19219
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.1066038608551025,
+      "learning_rate": 8.043728704162442e-06,
+      "loss": 0.3632,
+      "step": 19220
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 5.119612693786621,
+      "learning_rate": 8.043519522660786e-06,
+      "loss": 0.9697,
+      "step": 19221
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.219403028488159,
+      "learning_rate": 8.043310332696244e-06,
+      "loss": 0.5604,
+      "step": 19222
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.8998123407363892,
+      "learning_rate": 8.043101134269397e-06,
+      "loss": 0.7707,
+      "step": 19223
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.067811965942383,
+      "learning_rate": 8.042891927380823e-06,
+      "loss": 0.8703,
+      "step": 19224
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.9918524026870728,
+      "learning_rate": 8.042682712031109e-06,
+      "loss": 0.6249,
+      "step": 19225
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.1987528800964355,
+      "learning_rate": 8.042473488220833e-06,
+      "loss": 0.7589,
+      "step": 19226
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 3.7011306285858154,
+      "learning_rate": 8.042264255950579e-06,
+      "loss": 0.846,
+      "step": 19227
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.8314485549926758,
+      "learning_rate": 8.042055015220927e-06,
+      "loss": 0.7156,
+      "step": 19228
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.0800483226776123,
+      "learning_rate": 8.04184576603246e-06,
+      "loss": 0.6539,
+      "step": 19229
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 3.894385814666748,
+      "learning_rate": 8.04163650838576e-06,
+      "loss": 0.7751,
+      "step": 19230
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.0279619693756104,
+      "learning_rate": 8.041427242281407e-06,
+      "loss": 0.8484,
+      "step": 19231
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.6604783535003662,
+      "learning_rate": 8.041217967719985e-06,
+      "loss": 0.8502,
+      "step": 19232
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.95658540725708,
+      "learning_rate": 8.041008684702075e-06,
+      "loss": 0.539,
+      "step": 19233
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.703581690788269,
+      "learning_rate": 8.040799393228258e-06,
+      "loss": 0.7398,
+      "step": 19234
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.064363479614258,
+      "learning_rate": 8.040590093299117e-06,
+      "loss": 0.4843,
+      "step": 19235
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.2941386699676514,
+      "learning_rate": 8.040380784915234e-06,
+      "loss": 0.4591,
+      "step": 19236
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 4.573256492614746,
+      "learning_rate": 8.040171468077193e-06,
+      "loss": 0.5697,
+      "step": 19237
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.8957369327545166,
+      "learning_rate": 8.039962142785574e-06,
+      "loss": 0.8163,
+      "step": 19238
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.163590908050537,
+      "learning_rate": 8.039752809040957e-06,
+      "loss": 0.7459,
+      "step": 19239
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 3.531934976577759,
+      "learning_rate": 8.039543466843926e-06,
+      "loss": 0.915,
+      "step": 19240
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 4.499414920806885,
+      "learning_rate": 8.039334116195064e-06,
+      "loss": 1.1024,
+      "step": 19241
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.3669822216033936,
+      "learning_rate": 8.039124757094953e-06,
+      "loss": 0.5126,
+      "step": 19242
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.8991825580596924,
+      "learning_rate": 8.038915389544176e-06,
+      "loss": 0.7014,
+      "step": 19243
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.085852861404419,
+      "learning_rate": 8.038706013543311e-06,
+      "loss": 0.3637,
+      "step": 19244
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.246004104614258,
+      "learning_rate": 8.038496629092942e-06,
+      "loss": 0.7585,
+      "step": 19245
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 3.379948616027832,
+      "learning_rate": 8.038287236193655e-06,
+      "loss": 0.7228,
+      "step": 19246
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.407517671585083,
+      "learning_rate": 8.038077834846028e-06,
+      "loss": 0.4544,
+      "step": 19247
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.50613534450531,
+      "learning_rate": 8.037868425050646e-06,
+      "loss": 0.4696,
+      "step": 19248
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.5549120903015137,
+      "learning_rate": 8.037659006808087e-06,
+      "loss": 0.7351,
+      "step": 19249
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.5698089599609375,
+      "learning_rate": 8.03744958011894e-06,
+      "loss": 0.6991,
+      "step": 19250
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.4157321453094482,
+      "learning_rate": 8.03724014498378e-06,
+      "loss": 0.5435,
+      "step": 19251
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.4216420650482178,
+      "learning_rate": 8.037030701403194e-06,
+      "loss": 0.5242,
+      "step": 19252
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.9493985176086426,
+      "learning_rate": 8.036821249377765e-06,
+      "loss": 0.612,
+      "step": 19253
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.0368683338165283,
+      "learning_rate": 8.036611788908071e-06,
+      "loss": 0.5922,
+      "step": 19254
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.0443379878997803,
+      "learning_rate": 8.0364023199947e-06,
+      "loss": 0.8,
+      "step": 19255
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.1867361068725586,
+      "learning_rate": 8.036192842638231e-06,
+      "loss": 0.8497,
+      "step": 19256
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.6657543182373047,
+      "learning_rate": 8.035983356839245e-06,
+      "loss": 0.3626,
+      "step": 19257
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.768035888671875,
+      "learning_rate": 8.035773862598328e-06,
+      "loss": 0.5222,
+      "step": 19258
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.0824568271636963,
+      "learning_rate": 8.03556435991606e-06,
+      "loss": 0.6684,
+      "step": 19259
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.139322280883789,
+      "learning_rate": 8.03535484879303e-06,
+      "loss": 0.7353,
+      "step": 19260
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.9073781967163086,
+      "learning_rate": 8.03514532922981e-06,
+      "loss": 0.6832,
+      "step": 19261
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.2348499298095703,
+      "learning_rate": 8.034935801226988e-06,
+      "loss": 0.9794,
+      "step": 19262
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 3.475950241088867,
+      "learning_rate": 8.034726264785148e-06,
+      "loss": 0.5617,
+      "step": 19263
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.7366905212402344,
+      "learning_rate": 8.034516719904868e-06,
+      "loss": 0.8817,
+      "step": 19264
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 3.9533636569976807,
+      "learning_rate": 8.034307166586737e-06,
+      "loss": 0.6804,
+      "step": 19265
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.118861675262451,
+      "learning_rate": 8.034097604831335e-06,
+      "loss": 0.5461,
+      "step": 19266
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.4267613887786865,
+      "learning_rate": 8.033888034639243e-06,
+      "loss": 0.7904,
+      "step": 19267
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 3.146759033203125,
+      "learning_rate": 8.033678456011043e-06,
+      "loss": 0.6746,
+      "step": 19268
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.1562135219573975,
+      "learning_rate": 8.033468868947322e-06,
+      "loss": 0.6183,
+      "step": 19269
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.5372557640075684,
+      "learning_rate": 8.03325927344866e-06,
+      "loss": 0.6074,
+      "step": 19270
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.4781994819641113,
+      "learning_rate": 8.03304966951564e-06,
+      "loss": 0.4253,
+      "step": 19271
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.0984740257263184,
+      "learning_rate": 8.032840057148845e-06,
+      "loss": 0.6393,
+      "step": 19272
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 3.1258137226104736,
+      "learning_rate": 8.032630436348859e-06,
+      "loss": 0.7735,
+      "step": 19273
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.3582375049591064,
+      "learning_rate": 8.032420807116261e-06,
+      "loss": 0.4937,
+      "step": 19274
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.3283376693725586,
+      "learning_rate": 8.032211169451639e-06,
+      "loss": 0.6831,
+      "step": 19275
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 6.799741744995117,
+      "learning_rate": 8.032001523355574e-06,
+      "loss": 0.6301,
+      "step": 19276
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 4.435879230499268,
+      "learning_rate": 8.031791868828647e-06,
+      "loss": 0.817,
+      "step": 19277
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.142956018447876,
+      "learning_rate": 8.031582205871443e-06,
+      "loss": 0.5672,
+      "step": 19278
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.7172987461090088,
+      "learning_rate": 8.031372534484545e-06,
+      "loss": 0.598,
+      "step": 19279
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.6821157932281494,
+      "learning_rate": 8.031162854668536e-06,
+      "loss": 0.6583,
+      "step": 19280
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.3830759525299072,
+      "learning_rate": 8.030953166423998e-06,
+      "loss": 0.5114,
+      "step": 19281
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 3.651801824569702,
+      "learning_rate": 8.030743469751515e-06,
+      "loss": 0.5596,
+      "step": 19282
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.2466862201690674,
+      "learning_rate": 8.030533764651669e-06,
+      "loss": 0.7126,
+      "step": 19283
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 6.517943859100342,
+      "learning_rate": 8.030324051125044e-06,
+      "loss": 0.9208,
+      "step": 19284
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.6013875007629395,
+      "learning_rate": 8.030114329172222e-06,
+      "loss": 0.6402,
+      "step": 19285
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.3819289207458496,
+      "learning_rate": 8.02990459879379e-06,
+      "loss": 0.6504,
+      "step": 19286
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.8574841022491455,
+      "learning_rate": 8.029694859990326e-06,
+      "loss": 0.6984,
+      "step": 19287
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.993657350540161,
+      "learning_rate": 8.029485112762417e-06,
+      "loss": 0.6334,
+      "step": 19288
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.8567856550216675,
+      "learning_rate": 8.029275357110643e-06,
+      "loss": 0.5487,
+      "step": 19289
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.7779197692871094,
+      "learning_rate": 8.029065593035591e-06,
+      "loss": 0.4444,
+      "step": 19290
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.0173470973968506,
+      "learning_rate": 8.028855820537842e-06,
+      "loss": 0.624,
+      "step": 19291
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.263518810272217,
+      "learning_rate": 8.02864603961798e-06,
+      "loss": 0.7258,
+      "step": 19292
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 3.588526964187622,
+      "learning_rate": 8.028436250276585e-06,
+      "loss": 0.6993,
+      "step": 19293
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.7202239036560059,
+      "learning_rate": 8.028226452514247e-06,
+      "loss": 0.6569,
+      "step": 19294
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.4536304473876953,
+      "learning_rate": 8.028016646331542e-06,
+      "loss": 0.9365,
+      "step": 19295
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.0028676986694336,
+      "learning_rate": 8.027806831729059e-06,
+      "loss": 0.7687,
+      "step": 19296
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 3.074012041091919,
+      "learning_rate": 8.02759700870738e-06,
+      "loss": 0.7382,
+      "step": 19297
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.8282690048217773,
+      "learning_rate": 8.027387177267086e-06,
+      "loss": 0.7378,
+      "step": 19298
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.0470941066741943,
+      "learning_rate": 8.027177337408765e-06,
+      "loss": 0.7671,
+      "step": 19299
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 3.123342275619507,
+      "learning_rate": 8.026967489132995e-06,
+      "loss": 0.8087,
+      "step": 19300
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.911246657371521,
+      "learning_rate": 8.026757632440365e-06,
+      "loss": 0.6471,
+      "step": 19301
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.599554181098938,
+      "learning_rate": 8.026547767331454e-06,
+      "loss": 0.7438,
+      "step": 19302
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.7823330163955688,
+      "learning_rate": 8.026337893806848e-06,
+      "loss": 0.7492,
+      "step": 19303
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.037393093109131,
+      "learning_rate": 8.026128011867129e-06,
+      "loss": 0.5209,
+      "step": 19304
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 4.654261589050293,
+      "learning_rate": 8.025918121512883e-06,
+      "loss": 0.5141,
+      "step": 19305
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 5.172540664672852,
+      "learning_rate": 8.025708222744693e-06,
+      "loss": 0.8156,
+      "step": 19306
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 5.55525016784668,
+      "learning_rate": 8.02549831556314e-06,
+      "loss": 0.5673,
+      "step": 19307
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.804145574569702,
+      "learning_rate": 8.02528839996881e-06,
+      "loss": 0.5732,
+      "step": 19308
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.9548609256744385,
+      "learning_rate": 8.025078475962287e-06,
+      "loss": 0.6775,
+      "step": 19309
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.8959109783172607,
+      "learning_rate": 8.024868543544153e-06,
+      "loss": 0.7722,
+      "step": 19310
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.115112066268921,
+      "learning_rate": 8.024658602714994e-06,
+      "loss": 0.5189,
+      "step": 19311
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 3.379368782043457,
+      "learning_rate": 8.024448653475392e-06,
+      "loss": 0.8994,
+      "step": 19312
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.581974744796753,
+      "learning_rate": 8.02423869582593e-06,
+      "loss": 0.5155,
+      "step": 19313
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.1841742992401123,
+      "learning_rate": 8.024028729767195e-06,
+      "loss": 0.6214,
+      "step": 19314
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.4043238162994385,
+      "learning_rate": 8.023818755299767e-06,
+      "loss": 0.7092,
+      "step": 19315
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 4.162422180175781,
+      "learning_rate": 8.023608772424233e-06,
+      "loss": 0.9464,
+      "step": 19316
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 3.081224203109741,
+      "learning_rate": 8.023398781141177e-06,
+      "loss": 0.8421,
+      "step": 19317
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.8014471530914307,
+      "learning_rate": 8.02318878145118e-06,
+      "loss": 0.7012,
+      "step": 19318
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.053870439529419,
+      "learning_rate": 8.022978773354826e-06,
+      "loss": 0.6713,
+      "step": 19319
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.8250751495361328,
+      "learning_rate": 8.022768756852703e-06,
+      "loss": 0.8247,
+      "step": 19320
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.989130735397339,
+      "learning_rate": 8.022558731945392e-06,
+      "loss": 0.4824,
+      "step": 19321
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.0709359645843506,
+      "learning_rate": 8.022348698633477e-06,
+      "loss": 0.6511,
+      "step": 19322
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.9466660022735596,
+      "learning_rate": 8.022138656917542e-06,
+      "loss": 0.7576,
+      "step": 19323
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.4106996059417725,
+      "learning_rate": 8.021928606798173e-06,
+      "loss": 0.8003,
+      "step": 19324
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.407121181488037,
+      "learning_rate": 8.02171854827595e-06,
+      "loss": 0.7609,
+      "step": 19325
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.8153282403945923,
+      "learning_rate": 8.021508481351461e-06,
+      "loss": 0.5886,
+      "step": 19326
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.4656591415405273,
+      "learning_rate": 8.021298406025289e-06,
+      "loss": 0.8983,
+      "step": 19327
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.519927978515625,
+      "learning_rate": 8.02108832229802e-06,
+      "loss": 0.5435,
+      "step": 19328
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.803286552429199,
+      "learning_rate": 8.020878230170232e-06,
+      "loss": 0.7841,
+      "step": 19329
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.896401047706604,
+      "learning_rate": 8.020668129642516e-06,
+      "loss": 0.8851,
+      "step": 19330
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 3.271052360534668,
+      "learning_rate": 8.020458020715453e-06,
+      "loss": 0.6242,
+      "step": 19331
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.321781039237976,
+      "learning_rate": 8.020247903389627e-06,
+      "loss": 0.613,
+      "step": 19332
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 3.04621958732605,
+      "learning_rate": 8.020037777665623e-06,
+      "loss": 0.5502,
+      "step": 19333
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.4549877643585205,
+      "learning_rate": 8.019827643544027e-06,
+      "loss": 0.5733,
+      "step": 19334
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 4.294954299926758,
+      "learning_rate": 8.01961750102542e-06,
+      "loss": 0.4931,
+      "step": 19335
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 3.7357866764068604,
+      "learning_rate": 8.019407350110388e-06,
+      "loss": 0.7873,
+      "step": 19336
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.22635555267334,
+      "learning_rate": 8.019197190799516e-06,
+      "loss": 0.6637,
+      "step": 19337
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.081160068511963,
+      "learning_rate": 8.018987023093388e-06,
+      "loss": 0.5515,
+      "step": 19338
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.255697250366211,
+      "learning_rate": 8.018776846992586e-06,
+      "loss": 0.6319,
+      "step": 19339
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.856323480606079,
+      "learning_rate": 8.018566662497697e-06,
+      "loss": 0.5764,
+      "step": 19340
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.1100034713745117,
+      "learning_rate": 8.018356469609307e-06,
+      "loss": 0.8,
+      "step": 19341
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 3.035769462585449,
+      "learning_rate": 8.018146268327996e-06,
+      "loss": 0.7822,
+      "step": 19342
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.927046537399292,
+      "learning_rate": 8.017936058654352e-06,
+      "loss": 0.5912,
+      "step": 19343
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.0965116024017334,
+      "learning_rate": 8.017725840588959e-06,
+      "loss": 0.6932,
+      "step": 19344
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 3.045701503753662,
+      "learning_rate": 8.017515614132399e-06,
+      "loss": 0.5,
+      "step": 19345
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.4815689325332642,
+      "learning_rate": 8.017305379285259e-06,
+      "loss": 0.5173,
+      "step": 19346
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.974630355834961,
+      "learning_rate": 8.017095136048124e-06,
+      "loss": 0.6574,
+      "step": 19347
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.879828453063965,
+      "learning_rate": 8.016884884421577e-06,
+      "loss": 0.7636,
+      "step": 19348
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.9368960857391357,
+      "learning_rate": 8.016674624406205e-06,
+      "loss": 0.7066,
+      "step": 19349
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.103161096572876,
+      "learning_rate": 8.01646435600259e-06,
+      "loss": 0.696,
+      "step": 19350
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.691218852996826,
+      "learning_rate": 8.016254079211317e-06,
+      "loss": 0.604,
+      "step": 19351
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.4089175462722778,
+      "learning_rate": 8.016043794032971e-06,
+      "loss": 0.5947,
+      "step": 19352
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.5189456939697266,
+      "learning_rate": 8.015833500468137e-06,
+      "loss": 0.6889,
+      "step": 19353
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.385197639465332,
+      "learning_rate": 8.015623198517399e-06,
+      "loss": 0.6037,
+      "step": 19354
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.0581042766571045,
+      "learning_rate": 8.015412888181344e-06,
+      "loss": 0.7892,
+      "step": 19355
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.1382477283477783,
+      "learning_rate": 8.015202569460555e-06,
+      "loss": 0.856,
+      "step": 19356
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.8381808996200562,
+      "learning_rate": 8.014992242355617e-06,
+      "loss": 0.5595,
+      "step": 19357
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.508479356765747,
+      "learning_rate": 8.014781906867115e-06,
+      "loss": 0.9298,
+      "step": 19358
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.8965976238250732,
+      "learning_rate": 8.014571562995636e-06,
+      "loss": 0.5996,
+      "step": 19359
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.9966347217559814,
+      "learning_rate": 8.01436121074176e-06,
+      "loss": 0.5949,
+      "step": 19360
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.6235681772232056,
+      "learning_rate": 8.014150850106076e-06,
+      "loss": 0.5167,
+      "step": 19361
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.8247913122177124,
+      "learning_rate": 8.013940481089168e-06,
+      "loss": 0.5094,
+      "step": 19362
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.363596200942993,
+      "learning_rate": 8.01373010369162e-06,
+      "loss": 0.6489,
+      "step": 19363
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.9990363121032715,
+      "learning_rate": 8.013519717914017e-06,
+      "loss": 0.3556,
+      "step": 19364
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.4132378101348877,
+      "learning_rate": 8.013309323756944e-06,
+      "loss": 0.426,
+      "step": 19365
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.2505390644073486,
+      "learning_rate": 8.013098921220988e-06,
+      "loss": 0.6356,
+      "step": 19366
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.6578912734985352,
+      "learning_rate": 8.012888510306732e-06,
+      "loss": 0.945,
+      "step": 19367
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.26456356048584,
+      "learning_rate": 8.012678091014763e-06,
+      "loss": 0.5064,
+      "step": 19368
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.992796540260315,
+      "learning_rate": 8.012467663345661e-06,
+      "loss": 0.7865,
+      "step": 19369
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.9863933324813843,
+      "learning_rate": 8.01225722730002e-06,
+      "loss": 0.5606,
+      "step": 19370
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.7684075832366943,
+      "learning_rate": 8.012046782878417e-06,
+      "loss": 0.6915,
+      "step": 19371
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.1749939918518066,
+      "learning_rate": 8.011836330081442e-06,
+      "loss": 0.5231,
+      "step": 19372
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.8805395364761353,
+      "learning_rate": 8.011625868909675e-06,
+      "loss": 0.6697,
+      "step": 19373
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.8880326747894287,
+      "learning_rate": 8.011415399363709e-06,
+      "loss": 0.9693,
+      "step": 19374
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.7754757404327393,
+      "learning_rate": 8.011204921444122e-06,
+      "loss": 0.5165,
+      "step": 19375
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.9954155683517456,
+      "learning_rate": 8.010994435151503e-06,
+      "loss": 0.5717,
+      "step": 19376
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.2214977741241455,
+      "learning_rate": 8.010783940486436e-06,
+      "loss": 0.7065,
+      "step": 19377
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.0602986812591553,
+      "learning_rate": 8.010573437449507e-06,
+      "loss": 0.803,
+      "step": 19378
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.122919797897339,
+      "learning_rate": 8.010362926041302e-06,
+      "loss": 0.5866,
+      "step": 19379
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.026362419128418,
+      "learning_rate": 8.010152406262404e-06,
+      "loss": 0.6426,
+      "step": 19380
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.9567575454711914,
+      "learning_rate": 8.009941878113401e-06,
+      "loss": 0.7234,
+      "step": 19381
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.595674753189087,
+      "learning_rate": 8.009731341594876e-06,
+      "loss": 0.5011,
+      "step": 19382
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.4962995052337646,
+      "learning_rate": 8.009520796707418e-06,
+      "loss": 0.793,
+      "step": 19383
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.2262163162231445,
+      "learning_rate": 8.009310243451606e-06,
+      "loss": 0.7466,
+      "step": 19384
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.671834707260132,
+      "learning_rate": 8.009099681828032e-06,
+      "loss": 0.426,
+      "step": 19385
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.7432096004486084,
+      "learning_rate": 8.008889111837279e-06,
+      "loss": 0.9309,
+      "step": 19386
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 4.294921398162842,
+      "learning_rate": 8.008678533479932e-06,
+      "loss": 0.9758,
+      "step": 19387
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.090178966522217,
+      "learning_rate": 8.008467946756577e-06,
+      "loss": 0.7785,
+      "step": 19388
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.1473419666290283,
+      "learning_rate": 8.008257351667802e-06,
+      "loss": 0.6235,
+      "step": 19389
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.122807741165161,
+      "learning_rate": 8.008046748214187e-06,
+      "loss": 0.7354,
+      "step": 19390
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 5.639915943145752,
+      "learning_rate": 8.007836136396323e-06,
+      "loss": 0.761,
+      "step": 19391
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.844860553741455,
+      "learning_rate": 8.007625516214791e-06,
+      "loss": 0.8846,
+      "step": 19392
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.4148967266082764,
+      "learning_rate": 8.007414887670182e-06,
+      "loss": 0.6412,
+      "step": 19393
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.000640392303467,
+      "learning_rate": 8.007204250763077e-06,
+      "loss": 0.7714,
+      "step": 19394
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.805041790008545,
+      "learning_rate": 8.006993605494064e-06,
+      "loss": 0.6669,
+      "step": 19395
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 4.055415153503418,
+      "learning_rate": 8.006782951863727e-06,
+      "loss": 0.8028,
+      "step": 19396
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.0343406200408936,
+      "learning_rate": 8.006572289872654e-06,
+      "loss": 0.7507,
+      "step": 19397
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.733437418937683,
+      "learning_rate": 8.006361619521431e-06,
+      "loss": 0.6595,
+      "step": 19398
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.097100019454956,
+      "learning_rate": 8.006150940810643e-06,
+      "loss": 0.8177,
+      "step": 19399
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.6103339195251465,
+      "learning_rate": 8.005940253740873e-06,
+      "loss": 0.8736,
+      "step": 19400
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.011251926422119,
+      "learning_rate": 8.005729558312709e-06,
+      "loss": 0.772,
+      "step": 19401
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.3016417026519775,
+      "learning_rate": 8.00551885452674e-06,
+      "loss": 0.5023,
+      "step": 19402
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.9583113193511963,
+      "learning_rate": 8.005308142383546e-06,
+      "loss": 0.6491,
+      "step": 19403
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.9250478744506836,
+      "learning_rate": 8.005097421883717e-06,
+      "loss": 0.6151,
+      "step": 19404
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.8239574432373047,
+      "learning_rate": 8.004886693027838e-06,
+      "loss": 0.863,
+      "step": 19405
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.180797815322876,
+      "learning_rate": 8.004675955816494e-06,
+      "loss": 0.7667,
+      "step": 19406
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.658926010131836,
+      "learning_rate": 8.004465210250271e-06,
+      "loss": 0.7653,
+      "step": 19407
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.3007497787475586,
+      "learning_rate": 8.004254456329759e-06,
+      "loss": 0.6797,
+      "step": 19408
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.566049337387085,
+      "learning_rate": 8.004043694055539e-06,
+      "loss": 0.6085,
+      "step": 19409
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.118401288986206,
+      "learning_rate": 8.003832923428198e-06,
+      "loss": 0.7311,
+      "step": 19410
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.5150253772735596,
+      "learning_rate": 8.003622144448322e-06,
+      "loss": 0.4765,
+      "step": 19411
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.9445266723632812,
+      "learning_rate": 8.0034113571165e-06,
+      "loss": 0.8944,
+      "step": 19412
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.9848389625549316,
+      "learning_rate": 8.003200561433317e-06,
+      "loss": 0.7733,
+      "step": 19413
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.640150785446167,
+      "learning_rate": 8.002989757399356e-06,
+      "loss": 0.5808,
+      "step": 19414
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.4640052318573,
+      "learning_rate": 8.002778945015207e-06,
+      "loss": 0.668,
+      "step": 19415
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 4.036068439483643,
+      "learning_rate": 8.002568124281454e-06,
+      "loss": 0.9784,
+      "step": 19416
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.0636961460113525,
+      "learning_rate": 8.002357295198682e-06,
+      "loss": 0.7018,
+      "step": 19417
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.0835864543914795,
+      "learning_rate": 8.00214645776748e-06,
+      "loss": 0.7853,
+      "step": 19418
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.9569451808929443,
+      "learning_rate": 8.001935611988436e-06,
+      "loss": 0.7796,
+      "step": 19419
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.512611985206604,
+      "learning_rate": 8.001724757862132e-06,
+      "loss": 0.5627,
+      "step": 19420
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.8746583461761475,
+      "learning_rate": 8.001513895389155e-06,
+      "loss": 0.6859,
+      "step": 19421
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.2057688236236572,
+      "learning_rate": 8.001303024570094e-06,
+      "loss": 0.7293,
+      "step": 19422
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.6691319942474365,
+      "learning_rate": 8.001092145405533e-06,
+      "loss": 0.7421,
+      "step": 19423
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.2874796390533447,
+      "learning_rate": 8.000881257896058e-06,
+      "loss": 0.8539,
+      "step": 19424
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.8790342807769775,
+      "learning_rate": 8.000670362042257e-06,
+      "loss": 1.0023,
+      "step": 19425
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.8509581089019775,
+      "learning_rate": 8.000459457844715e-06,
+      "loss": 0.8142,
+      "step": 19426
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.920824408531189,
+      "learning_rate": 8.00024854530402e-06,
+      "loss": 0.7505,
+      "step": 19427
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.0199670791625977,
+      "learning_rate": 8.000037624420758e-06,
+      "loss": 0.8589,
+      "step": 19428
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.2467501163482666,
+      "learning_rate": 7.999826695195516e-06,
+      "loss": 0.6013,
+      "step": 19429
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.271923780441284,
+      "learning_rate": 7.999615757628879e-06,
+      "loss": 0.7629,
+      "step": 19430
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.9770054817199707,
+      "learning_rate": 7.999404811721432e-06,
+      "loss": 0.796,
+      "step": 19431
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.5708565711975098,
+      "learning_rate": 7.999193857473765e-06,
+      "loss": 0.7488,
+      "step": 19432
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.6414129734039307,
+      "learning_rate": 7.998982894886465e-06,
+      "loss": 0.6793,
+      "step": 19433
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.3057632446289062,
+      "learning_rate": 7.998771923960116e-06,
+      "loss": 0.7189,
+      "step": 19434
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.8585104942321777,
+      "learning_rate": 7.998560944695306e-06,
+      "loss": 0.6564,
+      "step": 19435
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.1352696418762207,
+      "learning_rate": 7.998349957092622e-06,
+      "loss": 0.6718,
+      "step": 19436
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.2172460556030273,
+      "learning_rate": 7.998138961152649e-06,
+      "loss": 0.7004,
+      "step": 19437
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.6340317726135254,
+      "learning_rate": 7.997927956875974e-06,
+      "loss": 0.7891,
+      "step": 19438
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.0376200675964355,
+      "learning_rate": 7.997716944263185e-06,
+      "loss": 0.6921,
+      "step": 19439
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.718959093093872,
+      "learning_rate": 7.997505923314867e-06,
+      "loss": 0.6945,
+      "step": 19440
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.230884313583374,
+      "learning_rate": 7.997294894031611e-06,
+      "loss": 0.6682,
+      "step": 19441
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.973170280456543,
+      "learning_rate": 7.997083856413998e-06,
+      "loss": 0.8322,
+      "step": 19442
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.19659423828125,
+      "learning_rate": 7.99687281046262e-06,
+      "loss": 0.7631,
+      "step": 19443
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.5029964447021484,
+      "learning_rate": 7.99666175617806e-06,
+      "loss": 0.7268,
+      "step": 19444
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.3953683376312256,
+      "learning_rate": 7.996450693560905e-06,
+      "loss": 0.8082,
+      "step": 19445
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.947349786758423,
+      "learning_rate": 7.996239622611744e-06,
+      "loss": 0.6861,
+      "step": 19446
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.363756537437439,
+      "learning_rate": 7.996028543331162e-06,
+      "loss": 0.6204,
+      "step": 19447
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.184985876083374,
+      "learning_rate": 7.995817455719749e-06,
+      "loss": 0.5008,
+      "step": 19448
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.528468132019043,
+      "learning_rate": 7.99560635977809e-06,
+      "loss": 0.7202,
+      "step": 19449
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.9522902965545654,
+      "learning_rate": 7.99539525550677e-06,
+      "loss": 0.8336,
+      "step": 19450
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.7134106159210205,
+      "learning_rate": 7.995184142906378e-06,
+      "loss": 0.8001,
+      "step": 19451
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.10979962348938,
+      "learning_rate": 7.994973021977503e-06,
+      "loss": 0.5351,
+      "step": 19452
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 7.410547733306885,
+      "learning_rate": 7.994761892720729e-06,
+      "loss": 0.5459,
+      "step": 19453
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.947471022605896,
+      "learning_rate": 7.994550755136643e-06,
+      "loss": 0.5481,
+      "step": 19454
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.1015021800994873,
+      "learning_rate": 7.994339609225834e-06,
+      "loss": 0.7457,
+      "step": 19455
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 4.630913734436035,
+      "learning_rate": 7.99412845498889e-06,
+      "loss": 0.7919,
+      "step": 19456
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.124600410461426,
+      "learning_rate": 7.993917292426392e-06,
+      "loss": 0.6757,
+      "step": 19457
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.04793643951416,
+      "learning_rate": 7.993706121538934e-06,
+      "loss": 0.8401,
+      "step": 19458
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.5189677476882935,
+      "learning_rate": 7.9934949423271e-06,
+      "loss": 0.3264,
+      "step": 19459
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.9055098295211792,
+      "learning_rate": 7.993283754791481e-06,
+      "loss": 0.8173,
+      "step": 19460
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.7937512397766113,
+      "learning_rate": 7.993072558932659e-06,
+      "loss": 0.6454,
+      "step": 19461
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.8577184677124023,
+      "learning_rate": 7.992861354751223e-06,
+      "loss": 0.5967,
+      "step": 19462
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.814024567604065,
+      "learning_rate": 7.992650142247761e-06,
+      "loss": 0.581,
+      "step": 19463
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.0221362113952637,
+      "learning_rate": 7.99243892142286e-06,
+      "loss": 1.0084,
+      "step": 19464
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.879833698272705,
+      "learning_rate": 7.992227692277107e-06,
+      "loss": 0.668,
+      "step": 19465
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.49971866607666,
+      "learning_rate": 7.992016454811091e-06,
+      "loss": 0.6895,
+      "step": 19466
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.505532741546631,
+      "learning_rate": 7.991805209025397e-06,
+      "loss": 0.7927,
+      "step": 19467
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.6080431938171387,
+      "learning_rate": 7.991593954920614e-06,
+      "loss": 0.5938,
+      "step": 19468
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.7447247505187988,
+      "learning_rate": 7.991382692497328e-06,
+      "loss": 0.8667,
+      "step": 19469
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.3240110874176025,
+      "learning_rate": 7.991171421756127e-06,
+      "loss": 0.621,
+      "step": 19470
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.365770936012268,
+      "learning_rate": 7.990960142697602e-06,
+      "loss": 0.935,
+      "step": 19471
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.4386630058288574,
+      "learning_rate": 7.990748855322334e-06,
+      "loss": 0.427,
+      "step": 19472
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.8314766883850098,
+      "learning_rate": 7.990537559630915e-06,
+      "loss": 0.6121,
+      "step": 19473
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.8006255626678467,
+      "learning_rate": 7.990326255623932e-06,
+      "loss": 0.6177,
+      "step": 19474
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.2859480381011963,
+      "learning_rate": 7.99011494330197e-06,
+      "loss": 0.5259,
+      "step": 19475
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.0430362224578857,
+      "learning_rate": 7.98990362266562e-06,
+      "loss": 0.4085,
+      "step": 19476
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.6790027618408203,
+      "learning_rate": 7.989692293715468e-06,
+      "loss": 0.6739,
+      "step": 19477
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.9296727180480957,
+      "learning_rate": 7.9894809564521e-06,
+      "loss": 0.7573,
+      "step": 19478
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 4.96384334564209,
+      "learning_rate": 7.989269610876108e-06,
+      "loss": 0.6706,
+      "step": 19479
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.6632739305496216,
+      "learning_rate": 7.989058256988076e-06,
+      "loss": 0.6017,
+      "step": 19480
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.1545355319976807,
+      "learning_rate": 7.988846894788592e-06,
+      "loss": 0.6728,
+      "step": 19481
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.762545585632324,
+      "learning_rate": 7.988635524278245e-06,
+      "loss": 0.5634,
+      "step": 19482
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.628087043762207,
+      "learning_rate": 7.988424145457623e-06,
+      "loss": 0.5935,
+      "step": 19483
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.105731248855591,
+      "learning_rate": 7.98821275832731e-06,
+      "loss": 0.7357,
+      "step": 19484
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.3895509243011475,
+      "learning_rate": 7.9880013628879e-06,
+      "loss": 0.6344,
+      "step": 19485
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.0647196769714355,
+      "learning_rate": 7.987789959139975e-06,
+      "loss": 0.5413,
+      "step": 19486
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.088679075241089,
+      "learning_rate": 7.987578547084128e-06,
+      "loss": 0.634,
+      "step": 19487
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.024069309234619,
+      "learning_rate": 7.987367126720942e-06,
+      "loss": 0.528,
+      "step": 19488
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.022108793258667,
+      "learning_rate": 7.987155698051008e-06,
+      "loss": 0.6108,
+      "step": 19489
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.4227800369262695,
+      "learning_rate": 7.986944261074913e-06,
+      "loss": 0.4811,
+      "step": 19490
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.6403526067733765,
+      "learning_rate": 7.986732815793244e-06,
+      "loss": 0.5166,
+      "step": 19491
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.106400489807129,
+      "learning_rate": 7.986521362206592e-06,
+      "loss": 0.5561,
+      "step": 19492
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.523498773574829,
+      "learning_rate": 7.986309900315541e-06,
+      "loss": 0.7878,
+      "step": 19493
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.325249671936035,
+      "learning_rate": 7.986098430120682e-06,
+      "loss": 0.6703,
+      "step": 19494
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.204493522644043,
+      "learning_rate": 7.985886951622601e-06,
+      "loss": 0.7246,
+      "step": 19495
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 4.430333137512207,
+      "learning_rate": 7.985675464821887e-06,
+      "loss": 0.6509,
+      "step": 19496
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.821730375289917,
+      "learning_rate": 7.985463969719128e-06,
+      "loss": 0.5265,
+      "step": 19497
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.5778322219848633,
+      "learning_rate": 7.985252466314912e-06,
+      "loss": 0.6754,
+      "step": 19498
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.829054594039917,
+      "learning_rate": 7.985040954609828e-06,
+      "loss": 0.7121,
+      "step": 19499
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.4934487342834473,
+      "learning_rate": 7.984829434604463e-06,
+      "loss": 0.6431,
+      "step": 19500
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.787660598754883,
+      "learning_rate": 7.984617906299405e-06,
+      "loss": 0.4623,
+      "step": 19501
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.045626163482666,
+      "learning_rate": 7.984406369695244e-06,
+      "loss": 0.5022,
+      "step": 19502
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.4114603996276855,
+      "learning_rate": 7.984194824792567e-06,
+      "loss": 0.8097,
+      "step": 19503
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.994791030883789,
+      "learning_rate": 7.98398327159196e-06,
+      "loss": 0.4758,
+      "step": 19504
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.351638078689575,
+      "learning_rate": 7.983771710094015e-06,
+      "loss": 0.4598,
+      "step": 19505
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.6684532165527344,
+      "learning_rate": 7.983560140299319e-06,
+      "loss": 0.7105,
+      "step": 19506
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.001549243927002,
+      "learning_rate": 7.983348562208457e-06,
+      "loss": 0.5786,
+      "step": 19507
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.7228283882141113,
+      "learning_rate": 7.983136975822022e-06,
+      "loss": 0.9265,
+      "step": 19508
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 4.048925399780273,
+      "learning_rate": 7.9829253811406e-06,
+      "loss": 1.1143,
+      "step": 19509
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.943964958190918,
+      "learning_rate": 7.98271377816478e-06,
+      "loss": 0.648,
+      "step": 19510
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.983510732650757,
+      "learning_rate": 7.982502166895153e-06,
+      "loss": 0.7769,
+      "step": 19511
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.100790023803711,
+      "learning_rate": 7.982290547332302e-06,
+      "loss": 0.7957,
+      "step": 19512
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.8249616622924805,
+      "learning_rate": 7.982078919476818e-06,
+      "loss": 0.8945,
+      "step": 19513
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.130969762802124,
+      "learning_rate": 7.981867283329292e-06,
+      "loss": 0.735,
+      "step": 19514
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.230483293533325,
+      "learning_rate": 7.981655638890308e-06,
+      "loss": 0.5043,
+      "step": 19515
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.1745442152023315,
+      "learning_rate": 7.981443986160457e-06,
+      "loss": 0.5117,
+      "step": 19516
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.601393222808838,
+      "learning_rate": 7.981232325140329e-06,
+      "loss": 0.8785,
+      "step": 19517
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.458401918411255,
+      "learning_rate": 7.981020655830508e-06,
+      "loss": 0.5237,
+      "step": 19518
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.1638238430023193,
+      "learning_rate": 7.980808978231586e-06,
+      "loss": 0.6407,
+      "step": 19519
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.0979576110839844,
+      "learning_rate": 7.980597292344151e-06,
+      "loss": 0.6803,
+      "step": 19520
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.3854994773864746,
+      "learning_rate": 7.980385598168792e-06,
+      "loss": 0.4859,
+      "step": 19521
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 5.07581090927124,
+      "learning_rate": 7.980173895706098e-06,
+      "loss": 1.1281,
+      "step": 19522
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.8776392936706543,
+      "learning_rate": 7.979962184956655e-06,
+      "loss": 0.488,
+      "step": 19523
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 4.271410942077637,
+      "learning_rate": 7.979750465921055e-06,
+      "loss": 0.7662,
+      "step": 19524
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.6108760833740234,
+      "learning_rate": 7.979538738599883e-06,
+      "loss": 0.8054,
+      "step": 19525
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.475423574447632,
+      "learning_rate": 7.979327002993733e-06,
+      "loss": 0.4978,
+      "step": 19526
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.1879420280456543,
+      "learning_rate": 7.979115259103188e-06,
+      "loss": 0.5464,
+      "step": 19527
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.0012941360473633,
+      "learning_rate": 7.978903506928842e-06,
+      "loss": 0.5712,
+      "step": 19528
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.061652898788452,
+      "learning_rate": 7.978691746471279e-06,
+      "loss": 0.3499,
+      "step": 19529
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.429758071899414,
+      "learning_rate": 7.97847997773109e-06,
+      "loss": 0.6499,
+      "step": 19530
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.2602059841156006,
+      "learning_rate": 7.978268200708865e-06,
+      "loss": 0.6549,
+      "step": 19531
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.91875159740448,
+      "learning_rate": 7.978056415405192e-06,
+      "loss": 0.6558,
+      "step": 19532
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.084426164627075,
+      "learning_rate": 7.977844621820658e-06,
+      "loss": 0.7266,
+      "step": 19533
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.9416282176971436,
+      "learning_rate": 7.977632819955856e-06,
+      "loss": 0.629,
+      "step": 19534
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.5083460807800293,
+      "learning_rate": 7.977421009811372e-06,
+      "loss": 0.5266,
+      "step": 19535
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.9991114139556885,
+      "learning_rate": 7.977209191387794e-06,
+      "loss": 0.5881,
+      "step": 19536
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 6.336739540100098,
+      "learning_rate": 7.976997364685714e-06,
+      "loss": 0.571,
+      "step": 19537
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.718412160873413,
+      "learning_rate": 7.97678552970572e-06,
+      "loss": 0.6923,
+      "step": 19538
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.3605740070343018,
+      "learning_rate": 7.9765736864484e-06,
+      "loss": 0.5595,
+      "step": 19539
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.271193504333496,
+      "learning_rate": 7.97636183491434e-06,
+      "loss": 0.3541,
+      "step": 19540
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.627063751220703,
+      "learning_rate": 7.976149975104138e-06,
+      "loss": 0.7307,
+      "step": 19541
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.366794466972351,
+      "learning_rate": 7.975938107018376e-06,
+      "loss": 0.7113,
+      "step": 19542
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.72776198387146,
+      "learning_rate": 7.975726230657646e-06,
+      "loss": 0.7076,
+      "step": 19543
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.3865160942077637,
+      "learning_rate": 7.975514346022534e-06,
+      "loss": 0.5984,
+      "step": 19544
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.0593414306640625,
+      "learning_rate": 7.975302453113631e-06,
+      "loss": 0.6773,
+      "step": 19545
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 5.1505913734436035,
+      "learning_rate": 7.975090551931527e-06,
+      "loss": 0.7004,
+      "step": 19546
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.6972970962524414,
+      "learning_rate": 7.974878642476813e-06,
+      "loss": 0.6679,
+      "step": 19547
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.1342999935150146,
+      "learning_rate": 7.974666724750074e-06,
+      "loss": 0.6956,
+      "step": 19548
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.899924635887146,
+      "learning_rate": 7.9744547987519e-06,
+      "loss": 0.7216,
+      "step": 19549
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.2894177436828613,
+      "learning_rate": 7.974242864482884e-06,
+      "loss": 0.7105,
+      "step": 19550
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.971914529800415,
+      "learning_rate": 7.974030921943612e-06,
+      "loss": 0.6649,
+      "step": 19551
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.9296422004699707,
+      "learning_rate": 7.973818971134672e-06,
+      "loss": 0.5307,
+      "step": 19552
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 5.076889514923096,
+      "learning_rate": 7.973607012056659e-06,
+      "loss": 0.5077,
+      "step": 19553
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.547903060913086,
+      "learning_rate": 7.973395044710157e-06,
+      "loss": 0.5425,
+      "step": 19554
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 5.90574836730957,
+      "learning_rate": 7.973183069095756e-06,
+      "loss": 1.1782,
+      "step": 19555
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.128443956375122,
+      "learning_rate": 7.97297108521405e-06,
+      "loss": 0.6192,
+      "step": 19556
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.1773366928100586,
+      "learning_rate": 7.972759093065621e-06,
+      "loss": 0.5287,
+      "step": 19557
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.1746561527252197,
+      "learning_rate": 7.972547092651065e-06,
+      "loss": 0.6484,
+      "step": 19558
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.623894691467285,
+      "learning_rate": 7.972335083970968e-06,
+      "loss": 0.6785,
+      "step": 19559
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.4445724487304688,
+      "learning_rate": 7.972123067025923e-06,
+      "loss": 0.7579,
+      "step": 19560
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.5580832958221436,
+      "learning_rate": 7.971911041816513e-06,
+      "loss": 0.6216,
+      "step": 19561
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.8730003833770752,
+      "learning_rate": 7.971699008343335e-06,
+      "loss": 0.8686,
+      "step": 19562
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.5772664546966553,
+      "learning_rate": 7.971486966606974e-06,
+      "loss": 0.6241,
+      "step": 19563
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.9622137546539307,
+      "learning_rate": 7.971274916608022e-06,
+      "loss": 0.6692,
+      "step": 19564
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.5428436994552612,
+      "learning_rate": 7.971062858347068e-06,
+      "loss": 0.9447,
+      "step": 19565
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 5.254523754119873,
+      "learning_rate": 7.970850791824698e-06,
+      "loss": 1.0581,
+      "step": 19566
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.8210175037384033,
+      "learning_rate": 7.970638717041508e-06,
+      "loss": 0.5047,
+      "step": 19567
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.3412740230560303,
+      "learning_rate": 7.970426633998083e-06,
+      "loss": 0.6393,
+      "step": 19568
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.893669605255127,
+      "learning_rate": 7.970214542695014e-06,
+      "loss": 0.6413,
+      "step": 19569
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.4559454917907715,
+      "learning_rate": 7.970002443132891e-06,
+      "loss": 0.6165,
+      "step": 19570
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.424224615097046,
+      "learning_rate": 7.969790335312306e-06,
+      "loss": 0.6389,
+      "step": 19571
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.933109164237976,
+      "learning_rate": 7.969578219233845e-06,
+      "loss": 0.7666,
+      "step": 19572
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.8172746896743774,
+      "learning_rate": 7.9693660948981e-06,
+      "loss": 0.4522,
+      "step": 19573
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.4097158908843994,
+      "learning_rate": 7.969153962305657e-06,
+      "loss": 0.7991,
+      "step": 19574
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.7187820672988892,
+      "learning_rate": 7.968941821457114e-06,
+      "loss": 0.2932,
+      "step": 19575
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.3642449378967285,
+      "learning_rate": 7.968729672353052e-06,
+      "loss": 0.5317,
+      "step": 19576
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.429802179336548,
+      "learning_rate": 7.968517514994067e-06,
+      "loss": 0.6605,
+      "step": 19577
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.060904026031494,
+      "learning_rate": 7.968305349380747e-06,
+      "loss": 0.5624,
+      "step": 19578
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.0483505725860596,
+      "learning_rate": 7.96809317551368e-06,
+      "loss": 0.7361,
+      "step": 19579
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.0211033821105957,
+      "learning_rate": 7.96788099339346e-06,
+      "loss": 0.2834,
+      "step": 19580
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.4314727783203125,
+      "learning_rate": 7.967668803020674e-06,
+      "loss": 0.524,
+      "step": 19581
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.4511067867279053,
+      "learning_rate": 7.967456604395912e-06,
+      "loss": 0.7765,
+      "step": 19582
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.8871382474899292,
+      "learning_rate": 7.967244397519766e-06,
+      "loss": 0.6,
+      "step": 19583
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.720801591873169,
+      "learning_rate": 7.967032182392824e-06,
+      "loss": 0.7328,
+      "step": 19584
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.268829345703125,
+      "learning_rate": 7.966819959015677e-06,
+      "loss": 0.7396,
+      "step": 19585
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.615691661834717,
+      "learning_rate": 7.966607727388915e-06,
+      "loss": 0.4775,
+      "step": 19586
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.7098662853240967,
+      "learning_rate": 7.966395487513128e-06,
+      "loss": 0.5082,
+      "step": 19587
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.3763203620910645,
+      "learning_rate": 7.966183239388907e-06,
+      "loss": 0.6053,
+      "step": 19588
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.883544445037842,
+      "learning_rate": 7.965970983016842e-06,
+      "loss": 0.7854,
+      "step": 19589
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.0148072242736816,
+      "learning_rate": 7.965758718397522e-06,
+      "loss": 0.5194,
+      "step": 19590
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.6464922428131104,
+      "learning_rate": 7.965546445531539e-06,
+      "loss": 0.5888,
+      "step": 19591
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.5117509365081787,
+      "learning_rate": 7.96533416441948e-06,
+      "loss": 0.5041,
+      "step": 19592
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.1023104190826416,
+      "learning_rate": 7.96512187506194e-06,
+      "loss": 0.7559,
+      "step": 19593
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.1444249153137207,
+      "learning_rate": 7.964909577459506e-06,
+      "loss": 0.4794,
+      "step": 19594
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.361372947692871,
+      "learning_rate": 7.964697271612768e-06,
+      "loss": 0.6822,
+      "step": 19595
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.3945868015289307,
+      "learning_rate": 7.96448495752232e-06,
+      "loss": 0.5769,
+      "step": 19596
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.0453479290008545,
+      "learning_rate": 7.964272635188748e-06,
+      "loss": 0.5882,
+      "step": 19597
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.438995122909546,
+      "learning_rate": 7.964060304612646e-06,
+      "loss": 0.6781,
+      "step": 19598
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.2413643598556519,
+      "learning_rate": 7.9638479657946e-06,
+      "loss": 0.4489,
+      "step": 19599
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 4.438167572021484,
+      "learning_rate": 7.963635618735203e-06,
+      "loss": 0.9123,
+      "step": 19600
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.9551877975463867,
+      "learning_rate": 7.963423263435048e-06,
+      "loss": 0.7734,
+      "step": 19601
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.7387986183166504,
+      "learning_rate": 7.963210899894723e-06,
+      "loss": 0.6618,
+      "step": 19602
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.9321529865264893,
+      "learning_rate": 7.962998528114817e-06,
+      "loss": 0.7938,
+      "step": 19603
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.180068254470825,
+      "learning_rate": 7.962786148095923e-06,
+      "loss": 0.7733,
+      "step": 19604
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.7629846334457397,
+      "learning_rate": 7.96257375983863e-06,
+      "loss": 0.524,
+      "step": 19605
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.531834363937378,
+      "learning_rate": 7.96236136334353e-06,
+      "loss": 0.5616,
+      "step": 19606
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.1468067169189453,
+      "learning_rate": 7.96214895861121e-06,
+      "loss": 0.7176,
+      "step": 19607
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.4548730850219727,
+      "learning_rate": 7.961936545642266e-06,
+      "loss": 0.9286,
+      "step": 19608
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.9905704259872437,
+      "learning_rate": 7.961724124437286e-06,
+      "loss": 0.6641,
+      "step": 19609
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.2283935546875,
+      "learning_rate": 7.961511694996859e-06,
+      "loss": 0.6869,
+      "step": 19610
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.8140971660614014,
+      "learning_rate": 7.961299257321579e-06,
+      "loss": 0.5419,
+      "step": 19611
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.3979005813598633,
+      "learning_rate": 7.961086811412036e-06,
+      "loss": 0.7574,
+      "step": 19612
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.0719897747039795,
+      "learning_rate": 7.960874357268818e-06,
+      "loss": 0.7867,
+      "step": 19613
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.4083130359649658,
+      "learning_rate": 7.960661894892518e-06,
+      "loss": 0.6201,
+      "step": 19614
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.4228036403656006,
+      "learning_rate": 7.960449424283726e-06,
+      "loss": 0.5965,
+      "step": 19615
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.9210397005081177,
+      "learning_rate": 7.960236945443035e-06,
+      "loss": 0.223,
+      "step": 19616
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.260072708129883,
+      "learning_rate": 7.960024458371031e-06,
+      "loss": 0.5492,
+      "step": 19617
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.7385716438293457,
+      "learning_rate": 7.959811963068309e-06,
+      "loss": 0.6709,
+      "step": 19618
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.9057424068450928,
+      "learning_rate": 7.95959945953546e-06,
+      "loss": 0.6199,
+      "step": 19619
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.993299722671509,
+      "learning_rate": 7.959386947773071e-06,
+      "loss": 0.6363,
+      "step": 19620
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.5570480823516846,
+      "learning_rate": 7.959174427781739e-06,
+      "loss": 0.6601,
+      "step": 19621
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.505525827407837,
+      "learning_rate": 7.958961899562049e-06,
+      "loss": 0.6782,
+      "step": 19622
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.5812888145446777,
+      "learning_rate": 7.958749363114594e-06,
+      "loss": 0.7196,
+      "step": 19623
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.7467315196990967,
+      "learning_rate": 7.958536818439968e-06,
+      "loss": 0.5298,
+      "step": 19624
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.6562421321868896,
+      "learning_rate": 7.958324265538758e-06,
+      "loss": 0.4528,
+      "step": 19625
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.3821675777435303,
+      "learning_rate": 7.958111704411555e-06,
+      "loss": 0.6845,
+      "step": 19626
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.472763776779175,
+      "learning_rate": 7.957899135058952e-06,
+      "loss": 0.7257,
+      "step": 19627
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.0494039058685303,
+      "learning_rate": 7.957686557481543e-06,
+      "loss": 0.1746,
+      "step": 19628
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.1947147846221924,
+      "learning_rate": 7.957473971679913e-06,
+      "loss": 0.7113,
+      "step": 19629
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.8691492080688477,
+      "learning_rate": 7.957261377654655e-06,
+      "loss": 0.4687,
+      "step": 19630
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.7303187847137451,
+      "learning_rate": 7.95704877540636e-06,
+      "loss": 0.581,
+      "step": 19631
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.439690351486206,
+      "learning_rate": 7.956836164935623e-06,
+      "loss": 0.4854,
+      "step": 19632
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.0353009700775146,
+      "learning_rate": 7.95662354624303e-06,
+      "loss": 0.678,
+      "step": 19633
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.5002987384796143,
+      "learning_rate": 7.956410919329176e-06,
+      "loss": 0.6027,
+      "step": 19634
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.793330669403076,
+      "learning_rate": 7.956198284194651e-06,
+      "loss": 0.4583,
+      "step": 19635
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.5085549354553223,
+      "learning_rate": 7.955985640840044e-06,
+      "loss": 0.3795,
+      "step": 19636
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.9168365001678467,
+      "learning_rate": 7.95577298926595e-06,
+      "loss": 0.9499,
+      "step": 19637
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.1705667972564697,
+      "learning_rate": 7.95556032947296e-06,
+      "loss": 0.4105,
+      "step": 19638
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.9858534336090088,
+      "learning_rate": 7.95534766146166e-06,
+      "loss": 0.9134,
+      "step": 19639
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.2770097255706787,
+      "learning_rate": 7.955134985232648e-06,
+      "loss": 0.8728,
+      "step": 19640
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.518033027648926,
+      "learning_rate": 7.954922300786511e-06,
+      "loss": 0.746,
+      "step": 19641
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.661496162414551,
+      "learning_rate": 7.954709608123843e-06,
+      "loss": 0.5118,
+      "step": 19642
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.1703341007232666,
+      "learning_rate": 7.954496907245235e-06,
+      "loss": 0.5987,
+      "step": 19643
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.4172847270965576,
+      "learning_rate": 7.954284198151278e-06,
+      "loss": 0.7949,
+      "step": 19644
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.0878489017486572,
+      "learning_rate": 7.954071480842562e-06,
+      "loss": 0.8749,
+      "step": 19645
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.1332437992095947,
+      "learning_rate": 7.953858755319681e-06,
+      "loss": 0.6478,
+      "step": 19646
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.7700519561767578,
+      "learning_rate": 7.953646021583225e-06,
+      "loss": 0.6605,
+      "step": 19647
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.060194253921509,
+      "learning_rate": 7.953433279633785e-06,
+      "loss": 0.679,
+      "step": 19648
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.9643150568008423,
+      "learning_rate": 7.953220529471954e-06,
+      "loss": 0.6737,
+      "step": 19649
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.2965104579925537,
+      "learning_rate": 7.953007771098323e-06,
+      "loss": 0.6172,
+      "step": 19650
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.4356613159179688,
+      "learning_rate": 7.952795004513484e-06,
+      "loss": 0.6628,
+      "step": 19651
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.5774612426757812,
+      "learning_rate": 7.952582229718027e-06,
+      "loss": 0.7463,
+      "step": 19652
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.4336960315704346,
+      "learning_rate": 7.952369446712546e-06,
+      "loss": 0.9662,
+      "step": 19653
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.9454922676086426,
+      "learning_rate": 7.952156655497632e-06,
+      "loss": 0.5344,
+      "step": 19654
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.767992377281189,
+      "learning_rate": 7.951943856073876e-06,
+      "loss": 0.7067,
+      "step": 19655
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.1596274375915527,
+      "learning_rate": 7.951731048441869e-06,
+      "loss": 0.6732,
+      "step": 19656
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.1919705867767334,
+      "learning_rate": 7.951518232602204e-06,
+      "loss": 0.7057,
+      "step": 19657
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 5.019277095794678,
+      "learning_rate": 7.951305408555475e-06,
+      "loss": 0.7381,
+      "step": 19658
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.4643032550811768,
+      "learning_rate": 7.951092576302268e-06,
+      "loss": 0.6397,
+      "step": 19659
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.003758668899536,
+      "learning_rate": 7.95087973584318e-06,
+      "loss": 0.3669,
+      "step": 19660
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.8410295248031616,
+      "learning_rate": 7.950666887178797e-06,
+      "loss": 0.7083,
+      "step": 19661
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.905318021774292,
+      "learning_rate": 7.950454030309717e-06,
+      "loss": 0.7465,
+      "step": 19662
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.915853500366211,
+      "learning_rate": 7.950241165236532e-06,
+      "loss": 1.0624,
+      "step": 19663
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.316776752471924,
+      "learning_rate": 7.950028291959831e-06,
+      "loss": 0.8421,
+      "step": 19664
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.7056727409362793,
+      "learning_rate": 7.949815410480204e-06,
+      "loss": 0.5351,
+      "step": 19665
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.2149893045425415,
+      "learning_rate": 7.949602520798246e-06,
+      "loss": 0.6897,
+      "step": 19666
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.5568623542785645,
+      "learning_rate": 7.94938962291455e-06,
+      "loss": 0.6741,
+      "step": 19667
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.6709673404693604,
+      "learning_rate": 7.949176716829705e-06,
+      "loss": 0.847,
+      "step": 19668
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.0742347240448,
+      "learning_rate": 7.948963802544305e-06,
+      "loss": 0.6018,
+      "step": 19669
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.848427414894104,
+      "learning_rate": 7.94875088005894e-06,
+      "loss": 0.8024,
+      "step": 19670
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.9696599245071411,
+      "learning_rate": 7.948537949374204e-06,
+      "loss": 0.5357,
+      "step": 19671
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.051215648651123,
+      "learning_rate": 7.94832501049069e-06,
+      "loss": 0.5996,
+      "step": 19672
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.835576295852661,
+      "learning_rate": 7.948112063408985e-06,
+      "loss": 0.4854,
+      "step": 19673
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.914305329322815,
+      "learning_rate": 7.947899108129688e-06,
+      "loss": 0.4729,
+      "step": 19674
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.6841679811477661,
+      "learning_rate": 7.947686144653388e-06,
+      "loss": 0.8378,
+      "step": 19675
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.674225091934204,
+      "learning_rate": 7.947473172980676e-06,
+      "loss": 0.4466,
+      "step": 19676
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.9985780715942383,
+      "learning_rate": 7.947260193112145e-06,
+      "loss": 0.5999,
+      "step": 19677
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.8177614212036133,
+      "learning_rate": 7.947047205048388e-06,
+      "loss": 0.5528,
+      "step": 19678
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.3735506534576416,
+      "learning_rate": 7.946834208789997e-06,
+      "loss": 0.5981,
+      "step": 19679
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.406416416168213,
+      "learning_rate": 7.946621204337563e-06,
+      "loss": 0.5215,
+      "step": 19680
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.0760915279388428,
+      "learning_rate": 7.94640819169168e-06,
+      "loss": 0.5963,
+      "step": 19681
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.6483936309814453,
+      "learning_rate": 7.946195170852939e-06,
+      "loss": 0.6507,
+      "step": 19682
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.6750450134277344,
+      "learning_rate": 7.945982141821934e-06,
+      "loss": 0.5037,
+      "step": 19683
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.2020931243896484,
+      "learning_rate": 7.945769104599255e-06,
+      "loss": 0.7178,
+      "step": 19684
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.176727533340454,
+      "learning_rate": 7.945556059185496e-06,
+      "loss": 0.8813,
+      "step": 19685
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.1653566360473633,
+      "learning_rate": 7.94534300558125e-06,
+      "loss": 0.7371,
+      "step": 19686
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.858354926109314,
+      "learning_rate": 7.945129943787107e-06,
+      "loss": 0.5445,
+      "step": 19687
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.523958444595337,
+      "learning_rate": 7.944916873803664e-06,
+      "loss": 0.4906,
+      "step": 19688
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.562401294708252,
+      "learning_rate": 7.944703795631507e-06,
+      "loss": 0.7041,
+      "step": 19689
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.2640540599823,
+      "learning_rate": 7.944490709271232e-06,
+      "loss": 0.6696,
+      "step": 19690
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.183478355407715,
+      "learning_rate": 7.944277614723431e-06,
+      "loss": 0.6975,
+      "step": 19691
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.657722234725952,
+      "learning_rate": 7.944064511988698e-06,
+      "loss": 0.6331,
+      "step": 19692
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.6072816848754883,
+      "learning_rate": 7.943851401067627e-06,
+      "loss": 0.6701,
+      "step": 19693
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.13030743598938,
+      "learning_rate": 7.943638281960804e-06,
+      "loss": 0.7006,
+      "step": 19694
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 6.090198993682861,
+      "learning_rate": 7.943425154668827e-06,
+      "loss": 0.8796,
+      "step": 19695
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.443934679031372,
+      "learning_rate": 7.943212019192288e-06,
+      "loss": 0.7384,
+      "step": 19696
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.366419553756714,
+      "learning_rate": 7.942998875531779e-06,
+      "loss": 0.661,
+      "step": 19697
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.2905070781707764,
+      "learning_rate": 7.942785723687891e-06,
+      "loss": 0.6326,
+      "step": 19698
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.742736339569092,
+      "learning_rate": 7.94257256366122e-06,
+      "loss": 0.5962,
+      "step": 19699
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.099071741104126,
+      "learning_rate": 7.942359395452358e-06,
+      "loss": 0.8418,
+      "step": 19700
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.3330607414245605,
+      "learning_rate": 7.942146219061892e-06,
+      "loss": 0.4941,
+      "step": 19701
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.428288459777832,
+      "learning_rate": 7.941933034490424e-06,
+      "loss": 0.6195,
+      "step": 19702
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.8528592586517334,
+      "learning_rate": 7.94171984173854e-06,
+      "loss": 0.6758,
+      "step": 19703
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.3379628658294678,
+      "learning_rate": 7.941506640806837e-06,
+      "loss": 0.6374,
+      "step": 19704
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.0869195461273193,
+      "learning_rate": 7.941293431695904e-06,
+      "loss": 0.5051,
+      "step": 19705
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.59086012840271,
+      "learning_rate": 7.941080214406337e-06,
+      "loss": 0.53,
+      "step": 19706
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.442681074142456,
+      "learning_rate": 7.940866988938727e-06,
+      "loss": 0.828,
+      "step": 19707
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.6118204593658447,
+      "learning_rate": 7.94065375529367e-06,
+      "loss": 0.5106,
+      "step": 19708
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.179378032684326,
+      "learning_rate": 7.940440513471753e-06,
+      "loss": 0.6118,
+      "step": 19709
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.7574303150177002,
+      "learning_rate": 7.940227263473574e-06,
+      "loss": 0.6161,
+      "step": 19710
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.303861379623413,
+      "learning_rate": 7.940014005299726e-06,
+      "loss": 0.7654,
+      "step": 19711
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.318417549133301,
+      "learning_rate": 7.939800738950798e-06,
+      "loss": 0.5989,
+      "step": 19712
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.803537607192993,
+      "learning_rate": 7.939587464427384e-06,
+      "loss": 0.5249,
+      "step": 19713
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.196983814239502,
+      "learning_rate": 7.939374181730081e-06,
+      "loss": 0.6277,
+      "step": 19714
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.2666633129119873,
+      "learning_rate": 7.939160890859481e-06,
+      "loss": 0.46,
+      "step": 19715
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.588790774345398,
+      "learning_rate": 7.938947591816174e-06,
+      "loss": 0.5374,
+      "step": 19716
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.3008203506469727,
+      "learning_rate": 7.938734284600753e-06,
+      "loss": 0.4683,
+      "step": 19717
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.518564224243164,
+      "learning_rate": 7.938520969213814e-06,
+      "loss": 0.8411,
+      "step": 19718
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.6261173486709595,
+      "learning_rate": 7.93830764565595e-06,
+      "loss": 0.6148,
+      "step": 19719
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.8620522022247314,
+      "learning_rate": 7.938094313927754e-06,
+      "loss": 0.4698,
+      "step": 19720
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.3169755935668945,
+      "learning_rate": 7.937880974029816e-06,
+      "loss": 0.5836,
+      "step": 19721
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.3830392360687256,
+      "learning_rate": 7.93766762596273e-06,
+      "loss": 1.0392,
+      "step": 19722
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.8611369132995605,
+      "learning_rate": 7.937454269727094e-06,
+      "loss": 0.6371,
+      "step": 19723
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.274498701095581,
+      "learning_rate": 7.937240905323498e-06,
+      "loss": 0.4597,
+      "step": 19724
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.7449886798858643,
+      "learning_rate": 7.937027532752535e-06,
+      "loss": 0.81,
+      "step": 19725
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.8169021606445312,
+      "learning_rate": 7.936814152014797e-06,
+      "loss": 0.6377,
+      "step": 19726
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.482199192047119,
+      "learning_rate": 7.93660076311088e-06,
+      "loss": 0.5305,
+      "step": 19727
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.0883641242980957,
+      "learning_rate": 7.936387366041378e-06,
+      "loss": 0.7886,
+      "step": 19728
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.409018874168396,
+      "learning_rate": 7.936173960806882e-06,
+      "loss": 0.6168,
+      "step": 19729
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.957037091255188,
+      "learning_rate": 7.935960547407985e-06,
+      "loss": 0.8307,
+      "step": 19730
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.132763624191284,
+      "learning_rate": 7.935747125845282e-06,
+      "loss": 0.6276,
+      "step": 19731
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 4.568691253662109,
+      "learning_rate": 7.935533696119365e-06,
+      "loss": 1.0901,
+      "step": 19732
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.242098331451416,
+      "learning_rate": 7.93532025823083e-06,
+      "loss": 0.7241,
+      "step": 19733
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.670912742614746,
+      "learning_rate": 7.935106812180268e-06,
+      "loss": 0.8347,
+      "step": 19734
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.1775119304656982,
+      "learning_rate": 7.934893357968273e-06,
+      "loss": 0.776,
+      "step": 19735
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.628225326538086,
+      "learning_rate": 7.934679895595441e-06,
+      "loss": 0.617,
+      "step": 19736
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.42378306388855,
+      "learning_rate": 7.934466425062363e-06,
+      "loss": 0.59,
+      "step": 19737
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.032970905303955,
+      "learning_rate": 7.934252946369632e-06,
+      "loss": 0.9371,
+      "step": 19738
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.9132492542266846,
+      "learning_rate": 7.934039459517844e-06,
+      "loss": 0.8651,
+      "step": 19739
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.0803310871124268,
+      "learning_rate": 7.933825964507591e-06,
+      "loss": 0.6636,
+      "step": 19740
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.908841371536255,
+      "learning_rate": 7.933612461339465e-06,
+      "loss": 0.6594,
+      "step": 19741
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.8035720586776733,
+      "learning_rate": 7.933398950014063e-06,
+      "loss": 0.3216,
+      "step": 19742
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.6644086837768555,
+      "learning_rate": 7.93318543053198e-06,
+      "loss": 0.7195,
+      "step": 19743
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.7141270637512207,
+      "learning_rate": 7.932971902893805e-06,
+      "loss": 0.6604,
+      "step": 19744
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.7339441776275635,
+      "learning_rate": 7.932758367100134e-06,
+      "loss": 0.5895,
+      "step": 19745
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.119055986404419,
+      "learning_rate": 7.93254482315156e-06,
+      "loss": 0.5779,
+      "step": 19746
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.074467420578003,
+      "learning_rate": 7.932331271048678e-06,
+      "loss": 0.673,
+      "step": 19747
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 4.643789291381836,
+      "learning_rate": 7.932117710792082e-06,
+      "loss": 0.4439,
+      "step": 19748
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.6054151058197021,
+      "learning_rate": 7.931904142382363e-06,
+      "loss": 0.624,
+      "step": 19749
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.6542426347732544,
+      "learning_rate": 7.931690565820118e-06,
+      "loss": 0.8251,
+      "step": 19750
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 7.005558013916016,
+      "learning_rate": 7.931476981105941e-06,
+      "loss": 0.4494,
+      "step": 19751
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.7101272344589233,
+      "learning_rate": 7.931263388240421e-06,
+      "loss": 0.7733,
+      "step": 19752
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.507807493209839,
+      "learning_rate": 7.931049787224159e-06,
+      "loss": 0.5947,
+      "step": 19753
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.4450573921203613,
+      "learning_rate": 7.930836178057746e-06,
+      "loss": 0.6131,
+      "step": 19754
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.6594531536102295,
+      "learning_rate": 7.930622560741774e-06,
+      "loss": 0.7292,
+      "step": 19755
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.6340131759643555,
+      "learning_rate": 7.93040893527684e-06,
+      "loss": 0.7587,
+      "step": 19756
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.5615038871765137,
+      "learning_rate": 7.930195301663532e-06,
+      "loss": 0.8693,
+      "step": 19757
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.3934032917022705,
+      "learning_rate": 7.929981659902453e-06,
+      "loss": 0.6084,
+      "step": 19758
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.835211753845215,
+      "learning_rate": 7.929768009994192e-06,
+      "loss": 0.4479,
+      "step": 19759
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.043372392654419,
+      "learning_rate": 7.929554351939342e-06,
+      "loss": 0.7404,
+      "step": 19760
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.8966251611709595,
+      "learning_rate": 7.9293406857385e-06,
+      "loss": 0.7455,
+      "step": 19761
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.8473007678985596,
+      "learning_rate": 7.929127011392258e-06,
+      "loss": 0.5714,
+      "step": 19762
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.274168014526367,
+      "learning_rate": 7.928913328901211e-06,
+      "loss": 0.6433,
+      "step": 19763
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.672589063644409,
+      "learning_rate": 7.928699638265955e-06,
+      "loss": 0.899,
+      "step": 19764
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.782985210418701,
+      "learning_rate": 7.928485939487082e-06,
+      "loss": 0.6422,
+      "step": 19765
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.5149614810943604,
+      "learning_rate": 7.928272232565186e-06,
+      "loss": 0.4977,
+      "step": 19766
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.670945644378662,
+      "learning_rate": 7.92805851750086e-06,
+      "loss": 0.5102,
+      "step": 19767
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.3168210983276367,
+      "learning_rate": 7.927844794294703e-06,
+      "loss": 0.8651,
+      "step": 19768
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.7885072231292725,
+      "learning_rate": 7.927631062947306e-06,
+      "loss": 0.4665,
+      "step": 19769
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.858361840248108,
+      "learning_rate": 7.927417323459263e-06,
+      "loss": 0.4292,
+      "step": 19770
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.4390746355056763,
+      "learning_rate": 7.927203575831169e-06,
+      "loss": 0.8364,
+      "step": 19771
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.4378381967544556,
+      "learning_rate": 7.92698982006362e-06,
+      "loss": 0.343,
+      "step": 19772
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.0030429363250732,
+      "learning_rate": 7.926776056157206e-06,
+      "loss": 0.8453,
+      "step": 19773
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.2629995346069336,
+      "learning_rate": 7.926562284112525e-06,
+      "loss": 0.6906,
+      "step": 19774
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 4.5495734214782715,
+      "learning_rate": 7.926348503930172e-06,
+      "loss": 0.7749,
+      "step": 19775
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.501300096511841,
+      "learning_rate": 7.926134715610739e-06,
+      "loss": 0.5973,
+      "step": 19776
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.9288955926895142,
+      "learning_rate": 7.92592091915482e-06,
+      "loss": 0.3322,
+      "step": 19777
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.5635018348693848,
+      "learning_rate": 7.925707114563016e-06,
+      "loss": 0.5022,
+      "step": 19778
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.6155846118927,
+      "learning_rate": 7.925493301835911e-06,
+      "loss": 0.7496,
+      "step": 19779
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.5361337661743164,
+      "learning_rate": 7.925279480974108e-06,
+      "loss": 0.9312,
+      "step": 19780
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.4738316535949707,
+      "learning_rate": 7.925065651978199e-06,
+      "loss": 0.5514,
+      "step": 19781
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.419337034225464,
+      "learning_rate": 7.924851814848777e-06,
+      "loss": 0.3048,
+      "step": 19782
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.5138928890228271,
+      "learning_rate": 7.924637969586438e-06,
+      "loss": 0.625,
+      "step": 19783
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 4.069387435913086,
+      "learning_rate": 7.924424116191776e-06,
+      "loss": 0.5456,
+      "step": 19784
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.8683130741119385,
+      "learning_rate": 7.924210254665385e-06,
+      "loss": 0.5684,
+      "step": 19785
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.1438944339752197,
+      "learning_rate": 7.923996385007862e-06,
+      "loss": 1.0042,
+      "step": 19786
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.71130633354187,
+      "learning_rate": 7.923782507219801e-06,
+      "loss": 0.9057,
+      "step": 19787
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.692547082901001,
+      "learning_rate": 7.923568621301796e-06,
+      "loss": 0.8423,
+      "step": 19788
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.6047873497009277,
+      "learning_rate": 7.92335472725444e-06,
+      "loss": 0.526,
+      "step": 19789
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.1894776821136475,
+      "learning_rate": 7.923140825078333e-06,
+      "loss": 0.6285,
+      "step": 19790
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.051053762435913,
+      "learning_rate": 7.922926914774064e-06,
+      "loss": 0.6595,
+      "step": 19791
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.340481758117676,
+      "learning_rate": 7.92271299634223e-06,
+      "loss": 0.6316,
+      "step": 19792
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.8653016090393066,
+      "learning_rate": 7.922499069783427e-06,
+      "loss": 0.8744,
+      "step": 19793
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.500368118286133,
+      "learning_rate": 7.92228513509825e-06,
+      "loss": 0.6088,
+      "step": 19794
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.0215444564819336,
+      "learning_rate": 7.92207119228729e-06,
+      "loss": 0.5819,
+      "step": 19795
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.5960817337036133,
+      "learning_rate": 7.921857241351146e-06,
+      "loss": 0.7833,
+      "step": 19796
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.2363033294677734,
+      "learning_rate": 7.921643282290411e-06,
+      "loss": 0.6335,
+      "step": 19797
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.7818193435668945,
+      "learning_rate": 7.921429315105682e-06,
+      "loss": 0.7798,
+      "step": 19798
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.6101982593536377,
+      "learning_rate": 7.92121533979755e-06,
+      "loss": 0.7037,
+      "step": 19799
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.6872036457061768,
+      "learning_rate": 7.921001356366616e-06,
+      "loss": 0.6276,
+      "step": 19800
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 13.83818244934082,
+      "learning_rate": 7.920787364813468e-06,
+      "loss": 0.6702,
+      "step": 19801
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.755146026611328,
+      "learning_rate": 7.920573365138707e-06,
+      "loss": 0.761,
+      "step": 19802
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.178051233291626,
+      "learning_rate": 7.920359357342923e-06,
+      "loss": 0.6441,
+      "step": 19803
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.34091055393219,
+      "learning_rate": 7.920145341426715e-06,
+      "loss": 0.3682,
+      "step": 19804
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.9141446352005005,
+      "learning_rate": 7.919931317390677e-06,
+      "loss": 0.4726,
+      "step": 19805
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.903489589691162,
+      "learning_rate": 7.919717285235404e-06,
+      "loss": 0.772,
+      "step": 19806
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.7099820375442505,
+      "learning_rate": 7.91950324496149e-06,
+      "loss": 0.8513,
+      "step": 19807
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.350666046142578,
+      "learning_rate": 7.919289196569532e-06,
+      "loss": 0.7481,
+      "step": 19808
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.746776580810547,
+      "learning_rate": 7.919075140060123e-06,
+      "loss": 0.6142,
+      "step": 19809
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.2507822513580322,
+      "learning_rate": 7.91886107543386e-06,
+      "loss": 0.434,
+      "step": 19810
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.4372613430023193,
+      "learning_rate": 7.91864700269134e-06,
+      "loss": 0.6236,
+      "step": 19811
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.9269354343414307,
+      "learning_rate": 7.918432921833154e-06,
+      "loss": 0.7344,
+      "step": 19812
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 4.5221781730651855,
+      "learning_rate": 7.9182188328599e-06,
+      "loss": 0.8774,
+      "step": 19813
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.3118278980255127,
+      "learning_rate": 7.918004735772171e-06,
+      "loss": 0.7874,
+      "step": 19814
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.5261220932006836,
+      "learning_rate": 7.917790630570566e-06,
+      "loss": 0.7577,
+      "step": 19815
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.2903945446014404,
+      "learning_rate": 7.917576517255678e-06,
+      "loss": 0.617,
+      "step": 19816
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.0002007484436035,
+      "learning_rate": 7.917362395828102e-06,
+      "loss": 0.6567,
+      "step": 19817
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.7868659496307373,
+      "learning_rate": 7.917148266288435e-06,
+      "loss": 0.782,
+      "step": 19818
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.1065948009490967,
+      "learning_rate": 7.91693412863727e-06,
+      "loss": 0.8806,
+      "step": 19819
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.073247194290161,
+      "learning_rate": 7.916719982875204e-06,
+      "loss": 0.5056,
+      "step": 19820
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.8266961574554443,
+      "learning_rate": 7.916505829002834e-06,
+      "loss": 0.8295,
+      "step": 19821
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 4.622527122497559,
+      "learning_rate": 7.916291667020753e-06,
+      "loss": 0.5913,
+      "step": 19822
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.224341869354248,
+      "learning_rate": 7.916077496929557e-06,
+      "loss": 0.5214,
+      "step": 19823
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.711437702178955,
+      "learning_rate": 7.915863318729842e-06,
+      "loss": 0.5715,
+      "step": 19824
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.4516041278839111,
+      "learning_rate": 7.915649132422204e-06,
+      "loss": 0.518,
+      "step": 19825
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.6092889308929443,
+      "learning_rate": 7.915434938007235e-06,
+      "loss": 0.6904,
+      "step": 19826
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.34679913520813,
+      "learning_rate": 7.915220735485538e-06,
+      "loss": 0.5425,
+      "step": 19827
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.3321523666381836,
+      "learning_rate": 7.915006524857702e-06,
+      "loss": 0.5153,
+      "step": 19828
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.5659024715423584,
+      "learning_rate": 7.914792306124326e-06,
+      "loss": 0.6114,
+      "step": 19829
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.3963136672973633,
+      "learning_rate": 7.914578079286003e-06,
+      "loss": 0.6012,
+      "step": 19830
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.8658230304718018,
+      "learning_rate": 7.914363844343332e-06,
+      "loss": 0.6344,
+      "step": 19831
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.0155491828918457,
+      "learning_rate": 7.914149601296905e-06,
+      "loss": 0.6283,
+      "step": 19832
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.9476585388183594,
+      "learning_rate": 7.91393535014732e-06,
+      "loss": 0.632,
+      "step": 19833
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.1533052921295166,
+      "learning_rate": 7.913721090895173e-06,
+      "loss": 0.6781,
+      "step": 19834
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.0572187900543213,
+      "learning_rate": 7.913506823541057e-06,
+      "loss": 0.7627,
+      "step": 19835
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.081787109375,
+      "learning_rate": 7.913292548085573e-06,
+      "loss": 0.7531,
+      "step": 19836
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.5003180503845215,
+      "learning_rate": 7.913078264529313e-06,
+      "loss": 0.6951,
+      "step": 19837
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.2855277061462402,
+      "learning_rate": 7.912863972872874e-06,
+      "loss": 0.8137,
+      "step": 19838
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.3622188568115234,
+      "learning_rate": 7.91264967311685e-06,
+      "loss": 0.6675,
+      "step": 19839
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.7331974506378174,
+      "learning_rate": 7.912435365261838e-06,
+      "loss": 0.4512,
+      "step": 19840
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.973356246948242,
+      "learning_rate": 7.912221049308435e-06,
+      "loss": 0.7805,
+      "step": 19841
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.714027166366577,
+      "learning_rate": 7.912006725257238e-06,
+      "loss": 0.6897,
+      "step": 19842
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.7987841367721558,
+      "learning_rate": 7.911792393108839e-06,
+      "loss": 0.7352,
+      "step": 19843
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.7689088582992554,
+      "learning_rate": 7.911578052863835e-06,
+      "loss": 0.8485,
+      "step": 19844
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.2083075046539307,
+      "learning_rate": 7.911363704522826e-06,
+      "loss": 0.8296,
+      "step": 19845
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.8214964866638184,
+      "learning_rate": 7.911149348086403e-06,
+      "loss": 0.5612,
+      "step": 19846
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.1397039890289307,
+      "learning_rate": 7.910934983555165e-06,
+      "loss": 0.7379,
+      "step": 19847
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.806692600250244,
+      "learning_rate": 7.910720610929708e-06,
+      "loss": 0.4731,
+      "step": 19848
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.502354621887207,
+      "learning_rate": 7.910506230210627e-06,
+      "loss": 0.846,
+      "step": 19849
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.1256489753723145,
+      "learning_rate": 7.910291841398518e-06,
+      "loss": 0.4877,
+      "step": 19850
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.056824207305908,
+      "learning_rate": 7.91007744449398e-06,
+      "loss": 0.4198,
+      "step": 19851
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.364809274673462,
+      "learning_rate": 7.909863039497604e-06,
+      "loss": 0.6667,
+      "step": 19852
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.2632761001586914,
+      "learning_rate": 7.90964862640999e-06,
+      "loss": 0.8489,
+      "step": 19853
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.160827159881592,
+      "learning_rate": 7.909434205231732e-06,
+      "loss": 0.6689,
+      "step": 19854
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.699194073677063,
+      "learning_rate": 7.909219775963428e-06,
+      "loss": 0.5854,
+      "step": 19855
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.8116648197174072,
+      "learning_rate": 7.909005338605673e-06,
+      "loss": 0.7008,
+      "step": 19856
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.6355355978012085,
+      "learning_rate": 7.908790893159065e-06,
+      "loss": 0.5391,
+      "step": 19857
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 4.526346206665039,
+      "learning_rate": 7.908576439624201e-06,
+      "loss": 0.8214,
+      "step": 19858
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.1549508571624756,
+      "learning_rate": 7.908361978001672e-06,
+      "loss": 0.8454,
+      "step": 19859
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.5789844989776611,
+      "learning_rate": 7.90814750829208e-06,
+      "loss": 0.5421,
+      "step": 19860
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.023115396499634,
+      "learning_rate": 7.907933030496018e-06,
+      "loss": 0.8281,
+      "step": 19861
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.305917263031006,
+      "learning_rate": 7.907718544614085e-06,
+      "loss": 0.6284,
+      "step": 19862
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.8745031356811523,
+      "learning_rate": 7.907504050646874e-06,
+      "loss": 0.3953,
+      "step": 19863
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.9759106636047363,
+      "learning_rate": 7.907289548594982e-06,
+      "loss": 0.4297,
+      "step": 19864
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.732600450515747,
+      "learning_rate": 7.907075038459012e-06,
+      "loss": 0.7092,
+      "step": 19865
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.554319381713867,
+      "learning_rate": 7.906860520239551e-06,
+      "loss": 0.634,
+      "step": 19866
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.85168194770813,
+      "learning_rate": 7.906645993937203e-06,
+      "loss": 0.7059,
+      "step": 19867
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.635467290878296,
+      "learning_rate": 7.90643145955256e-06,
+      "loss": 0.5621,
+      "step": 19868
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.1837100982666016,
+      "learning_rate": 7.906216917086219e-06,
+      "loss": 0.5794,
+      "step": 19869
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.3831489086151123,
+      "learning_rate": 7.906002366538777e-06,
+      "loss": 0.6971,
+      "step": 19870
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.227356195449829,
+      "learning_rate": 7.905787807910832e-06,
+      "loss": 0.9047,
+      "step": 19871
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.596153497695923,
+      "learning_rate": 7.905573241202981e-06,
+      "loss": 0.4919,
+      "step": 19872
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.446500539779663,
+      "learning_rate": 7.905358666415818e-06,
+      "loss": 0.684,
+      "step": 19873
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 5.8146209716796875,
+      "learning_rate": 7.905144083549941e-06,
+      "loss": 0.7361,
+      "step": 19874
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.01692795753479,
+      "learning_rate": 7.904929492605946e-06,
+      "loss": 0.5483,
+      "step": 19875
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 5.099205017089844,
+      "learning_rate": 7.90471489358443e-06,
+      "loss": 1.0548,
+      "step": 19876
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 4.107800006866455,
+      "learning_rate": 7.904500286485993e-06,
+      "loss": 1.0806,
+      "step": 19877
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.9725873470306396,
+      "learning_rate": 7.904285671311225e-06,
+      "loss": 0.8195,
+      "step": 19878
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 7.354637145996094,
+      "learning_rate": 7.904071048060729e-06,
+      "loss": 0.6865,
+      "step": 19879
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.7170445919036865,
+      "learning_rate": 7.903856416735098e-06,
+      "loss": 0.5428,
+      "step": 19880
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.8581087589263916,
+      "learning_rate": 7.903641777334931e-06,
+      "loss": 0.9946,
+      "step": 19881
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.263976573944092,
+      "learning_rate": 7.903427129860824e-06,
+      "loss": 0.7872,
+      "step": 19882
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.911960482597351,
+      "learning_rate": 7.903212474313374e-06,
+      "loss": 0.5165,
+      "step": 19883
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.20914888381958,
+      "learning_rate": 7.902997810693174e-06,
+      "loss": 0.6312,
+      "step": 19884
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.0447380542755127,
+      "learning_rate": 7.902783139000829e-06,
+      "loss": 0.7526,
+      "step": 19885
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.5728516578674316,
+      "learning_rate": 7.90256845923693e-06,
+      "loss": 0.5644,
+      "step": 19886
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.6044867038726807,
+      "learning_rate": 7.902353771402076e-06,
+      "loss": 0.5193,
+      "step": 19887
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.9197309017181396,
+      "learning_rate": 7.902139075496862e-06,
+      "loss": 0.825,
+      "step": 19888
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.311413288116455,
+      "learning_rate": 7.901924371521887e-06,
+      "loss": 0.8137,
+      "step": 19889
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.111480951309204,
+      "learning_rate": 7.901709659477749e-06,
+      "loss": 0.5928,
+      "step": 19890
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.7898826599121094,
+      "learning_rate": 7.901494939365042e-06,
+      "loss": 0.4492,
+      "step": 19891
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.2670509815216064,
+      "learning_rate": 7.901280211184364e-06,
+      "loss": 0.7358,
+      "step": 19892
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.9501317739486694,
+      "learning_rate": 7.901065474936314e-06,
+      "loss": 0.4971,
+      "step": 19893
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.236731767654419,
+      "learning_rate": 7.900850730621488e-06,
+      "loss": 0.7345,
+      "step": 19894
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 5.086175441741943,
+      "learning_rate": 7.900635978240481e-06,
+      "loss": 0.9893,
+      "step": 19895
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 6.869816303253174,
+      "learning_rate": 7.900421217793893e-06,
+      "loss": 1.1094,
+      "step": 19896
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.487713098526001,
+      "learning_rate": 7.900206449282319e-06,
+      "loss": 0.5367,
+      "step": 19897
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.4545328617095947,
+      "learning_rate": 7.899991672706358e-06,
+      "loss": 0.3547,
+      "step": 19898
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.313164472579956,
+      "learning_rate": 7.899776888066606e-06,
+      "loss": 0.7902,
+      "step": 19899
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.6634750366210938,
+      "learning_rate": 7.899562095363663e-06,
+      "loss": 0.593,
+      "step": 19900
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.5297431945800781,
+      "learning_rate": 7.899347294598121e-06,
+      "loss": 0.5681,
+      "step": 19901
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.3217172622680664,
+      "learning_rate": 7.899132485770583e-06,
+      "loss": 0.761,
+      "step": 19902
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 4.5306196212768555,
+      "learning_rate": 7.89891766888164e-06,
+      "loss": 1.1347,
+      "step": 19903
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.698089599609375,
+      "learning_rate": 7.898702843931895e-06,
+      "loss": 0.6866,
+      "step": 19904
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.5285282135009766,
+      "learning_rate": 7.898488010921944e-06,
+      "loss": 0.8511,
+      "step": 19905
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.4295120239257812,
+      "learning_rate": 7.898273169852383e-06,
+      "loss": 0.7662,
+      "step": 19906
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.364513635635376,
+      "learning_rate": 7.898058320723811e-06,
+      "loss": 0.6161,
+      "step": 19907
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.6749449968338013,
+      "learning_rate": 7.897843463536822e-06,
+      "loss": 0.7054,
+      "step": 19908
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.473571300506592,
+      "learning_rate": 7.897628598292019e-06,
+      "loss": 0.5754,
+      "step": 19909
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.0973825454711914,
+      "learning_rate": 7.897413724989995e-06,
+      "loss": 0.6964,
+      "step": 19910
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.912353754043579,
+      "learning_rate": 7.897198843631348e-06,
+      "loss": 0.857,
+      "step": 19911
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.3157958984375,
+      "learning_rate": 7.896983954216677e-06,
+      "loss": 0.8793,
+      "step": 19912
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.5406413078308105,
+      "learning_rate": 7.89676905674658e-06,
+      "loss": 0.6055,
+      "step": 19913
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.0355441570281982,
+      "learning_rate": 7.896554151221652e-06,
+      "loss": 0.7499,
+      "step": 19914
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.2329912185668945,
+      "learning_rate": 7.89633923764249e-06,
+      "loss": 0.7337,
+      "step": 19915
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.3648929595947266,
+      "learning_rate": 7.896124316009696e-06,
+      "loss": 0.5065,
+      "step": 19916
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.412123680114746,
+      "learning_rate": 7.895909386323865e-06,
+      "loss": 0.8176,
+      "step": 19917
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.1004185676574707,
+      "learning_rate": 7.895694448585596e-06,
+      "loss": 0.6272,
+      "step": 19918
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.117461681365967,
+      "learning_rate": 7.895479502795485e-06,
+      "loss": 0.5867,
+      "step": 19919
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.6523358821868896,
+      "learning_rate": 7.895264548954129e-06,
+      "loss": 0.7906,
+      "step": 19920
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.7012946605682373,
+      "learning_rate": 7.895049587062128e-06,
+      "loss": 0.9054,
+      "step": 19921
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.572306513786316,
+      "learning_rate": 7.894834617120078e-06,
+      "loss": 0.6385,
+      "step": 19922
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 4.192184925079346,
+      "learning_rate": 7.894619639128577e-06,
+      "loss": 0.6693,
+      "step": 19923
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.049199104309082,
+      "learning_rate": 7.894404653088223e-06,
+      "loss": 0.7005,
+      "step": 19924
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.1672098636627197,
+      "learning_rate": 7.894189658999613e-06,
+      "loss": 0.6937,
+      "step": 19925
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.475876808166504,
+      "learning_rate": 7.893974656863347e-06,
+      "loss": 0.6372,
+      "step": 19926
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.903083086013794,
+      "learning_rate": 7.893759646680023e-06,
+      "loss": 0.6561,
+      "step": 19927
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.2400310039520264,
+      "learning_rate": 7.893544628450236e-06,
+      "loss": 0.6002,
+      "step": 19928
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.2588534355163574,
+      "learning_rate": 7.893329602174586e-06,
+      "loss": 0.791,
+      "step": 19929
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 5.474737167358398,
+      "learning_rate": 7.893114567853668e-06,
+      "loss": 0.8206,
+      "step": 19930
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.25402569770813,
+      "learning_rate": 7.892899525488084e-06,
+      "loss": 0.7549,
+      "step": 19931
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.563807487487793,
+      "learning_rate": 7.892684475078431e-06,
+      "loss": 0.7821,
+      "step": 19932
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.606405258178711,
+      "learning_rate": 7.892469416625306e-06,
+      "loss": 0.5752,
+      "step": 19933
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.86464524269104,
+      "learning_rate": 7.892254350129305e-06,
+      "loss": 0.8704,
+      "step": 19934
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.9077563285827637,
+      "learning_rate": 7.892039275591031e-06,
+      "loss": 0.5701,
+      "step": 19935
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.643129587173462,
+      "learning_rate": 7.891824193011075e-06,
+      "loss": 0.8449,
+      "step": 19936
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.3797810077667236,
+      "learning_rate": 7.891609102390043e-06,
+      "loss": 0.7484,
+      "step": 19937
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.2199888229370117,
+      "learning_rate": 7.891394003728528e-06,
+      "loss": 0.6524,
+      "step": 19938
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.002997636795044,
+      "learning_rate": 7.89117889702713e-06,
+      "loss": 0.708,
+      "step": 19939
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.1461002826690674,
+      "learning_rate": 7.890963782286445e-06,
+      "loss": 0.6471,
+      "step": 19940
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.4546867609024048,
+      "learning_rate": 7.890748659507075e-06,
+      "loss": 0.475,
+      "step": 19941
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.5786585807800293,
+      "learning_rate": 7.890533528689614e-06,
+      "loss": 0.8854,
+      "step": 19942
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 4.20720911026001,
+      "learning_rate": 7.890318389834663e-06,
+      "loss": 0.7708,
+      "step": 19943
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.8896524906158447,
+      "learning_rate": 7.89010324294282e-06,
+      "loss": 0.4284,
+      "step": 19944
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.0772411823272705,
+      "learning_rate": 7.88988808801468e-06,
+      "loss": 0.8173,
+      "step": 19945
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.1159234046936035,
+      "learning_rate": 7.889672925050845e-06,
+      "loss": 0.5992,
+      "step": 19946
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.4040818214416504,
+      "learning_rate": 7.889457754051912e-06,
+      "loss": 0.8097,
+      "step": 19947
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.537069082260132,
+      "learning_rate": 7.88924257501848e-06,
+      "loss": 0.8316,
+      "step": 19948
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.370511293411255,
+      "learning_rate": 7.889027387951147e-06,
+      "loss": 0.7826,
+      "step": 19949
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.655551552772522,
+      "learning_rate": 7.88881219285051e-06,
+      "loss": 0.6569,
+      "step": 19950
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.3334128856658936,
+      "learning_rate": 7.88859698971717e-06,
+      "loss": 0.739,
+      "step": 19951
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.8089722394943237,
+      "learning_rate": 7.888381778551723e-06,
+      "loss": 0.7391,
+      "step": 19952
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.521450996398926,
+      "learning_rate": 7.888166559354768e-06,
+      "loss": 0.5338,
+      "step": 19953
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.3643133640289307,
+      "learning_rate": 7.887951332126905e-06,
+      "loss": 0.7457,
+      "step": 19954
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.7458415031433105,
+      "learning_rate": 7.887736096868728e-06,
+      "loss": 0.6904,
+      "step": 19955
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.115795612335205,
+      "learning_rate": 7.88752085358084e-06,
+      "loss": 0.5101,
+      "step": 19956
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.472414493560791,
+      "learning_rate": 7.887305602263841e-06,
+      "loss": 0.6203,
+      "step": 19957
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.0163307189941406,
+      "learning_rate": 7.887090342918324e-06,
+      "loss": 0.7216,
+      "step": 19958
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.751783609390259,
+      "learning_rate": 7.88687507554489e-06,
+      "loss": 0.5251,
+      "step": 19959
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.8114073276519775,
+      "learning_rate": 7.88665980014414e-06,
+      "loss": 0.725,
+      "step": 19960
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 4.053324222564697,
+      "learning_rate": 7.88644451671667e-06,
+      "loss": 0.6902,
+      "step": 19961
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.6748883724212646,
+      "learning_rate": 7.886229225263078e-06,
+      "loss": 0.7622,
+      "step": 19962
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.2513117790222168,
+      "learning_rate": 7.886013925783965e-06,
+      "loss": 0.4765,
+      "step": 19963
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 4.720029354095459,
+      "learning_rate": 7.885798618279927e-06,
+      "loss": 0.68,
+      "step": 19964
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.9560155868530273,
+      "learning_rate": 7.885583302751566e-06,
+      "loss": 0.76,
+      "step": 19965
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.8395791053771973,
+      "learning_rate": 7.885367979199476e-06,
+      "loss": 0.6772,
+      "step": 19966
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 3.1192684173583984,
+      "learning_rate": 7.885152647624258e-06,
+      "loss": 0.6073,
+      "step": 19967
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 6.5677170753479,
+      "learning_rate": 7.884937308026515e-06,
+      "loss": 0.5617,
+      "step": 19968
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 3.565431594848633,
+      "learning_rate": 7.88472196040684e-06,
+      "loss": 0.8118,
+      "step": 19969
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.5990426540374756,
+      "learning_rate": 7.884506604765834e-06,
+      "loss": 0.5978,
+      "step": 19970
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 4.496781826019287,
+      "learning_rate": 7.884291241104097e-06,
+      "loss": 0.7433,
+      "step": 19971
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 3.6933059692382812,
+      "learning_rate": 7.884075869422225e-06,
+      "loss": 0.9053,
+      "step": 19972
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.9317715167999268,
+      "learning_rate": 7.88386048972082e-06,
+      "loss": 0.6479,
+      "step": 19973
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.6654489040374756,
+      "learning_rate": 7.883645102000477e-06,
+      "loss": 0.5882,
+      "step": 19974
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.3539791107177734,
+      "learning_rate": 7.883429706261798e-06,
+      "loss": 0.8837,
+      "step": 19975
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.497398853302002,
+      "learning_rate": 7.883214302505382e-06,
+      "loss": 0.7509,
+      "step": 19976
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 3.0526514053344727,
+      "learning_rate": 7.882998890731826e-06,
+      "loss": 0.8856,
+      "step": 19977
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.9837929010391235,
+      "learning_rate": 7.88278347094173e-06,
+      "loss": 0.5778,
+      "step": 19978
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.022207260131836,
+      "learning_rate": 7.882568043135694e-06,
+      "loss": 0.5072,
+      "step": 19979
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.598742961883545,
+      "learning_rate": 7.882352607314317e-06,
+      "loss": 0.5927,
+      "step": 19980
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 3.505674362182617,
+      "learning_rate": 7.882137163478196e-06,
+      "loss": 0.7591,
+      "step": 19981
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.7191004753112793,
+      "learning_rate": 7.881921711627931e-06,
+      "loss": 0.742,
+      "step": 19982
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.6399850845336914,
+      "learning_rate": 7.881706251764122e-06,
+      "loss": 0.6724,
+      "step": 19983
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.0545434951782227,
+      "learning_rate": 7.881490783887366e-06,
+      "loss": 0.5175,
+      "step": 19984
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.5671541690826416,
+      "learning_rate": 7.881275307998266e-06,
+      "loss": 0.4651,
+      "step": 19985
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.6673340797424316,
+      "learning_rate": 7.881059824097417e-06,
+      "loss": 0.7197,
+      "step": 19986
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.2849764823913574,
+      "learning_rate": 7.88084433218542e-06,
+      "loss": 0.5275,
+      "step": 19987
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.9048774242401123,
+      "learning_rate": 7.880628832262873e-06,
+      "loss": 0.6962,
+      "step": 19988
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.106867790222168,
+      "learning_rate": 7.880413324330378e-06,
+      "loss": 0.7216,
+      "step": 19989
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.430228590965271,
+      "learning_rate": 7.880197808388532e-06,
+      "loss": 0.5076,
+      "step": 19990
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.8991550207138062,
+      "learning_rate": 7.879982284437936e-06,
+      "loss": 0.6479,
+      "step": 19991
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.156271457672119,
+      "learning_rate": 7.879766752479187e-06,
+      "loss": 0.6805,
+      "step": 19992
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 3.8266403675079346,
+      "learning_rate": 7.879551212512886e-06,
+      "loss": 0.7659,
+      "step": 19993
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.8848819732666016,
+      "learning_rate": 7.879335664539631e-06,
+      "loss": 0.6891,
+      "step": 19994
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 4.863030910491943,
+      "learning_rate": 7.879120108560024e-06,
+      "loss": 0.6816,
+      "step": 19995
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.5288580656051636,
+      "learning_rate": 7.878904544574659e-06,
+      "loss": 0.584,
+      "step": 19996
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.5572195053100586,
+      "learning_rate": 7.878688972584142e-06,
+      "loss": 0.7259,
+      "step": 19997
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 4.297033309936523,
+      "learning_rate": 7.878473392589068e-06,
+      "loss": 0.8756,
+      "step": 19998
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.9929126501083374,
+      "learning_rate": 7.878257804590037e-06,
+      "loss": 0.7452,
+      "step": 19999
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.384561777114868,
+      "learning_rate": 7.87804220858765e-06,
+      "loss": 0.714,
+      "step": 20000
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 3.3598616123199463,
+      "learning_rate": 7.877826604582508e-06,
+      "loss": 0.7965,
+      "step": 20001
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 3.180565357208252,
+      "learning_rate": 7.877610992575206e-06,
+      "loss": 0.5398,
+      "step": 20002
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.3528130054473877,
+      "learning_rate": 7.877395372566346e-06,
+      "loss": 0.6781,
+      "step": 20003
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.9136525392532349,
+      "learning_rate": 7.877179744556529e-06,
+      "loss": 0.4131,
+      "step": 20004
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.5150176286697388,
+      "learning_rate": 7.87696410854635e-06,
+      "loss": 0.5348,
+      "step": 20005
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 4.916192054748535,
+      "learning_rate": 7.876748464536412e-06,
+      "loss": 0.7377,
+      "step": 20006
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.174522876739502,
+      "learning_rate": 7.876532812527316e-06,
+      "loss": 0.6228,
+      "step": 20007
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.3723676204681396,
+      "learning_rate": 7.876317152519659e-06,
+      "loss": 0.7115,
+      "step": 20008
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 3.8339171409606934,
+      "learning_rate": 7.876101484514042e-06,
+      "loss": 1.0912,
+      "step": 20009
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.7578352689743042,
+      "learning_rate": 7.875885808511064e-06,
+      "loss": 0.7435,
+      "step": 20010
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.6023526191711426,
+      "learning_rate": 7.875670124511325e-06,
+      "loss": 0.6589,
+      "step": 20011
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.7095814943313599,
+      "learning_rate": 7.875454432515423e-06,
+      "loss": 0.5649,
+      "step": 20012
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.4944814443588257,
+      "learning_rate": 7.875238732523962e-06,
+      "loss": 0.7842,
+      "step": 20013
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.367832899093628,
+      "learning_rate": 7.875023024537538e-06,
+      "loss": 0.6626,
+      "step": 20014
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.288444757461548,
+      "learning_rate": 7.87480730855675e-06,
+      "loss": 0.6386,
+      "step": 20015
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.9464370012283325,
+      "learning_rate": 7.874591584582204e-06,
+      "loss": 0.6335,
+      "step": 20016
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.072744846343994,
+      "learning_rate": 7.874375852614492e-06,
+      "loss": 0.7803,
+      "step": 20017
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.7833446264266968,
+      "learning_rate": 7.874160112654219e-06,
+      "loss": 0.5944,
+      "step": 20018
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.1881256103515625,
+      "learning_rate": 7.873944364701983e-06,
+      "loss": 0.6063,
+      "step": 20019
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.046316146850586,
+      "learning_rate": 7.873728608758383e-06,
+      "loss": 0.5416,
+      "step": 20020
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.153646945953369,
+      "learning_rate": 7.873512844824024e-06,
+      "loss": 0.5741,
+      "step": 20021
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.883154273033142,
+      "learning_rate": 7.873297072899499e-06,
+      "loss": 0.5499,
+      "step": 20022
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.7941603660583496,
+      "learning_rate": 7.873081292985412e-06,
+      "loss": 0.8238,
+      "step": 20023
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.5367212295532227,
+      "learning_rate": 7.872865505082363e-06,
+      "loss": 0.376,
+      "step": 20024
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.106786012649536,
+      "learning_rate": 7.872649709190951e-06,
+      "loss": 0.1769,
+      "step": 20025
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.7957631349563599,
+      "learning_rate": 7.872433905311777e-06,
+      "loss": 0.674,
+      "step": 20026
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.1304209232330322,
+      "learning_rate": 7.872218093445439e-06,
+      "loss": 0.7984,
+      "step": 20027
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.310307264328003,
+      "learning_rate": 7.872002273592539e-06,
+      "loss": 0.5727,
+      "step": 20028
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.806547164916992,
+      "learning_rate": 7.871786445753675e-06,
+      "loss": 0.7954,
+      "step": 20029
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 3.595930337905884,
+      "learning_rate": 7.871570609929449e-06,
+      "loss": 0.5652,
+      "step": 20030
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.686647415161133,
+      "learning_rate": 7.871354766120462e-06,
+      "loss": 0.6344,
+      "step": 20031
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 4.2668890953063965,
+      "learning_rate": 7.871138914327315e-06,
+      "loss": 1.0496,
+      "step": 20032
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.6756527423858643,
+      "learning_rate": 7.870923054550603e-06,
+      "loss": 0.4752,
+      "step": 20033
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.071955919265747,
+      "learning_rate": 7.870707186790931e-06,
+      "loss": 0.5706,
+      "step": 20034
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 3.7178616523742676,
+      "learning_rate": 7.870491311048896e-06,
+      "loss": 0.6271,
+      "step": 20035
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.6911115646362305,
+      "learning_rate": 7.870275427325103e-06,
+      "loss": 0.5927,
+      "step": 20036
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.4237735271453857,
+      "learning_rate": 7.870059535620146e-06,
+      "loss": 0.7907,
+      "step": 20037
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.8541418313980103,
+      "learning_rate": 7.86984363593463e-06,
+      "loss": 0.7259,
+      "step": 20038
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.4293313026428223,
+      "learning_rate": 7.869627728269154e-06,
+      "loss": 0.55,
+      "step": 20039
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.72046959400177,
+      "learning_rate": 7.869411812624319e-06,
+      "loss": 0.6719,
+      "step": 20040
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.6881089210510254,
+      "learning_rate": 7.869195889000723e-06,
+      "loss": 0.6171,
+      "step": 20041
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.539674758911133,
+      "learning_rate": 7.86897995739897e-06,
+      "loss": 0.6705,
+      "step": 20042
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.1215429306030273,
+      "learning_rate": 7.868764017819657e-06,
+      "loss": 0.5505,
+      "step": 20043
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 5.120129585266113,
+      "learning_rate": 7.868548070263386e-06,
+      "loss": 0.5268,
+      "step": 20044
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.397735357284546,
+      "learning_rate": 7.868332114730758e-06,
+      "loss": 0.3382,
+      "step": 20045
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.1650915145874023,
+      "learning_rate": 7.868116151222373e-06,
+      "loss": 0.7751,
+      "step": 20046
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.6333929300308228,
+      "learning_rate": 7.867900179738829e-06,
+      "loss": 0.5695,
+      "step": 20047
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.6575626134872437,
+      "learning_rate": 7.867684200280732e-06,
+      "loss": 0.6451,
+      "step": 20048
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.6662938594818115,
+      "learning_rate": 7.86746821284868e-06,
+      "loss": 0.4065,
+      "step": 20049
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 3.0441832542419434,
+      "learning_rate": 7.867252217443272e-06,
+      "loss": 0.6119,
+      "step": 20050
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 3.9021196365356445,
+      "learning_rate": 7.867036214065109e-06,
+      "loss": 0.9229,
+      "step": 20051
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 3.014111280441284,
+      "learning_rate": 7.866820202714792e-06,
+      "loss": 0.5178,
+      "step": 20052
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 3.1309902667999268,
+      "learning_rate": 7.866604183392922e-06,
+      "loss": 0.7538,
+      "step": 20053
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 3.263105630874634,
+      "learning_rate": 7.8663881561001e-06,
+      "loss": 0.627,
+      "step": 20054
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.9815300703048706,
+      "learning_rate": 7.866172120836927e-06,
+      "loss": 0.7771,
+      "step": 20055
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 3.167097806930542,
+      "learning_rate": 7.865956077604002e-06,
+      "loss": 0.8825,
+      "step": 20056
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.8049113750457764,
+      "learning_rate": 7.865740026401927e-06,
+      "loss": 0.6063,
+      "step": 20057
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.7290701866149902,
+      "learning_rate": 7.865523967231304e-06,
+      "loss": 0.8452,
+      "step": 20058
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.3272910118103027,
+      "learning_rate": 7.86530790009273e-06,
+      "loss": 0.6714,
+      "step": 20059
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 5.1847243309021,
+      "learning_rate": 7.86509182498681e-06,
+      "loss": 0.7303,
+      "step": 20060
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.8363457918167114,
+      "learning_rate": 7.864875741914142e-06,
+      "loss": 0.3319,
+      "step": 20061
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 3.9525344371795654,
+      "learning_rate": 7.864659650875327e-06,
+      "loss": 0.6579,
+      "step": 20062
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.108985424041748,
+      "learning_rate": 7.864443551870969e-06,
+      "loss": 0.5722,
+      "step": 20063
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.300806999206543,
+      "learning_rate": 7.864227444901664e-06,
+      "loss": 0.7065,
+      "step": 20064
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.8190064430236816,
+      "learning_rate": 7.864011329968017e-06,
+      "loss": 0.9252,
+      "step": 20065
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.8141779899597168,
+      "learning_rate": 7.863795207070627e-06,
+      "loss": 0.5562,
+      "step": 20066
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.6869970560073853,
+      "learning_rate": 7.863579076210093e-06,
+      "loss": 0.6113,
+      "step": 20067
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.7811740636825562,
+      "learning_rate": 7.863362937387022e-06,
+      "loss": 0.915,
+      "step": 20068
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.3863712549209595,
+      "learning_rate": 7.86314679060201e-06,
+      "loss": 0.5328,
+      "step": 20069
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.0081586837768555,
+      "learning_rate": 7.862930635855659e-06,
+      "loss": 0.6341,
+      "step": 20070
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.964781403541565,
+      "learning_rate": 7.86271447314857e-06,
+      "loss": 0.5397,
+      "step": 20071
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 5.366875648498535,
+      "learning_rate": 7.862498302481345e-06,
+      "loss": 0.8295,
+      "step": 20072
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 4.286123752593994,
+      "learning_rate": 7.862282123854586e-06,
+      "loss": 0.8424,
+      "step": 20073
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.8705757856369019,
+      "learning_rate": 7.86206593726889e-06,
+      "loss": 0.7149,
+      "step": 20074
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.0449390411376953,
+      "learning_rate": 7.861849742724862e-06,
+      "loss": 0.5022,
+      "step": 20075
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 6.065934181213379,
+      "learning_rate": 7.861633540223102e-06,
+      "loss": 0.7882,
+      "step": 20076
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 3.5494303703308105,
+      "learning_rate": 7.861417329764211e-06,
+      "loss": 0.9322,
+      "step": 20077
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 4.770034313201904,
+      "learning_rate": 7.86120111134879e-06,
+      "loss": 0.3728,
+      "step": 20078
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 3.0722241401672363,
+      "learning_rate": 7.860984884977441e-06,
+      "loss": 0.6627,
+      "step": 20079
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.8726654052734375,
+      "learning_rate": 7.860768650650764e-06,
+      "loss": 0.7748,
+      "step": 20080
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 3.580815076828003,
+      "learning_rate": 7.860552408369362e-06,
+      "loss": 0.6764,
+      "step": 20081
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.318746328353882,
+      "learning_rate": 7.860336158133835e-06,
+      "loss": 0.6812,
+      "step": 20082
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.3876473903656006,
+      "learning_rate": 7.860119899944785e-06,
+      "loss": 0.5855,
+      "step": 20083
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 3.3672993183135986,
+      "learning_rate": 7.859903633802812e-06,
+      "loss": 0.5954,
+      "step": 20084
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.6886281967163086,
+      "learning_rate": 7.85968735970852e-06,
+      "loss": 0.8867,
+      "step": 20085
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.9647656679153442,
+      "learning_rate": 7.859471077662507e-06,
+      "loss": 0.4556,
+      "step": 20086
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.607886791229248,
+      "learning_rate": 7.859254787665377e-06,
+      "loss": 0.6215,
+      "step": 20087
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.4152135848999023,
+      "learning_rate": 7.859038489717729e-06,
+      "loss": 0.9097,
+      "step": 20088
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.6592040061950684,
+      "learning_rate": 7.858822183820166e-06,
+      "loss": 0.6567,
+      "step": 20089
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.8464667797088623,
+      "learning_rate": 7.858605869973293e-06,
+      "loss": 0.8713,
+      "step": 20090
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.505805015563965,
+      "learning_rate": 7.858389548177705e-06,
+      "loss": 0.6757,
+      "step": 20091
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.2587761878967285,
+      "learning_rate": 7.858173218434008e-06,
+      "loss": 0.6539,
+      "step": 20092
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.2814886569976807,
+      "learning_rate": 7.857956880742802e-06,
+      "loss": 0.7546,
+      "step": 20093
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.061072826385498,
+      "learning_rate": 7.857740535104686e-06,
+      "loss": 0.5045,
+      "step": 20094
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.6302920579910278,
+      "learning_rate": 7.857524181520266e-06,
+      "loss": 0.5893,
+      "step": 20095
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.4004864692687988,
+      "learning_rate": 7.857307819990142e-06,
+      "loss": 0.7449,
+      "step": 20096
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.4285995960235596,
+      "learning_rate": 7.857091450514915e-06,
+      "loss": 0.6322,
+      "step": 20097
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.185455560684204,
+      "learning_rate": 7.856875073095185e-06,
+      "loss": 0.4911,
+      "step": 20098
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.131399393081665,
+      "learning_rate": 7.856658687731558e-06,
+      "loss": 0.6951,
+      "step": 20099
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 4.891254425048828,
+      "learning_rate": 7.856442294424634e-06,
+      "loss": 0.6132,
+      "step": 20100
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.6132259368896484,
+      "learning_rate": 7.85622589317501e-06,
+      "loss": 0.331,
+      "step": 20101
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.886794328689575,
+      "learning_rate": 7.856009483983296e-06,
+      "loss": 0.689,
+      "step": 20102
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.230391263961792,
+      "learning_rate": 7.855793066850088e-06,
+      "loss": 0.4913,
+      "step": 20103
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 3.73793888092041,
+      "learning_rate": 7.85557664177599e-06,
+      "loss": 0.5677,
+      "step": 20104
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 4.5771613121032715,
+      "learning_rate": 7.855360208761602e-06,
+      "loss": 0.6319,
+      "step": 20105
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.4834704399108887,
+      "learning_rate": 7.855143767807528e-06,
+      "loss": 0.6924,
+      "step": 20106
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 3.2573130130767822,
+      "learning_rate": 7.854927318914366e-06,
+      "loss": 0.6853,
+      "step": 20107
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.4668383598327637,
+      "learning_rate": 7.854710862082724e-06,
+      "loss": 0.4811,
+      "step": 20108
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.949751377105713,
+      "learning_rate": 7.854494397313199e-06,
+      "loss": 0.8618,
+      "step": 20109
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 3.2647511959075928,
+      "learning_rate": 7.854277924606395e-06,
+      "loss": 0.6224,
+      "step": 20110
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.705031394958496,
+      "learning_rate": 7.854061443962912e-06,
+      "loss": 0.744,
+      "step": 20111
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.9561965465545654,
+      "learning_rate": 7.853844955383353e-06,
+      "loss": 0.6581,
+      "step": 20112
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.0871753692626953,
+      "learning_rate": 7.853628458868322e-06,
+      "loss": 0.4182,
+      "step": 20113
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.610628843307495,
+      "learning_rate": 7.853411954418419e-06,
+      "loss": 0.4728,
+      "step": 20114
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 6.981861114501953,
+      "learning_rate": 7.853195442034248e-06,
+      "loss": 0.7725,
+      "step": 20115
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.461881637573242,
+      "learning_rate": 7.852978921716405e-06,
+      "loss": 0.5605,
+      "step": 20116
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 3.2139675617218018,
+      "learning_rate": 7.8527623934655e-06,
+      "loss": 0.6473,
+      "step": 20117
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 3.587313175201416,
+      "learning_rate": 7.852545857282128e-06,
+      "loss": 0.898,
+      "step": 20118
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.9519639015197754,
+      "learning_rate": 7.852329313166897e-06,
+      "loss": 0.6178,
+      "step": 20119
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.6785566806793213,
+      "learning_rate": 7.852112761120408e-06,
+      "loss": 0.556,
+      "step": 20120
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.623288869857788,
+      "learning_rate": 7.851896201143258e-06,
+      "loss": 0.4716,
+      "step": 20121
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.8226337432861328,
+      "learning_rate": 7.851679633236055e-06,
+      "loss": 0.7733,
+      "step": 20122
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.468315362930298,
+      "learning_rate": 7.8514630573994e-06,
+      "loss": 0.5969,
+      "step": 20123
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.6928935050964355,
+      "learning_rate": 7.851246473633895e-06,
+      "loss": 0.6113,
+      "step": 20124
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.5505938529968262,
+      "learning_rate": 7.85102988194014e-06,
+      "loss": 0.581,
+      "step": 20125
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.5623574256896973,
+      "learning_rate": 7.85081328231874e-06,
+      "loss": 0.7326,
+      "step": 20126
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.9666723012924194,
+      "learning_rate": 7.850596674770295e-06,
+      "loss": 0.6253,
+      "step": 20127
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 3.949044942855835,
+      "learning_rate": 7.850380059295409e-06,
+      "loss": 0.7388,
+      "step": 20128
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 3.979527711868286,
+      "learning_rate": 7.850163435894683e-06,
+      "loss": 0.8204,
+      "step": 20129
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.3602707386016846,
+      "learning_rate": 7.84994680456872e-06,
+      "loss": 0.3636,
+      "step": 20130
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 4.291170120239258,
+      "learning_rate": 7.849730165318125e-06,
+      "loss": 0.7225,
+      "step": 20131
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.8934822082519531,
+      "learning_rate": 7.849513518143497e-06,
+      "loss": 0.9248,
+      "step": 20132
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 3.307020664215088,
+      "learning_rate": 7.84929686304544e-06,
+      "loss": 0.3962,
+      "step": 20133
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.4543495178222656,
+      "learning_rate": 7.849080200024553e-06,
+      "loss": 0.5157,
+      "step": 20134
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 3.550137519836426,
+      "learning_rate": 7.848863529081444e-06,
+      "loss": 0.7996,
+      "step": 20135
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.8477636575698853,
+      "learning_rate": 7.848646850216711e-06,
+      "loss": 0.7875,
+      "step": 20136
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.6093952655792236,
+      "learning_rate": 7.84843016343096e-06,
+      "loss": 0.6573,
+      "step": 20137
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.524315595626831,
+      "learning_rate": 7.84821346872479e-06,
+      "loss": 0.8734,
+      "step": 20138
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.859483003616333,
+      "learning_rate": 7.847996766098804e-06,
+      "loss": 0.4774,
+      "step": 20139
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.5350499153137207,
+      "learning_rate": 7.84778005555361e-06,
+      "loss": 0.7593,
+      "step": 20140
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.45322585105896,
+      "learning_rate": 7.847563337089804e-06,
+      "loss": 0.6797,
+      "step": 20141
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.8279056549072266,
+      "learning_rate": 7.847346610707991e-06,
+      "loss": 0.6531,
+      "step": 20142
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.3774569034576416,
+      "learning_rate": 7.847129876408774e-06,
+      "loss": 0.6268,
+      "step": 20143
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.6679341793060303,
+      "learning_rate": 7.846913134192755e-06,
+      "loss": 0.5585,
+      "step": 20144
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.934361219406128,
+      "learning_rate": 7.846696384060538e-06,
+      "loss": 0.2755,
+      "step": 20145
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.4536941051483154,
+      "learning_rate": 7.846479626012725e-06,
+      "loss": 0.3533,
+      "step": 20146
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.14371919631958,
+      "learning_rate": 7.846262860049917e-06,
+      "loss": 0.5826,
+      "step": 20147
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.7720372676849365,
+      "learning_rate": 7.846046086172719e-06,
+      "loss": 0.7217,
+      "step": 20148
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.140260934829712,
+      "learning_rate": 7.845829304381733e-06,
+      "loss": 0.3969,
+      "step": 20149
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.305182456970215,
+      "learning_rate": 7.84561251467756e-06,
+      "loss": 0.6727,
+      "step": 20150
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 3.019611358642578,
+      "learning_rate": 7.84539571706081e-06,
+      "loss": 0.7307,
+      "step": 20151
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 3.486387252807617,
+      "learning_rate": 7.845178911532074e-06,
+      "loss": 0.4689,
+      "step": 20152
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.0926895141601562,
+      "learning_rate": 7.844962098091963e-06,
+      "loss": 0.7175,
+      "step": 20153
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 4.108351707458496,
+      "learning_rate": 7.844745276741081e-06,
+      "loss": 0.4878,
+      "step": 20154
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 5.5609869956970215,
+      "learning_rate": 7.844528447480026e-06,
+      "loss": 0.574,
+      "step": 20155
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.8316502571105957,
+      "learning_rate": 7.844311610309402e-06,
+      "loss": 0.8937,
+      "step": 20156
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.9941962957382202,
+      "learning_rate": 7.844094765229816e-06,
+      "loss": 0.3797,
+      "step": 20157
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 3.8559224605560303,
+      "learning_rate": 7.843877912241867e-06,
+      "loss": 0.4943,
+      "step": 20158
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 3.02101993560791,
+      "learning_rate": 7.843661051346157e-06,
+      "loss": 0.7401,
+      "step": 20159
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 4.219636917114258,
+      "learning_rate": 7.84344418254329e-06,
+      "loss": 0.9172,
+      "step": 20160
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 3.9077606201171875,
+      "learning_rate": 7.843227305833873e-06,
+      "loss": 0.6843,
+      "step": 20161
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.932835578918457,
+      "learning_rate": 7.843010421218504e-06,
+      "loss": 0.5269,
+      "step": 20162
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.5922560691833496,
+      "learning_rate": 7.842793528697789e-06,
+      "loss": 0.8025,
+      "step": 20163
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.1846234798431396,
+      "learning_rate": 7.84257662827233e-06,
+      "loss": 0.6973,
+      "step": 20164
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.0063581466674805,
+      "learning_rate": 7.842359719942729e-06,
+      "loss": 0.6441,
+      "step": 20165
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.293126344680786,
+      "learning_rate": 7.842142803709593e-06,
+      "loss": 0.6222,
+      "step": 20166
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.5204296112060547,
+      "learning_rate": 7.84192587957352e-06,
+      "loss": 0.8959,
+      "step": 20167
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.4613375663757324,
+      "learning_rate": 7.841708947535117e-06,
+      "loss": 0.3887,
+      "step": 20168
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.6426137685775757,
+      "learning_rate": 7.841492007594984e-06,
+      "loss": 0.5651,
+      "step": 20169
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.245060682296753,
+      "learning_rate": 7.84127505975373e-06,
+      "loss": 0.8283,
+      "step": 20170
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 4.5739426612854,
+      "learning_rate": 7.841058104011951e-06,
+      "loss": 0.7789,
+      "step": 20171
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.529839038848877,
+      "learning_rate": 7.840841140370256e-06,
+      "loss": 0.856,
+      "step": 20172
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 3.6779067516326904,
+      "learning_rate": 7.840624168829245e-06,
+      "loss": 0.5516,
+      "step": 20173
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.8357553482055664,
+      "learning_rate": 7.840407189389523e-06,
+      "loss": 0.8441,
+      "step": 20174
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.6070748567581177,
+      "learning_rate": 7.840190202051692e-06,
+      "loss": 0.6591,
+      "step": 20175
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.1829752922058105,
+      "learning_rate": 7.839973206816356e-06,
+      "loss": 0.7945,
+      "step": 20176
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.4545748233795166,
+      "learning_rate": 7.839756203684118e-06,
+      "loss": 0.6981,
+      "step": 20177
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 3.305483818054199,
+      "learning_rate": 7.83953919265558e-06,
+      "loss": 0.6552,
+      "step": 20178
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 3.332674264907837,
+      "learning_rate": 7.83932217373135e-06,
+      "loss": 0.9107,
+      "step": 20179
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 3.302948474884033,
+      "learning_rate": 7.839105146912028e-06,
+      "loss": 0.4903,
+      "step": 20180
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 3.3325741291046143,
+      "learning_rate": 7.838888112198219e-06,
+      "loss": 0.5799,
+      "step": 20181
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.218614101409912,
+      "learning_rate": 7.838671069590525e-06,
+      "loss": 0.3461,
+      "step": 20182
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.8045793771743774,
+      "learning_rate": 7.83845401908955e-06,
+      "loss": 0.509,
+      "step": 20183
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.862070083618164,
+      "learning_rate": 7.838236960695898e-06,
+      "loss": 0.5927,
+      "step": 20184
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 4.4039506912231445,
+      "learning_rate": 7.83801989441017e-06,
+      "loss": 0.4558,
+      "step": 20185
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 3.305772304534912,
+      "learning_rate": 7.837802820232974e-06,
+      "loss": 0.5159,
+      "step": 20186
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.1055264472961426,
+      "learning_rate": 7.837585738164912e-06,
+      "loss": 0.5824,
+      "step": 20187
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.9395794868469238,
+      "learning_rate": 7.837368648206586e-06,
+      "loss": 0.8318,
+      "step": 20188
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 3.0640921592712402,
+      "learning_rate": 7.8371515503586e-06,
+      "loss": 0.5653,
+      "step": 20189
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.6268327236175537,
+      "learning_rate": 7.836934444621559e-06,
+      "loss": 0.7225,
+      "step": 20190
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 4.208700180053711,
+      "learning_rate": 7.836717330996067e-06,
+      "loss": 0.766,
+      "step": 20191
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.0553746223449707,
+      "learning_rate": 7.836500209482725e-06,
+      "loss": 0.6771,
+      "step": 20192
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 3.0674567222595215,
+      "learning_rate": 7.83628308008214e-06,
+      "loss": 0.3905,
+      "step": 20193
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.288869857788086,
+      "learning_rate": 7.836065942794913e-06,
+      "loss": 0.7368,
+      "step": 20194
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.6124160289764404,
+      "learning_rate": 7.83584879762165e-06,
+      "loss": 0.4563,
+      "step": 20195
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 4.173799514770508,
+      "learning_rate": 7.835631644562955e-06,
+      "loss": 0.7383,
+      "step": 20196
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.2168431282043457,
+      "learning_rate": 7.835414483619427e-06,
+      "loss": 0.8313,
+      "step": 20197
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 3.5628247261047363,
+      "learning_rate": 7.835197314791678e-06,
+      "loss": 0.5377,
+      "step": 20198
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 4.127208709716797,
+      "learning_rate": 7.834980138080305e-06,
+      "loss": 0.5611,
+      "step": 20199
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.3555169105529785,
+      "learning_rate": 7.834762953485913e-06,
+      "loss": 0.8216,
+      "step": 20200
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 3.340236186981201,
+      "learning_rate": 7.834545761009109e-06,
+      "loss": 0.6833,
+      "step": 20201
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.6652648448944092,
+      "learning_rate": 7.834328560650495e-06,
+      "loss": 0.4852,
+      "step": 20202
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 3.9651618003845215,
+      "learning_rate": 7.834111352410674e-06,
+      "loss": 0.6418,
+      "step": 20203
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.745912551879883,
+      "learning_rate": 7.833894136290252e-06,
+      "loss": 0.5334,
+      "step": 20204
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 3.7830941677093506,
+      "learning_rate": 7.833676912289831e-06,
+      "loss": 0.7664,
+      "step": 20205
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.6582647562026978,
+      "learning_rate": 7.833459680410016e-06,
+      "loss": 0.6288,
+      "step": 20206
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.2304368019104004,
+      "learning_rate": 7.833242440651413e-06,
+      "loss": 0.5538,
+      "step": 20207
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.8514623641967773,
+      "learning_rate": 7.833025193014624e-06,
+      "loss": 0.7818,
+      "step": 20208
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.047980308532715,
+      "learning_rate": 7.83280793750025e-06,
+      "loss": 0.4281,
+      "step": 20209
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.6441704034805298,
+      "learning_rate": 7.8325906741089e-06,
+      "loss": 0.444,
+      "step": 20210
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.5970041751861572,
+      "learning_rate": 7.832373402841178e-06,
+      "loss": 0.7773,
+      "step": 20211
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.389167070388794,
+      "learning_rate": 7.832156123697685e-06,
+      "loss": 0.8573,
+      "step": 20212
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.7369754314422607,
+      "learning_rate": 7.831938836679028e-06,
+      "loss": 0.4681,
+      "step": 20213
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.109215259552002,
+      "learning_rate": 7.831721541785808e-06,
+      "loss": 0.6439,
+      "step": 20214
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 3.720792770385742,
+      "learning_rate": 7.831504239018632e-06,
+      "loss": 0.5601,
+      "step": 20215
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.2965176105499268,
+      "learning_rate": 7.831286928378104e-06,
+      "loss": 0.5441,
+      "step": 20216
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.14414644241333,
+      "learning_rate": 7.831069609864827e-06,
+      "loss": 0.6067,
+      "step": 20217
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.6438170671463013,
+      "learning_rate": 7.830852283479408e-06,
+      "loss": 0.6838,
+      "step": 20218
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.4222609996795654,
+      "learning_rate": 7.830634949222447e-06,
+      "loss": 0.6776,
+      "step": 20219
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.8137269020080566,
+      "learning_rate": 7.830417607094551e-06,
+      "loss": 0.8159,
+      "step": 20220
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.527482748031616,
+      "learning_rate": 7.830200257096324e-06,
+      "loss": 0.5964,
+      "step": 20221
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 3.341334819793701,
+      "learning_rate": 7.82998289922837e-06,
+      "loss": 0.7263,
+      "step": 20222
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.719111919403076,
+      "learning_rate": 7.829765533491295e-06,
+      "loss": 0.6093,
+      "step": 20223
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 8.902339935302734,
+      "learning_rate": 7.8295481598857e-06,
+      "loss": 0.7271,
+      "step": 20224
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 3.0199034214019775,
+      "learning_rate": 7.829330778412194e-06,
+      "loss": 0.4394,
+      "step": 20225
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.4458757638931274,
+      "learning_rate": 7.829113389071377e-06,
+      "loss": 0.5805,
+      "step": 20226
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 3.144298553466797,
+      "learning_rate": 7.828895991863857e-06,
+      "loss": 0.7647,
+      "step": 20227
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 3.2150747776031494,
+      "learning_rate": 7.828678586790238e-06,
+      "loss": 0.529,
+      "step": 20228
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 3.0004115104675293,
+      "learning_rate": 7.82846117385112e-06,
+      "loss": 0.5876,
+      "step": 20229
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.0597410202026367,
+      "learning_rate": 7.828243753047113e-06,
+      "loss": 0.7541,
+      "step": 20230
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.608407735824585,
+      "learning_rate": 7.828026324378819e-06,
+      "loss": 0.6974,
+      "step": 20231
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.0858089923858643,
+      "learning_rate": 7.827808887846844e-06,
+      "loss": 0.3888,
+      "step": 20232
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.899632215499878,
+      "learning_rate": 7.827591443451792e-06,
+      "loss": 0.5742,
+      "step": 20233
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 5.119177341461182,
+      "learning_rate": 7.827373991194267e-06,
+      "loss": 0.8236,
+      "step": 20234
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 3.8906233310699463,
+      "learning_rate": 7.827156531074874e-06,
+      "loss": 0.821,
+      "step": 20235
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.31821346282959,
+      "learning_rate": 7.826939063094217e-06,
+      "loss": 0.6721,
+      "step": 20236
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 3.537370443344116,
+      "learning_rate": 7.826721587252902e-06,
+      "loss": 0.7344,
+      "step": 20237
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 3.657174825668335,
+      "learning_rate": 7.826504103551532e-06,
+      "loss": 0.6126,
+      "step": 20238
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.4002147912979126,
+      "learning_rate": 7.826286611990717e-06,
+      "loss": 0.6061,
+      "step": 20239
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.7841837406158447,
+      "learning_rate": 7.826069112571053e-06,
+      "loss": 0.7087,
+      "step": 20240
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.4778860807418823,
+      "learning_rate": 7.825851605293151e-06,
+      "loss": 0.6799,
+      "step": 20241
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.818920850753784,
+      "learning_rate": 7.825634090157616e-06,
+      "loss": 0.8752,
+      "step": 20242
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.8813937902450562,
+      "learning_rate": 7.825416567165049e-06,
+      "loss": 0.5705,
+      "step": 20243
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.670764684677124,
+      "learning_rate": 7.825199036316058e-06,
+      "loss": 0.5055,
+      "step": 20244
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 3.359802007675171,
+      "learning_rate": 7.824981497611247e-06,
+      "loss": 0.8947,
+      "step": 20245
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 3.923161506652832,
+      "learning_rate": 7.82476395105122e-06,
+      "loss": 0.7577,
+      "step": 20246
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.4037909507751465,
+      "learning_rate": 7.824546396636583e-06,
+      "loss": 0.8047,
+      "step": 20247
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.5265073776245117,
+      "learning_rate": 7.82432883436794e-06,
+      "loss": 0.5591,
+      "step": 20248
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.8290693759918213,
+      "learning_rate": 7.824111264245897e-06,
+      "loss": 0.4089,
+      "step": 20249
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.010608673095703,
+      "learning_rate": 7.823893686271058e-06,
+      "loss": 0.4158,
+      "step": 20250
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.6084847450256348,
+      "learning_rate": 7.823676100444029e-06,
+      "loss": 0.3968,
+      "step": 20251
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 3.6521074771881104,
+      "learning_rate": 7.823458506765415e-06,
+      "loss": 0.8766,
+      "step": 20252
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 4.140267372131348,
+      "learning_rate": 7.823240905235821e-06,
+      "loss": 0.6092,
+      "step": 20253
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 3.4666004180908203,
+      "learning_rate": 7.82302329585585e-06,
+      "loss": 0.7612,
+      "step": 20254
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.147115707397461,
+      "learning_rate": 7.82280567862611e-06,
+      "loss": 0.7106,
+      "step": 20255
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.518746256828308,
+      "learning_rate": 7.822588053547205e-06,
+      "loss": 0.5499,
+      "step": 20256
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.54885196685791,
+      "learning_rate": 7.822370420619739e-06,
+      "loss": 0.657,
+      "step": 20257
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.316230058670044,
+      "learning_rate": 7.822152779844318e-06,
+      "loss": 0.7203,
+      "step": 20258
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.8605924844741821,
+      "learning_rate": 7.821935131221547e-06,
+      "loss": 0.8297,
+      "step": 20259
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.641998291015625,
+      "learning_rate": 7.821717474752033e-06,
+      "loss": 0.7148,
+      "step": 20260
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 3.415175676345825,
+      "learning_rate": 7.821499810436379e-06,
+      "loss": 0.4243,
+      "step": 20261
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.9666780233383179,
+      "learning_rate": 7.821282138275191e-06,
+      "loss": 0.7697,
+      "step": 20262
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.9289066791534424,
+      "learning_rate": 7.821064458269074e-06,
+      "loss": 0.5654,
+      "step": 20263
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.2127885818481445,
+      "learning_rate": 7.820846770418635e-06,
+      "loss": 0.6767,
+      "step": 20264
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.9741793870925903,
+      "learning_rate": 7.820629074724476e-06,
+      "loss": 0.6103,
+      "step": 20265
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 3.9586782455444336,
+      "learning_rate": 7.820411371187206e-06,
+      "loss": 0.5066,
+      "step": 20266
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 3.353250026702881,
+      "learning_rate": 7.820193659807427e-06,
+      "loss": 0.7049,
+      "step": 20267
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.402822256088257,
+      "learning_rate": 7.819975940585747e-06,
+      "loss": 0.7434,
+      "step": 20268
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.9687135219573975,
+      "learning_rate": 7.819758213522768e-06,
+      "loss": 0.6154,
+      "step": 20269
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.316380023956299,
+      "learning_rate": 7.8195404786191e-06,
+      "loss": 0.5612,
+      "step": 20270
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 3.361978769302368,
+      "learning_rate": 7.819322735875345e-06,
+      "loss": 0.932,
+      "step": 20271
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.067272186279297,
+      "learning_rate": 7.81910498529211e-06,
+      "loss": 0.9,
+      "step": 20272
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.3146982192993164,
+      "learning_rate": 7.81888722687e-06,
+      "loss": 0.8067,
+      "step": 20273
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 3.897848129272461,
+      "learning_rate": 7.818669460609622e-06,
+      "loss": 0.8241,
+      "step": 20274
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 3.323915958404541,
+      "learning_rate": 7.818451686511578e-06,
+      "loss": 0.9304,
+      "step": 20275
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 5.257742881774902,
+      "learning_rate": 7.818233904576477e-06,
+      "loss": 1.0151,
+      "step": 20276
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.472109079360962,
+      "learning_rate": 7.818016114804922e-06,
+      "loss": 0.4579,
+      "step": 20277
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.265598773956299,
+      "learning_rate": 7.817798317197523e-06,
+      "loss": 0.8088,
+      "step": 20278
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.093515396118164,
+      "learning_rate": 7.817580511754879e-06,
+      "loss": 0.836,
+      "step": 20279
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.337489366531372,
+      "learning_rate": 7.8173626984776e-06,
+      "loss": 0.695,
+      "step": 20280
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 4.931076526641846,
+      "learning_rate": 7.817144877366292e-06,
+      "loss": 0.5135,
+      "step": 20281
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.3584842681884766,
+      "learning_rate": 7.816927048421557e-06,
+      "loss": 0.6888,
+      "step": 20282
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 3.317671775817871,
+      "learning_rate": 7.816709211644005e-06,
+      "loss": 0.706,
+      "step": 20283
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 3.8517708778381348,
+      "learning_rate": 7.816491367034239e-06,
+      "loss": 0.9012,
+      "step": 20284
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.8388426303863525,
+      "learning_rate": 7.816273514592866e-06,
+      "loss": 0.6185,
+      "step": 20285
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.124906301498413,
+      "learning_rate": 7.816055654320492e-06,
+      "loss": 0.5675,
+      "step": 20286
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.8356359004974365,
+      "learning_rate": 7.815837786217722e-06,
+      "loss": 0.5107,
+      "step": 20287
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.6304328441619873,
+      "learning_rate": 7.81561991028516e-06,
+      "loss": 0.8738,
+      "step": 20288
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.040081024169922,
+      "learning_rate": 7.815402026523416e-06,
+      "loss": 0.6761,
+      "step": 20289
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.728922963142395,
+      "learning_rate": 7.815184134933094e-06,
+      "loss": 0.4524,
+      "step": 20290
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.6202062368392944,
+      "learning_rate": 7.814966235514796e-06,
+      "loss": 0.5104,
+      "step": 20291
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.18819522857666,
+      "learning_rate": 7.814748328269134e-06,
+      "loss": 0.6016,
+      "step": 20292
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.174529790878296,
+      "learning_rate": 7.814530413196712e-06,
+      "loss": 0.6695,
+      "step": 20293
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.6549503803253174,
+      "learning_rate": 7.814312490298134e-06,
+      "loss": 0.6615,
+      "step": 20294
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.2201075553894043,
+      "learning_rate": 7.814094559574008e-06,
+      "loss": 0.7411,
+      "step": 20295
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.4442758560180664,
+      "learning_rate": 7.813876621024939e-06,
+      "loss": 0.6699,
+      "step": 20296
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.768685817718506,
+      "learning_rate": 7.813658674651532e-06,
+      "loss": 0.5971,
+      "step": 20297
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 3.018126964569092,
+      "learning_rate": 7.813440720454397e-06,
+      "loss": 0.9564,
+      "step": 20298
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.4470160007476807,
+      "learning_rate": 7.813222758434135e-06,
+      "loss": 0.7762,
+      "step": 20299
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.3260416984558105,
+      "learning_rate": 7.813004788591354e-06,
+      "loss": 0.8309,
+      "step": 20300
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.3837902545928955,
+      "learning_rate": 7.812786810926661e-06,
+      "loss": 0.8561,
+      "step": 20301
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.8862013816833496,
+      "learning_rate": 7.812568825440662e-06,
+      "loss": 0.6897,
+      "step": 20302
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 3.3876023292541504,
+      "learning_rate": 7.812350832133963e-06,
+      "loss": 0.6687,
+      "step": 20303
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.886735200881958,
+      "learning_rate": 7.81213283100717e-06,
+      "loss": 0.6418,
+      "step": 20304
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 3.141200542449951,
+      "learning_rate": 7.811914822060888e-06,
+      "loss": 1.0038,
+      "step": 20305
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 4.83597469329834,
+      "learning_rate": 7.811696805295725e-06,
+      "loss": 0.547,
+      "step": 20306
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.531794786453247,
+      "learning_rate": 7.811478780712286e-06,
+      "loss": 0.4876,
+      "step": 20307
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.5617098808288574,
+      "learning_rate": 7.811260748311177e-06,
+      "loss": 0.7068,
+      "step": 20308
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 4.451285362243652,
+      "learning_rate": 7.811042708093007e-06,
+      "loss": 0.6026,
+      "step": 20309
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 3.357724905014038,
+      "learning_rate": 7.81082466005838e-06,
+      "loss": 0.4091,
+      "step": 20310
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.7919561862945557,
+      "learning_rate": 7.810606604207903e-06,
+      "loss": 0.4701,
+      "step": 20311
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.6413780450820923,
+      "learning_rate": 7.81038854054218e-06,
+      "loss": 0.4545,
+      "step": 20312
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.552881956100464,
+      "learning_rate": 7.81017046906182e-06,
+      "loss": 0.324,
+      "step": 20313
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.609699010848999,
+      "learning_rate": 7.809952389767429e-06,
+      "loss": 0.8321,
+      "step": 20314
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.6659300327301025,
+      "learning_rate": 7.809734302659611e-06,
+      "loss": 0.5986,
+      "step": 20315
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.7014858722686768,
+      "learning_rate": 7.809516207738976e-06,
+      "loss": 0.6802,
+      "step": 20316
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 3.537653923034668,
+      "learning_rate": 7.809298105006128e-06,
+      "loss": 0.732,
+      "step": 20317
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.5821733474731445,
+      "learning_rate": 7.809079994461677e-06,
+      "loss": 0.5741,
+      "step": 20318
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 3.125537157058716,
+      "learning_rate": 7.808861876106226e-06,
+      "loss": 0.6507,
+      "step": 20319
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.858459949493408,
+      "learning_rate": 7.80864374994038e-06,
+      "loss": 0.7641,
+      "step": 20320
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.674710988998413,
+      "learning_rate": 7.808425615964751e-06,
+      "loss": 0.628,
+      "step": 20321
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 3.2086524963378906,
+      "learning_rate": 7.80820747417994e-06,
+      "loss": 0.5952,
+      "step": 20322
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.765979766845703,
+      "learning_rate": 7.807989324586556e-06,
+      "loss": 0.4562,
+      "step": 20323
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.411698818206787,
+      "learning_rate": 7.807771167185207e-06,
+      "loss": 0.3756,
+      "step": 20324
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.523118495941162,
+      "learning_rate": 7.807553001976498e-06,
+      "loss": 0.5499,
+      "step": 20325
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.5005464553833008,
+      "learning_rate": 7.807334828961035e-06,
+      "loss": 0.8494,
+      "step": 20326
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.271988868713379,
+      "learning_rate": 7.807116648139427e-06,
+      "loss": 0.7411,
+      "step": 20327
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.791964292526245,
+      "learning_rate": 7.806898459512278e-06,
+      "loss": 0.5596,
+      "step": 20328
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.928762674331665,
+      "learning_rate": 7.806680263080196e-06,
+      "loss": 0.5231,
+      "step": 20329
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 3.4469945430755615,
+      "learning_rate": 7.806462058843787e-06,
+      "loss": 0.6167,
+      "step": 20330
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.9652327299118042,
+      "learning_rate": 7.80624384680366e-06,
+      "loss": 0.9019,
+      "step": 20331
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.621561050415039,
+      "learning_rate": 7.80602562696042e-06,
+      "loss": 0.6023,
+      "step": 20332
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.469503164291382,
+      "learning_rate": 7.80580739931467e-06,
+      "loss": 0.6398,
+      "step": 20333
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.9883036613464355,
+      "learning_rate": 7.805589163867025e-06,
+      "loss": 0.5817,
+      "step": 20334
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.0234169960021973,
+      "learning_rate": 7.805370920618087e-06,
+      "loss": 0.5207,
+      "step": 20335
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.6665524244308472,
+      "learning_rate": 7.805152669568462e-06,
+      "loss": 0.7039,
+      "step": 20336
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.8397293090820312,
+      "learning_rate": 7.804934410718758e-06,
+      "loss": 0.5838,
+      "step": 20337
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.852515935897827,
+      "learning_rate": 7.804716144069585e-06,
+      "loss": 0.4456,
+      "step": 20338
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 3.2943928241729736,
+      "learning_rate": 7.804497869621544e-06,
+      "loss": 0.8334,
+      "step": 20339
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.8877370357513428,
+      "learning_rate": 7.804279587375246e-06,
+      "loss": 0.6613,
+      "step": 20340
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 3.580010175704956,
+      "learning_rate": 7.804061297331298e-06,
+      "loss": 0.6394,
+      "step": 20341
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.4142231941223145,
+      "learning_rate": 7.803842999490304e-06,
+      "loss": 0.4993,
+      "step": 20342
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 4.664874076843262,
+      "learning_rate": 7.803624693852873e-06,
+      "loss": 0.5919,
+      "step": 20343
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.766277313232422,
+      "learning_rate": 7.803406380419614e-06,
+      "loss": 0.7071,
+      "step": 20344
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.8315536975860596,
+      "learning_rate": 7.80318805919113e-06,
+      "loss": 0.8106,
+      "step": 20345
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.9879655838012695,
+      "learning_rate": 7.802969730168032e-06,
+      "loss": 0.8956,
+      "step": 20346
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.0443503856658936,
+      "learning_rate": 7.802751393350925e-06,
+      "loss": 0.5936,
+      "step": 20347
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 3.210543155670166,
+      "learning_rate": 7.802533048740414e-06,
+      "loss": 0.6941,
+      "step": 20348
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.1505539417266846,
+      "learning_rate": 7.80231469633711e-06,
+      "loss": 0.6417,
+      "step": 20349
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.390031099319458,
+      "learning_rate": 7.80209633614162e-06,
+      "loss": 0.6199,
+      "step": 20350
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 3.444068670272827,
+      "learning_rate": 7.801877968154548e-06,
+      "loss": 0.734,
+      "step": 20351
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.4984431266784668,
+      "learning_rate": 7.801659592376503e-06,
+      "loss": 0.7123,
+      "step": 20352
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.0705301761627197,
+      "learning_rate": 7.801441208808092e-06,
+      "loss": 0.6843,
+      "step": 20353
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.4313127994537354,
+      "learning_rate": 7.801222817449924e-06,
+      "loss": 0.7255,
+      "step": 20354
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.003089427947998,
+      "learning_rate": 7.801004418302604e-06,
+      "loss": 0.77,
+      "step": 20355
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 3.7664248943328857,
+      "learning_rate": 7.80078601136674e-06,
+      "loss": 0.5666,
+      "step": 20356
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.7883565425872803,
+      "learning_rate": 7.800567596642938e-06,
+      "loss": 0.7697,
+      "step": 20357
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.198314666748047,
+      "learning_rate": 7.80034917413181e-06,
+      "loss": 0.787,
+      "step": 20358
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.1166677474975586,
+      "learning_rate": 7.800130743833955e-06,
+      "loss": 0.6392,
+      "step": 20359
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 3.291278839111328,
+      "learning_rate": 7.79991230574999e-06,
+      "loss": 0.6169,
+      "step": 20360
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.7629244327545166,
+      "learning_rate": 7.799693859880514e-06,
+      "loss": 0.5818,
+      "step": 20361
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.646519184112549,
+      "learning_rate": 7.79947540622614e-06,
+      "loss": 0.6546,
+      "step": 20362
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.6100460290908813,
+      "learning_rate": 7.799256944787474e-06,
+      "loss": 0.4687,
+      "step": 20363
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 3.7564940452575684,
+      "learning_rate": 7.799038475565124e-06,
+      "loss": 0.5686,
+      "step": 20364
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.9792219400405884,
+      "learning_rate": 7.798819998559695e-06,
+      "loss": 0.5581,
+      "step": 20365
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 5.52041482925415,
+      "learning_rate": 7.798601513771794e-06,
+      "loss": 0.6757,
+      "step": 20366
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 3.166731119155884,
+      "learning_rate": 7.798383021202034e-06,
+      "loss": 0.7713,
+      "step": 20367
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.5980348587036133,
+      "learning_rate": 7.798164520851018e-06,
+      "loss": 0.4631,
+      "step": 20368
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.0571062564849854,
+      "learning_rate": 7.797946012719355e-06,
+      "loss": 0.682,
+      "step": 20369
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.1494593620300293,
+      "learning_rate": 7.797727496807652e-06,
+      "loss": 0.6886,
+      "step": 20370
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.7612241506576538,
+      "learning_rate": 7.797508973116517e-06,
+      "loss": 0.6311,
+      "step": 20371
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.032474994659424,
+      "learning_rate": 7.797290441646557e-06,
+      "loss": 0.4985,
+      "step": 20372
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.2463016510009766,
+      "learning_rate": 7.79707190239838e-06,
+      "loss": 0.5314,
+      "step": 20373
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.9932975769042969,
+      "learning_rate": 7.796853355372593e-06,
+      "loss": 0.812,
+      "step": 20374
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.3272857666015625,
+      "learning_rate": 7.796634800569806e-06,
+      "loss": 0.7451,
+      "step": 20375
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.5885367393493652,
+      "learning_rate": 7.796416237990624e-06,
+      "loss": 0.6377,
+      "step": 20376
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.294403314590454,
+      "learning_rate": 7.796197667635659e-06,
+      "loss": 0.816,
+      "step": 20377
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.8773910999298096,
+      "learning_rate": 7.795979089505515e-06,
+      "loss": 0.5096,
+      "step": 20378
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.7311737537384033,
+      "learning_rate": 7.795760503600797e-06,
+      "loss": 0.6255,
+      "step": 20379
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 3.8887321949005127,
+      "learning_rate": 7.79554190992212e-06,
+      "loss": 0.54,
+      "step": 20380
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.7431349754333496,
+      "learning_rate": 7.795323308470088e-06,
+      "loss": 0.5225,
+      "step": 20381
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 4.515003681182861,
+      "learning_rate": 7.795104699245307e-06,
+      "loss": 0.6152,
+      "step": 20382
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.530007839202881,
+      "learning_rate": 7.794886082248387e-06,
+      "loss": 0.5369,
+      "step": 20383
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.344179153442383,
+      "learning_rate": 7.794667457479937e-06,
+      "loss": 0.6808,
+      "step": 20384
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.6350884437561035,
+      "learning_rate": 7.794448824940564e-06,
+      "loss": 0.8037,
+      "step": 20385
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.4949332475662231,
+      "learning_rate": 7.794230184630874e-06,
+      "loss": 0.6851,
+      "step": 20386
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.96838641166687,
+      "learning_rate": 7.794011536551477e-06,
+      "loss": 0.8511,
+      "step": 20387
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.6689685583114624,
+      "learning_rate": 7.793792880702983e-06,
+      "loss": 0.5175,
+      "step": 20388
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.0474953651428223,
+      "learning_rate": 7.793574217085994e-06,
+      "loss": 0.65,
+      "step": 20389
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 3.5374720096588135,
+      "learning_rate": 7.793355545701125e-06,
+      "loss": 0.7348,
+      "step": 20390
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 3.8358049392700195,
+      "learning_rate": 7.793136866548978e-06,
+      "loss": 0.891,
+      "step": 20391
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.3353028297424316,
+      "learning_rate": 7.792918179630166e-06,
+      "loss": 0.6109,
+      "step": 20392
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 3.0677671432495117,
+      "learning_rate": 7.792699484945292e-06,
+      "loss": 0.7474,
+      "step": 20393
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.5454165935516357,
+      "learning_rate": 7.792480782494968e-06,
+      "loss": 0.549,
+      "step": 20394
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 3.6217455863952637,
+      "learning_rate": 7.792262072279802e-06,
+      "loss": 0.796,
+      "step": 20395
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.0075910091400146,
+      "learning_rate": 7.7920433543004e-06,
+      "loss": 0.7798,
+      "step": 20396
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 4.578409671783447,
+      "learning_rate": 7.791824628557374e-06,
+      "loss": 0.5991,
+      "step": 20397
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.661436676979065,
+      "learning_rate": 7.791605895051327e-06,
+      "loss": 0.4029,
+      "step": 20398
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.732675790786743,
+      "learning_rate": 7.79138715378287e-06,
+      "loss": 0.6702,
+      "step": 20399
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.672964096069336,
+      "learning_rate": 7.791168404752612e-06,
+      "loss": 0.5445,
+      "step": 20400
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.4926923513412476,
+      "learning_rate": 7.79094964796116e-06,
+      "loss": 0.7553,
+      "step": 20401
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.683448314666748,
+      "learning_rate": 7.790730883409121e-06,
+      "loss": 0.6654,
+      "step": 20402
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.272669792175293,
+      "learning_rate": 7.790512111097106e-06,
+      "loss": 0.6119,
+      "step": 20403
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.289865493774414,
+      "learning_rate": 7.790293331025722e-06,
+      "loss": 0.7823,
+      "step": 20404
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.241136074066162,
+      "learning_rate": 7.790074543195579e-06,
+      "loss": 0.5445,
+      "step": 20405
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.8691517114639282,
+      "learning_rate": 7.78985574760728e-06,
+      "loss": 0.6069,
+      "step": 20406
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.148667335510254,
+      "learning_rate": 7.789636944261441e-06,
+      "loss": 0.6465,
+      "step": 20407
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 3.6026058197021484,
+      "learning_rate": 7.789418133158664e-06,
+      "loss": 0.4844,
+      "step": 20408
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.8664226531982422,
+      "learning_rate": 7.789199314299563e-06,
+      "loss": 0.5318,
+      "step": 20409
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.0426928997039795,
+      "learning_rate": 7.788980487684742e-06,
+      "loss": 0.4638,
+      "step": 20410
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.9634828567504883,
+      "learning_rate": 7.788761653314812e-06,
+      "loss": 0.9203,
+      "step": 20411
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 5.295475482940674,
+      "learning_rate": 7.788542811190379e-06,
+      "loss": 0.546,
+      "step": 20412
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.1564435958862305,
+      "learning_rate": 7.788323961312053e-06,
+      "loss": 0.6164,
+      "step": 20413
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.4531350135803223,
+      "learning_rate": 7.788105103680444e-06,
+      "loss": 0.54,
+      "step": 20414
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.5116175413131714,
+      "learning_rate": 7.787886238296159e-06,
+      "loss": 0.6235,
+      "step": 20415
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.418729066848755,
+      "learning_rate": 7.787667365159806e-06,
+      "loss": 0.6374,
+      "step": 20416
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.1840832233428955,
+      "learning_rate": 7.787448484271993e-06,
+      "loss": 0.7126,
+      "step": 20417
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.6154797077178955,
+      "learning_rate": 7.787229595633333e-06,
+      "loss": 0.6777,
+      "step": 20418
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.271465539932251,
+      "learning_rate": 7.78701069924443e-06,
+      "loss": 0.5628,
+      "step": 20419
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.2695913314819336,
+      "learning_rate": 7.786791795105894e-06,
+      "loss": 0.7313,
+      "step": 20420
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.1348369121551514,
+      "learning_rate": 7.786572883218335e-06,
+      "loss": 0.6068,
+      "step": 20421
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.1485421657562256,
+      "learning_rate": 7.78635396358236e-06,
+      "loss": 0.719,
+      "step": 20422
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.6305723190307617,
+      "learning_rate": 7.786135036198577e-06,
+      "loss": 0.6688,
+      "step": 20423
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.5857465267181396,
+      "learning_rate": 7.785916101067598e-06,
+      "loss": 0.4585,
+      "step": 20424
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.2783883810043335,
+      "learning_rate": 7.785697158190031e-06,
+      "loss": 0.4495,
+      "step": 20425
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.6131079196929932,
+      "learning_rate": 7.785478207566481e-06,
+      "loss": 0.4994,
+      "step": 20426
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.2990593910217285,
+      "learning_rate": 7.785259249197562e-06,
+      "loss": 0.4239,
+      "step": 20427
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.0765578746795654,
+      "learning_rate": 7.78504028308388e-06,
+      "loss": 0.5632,
+      "step": 20428
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.8769850730895996,
+      "learning_rate": 7.784821309226043e-06,
+      "loss": 0.793,
+      "step": 20429
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.7627699375152588,
+      "learning_rate": 7.784602327624662e-06,
+      "loss": 0.605,
+      "step": 20430
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.8620730638504028,
+      "learning_rate": 7.784383338280344e-06,
+      "loss": 0.686,
+      "step": 20431
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.7507165670394897,
+      "learning_rate": 7.784164341193701e-06,
+      "loss": 0.5641,
+      "step": 20432
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 3.0119829177856445,
+      "learning_rate": 7.783945336365337e-06,
+      "loss": 0.7933,
+      "step": 20433
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 3.9176626205444336,
+      "learning_rate": 7.783726323795867e-06,
+      "loss": 0.6648,
+      "step": 20434
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.8305221796035767,
+      "learning_rate": 7.783507303485896e-06,
+      "loss": 0.5788,
+      "step": 20435
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.705275535583496,
+      "learning_rate": 7.783288275436034e-06,
+      "loss": 0.6604,
+      "step": 20436
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.8491407632827759,
+      "learning_rate": 7.783069239646888e-06,
+      "loss": 0.799,
+      "step": 20437
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.5395145416259766,
+      "learning_rate": 7.782850196119072e-06,
+      "loss": 0.6133,
+      "step": 20438
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.02363657951355,
+      "learning_rate": 7.78263114485319e-06,
+      "loss": 0.648,
+      "step": 20439
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.319362163543701,
+      "learning_rate": 7.782412085849853e-06,
+      "loss": 0.5552,
+      "step": 20440
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.346691370010376,
+      "learning_rate": 7.782193019109672e-06,
+      "loss": 0.5813,
+      "step": 20441
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.3957252502441406,
+      "learning_rate": 7.781973944633253e-06,
+      "loss": 0.6952,
+      "step": 20442
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 4.419788360595703,
+      "learning_rate": 7.781754862421206e-06,
+      "loss": 0.4683,
+      "step": 20443
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 3.056622266769409,
+      "learning_rate": 7.781535772474143e-06,
+      "loss": 0.6898,
+      "step": 20444
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.31807017326355,
+      "learning_rate": 7.78131667479267e-06,
+      "loss": 0.31,
+      "step": 20445
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.673333764076233,
+      "learning_rate": 7.781097569377396e-06,
+      "loss": 0.4776,
+      "step": 20446
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.629817247390747,
+      "learning_rate": 7.780878456228934e-06,
+      "loss": 0.4485,
+      "step": 20447
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.166518449783325,
+      "learning_rate": 7.780659335347889e-06,
+      "loss": 0.6532,
+      "step": 20448
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.7822529077529907,
+      "learning_rate": 7.780440206734871e-06,
+      "loss": 0.543,
+      "step": 20449
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 3.6255438327789307,
+      "learning_rate": 7.780221070390494e-06,
+      "loss": 1.0497,
+      "step": 20450
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.4296467304229736,
+      "learning_rate": 7.780001926315361e-06,
+      "loss": 0.5563,
+      "step": 20451
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.8450801372528076,
+      "learning_rate": 7.779782774510085e-06,
+      "loss": 0.7145,
+      "step": 20452
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.7036702632904053,
+      "learning_rate": 7.779563614975274e-06,
+      "loss": 0.6745,
+      "step": 20453
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.9394325017929077,
+      "learning_rate": 7.779344447711537e-06,
+      "loss": 0.5714,
+      "step": 20454
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.3034729957580566,
+      "learning_rate": 7.779125272719486e-06,
+      "loss": 0.8416,
+      "step": 20455
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.787050247192383,
+      "learning_rate": 7.778906089999726e-06,
+      "loss": 0.725,
+      "step": 20456
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 3.161644220352173,
+      "learning_rate": 7.778686899552872e-06,
+      "loss": 0.6224,
+      "step": 20457
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 3.2094626426696777,
+      "learning_rate": 7.77846770137953e-06,
+      "loss": 0.5973,
+      "step": 20458
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.2784523963928223,
+      "learning_rate": 7.778248495480309e-06,
+      "loss": 0.7223,
+      "step": 20459
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 3.785288095474243,
+      "learning_rate": 7.77802928185582e-06,
+      "loss": 0.6499,
+      "step": 20460
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 3.3508424758911133,
+      "learning_rate": 7.777810060506675e-06,
+      "loss": 0.7402,
+      "step": 20461
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.631513833999634,
+      "learning_rate": 7.777590831433478e-06,
+      "loss": 1.087,
+      "step": 20462
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.718475580215454,
+      "learning_rate": 7.77737159463684e-06,
+      "loss": 1.0043,
+      "step": 20463
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.337409734725952,
+      "learning_rate": 7.777152350117373e-06,
+      "loss": 0.5477,
+      "step": 20464
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.9281220436096191,
+      "learning_rate": 7.776933097875687e-06,
+      "loss": 0.8076,
+      "step": 20465
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.819088339805603,
+      "learning_rate": 7.776713837912391e-06,
+      "loss": 0.5714,
+      "step": 20466
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 4.331097602844238,
+      "learning_rate": 7.776494570228092e-06,
+      "loss": 0.7591,
+      "step": 20467
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.9750781059265137,
+      "learning_rate": 7.776275294823404e-06,
+      "loss": 0.8178,
+      "step": 20468
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.2301182746887207,
+      "learning_rate": 7.77605601169893e-06,
+      "loss": 0.7381,
+      "step": 20469
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.9368195533752441,
+      "learning_rate": 7.77583672085529e-06,
+      "loss": 0.6102,
+      "step": 20470
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.976237177848816,
+      "learning_rate": 7.775617422293084e-06,
+      "loss": 0.4689,
+      "step": 20471
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.0052530765533447,
+      "learning_rate": 7.775398116012927e-06,
+      "loss": 0.4473,
+      "step": 20472
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.8720699548721313,
+      "learning_rate": 7.77517880201543e-06,
+      "loss": 0.6219,
+      "step": 20473
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.425503969192505,
+      "learning_rate": 7.774959480301195e-06,
+      "loss": 0.5274,
+      "step": 20474
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.291544198989868,
+      "learning_rate": 7.77474015087084e-06,
+      "loss": 0.8319,
+      "step": 20475
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.238524913787842,
+      "learning_rate": 7.774520813724972e-06,
+      "loss": 0.7287,
+      "step": 20476
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.5585837364196777,
+      "learning_rate": 7.7743014688642e-06,
+      "loss": 0.7653,
+      "step": 20477
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 3.1511831283569336,
+      "learning_rate": 7.774082116289136e-06,
+      "loss": 0.534,
+      "step": 20478
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 3.038339138031006,
+      "learning_rate": 7.77386275600039e-06,
+      "loss": 0.6193,
+      "step": 20479
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.3857226371765137,
+      "learning_rate": 7.77364338799857e-06,
+      "loss": 0.8555,
+      "step": 20480
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.498441457748413,
+      "learning_rate": 7.773424012284285e-06,
+      "loss": 0.6852,
+      "step": 20481
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.7037595510482788,
+      "learning_rate": 7.77320462885815e-06,
+      "loss": 0.8648,
+      "step": 20482
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 3.0854132175445557,
+      "learning_rate": 7.77298523772077e-06,
+      "loss": 0.7621,
+      "step": 20483
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.587149143218994,
+      "learning_rate": 7.772765838872756e-06,
+      "loss": 0.8017,
+      "step": 20484
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.974086880683899,
+      "learning_rate": 7.772546432314718e-06,
+      "loss": 0.7689,
+      "step": 20485
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.6598273515701294,
+      "learning_rate": 7.77232701804727e-06,
+      "loss": 0.5243,
+      "step": 20486
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.5554002523422241,
+      "learning_rate": 7.772107596071017e-06,
+      "loss": 0.4447,
+      "step": 20487
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.3607614040374756,
+      "learning_rate": 7.771888166386571e-06,
+      "loss": 0.619,
+      "step": 20488
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.873509168624878,
+      "learning_rate": 7.771668728994543e-06,
+      "loss": 0.5871,
+      "step": 20489
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.8223017454147339,
+      "learning_rate": 7.771449283895544e-06,
+      "loss": 0.792,
+      "step": 20490
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 3.723175048828125,
+      "learning_rate": 7.771229831090182e-06,
+      "loss": 0.6504,
+      "step": 20491
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.833522081375122,
+      "learning_rate": 7.771010370579067e-06,
+      "loss": 0.8297,
+      "step": 20492
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 3.5879008769989014,
+      "learning_rate": 7.770790902362811e-06,
+      "loss": 0.7019,
+      "step": 20493
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.7944605350494385,
+      "learning_rate": 7.770571426442023e-06,
+      "loss": 0.8065,
+      "step": 20494
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.914276361465454,
+      "learning_rate": 7.770351942817314e-06,
+      "loss": 0.4722,
+      "step": 20495
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.1695891618728638,
+      "learning_rate": 7.770132451489294e-06,
+      "loss": 0.529,
+      "step": 20496
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.5734267234802246,
+      "learning_rate": 7.769912952458574e-06,
+      "loss": 0.6335,
+      "step": 20497
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.2955825328826904,
+      "learning_rate": 7.769693445725763e-06,
+      "loss": 0.5262,
+      "step": 20498
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.7037034034729004,
+      "learning_rate": 7.769473931291472e-06,
+      "loss": 0.6041,
+      "step": 20499
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.9105066061019897,
+      "learning_rate": 7.76925440915631e-06,
+      "loss": 0.735,
+      "step": 20500
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.9778358936309814,
+      "learning_rate": 7.769034879320892e-06,
+      "loss": 0.8337,
+      "step": 20501
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.2809338569641113,
+      "learning_rate": 7.768815341785824e-06,
+      "loss": 0.6006,
+      "step": 20502
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.520826578140259,
+      "learning_rate": 7.768595796551717e-06,
+      "loss": 0.543,
+      "step": 20503
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 3.5273237228393555,
+      "learning_rate": 7.768376243619184e-06,
+      "loss": 0.6343,
+      "step": 20504
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.1396501064300537,
+      "learning_rate": 7.768156682988831e-06,
+      "loss": 0.6376,
+      "step": 20505
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.7721964120864868,
+      "learning_rate": 7.767937114661273e-06,
+      "loss": 0.5049,
+      "step": 20506
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 3.0234265327453613,
+      "learning_rate": 7.767717538637118e-06,
+      "loss": 0.6185,
+      "step": 20507
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.754467010498047,
+      "learning_rate": 7.767497954916979e-06,
+      "loss": 0.3477,
+      "step": 20508
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 3.4793953895568848,
+      "learning_rate": 7.767278363501463e-06,
+      "loss": 0.5957,
+      "step": 20509
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.7476400136947632,
+      "learning_rate": 7.767058764391184e-06,
+      "loss": 0.6819,
+      "step": 20510
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 3.325458526611328,
+      "learning_rate": 7.76683915758675e-06,
+      "loss": 0.704,
+      "step": 20511
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.5165767669677734,
+      "learning_rate": 7.766619543088774e-06,
+      "loss": 0.6633,
+      "step": 20512
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.9399367570877075,
+      "learning_rate": 7.766399920897864e-06,
+      "loss": 0.6861,
+      "step": 20513
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.925755739212036,
+      "learning_rate": 7.766180291014633e-06,
+      "loss": 0.7027,
+      "step": 20514
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.9104312658309937,
+      "learning_rate": 7.76596065343969e-06,
+      "loss": 0.4582,
+      "step": 20515
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.075178861618042,
+      "learning_rate": 7.765741008173647e-06,
+      "loss": 0.6777,
+      "step": 20516
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.0246500968933105,
+      "learning_rate": 7.765521355217115e-06,
+      "loss": 0.5547,
+      "step": 20517
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.7106258869171143,
+      "learning_rate": 7.765301694570704e-06,
+      "loss": 0.5884,
+      "step": 20518
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.6251823902130127,
+      "learning_rate": 7.765082026235025e-06,
+      "loss": 0.4387,
+      "step": 20519
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.2196409702301025,
+      "learning_rate": 7.764862350210688e-06,
+      "loss": 0.7766,
+      "step": 20520
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 3.6979422569274902,
+      "learning_rate": 7.764642666498305e-06,
+      "loss": 0.7095,
+      "step": 20521
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 3.155904531478882,
+      "learning_rate": 7.764422975098486e-06,
+      "loss": 0.9333,
+      "step": 20522
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.5395071506500244,
+      "learning_rate": 7.764203276011843e-06,
+      "loss": 0.8754,
+      "step": 20523
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.1976706981658936,
+      "learning_rate": 7.763983569238984e-06,
+      "loss": 0.9213,
+      "step": 20524
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.7214782238006592,
+      "learning_rate": 7.763763854780524e-06,
+      "loss": 0.5964,
+      "step": 20525
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 4.182620048522949,
+      "learning_rate": 7.763544132637074e-06,
+      "loss": 0.6391,
+      "step": 20526
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 3.1025021076202393,
+      "learning_rate": 7.763324402809238e-06,
+      "loss": 0.7144,
+      "step": 20527
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.2030653953552246,
+      "learning_rate": 7.763104665297635e-06,
+      "loss": 0.6155,
+      "step": 20528
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.0972962379455566,
+      "learning_rate": 7.762884920102874e-06,
+      "loss": 0.5589,
+      "step": 20529
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 4.081942081451416,
+      "learning_rate": 7.762665167225563e-06,
+      "loss": 0.5504,
+      "step": 20530
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.9201359748840332,
+      "learning_rate": 7.762445406666317e-06,
+      "loss": 0.5293,
+      "step": 20531
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.6109254360198975,
+      "learning_rate": 7.762225638425744e-06,
+      "loss": 0.5676,
+      "step": 20532
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.7697489261627197,
+      "learning_rate": 7.762005862504456e-06,
+      "loss": 0.6819,
+      "step": 20533
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.9323822259902954,
+      "learning_rate": 7.761786078903065e-06,
+      "loss": 0.8457,
+      "step": 20534
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.3822736740112305,
+      "learning_rate": 7.761566287622181e-06,
+      "loss": 0.6242,
+      "step": 20535
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.9963712692260742,
+      "learning_rate": 7.761346488662415e-06,
+      "loss": 0.8284,
+      "step": 20536
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.5372071266174316,
+      "learning_rate": 7.76112668202438e-06,
+      "loss": 0.494,
+      "step": 20537
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.2339060306549072,
+      "learning_rate": 7.760906867708686e-06,
+      "loss": 0.5532,
+      "step": 20538
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.8275920152664185,
+      "learning_rate": 7.760687045715945e-06,
+      "loss": 0.5823,
+      "step": 20539
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.8351378440856934,
+      "learning_rate": 7.760467216046766e-06,
+      "loss": 0.5535,
+      "step": 20540
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.3434665203094482,
+      "learning_rate": 7.760247378701762e-06,
+      "loss": 0.6304,
+      "step": 20541
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 4.552968502044678,
+      "learning_rate": 7.760027533681544e-06,
+      "loss": 0.5372,
+      "step": 20542
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 6.046835899353027,
+      "learning_rate": 7.759807680986724e-06,
+      "loss": 0.6455,
+      "step": 20543
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.6202049255371094,
+      "learning_rate": 7.759587820617914e-06,
+      "loss": 0.6495,
+      "step": 20544
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.071044445037842,
+      "learning_rate": 7.759367952575721e-06,
+      "loss": 0.5949,
+      "step": 20545
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.0035457611083984,
+      "learning_rate": 7.75914807686076e-06,
+      "loss": 0.594,
+      "step": 20546
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.1235156059265137,
+      "learning_rate": 7.758928193473644e-06,
+      "loss": 0.3698,
+      "step": 20547
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 3.5858724117279053,
+      "learning_rate": 7.758708302414982e-06,
+      "loss": 0.7607,
+      "step": 20548
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.656423568725586,
+      "learning_rate": 7.758488403685383e-06,
+      "loss": 0.5354,
+      "step": 20549
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.0938560962677,
+      "learning_rate": 7.758268497285463e-06,
+      "loss": 0.7045,
+      "step": 20550
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.312714099884033,
+      "learning_rate": 7.75804858321583e-06,
+      "loss": 0.9068,
+      "step": 20551
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.7270570993423462,
+      "learning_rate": 7.7578286614771e-06,
+      "loss": 0.634,
+      "step": 20552
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 3.217229127883911,
+      "learning_rate": 7.75760873206988e-06,
+      "loss": 0.7275,
+      "step": 20553
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.8501460552215576,
+      "learning_rate": 7.757388794994782e-06,
+      "loss": 0.9128,
+      "step": 20554
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.8986637592315674,
+      "learning_rate": 7.75716885025242e-06,
+      "loss": 0.6037,
+      "step": 20555
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.495539903640747,
+      "learning_rate": 7.756948897843404e-06,
+      "loss": 0.4892,
+      "step": 20556
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 3.379456043243408,
+      "learning_rate": 7.756728937768346e-06,
+      "loss": 0.6818,
+      "step": 20557
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.1500253677368164,
+      "learning_rate": 7.756508970027858e-06,
+      "loss": 0.8097,
+      "step": 20558
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.033188581466675,
+      "learning_rate": 7.756288994622551e-06,
+      "loss": 0.8569,
+      "step": 20559
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 3.143285036087036,
+      "learning_rate": 7.756069011553035e-06,
+      "loss": 0.5608,
+      "step": 20560
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.5423401594161987,
+      "learning_rate": 7.755849020819926e-06,
+      "loss": 0.8069,
+      "step": 20561
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.0425002574920654,
+      "learning_rate": 7.755629022423832e-06,
+      "loss": 0.7295,
+      "step": 20562
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.1350603103637695,
+      "learning_rate": 7.755409016365366e-06,
+      "loss": 0.5496,
+      "step": 20563
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.8624277114868164,
+      "learning_rate": 7.75518900264514e-06,
+      "loss": 0.4403,
+      "step": 20564
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 3.2319118976593018,
+      "learning_rate": 7.754968981263766e-06,
+      "loss": 1.054,
+      "step": 20565
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.0268659591674805,
+      "learning_rate": 7.754748952221853e-06,
+      "loss": 0.7819,
+      "step": 20566
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.7010014057159424,
+      "learning_rate": 7.754528915520018e-06,
+      "loss": 0.706,
+      "step": 20567
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.024919033050537,
+      "learning_rate": 7.754308871158868e-06,
+      "loss": 0.2586,
+      "step": 20568
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.849483609199524,
+      "learning_rate": 7.754088819139019e-06,
+      "loss": 0.8056,
+      "step": 20569
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.020887851715088,
+      "learning_rate": 7.753868759461078e-06,
+      "loss": 0.5906,
+      "step": 20570
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.032740354537964,
+      "learning_rate": 7.753648692125662e-06,
+      "loss": 0.7087,
+      "step": 20571
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.0422122478485107,
+      "learning_rate": 7.753428617133378e-06,
+      "loss": 0.4857,
+      "step": 20572
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.440660238265991,
+      "learning_rate": 7.753208534484843e-06,
+      "loss": 0.7724,
+      "step": 20573
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.4298675060272217,
+      "learning_rate": 7.752988444180665e-06,
+      "loss": 0.4122,
+      "step": 20574
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.863409161567688,
+      "learning_rate": 7.752768346221458e-06,
+      "loss": 0.5347,
+      "step": 20575
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.1186370849609375,
+      "learning_rate": 7.752548240607832e-06,
+      "loss": 0.5214,
+      "step": 20576
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 3.615614414215088,
+      "learning_rate": 7.752328127340403e-06,
+      "loss": 1.0019,
+      "step": 20577
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.286247491836548,
+      "learning_rate": 7.752108006419778e-06,
+      "loss": 0.574,
+      "step": 20578
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 5.54217004776001,
+      "learning_rate": 7.751887877846573e-06,
+      "loss": 0.8948,
+      "step": 20579
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.118633985519409,
+      "learning_rate": 7.751667741621398e-06,
+      "loss": 0.3143,
+      "step": 20580
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 3.76619029045105,
+      "learning_rate": 7.751447597744868e-06,
+      "loss": 0.3653,
+      "step": 20581
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.82789945602417,
+      "learning_rate": 7.75122744621759e-06,
+      "loss": 0.4535,
+      "step": 20582
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.9977695941925049,
+      "learning_rate": 7.75100728704018e-06,
+      "loss": 0.593,
+      "step": 20583
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 3.5615391731262207,
+      "learning_rate": 7.75078712021325e-06,
+      "loss": 0.7634,
+      "step": 20584
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.4295847415924072,
+      "learning_rate": 7.750566945737412e-06,
+      "loss": 0.5471,
+      "step": 20585
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.8463364839553833,
+      "learning_rate": 7.750346763613277e-06,
+      "loss": 0.5163,
+      "step": 20586
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.5210959911346436,
+      "learning_rate": 7.750126573841457e-06,
+      "loss": 0.7521,
+      "step": 20587
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 4.745448112487793,
+      "learning_rate": 7.749906376422566e-06,
+      "loss": 0.8984,
+      "step": 20588
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.368396282196045,
+      "learning_rate": 7.749686171357216e-06,
+      "loss": 0.6185,
+      "step": 20589
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.2764179706573486,
+      "learning_rate": 7.749465958646018e-06,
+      "loss": 0.6867,
+      "step": 20590
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.0300536155700684,
+      "learning_rate": 7.749245738289585e-06,
+      "loss": 0.6897,
+      "step": 20591
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.5509212017059326,
+      "learning_rate": 7.749025510288531e-06,
+      "loss": 0.8508,
+      "step": 20592
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 3.6939809322357178,
+      "learning_rate": 7.748805274643465e-06,
+      "loss": 0.7897,
+      "step": 20593
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.421194314956665,
+      "learning_rate": 7.748585031355003e-06,
+      "loss": 0.4565,
+      "step": 20594
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.2426388263702393,
+      "learning_rate": 7.748364780423754e-06,
+      "loss": 0.7445,
+      "step": 20595
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 3.4253275394439697,
+      "learning_rate": 7.748144521850333e-06,
+      "loss": 0.8004,
+      "step": 20596
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.0098788738250732,
+      "learning_rate": 7.747924255635351e-06,
+      "loss": 0.4226,
+      "step": 20597
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 3.2407078742980957,
+      "learning_rate": 7.747703981779421e-06,
+      "loss": 0.7413,
+      "step": 20598
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.2676942348480225,
+      "learning_rate": 7.747483700283156e-06,
+      "loss": 0.7329,
+      "step": 20599
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 4.024285793304443,
+      "learning_rate": 7.747263411147168e-06,
+      "loss": 0.495,
+      "step": 20600
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.9907715320587158,
+      "learning_rate": 7.74704311437207e-06,
+      "loss": 0.7305,
+      "step": 20601
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 3.5853755474090576,
+      "learning_rate": 7.746822809958475e-06,
+      "loss": 0.7805,
+      "step": 20602
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.805587649345398,
+      "learning_rate": 7.746602497906992e-06,
+      "loss": 0.5261,
+      "step": 20603
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 3.6878244876861572,
+      "learning_rate": 7.74638217821824e-06,
+      "loss": 0.6361,
+      "step": 20604
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 4.179812431335449,
+      "learning_rate": 7.746161850892824e-06,
+      "loss": 0.8174,
+      "step": 20605
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.9240808486938477,
+      "learning_rate": 7.745941515931362e-06,
+      "loss": 0.5221,
+      "step": 20606
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 4.942934989929199,
+      "learning_rate": 7.745721173334466e-06,
+      "loss": 0.6521,
+      "step": 20607
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 3.1716225147247314,
+      "learning_rate": 7.74550082310275e-06,
+      "loss": 0.8377,
+      "step": 20608
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.4471590518951416,
+      "learning_rate": 7.745280465236821e-06,
+      "loss": 0.6904,
+      "step": 20609
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.3601009845733643,
+      "learning_rate": 7.745060099737296e-06,
+      "loss": 0.8546,
+      "step": 20610
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.01908278465271,
+      "learning_rate": 7.744839726604789e-06,
+      "loss": 0.5127,
+      "step": 20611
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.74457049369812,
+      "learning_rate": 7.744619345839909e-06,
+      "loss": 0.449,
+      "step": 20612
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.9930284023284912,
+      "learning_rate": 7.744398957443273e-06,
+      "loss": 0.7682,
+      "step": 20613
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.715036153793335,
+      "learning_rate": 7.74417856141549e-06,
+      "loss": 0.6076,
+      "step": 20614
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.2396838665008545,
+      "learning_rate": 7.743958157757177e-06,
+      "loss": 0.3428,
+      "step": 20615
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.777397871017456,
+      "learning_rate": 7.74373774646894e-06,
+      "loss": 0.4622,
+      "step": 20616
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 4.4294257164001465,
+      "learning_rate": 7.7435173275514e-06,
+      "loss": 0.8623,
+      "step": 20617
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 5.031561851501465,
+      "learning_rate": 7.743296901005163e-06,
+      "loss": 0.8142,
+      "step": 20618
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 3.199636220932007,
+      "learning_rate": 7.743076466830845e-06,
+      "loss": 0.8466,
+      "step": 20619
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.559246301651001,
+      "learning_rate": 7.742856025029063e-06,
+      "loss": 0.8199,
+      "step": 20620
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 3.2837812900543213,
+      "learning_rate": 7.742635575600421e-06,
+      "loss": 0.8914,
+      "step": 20621
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.4752840995788574,
+      "learning_rate": 7.742415118545539e-06,
+      "loss": 0.56,
+      "step": 20622
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.822439670562744,
+      "learning_rate": 7.74219465386503e-06,
+      "loss": 0.7375,
+      "step": 20623
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.235159397125244,
+      "learning_rate": 7.741974181559502e-06,
+      "loss": 0.8839,
+      "step": 20624
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.3202857971191406,
+      "learning_rate": 7.741753701629571e-06,
+      "loss": 0.5882,
+      "step": 20625
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.569788932800293,
+      "learning_rate": 7.741533214075852e-06,
+      "loss": 0.6269,
+      "step": 20626
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 3.4385690689086914,
+      "learning_rate": 7.741312718898954e-06,
+      "loss": 0.5462,
+      "step": 20627
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 3.250458240509033,
+      "learning_rate": 7.741092216099493e-06,
+      "loss": 0.6391,
+      "step": 20628
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.7968435287475586,
+      "learning_rate": 7.740871705678081e-06,
+      "loss": 0.6272,
+      "step": 20629
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.6625723838806152,
+      "learning_rate": 7.740651187635333e-06,
+      "loss": 0.6563,
+      "step": 20630
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 3.161463975906372,
+      "learning_rate": 7.740430661971859e-06,
+      "loss": 0.675,
+      "step": 20631
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.3567447662353516,
+      "learning_rate": 7.740210128688274e-06,
+      "loss": 0.6284,
+      "step": 20632
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.9040659666061401,
+      "learning_rate": 7.739989587785192e-06,
+      "loss": 0.5784,
+      "step": 20633
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.3012120723724365,
+      "learning_rate": 7.739769039263228e-06,
+      "loss": 0.5292,
+      "step": 20634
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 3.1934938430786133,
+      "learning_rate": 7.73954848312299e-06,
+      "loss": 0.6356,
+      "step": 20635
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 3.921104669570923,
+      "learning_rate": 7.739327919365093e-06,
+      "loss": 0.6329,
+      "step": 20636
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 3.7951722145080566,
+      "learning_rate": 7.739107347990152e-06,
+      "loss": 0.8108,
+      "step": 20637
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.050964117050171,
+      "learning_rate": 7.73888676899878e-06,
+      "loss": 0.4773,
+      "step": 20638
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.7637579441070557,
+      "learning_rate": 7.738666182391587e-06,
+      "loss": 0.5573,
+      "step": 20639
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.7637593746185303,
+      "learning_rate": 7.738445588169194e-06,
+      "loss": 0.8448,
+      "step": 20640
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.690159797668457,
+      "learning_rate": 7.738224986332207e-06,
+      "loss": 0.7166,
+      "step": 20641
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.6330362558364868,
+      "learning_rate": 7.738004376881242e-06,
+      "loss": 0.5488,
+      "step": 20642
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.161701202392578,
+      "learning_rate": 7.737783759816913e-06,
+      "loss": 0.6363,
+      "step": 20643
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.298645257949829,
+      "learning_rate": 7.737563135139833e-06,
+      "loss": 0.6822,
+      "step": 20644
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 3.4931769371032715,
+      "learning_rate": 7.737342502850616e-06,
+      "loss": 0.8489,
+      "step": 20645
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.599491834640503,
+      "learning_rate": 7.737121862949875e-06,
+      "loss": 0.7092,
+      "step": 20646
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 4.770122051239014,
+      "learning_rate": 7.736901215438222e-06,
+      "loss": 0.6152,
+      "step": 20647
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 3.268825054168701,
+      "learning_rate": 7.736680560316273e-06,
+      "loss": 0.6861,
+      "step": 20648
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 3.4572246074676514,
+      "learning_rate": 7.73645989758464e-06,
+      "loss": 0.5398,
+      "step": 20649
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 3.416294813156128,
+      "learning_rate": 7.736239227243938e-06,
+      "loss": 0.6702,
+      "step": 20650
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.8468170166015625,
+      "learning_rate": 7.736018549294779e-06,
+      "loss": 0.5316,
+      "step": 20651
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.9016190767288208,
+      "learning_rate": 7.735797863737777e-06,
+      "loss": 0.445,
+      "step": 20652
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.3370754718780518,
+      "learning_rate": 7.735577170573546e-06,
+      "loss": 0.717,
+      "step": 20653
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.5357136726379395,
+      "learning_rate": 7.735356469802701e-06,
+      "loss": 0.7386,
+      "step": 20654
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.3519814014434814,
+      "learning_rate": 7.735135761425854e-06,
+      "loss": 0.6737,
+      "step": 20655
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.9755631685256958,
+      "learning_rate": 7.734915045443618e-06,
+      "loss": 0.7546,
+      "step": 20656
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 7.163577079772949,
+      "learning_rate": 7.734694321856609e-06,
+      "loss": 0.5164,
+      "step": 20657
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.236675977706909,
+      "learning_rate": 7.734473590665439e-06,
+      "loss": 0.7218,
+      "step": 20658
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.5263701677322388,
+      "learning_rate": 7.734252851870722e-06,
+      "loss": 0.9604,
+      "step": 20659
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.2550017833709717,
+      "learning_rate": 7.734032105473073e-06,
+      "loss": 0.6877,
+      "step": 20660
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.9932608604431152,
+      "learning_rate": 7.733811351473103e-06,
+      "loss": 0.4442,
+      "step": 20661
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.6059256792068481,
+      "learning_rate": 7.73359058987143e-06,
+      "loss": 0.6269,
+      "step": 20662
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 4.299672603607178,
+      "learning_rate": 7.733369820668664e-06,
+      "loss": 0.6437,
+      "step": 20663
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.564622640609741,
+      "learning_rate": 7.733149043865422e-06,
+      "loss": 0.8433,
+      "step": 20664
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 3.8665120601654053,
+      "learning_rate": 7.732928259462316e-06,
+      "loss": 0.7885,
+      "step": 20665
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.8139735460281372,
+      "learning_rate": 7.732707467459959e-06,
+      "loss": 0.792,
+      "step": 20666
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 3.8287265300750732,
+      "learning_rate": 7.732486667858966e-06,
+      "loss": 0.7861,
+      "step": 20667
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.7974481582641602,
+      "learning_rate": 7.732265860659952e-06,
+      "loss": 0.5178,
+      "step": 20668
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.8816841840744019,
+      "learning_rate": 7.73204504586353e-06,
+      "loss": 0.7351,
+      "step": 20669
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.8098536729812622,
+      "learning_rate": 7.731824223470314e-06,
+      "loss": 0.632,
+      "step": 20670
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.7252613306045532,
+      "learning_rate": 7.73160339348092e-06,
+      "loss": 0.3333,
+      "step": 20671
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.5694963932037354,
+      "learning_rate": 7.731382555895959e-06,
+      "loss": 0.6396,
+      "step": 20672
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.899856448173523,
+      "learning_rate": 7.731161710716047e-06,
+      "loss": 0.7067,
+      "step": 20673
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 4.4752960205078125,
+      "learning_rate": 7.730940857941795e-06,
+      "loss": 0.7676,
+      "step": 20674
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.6290628910064697,
+      "learning_rate": 7.730719997573822e-06,
+      "loss": 0.6785,
+      "step": 20675
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.792484998703003,
+      "learning_rate": 7.730499129612738e-06,
+      "loss": 0.7186,
+      "step": 20676
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 3.4575703144073486,
+      "learning_rate": 7.73027825405916e-06,
+      "loss": 0.5919,
+      "step": 20677
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.190735340118408,
+      "learning_rate": 7.730057370913701e-06,
+      "loss": 0.5129,
+      "step": 20678
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.524661660194397,
+      "learning_rate": 7.729836480176974e-06,
+      "loss": 0.6455,
+      "step": 20679
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 3.155254364013672,
+      "learning_rate": 7.729615581849595e-06,
+      "loss": 0.5808,
+      "step": 20680
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.7606828212738037,
+      "learning_rate": 7.729394675932178e-06,
+      "loss": 0.5745,
+      "step": 20681
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 3.081198215484619,
+      "learning_rate": 7.729173762425337e-06,
+      "loss": 0.5468,
+      "step": 20682
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.633712887763977,
+      "learning_rate": 7.728952841329685e-06,
+      "loss": 0.6204,
+      "step": 20683
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.574099540710449,
+      "learning_rate": 7.728731912645838e-06,
+      "loss": 0.7842,
+      "step": 20684
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.0247514247894287,
+      "learning_rate": 7.72851097637441e-06,
+      "loss": 0.9025,
+      "step": 20685
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 3.5149548053741455,
+      "learning_rate": 7.728290032516015e-06,
+      "loss": 0.7168,
+      "step": 20686
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.6317179203033447,
+      "learning_rate": 7.728069081071268e-06,
+      "loss": 0.6955,
+      "step": 20687
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.3509910106658936,
+      "learning_rate": 7.727848122040782e-06,
+      "loss": 0.432,
+      "step": 20688
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.916178584098816,
+      "learning_rate": 7.727627155425173e-06,
+      "loss": 0.73,
+      "step": 20689
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 3.7058351039886475,
+      "learning_rate": 7.727406181225052e-06,
+      "loss": 0.9895,
+      "step": 20690
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 3.3698666095733643,
+      "learning_rate": 7.72718519944104e-06,
+      "loss": 0.6524,
+      "step": 20691
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.1214866638183594,
+      "learning_rate": 7.726964210073745e-06,
+      "loss": 0.6108,
+      "step": 20692
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.4736292362213135,
+      "learning_rate": 7.726743213123785e-06,
+      "loss": 0.5474,
+      "step": 20693
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.680001974105835,
+      "learning_rate": 7.726522208591773e-06,
+      "loss": 0.7963,
+      "step": 20694
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.9447872638702393,
+      "learning_rate": 7.726301196478325e-06,
+      "loss": 0.7172,
+      "step": 20695
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 3.783907651901245,
+      "learning_rate": 7.726080176784056e-06,
+      "loss": 0.6452,
+      "step": 20696
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.847465991973877,
+      "learning_rate": 7.725859149509577e-06,
+      "loss": 0.4485,
+      "step": 20697
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.8136188983917236,
+      "learning_rate": 7.725638114655505e-06,
+      "loss": 0.5047,
+      "step": 20698
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.2919697761535645,
+      "learning_rate": 7.725417072222454e-06,
+      "loss": 0.5595,
+      "step": 20699
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.5151338577270508,
+      "learning_rate": 7.725196022211041e-06,
+      "loss": 0.6564,
+      "step": 20700
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.9766523838043213,
+      "learning_rate": 7.724974964621878e-06,
+      "loss": 0.4937,
+      "step": 20701
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.2251644134521484,
+      "learning_rate": 7.72475389945558e-06,
+      "loss": 0.6425,
+      "step": 20702
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.4703432321548462,
+      "learning_rate": 7.724532826712762e-06,
+      "loss": 0.633,
+      "step": 20703
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 3.63674259185791,
+      "learning_rate": 7.724311746394039e-06,
+      "loss": 0.7233,
+      "step": 20704
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 3.4111175537109375,
+      "learning_rate": 7.724090658500026e-06,
+      "loss": 0.9196,
+      "step": 20705
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.916953682899475,
+      "learning_rate": 7.723869563031337e-06,
+      "loss": 0.7009,
+      "step": 20706
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.7190167903900146,
+      "learning_rate": 7.723648459988587e-06,
+      "loss": 0.4245,
+      "step": 20707
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.77443528175354,
+      "learning_rate": 7.72342734937239e-06,
+      "loss": 0.7903,
+      "step": 20708
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.9421768188476562,
+      "learning_rate": 7.723206231183363e-06,
+      "loss": 0.871,
+      "step": 20709
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.4647632837295532,
+      "learning_rate": 7.72298510542212e-06,
+      "loss": 0.5092,
+      "step": 20710
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.147374153137207,
+      "learning_rate": 7.722763972089275e-06,
+      "loss": 0.6608,
+      "step": 20711
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.4630324840545654,
+      "learning_rate": 7.722542831185443e-06,
+      "loss": 0.6098,
+      "step": 20712
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.836613416671753,
+      "learning_rate": 7.72232168271124e-06,
+      "loss": 0.6506,
+      "step": 20713
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.4735045433044434,
+      "learning_rate": 7.72210052666728e-06,
+      "loss": 0.5059,
+      "step": 20714
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.1243650913238525,
+      "learning_rate": 7.721879363054177e-06,
+      "loss": 0.8491,
+      "step": 20715
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.900286316871643,
+      "learning_rate": 7.721658191872547e-06,
+      "loss": 0.6951,
+      "step": 20716
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.5714643001556396,
+      "learning_rate": 7.721437013123008e-06,
+      "loss": 0.6353,
+      "step": 20717
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.725053071975708,
+      "learning_rate": 7.72121582680617e-06,
+      "loss": 0.5611,
+      "step": 20718
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.6671228408813477,
+      "learning_rate": 7.720994632922649e-06,
+      "loss": 0.4168,
+      "step": 20719
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.141132116317749,
+      "learning_rate": 7.720773431473063e-06,
+      "loss": 0.6809,
+      "step": 20720
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.886843681335449,
+      "learning_rate": 7.720552222458025e-06,
+      "loss": 0.3842,
+      "step": 20721
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.7977657318115234,
+      "learning_rate": 7.72033100587815e-06,
+      "loss": 0.2433,
+      "step": 20722
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 3.2402184009552,
+      "learning_rate": 7.720109781734055e-06,
+      "loss": 0.7721,
+      "step": 20723
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.7379313707351685,
+      "learning_rate": 7.719888550026353e-06,
+      "loss": 0.552,
+      "step": 20724
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.492457628250122,
+      "learning_rate": 7.71966731075566e-06,
+      "loss": 0.6639,
+      "step": 20725
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.1225876808166504,
+      "learning_rate": 7.71944606392259e-06,
+      "loss": 0.7589,
+      "step": 20726
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.3934714794158936,
+      "learning_rate": 7.71922480952776e-06,
+      "loss": 0.6348,
+      "step": 20727
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.231597661972046,
+      "learning_rate": 7.719003547571784e-06,
+      "loss": 0.4889,
+      "step": 20728
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.6849355697631836,
+      "learning_rate": 7.718782278055277e-06,
+      "loss": 0.7607,
+      "step": 20729
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 3.80342173576355,
+      "learning_rate": 7.718561000978858e-06,
+      "loss": 0.8349,
+      "step": 20730
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.9867836236953735,
+      "learning_rate": 7.718339716343135e-06,
+      "loss": 0.7513,
+      "step": 20731
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.352489471435547,
+      "learning_rate": 7.71811842414873e-06,
+      "loss": 0.9963,
+      "step": 20732
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.144692897796631,
+      "learning_rate": 7.717897124396256e-06,
+      "loss": 0.7493,
+      "step": 20733
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.19754695892334,
+      "learning_rate": 7.717675817086327e-06,
+      "loss": 0.5987,
+      "step": 20734
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 3.2426490783691406,
+      "learning_rate": 7.71745450221956e-06,
+      "loss": 0.6133,
+      "step": 20735
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.4099690914154053,
+      "learning_rate": 7.717233179796572e-06,
+      "loss": 0.5351,
+      "step": 20736
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 3.821291923522949,
+      "learning_rate": 7.717011849817974e-06,
+      "loss": 0.851,
+      "step": 20737
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 3.2562360763549805,
+      "learning_rate": 7.716790512284384e-06,
+      "loss": 0.5346,
+      "step": 20738
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.9884599447250366,
+      "learning_rate": 7.71656916719642e-06,
+      "loss": 0.7328,
+      "step": 20739
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 3.7095701694488525,
+      "learning_rate": 7.716347814554692e-06,
+      "loss": 0.5834,
+      "step": 20740
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.9166854619979858,
+      "learning_rate": 7.716126454359818e-06,
+      "loss": 0.6457,
+      "step": 20741
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.9895323514938354,
+      "learning_rate": 7.715905086612416e-06,
+      "loss": 0.6104,
+      "step": 20742
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 18.056804656982422,
+      "learning_rate": 7.715683711313099e-06,
+      "loss": 0.732,
+      "step": 20743
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.8108055591583252,
+      "learning_rate": 7.715462328462482e-06,
+      "loss": 0.8223,
+      "step": 20744
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.7306792736053467,
+      "learning_rate": 7.71524093806118e-06,
+      "loss": 0.6285,
+      "step": 20745
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 3.1129181385040283,
+      "learning_rate": 7.715019540109812e-06,
+      "loss": 0.6538,
+      "step": 20746
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.3207263946533203,
+      "learning_rate": 7.714798134608993e-06,
+      "loss": 0.6148,
+      "step": 20747
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.6005653142929077,
+      "learning_rate": 7.714576721559334e-06,
+      "loss": 0.551,
+      "step": 20748
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.846726417541504,
+      "learning_rate": 7.714355300961457e-06,
+      "loss": 0.6074,
+      "step": 20749
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 3.500657320022583,
+      "learning_rate": 7.714133872815974e-06,
+      "loss": 0.7073,
+      "step": 20750
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.444789171218872,
+      "learning_rate": 7.713912437123501e-06,
+      "loss": 0.5442,
+      "step": 20751
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.338083267211914,
+      "learning_rate": 7.713690993884653e-06,
+      "loss": 0.6406,
+      "step": 20752
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 3.236943244934082,
+      "learning_rate": 7.71346954310005e-06,
+      "loss": 0.896,
+      "step": 20753
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.3256492614746094,
+      "learning_rate": 7.713248084770302e-06,
+      "loss": 0.6127,
+      "step": 20754
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.506836175918579,
+      "learning_rate": 7.713026618896029e-06,
+      "loss": 0.5694,
+      "step": 20755
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.8251004219055176,
+      "learning_rate": 7.712805145477845e-06,
+      "loss": 0.7659,
+      "step": 20756
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.8261868953704834,
+      "learning_rate": 7.712583664516366e-06,
+      "loss": 0.7772,
+      "step": 20757
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.783597707748413,
+      "learning_rate": 7.712362176012207e-06,
+      "loss": 0.7957,
+      "step": 20758
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 3.0443413257598877,
+      "learning_rate": 7.712140679965984e-06,
+      "loss": 0.6598,
+      "step": 20759
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 4.433345794677734,
+      "learning_rate": 7.711919176378316e-06,
+      "loss": 0.5051,
+      "step": 20760
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.8876917362213135,
+      "learning_rate": 7.711697665249817e-06,
+      "loss": 0.6209,
+      "step": 20761
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.3198082447052,
+      "learning_rate": 7.711476146581101e-06,
+      "loss": 0.3148,
+      "step": 20762
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.02250075340271,
+      "learning_rate": 7.711254620372786e-06,
+      "loss": 0.7635,
+      "step": 20763
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 4.268056392669678,
+      "learning_rate": 7.711033086625488e-06,
+      "loss": 1.0796,
+      "step": 20764
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.178239583969116,
+      "learning_rate": 7.710811545339823e-06,
+      "loss": 0.7627,
+      "step": 20765
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 4.814976692199707,
+      "learning_rate": 7.710589996516407e-06,
+      "loss": 0.5846,
+      "step": 20766
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 3.551161766052246,
+      "learning_rate": 7.710368440155854e-06,
+      "loss": 0.5376,
+      "step": 20767
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 3.0007996559143066,
+      "learning_rate": 7.710146876258782e-06,
+      "loss": 0.5244,
+      "step": 20768
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.3699355125427246,
+      "learning_rate": 7.709925304825808e-06,
+      "loss": 0.8804,
+      "step": 20769
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.437424659729004,
+      "learning_rate": 7.709703725857545e-06,
+      "loss": 0.7365,
+      "step": 20770
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 4.505810737609863,
+      "learning_rate": 7.709482139354611e-06,
+      "loss": 1.0125,
+      "step": 20771
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.6168137788772583,
+      "learning_rate": 7.709260545317624e-06,
+      "loss": 0.5756,
+      "step": 20772
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 3.7197670936584473,
+      "learning_rate": 7.709038943747197e-06,
+      "loss": 0.8041,
+      "step": 20773
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 4.112822532653809,
+      "learning_rate": 7.708817334643949e-06,
+      "loss": 0.664,
+      "step": 20774
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.638413190841675,
+      "learning_rate": 7.708595718008493e-06,
+      "loss": 0.9348,
+      "step": 20775
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.5304303169250488,
+      "learning_rate": 7.708374093841448e-06,
+      "loss": 0.7558,
+      "step": 20776
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.309529423713684,
+      "learning_rate": 7.70815246214343e-06,
+      "loss": 0.5885,
+      "step": 20777
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 3.013201951980591,
+      "learning_rate": 7.707930822915053e-06,
+      "loss": 0.7954,
+      "step": 20778
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.2618064880371094,
+      "learning_rate": 7.707709176156934e-06,
+      "loss": 0.6303,
+      "step": 20779
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.7258687019348145,
+      "learning_rate": 7.707487521869692e-06,
+      "loss": 0.8541,
+      "step": 20780
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 4.033435821533203,
+      "learning_rate": 7.70726586005394e-06,
+      "loss": 0.7762,
+      "step": 20781
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 3.852820634841919,
+      "learning_rate": 7.707044190710298e-06,
+      "loss": 0.577,
+      "step": 20782
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.8162131309509277,
+      "learning_rate": 7.706822513839379e-06,
+      "loss": 0.6826,
+      "step": 20783
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.960458278656006,
+      "learning_rate": 7.7066008294418e-06,
+      "loss": 0.8629,
+      "step": 20784
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.6318516731262207,
+      "learning_rate": 7.706379137518178e-06,
+      "loss": 0.5608,
+      "step": 20785
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.9327261447906494,
+      "learning_rate": 7.706157438069131e-06,
+      "loss": 0.7037,
+      "step": 20786
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.2301442623138428,
+      "learning_rate": 7.705935731095271e-06,
+      "loss": 0.4719,
+      "step": 20787
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.2570271492004395,
+      "learning_rate": 7.70571401659722e-06,
+      "loss": 0.7383,
+      "step": 20788
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.9982800483703613,
+      "learning_rate": 7.70549229457559e-06,
+      "loss": 0.4746,
+      "step": 20789
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.713773250579834,
+      "learning_rate": 7.705270565031002e-06,
+      "loss": 0.6634,
+      "step": 20790
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.465327262878418,
+      "learning_rate": 7.705048827964068e-06,
+      "loss": 0.5201,
+      "step": 20791
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.364751100540161,
+      "learning_rate": 7.704827083375406e-06,
+      "loss": 0.7579,
+      "step": 20792
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.452162742614746,
+      "learning_rate": 7.704605331265635e-06,
+      "loss": 0.6741,
+      "step": 20793
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.4124493598937988,
+      "learning_rate": 7.70438357163537e-06,
+      "loss": 0.4927,
+      "step": 20794
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.0096070766448975,
+      "learning_rate": 7.704161804485227e-06,
+      "loss": 0.7428,
+      "step": 20795
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 3.589825391769409,
+      "learning_rate": 7.703940029815821e-06,
+      "loss": 0.4741,
+      "step": 20796
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.8199682235717773,
+      "learning_rate": 7.703718247627771e-06,
+      "loss": 0.6283,
+      "step": 20797
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.51894211769104,
+      "learning_rate": 7.703496457921696e-06,
+      "loss": 0.603,
+      "step": 20798
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.19205904006958,
+      "learning_rate": 7.703274660698208e-06,
+      "loss": 0.6542,
+      "step": 20799
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.8801825046539307,
+      "learning_rate": 7.703052855957925e-06,
+      "loss": 0.6316,
+      "step": 20800
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.6196138858795166,
+      "learning_rate": 7.702831043701467e-06,
+      "loss": 0.422,
+      "step": 20801
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 3.339613676071167,
+      "learning_rate": 7.702609223929447e-06,
+      "loss": 0.5528,
+      "step": 20802
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 4.2406229972839355,
+      "learning_rate": 7.702387396642484e-06,
+      "loss": 0.5203,
+      "step": 20803
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.5856449604034424,
+      "learning_rate": 7.702165561841192e-06,
+      "loss": 0.4646,
+      "step": 20804
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.3018136024475098,
+      "learning_rate": 7.701943719526192e-06,
+      "loss": 0.9079,
+      "step": 20805
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.706894874572754,
+      "learning_rate": 7.701721869698097e-06,
+      "loss": 0.7785,
+      "step": 20806
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 3.26584529876709,
+      "learning_rate": 7.701500012357529e-06,
+      "loss": 0.5235,
+      "step": 20807
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 3.3577117919921875,
+      "learning_rate": 7.701278147505097e-06,
+      "loss": 0.5297,
+      "step": 20808
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 3.2990882396698,
+      "learning_rate": 7.701056275141425e-06,
+      "loss": 0.8188,
+      "step": 20809
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.253483295440674,
+      "learning_rate": 7.700834395267126e-06,
+      "loss": 0.8103,
+      "step": 20810
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.037351608276367,
+      "learning_rate": 7.700612507882819e-06,
+      "loss": 0.7274,
+      "step": 20811
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 3.11616849899292,
+      "learning_rate": 7.700390612989122e-06,
+      "loss": 0.801,
+      "step": 20812
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.536139965057373,
+      "learning_rate": 7.700168710586648e-06,
+      "loss": 0.8195,
+      "step": 20813
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 3.2096469402313232,
+      "learning_rate": 7.699946800676016e-06,
+      "loss": 0.4644,
+      "step": 20814
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.0764167308807373,
+      "learning_rate": 7.699724883257844e-06,
+      "loss": 0.778,
+      "step": 20815
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.152019739151001,
+      "learning_rate": 7.69950295833275e-06,
+      "loss": 0.7311,
+      "step": 20816
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 3.3136367797851562,
+      "learning_rate": 7.699281025901348e-06,
+      "loss": 0.7849,
+      "step": 20817
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.9753541946411133,
+      "learning_rate": 7.699059085964257e-06,
+      "loss": 0.6381,
+      "step": 20818
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.562516927719116,
+      "learning_rate": 7.698837138522094e-06,
+      "loss": 0.5931,
+      "step": 20819
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 3.299696207046509,
+      "learning_rate": 7.698615183575476e-06,
+      "loss": 0.7279,
+      "step": 20820
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 3.521930694580078,
+      "learning_rate": 7.698393221125018e-06,
+      "loss": 0.6397,
+      "step": 20821
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.791285753250122,
+      "learning_rate": 7.698171251171343e-06,
+      "loss": 0.5764,
+      "step": 20822
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.948068380355835,
+      "learning_rate": 7.697949273715062e-06,
+      "loss": 0.4113,
+      "step": 20823
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.8283963203430176,
+      "learning_rate": 7.697727288756796e-06,
+      "loss": 0.6728,
+      "step": 20824
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.6040709018707275,
+      "learning_rate": 7.697505296297159e-06,
+      "loss": 0.7882,
+      "step": 20825
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 3.0693187713623047,
+      "learning_rate": 7.697283296336772e-06,
+      "loss": 0.6315,
+      "step": 20826
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.0066299438476562,
+      "learning_rate": 7.697061288876251e-06,
+      "loss": 0.4268,
+      "step": 20827
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.8724697828292847,
+      "learning_rate": 7.696839273916212e-06,
+      "loss": 0.6871,
+      "step": 20828
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.2600605487823486,
+      "learning_rate": 7.696617251457272e-06,
+      "loss": 0.7439,
+      "step": 20829
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.371492624282837,
+      "learning_rate": 7.69639522150005e-06,
+      "loss": 0.6905,
+      "step": 20830
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.3574886322021484,
+      "learning_rate": 7.696173184045164e-06,
+      "loss": 0.4673,
+      "step": 20831
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 3.021472454071045,
+      "learning_rate": 7.69595113909323e-06,
+      "loss": 0.4662,
+      "step": 20832
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.6181877851486206,
+      "learning_rate": 7.695729086644868e-06,
+      "loss": 0.4465,
+      "step": 20833
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.0743253231048584,
+      "learning_rate": 7.695507026700691e-06,
+      "loss": 0.4652,
+      "step": 20834
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 4.573472023010254,
+      "learning_rate": 7.695284959261318e-06,
+      "loss": 0.8703,
+      "step": 20835
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.960485816001892,
+      "learning_rate": 7.695062884327369e-06,
+      "loss": 0.7165,
+      "step": 20836
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.0509724617004395,
+      "learning_rate": 7.69484080189946e-06,
+      "loss": 0.4918,
+      "step": 20837
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.093266248703003,
+      "learning_rate": 7.694618711978206e-06,
+      "loss": 0.5199,
+      "step": 20838
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.7714951038360596,
+      "learning_rate": 7.694396614564227e-06,
+      "loss": 0.6562,
+      "step": 20839
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.9356086254119873,
+      "learning_rate": 7.694174509658141e-06,
+      "loss": 0.7727,
+      "step": 20840
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.723447322845459,
+      "learning_rate": 7.693952397260566e-06,
+      "loss": 0.6592,
+      "step": 20841
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.890007734298706,
+      "learning_rate": 7.693730277372119e-06,
+      "loss": 0.8882,
+      "step": 20842
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.5761250257492065,
+      "learning_rate": 7.693508149993415e-06,
+      "loss": 0.6768,
+      "step": 20843
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.7260931730270386,
+      "learning_rate": 7.693286015125074e-06,
+      "loss": 0.7668,
+      "step": 20844
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.9933009147644043,
+      "learning_rate": 7.693063872767715e-06,
+      "loss": 0.5259,
+      "step": 20845
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.67958927154541,
+      "learning_rate": 7.692841722921954e-06,
+      "loss": 0.6045,
+      "step": 20846
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.797853469848633,
+      "learning_rate": 7.692619565588409e-06,
+      "loss": 0.4793,
+      "step": 20847
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.2010245323181152,
+      "learning_rate": 7.692397400767697e-06,
+      "loss": 0.5128,
+      "step": 20848
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 4.043004989624023,
+      "learning_rate": 7.692175228460436e-06,
+      "loss": 0.8486,
+      "step": 20849
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.092534065246582,
+      "learning_rate": 7.691953048667245e-06,
+      "loss": 0.5018,
+      "step": 20850
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.3711588382720947,
+      "learning_rate": 7.691730861388741e-06,
+      "loss": 0.8411,
+      "step": 20851
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.1905317306518555,
+      "learning_rate": 7.69150866662554e-06,
+      "loss": 0.7154,
+      "step": 20852
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 3.67134428024292,
+      "learning_rate": 7.691286464378265e-06,
+      "loss": 0.5403,
+      "step": 20853
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.8279167413711548,
+      "learning_rate": 7.69106425464753e-06,
+      "loss": 0.4788,
+      "step": 20854
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 3.870884895324707,
+      "learning_rate": 7.690842037433951e-06,
+      "loss": 0.886,
+      "step": 20855
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.165823221206665,
+      "learning_rate": 7.69061981273815e-06,
+      "loss": 0.8921,
+      "step": 20856
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.2710928916931152,
+      "learning_rate": 7.690397580560743e-06,
+      "loss": 0.642,
+      "step": 20857
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.268385887145996,
+      "learning_rate": 7.690175340902346e-06,
+      "loss": 0.7171,
+      "step": 20858
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.1814496517181396,
+      "learning_rate": 7.689953093763582e-06,
+      "loss": 0.5449,
+      "step": 20859
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 3.838749647140503,
+      "learning_rate": 7.689730839145065e-06,
+      "loss": 0.7063,
+      "step": 20860
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.8845736980438232,
+      "learning_rate": 7.689508577047416e-06,
+      "loss": 0.7381,
+      "step": 20861
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.7596149444580078,
+      "learning_rate": 7.689286307471247e-06,
+      "loss": 0.8178,
+      "step": 20862
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.9994220733642578,
+      "learning_rate": 7.689064030417184e-06,
+      "loss": 0.8451,
+      "step": 20863
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 3.970029592514038,
+      "learning_rate": 7.68884174588584e-06,
+      "loss": 0.6037,
+      "step": 20864
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.3723251819610596,
+      "learning_rate": 7.688619453877834e-06,
+      "loss": 0.623,
+      "step": 20865
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.6913954019546509,
+      "learning_rate": 7.688397154393785e-06,
+      "loss": 0.5044,
+      "step": 20866
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.9707027673721313,
+      "learning_rate": 7.68817484743431e-06,
+      "loss": 0.7424,
+      "step": 20867
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.7937709093093872,
+      "learning_rate": 7.687952533000028e-06,
+      "loss": 0.8513,
+      "step": 20868
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.5880296230316162,
+      "learning_rate": 7.687730211091557e-06,
+      "loss": 0.4676,
+      "step": 20869
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.3604702949523926,
+      "learning_rate": 7.687507881709517e-06,
+      "loss": 0.6719,
+      "step": 20870
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.2012903690338135,
+      "learning_rate": 7.687285544854522e-06,
+      "loss": 0.448,
+      "step": 20871
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.027578592300415,
+      "learning_rate": 7.687063200527192e-06,
+      "loss": 0.6448,
+      "step": 20872
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.165154218673706,
+      "learning_rate": 7.686840848728148e-06,
+      "loss": 0.7419,
+      "step": 20873
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.126614809036255,
+      "learning_rate": 7.686618489458005e-06,
+      "loss": 0.6263,
+      "step": 20874
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.2526867389678955,
+      "learning_rate": 7.686396122717382e-06,
+      "loss": 0.5809,
+      "step": 20875
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 3.4999730587005615,
+      "learning_rate": 7.686173748506897e-06,
+      "loss": 0.5965,
+      "step": 20876
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.951436996459961,
+      "learning_rate": 7.68595136682717e-06,
+      "loss": 0.7458,
+      "step": 20877
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 4.506922721862793,
+      "learning_rate": 7.685728977678821e-06,
+      "loss": 0.6511,
+      "step": 20878
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.2902238368988037,
+      "learning_rate": 7.685506581062463e-06,
+      "loss": 0.7342,
+      "step": 20879
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.30181884765625,
+      "learning_rate": 7.685284176978717e-06,
+      "loss": 0.6259,
+      "step": 20880
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.016700029373169,
+      "learning_rate": 7.685061765428203e-06,
+      "loss": 0.5226,
+      "step": 20881
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.20284366607666,
+      "learning_rate": 7.684839346411538e-06,
+      "loss": 0.5818,
+      "step": 20882
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.2430200576782227,
+      "learning_rate": 7.68461691992934e-06,
+      "loss": 0.6674,
+      "step": 20883
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.9817531108856201,
+      "learning_rate": 7.684394485982227e-06,
+      "loss": 0.6322,
+      "step": 20884
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 3.147655725479126,
+      "learning_rate": 7.684172044570821e-06,
+      "loss": 0.5871,
+      "step": 20885
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.6291098594665527,
+      "learning_rate": 7.683949595695737e-06,
+      "loss": 0.5654,
+      "step": 20886
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 3.544793128967285,
+      "learning_rate": 7.683727139357594e-06,
+      "loss": 0.6934,
+      "step": 20887
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.249922275543213,
+      "learning_rate": 7.683504675557012e-06,
+      "loss": 0.6339,
+      "step": 20888
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.1136960983276367,
+      "learning_rate": 7.683282204294608e-06,
+      "loss": 0.6922,
+      "step": 20889
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 4.919315338134766,
+      "learning_rate": 7.683059725571002e-06,
+      "loss": 0.8559,
+      "step": 20890
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.510404109954834,
+      "learning_rate": 7.68283723938681e-06,
+      "loss": 0.582,
+      "step": 20891
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.009674072265625,
+      "learning_rate": 7.682614745742653e-06,
+      "loss": 0.7575,
+      "step": 20892
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.4183263778686523,
+      "learning_rate": 7.682392244639152e-06,
+      "loss": 0.5056,
+      "step": 20893
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 3.1205830574035645,
+      "learning_rate": 7.682169736076921e-06,
+      "loss": 0.6312,
+      "step": 20894
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.9406800270080566,
+      "learning_rate": 7.681947220056583e-06,
+      "loss": 0.868,
+      "step": 20895
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.6413320302963257,
+      "learning_rate": 7.681724696578753e-06,
+      "loss": 0.6984,
+      "step": 20896
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.5021355152130127,
+      "learning_rate": 7.68150216564405e-06,
+      "loss": 0.4998,
+      "step": 20897
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 3.362860679626465,
+      "learning_rate": 7.681279627253095e-06,
+      "loss": 0.6719,
+      "step": 20898
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.76986563205719,
+      "learning_rate": 7.681057081406506e-06,
+      "loss": 0.7952,
+      "step": 20899
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.125101327896118,
+      "learning_rate": 7.680834528104901e-06,
+      "loss": 0.5639,
+      "step": 20900
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.6264688968658447,
+      "learning_rate": 7.6806119673489e-06,
+      "loss": 0.7493,
+      "step": 20901
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.0325818061828613,
+      "learning_rate": 7.680389399139121e-06,
+      "loss": 0.5966,
+      "step": 20902
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.9123382568359375,
+      "learning_rate": 7.680166823476183e-06,
+      "loss": 0.847,
+      "step": 20903
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 3.210365056991577,
+      "learning_rate": 7.679944240360706e-06,
+      "loss": 0.5628,
+      "step": 20904
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.9912409782409668,
+      "learning_rate": 7.679721649793307e-06,
+      "loss": 0.7302,
+      "step": 20905
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 3.2662527561187744,
+      "learning_rate": 7.679499051774607e-06,
+      "loss": 0.7865,
+      "step": 20906
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.429133176803589,
+      "learning_rate": 7.679276446305223e-06,
+      "loss": 0.6821,
+      "step": 20907
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.5669912099838257,
+      "learning_rate": 7.679053833385775e-06,
+      "loss": 0.7668,
+      "step": 20908
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.1950628757476807,
+      "learning_rate": 7.678831213016882e-06,
+      "loss": 0.6617,
+      "step": 20909
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.6585309505462646,
+      "learning_rate": 7.678608585199163e-06,
+      "loss": 0.5777,
+      "step": 20910
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.8837802410125732,
+      "learning_rate": 7.678385949933235e-06,
+      "loss": 0.7292,
+      "step": 20911
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.498321533203125,
+      "learning_rate": 7.678163307219722e-06,
+      "loss": 0.6584,
+      "step": 20912
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 4.2231526374816895,
+      "learning_rate": 7.677940657059238e-06,
+      "loss": 0.8297,
+      "step": 20913
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.8557896614074707,
+      "learning_rate": 7.677717999452405e-06,
+      "loss": 0.7915,
+      "step": 20914
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.4142775535583496,
+      "learning_rate": 7.677495334399843e-06,
+      "loss": 0.4982,
+      "step": 20915
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.7031822204589844,
+      "learning_rate": 7.677272661902167e-06,
+      "loss": 0.7284,
+      "step": 20916
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.6618688106536865,
+      "learning_rate": 7.67704998196e-06,
+      "loss": 0.7101,
+      "step": 20917
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.0040347576141357,
+      "learning_rate": 7.676827294573959e-06,
+      "loss": 0.6749,
+      "step": 20918
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.1150412559509277,
+      "learning_rate": 7.676604599744665e-06,
+      "loss": 0.9837,
+      "step": 20919
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.580313205718994,
+      "learning_rate": 7.676381897472736e-06,
+      "loss": 0.7581,
+      "step": 20920
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.8800163269042969,
+      "learning_rate": 7.676159187758792e-06,
+      "loss": 0.7806,
+      "step": 20921
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.386566400527954,
+      "learning_rate": 7.67593647060345e-06,
+      "loss": 0.6987,
+      "step": 20922
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 3.243743419647217,
+      "learning_rate": 7.675713746007332e-06,
+      "loss": 0.7589,
+      "step": 20923
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.7471104860305786,
+      "learning_rate": 7.675491013971056e-06,
+      "loss": 0.5223,
+      "step": 20924
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.2337872982025146,
+      "learning_rate": 7.675268274495243e-06,
+      "loss": 0.5356,
+      "step": 20925
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.6289000511169434,
+      "learning_rate": 7.67504552758051e-06,
+      "loss": 0.5834,
+      "step": 20926
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.5355610847473145,
+      "learning_rate": 7.67482277322748e-06,
+      "loss": 0.6433,
+      "step": 20927
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 3.573796272277832,
+      "learning_rate": 7.674600011436767e-06,
+      "loss": 0.4996,
+      "step": 20928
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.8623759746551514,
+      "learning_rate": 7.674377242208994e-06,
+      "loss": 0.6832,
+      "step": 20929
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 5.292903900146484,
+      "learning_rate": 7.67415446554478e-06,
+      "loss": 0.6766,
+      "step": 20930
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.836420774459839,
+      "learning_rate": 7.673931681444744e-06,
+      "loss": 0.7732,
+      "step": 20931
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 3.3555831909179688,
+      "learning_rate": 7.673708889909505e-06,
+      "loss": 0.7121,
+      "step": 20932
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 3.4267661571502686,
+      "learning_rate": 7.673486090939685e-06,
+      "loss": 0.4377,
+      "step": 20933
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.1092684268951416,
+      "learning_rate": 7.673263284535901e-06,
+      "loss": 0.5996,
+      "step": 20934
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.5682517290115356,
+      "learning_rate": 7.673040470698772e-06,
+      "loss": 0.5462,
+      "step": 20935
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.6395068168640137,
+      "learning_rate": 7.672817649428921e-06,
+      "loss": 0.5334,
+      "step": 20936
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.9470891952514648,
+      "learning_rate": 7.672594820726964e-06,
+      "loss": 0.6083,
+      "step": 20937
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.8675649166107178,
+      "learning_rate": 7.672371984593522e-06,
+      "loss": 0.6578,
+      "step": 20938
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.1006476879119873,
+      "learning_rate": 7.672149141029215e-06,
+      "loss": 0.466,
+      "step": 20939
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.3169504404067993,
+      "learning_rate": 7.671926290034662e-06,
+      "loss": 0.6076,
+      "step": 20940
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.681230068206787,
+      "learning_rate": 7.671703431610483e-06,
+      "loss": 0.453,
+      "step": 20941
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.9928950071334839,
+      "learning_rate": 7.671480565757296e-06,
+      "loss": 0.5467,
+      "step": 20942
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.7926640510559082,
+      "learning_rate": 7.671257692475724e-06,
+      "loss": 0.8934,
+      "step": 20943
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.73039174079895,
+      "learning_rate": 7.671034811766384e-06,
+      "loss": 0.5105,
+      "step": 20944
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.6849032640457153,
+      "learning_rate": 7.670811923629899e-06,
+      "loss": 0.6491,
+      "step": 20945
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.802220106124878,
+      "learning_rate": 7.670589028066885e-06,
+      "loss": 0.6842,
+      "step": 20946
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.0864176750183105,
+      "learning_rate": 7.670366125077964e-06,
+      "loss": 0.5909,
+      "step": 20947
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.9590866565704346,
+      "learning_rate": 7.670143214663755e-06,
+      "loss": 0.7438,
+      "step": 20948
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 3.604438304901123,
+      "learning_rate": 7.669920296824877e-06,
+      "loss": 0.9254,
+      "step": 20949
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.4665801525115967,
+      "learning_rate": 7.66969737156195e-06,
+      "loss": 0.9357,
+      "step": 20950
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.185803174972534,
+      "learning_rate": 7.669474438875597e-06,
+      "loss": 0.5749,
+      "step": 20951
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 4.0760931968688965,
+      "learning_rate": 7.669251498766434e-06,
+      "loss": 0.5735,
+      "step": 20952
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.6504515409469604,
+      "learning_rate": 7.669028551235082e-06,
+      "loss": 0.6234,
+      "step": 20953
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 6.796932220458984,
+      "learning_rate": 7.668805596282164e-06,
+      "loss": 0.8917,
+      "step": 20954
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.1146938800811768,
+      "learning_rate": 7.668582633908298e-06,
+      "loss": 0.8735,
+      "step": 20955
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.084348440170288,
+      "learning_rate": 7.6683596641141e-06,
+      "loss": 0.5584,
+      "step": 20956
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.6734424829483032,
+      "learning_rate": 7.668136686900193e-06,
+      "loss": 0.7561,
+      "step": 20957
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.628740668296814,
+      "learning_rate": 7.6679137022672e-06,
+      "loss": 0.7816,
+      "step": 20958
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.874525547027588,
+      "learning_rate": 7.667690710215739e-06,
+      "loss": 0.6293,
+      "step": 20959
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.001246690750122,
+      "learning_rate": 7.667467710746427e-06,
+      "loss": 0.767,
+      "step": 20960
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.243844509124756,
+      "learning_rate": 7.667244703859888e-06,
+      "loss": 0.4339,
+      "step": 20961
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.8545498847961426,
+      "learning_rate": 7.66702168955674e-06,
+      "loss": 0.6534,
+      "step": 20962
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.8333957195281982,
+      "learning_rate": 7.666798667837603e-06,
+      "loss": 0.6031,
+      "step": 20963
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.6024426221847534,
+      "learning_rate": 7.6665756387031e-06,
+      "loss": 0.6132,
+      "step": 20964
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.4795126914978027,
+      "learning_rate": 7.666352602153849e-06,
+      "loss": 0.4845,
+      "step": 20965
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.9149844646453857,
+      "learning_rate": 7.666129558190469e-06,
+      "loss": 0.6292,
+      "step": 20966
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.9048172235488892,
+      "learning_rate": 7.66590650681358e-06,
+      "loss": 0.7179,
+      "step": 20967
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.7450785636901855,
+      "learning_rate": 7.665683448023807e-06,
+      "loss": 0.6438,
+      "step": 20968
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.842646837234497,
+      "learning_rate": 7.665460381821765e-06,
+      "loss": 0.6237,
+      "step": 20969
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 3.7622838020324707,
+      "learning_rate": 7.665237308208076e-06,
+      "loss": 0.6881,
+      "step": 20970
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.7608604431152344,
+      "learning_rate": 7.665014227183361e-06,
+      "loss": 0.6489,
+      "step": 20971
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.8311898708343506,
+      "learning_rate": 7.66479113874824e-06,
+      "loss": 0.6588,
+      "step": 20972
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.737727403640747,
+      "learning_rate": 7.664568042903334e-06,
+      "loss": 0.4707,
+      "step": 20973
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.8575363159179688,
+      "learning_rate": 7.66434493964926e-06,
+      "loss": 0.6693,
+      "step": 20974
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.8287986516952515,
+      "learning_rate": 7.664121828986642e-06,
+      "loss": 0.9261,
+      "step": 20975
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.38057017326355,
+      "learning_rate": 7.6638987109161e-06,
+      "loss": 0.7786,
+      "step": 20976
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.859689474105835,
+      "learning_rate": 7.663675585438252e-06,
+      "loss": 0.7945,
+      "step": 20977
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.5764938592910767,
+      "learning_rate": 7.663452452553721e-06,
+      "loss": 0.5352,
+      "step": 20978
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 3.9031715393066406,
+      "learning_rate": 7.663229312263125e-06,
+      "loss": 0.924,
+      "step": 20979
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.197937488555908,
+      "learning_rate": 7.663006164567088e-06,
+      "loss": 0.5429,
+      "step": 20980
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.219064950942993,
+      "learning_rate": 7.662783009466227e-06,
+      "loss": 0.5692,
+      "step": 20981
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.734513521194458,
+      "learning_rate": 7.662559846961166e-06,
+      "loss": 0.5933,
+      "step": 20982
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.0007107257843018,
+      "learning_rate": 7.662336677052521e-06,
+      "loss": 0.6578,
+      "step": 20983
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.768466591835022,
+      "learning_rate": 7.662113499740917e-06,
+      "loss": 0.3379,
+      "step": 20984
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.598863124847412,
+      "learning_rate": 7.661890315026972e-06,
+      "loss": 0.5865,
+      "step": 20985
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.9169151782989502,
+      "learning_rate": 7.661667122911307e-06,
+      "loss": 0.6231,
+      "step": 20986
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.9933271408081055,
+      "learning_rate": 7.661443923394543e-06,
+      "loss": 0.7219,
+      "step": 20987
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.6086153984069824,
+      "learning_rate": 7.6612207164773e-06,
+      "loss": 0.7101,
+      "step": 20988
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 3.075366735458374,
+      "learning_rate": 7.6609975021602e-06,
+      "loss": 0.7655,
+      "step": 20989
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.2539842128753662,
+      "learning_rate": 7.66077428044386e-06,
+      "loss": 0.6807,
+      "step": 20990
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.4000070095062256,
+      "learning_rate": 7.660551051328908e-06,
+      "loss": 0.626,
+      "step": 20991
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 3.302431344985962,
+      "learning_rate": 7.660327814815958e-06,
+      "loss": 0.8264,
+      "step": 20992
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.425612688064575,
+      "learning_rate": 7.660104570905632e-06,
+      "loss": 0.5779,
+      "step": 20993
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 4.102215766906738,
+      "learning_rate": 7.659881319598552e-06,
+      "loss": 0.6065,
+      "step": 20994
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 3.295070171356201,
+      "learning_rate": 7.659658060895339e-06,
+      "loss": 0.8203,
+      "step": 20995
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.7683587074279785,
+      "learning_rate": 7.659434794796615e-06,
+      "loss": 0.6371,
+      "step": 20996
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 3.7726986408233643,
+      "learning_rate": 7.659211521302996e-06,
+      "loss": 0.8624,
+      "step": 20997
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.2976815700531006,
+      "learning_rate": 7.658988240415108e-06,
+      "loss": 0.643,
+      "step": 20998
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.898410439491272,
+      "learning_rate": 7.658764952133568e-06,
+      "loss": 0.8567,
+      "step": 20999
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.080235719680786,
+      "learning_rate": 7.658541656459e-06,
+      "loss": 0.7176,
+      "step": 21000
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.3175114393234253,
+      "learning_rate": 7.658318353392024e-06,
+      "loss": 0.3939,
+      "step": 21001
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.5928955078125,
+      "learning_rate": 7.658095042933261e-06,
+      "loss": 0.9056,
+      "step": 21002
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.691776990890503,
+      "learning_rate": 7.657871725083328e-06,
+      "loss": 0.585,
+      "step": 21003
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.5498939752578735,
+      "learning_rate": 7.657648399842852e-06,
+      "loss": 0.4932,
+      "step": 21004
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 3.2244954109191895,
+      "learning_rate": 7.65742506721245e-06,
+      "loss": 0.7022,
+      "step": 21005
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 3.331295967102051,
+      "learning_rate": 7.657201727192745e-06,
+      "loss": 0.4544,
+      "step": 21006
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.6191184520721436,
+      "learning_rate": 7.656978379784359e-06,
+      "loss": 0.7303,
+      "step": 21007
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.8282651901245117,
+      "learning_rate": 7.65675502498791e-06,
+      "loss": 0.4076,
+      "step": 21008
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.1516120433807373,
+      "learning_rate": 7.656531662804022e-06,
+      "loss": 0.6395,
+      "step": 21009
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.7840631008148193,
+      "learning_rate": 7.656308293233313e-06,
+      "loss": 0.5861,
+      "step": 21010
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.8281986713409424,
+      "learning_rate": 7.656084916276406e-06,
+      "loss": 0.3769,
+      "step": 21011
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.7034828662872314,
+      "learning_rate": 7.655861531933922e-06,
+      "loss": 0.7739,
+      "step": 21012
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 6.5503435134887695,
+      "learning_rate": 7.655638140206481e-06,
+      "loss": 0.4854,
+      "step": 21013
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.7435364723205566,
+      "learning_rate": 7.655414741094704e-06,
+      "loss": 0.7464,
+      "step": 21014
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.5657756328582764,
+      "learning_rate": 7.655191334599217e-06,
+      "loss": 0.5929,
+      "step": 21015
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.4662413597106934,
+      "learning_rate": 7.654967920720635e-06,
+      "loss": 0.5772,
+      "step": 21016
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.693342447280884,
+      "learning_rate": 7.654744499459583e-06,
+      "loss": 0.7156,
+      "step": 21017
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.7414886951446533,
+      "learning_rate": 7.654521070816681e-06,
+      "loss": 0.7073,
+      "step": 21018
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.814866304397583,
+      "learning_rate": 7.65429763479255e-06,
+      "loss": 0.6278,
+      "step": 21019
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.7637280225753784,
+      "learning_rate": 7.654074191387812e-06,
+      "loss": 0.6169,
+      "step": 21020
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.152334213256836,
+      "learning_rate": 7.653850740603088e-06,
+      "loss": 0.5243,
+      "step": 21021
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.9410111904144287,
+      "learning_rate": 7.653627282438997e-06,
+      "loss": 0.6905,
+      "step": 21022
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.637481689453125,
+      "learning_rate": 7.653403816896166e-06,
+      "loss": 0.5882,
+      "step": 21023
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.5599591732025146,
+      "learning_rate": 7.653180343975211e-06,
+      "loss": 0.5485,
+      "step": 21024
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 3.5463130474090576,
+      "learning_rate": 7.652956863676754e-06,
+      "loss": 0.7891,
+      "step": 21025
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.4579315185546875,
+      "learning_rate": 7.65273337600142e-06,
+      "loss": 0.6576,
+      "step": 21026
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.066098213195801,
+      "learning_rate": 7.652509880949828e-06,
+      "loss": 0.4657,
+      "step": 21027
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.9122016429901123,
+      "learning_rate": 7.6522863785226e-06,
+      "loss": 0.7816,
+      "step": 21028
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 3.6628994941711426,
+      "learning_rate": 7.652062868720357e-06,
+      "loss": 0.7913,
+      "step": 21029
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.6135354042053223,
+      "learning_rate": 7.65183935154372e-06,
+      "loss": 0.6966,
+      "step": 21030
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.4668800830841064,
+      "learning_rate": 7.65161582699331e-06,
+      "loss": 0.56,
+      "step": 21031
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.1166656017303467,
+      "learning_rate": 7.651392295069752e-06,
+      "loss": 0.7246,
+      "step": 21032
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.6972413063049316,
+      "learning_rate": 7.651168755773663e-06,
+      "loss": 0.7953,
+      "step": 21033
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.3402318954467773,
+      "learning_rate": 7.65094520910567e-06,
+      "loss": 0.582,
+      "step": 21034
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 3.505838394165039,
+      "learning_rate": 7.65072165506639e-06,
+      "loss": 0.8222,
+      "step": 21035
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.6441731452941895,
+      "learning_rate": 7.650498093656445e-06,
+      "loss": 0.8326,
+      "step": 21036
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.4090476036071777,
+      "learning_rate": 7.650274524876457e-06,
+      "loss": 0.7171,
+      "step": 21037
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 3.025081157684326,
+      "learning_rate": 7.65005094872705e-06,
+      "loss": 0.984,
+      "step": 21038
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.9968171119689941,
+      "learning_rate": 7.649827365208845e-06,
+      "loss": 0.5886,
+      "step": 21039
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.9199061393737793,
+      "learning_rate": 7.649603774322462e-06,
+      "loss": 0.6648,
+      "step": 21040
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.1401619911193848,
+      "learning_rate": 7.64938017606852e-06,
+      "loss": 0.5902,
+      "step": 21041
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.399766206741333,
+      "learning_rate": 7.64915657044765e-06,
+      "loss": 0.8673,
+      "step": 21042
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.5283515453338623,
+      "learning_rate": 7.648932957460465e-06,
+      "loss": 0.5891,
+      "step": 21043
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.473539113998413,
+      "learning_rate": 7.64870933710759e-06,
+      "loss": 0.7104,
+      "step": 21044
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.5056500434875488,
+      "learning_rate": 7.648485709389646e-06,
+      "loss": 0.7914,
+      "step": 21045
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.5747811794281006,
+      "learning_rate": 7.648262074307255e-06,
+      "loss": 0.8611,
+      "step": 21046
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.1135387420654297,
+      "learning_rate": 7.64803843186104e-06,
+      "loss": 0.5139,
+      "step": 21047
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 3.5900402069091797,
+      "learning_rate": 7.647814782051622e-06,
+      "loss": 0.7785,
+      "step": 21048
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.877307653427124,
+      "learning_rate": 7.647591124879622e-06,
+      "loss": 0.7816,
+      "step": 21049
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.0270578861236572,
+      "learning_rate": 7.647367460345665e-06,
+      "loss": 0.7556,
+      "step": 21050
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.240365982055664,
+      "learning_rate": 7.64714378845037e-06,
+      "loss": 0.6108,
+      "step": 21051
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.322235107421875,
+      "learning_rate": 7.64692010919436e-06,
+      "loss": 0.6474,
+      "step": 21052
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.8179049491882324,
+      "learning_rate": 7.646696422578257e-06,
+      "loss": 0.5596,
+      "step": 21053
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 3.18656587600708,
+      "learning_rate": 7.646472728602681e-06,
+      "loss": 0.6431,
+      "step": 21054
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.6502559185028076,
+      "learning_rate": 7.646249027268258e-06,
+      "loss": 0.5584,
+      "step": 21055
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.4337079524993896,
+      "learning_rate": 7.646025318575606e-06,
+      "loss": 0.7386,
+      "step": 21056
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.1728389263153076,
+      "learning_rate": 7.645801602525349e-06,
+      "loss": 0.6699,
+      "step": 21057
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.501447319984436,
+      "learning_rate": 7.64557787911811e-06,
+      "loss": 0.7974,
+      "step": 21058
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.2736406326293945,
+      "learning_rate": 7.64535414835451e-06,
+      "loss": 0.6382,
+      "step": 21059
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 3.465796709060669,
+      "learning_rate": 7.64513041023517e-06,
+      "loss": 0.6559,
+      "step": 21060
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.7176573276519775,
+      "learning_rate": 7.644906664760713e-06,
+      "loss": 0.8495,
+      "step": 21061
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.5210005044937134,
+      "learning_rate": 7.644682911931762e-06,
+      "loss": 0.8009,
+      "step": 21062
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.1958467960357666,
+      "learning_rate": 7.64445915174894e-06,
+      "loss": 0.731,
+      "step": 21063
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 3.0095646381378174,
+      "learning_rate": 7.644235384212865e-06,
+      "loss": 0.5792,
+      "step": 21064
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.598154306411743,
+      "learning_rate": 7.644011609324162e-06,
+      "loss": 0.5453,
+      "step": 21065
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.356919050216675,
+      "learning_rate": 7.643787827083455e-06,
+      "loss": 0.6392,
+      "step": 21066
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.2851336002349854,
+      "learning_rate": 7.643564037491365e-06,
+      "loss": 0.6886,
+      "step": 21067
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 3.2955868244171143,
+      "learning_rate": 7.643340240548512e-06,
+      "loss": 0.7126,
+      "step": 21068
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.0030441284179688,
+      "learning_rate": 7.64311643625552e-06,
+      "loss": 0.6151,
+      "step": 21069
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.3521981239318848,
+      "learning_rate": 7.642892624613013e-06,
+      "loss": 0.844,
+      "step": 21070
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.9645813703536987,
+      "learning_rate": 7.64266880562161e-06,
+      "loss": 0.8229,
+      "step": 21071
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.2705271244049072,
+      "learning_rate": 7.642444979281936e-06,
+      "loss": 0.3738,
+      "step": 21072
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 4.36628532409668,
+      "learning_rate": 7.642221145594611e-06,
+      "loss": 0.999,
+      "step": 21073
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.3922927379608154,
+      "learning_rate": 7.64199730456026e-06,
+      "loss": 0.6639,
+      "step": 21074
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.7788994312286377,
+      "learning_rate": 7.641773456179504e-06,
+      "loss": 0.6386,
+      "step": 21075
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.412914276123047,
+      "learning_rate": 7.641549600452966e-06,
+      "loss": 0.7555,
+      "step": 21076
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.65824031829834,
+      "learning_rate": 7.641325737381267e-06,
+      "loss": 0.8963,
+      "step": 21077
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 3.7071633338928223,
+      "learning_rate": 7.641101866965032e-06,
+      "loss": 0.6265,
+      "step": 21078
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.4332075119018555,
+      "learning_rate": 7.640877989204882e-06,
+      "loss": 0.5714,
+      "step": 21079
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.2155351638793945,
+      "learning_rate": 7.640654104101439e-06,
+      "loss": 0.6739,
+      "step": 21080
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.7996331453323364,
+      "learning_rate": 7.640430211655325e-06,
+      "loss": 0.7248,
+      "step": 21081
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 5.86279296875,
+      "learning_rate": 7.640206311867165e-06,
+      "loss": 0.6677,
+      "step": 21082
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 3.701063871383667,
+      "learning_rate": 7.639982404737581e-06,
+      "loss": 1.0228,
+      "step": 21083
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 3.7989068031311035,
+      "learning_rate": 7.639758490267192e-06,
+      "loss": 0.5854,
+      "step": 21084
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.053622245788574,
+      "learning_rate": 7.639534568456626e-06,
+      "loss": 0.6721,
+      "step": 21085
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 4.395009517669678,
+      "learning_rate": 7.639310639306503e-06,
+      "loss": 0.5352,
+      "step": 21086
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.749826669692993,
+      "learning_rate": 7.639086702817446e-06,
+      "loss": 0.8166,
+      "step": 21087
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.779425621032715,
+      "learning_rate": 7.638862758990076e-06,
+      "loss": 0.8128,
+      "step": 21088
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.4611988067626953,
+      "learning_rate": 7.63863880782502e-06,
+      "loss": 0.5817,
+      "step": 21089
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 3.1515235900878906,
+      "learning_rate": 7.638414849322895e-06,
+      "loss": 0.8359,
+      "step": 21090
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.883671522140503,
+      "learning_rate": 7.63819088348433e-06,
+      "loss": 0.5968,
+      "step": 21091
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.3315329551696777,
+      "learning_rate": 7.637966910309941e-06,
+      "loss": 0.849,
+      "step": 21092
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.0969865322113037,
+      "learning_rate": 7.637742929800356e-06,
+      "loss": 0.674,
+      "step": 21093
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.5034334659576416,
+      "learning_rate": 7.637518941956194e-06,
+      "loss": 0.878,
+      "step": 21094
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.2737770080566406,
+      "learning_rate": 7.637294946778083e-06,
+      "loss": 0.6324,
+      "step": 21095
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.2249958515167236,
+      "learning_rate": 7.63707094426664e-06,
+      "loss": 0.6861,
+      "step": 21096
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.2492363452911377,
+      "learning_rate": 7.636846934422492e-06,
+      "loss": 0.6695,
+      "step": 21097
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.4629218578338623,
+      "learning_rate": 7.636622917246259e-06,
+      "loss": 0.5366,
+      "step": 21098
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 3.6749632358551025,
+      "learning_rate": 7.636398892738569e-06,
+      "loss": 0.7922,
+      "step": 21099
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.7128639221191406,
+      "learning_rate": 7.636174860900037e-06,
+      "loss": 0.6616,
+      "step": 21100
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 3.3045806884765625,
+      "learning_rate": 7.635950821731293e-06,
+      "loss": 0.6741,
+      "step": 21101
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 3.2341535091400146,
+      "learning_rate": 7.635726775232956e-06,
+      "loss": 0.7986,
+      "step": 21102
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 4.069141387939453,
+      "learning_rate": 7.63550272140565e-06,
+      "loss": 0.5945,
+      "step": 21103
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.3932831287384033,
+      "learning_rate": 7.635278660249998e-06,
+      "loss": 0.5507,
+      "step": 21104
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.971996784210205,
+      "learning_rate": 7.635054591766624e-06,
+      "loss": 0.5913,
+      "step": 21105
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.731468677520752,
+      "learning_rate": 7.634830515956151e-06,
+      "loss": 0.9125,
+      "step": 21106
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.727865695953369,
+      "learning_rate": 7.6346064328192e-06,
+      "loss": 0.6648,
+      "step": 21107
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.63288414478302,
+      "learning_rate": 7.634382342356397e-06,
+      "loss": 0.6581,
+      "step": 21108
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.9783955812454224,
+      "learning_rate": 7.634158244568362e-06,
+      "loss": 0.4616,
+      "step": 21109
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 4.686118125915527,
+      "learning_rate": 7.633934139455719e-06,
+      "loss": 0.6732,
+      "step": 21110
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 3.718306303024292,
+      "learning_rate": 7.633710027019094e-06,
+      "loss": 0.9768,
+      "step": 21111
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.416574001312256,
+      "learning_rate": 7.633485907259107e-06,
+      "loss": 0.5478,
+      "step": 21112
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.719066858291626,
+      "learning_rate": 7.633261780176381e-06,
+      "loss": 0.9323,
+      "step": 21113
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.6679911613464355,
+      "learning_rate": 7.633037645771543e-06,
+      "loss": 0.611,
+      "step": 21114
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.9050047397613525,
+      "learning_rate": 7.632813504045212e-06,
+      "loss": 0.5647,
+      "step": 21115
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.0938827991485596,
+      "learning_rate": 7.632589354998014e-06,
+      "loss": 0.6468,
+      "step": 21116
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.875744581222534,
+      "learning_rate": 7.63236519863057e-06,
+      "loss": 0.5147,
+      "step": 21117
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 3.232846736907959,
+      "learning_rate": 7.632141034943505e-06,
+      "loss": 0.5154,
+      "step": 21118
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.188053846359253,
+      "learning_rate": 7.631916863937442e-06,
+      "loss": 0.3486,
+      "step": 21119
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.543545961380005,
+      "learning_rate": 7.631692685613004e-06,
+      "loss": 0.7463,
+      "step": 21120
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 3.6713054180145264,
+      "learning_rate": 7.631468499970815e-06,
+      "loss": 0.6415,
+      "step": 21121
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 3.61769700050354,
+      "learning_rate": 7.631244307011499e-06,
+      "loss": 0.8864,
+      "step": 21122
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.7439820766448975,
+      "learning_rate": 7.631020106735675e-06,
+      "loss": 0.8361,
+      "step": 21123
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.4998555183410645,
+      "learning_rate": 7.630795899143972e-06,
+      "loss": 0.565,
+      "step": 21124
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.033450126647949,
+      "learning_rate": 7.63057168423701e-06,
+      "loss": 0.5054,
+      "step": 21125
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.170262098312378,
+      "learning_rate": 7.630347462015414e-06,
+      "loss": 0.7179,
+      "step": 21126
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.1140527725219727,
+      "learning_rate": 7.630123232479806e-06,
+      "loss": 0.5842,
+      "step": 21127
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 3.318732976913452,
+      "learning_rate": 7.629898995630813e-06,
+      "loss": 0.4768,
+      "step": 21128
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.8597283363342285,
+      "learning_rate": 7.629674751469055e-06,
+      "loss": 0.6551,
+      "step": 21129
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.0764694213867188,
+      "learning_rate": 7.629450499995157e-06,
+      "loss": 0.5329,
+      "step": 21130
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.3694510459899902,
+      "learning_rate": 7.629226241209742e-06,
+      "loss": 0.6528,
+      "step": 21131
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 3.553028106689453,
+      "learning_rate": 7.629001975113432e-06,
+      "loss": 0.6133,
+      "step": 21132
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.7097818851470947,
+      "learning_rate": 7.628777701706854e-06,
+      "loss": 0.5638,
+      "step": 21133
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.2578532695770264,
+      "learning_rate": 7.62855342099063e-06,
+      "loss": 0.6431,
+      "step": 21134
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.8397232294082642,
+      "learning_rate": 7.628329132965382e-06,
+      "loss": 0.8058,
+      "step": 21135
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.6803765296936035,
+      "learning_rate": 7.628104837631737e-06,
+      "loss": 0.6889,
+      "step": 21136
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.581573724746704,
+      "learning_rate": 7.627880534990317e-06,
+      "loss": 0.7044,
+      "step": 21137
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.450467824935913,
+      "learning_rate": 7.627656225041744e-06,
+      "loss": 0.6723,
+      "step": 21138
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.7352652549743652,
+      "learning_rate": 7.627431907786645e-06,
+      "loss": 0.9885,
+      "step": 21139
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.8317222595214844,
+      "learning_rate": 7.627207583225642e-06,
+      "loss": 0.6619,
+      "step": 21140
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.046562910079956,
+      "learning_rate": 7.626983251359359e-06,
+      "loss": 0.5869,
+      "step": 21141
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 4.8849287033081055,
+      "learning_rate": 7.626758912188419e-06,
+      "loss": 0.7005,
+      "step": 21142
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.0826637744903564,
+      "learning_rate": 7.626534565713447e-06,
+      "loss": 0.5807,
+      "step": 21143
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.100834608078003,
+      "learning_rate": 7.626310211935066e-06,
+      "loss": 0.8651,
+      "step": 21144
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 3.2464349269866943,
+      "learning_rate": 7.6260858508539e-06,
+      "loss": 0.9242,
+      "step": 21145
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.6249070167541504,
+      "learning_rate": 7.6258614824705725e-06,
+      "loss": 0.6511,
+      "step": 21146
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.0573604106903076,
+      "learning_rate": 7.625637106785709e-06,
+      "loss": 0.7318,
+      "step": 21147
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 3.0372731685638428,
+      "learning_rate": 7.625412723799932e-06,
+      "loss": 0.7775,
+      "step": 21148
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.4351155757904053,
+      "learning_rate": 7.625188333513865e-06,
+      "loss": 0.5995,
+      "step": 21149
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.691246271133423,
+      "learning_rate": 7.624963935928134e-06,
+      "loss": 0.5404,
+      "step": 21150
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 4.4234442710876465,
+      "learning_rate": 7.62473953104336e-06,
+      "loss": 0.8313,
+      "step": 21151
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.0589776039123535,
+      "learning_rate": 7.62451511886017e-06,
+      "loss": 0.8195,
+      "step": 21152
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 3.8181262016296387,
+      "learning_rate": 7.624290699379186e-06,
+      "loss": 0.6343,
+      "step": 21153
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.6691136360168457,
+      "learning_rate": 7.624066272601033e-06,
+      "loss": 0.4595,
+      "step": 21154
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.7576632499694824,
+      "learning_rate": 7.623841838526334e-06,
+      "loss": 0.6965,
+      "step": 21155
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.8907595872879028,
+      "learning_rate": 7.623617397155714e-06,
+      "loss": 0.8724,
+      "step": 21156
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.6497993469238281,
+      "learning_rate": 7.623392948489796e-06,
+      "loss": 0.7591,
+      "step": 21157
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.078411817550659,
+      "learning_rate": 7.623168492529207e-06,
+      "loss": 0.6143,
+      "step": 21158
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.5999252796173096,
+      "learning_rate": 7.622944029274568e-06,
+      "loss": 0.6953,
+      "step": 21159
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.6350600719451904,
+      "learning_rate": 7.6227195587265034e-06,
+      "loss": 0.6365,
+      "step": 21160
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.503479242324829,
+      "learning_rate": 7.622495080885639e-06,
+      "loss": 0.7873,
+      "step": 21161
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 3.6785624027252197,
+      "learning_rate": 7.622270595752599e-06,
+      "loss": 0.8544,
+      "step": 21162
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.8409690856933594,
+      "learning_rate": 7.622046103328007e-06,
+      "loss": 0.9292,
+      "step": 21163
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 3.891207218170166,
+      "learning_rate": 7.621821603612485e-06,
+      "loss": 0.7105,
+      "step": 21164
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.926771640777588,
+      "learning_rate": 7.62159709660666e-06,
+      "loss": 0.4465,
+      "step": 21165
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.359145164489746,
+      "learning_rate": 7.621372582311155e-06,
+      "loss": 0.6756,
+      "step": 21166
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.7711637020111084,
+      "learning_rate": 7.621148060726598e-06,
+      "loss": 0.442,
+      "step": 21167
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.5008116960525513,
+      "learning_rate": 7.620923531853607e-06,
+      "loss": 0.8462,
+      "step": 21168
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.208637237548828,
+      "learning_rate": 7.62069899569281e-06,
+      "loss": 1.0143,
+      "step": 21169
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.176661491394043,
+      "learning_rate": 7.62047445224483e-06,
+      "loss": 0.5924,
+      "step": 21170
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.465338945388794,
+      "learning_rate": 7.620249901510294e-06,
+      "loss": 0.5965,
+      "step": 21171
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 3.127472400665283,
+      "learning_rate": 7.620025343489824e-06,
+      "loss": 0.8379,
+      "step": 21172
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.5702946186065674,
+      "learning_rate": 7.619800778184045e-06,
+      "loss": 0.7089,
+      "step": 21173
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 3.170644998550415,
+      "learning_rate": 7.619576205593582e-06,
+      "loss": 0.5624,
+      "step": 21174
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.8161606788635254,
+      "learning_rate": 7.619351625719058e-06,
+      "loss": 0.5966,
+      "step": 21175
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.7468771934509277,
+      "learning_rate": 7.619127038561098e-06,
+      "loss": 0.8677,
+      "step": 21176
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 3.192828416824341,
+      "learning_rate": 7.618902444120328e-06,
+      "loss": 0.6071,
+      "step": 21177
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.3125123977661133,
+      "learning_rate": 7.6186778423973706e-06,
+      "loss": 0.2729,
+      "step": 21178
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.4266961812973022,
+      "learning_rate": 7.6184532333928506e-06,
+      "loss": 0.6332,
+      "step": 21179
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.4413692951202393,
+      "learning_rate": 7.618228617107395e-06,
+      "loss": 0.8365,
+      "step": 21180
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.57185959815979,
+      "learning_rate": 7.618003993541625e-06,
+      "loss": 0.9338,
+      "step": 21181
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.4930963516235352,
+      "learning_rate": 7.6177793626961675e-06,
+      "loss": 0.7284,
+      "step": 21182
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.5839271545410156,
+      "learning_rate": 7.617554724571645e-06,
+      "loss": 0.8112,
+      "step": 21183
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 3.6292951107025146,
+      "learning_rate": 7.617330079168684e-06,
+      "loss": 0.5828,
+      "step": 21184
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.3938090801239014,
+      "learning_rate": 7.61710542648791e-06,
+      "loss": 0.561,
+      "step": 21185
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.5370157957077026,
+      "learning_rate": 7.6168807665299425e-06,
+      "loss": 0.7128,
+      "step": 21186
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.8331546783447266,
+      "learning_rate": 7.616656099295412e-06,
+      "loss": 0.7796,
+      "step": 21187
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 6.152624130249023,
+      "learning_rate": 7.616431424784942e-06,
+      "loss": 0.5121,
+      "step": 21188
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.1488497257232666,
+      "learning_rate": 7.616206742999155e-06,
+      "loss": 0.5703,
+      "step": 21189
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 3.5848727226257324,
+      "learning_rate": 7.615982053938677e-06,
+      "loss": 0.616,
+      "step": 21190
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.1754744052886963,
+      "learning_rate": 7.615757357604135e-06,
+      "loss": 0.7384,
+      "step": 21191
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.2593140602111816,
+      "learning_rate": 7.6155326539961505e-06,
+      "loss": 0.74,
+      "step": 21192
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.7460196018218994,
+      "learning_rate": 7.615307943115348e-06,
+      "loss": 0.8168,
+      "step": 21193
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.8331282138824463,
+      "learning_rate": 7.615083224962356e-06,
+      "loss": 0.6386,
+      "step": 21194
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 3.5134224891662598,
+      "learning_rate": 7.6148584995377965e-06,
+      "loss": 0.5915,
+      "step": 21195
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.571110963821411,
+      "learning_rate": 7.6146337668422944e-06,
+      "loss": 0.8913,
+      "step": 21196
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.2786262035369873,
+      "learning_rate": 7.614409026876475e-06,
+      "loss": 0.6508,
+      "step": 21197
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.7743568420410156,
+      "learning_rate": 7.6141842796409635e-06,
+      "loss": 0.6445,
+      "step": 21198
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.0075109004974365,
+      "learning_rate": 7.6139595251363874e-06,
+      "loss": 0.5992,
+      "step": 21199
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.8610103130340576,
+      "learning_rate": 7.613734763363367e-06,
+      "loss": 0.5974,
+      "step": 21200
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 3.379344940185547,
+      "learning_rate": 7.613509994322529e-06,
+      "loss": 0.6755,
+      "step": 21201
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.353846788406372,
+      "learning_rate": 7.613285218014499e-06,
+      "loss": 0.837,
+      "step": 21202
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.947901487350464,
+      "learning_rate": 7.613060434439903e-06,
+      "loss": 0.637,
+      "step": 21203
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.9858856201171875,
+      "learning_rate": 7.612835643599363e-06,
+      "loss": 0.7428,
+      "step": 21204
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.8162178993225098,
+      "learning_rate": 7.612610845493507e-06,
+      "loss": 0.6967,
+      "step": 21205
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 3.419656753540039,
+      "learning_rate": 7.612386040122957e-06,
+      "loss": 0.5572,
+      "step": 21206
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 4.4936089515686035,
+      "learning_rate": 7.612161227488341e-06,
+      "loss": 0.7795,
+      "step": 21207
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 3.4342803955078125,
+      "learning_rate": 7.611936407590286e-06,
+      "loss": 0.7368,
+      "step": 21208
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 3.012481689453125,
+      "learning_rate": 7.6117115804294105e-06,
+      "loss": 0.7129,
+      "step": 21209
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.6591392755508423,
+      "learning_rate": 7.611486746006347e-06,
+      "loss": 0.8453,
+      "step": 21210
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 3.996309518814087,
+      "learning_rate": 7.611261904321714e-06,
+      "loss": 0.688,
+      "step": 21211
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.5395398139953613,
+      "learning_rate": 7.611037055376142e-06,
+      "loss": 0.6589,
+      "step": 21212
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.2945492267608643,
+      "learning_rate": 7.610812199170252e-06,
+      "loss": 0.8156,
+      "step": 21213
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.8577381372451782,
+      "learning_rate": 7.610587335704674e-06,
+      "loss": 0.8026,
+      "step": 21214
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.0798728466033936,
+      "learning_rate": 7.6103624649800275e-06,
+      "loss": 0.665,
+      "step": 21215
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.6635173559188843,
+      "learning_rate": 7.610137586996942e-06,
+      "loss": 0.5986,
+      "step": 21216
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.7257699966430664,
+      "learning_rate": 7.609912701756043e-06,
+      "loss": 0.5081,
+      "step": 21217
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.278618097305298,
+      "learning_rate": 7.609687809257953e-06,
+      "loss": 0.651,
+      "step": 21218
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 4.402651309967041,
+      "learning_rate": 7.6094629095033e-06,
+      "loss": 0.9987,
+      "step": 21219
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 3.593052864074707,
+      "learning_rate": 7.609238002492707e-06,
+      "loss": 0.8301,
+      "step": 21220
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.0193722248077393,
+      "learning_rate": 7.609013088226801e-06,
+      "loss": 0.4774,
+      "step": 21221
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.9464319944381714,
+      "learning_rate": 7.608788166706207e-06,
+      "loss": 0.6616,
+      "step": 21222
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 3.0792086124420166,
+      "learning_rate": 7.60856323793155e-06,
+      "loss": 0.6851,
+      "step": 21223
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.0781803131103516,
+      "learning_rate": 7.608338301903458e-06,
+      "loss": 0.4669,
+      "step": 21224
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.187346935272217,
+      "learning_rate": 7.608113358622551e-06,
+      "loss": 0.4919,
+      "step": 21225
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.1502039432525635,
+      "learning_rate": 7.6078884080894585e-06,
+      "loss": 0.6434,
+      "step": 21226
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.820966958999634,
+      "learning_rate": 7.607663450304807e-06,
+      "loss": 0.488,
+      "step": 21227
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.2888858318328857,
+      "learning_rate": 7.607438485269218e-06,
+      "loss": 0.6572,
+      "step": 21228
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 4.120985984802246,
+      "learning_rate": 7.607213512983321e-06,
+      "loss": 0.723,
+      "step": 21229
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.5626262426376343,
+      "learning_rate": 7.60698853344774e-06,
+      "loss": 0.5205,
+      "step": 21230
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.9122835397720337,
+      "learning_rate": 7.606763546663101e-06,
+      "loss": 0.7811,
+      "step": 21231
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 4.000810146331787,
+      "learning_rate": 7.606538552630027e-06,
+      "loss": 0.585,
+      "step": 21232
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.995323896408081,
+      "learning_rate": 7.606313551349147e-06,
+      "loss": 0.6081,
+      "step": 21233
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.5182249546051025,
+      "learning_rate": 7.606088542821085e-06,
+      "loss": 0.7027,
+      "step": 21234
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 3.558872938156128,
+      "learning_rate": 7.605863527046466e-06,
+      "loss": 0.6046,
+      "step": 21235
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 3.4765310287475586,
+      "learning_rate": 7.60563850402592e-06,
+      "loss": 0.6309,
+      "step": 21236
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.390228748321533,
+      "learning_rate": 7.605413473760066e-06,
+      "loss": 0.7977,
+      "step": 21237
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.320718288421631,
+      "learning_rate": 7.605188436249535e-06,
+      "loss": 0.9,
+      "step": 21238
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 4.431390285491943,
+      "learning_rate": 7.604963391494951e-06,
+      "loss": 0.6926,
+      "step": 21239
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.922980546951294,
+      "learning_rate": 7.604738339496938e-06,
+      "loss": 0.7476,
+      "step": 21240
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.083958148956299,
+      "learning_rate": 7.604513280256125e-06,
+      "loss": 0.6538,
+      "step": 21241
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.474946975708008,
+      "learning_rate": 7.604288213773136e-06,
+      "loss": 0.9043,
+      "step": 21242
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.6139767169952393,
+      "learning_rate": 7.604063140048596e-06,
+      "loss": 0.8018,
+      "step": 21243
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.502408504486084,
+      "learning_rate": 7.603838059083133e-06,
+      "loss": 0.6604,
+      "step": 21244
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.6823135614395142,
+      "learning_rate": 7.603612970877372e-06,
+      "loss": 0.6692,
+      "step": 21245
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.207671642303467,
+      "learning_rate": 7.603387875431938e-06,
+      "loss": 0.653,
+      "step": 21246
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.880833148956299,
+      "learning_rate": 7.603162772747457e-06,
+      "loss": 0.6708,
+      "step": 21247
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 4.692983150482178,
+      "learning_rate": 7.602937662824556e-06,
+      "loss": 0.8798,
+      "step": 21248
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.8931546211242676,
+      "learning_rate": 7.6027125456638605e-06,
+      "loss": 0.5145,
+      "step": 21249
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.7688145637512207,
+      "learning_rate": 7.602487421265997e-06,
+      "loss": 0.7893,
+      "step": 21250
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.449608564376831,
+      "learning_rate": 7.60226228963159e-06,
+      "loss": 0.6992,
+      "step": 21251
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.6631033420562744,
+      "learning_rate": 7.602037150761268e-06,
+      "loss": 0.5561,
+      "step": 21252
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 3.3822686672210693,
+      "learning_rate": 7.601812004655653e-06,
+      "loss": 0.7257,
+      "step": 21253
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.844668984413147,
+      "learning_rate": 7.601586851315376e-06,
+      "loss": 0.5302,
+      "step": 21254
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.4889416694641113,
+      "learning_rate": 7.601361690741059e-06,
+      "loss": 0.5227,
+      "step": 21255
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 3.4849600791931152,
+      "learning_rate": 7.6011365229333304e-06,
+      "loss": 0.938,
+      "step": 21256
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.7285401821136475,
+      "learning_rate": 7.600911347892814e-06,
+      "loss": 0.4577,
+      "step": 21257
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.6289664506912231,
+      "learning_rate": 7.6006861656201395e-06,
+      "loss": 0.7238,
+      "step": 21258
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.283169746398926,
+      "learning_rate": 7.6004609761159285e-06,
+      "loss": 0.7494,
+      "step": 21259
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 3.0210249423980713,
+      "learning_rate": 7.6002357793808124e-06,
+      "loss": 0.8828,
+      "step": 21260
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.4125189781188965,
+      "learning_rate": 7.600010575415413e-06,
+      "loss": 0.7702,
+      "step": 21261
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.0658364295959473,
+      "learning_rate": 7.599785364220359e-06,
+      "loss": 0.731,
+      "step": 21262
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.9769113063812256,
+      "learning_rate": 7.599560145796277e-06,
+      "loss": 0.7768,
+      "step": 21263
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.177555799484253,
+      "learning_rate": 7.59933492014379e-06,
+      "loss": 0.5077,
+      "step": 21264
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.9935922622680664,
+      "learning_rate": 7.599109687263526e-06,
+      "loss": 0.6348,
+      "step": 21265
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.745420217514038,
+      "learning_rate": 7.598884447156113e-06,
+      "loss": 0.7312,
+      "step": 21266
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.4306259155273438,
+      "learning_rate": 7.598659199822174e-06,
+      "loss": 0.6517,
+      "step": 21267
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 3.018037796020508,
+      "learning_rate": 7.598433945262339e-06,
+      "loss": 0.7312,
+      "step": 21268
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.535804510116577,
+      "learning_rate": 7.5982086834772325e-06,
+      "loss": 0.8569,
+      "step": 21269
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.0866401195526123,
+      "learning_rate": 7.59798341446748e-06,
+      "loss": 0.5486,
+      "step": 21270
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.5379602909088135,
+      "learning_rate": 7.597758138233711e-06,
+      "loss": 0.6923,
+      "step": 21271
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.8367654085159302,
+      "learning_rate": 7.5975328547765484e-06,
+      "loss": 0.6398,
+      "step": 21272
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.137965440750122,
+      "learning_rate": 7.5973075640966185e-06,
+      "loss": 0.4437,
+      "step": 21273
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.9042391777038574,
+      "learning_rate": 7.597082266194551e-06,
+      "loss": 0.7045,
+      "step": 21274
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.725920557975769,
+      "learning_rate": 7.5968569610709705e-06,
+      "loss": 0.6802,
+      "step": 21275
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.9280166625976562,
+      "learning_rate": 7.596631648726503e-06,
+      "loss": 0.726,
+      "step": 21276
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.8327584266662598,
+      "learning_rate": 7.596406329161776e-06,
+      "loss": 0.455,
+      "step": 21277
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.1418230533599854,
+      "learning_rate": 7.5961810023774156e-06,
+      "loss": 0.4222,
+      "step": 21278
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 5.806585788726807,
+      "learning_rate": 7.595955668374048e-06,
+      "loss": 0.6388,
+      "step": 21279
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.907855749130249,
+      "learning_rate": 7.595730327152301e-06,
+      "loss": 0.7451,
+      "step": 21280
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 4.738418102264404,
+      "learning_rate": 7.5955049787128e-06,
+      "loss": 0.6973,
+      "step": 21281
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 3.436323404312134,
+      "learning_rate": 7.595279623056171e-06,
+      "loss": 0.8281,
+      "step": 21282
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.4422335624694824,
+      "learning_rate": 7.595054260183044e-06,
+      "loss": 0.7627,
+      "step": 21283
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.782233238220215,
+      "learning_rate": 7.594828890094041e-06,
+      "loss": 0.6561,
+      "step": 21284
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.903146505355835,
+      "learning_rate": 7.5946035127897914e-06,
+      "loss": 0.8507,
+      "step": 21285
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 3.9738850593566895,
+      "learning_rate": 7.594378128270922e-06,
+      "loss": 0.6459,
+      "step": 21286
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.8124244213104248,
+      "learning_rate": 7.594152736538057e-06,
+      "loss": 0.5366,
+      "step": 21287
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.052027940750122,
+      "learning_rate": 7.593927337591827e-06,
+      "loss": 0.3544,
+      "step": 21288
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.701014280319214,
+      "learning_rate": 7.593701931432855e-06,
+      "loss": 0.3619,
+      "step": 21289
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.289492130279541,
+      "learning_rate": 7.593476518061772e-06,
+      "loss": 0.7847,
+      "step": 21290
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.7622030973434448,
+      "learning_rate": 7.593251097479202e-06,
+      "loss": 0.5854,
+      "step": 21291
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.839707136154175,
+      "learning_rate": 7.59302566968577e-06,
+      "loss": 1.0244,
+      "step": 21292
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.039644241333008,
+      "learning_rate": 7.592800234682107e-06,
+      "loss": 0.7575,
+      "step": 21293
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.5927963256835938,
+      "learning_rate": 7.592574792468836e-06,
+      "loss": 0.6023,
+      "step": 21294
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.1060118675231934,
+      "learning_rate": 7.592349343046587e-06,
+      "loss": 0.7302,
+      "step": 21295
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 4.307034969329834,
+      "learning_rate": 7.592123886415985e-06,
+      "loss": 0.8109,
+      "step": 21296
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 4.053701400756836,
+      "learning_rate": 7.591898422577658e-06,
+      "loss": 0.4758,
+      "step": 21297
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.9536538124084473,
+      "learning_rate": 7.591672951532233e-06,
+      "loss": 0.5806,
+      "step": 21298
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 3.0547242164611816,
+      "learning_rate": 7.5914474732803345e-06,
+      "loss": 0.7184,
+      "step": 21299
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.7661776542663574,
+      "learning_rate": 7.591221987822592e-06,
+      "loss": 0.4501,
+      "step": 21300
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 5.302207946777344,
+      "learning_rate": 7.590996495159633e-06,
+      "loss": 0.8124,
+      "step": 21301
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.9440265893936157,
+      "learning_rate": 7.5907709952920825e-06,
+      "loss": 0.7991,
+      "step": 21302
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.2157633304595947,
+      "learning_rate": 7.5905454882205685e-06,
+      "loss": 0.5406,
+      "step": 21303
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.680393934249878,
+      "learning_rate": 7.5903199739457185e-06,
+      "loss": 0.554,
+      "step": 21304
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.637166976928711,
+      "learning_rate": 7.5900944524681595e-06,
+      "loss": 0.6892,
+      "step": 21305
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 3.586308717727661,
+      "learning_rate": 7.589868923788517e-06,
+      "loss": 0.5264,
+      "step": 21306
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.234287738800049,
+      "learning_rate": 7.58964338790742e-06,
+      "loss": 0.9993,
+      "step": 21307
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.4750218391418457,
+      "learning_rate": 7.5894178448254925e-06,
+      "loss": 0.4155,
+      "step": 21308
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.164844274520874,
+      "learning_rate": 7.589192294543366e-06,
+      "loss": 0.4436,
+      "step": 21309
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.548797607421875,
+      "learning_rate": 7.5889667370616645e-06,
+      "loss": 0.4945,
+      "step": 21310
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.5886517763137817,
+      "learning_rate": 7.588741172381018e-06,
+      "loss": 0.8144,
+      "step": 21311
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 5.11414098739624,
+      "learning_rate": 7.588515600502051e-06,
+      "loss": 0.6324,
+      "step": 21312
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.4065423011779785,
+      "learning_rate": 7.588290021425393e-06,
+      "loss": 0.6717,
+      "step": 21313
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 3.8846426010131836,
+      "learning_rate": 7.58806443515167e-06,
+      "loss": 0.6848,
+      "step": 21314
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.3176658153533936,
+      "learning_rate": 7.587838841681508e-06,
+      "loss": 0.6293,
+      "step": 21315
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 3.41329026222229,
+      "learning_rate": 7.587613241015536e-06,
+      "loss": 0.7207,
+      "step": 21316
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.8366732597351074,
+      "learning_rate": 7.587387633154382e-06,
+      "loss": 0.6977,
+      "step": 21317
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 3.2151854038238525,
+      "learning_rate": 7.587162018098669e-06,
+      "loss": 0.7183,
+      "step": 21318
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.9975428581237793,
+      "learning_rate": 7.58693639584903e-06,
+      "loss": 0.415,
+      "step": 21319
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 3.7102773189544678,
+      "learning_rate": 7.5867107664060905e-06,
+      "loss": 0.6295,
+      "step": 21320
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 3.037698268890381,
+      "learning_rate": 7.586485129770476e-06,
+      "loss": 0.7015,
+      "step": 21321
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 3.7152366638183594,
+      "learning_rate": 7.586259485942817e-06,
+      "loss": 0.5265,
+      "step": 21322
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 4.2465410232543945,
+      "learning_rate": 7.586033834923738e-06,
+      "loss": 0.378,
+      "step": 21323
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.7109872102737427,
+      "learning_rate": 7.585808176713868e-06,
+      "loss": 0.6592,
+      "step": 21324
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 3.2848408222198486,
+      "learning_rate": 7.585582511313834e-06,
+      "loss": 0.7578,
+      "step": 21325
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.9535303115844727,
+      "learning_rate": 7.585356838724263e-06,
+      "loss": 0.5198,
+      "step": 21326
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.7103097438812256,
+      "learning_rate": 7.5851311589457845e-06,
+      "loss": 0.5846,
+      "step": 21327
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.7081880569458008,
+      "learning_rate": 7.584905471979023e-06,
+      "loss": 0.7053,
+      "step": 21328
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.5178871154785156,
+      "learning_rate": 7.584679777824609e-06,
+      "loss": 0.6593,
+      "step": 21329
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.9450236558914185,
+      "learning_rate": 7.584454076483168e-06,
+      "loss": 0.4988,
+      "step": 21330
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.264350175857544,
+      "learning_rate": 7.584228367955329e-06,
+      "loss": 0.7861,
+      "step": 21331
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.271697759628296,
+      "learning_rate": 7.584002652241717e-06,
+      "loss": 0.559,
+      "step": 21332
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.972856879234314,
+      "learning_rate": 7.583776929342965e-06,
+      "loss": 0.6447,
+      "step": 21333
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.183814525604248,
+      "learning_rate": 7.5835511992596954e-06,
+      "loss": 0.5614,
+      "step": 21334
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.822885036468506,
+      "learning_rate": 7.583325461992538e-06,
+      "loss": 0.6923,
+      "step": 21335
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.7741284370422363,
+      "learning_rate": 7.58309971754212e-06,
+      "loss": 0.7148,
+      "step": 21336
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.6011550426483154,
+      "learning_rate": 7.58287396590907e-06,
+      "loss": 0.7616,
+      "step": 21337
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.316654920578003,
+      "learning_rate": 7.582648207094013e-06,
+      "loss": 0.786,
+      "step": 21338
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 3.0655627250671387,
+      "learning_rate": 7.582422441097581e-06,
+      "loss": 1.0892,
+      "step": 21339
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.5225049257278442,
+      "learning_rate": 7.582196667920399e-06,
+      "loss": 0.4583,
+      "step": 21340
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.298642873764038,
+      "learning_rate": 7.581970887563095e-06,
+      "loss": 0.749,
+      "step": 21341
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.615938663482666,
+      "learning_rate": 7.581745100026299e-06,
+      "loss": 0.7685,
+      "step": 21342
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.39532732963562,
+      "learning_rate": 7.581519305310634e-06,
+      "loss": 0.6764,
+      "step": 21343
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.7607372999191284,
+      "learning_rate": 7.581293503416734e-06,
+      "loss": 0.7188,
+      "step": 21344
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.0193514823913574,
+      "learning_rate": 7.581067694345221e-06,
+      "loss": 0.5965,
+      "step": 21345
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.9820733070373535,
+      "learning_rate": 7.580841878096728e-06,
+      "loss": 0.5842,
+      "step": 21346
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.065446138381958,
+      "learning_rate": 7.580616054671879e-06,
+      "loss": 0.732,
+      "step": 21347
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 3.997748374938965,
+      "learning_rate": 7.580390224071302e-06,
+      "loss": 0.7606,
+      "step": 21348
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.1817781925201416,
+      "learning_rate": 7.580164386295629e-06,
+      "loss": 0.6938,
+      "step": 21349
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 3.8345425128936768,
+      "learning_rate": 7.579938541345484e-06,
+      "loss": 0.6686,
+      "step": 21350
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 3.5901987552642822,
+      "learning_rate": 7.579712689221498e-06,
+      "loss": 0.9282,
+      "step": 21351
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.8838863372802734,
+      "learning_rate": 7.579486829924296e-06,
+      "loss": 0.6336,
+      "step": 21352
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.6440482139587402,
+      "learning_rate": 7.579260963454508e-06,
+      "loss": 0.6,
+      "step": 21353
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 3.2533040046691895,
+      "learning_rate": 7.579035089812761e-06,
+      "loss": 0.6574,
+      "step": 21354
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.6705015897750854,
+      "learning_rate": 7.578809208999683e-06,
+      "loss": 0.6677,
+      "step": 21355
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.5669089555740356,
+      "learning_rate": 7.5785833210159044e-06,
+      "loss": 0.4194,
+      "step": 21356
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.6525698900222778,
+      "learning_rate": 7.578357425862051e-06,
+      "loss": 0.5564,
+      "step": 21357
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 3.6081442832946777,
+      "learning_rate": 7.57813152353875e-06,
+      "loss": 1.0931,
+      "step": 21358
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.485284447669983,
+      "learning_rate": 7.577905614046632e-06,
+      "loss": 0.5261,
+      "step": 21359
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.5452041625976562,
+      "learning_rate": 7.577679697386324e-06,
+      "loss": 0.7574,
+      "step": 21360
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.0416669845581055,
+      "learning_rate": 7.5774537735584545e-06,
+      "loss": 0.5393,
+      "step": 21361
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.7729804515838623,
+      "learning_rate": 7.577227842563652e-06,
+      "loss": 0.8218,
+      "step": 21362
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.90938663482666,
+      "learning_rate": 7.5770019044025445e-06,
+      "loss": 0.393,
+      "step": 21363
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.7321147918701172,
+      "learning_rate": 7.576775959075759e-06,
+      "loss": 0.5093,
+      "step": 21364
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.7113556861877441,
+      "learning_rate": 7.576550006583927e-06,
+      "loss": 0.5811,
+      "step": 21365
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.128267288208008,
+      "learning_rate": 7.576324046927672e-06,
+      "loss": 0.6992,
+      "step": 21366
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.8590868711471558,
+      "learning_rate": 7.5760980801076265e-06,
+      "loss": 0.6926,
+      "step": 21367
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 3.0319294929504395,
+      "learning_rate": 7.575872106124418e-06,
+      "loss": 0.9058,
+      "step": 21368
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.5808231830596924,
+      "learning_rate": 7.57564612497867e-06,
+      "loss": 0.6986,
+      "step": 21369
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.760836124420166,
+      "learning_rate": 7.575420136671019e-06,
+      "loss": 0.7053,
+      "step": 21370
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.5090872049331665,
+      "learning_rate": 7.575194141202089e-06,
+      "loss": 0.7225,
+      "step": 21371
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 4.2061448097229,
+      "learning_rate": 7.574968138572507e-06,
+      "loss": 0.6719,
+      "step": 21372
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.6838494539260864,
+      "learning_rate": 7.574742128782906e-06,
+      "loss": 0.6805,
+      "step": 21373
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.792860507965088,
+      "learning_rate": 7.57451611183391e-06,
+      "loss": 0.5204,
+      "step": 21374
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 3.008782148361206,
+      "learning_rate": 7.574290087726148e-06,
+      "loss": 0.5236,
+      "step": 21375
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.422597646713257,
+      "learning_rate": 7.574064056460251e-06,
+      "loss": 0.6986,
+      "step": 21376
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.0127031803131104,
+      "learning_rate": 7.573838018036846e-06,
+      "loss": 0.4386,
+      "step": 21377
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.5789875984191895,
+      "learning_rate": 7.573611972456561e-06,
+      "loss": 0.7863,
+      "step": 21378
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 3.859248161315918,
+      "learning_rate": 7.573385919720025e-06,
+      "loss": 0.6779,
+      "step": 21379
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.5424189567565918,
+      "learning_rate": 7.573159859827868e-06,
+      "loss": 0.4945,
+      "step": 21380
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 3.538836717605591,
+      "learning_rate": 7.572933792780717e-06,
+      "loss": 0.8673,
+      "step": 21381
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.5753298997879028,
+      "learning_rate": 7.572707718579201e-06,
+      "loss": 0.6329,
+      "step": 21382
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 3.455077648162842,
+      "learning_rate": 7.572481637223949e-06,
+      "loss": 0.7043,
+      "step": 21383
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.74637770652771,
+      "learning_rate": 7.572255548715588e-06,
+      "loss": 0.6533,
+      "step": 21384
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.352991819381714,
+      "learning_rate": 7.5720294530547485e-06,
+      "loss": 0.6324,
+      "step": 21385
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.033825397491455,
+      "learning_rate": 7.57180335024206e-06,
+      "loss": 0.746,
+      "step": 21386
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.7931344509124756,
+      "learning_rate": 7.571577240278147e-06,
+      "loss": 0.5807,
+      "step": 21387
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.895577073097229,
+      "learning_rate": 7.571351123163644e-06,
+      "loss": 0.558,
+      "step": 21388
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 3.6218185424804688,
+      "learning_rate": 7.571124998899173e-06,
+      "loss": 0.7474,
+      "step": 21389
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.7660610675811768,
+      "learning_rate": 7.57089886748537e-06,
+      "loss": 0.5318,
+      "step": 21390
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.351956605911255,
+      "learning_rate": 7.570672728922859e-06,
+      "loss": 0.4256,
+      "step": 21391
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.8883854150772095,
+      "learning_rate": 7.5704465832122695e-06,
+      "loss": 0.5944,
+      "step": 21392
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 3.489297866821289,
+      "learning_rate": 7.5702204303542315e-06,
+      "loss": 0.5043,
+      "step": 21393
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.5289254188537598,
+      "learning_rate": 7.569994270349374e-06,
+      "loss": 0.6432,
+      "step": 21394
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.7472736835479736,
+      "learning_rate": 7.569768103198325e-06,
+      "loss": 0.7551,
+      "step": 21395
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.7115663290023804,
+      "learning_rate": 7.569541928901714e-06,
+      "loss": 0.7944,
+      "step": 21396
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 3.012439012527466,
+      "learning_rate": 7.569315747460169e-06,
+      "loss": 0.5754,
+      "step": 21397
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.597429871559143,
+      "learning_rate": 7.569089558874318e-06,
+      "loss": 0.6101,
+      "step": 21398
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.1676065921783447,
+      "learning_rate": 7.568863363144792e-06,
+      "loss": 0.7427,
+      "step": 21399
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 3.6104142665863037,
+      "learning_rate": 7.568637160272219e-06,
+      "loss": 0.8816,
+      "step": 21400
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.3095399141311646,
+      "learning_rate": 7.56841095025723e-06,
+      "loss": 0.6673,
+      "step": 21401
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 5.552069664001465,
+      "learning_rate": 7.568184733100452e-06,
+      "loss": 0.8746,
+      "step": 21402
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 3.0966804027557373,
+      "learning_rate": 7.567958508802513e-06,
+      "loss": 0.5439,
+      "step": 21403
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.282973289489746,
+      "learning_rate": 7.567732277364043e-06,
+      "loss": 0.5452,
+      "step": 21404
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 4.073455333709717,
+      "learning_rate": 7.567506038785675e-06,
+      "loss": 1.0152,
+      "step": 21405
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.811026096343994,
+      "learning_rate": 7.5672797930680305e-06,
+      "loss": 0.6721,
+      "step": 21406
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.4445478916168213,
+      "learning_rate": 7.567053540211744e-06,
+      "loss": 0.7729,
+      "step": 21407
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.586757183074951,
+      "learning_rate": 7.5668272802174435e-06,
+      "loss": 0.7384,
+      "step": 21408
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.318345308303833,
+      "learning_rate": 7.566601013085757e-06,
+      "loss": 0.6245,
+      "step": 21409
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.1743416786193848,
+      "learning_rate": 7.566374738817316e-06,
+      "loss": 0.5863,
+      "step": 21410
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.7347187995910645,
+      "learning_rate": 7.5661484574127475e-06,
+      "loss": 0.556,
+      "step": 21411
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.216068744659424,
+      "learning_rate": 7.565922168872682e-06,
+      "loss": 0.456,
+      "step": 21412
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.3726956844329834,
+      "learning_rate": 7.565695873197747e-06,
+      "loss": 0.7086,
+      "step": 21413
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.9365851879119873,
+      "learning_rate": 7.565469570388574e-06,
+      "loss": 0.6597,
+      "step": 21414
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.1102771759033203,
+      "learning_rate": 7.5652432604457916e-06,
+      "loss": 0.7023,
+      "step": 21415
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.976646661758423,
+      "learning_rate": 7.565016943370027e-06,
+      "loss": 0.3753,
+      "step": 21416
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.5087674856185913,
+      "learning_rate": 7.56479061916191e-06,
+      "loss": 0.4993,
+      "step": 21417
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.6324212551116943,
+      "learning_rate": 7.5645642878220735e-06,
+      "loss": 0.703,
+      "step": 21418
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.8989346027374268,
+      "learning_rate": 7.564337949351145e-06,
+      "loss": 0.585,
+      "step": 21419
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.729942798614502,
+      "learning_rate": 7.564111603749752e-06,
+      "loss": 0.9066,
+      "step": 21420
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.4661445617675781,
+      "learning_rate": 7.563885251018526e-06,
+      "loss": 0.59,
+      "step": 21421
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.2176413536071777,
+      "learning_rate": 7.563658891158095e-06,
+      "loss": 0.74,
+      "step": 21422
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.6601316928863525,
+      "learning_rate": 7.563432524169087e-06,
+      "loss": 0.6076,
+      "step": 21423
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 4.579569339752197,
+      "learning_rate": 7.563206150052136e-06,
+      "loss": 0.6676,
+      "step": 21424
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.5081746578216553,
+      "learning_rate": 7.5629797688078676e-06,
+      "loss": 0.8097,
+      "step": 21425
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 3.099705457687378,
+      "learning_rate": 7.562753380436914e-06,
+      "loss": 0.6078,
+      "step": 21426
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.757692813873291,
+      "learning_rate": 7.562526984939903e-06,
+      "loss": 0.6475,
+      "step": 21427
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.8880596160888672,
+      "learning_rate": 7.562300582317463e-06,
+      "loss": 0.7988,
+      "step": 21428
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.675874948501587,
+      "learning_rate": 7.562074172570226e-06,
+      "loss": 0.5303,
+      "step": 21429
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.7933889627456665,
+      "learning_rate": 7.56184775569882e-06,
+      "loss": 0.575,
+      "step": 21430
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 3.3883378505706787,
+      "learning_rate": 7.561621331703877e-06,
+      "loss": 0.8279,
+      "step": 21431
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.7580628395080566,
+      "learning_rate": 7.561394900586022e-06,
+      "loss": 0.7159,
+      "step": 21432
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.0648860931396484,
+      "learning_rate": 7.561168462345889e-06,
+      "loss": 0.635,
+      "step": 21433
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.747349977493286,
+      "learning_rate": 7.560942016984106e-06,
+      "loss": 0.6145,
+      "step": 21434
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.9484797716140747,
+      "learning_rate": 7.560715564501302e-06,
+      "loss": 0.5445,
+      "step": 21435
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.466982841491699,
+      "learning_rate": 7.560489104898107e-06,
+      "loss": 0.7249,
+      "step": 21436
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.1921467781066895,
+      "learning_rate": 7.560262638175151e-06,
+      "loss": 0.8026,
+      "step": 21437
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.203507661819458,
+      "learning_rate": 7.560036164333064e-06,
+      "loss": 0.7715,
+      "step": 21438
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.2736010551452637,
+      "learning_rate": 7.559809683372476e-06,
+      "loss": 0.7035,
+      "step": 21439
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.8555405139923096,
+      "learning_rate": 7.559583195294013e-06,
+      "loss": 0.8035,
+      "step": 21440
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.948065996170044,
+      "learning_rate": 7.559356700098313e-06,
+      "loss": 0.5711,
+      "step": 21441
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.615377187728882,
+      "learning_rate": 7.559130197785998e-06,
+      "loss": 0.8139,
+      "step": 21442
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.3197202682495117,
+      "learning_rate": 7.558903688357701e-06,
+      "loss": 0.5502,
+      "step": 21443
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.8929009437561035,
+      "learning_rate": 7.558677171814051e-06,
+      "loss": 0.5854,
+      "step": 21444
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.8294074535369873,
+      "learning_rate": 7.558450648155678e-06,
+      "loss": 0.5773,
+      "step": 21445
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.425018310546875,
+      "learning_rate": 7.558224117383213e-06,
+      "loss": 0.875,
+      "step": 21446
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.498467206954956,
+      "learning_rate": 7.557997579497285e-06,
+      "loss": 0.7431,
+      "step": 21447
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.020859479904175,
+      "learning_rate": 7.557771034498524e-06,
+      "loss": 0.5777,
+      "step": 21448
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.0291128158569336,
+      "learning_rate": 7.55754448238756e-06,
+      "loss": 0.5425,
+      "step": 21449
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.5910961627960205,
+      "learning_rate": 7.557317923165022e-06,
+      "loss": 0.5192,
+      "step": 21450
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.045290470123291,
+      "learning_rate": 7.55709135683154e-06,
+      "loss": 0.8461,
+      "step": 21451
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.9720687866210938,
+      "learning_rate": 7.556864783387747e-06,
+      "loss": 0.6249,
+      "step": 21452
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.637991428375244,
+      "learning_rate": 7.556638202834271e-06,
+      "loss": 0.8005,
+      "step": 21453
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.4027867317199707,
+      "learning_rate": 7.5564116151717415e-06,
+      "loss": 0.5399,
+      "step": 21454
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.13208270072937,
+      "learning_rate": 7.556185020400788e-06,
+      "loss": 0.6211,
+      "step": 21455
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.4206955432891846,
+      "learning_rate": 7.555958418522042e-06,
+      "loss": 0.903,
+      "step": 21456
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.0844686031341553,
+      "learning_rate": 7.555731809536133e-06,
+      "loss": 0.855,
+      "step": 21457
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.97005033493042,
+      "learning_rate": 7.555505193443693e-06,
+      "loss": 0.8112,
+      "step": 21458
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.568751335144043,
+      "learning_rate": 7.555278570245348e-06,
+      "loss": 0.4105,
+      "step": 21459
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 6.036579132080078,
+      "learning_rate": 7.555051939941731e-06,
+      "loss": 0.7318,
+      "step": 21460
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 3.259596109390259,
+      "learning_rate": 7.554825302533472e-06,
+      "loss": 0.9888,
+      "step": 21461
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.0862300395965576,
+      "learning_rate": 7.554598658021202e-06,
+      "loss": 0.8025,
+      "step": 21462
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.2453320026397705,
+      "learning_rate": 7.554372006405548e-06,
+      "loss": 0.6358,
+      "step": 21463
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.952050805091858,
+      "learning_rate": 7.554145347687143e-06,
+      "loss": 0.7709,
+      "step": 21464
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.1508584022521973,
+      "learning_rate": 7.553918681866617e-06,
+      "loss": 0.7206,
+      "step": 21465
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 3.1777737140655518,
+      "learning_rate": 7.5536920089446e-06,
+      "loss": 0.4963,
+      "step": 21466
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.7089821100234985,
+      "learning_rate": 7.553465328921723e-06,
+      "loss": 0.8861,
+      "step": 21467
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.987426996231079,
+      "learning_rate": 7.553238641798615e-06,
+      "loss": 0.6939,
+      "step": 21468
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.280042886734009,
+      "learning_rate": 7.553011947575906e-06,
+      "loss": 0.7327,
+      "step": 21469
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 3.322246789932251,
+      "learning_rate": 7.552785246254226e-06,
+      "loss": 0.7255,
+      "step": 21470
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.838895082473755,
+      "learning_rate": 7.552558537834208e-06,
+      "loss": 1.0686,
+      "step": 21471
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.2536540031433105,
+      "learning_rate": 7.552331822316481e-06,
+      "loss": 0.7086,
+      "step": 21472
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.923155426979065,
+      "learning_rate": 7.5521050997016745e-06,
+      "loss": 0.5697,
+      "step": 21473
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 3.988983392715454,
+      "learning_rate": 7.5518783699904215e-06,
+      "loss": 0.5403,
+      "step": 21474
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.1620731353759766,
+      "learning_rate": 7.55165163318335e-06,
+      "loss": 0.7229,
+      "step": 21475
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.0979342460632324,
+      "learning_rate": 7.551424889281091e-06,
+      "loss": 0.5339,
+      "step": 21476
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.471140146255493,
+      "learning_rate": 7.551198138284275e-06,
+      "loss": 0.6584,
+      "step": 21477
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.8951884508132935,
+      "learning_rate": 7.550971380193533e-06,
+      "loss": 0.7074,
+      "step": 21478
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.0247092247009277,
+      "learning_rate": 7.550744615009493e-06,
+      "loss": 0.9962,
+      "step": 21479
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.4912680387496948,
+      "learning_rate": 7.5505178427327895e-06,
+      "loss": 0.5957,
+      "step": 21480
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.6121232509613037,
+      "learning_rate": 7.550291063364053e-06,
+      "loss": 0.5171,
+      "step": 21481
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.122260570526123,
+      "learning_rate": 7.55006427690391e-06,
+      "loss": 0.6398,
+      "step": 21482
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.907405138015747,
+      "learning_rate": 7.549837483352994e-06,
+      "loss": 0.8071,
+      "step": 21483
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.6621023416519165,
+      "learning_rate": 7.549610682711936e-06,
+      "loss": 0.7508,
+      "step": 21484
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.8050602674484253,
+      "learning_rate": 7.5493838749813646e-06,
+      "loss": 0.5672,
+      "step": 21485
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.9439053535461426,
+      "learning_rate": 7.549157060161912e-06,
+      "loss": 0.7845,
+      "step": 21486
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.058872938156128,
+      "learning_rate": 7.5489302382542095e-06,
+      "loss": 0.6276,
+      "step": 21487
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 4.001720905303955,
+      "learning_rate": 7.5487034092588865e-06,
+      "loss": 0.8058,
+      "step": 21488
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.450606107711792,
+      "learning_rate": 7.548476573176573e-06,
+      "loss": 0.7928,
+      "step": 21489
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.2242166996002197,
+      "learning_rate": 7.548249730007903e-06,
+      "loss": 0.7606,
+      "step": 21490
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.8811777830123901,
+      "learning_rate": 7.548022879753503e-06,
+      "loss": 0.6425,
+      "step": 21491
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.2171506881713867,
+      "learning_rate": 7.547796022414007e-06,
+      "loss": 0.6589,
+      "step": 21492
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.6993070840835571,
+      "learning_rate": 7.5475691579900445e-06,
+      "loss": 0.5418,
+      "step": 21493
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.989243745803833,
+      "learning_rate": 7.547342286482246e-06,
+      "loss": 0.6916,
+      "step": 21494
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 3.4172165393829346,
+      "learning_rate": 7.547115407891244e-06,
+      "loss": 0.7171,
+      "step": 21495
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.828469753265381,
+      "learning_rate": 7.5468885222176665e-06,
+      "loss": 0.8321,
+      "step": 21496
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.948742628097534,
+      "learning_rate": 7.546661629462148e-06,
+      "loss": 0.7238,
+      "step": 21497
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.5779080390930176,
+      "learning_rate": 7.546434729625316e-06,
+      "loss": 0.6053,
+      "step": 21498
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.9597907066345215,
+      "learning_rate": 7.546207822707804e-06,
+      "loss": 0.5265,
+      "step": 21499
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.553638219833374,
+      "learning_rate": 7.545980908710242e-06,
+      "loss": 0.4838,
+      "step": 21500
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.826974630355835,
+      "learning_rate": 7.545753987633259e-06,
+      "loss": 0.7402,
+      "step": 21501
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.83316969871521,
+      "learning_rate": 7.545527059477489e-06,
+      "loss": 0.7668,
+      "step": 21502
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.4007859230041504,
+      "learning_rate": 7.545300124243562e-06,
+      "loss": 0.6327,
+      "step": 21503
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.3021082878112793,
+      "learning_rate": 7.545073181932109e-06,
+      "loss": 0.87,
+      "step": 21504
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 4.106975078582764,
+      "learning_rate": 7.5448462325437614e-06,
+      "loss": 0.9835,
+      "step": 21505
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.1496307849884033,
+      "learning_rate": 7.544619276079148e-06,
+      "loss": 0.8853,
+      "step": 21506
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.4596139192581177,
+      "learning_rate": 7.544392312538903e-06,
+      "loss": 0.7741,
+      "step": 21507
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.8525030612945557,
+      "learning_rate": 7.544165341923655e-06,
+      "loss": 0.9399,
+      "step": 21508
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.6498968601226807,
+      "learning_rate": 7.543938364234038e-06,
+      "loss": 0.8205,
+      "step": 21509
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 3.664552927017212,
+      "learning_rate": 7.5437113794706794e-06,
+      "loss": 0.7035,
+      "step": 21510
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.2712544202804565,
+      "learning_rate": 7.5434843876342135e-06,
+      "loss": 0.2971,
+      "step": 21511
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.8301748037338257,
+      "learning_rate": 7.543257388725269e-06,
+      "loss": 0.6779,
+      "step": 21512
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.8200387954711914,
+      "learning_rate": 7.543030382744481e-06,
+      "loss": 0.818,
+      "step": 21513
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.8247456550598145,
+      "learning_rate": 7.542803369692476e-06,
+      "loss": 0.5924,
+      "step": 21514
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 3.074075937271118,
+      "learning_rate": 7.542576349569889e-06,
+      "loss": 0.6709,
+      "step": 21515
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 4.519360542297363,
+      "learning_rate": 7.5423493223773495e-06,
+      "loss": 0.8624,
+      "step": 21516
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.0273873805999756,
+      "learning_rate": 7.5421222881154876e-06,
+      "loss": 0.7695,
+      "step": 21517
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 5.636937618255615,
+      "learning_rate": 7.541895246784936e-06,
+      "loss": 0.7557,
+      "step": 21518
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.207686424255371,
+      "learning_rate": 7.541668198386327e-06,
+      "loss": 0.7145,
+      "step": 21519
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.6203728914260864,
+      "learning_rate": 7.541441142920289e-06,
+      "loss": 0.811,
+      "step": 21520
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.861990213394165,
+      "learning_rate": 7.541214080387457e-06,
+      "loss": 0.6628,
+      "step": 21521
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 3.3340325355529785,
+      "learning_rate": 7.540987010788461e-06,
+      "loss": 0.7273,
+      "step": 21522
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.5250394344329834,
+      "learning_rate": 7.54075993412393e-06,
+      "loss": 0.6828,
+      "step": 21523
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.257650375366211,
+      "learning_rate": 7.5405328503945e-06,
+      "loss": 0.6291,
+      "step": 21524
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.3878989219665527,
+      "learning_rate": 7.540305759600797e-06,
+      "loss": 0.4525,
+      "step": 21525
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.7054388523101807,
+      "learning_rate": 7.540078661743457e-06,
+      "loss": 0.4741,
+      "step": 21526
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.9985765218734741,
+      "learning_rate": 7.53985155682311e-06,
+      "loss": 0.74,
+      "step": 21527
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.818652629852295,
+      "learning_rate": 7.539624444840386e-06,
+      "loss": 0.626,
+      "step": 21528
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 3.249854564666748,
+      "learning_rate": 7.5393973257959184e-06,
+      "loss": 0.5072,
+      "step": 21529
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.9792304039001465,
+      "learning_rate": 7.53917019969034e-06,
+      "loss": 0.8544,
+      "step": 21530
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.2075393199920654,
+      "learning_rate": 7.538943066524277e-06,
+      "loss": 0.7058,
+      "step": 21531
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.7203655242919922,
+      "learning_rate": 7.5387159262983655e-06,
+      "loss": 0.4256,
+      "step": 21532
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.9012322425842285,
+      "learning_rate": 7.538488779013237e-06,
+      "loss": 0.8128,
+      "step": 21533
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 3.3336291313171387,
+      "learning_rate": 7.538261624669523e-06,
+      "loss": 0.6726,
+      "step": 21534
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.229048252105713,
+      "learning_rate": 7.538034463267853e-06,
+      "loss": 0.3715,
+      "step": 21535
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 3.2293407917022705,
+      "learning_rate": 7.53780729480886e-06,
+      "loss": 0.8944,
+      "step": 21536
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.168168067932129,
+      "learning_rate": 7.537580119293175e-06,
+      "loss": 0.6897,
+      "step": 21537
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.2439701557159424,
+      "learning_rate": 7.537352936721432e-06,
+      "loss": 0.7009,
+      "step": 21538
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.5325191020965576,
+      "learning_rate": 7.537125747094259e-06,
+      "loss": 0.6692,
+      "step": 21539
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.495138645172119,
+      "learning_rate": 7.536898550412291e-06,
+      "loss": 0.4185,
+      "step": 21540
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.5694923400878906,
+      "learning_rate": 7.536671346676158e-06,
+      "loss": 0.7571,
+      "step": 21541
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.203303575515747,
+      "learning_rate": 7.536444135886491e-06,
+      "loss": 0.5615,
+      "step": 21542
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.8961281776428223,
+      "learning_rate": 7.536216918043926e-06,
+      "loss": 0.5884,
+      "step": 21543
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 3.381880521774292,
+      "learning_rate": 7.535989693149091e-06,
+      "loss": 0.7012,
+      "step": 21544
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 3.1783673763275146,
+      "learning_rate": 7.535762461202617e-06,
+      "loss": 1.0139,
+      "step": 21545
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.636728048324585,
+      "learning_rate": 7.535535222205137e-06,
+      "loss": 0.5932,
+      "step": 21546
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.6768293380737305,
+      "learning_rate": 7.5353079761572845e-06,
+      "loss": 0.745,
+      "step": 21547
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.5614776611328125,
+      "learning_rate": 7.535080723059691e-06,
+      "loss": 0.5159,
+      "step": 21548
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.2045702934265137,
+      "learning_rate": 7.534853462912988e-06,
+      "loss": 0.6571,
+      "step": 21549
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 3.203449010848999,
+      "learning_rate": 7.534626195717807e-06,
+      "loss": 0.5257,
+      "step": 21550
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.907119035720825,
+      "learning_rate": 7.534398921474779e-06,
+      "loss": 0.4954,
+      "step": 21551
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.383514165878296,
+      "learning_rate": 7.534171640184536e-06,
+      "loss": 0.8249,
+      "step": 21552
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 4.180952072143555,
+      "learning_rate": 7.533944351847713e-06,
+      "loss": 0.5375,
+      "step": 21553
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.240567922592163,
+      "learning_rate": 7.53371705646494e-06,
+      "loss": 0.8386,
+      "step": 21554
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 3.3466532230377197,
+      "learning_rate": 7.533489754036847e-06,
+      "loss": 0.7993,
+      "step": 21555
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 4.063970565795898,
+      "learning_rate": 7.53326244456407e-06,
+      "loss": 0.8901,
+      "step": 21556
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 3.189168930053711,
+      "learning_rate": 7.533035128047239e-06,
+      "loss": 0.7568,
+      "step": 21557
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.0794312953948975,
+      "learning_rate": 7.5328078044869855e-06,
+      "loss": 0.7018,
+      "step": 21558
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.8932275772094727,
+      "learning_rate": 7.5325804738839436e-06,
+      "loss": 0.9335,
+      "step": 21559
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 3.126844644546509,
+      "learning_rate": 7.5323531362387425e-06,
+      "loss": 0.9046,
+      "step": 21560
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 6.372910022735596,
+      "learning_rate": 7.532125791552016e-06,
+      "loss": 0.6907,
+      "step": 21561
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 3.057450532913208,
+      "learning_rate": 7.531898439824397e-06,
+      "loss": 0.9776,
+      "step": 21562
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.043214797973633,
+      "learning_rate": 7.531671081056516e-06,
+      "loss": 0.5998,
+      "step": 21563
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.7417309284210205,
+      "learning_rate": 7.531443715249007e-06,
+      "loss": 0.6456,
+      "step": 21564
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.7571613788604736,
+      "learning_rate": 7.531216342402501e-06,
+      "loss": 0.7792,
+      "step": 21565
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.119497060775757,
+      "learning_rate": 7.530988962517631e-06,
+      "loss": 0.4145,
+      "step": 21566
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.0675289630889893,
+      "learning_rate": 7.530761575595029e-06,
+      "loss": 0.6439,
+      "step": 21567
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.4092695713043213,
+      "learning_rate": 7.530534181635328e-06,
+      "loss": 0.6494,
+      "step": 21568
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.6898915767669678,
+      "learning_rate": 7.530306780639157e-06,
+      "loss": 0.7706,
+      "step": 21569
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.787187337875366,
+      "learning_rate": 7.530079372607152e-06,
+      "loss": 0.5278,
+      "step": 21570
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.270846366882324,
+      "learning_rate": 7.529851957539944e-06,
+      "loss": 0.7733,
+      "step": 21571
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.7998526096343994,
+      "learning_rate": 7.529624535438165e-06,
+      "loss": 0.9041,
+      "step": 21572
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.4548921585083008,
+      "learning_rate": 7.529397106302447e-06,
+      "loss": 0.6272,
+      "step": 21573
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 3.4243922233581543,
+      "learning_rate": 7.529169670133425e-06,
+      "loss": 0.7597,
+      "step": 21574
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.7352626323699951,
+      "learning_rate": 7.528942226931728e-06,
+      "loss": 0.6687,
+      "step": 21575
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 3.7132139205932617,
+      "learning_rate": 7.528714776697991e-06,
+      "loss": 0.7222,
+      "step": 21576
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 4.130342960357666,
+      "learning_rate": 7.528487319432846e-06,
+      "loss": 0.5754,
+      "step": 21577
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 4.095432758331299,
+      "learning_rate": 7.528259855136924e-06,
+      "loss": 0.7039,
+      "step": 21578
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.2691593170166016,
+      "learning_rate": 7.5280323838108596e-06,
+      "loss": 0.6427,
+      "step": 21579
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.958646774291992,
+      "learning_rate": 7.527804905455282e-06,
+      "loss": 0.5312,
+      "step": 21580
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.3836212158203125,
+      "learning_rate": 7.5275774200708284e-06,
+      "loss": 0.5707,
+      "step": 21581
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.6913530826568604,
+      "learning_rate": 7.527349927658126e-06,
+      "loss": 0.4953,
+      "step": 21582
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 5.0811567306518555,
+      "learning_rate": 7.527122428217813e-06,
+      "loss": 0.4685,
+      "step": 21583
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.9068500995635986,
+      "learning_rate": 7.526894921750517e-06,
+      "loss": 0.4921,
+      "step": 21584
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.3317296504974365,
+      "learning_rate": 7.5266674082568736e-06,
+      "loss": 0.5854,
+      "step": 21585
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 3.818601608276367,
+      "learning_rate": 7.526439887737516e-06,
+      "loss": 0.6564,
+      "step": 21586
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.2648298740386963,
+      "learning_rate": 7.526212360193074e-06,
+      "loss": 0.6498,
+      "step": 21587
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.203047513961792,
+      "learning_rate": 7.525984825624182e-06,
+      "loss": 0.7189,
+      "step": 21588
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.7753639221191406,
+      "learning_rate": 7.525757284031474e-06,
+      "loss": 0.7492,
+      "step": 21589
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 3.535604238510132,
+      "learning_rate": 7.52552973541558e-06,
+      "loss": 0.8351,
+      "step": 21590
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.6948094367980957,
+      "learning_rate": 7.525302179777133e-06,
+      "loss": 0.7316,
+      "step": 21591
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 4.569797515869141,
+      "learning_rate": 7.525074617116767e-06,
+      "loss": 0.7015,
+      "step": 21592
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.568392276763916,
+      "learning_rate": 7.5248470474351145e-06,
+      "loss": 0.5207,
+      "step": 21593
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 3.2281107902526855,
+      "learning_rate": 7.524619470732808e-06,
+      "loss": 0.6104,
+      "step": 21594
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.274155616760254,
+      "learning_rate": 7.524391887010481e-06,
+      "loss": 0.7944,
+      "step": 21595
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.188901662826538,
+      "learning_rate": 7.524164296268765e-06,
+      "loss": 0.6491,
+      "step": 21596
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.379648208618164,
+      "learning_rate": 7.523936698508296e-06,
+      "loss": 0.7962,
+      "step": 21597
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.7559518814086914,
+      "learning_rate": 7.523709093729703e-06,
+      "loss": 0.5615,
+      "step": 21598
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.4603943824768066,
+      "learning_rate": 7.523481481933621e-06,
+      "loss": 0.657,
+      "step": 21599
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.3095498085021973,
+      "learning_rate": 7.523253863120681e-06,
+      "loss": 0.6544,
+      "step": 21600
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 3.783827543258667,
+      "learning_rate": 7.523026237291517e-06,
+      "loss": 0.4078,
+      "step": 21601
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.459800362586975,
+      "learning_rate": 7.522798604446764e-06,
+      "loss": 0.4714,
+      "step": 21602
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 4.855252265930176,
+      "learning_rate": 7.522570964587051e-06,
+      "loss": 0.5562,
+      "step": 21603
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.9773898124694824,
+      "learning_rate": 7.522343317713015e-06,
+      "loss": 0.5696,
+      "step": 21604
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.5575666427612305,
+      "learning_rate": 7.522115663825287e-06,
+      "loss": 0.4213,
+      "step": 21605
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.2387046813964844,
+      "learning_rate": 7.5218880029245e-06,
+      "loss": 0.6236,
+      "step": 21606
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.3435146808624268,
+      "learning_rate": 7.521660335011286e-06,
+      "loss": 0.6332,
+      "step": 21607
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.106149196624756,
+      "learning_rate": 7.52143266008628e-06,
+      "loss": 0.6506,
+      "step": 21608
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.5868380069732666,
+      "learning_rate": 7.521204978150114e-06,
+      "loss": 0.7491,
+      "step": 21609
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.8762784004211426,
+      "learning_rate": 7.5209772892034215e-06,
+      "loss": 0.5845,
+      "step": 21610
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.2372913360595703,
+      "learning_rate": 7.520749593246837e-06,
+      "loss": 0.5772,
+      "step": 21611
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.086463451385498,
+      "learning_rate": 7.5205218902809894e-06,
+      "loss": 0.4922,
+      "step": 21612
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 3.9653549194335938,
+      "learning_rate": 7.520294180306518e-06,
+      "loss": 0.8101,
+      "step": 21613
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 7.9977240562438965,
+      "learning_rate": 7.52006646332405e-06,
+      "loss": 0.8415,
+      "step": 21614
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.6017260551452637,
+      "learning_rate": 7.519838739334222e-06,
+      "loss": 0.4823,
+      "step": 21615
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.6519834995269775,
+      "learning_rate": 7.519611008337666e-06,
+      "loss": 0.5927,
+      "step": 21616
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.1824495792388916,
+      "learning_rate": 7.519383270335017e-06,
+      "loss": 0.9416,
+      "step": 21617
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.7699041366577148,
+      "learning_rate": 7.519155525326906e-06,
+      "loss": 0.7656,
+      "step": 21618
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.134640693664551,
+      "learning_rate": 7.518927773313966e-06,
+      "loss": 0.6989,
+      "step": 21619
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.5890170335769653,
+      "learning_rate": 7.518700014296833e-06,
+      "loss": 0.6544,
+      "step": 21620
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.1880197525024414,
+      "learning_rate": 7.518472248276139e-06,
+      "loss": 0.719,
+      "step": 21621
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.5398263931274414,
+      "learning_rate": 7.5182444752525166e-06,
+      "loss": 0.5869,
+      "step": 21622
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 4.1946587562561035,
+      "learning_rate": 7.5180166952266e-06,
+      "loss": 0.6686,
+      "step": 21623
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 3.078634262084961,
+      "learning_rate": 7.517788908199021e-06,
+      "loss": 0.529,
+      "step": 21624
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.7370049953460693,
+      "learning_rate": 7.517561114170415e-06,
+      "loss": 0.4057,
+      "step": 21625
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 3.1794629096984863,
+      "learning_rate": 7.517333313141415e-06,
+      "loss": 0.6453,
+      "step": 21626
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.8647241592407227,
+      "learning_rate": 7.517105505112654e-06,
+      "loss": 0.788,
+      "step": 21627
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.0428948402404785,
+      "learning_rate": 7.516877690084765e-06,
+      "loss": 0.31,
+      "step": 21628
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.5453007221221924,
+      "learning_rate": 7.516649868058384e-06,
+      "loss": 0.7156,
+      "step": 21629
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.6037116050720215,
+      "learning_rate": 7.516422039034139e-06,
+      "loss": 0.5025,
+      "step": 21630
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.6658663749694824,
+      "learning_rate": 7.51619420301267e-06,
+      "loss": 0.8247,
+      "step": 21631
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.4393723011016846,
+      "learning_rate": 7.515966359994607e-06,
+      "loss": 0.7687,
+      "step": 21632
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 3.6340746879577637,
+      "learning_rate": 7.515738509980581e-06,
+      "loss": 0.5747,
+      "step": 21633
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.1475131511688232,
+      "learning_rate": 7.515510652971232e-06,
+      "loss": 0.7533,
+      "step": 21634
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.3205006122589111,
+      "learning_rate": 7.5152827889671885e-06,
+      "loss": 0.6229,
+      "step": 21635
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.8283586502075195,
+      "learning_rate": 7.515054917969087e-06,
+      "loss": 0.6061,
+      "step": 21636
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 4.62929630279541,
+      "learning_rate": 7.514827039977559e-06,
+      "loss": 0.4363,
+      "step": 21637
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.880523204803467,
+      "learning_rate": 7.51459915499324e-06,
+      "loss": 0.4627,
+      "step": 21638
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.3614354133605957,
+      "learning_rate": 7.514371263016763e-06,
+      "loss": 0.6176,
+      "step": 21639
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.036308526992798,
+      "learning_rate": 7.51414336404876e-06,
+      "loss": 0.5687,
+      "step": 21640
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.0205440521240234,
+      "learning_rate": 7.513915458089868e-06,
+      "loss": 0.9734,
+      "step": 21641
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 3.296797037124634,
+      "learning_rate": 7.513687545140717e-06,
+      "loss": 0.6223,
+      "step": 21642
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.286351442337036,
+      "learning_rate": 7.513459625201942e-06,
+      "loss": 0.5304,
+      "step": 21643
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.079146385192871,
+      "learning_rate": 7.513231698274179e-06,
+      "loss": 0.8571,
+      "step": 21644
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.5611913204193115,
+      "learning_rate": 7.51300376435806e-06,
+      "loss": 0.5026,
+      "step": 21645
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 4.462684631347656,
+      "learning_rate": 7.512775823454219e-06,
+      "loss": 0.4582,
+      "step": 21646
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 3.2924511432647705,
+      "learning_rate": 7.512547875563289e-06,
+      "loss": 0.8667,
+      "step": 21647
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.473539113998413,
+      "learning_rate": 7.5123199206859056e-06,
+      "loss": 0.6995,
+      "step": 21648
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.7395410537719727,
+      "learning_rate": 7.5120919588227005e-06,
+      "loss": 0.5731,
+      "step": 21649
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.8985016345977783,
+      "learning_rate": 7.51186398997431e-06,
+      "loss": 0.6464,
+      "step": 21650
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.301702857017517,
+      "learning_rate": 7.511636014141366e-06,
+      "loss": 0.3807,
+      "step": 21651
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.8234045505523682,
+      "learning_rate": 7.511408031324503e-06,
+      "loss": 0.6752,
+      "step": 21652
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.4223321676254272,
+      "learning_rate": 7.511180041524355e-06,
+      "loss": 0.6556,
+      "step": 21653
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 3.8600969314575195,
+      "learning_rate": 7.510952044741555e-06,
+      "loss": 0.9251,
+      "step": 21654
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.3147432804107666,
+      "learning_rate": 7.51072404097674e-06,
+      "loss": 0.649,
+      "step": 21655
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 3.162747859954834,
+      "learning_rate": 7.510496030230541e-06,
+      "loss": 0.5025,
+      "step": 21656
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 3.1053049564361572,
+      "learning_rate": 7.510268012503593e-06,
+      "loss": 0.623,
+      "step": 21657
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.430166721343994,
+      "learning_rate": 7.51003998779653e-06,
+      "loss": 0.6887,
+      "step": 21658
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.7104922533035278,
+      "learning_rate": 7.509811956109987e-06,
+      "loss": 0.7654,
+      "step": 21659
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.8323822021484375,
+      "learning_rate": 7.509583917444596e-06,
+      "loss": 0.7124,
+      "step": 21660
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.254655361175537,
+      "learning_rate": 7.509355871800993e-06,
+      "loss": 0.734,
+      "step": 21661
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.6964586973190308,
+      "learning_rate": 7.509127819179811e-06,
+      "loss": 0.5932,
+      "step": 21662
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.001030445098877,
+      "learning_rate": 7.5088997595816845e-06,
+      "loss": 0.3434,
+      "step": 21663
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 3.032038688659668,
+      "learning_rate": 7.508671693007247e-06,
+      "loss": 0.7004,
+      "step": 21664
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.641753911972046,
+      "learning_rate": 7.5084436194571334e-06,
+      "loss": 0.7809,
+      "step": 21665
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.7793588638305664,
+      "learning_rate": 7.5082155389319774e-06,
+      "loss": 0.8407,
+      "step": 21666
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.6587579250335693,
+      "learning_rate": 7.507987451432415e-06,
+      "loss": 0.5818,
+      "step": 21667
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.8807687759399414,
+      "learning_rate": 7.507759356959079e-06,
+      "loss": 0.4846,
+      "step": 21668
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.4807827472686768,
+      "learning_rate": 7.507531255512603e-06,
+      "loss": 0.7881,
+      "step": 21669
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.278752326965332,
+      "learning_rate": 7.507303147093621e-06,
+      "loss": 0.8341,
+      "step": 21670
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.4960306882858276,
+      "learning_rate": 7.50707503170277e-06,
+      "loss": 0.5891,
+      "step": 21671
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.3850607872009277,
+      "learning_rate": 7.506846909340681e-06,
+      "loss": 0.4108,
+      "step": 21672
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.415358543395996,
+      "learning_rate": 7.50661878000799e-06,
+      "loss": 0.5791,
+      "step": 21673
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.9598467350006104,
+      "learning_rate": 7.506390643705333e-06,
+      "loss": 0.7379,
+      "step": 21674
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.0193068981170654,
+      "learning_rate": 7.5061625004333405e-06,
+      "loss": 0.9399,
+      "step": 21675
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 3.336423873901367,
+      "learning_rate": 7.50593435019265e-06,
+      "loss": 0.5735,
+      "step": 21676
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.8266677856445312,
+      "learning_rate": 7.5057061929838936e-06,
+      "loss": 0.7191,
+      "step": 21677
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.195754289627075,
+      "learning_rate": 7.505478028807708e-06,
+      "loss": 0.713,
+      "step": 21678
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.9154794216156006,
+      "learning_rate": 7.505249857664727e-06,
+      "loss": 0.581,
+      "step": 21679
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.5045632123947144,
+      "learning_rate": 7.505021679555584e-06,
+      "loss": 0.7242,
+      "step": 21680
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.0591676235198975,
+      "learning_rate": 7.504793494480913e-06,
+      "loss": 0.5682,
+      "step": 21681
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.7090682983398438,
+      "learning_rate": 7.504565302441351e-06,
+      "loss": 0.5553,
+      "step": 21682
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.396942377090454,
+      "learning_rate": 7.504337103437531e-06,
+      "loss": 0.668,
+      "step": 21683
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.6875994205474854,
+      "learning_rate": 7.5041088974700875e-06,
+      "loss": 0.626,
+      "step": 21684
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.5786328315734863,
+      "learning_rate": 7.503880684539655e-06,
+      "loss": 0.592,
+      "step": 21685
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.4402992725372314,
+      "learning_rate": 7.503652464646869e-06,
+      "loss": 0.6049,
+      "step": 21686
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.005084276199341,
+      "learning_rate": 7.503424237792363e-06,
+      "loss": 0.625,
+      "step": 21687
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.9403009414672852,
+      "learning_rate": 7.503196003976772e-06,
+      "loss": 0.4354,
+      "step": 21688
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.4398481845855713,
+      "learning_rate": 7.50296776320073e-06,
+      "loss": 0.6742,
+      "step": 21689
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.4151785373687744,
+      "learning_rate": 7.502739515464873e-06,
+      "loss": 0.6603,
+      "step": 21690
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 4.4311628341674805,
+      "learning_rate": 7.5025112607698356e-06,
+      "loss": 0.5324,
+      "step": 21691
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.2440333366394043,
+      "learning_rate": 7.50228299911625e-06,
+      "loss": 0.6232,
+      "step": 21692
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.61308217048645,
+      "learning_rate": 7.502054730504754e-06,
+      "loss": 0.7793,
+      "step": 21693
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.9855825901031494,
+      "learning_rate": 7.501826454935981e-06,
+      "loss": 0.5403,
+      "step": 21694
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.3833377361297607,
+      "learning_rate": 7.5015981724105645e-06,
+      "loss": 0.7307,
+      "step": 21695
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.2614295482635498,
+      "learning_rate": 7.501369882929142e-06,
+      "loss": 0.5064,
+      "step": 21696
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.1390438079833984,
+      "learning_rate": 7.501141586492348e-06,
+      "loss": 0.5588,
+      "step": 21697
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 3.4781885147094727,
+      "learning_rate": 7.500913283100814e-06,
+      "loss": 0.5597,
+      "step": 21698
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 5.872242450714111,
+      "learning_rate": 7.500684972755179e-06,
+      "loss": 0.5288,
+      "step": 21699
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.901754856109619,
+      "learning_rate": 7.500456655456074e-06,
+      "loss": 0.7691,
+      "step": 21700
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.4328176975250244,
+      "learning_rate": 7.500228331204137e-06,
+      "loss": 0.472,
+      "step": 21701
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.3311302661895752,
+      "learning_rate": 7.500000000000001e-06,
+      "loss": 0.5618,
+      "step": 21702
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.708162784576416,
+      "learning_rate": 7.4997716618443015e-06,
+      "loss": 0.5697,
+      "step": 21703
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 3.7281501293182373,
+      "learning_rate": 7.499543316737673e-06,
+      "loss": 0.7141,
+      "step": 21704
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.1227598190307617,
+      "learning_rate": 7.499314964680752e-06,
+      "loss": 0.3497,
+      "step": 21705
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 3.7757728099823,
+      "learning_rate": 7.499086605674172e-06,
+      "loss": 0.6245,
+      "step": 21706
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 3.8596808910369873,
+      "learning_rate": 7.498858239718567e-06,
+      "loss": 0.641,
+      "step": 21707
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.5434057712554932,
+      "learning_rate": 7.498629866814575e-06,
+      "loss": 0.449,
+      "step": 21708
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.241222620010376,
+      "learning_rate": 7.49840148696283e-06,
+      "loss": 0.7233,
+      "step": 21709
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.808470606803894,
+      "learning_rate": 7.498173100163964e-06,
+      "loss": 0.4371,
+      "step": 21710
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.58408784866333,
+      "learning_rate": 7.497944706418616e-06,
+      "loss": 0.6414,
+      "step": 21711
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.647177815437317,
+      "learning_rate": 7.49771630572742e-06,
+      "loss": 0.6984,
+      "step": 21712
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.5216608047485352,
+      "learning_rate": 7.497487898091009e-06,
+      "loss": 0.6128,
+      "step": 21713
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.8579826354980469,
+      "learning_rate": 7.49725948351002e-06,
+      "loss": 0.5404,
+      "step": 21714
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.3778085708618164,
+      "learning_rate": 7.4970310619850895e-06,
+      "loss": 0.6965,
+      "step": 21715
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.2581212520599365,
+      "learning_rate": 7.49680263351685e-06,
+      "loss": 0.5956,
+      "step": 21716
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 3.0340445041656494,
+      "learning_rate": 7.496574198105937e-06,
+      "loss": 0.676,
+      "step": 21717
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.510542869567871,
+      "learning_rate": 7.4963457557529875e-06,
+      "loss": 0.5348,
+      "step": 21718
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 4.445478439331055,
+      "learning_rate": 7.4961173064586345e-06,
+      "loss": 0.7235,
+      "step": 21719
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.6052443981170654,
+      "learning_rate": 7.495888850223516e-06,
+      "loss": 0.5785,
+      "step": 21720
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.7232468128204346,
+      "learning_rate": 7.495660387048264e-06,
+      "loss": 0.8172,
+      "step": 21721
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.8977556228637695,
+      "learning_rate": 7.495431916933516e-06,
+      "loss": 0.6171,
+      "step": 21722
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 3.356306314468384,
+      "learning_rate": 7.4952034398799064e-06,
+      "loss": 0.8065,
+      "step": 21723
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 3.199136257171631,
+      "learning_rate": 7.494974955888071e-06,
+      "loss": 0.6722,
+      "step": 21724
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.8474940061569214,
+      "learning_rate": 7.494746464958644e-06,
+      "loss": 0.8945,
+      "step": 21725
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.4086393117904663,
+      "learning_rate": 7.494517967092263e-06,
+      "loss": 0.4577,
+      "step": 21726
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 5.787997245788574,
+      "learning_rate": 7.494289462289561e-06,
+      "loss": 0.5285,
+      "step": 21727
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.3544317483901978,
+      "learning_rate": 7.494060950551175e-06,
+      "loss": 0.7605,
+      "step": 21728
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.561150550842285,
+      "learning_rate": 7.49383243187774e-06,
+      "loss": 0.7114,
+      "step": 21729
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 3.236222267150879,
+      "learning_rate": 7.493603906269891e-06,
+      "loss": 0.7576,
+      "step": 21730
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 3.357163190841675,
+      "learning_rate": 7.493375373728263e-06,
+      "loss": 0.7026,
+      "step": 21731
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 3.0726256370544434,
+      "learning_rate": 7.493146834253493e-06,
+      "loss": 0.7236,
+      "step": 21732
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.478693962097168,
+      "learning_rate": 7.4929182878462145e-06,
+      "loss": 0.387,
+      "step": 21733
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 3.3568825721740723,
+      "learning_rate": 7.492689734507066e-06,
+      "loss": 0.7337,
+      "step": 21734
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.1854584217071533,
+      "learning_rate": 7.4924611742366795e-06,
+      "loss": 0.5098,
+      "step": 21735
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.994603157043457,
+      "learning_rate": 7.492232607035693e-06,
+      "loss": 0.6853,
+      "step": 21736
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 3.94763445854187,
+      "learning_rate": 7.49200403290474e-06,
+      "loss": 0.7919,
+      "step": 21737
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 3.525658369064331,
+      "learning_rate": 7.491775451844459e-06,
+      "loss": 0.7131,
+      "step": 21738
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.9153072834014893,
+      "learning_rate": 7.4915468638554826e-06,
+      "loss": 0.5793,
+      "step": 21739
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 4.029727935791016,
+      "learning_rate": 7.491318268938448e-06,
+      "loss": 0.9589,
+      "step": 21740
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 3.0474605560302734,
+      "learning_rate": 7.491089667093991e-06,
+      "loss": 0.6998,
+      "step": 21741
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.1166939735412598,
+      "learning_rate": 7.490861058322747e-06,
+      "loss": 0.6082,
+      "step": 21742
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.865730881690979,
+      "learning_rate": 7.4906324426253505e-06,
+      "loss": 0.5307,
+      "step": 21743
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.6896384954452515,
+      "learning_rate": 7.490403820002439e-06,
+      "loss": 0.6311,
+      "step": 21744
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 5.6349897384643555,
+      "learning_rate": 7.490175190454647e-06,
+      "loss": 0.5128,
+      "step": 21745
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.9097609519958496,
+      "learning_rate": 7.4899465539826114e-06,
+      "loss": 0.633,
+      "step": 21746
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.8825631141662598,
+      "learning_rate": 7.489717910586966e-06,
+      "loss": 0.5762,
+      "step": 21747
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 3.5015485286712646,
+      "learning_rate": 7.489489260268349e-06,
+      "loss": 0.9431,
+      "step": 21748
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 3.5134992599487305,
+      "learning_rate": 7.489260603027394e-06,
+      "loss": 0.8326,
+      "step": 21749
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 3.7098662853240967,
+      "learning_rate": 7.489031938864738e-06,
+      "loss": 0.6416,
+      "step": 21750
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.050499677658081,
+      "learning_rate": 7.488803267781017e-06,
+      "loss": 0.6117,
+      "step": 21751
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 3.4144434928894043,
+      "learning_rate": 7.488574589776865e-06,
+      "loss": 0.4788,
+      "step": 21752
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.9519941806793213,
+      "learning_rate": 7.488345904852922e-06,
+      "loss": 0.5122,
+      "step": 21753
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.9465243816375732,
+      "learning_rate": 7.4881172130098205e-06,
+      "loss": 0.5258,
+      "step": 21754
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 3.8968346118927,
+      "learning_rate": 7.487888514248194e-06,
+      "loss": 0.5709,
+      "step": 21755
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.5358030796051025,
+      "learning_rate": 7.487659808568684e-06,
+      "loss": 0.8209,
+      "step": 21756
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.2852299213409424,
+      "learning_rate": 7.487431095971924e-06,
+      "loss": 0.5536,
+      "step": 21757
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.7881933450698853,
+      "learning_rate": 7.48720237645855e-06,
+      "loss": 0.7282,
+      "step": 21758
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.7748613357543945,
+      "learning_rate": 7.486973650029198e-06,
+      "loss": 0.7372,
+      "step": 21759
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.382367730140686,
+      "learning_rate": 7.486744916684503e-06,
+      "loss": 0.5798,
+      "step": 21760
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.0153589248657227,
+      "learning_rate": 7.486516176425103e-06,
+      "loss": 0.718,
+      "step": 21761
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.3303260803222656,
+      "learning_rate": 7.486287429251632e-06,
+      "loss": 0.489,
+      "step": 21762
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.593465805053711,
+      "learning_rate": 7.486058675164728e-06,
+      "loss": 0.5692,
+      "step": 21763
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.162709951400757,
+      "learning_rate": 7.485829914165026e-06,
+      "loss": 0.4994,
+      "step": 21764
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.8427460193634033,
+      "learning_rate": 7.485601146253161e-06,
+      "loss": 0.736,
+      "step": 21765
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.7323609590530396,
+      "learning_rate": 7.485372371429771e-06,
+      "loss": 0.6412,
+      "step": 21766
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.0296337604522705,
+      "learning_rate": 7.485143589695491e-06,
+      "loss": 0.6832,
+      "step": 21767
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 4.660975456237793,
+      "learning_rate": 7.484914801050959e-06,
+      "loss": 0.5543,
+      "step": 21768
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.9037572145462036,
+      "learning_rate": 7.484686005496808e-06,
+      "loss": 0.922,
+      "step": 21769
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.8151254653930664,
+      "learning_rate": 7.484457203033678e-06,
+      "loss": 0.7396,
+      "step": 21770
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 3.000678062438965,
+      "learning_rate": 7.484228393662203e-06,
+      "loss": 0.757,
+      "step": 21771
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.9451894760131836,
+      "learning_rate": 7.4839995773830184e-06,
+      "loss": 0.6248,
+      "step": 21772
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 3.720824956893921,
+      "learning_rate": 7.483770754196762e-06,
+      "loss": 0.538,
+      "step": 21773
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.735116481781006,
+      "learning_rate": 7.4835419241040695e-06,
+      "loss": 0.7997,
+      "step": 21774
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.9221025705337524,
+      "learning_rate": 7.483313087105577e-06,
+      "loss": 0.8824,
+      "step": 21775
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.2913014888763428,
+      "learning_rate": 7.483084243201921e-06,
+      "loss": 0.7747,
+      "step": 21776
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.0778472423553467,
+      "learning_rate": 7.482855392393737e-06,
+      "loss": 0.8181,
+      "step": 21777
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 3.7041590213775635,
+      "learning_rate": 7.482626534681665e-06,
+      "loss": 0.8264,
+      "step": 21778
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 4.063631057739258,
+      "learning_rate": 7.482397670066337e-06,
+      "loss": 0.7015,
+      "step": 21779
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.0192067623138428,
+      "learning_rate": 7.48216879854839e-06,
+      "loss": 0.4543,
+      "step": 21780
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 3.635946035385132,
+      "learning_rate": 7.481939920128464e-06,
+      "loss": 0.5508,
+      "step": 21781
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 3.0443966388702393,
+      "learning_rate": 7.481711034807191e-06,
+      "loss": 0.6075,
+      "step": 21782
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 3.5231521129608154,
+      "learning_rate": 7.48148214258521e-06,
+      "loss": 0.8514,
+      "step": 21783
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.6394009590148926,
+      "learning_rate": 7.481253243463156e-06,
+      "loss": 0.9281,
+      "step": 21784
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.9834448099136353,
+      "learning_rate": 7.481024337441667e-06,
+      "loss": 0.3983,
+      "step": 21785
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.521538019180298,
+      "learning_rate": 7.480795424521377e-06,
+      "loss": 0.8267,
+      "step": 21786
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 3.0577549934387207,
+      "learning_rate": 7.480566504702926e-06,
+      "loss": 0.5753,
+      "step": 21787
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.8833718299865723,
+      "learning_rate": 7.480337577986948e-06,
+      "loss": 0.5118,
+      "step": 21788
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.1010429859161377,
+      "learning_rate": 7.480108644374081e-06,
+      "loss": 0.5862,
+      "step": 21789
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.138516426086426,
+      "learning_rate": 7.47987970386496e-06,
+      "loss": 0.5347,
+      "step": 21790
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 4.256069660186768,
+      "learning_rate": 7.479650756460224e-06,
+      "loss": 0.756,
+      "step": 21791
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.424938201904297,
+      "learning_rate": 7.479421802160507e-06,
+      "loss": 0.6747,
+      "step": 21792
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.6518369913101196,
+      "learning_rate": 7.479192840966447e-06,
+      "loss": 0.5384,
+      "step": 21793
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 3.0464181900024414,
+      "learning_rate": 7.47896387287868e-06,
+      "loss": 0.5999,
+      "step": 21794
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.633948564529419,
+      "learning_rate": 7.478734897897844e-06,
+      "loss": 0.5944,
+      "step": 21795
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.9456305503845215,
+      "learning_rate": 7.4785059160245745e-06,
+      "loss": 0.8811,
+      "step": 21796
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 3.658378839492798,
+      "learning_rate": 7.478276927259509e-06,
+      "loss": 0.673,
+      "step": 21797
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 3.6557326316833496,
+      "learning_rate": 7.478047931603283e-06,
+      "loss": 0.5743,
+      "step": 21798
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.7062722444534302,
+      "learning_rate": 7.477818929056533e-06,
+      "loss": 0.5746,
+      "step": 21799
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.8650398254394531,
+      "learning_rate": 7.477589919619898e-06,
+      "loss": 0.7434,
+      "step": 21800
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.923332691192627,
+      "learning_rate": 7.477360903294013e-06,
+      "loss": 0.6677,
+      "step": 21801
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 3.3612701892852783,
+      "learning_rate": 7.4771318800795165e-06,
+      "loss": 0.7296,
+      "step": 21802
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.890807867050171,
+      "learning_rate": 7.476902849977043e-06,
+      "loss": 0.7414,
+      "step": 21803
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 3.3326797485351562,
+      "learning_rate": 7.4766738129872315e-06,
+      "loss": 0.6737,
+      "step": 21804
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.721074104309082,
+      "learning_rate": 7.4764447691107176e-06,
+      "loss": 0.4958,
+      "step": 21805
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.8313082456588745,
+      "learning_rate": 7.476215718348139e-06,
+      "loss": 0.7911,
+      "step": 21806
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.1307883262634277,
+      "learning_rate": 7.475986660700131e-06,
+      "loss": 0.5401,
+      "step": 21807
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.5575780868530273,
+      "learning_rate": 7.475757596167333e-06,
+      "loss": 0.5634,
+      "step": 21808
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.3135600090026855,
+      "learning_rate": 7.475528524750379e-06,
+      "loss": 0.7019,
+      "step": 21809
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.172891855239868,
+      "learning_rate": 7.4752994464499085e-06,
+      "loss": 0.8644,
+      "step": 21810
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 3.0335750579833984,
+      "learning_rate": 7.475070361266557e-06,
+      "loss": 0.7022,
+      "step": 21811
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.1650640964508057,
+      "learning_rate": 7.474841269200963e-06,
+      "loss": 0.6292,
+      "step": 21812
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 3.173495292663574,
+      "learning_rate": 7.474612170253761e-06,
+      "loss": 0.7891,
+      "step": 21813
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.8933119773864746,
+      "learning_rate": 7.474383064425591e-06,
+      "loss": 0.7319,
+      "step": 21814
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.0908634662628174,
+      "learning_rate": 7.474153951717088e-06,
+      "loss": 0.749,
+      "step": 21815
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 3.2043545246124268,
+      "learning_rate": 7.47392483212889e-06,
+      "loss": 0.6238,
+      "step": 21816
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.8168416023254395,
+      "learning_rate": 7.473695705661632e-06,
+      "loss": 0.6854,
+      "step": 21817
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.6367865800857544,
+      "learning_rate": 7.473466572315955e-06,
+      "loss": 0.6272,
+      "step": 21818
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.4293150901794434,
+      "learning_rate": 7.473237432092493e-06,
+      "loss": 0.5402,
+      "step": 21819
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.129437208175659,
+      "learning_rate": 7.473008284991885e-06,
+      "loss": 0.6345,
+      "step": 21820
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.394683599472046,
+      "learning_rate": 7.472779131014767e-06,
+      "loss": 0.7812,
+      "step": 21821
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 3.456350564956665,
+      "learning_rate": 7.472549970161776e-06,
+      "loss": 0.8743,
+      "step": 21822
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 3.2669525146484375,
+      "learning_rate": 7.4723208024335516e-06,
+      "loss": 0.5182,
+      "step": 21823
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.383666753768921,
+      "learning_rate": 7.472091627830728e-06,
+      "loss": 0.7994,
+      "step": 21824
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.8805134296417236,
+      "learning_rate": 7.471862446353944e-06,
+      "loss": 0.6906,
+      "step": 21825
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 3.8683009147644043,
+      "learning_rate": 7.471633258003834e-06,
+      "loss": 0.6821,
+      "step": 21826
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.248668670654297,
+      "learning_rate": 7.4714040627810405e-06,
+      "loss": 0.5032,
+      "step": 21827
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 3.2165603637695312,
+      "learning_rate": 7.471174860686198e-06,
+      "loss": 0.8338,
+      "step": 21828
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.140418291091919,
+      "learning_rate": 7.470945651719943e-06,
+      "loss": 0.5921,
+      "step": 21829
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 3.763110876083374,
+      "learning_rate": 7.470716435882914e-06,
+      "loss": 0.7474,
+      "step": 21830
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.5416202545166016,
+      "learning_rate": 7.470487213175749e-06,
+      "loss": 0.5455,
+      "step": 21831
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 3.3610527515411377,
+      "learning_rate": 7.470257983599084e-06,
+      "loss": 0.6779,
+      "step": 21832
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.98034405708313,
+      "learning_rate": 7.470028747153557e-06,
+      "loss": 0.889,
+      "step": 21833
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 3.250208854675293,
+      "learning_rate": 7.469799503839805e-06,
+      "loss": 0.7067,
+      "step": 21834
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.5755921602249146,
+      "learning_rate": 7.4695702536584665e-06,
+      "loss": 0.5531,
+      "step": 21835
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.9688642024993896,
+      "learning_rate": 7.469340996610177e-06,
+      "loss": 0.598,
+      "step": 21836
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.9766809940338135,
+      "learning_rate": 7.469111732695575e-06,
+      "loss": 0.5025,
+      "step": 21837
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.345093011856079,
+      "learning_rate": 7.4688824619153e-06,
+      "loss": 0.6215,
+      "step": 21838
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 3.074387550354004,
+      "learning_rate": 7.468653184269987e-06,
+      "loss": 0.821,
+      "step": 21839
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.807245135307312,
+      "learning_rate": 7.468423899760275e-06,
+      "loss": 0.7065,
+      "step": 21840
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.168591022491455,
+      "learning_rate": 7.468194608386799e-06,
+      "loss": 0.5087,
+      "step": 21841
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.7110410928726196,
+      "learning_rate": 7.4679653101502e-06,
+      "loss": 0.6249,
+      "step": 21842
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 3.076768398284912,
+      "learning_rate": 7.467736005051114e-06,
+      "loss": 0.4246,
+      "step": 21843
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.2520029544830322,
+      "learning_rate": 7.467506693090177e-06,
+      "loss": 0.3379,
+      "step": 21844
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 3.3793208599090576,
+      "learning_rate": 7.467277374268029e-06,
+      "loss": 0.7845,
+      "step": 21845
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 3.3969061374664307,
+      "learning_rate": 7.467048048585308e-06,
+      "loss": 0.6652,
+      "step": 21846
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.7967300415039062,
+      "learning_rate": 7.466818716042649e-06,
+      "loss": 0.7624,
+      "step": 21847
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 3.1452364921569824,
+      "learning_rate": 7.466589376640692e-06,
+      "loss": 0.5442,
+      "step": 21848
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.616131544113159,
+      "learning_rate": 7.466360030380073e-06,
+      "loss": 0.6274,
+      "step": 21849
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.276346445083618,
+      "learning_rate": 7.466130677261431e-06,
+      "loss": 0.5212,
+      "step": 21850
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.2997632026672363,
+      "learning_rate": 7.465901317285406e-06,
+      "loss": 0.6815,
+      "step": 21851
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 7.403448581695557,
+      "learning_rate": 7.465671950452631e-06,
+      "loss": 0.8772,
+      "step": 21852
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 3.836068630218506,
+      "learning_rate": 7.465442576763746e-06,
+      "loss": 0.7176,
+      "step": 21853
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.13277006149292,
+      "learning_rate": 7.4652131962193884e-06,
+      "loss": 0.6498,
+      "step": 21854
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 3.29744815826416,
+      "learning_rate": 7.464983808820197e-06,
+      "loss": 0.589,
+      "step": 21855
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.056018590927124,
+      "learning_rate": 7.464754414566808e-06,
+      "loss": 0.6712,
+      "step": 21856
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.3531718254089355,
+      "learning_rate": 7.464525013459862e-06,
+      "loss": 0.4144,
+      "step": 21857
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 3.690178632736206,
+      "learning_rate": 7.464295605499994e-06,
+      "loss": 0.8018,
+      "step": 21858
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.4334535598754883,
+      "learning_rate": 7.464066190687844e-06,
+      "loss": 0.9272,
+      "step": 21859
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 3.390990972518921,
+      "learning_rate": 7.463836769024048e-06,
+      "loss": 0.5305,
+      "step": 21860
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.6679956912994385,
+      "learning_rate": 7.463607340509245e-06,
+      "loss": 0.6912,
+      "step": 21861
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 3.423377752304077,
+      "learning_rate": 7.463377905144074e-06,
+      "loss": 0.5685,
+      "step": 21862
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.3242857456207275,
+      "learning_rate": 7.463148462929171e-06,
+      "loss": 0.3773,
+      "step": 21863
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.9592667818069458,
+      "learning_rate": 7.462919013865174e-06,
+      "loss": 0.7807,
+      "step": 21864
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.9395818710327148,
+      "learning_rate": 7.462689557952722e-06,
+      "loss": 0.3957,
+      "step": 21865
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 8.134153366088867,
+      "learning_rate": 7.462460095192454e-06,
+      "loss": 0.7046,
+      "step": 21866
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.8039711713790894,
+      "learning_rate": 7.462230625585007e-06,
+      "loss": 0.5318,
+      "step": 21867
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 3.8701000213623047,
+      "learning_rate": 7.462001149131017e-06,
+      "loss": 0.7808,
+      "step": 21868
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.4004316329956055,
+      "learning_rate": 7.461771665831127e-06,
+      "loss": 0.5874,
+      "step": 21869
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.8587843179702759,
+      "learning_rate": 7.46154217568597e-06,
+      "loss": 0.6777,
+      "step": 21870
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.2947795391082764,
+      "learning_rate": 7.461312678696189e-06,
+      "loss": 0.5702,
+      "step": 21871
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.3943936824798584,
+      "learning_rate": 7.4610831748624165e-06,
+      "loss": 0.5589,
+      "step": 21872
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.13407564163208,
+      "learning_rate": 7.460853664185295e-06,
+      "loss": 0.6067,
+      "step": 21873
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.2028863430023193,
+      "learning_rate": 7.460624146665462e-06,
+      "loss": 0.7265,
+      "step": 21874
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.2126505374908447,
+      "learning_rate": 7.460394622303554e-06,
+      "loss": 0.6531,
+      "step": 21875
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.07574462890625,
+      "learning_rate": 7.46016509110021e-06,
+      "loss": 0.4922,
+      "step": 21876
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 3.8933043479919434,
+      "learning_rate": 7.459935553056068e-06,
+      "loss": 0.5594,
+      "step": 21877
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.8870524168014526,
+      "learning_rate": 7.459706008171768e-06,
+      "loss": 0.9107,
+      "step": 21878
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.3836262226104736,
+      "learning_rate": 7.459476456447946e-06,
+      "loss": 0.4795,
+      "step": 21879
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 3.390458822250366,
+      "learning_rate": 7.459246897885242e-06,
+      "loss": 0.6543,
+      "step": 21880
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.131455421447754,
+      "learning_rate": 7.459017332484294e-06,
+      "loss": 0.4916,
+      "step": 21881
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 3.528383731842041,
+      "learning_rate": 7.45878776024574e-06,
+      "loss": 0.781,
+      "step": 21882
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.3225414752960205,
+      "learning_rate": 7.458558181170218e-06,
+      "loss": 0.7136,
+      "step": 21883
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.3476290702819824,
+      "learning_rate": 7.458328595258366e-06,
+      "loss": 0.7425,
+      "step": 21884
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.5715811252593994,
+      "learning_rate": 7.458099002510824e-06,
+      "loss": 0.6117,
+      "step": 21885
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.1547248363494873,
+      "learning_rate": 7.4578694029282295e-06,
+      "loss": 0.7255,
+      "step": 21886
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.163024425506592,
+      "learning_rate": 7.4576397965112194e-06,
+      "loss": 0.8546,
+      "step": 21887
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.8183581829071045,
+      "learning_rate": 7.457410183260435e-06,
+      "loss": 0.6921,
+      "step": 21888
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.513427257537842,
+      "learning_rate": 7.4571805631765135e-06,
+      "loss": 0.7587,
+      "step": 21889
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.4167652130126953,
+      "learning_rate": 7.456950936260092e-06,
+      "loss": 0.6714,
+      "step": 21890
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.9419307708740234,
+      "learning_rate": 7.4567213025118126e-06,
+      "loss": 0.6428,
+      "step": 21891
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.1179637908935547,
+      "learning_rate": 7.45649166193231e-06,
+      "loss": 0.5737,
+      "step": 21892
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 3.0361523628234863,
+      "learning_rate": 7.456262014522225e-06,
+      "loss": 0.5854,
+      "step": 21893
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.3525805473327637,
+      "learning_rate": 7.456032360282195e-06,
+      "loss": 0.5989,
+      "step": 21894
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.1307637691497803,
+      "learning_rate": 7.45580269921286e-06,
+      "loss": 0.6443,
+      "step": 21895
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.2828493118286133,
+      "learning_rate": 7.455573031314857e-06,
+      "loss": 0.3681,
+      "step": 21896
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.399813175201416,
+      "learning_rate": 7.455343356588825e-06,
+      "loss": 0.5329,
+      "step": 21897
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.5532517433166504,
+      "learning_rate": 7.455113675035402e-06,
+      "loss": 0.7569,
+      "step": 21898
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 3.3370814323425293,
+      "learning_rate": 7.454883986655229e-06,
+      "loss": 0.7661,
+      "step": 21899
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.242971658706665,
+      "learning_rate": 7.454654291448944e-06,
+      "loss": 0.6733,
+      "step": 21900
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 3.002121925354004,
+      "learning_rate": 7.454424589417183e-06,
+      "loss": 0.7105,
+      "step": 21901
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.2017362117767334,
+      "learning_rate": 7.454194880560588e-06,
+      "loss": 0.3988,
+      "step": 21902
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.016514778137207,
+      "learning_rate": 7.4539651648797966e-06,
+      "loss": 0.4821,
+      "step": 21903
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.984800100326538,
+      "learning_rate": 7.4537354423754456e-06,
+      "loss": 0.894,
+      "step": 21904
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.38700008392334,
+      "learning_rate": 7.453505713048177e-06,
+      "loss": 0.586,
+      "step": 21905
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.399319887161255,
+      "learning_rate": 7.453275976898629e-06,
+      "loss": 0.3716,
+      "step": 21906
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 3.2091476917266846,
+      "learning_rate": 7.4530462339274375e-06,
+      "loss": 0.4812,
+      "step": 21907
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.9545475244522095,
+      "learning_rate": 7.452816484135243e-06,
+      "loss": 0.7262,
+      "step": 21908
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 3.988795280456543,
+      "learning_rate": 7.452586727522686e-06,
+      "loss": 0.6625,
+      "step": 21909
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.6702356338500977,
+      "learning_rate": 7.452356964090404e-06,
+      "loss": 0.6361,
+      "step": 21910
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.896691083908081,
+      "learning_rate": 7.452127193839036e-06,
+      "loss": 0.6265,
+      "step": 21911
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 3.0893607139587402,
+      "learning_rate": 7.451897416769221e-06,
+      "loss": 0.7958,
+      "step": 21912
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.5984928607940674,
+      "learning_rate": 7.4516676328815975e-06,
+      "loss": 0.6394,
+      "step": 21913
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.9664227962493896,
+      "learning_rate": 7.451437842176806e-06,
+      "loss": 0.5775,
+      "step": 21914
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.786895751953125,
+      "learning_rate": 7.451208044655482e-06,
+      "loss": 0.5615,
+      "step": 21915
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.890470504760742,
+      "learning_rate": 7.450978240318268e-06,
+      "loss": 0.8212,
+      "step": 21916
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 3.1377298831939697,
+      "learning_rate": 7.4507484291658e-06,
+      "loss": 0.8735,
+      "step": 21917
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.045923948287964,
+      "learning_rate": 7.450518611198721e-06,
+      "loss": 0.8583,
+      "step": 21918
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.1274309158325195,
+      "learning_rate": 7.450288786417665e-06,
+      "loss": 0.807,
+      "step": 21919
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.2171590328216553,
+      "learning_rate": 7.450058954823276e-06,
+      "loss": 0.5746,
+      "step": 21920
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 3.4294674396514893,
+      "learning_rate": 7.449829116416191e-06,
+      "loss": 0.6372,
+      "step": 21921
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.120854377746582,
+      "learning_rate": 7.449599271197048e-06,
+      "loss": 1.005,
+      "step": 21922
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.8672206401824951,
+      "learning_rate": 7.449369419166486e-06,
+      "loss": 0.7034,
+      "step": 21923
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.479992628097534,
+      "learning_rate": 7.449139560325147e-06,
+      "loss": 0.6159,
+      "step": 21924
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.497415781021118,
+      "learning_rate": 7.448909694673668e-06,
+      "loss": 0.6083,
+      "step": 21925
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.0837199687957764,
+      "learning_rate": 7.4486798222126885e-06,
+      "loss": 0.5196,
+      "step": 21926
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 3.82438588142395,
+      "learning_rate": 7.448449942942848e-06,
+      "loss": 0.6555,
+      "step": 21927
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.0146737098693848,
+      "learning_rate": 7.448220056864784e-06,
+      "loss": 0.7169,
+      "step": 21928
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 3.0756309032440186,
+      "learning_rate": 7.447990163979138e-06,
+      "loss": 0.9201,
+      "step": 21929
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 4.0194783210754395,
+      "learning_rate": 7.4477602642865495e-06,
+      "loss": 0.6068,
+      "step": 21930
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.351471185684204,
+      "learning_rate": 7.447530357787655e-06,
+      "loss": 0.7171,
+      "step": 21931
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.7802743911743164,
+      "learning_rate": 7.447300444483097e-06,
+      "loss": 0.8858,
+      "step": 21932
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.0136988162994385,
+      "learning_rate": 7.447070524373512e-06,
+      "loss": 0.5423,
+      "step": 21933
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.3378946781158447,
+      "learning_rate": 7.446840597459542e-06,
+      "loss": 0.553,
+      "step": 21934
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.1946396827697754,
+      "learning_rate": 7.446610663741823e-06,
+      "loss": 0.5777,
+      "step": 21935
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 3.6629257202148438,
+      "learning_rate": 7.4463807232209986e-06,
+      "loss": 0.8494,
+      "step": 21936
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.9228508472442627,
+      "learning_rate": 7.446150775897704e-06,
+      "loss": 0.6195,
+      "step": 21937
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.317768096923828,
+      "learning_rate": 7.44592082177258e-06,
+      "loss": 0.4672,
+      "step": 21938
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.949799060821533,
+      "learning_rate": 7.445690860846267e-06,
+      "loss": 0.7286,
+      "step": 21939
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 3.3132824897766113,
+      "learning_rate": 7.445460893119405e-06,
+      "loss": 0.8051,
+      "step": 21940
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.8887907266616821,
+      "learning_rate": 7.445230918592632e-06,
+      "loss": 0.5609,
+      "step": 21941
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.9911447763442993,
+      "learning_rate": 7.445000937266587e-06,
+      "loss": 0.6104,
+      "step": 21942
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 5.631313323974609,
+      "learning_rate": 7.444770949141911e-06,
+      "loss": 0.7819,
+      "step": 21943
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.8794827461242676,
+      "learning_rate": 7.444540954219242e-06,
+      "loss": 0.6315,
+      "step": 21944
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.8837578296661377,
+      "learning_rate": 7.444310952499221e-06,
+      "loss": 0.991,
+      "step": 21945
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.9977777004241943,
+      "learning_rate": 7.444080943982487e-06,
+      "loss": 0.6954,
+      "step": 21946
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.767627000808716,
+      "learning_rate": 7.443850928669679e-06,
+      "loss": 0.6331,
+      "step": 21947
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.890399217605591,
+      "learning_rate": 7.443620906561438e-06,
+      "loss": 0.8958,
+      "step": 21948
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 6.607972621917725,
+      "learning_rate": 7.443390877658401e-06,
+      "loss": 0.5235,
+      "step": 21949
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.5435720682144165,
+      "learning_rate": 7.44316084196121e-06,
+      "loss": 0.6026,
+      "step": 21950
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.857376217842102,
+      "learning_rate": 7.442930799470505e-06,
+      "loss": 0.4833,
+      "step": 21951
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.9439250230789185,
+      "learning_rate": 7.442700750186923e-06,
+      "loss": 0.7151,
+      "step": 21952
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.9890096187591553,
+      "learning_rate": 7.442470694111105e-06,
+      "loss": 0.7125,
+      "step": 21953
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.8987252712249756,
+      "learning_rate": 7.442240631243693e-06,
+      "loss": 0.6782,
+      "step": 21954
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.2391321659088135,
+      "learning_rate": 7.442010561585324e-06,
+      "loss": 0.8641,
+      "step": 21955
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 3.3363444805145264,
+      "learning_rate": 7.441780485136638e-06,
+      "loss": 0.9386,
+      "step": 21956
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.122795581817627,
+      "learning_rate": 7.441550401898277e-06,
+      "loss": 0.6844,
+      "step": 21957
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.9302788972854614,
+      "learning_rate": 7.441320311870875e-06,
+      "loss": 0.8685,
+      "step": 21958
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.39266300201416,
+      "learning_rate": 7.441090215055078e-06,
+      "loss": 0.6314,
+      "step": 21959
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.279669761657715,
+      "learning_rate": 7.4408601114515245e-06,
+      "loss": 0.8323,
+      "step": 21960
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.873448610305786,
+      "learning_rate": 7.440630001060852e-06,
+      "loss": 0.6125,
+      "step": 21961
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.8501242399215698,
+      "learning_rate": 7.440399883883702e-06,
+      "loss": 0.6469,
+      "step": 21962
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.8413225412368774,
+      "learning_rate": 7.440169759920714e-06,
+      "loss": 0.8028,
+      "step": 21963
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 5.015364646911621,
+      "learning_rate": 7.439939629172529e-06,
+      "loss": 0.5266,
+      "step": 21964
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.197067975997925,
+      "learning_rate": 7.439709491639786e-06,
+      "loss": 0.5064,
+      "step": 21965
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.3389017581939697,
+      "learning_rate": 7.439479347323124e-06,
+      "loss": 0.6297,
+      "step": 21966
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.2463064193725586,
+      "learning_rate": 7.439249196223184e-06,
+      "loss": 0.7722,
+      "step": 21967
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.63280987739563,
+      "learning_rate": 7.439019038340606e-06,
+      "loss": 0.3737,
+      "step": 21968
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.2484283447265625,
+      "learning_rate": 7.4387888736760284e-06,
+      "loss": 0.6403,
+      "step": 21969
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 3.7274820804595947,
+      "learning_rate": 7.438558702230093e-06,
+      "loss": 0.5436,
+      "step": 21970
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 3.1627633571624756,
+      "learning_rate": 7.43832852400344e-06,
+      "loss": 0.834,
+      "step": 21971
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 3.4637999534606934,
+      "learning_rate": 7.4380983389967084e-06,
+      "loss": 0.7598,
+      "step": 21972
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.8448898792266846,
+      "learning_rate": 7.4378681472105395e-06,
+      "loss": 0.6859,
+      "step": 21973
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.7418053150177002,
+      "learning_rate": 7.437637948645572e-06,
+      "loss": 0.6926,
+      "step": 21974
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 3.3202195167541504,
+      "learning_rate": 7.437407743302446e-06,
+      "loss": 0.8549,
+      "step": 21975
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.645972490310669,
+      "learning_rate": 7.4371775311818025e-06,
+      "loss": 0.7254,
+      "step": 21976
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.6537764072418213,
+      "learning_rate": 7.436947312284282e-06,
+      "loss": 0.6843,
+      "step": 21977
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.615288257598877,
+      "learning_rate": 7.436717086610523e-06,
+      "loss": 0.9508,
+      "step": 21978
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.5794973373413086,
+      "learning_rate": 7.4364868541611665e-06,
+      "loss": 0.5688,
+      "step": 21979
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 3.331441640853882,
+      "learning_rate": 7.436256614936853e-06,
+      "loss": 0.4842,
+      "step": 21980
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.9565906524658203,
+      "learning_rate": 7.436026368938223e-06,
+      "loss": 0.4936,
+      "step": 21981
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.373781681060791,
+      "learning_rate": 7.435796116165917e-06,
+      "loss": 0.8474,
+      "step": 21982
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.064307451248169,
+      "learning_rate": 7.435565856620572e-06,
+      "loss": 0.5016,
+      "step": 21983
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 3.8630764484405518,
+      "learning_rate": 7.4353355903028316e-06,
+      "loss": 0.5843,
+      "step": 21984
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.326577067375183,
+      "learning_rate": 7.435105317213336e-06,
+      "loss": 0.4758,
+      "step": 21985
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.099426031112671,
+      "learning_rate": 7.4348750373527246e-06,
+      "loss": 0.5734,
+      "step": 21986
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.7919907569885254,
+      "learning_rate": 7.434644750721638e-06,
+      "loss": 0.6675,
+      "step": 21987
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.186790704727173,
+      "learning_rate": 7.434414457320715e-06,
+      "loss": 0.7585,
+      "step": 21988
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.8887441158294678,
+      "learning_rate": 7.434184157150599e-06,
+      "loss": 0.781,
+      "step": 21989
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.097787618637085,
+      "learning_rate": 7.433953850211928e-06,
+      "loss": 0.8316,
+      "step": 21990
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.6560559272766113,
+      "learning_rate": 7.433723536505343e-06,
+      "loss": 0.4461,
+      "step": 21991
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.22196102142334,
+      "learning_rate": 7.433493216031484e-06,
+      "loss": 0.8857,
+      "step": 21992
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.9327120780944824,
+      "learning_rate": 7.433262888790993e-06,
+      "loss": 0.7561,
+      "step": 21993
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 3.477830648422241,
+      "learning_rate": 7.433032554784509e-06,
+      "loss": 0.7431,
+      "step": 21994
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.855273723602295,
+      "learning_rate": 7.432802214012674e-06,
+      "loss": 0.7166,
+      "step": 21995
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.4056568145751953,
+      "learning_rate": 7.432571866476127e-06,
+      "loss": 0.782,
+      "step": 21996
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.5708203315734863,
+      "learning_rate": 7.432341512175508e-06,
+      "loss": 0.7149,
+      "step": 21997
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.949779510498047,
+      "learning_rate": 7.432111151111458e-06,
+      "loss": 0.9208,
+      "step": 21998
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.6354610919952393,
+      "learning_rate": 7.431880783284619e-06,
+      "loss": 0.6984,
+      "step": 21999
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.4428913593292236,
+      "learning_rate": 7.431650408695631e-06,
+      "loss": 0.6955,
+      "step": 22000
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.9621994495391846,
+      "learning_rate": 7.431420027345133e-06,
+      "loss": 0.7738,
+      "step": 22001
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.6825915575027466,
+      "learning_rate": 7.431189639233767e-06,
+      "loss": 0.6918,
+      "step": 22002
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 3.023866653442383,
+      "learning_rate": 7.430959244362175e-06,
+      "loss": 0.5158,
+      "step": 22003
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.5109137296676636,
+      "learning_rate": 7.430728842730995e-06,
+      "loss": 0.5634,
+      "step": 22004
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.5667905807495117,
+      "learning_rate": 7.430498434340867e-06,
+      "loss": 0.519,
+      "step": 22005
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.4299001693725586,
+      "learning_rate": 7.430268019192435e-06,
+      "loss": 0.8318,
+      "step": 22006
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 5.367792129516602,
+      "learning_rate": 7.430037597286337e-06,
+      "loss": 0.7706,
+      "step": 22007
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.1438136100769043,
+      "learning_rate": 7.429807168623216e-06,
+      "loss": 0.7194,
+      "step": 22008
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.6686153411865234,
+      "learning_rate": 7.429576733203709e-06,
+      "loss": 0.5851,
+      "step": 22009
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.410003900527954,
+      "learning_rate": 7.429346291028462e-06,
+      "loss": 0.6877,
+      "step": 22010
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 3.5999460220336914,
+      "learning_rate": 7.429115842098112e-06,
+      "loss": 0.6086,
+      "step": 22011
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 3.072564125061035,
+      "learning_rate": 7.4288853864133015e-06,
+      "loss": 0.7411,
+      "step": 22012
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 4.373300552368164,
+      "learning_rate": 7.4286549239746696e-06,
+      "loss": 0.6057,
+      "step": 22013
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.1879444122314453,
+      "learning_rate": 7.428424454782859e-06,
+      "loss": 0.6162,
+      "step": 22014
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 4.751893043518066,
+      "learning_rate": 7.4281939788385095e-06,
+      "loss": 0.6177,
+      "step": 22015
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 3.7988460063934326,
+      "learning_rate": 7.427963496142262e-06,
+      "loss": 0.5066,
+      "step": 22016
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.120452880859375,
+      "learning_rate": 7.427733006694757e-06,
+      "loss": 0.7401,
+      "step": 22017
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.374342203140259,
+      "learning_rate": 7.427502510496638e-06,
+      "loss": 0.6437,
+      "step": 22018
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.4074381589889526,
+      "learning_rate": 7.427272007548541e-06,
+      "loss": 0.569,
+      "step": 22019
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.7868504524230957,
+      "learning_rate": 7.427041497851111e-06,
+      "loss": 0.727,
+      "step": 22020
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.068549871444702,
+      "learning_rate": 7.426810981404988e-06,
+      "loss": 0.5885,
+      "step": 22021
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.6883630752563477,
+      "learning_rate": 7.426580458210812e-06,
+      "loss": 0.6228,
+      "step": 22022
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 3.161106824874878,
+      "learning_rate": 7.426349928269226e-06,
+      "loss": 0.8718,
+      "step": 22023
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.3117454051971436,
+      "learning_rate": 7.42611939158087e-06,
+      "loss": 0.4957,
+      "step": 22024
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.980988621711731,
+      "learning_rate": 7.425888848146384e-06,
+      "loss": 0.4217,
+      "step": 22025
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.7431150674819946,
+      "learning_rate": 7.42565829796641e-06,
+      "loss": 0.7253,
+      "step": 22026
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.7509288787841797,
+      "learning_rate": 7.425427741041588e-06,
+      "loss": 0.9179,
+      "step": 22027
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.7825441360473633,
+      "learning_rate": 7.425197177372563e-06,
+      "loss": 0.8286,
+      "step": 22028
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.2539076805114746,
+      "learning_rate": 7.4249666069599715e-06,
+      "loss": 0.6425,
+      "step": 22029
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.296330690383911,
+      "learning_rate": 7.424736029804454e-06,
+      "loss": 0.6652,
+      "step": 22030
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.902323842048645,
+      "learning_rate": 7.4245054459066556e-06,
+      "loss": 0.5829,
+      "step": 22031
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.5889241695404053,
+      "learning_rate": 7.424274855267216e-06,
+      "loss": 0.7456,
+      "step": 22032
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.295186996459961,
+      "learning_rate": 7.424044257886777e-06,
+      "loss": 0.6706,
+      "step": 22033
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.908461570739746,
+      "learning_rate": 7.423813653765978e-06,
+      "loss": 0.874,
+      "step": 22034
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.8162827491760254,
+      "learning_rate": 7.423583042905463e-06,
+      "loss": 0.5718,
+      "step": 22035
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.8396447896957397,
+      "learning_rate": 7.423352425305868e-06,
+      "loss": 0.7195,
+      "step": 22036
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.9379501342773438,
+      "learning_rate": 7.423121800967841e-06,
+      "loss": 0.5554,
+      "step": 22037
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.626303195953369,
+      "learning_rate": 7.422891169892018e-06,
+      "loss": 0.4212,
+      "step": 22038
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 3.9554452896118164,
+      "learning_rate": 7.422660532079042e-06,
+      "loss": 0.6139,
+      "step": 22039
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 4.435739040374756,
+      "learning_rate": 7.422429887529556e-06,
+      "loss": 0.6355,
+      "step": 22040
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 3.1109068393707275,
+      "learning_rate": 7.422199236244199e-06,
+      "loss": 0.6362,
+      "step": 22041
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.130417823791504,
+      "learning_rate": 7.421968578223614e-06,
+      "loss": 0.6324,
+      "step": 22042
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.2896296977996826,
+      "learning_rate": 7.421737913468441e-06,
+      "loss": 0.3462,
+      "step": 22043
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.687687635421753,
+      "learning_rate": 7.421507241979322e-06,
+      "loss": 0.4333,
+      "step": 22044
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.23661732673645,
+      "learning_rate": 7.4212765637568995e-06,
+      "loss": 0.6766,
+      "step": 22045
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.6223639249801636,
+      "learning_rate": 7.421045878801813e-06,
+      "loss": 0.725,
+      "step": 22046
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 3.0617716312408447,
+      "learning_rate": 7.420815187114705e-06,
+      "loss": 0.855,
+      "step": 22047
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.4097626209259033,
+      "learning_rate": 7.420584488696216e-06,
+      "loss": 0.6261,
+      "step": 22048
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.2286527156829834,
+      "learning_rate": 7.420353783546991e-06,
+      "loss": 0.7825,
+      "step": 22049
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.5725046396255493,
+      "learning_rate": 7.420123071667667e-06,
+      "loss": 0.6561,
+      "step": 22050
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.489655017852783,
+      "learning_rate": 7.419892353058887e-06,
+      "loss": 0.7565,
+      "step": 22051
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.3904662132263184,
+      "learning_rate": 7.419661627721293e-06,
+      "loss": 0.777,
+      "step": 22052
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.6894617080688477,
+      "learning_rate": 7.419430895655526e-06,
+      "loss": 0.5926,
+      "step": 22053
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.309516668319702,
+      "learning_rate": 7.41920015686223e-06,
+      "loss": 0.6368,
+      "step": 22054
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.9916270971298218,
+      "learning_rate": 7.418969411342043e-06,
+      "loss": 0.688,
+      "step": 22055
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.929627776145935,
+      "learning_rate": 7.418738659095609e-06,
+      "loss": 0.5237,
+      "step": 22056
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.8789329528808594,
+      "learning_rate": 7.418507900123568e-06,
+      "loss": 0.5901,
+      "step": 22057
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.8379015922546387,
+      "learning_rate": 7.418277134426563e-06,
+      "loss": 0.7269,
+      "step": 22058
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 3.030360221862793,
+      "learning_rate": 7.418046362005236e-06,
+      "loss": 0.6548,
+      "step": 22059
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.9215432405471802,
+      "learning_rate": 7.417815582860226e-06,
+      "loss": 0.6546,
+      "step": 22060
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.9727829694747925,
+      "learning_rate": 7.417584796992177e-06,
+      "loss": 0.7191,
+      "step": 22061
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.295224666595459,
+      "learning_rate": 7.417354004401732e-06,
+      "loss": 0.934,
+      "step": 22062
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.6208219528198242,
+      "learning_rate": 7.4171232050895304e-06,
+      "loss": 0.5704,
+      "step": 22063
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.882683515548706,
+      "learning_rate": 7.416892399056214e-06,
+      "loss": 0.5797,
+      "step": 22064
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.3367698192596436,
+      "learning_rate": 7.416661586302426e-06,
+      "loss": 0.6956,
+      "step": 22065
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.4372143745422363,
+      "learning_rate": 7.416430766828807e-06,
+      "loss": 0.4213,
+      "step": 22066
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.5752650499343872,
+      "learning_rate": 7.416199940635999e-06,
+      "loss": 0.3588,
+      "step": 22067
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.4788408279418945,
+      "learning_rate": 7.415969107724646e-06,
+      "loss": 0.5558,
+      "step": 22068
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.697438955307007,
+      "learning_rate": 7.415738268095385e-06,
+      "loss": 0.6005,
+      "step": 22069
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.4131178855895996,
+      "learning_rate": 7.415507421748862e-06,
+      "loss": 0.4848,
+      "step": 22070
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.8383493423461914,
+      "learning_rate": 7.415276568685718e-06,
+      "loss": 0.7601,
+      "step": 22071
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.7314926385879517,
+      "learning_rate": 7.415045708906594e-06,
+      "loss": 0.5806,
+      "step": 22072
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 3.6681549549102783,
+      "learning_rate": 7.414814842412133e-06,
+      "loss": 0.6958,
+      "step": 22073
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.779874324798584,
+      "learning_rate": 7.414583969202978e-06,
+      "loss": 0.8717,
+      "step": 22074
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 3.1358368396759033,
+      "learning_rate": 7.4143530892797674e-06,
+      "loss": 0.5952,
+      "step": 22075
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.088404655456543,
+      "learning_rate": 7.414122202643146e-06,
+      "loss": 0.4894,
+      "step": 22076
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.8027613162994385,
+      "learning_rate": 7.413891309293754e-06,
+      "loss": 0.4699,
+      "step": 22077
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.4223597049713135,
+      "learning_rate": 7.4136604092322375e-06,
+      "loss": 0.6897,
+      "step": 22078
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.5110526084899902,
+      "learning_rate": 7.413429502459233e-06,
+      "loss": 0.6617,
+      "step": 22079
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.110537528991699,
+      "learning_rate": 7.4131985889753855e-06,
+      "loss": 0.9378,
+      "step": 22080
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 9.072430610656738,
+      "learning_rate": 7.412967668781337e-06,
+      "loss": 0.6484,
+      "step": 22081
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.2216062545776367,
+      "learning_rate": 7.41273674187773e-06,
+      "loss": 0.6764,
+      "step": 22082
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.9374455213546753,
+      "learning_rate": 7.412505808265206e-06,
+      "loss": 0.5407,
+      "step": 22083
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.129293203353882,
+      "learning_rate": 7.412274867944406e-06,
+      "loss": 0.546,
+      "step": 22084
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 3.227637767791748,
+      "learning_rate": 7.412043920915974e-06,
+      "loss": 0.4662,
+      "step": 22085
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.610255002975464,
+      "learning_rate": 7.411812967180552e-06,
+      "loss": 0.6846,
+      "step": 22086
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.6274458169937134,
+      "learning_rate": 7.41158200673878e-06,
+      "loss": 0.4315,
+      "step": 22087
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 3.339799404144287,
+      "learning_rate": 7.411351039591303e-06,
+      "loss": 0.5786,
+      "step": 22088
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.225076198577881,
+      "learning_rate": 7.411120065738761e-06,
+      "loss": 0.7201,
+      "step": 22089
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 9.61667537689209,
+      "learning_rate": 7.4108890851817984e-06,
+      "loss": 0.7089,
+      "step": 22090
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.1261167526245117,
+      "learning_rate": 7.410658097921055e-06,
+      "loss": 0.6187,
+      "step": 22091
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 3.986736536026001,
+      "learning_rate": 7.410427103957176e-06,
+      "loss": 0.3671,
+      "step": 22092
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.8192436695098877,
+      "learning_rate": 7.410196103290802e-06,
+      "loss": 0.5066,
+      "step": 22093
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 3.370683431625366,
+      "learning_rate": 7.409965095922576e-06,
+      "loss": 0.7977,
+      "step": 22094
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 3.0305209159851074,
+      "learning_rate": 7.409734081853138e-06,
+      "loss": 0.791,
+      "step": 22095
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.940885305404663,
+      "learning_rate": 7.4095030610831345e-06,
+      "loss": 0.9032,
+      "step": 22096
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.6293505430221558,
+      "learning_rate": 7.409272033613204e-06,
+      "loss": 0.5893,
+      "step": 22097
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.99159836769104,
+      "learning_rate": 7.409040999443991e-06,
+      "loss": 0.6925,
+      "step": 22098
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.3856430053710938,
+      "learning_rate": 7.4088099585761374e-06,
+      "loss": 0.425,
+      "step": 22099
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.180232048034668,
+      "learning_rate": 7.4085789110102855e-06,
+      "loss": 0.5132,
+      "step": 22100
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.7156645059585571,
+      "learning_rate": 7.408347856747078e-06,
+      "loss": 0.8635,
+      "step": 22101
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 3.0108611583709717,
+      "learning_rate": 7.4081167957871565e-06,
+      "loss": 0.7928,
+      "step": 22102
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 4.227993488311768,
+      "learning_rate": 7.407885728131165e-06,
+      "loss": 0.6765,
+      "step": 22103
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.9593555927276611,
+      "learning_rate": 7.407654653779746e-06,
+      "loss": 0.5511,
+      "step": 22104
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.5390279293060303,
+      "learning_rate": 7.407423572733541e-06,
+      "loss": 0.5839,
+      "step": 22105
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.134197473526001,
+      "learning_rate": 7.407192484993193e-06,
+      "loss": 0.6457,
+      "step": 22106
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.3577449321746826,
+      "learning_rate": 7.406961390559343e-06,
+      "loss": 0.8331,
+      "step": 22107
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.6427066326141357,
+      "learning_rate": 7.406730289432636e-06,
+      "loss": 0.6774,
+      "step": 22108
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 3.478713274002075,
+      "learning_rate": 7.406499181613715e-06,
+      "loss": 0.7501,
+      "step": 22109
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.8472399711608887,
+      "learning_rate": 7.4062680671032195e-06,
+      "loss": 0.7415,
+      "step": 22110
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 3.553994655609131,
+      "learning_rate": 7.406036945901794e-06,
+      "loss": 0.508,
+      "step": 22111
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.127704381942749,
+      "learning_rate": 7.405805818010082e-06,
+      "loss": 0.5169,
+      "step": 22112
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 3.921394109725952,
+      "learning_rate": 7.4055746834287245e-06,
+      "loss": 0.7251,
+      "step": 22113
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.032636880874634,
+      "learning_rate": 7.405343542158365e-06,
+      "loss": 0.7298,
+      "step": 22114
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.9113677740097046,
+      "learning_rate": 7.405112394199647e-06,
+      "loss": 0.5758,
+      "step": 22115
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.5402822494506836,
+      "learning_rate": 7.404881239553212e-06,
+      "loss": 0.4502,
+      "step": 22116
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.802188754081726,
+      "learning_rate": 7.4046500782197025e-06,
+      "loss": 0.6653,
+      "step": 22117
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.4602646827697754,
+      "learning_rate": 7.404418910199763e-06,
+      "loss": 0.6481,
+      "step": 22118
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.1859817504882812,
+      "learning_rate": 7.404187735494035e-06,
+      "loss": 0.6664,
+      "step": 22119
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.4993884563446045,
+      "learning_rate": 7.403956554103161e-06,
+      "loss": 0.68,
+      "step": 22120
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.2634074687957764,
+      "learning_rate": 7.403725366027784e-06,
+      "loss": 0.5304,
+      "step": 22121
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 4.178030967712402,
+      "learning_rate": 7.403494171268548e-06,
+      "loss": 0.5905,
+      "step": 22122
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.014578342437744,
+      "learning_rate": 7.403262969826095e-06,
+      "loss": 0.4542,
+      "step": 22123
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.2277071475982666,
+      "learning_rate": 7.403031761701068e-06,
+      "loss": 0.436,
+      "step": 22124
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.6234872341156006,
+      "learning_rate": 7.402800546894109e-06,
+      "loss": 0.7756,
+      "step": 22125
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.859502911567688,
+      "learning_rate": 7.402569325405863e-06,
+      "loss": 0.8122,
+      "step": 22126
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.2648353576660156,
+      "learning_rate": 7.402338097236972e-06,
+      "loss": 0.5437,
+      "step": 22127
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.5920326709747314,
+      "learning_rate": 7.402106862388078e-06,
+      "loss": 0.6435,
+      "step": 22128
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.8075296878814697,
+      "learning_rate": 7.401875620859824e-06,
+      "loss": 0.647,
+      "step": 22129
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.271912097930908,
+      "learning_rate": 7.401644372652854e-06,
+      "loss": 0.7549,
+      "step": 22130
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.3243567943573,
+      "learning_rate": 7.401413117767811e-06,
+      "loss": 0.4257,
+      "step": 22131
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 4.207424640655518,
+      "learning_rate": 7.4011818562053375e-06,
+      "loss": 1.0019,
+      "step": 22132
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.1650514602661133,
+      "learning_rate": 7.400950587966077e-06,
+      "loss": 0.7702,
+      "step": 22133
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.9432368278503418,
+      "learning_rate": 7.400719313050673e-06,
+      "loss": 0.4523,
+      "step": 22134
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.9207361936569214,
+      "learning_rate": 7.400488031459766e-06,
+      "loss": 0.4989,
+      "step": 22135
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 3.6934421062469482,
+      "learning_rate": 7.400256743194003e-06,
+      "loss": 0.6297,
+      "step": 22136
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.9829721450805664,
+      "learning_rate": 7.400025448254025e-06,
+      "loss": 0.4875,
+      "step": 22137
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.6755048036575317,
+      "learning_rate": 7.399794146640473e-06,
+      "loss": 0.6787,
+      "step": 22138
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.8814997673034668,
+      "learning_rate": 7.399562838353995e-06,
+      "loss": 0.5442,
+      "step": 22139
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.3116822242736816,
+      "learning_rate": 7.39933152339523e-06,
+      "loss": 0.7772,
+      "step": 22140
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.0107216835021973,
+      "learning_rate": 7.3991002017648235e-06,
+      "loss": 0.4742,
+      "step": 22141
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.988110065460205,
+      "learning_rate": 7.398868873463418e-06,
+      "loss": 0.6512,
+      "step": 22142
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.1262409687042236,
+      "learning_rate": 7.398637538491658e-06,
+      "loss": 0.4138,
+      "step": 22143
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.2709200382232666,
+      "learning_rate": 7.398406196850184e-06,
+      "loss": 0.4903,
+      "step": 22144
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.351260781288147,
+      "learning_rate": 7.398174848539641e-06,
+      "loss": 0.6774,
+      "step": 22145
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.7341171503067017,
+      "learning_rate": 7.397943493560673e-06,
+      "loss": 0.6042,
+      "step": 22146
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.073716878890991,
+      "learning_rate": 7.397712131913921e-06,
+      "loss": 0.6914,
+      "step": 22147
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.790780782699585,
+      "learning_rate": 7.397480763600031e-06,
+      "loss": 0.7871,
+      "step": 22148
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.399179458618164,
+      "learning_rate": 7.397249388619645e-06,
+      "loss": 0.7632,
+      "step": 22149
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.9793778657913208,
+      "learning_rate": 7.397018006973407e-06,
+      "loss": 0.7042,
+      "step": 22150
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.8889129161834717,
+      "learning_rate": 7.396786618661958e-06,
+      "loss": 0.5379,
+      "step": 22151
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.6973474025726318,
+      "learning_rate": 7.396555223685944e-06,
+      "loss": 0.4936,
+      "step": 22152
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.271246910095215,
+      "learning_rate": 7.396323822046008e-06,
+      "loss": 0.4578,
+      "step": 22153
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.5674196481704712,
+      "learning_rate": 7.396092413742793e-06,
+      "loss": 0.4951,
+      "step": 22154
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.24885630607605,
+      "learning_rate": 7.3958609987769435e-06,
+      "loss": 0.4848,
+      "step": 22155
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.9113025665283203,
+      "learning_rate": 7.3956295771491015e-06,
+      "loss": 0.74,
+      "step": 22156
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.7386558055877686,
+      "learning_rate": 7.395398148859911e-06,
+      "loss": 0.4,
+      "step": 22157
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.6799824237823486,
+      "learning_rate": 7.3951667139100156e-06,
+      "loss": 0.7468,
+      "step": 22158
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.4219138622283936,
+      "learning_rate": 7.394935272300059e-06,
+      "loss": 0.6557,
+      "step": 22159
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.0966250896453857,
+      "learning_rate": 7.394703824030685e-06,
+      "loss": 0.4179,
+      "step": 22160
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.228757619857788,
+      "learning_rate": 7.394472369102535e-06,
+      "loss": 0.8661,
+      "step": 22161
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 3.645111083984375,
+      "learning_rate": 7.3942409075162544e-06,
+      "loss": 0.6023,
+      "step": 22162
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.625243067741394,
+      "learning_rate": 7.3940094392724895e-06,
+      "loss": 0.717,
+      "step": 22163
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 3.401413917541504,
+      "learning_rate": 7.393777964371879e-06,
+      "loss": 0.8234,
+      "step": 22164
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.7936639785766602,
+      "learning_rate": 7.393546482815068e-06,
+      "loss": 0.681,
+      "step": 22165
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 3.0323352813720703,
+      "learning_rate": 7.393314994602703e-06,
+      "loss": 0.5914,
+      "step": 22166
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 3.334520101547241,
+      "learning_rate": 7.393083499735425e-06,
+      "loss": 1.07,
+      "step": 22167
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 3.0074563026428223,
+      "learning_rate": 7.392851998213877e-06,
+      "loss": 0.6338,
+      "step": 22168
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.8056061267852783,
+      "learning_rate": 7.392620490038704e-06,
+      "loss": 0.7558,
+      "step": 22169
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.263460159301758,
+      "learning_rate": 7.392388975210551e-06,
+      "loss": 0.8241,
+      "step": 22170
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.508333683013916,
+      "learning_rate": 7.39215745373006e-06,
+      "loss": 0.6263,
+      "step": 22171
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.1795756816864014,
+      "learning_rate": 7.391925925597876e-06,
+      "loss": 0.4553,
+      "step": 22172
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 3.551374912261963,
+      "learning_rate": 7.391694390814641e-06,
+      "loss": 0.5275,
+      "step": 22173
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.4169764518737793,
+      "learning_rate": 7.391462849380999e-06,
+      "loss": 0.5675,
+      "step": 22174
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.374080181121826,
+      "learning_rate": 7.391231301297595e-06,
+      "loss": 0.4267,
+      "step": 22175
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.0610573291778564,
+      "learning_rate": 7.390999746565074e-06,
+      "loss": 0.6632,
+      "step": 22176
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.5496602058410645,
+      "learning_rate": 7.390768185184079e-06,
+      "loss": 0.4051,
+      "step": 22177
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.5493760108947754,
+      "learning_rate": 7.390536617155251e-06,
+      "loss": 0.7701,
+      "step": 22178
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 3.32127046585083,
+      "learning_rate": 7.390305042479237e-06,
+      "loss": 0.5777,
+      "step": 22179
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 4.0615105628967285,
+      "learning_rate": 7.390073461156681e-06,
+      "loss": 0.5356,
+      "step": 22180
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.19079327583313,
+      "learning_rate": 7.389841873188225e-06,
+      "loss": 0.6328,
+      "step": 22181
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.6910990476608276,
+      "learning_rate": 7.3896102785745125e-06,
+      "loss": 0.7551,
+      "step": 22182
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.50729238986969,
+      "learning_rate": 7.389378677316191e-06,
+      "loss": 0.7363,
+      "step": 22183
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.2559092044830322,
+      "learning_rate": 7.389147069413903e-06,
+      "loss": 0.5865,
+      "step": 22184
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.610460638999939,
+      "learning_rate": 7.388915454868292e-06,
+      "loss": 0.4355,
+      "step": 22185
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.4321155548095703,
+      "learning_rate": 7.388683833680001e-06,
+      "loss": 0.6987,
+      "step": 22186
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.906266450881958,
+      "learning_rate": 7.3884522058496755e-06,
+      "loss": 0.7097,
+      "step": 22187
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 4.855729579925537,
+      "learning_rate": 7.388220571377959e-06,
+      "loss": 0.6685,
+      "step": 22188
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.5186171531677246,
+      "learning_rate": 7.387988930265495e-06,
+      "loss": 0.3804,
+      "step": 22189
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.5761436223983765,
+      "learning_rate": 7.38775728251293e-06,
+      "loss": 0.7379,
+      "step": 22190
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.314002513885498,
+      "learning_rate": 7.3875256281209065e-06,
+      "loss": 0.5401,
+      "step": 22191
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.3107478618621826,
+      "learning_rate": 7.387293967090065e-06,
+      "loss": 0.6369,
+      "step": 22192
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 4.7061944007873535,
+      "learning_rate": 7.387062299421057e-06,
+      "loss": 0.4642,
+      "step": 22193
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.9854975938796997,
+      "learning_rate": 7.386830625114522e-06,
+      "loss": 0.5526,
+      "step": 22194
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.4858899116516113,
+      "learning_rate": 7.386598944171105e-06,
+      "loss": 0.589,
+      "step": 22195
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.6698967218399048,
+      "learning_rate": 7.386367256591451e-06,
+      "loss": 0.5533,
+      "step": 22196
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.7871869802474976,
+      "learning_rate": 7.386135562376203e-06,
+      "loss": 0.6877,
+      "step": 22197
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.6049261093139648,
+      "learning_rate": 7.385903861526006e-06,
+      "loss": 0.4068,
+      "step": 22198
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 5.191353797912598,
+      "learning_rate": 7.3856721540415045e-06,
+      "loss": 0.4654,
+      "step": 22199
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.6617326736450195,
+      "learning_rate": 7.385440439923342e-06,
+      "loss": 0.7658,
+      "step": 22200
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.1591005325317383,
+      "learning_rate": 7.385208719172164e-06,
+      "loss": 0.7221,
+      "step": 22201
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.967918872833252,
+      "learning_rate": 7.384976991788612e-06,
+      "loss": 0.6044,
+      "step": 22202
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.271145820617676,
+      "learning_rate": 7.384745257773333e-06,
+      "loss": 0.4222,
+      "step": 22203
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.00565242767334,
+      "learning_rate": 7.384513517126972e-06,
+      "loss": 0.6053,
+      "step": 22204
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 3.1787989139556885,
+      "learning_rate": 7.384281769850172e-06,
+      "loss": 0.9022,
+      "step": 22205
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.1828269958496094,
+      "learning_rate": 7.384050015943577e-06,
+      "loss": 0.6128,
+      "step": 22206
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.3003120422363281,
+      "learning_rate": 7.383818255407833e-06,
+      "loss": 0.6204,
+      "step": 22207
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 3.885012626647949,
+      "learning_rate": 7.383586488243583e-06,
+      "loss": 0.7309,
+      "step": 22208
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.3897390365600586,
+      "learning_rate": 7.383354714451471e-06,
+      "loss": 0.7864,
+      "step": 22209
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.2229671478271484,
+      "learning_rate": 7.383122934032143e-06,
+      "loss": 0.7222,
+      "step": 22210
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 3.2373385429382324,
+      "learning_rate": 7.382891146986243e-06,
+      "loss": 0.5382,
+      "step": 22211
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.8998684883117676,
+      "learning_rate": 7.382659353314415e-06,
+      "loss": 0.6403,
+      "step": 22212
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.1356124877929688,
+      "learning_rate": 7.382427553017304e-06,
+      "loss": 0.6788,
+      "step": 22213
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.1993396282196045,
+      "learning_rate": 7.382195746095554e-06,
+      "loss": 0.8255,
+      "step": 22214
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 3.2077443599700928,
+      "learning_rate": 7.381963932549811e-06,
+      "loss": 0.5999,
+      "step": 22215
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.291947841644287,
+      "learning_rate": 7.381732112380718e-06,
+      "loss": 0.2155,
+      "step": 22216
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.998417377471924,
+      "learning_rate": 7.3815002855889205e-06,
+      "loss": 0.6611,
+      "step": 22217
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.209892511367798,
+      "learning_rate": 7.381268452175063e-06,
+      "loss": 0.5602,
+      "step": 22218
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.3251564502716064,
+      "learning_rate": 7.381036612139789e-06,
+      "loss": 0.6672,
+      "step": 22219
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.037022829055786,
+      "learning_rate": 7.3808047654837455e-06,
+      "loss": 0.6061,
+      "step": 22220
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 3.7253549098968506,
+      "learning_rate": 7.380572912207576e-06,
+      "loss": 0.7584,
+      "step": 22221
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.1850438117980957,
+      "learning_rate": 7.380341052311924e-06,
+      "loss": 0.6183,
+      "step": 22222
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.541274309158325,
+      "learning_rate": 7.380109185797434e-06,
+      "loss": 0.7009,
+      "step": 22223
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.8497776985168457,
+      "learning_rate": 7.379877312664754e-06,
+      "loss": 0.5061,
+      "step": 22224
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.063720941543579,
+      "learning_rate": 7.379645432914526e-06,
+      "loss": 0.5998,
+      "step": 22225
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.9858827590942383,
+      "learning_rate": 7.379413546547396e-06,
+      "loss": 0.7331,
+      "step": 22226
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.0251739025115967,
+      "learning_rate": 7.379181653564007e-06,
+      "loss": 0.8455,
+      "step": 22227
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.9885867834091187,
+      "learning_rate": 7.378949753965005e-06,
+      "loss": 0.7367,
+      "step": 22228
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.8701698780059814,
+      "learning_rate": 7.378717847751037e-06,
+      "loss": 0.7839,
+      "step": 22229
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.950331211090088,
+      "learning_rate": 7.378485934922745e-06,
+      "loss": 0.6035,
+      "step": 22230
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.8572709560394287,
+      "learning_rate": 7.378254015480772e-06,
+      "loss": 0.7902,
+      "step": 22231
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.7680147886276245,
+      "learning_rate": 7.378022089425769e-06,
+      "loss": 0.9433,
+      "step": 22232
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.9673688411712646,
+      "learning_rate": 7.377790156758375e-06,
+      "loss": 0.5308,
+      "step": 22233
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.5586011409759521,
+      "learning_rate": 7.377558217479239e-06,
+      "loss": 0.7497,
+      "step": 22234
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.8146778345108032,
+      "learning_rate": 7.3773262715890045e-06,
+      "loss": 0.6599,
+      "step": 22235
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.59478497505188,
+      "learning_rate": 7.377094319088316e-06,
+      "loss": 0.6777,
+      "step": 22236
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 4.056361675262451,
+      "learning_rate": 7.3768623599778165e-06,
+      "loss": 0.7369,
+      "step": 22237
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.5746164321899414,
+      "learning_rate": 7.376630394258155e-06,
+      "loss": 0.6322,
+      "step": 22238
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.1458935737609863,
+      "learning_rate": 7.376398421929975e-06,
+      "loss": 0.6109,
+      "step": 22239
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 3.5737056732177734,
+      "learning_rate": 7.376166442993919e-06,
+      "loss": 0.7472,
+      "step": 22240
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.9848041534423828,
+      "learning_rate": 7.3759344574506375e-06,
+      "loss": 0.6192,
+      "step": 22241
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.5912858247756958,
+      "learning_rate": 7.37570246530077e-06,
+      "loss": 0.4916,
+      "step": 22242
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 3.296734571456909,
+      "learning_rate": 7.375470466544964e-06,
+      "loss": 0.9277,
+      "step": 22243
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.987881064414978,
+      "learning_rate": 7.375238461183864e-06,
+      "loss": 0.7432,
+      "step": 22244
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.466881513595581,
+      "learning_rate": 7.375006449218118e-06,
+      "loss": 0.7591,
+      "step": 22245
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.7709269523620605,
+      "learning_rate": 7.374774430648367e-06,
+      "loss": 0.9379,
+      "step": 22246
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 5.203693389892578,
+      "learning_rate": 7.374542405475259e-06,
+      "loss": 0.6657,
+      "step": 22247
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 4.223521709442139,
+      "learning_rate": 7.374310373699437e-06,
+      "loss": 0.7413,
+      "step": 22248
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.3041415214538574,
+      "learning_rate": 7.374078335321548e-06,
+      "loss": 0.4307,
+      "step": 22249
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.9127147197723389,
+      "learning_rate": 7.373846290342237e-06,
+      "loss": 0.6001,
+      "step": 22250
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.358125925064087,
+      "learning_rate": 7.373614238762148e-06,
+      "loss": 0.3197,
+      "step": 22251
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.4365248680114746,
+      "learning_rate": 7.3733821805819275e-06,
+      "loss": 0.8253,
+      "step": 22252
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.7704317569732666,
+      "learning_rate": 7.373150115802218e-06,
+      "loss": 0.5375,
+      "step": 22253
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.177894115447998,
+      "learning_rate": 7.372918044423669e-06,
+      "loss": 0.4704,
+      "step": 22254
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.3694770336151123,
+      "learning_rate": 7.372685966446924e-06,
+      "loss": 0.4552,
+      "step": 22255
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 3.7259771823883057,
+      "learning_rate": 7.372453881872628e-06,
+      "loss": 1.0394,
+      "step": 22256
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.9354286193847656,
+      "learning_rate": 7.372221790701426e-06,
+      "loss": 0.7716,
+      "step": 22257
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.2372896671295166,
+      "learning_rate": 7.371989692933964e-06,
+      "loss": 0.5569,
+      "step": 22258
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.015252113342285,
+      "learning_rate": 7.371757588570887e-06,
+      "loss": 0.5613,
+      "step": 22259
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.9582058191299438,
+      "learning_rate": 7.371525477612842e-06,
+      "loss": 0.6692,
+      "step": 22260
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.8239854574203491,
+      "learning_rate": 7.371293360060472e-06,
+      "loss": 0.6166,
+      "step": 22261
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.234135866165161,
+      "learning_rate": 7.371061235914423e-06,
+      "loss": 0.7978,
+      "step": 22262
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.852034091949463,
+      "learning_rate": 7.370829105175341e-06,
+      "loss": 0.5056,
+      "step": 22263
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.221137762069702,
+      "learning_rate": 7.370596967843872e-06,
+      "loss": 0.6133,
+      "step": 22264
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.7956762313842773,
+      "learning_rate": 7.37036482392066e-06,
+      "loss": 0.5977,
+      "step": 22265
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 3.271456003189087,
+      "learning_rate": 7.3701326734063516e-06,
+      "loss": 0.7108,
+      "step": 22266
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.5298740863800049,
+      "learning_rate": 7.369900516301594e-06,
+      "loss": 0.7142,
+      "step": 22267
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.180826425552368,
+      "learning_rate": 7.369668352607029e-06,
+      "loss": 0.881,
+      "step": 22268
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.8584116697311401,
+      "learning_rate": 7.369436182323304e-06,
+      "loss": 0.7086,
+      "step": 22269
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 3.062570571899414,
+      "learning_rate": 7.369204005451065e-06,
+      "loss": 0.7018,
+      "step": 22270
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.2135467529296875,
+      "learning_rate": 7.368971821990958e-06,
+      "loss": 0.6245,
+      "step": 22271
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.923597574234009,
+      "learning_rate": 7.368739631943627e-06,
+      "loss": 0.8405,
+      "step": 22272
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.4474995136260986,
+      "learning_rate": 7.368507435309718e-06,
+      "loss": 0.5261,
+      "step": 22273
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.0204031467437744,
+      "learning_rate": 7.368275232089876e-06,
+      "loss": 0.554,
+      "step": 22274
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.0860729217529297,
+      "learning_rate": 7.36804302228475e-06,
+      "loss": 0.5801,
+      "step": 22275
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.8259758949279785,
+      "learning_rate": 7.367810805894983e-06,
+      "loss": 0.7971,
+      "step": 22276
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.087712287902832,
+      "learning_rate": 7.367578582921221e-06,
+      "loss": 0.652,
+      "step": 22277
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 3.0839333534240723,
+      "learning_rate": 7.3673463533641095e-06,
+      "loss": 0.6839,
+      "step": 22278
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.405174970626831,
+      "learning_rate": 7.367114117224296e-06,
+      "loss": 0.6453,
+      "step": 22279
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.419450283050537,
+      "learning_rate": 7.3668818745024225e-06,
+      "loss": 0.5856,
+      "step": 22280
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 7.719054698944092,
+      "learning_rate": 7.366649625199139e-06,
+      "loss": 0.4672,
+      "step": 22281
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.5532689094543457,
+      "learning_rate": 7.366417369315089e-06,
+      "loss": 0.6184,
+      "step": 22282
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.780382752418518,
+      "learning_rate": 7.366185106850918e-06,
+      "loss": 0.6087,
+      "step": 22283
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.411081075668335,
+      "learning_rate": 7.365952837807273e-06,
+      "loss": 0.5644,
+      "step": 22284
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.147761106491089,
+      "learning_rate": 7.365720562184799e-06,
+      "loss": 0.5598,
+      "step": 22285
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 3.0001919269561768,
+      "learning_rate": 7.365488279984144e-06,
+      "loss": 0.6233,
+      "step": 22286
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.563884735107422,
+      "learning_rate": 7.365255991205951e-06,
+      "loss": 0.8239,
+      "step": 22287
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.6611692905426025,
+      "learning_rate": 7.365023695850868e-06,
+      "loss": 0.5697,
+      "step": 22288
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 3.1399006843566895,
+      "learning_rate": 7.36479139391954e-06,
+      "loss": 0.6433,
+      "step": 22289
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.993163585662842,
+      "learning_rate": 7.364559085412611e-06,
+      "loss": 0.7944,
+      "step": 22290
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.8849403858184814,
+      "learning_rate": 7.364326770330732e-06,
+      "loss": 0.7128,
+      "step": 22291
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.1353094577789307,
+      "learning_rate": 7.364094448674544e-06,
+      "loss": 0.9258,
+      "step": 22292
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.8214473724365234,
+      "learning_rate": 7.363862120444694e-06,
+      "loss": 0.3099,
+      "step": 22293
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.182727813720703,
+      "learning_rate": 7.363629785641831e-06,
+      "loss": 0.6957,
+      "step": 22294
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 3.0512826442718506,
+      "learning_rate": 7.363397444266598e-06,
+      "loss": 0.268,
+      "step": 22295
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.953588604927063,
+      "learning_rate": 7.363165096319642e-06,
+      "loss": 0.6089,
+      "step": 22296
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.8392488956451416,
+      "learning_rate": 7.3629327418016095e-06,
+      "loss": 0.9565,
+      "step": 22297
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.107961893081665,
+      "learning_rate": 7.362700380713146e-06,
+      "loss": 0.6466,
+      "step": 22298
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.924023389816284,
+      "learning_rate": 7.362468013054897e-06,
+      "loss": 0.9961,
+      "step": 22299
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.8182899951934814,
+      "learning_rate": 7.362235638827512e-06,
+      "loss": 0.5601,
+      "step": 22300
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.8625648021697998,
+      "learning_rate": 7.362003258031632e-06,
+      "loss": 0.4996,
+      "step": 22301
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.7419936656951904,
+      "learning_rate": 7.361770870667908e-06,
+      "loss": 0.6089,
+      "step": 22302
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 4.040568828582764,
+      "learning_rate": 7.361538476736982e-06,
+      "loss": 0.5553,
+      "step": 22303
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.9151830673217773,
+      "learning_rate": 7.3613060762395025e-06,
+      "loss": 0.8625,
+      "step": 22304
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.9589457511901855,
+      "learning_rate": 7.361073669176116e-06,
+      "loss": 0.5358,
+      "step": 22305
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.9199761152267456,
+      "learning_rate": 7.360841255547468e-06,
+      "loss": 0.7147,
+      "step": 22306
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 4.884955406188965,
+      "learning_rate": 7.360608835354205e-06,
+      "loss": 0.5012,
+      "step": 22307
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 3.496164560317993,
+      "learning_rate": 7.360376408596973e-06,
+      "loss": 0.8008,
+      "step": 22308
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.2627038955688477,
+      "learning_rate": 7.360143975276418e-06,
+      "loss": 0.6434,
+      "step": 22309
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.2262074947357178,
+      "learning_rate": 7.359911535393187e-06,
+      "loss": 0.7842,
+      "step": 22310
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.7045612335205078,
+      "learning_rate": 7.359679088947926e-06,
+      "loss": 0.6004,
+      "step": 22311
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.7605892419815063,
+      "learning_rate": 7.3594466359412815e-06,
+      "loss": 0.6415,
+      "step": 22312
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.5188326835632324,
+      "learning_rate": 7.359214176373901e-06,
+      "loss": 0.6416,
+      "step": 22313
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.28045916557312,
+      "learning_rate": 7.3589817102464266e-06,
+      "loss": 0.7471,
+      "step": 22314
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.2493278980255127,
+      "learning_rate": 7.35874923755951e-06,
+      "loss": 0.5602,
+      "step": 22315
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.7328134775161743,
+      "learning_rate": 7.358516758313796e-06,
+      "loss": 0.5308,
+      "step": 22316
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.972420334815979,
+      "learning_rate": 7.358284272509929e-06,
+      "loss": 0.8632,
+      "step": 22317
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.1307833194732666,
+      "learning_rate": 7.358051780148558e-06,
+      "loss": 0.7693,
+      "step": 22318
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.3595638275146484,
+      "learning_rate": 7.357819281230327e-06,
+      "loss": 0.545,
+      "step": 22319
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.0560498237609863,
+      "learning_rate": 7.357586775755885e-06,
+      "loss": 0.5754,
+      "step": 22320
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.197248935699463,
+      "learning_rate": 7.357354263725877e-06,
+      "loss": 0.7911,
+      "step": 22321
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.6678873300552368,
+      "learning_rate": 7.3571217451409495e-06,
+      "loss": 0.7475,
+      "step": 22322
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.63737952709198,
+      "learning_rate": 7.356889220001749e-06,
+      "loss": 0.6011,
+      "step": 22323
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.055211305618286,
+      "learning_rate": 7.356656688308923e-06,
+      "loss": 0.747,
+      "step": 22324
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.750467538833618,
+      "learning_rate": 7.356424150063118e-06,
+      "loss": 0.5218,
+      "step": 22325
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.6648569107055664,
+      "learning_rate": 7.35619160526498e-06,
+      "loss": 0.7568,
+      "step": 22326
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 3.63238787651062,
+      "learning_rate": 7.3559590539151555e-06,
+      "loss": 0.64,
+      "step": 22327
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.9466792345046997,
+      "learning_rate": 7.355726496014291e-06,
+      "loss": 0.4125,
+      "step": 22328
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.2329277992248535,
+      "learning_rate": 7.355493931563035e-06,
+      "loss": 0.6405,
+      "step": 22329
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.2393598556518555,
+      "learning_rate": 7.355261360562032e-06,
+      "loss": 0.5779,
+      "step": 22330
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.794344186782837,
+      "learning_rate": 7.355028783011929e-06,
+      "loss": 0.5897,
+      "step": 22331
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.2664822340011597,
+      "learning_rate": 7.3547961989133745e-06,
+      "loss": 0.364,
+      "step": 22332
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.3368043899536133,
+      "learning_rate": 7.354563608267013e-06,
+      "loss": 0.6343,
+      "step": 22333
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.1337826251983643,
+      "learning_rate": 7.354331011073492e-06,
+      "loss": 0.6004,
+      "step": 22334
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.0933680534362793,
+      "learning_rate": 7.354098407333458e-06,
+      "loss": 0.4457,
+      "step": 22335
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.160259485244751,
+      "learning_rate": 7.353865797047559e-06,
+      "loss": 0.8063,
+      "step": 22336
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.3789222240447998,
+      "learning_rate": 7.35363318021644e-06,
+      "loss": 0.5164,
+      "step": 22337
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.305047035217285,
+      "learning_rate": 7.353400556840749e-06,
+      "loss": 0.7991,
+      "step": 22338
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.2121212482452393,
+      "learning_rate": 7.353167926921134e-06,
+      "loss": 0.7514,
+      "step": 22339
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.8178977966308594,
+      "learning_rate": 7.35293529045824e-06,
+      "loss": 0.6465,
+      "step": 22340
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.7390626668930054,
+      "learning_rate": 7.3527026474527144e-06,
+      "loss": 0.8145,
+      "step": 22341
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 3.274026870727539,
+      "learning_rate": 7.352469997905204e-06,
+      "loss": 0.6447,
+      "step": 22342
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.968595266342163,
+      "learning_rate": 7.3522373418163555e-06,
+      "loss": 0.6112,
+      "step": 22343
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 3.9846112728118896,
+      "learning_rate": 7.352004679186816e-06,
+      "loss": 0.7134,
+      "step": 22344
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 3.934462308883667,
+      "learning_rate": 7.351772010017232e-06,
+      "loss": 0.5673,
+      "step": 22345
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.7691954374313354,
+      "learning_rate": 7.351539334308252e-06,
+      "loss": 0.7592,
+      "step": 22346
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.927505373954773,
+      "learning_rate": 7.351306652060522e-06,
+      "loss": 0.6423,
+      "step": 22347
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.7691309452056885,
+      "learning_rate": 7.351073963274689e-06,
+      "loss": 0.6152,
+      "step": 22348
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.8349087238311768,
+      "learning_rate": 7.3508412679514e-06,
+      "loss": 0.6841,
+      "step": 22349
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.1220626831054688,
+      "learning_rate": 7.350608566091302e-06,
+      "loss": 0.4894,
+      "step": 22350
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 3.719541072845459,
+      "learning_rate": 7.350375857695042e-06,
+      "loss": 0.4649,
+      "step": 22351
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.008359670639038,
+      "learning_rate": 7.3501431427632675e-06,
+      "loss": 0.8188,
+      "step": 22352
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.0809237957000732,
+      "learning_rate": 7.349910421296624e-06,
+      "loss": 0.663,
+      "step": 22353
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.0111207962036133,
+      "learning_rate": 7.3496776932957614e-06,
+      "loss": 0.6491,
+      "step": 22354
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.5798068046569824,
+      "learning_rate": 7.349444958761325e-06,
+      "loss": 0.6288,
+      "step": 22355
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.7968902587890625,
+      "learning_rate": 7.349212217693962e-06,
+      "loss": 0.5058,
+      "step": 22356
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.9787986278533936,
+      "learning_rate": 7.348979470094321e-06,
+      "loss": 0.5338,
+      "step": 22357
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.9445408582687378,
+      "learning_rate": 7.348746715963047e-06,
+      "loss": 0.5641,
+      "step": 22358
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.1131908893585205,
+      "learning_rate": 7.348513955300788e-06,
+      "loss": 0.4148,
+      "step": 22359
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.170833110809326,
+      "learning_rate": 7.34828118810819e-06,
+      "loss": 0.8608,
+      "step": 22360
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.3126235008239746,
+      "learning_rate": 7.348048414385904e-06,
+      "loss": 0.822,
+      "step": 22361
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.3045467138290405,
+      "learning_rate": 7.347815634134575e-06,
+      "loss": 0.511,
+      "step": 22362
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 4.192430019378662,
+      "learning_rate": 7.347582847354851e-06,
+      "loss": 0.762,
+      "step": 22363
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.867285966873169,
+      "learning_rate": 7.347350054047377e-06,
+      "loss": 0.4549,
+      "step": 22364
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.604304313659668,
+      "learning_rate": 7.347117254212801e-06,
+      "loss": 0.3277,
+      "step": 22365
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.7183825969696045,
+      "learning_rate": 7.346884447851772e-06,
+      "loss": 0.5571,
+      "step": 22366
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.3136374950408936,
+      "learning_rate": 7.3466516349649365e-06,
+      "loss": 0.8867,
+      "step": 22367
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.4801011085510254,
+      "learning_rate": 7.346418815552943e-06,
+      "loss": 0.338,
+      "step": 22368
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.7341076135635376,
+      "learning_rate": 7.346185989616437e-06,
+      "loss": 0.8071,
+      "step": 22369
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.48494553565979,
+      "learning_rate": 7.345953157156066e-06,
+      "loss": 0.706,
+      "step": 22370
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.903769612312317,
+      "learning_rate": 7.345720318172479e-06,
+      "loss": 0.7935,
+      "step": 22371
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 3.209545373916626,
+      "learning_rate": 7.345487472666322e-06,
+      "loss": 0.9104,
+      "step": 22372
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.891255497932434,
+      "learning_rate": 7.3452546206382435e-06,
+      "loss": 0.8501,
+      "step": 22373
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.7004051208496094,
+      "learning_rate": 7.345021762088891e-06,
+      "loss": 0.985,
+      "step": 22374
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 3.269895553588867,
+      "learning_rate": 7.344788897018909e-06,
+      "loss": 0.5197,
+      "step": 22375
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.8613126277923584,
+      "learning_rate": 7.344556025428949e-06,
+      "loss": 0.7433,
+      "step": 22376
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.308387041091919,
+      "learning_rate": 7.344323147319657e-06,
+      "loss": 0.6067,
+      "step": 22377
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.2465198040008545,
+      "learning_rate": 7.344090262691681e-06,
+      "loss": 0.8459,
+      "step": 22378
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.6719260215759277,
+      "learning_rate": 7.3438573715456685e-06,
+      "loss": 0.7228,
+      "step": 22379
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.3556158542633057,
+      "learning_rate": 7.343624473882265e-06,
+      "loss": 0.6111,
+      "step": 22380
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.9736933708190918,
+      "learning_rate": 7.343391569702121e-06,
+      "loss": 0.6561,
+      "step": 22381
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.3298635482788086,
+      "learning_rate": 7.343158659005882e-06,
+      "loss": 0.6015,
+      "step": 22382
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.3261382579803467,
+      "learning_rate": 7.342925741794198e-06,
+      "loss": 0.6014,
+      "step": 22383
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.367633581161499,
+      "learning_rate": 7.3426928180677135e-06,
+      "loss": 0.7464,
+      "step": 22384
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.1572821140289307,
+      "learning_rate": 7.342459887827078e-06,
+      "loss": 0.6254,
+      "step": 22385
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.3438806533813477,
+      "learning_rate": 7.3422269510729395e-06,
+      "loss": 0.6477,
+      "step": 22386
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.0806641578674316,
+      "learning_rate": 7.341994007805946e-06,
+      "loss": 0.6067,
+      "step": 22387
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 3.656966209411621,
+      "learning_rate": 7.341761058026744e-06,
+      "loss": 0.7682,
+      "step": 22388
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 5.088823318481445,
+      "learning_rate": 7.341528101735982e-06,
+      "loss": 0.4581,
+      "step": 22389
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.3350367546081543,
+      "learning_rate": 7.341295138934306e-06,
+      "loss": 0.6882,
+      "step": 22390
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.8858002424240112,
+      "learning_rate": 7.341062169622369e-06,
+      "loss": 0.7504,
+      "step": 22391
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 3.3271398544311523,
+      "learning_rate": 7.340829193800812e-06,
+      "loss": 0.6165,
+      "step": 22392
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.971270799636841,
+      "learning_rate": 7.340596211470286e-06,
+      "loss": 0.8174,
+      "step": 22393
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.715437889099121,
+      "learning_rate": 7.340363222631439e-06,
+      "loss": 0.8261,
+      "step": 22394
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 3.084110736846924,
+      "learning_rate": 7.340130227284918e-06,
+      "loss": 0.5378,
+      "step": 22395
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 3.3107151985168457,
+      "learning_rate": 7.339897225431373e-06,
+      "loss": 0.6053,
+      "step": 22396
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 3.241258144378662,
+      "learning_rate": 7.339664217071449e-06,
+      "loss": 0.3642,
+      "step": 22397
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.32936954498291,
+      "learning_rate": 7.339431202205796e-06,
+      "loss": 0.5147,
+      "step": 22398
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.2949845790863037,
+      "learning_rate": 7.339198180835061e-06,
+      "loss": 0.6788,
+      "step": 22399
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.6698113679885864,
+      "learning_rate": 7.338965152959893e-06,
+      "loss": 0.638,
+      "step": 22400
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.5072827339172363,
+      "learning_rate": 7.338732118580938e-06,
+      "loss": 0.688,
+      "step": 22401
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.3966708183288574,
+      "learning_rate": 7.3384990776988454e-06,
+      "loss": 0.584,
+      "step": 22402
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.7215570211410522,
+      "learning_rate": 7.338266030314263e-06,
+      "loss": 0.4838,
+      "step": 22403
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.7632099390029907,
+      "learning_rate": 7.338032976427839e-06,
+      "loss": 0.5391,
+      "step": 22404
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 3.617063522338867,
+      "learning_rate": 7.33779991604022e-06,
+      "loss": 0.7203,
+      "step": 22405
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.472465991973877,
+      "learning_rate": 7.337566849152056e-06,
+      "loss": 0.6558,
+      "step": 22406
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.7895550727844238,
+      "learning_rate": 7.337333775763994e-06,
+      "loss": 0.5382,
+      "step": 22407
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 3.559030532836914,
+      "learning_rate": 7.337100695876682e-06,
+      "loss": 0.5741,
+      "step": 22408
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.5707666873931885,
+      "learning_rate": 7.33686760949077e-06,
+      "loss": 0.8045,
+      "step": 22409
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.9563381671905518,
+      "learning_rate": 7.336634516606904e-06,
+      "loss": 0.6108,
+      "step": 22410
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.6707377433776855,
+      "learning_rate": 7.336401417225731e-06,
+      "loss": 0.8742,
+      "step": 22411
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.9247136116027832,
+      "learning_rate": 7.336168311347903e-06,
+      "loss": 0.6731,
+      "step": 22412
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.3693017959594727,
+      "learning_rate": 7.335935198974065e-06,
+      "loss": 0.505,
+      "step": 22413
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.7884316444396973,
+      "learning_rate": 7.335702080104868e-06,
+      "loss": 0.5389,
+      "step": 22414
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 4.132223606109619,
+      "learning_rate": 7.335468954740957e-06,
+      "loss": 0.673,
+      "step": 22415
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.604191303253174,
+      "learning_rate": 7.33523582288298e-06,
+      "loss": 0.6908,
+      "step": 22416
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.411402463912964,
+      "learning_rate": 7.3350026845315895e-06,
+      "loss": 0.4473,
+      "step": 22417
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.5495851039886475,
+      "learning_rate": 7.334769539687431e-06,
+      "loss": 0.6323,
+      "step": 22418
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 3.068186044692993,
+      "learning_rate": 7.334536388351152e-06,
+      "loss": 0.8789,
+      "step": 22419
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 4.036111354827881,
+      "learning_rate": 7.334303230523403e-06,
+      "loss": 0.9707,
+      "step": 22420
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.2655985355377197,
+      "learning_rate": 7.334070066204831e-06,
+      "loss": 0.6497,
+      "step": 22421
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.7272238731384277,
+      "learning_rate": 7.3338368953960824e-06,
+      "loss": 0.6426,
+      "step": 22422
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.1702940464019775,
+      "learning_rate": 7.33360371809781e-06,
+      "loss": 0.4851,
+      "step": 22423
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.183220624923706,
+      "learning_rate": 7.333370534310659e-06,
+      "loss": 0.5109,
+      "step": 22424
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.8700090646743774,
+      "learning_rate": 7.333137344035279e-06,
+      "loss": 0.7669,
+      "step": 22425
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.3227672576904297,
+      "learning_rate": 7.332904147272317e-06,
+      "loss": 0.5707,
+      "step": 22426
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.104294538497925,
+      "learning_rate": 7.332670944022424e-06,
+      "loss": 0.4184,
+      "step": 22427
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.601097822189331,
+      "learning_rate": 7.3324377342862465e-06,
+      "loss": 0.4762,
+      "step": 22428
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.2367420196533203,
+      "learning_rate": 7.332204518064433e-06,
+      "loss": 0.4493,
+      "step": 22429
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.057227849960327,
+      "learning_rate": 7.3319712953576326e-06,
+      "loss": 0.7702,
+      "step": 22430
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 3.9429514408111572,
+      "learning_rate": 7.331738066166494e-06,
+      "loss": 0.8773,
+      "step": 22431
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 3.3932971954345703,
+      "learning_rate": 7.3315048304916645e-06,
+      "loss": 0.6537,
+      "step": 22432
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.9639455080032349,
+      "learning_rate": 7.331271588333795e-06,
+      "loss": 0.7159,
+      "step": 22433
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.247159957885742,
+      "learning_rate": 7.33103833969353e-06,
+      "loss": 0.6987,
+      "step": 22434
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.052490472793579,
+      "learning_rate": 7.330805084571523e-06,
+      "loss": 0.5605,
+      "step": 22435
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.214425802230835,
+      "learning_rate": 7.330571822968419e-06,
+      "loss": 0.5961,
+      "step": 22436
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.6478116512298584,
+      "learning_rate": 7.330338554884868e-06,
+      "loss": 0.5967,
+      "step": 22437
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.4285221099853516,
+      "learning_rate": 7.330105280321518e-06,
+      "loss": 0.6259,
+      "step": 22438
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.9687784910202026,
+      "learning_rate": 7.329871999279019e-06,
+      "loss": 0.6445,
+      "step": 22439
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.795053720474243,
+      "learning_rate": 7.329638711758018e-06,
+      "loss": 0.6465,
+      "step": 22440
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.214780330657959,
+      "learning_rate": 7.329405417759165e-06,
+      "loss": 0.446,
+      "step": 22441
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.9241433143615723,
+      "learning_rate": 7.3291721172831085e-06,
+      "loss": 0.6943,
+      "step": 22442
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.1490604877471924,
+      "learning_rate": 7.328938810330497e-06,
+      "loss": 0.5661,
+      "step": 22443
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.278216600418091,
+      "learning_rate": 7.328705496901978e-06,
+      "loss": 0.4434,
+      "step": 22444
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.6549272537231445,
+      "learning_rate": 7.328472176998201e-06,
+      "loss": 0.5713,
+      "step": 22445
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 4.435177803039551,
+      "learning_rate": 7.328238850619815e-06,
+      "loss": 0.6501,
+      "step": 22446
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.8666727542877197,
+      "learning_rate": 7.3280055177674715e-06,
+      "loss": 0.5195,
+      "step": 22447
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.4501309394836426,
+      "learning_rate": 7.327772178441814e-06,
+      "loss": 0.6543,
+      "step": 22448
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 3.6691062450408936,
+      "learning_rate": 7.327538832643496e-06,
+      "loss": 0.3767,
+      "step": 22449
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.2323062419891357,
+      "learning_rate": 7.3273054803731634e-06,
+      "loss": 0.8491,
+      "step": 22450
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.8710637092590332,
+      "learning_rate": 7.327072121631467e-06,
+      "loss": 0.3749,
+      "step": 22451
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.7201581001281738,
+      "learning_rate": 7.326838756419055e-06,
+      "loss": 0.4115,
+      "step": 22452
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 3.4518728256225586,
+      "learning_rate": 7.326605384736573e-06,
+      "loss": 0.8895,
+      "step": 22453
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.288445234298706,
+      "learning_rate": 7.3263720065846765e-06,
+      "loss": 0.53,
+      "step": 22454
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.289069652557373,
+      "learning_rate": 7.326138621964009e-06,
+      "loss": 0.4122,
+      "step": 22455
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.7791601419448853,
+      "learning_rate": 7.325905230875223e-06,
+      "loss": 0.7297,
+      "step": 22456
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.4949893951416016,
+      "learning_rate": 7.325671833318964e-06,
+      "loss": 0.4848,
+      "step": 22457
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.082843542098999,
+      "learning_rate": 7.325438429295884e-06,
+      "loss": 0.6926,
+      "step": 22458
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.6286144256591797,
+      "learning_rate": 7.32520501880663e-06,
+      "loss": 0.5407,
+      "step": 22459
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.7027711868286133,
+      "learning_rate": 7.3249716018518525e-06,
+      "loss": 0.8902,
+      "step": 22460
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.8996851444244385,
+      "learning_rate": 7.3247381784322e-06,
+      "loss": 0.5971,
+      "step": 22461
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 3.5036203861236572,
+      "learning_rate": 7.3245047485483215e-06,
+      "loss": 0.6524,
+      "step": 22462
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.4149110317230225,
+      "learning_rate": 7.324271312200866e-06,
+      "loss": 0.939,
+      "step": 22463
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.543034553527832,
+      "learning_rate": 7.324037869390484e-06,
+      "loss": 0.6593,
+      "step": 22464
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 3.035862445831299,
+      "learning_rate": 7.32380442011782e-06,
+      "loss": 0.4905,
+      "step": 22465
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.6162517070770264,
+      "learning_rate": 7.3235709643835295e-06,
+      "loss": 0.8635,
+      "step": 22466
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.893944501876831,
+      "learning_rate": 7.323337502188256e-06,
+      "loss": 0.6822,
+      "step": 22467
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.4562650918960571,
+      "learning_rate": 7.323104033532653e-06,
+      "loss": 0.5809,
+      "step": 22468
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.3198304176330566,
+      "learning_rate": 7.322870558417368e-06,
+      "loss": 0.7663,
+      "step": 22469
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.6324949264526367,
+      "learning_rate": 7.322637076843049e-06,
+      "loss": 0.7432,
+      "step": 22470
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.253488540649414,
+      "learning_rate": 7.322403588810347e-06,
+      "loss": 0.3968,
+      "step": 22471
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.222012996673584,
+      "learning_rate": 7.322170094319912e-06,
+      "loss": 0.5371,
+      "step": 22472
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.1850268840789795,
+      "learning_rate": 7.32193659337239e-06,
+      "loss": 0.8159,
+      "step": 22473
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.3074512481689453,
+      "learning_rate": 7.321703085968432e-06,
+      "loss": 0.9877,
+      "step": 22474
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.4699857234954834,
+      "learning_rate": 7.321469572108689e-06,
+      "loss": 0.8245,
+      "step": 22475
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.2390942573547363,
+      "learning_rate": 7.321236051793809e-06,
+      "loss": 0.7423,
+      "step": 22476
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.9384794235229492,
+      "learning_rate": 7.321002525024439e-06,
+      "loss": 0.44,
+      "step": 22477
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 3.8570005893707275,
+      "learning_rate": 7.320768991801231e-06,
+      "loss": 0.6392,
+      "step": 22478
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.9518051147460938,
+      "learning_rate": 7.320535452124834e-06,
+      "loss": 0.603,
+      "step": 22479
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.700199842453003,
+      "learning_rate": 7.320301905995898e-06,
+      "loss": 0.6906,
+      "step": 22480
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.7508502006530762,
+      "learning_rate": 7.320068353415071e-06,
+      "loss": 0.574,
+      "step": 22481
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.6137475967407227,
+      "learning_rate": 7.3198347943830025e-06,
+      "loss": 0.6424,
+      "step": 22482
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.9719908237457275,
+      "learning_rate": 7.3196012289003435e-06,
+      "loss": 0.7631,
+      "step": 22483
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.6568775177001953,
+      "learning_rate": 7.319367656967742e-06,
+      "loss": 0.5355,
+      "step": 22484
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.1253507137298584,
+      "learning_rate": 7.319134078585848e-06,
+      "loss": 0.7495,
+      "step": 22485
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.6434881687164307,
+      "learning_rate": 7.31890049375531e-06,
+      "loss": 0.7424,
+      "step": 22486
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 3.141826868057251,
+      "learning_rate": 7.31866690247678e-06,
+      "loss": 0.572,
+      "step": 22487
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.457931399345398,
+      "learning_rate": 7.318433304750905e-06,
+      "loss": 0.6341,
+      "step": 22488
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.261225700378418,
+      "learning_rate": 7.3181997005783345e-06,
+      "loss": 0.685,
+      "step": 22489
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.8994024991989136,
+      "learning_rate": 7.317966089959721e-06,
+      "loss": 0.6482,
+      "step": 22490
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.3899080753326416,
+      "learning_rate": 7.317732472895711e-06,
+      "loss": 0.6435,
+      "step": 22491
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 4.318205833435059,
+      "learning_rate": 7.317498849386956e-06,
+      "loss": 0.9664,
+      "step": 22492
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.0607290267944336,
+      "learning_rate": 7.317265219434104e-06,
+      "loss": 0.4217,
+      "step": 22493
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 14.971044540405273,
+      "learning_rate": 7.317031583037806e-06,
+      "loss": 0.7633,
+      "step": 22494
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.892175555229187,
+      "learning_rate": 7.316797940198711e-06,
+      "loss": 0.6881,
+      "step": 22495
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.653388500213623,
+      "learning_rate": 7.316564290917469e-06,
+      "loss": 0.7757,
+      "step": 22496
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.9883990287780762,
+      "learning_rate": 7.316330635194729e-06,
+      "loss": 0.6282,
+      "step": 22497
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.104811191558838,
+      "learning_rate": 7.316096973031141e-06,
+      "loss": 0.7713,
+      "step": 22498
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 3.3856091499328613,
+      "learning_rate": 7.3158633044273565e-06,
+      "loss": 0.6059,
+      "step": 22499
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.5789594650268555,
+      "learning_rate": 7.315629629384022e-06,
+      "loss": 0.8175,
+      "step": 22500
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.5752407312393188,
+      "learning_rate": 7.31539594790179e-06,
+      "loss": 0.7616,
+      "step": 22501
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.6521966457366943,
+      "learning_rate": 7.31516225998131e-06,
+      "loss": 0.6999,
+      "step": 22502
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.393228769302368,
+      "learning_rate": 7.314928565623229e-06,
+      "loss": 0.5778,
+      "step": 22503
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.1088550090789795,
+      "learning_rate": 7.3146948648282005e-06,
+      "loss": 0.5944,
+      "step": 22504
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 3.60052490234375,
+      "learning_rate": 7.3144611575968715e-06,
+      "loss": 0.4157,
+      "step": 22505
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.6648919582366943,
+      "learning_rate": 7.314227443929895e-06,
+      "loss": 0.7394,
+      "step": 22506
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.8828332424163818,
+      "learning_rate": 7.313993723827915e-06,
+      "loss": 0.7612,
+      "step": 22507
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.7950655221939087,
+      "learning_rate": 7.313759997291588e-06,
+      "loss": 0.555,
+      "step": 22508
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 3.0016427040100098,
+      "learning_rate": 7.3135262643215624e-06,
+      "loss": 0.4335,
+      "step": 22509
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 5.2894206047058105,
+      "learning_rate": 7.313292524918485e-06,
+      "loss": 0.689,
+      "step": 22510
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.4086670875549316,
+      "learning_rate": 7.3130587790830075e-06,
+      "loss": 0.6862,
+      "step": 22511
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.978700041770935,
+      "learning_rate": 7.3128250268157804e-06,
+      "loss": 0.5296,
+      "step": 22512
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 3.5122599601745605,
+      "learning_rate": 7.312591268117455e-06,
+      "loss": 0.9115,
+      "step": 22513
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.0906198024749756,
+      "learning_rate": 7.312357502988677e-06,
+      "loss": 0.875,
+      "step": 22514
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.395672559738159,
+      "learning_rate": 7.312123731430101e-06,
+      "loss": 0.3861,
+      "step": 22515
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 3.063053846359253,
+      "learning_rate": 7.3118899534423745e-06,
+      "loss": 0.8883,
+      "step": 22516
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.489093542098999,
+      "learning_rate": 7.3116561690261474e-06,
+      "loss": 0.9453,
+      "step": 22517
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.8499375581741333,
+      "learning_rate": 7.31142237818207e-06,
+      "loss": 0.5183,
+      "step": 22518
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.0943214893341064,
+      "learning_rate": 7.311188580910794e-06,
+      "loss": 0.625,
+      "step": 22519
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.909531354904175,
+      "learning_rate": 7.310954777212968e-06,
+      "loss": 0.6378,
+      "step": 22520
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.0024526119232178,
+      "learning_rate": 7.310720967089243e-06,
+      "loss": 0.7275,
+      "step": 22521
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.371192216873169,
+      "learning_rate": 7.310487150540269e-06,
+      "loss": 0.7231,
+      "step": 22522
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.3796422481536865,
+      "learning_rate": 7.310253327566695e-06,
+      "loss": 0.7651,
+      "step": 22523
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.4144234657287598,
+      "learning_rate": 7.310019498169171e-06,
+      "loss": 0.5212,
+      "step": 22524
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.9826674461364746,
+      "learning_rate": 7.309785662348349e-06,
+      "loss": 0.4604,
+      "step": 22525
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.0665321350097656,
+      "learning_rate": 7.309551820104878e-06,
+      "loss": 0.7888,
+      "step": 22526
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.9479575157165527,
+      "learning_rate": 7.309317971439409e-06,
+      "loss": 0.6069,
+      "step": 22527
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 3.8600733280181885,
+      "learning_rate": 7.3090841163525915e-06,
+      "loss": 0.7174,
+      "step": 22528
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 3.5605854988098145,
+      "learning_rate": 7.308850254845077e-06,
+      "loss": 0.5233,
+      "step": 22529
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.6864479780197144,
+      "learning_rate": 7.3086163869175134e-06,
+      "loss": 0.6654,
+      "step": 22530
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.9295129776000977,
+      "learning_rate": 7.308382512570553e-06,
+      "loss": 0.4623,
+      "step": 22531
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.575725793838501,
+      "learning_rate": 7.308148631804846e-06,
+      "loss": 0.509,
+      "step": 22532
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.3158211708068848,
+      "learning_rate": 7.307914744621043e-06,
+      "loss": 0.7987,
+      "step": 22533
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.6666734218597412,
+      "learning_rate": 7.307680851019791e-06,
+      "loss": 0.6973,
+      "step": 22534
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.118546962738037,
+      "learning_rate": 7.307446951001747e-06,
+      "loss": 0.6225,
+      "step": 22535
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 6.649165153503418,
+      "learning_rate": 7.307213044567554e-06,
+      "loss": 0.5321,
+      "step": 22536
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.674606204032898,
+      "learning_rate": 7.3069791317178675e-06,
+      "loss": 0.2978,
+      "step": 22537
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.6483373641967773,
+      "learning_rate": 7.3067452124533345e-06,
+      "loss": 0.7787,
+      "step": 22538
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 4.030699729919434,
+      "learning_rate": 7.306511286774609e-06,
+      "loss": 0.8493,
+      "step": 22539
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.7306690216064453,
+      "learning_rate": 7.306277354682338e-06,
+      "loss": 0.5376,
+      "step": 22540
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.7228648662567139,
+      "learning_rate": 7.3060434161771755e-06,
+      "loss": 0.435,
+      "step": 22541
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.0767571926116943,
+      "learning_rate": 7.3058094712597695e-06,
+      "loss": 0.7182,
+      "step": 22542
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.522626519203186,
+      "learning_rate": 7.305575519930771e-06,
+      "loss": 0.5627,
+      "step": 22543
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.628631830215454,
+      "learning_rate": 7.3053415621908305e-06,
+      "loss": 0.7021,
+      "step": 22544
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.905877947807312,
+      "learning_rate": 7.305107598040599e-06,
+      "loss": 0.7406,
+      "step": 22545
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 4.801685333251953,
+      "learning_rate": 7.304873627480726e-06,
+      "loss": 0.8259,
+      "step": 22546
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.4093197584152222,
+      "learning_rate": 7.3046396505118645e-06,
+      "loss": 0.7018,
+      "step": 22547
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.640798807144165,
+      "learning_rate": 7.3044056671346615e-06,
+      "loss": 0.6639,
+      "step": 22548
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.9576314687728882,
+      "learning_rate": 7.30417167734977e-06,
+      "loss": 0.5268,
+      "step": 22549
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.8871301412582397,
+      "learning_rate": 7.303937681157842e-06,
+      "loss": 0.7757,
+      "step": 22550
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.714376926422119,
+      "learning_rate": 7.303703678559524e-06,
+      "loss": 0.3895,
+      "step": 22551
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.7284348011016846,
+      "learning_rate": 7.303469669555471e-06,
+      "loss": 0.6226,
+      "step": 22552
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 3.4578843116760254,
+      "learning_rate": 7.30323565414633e-06,
+      "loss": 0.5473,
+      "step": 22553
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.5556952953338623,
+      "learning_rate": 7.303001632332755e-06,
+      "loss": 0.7641,
+      "step": 22554
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.9081671237945557,
+      "learning_rate": 7.302767604115395e-06,
+      "loss": 0.6554,
+      "step": 22555
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.5914586782455444,
+      "learning_rate": 7.3025335694949e-06,
+      "loss": 0.3813,
+      "step": 22556
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.6197147369384766,
+      "learning_rate": 7.302299528471923e-06,
+      "loss": 0.4094,
+      "step": 22557
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 3.2529807090759277,
+      "learning_rate": 7.3020654810471115e-06,
+      "loss": 0.7969,
+      "step": 22558
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.7976322174072266,
+      "learning_rate": 7.3018314272211195e-06,
+      "loss": 0.8865,
+      "step": 22559
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.8306149244308472,
+      "learning_rate": 7.301597366994598e-06,
+      "loss": 0.6408,
+      "step": 22560
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.8751723766326904,
+      "learning_rate": 7.301363300368194e-06,
+      "loss": 0.654,
+      "step": 22561
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.4953670501708984,
+      "learning_rate": 7.301129227342562e-06,
+      "loss": 0.7801,
+      "step": 22562
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.338931083679199,
+      "learning_rate": 7.300895147918352e-06,
+      "loss": 0.7037,
+      "step": 22563
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.481708288192749,
+      "learning_rate": 7.3006610620962134e-06,
+      "loss": 0.6677,
+      "step": 22564
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.239241600036621,
+      "learning_rate": 7.3004269698768e-06,
+      "loss": 0.7016,
+      "step": 22565
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 4.063557147979736,
+      "learning_rate": 7.300192871260758e-06,
+      "loss": 0.7031,
+      "step": 22566
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.828047752380371,
+      "learning_rate": 7.299958766248744e-06,
+      "loss": 0.6764,
+      "step": 22567
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.1417438983917236,
+      "learning_rate": 7.299724654841405e-06,
+      "loss": 0.4625,
+      "step": 22568
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.4627130031585693,
+      "learning_rate": 7.299490537039393e-06,
+      "loss": 0.5864,
+      "step": 22569
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.0531363487243652,
+      "learning_rate": 7.29925641284336e-06,
+      "loss": 0.5945,
+      "step": 22570
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.793710470199585,
+      "learning_rate": 7.299022282253956e-06,
+      "loss": 0.6974,
+      "step": 22571
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.7811788320541382,
+      "learning_rate": 7.298788145271832e-06,
+      "loss": 0.5312,
+      "step": 22572
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.55077064037323,
+      "learning_rate": 7.29855400189764e-06,
+      "loss": 0.7654,
+      "step": 22573
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.508044719696045,
+      "learning_rate": 7.29831985213203e-06,
+      "loss": 0.6538,
+      "step": 22574
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 3.4775168895721436,
+      "learning_rate": 7.298085695975653e-06,
+      "loss": 0.5189,
+      "step": 22575
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.3424220085144043,
+      "learning_rate": 7.297851533429161e-06,
+      "loss": 0.5631,
+      "step": 22576
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.0313427448272705,
+      "learning_rate": 7.2976173644932045e-06,
+      "loss": 0.831,
+      "step": 22577
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.8422999382019043,
+      "learning_rate": 7.297383189168436e-06,
+      "loss": 0.6259,
+      "step": 22578
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.456590175628662,
+      "learning_rate": 7.2971490074555025e-06,
+      "loss": 0.6257,
+      "step": 22579
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.9128479957580566,
+      "learning_rate": 7.296914819355061e-06,
+      "loss": 0.5867,
+      "step": 22580
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.155898094177246,
+      "learning_rate": 7.296680624867759e-06,
+      "loss": 0.6892,
+      "step": 22581
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.9660192728042603,
+      "learning_rate": 7.296446423994248e-06,
+      "loss": 0.6389,
+      "step": 22582
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 3.443722724914551,
+      "learning_rate": 7.296212216735181e-06,
+      "loss": 0.6319,
+      "step": 22583
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 3.180048704147339,
+      "learning_rate": 7.295978003091206e-06,
+      "loss": 0.4599,
+      "step": 22584
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.452989339828491,
+      "learning_rate": 7.295743783062978e-06,
+      "loss": 0.6025,
+      "step": 22585
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 3.07185697555542,
+      "learning_rate": 7.295509556651147e-06,
+      "loss": 0.7297,
+      "step": 22586
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.139529228210449,
+      "learning_rate": 7.295275323856362e-06,
+      "loss": 0.7186,
+      "step": 22587
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.8665354251861572,
+      "learning_rate": 7.295041084679277e-06,
+      "loss": 0.6242,
+      "step": 22588
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.976241111755371,
+      "learning_rate": 7.294806839120543e-06,
+      "loss": 0.6987,
+      "step": 22589
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.25427508354187,
+      "learning_rate": 7.29457258718081e-06,
+      "loss": 0.3544,
+      "step": 22590
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.770425319671631,
+      "learning_rate": 7.29433832886073e-06,
+      "loss": 0.5675,
+      "step": 22591
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.350796699523926,
+      "learning_rate": 7.294104064160956e-06,
+      "loss": 0.4933,
+      "step": 22592
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.373997211456299,
+      "learning_rate": 7.2938697930821365e-06,
+      "loss": 0.5792,
+      "step": 22593
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.7892253398895264,
+      "learning_rate": 7.293635515624925e-06,
+      "loss": 0.9037,
+      "step": 22594
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 3.535196542739868,
+      "learning_rate": 7.293401231789973e-06,
+      "loss": 0.6312,
+      "step": 22595
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.4582762718200684,
+      "learning_rate": 7.293166941577931e-06,
+      "loss": 0.6133,
+      "step": 22596
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.2356908321380615,
+      "learning_rate": 7.29293264498945e-06,
+      "loss": 0.5934,
+      "step": 22597
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 4.834668159484863,
+      "learning_rate": 7.292698342025183e-06,
+      "loss": 0.4792,
+      "step": 22598
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 5.535478591918945,
+      "learning_rate": 7.29246403268578e-06,
+      "loss": 0.449,
+      "step": 22599
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.351490020751953,
+      "learning_rate": 7.292229716971893e-06,
+      "loss": 0.5486,
+      "step": 22600
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.985201120376587,
+      "learning_rate": 7.291995394884175e-06,
+      "loss": 0.6296,
+      "step": 22601
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.1462318897247314,
+      "learning_rate": 7.291761066423276e-06,
+      "loss": 0.7579,
+      "step": 22602
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.643132209777832,
+      "learning_rate": 7.291526731589848e-06,
+      "loss": 0.4637,
+      "step": 22603
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.3232438564300537,
+      "learning_rate": 7.291292390384543e-06,
+      "loss": 0.7961,
+      "step": 22604
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 3.6383419036865234,
+      "learning_rate": 7.291058042808012e-06,
+      "loss": 0.4989,
+      "step": 22605
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 3.4861485958099365,
+      "learning_rate": 7.290823688860907e-06,
+      "loss": 0.4745,
+      "step": 22606
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 5.236969947814941,
+      "learning_rate": 7.290589328543878e-06,
+      "loss": 0.7008,
+      "step": 22607
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.9295339584350586,
+      "learning_rate": 7.2903549618575795e-06,
+      "loss": 0.7223,
+      "step": 22608
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 3.044086456298828,
+      "learning_rate": 7.2901205888026616e-06,
+      "loss": 0.6229,
+      "step": 22609
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.356010675430298,
+      "learning_rate": 7.289886209379776e-06,
+      "loss": 0.7164,
+      "step": 22610
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.348707675933838,
+      "learning_rate": 7.289651823589575e-06,
+      "loss": 0.4459,
+      "step": 22611
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.5562946796417236,
+      "learning_rate": 7.2894174314327104e-06,
+      "loss": 0.5783,
+      "step": 22612
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 3.766282558441162,
+      "learning_rate": 7.289183032909833e-06,
+      "loss": 0.7792,
+      "step": 22613
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.029292106628418,
+      "learning_rate": 7.288948628021595e-06,
+      "loss": 0.7468,
+      "step": 22614
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.9305874109268188,
+      "learning_rate": 7.288714216768649e-06,
+      "loss": 0.4942,
+      "step": 22615
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.041120767593384,
+      "learning_rate": 7.288479799151644e-06,
+      "loss": 0.6928,
+      "step": 22616
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 3.6806676387786865,
+      "learning_rate": 7.288245375171237e-06,
+      "loss": 0.6518,
+      "step": 22617
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.031489133834839,
+      "learning_rate": 7.288010944828076e-06,
+      "loss": 0.5473,
+      "step": 22618
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.150587558746338,
+      "learning_rate": 7.287776508122811e-06,
+      "loss": 0.4793,
+      "step": 22619
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.0834927558898926,
+      "learning_rate": 7.287542065056099e-06,
+      "loss": 0.5957,
+      "step": 22620
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.4701944589614868,
+      "learning_rate": 7.287307615628589e-06,
+      "loss": 0.6835,
+      "step": 22621
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 3.216219186782837,
+      "learning_rate": 7.287073159840934e-06,
+      "loss": 0.7773,
+      "step": 22622
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.6172094345092773,
+      "learning_rate": 7.2868386976937856e-06,
+      "loss": 0.7166,
+      "step": 22623
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.1354055404663086,
+      "learning_rate": 7.286604229187794e-06,
+      "loss": 0.7629,
+      "step": 22624
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.6884883642196655,
+      "learning_rate": 7.286369754323614e-06,
+      "loss": 0.7299,
+      "step": 22625
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.4136626720428467,
+      "learning_rate": 7.2861352731018955e-06,
+      "loss": 0.6724,
+      "step": 22626
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 4.139815807342529,
+      "learning_rate": 7.285900785523291e-06,
+      "loss": 0.7966,
+      "step": 22627
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.983558416366577,
+      "learning_rate": 7.285666291588453e-06,
+      "loss": 0.4968,
+      "step": 22628
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 3.4567902088165283,
+      "learning_rate": 7.285431791298032e-06,
+      "loss": 0.6868,
+      "step": 22629
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.9736541509628296,
+      "learning_rate": 7.285197284652683e-06,
+      "loss": 0.5713,
+      "step": 22630
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 3.726125717163086,
+      "learning_rate": 7.284962771653056e-06,
+      "loss": 0.7307,
+      "step": 22631
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.9797656536102295,
+      "learning_rate": 7.284728252299803e-06,
+      "loss": 0.4758,
+      "step": 22632
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.4534947872161865,
+      "learning_rate": 7.284493726593578e-06,
+      "loss": 0.4431,
+      "step": 22633
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.823293924331665,
+      "learning_rate": 7.284259194535031e-06,
+      "loss": 0.7771,
+      "step": 22634
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.528218388557434,
+      "learning_rate": 7.284024656124814e-06,
+      "loss": 0.5941,
+      "step": 22635
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.1631786823272705,
+      "learning_rate": 7.28379011136358e-06,
+      "loss": 0.6459,
+      "step": 22636
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.6397823095321655,
+      "learning_rate": 7.283555560251983e-06,
+      "loss": 0.5054,
+      "step": 22637
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.645078420639038,
+      "learning_rate": 7.283321002790672e-06,
+      "loss": 0.6401,
+      "step": 22638
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 3.0482165813446045,
+      "learning_rate": 7.283086438980302e-06,
+      "loss": 0.685,
+      "step": 22639
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.538543224334717,
+      "learning_rate": 7.282851868821521e-06,
+      "loss": 0.505,
+      "step": 22640
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.887634038925171,
+      "learning_rate": 7.282617292314986e-06,
+      "loss": 0.9224,
+      "step": 22641
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 3.245713472366333,
+      "learning_rate": 7.282382709461347e-06,
+      "loss": 0.8731,
+      "step": 22642
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 3.5262441635131836,
+      "learning_rate": 7.282148120261257e-06,
+      "loss": 0.8072,
+      "step": 22643
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 3.3949801921844482,
+      "learning_rate": 7.281913524715369e-06,
+      "loss": 0.639,
+      "step": 22644
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.1377081871032715,
+      "learning_rate": 7.281678922824334e-06,
+      "loss": 0.4991,
+      "step": 22645
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 3.0146899223327637,
+      "learning_rate": 7.281444314588804e-06,
+      "loss": 0.7408,
+      "step": 22646
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.4100074768066406,
+      "learning_rate": 7.281209700009432e-06,
+      "loss": 0.7376,
+      "step": 22647
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.9246331453323364,
+      "learning_rate": 7.28097507908687e-06,
+      "loss": 0.5892,
+      "step": 22648
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.8154956102371216,
+      "learning_rate": 7.280740451821772e-06,
+      "loss": 0.7641,
+      "step": 22649
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 3.0683345794677734,
+      "learning_rate": 7.280505818214787e-06,
+      "loss": 0.6769,
+      "step": 22650
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.751311779022217,
+      "learning_rate": 7.280271178266571e-06,
+      "loss": 0.6861,
+      "step": 22651
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.6652438640594482,
+      "learning_rate": 7.280036531977775e-06,
+      "loss": 0.4364,
+      "step": 22652
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.8167788982391357,
+      "learning_rate": 7.279801879349052e-06,
+      "loss": 0.5618,
+      "step": 22653
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.623444080352783,
+      "learning_rate": 7.2795672203810545e-06,
+      "loss": 0.4062,
+      "step": 22654
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.478689193725586,
+      "learning_rate": 7.279332555074433e-06,
+      "loss": 0.5423,
+      "step": 22655
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 3.4701390266418457,
+      "learning_rate": 7.279097883429842e-06,
+      "loss": 0.5312,
+      "step": 22656
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 3.612833023071289,
+      "learning_rate": 7.278863205447935e-06,
+      "loss": 0.6918,
+      "step": 22657
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.4988983869552612,
+      "learning_rate": 7.27862852112936e-06,
+      "loss": 0.5349,
+      "step": 22658
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 4.859195709228516,
+      "learning_rate": 7.278393830474775e-06,
+      "loss": 0.6891,
+      "step": 22659
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.4034664630889893,
+      "learning_rate": 7.278159133484829e-06,
+      "loss": 0.6471,
+      "step": 22660
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.7219831943511963,
+      "learning_rate": 7.2779244301601774e-06,
+      "loss": 0.4941,
+      "step": 22661
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.932689905166626,
+      "learning_rate": 7.27768972050147e-06,
+      "loss": 0.673,
+      "step": 22662
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.15419864654541,
+      "learning_rate": 7.277455004509362e-06,
+      "loss": 0.7326,
+      "step": 22663
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.3654091358184814,
+      "learning_rate": 7.277220282184504e-06,
+      "loss": 0.8256,
+      "step": 22664
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.1458544731140137,
+      "learning_rate": 7.2769855535275494e-06,
+      "loss": 0.6289,
+      "step": 22665
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.017644166946411,
+      "learning_rate": 7.276750818539151e-06,
+      "loss": 0.6032,
+      "step": 22666
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 3.1379613876342773,
+      "learning_rate": 7.276516077219962e-06,
+      "loss": 0.8796,
+      "step": 22667
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.5101566314697266,
+      "learning_rate": 7.276281329570634e-06,
+      "loss": 0.6888,
+      "step": 22668
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 4.302816867828369,
+      "learning_rate": 7.27604657559182e-06,
+      "loss": 0.6327,
+      "step": 22669
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.5891659259796143,
+      "learning_rate": 7.275811815284173e-06,
+      "loss": 0.6532,
+      "step": 22670
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 3.5796709060668945,
+      "learning_rate": 7.275577048648345e-06,
+      "loss": 0.6193,
+      "step": 22671
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.9854625463485718,
+      "learning_rate": 7.275342275684992e-06,
+      "loss": 0.858,
+      "step": 22672
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 3.1019222736358643,
+      "learning_rate": 7.275107496394764e-06,
+      "loss": 0.7317,
+      "step": 22673
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.9183027744293213,
+      "learning_rate": 7.274872710778313e-06,
+      "loss": 0.6359,
+      "step": 22674
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.2845964431762695,
+      "learning_rate": 7.274637918836294e-06,
+      "loss": 0.4143,
+      "step": 22675
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.8927114009857178,
+      "learning_rate": 7.27440312056936e-06,
+      "loss": 0.359,
+      "step": 22676
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.8708173036575317,
+      "learning_rate": 7.274168315978161e-06,
+      "loss": 0.6492,
+      "step": 22677
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.006718635559082,
+      "learning_rate": 7.273933505063353e-06,
+      "loss": 0.7529,
+      "step": 22678
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.1853139400482178,
+      "learning_rate": 7.273698687825588e-06,
+      "loss": 0.4407,
+      "step": 22679
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.56908917427063,
+      "learning_rate": 7.273463864265518e-06,
+      "loss": 0.7641,
+      "step": 22680
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.076514482498169,
+      "learning_rate": 7.2732290343837976e-06,
+      "loss": 0.5657,
+      "step": 22681
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 6.910025596618652,
+      "learning_rate": 7.272994198181078e-06,
+      "loss": 0.6398,
+      "step": 22682
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.163489580154419,
+      "learning_rate": 7.272759355658013e-06,
+      "loss": 0.4836,
+      "step": 22683
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.3904571533203125,
+      "learning_rate": 7.2725245068152565e-06,
+      "loss": 0.5959,
+      "step": 22684
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.002150535583496,
+      "learning_rate": 7.272289651653459e-06,
+      "loss": 0.6679,
+      "step": 22685
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.1195197105407715,
+      "learning_rate": 7.2720547901732774e-06,
+      "loss": 0.631,
+      "step": 22686
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.7358717918395996,
+      "learning_rate": 7.271819922375361e-06,
+      "loss": 0.5621,
+      "step": 22687
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.6749470233917236,
+      "learning_rate": 7.271585048260364e-06,
+      "loss": 0.4731,
+      "step": 22688
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.1629669666290283,
+      "learning_rate": 7.271350167828941e-06,
+      "loss": 0.669,
+      "step": 22689
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.6812365055084229,
+      "learning_rate": 7.271115281081744e-06,
+      "loss": 0.6169,
+      "step": 22690
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.481485366821289,
+      "learning_rate": 7.2708803880194256e-06,
+      "loss": 0.5089,
+      "step": 22691
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.7332944869995117,
+      "learning_rate": 7.270645488642641e-06,
+      "loss": 0.6114,
+      "step": 22692
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.2685725688934326,
+      "learning_rate": 7.27041058295204e-06,
+      "loss": 0.724,
+      "step": 22693
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.8756659030914307,
+      "learning_rate": 7.270175670948278e-06,
+      "loss": 0.6504,
+      "step": 22694
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.929107427597046,
+      "learning_rate": 7.2699407526320095e-06,
+      "loss": 0.3214,
+      "step": 22695
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.5599781274795532,
+      "learning_rate": 7.269705828003885e-06,
+      "loss": 0.4388,
+      "step": 22696
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.264341115951538,
+      "learning_rate": 7.2694708970645585e-06,
+      "loss": 0.6935,
+      "step": 22697
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.714907169342041,
+      "learning_rate": 7.269235959814684e-06,
+      "loss": 0.5759,
+      "step": 22698
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.522714614868164,
+      "learning_rate": 7.269001016254915e-06,
+      "loss": 0.5605,
+      "step": 22699
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.11510968208313,
+      "learning_rate": 7.268766066385902e-06,
+      "loss": 0.6004,
+      "step": 22700
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.1409547328948975,
+      "learning_rate": 7.268531110208302e-06,
+      "loss": 0.6659,
+      "step": 22701
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.4991652965545654,
+      "learning_rate": 7.2682961477227665e-06,
+      "loss": 0.8097,
+      "step": 22702
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.9366223812103271,
+      "learning_rate": 7.26806117892995e-06,
+      "loss": 0.7336,
+      "step": 22703
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.2163636684417725,
+      "learning_rate": 7.267826203830504e-06,
+      "loss": 0.6065,
+      "step": 22704
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.7677340507507324,
+      "learning_rate": 7.267591222425084e-06,
+      "loss": 0.5822,
+      "step": 22705
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.4513649940490723,
+      "learning_rate": 7.267356234714341e-06,
+      "loss": 0.6664,
+      "step": 22706
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.600081205368042,
+      "learning_rate": 7.267121240698931e-06,
+      "loss": 0.6009,
+      "step": 22707
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.6109378337860107,
+      "learning_rate": 7.2668862403795045e-06,
+      "loss": 0.4149,
+      "step": 22708
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.006502389907837,
+      "learning_rate": 7.266651233756719e-06,
+      "loss": 0.7095,
+      "step": 22709
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.390458106994629,
+      "learning_rate": 7.266416220831223e-06,
+      "loss": 0.5229,
+      "step": 22710
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.7429027557373047,
+      "learning_rate": 7.266181201603673e-06,
+      "loss": 0.5627,
+      "step": 22711
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 3.2788615226745605,
+      "learning_rate": 7.265946176074722e-06,
+      "loss": 0.6257,
+      "step": 22712
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.4227683544158936,
+      "learning_rate": 7.265711144245025e-06,
+      "loss": 0.6071,
+      "step": 22713
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.622176170349121,
+      "learning_rate": 7.2654761061152326e-06,
+      "loss": 0.486,
+      "step": 22714
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 3.813441753387451,
+      "learning_rate": 7.265241061685999e-06,
+      "loss": 0.7535,
+      "step": 22715
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.9905357360839844,
+      "learning_rate": 7.26500601095798e-06,
+      "loss": 0.6507,
+      "step": 22716
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 8.921342849731445,
+      "learning_rate": 7.264770953931828e-06,
+      "loss": 0.5985,
+      "step": 22717
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 3.400907516479492,
+      "learning_rate": 7.264535890608197e-06,
+      "loss": 0.8067,
+      "step": 22718
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.931921124458313,
+      "learning_rate": 7.264300820987739e-06,
+      "loss": 0.7392,
+      "step": 22719
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.6208415031433105,
+      "learning_rate": 7.264065745071109e-06,
+      "loss": 0.6552,
+      "step": 22720
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.768441677093506,
+      "learning_rate": 7.263830662858959e-06,
+      "loss": 0.667,
+      "step": 22721
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.340106248855591,
+      "learning_rate": 7.263595574351945e-06,
+      "loss": 0.5919,
+      "step": 22722
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.8115034103393555,
+      "learning_rate": 7.263360479550718e-06,
+      "loss": 0.462,
+      "step": 22723
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.9909329414367676,
+      "learning_rate": 7.263125378455935e-06,
+      "loss": 0.4932,
+      "step": 22724
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 3.1548473834991455,
+      "learning_rate": 7.262890271068249e-06,
+      "loss": 0.554,
+      "step": 22725
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.890037178993225,
+      "learning_rate": 7.262655157388312e-06,
+      "loss": 0.6104,
+      "step": 22726
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.842624306678772,
+      "learning_rate": 7.262420037416778e-06,
+      "loss": 0.6099,
+      "step": 22727
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.7406113147735596,
+      "learning_rate": 7.262184911154303e-06,
+      "loss": 0.8231,
+      "step": 22728
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 3.249222993850708,
+      "learning_rate": 7.261949778601538e-06,
+      "loss": 0.4433,
+      "step": 22729
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.930414080619812,
+      "learning_rate": 7.261714639759137e-06,
+      "loss": 0.8167,
+      "step": 22730
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 3.5728328227996826,
+      "learning_rate": 7.261479494627757e-06,
+      "loss": 0.5808,
+      "step": 22731
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.163400650024414,
+      "learning_rate": 7.261244343208048e-06,
+      "loss": 0.5728,
+      "step": 22732
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 3.179952383041382,
+      "learning_rate": 7.261009185500666e-06,
+      "loss": 0.8395,
+      "step": 22733
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.188532590866089,
+      "learning_rate": 7.260774021506265e-06,
+      "loss": 0.7161,
+      "step": 22734
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.1191773414611816,
+      "learning_rate": 7.260538851225499e-06,
+      "loss": 0.7866,
+      "step": 22735
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.07684326171875,
+      "learning_rate": 7.260303674659018e-06,
+      "loss": 0.6321,
+      "step": 22736
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.724080801010132,
+      "learning_rate": 7.260068491807482e-06,
+      "loss": 0.6991,
+      "step": 22737
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.8848658800125122,
+      "learning_rate": 7.2598333026715415e-06,
+      "loss": 0.741,
+      "step": 22738
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.1233890056610107,
+      "learning_rate": 7.259598107251852e-06,
+      "loss": 0.6452,
+      "step": 22739
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 3.4437255859375,
+      "learning_rate": 7.259362905549065e-06,
+      "loss": 0.6091,
+      "step": 22740
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.0643467903137207,
+      "learning_rate": 7.259127697563838e-06,
+      "loss": 0.7658,
+      "step": 22741
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.6443560123443604,
+      "learning_rate": 7.258892483296821e-06,
+      "loss": 0.4812,
+      "step": 22742
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.6704256534576416,
+      "learning_rate": 7.258657262748672e-06,
+      "loss": 0.8243,
+      "step": 22743
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.7229948043823242,
+      "learning_rate": 7.258422035920043e-06,
+      "loss": 0.5839,
+      "step": 22744
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.257718086242676,
+      "learning_rate": 7.258186802811588e-06,
+      "loss": 0.5661,
+      "step": 22745
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.3568148612976074,
+      "learning_rate": 7.2579515634239615e-06,
+      "loss": 0.5703,
+      "step": 22746
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.0520167350769043,
+      "learning_rate": 7.257716317757817e-06,
+      "loss": 0.6795,
+      "step": 22747
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.3289151191711426,
+      "learning_rate": 7.2574810658138096e-06,
+      "loss": 0.849,
+      "step": 22748
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 5.502601146697998,
+      "learning_rate": 7.257245807592595e-06,
+      "loss": 0.8399,
+      "step": 22749
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.168489456176758,
+      "learning_rate": 7.257010543094823e-06,
+      "loss": 0.6324,
+      "step": 22750
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.9126211404800415,
+      "learning_rate": 7.256775272321149e-06,
+      "loss": 0.7376,
+      "step": 22751
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.939697742462158,
+      "learning_rate": 7.256539995272231e-06,
+      "loss": 0.6413,
+      "step": 22752
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.6138558387756348,
+      "learning_rate": 7.256304711948719e-06,
+      "loss": 0.6754,
+      "step": 22753
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.8972177505493164,
+      "learning_rate": 7.25606942235127e-06,
+      "loss": 0.5827,
+      "step": 22754
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.0019643306732178,
+      "learning_rate": 7.255834126480537e-06,
+      "loss": 0.5636,
+      "step": 22755
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.75736665725708,
+      "learning_rate": 7.255598824337173e-06,
+      "loss": 0.6863,
+      "step": 22756
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.4579861164093018,
+      "learning_rate": 7.255363515921835e-06,
+      "loss": 0.7687,
+      "step": 22757
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 3.6300106048583984,
+      "learning_rate": 7.255128201235176e-06,
+      "loss": 0.9054,
+      "step": 22758
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.4072318077087402,
+      "learning_rate": 7.25489288027785e-06,
+      "loss": 0.7523,
+      "step": 22759
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.7340104579925537,
+      "learning_rate": 7.254657553050512e-06,
+      "loss": 0.8201,
+      "step": 22760
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.2638955116271973,
+      "learning_rate": 7.254422219553814e-06,
+      "loss": 0.6727,
+      "step": 22761
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.495685338973999,
+      "learning_rate": 7.254186879788413e-06,
+      "loss": 0.5882,
+      "step": 22762
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 3.7185075283050537,
+      "learning_rate": 7.253951533754964e-06,
+      "loss": 0.74,
+      "step": 22763
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.6109579801559448,
+      "learning_rate": 7.253716181454119e-06,
+      "loss": 0.4768,
+      "step": 22764
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.5963306427001953,
+      "learning_rate": 7.253480822886534e-06,
+      "loss": 0.7268,
+      "step": 22765
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 3.811107635498047,
+      "learning_rate": 7.253245458052863e-06,
+      "loss": 0.8146,
+      "step": 22766
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.7425777912139893,
+      "learning_rate": 7.253010086953761e-06,
+      "loss": 0.3871,
+      "step": 22767
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.1843485832214355,
+      "learning_rate": 7.252774709589883e-06,
+      "loss": 0.6551,
+      "step": 22768
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.863886594772339,
+      "learning_rate": 7.25253932596188e-06,
+      "loss": 0.8262,
+      "step": 22769
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.2811355590820312,
+      "learning_rate": 7.25230393607041e-06,
+      "loss": 0.3631,
+      "step": 22770
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 4.191816806793213,
+      "learning_rate": 7.252068539916126e-06,
+      "loss": 1.0251,
+      "step": 22771
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.3961305618286133,
+      "learning_rate": 7.251833137499683e-06,
+      "loss": 0.6215,
+      "step": 22772
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.7008581161499023,
+      "learning_rate": 7.251597728821735e-06,
+      "loss": 0.5764,
+      "step": 22773
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.8706305027008057,
+      "learning_rate": 7.251362313882939e-06,
+      "loss": 0.8935,
+      "step": 22774
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.017179012298584,
+      "learning_rate": 7.251126892683947e-06,
+      "loss": 0.5867,
+      "step": 22775
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 4.059394836425781,
+      "learning_rate": 7.250891465225415e-06,
+      "loss": 0.7125,
+      "step": 22776
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.971449851989746,
+      "learning_rate": 7.250656031507996e-06,
+      "loss": 0.6579,
+      "step": 22777
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.1605541706085205,
+      "learning_rate": 7.2504205915323466e-06,
+      "loss": 0.5306,
+      "step": 22778
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.058950424194336,
+      "learning_rate": 7.25018514529912e-06,
+      "loss": 0.653,
+      "step": 22779
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 3.133185386657715,
+      "learning_rate": 7.249949692808972e-06,
+      "loss": 0.7014,
+      "step": 22780
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.9898834228515625,
+      "learning_rate": 7.249714234062556e-06,
+      "loss": 0.5905,
+      "step": 22781
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.190819025039673,
+      "learning_rate": 7.2494787690605275e-06,
+      "loss": 0.5253,
+      "step": 22782
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.696316719055176,
+      "learning_rate": 7.249243297803541e-06,
+      "loss": 0.5798,
+      "step": 22783
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.9620254039764404,
+      "learning_rate": 7.249007820292252e-06,
+      "loss": 0.6969,
+      "step": 22784
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.777536153793335,
+      "learning_rate": 7.248772336527314e-06,
+      "loss": 1.0014,
+      "step": 22785
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 3.3955719470977783,
+      "learning_rate": 7.248536846509384e-06,
+      "loss": 0.7322,
+      "step": 22786
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.8384976387023926,
+      "learning_rate": 7.2483013502391154e-06,
+      "loss": 0.5795,
+      "step": 22787
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.9734899997711182,
+      "learning_rate": 7.248065847717162e-06,
+      "loss": 0.699,
+      "step": 22788
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.6382964849472046,
+      "learning_rate": 7.247830338944179e-06,
+      "loss": 0.4812,
+      "step": 22789
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 3.659113645553589,
+      "learning_rate": 7.247594823920824e-06,
+      "loss": 0.6817,
+      "step": 22790
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.7683244943618774,
+      "learning_rate": 7.247359302647748e-06,
+      "loss": 0.4074,
+      "step": 22791
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.28436017036438,
+      "learning_rate": 7.247123775125607e-06,
+      "loss": 0.7537,
+      "step": 22792
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.029923915863037,
+      "learning_rate": 7.246888241355058e-06,
+      "loss": 0.6028,
+      "step": 22793
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.3733601570129395,
+      "learning_rate": 7.246652701336754e-06,
+      "loss": 0.5104,
+      "step": 22794
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 3.7854721546173096,
+      "learning_rate": 7.246417155071351e-06,
+      "loss": 0.802,
+      "step": 22795
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.8181523084640503,
+      "learning_rate": 7.2461816025595035e-06,
+      "loss": 0.5456,
+      "step": 22796
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.809693455696106,
+      "learning_rate": 7.2459460438018655e-06,
+      "loss": 0.4445,
+      "step": 22797
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.109555721282959,
+      "learning_rate": 7.245710478799093e-06,
+      "loss": 0.5588,
+      "step": 22798
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.367002487182617,
+      "learning_rate": 7.245474907551841e-06,
+      "loss": 0.6642,
+      "step": 22799
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.810168743133545,
+      "learning_rate": 7.245239330060765e-06,
+      "loss": 0.7177,
+      "step": 22800
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.9341015815734863,
+      "learning_rate": 7.24500374632652e-06,
+      "loss": 0.6132,
+      "step": 22801
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.1592068672180176,
+      "learning_rate": 7.24476815634976e-06,
+      "loss": 0.3679,
+      "step": 22802
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.4759488105773926,
+      "learning_rate": 7.24453256013114e-06,
+      "loss": 0.799,
+      "step": 22803
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 4.506481170654297,
+      "learning_rate": 7.244296957671315e-06,
+      "loss": 0.4876,
+      "step": 22804
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.3523340225219727,
+      "learning_rate": 7.244061348970942e-06,
+      "loss": 0.5892,
+      "step": 22805
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.3770649433135986,
+      "learning_rate": 7.2438257340306765e-06,
+      "loss": 0.483,
+      "step": 22806
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 3.1671817302703857,
+      "learning_rate": 7.24359011285117e-06,
+      "loss": 0.5391,
+      "step": 22807
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.426389694213867,
+      "learning_rate": 7.24335448543308e-06,
+      "loss": 0.8938,
+      "step": 22808
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.1656651496887207,
+      "learning_rate": 7.243118851777064e-06,
+      "loss": 0.4903,
+      "step": 22809
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.816540241241455,
+      "learning_rate": 7.242883211883772e-06,
+      "loss": 0.8299,
+      "step": 22810
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.3026111125946045,
+      "learning_rate": 7.242647565753864e-06,
+      "loss": 0.8783,
+      "step": 22811
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.2528436183929443,
+      "learning_rate": 7.242411913387991e-06,
+      "loss": 0.4452,
+      "step": 22812
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.5716114044189453,
+      "learning_rate": 7.242176254786813e-06,
+      "loss": 0.897,
+      "step": 22813
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.2570347785949707,
+      "learning_rate": 7.241940589950981e-06,
+      "loss": 0.7029,
+      "step": 22814
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.6440730094909668,
+      "learning_rate": 7.241704918881151e-06,
+      "loss": 0.5604,
+      "step": 22815
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.3179876804351807,
+      "learning_rate": 7.241469241577983e-06,
+      "loss": 0.5735,
+      "step": 22816
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.7307453155517578,
+      "learning_rate": 7.2412335580421265e-06,
+      "loss": 0.6812,
+      "step": 22817
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 3.413322925567627,
+      "learning_rate": 7.240997868274239e-06,
+      "loss": 0.8219,
+      "step": 22818
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 3.504966974258423,
+      "learning_rate": 7.240762172274976e-06,
+      "loss": 0.7688,
+      "step": 22819
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.9983198642730713,
+      "learning_rate": 7.240526470044993e-06,
+      "loss": 0.7283,
+      "step": 22820
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.2126973867416382,
+      "learning_rate": 7.240290761584946e-06,
+      "loss": 0.3064,
+      "step": 22821
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.7919929027557373,
+      "learning_rate": 7.240055046895489e-06,
+      "loss": 0.5966,
+      "step": 22822
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.998719573020935,
+      "learning_rate": 7.239819325977276e-06,
+      "loss": 0.6058,
+      "step": 22823
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.925626039505005,
+      "learning_rate": 7.239583598830967e-06,
+      "loss": 0.5868,
+      "step": 22824
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 3.9883859157562256,
+      "learning_rate": 7.239347865457215e-06,
+      "loss": 0.6332,
+      "step": 22825
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 4.867415904998779,
+      "learning_rate": 7.239112125856674e-06,
+      "loss": 0.5995,
+      "step": 22826
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.041402578353882,
+      "learning_rate": 7.2388763800300025e-06,
+      "loss": 0.5548,
+      "step": 22827
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.394104242324829,
+      "learning_rate": 7.238640627977854e-06,
+      "loss": 0.9453,
+      "step": 22828
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.010981559753418,
+      "learning_rate": 7.238404869700885e-06,
+      "loss": 0.6498,
+      "step": 22829
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.7955296039581299,
+      "learning_rate": 7.2381691051997495e-06,
+      "loss": 0.6152,
+      "step": 22830
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.662816047668457,
+      "learning_rate": 7.237933334475105e-06,
+      "loss": 0.5918,
+      "step": 22831
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.1954755783081055,
+      "learning_rate": 7.2376975575276055e-06,
+      "loss": 0.5796,
+      "step": 22832
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.2992188930511475,
+      "learning_rate": 7.237461774357906e-06,
+      "loss": 0.7378,
+      "step": 22833
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.662330985069275,
+      "learning_rate": 7.237225984966666e-06,
+      "loss": 0.7392,
+      "step": 22834
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.733826160430908,
+      "learning_rate": 7.236990189354538e-06,
+      "loss": 0.7857,
+      "step": 22835
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.3699727058410645,
+      "learning_rate": 7.236754387522178e-06,
+      "loss": 0.7262,
+      "step": 22836
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.9817252159118652,
+      "learning_rate": 7.236518579470242e-06,
+      "loss": 0.5845,
+      "step": 22837
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.73538076877594,
+      "learning_rate": 7.236282765199386e-06,
+      "loss": 0.5544,
+      "step": 22838
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.0175857543945312,
+      "learning_rate": 7.236046944710266e-06,
+      "loss": 0.4674,
+      "step": 22839
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.078242063522339,
+      "learning_rate": 7.235811118003535e-06,
+      "loss": 0.7479,
+      "step": 22840
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.9059581756591797,
+      "learning_rate": 7.235575285079852e-06,
+      "loss": 0.7152,
+      "step": 22841
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.480181932449341,
+      "learning_rate": 7.235339445939872e-06,
+      "loss": 0.66,
+      "step": 22842
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 4.24493932723999,
+      "learning_rate": 7.235103600584249e-06,
+      "loss": 0.6724,
+      "step": 22843
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 3.023698091506958,
+      "learning_rate": 7.234867749013641e-06,
+      "loss": 0.4435,
+      "step": 22844
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 4.479062557220459,
+      "learning_rate": 7.234631891228703e-06,
+      "loss": 0.5089,
+      "step": 22845
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 3.9808268547058105,
+      "learning_rate": 7.234396027230092e-06,
+      "loss": 0.9449,
+      "step": 22846
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 3.2969539165496826,
+      "learning_rate": 7.234160157018461e-06,
+      "loss": 0.6313,
+      "step": 22847
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 3.973501682281494,
+      "learning_rate": 7.233924280594468e-06,
+      "loss": 1.1712,
+      "step": 22848
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.228970766067505,
+      "learning_rate": 7.233688397958769e-06,
+      "loss": 0.6563,
+      "step": 22849
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.368365526199341,
+      "learning_rate": 7.233452509112019e-06,
+      "loss": 0.8563,
+      "step": 22850
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.5174596309661865,
+      "learning_rate": 7.233216614054874e-06,
+      "loss": 0.5272,
+      "step": 22851
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 4.041828632354736,
+      "learning_rate": 7.23298071278799e-06,
+      "loss": 1.0172,
+      "step": 22852
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.1582231521606445,
+      "learning_rate": 7.232744805312023e-06,
+      "loss": 0.9103,
+      "step": 22853
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.884889841079712,
+      "learning_rate": 7.23250889162763e-06,
+      "loss": 0.4575,
+      "step": 22854
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.8003891706466675,
+      "learning_rate": 7.232272971735466e-06,
+      "loss": 0.6721,
+      "step": 22855
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.7805590629577637,
+      "learning_rate": 7.232037045636187e-06,
+      "loss": 0.6251,
+      "step": 22856
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.8702980279922485,
+      "learning_rate": 7.231801113330448e-06,
+      "loss": 0.662,
+      "step": 22857
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.39701247215271,
+      "learning_rate": 7.231565174818908e-06,
+      "loss": 0.6737,
+      "step": 22858
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.952902913093567,
+      "learning_rate": 7.23132923010222e-06,
+      "loss": 0.6271,
+      "step": 22859
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 5.92567253112793,
+      "learning_rate": 7.2310932791810406e-06,
+      "loss": 0.4996,
+      "step": 22860
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.8867290019989014,
+      "learning_rate": 7.230857322056027e-06,
+      "loss": 0.6774,
+      "step": 22861
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.651416301727295,
+      "learning_rate": 7.230621358727835e-06,
+      "loss": 0.6859,
+      "step": 22862
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.0372154712677,
+      "learning_rate": 7.230385389197121e-06,
+      "loss": 0.4768,
+      "step": 22863
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 3.263126850128174,
+      "learning_rate": 7.230149413464541e-06,
+      "loss": 0.7799,
+      "step": 22864
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 3.27435302734375,
+      "learning_rate": 7.229913431530749e-06,
+      "loss": 0.5476,
+      "step": 22865
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.4888546466827393,
+      "learning_rate": 7.2296774433964055e-06,
+      "loss": 0.6811,
+      "step": 22866
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 4.65047550201416,
+      "learning_rate": 7.229441449062164e-06,
+      "loss": 0.7222,
+      "step": 22867
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.713830590248108,
+      "learning_rate": 7.22920544852868e-06,
+      "loss": 0.7233,
+      "step": 22868
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.7907692193984985,
+      "learning_rate": 7.228969441796612e-06,
+      "loss": 0.6606,
+      "step": 22869
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 4.198525428771973,
+      "learning_rate": 7.228733428866614e-06,
+      "loss": 0.6143,
+      "step": 22870
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.8094563484191895,
+      "learning_rate": 7.2284974097393434e-06,
+      "loss": 0.7437,
+      "step": 22871
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 3.3485729694366455,
+      "learning_rate": 7.228261384415457e-06,
+      "loss": 0.4092,
+      "step": 22872
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.5142719745635986,
+      "learning_rate": 7.228025352895609e-06,
+      "loss": 0.7422,
+      "step": 22873
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.8529117107391357,
+      "learning_rate": 7.227789315180458e-06,
+      "loss": 0.4766,
+      "step": 22874
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.1724367141723633,
+      "learning_rate": 7.227553271270659e-06,
+      "loss": 0.5442,
+      "step": 22875
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 3.305868625640869,
+      "learning_rate": 7.22731722116687e-06,
+      "loss": 0.7746,
+      "step": 22876
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.1703288555145264,
+      "learning_rate": 7.227081164869745e-06,
+      "loss": 0.7493,
+      "step": 22877
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.2422678470611572,
+      "learning_rate": 7.226845102379941e-06,
+      "loss": 0.7693,
+      "step": 22878
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.1597907543182373,
+      "learning_rate": 7.2266090336981165e-06,
+      "loss": 0.6994,
+      "step": 22879
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.309458017349243,
+      "learning_rate": 7.226372958824926e-06,
+      "loss": 0.6436,
+      "step": 22880
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.1490156650543213,
+      "learning_rate": 7.226136877761026e-06,
+      "loss": 0.5248,
+      "step": 22881
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 3.6690332889556885,
+      "learning_rate": 7.225900790507073e-06,
+      "loss": 0.5626,
+      "step": 22882
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.888940453529358,
+      "learning_rate": 7.225664697063724e-06,
+      "loss": 0.8505,
+      "step": 22883
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.7686896324157715,
+      "learning_rate": 7.225428597431635e-06,
+      "loss": 0.7677,
+      "step": 22884
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 3.9899723529815674,
+      "learning_rate": 7.225192491611464e-06,
+      "loss": 0.663,
+      "step": 22885
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.6425626277923584,
+      "learning_rate": 7.224956379603865e-06,
+      "loss": 0.8682,
+      "step": 22886
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 3.573251724243164,
+      "learning_rate": 7.224720261409496e-06,
+      "loss": 0.6324,
+      "step": 22887
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 3.1698641777038574,
+      "learning_rate": 7.224484137029013e-06,
+      "loss": 0.6027,
+      "step": 22888
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 4.600076675415039,
+      "learning_rate": 7.224248006463074e-06,
+      "loss": 0.5007,
+      "step": 22889
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.6427650451660156,
+      "learning_rate": 7.224011869712334e-06,
+      "loss": 0.7101,
+      "step": 22890
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.8544907569885254,
+      "learning_rate": 7.22377572677745e-06,
+      "loss": 0.5641,
+      "step": 22891
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.8916337490081787,
+      "learning_rate": 7.223539577659079e-06,
+      "loss": 0.5186,
+      "step": 22892
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.681907892227173,
+      "learning_rate": 7.223303422357877e-06,
+      "loss": 0.552,
+      "step": 22893
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.4721252918243408,
+      "learning_rate": 7.223067260874501e-06,
+      "loss": 0.663,
+      "step": 22894
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.179955244064331,
+      "learning_rate": 7.222831093209608e-06,
+      "loss": 0.7211,
+      "step": 22895
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.2737293243408203,
+      "learning_rate": 7.222594919363855e-06,
+      "loss": 0.7273,
+      "step": 22896
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 3.4058470726013184,
+      "learning_rate": 7.222358739337897e-06,
+      "loss": 0.4026,
+      "step": 22897
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.2684366703033447,
+      "learning_rate": 7.222122553132393e-06,
+      "loss": 0.3837,
+      "step": 22898
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.4921343326568604,
+      "learning_rate": 7.221886360747997e-06,
+      "loss": 0.7161,
+      "step": 22899
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.339838743209839,
+      "learning_rate": 7.22165016218537e-06,
+      "loss": 0.5106,
+      "step": 22900
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.5153307914733887,
+      "learning_rate": 7.221413957445164e-06,
+      "loss": 0.4895,
+      "step": 22901
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.1304731369018555,
+      "learning_rate": 7.221177746528039e-06,
+      "loss": 0.6005,
+      "step": 22902
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.3916704654693604,
+      "learning_rate": 7.22094152943465e-06,
+      "loss": 0.6442,
+      "step": 22903
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.3813506364822388,
+      "learning_rate": 7.220705306165653e-06,
+      "loss": 0.3355,
+      "step": 22904
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.8864680528640747,
+      "learning_rate": 7.220469076721709e-06,
+      "loss": 0.6005,
+      "step": 22905
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.7221248149871826,
+      "learning_rate": 7.220232841103472e-06,
+      "loss": 0.7889,
+      "step": 22906
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.5783474445343018,
+      "learning_rate": 7.219996599311598e-06,
+      "loss": 0.6489,
+      "step": 22907
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 3.842149496078491,
+      "learning_rate": 7.219760351346746e-06,
+      "loss": 0.8377,
+      "step": 22908
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.401082992553711,
+      "learning_rate": 7.219524097209572e-06,
+      "loss": 0.6323,
+      "step": 22909
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.7870452404022217,
+      "learning_rate": 7.219287836900732e-06,
+      "loss": 0.5514,
+      "step": 22910
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.086604356765747,
+      "learning_rate": 7.219051570420885e-06,
+      "loss": 0.748,
+      "step": 22911
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.5424797534942627,
+      "learning_rate": 7.2188152977706875e-06,
+      "loss": 0.8262,
+      "step": 22912
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.2964513301849365,
+      "learning_rate": 7.218579018950794e-06,
+      "loss": 0.5941,
+      "step": 22913
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.9475356340408325,
+      "learning_rate": 7.218342733961864e-06,
+      "loss": 0.7418,
+      "step": 22914
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 3.4568088054656982,
+      "learning_rate": 7.218106442804553e-06,
+      "loss": 0.6543,
+      "step": 22915
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.824627161026001,
+      "learning_rate": 7.217870145479521e-06,
+      "loss": 0.689,
+      "step": 22916
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.6839354038238525,
+      "learning_rate": 7.2176338419874214e-06,
+      "loss": 0.823,
+      "step": 22917
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 3.2969183921813965,
+      "learning_rate": 7.2173975323289136e-06,
+      "loss": 0.3987,
+      "step": 22918
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.967511773109436,
+      "learning_rate": 7.217161216504655e-06,
+      "loss": 0.4482,
+      "step": 22919
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.9426583051681519,
+      "learning_rate": 7.2169248945152995e-06,
+      "loss": 0.7288,
+      "step": 22920
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.561419725418091,
+      "learning_rate": 7.216688566361506e-06,
+      "loss": 0.5274,
+      "step": 22921
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 5.964693546295166,
+      "learning_rate": 7.216452232043934e-06,
+      "loss": 0.732,
+      "step": 22922
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.253309726715088,
+      "learning_rate": 7.216215891563238e-06,
+      "loss": 0.7756,
+      "step": 22923
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.375420093536377,
+      "learning_rate": 7.2159795449200755e-06,
+      "loss": 0.7746,
+      "step": 22924
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.375297784805298,
+      "learning_rate": 7.2157431921151035e-06,
+      "loss": 0.5472,
+      "step": 22925
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 4.039697647094727,
+      "learning_rate": 7.215506833148981e-06,
+      "loss": 0.4427,
+      "step": 22926
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.3938419818878174,
+      "learning_rate": 7.215270468022362e-06,
+      "loss": 0.69,
+      "step": 22927
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.9477354288101196,
+      "learning_rate": 7.215034096735907e-06,
+      "loss": 0.6461,
+      "step": 22928
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.398268222808838,
+      "learning_rate": 7.214797719290272e-06,
+      "loss": 0.5606,
+      "step": 22929
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 5.671853542327881,
+      "learning_rate": 7.214561335686114e-06,
+      "loss": 0.7042,
+      "step": 22930
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 3.1818532943725586,
+      "learning_rate": 7.214324945924092e-06,
+      "loss": 0.7016,
+      "step": 22931
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.1791220903396606,
+      "learning_rate": 7.21408855000486e-06,
+      "loss": 0.534,
+      "step": 22932
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 3.647385835647583,
+      "learning_rate": 7.213852147929077e-06,
+      "loss": 0.5787,
+      "step": 22933
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.1250486373901367,
+      "learning_rate": 7.213615739697401e-06,
+      "loss": 0.7286,
+      "step": 22934
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.2522568702697754,
+      "learning_rate": 7.213379325310489e-06,
+      "loss": 0.6803,
+      "step": 22935
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.5236659049987793,
+      "learning_rate": 7.2131429047689986e-06,
+      "loss": 0.6639,
+      "step": 22936
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.107475519180298,
+      "learning_rate": 7.212906478073586e-06,
+      "loss": 0.8345,
+      "step": 22937
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 3.0682742595672607,
+      "learning_rate": 7.21267004522491e-06,
+      "loss": 0.6598,
+      "step": 22938
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.7214432954788208,
+      "learning_rate": 7.212433606223628e-06,
+      "loss": 0.7539,
+      "step": 22939
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.7186676263809204,
+      "learning_rate": 7.2121971610703956e-06,
+      "loss": 0.6111,
+      "step": 22940
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.7445595264434814,
+      "learning_rate": 7.211960709765873e-06,
+      "loss": 0.6981,
+      "step": 22941
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.9284732341766357,
+      "learning_rate": 7.211724252310716e-06,
+      "loss": 0.754,
+      "step": 22942
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.9013657569885254,
+      "learning_rate": 7.211487788705582e-06,
+      "loss": 0.6524,
+      "step": 22943
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.394364356994629,
+      "learning_rate": 7.211251318951129e-06,
+      "loss": 0.5878,
+      "step": 22944
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.6965458393096924,
+      "learning_rate": 7.211014843048014e-06,
+      "loss": 0.792,
+      "step": 22945
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.4287461042404175,
+      "learning_rate": 7.210778360996895e-06,
+      "loss": 0.5227,
+      "step": 22946
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.662815809249878,
+      "learning_rate": 7.21054187279843e-06,
+      "loss": 0.6844,
+      "step": 22947
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.91978120803833,
+      "learning_rate": 7.210305378453276e-06,
+      "loss": 0.7676,
+      "step": 22948
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 3.3590800762176514,
+      "learning_rate": 7.210068877962091e-06,
+      "loss": 0.7993,
+      "step": 22949
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.9755626916885376,
+      "learning_rate": 7.209832371325532e-06,
+      "loss": 0.6016,
+      "step": 22950
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.144613027572632,
+      "learning_rate": 7.209595858544258e-06,
+      "loss": 0.7112,
+      "step": 22951
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 3.512033462524414,
+      "learning_rate": 7.209359339618924e-06,
+      "loss": 0.5997,
+      "step": 22952
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.317380666732788,
+      "learning_rate": 7.209122814550191e-06,
+      "loss": 0.5333,
+      "step": 22953
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.9897617101669312,
+      "learning_rate": 7.208886283338714e-06,
+      "loss": 0.7506,
+      "step": 22954
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 3.992898941040039,
+      "learning_rate": 7.208649745985151e-06,
+      "loss": 0.6216,
+      "step": 22955
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.269183874130249,
+      "learning_rate": 7.208413202490161e-06,
+      "loss": 0.7217,
+      "step": 22956
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.050224781036377,
+      "learning_rate": 7.208176652854402e-06,
+      "loss": 0.6274,
+      "step": 22957
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.5735771656036377,
+      "learning_rate": 7.207940097078531e-06,
+      "loss": 0.6464,
+      "step": 22958
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.548177719116211,
+      "learning_rate": 7.207703535163205e-06,
+      "loss": 0.4112,
+      "step": 22959
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.4665467739105225,
+      "learning_rate": 7.207466967109083e-06,
+      "loss": 0.5147,
+      "step": 22960
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.7342586517333984,
+      "learning_rate": 7.207230392916822e-06,
+      "loss": 0.407,
+      "step": 22961
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.8393912315368652,
+      "learning_rate": 7.20699381258708e-06,
+      "loss": 0.582,
+      "step": 22962
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.585826873779297,
+      "learning_rate": 7.206757226120516e-06,
+      "loss": 0.724,
+      "step": 22963
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.260413646697998,
+      "learning_rate": 7.2065206335177855e-06,
+      "loss": 0.7659,
+      "step": 22964
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 3.906127691268921,
+      "learning_rate": 7.206284034779546e-06,
+      "loss": 0.5102,
+      "step": 22965
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.0389654636383057,
+      "learning_rate": 7.20604742990646e-06,
+      "loss": 0.7113,
+      "step": 22966
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.3864736557006836,
+      "learning_rate": 7.205810818899181e-06,
+      "loss": 0.6935,
+      "step": 22967
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.5728650093078613,
+      "learning_rate": 7.20557420175837e-06,
+      "loss": 0.6119,
+      "step": 22968
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.951695680618286,
+      "learning_rate": 7.205337578484682e-06,
+      "loss": 0.7672,
+      "step": 22969
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.6307101249694824,
+      "learning_rate": 7.205100949078777e-06,
+      "loss": 0.4434,
+      "step": 22970
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.1362130641937256,
+      "learning_rate": 7.204864313541312e-06,
+      "loss": 0.5217,
+      "step": 22971
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.2362587451934814,
+      "learning_rate": 7.204627671872946e-06,
+      "loss": 0.7077,
+      "step": 22972
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.895888566970825,
+      "learning_rate": 7.204391024074335e-06,
+      "loss": 0.5082,
+      "step": 22973
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.1893913745880127,
+      "learning_rate": 7.204154370146139e-06,
+      "loss": 0.5491,
+      "step": 22974
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 5.109662055969238,
+      "learning_rate": 7.2039177100890145e-06,
+      "loss": 0.7187,
+      "step": 22975
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 3.502851963043213,
+      "learning_rate": 7.203681043903622e-06,
+      "loss": 0.7905,
+      "step": 22976
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 3.319554567337036,
+      "learning_rate": 7.203444371590618e-06,
+      "loss": 0.7062,
+      "step": 22977
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.1932473182678223,
+      "learning_rate": 7.20320769315066e-06,
+      "loss": 0.5734,
+      "step": 22978
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 3.5682690143585205,
+      "learning_rate": 7.202971008584406e-06,
+      "loss": 0.9954,
+      "step": 22979
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.901371479034424,
+      "learning_rate": 7.202734317892516e-06,
+      "loss": 0.7452,
+      "step": 22980
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.0132534503936768,
+      "learning_rate": 7.202497621075647e-06,
+      "loss": 0.715,
+      "step": 22981
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.869433879852295,
+      "learning_rate": 7.2022609181344575e-06,
+      "loss": 0.7334,
+      "step": 22982
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.1777658462524414,
+      "learning_rate": 7.202024209069604e-06,
+      "loss": 0.7006,
+      "step": 22983
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 3.501204490661621,
+      "learning_rate": 7.2017874938817465e-06,
+      "loss": 0.5716,
+      "step": 22984
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.6896698474884033,
+      "learning_rate": 7.201550772571543e-06,
+      "loss": 0.5737,
+      "step": 22985
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.362412691116333,
+      "learning_rate": 7.201314045139651e-06,
+      "loss": 0.8272,
+      "step": 22986
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.1570236682891846,
+      "learning_rate": 7.20107731158673e-06,
+      "loss": 0.63,
+      "step": 22987
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.578016519546509,
+      "learning_rate": 7.200840571913437e-06,
+      "loss": 0.6417,
+      "step": 22988
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.178938150405884,
+      "learning_rate": 7.200603826120431e-06,
+      "loss": 0.4159,
+      "step": 22989
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.2689759731292725,
+      "learning_rate": 7.200367074208371e-06,
+      "loss": 0.6996,
+      "step": 22990
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.784456253051758,
+      "learning_rate": 7.2001303161779135e-06,
+      "loss": 0.7648,
+      "step": 22991
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.2240545749664307,
+      "learning_rate": 7.199893552029719e-06,
+      "loss": 0.5959,
+      "step": 22992
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.8260178565979004,
+      "learning_rate": 7.1996567817644435e-06,
+      "loss": 0.6457,
+      "step": 22993
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.256542444229126,
+      "learning_rate": 7.199420005382747e-06,
+      "loss": 0.6952,
+      "step": 22994
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 4.002071380615234,
+      "learning_rate": 7.199183222885287e-06,
+      "loss": 0.6307,
+      "step": 22995
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 4.328883171081543,
+      "learning_rate": 7.198946434272721e-06,
+      "loss": 0.963,
+      "step": 22996
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.525756359100342,
+      "learning_rate": 7.198709639545711e-06,
+      "loss": 0.6868,
+      "step": 22997
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.220348596572876,
+      "learning_rate": 7.1984728387049115e-06,
+      "loss": 0.6317,
+      "step": 22998
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 3.2356276512145996,
+      "learning_rate": 7.198236031750984e-06,
+      "loss": 0.6804,
+      "step": 22999
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.2578577995300293,
+      "learning_rate": 7.197999218684585e-06,
+      "loss": 0.5455,
+      "step": 23000
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.9132564067840576,
+      "learning_rate": 7.197762399506374e-06,
+      "loss": 0.3309,
+      "step": 23001
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.9350101947784424,
+      "learning_rate": 7.197525574217008e-06,
+      "loss": 0.9054,
+      "step": 23002
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 3.301637649536133,
+      "learning_rate": 7.197288742817148e-06,
+      "loss": 0.6593,
+      "step": 23003
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.7184882164001465,
+      "learning_rate": 7.1970519053074505e-06,
+      "loss": 0.3698,
+      "step": 23004
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 3.531738758087158,
+      "learning_rate": 7.196815061688575e-06,
+      "loss": 0.7013,
+      "step": 23005
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.5703954696655273,
+      "learning_rate": 7.196578211961177e-06,
+      "loss": 0.5931,
+      "step": 23006
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.7190059423446655,
+      "learning_rate": 7.196341356125921e-06,
+      "loss": 0.3665,
+      "step": 23007
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.8446743488311768,
+      "learning_rate": 7.196104494183461e-06,
+      "loss": 0.8024,
+      "step": 23008
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.708988904953003,
+      "learning_rate": 7.195867626134458e-06,
+      "loss": 0.5919,
+      "step": 23009
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.925919532775879,
+      "learning_rate": 7.1956307519795695e-06,
+      "loss": 0.8034,
+      "step": 23010
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.1618239879608154,
+      "learning_rate": 7.195393871719454e-06,
+      "loss": 0.5737,
+      "step": 23011
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.4454245567321777,
+      "learning_rate": 7.195156985354771e-06,
+      "loss": 0.6231,
+      "step": 23012
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.4675498008728027,
+      "learning_rate": 7.194920092886179e-06,
+      "loss": 0.688,
+      "step": 23013
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.4985079765319824,
+      "learning_rate": 7.194683194314336e-06,
+      "loss": 0.827,
+      "step": 23014
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.691073417663574,
+      "learning_rate": 7.194446289639902e-06,
+      "loss": 0.7525,
+      "step": 23015
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 3.0749704837799072,
+      "learning_rate": 7.194209378863533e-06,
+      "loss": 0.5271,
+      "step": 23016
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.699605107307434,
+      "learning_rate": 7.1939724619858905e-06,
+      "loss": 0.7145,
+      "step": 23017
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 3.351555585861206,
+      "learning_rate": 7.193735539007632e-06,
+      "loss": 0.7535,
+      "step": 23018
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.849305272102356,
+      "learning_rate": 7.193498609929419e-06,
+      "loss": 0.6907,
+      "step": 23019
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.184070110321045,
+      "learning_rate": 7.193261674751906e-06,
+      "loss": 0.4972,
+      "step": 23020
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 3.244166612625122,
+      "learning_rate": 7.193024733475755e-06,
+      "loss": 0.495,
+      "step": 23021
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.5826416015625,
+      "learning_rate": 7.192787786101622e-06,
+      "loss": 0.7786,
+      "step": 23022
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.085651159286499,
+      "learning_rate": 7.1925508326301696e-06,
+      "loss": 0.7373,
+      "step": 23023
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.9179394245147705,
+      "learning_rate": 7.192313873062055e-06,
+      "loss": 0.475,
+      "step": 23024
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.4556366205215454,
+      "learning_rate": 7.1920769073979356e-06,
+      "loss": 0.52,
+      "step": 23025
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.7478222846984863,
+      "learning_rate": 7.191839935638471e-06,
+      "loss": 0.9626,
+      "step": 23026
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.848185658454895,
+      "learning_rate": 7.191602957784321e-06,
+      "loss": 0.6372,
+      "step": 23027
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 4.39971923828125,
+      "learning_rate": 7.191365973836144e-06,
+      "loss": 0.7915,
+      "step": 23028
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.612882137298584,
+      "learning_rate": 7.191128983794599e-06,
+      "loss": 0.476,
+      "step": 23029
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 3.1700451374053955,
+      "learning_rate": 7.190891987660347e-06,
+      "loss": 0.4007,
+      "step": 23030
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.552071452140808,
+      "learning_rate": 7.1906549854340435e-06,
+      "loss": 0.5866,
+      "step": 23031
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.2171390056610107,
+      "learning_rate": 7.1904179771163486e-06,
+      "loss": 0.7268,
+      "step": 23032
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.0051088333129883,
+      "learning_rate": 7.1901809627079234e-06,
+      "loss": 0.531,
+      "step": 23033
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.154240846633911,
+      "learning_rate": 7.189943942209424e-06,
+      "loss": 0.7097,
+      "step": 23034
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 3.156074285507202,
+      "learning_rate": 7.189706915621513e-06,
+      "loss": 0.7373,
+      "step": 23035
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 6.295186996459961,
+      "learning_rate": 7.189469882944844e-06,
+      "loss": 0.5328,
+      "step": 23036
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.6678715944290161,
+      "learning_rate": 7.189232844180081e-06,
+      "loss": 0.7863,
+      "step": 23037
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.3178181648254395,
+      "learning_rate": 7.188995799327883e-06,
+      "loss": 0.7788,
+      "step": 23038
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.9611800909042358,
+      "learning_rate": 7.188758748388906e-06,
+      "loss": 0.9524,
+      "step": 23039
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.6970431804656982,
+      "learning_rate": 7.1885216913638115e-06,
+      "loss": 0.5261,
+      "step": 23040
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 3.101238489151001,
+      "learning_rate": 7.188284628253258e-06,
+      "loss": 0.5874,
+      "step": 23041
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.6122359037399292,
+      "learning_rate": 7.188047559057905e-06,
+      "loss": 0.5955,
+      "step": 23042
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 4.028682708740234,
+      "learning_rate": 7.18781048377841e-06,
+      "loss": 0.5595,
+      "step": 23043
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.733490228652954,
+      "learning_rate": 7.187573402415435e-06,
+      "loss": 0.721,
+      "step": 23044
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 3.5939016342163086,
+      "learning_rate": 7.187336314969638e-06,
+      "loss": 0.9443,
+      "step": 23045
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.8844397068023682,
+      "learning_rate": 7.187099221441677e-06,
+      "loss": 0.7767,
+      "step": 23046
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.9135737419128418,
+      "learning_rate": 7.186862121832212e-06,
+      "loss": 0.7923,
+      "step": 23047
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 3.937919855117798,
+      "learning_rate": 7.186625016141905e-06,
+      "loss": 0.6076,
+      "step": 23048
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.1073923110961914,
+      "learning_rate": 7.186387904371411e-06,
+      "loss": 0.4483,
+      "step": 23049
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.1087474822998047,
+      "learning_rate": 7.186150786521391e-06,
+      "loss": 0.8292,
+      "step": 23050
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.215707302093506,
+      "learning_rate": 7.185913662592505e-06,
+      "loss": 0.8833,
+      "step": 23051
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.257782459259033,
+      "learning_rate": 7.185676532585413e-06,
+      "loss": 0.8034,
+      "step": 23052
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 3.3575193881988525,
+      "learning_rate": 7.185439396500773e-06,
+      "loss": 0.7095,
+      "step": 23053
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 3.0439136028289795,
+      "learning_rate": 7.185202254339245e-06,
+      "loss": 0.4822,
+      "step": 23054
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.405132532119751,
+      "learning_rate": 7.1849651061014865e-06,
+      "loss": 0.2735,
+      "step": 23055
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.889587163925171,
+      "learning_rate": 7.184727951788159e-06,
+      "loss": 0.7149,
+      "step": 23056
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 3.7485744953155518,
+      "learning_rate": 7.184490791399923e-06,
+      "loss": 1.0604,
+      "step": 23057
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.990889310836792,
+      "learning_rate": 7.184253624937435e-06,
+      "loss": 0.6463,
+      "step": 23058
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.075706720352173,
+      "learning_rate": 7.1840164524013565e-06,
+      "loss": 0.8522,
+      "step": 23059
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.608903646469116,
+      "learning_rate": 7.183779273792347e-06,
+      "loss": 0.5427,
+      "step": 23060
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.3659521341323853,
+      "learning_rate": 7.183542089111065e-06,
+      "loss": 0.2721,
+      "step": 23061
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.2694802284240723,
+      "learning_rate": 7.18330489835817e-06,
+      "loss": 0.5147,
+      "step": 23062
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.9037328958511353,
+      "learning_rate": 7.183067701534323e-06,
+      "loss": 0.7834,
+      "step": 23063
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.8974783420562744,
+      "learning_rate": 7.182830498640181e-06,
+      "loss": 0.7766,
+      "step": 23064
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.7717481851577759,
+      "learning_rate": 7.182593289676406e-06,
+      "loss": 0.6927,
+      "step": 23065
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 3.1944382190704346,
+      "learning_rate": 7.1823560746436575e-06,
+      "loss": 0.51,
+      "step": 23066
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.48317289352417,
+      "learning_rate": 7.1821188535425914e-06,
+      "loss": 0.5727,
+      "step": 23067
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.992323398590088,
+      "learning_rate": 7.1818816263738726e-06,
+      "loss": 0.5912,
+      "step": 23068
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 3.9003450870513916,
+      "learning_rate": 7.181644393138157e-06,
+      "loss": 0.7713,
+      "step": 23069
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 3.120371103286743,
+      "learning_rate": 7.1814071538361075e-06,
+      "loss": 0.9251,
+      "step": 23070
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.8630876541137695,
+      "learning_rate": 7.181169908468381e-06,
+      "loss": 0.5757,
+      "step": 23071
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.578931450843811,
+      "learning_rate": 7.1809326570356376e-06,
+      "loss": 0.6116,
+      "step": 23072
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.0563735961914062,
+      "learning_rate": 7.180695399538537e-06,
+      "loss": 0.4789,
+      "step": 23073
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.8774707317352295,
+      "learning_rate": 7.1804581359777415e-06,
+      "loss": 0.6243,
+      "step": 23074
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.84724497795105,
+      "learning_rate": 7.180220866353906e-06,
+      "loss": 0.515,
+      "step": 23075
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.255556344985962,
+      "learning_rate": 7.179983590667695e-06,
+      "loss": 0.6812,
+      "step": 23076
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.35644793510437,
+      "learning_rate": 7.179746308919765e-06,
+      "loss": 0.535,
+      "step": 23077
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.7764747142791748,
+      "learning_rate": 7.1795090211107776e-06,
+      "loss": 0.5984,
+      "step": 23078
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.456233501434326,
+      "learning_rate": 7.179271727241392e-06,
+      "loss": 0.6326,
+      "step": 23079
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.183655023574829,
+      "learning_rate": 7.1790344273122696e-06,
+      "loss": 0.7096,
+      "step": 23080
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.130850076675415,
+      "learning_rate": 7.178797121324067e-06,
+      "loss": 0.5317,
+      "step": 23081
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.794181227684021,
+      "learning_rate": 7.178559809277445e-06,
+      "loss": 0.6745,
+      "step": 23082
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.4854836463928223,
+      "learning_rate": 7.1783224911730665e-06,
+      "loss": 0.4917,
+      "step": 23083
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.352006196975708,
+      "learning_rate": 7.178085167011588e-06,
+      "loss": 0.6595,
+      "step": 23084
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.905273199081421,
+      "learning_rate": 7.177847836793671e-06,
+      "loss": 0.4888,
+      "step": 23085
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 3.047839879989624,
+      "learning_rate": 7.177610500519976e-06,
+      "loss": 0.6464,
+      "step": 23086
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.310967206954956,
+      "learning_rate": 7.177373158191159e-06,
+      "loss": 0.861,
+      "step": 23087
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.549422025680542,
+      "learning_rate": 7.177135809807885e-06,
+      "loss": 0.6942,
+      "step": 23088
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.9404165744781494,
+      "learning_rate": 7.176898455370811e-06,
+      "loss": 0.6778,
+      "step": 23089
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 3.7815065383911133,
+      "learning_rate": 7.176661094880599e-06,
+      "loss": 0.5906,
+      "step": 23090
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.3294596672058105,
+      "learning_rate": 7.176423728337907e-06,
+      "loss": 0.682,
+      "step": 23091
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.6998769044876099,
+      "learning_rate": 7.1761863557433965e-06,
+      "loss": 0.5366,
+      "step": 23092
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.4851505756378174,
+      "learning_rate": 7.175948977097727e-06,
+      "loss": 0.5471,
+      "step": 23093
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.7161645889282227,
+      "learning_rate": 7.175711592401558e-06,
+      "loss": 0.5951,
+      "step": 23094
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 3.5233449935913086,
+      "learning_rate": 7.17547420165555e-06,
+      "loss": 0.6795,
+      "step": 23095
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.695329189300537,
+      "learning_rate": 7.175236804860364e-06,
+      "loss": 0.5111,
+      "step": 23096
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.297673463821411,
+      "learning_rate": 7.174999402016658e-06,
+      "loss": 0.5433,
+      "step": 23097
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.19856333732605,
+      "learning_rate": 7.174761993125094e-06,
+      "loss": 0.5501,
+      "step": 23098
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 4.053053855895996,
+      "learning_rate": 7.1745245781863325e-06,
+      "loss": 0.9638,
+      "step": 23099
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.4835898876190186,
+      "learning_rate": 7.174287157201032e-06,
+      "loss": 0.7541,
+      "step": 23100
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.9680246114730835,
+      "learning_rate": 7.174049730169854e-06,
+      "loss": 0.6507,
+      "step": 23101
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.746375799179077,
+      "learning_rate": 7.173812297093457e-06,
+      "loss": 0.6633,
+      "step": 23102
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.166611671447754,
+      "learning_rate": 7.173574857972503e-06,
+      "loss": 0.615,
+      "step": 23103
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.8688693046569824,
+      "learning_rate": 7.173337412807652e-06,
+      "loss": 0.5412,
+      "step": 23104
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.015742301940918,
+      "learning_rate": 7.173099961599563e-06,
+      "loss": 0.4562,
+      "step": 23105
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.914568305015564,
+      "learning_rate": 7.172862504348896e-06,
+      "loss": 0.2148,
+      "step": 23106
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.377168893814087,
+      "learning_rate": 7.172625041056314e-06,
+      "loss": 0.7267,
+      "step": 23107
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.4438865184783936,
+      "learning_rate": 7.172387571722476e-06,
+      "loss": 0.7966,
+      "step": 23108
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.4977633953094482,
+      "learning_rate": 7.1721500963480415e-06,
+      "loss": 1.0712,
+      "step": 23109
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.5876963138580322,
+      "learning_rate": 7.171912614933672e-06,
+      "loss": 0.7039,
+      "step": 23110
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.2006373405456543,
+      "learning_rate": 7.171675127480026e-06,
+      "loss": 0.597,
+      "step": 23111
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.7438334226608276,
+      "learning_rate": 7.171437633987763e-06,
+      "loss": 0.6389,
+      "step": 23112
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.6830552816390991,
+      "learning_rate": 7.1712001344575485e-06,
+      "loss": 0.5797,
+      "step": 23113
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.264241933822632,
+      "learning_rate": 7.170962628890039e-06,
+      "loss": 0.8377,
+      "step": 23114
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.9974440336227417,
+      "learning_rate": 7.170725117285896e-06,
+      "loss": 0.5853,
+      "step": 23115
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.8307567834854126,
+      "learning_rate": 7.170487599645779e-06,
+      "loss": 0.7369,
+      "step": 23116
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 4.823836803436279,
+      "learning_rate": 7.17025007597035e-06,
+      "loss": 0.7405,
+      "step": 23117
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 3.109677791595459,
+      "learning_rate": 7.1700125462602665e-06,
+      "loss": 0.8206,
+      "step": 23118
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.503307342529297,
+      "learning_rate": 7.169775010516192e-06,
+      "loss": 0.654,
+      "step": 23119
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.616849422454834,
+      "learning_rate": 7.169537468738786e-06,
+      "loss": 0.4798,
+      "step": 23120
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.88630747795105,
+      "learning_rate": 7.16929992092871e-06,
+      "loss": 0.7122,
+      "step": 23121
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 3.7857649326324463,
+      "learning_rate": 7.169062367086624e-06,
+      "loss": 0.6882,
+      "step": 23122
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.2647860050201416,
+      "learning_rate": 7.168824807213186e-06,
+      "loss": 0.7861,
+      "step": 23123
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.331385850906372,
+      "learning_rate": 7.1685872413090595e-06,
+      "loss": 0.9108,
+      "step": 23124
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.933314561843872,
+      "learning_rate": 7.168349669374905e-06,
+      "loss": 0.7194,
+      "step": 23125
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 3.890326738357544,
+      "learning_rate": 7.168112091411382e-06,
+      "loss": 0.7936,
+      "step": 23126
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 3.3408761024475098,
+      "learning_rate": 7.167874507419151e-06,
+      "loss": 0.3665,
+      "step": 23127
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.909907102584839,
+      "learning_rate": 7.167636917398873e-06,
+      "loss": 0.6078,
+      "step": 23128
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.4274706840515137,
+      "learning_rate": 7.167399321351209e-06,
+      "loss": 0.5739,
+      "step": 23129
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.7752108573913574,
+      "learning_rate": 7.167161719276819e-06,
+      "loss": 0.8321,
+      "step": 23130
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.5774593353271484,
+      "learning_rate": 7.166924111176366e-06,
+      "loss": 0.8568,
+      "step": 23131
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.0616443157196045,
+      "learning_rate": 7.166686497050507e-06,
+      "loss": 0.8246,
+      "step": 23132
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.633200764656067,
+      "learning_rate": 7.166448876899905e-06,
+      "loss": 0.5393,
+      "step": 23133
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.277876377105713,
+      "learning_rate": 7.16621125072522e-06,
+      "loss": 0.5839,
+      "step": 23134
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.6013243198394775,
+      "learning_rate": 7.165973618527113e-06,
+      "loss": 0.6713,
+      "step": 23135
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.587825894355774,
+      "learning_rate": 7.165735980306245e-06,
+      "loss": 0.6858,
+      "step": 23136
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 4.020458698272705,
+      "learning_rate": 7.165498336063277e-06,
+      "loss": 0.8561,
+      "step": 23137
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.44084095954895,
+      "learning_rate": 7.165260685798868e-06,
+      "loss": 0.4967,
+      "step": 23138
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.930539846420288,
+      "learning_rate": 7.16502302951368e-06,
+      "loss": 0.8573,
+      "step": 23139
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.2023415565490723,
+      "learning_rate": 7.1647853672083756e-06,
+      "loss": 0.812,
+      "step": 23140
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 3.345968246459961,
+      "learning_rate": 7.1645476988836135e-06,
+      "loss": 0.7129,
+      "step": 23141
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.5658366680145264,
+      "learning_rate": 7.164310024540056e-06,
+      "loss": 0.7138,
+      "step": 23142
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.2915749549865723,
+      "learning_rate": 7.164072344178361e-06,
+      "loss": 0.7759,
+      "step": 23143
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.971457839012146,
+      "learning_rate": 7.163834657799193e-06,
+      "loss": 0.6751,
+      "step": 23144
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.067340135574341,
+      "learning_rate": 7.163596965403211e-06,
+      "loss": 0.6027,
+      "step": 23145
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.1491754055023193,
+      "learning_rate": 7.163359266991077e-06,
+      "loss": 1.0787,
+      "step": 23146
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.159005641937256,
+      "learning_rate": 7.163121562563451e-06,
+      "loss": 0.445,
+      "step": 23147
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.2783970832824707,
+      "learning_rate": 7.162883852120992e-06,
+      "loss": 0.5758,
+      "step": 23148
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.683901786804199,
+      "learning_rate": 7.162646135664366e-06,
+      "loss": 0.6053,
+      "step": 23149
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.7425950765609741,
+      "learning_rate": 7.162408413194231e-06,
+      "loss": 0.6337,
+      "step": 23150
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.087120294570923,
+      "learning_rate": 7.162170684711247e-06,
+      "loss": 0.4475,
+      "step": 23151
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.420745372772217,
+      "learning_rate": 7.1619329502160776e-06,
+      "loss": 0.5335,
+      "step": 23152
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.8839080333709717,
+      "learning_rate": 7.161695209709382e-06,
+      "loss": 0.5214,
+      "step": 23153
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.770293712615967,
+      "learning_rate": 7.161457463191821e-06,
+      "loss": 0.7016,
+      "step": 23154
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.172456741333008,
+      "learning_rate": 7.161219710664058e-06,
+      "loss": 0.6644,
+      "step": 23155
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.8734889030456543,
+      "learning_rate": 7.160981952126751e-06,
+      "loss": 0.5921,
+      "step": 23156
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.326585292816162,
+      "learning_rate": 7.160744187580564e-06,
+      "loss": 0.5711,
+      "step": 23157
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.5026421546936035,
+      "learning_rate": 7.160506417026155e-06,
+      "loss": 0.5527,
+      "step": 23158
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.4107933044433594,
+      "learning_rate": 7.160268640464189e-06,
+      "loss": 0.69,
+      "step": 23159
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.24715256690979,
+      "learning_rate": 7.160030857895324e-06,
+      "loss": 0.5964,
+      "step": 23160
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 3.8967514038085938,
+      "learning_rate": 7.159793069320223e-06,
+      "loss": 0.7291,
+      "step": 23161
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.844149112701416,
+      "learning_rate": 7.159555274739547e-06,
+      "loss": 0.7923,
+      "step": 23162
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 3.298642635345459,
+      "learning_rate": 7.1593174741539566e-06,
+      "loss": 0.6201,
+      "step": 23163
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.6805329322814941,
+      "learning_rate": 7.159079667564113e-06,
+      "loss": 0.6312,
+      "step": 23164
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.7985613346099854,
+      "learning_rate": 7.158841854970677e-06,
+      "loss": 0.6603,
+      "step": 23165
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.6899197101593018,
+      "learning_rate": 7.15860403637431e-06,
+      "loss": 0.725,
+      "step": 23166
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.850614309310913,
+      "learning_rate": 7.158366211775676e-06,
+      "loss": 0.7195,
+      "step": 23167
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.19757342338562,
+      "learning_rate": 7.158128381175432e-06,
+      "loss": 0.8389,
+      "step": 23168
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 3.169414520263672,
+      "learning_rate": 7.157890544574242e-06,
+      "loss": 0.768,
+      "step": 23169
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.0387909412384033,
+      "learning_rate": 7.157652701972767e-06,
+      "loss": 0.7019,
+      "step": 23170
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.233560800552368,
+      "learning_rate": 7.1574148533716694e-06,
+      "loss": 0.6398,
+      "step": 23171
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.5393717288970947,
+      "learning_rate": 7.157176998771607e-06,
+      "loss": 0.5752,
+      "step": 23172
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 3.457228422164917,
+      "learning_rate": 7.156939138173245e-06,
+      "loss": 0.8064,
+      "step": 23173
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.738083004951477,
+      "learning_rate": 7.1567012715772424e-06,
+      "loss": 0.6645,
+      "step": 23174
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.382361650466919,
+      "learning_rate": 7.156463398984261e-06,
+      "loss": 0.4973,
+      "step": 23175
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.8941409587860107,
+      "learning_rate": 7.1562255203949634e-06,
+      "loss": 0.7155,
+      "step": 23176
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.7301323413848877,
+      "learning_rate": 7.155987635810011e-06,
+      "loss": 0.7427,
+      "step": 23177
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.9172589778900146,
+      "learning_rate": 7.1557497452300655e-06,
+      "loss": 0.6632,
+      "step": 23178
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.2751054763793945,
+      "learning_rate": 7.155511848655785e-06,
+      "loss": 0.7706,
+      "step": 23179
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.4200656414031982,
+      "learning_rate": 7.155273946087834e-06,
+      "loss": 0.4352,
+      "step": 23180
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.5413475036621094,
+      "learning_rate": 7.155036037526874e-06,
+      "loss": 0.5957,
+      "step": 23181
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 3.5019519329071045,
+      "learning_rate": 7.154798122973567e-06,
+      "loss": 0.4901,
+      "step": 23182
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.5313827991485596,
+      "learning_rate": 7.154560202428574e-06,
+      "loss": 0.6056,
+      "step": 23183
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.0283854007720947,
+      "learning_rate": 7.154322275892553e-06,
+      "loss": 0.7719,
+      "step": 23184
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.4616892337799072,
+      "learning_rate": 7.154084343366172e-06,
+      "loss": 0.7754,
+      "step": 23185
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.8793754577636719,
+      "learning_rate": 7.153846404850089e-06,
+      "loss": 0.6617,
+      "step": 23186
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.4627771377563477,
+      "learning_rate": 7.153608460344965e-06,
+      "loss": 0.7033,
+      "step": 23187
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.5598015785217285,
+      "learning_rate": 7.1533705098514626e-06,
+      "loss": 0.5877,
+      "step": 23188
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.280898332595825,
+      "learning_rate": 7.153132553370243e-06,
+      "loss": 0.7069,
+      "step": 23189
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 3.5747365951538086,
+      "learning_rate": 7.15289459090197e-06,
+      "loss": 0.6688,
+      "step": 23190
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.7561864852905273,
+      "learning_rate": 7.1526566224473026e-06,
+      "loss": 0.5431,
+      "step": 23191
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.37829852104187,
+      "learning_rate": 7.1524186480069035e-06,
+      "loss": 0.8506,
+      "step": 23192
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.712142825126648,
+      "learning_rate": 7.152180667581434e-06,
+      "loss": 0.6829,
+      "step": 23193
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.9258016347885132,
+      "learning_rate": 7.1519426811715574e-06,
+      "loss": 0.3361,
+      "step": 23194
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.5408631563186646,
+      "learning_rate": 7.151704688777934e-06,
+      "loss": 0.8525,
+      "step": 23195
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.57787823677063,
+      "learning_rate": 7.151466690401226e-06,
+      "loss": 0.6562,
+      "step": 23196
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.2082085609436035,
+      "learning_rate": 7.151228686042095e-06,
+      "loss": 0.8235,
+      "step": 23197
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.983121871948242,
+      "learning_rate": 7.1509906757012035e-06,
+      "loss": 0.6636,
+      "step": 23198
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.7905545234680176,
+      "learning_rate": 7.150752659379211e-06,
+      "loss": 0.6783,
+      "step": 23199
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 3.5300140380859375,
+      "learning_rate": 7.150514637076782e-06,
+      "loss": 0.6414,
+      "step": 23200
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.7054980993270874,
+      "learning_rate": 7.150276608794578e-06,
+      "loss": 0.5886,
+      "step": 23201
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 3.6414997577667236,
+      "learning_rate": 7.15003857453326e-06,
+      "loss": 0.2499,
+      "step": 23202
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.2598814964294434,
+      "learning_rate": 7.1498005342934895e-06,
+      "loss": 0.7448,
+      "step": 23203
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 3.8712668418884277,
+      "learning_rate": 7.149562488075929e-06,
+      "loss": 0.5404,
+      "step": 23204
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.372981071472168,
+      "learning_rate": 7.149324435881242e-06,
+      "loss": 0.3786,
+      "step": 23205
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.5270017385482788,
+      "learning_rate": 7.149086377710087e-06,
+      "loss": 0.5941,
+      "step": 23206
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 3.6686272621154785,
+      "learning_rate": 7.148848313563129e-06,
+      "loss": 0.7639,
+      "step": 23207
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 3.813493490219116,
+      "learning_rate": 7.1486102434410285e-06,
+      "loss": 0.6438,
+      "step": 23208
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.962719440460205,
+      "learning_rate": 7.148372167344447e-06,
+      "loss": 0.6182,
+      "step": 23209
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.9560692310333252,
+      "learning_rate": 7.1481340852740474e-06,
+      "loss": 0.4499,
+      "step": 23210
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.9096325635910034,
+      "learning_rate": 7.147895997230494e-06,
+      "loss": 0.7288,
+      "step": 23211
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.8964931964874268,
+      "learning_rate": 7.147657903214445e-06,
+      "loss": 0.7773,
+      "step": 23212
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 3.133646011352539,
+      "learning_rate": 7.147419803226563e-06,
+      "loss": 0.5389,
+      "step": 23213
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.6919710636138916,
+      "learning_rate": 7.1471816972675116e-06,
+      "loss": 0.5056,
+      "step": 23214
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.6815277338027954,
+      "learning_rate": 7.146943585337952e-06,
+      "loss": 0.6541,
+      "step": 23215
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.4349100589752197,
+      "learning_rate": 7.146705467438547e-06,
+      "loss": 0.6387,
+      "step": 23216
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 5.619069576263428,
+      "learning_rate": 7.146467343569958e-06,
+      "loss": 0.6895,
+      "step": 23217
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.9320640563964844,
+      "learning_rate": 7.146229213732848e-06,
+      "loss": 0.5496,
+      "step": 23218
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.346966028213501,
+      "learning_rate": 7.1459910779278776e-06,
+      "loss": 0.9421,
+      "step": 23219
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.5898162126541138,
+      "learning_rate": 7.14575293615571e-06,
+      "loss": 0.6543,
+      "step": 23220
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.041095018386841,
+      "learning_rate": 7.145514788417007e-06,
+      "loss": 0.5749,
+      "step": 23221
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.18169903755188,
+      "learning_rate": 7.145276634712432e-06,
+      "loss": 0.7035,
+      "step": 23222
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.417492389678955,
+      "learning_rate": 7.145038475042647e-06,
+      "loss": 0.6519,
+      "step": 23223
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 3.109987258911133,
+      "learning_rate": 7.144800309408313e-06,
+      "loss": 0.8732,
+      "step": 23224
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.141774892807007,
+      "learning_rate": 7.144562137810092e-06,
+      "loss": 0.8926,
+      "step": 23225
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.4921226501464844,
+      "learning_rate": 7.1443239602486466e-06,
+      "loss": 0.3684,
+      "step": 23226
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.2867629528045654,
+      "learning_rate": 7.144085776724641e-06,
+      "loss": 0.6115,
+      "step": 23227
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.838780403137207,
+      "learning_rate": 7.1438475872387354e-06,
+      "loss": 0.8855,
+      "step": 23228
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.711729049682617,
+      "learning_rate": 7.1436093917915925e-06,
+      "loss": 0.8422,
+      "step": 23229
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 3.502176284790039,
+      "learning_rate": 7.143371190383875e-06,
+      "loss": 0.9413,
+      "step": 23230
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 3.412261962890625,
+      "learning_rate": 7.143132983016245e-06,
+      "loss": 0.6469,
+      "step": 23231
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.927839517593384,
+      "learning_rate": 7.142894769689366e-06,
+      "loss": 0.643,
+      "step": 23232
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.2510125637054443,
+      "learning_rate": 7.1426565504038996e-06,
+      "loss": 0.9566,
+      "step": 23233
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.681381344795227,
+      "learning_rate": 7.1424183251605075e-06,
+      "loss": 0.3392,
+      "step": 23234
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.1346049308776855,
+      "learning_rate": 7.1421800939598515e-06,
+      "loss": 0.4751,
+      "step": 23235
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.678821325302124,
+      "learning_rate": 7.1419418568025955e-06,
+      "loss": 0.7903,
+      "step": 23236
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.119126796722412,
+      "learning_rate": 7.141703613689401e-06,
+      "loss": 0.6438,
+      "step": 23237
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.480886459350586,
+      "learning_rate": 7.141465364620933e-06,
+      "loss": 0.7827,
+      "step": 23238
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 3.483642339706421,
+      "learning_rate": 7.141227109597851e-06,
+      "loss": 0.6923,
+      "step": 23239
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.615825653076172,
+      "learning_rate": 7.140988848620819e-06,
+      "loss": 0.6538,
+      "step": 23240
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.079495906829834,
+      "learning_rate": 7.1407505816904975e-06,
+      "loss": 0.6651,
+      "step": 23241
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.765740990638733,
+      "learning_rate": 7.1405123088075526e-06,
+      "loss": 0.6928,
+      "step": 23242
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.6166932582855225,
+      "learning_rate": 7.1402740299726435e-06,
+      "loss": 0.7601,
+      "step": 23243
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.8696608543395996,
+      "learning_rate": 7.140035745186434e-06,
+      "loss": 0.7577,
+      "step": 23244
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 10.335380554199219,
+      "learning_rate": 7.139797454449588e-06,
+      "loss": 0.7461,
+      "step": 23245
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.351466417312622,
+      "learning_rate": 7.139559157762767e-06,
+      "loss": 0.3993,
+      "step": 23246
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 4.435137748718262,
+      "learning_rate": 7.139320855126633e-06,
+      "loss": 0.5616,
+      "step": 23247
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.276932716369629,
+      "learning_rate": 7.139082546541848e-06,
+      "loss": 0.4919,
+      "step": 23248
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.7902600765228271,
+      "learning_rate": 7.1388442320090765e-06,
+      "loss": 0.58,
+      "step": 23249
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.8381975889205933,
+      "learning_rate": 7.138605911528979e-06,
+      "loss": 0.6587,
+      "step": 23250
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 3.0522704124450684,
+      "learning_rate": 7.138367585102221e-06,
+      "loss": 0.9369,
+      "step": 23251
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 3.236083745956421,
+      "learning_rate": 7.138129252729464e-06,
+      "loss": 0.5784,
+      "step": 23252
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 4.191744804382324,
+      "learning_rate": 7.1378909144113715e-06,
+      "loss": 0.8077,
+      "step": 23253
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.6368160247802734,
+      "learning_rate": 7.137652570148604e-06,
+      "loss": 0.773,
+      "step": 23254
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.2049286365509033,
+      "learning_rate": 7.137414219941825e-06,
+      "loss": 0.4479,
+      "step": 23255
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.5923428535461426,
+      "learning_rate": 7.137175863791699e-06,
+      "loss": 0.4179,
+      "step": 23256
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.3562467098236084,
+      "learning_rate": 7.1369375016988875e-06,
+      "loss": 0.6872,
+      "step": 23257
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.0595381259918213,
+      "learning_rate": 7.136699133664053e-06,
+      "loss": 0.4102,
+      "step": 23258
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.516798257827759,
+      "learning_rate": 7.1364607596878575e-06,
+      "loss": 0.5399,
+      "step": 23259
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.042243480682373,
+      "learning_rate": 7.1362223797709655e-06,
+      "loss": 0.6901,
+      "step": 23260
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.6329269409179688,
+      "learning_rate": 7.13598399391404e-06,
+      "loss": 0.8552,
+      "step": 23261
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.7315219640731812,
+      "learning_rate": 7.135745602117742e-06,
+      "loss": 0.5242,
+      "step": 23262
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.9668829441070557,
+      "learning_rate": 7.135507204382737e-06,
+      "loss": 0.6577,
+      "step": 23263
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 3.2308828830718994,
+      "learning_rate": 7.1352688007096844e-06,
+      "loss": 0.819,
+      "step": 23264
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.7929551601409912,
+      "learning_rate": 7.135030391099251e-06,
+      "loss": 0.6218,
+      "step": 23265
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.999929666519165,
+      "learning_rate": 7.134791975552098e-06,
+      "loss": 0.6594,
+      "step": 23266
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 3.9164206981658936,
+      "learning_rate": 7.1345535540688885e-06,
+      "loss": 0.7731,
+      "step": 23267
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.530271530151367,
+      "learning_rate": 7.134315126650284e-06,
+      "loss": 0.886,
+      "step": 23268
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.50498628616333,
+      "learning_rate": 7.134076693296949e-06,
+      "loss": 0.7605,
+      "step": 23269
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 4.186663627624512,
+      "learning_rate": 7.1338382540095455e-06,
+      "loss": 0.5478,
+      "step": 23270
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.6620492935180664,
+      "learning_rate": 7.133599808788737e-06,
+      "loss": 0.4103,
+      "step": 23271
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.5197653770446777,
+      "learning_rate": 7.133361357635188e-06,
+      "loss": 0.6839,
+      "step": 23272
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.851616859436035,
+      "learning_rate": 7.133122900549559e-06,
+      "loss": 0.6064,
+      "step": 23273
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 3.9210922718048096,
+      "learning_rate": 7.132884437532515e-06,
+      "loss": 0.6005,
+      "step": 23274
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.7166591882705688,
+      "learning_rate": 7.1326459685847195e-06,
+      "loss": 0.4829,
+      "step": 23275
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.2346982955932617,
+      "learning_rate": 7.132407493706833e-06,
+      "loss": 0.6494,
+      "step": 23276
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.50639271736145,
+      "learning_rate": 7.132169012899521e-06,
+      "loss": 0.7293,
+      "step": 23277
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.2238974571228027,
+      "learning_rate": 7.1319305261634454e-06,
+      "loss": 0.6874,
+      "step": 23278
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.893775224685669,
+      "learning_rate": 7.131692033499269e-06,
+      "loss": 0.6487,
+      "step": 23279
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.8393890857696533,
+      "learning_rate": 7.1314535349076556e-06,
+      "loss": 0.3316,
+      "step": 23280
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.829911708831787,
+      "learning_rate": 7.131215030389269e-06,
+      "loss": 0.5959,
+      "step": 23281
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.774653196334839,
+      "learning_rate": 7.130976519944771e-06,
+      "loss": 0.5605,
+      "step": 23282
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.0850064754486084,
+      "learning_rate": 7.130738003574826e-06,
+      "loss": 0.4657,
+      "step": 23283
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 3.285487413406372,
+      "learning_rate": 7.130499481280098e-06,
+      "loss": 0.7672,
+      "step": 23284
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.745283603668213,
+      "learning_rate": 7.1302609530612475e-06,
+      "loss": 0.5592,
+      "step": 23285
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 3.3387296199798584,
+      "learning_rate": 7.13002241891894e-06,
+      "loss": 0.6531,
+      "step": 23286
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.858513593673706,
+      "learning_rate": 7.129783878853837e-06,
+      "loss": 0.5231,
+      "step": 23287
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 3.4289515018463135,
+      "learning_rate": 7.129545332866605e-06,
+      "loss": 0.7452,
+      "step": 23288
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 3.6565136909484863,
+      "learning_rate": 7.129306780957905e-06,
+      "loss": 0.7063,
+      "step": 23289
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.1748433113098145,
+      "learning_rate": 7.129068223128398e-06,
+      "loss": 0.7695,
+      "step": 23290
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 4.048457622528076,
+      "learning_rate": 7.128829659378751e-06,
+      "loss": 0.5809,
+      "step": 23291
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.4987378120422363,
+      "learning_rate": 7.128591089709626e-06,
+      "loss": 0.6337,
+      "step": 23292
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.382611036300659,
+      "learning_rate": 7.128352514121688e-06,
+      "loss": 0.4998,
+      "step": 23293
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.4296908378601074,
+      "learning_rate": 7.128113932615598e-06,
+      "loss": 0.8519,
+      "step": 23294
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.7819185256958008,
+      "learning_rate": 7.1278753451920205e-06,
+      "loss": 0.5421,
+      "step": 23295
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.428534984588623,
+      "learning_rate": 7.127636751851619e-06,
+      "loss": 0.88,
+      "step": 23296
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.0616798400878906,
+      "learning_rate": 7.127398152595056e-06,
+      "loss": 0.6538,
+      "step": 23297
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.5228753089904785,
+      "learning_rate": 7.127159547422995e-06,
+      "loss": 0.6838,
+      "step": 23298
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 3.4235453605651855,
+      "learning_rate": 7.126920936336101e-06,
+      "loss": 0.4589,
+      "step": 23299
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.2755208015441895,
+      "learning_rate": 7.126682319335038e-06,
+      "loss": 0.6887,
+      "step": 23300
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.6492063999176025,
+      "learning_rate": 7.1264436964204665e-06,
+      "loss": 0.737,
+      "step": 23301
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 3.5468833446502686,
+      "learning_rate": 7.126205067593051e-06,
+      "loss": 0.5293,
+      "step": 23302
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.7477349042892456,
+      "learning_rate": 7.125966432853457e-06,
+      "loss": 0.5662,
+      "step": 23303
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.8732891082763672,
+      "learning_rate": 7.125727792202346e-06,
+      "loss": 0.6264,
+      "step": 23304
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.1560511589050293,
+      "learning_rate": 7.125489145640384e-06,
+      "loss": 0.7641,
+      "step": 23305
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.893090009689331,
+      "learning_rate": 7.125250493168232e-06,
+      "loss": 0.742,
+      "step": 23306
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.6386539936065674,
+      "learning_rate": 7.125011834786554e-06,
+      "loss": 0.7321,
+      "step": 23307
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.692032814025879,
+      "learning_rate": 7.124773170496015e-06,
+      "loss": 0.4996,
+      "step": 23308
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.4686100482940674,
+      "learning_rate": 7.124534500297277e-06,
+      "loss": 0.6141,
+      "step": 23309
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.9783908128738403,
+      "learning_rate": 7.124295824191006e-06,
+      "loss": 0.6601,
+      "step": 23310
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.7446297407150269,
+      "learning_rate": 7.124057142177861e-06,
+      "loss": 0.7598,
+      "step": 23311
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.2999067306518555,
+      "learning_rate": 7.12381845425851e-06,
+      "loss": 0.4268,
+      "step": 23312
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.5296337604522705,
+      "learning_rate": 7.123579760433617e-06,
+      "loss": 0.4662,
+      "step": 23313
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.8520843982696533,
+      "learning_rate": 7.123341060703843e-06,
+      "loss": 0.6287,
+      "step": 23314
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.5854041576385498,
+      "learning_rate": 7.123102355069853e-06,
+      "loss": 0.7374,
+      "step": 23315
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.1692698001861572,
+      "learning_rate": 7.122863643532311e-06,
+      "loss": 0.6156,
+      "step": 23316
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.5826250314712524,
+      "learning_rate": 7.12262492609188e-06,
+      "loss": 0.4768,
+      "step": 23317
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.1335718631744385,
+      "learning_rate": 7.1223862027492255e-06,
+      "loss": 0.4959,
+      "step": 23318
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.7334823608398438,
+      "learning_rate": 7.1221474735050075e-06,
+      "loss": 0.64,
+      "step": 23319
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.3065671920776367,
+      "learning_rate": 7.121908738359895e-06,
+      "loss": 0.7802,
+      "step": 23320
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.0363152027130127,
+      "learning_rate": 7.121669997314548e-06,
+      "loss": 0.6127,
+      "step": 23321
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.554917335510254,
+      "learning_rate": 7.1214312503696305e-06,
+      "loss": 0.5377,
+      "step": 23322
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 3.0197465419769287,
+      "learning_rate": 7.121192497525808e-06,
+      "loss": 0.7664,
+      "step": 23323
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.8761093616485596,
+      "learning_rate": 7.120953738783746e-06,
+      "loss": 0.6125,
+      "step": 23324
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.686673641204834,
+      "learning_rate": 7.120714974144104e-06,
+      "loss": 0.5724,
+      "step": 23325
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 3.031374216079712,
+      "learning_rate": 7.120476203607549e-06,
+      "loss": 0.5352,
+      "step": 23326
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.6959115266799927,
+      "learning_rate": 7.120237427174743e-06,
+      "loss": 0.5494,
+      "step": 23327
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.8340349197387695,
+      "learning_rate": 7.119998644846352e-06,
+      "loss": 0.4231,
+      "step": 23328
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.061582326889038,
+      "learning_rate": 7.1197598566230385e-06,
+      "loss": 0.6994,
+      "step": 23329
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.2303194999694824,
+      "learning_rate": 7.1195210625054675e-06,
+      "loss": 0.6361,
+      "step": 23330
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.839374303817749,
+      "learning_rate": 7.1192822624943005e-06,
+      "loss": 0.5836,
+      "step": 23331
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.9814658164978027,
+      "learning_rate": 7.119043456590204e-06,
+      "loss": 0.757,
+      "step": 23332
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 3.495537042617798,
+      "learning_rate": 7.118804644793843e-06,
+      "loss": 0.6113,
+      "step": 23333
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 3.6905012130737305,
+      "learning_rate": 7.11856582710588e-06,
+      "loss": 1.1112,
+      "step": 23334
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.7126684188842773,
+      "learning_rate": 7.118327003526978e-06,
+      "loss": 0.7298,
+      "step": 23335
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 3.336960554122925,
+      "learning_rate": 7.1180881740578025e-06,
+      "loss": 0.8274,
+      "step": 23336
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 3.1031410694122314,
+      "learning_rate": 7.117849338699018e-06,
+      "loss": 0.6215,
+      "step": 23337
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.589255452156067,
+      "learning_rate": 7.117610497451286e-06,
+      "loss": 0.6075,
+      "step": 23338
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 4.700674057006836,
+      "learning_rate": 7.117371650315274e-06,
+      "loss": 0.791,
+      "step": 23339
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.93442440032959,
+      "learning_rate": 7.117132797291645e-06,
+      "loss": 0.4878,
+      "step": 23340
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.3556100130081177,
+      "learning_rate": 7.116893938381061e-06,
+      "loss": 0.4742,
+      "step": 23341
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.485426902770996,
+      "learning_rate": 7.116655073584188e-06,
+      "loss": 0.621,
+      "step": 23342
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.1722819805145264,
+      "learning_rate": 7.116416202901691e-06,
+      "loss": 0.7533,
+      "step": 23343
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.789943814277649,
+      "learning_rate": 7.116177326334235e-06,
+      "loss": 0.3533,
+      "step": 23344
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.8350684642791748,
+      "learning_rate": 7.115938443882479e-06,
+      "loss": 0.8694,
+      "step": 23345
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.9542276859283447,
+      "learning_rate": 7.115699555547094e-06,
+      "loss": 0.86,
+      "step": 23346
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.7698343992233276,
+      "learning_rate": 7.1154606613287405e-06,
+      "loss": 0.7228,
+      "step": 23347
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.2947325706481934,
+      "learning_rate": 7.115221761228083e-06,
+      "loss": 0.5186,
+      "step": 23348
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.452608346939087,
+      "learning_rate": 7.114982855245786e-06,
+      "loss": 0.5633,
+      "step": 23349
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.5926096439361572,
+      "learning_rate": 7.114743943382514e-06,
+      "loss": 0.6664,
+      "step": 23350
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 3.1416800022125244,
+      "learning_rate": 7.114505025638931e-06,
+      "loss": 0.4501,
+      "step": 23351
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.8833318948745728,
+      "learning_rate": 7.114266102015701e-06,
+      "loss": 0.6698,
+      "step": 23352
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 3.7358081340789795,
+      "learning_rate": 7.11402717251349e-06,
+      "loss": 0.6731,
+      "step": 23353
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.618974208831787,
+      "learning_rate": 7.113788237132961e-06,
+      "loss": 0.792,
+      "step": 23354
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.3901853561401367,
+      "learning_rate": 7.1135492958747796e-06,
+      "loss": 0.7679,
+      "step": 23355
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.4262588024139404,
+      "learning_rate": 7.11331034873961e-06,
+      "loss": 0.5823,
+      "step": 23356
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.809407114982605,
+      "learning_rate": 7.113071395728114e-06,
+      "loss": 0.5657,
+      "step": 23357
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.935910940170288,
+      "learning_rate": 7.112832436840958e-06,
+      "loss": 0.6147,
+      "step": 23358
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.3296592235565186,
+      "learning_rate": 7.112593472078807e-06,
+      "loss": 0.6814,
+      "step": 23359
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 3.039774179458618,
+      "learning_rate": 7.1123545014423256e-06,
+      "loss": 0.6813,
+      "step": 23360
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.467752695083618,
+      "learning_rate": 7.112115524932177e-06,
+      "loss": 0.6604,
+      "step": 23361
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 5.810988903045654,
+      "learning_rate": 7.111876542549027e-06,
+      "loss": 0.7785,
+      "step": 23362
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.166354179382324,
+      "learning_rate": 7.111637554293539e-06,
+      "loss": 0.4534,
+      "step": 23363
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.1880593299865723,
+      "learning_rate": 7.1113985601663785e-06,
+      "loss": 0.4232,
+      "step": 23364
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.6816587448120117,
+      "learning_rate": 7.1111595601682085e-06,
+      "loss": 0.8071,
+      "step": 23365
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.4454386234283447,
+      "learning_rate": 7.110920554299694e-06,
+      "loss": 0.4175,
+      "step": 23366
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.2665538787841797,
+      "learning_rate": 7.110681542561502e-06,
+      "loss": 0.4927,
+      "step": 23367
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.5461088418960571,
+      "learning_rate": 7.1104425249542955e-06,
+      "loss": 0.659,
+      "step": 23368
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.35099458694458,
+      "learning_rate": 7.110203501478737e-06,
+      "loss": 0.5233,
+      "step": 23369
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.853475570678711,
+      "learning_rate": 7.109964472135495e-06,
+      "loss": 0.3379,
+      "step": 23370
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.9114196300506592,
+      "learning_rate": 7.109725436925232e-06,
+      "loss": 0.6624,
+      "step": 23371
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.179893732070923,
+      "learning_rate": 7.10948639584861e-06,
+      "loss": 0.7169,
+      "step": 23372
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.3268394470214844,
+      "learning_rate": 7.1092473489063e-06,
+      "loss": 0.9005,
+      "step": 23373
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.4430124759674072,
+      "learning_rate": 7.109008296098963e-06,
+      "loss": 0.6395,
+      "step": 23374
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.274216413497925,
+      "learning_rate": 7.108769237427262e-06,
+      "loss": 0.6741,
+      "step": 23375
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.7487144470214844,
+      "learning_rate": 7.108530172891865e-06,
+      "loss": 0.7761,
+      "step": 23376
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.95948326587677,
+      "learning_rate": 7.108291102493437e-06,
+      "loss": 0.6622,
+      "step": 23377
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.9906558990478516,
+      "learning_rate": 7.108052026232638e-06,
+      "loss": 0.587,
+      "step": 23378
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.947556972503662,
+      "learning_rate": 7.107812944110138e-06,
+      "loss": 0.5181,
+      "step": 23379
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.1774067878723145,
+      "learning_rate": 7.1075738561266e-06,
+      "loss": 0.9605,
+      "step": 23380
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 4.203575134277344,
+      "learning_rate": 7.107334762282687e-06,
+      "loss": 0.4506,
+      "step": 23381
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.789088249206543,
+      "learning_rate": 7.107095662579066e-06,
+      "loss": 0.7697,
+      "step": 23382
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 3.751852035522461,
+      "learning_rate": 7.1068565570164015e-06,
+      "loss": 0.6788,
+      "step": 23383
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.2167246341705322,
+      "learning_rate": 7.106617445595359e-06,
+      "loss": 0.7767,
+      "step": 23384
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.2946417331695557,
+      "learning_rate": 7.106378328316602e-06,
+      "loss": 0.5339,
+      "step": 23385
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.60104501247406,
+      "learning_rate": 7.106139205180795e-06,
+      "loss": 0.489,
+      "step": 23386
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.962076425552368,
+      "learning_rate": 7.105900076188605e-06,
+      "loss": 0.8608,
+      "step": 23387
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.260652542114258,
+      "learning_rate": 7.105660941340696e-06,
+      "loss": 0.5707,
+      "step": 23388
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.9846528768539429,
+      "learning_rate": 7.105421800637733e-06,
+      "loss": 0.7012,
+      "step": 23389
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.696594476699829,
+      "learning_rate": 7.105182654080379e-06,
+      "loss": 0.4909,
+      "step": 23390
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.3376705646514893,
+      "learning_rate": 7.104943501669302e-06,
+      "loss": 0.5174,
+      "step": 23391
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 4.150249481201172,
+      "learning_rate": 7.104704343405164e-06,
+      "loss": 0.4306,
+      "step": 23392
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.109917402267456,
+      "learning_rate": 7.104465179288634e-06,
+      "loss": 0.7723,
+      "step": 23393
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.0533454418182373,
+      "learning_rate": 7.104226009320373e-06,
+      "loss": 0.7081,
+      "step": 23394
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 3.4010493755340576,
+      "learning_rate": 7.103986833501049e-06,
+      "loss": 0.708,
+      "step": 23395
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.5577328205108643,
+      "learning_rate": 7.103747651831325e-06,
+      "loss": 0.8101,
+      "step": 23396
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.8834972381591797,
+      "learning_rate": 7.1035084643118666e-06,
+      "loss": 0.76,
+      "step": 23397
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 3.0468945503234863,
+      "learning_rate": 7.10326927094334e-06,
+      "loss": 0.7007,
+      "step": 23398
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 4.925734043121338,
+      "learning_rate": 7.103030071726409e-06,
+      "loss": 0.8614,
+      "step": 23399
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.3547000885009766,
+      "learning_rate": 7.10279086666174e-06,
+      "loss": 0.68,
+      "step": 23400
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.6789684295654297,
+      "learning_rate": 7.102551655749995e-06,
+      "loss": 0.7352,
+      "step": 23401
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.675685167312622,
+      "learning_rate": 7.102312438991843e-06,
+      "loss": 0.8833,
+      "step": 23402
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.8553106784820557,
+      "learning_rate": 7.1020732163879466e-06,
+      "loss": 0.5813,
+      "step": 23403
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.9805368185043335,
+      "learning_rate": 7.101833987938973e-06,
+      "loss": 0.6295,
+      "step": 23404
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.022392749786377,
+      "learning_rate": 7.101594753645588e-06,
+      "loss": 0.9395,
+      "step": 23405
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 3.3778319358825684,
+      "learning_rate": 7.101355513508453e-06,
+      "loss": 0.6689,
+      "step": 23406
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.602959632873535,
+      "learning_rate": 7.101116267528236e-06,
+      "loss": 0.8429,
+      "step": 23407
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.2193355560302734,
+      "learning_rate": 7.100877015705603e-06,
+      "loss": 0.6228,
+      "step": 23408
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.8427820205688477,
+      "learning_rate": 7.100637758041217e-06,
+      "loss": 0.7507,
+      "step": 23409
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.5487208366394043,
+      "learning_rate": 7.100398494535744e-06,
+      "loss": 0.7788,
+      "step": 23410
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 3.8756518363952637,
+      "learning_rate": 7.10015922518985e-06,
+      "loss": 0.9043,
+      "step": 23411
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 3.131970167160034,
+      "learning_rate": 7.099919950004199e-06,
+      "loss": 0.7742,
+      "step": 23412
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.583113670349121,
+      "learning_rate": 7.099680668979459e-06,
+      "loss": 0.7984,
+      "step": 23413
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.902883529663086,
+      "learning_rate": 7.099441382116292e-06,
+      "loss": 0.7684,
+      "step": 23414
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.944608449935913,
+      "learning_rate": 7.0992020894153655e-06,
+      "loss": 0.7437,
+      "step": 23415
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.1438238620758057,
+      "learning_rate": 7.098962790877345e-06,
+      "loss": 0.523,
+      "step": 23416
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.561443328857422,
+      "learning_rate": 7.0987234865028955e-06,
+      "loss": 0.5455,
+      "step": 23417
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 3.2222111225128174,
+      "learning_rate": 7.098484176292681e-06,
+      "loss": 0.604,
+      "step": 23418
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.137556791305542,
+      "learning_rate": 7.09824486024737e-06,
+      "loss": 0.8144,
+      "step": 23419
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 3.7894632816314697,
+      "learning_rate": 7.098005538367624e-06,
+      "loss": 0.7682,
+      "step": 23420
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.291992425918579,
+      "learning_rate": 7.09776621065411e-06,
+      "loss": 0.4472,
+      "step": 23421
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 3.1136045455932617,
+      "learning_rate": 7.0975268771074955e-06,
+      "loss": 0.5714,
+      "step": 23422
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.075509786605835,
+      "learning_rate": 7.097287537728445e-06,
+      "loss": 0.4895,
+      "step": 23423
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.7252719402313232,
+      "learning_rate": 7.097048192517623e-06,
+      "loss": 0.6505,
+      "step": 23424
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.48848295211792,
+      "learning_rate": 7.096808841475695e-06,
+      "loss": 0.574,
+      "step": 23425
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.7917169332504272,
+      "learning_rate": 7.096569484603327e-06,
+      "loss": 0.7269,
+      "step": 23426
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.450418710708618,
+      "learning_rate": 7.096330121901185e-06,
+      "loss": 0.5744,
+      "step": 23427
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.9152787923812866,
+      "learning_rate": 7.096090753369934e-06,
+      "loss": 0.7805,
+      "step": 23428
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.8126152753829956,
+      "learning_rate": 7.09585137901024e-06,
+      "loss": 0.567,
+      "step": 23429
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.9517616033554077,
+      "learning_rate": 7.095611998822769e-06,
+      "loss": 0.613,
+      "step": 23430
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.4359309673309326,
+      "learning_rate": 7.0953726128081855e-06,
+      "loss": 0.8611,
+      "step": 23431
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.504579782485962,
+      "learning_rate": 7.0951332209671565e-06,
+      "loss": 0.8935,
+      "step": 23432
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 4.120340347290039,
+      "learning_rate": 7.094893823300345e-06,
+      "loss": 0.582,
+      "step": 23433
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.695543050765991,
+      "learning_rate": 7.09465441980842e-06,
+      "loss": 0.6359,
+      "step": 23434
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.7008190155029297,
+      "learning_rate": 7.094415010492046e-06,
+      "loss": 0.5744,
+      "step": 23435
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.124042272567749,
+      "learning_rate": 7.094175595351887e-06,
+      "loss": 0.7094,
+      "step": 23436
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.870013952255249,
+      "learning_rate": 7.093936174388611e-06,
+      "loss": 0.6051,
+      "step": 23437
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.8560266494750977,
+      "learning_rate": 7.093696747602884e-06,
+      "loss": 0.58,
+      "step": 23438
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.1188998222351074,
+      "learning_rate": 7.09345731499537e-06,
+      "loss": 0.6572,
+      "step": 23439
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.6093989610671997,
+      "learning_rate": 7.093217876566735e-06,
+      "loss": 0.7137,
+      "step": 23440
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 3.2493340969085693,
+      "learning_rate": 7.092978432317645e-06,
+      "loss": 0.5926,
+      "step": 23441
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 3.361032247543335,
+      "learning_rate": 7.092738982248767e-06,
+      "loss": 0.7279,
+      "step": 23442
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.706144094467163,
+      "learning_rate": 7.0924995263607645e-06,
+      "loss": 0.5628,
+      "step": 23443
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 4.004696369171143,
+      "learning_rate": 7.092260064654306e-06,
+      "loss": 1.0738,
+      "step": 23444
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.8319694995880127,
+      "learning_rate": 7.092020597130055e-06,
+      "loss": 0.7052,
+      "step": 23445
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.4333221912384033,
+      "learning_rate": 7.091781123788679e-06,
+      "loss": 0.6064,
+      "step": 23446
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 3.3196752071380615,
+      "learning_rate": 7.0915416446308435e-06,
+      "loss": 0.6722,
+      "step": 23447
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.024074077606201,
+      "learning_rate": 7.091302159657214e-06,
+      "loss": 0.5924,
+      "step": 23448
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.334056854248047,
+      "learning_rate": 7.0910626688684555e-06,
+      "loss": 0.5018,
+      "step": 23449
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 3.061462163925171,
+      "learning_rate": 7.090823172265236e-06,
+      "loss": 0.5302,
+      "step": 23450
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.964034080505371,
+      "learning_rate": 7.090583669848221e-06,
+      "loss": 0.6914,
+      "step": 23451
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.6710081100463867,
+      "learning_rate": 7.090344161618076e-06,
+      "loss": 0.5291,
+      "step": 23452
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.6252684593200684,
+      "learning_rate": 7.090104647575465e-06,
+      "loss": 0.605,
+      "step": 23453
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 23.132139205932617,
+      "learning_rate": 7.089865127721059e-06,
+      "loss": 0.5479,
+      "step": 23454
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.713228940963745,
+      "learning_rate": 7.089625602055519e-06,
+      "loss": 0.6794,
+      "step": 23455
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.7483246326446533,
+      "learning_rate": 7.089386070579513e-06,
+      "loss": 0.6124,
+      "step": 23456
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.496936082839966,
+      "learning_rate": 7.089146533293707e-06,
+      "loss": 0.726,
+      "step": 23457
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.6315228939056396,
+      "learning_rate": 7.0889069901987675e-06,
+      "loss": 0.5538,
+      "step": 23458
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 3.2875499725341797,
+      "learning_rate": 7.088667441295361e-06,
+      "loss": 0.4883,
+      "step": 23459
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.098391532897949,
+      "learning_rate": 7.0884278865841515e-06,
+      "loss": 0.7656,
+      "step": 23460
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.514202117919922,
+      "learning_rate": 7.088188326065807e-06,
+      "loss": 0.6948,
+      "step": 23461
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.8755030632019043,
+      "learning_rate": 7.087948759740993e-06,
+      "loss": 0.5778,
+      "step": 23462
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.858810305595398,
+      "learning_rate": 7.087709187610374e-06,
+      "loss": 0.6425,
+      "step": 23463
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 3.189342498779297,
+      "learning_rate": 7.08746960967462e-06,
+      "loss": 0.6932,
+      "step": 23464
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.1504979133605957,
+      "learning_rate": 7.087230025934394e-06,
+      "loss": 0.5543,
+      "step": 23465
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.807246208190918,
+      "learning_rate": 7.086990436390364e-06,
+      "loss": 0.6907,
+      "step": 23466
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.2472336292266846,
+      "learning_rate": 7.086750841043195e-06,
+      "loss": 0.727,
+      "step": 23467
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.5697877407073975,
+      "learning_rate": 7.086511239893554e-06,
+      "loss": 0.5028,
+      "step": 23468
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.1031172275543213,
+      "learning_rate": 7.0862716329421054e-06,
+      "loss": 0.6811,
+      "step": 23469
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.7197149991989136,
+      "learning_rate": 7.0860320201895195e-06,
+      "loss": 0.554,
+      "step": 23470
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.9571683406829834,
+      "learning_rate": 7.085792401636458e-06,
+      "loss": 0.5939,
+      "step": 23471
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.8786752223968506,
+      "learning_rate": 7.08555277728359e-06,
+      "loss": 0.5075,
+      "step": 23472
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.7084767818450928,
+      "learning_rate": 7.08531314713158e-06,
+      "loss": 0.4811,
+      "step": 23473
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.748682975769043,
+      "learning_rate": 7.085073511181095e-06,
+      "loss": 0.5912,
+      "step": 23474
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.457787275314331,
+      "learning_rate": 7.084833869432804e-06,
+      "loss": 0.8216,
+      "step": 23475
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 3.5155138969421387,
+      "learning_rate": 7.084594221887369e-06,
+      "loss": 0.5571,
+      "step": 23476
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.175151824951172,
+      "learning_rate": 7.0843545685454586e-06,
+      "loss": 0.6067,
+      "step": 23477
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 3.3398234844207764,
+      "learning_rate": 7.08411490940774e-06,
+      "loss": 0.7229,
+      "step": 23478
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.1437268257141113,
+      "learning_rate": 7.083875244474877e-06,
+      "loss": 0.5518,
+      "step": 23479
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.599993109703064,
+      "learning_rate": 7.083635573747539e-06,
+      "loss": 0.3776,
+      "step": 23480
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.5113346576690674,
+      "learning_rate": 7.08339589722639e-06,
+      "loss": 0.5091,
+      "step": 23481
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.5078914165496826,
+      "learning_rate": 7.083156214912098e-06,
+      "loss": 0.7037,
+      "step": 23482
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 4.390704154968262,
+      "learning_rate": 7.082916526805329e-06,
+      "loss": 0.8543,
+      "step": 23483
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.816349744796753,
+      "learning_rate": 7.082676832906749e-06,
+      "loss": 0.6987,
+      "step": 23484
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.985485792160034,
+      "learning_rate": 7.082437133217025e-06,
+      "loss": 0.7323,
+      "step": 23485
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 4.516984939575195,
+      "learning_rate": 7.0821974277368245e-06,
+      "loss": 0.9691,
+      "step": 23486
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 3.873913049697876,
+      "learning_rate": 7.081957716466812e-06,
+      "loss": 0.6006,
+      "step": 23487
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.8435020446777344,
+      "learning_rate": 7.081717999407653e-06,
+      "loss": 0.56,
+      "step": 23488
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.6651958227157593,
+      "learning_rate": 7.081478276560019e-06,
+      "loss": 0.6987,
+      "step": 23489
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.902209520339966,
+      "learning_rate": 7.081238547924574e-06,
+      "loss": 0.7304,
+      "step": 23490
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.825120449066162,
+      "learning_rate": 7.0809988135019825e-06,
+      "loss": 0.4731,
+      "step": 23491
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.6057190895080566,
+      "learning_rate": 7.0807590732929134e-06,
+      "loss": 0.4421,
+      "step": 23492
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.911644697189331,
+      "learning_rate": 7.080519327298033e-06,
+      "loss": 0.759,
+      "step": 23493
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.7009732723236084,
+      "learning_rate": 7.080279575518006e-06,
+      "loss": 0.7663,
+      "step": 23494
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.7183809280395508,
+      "learning_rate": 7.080039817953503e-06,
+      "loss": 0.567,
+      "step": 23495
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 3.1933324337005615,
+      "learning_rate": 7.079800054605188e-06,
+      "loss": 0.9547,
+      "step": 23496
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.057040214538574,
+      "learning_rate": 7.0795602854737275e-06,
+      "loss": 0.6615,
+      "step": 23497
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.9760876893997192,
+      "learning_rate": 7.079320510559791e-06,
+      "loss": 0.4046,
+      "step": 23498
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.5061269998550415,
+      "learning_rate": 7.079080729864041e-06,
+      "loss": 0.5744,
+      "step": 23499
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.7041890621185303,
+      "learning_rate": 7.0788409433871464e-06,
+      "loss": 0.6235,
+      "step": 23500
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.5168358087539673,
+      "learning_rate": 7.078601151129775e-06,
+      "loss": 0.5821,
+      "step": 23501
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.7504717111587524,
+      "learning_rate": 7.078361353092591e-06,
+      "loss": 0.72,
+      "step": 23502
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.913886308670044,
+      "learning_rate": 7.078121549276264e-06,
+      "loss": 0.568,
+      "step": 23503
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.15830135345459,
+      "learning_rate": 7.077881739681458e-06,
+      "loss": 0.6637,
+      "step": 23504
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.8537805080413818,
+      "learning_rate": 7.0776419243088424e-06,
+      "loss": 0.6605,
+      "step": 23505
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.003378391265869,
+      "learning_rate": 7.077402103159082e-06,
+      "loss": 0.6623,
+      "step": 23506
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 3.1260361671447754,
+      "learning_rate": 7.077162276232845e-06,
+      "loss": 1.1163,
+      "step": 23507
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 3.280487060546875,
+      "learning_rate": 7.0769224435307984e-06,
+      "loss": 0.7609,
+      "step": 23508
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.8926278352737427,
+      "learning_rate": 7.0766826050536086e-06,
+      "loss": 0.6107,
+      "step": 23509
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.294883966445923,
+      "learning_rate": 7.076442760801943e-06,
+      "loss": 0.6538,
+      "step": 23510
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.408970594406128,
+      "learning_rate": 7.0762029107764665e-06,
+      "loss": 0.7266,
+      "step": 23511
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.012270212173462,
+      "learning_rate": 7.0759630549778476e-06,
+      "loss": 0.7567,
+      "step": 23512
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.815698266029358,
+      "learning_rate": 7.075723193406754e-06,
+      "loss": 0.6909,
+      "step": 23513
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.079238176345825,
+      "learning_rate": 7.07548332606385e-06,
+      "loss": 0.3636,
+      "step": 23514
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 3.0220694541931152,
+      "learning_rate": 7.075243452949805e-06,
+      "loss": 0.7354,
+      "step": 23515
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.8385519981384277,
+      "learning_rate": 7.075003574065286e-06,
+      "loss": 0.7418,
+      "step": 23516
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.9338047504425049,
+      "learning_rate": 7.0747636894109595e-06,
+      "loss": 0.509,
+      "step": 23517
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.9114887714385986,
+      "learning_rate": 7.074523798987491e-06,
+      "loss": 0.8271,
+      "step": 23518
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 3.8774218559265137,
+      "learning_rate": 7.074283902795552e-06,
+      "loss": 0.6438,
+      "step": 23519
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.8538691997528076,
+      "learning_rate": 7.0740440008358046e-06,
+      "loss": 0.5569,
+      "step": 23520
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.7250572443008423,
+      "learning_rate": 7.073804093108918e-06,
+      "loss": 0.6257,
+      "step": 23521
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.5659748315811157,
+      "learning_rate": 7.0735641796155584e-06,
+      "loss": 0.6883,
+      "step": 23522
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.6530064344406128,
+      "learning_rate": 7.073324260356394e-06,
+      "loss": 0.728,
+      "step": 23523
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.3329315185546875,
+      "learning_rate": 7.073084335332091e-06,
+      "loss": 0.6317,
+      "step": 23524
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.8663980960845947,
+      "learning_rate": 7.072844404543317e-06,
+      "loss": 0.6402,
+      "step": 23525
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.272982358932495,
+      "learning_rate": 7.07260446799074e-06,
+      "loss": 0.6618,
+      "step": 23526
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.4448840618133545,
+      "learning_rate": 7.072364525675027e-06,
+      "loss": 0.5795,
+      "step": 23527
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.3850704431533813,
+      "learning_rate": 7.072124577596844e-06,
+      "loss": 0.4458,
+      "step": 23528
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.0955426692962646,
+      "learning_rate": 7.0718846237568585e-06,
+      "loss": 0.6526,
+      "step": 23529
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.2667713165283203,
+      "learning_rate": 7.071644664155739e-06,
+      "loss": 0.5431,
+      "step": 23530
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.92677640914917,
+      "learning_rate": 7.071404698794151e-06,
+      "loss": 0.7794,
+      "step": 23531
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.2705719470977783,
+      "learning_rate": 7.071164727672762e-06,
+      "loss": 0.7982,
+      "step": 23532
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 3.2638187408447266,
+      "learning_rate": 7.070924750792241e-06,
+      "loss": 0.5658,
+      "step": 23533
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.9516485929489136,
+      "learning_rate": 7.070684768153251e-06,
+      "loss": 0.5867,
+      "step": 23534
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.447479724884033,
+      "learning_rate": 7.070444779756465e-06,
+      "loss": 0.6295,
+      "step": 23535
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.7786056995391846,
+      "learning_rate": 7.070204785602548e-06,
+      "loss": 0.6366,
+      "step": 23536
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.669642448425293,
+      "learning_rate": 7.069964785692166e-06,
+      "loss": 0.8755,
+      "step": 23537
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.919520378112793,
+      "learning_rate": 7.069724780025987e-06,
+      "loss": 0.5328,
+      "step": 23538
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.214888095855713,
+      "learning_rate": 7.0694847686046795e-06,
+      "loss": 0.8229,
+      "step": 23539
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 3.182088613510132,
+      "learning_rate": 7.06924475142891e-06,
+      "loss": 0.7044,
+      "step": 23540
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 3.4985551834106445,
+      "learning_rate": 7.069004728499347e-06,
+      "loss": 0.8972,
+      "step": 23541
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.9596474170684814,
+      "learning_rate": 7.068764699816655e-06,
+      "loss": 0.4877,
+      "step": 23542
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.013390302658081,
+      "learning_rate": 7.068524665381505e-06,
+      "loss": 0.7511,
+      "step": 23543
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.6129205226898193,
+      "learning_rate": 7.06828462519456e-06,
+      "loss": 0.5642,
+      "step": 23544
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.614938497543335,
+      "learning_rate": 7.068044579256493e-06,
+      "loss": 0.5838,
+      "step": 23545
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 3.852632522583008,
+      "learning_rate": 7.067804527567967e-06,
+      "loss": 0.8269,
+      "step": 23546
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.358272075653076,
+      "learning_rate": 7.067564470129653e-06,
+      "loss": 0.5751,
+      "step": 23547
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.1226377487182617,
+      "learning_rate": 7.0673244069422165e-06,
+      "loss": 0.5868,
+      "step": 23548
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.4228241443634033,
+      "learning_rate": 7.067084338006325e-06,
+      "loss": 0.5722,
+      "step": 23549
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.5239428281784058,
+      "learning_rate": 7.066844263322645e-06,
+      "loss": 0.564,
+      "step": 23550
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.9428391456604004,
+      "learning_rate": 7.0666041828918464e-06,
+      "loss": 0.6856,
+      "step": 23551
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.8296973705291748,
+      "learning_rate": 7.066364096714595e-06,
+      "loss": 0.568,
+      "step": 23552
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.648611068725586,
+      "learning_rate": 7.06612400479156e-06,
+      "loss": 0.6817,
+      "step": 23553
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.3723746538162231,
+      "learning_rate": 7.065883907123409e-06,
+      "loss": 0.4628,
+      "step": 23554
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 3.540827751159668,
+      "learning_rate": 7.0656438037108075e-06,
+      "loss": 0.6428,
+      "step": 23555
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.6464335918426514,
+      "learning_rate": 7.065403694554424e-06,
+      "loss": 0.7305,
+      "step": 23556
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.385038375854492,
+      "learning_rate": 7.0651635796549266e-06,
+      "loss": 0.4351,
+      "step": 23557
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.7588605880737305,
+      "learning_rate": 7.064923459012984e-06,
+      "loss": 0.3872,
+      "step": 23558
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.2181222438812256,
+      "learning_rate": 7.064683332629262e-06,
+      "loss": 0.7475,
+      "step": 23559
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.8326442241668701,
+      "learning_rate": 7.06444320050443e-06,
+      "loss": 0.6744,
+      "step": 23560
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.7602654695510864,
+      "learning_rate": 7.064203062639154e-06,
+      "loss": 0.6389,
+      "step": 23561
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 3.864189386367798,
+      "learning_rate": 7.063962919034102e-06,
+      "loss": 0.7683,
+      "step": 23562
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.645299196243286,
+      "learning_rate": 7.063722769689944e-06,
+      "loss": 0.6597,
+      "step": 23563
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.8330461978912354,
+      "learning_rate": 7.063482614607345e-06,
+      "loss": 0.4525,
+      "step": 23564
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 4.68785285949707,
+      "learning_rate": 7.0632424537869725e-06,
+      "loss": 0.726,
+      "step": 23565
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 4.207131862640381,
+      "learning_rate": 7.063002287229498e-06,
+      "loss": 0.6432,
+      "step": 23566
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.4795093536376953,
+      "learning_rate": 7.062762114935586e-06,
+      "loss": 0.5315,
+      "step": 23567
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.198209047317505,
+      "learning_rate": 7.062521936905905e-06,
+      "loss": 0.7164,
+      "step": 23568
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.9891746044158936,
+      "learning_rate": 7.062281753141124e-06,
+      "loss": 0.565,
+      "step": 23569
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.8342660665512085,
+      "learning_rate": 7.062041563641909e-06,
+      "loss": 0.6945,
+      "step": 23570
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 3.509702682495117,
+      "learning_rate": 7.0618013684089295e-06,
+      "loss": 0.4966,
+      "step": 23571
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.788267731666565,
+      "learning_rate": 7.061561167442853e-06,
+      "loss": 0.3935,
+      "step": 23572
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.7653069496154785,
+      "learning_rate": 7.0613209607443466e-06,
+      "loss": 0.549,
+      "step": 23573
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.127697467803955,
+      "learning_rate": 7.061080748314079e-06,
+      "loss": 0.3752,
+      "step": 23574
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 3.0464699268341064,
+      "learning_rate": 7.060840530152717e-06,
+      "loss": 0.6836,
+      "step": 23575
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.02414870262146,
+      "learning_rate": 7.060600306260931e-06,
+      "loss": 0.4148,
+      "step": 23576
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.387406349182129,
+      "learning_rate": 7.060360076639387e-06,
+      "loss": 0.4468,
+      "step": 23577
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.1348628997802734,
+      "learning_rate": 7.0601198412887525e-06,
+      "loss": 0.8047,
+      "step": 23578
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.8504709005355835,
+      "learning_rate": 7.059879600209698e-06,
+      "loss": 0.5494,
+      "step": 23579
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.7013545036315918,
+      "learning_rate": 7.05963935340289e-06,
+      "loss": 0.4415,
+      "step": 23580
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 3.0997531414031982,
+      "learning_rate": 7.059399100868996e-06,
+      "loss": 0.7041,
+      "step": 23581
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.743257999420166,
+      "learning_rate": 7.0591588426086835e-06,
+      "loss": 0.6631,
+      "step": 23582
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.7911932468414307,
+      "learning_rate": 7.058918578622622e-06,
+      "loss": 0.3715,
+      "step": 23583
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 3.3108432292938232,
+      "learning_rate": 7.05867830891148e-06,
+      "loss": 0.695,
+      "step": 23584
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 3.1357266902923584,
+      "learning_rate": 7.0584380334759235e-06,
+      "loss": 0.6943,
+      "step": 23585
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.985772132873535,
+      "learning_rate": 7.058197752316623e-06,
+      "loss": 0.5057,
+      "step": 23586
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.9515198469161987,
+      "learning_rate": 7.057957465434246e-06,
+      "loss": 0.6499,
+      "step": 23587
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.938426971435547,
+      "learning_rate": 7.057717172829459e-06,
+      "loss": 0.5206,
+      "step": 23588
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.236969232559204,
+      "learning_rate": 7.057476874502933e-06,
+      "loss": 0.5957,
+      "step": 23589
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.64546799659729,
+      "learning_rate": 7.057236570455334e-06,
+      "loss": 0.714,
+      "step": 23590
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.191052198410034,
+      "learning_rate": 7.05699626068733e-06,
+      "loss": 0.4745,
+      "step": 23591
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.987768292427063,
+      "learning_rate": 7.05675594519959e-06,
+      "loss": 0.7261,
+      "step": 23592
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.6440869569778442,
+      "learning_rate": 7.056515623992783e-06,
+      "loss": 0.656,
+      "step": 23593
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.3304240703582764,
+      "learning_rate": 7.0562752970675765e-06,
+      "loss": 0.6112,
+      "step": 23594
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.161696672439575,
+      "learning_rate": 7.056034964424637e-06,
+      "loss": 0.7563,
+      "step": 23595
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 3.361480236053467,
+      "learning_rate": 7.055794626064636e-06,
+      "loss": 0.4962,
+      "step": 23596
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.7576450109481812,
+      "learning_rate": 7.05555428198824e-06,
+      "loss": 0.7345,
+      "step": 23597
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.2642922401428223,
+      "learning_rate": 7.055313932196117e-06,
+      "loss": 0.5219,
+      "step": 23598
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.3235538005828857,
+      "learning_rate": 7.055073576688938e-06,
+      "loss": 0.3959,
+      "step": 23599
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.911231279373169,
+      "learning_rate": 7.054833215467367e-06,
+      "loss": 0.8868,
+      "step": 23600
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 3.003286361694336,
+      "learning_rate": 7.054592848532076e-06,
+      "loss": 0.5761,
+      "step": 23601
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.3978755474090576,
+      "learning_rate": 7.05435247588373e-06,
+      "loss": 0.7392,
+      "step": 23602
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.351050853729248,
+      "learning_rate": 7.0541120975230014e-06,
+      "loss": 0.666,
+      "step": 23603
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.1916663646698,
+      "learning_rate": 7.053871713450555e-06,
+      "loss": 0.7591,
+      "step": 23604
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.157621145248413,
+      "learning_rate": 7.05363132366706e-06,
+      "loss": 0.8584,
+      "step": 23605
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.4710514545440674,
+      "learning_rate": 7.053390928173188e-06,
+      "loss": 0.3499,
+      "step": 23606
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.3054819107055664,
+      "learning_rate": 7.0531505269696045e-06,
+      "loss": 0.7199,
+      "step": 23607
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.8221293687820435,
+      "learning_rate": 7.052910120056978e-06,
+      "loss": 0.4189,
+      "step": 23608
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.3158272504806519,
+      "learning_rate": 7.052669707435978e-06,
+      "loss": 0.2571,
+      "step": 23609
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.2888009548187256,
+      "learning_rate": 7.052429289107272e-06,
+      "loss": 0.7489,
+      "step": 23610
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.3605856895446777,
+      "learning_rate": 7.05218886507153e-06,
+      "loss": 0.6668,
+      "step": 23611
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.838818907737732,
+      "learning_rate": 7.051948435329417e-06,
+      "loss": 0.6036,
+      "step": 23612
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.44891619682312,
+      "learning_rate": 7.051707999881605e-06,
+      "loss": 0.7907,
+      "step": 23613
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.9951372146606445,
+      "learning_rate": 7.051467558728763e-06,
+      "loss": 0.6258,
+      "step": 23614
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.3463797569274902,
+      "learning_rate": 7.051227111871558e-06,
+      "loss": 0.6946,
+      "step": 23615
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.969342827796936,
+      "learning_rate": 7.050986659310658e-06,
+      "loss": 0.5533,
+      "step": 23616
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.0321145057678223,
+      "learning_rate": 7.050746201046732e-06,
+      "loss": 0.7135,
+      "step": 23617
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.0987508296966553,
+      "learning_rate": 7.05050573708045e-06,
+      "loss": 0.6197,
+      "step": 23618
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.923649787902832,
+      "learning_rate": 7.050265267412479e-06,
+      "loss": 0.6092,
+      "step": 23619
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.0151047706604004,
+      "learning_rate": 7.05002479204349e-06,
+      "loss": 0.7931,
+      "step": 23620
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 3.437934398651123,
+      "learning_rate": 7.049784310974148e-06,
+      "loss": 0.8271,
+      "step": 23621
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.901315689086914,
+      "learning_rate": 7.049543824205125e-06,
+      "loss": 0.6584,
+      "step": 23622
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.9931319952011108,
+      "learning_rate": 7.049303331737089e-06,
+      "loss": 0.4745,
+      "step": 23623
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.24446177482605,
+      "learning_rate": 7.049062833570706e-06,
+      "loss": 0.7429,
+      "step": 23624
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.5896492004394531,
+      "learning_rate": 7.048822329706648e-06,
+      "loss": 0.5463,
+      "step": 23625
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.4816070795059204,
+      "learning_rate": 7.048581820145582e-06,
+      "loss": 0.6317,
+      "step": 23626
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.1487597227096558,
+      "learning_rate": 7.048341304888177e-06,
+      "loss": 0.4824,
+      "step": 23627
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.882551670074463,
+      "learning_rate": 7.048100783935104e-06,
+      "loss": 0.936,
+      "step": 23628
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.8229873180389404,
+      "learning_rate": 7.047860257287029e-06,
+      "loss": 0.6521,
+      "step": 23629
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.933051109313965,
+      "learning_rate": 7.047619724944622e-06,
+      "loss": 0.7262,
+      "step": 23630
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.127440929412842,
+      "learning_rate": 7.047379186908551e-06,
+      "loss": 0.4979,
+      "step": 23631
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.452791452407837,
+      "learning_rate": 7.047138643179487e-06,
+      "loss": 0.6987,
+      "step": 23632
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.3297271728515625,
+      "learning_rate": 7.046898093758097e-06,
+      "loss": 0.6021,
+      "step": 23633
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.7386608123779297,
+      "learning_rate": 7.046657538645048e-06,
+      "loss": 0.8676,
+      "step": 23634
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.6988856792449951,
+      "learning_rate": 7.046416977841013e-06,
+      "loss": 0.6072,
+      "step": 23635
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.1546733379364014,
+      "learning_rate": 7.046176411346657e-06,
+      "loss": 0.7191,
+      "step": 23636
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.9997707605361938,
+      "learning_rate": 7.045935839162653e-06,
+      "loss": 0.6081,
+      "step": 23637
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.045804977416992,
+      "learning_rate": 7.0456952612896664e-06,
+      "loss": 0.6364,
+      "step": 23638
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.965425729751587,
+      "learning_rate": 7.045454677728369e-06,
+      "loss": 0.5998,
+      "step": 23639
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 3.119037628173828,
+      "learning_rate": 7.0452140884794285e-06,
+      "loss": 0.5694,
+      "step": 23640
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.8451378345489502,
+      "learning_rate": 7.044973493543512e-06,
+      "loss": 0.6901,
+      "step": 23641
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.7707178592681885,
+      "learning_rate": 7.044732892921292e-06,
+      "loss": 0.6235,
+      "step": 23642
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 3.0362889766693115,
+      "learning_rate": 7.044492286613436e-06,
+      "loss": 0.9146,
+      "step": 23643
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.7413620948791504,
+      "learning_rate": 7.044251674620611e-06,
+      "loss": 0.7996,
+      "step": 23644
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.668671131134033,
+      "learning_rate": 7.04401105694349e-06,
+      "loss": 0.4188,
+      "step": 23645
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.6738734245300293,
+      "learning_rate": 7.0437704335827375e-06,
+      "loss": 0.4803,
+      "step": 23646
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.859838843345642,
+      "learning_rate": 7.043529804539027e-06,
+      "loss": 0.5461,
+      "step": 23647
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.180854558944702,
+      "learning_rate": 7.0432891698130245e-06,
+      "loss": 0.5994,
+      "step": 23648
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.8070926666259766,
+      "learning_rate": 7.0430485294054005e-06,
+      "loss": 0.5452,
+      "step": 23649
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.6682336330413818,
+      "learning_rate": 7.042807883316825e-06,
+      "loss": 0.7214,
+      "step": 23650
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.3666576147079468,
+      "learning_rate": 7.0425672315479655e-06,
+      "loss": 0.5555,
+      "step": 23651
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.575263261795044,
+      "learning_rate": 7.04232657409949e-06,
+      "loss": 0.7101,
+      "step": 23652
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.7151153087615967,
+      "learning_rate": 7.042085910972072e-06,
+      "loss": 0.8007,
+      "step": 23653
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.7874534130096436,
+      "learning_rate": 7.041845242166376e-06,
+      "loss": 0.6236,
+      "step": 23654
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 7.049840927124023,
+      "learning_rate": 7.041604567683074e-06,
+      "loss": 0.4216,
+      "step": 23655
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 3.1760945320129395,
+      "learning_rate": 7.041363887522834e-06,
+      "loss": 0.7944,
+      "step": 23656
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.9349817037582397,
+      "learning_rate": 7.041123201686326e-06,
+      "loss": 0.6407,
+      "step": 23657
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.7942105531692505,
+      "learning_rate": 7.04088251017422e-06,
+      "loss": 0.6701,
+      "step": 23658
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.237494707107544,
+      "learning_rate": 7.0406418129871835e-06,
+      "loss": 0.6734,
+      "step": 23659
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.9052386283874512,
+      "learning_rate": 7.040401110125887e-06,
+      "loss": 0.967,
+      "step": 23660
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.59796142578125,
+      "learning_rate": 7.040160401590998e-06,
+      "loss": 0.4918,
+      "step": 23661
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.098163604736328,
+      "learning_rate": 7.03991968738319e-06,
+      "loss": 0.6341,
+      "step": 23662
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.9703274965286255,
+      "learning_rate": 7.0396789675031276e-06,
+      "loss": 0.6655,
+      "step": 23663
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.7212167978286743,
+      "learning_rate": 7.0394382419514814e-06,
+      "loss": 0.509,
+      "step": 23664
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.8216042518615723,
+      "learning_rate": 7.039197510728923e-06,
+      "loss": 0.6397,
+      "step": 23665
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 3.51391339302063,
+      "learning_rate": 7.038956773836119e-06,
+      "loss": 0.6741,
+      "step": 23666
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.8393101692199707,
+      "learning_rate": 7.03871603127374e-06,
+      "loss": 0.4958,
+      "step": 23667
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 4.068389892578125,
+      "learning_rate": 7.0384752830424565e-06,
+      "loss": 0.6961,
+      "step": 23668
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 3.186141014099121,
+      "learning_rate": 7.038234529142935e-06,
+      "loss": 0.6037,
+      "step": 23669
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.974827766418457,
+      "learning_rate": 7.0379937695758485e-06,
+      "loss": 0.6253,
+      "step": 23670
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 3.3161473274230957,
+      "learning_rate": 7.037753004341865e-06,
+      "loss": 0.67,
+      "step": 23671
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 3.466601610183716,
+      "learning_rate": 7.037512233441652e-06,
+      "loss": 0.6005,
+      "step": 23672
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.3006093502044678,
+      "learning_rate": 7.037271456875881e-06,
+      "loss": 0.5114,
+      "step": 23673
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 3.036454439163208,
+      "learning_rate": 7.037030674645223e-06,
+      "loss": 0.7144,
+      "step": 23674
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.7034403085708618,
+      "learning_rate": 7.036789886750345e-06,
+      "loss": 0.4956,
+      "step": 23675
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.2995171546936035,
+      "learning_rate": 7.036549093191918e-06,
+      "loss": 0.8195,
+      "step": 23676
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.192901849746704,
+      "learning_rate": 7.036308293970609e-06,
+      "loss": 0.4715,
+      "step": 23677
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.9712156057357788,
+      "learning_rate": 7.036067489087091e-06,
+      "loss": 0.7063,
+      "step": 23678
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.6516412496566772,
+      "learning_rate": 7.03582667854203e-06,
+      "loss": 0.5205,
+      "step": 23679
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.8819658756256104,
+      "learning_rate": 7.035585862336101e-06,
+      "loss": 0.6004,
+      "step": 23680
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.311305046081543,
+      "learning_rate": 7.035345040469968e-06,
+      "loss": 0.5906,
+      "step": 23681
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.8676837682724,
+      "learning_rate": 7.035104212944304e-06,
+      "loss": 0.6029,
+      "step": 23682
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.301882743835449,
+      "learning_rate": 7.034863379759776e-06,
+      "loss": 0.7332,
+      "step": 23683
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.4705305099487305,
+      "learning_rate": 7.034622540917057e-06,
+      "loss": 0.6229,
+      "step": 23684
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.5424396991729736,
+      "learning_rate": 7.034381696416814e-06,
+      "loss": 0.5821,
+      "step": 23685
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.8998087644577026,
+      "learning_rate": 7.0341408462597175e-06,
+      "loss": 0.471,
+      "step": 23686
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.3314051628112793,
+      "learning_rate": 7.033899990446437e-06,
+      "loss": 0.51,
+      "step": 23687
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.4607255458831787,
+      "learning_rate": 7.033659128977642e-06,
+      "loss": 0.6007,
+      "step": 23688
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.9503440856933594,
+      "learning_rate": 7.033418261854004e-06,
+      "loss": 0.7185,
+      "step": 23689
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.7342894077301025,
+      "learning_rate": 7.033177389076192e-06,
+      "loss": 0.4658,
+      "step": 23690
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.087099552154541,
+      "learning_rate": 7.032936510644875e-06,
+      "loss": 0.8703,
+      "step": 23691
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.03425931930542,
+      "learning_rate": 7.032695626560723e-06,
+      "loss": 0.6901,
+      "step": 23692
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.3098254203796387,
+      "learning_rate": 7.0324547368244054e-06,
+      "loss": 0.6535,
+      "step": 23693
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 4.877508163452148,
+      "learning_rate": 7.032213841436593e-06,
+      "loss": 0.7697,
+      "step": 23694
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.3684844970703125,
+      "learning_rate": 7.031972940397955e-06,
+      "loss": 0.7064,
+      "step": 23695
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.2802865505218506,
+      "learning_rate": 7.031732033709162e-06,
+      "loss": 0.4689,
+      "step": 23696
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.396404981613159,
+      "learning_rate": 7.031491121370882e-06,
+      "loss": 0.7198,
+      "step": 23697
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.3935580253601074,
+      "learning_rate": 7.031250203383787e-06,
+      "loss": 0.4576,
+      "step": 23698
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.465775728225708,
+      "learning_rate": 7.031009279748546e-06,
+      "loss": 0.4274,
+      "step": 23699
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.0815155506134033,
+      "learning_rate": 7.030768350465829e-06,
+      "loss": 0.5061,
+      "step": 23700
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.623763084411621,
+      "learning_rate": 7.030527415536306e-06,
+      "loss": 0.7799,
+      "step": 23701
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.3997321128845215,
+      "learning_rate": 7.030286474960648e-06,
+      "loss": 0.6932,
+      "step": 23702
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.5293710231781006,
+      "learning_rate": 7.030045528739523e-06,
+      "loss": 0.3472,
+      "step": 23703
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.0324878692626953,
+      "learning_rate": 7.029804576873602e-06,
+      "loss": 0.8071,
+      "step": 23704
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.4143428802490234,
+      "learning_rate": 7.029563619363555e-06,
+      "loss": 0.8669,
+      "step": 23705
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.0491673946380615,
+      "learning_rate": 7.029322656210051e-06,
+      "loss": 0.7602,
+      "step": 23706
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.6888508796691895,
+      "learning_rate": 7.0290816874137615e-06,
+      "loss": 0.8269,
+      "step": 23707
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.001094102859497,
+      "learning_rate": 7.028840712975356e-06,
+      "loss": 0.6732,
+      "step": 23708
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.8319664001464844,
+      "learning_rate": 7.028599732895504e-06,
+      "loss": 0.9469,
+      "step": 23709
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.10512113571167,
+      "learning_rate": 7.028358747174876e-06,
+      "loss": 0.5829,
+      "step": 23710
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 3.3394596576690674,
+      "learning_rate": 7.028117755814143e-06,
+      "loss": 0.7108,
+      "step": 23711
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.503692388534546,
+      "learning_rate": 7.027876758813974e-06,
+      "loss": 0.5298,
+      "step": 23712
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.6977362632751465,
+      "learning_rate": 7.027635756175039e-06,
+      "loss": 0.74,
+      "step": 23713
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.121720552444458,
+      "learning_rate": 7.027394747898009e-06,
+      "loss": 0.6738,
+      "step": 23714
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.0543079376220703,
+      "learning_rate": 7.0271537339835535e-06,
+      "loss": 0.5959,
+      "step": 23715
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.0780296325683594,
+      "learning_rate": 7.026912714432343e-06,
+      "loss": 0.612,
+      "step": 23716
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.593139171600342,
+      "learning_rate": 7.026671689245047e-06,
+      "loss": 0.5414,
+      "step": 23717
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.7893726825714111,
+      "learning_rate": 7.026430658422336e-06,
+      "loss": 0.6142,
+      "step": 23718
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.5103251934051514,
+      "learning_rate": 7.026189621964882e-06,
+      "loss": 0.7124,
+      "step": 23719
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.542447805404663,
+      "learning_rate": 7.025948579873352e-06,
+      "loss": 0.7239,
+      "step": 23720
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.900020718574524,
+      "learning_rate": 7.025707532148419e-06,
+      "loss": 0.5615,
+      "step": 23721
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.3738462924957275,
+      "learning_rate": 7.025466478790752e-06,
+      "loss": 0.4645,
+      "step": 23722
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 3.8158435821533203,
+      "learning_rate": 7.025225419801021e-06,
+      "loss": 0.7051,
+      "step": 23723
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.0812084674835205,
+      "learning_rate": 7.024984355179898e-06,
+      "loss": 0.4998,
+      "step": 23724
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.9794375896453857,
+      "learning_rate": 7.0247432849280505e-06,
+      "loss": 0.9025,
+      "step": 23725
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.883358955383301,
+      "learning_rate": 7.024502209046152e-06,
+      "loss": 0.6648,
+      "step": 23726
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.5733108520507812,
+      "learning_rate": 7.024261127534869e-06,
+      "loss": 0.3334,
+      "step": 23727
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 4.750999927520752,
+      "learning_rate": 7.0240200403948764e-06,
+      "loss": 0.5723,
+      "step": 23728
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.770638942718506,
+      "learning_rate": 7.023778947626841e-06,
+      "loss": 0.7224,
+      "step": 23729
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.8468080759048462,
+      "learning_rate": 7.023537849231434e-06,
+      "loss": 0.6486,
+      "step": 23730
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.013715982437134,
+      "learning_rate": 7.023296745209328e-06,
+      "loss": 0.3615,
+      "step": 23731
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 3.5771195888519287,
+      "learning_rate": 7.023055635561191e-06,
+      "loss": 0.4633,
+      "step": 23732
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 3.3012142181396484,
+      "learning_rate": 7.022814520287694e-06,
+      "loss": 0.6172,
+      "step": 23733
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.416987895965576,
+      "learning_rate": 7.022573399389507e-06,
+      "loss": 0.705,
+      "step": 23734
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.826401948928833,
+      "learning_rate": 7.0223322728673014e-06,
+      "loss": 0.5116,
+      "step": 23735
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 3.7427096366882324,
+      "learning_rate": 7.022091140721747e-06,
+      "loss": 0.8352,
+      "step": 23736
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 6.840424537658691,
+      "learning_rate": 7.021850002953516e-06,
+      "loss": 0.8141,
+      "step": 23737
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.325981855392456,
+      "learning_rate": 7.021608859563277e-06,
+      "loss": 0.9256,
+      "step": 23738
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.7155330181121826,
+      "learning_rate": 7.021367710551701e-06,
+      "loss": 0.7235,
+      "step": 23739
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.1607283353805542,
+      "learning_rate": 7.021126555919459e-06,
+      "loss": 0.4199,
+      "step": 23740
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.0589241981506348,
+      "learning_rate": 7.0208853956672185e-06,
+      "loss": 0.7205,
+      "step": 23741
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 3.090251922607422,
+      "learning_rate": 7.020644229795656e-06,
+      "loss": 0.524,
+      "step": 23742
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 3.9083619117736816,
+      "learning_rate": 7.020403058305437e-06,
+      "loss": 0.6636,
+      "step": 23743
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.3948445320129395,
+      "learning_rate": 7.020161881197235e-06,
+      "loss": 0.6446,
+      "step": 23744
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.156224489212036,
+      "learning_rate": 7.019920698471719e-06,
+      "loss": 0.5674,
+      "step": 23745
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.3471405506134033,
+      "learning_rate": 7.01967951012956e-06,
+      "loss": 0.5357,
+      "step": 23746
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.4344611167907715,
+      "learning_rate": 7.01943831617143e-06,
+      "loss": 0.7025,
+      "step": 23747
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.2364561557769775,
+      "learning_rate": 7.0191971165979966e-06,
+      "loss": 0.6746,
+      "step": 23748
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.0793263912200928,
+      "learning_rate": 7.018955911409933e-06,
+      "loss": 0.8048,
+      "step": 23749
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 3.777608633041382,
+      "learning_rate": 7.018714700607911e-06,
+      "loss": 0.4309,
+      "step": 23750
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 3.017949104309082,
+      "learning_rate": 7.018473484192598e-06,
+      "loss": 0.6836,
+      "step": 23751
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.802962064743042,
+      "learning_rate": 7.018232262164668e-06,
+      "loss": 0.7281,
+      "step": 23752
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.586969017982483,
+      "learning_rate": 7.017991034524789e-06,
+      "loss": 0.5529,
+      "step": 23753
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.091341733932495,
+      "learning_rate": 7.0177498012736334e-06,
+      "loss": 0.7877,
+      "step": 23754
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.8465756177902222,
+      "learning_rate": 7.017508562411871e-06,
+      "loss": 0.6932,
+      "step": 23755
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.58441424369812,
+      "learning_rate": 7.017267317940173e-06,
+      "loss": 0.653,
+      "step": 23756
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.1986143589019775,
+      "learning_rate": 7.0170260678592104e-06,
+      "loss": 0.7605,
+      "step": 23757
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.290405511856079,
+      "learning_rate": 7.016784812169654e-06,
+      "loss": 0.7501,
+      "step": 23758
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.893700122833252,
+      "learning_rate": 7.016543550872175e-06,
+      "loss": 0.5571,
+      "step": 23759
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.4653737545013428,
+      "learning_rate": 7.016302283967442e-06,
+      "loss": 0.6011,
+      "step": 23760
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 4.459075450897217,
+      "learning_rate": 7.01606101145613e-06,
+      "loss": 0.6031,
+      "step": 23761
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.809389591217041,
+      "learning_rate": 7.0158197333389065e-06,
+      "loss": 0.6088,
+      "step": 23762
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.777589797973633,
+      "learning_rate": 7.015578449616444e-06,
+      "loss": 0.6695,
+      "step": 23763
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.9696052074432373,
+      "learning_rate": 7.015337160289411e-06,
+      "loss": 0.5211,
+      "step": 23764
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 3.6648406982421875,
+      "learning_rate": 7.015095865358482e-06,
+      "loss": 0.8463,
+      "step": 23765
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.1889162063598633,
+      "learning_rate": 7.014854564824327e-06,
+      "loss": 0.502,
+      "step": 23766
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.8963531255722046,
+      "learning_rate": 7.014613258687615e-06,
+      "loss": 0.5227,
+      "step": 23767
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.3158257007598877,
+      "learning_rate": 7.014371946949017e-06,
+      "loss": 0.6943,
+      "step": 23768
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.837951183319092,
+      "learning_rate": 7.014130629609207e-06,
+      "loss": 0.8469,
+      "step": 23769
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.748727560043335,
+      "learning_rate": 7.013889306668854e-06,
+      "loss": 0.8516,
+      "step": 23770
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.9525787830352783,
+      "learning_rate": 7.013647978128629e-06,
+      "loss": 0.7931,
+      "step": 23771
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.400479793548584,
+      "learning_rate": 7.013406643989203e-06,
+      "loss": 0.6186,
+      "step": 23772
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.6941555738449097,
+      "learning_rate": 7.013165304251248e-06,
+      "loss": 0.7771,
+      "step": 23773
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.226649045944214,
+      "learning_rate": 7.012923958915435e-06,
+      "loss": 0.699,
+      "step": 23774
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 3.5061821937561035,
+      "learning_rate": 7.012682607982434e-06,
+      "loss": 0.7281,
+      "step": 23775
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.348716974258423,
+      "learning_rate": 7.012441251452916e-06,
+      "loss": 0.6804,
+      "step": 23776
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.218143939971924,
+      "learning_rate": 7.012199889327553e-06,
+      "loss": 0.5884,
+      "step": 23777
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.6189311742782593,
+      "learning_rate": 7.0119585216070166e-06,
+      "loss": 0.8365,
+      "step": 23778
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.4958550930023193,
+      "learning_rate": 7.011717148291976e-06,
+      "loss": 0.7825,
+      "step": 23779
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.9152408838272095,
+      "learning_rate": 7.011475769383105e-06,
+      "loss": 0.4478,
+      "step": 23780
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.0960066318511963,
+      "learning_rate": 7.011234384881072e-06,
+      "loss": 0.5295,
+      "step": 23781
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.768978238105774,
+      "learning_rate": 7.010992994786551e-06,
+      "loss": 0.4539,
+      "step": 23782
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.7541847229003906,
+      "learning_rate": 7.010751599100211e-06,
+      "loss": 0.465,
+      "step": 23783
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.480354070663452,
+      "learning_rate": 7.0105101978227245e-06,
+      "loss": 0.7111,
+      "step": 23784
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.391226291656494,
+      "learning_rate": 7.010268790954762e-06,
+      "loss": 0.5606,
+      "step": 23785
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.196659564971924,
+      "learning_rate": 7.0100273784969955e-06,
+      "loss": 0.5994,
+      "step": 23786
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.277092218399048,
+      "learning_rate": 7.009785960450095e-06,
+      "loss": 0.5233,
+      "step": 23787
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 3.280996799468994,
+      "learning_rate": 7.0095445368147315e-06,
+      "loss": 0.8918,
+      "step": 23788
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.7632819414138794,
+      "learning_rate": 7.009303107591578e-06,
+      "loss": 0.8075,
+      "step": 23789
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.2908613681793213,
+      "learning_rate": 7.009061672781308e-06,
+      "loss": 0.8298,
+      "step": 23790
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.7796610593795776,
+      "learning_rate": 7.008820232384588e-06,
+      "loss": 0.6695,
+      "step": 23791
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.3738677501678467,
+      "learning_rate": 7.008578786402092e-06,
+      "loss": 0.653,
+      "step": 23792
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.4435503482818604,
+      "learning_rate": 7.00833733483449e-06,
+      "loss": 0.8101,
+      "step": 23793
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.947300910949707,
+      "learning_rate": 7.0080958776824545e-06,
+      "loss": 0.7398,
+      "step": 23794
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 3.3917880058288574,
+      "learning_rate": 7.007854414946658e-06,
+      "loss": 0.9109,
+      "step": 23795
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.6942365169525146,
+      "learning_rate": 7.007612946627769e-06,
+      "loss": 0.6618,
+      "step": 23796
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.4899170398712158,
+      "learning_rate": 7.007371472726459e-06,
+      "loss": 0.7751,
+      "step": 23797
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.9578396081924438,
+      "learning_rate": 7.007129993243404e-06,
+      "loss": 0.556,
+      "step": 23798
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 3.645447254180908,
+      "learning_rate": 7.006888508179271e-06,
+      "loss": 0.8162,
+      "step": 23799
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 3.113738536834717,
+      "learning_rate": 7.006647017534733e-06,
+      "loss": 0.6167,
+      "step": 23800
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.5207034349441528,
+      "learning_rate": 7.006405521310462e-06,
+      "loss": 0.4453,
+      "step": 23801
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.830176830291748,
+      "learning_rate": 7.006164019507128e-06,
+      "loss": 0.5556,
+      "step": 23802
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.7927980422973633,
+      "learning_rate": 7.005922512125403e-06,
+      "loss": 0.778,
+      "step": 23803
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.9223588705062866,
+      "learning_rate": 7.005680999165959e-06,
+      "loss": 0.6015,
+      "step": 23804
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 3.4554786682128906,
+      "learning_rate": 7.005439480629468e-06,
+      "loss": 0.6793,
+      "step": 23805
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 3.5953946113586426,
+      "learning_rate": 7.005197956516602e-06,
+      "loss": 0.9215,
+      "step": 23806
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 4.4624433517456055,
+      "learning_rate": 7.004956426828031e-06,
+      "loss": 0.7424,
+      "step": 23807
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.8154730796813965,
+      "learning_rate": 7.004714891564427e-06,
+      "loss": 0.8056,
+      "step": 23808
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.448498010635376,
+      "learning_rate": 7.004473350726461e-06,
+      "loss": 0.5172,
+      "step": 23809
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.665095329284668,
+      "learning_rate": 7.004231804314806e-06,
+      "loss": 0.7598,
+      "step": 23810
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.9343347549438477,
+      "learning_rate": 7.003990252330134e-06,
+      "loss": 0.5252,
+      "step": 23811
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.942058563232422,
+      "learning_rate": 7.003748694773115e-06,
+      "loss": 0.7427,
+      "step": 23812
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.02236008644104,
+      "learning_rate": 7.003507131644422e-06,
+      "loss": 0.6172,
+      "step": 23813
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.161285161972046,
+      "learning_rate": 7.003265562944727e-06,
+      "loss": 0.7104,
+      "step": 23814
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.9083970785140991,
+      "learning_rate": 7.0030239886747e-06,
+      "loss": 0.8087,
+      "step": 23815
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.5144168138504028,
+      "learning_rate": 7.0027824088350125e-06,
+      "loss": 0.8132,
+      "step": 23816
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.5993075370788574,
+      "learning_rate": 7.0025408234263395e-06,
+      "loss": 0.4402,
+      "step": 23817
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.4343667030334473,
+      "learning_rate": 7.00229923244935e-06,
+      "loss": 0.5014,
+      "step": 23818
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.0813920497894287,
+      "learning_rate": 7.002057635904714e-06,
+      "loss": 0.8266,
+      "step": 23819
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.937685012817383,
+      "learning_rate": 7.001816033793109e-06,
+      "loss": 0.7197,
+      "step": 23820
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.30340838432312,
+      "learning_rate": 7.001574426115202e-06,
+      "loss": 0.3949,
+      "step": 23821
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 3.4134552478790283,
+      "learning_rate": 7.001332812871667e-06,
+      "loss": 0.5018,
+      "step": 23822
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.8085553646087646,
+      "learning_rate": 7.001091194063174e-06,
+      "loss": 0.54,
+      "step": 23823
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.334122896194458,
+      "learning_rate": 7.000849569690397e-06,
+      "loss": 0.8129,
+      "step": 23824
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.2749335765838623,
+      "learning_rate": 7.000607939754007e-06,
+      "loss": 0.6836,
+      "step": 23825
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.790670156478882,
+      "learning_rate": 7.000366304254675e-06,
+      "loss": 0.5709,
+      "step": 23826
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.142627239227295,
+      "learning_rate": 7.000124663193074e-06,
+      "loss": 0.7227,
+      "step": 23827
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.9548040628433228,
+      "learning_rate": 6.999883016569876e-06,
+      "loss": 0.8502,
+      "step": 23828
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.3298637866973877,
+      "learning_rate": 6.999641364385751e-06,
+      "loss": 0.5644,
+      "step": 23829
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.5458213090896606,
+      "learning_rate": 6.999399706641374e-06,
+      "loss": 0.5743,
+      "step": 23830
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.7075755596160889,
+      "learning_rate": 6.999158043337415e-06,
+      "loss": 0.6253,
+      "step": 23831
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 3.1360912322998047,
+      "learning_rate": 6.998916374474547e-06,
+      "loss": 0.6572,
+      "step": 23832
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.524568557739258,
+      "learning_rate": 6.998674700053441e-06,
+      "loss": 0.5912,
+      "step": 23833
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.3878538608551025,
+      "learning_rate": 6.998433020074768e-06,
+      "loss": 0.7231,
+      "step": 23834
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.1075258255004883,
+      "learning_rate": 6.998191334539203e-06,
+      "loss": 0.5736,
+      "step": 23835
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.8837385177612305,
+      "learning_rate": 6.997949643447416e-06,
+      "loss": 0.5161,
+      "step": 23836
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.08052134513855,
+      "learning_rate": 6.99770794680008e-06,
+      "loss": 0.7446,
+      "step": 23837
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.313338041305542,
+      "learning_rate": 6.997466244597866e-06,
+      "loss": 0.8003,
+      "step": 23838
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.5966242551803589,
+      "learning_rate": 6.997224536841446e-06,
+      "loss": 0.425,
+      "step": 23839
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.0451042652130127,
+      "learning_rate": 6.996982823531493e-06,
+      "loss": 0.4789,
+      "step": 23840
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.5975213050842285,
+      "learning_rate": 6.9967411046686795e-06,
+      "loss": 0.7279,
+      "step": 23841
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.5038648843765259,
+      "learning_rate": 6.996499380253676e-06,
+      "loss": 0.4438,
+      "step": 23842
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.5477278232574463,
+      "learning_rate": 6.9962576502871575e-06,
+      "loss": 0.6921,
+      "step": 23843
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.6489572525024414,
+      "learning_rate": 6.996015914769792e-06,
+      "loss": 0.6662,
+      "step": 23844
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.185171365737915,
+      "learning_rate": 6.995774173702257e-06,
+      "loss": 0.6819,
+      "step": 23845
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.7544196844100952,
+      "learning_rate": 6.9955324270852185e-06,
+      "loss": 0.5489,
+      "step": 23846
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 3.0317280292510986,
+      "learning_rate": 6.995290674919354e-06,
+      "loss": 0.5461,
+      "step": 23847
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 3.2871317863464355,
+      "learning_rate": 6.995048917205333e-06,
+      "loss": 0.6105,
+      "step": 23848
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.8433473110198975,
+      "learning_rate": 6.994807153943826e-06,
+      "loss": 0.6551,
+      "step": 23849
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.6675686836242676,
+      "learning_rate": 6.99456538513551e-06,
+      "loss": 0.5807,
+      "step": 23850
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.963857889175415,
+      "learning_rate": 6.994323610781055e-06,
+      "loss": 0.6606,
+      "step": 23851
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.7657743692398071,
+      "learning_rate": 6.994081830881133e-06,
+      "loss": 0.6372,
+      "step": 23852
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.478905200958252,
+      "learning_rate": 6.993840045436415e-06,
+      "loss": 0.6552,
+      "step": 23853
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 4.141197204589844,
+      "learning_rate": 6.9935982544475765e-06,
+      "loss": 0.8345,
+      "step": 23854
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.232187032699585,
+      "learning_rate": 6.993356457915288e-06,
+      "loss": 0.7559,
+      "step": 23855
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 3.6062662601470947,
+      "learning_rate": 6.993114655840221e-06,
+      "loss": 0.6822,
+      "step": 23856
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.911862850189209,
+      "learning_rate": 6.992872848223049e-06,
+      "loss": 0.5302,
+      "step": 23857
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.770079255104065,
+      "learning_rate": 6.992631035064443e-06,
+      "loss": 0.9473,
+      "step": 23858
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.8072025775909424,
+      "learning_rate": 6.992389216365079e-06,
+      "loss": 0.7496,
+      "step": 23859
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.9854035377502441,
+      "learning_rate": 6.9921473921256255e-06,
+      "loss": 0.9454,
+      "step": 23860
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.73183536529541,
+      "learning_rate": 6.991905562346757e-06,
+      "loss": 0.6457,
+      "step": 23861
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.8254494667053223,
+      "learning_rate": 6.9916637270291444e-06,
+      "loss": 0.5673,
+      "step": 23862
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 3.3800606727600098,
+      "learning_rate": 6.991421886173462e-06,
+      "loss": 0.787,
+      "step": 23863
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.5607216358184814,
+      "learning_rate": 6.99118003978038e-06,
+      "loss": 0.895,
+      "step": 23864
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.9568769931793213,
+      "learning_rate": 6.9909381878505735e-06,
+      "loss": 0.5827,
+      "step": 23865
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.948270559310913,
+      "learning_rate": 6.990696330384713e-06,
+      "loss": 0.714,
+      "step": 23866
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.894009590148926,
+      "learning_rate": 6.990454467383472e-06,
+      "loss": 0.6954,
+      "step": 23867
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.3888683319091797,
+      "learning_rate": 6.990212598847523e-06,
+      "loss": 0.8401,
+      "step": 23868
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.5325332880020142,
+      "learning_rate": 6.989970724777538e-06,
+      "loss": 0.7332,
+      "step": 23869
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.2091028690338135,
+      "learning_rate": 6.9897288451741895e-06,
+      "loss": 0.7464,
+      "step": 23870
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 3.0040924549102783,
+      "learning_rate": 6.989486960038151e-06,
+      "loss": 0.7619,
+      "step": 23871
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.0044519901275635,
+      "learning_rate": 6.989245069370094e-06,
+      "loss": 0.6654,
+      "step": 23872
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 3.375568151473999,
+      "learning_rate": 6.989003173170693e-06,
+      "loss": 0.586,
+      "step": 23873
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 4.622779846191406,
+      "learning_rate": 6.988761271440619e-06,
+      "loss": 0.4838,
+      "step": 23874
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.439157247543335,
+      "learning_rate": 6.988519364180544e-06,
+      "loss": 0.8479,
+      "step": 23875
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 3.5022666454315186,
+      "learning_rate": 6.988277451391142e-06,
+      "loss": 0.5416,
+      "step": 23876
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.7266844511032104,
+      "learning_rate": 6.988035533073085e-06,
+      "loss": 0.6702,
+      "step": 23877
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 3.028578758239746,
+      "learning_rate": 6.987793609227046e-06,
+      "loss": 0.6073,
+      "step": 23878
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.6789404153823853,
+      "learning_rate": 6.987551679853698e-06,
+      "loss": 0.7428,
+      "step": 23879
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 3.577557325363159,
+      "learning_rate": 6.987309744953712e-06,
+      "loss": 0.5246,
+      "step": 23880
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.743198275566101,
+      "learning_rate": 6.987067804527763e-06,
+      "loss": 0.9357,
+      "step": 23881
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.4660404920578003,
+      "learning_rate": 6.986825858576524e-06,
+      "loss": 0.5265,
+      "step": 23882
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.977392315864563,
+      "learning_rate": 6.986583907100665e-06,
+      "loss": 0.7524,
+      "step": 23883
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.5117766857147217,
+      "learning_rate": 6.986341950100861e-06,
+      "loss": 0.5936,
+      "step": 23884
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.2355265617370605,
+      "learning_rate": 6.986099987577783e-06,
+      "loss": 0.7132,
+      "step": 23885
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 3.720367431640625,
+      "learning_rate": 6.985858019532107e-06,
+      "loss": 0.8699,
+      "step": 23886
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.720358967781067,
+      "learning_rate": 6.985616045964503e-06,
+      "loss": 0.9118,
+      "step": 23887
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.8389309644699097,
+      "learning_rate": 6.985374066875644e-06,
+      "loss": 0.7961,
+      "step": 23888
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.5940310955047607,
+      "learning_rate": 6.985132082266204e-06,
+      "loss": 0.4239,
+      "step": 23889
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 3.089404344558716,
+      "learning_rate": 6.984890092136853e-06,
+      "loss": 0.7068,
+      "step": 23890
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.712184429168701,
+      "learning_rate": 6.984648096488269e-06,
+      "loss": 0.7048,
+      "step": 23891
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.9298744201660156,
+      "learning_rate": 6.984406095321121e-06,
+      "loss": 0.6721,
+      "step": 23892
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 3.5991339683532715,
+      "learning_rate": 6.9841640886360855e-06,
+      "loss": 0.4895,
+      "step": 23893
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.457425355911255,
+      "learning_rate": 6.98392207643383e-06,
+      "loss": 0.6558,
+      "step": 23894
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.7240252494812012,
+      "learning_rate": 6.983680058715031e-06,
+      "loss": 0.6162,
+      "step": 23895
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.8429224491119385,
+      "learning_rate": 6.983438035480361e-06,
+      "loss": 0.6312,
+      "step": 23896
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.2065865993499756,
+      "learning_rate": 6.9831960067304935e-06,
+      "loss": 0.6332,
+      "step": 23897
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.1527864933013916,
+      "learning_rate": 6.9829539724660996e-06,
+      "loss": 0.8354,
+      "step": 23898
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.5734812021255493,
+      "learning_rate": 6.982711932687856e-06,
+      "loss": 0.7523,
+      "step": 23899
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.9937121868133545,
+      "learning_rate": 6.98246988739643e-06,
+      "loss": 0.4299,
+      "step": 23900
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.4392597675323486,
+      "learning_rate": 6.982227836592499e-06,
+      "loss": 0.7947,
+      "step": 23901
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 3.023366928100586,
+      "learning_rate": 6.981985780276736e-06,
+      "loss": 0.7831,
+      "step": 23902
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.5871145725250244,
+      "learning_rate": 6.981743718449812e-06,
+      "loss": 0.4974,
+      "step": 23903
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.943439483642578,
+      "learning_rate": 6.9815016511124014e-06,
+      "loss": 0.5389,
+      "step": 23904
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.4804149866104126,
+      "learning_rate": 6.981259578265177e-06,
+      "loss": 0.6931,
+      "step": 23905
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.4468499422073364,
+      "learning_rate": 6.981017499908812e-06,
+      "loss": 0.6499,
+      "step": 23906
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.394965410232544,
+      "learning_rate": 6.98077541604398e-06,
+      "loss": 0.649,
+      "step": 23907
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.8177409172058105,
+      "learning_rate": 6.980533326671353e-06,
+      "loss": 0.8201,
+      "step": 23908
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.166267156600952,
+      "learning_rate": 6.980291231791605e-06,
+      "loss": 0.7556,
+      "step": 23909
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.6052396297454834,
+      "learning_rate": 6.9800491314054074e-06,
+      "loss": 0.7156,
+      "step": 23910
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.1440579891204834,
+      "learning_rate": 6.979807025513437e-06,
+      "loss": 0.7031,
+      "step": 23911
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.7275152206420898,
+      "learning_rate": 6.979564914116363e-06,
+      "loss": 0.5933,
+      "step": 23912
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 3.526348829269409,
+      "learning_rate": 6.9793227972148624e-06,
+      "loss": 0.4839,
+      "step": 23913
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.9029945135116577,
+      "learning_rate": 6.979080674809605e-06,
+      "loss": 0.5072,
+      "step": 23914
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.4497032165527344,
+      "learning_rate": 6.978838546901267e-06,
+      "loss": 0.7877,
+      "step": 23915
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.4603190422058105,
+      "learning_rate": 6.978596413490519e-06,
+      "loss": 0.4844,
+      "step": 23916
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.5407674312591553,
+      "learning_rate": 6.978354274578036e-06,
+      "loss": 0.6713,
+      "step": 23917
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.621803641319275,
+      "learning_rate": 6.978112130164492e-06,
+      "loss": 0.8269,
+      "step": 23918
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.8187774419784546,
+      "learning_rate": 6.977869980250558e-06,
+      "loss": 0.7817,
+      "step": 23919
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.1630282402038574,
+      "learning_rate": 6.9776278248369076e-06,
+      "loss": 0.7527,
+      "step": 23920
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.2181758880615234,
+      "learning_rate": 6.977385663924217e-06,
+      "loss": 0.6945,
+      "step": 23921
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.86189341545105,
+      "learning_rate": 6.9771434975131565e-06,
+      "loss": 0.5429,
+      "step": 23922
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.8419643640518188,
+      "learning_rate": 6.976901325604402e-06,
+      "loss": 0.6551,
+      "step": 23923
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.148797035217285,
+      "learning_rate": 6.976659148198624e-06,
+      "loss": 0.375,
+      "step": 23924
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.8017406463623047,
+      "learning_rate": 6.976416965296497e-06,
+      "loss": 0.5299,
+      "step": 23925
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.779670000076294,
+      "learning_rate": 6.976174776898695e-06,
+      "loss": 0.8356,
+      "step": 23926
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.7421650886535645,
+      "learning_rate": 6.975932583005893e-06,
+      "loss": 0.5462,
+      "step": 23927
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.307939291000366,
+      "learning_rate": 6.975690383618761e-06,
+      "loss": 0.6463,
+      "step": 23928
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.8716421127319336,
+      "learning_rate": 6.975448178737976e-06,
+      "loss": 0.5496,
+      "step": 23929
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.8422656059265137,
+      "learning_rate": 6.9752059683642074e-06,
+      "loss": 0.7377,
+      "step": 23930
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 3.255725383758545,
+      "learning_rate": 6.9749637524981315e-06,
+      "loss": 0.7364,
+      "step": 23931
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.3796281814575195,
+      "learning_rate": 6.974721531140422e-06,
+      "loss": 0.5729,
+      "step": 23932
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.9831923246383667,
+      "learning_rate": 6.9744793042917505e-06,
+      "loss": 0.4744,
+      "step": 23933
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 3.1743884086608887,
+      "learning_rate": 6.974237071952793e-06,
+      "loss": 0.5295,
+      "step": 23934
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.8179881572723389,
+      "learning_rate": 6.973994834124222e-06,
+      "loss": 0.5384,
+      "step": 23935
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.971261978149414,
+      "learning_rate": 6.973752590806709e-06,
+      "loss": 0.6391,
+      "step": 23936
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.579354763031006,
+      "learning_rate": 6.973510342000931e-06,
+      "loss": 0.507,
+      "step": 23937
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.850571870803833,
+      "learning_rate": 6.973268087707559e-06,
+      "loss": 0.6239,
+      "step": 23938
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.061204671859741,
+      "learning_rate": 6.973025827927269e-06,
+      "loss": 0.4112,
+      "step": 23939
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.085812568664551,
+      "learning_rate": 6.972783562660732e-06,
+      "loss": 0.5459,
+      "step": 23940
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.626923084259033,
+      "learning_rate": 6.972541291908623e-06,
+      "loss": 0.8484,
+      "step": 23941
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.47556734085083,
+      "learning_rate": 6.972299015671616e-06,
+      "loss": 0.7904,
+      "step": 23942
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.6159000396728516,
+      "learning_rate": 6.9720567339503856e-06,
+      "loss": 0.8851,
+      "step": 23943
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.350996494293213,
+      "learning_rate": 6.971814446745602e-06,
+      "loss": 0.6426,
+      "step": 23944
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.6297391653060913,
+      "learning_rate": 6.971572154057942e-06,
+      "loss": 0.8367,
+      "step": 23945
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 4.09368896484375,
+      "learning_rate": 6.9713298558880785e-06,
+      "loss": 0.7983,
+      "step": 23946
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 3.314077377319336,
+      "learning_rate": 6.971087552236685e-06,
+      "loss": 0.5916,
+      "step": 23947
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.5940343141555786,
+      "learning_rate": 6.970845243104436e-06,
+      "loss": 0.5968,
+      "step": 23948
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.139340400695801,
+      "learning_rate": 6.9706029284920044e-06,
+      "loss": 0.7912,
+      "step": 23949
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 3.22480845451355,
+      "learning_rate": 6.970360608400064e-06,
+      "loss": 0.7842,
+      "step": 23950
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 3.7360692024230957,
+      "learning_rate": 6.970118282829288e-06,
+      "loss": 0.4788,
+      "step": 23951
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.6141375303268433,
+      "learning_rate": 6.969875951780352e-06,
+      "loss": 0.6576,
+      "step": 23952
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.261909008026123,
+      "learning_rate": 6.9696336152539276e-06,
+      "loss": 0.738,
+      "step": 23953
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.233429193496704,
+      "learning_rate": 6.969391273250692e-06,
+      "loss": 0.4775,
+      "step": 23954
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.2095086574554443,
+      "learning_rate": 6.969148925771316e-06,
+      "loss": 0.6574,
+      "step": 23955
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.6874661445617676,
+      "learning_rate": 6.968906572816475e-06,
+      "loss": 0.6247,
+      "step": 23956
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.2839322090148926,
+      "learning_rate": 6.968664214386842e-06,
+      "loss": 0.6129,
+      "step": 23957
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.163109540939331,
+      "learning_rate": 6.968421850483091e-06,
+      "loss": 0.5635,
+      "step": 23958
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.8564387559890747,
+      "learning_rate": 6.968179481105897e-06,
+      "loss": 0.6178,
+      "step": 23959
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.9338431358337402,
+      "learning_rate": 6.967937106255933e-06,
+      "loss": 0.8196,
+      "step": 23960
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.4566750526428223,
+      "learning_rate": 6.967694725933872e-06,
+      "loss": 0.5826,
+      "step": 23961
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.966649055480957,
+      "learning_rate": 6.967452340140391e-06,
+      "loss": 0.652,
+      "step": 23962
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 3.198650598526001,
+      "learning_rate": 6.96720994887616e-06,
+      "loss": 0.4509,
+      "step": 23963
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.360778570175171,
+      "learning_rate": 6.966967552141856e-06,
+      "loss": 0.6106,
+      "step": 23964
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.3431127071380615,
+      "learning_rate": 6.966725149938153e-06,
+      "loss": 0.6863,
+      "step": 23965
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.7642120122909546,
+      "learning_rate": 6.966482742265723e-06,
+      "loss": 0.696,
+      "step": 23966
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 3.7637674808502197,
+      "learning_rate": 6.966240329125243e-06,
+      "loss": 0.8285,
+      "step": 23967
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 3.103217601776123,
+      "learning_rate": 6.965997910517382e-06,
+      "loss": 0.686,
+      "step": 23968
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.0537002086639404,
+      "learning_rate": 6.96575548644282e-06,
+      "loss": 0.9218,
+      "step": 23969
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 5.3775529861450195,
+      "learning_rate": 6.965513056902227e-06,
+      "loss": 0.6711,
+      "step": 23970
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.4909017086029053,
+      "learning_rate": 6.965270621896277e-06,
+      "loss": 0.7909,
+      "step": 23971
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 3.69124436378479,
+      "learning_rate": 6.965028181425648e-06,
+      "loss": 0.7037,
+      "step": 23972
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 7.845826625823975,
+      "learning_rate": 6.964785735491011e-06,
+      "loss": 0.58,
+      "step": 23973
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 6.118736743927002,
+      "learning_rate": 6.964543284093041e-06,
+      "loss": 0.6804,
+      "step": 23974
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.88444447517395,
+      "learning_rate": 6.964300827232413e-06,
+      "loss": 0.7357,
+      "step": 23975
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 3.23241925239563,
+      "learning_rate": 6.964058364909799e-06,
+      "loss": 0.5116,
+      "step": 23976
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.4422942399978638,
+      "learning_rate": 6.9638158971258756e-06,
+      "loss": 0.5644,
+      "step": 23977
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.272024393081665,
+      "learning_rate": 6.9635734238813144e-06,
+      "loss": 0.5491,
+      "step": 23978
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.7197887897491455,
+      "learning_rate": 6.963330945176792e-06,
+      "loss": 0.5665,
+      "step": 23979
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.8763654232025146,
+      "learning_rate": 6.9630884610129815e-06,
+      "loss": 0.7495,
+      "step": 23980
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.8864175081253052,
+      "learning_rate": 6.962845971390557e-06,
+      "loss": 0.722,
+      "step": 23981
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.552361011505127,
+      "learning_rate": 6.962603476310193e-06,
+      "loss": 0.78,
+      "step": 23982
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.1770081520080566,
+      "learning_rate": 6.962360975772564e-06,
+      "loss": 0.724,
+      "step": 23983
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.7903387546539307,
+      "learning_rate": 6.962118469778345e-06,
+      "loss": 0.4969,
+      "step": 23984
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.0161147117614746,
+      "learning_rate": 6.9618759583282105e-06,
+      "loss": 0.6902,
+      "step": 23985
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.6768242120742798,
+      "learning_rate": 6.961633441422833e-06,
+      "loss": 0.7644,
+      "step": 23986
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.8948086500167847,
+      "learning_rate": 6.961390919062887e-06,
+      "loss": 0.7057,
+      "step": 23987
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.4552700519561768,
+      "learning_rate": 6.961148391249046e-06,
+      "loss": 0.4458,
+      "step": 23988
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.853069305419922,
+      "learning_rate": 6.960905857981989e-06,
+      "loss": 0.4375,
+      "step": 23989
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.4310944080352783,
+      "learning_rate": 6.960663319262386e-06,
+      "loss": 0.724,
+      "step": 23990
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 3.362457513809204,
+      "learning_rate": 6.960420775090914e-06,
+      "loss": 0.4937,
+      "step": 23991
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 5.219332218170166,
+      "learning_rate": 6.960178225468245e-06,
+      "loss": 0.8758,
+      "step": 23992
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.9751231670379639,
+      "learning_rate": 6.959935670395054e-06,
+      "loss": 0.6934,
+      "step": 23993
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.2444682121276855,
+      "learning_rate": 6.959693109872017e-06,
+      "loss": 0.4854,
+      "step": 23994
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 5.523433208465576,
+      "learning_rate": 6.959450543899808e-06,
+      "loss": 0.5083,
+      "step": 23995
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.7338263988494873,
+      "learning_rate": 6.9592079724791e-06,
+      "loss": 0.6436,
+      "step": 23996
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.003952741622925,
+      "learning_rate": 6.958965395610571e-06,
+      "loss": 0.52,
+      "step": 23997
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.9795868396759033,
+      "learning_rate": 6.95872281329489e-06,
+      "loss": 0.5953,
+      "step": 23998
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.7994663715362549,
+      "learning_rate": 6.958480225532736e-06,
+      "loss": 0.796,
+      "step": 23999
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.8874067068099976,
+      "learning_rate": 6.958237632324783e-06,
+      "loss": 0.4824,
+      "step": 24000
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 3.2240653038024902,
+      "learning_rate": 6.957995033671704e-06,
+      "loss": 0.5777,
+      "step": 24001
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.8468300104141235,
+      "learning_rate": 6.957752429574173e-06,
+      "loss": 0.5614,
+      "step": 24002
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.567986011505127,
+      "learning_rate": 6.957509820032867e-06,
+      "loss": 0.7551,
+      "step": 24003
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.269017219543457,
+      "learning_rate": 6.95726720504846e-06,
+      "loss": 0.6946,
+      "step": 24004
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.8913849592208862,
+      "learning_rate": 6.9570245846216255e-06,
+      "loss": 0.727,
+      "step": 24005
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.2351295948028564,
+      "learning_rate": 6.956781958753039e-06,
+      "loss": 0.5157,
+      "step": 24006
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.5913749933242798,
+      "learning_rate": 6.956539327443375e-06,
+      "loss": 0.2314,
+      "step": 24007
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 4.103354454040527,
+      "learning_rate": 6.956296690693309e-06,
+      "loss": 0.8908,
+      "step": 24008
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.2704522609710693,
+      "learning_rate": 6.956054048503513e-06,
+      "loss": 0.7581,
+      "step": 24009
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.6476680040359497,
+      "learning_rate": 6.955811400874663e-06,
+      "loss": 0.5666,
+      "step": 24010
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.8203363418579102,
+      "learning_rate": 6.955568747807436e-06,
+      "loss": 0.4912,
+      "step": 24011
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 4.2396392822265625,
+      "learning_rate": 6.955326089302503e-06,
+      "loss": 0.7436,
+      "step": 24012
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 4.1352458000183105,
+      "learning_rate": 6.955083425360543e-06,
+      "loss": 0.8648,
+      "step": 24013
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.542722463607788,
+      "learning_rate": 6.954840755982226e-06,
+      "loss": 0.7319,
+      "step": 24014
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.0094594955444336,
+      "learning_rate": 6.954598081168231e-06,
+      "loss": 0.572,
+      "step": 24015
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.7210136651992798,
+      "learning_rate": 6.9543554009192305e-06,
+      "loss": 0.4396,
+      "step": 24016
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.1628618240356445,
+      "learning_rate": 6.954112715235899e-06,
+      "loss": 0.6797,
+      "step": 24017
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.3668584823608398,
+      "learning_rate": 6.9538700241189126e-06,
+      "loss": 0.7428,
+      "step": 24018
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.5869433879852295,
+      "learning_rate": 6.953627327568946e-06,
+      "loss": 0.8383,
+      "step": 24019
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.7862122058868408,
+      "learning_rate": 6.9533846255866735e-06,
+      "loss": 0.6859,
+      "step": 24020
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.591648578643799,
+      "learning_rate": 6.95314191817277e-06,
+      "loss": 0.6419,
+      "step": 24021
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.831181764602661,
+      "learning_rate": 6.952899205327909e-06,
+      "loss": 0.8309,
+      "step": 24022
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.262012481689453,
+      "learning_rate": 6.952656487052768e-06,
+      "loss": 0.7274,
+      "step": 24023
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.496803045272827,
+      "learning_rate": 6.952413763348022e-06,
+      "loss": 0.6001,
+      "step": 24024
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.5954307317733765,
+      "learning_rate": 6.952171034214343e-06,
+      "loss": 0.4996,
+      "step": 24025
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.23442006111145,
+      "learning_rate": 6.951928299652408e-06,
+      "loss": 0.4983,
+      "step": 24026
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 3.245424509048462,
+      "learning_rate": 6.951685559662893e-06,
+      "loss": 0.6897,
+      "step": 24027
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.723639965057373,
+      "learning_rate": 6.951442814246469e-06,
+      "loss": 0.5074,
+      "step": 24028
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.037565231323242,
+      "learning_rate": 6.9512000634038166e-06,
+      "loss": 0.6659,
+      "step": 24029
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.6225991249084473,
+      "learning_rate": 6.950957307135606e-06,
+      "loss": 0.5803,
+      "step": 24030
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 3.423546314239502,
+      "learning_rate": 6.950714545442514e-06,
+      "loss": 0.7506,
+      "step": 24031
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 4.322084426879883,
+      "learning_rate": 6.9504717783252154e-06,
+      "loss": 0.6749,
+      "step": 24032
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.779609441757202,
+      "learning_rate": 6.950229005784386e-06,
+      "loss": 0.5483,
+      "step": 24033
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.22491192817688,
+      "learning_rate": 6.9499862278207e-06,
+      "loss": 0.7354,
+      "step": 24034
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.100403308868408,
+      "learning_rate": 6.949743444434833e-06,
+      "loss": 0.7555,
+      "step": 24035
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.7763748168945312,
+      "learning_rate": 6.9495006556274605e-06,
+      "loss": 0.5301,
+      "step": 24036
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.5951690673828125,
+      "learning_rate": 6.949257861399257e-06,
+      "loss": 0.5767,
+      "step": 24037
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 3.0070548057556152,
+      "learning_rate": 6.949015061750897e-06,
+      "loss": 0.6815,
+      "step": 24038
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.726744532585144,
+      "learning_rate": 6.9487722566830554e-06,
+      "loss": 0.5828,
+      "step": 24039
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.4025893211364746,
+      "learning_rate": 6.948529446196409e-06,
+      "loss": 0.6523,
+      "step": 24040
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.794166088104248,
+      "learning_rate": 6.948286630291632e-06,
+      "loss": 0.7383,
+      "step": 24041
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 4.033216953277588,
+      "learning_rate": 6.9480438089694e-06,
+      "loss": 0.3826,
+      "step": 24042
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.233614444732666,
+      "learning_rate": 6.947800982230387e-06,
+      "loss": 0.671,
+      "step": 24043
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.9904859066009521,
+      "learning_rate": 6.94755815007527e-06,
+      "loss": 0.6345,
+      "step": 24044
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.5545631647109985,
+      "learning_rate": 6.947315312504723e-06,
+      "loss": 0.5825,
+      "step": 24045
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.5692031383514404,
+      "learning_rate": 6.9470724695194225e-06,
+      "loss": 0.5577,
+      "step": 24046
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.0152153968811035,
+      "learning_rate": 6.946829621120041e-06,
+      "loss": 0.534,
+      "step": 24047
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.0283889770507812,
+      "learning_rate": 6.9465867673072575e-06,
+      "loss": 0.592,
+      "step": 24048
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.5836974382400513,
+      "learning_rate": 6.946343908081744e-06,
+      "loss": 0.5371,
+      "step": 24049
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.8949809074401855,
+      "learning_rate": 6.946101043444178e-06,
+      "loss": 0.5187,
+      "step": 24050
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 4.022388935089111,
+      "learning_rate": 6.945858173395234e-06,
+      "loss": 0.9053,
+      "step": 24051
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.1326918601989746,
+      "learning_rate": 6.945615297935588e-06,
+      "loss": 0.5142,
+      "step": 24052
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.550745725631714,
+      "learning_rate": 6.945372417065913e-06,
+      "loss": 0.7234,
+      "step": 24053
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.31811785697937,
+      "learning_rate": 6.945129530786887e-06,
+      "loss": 0.6454,
+      "step": 24054
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.8099322319030762,
+      "learning_rate": 6.944886639099184e-06,
+      "loss": 0.8607,
+      "step": 24055
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.6355856657028198,
+      "learning_rate": 6.9446437420034805e-06,
+      "loss": 0.6618,
+      "step": 24056
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 3.1410999298095703,
+      "learning_rate": 6.944400839500452e-06,
+      "loss": 0.6894,
+      "step": 24057
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.250229835510254,
+      "learning_rate": 6.944157931590771e-06,
+      "loss": 0.7036,
+      "step": 24058
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.9331610202789307,
+      "learning_rate": 6.943915018275116e-06,
+      "loss": 0.6973,
+      "step": 24059
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.8350764513015747,
+      "learning_rate": 6.943672099554162e-06,
+      "loss": 0.5532,
+      "step": 24060
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 3.089040756225586,
+      "learning_rate": 6.943429175428584e-06,
+      "loss": 0.6762,
+      "step": 24061
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.7535895109176636,
+      "learning_rate": 6.9431862458990575e-06,
+      "loss": 0.7929,
+      "step": 24062
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.0838239192962646,
+      "learning_rate": 6.942943310966257e-06,
+      "loss": 0.531,
+      "step": 24063
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.194800853729248,
+      "learning_rate": 6.942700370630859e-06,
+      "loss": 0.5021,
+      "step": 24064
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.902505874633789,
+      "learning_rate": 6.942457424893541e-06,
+      "loss": 0.5802,
+      "step": 24065
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 3.008737802505493,
+      "learning_rate": 6.942214473754975e-06,
+      "loss": 0.7231,
+      "step": 24066
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.6255979537963867,
+      "learning_rate": 6.941971517215839e-06,
+      "loss": 0.7456,
+      "step": 24067
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.0314414501190186,
+      "learning_rate": 6.941728555276807e-06,
+      "loss": 0.8047,
+      "step": 24068
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.4726905822753906,
+      "learning_rate": 6.9414855879385556e-06,
+      "loss": 0.5378,
+      "step": 24069
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.1595053672790527,
+      "learning_rate": 6.94124261520176e-06,
+      "loss": 0.5415,
+      "step": 24070
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.7947299480438232,
+      "learning_rate": 6.940999637067096e-06,
+      "loss": 0.7153,
+      "step": 24071
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.849644899368286,
+      "learning_rate": 6.940756653535239e-06,
+      "loss": 0.7204,
+      "step": 24072
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.611095666885376,
+      "learning_rate": 6.940513664606864e-06,
+      "loss": 0.4773,
+      "step": 24073
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.8727108240127563,
+      "learning_rate": 6.940270670282648e-06,
+      "loss": 0.7807,
+      "step": 24074
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 3.059140205383301,
+      "learning_rate": 6.940027670563266e-06,
+      "loss": 0.8523,
+      "step": 24075
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.115328073501587,
+      "learning_rate": 6.939784665449395e-06,
+      "loss": 0.5712,
+      "step": 24076
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 3.530534029006958,
+      "learning_rate": 6.939541654941708e-06,
+      "loss": 0.5796,
+      "step": 24077
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.980729818344116,
+      "learning_rate": 6.939298639040883e-06,
+      "loss": 0.7542,
+      "step": 24078
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.024796962738037,
+      "learning_rate": 6.939055617747595e-06,
+      "loss": 0.629,
+      "step": 24079
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.821648120880127,
+      "learning_rate": 6.9388125910625195e-06,
+      "loss": 0.8138,
+      "step": 24080
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.348602294921875,
+      "learning_rate": 6.9385695589863324e-06,
+      "loss": 0.549,
+      "step": 24081
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.471459150314331,
+      "learning_rate": 6.938326521519709e-06,
+      "loss": 0.4854,
+      "step": 24082
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.978824496269226,
+      "learning_rate": 6.938083478663327e-06,
+      "loss": 0.3458,
+      "step": 24083
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.543229103088379,
+      "learning_rate": 6.937840430417859e-06,
+      "loss": 0.8289,
+      "step": 24084
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.174600601196289,
+      "learning_rate": 6.937597376783984e-06,
+      "loss": 0.5259,
+      "step": 24085
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.8722598552703857,
+      "learning_rate": 6.937354317762377e-06,
+      "loss": 0.5408,
+      "step": 24086
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.4309110641479492,
+      "learning_rate": 6.937111253353712e-06,
+      "loss": 0.5825,
+      "step": 24087
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.9474555253982544,
+      "learning_rate": 6.936868183558668e-06,
+      "loss": 0.5228,
+      "step": 24088
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.1017844676971436,
+      "learning_rate": 6.936625108377918e-06,
+      "loss": 0.6285,
+      "step": 24089
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.883955717086792,
+      "learning_rate": 6.93638202781214e-06,
+      "loss": 0.7725,
+      "step": 24090
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.0012009143829346,
+      "learning_rate": 6.936138941862008e-06,
+      "loss": 0.5642,
+      "step": 24091
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 3.908066511154175,
+      "learning_rate": 6.9358958505281995e-06,
+      "loss": 0.961,
+      "step": 24092
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.2425711154937744,
+      "learning_rate": 6.9356527538113885e-06,
+      "loss": 0.706,
+      "step": 24093
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 3.607379674911499,
+      "learning_rate": 6.935409651712252e-06,
+      "loss": 0.7832,
+      "step": 24094
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.55595326423645,
+      "learning_rate": 6.9351665442314686e-06,
+      "loss": 0.7101,
+      "step": 24095
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.4465930461883545,
+      "learning_rate": 6.9349234313697115e-06,
+      "loss": 0.6137,
+      "step": 24096
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.9268797636032104,
+      "learning_rate": 6.934680313127655e-06,
+      "loss": 0.7504,
+      "step": 24097
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.1265969276428223,
+      "learning_rate": 6.93443718950598e-06,
+      "loss": 0.6721,
+      "step": 24098
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.7702999114990234,
+      "learning_rate": 6.934194060505358e-06,
+      "loss": 0.9837,
+      "step": 24099
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.2812219858169556,
+      "learning_rate": 6.933950926126468e-06,
+      "loss": 0.5983,
+      "step": 24100
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.7021498680114746,
+      "learning_rate": 6.933707786369985e-06,
+      "loss": 0.5719,
+      "step": 24101
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.4256843328475952,
+      "learning_rate": 6.9334646412365845e-06,
+      "loss": 0.2625,
+      "step": 24102
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.4092862606048584,
+      "learning_rate": 6.933221490726944e-06,
+      "loss": 0.4332,
+      "step": 24103
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.020763874053955,
+      "learning_rate": 6.9329783348417355e-06,
+      "loss": 0.6713,
+      "step": 24104
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 4.266318321228027,
+      "learning_rate": 6.9327351735816416e-06,
+      "loss": 0.5836,
+      "step": 24105
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 3.028085947036743,
+      "learning_rate": 6.9324920069473345e-06,
+      "loss": 0.5806,
+      "step": 24106
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.6854355335235596,
+      "learning_rate": 6.932248834939491e-06,
+      "loss": 0.7482,
+      "step": 24107
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.732967495918274,
+      "learning_rate": 6.932005657558787e-06,
+      "loss": 0.4791,
+      "step": 24108
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.7772634029388428,
+      "learning_rate": 6.9317624748059e-06,
+      "loss": 0.4587,
+      "step": 24109
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.6793403625488281,
+      "learning_rate": 6.9315192866815036e-06,
+      "loss": 0.7567,
+      "step": 24110
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.7298836708068848,
+      "learning_rate": 6.9312760931862765e-06,
+      "loss": 0.7026,
+      "step": 24111
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.9619617462158203,
+      "learning_rate": 6.931032894320894e-06,
+      "loss": 0.483,
+      "step": 24112
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.7175135612487793,
+      "learning_rate": 6.9307896900860326e-06,
+      "loss": 0.4653,
+      "step": 24113
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.312021017074585,
+      "learning_rate": 6.930546480482368e-06,
+      "loss": 0.5529,
+      "step": 24114
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 3.1930789947509766,
+      "learning_rate": 6.930303265510577e-06,
+      "loss": 0.8404,
+      "step": 24115
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.9440475702285767,
+      "learning_rate": 6.930060045171336e-06,
+      "loss": 0.6916,
+      "step": 24116
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.1955881118774414,
+      "learning_rate": 6.92981681946532e-06,
+      "loss": 0.7716,
+      "step": 24117
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.5757336616516113,
+      "learning_rate": 6.929573588393208e-06,
+      "loss": 0.557,
+      "step": 24118
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.7522543668746948,
+      "learning_rate": 6.929330351955673e-06,
+      "loss": 0.8065,
+      "step": 24119
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 5.242420673370361,
+      "learning_rate": 6.929087110153394e-06,
+      "loss": 0.7514,
+      "step": 24120
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.2236294746398926,
+      "learning_rate": 6.928843862987047e-06,
+      "loss": 0.4834,
+      "step": 24121
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.5446741580963135,
+      "learning_rate": 6.928600610457307e-06,
+      "loss": 0.53,
+      "step": 24122
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.1227729320526123,
+      "learning_rate": 6.9283573525648514e-06,
+      "loss": 0.624,
+      "step": 24123
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 3.6722898483276367,
+      "learning_rate": 6.928114089310355e-06,
+      "loss": 0.667,
+      "step": 24124
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.8057560920715332,
+      "learning_rate": 6.9278708206944965e-06,
+      "loss": 0.5764,
+      "step": 24125
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 3.4384589195251465,
+      "learning_rate": 6.927627546717952e-06,
+      "loss": 0.7274,
+      "step": 24126
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.629666328430176,
+      "learning_rate": 6.9273842673813975e-06,
+      "loss": 0.4617,
+      "step": 24127
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.5777678489685059,
+      "learning_rate": 6.92714098268551e-06,
+      "loss": 0.4576,
+      "step": 24128
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.8423722982406616,
+      "learning_rate": 6.9268976926309635e-06,
+      "loss": 0.3716,
+      "step": 24129
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.9287948608398438,
+      "learning_rate": 6.926654397218438e-06,
+      "loss": 0.4806,
+      "step": 24130
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.6198573112487793,
+      "learning_rate": 6.926411096448607e-06,
+      "loss": 0.583,
+      "step": 24131
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.4996647834777832,
+      "learning_rate": 6.926167790322149e-06,
+      "loss": 0.5273,
+      "step": 24132
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.366382598876953,
+      "learning_rate": 6.925924478839741e-06,
+      "loss": 0.6779,
+      "step": 24133
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 3.028968572616577,
+      "learning_rate": 6.925681162002056e-06,
+      "loss": 0.6315,
+      "step": 24134
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.5588202476501465,
+      "learning_rate": 6.925437839809774e-06,
+      "loss": 0.6097,
+      "step": 24135
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.4008989334106445,
+      "learning_rate": 6.925194512263571e-06,
+      "loss": 0.536,
+      "step": 24136
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.0636343955993652,
+      "learning_rate": 6.924951179364124e-06,
+      "loss": 0.5511,
+      "step": 24137
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.561107635498047,
+      "learning_rate": 6.924707841112108e-06,
+      "loss": 0.7055,
+      "step": 24138
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.0388615131378174,
+      "learning_rate": 6.924464497508201e-06,
+      "loss": 0.8937,
+      "step": 24139
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.7255738973617554,
+      "learning_rate": 6.92422114855308e-06,
+      "loss": 0.555,
+      "step": 24140
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.538464307785034,
+      "learning_rate": 6.92397779424742e-06,
+      "loss": 0.6166,
+      "step": 24141
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.758397102355957,
+      "learning_rate": 6.923734434591898e-06,
+      "loss": 0.6451,
+      "step": 24142
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.0621511936187744,
+      "learning_rate": 6.923491069587192e-06,
+      "loss": 0.6002,
+      "step": 24143
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.555138349533081,
+      "learning_rate": 6.923247699233976e-06,
+      "loss": 0.5414,
+      "step": 24144
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.620866060256958,
+      "learning_rate": 6.9230043235329296e-06,
+      "loss": 0.3516,
+      "step": 24145
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.0346004962921143,
+      "learning_rate": 6.922760942484729e-06,
+      "loss": 0.8164,
+      "step": 24146
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.568779706954956,
+      "learning_rate": 6.9225175560900506e-06,
+      "loss": 0.6994,
+      "step": 24147
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 3.2596256732940674,
+      "learning_rate": 6.92227416434957e-06,
+      "loss": 0.476,
+      "step": 24148
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 4.275887489318848,
+      "learning_rate": 6.922030767263966e-06,
+      "loss": 0.7301,
+      "step": 24149
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.3690810203552246,
+      "learning_rate": 6.921787364833915e-06,
+      "loss": 0.3476,
+      "step": 24150
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.733104944229126,
+      "learning_rate": 6.9215439570600926e-06,
+      "loss": 0.6864,
+      "step": 24151
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 3.1294360160827637,
+      "learning_rate": 6.921300543943177e-06,
+      "loss": 0.6368,
+      "step": 24152
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.4047999382019043,
+      "learning_rate": 6.921057125483843e-06,
+      "loss": 0.6534,
+      "step": 24153
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.0696845054626465,
+      "learning_rate": 6.920813701682769e-06,
+      "loss": 0.6445,
+      "step": 24154
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.9378889799118042,
+      "learning_rate": 6.920570272540632e-06,
+      "loss": 0.5918,
+      "step": 24155
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.133357524871826,
+      "learning_rate": 6.920326838058109e-06,
+      "loss": 0.5505,
+      "step": 24156
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.01227068901062,
+      "learning_rate": 6.920083398235875e-06,
+      "loss": 0.4701,
+      "step": 24157
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.7824106216430664,
+      "learning_rate": 6.91983995307461e-06,
+      "loss": 0.6366,
+      "step": 24158
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.885991096496582,
+      "learning_rate": 6.919596502574989e-06,
+      "loss": 0.796,
+      "step": 24159
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.6546263694763184,
+      "learning_rate": 6.919353046737689e-06,
+      "loss": 0.5778,
+      "step": 24160
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.9782823324203491,
+      "learning_rate": 6.919109585563387e-06,
+      "loss": 0.6412,
+      "step": 24161
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.131950855255127,
+      "learning_rate": 6.91886611905276e-06,
+      "loss": 0.7207,
+      "step": 24162
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 3.063060998916626,
+      "learning_rate": 6.918622647206485e-06,
+      "loss": 0.6738,
+      "step": 24163
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.284233331680298,
+      "learning_rate": 6.91837917002524e-06,
+      "loss": 0.697,
+      "step": 24164
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.0585317611694336,
+      "learning_rate": 6.9181356875097e-06,
+      "loss": 0.6533,
+      "step": 24165
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.3615198135375977,
+      "learning_rate": 6.917892199660544e-06,
+      "loss": 0.7373,
+      "step": 24166
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.1358230113983154,
+      "learning_rate": 6.917648706478448e-06,
+      "loss": 0.6131,
+      "step": 24167
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.9358850717544556,
+      "learning_rate": 6.9174052079640895e-06,
+      "loss": 0.4624,
+      "step": 24168
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.967534065246582,
+      "learning_rate": 6.917161704118145e-06,
+      "loss": 0.6371,
+      "step": 24169
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.8036450147628784,
+      "learning_rate": 6.916918194941294e-06,
+      "loss": 0.6472,
+      "step": 24170
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.0519673824310303,
+      "learning_rate": 6.916674680434209e-06,
+      "loss": 0.6398,
+      "step": 24171
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 3.5585057735443115,
+      "learning_rate": 6.916431160597571e-06,
+      "loss": 0.8573,
+      "step": 24172
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.6831684112548828,
+      "learning_rate": 6.916187635432055e-06,
+      "loss": 0.6771,
+      "step": 24173
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.1828272342681885,
+      "learning_rate": 6.915944104938339e-06,
+      "loss": 0.4884,
+      "step": 24174
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.3519575595855713,
+      "learning_rate": 6.915700569117101e-06,
+      "loss": 0.7487,
+      "step": 24175
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.0273964405059814,
+      "learning_rate": 6.915457027969017e-06,
+      "loss": 0.5975,
+      "step": 24176
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.2594635486602783,
+      "learning_rate": 6.915213481494765e-06,
+      "loss": 0.6164,
+      "step": 24177
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 3.444706916809082,
+      "learning_rate": 6.914969929695021e-06,
+      "loss": 0.4391,
+      "step": 24178
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.4551353454589844,
+      "learning_rate": 6.914726372570462e-06,
+      "loss": 0.6493,
+      "step": 24179
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.2193455696105957,
+      "learning_rate": 6.9144828101217675e-06,
+      "loss": 0.6329,
+      "step": 24180
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 3.262319803237915,
+      "learning_rate": 6.914239242349614e-06,
+      "loss": 0.8152,
+      "step": 24181
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.9918203353881836,
+      "learning_rate": 6.913995669254679e-06,
+      "loss": 0.7508,
+      "step": 24182
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.20741605758667,
+      "learning_rate": 6.913752090837636e-06,
+      "loss": 0.6672,
+      "step": 24183
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.081369400024414,
+      "learning_rate": 6.9135085070991685e-06,
+      "loss": 0.6532,
+      "step": 24184
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 3.0764951705932617,
+      "learning_rate": 6.913264918039948e-06,
+      "loss": 0.6114,
+      "step": 24185
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 3.461674928665161,
+      "learning_rate": 6.913021323660656e-06,
+      "loss": 0.6713,
+      "step": 24186
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 5.977625370025635,
+      "learning_rate": 6.912777723961968e-06,
+      "loss": 0.9339,
+      "step": 24187
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.5660386085510254,
+      "learning_rate": 6.912534118944562e-06,
+      "loss": 0.6792,
+      "step": 24188
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 3.7767889499664307,
+      "learning_rate": 6.9122905086091155e-06,
+      "loss": 0.5497,
+      "step": 24189
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.955493450164795,
+      "learning_rate": 6.912046892956305e-06,
+      "loss": 0.7579,
+      "step": 24190
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.952465295791626,
+      "learning_rate": 6.911803271986808e-06,
+      "loss": 0.795,
+      "step": 24191
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.4923713207244873,
+      "learning_rate": 6.911559645701303e-06,
+      "loss": 0.4485,
+      "step": 24192
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.9821488857269287,
+      "learning_rate": 6.911316014100465e-06,
+      "loss": 0.7648,
+      "step": 24193
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 3.4029436111450195,
+      "learning_rate": 6.911072377184975e-06,
+      "loss": 0.9537,
+      "step": 24194
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 4.163467884063721,
+      "learning_rate": 6.910828734955507e-06,
+      "loss": 0.5058,
+      "step": 24195
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 4.609709739685059,
+      "learning_rate": 6.9105850874127415e-06,
+      "loss": 0.6621,
+      "step": 24196
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.6875842809677124,
+      "learning_rate": 6.9103414345573535e-06,
+      "loss": 0.7387,
+      "step": 24197
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.825076937675476,
+      "learning_rate": 6.910097776390024e-06,
+      "loss": 0.7729,
+      "step": 24198
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.9728000164031982,
+      "learning_rate": 6.909854112911425e-06,
+      "loss": 0.9413,
+      "step": 24199
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 3.6856956481933594,
+      "learning_rate": 6.909610444122239e-06,
+      "loss": 0.6222,
+      "step": 24200
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.4431456327438354,
+      "learning_rate": 6.909366770023142e-06,
+      "loss": 0.3678,
+      "step": 24201
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.8115205764770508,
+      "learning_rate": 6.90912309061481e-06,
+      "loss": 0.6294,
+      "step": 24202
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.2768466472625732,
+      "learning_rate": 6.9088794058979225e-06,
+      "loss": 0.5002,
+      "step": 24203
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.98344087600708,
+      "learning_rate": 6.908635715873157e-06,
+      "loss": 0.633,
+      "step": 24204
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.3683788776397705,
+      "learning_rate": 6.908392020541188e-06,
+      "loss": 0.5366,
+      "step": 24205
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.0296924114227295,
+      "learning_rate": 6.908148319902699e-06,
+      "loss": 0.5916,
+      "step": 24206
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 4.785651683807373,
+      "learning_rate": 6.907904613958363e-06,
+      "loss": 0.8006,
+      "step": 24207
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.796875,
+      "learning_rate": 6.90766090270886e-06,
+      "loss": 0.7312,
+      "step": 24208
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 3.8717620372772217,
+      "learning_rate": 6.907417186154864e-06,
+      "loss": 0.7742,
+      "step": 24209
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.5982568264007568,
+      "learning_rate": 6.907173464297057e-06,
+      "loss": 0.7466,
+      "step": 24210
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.36004900932312,
+      "learning_rate": 6.906929737136115e-06,
+      "loss": 0.6597,
+      "step": 24211
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.114522933959961,
+      "learning_rate": 6.906686004672716e-06,
+      "loss": 0.6849,
+      "step": 24212
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.4766377210617065,
+      "learning_rate": 6.9064422669075385e-06,
+      "loss": 0.6795,
+      "step": 24213
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.723893642425537,
+      "learning_rate": 6.906198523841257e-06,
+      "loss": 0.7988,
+      "step": 24214
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.543569564819336,
+      "learning_rate": 6.9059547754745515e-06,
+      "loss": 0.7069,
+      "step": 24215
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.4635860919952393,
+      "learning_rate": 6.9057110218081015e-06,
+      "loss": 0.683,
+      "step": 24216
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 4.903170108795166,
+      "learning_rate": 6.905467262842582e-06,
+      "loss": 0.4451,
+      "step": 24217
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 3.1047301292419434,
+      "learning_rate": 6.905223498578672e-06,
+      "loss": 0.5883,
+      "step": 24218
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.5079164505004883,
+      "learning_rate": 6.9049797290170495e-06,
+      "loss": 0.8399,
+      "step": 24219
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.235898733139038,
+      "learning_rate": 6.904735954158392e-06,
+      "loss": 0.5923,
+      "step": 24220
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.7921037673950195,
+      "learning_rate": 6.904492174003377e-06,
+      "loss": 0.4957,
+      "step": 24221
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.689354658126831,
+      "learning_rate": 6.904248388552683e-06,
+      "loss": 0.516,
+      "step": 24222
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.2014782428741455,
+      "learning_rate": 6.904004597806987e-06,
+      "loss": 0.7367,
+      "step": 24223
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 3.234894275665283,
+      "learning_rate": 6.9037608017669675e-06,
+      "loss": 0.7854,
+      "step": 24224
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.093029737472534,
+      "learning_rate": 6.903517000433303e-06,
+      "loss": 0.568,
+      "step": 24225
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 3.1903200149536133,
+      "learning_rate": 6.90327319380667e-06,
+      "loss": 0.6133,
+      "step": 24226
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.7966241836547852,
+      "learning_rate": 6.903029381887748e-06,
+      "loss": 0.641,
+      "step": 24227
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.5202860832214355,
+      "learning_rate": 6.902785564677214e-06,
+      "loss": 0.5835,
+      "step": 24228
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 3.1791188716888428,
+      "learning_rate": 6.902541742175745e-06,
+      "loss": 0.8007,
+      "step": 24229
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 3.0116426944732666,
+      "learning_rate": 6.902297914384022e-06,
+      "loss": 0.7869,
+      "step": 24230
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.133852481842041,
+      "learning_rate": 6.90205408130272e-06,
+      "loss": 0.587,
+      "step": 24231
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 4.259470462799072,
+      "learning_rate": 6.901810242932518e-06,
+      "loss": 0.3677,
+      "step": 24232
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.6235241889953613,
+      "learning_rate": 6.9015663992740955e-06,
+      "loss": 0.4989,
+      "step": 24233
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.842943787574768,
+      "learning_rate": 6.901322550328129e-06,
+      "loss": 0.6051,
+      "step": 24234
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 5.280032157897949,
+      "learning_rate": 6.901078696095294e-06,
+      "loss": 0.9858,
+      "step": 24235
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 3.1934256553649902,
+      "learning_rate": 6.900834836576274e-06,
+      "loss": 0.8574,
+      "step": 24236
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.667132616043091,
+      "learning_rate": 6.900590971771744e-06,
+      "loss": 0.7637,
+      "step": 24237
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.215100049972534,
+      "learning_rate": 6.900347101682382e-06,
+      "loss": 0.6564,
+      "step": 24238
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.737004518508911,
+      "learning_rate": 6.900103226308867e-06,
+      "loss": 0.5829,
+      "step": 24239
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.2693822383880615,
+      "learning_rate": 6.899859345651876e-06,
+      "loss": 0.57,
+      "step": 24240
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.5877726078033447,
+      "learning_rate": 6.899615459712088e-06,
+      "loss": 0.5584,
+      "step": 24241
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.418914318084717,
+      "learning_rate": 6.899371568490182e-06,
+      "loss": 0.5299,
+      "step": 24242
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 3.2305104732513428,
+      "learning_rate": 6.899127671986834e-06,
+      "loss": 0.5078,
+      "step": 24243
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.1781163215637207,
+      "learning_rate": 6.898883770202724e-06,
+      "loss": 0.6538,
+      "step": 24244
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.9964247941970825,
+      "learning_rate": 6.8986398631385295e-06,
+      "loss": 0.7235,
+      "step": 24245
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.9820654392242432,
+      "learning_rate": 6.898395950794928e-06,
+      "loss": 0.6926,
+      "step": 24246
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.348146438598633,
+      "learning_rate": 6.898152033172599e-06,
+      "loss": 0.5535,
+      "step": 24247
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.423156499862671,
+      "learning_rate": 6.897908110272221e-06,
+      "loss": 0.5038,
+      "step": 24248
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.4326813220977783,
+      "learning_rate": 6.897664182094471e-06,
+      "loss": 0.651,
+      "step": 24249
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 3.237501621246338,
+      "learning_rate": 6.897420248640029e-06,
+      "loss": 0.6441,
+      "step": 24250
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 3.3046321868896484,
+      "learning_rate": 6.89717630990957e-06,
+      "loss": 0.821,
+      "step": 24251
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.709876537322998,
+      "learning_rate": 6.896932365903776e-06,
+      "loss": 0.8473,
+      "step": 24252
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 3.5200626850128174,
+      "learning_rate": 6.896688416623323e-06,
+      "loss": 0.7743,
+      "step": 24253
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.4382526874542236,
+      "learning_rate": 6.8964444620688895e-06,
+      "loss": 0.5668,
+      "step": 24254
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.1454193592071533,
+      "learning_rate": 6.896200502241156e-06,
+      "loss": 0.7103,
+      "step": 24255
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.604517936706543,
+      "learning_rate": 6.895956537140798e-06,
+      "loss": 0.6188,
+      "step": 24256
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.1528098583221436,
+      "learning_rate": 6.895712566768494e-06,
+      "loss": 0.6379,
+      "step": 24257
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 4.2959818840026855,
+      "learning_rate": 6.895468591124925e-06,
+      "loss": 0.7787,
+      "step": 24258
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.882627487182617,
+      "learning_rate": 6.895224610210768e-06,
+      "loss": 0.6009,
+      "step": 24259
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.6986875534057617,
+      "learning_rate": 6.894980624026701e-06,
+      "loss": 0.6843,
+      "step": 24260
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.8345788717269897,
+      "learning_rate": 6.894736632573403e-06,
+      "loss": 0.5067,
+      "step": 24261
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.701102375984192,
+      "learning_rate": 6.894492635851553e-06,
+      "loss": 0.8415,
+      "step": 24262
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.2369565963745117,
+      "learning_rate": 6.894248633861826e-06,
+      "loss": 0.6881,
+      "step": 24263
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.5975916385650635,
+      "learning_rate": 6.894004626604907e-06,
+      "loss": 0.7995,
+      "step": 24264
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 3.7878329753875732,
+      "learning_rate": 6.893760614081467e-06,
+      "loss": 0.6097,
+      "step": 24265
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.914501190185547,
+      "learning_rate": 6.893516596292189e-06,
+      "loss": 0.8381,
+      "step": 24266
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.616591215133667,
+      "learning_rate": 6.893272573237751e-06,
+      "loss": 0.8977,
+      "step": 24267
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.9205213785171509,
+      "learning_rate": 6.893028544918832e-06,
+      "loss": 0.7989,
+      "step": 24268
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.518075704574585,
+      "learning_rate": 6.892784511336109e-06,
+      "loss": 0.8189,
+      "step": 24269
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 3.520869016647339,
+      "learning_rate": 6.892540472490263e-06,
+      "loss": 0.7267,
+      "step": 24270
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 3.391334056854248,
+      "learning_rate": 6.892296428381968e-06,
+      "loss": 0.6371,
+      "step": 24271
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.6399571895599365,
+      "learning_rate": 6.892052379011907e-06,
+      "loss": 0.6479,
+      "step": 24272
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.3615376949310303,
+      "learning_rate": 6.891808324380757e-06,
+      "loss": 0.5199,
+      "step": 24273
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.4123260974884033,
+      "learning_rate": 6.891564264489198e-06,
+      "loss": 0.7693,
+      "step": 24274
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 3.1380767822265625,
+      "learning_rate": 6.8913201993379065e-06,
+      "loss": 0.7153,
+      "step": 24275
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.7953171730041504,
+      "learning_rate": 6.891076128927561e-06,
+      "loss": 0.7275,
+      "step": 24276
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.2597787380218506,
+      "learning_rate": 6.8908320532588425e-06,
+      "loss": 0.6315,
+      "step": 24277
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 3.401829242706299,
+      "learning_rate": 6.890587972332428e-06,
+      "loss": 0.7256,
+      "step": 24278
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.0848100185394287,
+      "learning_rate": 6.890343886148996e-06,
+      "loss": 0.6607,
+      "step": 24279
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.243734359741211,
+      "learning_rate": 6.890099794709226e-06,
+      "loss": 0.5662,
+      "step": 24280
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 3.4042813777923584,
+      "learning_rate": 6.889855698013796e-06,
+      "loss": 0.7128,
+      "step": 24281
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.569805145263672,
+      "learning_rate": 6.889611596063387e-06,
+      "loss": 0.6723,
+      "step": 24282
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 3.0065643787384033,
+      "learning_rate": 6.889367488858675e-06,
+      "loss": 0.6341,
+      "step": 24283
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.9111764430999756,
+      "learning_rate": 6.88912337640034e-06,
+      "loss": 0.7446,
+      "step": 24284
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.2426698207855225,
+      "learning_rate": 6.888879258689061e-06,
+      "loss": 0.7299,
+      "step": 24285
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.554840087890625,
+      "learning_rate": 6.8886351357255156e-06,
+      "loss": 0.6518,
+      "step": 24286
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.9997265338897705,
+      "learning_rate": 6.888391007510382e-06,
+      "loss": 0.5734,
+      "step": 24287
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.7962125539779663,
+      "learning_rate": 6.888146874044343e-06,
+      "loss": 0.8143,
+      "step": 24288
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.353752374649048,
+      "learning_rate": 6.887902735328074e-06,
+      "loss": 0.5946,
+      "step": 24289
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 3.4316911697387695,
+      "learning_rate": 6.887658591362254e-06,
+      "loss": 0.5907,
+      "step": 24290
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.4237372875213623,
+      "learning_rate": 6.887414442147563e-06,
+      "loss": 0.6963,
+      "step": 24291
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.5702641010284424,
+      "learning_rate": 6.887170287684679e-06,
+      "loss": 0.5858,
+      "step": 24292
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.7127066850662231,
+      "learning_rate": 6.886926127974282e-06,
+      "loss": 0.6342,
+      "step": 24293
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.750187873840332,
+      "learning_rate": 6.886681963017051e-06,
+      "loss": 0.6393,
+      "step": 24294
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 4.8025641441345215,
+      "learning_rate": 6.886437792813664e-06,
+      "loss": 0.7809,
+      "step": 24295
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 3.5786759853363037,
+      "learning_rate": 6.886193617364798e-06,
+      "loss": 0.6148,
+      "step": 24296
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.772072196006775,
+      "learning_rate": 6.885949436671137e-06,
+      "loss": 0.6232,
+      "step": 24297
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.0980403423309326,
+      "learning_rate": 6.885705250733355e-06,
+      "loss": 0.4847,
+      "step": 24298
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.517029285430908,
+      "learning_rate": 6.885461059552134e-06,
+      "loss": 0.6857,
+      "step": 24299
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.0747716426849365,
+      "learning_rate": 6.885216863128152e-06,
+      "loss": 0.7475,
+      "step": 24300
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.5627802610397339,
+      "learning_rate": 6.8849726614620895e-06,
+      "loss": 0.6233,
+      "step": 24301
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.4857231378555298,
+      "learning_rate": 6.8847284545546215e-06,
+      "loss": 0.3726,
+      "step": 24302
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 3.560492753982544,
+      "learning_rate": 6.88448424240643e-06,
+      "loss": 0.8986,
+      "step": 24303
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 4.543764591217041,
+      "learning_rate": 6.884240025018196e-06,
+      "loss": 1.1337,
+      "step": 24304
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.6342225074768066,
+      "learning_rate": 6.883995802390595e-06,
+      "loss": 0.4716,
+      "step": 24305
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.8149654865264893,
+      "learning_rate": 6.883751574524307e-06,
+      "loss": 0.7202,
+      "step": 24306
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.194201946258545,
+      "learning_rate": 6.883507341420014e-06,
+      "loss": 0.6009,
+      "step": 24307
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.045651912689209,
+      "learning_rate": 6.883263103078389e-06,
+      "loss": 0.6221,
+      "step": 24308
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.8208937644958496,
+      "learning_rate": 6.883018859500117e-06,
+      "loss": 0.6579,
+      "step": 24309
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.4449219703674316,
+      "learning_rate": 6.882774610685875e-06,
+      "loss": 0.7856,
+      "step": 24310
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.19376802444458,
+      "learning_rate": 6.882530356636341e-06,
+      "loss": 0.6165,
+      "step": 24311
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.367048740386963,
+      "learning_rate": 6.882286097352196e-06,
+      "loss": 0.5031,
+      "step": 24312
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.9691722393035889,
+      "learning_rate": 6.882041832834119e-06,
+      "loss": 0.5872,
+      "step": 24313
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.4570422172546387,
+      "learning_rate": 6.881797563082789e-06,
+      "loss": 0.4955,
+      "step": 24314
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.91349196434021,
+      "learning_rate": 6.881553288098884e-06,
+      "loss": 0.6416,
+      "step": 24315
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.695874810218811,
+      "learning_rate": 6.881309007883083e-06,
+      "loss": 0.6486,
+      "step": 24316
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.0932374000549316,
+      "learning_rate": 6.881064722436067e-06,
+      "loss": 0.5847,
+      "step": 24317
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.9082775115966797,
+      "learning_rate": 6.880820431758516e-06,
+      "loss": 0.7464,
+      "step": 24318
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.651892900466919,
+      "learning_rate": 6.880576135851107e-06,
+      "loss": 0.8043,
+      "step": 24319
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.300701141357422,
+      "learning_rate": 6.880331834714521e-06,
+      "loss": 0.7361,
+      "step": 24320
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 3.2874772548675537,
+      "learning_rate": 6.880087528349436e-06,
+      "loss": 1.0826,
+      "step": 24321
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.1115355491638184,
+      "learning_rate": 6.879843216756532e-06,
+      "loss": 0.6325,
+      "step": 24322
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.3597424030303955,
+      "learning_rate": 6.879598899936489e-06,
+      "loss": 0.33,
+      "step": 24323
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 4.3004889488220215,
+      "learning_rate": 6.879354577889984e-06,
+      "loss": 0.7526,
+      "step": 24324
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.2973198890686035,
+      "learning_rate": 6.879110250617698e-06,
+      "loss": 0.6619,
+      "step": 24325
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.1604111194610596,
+      "learning_rate": 6.878865918120312e-06,
+      "loss": 0.6461,
+      "step": 24326
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.09552264213562,
+      "learning_rate": 6.8786215803985015e-06,
+      "loss": 0.4869,
+      "step": 24327
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.8670107126235962,
+      "learning_rate": 6.8783772374529504e-06,
+      "loss": 0.6096,
+      "step": 24328
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.4977283477783203,
+      "learning_rate": 6.878132889284335e-06,
+      "loss": 0.4758,
+      "step": 24329
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 3.191539764404297,
+      "learning_rate": 6.877888535893335e-06,
+      "loss": 0.7631,
+      "step": 24330
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.4240264892578125,
+      "learning_rate": 6.877644177280631e-06,
+      "loss": 0.5063,
+      "step": 24331
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 3.7242679595947266,
+      "learning_rate": 6.877399813446901e-06,
+      "loss": 0.7233,
+      "step": 24332
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.0400679111480713,
+      "learning_rate": 6.8771554443928265e-06,
+      "loss": 0.5625,
+      "step": 24333
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 4.172549724578857,
+      "learning_rate": 6.876911070119086e-06,
+      "loss": 0.6218,
+      "step": 24334
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.513255000114441,
+      "learning_rate": 6.876666690626359e-06,
+      "loss": 0.5127,
+      "step": 24335
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.991668701171875,
+      "learning_rate": 6.876422305915325e-06,
+      "loss": 0.759,
+      "step": 24336
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 3.064429521560669,
+      "learning_rate": 6.876177915986661e-06,
+      "loss": 0.5619,
+      "step": 24337
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.6198253631591797,
+      "learning_rate": 6.875933520841051e-06,
+      "loss": 0.5508,
+      "step": 24338
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 3.7232394218444824,
+      "learning_rate": 6.875689120479173e-06,
+      "loss": 0.5342,
+      "step": 24339
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.839707374572754,
+      "learning_rate": 6.875444714901706e-06,
+      "loss": 0.5923,
+      "step": 24340
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.4205310344696045,
+      "learning_rate": 6.875200304109329e-06,
+      "loss": 0.6318,
+      "step": 24341
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 3.0281519889831543,
+      "learning_rate": 6.874955888102723e-06,
+      "loss": 0.6757,
+      "step": 24342
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.783766031265259,
+      "learning_rate": 6.874711466882568e-06,
+      "loss": 0.9073,
+      "step": 24343
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.8792388439178467,
+      "learning_rate": 6.8744670404495414e-06,
+      "loss": 0.7248,
+      "step": 24344
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 3.39489483833313,
+      "learning_rate": 6.874222608804325e-06,
+      "loss": 0.4661,
+      "step": 24345
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.9493286609649658,
+      "learning_rate": 6.873978171947597e-06,
+      "loss": 0.6876,
+      "step": 24346
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.3494808673858643,
+      "learning_rate": 6.873733729880039e-06,
+      "loss": 0.5985,
+      "step": 24347
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.333980083465576,
+      "learning_rate": 6.873489282602327e-06,
+      "loss": 0.7539,
+      "step": 24348
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 8.306413650512695,
+      "learning_rate": 6.873244830115145e-06,
+      "loss": 0.497,
+      "step": 24349
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.9526973962783813,
+      "learning_rate": 6.87300037241917e-06,
+      "loss": 0.6126,
+      "step": 24350
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.1446053981781006,
+      "learning_rate": 6.872755909515083e-06,
+      "loss": 0.872,
+      "step": 24351
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.491098165512085,
+      "learning_rate": 6.872511441403563e-06,
+      "loss": 0.5563,
+      "step": 24352
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.51277756690979,
+      "learning_rate": 6.87226696808529e-06,
+      "loss": 0.6704,
+      "step": 24353
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 3.2424283027648926,
+      "learning_rate": 6.872022489560945e-06,
+      "loss": 0.4023,
+      "step": 24354
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.8025469779968262,
+      "learning_rate": 6.871778005831207e-06,
+      "loss": 0.4986,
+      "step": 24355
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.021639585494995,
+      "learning_rate": 6.871533516896753e-06,
+      "loss": 0.7495,
+      "step": 24356
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 3.3881118297576904,
+      "learning_rate": 6.871289022758267e-06,
+      "loss": 0.6987,
+      "step": 24357
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 3.5923967361450195,
+      "learning_rate": 6.871044523416427e-06,
+      "loss": 0.616,
+      "step": 24358
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.0252554416656494,
+      "learning_rate": 6.870800018871914e-06,
+      "loss": 0.6554,
+      "step": 24359
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.3012731075286865,
+      "learning_rate": 6.870555509125406e-06,
+      "loss": 0.7777,
+      "step": 24360
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.4859857559204102,
+      "learning_rate": 6.870310994177585e-06,
+      "loss": 0.6397,
+      "step": 24361
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.5775368213653564,
+      "learning_rate": 6.870066474029129e-06,
+      "loss": 0.4519,
+      "step": 24362
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.017573833465576,
+      "learning_rate": 6.869821948680718e-06,
+      "loss": 0.7396,
+      "step": 24363
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.8538925647735596,
+      "learning_rate": 6.869577418133032e-06,
+      "loss": 0.6166,
+      "step": 24364
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.0025229454040527,
+      "learning_rate": 6.869332882386754e-06,
+      "loss": 0.6278,
+      "step": 24365
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 3.41434383392334,
+      "learning_rate": 6.8690883414425594e-06,
+      "loss": 0.5164,
+      "step": 24366
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.372114896774292,
+      "learning_rate": 6.868843795301131e-06,
+      "loss": 0.5381,
+      "step": 24367
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 3.2949585914611816,
+      "learning_rate": 6.868599243963149e-06,
+      "loss": 0.8466,
+      "step": 24368
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.9711354970932007,
+      "learning_rate": 6.868354687429291e-06,
+      "loss": 0.5443,
+      "step": 24369
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.9632397890090942,
+      "learning_rate": 6.868110125700239e-06,
+      "loss": 0.6588,
+      "step": 24370
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.023730754852295,
+      "learning_rate": 6.867865558776673e-06,
+      "loss": 0.7427,
+      "step": 24371
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.03936767578125,
+      "learning_rate": 6.867620986659273e-06,
+      "loss": 0.7477,
+      "step": 24372
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.9593287706375122,
+      "learning_rate": 6.8673764093487175e-06,
+      "loss": 0.5208,
+      "step": 24373
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.987312912940979,
+      "learning_rate": 6.867131826845688e-06,
+      "loss": 0.6121,
+      "step": 24374
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.7151458263397217,
+      "learning_rate": 6.866887239150865e-06,
+      "loss": 0.5001,
+      "step": 24375
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.8342552185058594,
+      "learning_rate": 6.866642646264927e-06,
+      "loss": 0.4955,
+      "step": 24376
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 3.3807506561279297,
+      "learning_rate": 6.866398048188556e-06,
+      "loss": 0.6439,
+      "step": 24377
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 3.2661056518554688,
+      "learning_rate": 6.8661534449224296e-06,
+      "loss": 0.7923,
+      "step": 24378
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 5.69654655456543,
+      "learning_rate": 6.865908836467232e-06,
+      "loss": 0.6123,
+      "step": 24379
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.1276469230651855,
+      "learning_rate": 6.865664222823639e-06,
+      "loss": 0.6648,
+      "step": 24380
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.661854863166809,
+      "learning_rate": 6.865419603992333e-06,
+      "loss": 0.6039,
+      "step": 24381
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.073780059814453,
+      "learning_rate": 6.8651749799739955e-06,
+      "loss": 0.634,
+      "step": 24382
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 3.762540102005005,
+      "learning_rate": 6.8649303507693044e-06,
+      "loss": 0.8016,
+      "step": 24383
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.19301438331604,
+      "learning_rate": 6.8646857163789405e-06,
+      "loss": 0.6182,
+      "step": 24384
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.9745779037475586,
+      "learning_rate": 6.864441076803583e-06,
+      "loss": 0.7775,
+      "step": 24385
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.4047791957855225,
+      "learning_rate": 6.8641964320439145e-06,
+      "loss": 0.6746,
+      "step": 24386
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.9357064962387085,
+      "learning_rate": 6.863951782100613e-06,
+      "loss": 0.7064,
+      "step": 24387
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.6506097316741943,
+      "learning_rate": 6.86370712697436e-06,
+      "loss": 0.6536,
+      "step": 24388
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 3.3432490825653076,
+      "learning_rate": 6.863462466665835e-06,
+      "loss": 0.7279,
+      "step": 24389
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.055929660797119,
+      "learning_rate": 6.8632178011757215e-06,
+      "loss": 0.3849,
+      "step": 24390
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 4.605380535125732,
+      "learning_rate": 6.862973130504695e-06,
+      "loss": 0.6239,
+      "step": 24391
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.3112528324127197,
+      "learning_rate": 6.86272845465344e-06,
+      "loss": 0.6959,
+      "step": 24392
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.559976577758789,
+      "learning_rate": 6.862483773622634e-06,
+      "loss": 0.6735,
+      "step": 24393
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.8793572187423706,
+      "learning_rate": 6.862239087412958e-06,
+      "loss": 0.7437,
+      "step": 24394
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.3839430809020996,
+      "learning_rate": 6.8619943960250935e-06,
+      "loss": 0.579,
+      "step": 24395
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.903412103652954,
+      "learning_rate": 6.8617496994597185e-06,
+      "loss": 0.3785,
+      "step": 24396
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 3.6724088191986084,
+      "learning_rate": 6.861504997717516e-06,
+      "loss": 0.7565,
+      "step": 24397
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.6539111137390137,
+      "learning_rate": 6.861260290799166e-06,
+      "loss": 0.6678,
+      "step": 24398
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.0780296325683594,
+      "learning_rate": 6.861015578705348e-06,
+      "loss": 0.6797,
+      "step": 24399
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 5.700173377990723,
+      "learning_rate": 6.860770861436742e-06,
+      "loss": 0.8755,
+      "step": 24400
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.180320978164673,
+      "learning_rate": 6.86052613899403e-06,
+      "loss": 0.693,
+      "step": 24401
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.211416244506836,
+      "learning_rate": 6.860281411377892e-06,
+      "loss": 0.5864,
+      "step": 24402
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 3.236143112182617,
+      "learning_rate": 6.860036678589008e-06,
+      "loss": 0.6134,
+      "step": 24403
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 7.795481204986572,
+      "learning_rate": 6.859791940628059e-06,
+      "loss": 0.5198,
+      "step": 24404
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.530592679977417,
+      "learning_rate": 6.859547197495725e-06,
+      "loss": 0.4143,
+      "step": 24405
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.808376431465149,
+      "learning_rate": 6.859302449192688e-06,
+      "loss": 0.6221,
+      "step": 24406
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.419235944747925,
+      "learning_rate": 6.859057695719626e-06,
+      "loss": 0.7653,
+      "step": 24407
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.9462541341781616,
+      "learning_rate": 6.8588129370772216e-06,
+      "loss": 0.6577,
+      "step": 24408
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.9708409309387207,
+      "learning_rate": 6.858568173266153e-06,
+      "loss": 0.6232,
+      "step": 24409
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 3.697927951812744,
+      "learning_rate": 6.858323404287104e-06,
+      "loss": 0.4551,
+      "step": 24410
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.500232219696045,
+      "learning_rate": 6.858078630140754e-06,
+      "loss": 0.7834,
+      "step": 24411
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 3.077918767929077,
+      "learning_rate": 6.857833850827783e-06,
+      "loss": 0.6796,
+      "step": 24412
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.6413230895996094,
+      "learning_rate": 6.8575890663488735e-06,
+      "loss": 0.8158,
+      "step": 24413
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.823744058609009,
+      "learning_rate": 6.857344276704703e-06,
+      "loss": 0.3825,
+      "step": 24414
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.9244680404663086,
+      "learning_rate": 6.857099481895953e-06,
+      "loss": 0.594,
+      "step": 24415
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.7710002660751343,
+      "learning_rate": 6.856854681923307e-06,
+      "loss": 0.6569,
+      "step": 24416
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.7148494720458984,
+      "learning_rate": 6.856609876787444e-06,
+      "loss": 0.4739,
+      "step": 24417
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.941309690475464,
+      "learning_rate": 6.856365066489042e-06,
+      "loss": 0.9948,
+      "step": 24418
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.663245916366577,
+      "learning_rate": 6.856120251028786e-06,
+      "loss": 0.742,
+      "step": 24419
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.8552069664001465,
+      "learning_rate": 6.855875430407354e-06,
+      "loss": 0.6779,
+      "step": 24420
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.612260341644287,
+      "learning_rate": 6.8556306046254285e-06,
+      "loss": 0.3626,
+      "step": 24421
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.850889205932617,
+      "learning_rate": 6.855385773683689e-06,
+      "loss": 0.6455,
+      "step": 24422
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.2882187366485596,
+      "learning_rate": 6.855140937582816e-06,
+      "loss": 0.372,
+      "step": 24423
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 4.886592864990234,
+      "learning_rate": 6.8548960963234925e-06,
+      "loss": 0.6637,
+      "step": 24424
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.2968149185180664,
+      "learning_rate": 6.854651249906397e-06,
+      "loss": 0.7403,
+      "step": 24425
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.7997355461120605,
+      "learning_rate": 6.854406398332211e-06,
+      "loss": 0.5772,
+      "step": 24426
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.728189468383789,
+      "learning_rate": 6.854161541601615e-06,
+      "loss": 0.7137,
+      "step": 24427
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.5484538078308105,
+      "learning_rate": 6.853916679715291e-06,
+      "loss": 0.7103,
+      "step": 24428
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.9843575954437256,
+      "learning_rate": 6.85367181267392e-06,
+      "loss": 0.5189,
+      "step": 24429
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.2870583534240723,
+      "learning_rate": 6.853426940478181e-06,
+      "loss": 0.8436,
+      "step": 24430
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 3.202427387237549,
+      "learning_rate": 6.853182063128756e-06,
+      "loss": 0.6984,
+      "step": 24431
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.8983656167984009,
+      "learning_rate": 6.852937180626326e-06,
+      "loss": 0.5814,
+      "step": 24432
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.843937635421753,
+      "learning_rate": 6.852692292971572e-06,
+      "loss": 0.577,
+      "step": 24433
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.782772183418274,
+      "learning_rate": 6.852447400165174e-06,
+      "loss": 0.7854,
+      "step": 24434
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.547356605529785,
+      "learning_rate": 6.8522025022078155e-06,
+      "loss": 0.4627,
+      "step": 24435
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 5.187408447265625,
+      "learning_rate": 6.851957599100174e-06,
+      "loss": 0.6067,
+      "step": 24436
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.0803189277648926,
+      "learning_rate": 6.851712690842933e-06,
+      "loss": 0.628,
+      "step": 24437
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.908075213432312,
+      "learning_rate": 6.851467777436773e-06,
+      "loss": 0.6799,
+      "step": 24438
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.2319955825805664,
+      "learning_rate": 6.851222858882374e-06,
+      "loss": 0.5351,
+      "step": 24439
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.302942991256714,
+      "learning_rate": 6.850977935180417e-06,
+      "loss": 0.6517,
+      "step": 24440
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.231398105621338,
+      "learning_rate": 6.8507330063315854e-06,
+      "loss": 0.77,
+      "step": 24441
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.419917583465576,
+      "learning_rate": 6.850488072336558e-06,
+      "loss": 0.4466,
+      "step": 24442
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 3.4029674530029297,
+      "learning_rate": 6.850243133196016e-06,
+      "loss": 0.6976,
+      "step": 24443
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.8786587715148926,
+      "learning_rate": 6.849998188910643e-06,
+      "loss": 0.7718,
+      "step": 24444
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 4.0633745193481445,
+      "learning_rate": 6.849753239481116e-06,
+      "loss": 0.7694,
+      "step": 24445
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.7972981929779053,
+      "learning_rate": 6.849508284908119e-06,
+      "loss": 0.6293,
+      "step": 24446
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 5.151513576507568,
+      "learning_rate": 6.849263325192331e-06,
+      "loss": 0.8446,
+      "step": 24447
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 6.487739086151123,
+      "learning_rate": 6.849018360334436e-06,
+      "loss": 0.7574,
+      "step": 24448
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 3.9944205284118652,
+      "learning_rate": 6.8487733903351114e-06,
+      "loss": 0.9183,
+      "step": 24449
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.293163776397705,
+      "learning_rate": 6.848528415195043e-06,
+      "loss": 0.5548,
+      "step": 24450
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.161597728729248,
+      "learning_rate": 6.848283434914908e-06,
+      "loss": 0.6439,
+      "step": 24451
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.473567008972168,
+      "learning_rate": 6.84803844949539e-06,
+      "loss": 0.761,
+      "step": 24452
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.957507848739624,
+      "learning_rate": 6.84779345893717e-06,
+      "loss": 0.7701,
+      "step": 24453
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.024710178375244,
+      "learning_rate": 6.847548463240927e-06,
+      "loss": 0.717,
+      "step": 24454
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.7737209796905518,
+      "learning_rate": 6.8473034624073455e-06,
+      "loss": 0.2717,
+      "step": 24455
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.8466796875,
+      "learning_rate": 6.847058456437103e-06,
+      "loss": 0.8263,
+      "step": 24456
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.4378273487091064,
+      "learning_rate": 6.846813445330885e-06,
+      "loss": 0.7337,
+      "step": 24457
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.2476673126220703,
+      "learning_rate": 6.84656842908937e-06,
+      "loss": 0.657,
+      "step": 24458
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 3.5845353603363037,
+      "learning_rate": 6.846323407713238e-06,
+      "loss": 0.5907,
+      "step": 24459
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.243924140930176,
+      "learning_rate": 6.846078381203174e-06,
+      "loss": 0.649,
+      "step": 24460
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 3.1292574405670166,
+      "learning_rate": 6.845833349559859e-06,
+      "loss": 0.6406,
+      "step": 24461
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.288959264755249,
+      "learning_rate": 6.845588312783971e-06,
+      "loss": 0.6151,
+      "step": 24462
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.689725160598755,
+      "learning_rate": 6.845343270876192e-06,
+      "loss": 0.7313,
+      "step": 24463
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.0008583068847656,
+      "learning_rate": 6.845098223837207e-06,
+      "loss": 0.5734,
+      "step": 24464
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.3827358484268188,
+      "learning_rate": 6.844853171667695e-06,
+      "loss": 0.7239,
+      "step": 24465
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.6894967555999756,
+      "learning_rate": 6.844608114368337e-06,
+      "loss": 0.7492,
+      "step": 24466
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 3.0175817012786865,
+      "learning_rate": 6.844363051939814e-06,
+      "loss": 0.7455,
+      "step": 24467
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.0263967514038086,
+      "learning_rate": 6.84411798438281e-06,
+      "loss": 0.5608,
+      "step": 24468
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.8198015689849854,
+      "learning_rate": 6.843872911698003e-06,
+      "loss": 0.6494,
+      "step": 24469
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.2328648567199707,
+      "learning_rate": 6.8436278338860766e-06,
+      "loss": 0.3907,
+      "step": 24470
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.9074666500091553,
+      "learning_rate": 6.843382750947711e-06,
+      "loss": 0.628,
+      "step": 24471
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.638897180557251,
+      "learning_rate": 6.84313766288359e-06,
+      "loss": 0.4034,
+      "step": 24472
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.1615288257598877,
+      "learning_rate": 6.842892569694392e-06,
+      "loss": 0.6589,
+      "step": 24473
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.9215196371078491,
+      "learning_rate": 6.842647471380803e-06,
+      "loss": 0.4618,
+      "step": 24474
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.013211727142334,
+      "learning_rate": 6.8424023679434995e-06,
+      "loss": 0.8437,
+      "step": 24475
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.2048399448394775,
+      "learning_rate": 6.842157259383165e-06,
+      "loss": 0.4495,
+      "step": 24476
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.0462872982025146,
+      "learning_rate": 6.841912145700482e-06,
+      "loss": 0.6106,
+      "step": 24477
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 3.3047971725463867,
+      "learning_rate": 6.841667026896132e-06,
+      "loss": 1.0362,
+      "step": 24478
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.2214300632476807,
+      "learning_rate": 6.841421902970795e-06,
+      "loss": 0.6989,
+      "step": 24479
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.7768199443817139,
+      "learning_rate": 6.841176773925152e-06,
+      "loss": 0.5507,
+      "step": 24480
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.439089775085449,
+      "learning_rate": 6.840931639759888e-06,
+      "loss": 0.7752,
+      "step": 24481
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.017791986465454,
+      "learning_rate": 6.840686500475683e-06,
+      "loss": 0.5148,
+      "step": 24482
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 3.592616081237793,
+      "learning_rate": 6.840441356073218e-06,
+      "loss": 0.5783,
+      "step": 24483
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.7943873405456543,
+      "learning_rate": 6.8401962065531745e-06,
+      "loss": 0.5761,
+      "step": 24484
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.3900740146636963,
+      "learning_rate": 6.839951051916235e-06,
+      "loss": 0.6823,
+      "step": 24485
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.5188411474227905,
+      "learning_rate": 6.839705892163081e-06,
+      "loss": 0.5036,
+      "step": 24486
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 4.378026962280273,
+      "learning_rate": 6.839460727294394e-06,
+      "loss": 0.6112,
+      "step": 24487
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.081404685974121,
+      "learning_rate": 6.839215557310854e-06,
+      "loss": 0.3784,
+      "step": 24488
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.2619614601135254,
+      "learning_rate": 6.838970382213147e-06,
+      "loss": 0.7985,
+      "step": 24489
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 3.4499118328094482,
+      "learning_rate": 6.838725202001952e-06,
+      "loss": 0.5478,
+      "step": 24490
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.490015983581543,
+      "learning_rate": 6.838480016677951e-06,
+      "loss": 0.5839,
+      "step": 24491
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.9846689701080322,
+      "learning_rate": 6.838234826241825e-06,
+      "loss": 0.6512,
+      "step": 24492
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.0036075115203857,
+      "learning_rate": 6.8379896306942564e-06,
+      "loss": 0.4975,
+      "step": 24493
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 3.225700616836548,
+      "learning_rate": 6.837744430035927e-06,
+      "loss": 0.507,
+      "step": 24494
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.7466845512390137,
+      "learning_rate": 6.83749922426752e-06,
+      "loss": 0.7011,
+      "step": 24495
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.1886725425720215,
+      "learning_rate": 6.837254013389716e-06,
+      "loss": 0.6754,
+      "step": 24496
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.9018884897232056,
+      "learning_rate": 6.8370087974031965e-06,
+      "loss": 0.6632,
+      "step": 24497
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 3.6734237670898438,
+      "learning_rate": 6.836763576308642e-06,
+      "loss": 0.7946,
+      "step": 24498
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.4649100303649902,
+      "learning_rate": 6.836518350106739e-06,
+      "loss": 0.5034,
+      "step": 24499
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.248236656188965,
+      "learning_rate": 6.836273118798164e-06,
+      "loss": 0.6632,
+      "step": 24500
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.6908159255981445,
+      "learning_rate": 6.836027882383602e-06,
+      "loss": 0.6318,
+      "step": 24501
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.653320789337158,
+      "learning_rate": 6.835782640863735e-06,
+      "loss": 0.6031,
+      "step": 24502
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.3566598892211914,
+      "learning_rate": 6.8355373942392435e-06,
+      "loss": 0.6907,
+      "step": 24503
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.8361406326293945,
+      "learning_rate": 6.835292142510811e-06,
+      "loss": 0.8219,
+      "step": 24504
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.7855701446533203,
+      "learning_rate": 6.8350468856791175e-06,
+      "loss": 0.5777,
+      "step": 24505
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 3.1572535037994385,
+      "learning_rate": 6.834801623744847e-06,
+      "loss": 0.6934,
+      "step": 24506
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 4.144914627075195,
+      "learning_rate": 6.834556356708679e-06,
+      "loss": 0.9813,
+      "step": 24507
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 3.5710842609405518,
+      "learning_rate": 6.834311084571299e-06,
+      "loss": 0.515,
+      "step": 24508
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.3355021476745605,
+      "learning_rate": 6.834065807333387e-06,
+      "loss": 0.5871,
+      "step": 24509
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 3.306875228881836,
+      "learning_rate": 6.833820524995622e-06,
+      "loss": 0.7023,
+      "step": 24510
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 4.192298889160156,
+      "learning_rate": 6.833575237558692e-06,
+      "loss": 0.9587,
+      "step": 24511
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.258354902267456,
+      "learning_rate": 6.833329945023275e-06,
+      "loss": 0.6982,
+      "step": 24512
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.7285804748535156,
+      "learning_rate": 6.833084647390055e-06,
+      "loss": 0.4742,
+      "step": 24513
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.769876480102539,
+      "learning_rate": 6.832839344659712e-06,
+      "loss": 0.8314,
+      "step": 24514
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 4.645458221435547,
+      "learning_rate": 6.832594036832931e-06,
+      "loss": 0.6265,
+      "step": 24515
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 6.094618320465088,
+      "learning_rate": 6.8323487239103915e-06,
+      "loss": 0.7613,
+      "step": 24516
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.211217164993286,
+      "learning_rate": 6.832103405892777e-06,
+      "loss": 0.5698,
+      "step": 24517
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.844726800918579,
+      "learning_rate": 6.831858082780769e-06,
+      "loss": 0.6549,
+      "step": 24518
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 3.4754929542541504,
+      "learning_rate": 6.83161275457505e-06,
+      "loss": 0.6312,
+      "step": 24519
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.7747706174850464,
+      "learning_rate": 6.8313674212763e-06,
+      "loss": 0.7765,
+      "step": 24520
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.760920763015747,
+      "learning_rate": 6.831122082885206e-06,
+      "loss": 0.6318,
+      "step": 24521
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.8257296085357666,
+      "learning_rate": 6.830876739402448e-06,
+      "loss": 0.7283,
+      "step": 24522
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.2094736099243164,
+      "learning_rate": 6.830631390828706e-06,
+      "loss": 0.6174,
+      "step": 24523
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 4.3594279289245605,
+      "learning_rate": 6.830386037164663e-06,
+      "loss": 0.8282,
+      "step": 24524
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.838545799255371,
+      "learning_rate": 6.830140678411004e-06,
+      "loss": 0.5833,
+      "step": 24525
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 3.488757371902466,
+      "learning_rate": 6.829895314568408e-06,
+      "loss": 0.6838,
+      "step": 24526
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.4477059841156006,
+      "learning_rate": 6.82964994563756e-06,
+      "loss": 0.6321,
+      "step": 24527
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.673203945159912,
+      "learning_rate": 6.82940457161914e-06,
+      "loss": 0.7002,
+      "step": 24528
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.4818227291107178,
+      "learning_rate": 6.829159192513832e-06,
+      "loss": 0.7611,
+      "step": 24529
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.720585346221924,
+      "learning_rate": 6.828913808322317e-06,
+      "loss": 0.7284,
+      "step": 24530
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.9120378494262695,
+      "learning_rate": 6.828668419045277e-06,
+      "loss": 0.499,
+      "step": 24531
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 3.381690502166748,
+      "learning_rate": 6.828423024683396e-06,
+      "loss": 0.449,
+      "step": 24532
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.231722593307495,
+      "learning_rate": 6.828177625237355e-06,
+      "loss": 0.669,
+      "step": 24533
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.679637908935547,
+      "learning_rate": 6.827932220707838e-06,
+      "loss": 0.6168,
+      "step": 24534
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.7446603775024414,
+      "learning_rate": 6.827686811095525e-06,
+      "loss": 0.7454,
+      "step": 24535
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.342512369155884,
+      "learning_rate": 6.827441396401101e-06,
+      "loss": 0.6167,
+      "step": 24536
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.4231723546981812,
+      "learning_rate": 6.827195976625246e-06,
+      "loss": 0.7224,
+      "step": 24537
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.550833821296692,
+      "learning_rate": 6.826950551768644e-06,
+      "loss": 0.6467,
+      "step": 24538
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.1166651248931885,
+      "learning_rate": 6.8267051218319766e-06,
+      "loss": 0.7139,
+      "step": 24539
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 3.3331823348999023,
+      "learning_rate": 6.826459686815927e-06,
+      "loss": 0.7701,
+      "step": 24540
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.6964119672775269,
+      "learning_rate": 6.826214246721176e-06,
+      "loss": 0.8038,
+      "step": 24541
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.7643524408340454,
+      "learning_rate": 6.825968801548408e-06,
+      "loss": 0.7197,
+      "step": 24542
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.7871217727661133,
+      "learning_rate": 6.825723351298305e-06,
+      "loss": 0.873,
+      "step": 24543
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.891396999359131,
+      "learning_rate": 6.8254778959715486e-06,
+      "loss": 0.7082,
+      "step": 24544
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 3.369401693344116,
+      "learning_rate": 6.825232435568823e-06,
+      "loss": 0.7811,
+      "step": 24545
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.783696174621582,
+      "learning_rate": 6.8249869700908085e-06,
+      "loss": 0.6283,
+      "step": 24546
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.4845051765441895,
+      "learning_rate": 6.82474149953819e-06,
+      "loss": 0.7883,
+      "step": 24547
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.2035582065582275,
+      "learning_rate": 6.824496023911649e-06,
+      "loss": 0.4867,
+      "step": 24548
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 3.375690460205078,
+      "learning_rate": 6.8242505432118665e-06,
+      "loss": 0.6482,
+      "step": 24549
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.7630653381347656,
+      "learning_rate": 6.824005057439529e-06,
+      "loss": 0.6362,
+      "step": 24550
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 3.907939910888672,
+      "learning_rate": 6.823759566595315e-06,
+      "loss": 0.7039,
+      "step": 24551
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.9033074378967285,
+      "learning_rate": 6.8235140706799095e-06,
+      "loss": 0.7007,
+      "step": 24552
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.4369444847106934,
+      "learning_rate": 6.823268569693995e-06,
+      "loss": 0.5145,
+      "step": 24553
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.868838310241699,
+      "learning_rate": 6.823023063638252e-06,
+      "loss": 0.6138,
+      "step": 24554
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.1994521617889404,
+      "learning_rate": 6.8227775525133655e-06,
+      "loss": 0.738,
+      "step": 24555
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.9063822031021118,
+      "learning_rate": 6.822532036320018e-06,
+      "loss": 0.7333,
+      "step": 24556
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.8430612087249756,
+      "learning_rate": 6.822286515058891e-06,
+      "loss": 0.541,
+      "step": 24557
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.103663921356201,
+      "learning_rate": 6.8220409887306685e-06,
+      "loss": 0.4813,
+      "step": 24558
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.381279468536377,
+      "learning_rate": 6.821795457336032e-06,
+      "loss": 0.7118,
+      "step": 24559
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.346686601638794,
+      "learning_rate": 6.821549920875665e-06,
+      "loss": 0.5368,
+      "step": 24560
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.368720769882202,
+      "learning_rate": 6.82130437935025e-06,
+      "loss": 0.6947,
+      "step": 24561
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.5986775159835815,
+      "learning_rate": 6.82105883276047e-06,
+      "loss": 0.5432,
+      "step": 24562
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.2923543453216553,
+      "learning_rate": 6.820813281107007e-06,
+      "loss": 0.7192,
+      "step": 24563
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.4290015697479248,
+      "learning_rate": 6.8205677243905455e-06,
+      "loss": 0.5843,
+      "step": 24564
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.6727826595306396,
+      "learning_rate": 6.820322162611766e-06,
+      "loss": 0.4615,
+      "step": 24565
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.1054794788360596,
+      "learning_rate": 6.820076595771353e-06,
+      "loss": 0.7054,
+      "step": 24566
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.8062245845794678,
+      "learning_rate": 6.819831023869989e-06,
+      "loss": 0.7426,
+      "step": 24567
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.6012773513793945,
+      "learning_rate": 6.819585446908356e-06,
+      "loss": 0.6265,
+      "step": 24568
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.4215147495269775,
+      "learning_rate": 6.819339864887138e-06,
+      "loss": 0.5865,
+      "step": 24569
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 3.1412885189056396,
+      "learning_rate": 6.819094277807018e-06,
+      "loss": 0.531,
+      "step": 24570
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 3.1103005409240723,
+      "learning_rate": 6.818848685668677e-06,
+      "loss": 0.8228,
+      "step": 24571
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.8993558883666992,
+      "learning_rate": 6.818603088472799e-06,
+      "loss": 0.7273,
+      "step": 24572
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.2201905250549316,
+      "learning_rate": 6.818357486220068e-06,
+      "loss": 0.6545,
+      "step": 24573
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.8457739353179932,
+      "learning_rate": 6.818111878911165e-06,
+      "loss": 0.7888,
+      "step": 24574
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.460638999938965,
+      "learning_rate": 6.817866266546776e-06,
+      "loss": 0.599,
+      "step": 24575
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 3.1828315258026123,
+      "learning_rate": 6.81762064912758e-06,
+      "loss": 0.57,
+      "step": 24576
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.8530209064483643,
+      "learning_rate": 6.817375026654263e-06,
+      "loss": 0.6635,
+      "step": 24577
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.420440912246704,
+      "learning_rate": 6.817129399127506e-06,
+      "loss": 0.6306,
+      "step": 24578
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 3.3244903087615967,
+      "learning_rate": 6.8168837665479936e-06,
+      "loss": 0.8507,
+      "step": 24579
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.782252311706543,
+      "learning_rate": 6.816638128916408e-06,
+      "loss": 0.7114,
+      "step": 24580
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.7698490619659424,
+      "learning_rate": 6.816392486233431e-06,
+      "loss": 0.5067,
+      "step": 24581
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 3.1977932453155518,
+      "learning_rate": 6.8161468384997475e-06,
+      "loss": 0.6695,
+      "step": 24582
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 3.174830436706543,
+      "learning_rate": 6.815901185716041e-06,
+      "loss": 0.6203,
+      "step": 24583
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 3.0920419692993164,
+      "learning_rate": 6.815655527882993e-06,
+      "loss": 0.5878,
+      "step": 24584
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.9852319955825806,
+      "learning_rate": 6.815409865001287e-06,
+      "loss": 0.5146,
+      "step": 24585
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.0276196002960205,
+      "learning_rate": 6.8151641970716065e-06,
+      "loss": 0.5645,
+      "step": 24586
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 4.362538814544678,
+      "learning_rate": 6.8149185240946335e-06,
+      "loss": 0.4362,
+      "step": 24587
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.9252688884735107,
+      "learning_rate": 6.8146728460710535e-06,
+      "loss": 0.748,
+      "step": 24588
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.214970111846924,
+      "learning_rate": 6.8144271630015466e-06,
+      "loss": 0.7834,
+      "step": 24589
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.575834035873413,
+      "learning_rate": 6.814181474886799e-06,
+      "loss": 0.5546,
+      "step": 24590
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.7427316904067993,
+      "learning_rate": 6.8139357817274905e-06,
+      "loss": 0.7862,
+      "step": 24591
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.038316249847412,
+      "learning_rate": 6.813690083524307e-06,
+      "loss": 0.4388,
+      "step": 24592
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.188256025314331,
+      "learning_rate": 6.8134443802779305e-06,
+      "loss": 0.6316,
+      "step": 24593
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.8266687393188477,
+      "learning_rate": 6.813198671989045e-06,
+      "loss": 0.7154,
+      "step": 24594
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.6023054122924805,
+      "learning_rate": 6.812952958658333e-06,
+      "loss": 0.4491,
+      "step": 24595
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.0616044998168945,
+      "learning_rate": 6.812707240286478e-06,
+      "loss": 0.8885,
+      "step": 24596
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.17802095413208,
+      "learning_rate": 6.8124615168741636e-06,
+      "loss": 0.6579,
+      "step": 24597
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.6958485841751099,
+      "learning_rate": 6.812215788422072e-06,
+      "loss": 0.5427,
+      "step": 24598
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.8264541625976562,
+      "learning_rate": 6.811970054930888e-06,
+      "loss": 0.6046,
+      "step": 24599
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.175161361694336,
+      "learning_rate": 6.811724316401294e-06,
+      "loss": 0.5778,
+      "step": 24600
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.099382162094116,
+      "learning_rate": 6.811478572833973e-06,
+      "loss": 0.6946,
+      "step": 24601
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.8639967441558838,
+      "learning_rate": 6.8112328242296075e-06,
+      "loss": 0.74,
+      "step": 24602
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.896437406539917,
+      "learning_rate": 6.810987070588882e-06,
+      "loss": 0.5412,
+      "step": 24603
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.5539638996124268,
+      "learning_rate": 6.810741311912482e-06,
+      "loss": 0.6735,
+      "step": 24604
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.5465736389160156,
+      "learning_rate": 6.810495548201088e-06,
+      "loss": 0.7487,
+      "step": 24605
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.8028573989868164,
+      "learning_rate": 6.810249779455383e-06,
+      "loss": 0.8349,
+      "step": 24606
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.078744649887085,
+      "learning_rate": 6.810004005676053e-06,
+      "loss": 0.622,
+      "step": 24607
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.553554892539978,
+      "learning_rate": 6.80975822686378e-06,
+      "loss": 0.7449,
+      "step": 24608
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 3.2509748935699463,
+      "learning_rate": 6.809512443019246e-06,
+      "loss": 0.5376,
+      "step": 24609
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.4521701335906982,
+      "learning_rate": 6.809266654143136e-06,
+      "loss": 0.7812,
+      "step": 24610
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 3.313096523284912,
+      "learning_rate": 6.809020860236133e-06,
+      "loss": 0.6839,
+      "step": 24611
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 4.136160850524902,
+      "learning_rate": 6.808775061298921e-06,
+      "loss": 0.9524,
+      "step": 24612
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.1032559871673584,
+      "learning_rate": 6.808529257332184e-06,
+      "loss": 0.7928,
+      "step": 24613
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 4.57716703414917,
+      "learning_rate": 6.808283448336603e-06,
+      "loss": 0.715,
+      "step": 24614
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.9090790748596191,
+      "learning_rate": 6.808037634312865e-06,
+      "loss": 0.8957,
+      "step": 24615
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.335061550140381,
+      "learning_rate": 6.807791815261649e-06,
+      "loss": 0.7759,
+      "step": 24616
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.646564245223999,
+      "learning_rate": 6.807545991183644e-06,
+      "loss": 0.7193,
+      "step": 24617
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.6514008045196533,
+      "learning_rate": 6.807300162079528e-06,
+      "loss": 0.6557,
+      "step": 24618
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.5516109466552734,
+      "learning_rate": 6.80705432794999e-06,
+      "loss": 0.6654,
+      "step": 24619
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.9293618202209473,
+      "learning_rate": 6.80680848879571e-06,
+      "loss": 0.5667,
+      "step": 24620
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.3093996047973633,
+      "learning_rate": 6.8065626446173725e-06,
+      "loss": 0.789,
+      "step": 24621
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 4.8188557624816895,
+      "learning_rate": 6.80631679541566e-06,
+      "loss": 0.4638,
+      "step": 24622
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.117839813232422,
+      "learning_rate": 6.806070941191257e-06,
+      "loss": 0.4718,
+      "step": 24623
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.3041181564331055,
+      "learning_rate": 6.805825081944849e-06,
+      "loss": 0.4796,
+      "step": 24624
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 3.0808982849121094,
+      "learning_rate": 6.8055792176771164e-06,
+      "loss": 0.5286,
+      "step": 24625
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.9912145137786865,
+      "learning_rate": 6.805333348388745e-06,
+      "loss": 0.8925,
+      "step": 24626
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.5367766618728638,
+      "learning_rate": 6.8050874740804185e-06,
+      "loss": 0.6996,
+      "step": 24627
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.6224005222320557,
+      "learning_rate": 6.80484159475282e-06,
+      "loss": 0.5601,
+      "step": 24628
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.6336922645568848,
+      "learning_rate": 6.804595710406632e-06,
+      "loss": 0.4952,
+      "step": 24629
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.9465103149414062,
+      "learning_rate": 6.804349821042539e-06,
+      "loss": 0.5049,
+      "step": 24630
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.35333514213562,
+      "learning_rate": 6.804103926661228e-06,
+      "loss": 0.5298,
+      "step": 24631
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.640756845474243,
+      "learning_rate": 6.803858027263376e-06,
+      "loss": 0.5416,
+      "step": 24632
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.028285026550293,
+      "learning_rate": 6.803612122849672e-06,
+      "loss": 0.4532,
+      "step": 24633
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.104505777359009,
+      "learning_rate": 6.803366213420799e-06,
+      "loss": 0.6535,
+      "step": 24634
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 3.2074596881866455,
+      "learning_rate": 6.8031202989774395e-06,
+      "loss": 0.6959,
+      "step": 24635
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.5335676670074463,
+      "learning_rate": 6.802874379520279e-06,
+      "loss": 0.5298,
+      "step": 24636
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.826326370239258,
+      "learning_rate": 6.802628455049999e-06,
+      "loss": 0.801,
+      "step": 24637
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 3.6443932056427,
+      "learning_rate": 6.8023825255672846e-06,
+      "loss": 0.5452,
+      "step": 24638
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 5.2939300537109375,
+      "learning_rate": 6.80213659107282e-06,
+      "loss": 0.5097,
+      "step": 24639
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.8741447925567627,
+      "learning_rate": 6.801890651567289e-06,
+      "loss": 0.7863,
+      "step": 24640
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 3.4880964756011963,
+      "learning_rate": 6.801644707051375e-06,
+      "loss": 0.7338,
+      "step": 24641
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.602397918701172,
+      "learning_rate": 6.80139875752576e-06,
+      "loss": 0.6622,
+      "step": 24642
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 4.369588851928711,
+      "learning_rate": 6.801152802991131e-06,
+      "loss": 0.4689,
+      "step": 24643
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 3.989502191543579,
+      "learning_rate": 6.800906843448172e-06,
+      "loss": 0.945,
+      "step": 24644
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 3.4909486770629883,
+      "learning_rate": 6.800660878897565e-06,
+      "loss": 0.6949,
+      "step": 24645
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 3.224749803543091,
+      "learning_rate": 6.8004149093399955e-06,
+      "loss": 0.6724,
+      "step": 24646
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 4.511003017425537,
+      "learning_rate": 6.8001689347761455e-06,
+      "loss": 0.4611,
+      "step": 24647
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.7169044017791748,
+      "learning_rate": 6.7999229552067e-06,
+      "loss": 0.3557,
+      "step": 24648
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.3053138256073,
+      "learning_rate": 6.799676970632343e-06,
+      "loss": 0.6033,
+      "step": 24649
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.6387813091278076,
+      "learning_rate": 6.79943098105376e-06,
+      "loss": 0.5429,
+      "step": 24650
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.935682773590088,
+      "learning_rate": 6.799184986471632e-06,
+      "loss": 0.745,
+      "step": 24651
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.6106932163238525,
+      "learning_rate": 6.798938986886645e-06,
+      "loss": 0.59,
+      "step": 24652
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.9458222389221191,
+      "learning_rate": 6.798692982299482e-06,
+      "loss": 0.8309,
+      "step": 24653
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.575258255004883,
+      "learning_rate": 6.798446972710827e-06,
+      "loss": 0.532,
+      "step": 24654
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.9288673400878906,
+      "learning_rate": 6.798200958121366e-06,
+      "loss": 0.5398,
+      "step": 24655
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.8330318927764893,
+      "learning_rate": 6.7979549385317815e-06,
+      "loss": 0.5423,
+      "step": 24656
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.2930004596710205,
+      "learning_rate": 6.797708913942758e-06,
+      "loss": 0.6808,
+      "step": 24657
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.321308135986328,
+      "learning_rate": 6.79746288435498e-06,
+      "loss": 0.6265,
+      "step": 24658
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.9355435371398926,
+      "learning_rate": 6.797216849769131e-06,
+      "loss": 0.5146,
+      "step": 24659
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.1103386878967285,
+      "learning_rate": 6.796970810185893e-06,
+      "loss": 0.6239,
+      "step": 24660
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 4.771523475646973,
+      "learning_rate": 6.796724765605954e-06,
+      "loss": 0.562,
+      "step": 24661
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.1698694229125977,
+      "learning_rate": 6.796478716029997e-06,
+      "loss": 0.7371,
+      "step": 24662
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.7676169872283936,
+      "learning_rate": 6.796232661458704e-06,
+      "loss": 0.6433,
+      "step": 24663
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.955643892288208,
+      "learning_rate": 6.795986601892762e-06,
+      "loss": 0.5747,
+      "step": 24664
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 4.3075103759765625,
+      "learning_rate": 6.795740537332854e-06,
+      "loss": 0.9317,
+      "step": 24665
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 4.524974822998047,
+      "learning_rate": 6.795494467779664e-06,
+      "loss": 0.597,
+      "step": 24666
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.5104191303253174,
+      "learning_rate": 6.795248393233878e-06,
+      "loss": 0.5759,
+      "step": 24667
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.0915067195892334,
+      "learning_rate": 6.795002313696178e-06,
+      "loss": 0.4837,
+      "step": 24668
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.6952545642852783,
+      "learning_rate": 6.794756229167248e-06,
+      "loss": 0.3944,
+      "step": 24669
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 3.2414636611938477,
+      "learning_rate": 6.794510139647773e-06,
+      "loss": 0.8483,
+      "step": 24670
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 4.401352405548096,
+      "learning_rate": 6.794264045138439e-06,
+      "loss": 0.6695,
+      "step": 24671
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.394314765930176,
+      "learning_rate": 6.7940179456399265e-06,
+      "loss": 0.6271,
+      "step": 24672
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 3.469022274017334,
+      "learning_rate": 6.793771841152924e-06,
+      "loss": 0.7061,
+      "step": 24673
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.1557424068450928,
+      "learning_rate": 6.793525731678113e-06,
+      "loss": 0.7781,
+      "step": 24674
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 3.083348274230957,
+      "learning_rate": 6.79327961721618e-06,
+      "loss": 0.3712,
+      "step": 24675
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.3794360160827637,
+      "learning_rate": 6.793033497767808e-06,
+      "loss": 0.7785,
+      "step": 24676
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 3.596817970275879,
+      "learning_rate": 6.792787373333681e-06,
+      "loss": 0.6113,
+      "step": 24677
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 3.3833868503570557,
+      "learning_rate": 6.7925412439144834e-06,
+      "loss": 0.5314,
+      "step": 24678
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.9193954467773438,
+      "learning_rate": 6.792295109510902e-06,
+      "loss": 0.8405,
+      "step": 24679
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 3.1098999977111816,
+      "learning_rate": 6.792048970123617e-06,
+      "loss": 0.9468,
+      "step": 24680
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.007444381713867,
+      "learning_rate": 6.791802825753317e-06,
+      "loss": 0.5751,
+      "step": 24681
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.698025941848755,
+      "learning_rate": 6.791556676400683e-06,
+      "loss": 0.6888,
+      "step": 24682
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.131448268890381,
+      "learning_rate": 6.791310522066402e-06,
+      "loss": 0.5839,
+      "step": 24683
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.3785037994384766,
+      "learning_rate": 6.791064362751157e-06,
+      "loss": 0.5543,
+      "step": 24684
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.934502363204956,
+      "learning_rate": 6.790818198455633e-06,
+      "loss": 0.7658,
+      "step": 24685
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.9759061336517334,
+      "learning_rate": 6.790572029180515e-06,
+      "loss": 0.8131,
+      "step": 24686
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.211751699447632,
+      "learning_rate": 6.790325854926487e-06,
+      "loss": 0.7856,
+      "step": 24687
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.4434709548950195,
+      "learning_rate": 6.790079675694232e-06,
+      "loss": 0.7334,
+      "step": 24688
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.5625633001327515,
+      "learning_rate": 6.789833491484438e-06,
+      "loss": 0.5525,
+      "step": 24689
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.2762722969055176,
+      "learning_rate": 6.789587302297786e-06,
+      "loss": 0.6721,
+      "step": 24690
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.38271164894104,
+      "learning_rate": 6.789341108134964e-06,
+      "loss": 0.3939,
+      "step": 24691
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.804452896118164,
+      "learning_rate": 6.7890949089966525e-06,
+      "loss": 0.5555,
+      "step": 24692
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.6526341438293457,
+      "learning_rate": 6.78884870488354e-06,
+      "loss": 0.6696,
+      "step": 24693
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.7824409008026123,
+      "learning_rate": 6.788602495796308e-06,
+      "loss": 0.615,
+      "step": 24694
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.914039134979248,
+      "learning_rate": 6.788356281735643e-06,
+      "loss": 0.5612,
+      "step": 24695
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.246311902999878,
+      "learning_rate": 6.788110062702229e-06,
+      "loss": 0.6421,
+      "step": 24696
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.9248396158218384,
+      "learning_rate": 6.787863838696751e-06,
+      "loss": 0.4812,
+      "step": 24697
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 3.257117748260498,
+      "learning_rate": 6.787617609719894e-06,
+      "loss": 0.9495,
+      "step": 24698
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.0476343631744385,
+      "learning_rate": 6.787371375772341e-06,
+      "loss": 0.6924,
+      "step": 24699
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.944503664970398,
+      "learning_rate": 6.7871251368547785e-06,
+      "loss": 0.8363,
+      "step": 24700
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.5863858461380005,
+      "learning_rate": 6.78687889296789e-06,
+      "loss": 0.4197,
+      "step": 24701
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.3815350532531738,
+      "learning_rate": 6.786632644112361e-06,
+      "loss": 0.5723,
+      "step": 24702
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.6145061254501343,
+      "learning_rate": 6.7863863902888746e-06,
+      "loss": 0.4646,
+      "step": 24703
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.233161687850952,
+      "learning_rate": 6.786140131498118e-06,
+      "loss": 0.4465,
+      "step": 24704
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.300013303756714,
+      "learning_rate": 6.785893867740776e-06,
+      "loss": 0.4935,
+      "step": 24705
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.458005905151367,
+      "learning_rate": 6.7856475990175305e-06,
+      "loss": 0.5646,
+      "step": 24706
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 4.923053741455078,
+      "learning_rate": 6.785401325329069e-06,
+      "loss": 0.7733,
+      "step": 24707
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.7594032287597656,
+      "learning_rate": 6.785155046676074e-06,
+      "loss": 0.7341,
+      "step": 24708
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.8721433877944946,
+      "learning_rate": 6.784908763059233e-06,
+      "loss": 0.5984,
+      "step": 24709
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.4561991691589355,
+      "learning_rate": 6.784662474479229e-06,
+      "loss": 0.6386,
+      "step": 24710
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.8700653314590454,
+      "learning_rate": 6.784416180936746e-06,
+      "loss": 0.4965,
+      "step": 24711
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.242515802383423,
+      "learning_rate": 6.7841698824324705e-06,
+      "loss": 0.6492,
+      "step": 24712
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 3.3310420513153076,
+      "learning_rate": 6.783923578967087e-06,
+      "loss": 0.6353,
+      "step": 24713
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.4257888793945312,
+      "learning_rate": 6.78367727054128e-06,
+      "loss": 0.7997,
+      "step": 24714
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 3.768700361251831,
+      "learning_rate": 6.783430957155735e-06,
+      "loss": 0.8966,
+      "step": 24715
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.4450137615203857,
+      "learning_rate": 6.7831846388111375e-06,
+      "loss": 0.4788,
+      "step": 24716
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.9507561922073364,
+      "learning_rate": 6.7829383155081716e-06,
+      "loss": 0.3796,
+      "step": 24717
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.8014984130859375,
+      "learning_rate": 6.782691987247521e-06,
+      "loss": 0.6894,
+      "step": 24718
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 3.9625377655029297,
+      "learning_rate": 6.782445654029873e-06,
+      "loss": 0.7563,
+      "step": 24719
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.1782310009002686,
+      "learning_rate": 6.782199315855911e-06,
+      "loss": 0.7744,
+      "step": 24720
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 3.6960530281066895,
+      "learning_rate": 6.781952972726319e-06,
+      "loss": 0.3101,
+      "step": 24721
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.4821139574050903,
+      "learning_rate": 6.781706624641784e-06,
+      "loss": 0.6291,
+      "step": 24722
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.803889513015747,
+      "learning_rate": 6.781460271602991e-06,
+      "loss": 0.7161,
+      "step": 24723
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 3.043485403060913,
+      "learning_rate": 6.781213913610622e-06,
+      "loss": 0.5837,
+      "step": 24724
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.2219913005828857,
+      "learning_rate": 6.7809675506653674e-06,
+      "loss": 0.7796,
+      "step": 24725
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.283568859100342,
+      "learning_rate": 6.780721182767908e-06,
+      "loss": 0.4651,
+      "step": 24726
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.0638797283172607,
+      "learning_rate": 6.780474809918931e-06,
+      "loss": 0.9784,
+      "step": 24727
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.104712724685669,
+      "learning_rate": 6.780228432119119e-06,
+      "loss": 0.6652,
+      "step": 24728
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.9077517986297607,
+      "learning_rate": 6.779982049369159e-06,
+      "loss": 0.5927,
+      "step": 24729
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.556021809577942,
+      "learning_rate": 6.779735661669736e-06,
+      "loss": 0.5962,
+      "step": 24730
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.606140375137329,
+      "learning_rate": 6.779489269021535e-06,
+      "loss": 0.7633,
+      "step": 24731
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.1338839530944824,
+      "learning_rate": 6.7792428714252415e-06,
+      "loss": 0.7088,
+      "step": 24732
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.8887910842895508,
+      "learning_rate": 6.778996468881539e-06,
+      "loss": 0.4929,
+      "step": 24733
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.6519994735717773,
+      "learning_rate": 6.778750061391113e-06,
+      "loss": 0.6738,
+      "step": 24734
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.6339070796966553,
+      "learning_rate": 6.778503648954651e-06,
+      "loss": 0.8151,
+      "step": 24735
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 4.582214832305908,
+      "learning_rate": 6.778257231572836e-06,
+      "loss": 0.6234,
+      "step": 24736
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.5471808910369873,
+      "learning_rate": 6.7780108092463535e-06,
+      "loss": 0.7429,
+      "step": 24737
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.6828078031539917,
+      "learning_rate": 6.7777643819758895e-06,
+      "loss": 0.5253,
+      "step": 24738
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.0348284244537354,
+      "learning_rate": 6.7775179497621294e-06,
+      "loss": 0.6599,
+      "step": 24739
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 3.168656587600708,
+      "learning_rate": 6.777271512605757e-06,
+      "loss": 0.8162,
+      "step": 24740
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.3387248516082764,
+      "learning_rate": 6.777025070507457e-06,
+      "loss": 0.8107,
+      "step": 24741
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 3.6847333908081055,
+      "learning_rate": 6.776778623467918e-06,
+      "loss": 0.3765,
+      "step": 24742
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.6850221157073975,
+      "learning_rate": 6.776532171487822e-06,
+      "loss": 0.5904,
+      "step": 24743
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.7110753059387207,
+      "learning_rate": 6.776285714567856e-06,
+      "loss": 0.4223,
+      "step": 24744
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 3.467331647872925,
+      "learning_rate": 6.776039252708706e-06,
+      "loss": 0.6326,
+      "step": 24745
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.3550455570220947,
+      "learning_rate": 6.775792785911054e-06,
+      "loss": 0.8505,
+      "step": 24746
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.5104572772979736,
+      "learning_rate": 6.775546314175588e-06,
+      "loss": 0.5929,
+      "step": 24747
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.7394955158233643,
+      "learning_rate": 6.775299837502993e-06,
+      "loss": 0.686,
+      "step": 24748
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 3.4877305030822754,
+      "learning_rate": 6.775053355893954e-06,
+      "loss": 0.8029,
+      "step": 24749
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.6323609352111816,
+      "learning_rate": 6.774806869349159e-06,
+      "loss": 0.7056,
+      "step": 24750
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.0183897018432617,
+      "learning_rate": 6.774560377869288e-06,
+      "loss": 0.6381,
+      "step": 24751
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.9690299034118652,
+      "learning_rate": 6.7743138814550315e-06,
+      "loss": 0.4165,
+      "step": 24752
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.069078207015991,
+      "learning_rate": 6.774067380107071e-06,
+      "loss": 0.5959,
+      "step": 24753
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.1037778854370117,
+      "learning_rate": 6.7738208738260935e-06,
+      "loss": 0.7392,
+      "step": 24754
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.1617870330810547,
+      "learning_rate": 6.773574362612786e-06,
+      "loss": 0.7074,
+      "step": 24755
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 3.538463592529297,
+      "learning_rate": 6.773327846467832e-06,
+      "loss": 0.6283,
+      "step": 24756
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.8489412069320679,
+      "learning_rate": 6.773081325391918e-06,
+      "loss": 0.5538,
+      "step": 24757
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.786651372909546,
+      "learning_rate": 6.772834799385729e-06,
+      "loss": 0.7498,
+      "step": 24758
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.1100053787231445,
+      "learning_rate": 6.77258826844995e-06,
+      "loss": 0.4776,
+      "step": 24759
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.9941792488098145,
+      "learning_rate": 6.772341732585268e-06,
+      "loss": 0.6707,
+      "step": 24760
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.568655014038086,
+      "learning_rate": 6.772095191792367e-06,
+      "loss": 0.5827,
+      "step": 24761
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.6978037357330322,
+      "learning_rate": 6.771848646071932e-06,
+      "loss": 0.6443,
+      "step": 24762
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.79508376121521,
+      "learning_rate": 6.771602095424652e-06,
+      "loss": 0.6908,
+      "step": 24763
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.6558592319488525,
+      "learning_rate": 6.771355539851208e-06,
+      "loss": 0.7245,
+      "step": 24764
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 4.3418192863464355,
+      "learning_rate": 6.771108979352288e-06,
+      "loss": 0.9083,
+      "step": 24765
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.877334713935852,
+      "learning_rate": 6.770862413928578e-06,
+      "loss": 0.6609,
+      "step": 24766
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 3.4642882347106934,
+      "learning_rate": 6.770615843580763e-06,
+      "loss": 0.5675,
+      "step": 24767
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.4877088069915771,
+      "learning_rate": 6.770369268309529e-06,
+      "loss": 0.4949,
+      "step": 24768
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 3.3841640949249268,
+      "learning_rate": 6.7701226881155615e-06,
+      "loss": 0.6716,
+      "step": 24769
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.9971261024475098,
+      "learning_rate": 6.7698761029995456e-06,
+      "loss": 0.7023,
+      "step": 24770
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.6432119607925415,
+      "learning_rate": 6.769629512962167e-06,
+      "loss": 0.6525,
+      "step": 24771
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.2586028575897217,
+      "learning_rate": 6.769382918004112e-06,
+      "loss": 0.8428,
+      "step": 24772
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.816449761390686,
+      "learning_rate": 6.769136318126066e-06,
+      "loss": 0.467,
+      "step": 24773
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.5115303993225098,
+      "learning_rate": 6.768889713328715e-06,
+      "loss": 0.574,
+      "step": 24774
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.6493310928344727,
+      "learning_rate": 6.768643103612742e-06,
+      "loss": 0.6468,
+      "step": 24775
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.400766134262085,
+      "learning_rate": 6.768396488978837e-06,
+      "loss": 0.4963,
+      "step": 24776
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.673454523086548,
+      "learning_rate": 6.768149869427684e-06,
+      "loss": 0.6215,
+      "step": 24777
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.9862520694732666,
+      "learning_rate": 6.767903244959968e-06,
+      "loss": 0.6992,
+      "step": 24778
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.6307580471038818,
+      "learning_rate": 6.767656615576375e-06,
+      "loss": 0.6602,
+      "step": 24779
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.7113085985183716,
+      "learning_rate": 6.767409981277592e-06,
+      "loss": 0.4263,
+      "step": 24780
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.0998129844665527,
+      "learning_rate": 6.767163342064304e-06,
+      "loss": 0.8181,
+      "step": 24781
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.5532708168029785,
+      "learning_rate": 6.766916697937196e-06,
+      "loss": 0.9127,
+      "step": 24782
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.810777425765991,
+      "learning_rate": 6.766670048896955e-06,
+      "loss": 0.6117,
+      "step": 24783
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.8208634853363037,
+      "learning_rate": 6.766423394944266e-06,
+      "loss": 0.5511,
+      "step": 24784
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.7408010959625244,
+      "learning_rate": 6.766176736079815e-06,
+      "loss": 0.6731,
+      "step": 24785
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.778777837753296,
+      "learning_rate": 6.765930072304289e-06,
+      "loss": 0.6799,
+      "step": 24786
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.6790605783462524,
+      "learning_rate": 6.765683403618372e-06,
+      "loss": 0.6232,
+      "step": 24787
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.0085456371307373,
+      "learning_rate": 6.765436730022752e-06,
+      "loss": 0.5945,
+      "step": 24788
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.8868200778961182,
+      "learning_rate": 6.765190051518114e-06,
+      "loss": 0.5932,
+      "step": 24789
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.5021424293518066,
+      "learning_rate": 6.764943368105142e-06,
+      "loss": 0.533,
+      "step": 24790
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.8600213527679443,
+      "learning_rate": 6.764696679784525e-06,
+      "loss": 0.7826,
+      "step": 24791
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.8652021884918213,
+      "learning_rate": 6.764449986556947e-06,
+      "loss": 0.8308,
+      "step": 24792
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.738281488418579,
+      "learning_rate": 6.764203288423094e-06,
+      "loss": 0.5299,
+      "step": 24793
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.9801833629608154,
+      "learning_rate": 6.7639565853836554e-06,
+      "loss": 0.7409,
+      "step": 24794
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.6391985416412354,
+      "learning_rate": 6.7637098774393105e-06,
+      "loss": 0.3367,
+      "step": 24795
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.813422441482544,
+      "learning_rate": 6.763463164590751e-06,
+      "loss": 0.5814,
+      "step": 24796
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.2637486457824707,
+      "learning_rate": 6.7632164468386605e-06,
+      "loss": 0.5359,
+      "step": 24797
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.4192216396331787,
+      "learning_rate": 6.7629697241837275e-06,
+      "loss": 0.5677,
+      "step": 24798
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.326031446456909,
+      "learning_rate": 6.762722996626634e-06,
+      "loss": 0.5905,
+      "step": 24799
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.9328054189682007,
+      "learning_rate": 6.7624762641680695e-06,
+      "loss": 0.6877,
+      "step": 24800
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.7273707389831543,
+      "learning_rate": 6.7622295268087165e-06,
+      "loss": 0.8327,
+      "step": 24801
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.626732110977173,
+      "learning_rate": 6.761982784549266e-06,
+      "loss": 0.7262,
+      "step": 24802
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.49127197265625,
+      "learning_rate": 6.761736037390401e-06,
+      "loss": 0.5856,
+      "step": 24803
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.620295524597168,
+      "learning_rate": 6.761489285332806e-06,
+      "loss": 0.5046,
+      "step": 24804
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.8609817028045654,
+      "learning_rate": 6.761242528377171e-06,
+      "loss": 0.8878,
+      "step": 24805
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.9640345573425293,
+      "learning_rate": 6.760995766524181e-06,
+      "loss": 0.4733,
+      "step": 24806
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.332370400428772,
+      "learning_rate": 6.760748999774519e-06,
+      "loss": 0.6078,
+      "step": 24807
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.8540303707122803,
+      "learning_rate": 6.760502228128874e-06,
+      "loss": 0.5811,
+      "step": 24808
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 3.0281357765197754,
+      "learning_rate": 6.760255451587934e-06,
+      "loss": 0.4418,
+      "step": 24809
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.1814489364624023,
+      "learning_rate": 6.760008670152381e-06,
+      "loss": 0.5132,
+      "step": 24810
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.2172927856445312,
+      "learning_rate": 6.759761883822904e-06,
+      "loss": 0.7078,
+      "step": 24811
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.1257898807525635,
+      "learning_rate": 6.759515092600189e-06,
+      "loss": 0.6011,
+      "step": 24812
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.8507161140441895,
+      "learning_rate": 6.759268296484921e-06,
+      "loss": 0.5891,
+      "step": 24813
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.8989672660827637,
+      "learning_rate": 6.759021495477788e-06,
+      "loss": 0.8367,
+      "step": 24814
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 4.644377708435059,
+      "learning_rate": 6.7587746895794726e-06,
+      "loss": 0.9065,
+      "step": 24815
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.9436943531036377,
+      "learning_rate": 6.758527878790665e-06,
+      "loss": 0.5059,
+      "step": 24816
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.689960241317749,
+      "learning_rate": 6.758281063112051e-06,
+      "loss": 0.5256,
+      "step": 24817
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.443295955657959,
+      "learning_rate": 6.758034242544316e-06,
+      "loss": 0.6482,
+      "step": 24818
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 3.7434604167938232,
+      "learning_rate": 6.757787417088145e-06,
+      "loss": 0.6742,
+      "step": 24819
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.6968330144882202,
+      "learning_rate": 6.757540586744227e-06,
+      "loss": 0.6356,
+      "step": 24820
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.2694945335388184,
+      "learning_rate": 6.757293751513246e-06,
+      "loss": 0.7336,
+      "step": 24821
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.3244919776916504,
+      "learning_rate": 6.75704691139589e-06,
+      "loss": 0.8514,
+      "step": 24822
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.496429443359375,
+      "learning_rate": 6.756800066392845e-06,
+      "loss": 0.6396,
+      "step": 24823
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.837651252746582,
+      "learning_rate": 6.756553216504796e-06,
+      "loss": 0.7659,
+      "step": 24824
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.9496021270751953,
+      "learning_rate": 6.75630636173243e-06,
+      "loss": 0.5264,
+      "step": 24825
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 3.5137765407562256,
+      "learning_rate": 6.756059502076435e-06,
+      "loss": 0.8018,
+      "step": 24826
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.9096273183822632,
+      "learning_rate": 6.755812637537496e-06,
+      "loss": 0.7576,
+      "step": 24827
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.2705326080322266,
+      "learning_rate": 6.7555657681163015e-06,
+      "loss": 0.5906,
+      "step": 24828
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.032541513442993,
+      "learning_rate": 6.755318893813535e-06,
+      "loss": 0.4146,
+      "step": 24829
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 5.322177410125732,
+      "learning_rate": 6.7550720146298844e-06,
+      "loss": 0.5574,
+      "step": 24830
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.0271687507629395,
+      "learning_rate": 6.754825130566036e-06,
+      "loss": 0.7399,
+      "step": 24831
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.312324047088623,
+      "learning_rate": 6.754578241622676e-06,
+      "loss": 0.6641,
+      "step": 24832
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.172414541244507,
+      "learning_rate": 6.754331347800491e-06,
+      "loss": 0.731,
+      "step": 24833
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 3.383974552154541,
+      "learning_rate": 6.754084449100168e-06,
+      "loss": 0.757,
+      "step": 24834
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 3.870549201965332,
+      "learning_rate": 6.753837545522394e-06,
+      "loss": 0.6843,
+      "step": 24835
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 4.911535739898682,
+      "learning_rate": 6.753590637067853e-06,
+      "loss": 0.5814,
+      "step": 24836
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.890566349029541,
+      "learning_rate": 6.7533437237372334e-06,
+      "loss": 0.783,
+      "step": 24837
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.7660714387893677,
+      "learning_rate": 6.7530968055312226e-06,
+      "loss": 0.6032,
+      "step": 24838
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.8847732543945312,
+      "learning_rate": 6.7528498824505075e-06,
+      "loss": 0.7247,
+      "step": 24839
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.7692978382110596,
+      "learning_rate": 6.7526029544957725e-06,
+      "loss": 0.5611,
+      "step": 24840
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.8770297765731812,
+      "learning_rate": 6.7523560216677054e-06,
+      "loss": 0.4999,
+      "step": 24841
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.3413329124450684,
+      "learning_rate": 6.7521090839669924e-06,
+      "loss": 0.7563,
+      "step": 24842
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 3.055422782897949,
+      "learning_rate": 6.751862141394321e-06,
+      "loss": 0.5603,
+      "step": 24843
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.2237160205841064,
+      "learning_rate": 6.751615193950376e-06,
+      "loss": 0.7877,
+      "step": 24844
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.405677080154419,
+      "learning_rate": 6.751368241635846e-06,
+      "loss": 0.6717,
+      "step": 24845
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.8138492107391357,
+      "learning_rate": 6.751121284451416e-06,
+      "loss": 0.6927,
+      "step": 24846
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.0257363319396973,
+      "learning_rate": 6.750874322397775e-06,
+      "loss": 0.5217,
+      "step": 24847
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.7906187772750854,
+      "learning_rate": 6.750627355475608e-06,
+      "loss": 0.67,
+      "step": 24848
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.6364076137542725,
+      "learning_rate": 6.7503803836856015e-06,
+      "loss": 0.5839,
+      "step": 24849
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.1598429679870605,
+      "learning_rate": 6.750133407028444e-06,
+      "loss": 0.5318,
+      "step": 24850
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 3.887377977371216,
+      "learning_rate": 6.749886425504821e-06,
+      "loss": 0.7425,
+      "step": 24851
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.9626340866088867,
+      "learning_rate": 6.7496394391154186e-06,
+      "loss": 0.5924,
+      "step": 24852
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.3244820833206177,
+      "learning_rate": 6.7493924478609255e-06,
+      "loss": 0.6054,
+      "step": 24853
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.4763853549957275,
+      "learning_rate": 6.749145451742026e-06,
+      "loss": 0.5548,
+      "step": 24854
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.6192214488983154,
+      "learning_rate": 6.748898450759409e-06,
+      "loss": 0.7014,
+      "step": 24855
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.728814721107483,
+      "learning_rate": 6.7486514449137596e-06,
+      "loss": 0.5498,
+      "step": 24856
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.5291740894317627,
+      "learning_rate": 6.748404434205767e-06,
+      "loss": 0.7674,
+      "step": 24857
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.176205635070801,
+      "learning_rate": 6.748157418636115e-06,
+      "loss": 0.6216,
+      "step": 24858
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 4.033459186553955,
+      "learning_rate": 6.747910398205494e-06,
+      "loss": 0.6802,
+      "step": 24859
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.643404245376587,
+      "learning_rate": 6.7476633729145875e-06,
+      "loss": 0.5498,
+      "step": 24860
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.208043098449707,
+      "learning_rate": 6.747416342764085e-06,
+      "loss": 0.5773,
+      "step": 24861
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 4.261011123657227,
+      "learning_rate": 6.747169307754672e-06,
+      "loss": 0.666,
+      "step": 24862
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.221311092376709,
+      "learning_rate": 6.746922267887034e-06,
+      "loss": 0.6333,
+      "step": 24863
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.2994436025619507,
+      "learning_rate": 6.746675223161862e-06,
+      "loss": 0.8617,
+      "step": 24864
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.624818801879883,
+      "learning_rate": 6.746428173579839e-06,
+      "loss": 0.8202,
+      "step": 24865
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 3.3902316093444824,
+      "learning_rate": 6.746181119141654e-06,
+      "loss": 0.6083,
+      "step": 24866
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 5.094982624053955,
+      "learning_rate": 6.745934059847993e-06,
+      "loss": 0.8929,
+      "step": 24867
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.6983565092086792,
+      "learning_rate": 6.745686995699544e-06,
+      "loss": 0.6239,
+      "step": 24868
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.6083738803863525,
+      "learning_rate": 6.745439926696993e-06,
+      "loss": 0.7615,
+      "step": 24869
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.8544526100158691,
+      "learning_rate": 6.745192852841027e-06,
+      "loss": 0.4682,
+      "step": 24870
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.3837709426879883,
+      "learning_rate": 6.744945774132334e-06,
+      "loss": 0.9667,
+      "step": 24871
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.9276098012924194,
+      "learning_rate": 6.744698690571601e-06,
+      "loss": 0.6599,
+      "step": 24872
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.2125587463378906,
+      "learning_rate": 6.744451602159514e-06,
+      "loss": 0.6777,
+      "step": 24873
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 4.604323863983154,
+      "learning_rate": 6.744204508896761e-06,
+      "loss": 0.7131,
+      "step": 24874
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.5518243312835693,
+      "learning_rate": 6.743957410784028e-06,
+      "loss": 0.6777,
+      "step": 24875
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.7561694383621216,
+      "learning_rate": 6.743710307822002e-06,
+      "loss": 0.4715,
+      "step": 24876
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 3.380063772201538,
+      "learning_rate": 6.743463200011372e-06,
+      "loss": 0.256,
+      "step": 24877
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.6333823204040527,
+      "learning_rate": 6.7432160873528245e-06,
+      "loss": 0.8648,
+      "step": 24878
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.738219738006592,
+      "learning_rate": 6.742968969847045e-06,
+      "loss": 0.717,
+      "step": 24879
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.813153624534607,
+      "learning_rate": 6.742721847494722e-06,
+      "loss": 0.6954,
+      "step": 24880
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.6269456148147583,
+      "learning_rate": 6.742474720296542e-06,
+      "loss": 0.7083,
+      "step": 24881
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.6420838832855225,
+      "learning_rate": 6.742227588253194e-06,
+      "loss": 0.6905,
+      "step": 24882
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.145275354385376,
+      "learning_rate": 6.741980451365363e-06,
+      "loss": 0.52,
+      "step": 24883
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.211088180541992,
+      "learning_rate": 6.741733309633737e-06,
+      "loss": 0.6345,
+      "step": 24884
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.685361623764038,
+      "learning_rate": 6.741486163059002e-06,
+      "loss": 0.5733,
+      "step": 24885
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.4779913425445557,
+      "learning_rate": 6.741239011641847e-06,
+      "loss": 0.8112,
+      "step": 24886
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.135141372680664,
+      "learning_rate": 6.740991855382959e-06,
+      "loss": 0.7301,
+      "step": 24887
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 3.7260425090789795,
+      "learning_rate": 6.7407446942830244e-06,
+      "loss": 0.6472,
+      "step": 24888
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.361193895339966,
+      "learning_rate": 6.740497528342731e-06,
+      "loss": 0.7212,
+      "step": 24889
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.893873691558838,
+      "learning_rate": 6.740250357562767e-06,
+      "loss": 0.9047,
+      "step": 24890
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.9327120780944824,
+      "learning_rate": 6.740003181943817e-06,
+      "loss": 0.7864,
+      "step": 24891
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.5375277996063232,
+      "learning_rate": 6.7397560014865715e-06,
+      "loss": 0.8091,
+      "step": 24892
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.9576092958450317,
+      "learning_rate": 6.739508816191715e-06,
+      "loss": 0.7183,
+      "step": 24893
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.8313188552856445,
+      "learning_rate": 6.739261626059937e-06,
+      "loss": 0.5647,
+      "step": 24894
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.2879718542099,
+      "learning_rate": 6.739014431091924e-06,
+      "loss": 0.6551,
+      "step": 24895
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.7565158605575562,
+      "learning_rate": 6.738767231288363e-06,
+      "loss": 0.72,
+      "step": 24896
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.5704419612884521,
+      "learning_rate": 6.738520026649941e-06,
+      "loss": 0.461,
+      "step": 24897
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.068125009536743,
+      "learning_rate": 6.738272817177347e-06,
+      "loss": 0.4272,
+      "step": 24898
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.3200581073760986,
+      "learning_rate": 6.738025602871267e-06,
+      "loss": 0.6136,
+      "step": 24899
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.7006804943084717,
+      "learning_rate": 6.737778383732389e-06,
+      "loss": 0.5277,
+      "step": 24900
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.9923231601715088,
+      "learning_rate": 6.7375311597614e-06,
+      "loss": 0.6442,
+      "step": 24901
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 3.2256479263305664,
+      "learning_rate": 6.7372839309589885e-06,
+      "loss": 0.7738,
+      "step": 24902
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.6449092626571655,
+      "learning_rate": 6.7370366973258405e-06,
+      "loss": 0.558,
+      "step": 24903
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.935460090637207,
+      "learning_rate": 6.736789458862645e-06,
+      "loss": 0.6496,
+      "step": 24904
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.9183071851730347,
+      "learning_rate": 6.736542215570088e-06,
+      "loss": 0.4655,
+      "step": 24905
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.767775058746338,
+      "learning_rate": 6.736294967448857e-06,
+      "loss": 0.7819,
+      "step": 24906
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.5841481685638428,
+      "learning_rate": 6.736047714499641e-06,
+      "loss": 0.5239,
+      "step": 24907
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 3.016620635986328,
+      "learning_rate": 6.735800456723126e-06,
+      "loss": 0.4945,
+      "step": 24908
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.9244182109832764,
+      "learning_rate": 6.7355531941200015e-06,
+      "loss": 0.5721,
+      "step": 24909
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.7672407627105713,
+      "learning_rate": 6.735305926690953e-06,
+      "loss": 0.8373,
+      "step": 24910
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 8.405647277832031,
+      "learning_rate": 6.735058654436668e-06,
+      "loss": 0.5196,
+      "step": 24911
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.4729225635528564,
+      "learning_rate": 6.734811377357835e-06,
+      "loss": 0.5642,
+      "step": 24912
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.749032735824585,
+      "learning_rate": 6.734564095455143e-06,
+      "loss": 0.7003,
+      "step": 24913
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.8142338991165161,
+      "learning_rate": 6.734316808729277e-06,
+      "loss": 0.5455,
+      "step": 24914
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 3.1417396068573,
+      "learning_rate": 6.734069517180925e-06,
+      "loss": 0.6479,
+      "step": 24915
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.7352514266967773,
+      "learning_rate": 6.733822220810776e-06,
+      "loss": 0.659,
+      "step": 24916
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.669266700744629,
+      "learning_rate": 6.733574919619517e-06,
+      "loss": 0.6632,
+      "step": 24917
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.9424307346343994,
+      "learning_rate": 6.733327613607834e-06,
+      "loss": 0.6543,
+      "step": 24918
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.8552274703979492,
+      "learning_rate": 6.733080302776418e-06,
+      "loss": 0.5271,
+      "step": 24919
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.5855698585510254,
+      "learning_rate": 6.732832987125954e-06,
+      "loss": 0.9967,
+      "step": 24920
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.0257585048675537,
+      "learning_rate": 6.732585666657132e-06,
+      "loss": 0.6125,
+      "step": 24921
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 3.0721192359924316,
+      "learning_rate": 6.732338341370638e-06,
+      "loss": 1.0742,
+      "step": 24922
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.9935674667358398,
+      "learning_rate": 6.732091011267159e-06,
+      "loss": 0.875,
+      "step": 24923
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 3.2510621547698975,
+      "learning_rate": 6.7318436763473836e-06,
+      "loss": 0.598,
+      "step": 24924
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.3602852821350098,
+      "learning_rate": 6.731596336611999e-06,
+      "loss": 0.7371,
+      "step": 24925
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.6108767986297607,
+      "learning_rate": 6.731348992061696e-06,
+      "loss": 0.4172,
+      "step": 24926
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 3.1276254653930664,
+      "learning_rate": 6.7311016426971586e-06,
+      "loss": 0.7559,
+      "step": 24927
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.797205924987793,
+      "learning_rate": 6.730854288519077e-06,
+      "loss": 0.6551,
+      "step": 24928
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.1487972736358643,
+      "learning_rate": 6.730606929528137e-06,
+      "loss": 0.3523,
+      "step": 24929
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 3.8354270458221436,
+      "learning_rate": 6.730359565725028e-06,
+      "loss": 0.5878,
+      "step": 24930
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.113215923309326,
+      "learning_rate": 6.730112197110436e-06,
+      "loss": 0.5203,
+      "step": 24931
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 3.6717851161956787,
+      "learning_rate": 6.729864823685052e-06,
+      "loss": 0.9438,
+      "step": 24932
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 3.185976505279541,
+      "learning_rate": 6.7296174454495614e-06,
+      "loss": 0.7991,
+      "step": 24933
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.3022851943969727,
+      "learning_rate": 6.729370062404652e-06,
+      "loss": 0.66,
+      "step": 24934
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.960842490196228,
+      "learning_rate": 6.729122674551012e-06,
+      "loss": 0.4334,
+      "step": 24935
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 3.078580379486084,
+      "learning_rate": 6.728875281889331e-06,
+      "loss": 0.4392,
+      "step": 24936
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.4278342723846436,
+      "learning_rate": 6.728627884420294e-06,
+      "loss": 0.4936,
+      "step": 24937
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.5204861164093018,
+      "learning_rate": 6.728380482144589e-06,
+      "loss": 0.6348,
+      "step": 24938
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.2059645652770996,
+      "learning_rate": 6.728133075062909e-06,
+      "loss": 0.7208,
+      "step": 24939
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.8322359323501587,
+      "learning_rate": 6.727885663175937e-06,
+      "loss": 0.6997,
+      "step": 24940
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.7908542156219482,
+      "learning_rate": 6.727638246484361e-06,
+      "loss": 0.8611,
+      "step": 24941
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.7838951349258423,
+      "learning_rate": 6.727390824988872e-06,
+      "loss": 0.8552,
+      "step": 24942
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.4598846435546875,
+      "learning_rate": 6.727143398690156e-06,
+      "loss": 0.5719,
+      "step": 24943
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.244112253189087,
+      "learning_rate": 6.7268959675889e-06,
+      "loss": 0.512,
+      "step": 24944
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.9962798357009888,
+      "learning_rate": 6.726648531685794e-06,
+      "loss": 0.4994,
+      "step": 24945
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.756762981414795,
+      "learning_rate": 6.726401090981525e-06,
+      "loss": 0.5773,
+      "step": 24946
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.7991960048675537,
+      "learning_rate": 6.726153645476781e-06,
+      "loss": 0.5578,
+      "step": 24947
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 3.2264904975891113,
+      "learning_rate": 6.72590619517225e-06,
+      "loss": 0.9064,
+      "step": 24948
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.704755187034607,
+      "learning_rate": 6.725658740068621e-06,
+      "loss": 0.5406,
+      "step": 24949
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.5836098194122314,
+      "learning_rate": 6.725411280166583e-06,
+      "loss": 0.6636,
+      "step": 24950
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.9915680885314941,
+      "learning_rate": 6.725163815466821e-06,
+      "loss": 0.7613,
+      "step": 24951
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.502760648727417,
+      "learning_rate": 6.724916345970025e-06,
+      "loss": 0.7173,
+      "step": 24952
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 5.991925239562988,
+      "learning_rate": 6.724668871676882e-06,
+      "loss": 0.8926,
+      "step": 24953
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 3.1409926414489746,
+      "learning_rate": 6.724421392588083e-06,
+      "loss": 0.6506,
+      "step": 24954
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.4921157360076904,
+      "learning_rate": 6.724173908704313e-06,
+      "loss": 0.5999,
+      "step": 24955
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.588552713394165,
+      "learning_rate": 6.723926420026261e-06,
+      "loss": 0.6093,
+      "step": 24956
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.4780101776123047,
+      "learning_rate": 6.723678926554616e-06,
+      "loss": 0.6067,
+      "step": 24957
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 4.468582630157471,
+      "learning_rate": 6.723431428290064e-06,
+      "loss": 0.9763,
+      "step": 24958
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.7624361515045166,
+      "learning_rate": 6.723183925233295e-06,
+      "loss": 0.4412,
+      "step": 24959
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 3.389622926712036,
+      "learning_rate": 6.7229364173849976e-06,
+      "loss": 0.6737,
+      "step": 24960
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.8921622037887573,
+      "learning_rate": 6.7226889047458595e-06,
+      "loss": 0.698,
+      "step": 24961
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.2084264755249023,
+      "learning_rate": 6.72244138731657e-06,
+      "loss": 0.8228,
+      "step": 24962
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.4323415756225586,
+      "learning_rate": 6.722193865097814e-06,
+      "loss": 0.5102,
+      "step": 24963
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 3.472073793411255,
+      "learning_rate": 6.721946338090285e-06,
+      "loss": 0.9389,
+      "step": 24964
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 3.998314142227173,
+      "learning_rate": 6.721698806294666e-06,
+      "loss": 0.4347,
+      "step": 24965
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.151442527770996,
+      "learning_rate": 6.721451269711648e-06,
+      "loss": 0.7191,
+      "step": 24966
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.199915647506714,
+      "learning_rate": 6.721203728341919e-06,
+      "loss": 0.6201,
+      "step": 24967
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.9373027086257935,
+      "learning_rate": 6.720956182186167e-06,
+      "loss": 0.7998,
+      "step": 24968
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.9799883365631104,
+      "learning_rate": 6.720708631245081e-06,
+      "loss": 0.4868,
+      "step": 24969
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.6939834356307983,
+      "learning_rate": 6.720461075519349e-06,
+      "loss": 0.7182,
+      "step": 24970
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 3.593538522720337,
+      "learning_rate": 6.720213515009659e-06,
+      "loss": 0.854,
+      "step": 24971
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.2280325889587402,
+      "learning_rate": 6.7199659497167e-06,
+      "loss": 0.5967,
+      "step": 24972
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.773779273033142,
+      "learning_rate": 6.71971837964116e-06,
+      "loss": 0.7866,
+      "step": 24973
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.3632051944732666,
+      "learning_rate": 6.719470804783727e-06,
+      "loss": 0.4531,
+      "step": 24974
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 3.307130813598633,
+      "learning_rate": 6.71922322514509e-06,
+      "loss": 0.8205,
+      "step": 24975
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.262627601623535,
+      "learning_rate": 6.718975640725938e-06,
+      "loss": 0.6181,
+      "step": 24976
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.5350428819656372,
+      "learning_rate": 6.718728051526957e-06,
+      "loss": 0.7011,
+      "step": 24977
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.603358507156372,
+      "learning_rate": 6.718480457548837e-06,
+      "loss": 0.6139,
+      "step": 24978
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.5903679132461548,
+      "learning_rate": 6.718232858792267e-06,
+      "loss": 0.2716,
+      "step": 24979
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.7289077043533325,
+      "learning_rate": 6.717985255257936e-06,
+      "loss": 0.5688,
+      "step": 24980
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.428624153137207,
+      "learning_rate": 6.717737646946532e-06,
+      "loss": 0.6581,
+      "step": 24981
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.639319658279419,
+      "learning_rate": 6.7174900338587426e-06,
+      "loss": 0.8135,
+      "step": 24982
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 4.470683574676514,
+      "learning_rate": 6.717242415995255e-06,
+      "loss": 0.6778,
+      "step": 24983
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 3.0913338661193848,
+      "learning_rate": 6.716994793356761e-06,
+      "loss": 0.705,
+      "step": 24984
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.5150516033172607,
+      "learning_rate": 6.716747165943949e-06,
+      "loss": 0.5425,
+      "step": 24985
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.7744574546813965,
+      "learning_rate": 6.716499533757505e-06,
+      "loss": 0.5433,
+      "step": 24986
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.6904546022415161,
+      "learning_rate": 6.716251896798118e-06,
+      "loss": 0.6426,
+      "step": 24987
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 3.1196045875549316,
+      "learning_rate": 6.716004255066478e-06,
+      "loss": 0.8081,
+      "step": 24988
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.182192325592041,
+      "learning_rate": 6.715756608563273e-06,
+      "loss": 0.8614,
+      "step": 24989
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 3.078479290008545,
+      "learning_rate": 6.715508957289192e-06,
+      "loss": 0.8643,
+      "step": 24990
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.033670425415039,
+      "learning_rate": 6.715261301244923e-06,
+      "loss": 0.4979,
+      "step": 24991
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.8317201137542725,
+      "learning_rate": 6.715013640431154e-06,
+      "loss": 0.6719,
+      "step": 24992
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.2106823921203613,
+      "learning_rate": 6.714765974848576e-06,
+      "loss": 0.7977,
+      "step": 24993
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.8623619079589844,
+      "learning_rate": 6.714518304497874e-06,
+      "loss": 0.5106,
+      "step": 24994
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.9804129600524902,
+      "learning_rate": 6.714270629379742e-06,
+      "loss": 0.6126,
+      "step": 24995
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.7869324684143066,
+      "learning_rate": 6.7140229494948635e-06,
+      "loss": 0.5285,
+      "step": 24996
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.4257237911224365,
+      "learning_rate": 6.71377526484393e-06,
+      "loss": 0.5573,
+      "step": 24997
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 3.8139498233795166,
+      "learning_rate": 6.713527575427629e-06,
+      "loss": 0.8357,
+      "step": 24998
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.4170138835906982,
+      "learning_rate": 6.713279881246649e-06,
+      "loss": 0.5688,
+      "step": 24999
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.6935852766036987,
+      "learning_rate": 6.713032182301681e-06,
+      "loss": 0.3655,
+      "step": 25000
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.083087921142578,
+      "learning_rate": 6.712784478593412e-06,
+      "loss": 0.4129,
+      "step": 25001
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 3.0415685176849365,
+      "learning_rate": 6.712536770122531e-06,
+      "loss": 0.7891,
+      "step": 25002
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.519524097442627,
+      "learning_rate": 6.712289056889726e-06,
+      "loss": 0.4544,
+      "step": 25003
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.0867486000061035,
+      "learning_rate": 6.7120413388956875e-06,
+      "loss": 0.7224,
+      "step": 25004
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.5090341567993164,
+      "learning_rate": 6.7117936161411025e-06,
+      "loss": 0.668,
+      "step": 25005
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.2286877632141113,
+      "learning_rate": 6.711545888626662e-06,
+      "loss": 0.6982,
+      "step": 25006
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.8913893699645996,
+      "learning_rate": 6.711298156353052e-06,
+      "loss": 0.6543,
+      "step": 25007
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.563596725463867,
+      "learning_rate": 6.711050419320963e-06,
+      "loss": 0.5612,
+      "step": 25008
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.9107892513275146,
+      "learning_rate": 6.710802677531084e-06,
+      "loss": 0.7122,
+      "step": 25009
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 3.298081874847412,
+      "learning_rate": 6.710554930984104e-06,
+      "loss": 0.6221,
+      "step": 25010
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.3678030967712402,
+      "learning_rate": 6.710307179680712e-06,
+      "loss": 0.4753,
+      "step": 25011
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.252078056335449,
+      "learning_rate": 6.710059423621595e-06,
+      "loss": 0.7971,
+      "step": 25012
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.110028028488159,
+      "learning_rate": 6.709811662807445e-06,
+      "loss": 0.5681,
+      "step": 25013
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.6930042505264282,
+      "learning_rate": 6.709563897238949e-06,
+      "loss": 0.6526,
+      "step": 25014
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.3385939598083496,
+      "learning_rate": 6.709316126916795e-06,
+      "loss": 0.7335,
+      "step": 25015
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.495847225189209,
+      "learning_rate": 6.7090683518416745e-06,
+      "loss": 0.5953,
+      "step": 25016
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.3718669414520264,
+      "learning_rate": 6.708820572014274e-06,
+      "loss": 0.5214,
+      "step": 25017
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.5131185054779053,
+      "learning_rate": 6.708572787435284e-06,
+      "loss": 0.7592,
+      "step": 25018
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.23630690574646,
+      "learning_rate": 6.708324998105393e-06,
+      "loss": 0.588,
+      "step": 25019
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 3.3597354888916016,
+      "learning_rate": 6.708077204025291e-06,
+      "loss": 0.5385,
+      "step": 25020
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.0835485458374023,
+      "learning_rate": 6.707829405195665e-06,
+      "loss": 0.4379,
+      "step": 25021
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 4.015723705291748,
+      "learning_rate": 6.707581601617205e-06,
+      "loss": 0.7716,
+      "step": 25022
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 3.4214348793029785,
+      "learning_rate": 6.707333793290603e-06,
+      "loss": 0.9076,
+      "step": 25023
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 3.7308294773101807,
+      "learning_rate": 6.707085980216542e-06,
+      "loss": 0.636,
+      "step": 25024
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.8680167198181152,
+      "learning_rate": 6.706838162395716e-06,
+      "loss": 0.8948,
+      "step": 25025
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.478114366531372,
+      "learning_rate": 6.7065903398288125e-06,
+      "loss": 0.4818,
+      "step": 25026
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.7881256341934204,
+      "learning_rate": 6.7063425125165195e-06,
+      "loss": 0.6045,
+      "step": 25027
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.1549293994903564,
+      "learning_rate": 6.706094680459528e-06,
+      "loss": 0.5859,
+      "step": 25028
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 5.2634172439575195,
+      "learning_rate": 6.705846843658526e-06,
+      "loss": 0.7471,
+      "step": 25029
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.5216821432113647,
+      "learning_rate": 6.705599002114202e-06,
+      "loss": 0.4979,
+      "step": 25030
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.9460688829421997,
+      "learning_rate": 6.705351155827249e-06,
+      "loss": 0.6177,
+      "step": 25031
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.591925859451294,
+      "learning_rate": 6.70510330479835e-06,
+      "loss": 0.8623,
+      "step": 25032
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.060530185699463,
+      "learning_rate": 6.7048554490282e-06,
+      "loss": 0.5392,
+      "step": 25033
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.661532998085022,
+      "learning_rate": 6.704607588517485e-06,
+      "loss": 0.5233,
+      "step": 25034
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 3.4585447311401367,
+      "learning_rate": 6.704359723266894e-06,
+      "loss": 0.7916,
+      "step": 25035
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.9553272724151611,
+      "learning_rate": 6.7041118532771174e-06,
+      "loss": 0.6368,
+      "step": 25036
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.6051926612854004,
+      "learning_rate": 6.703863978548846e-06,
+      "loss": 0.7063,
+      "step": 25037
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.8452606201171875,
+      "learning_rate": 6.703616099082765e-06,
+      "loss": 0.3684,
+      "step": 25038
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.9319217205047607,
+      "learning_rate": 6.703368214879565e-06,
+      "loss": 0.667,
+      "step": 25039
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.958585023880005,
+      "learning_rate": 6.703120325939938e-06,
+      "loss": 0.586,
+      "step": 25040
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.758565902709961,
+      "learning_rate": 6.70287243226457e-06,
+      "loss": 0.4725,
+      "step": 25041
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.789374589920044,
+      "learning_rate": 6.702624533854153e-06,
+      "loss": 0.808,
+      "step": 25042
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.26047420501709,
+      "learning_rate": 6.702376630709374e-06,
+      "loss": 0.7165,
+      "step": 25043
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 3.5279197692871094,
+      "learning_rate": 6.702128722830924e-06,
+      "loss": 0.366,
+      "step": 25044
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.9477148056030273,
+      "learning_rate": 6.701880810219492e-06,
+      "loss": 0.6862,
+      "step": 25045
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.9435570240020752,
+      "learning_rate": 6.701632892875766e-06,
+      "loss": 0.808,
+      "step": 25046
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 3.237405776977539,
+      "learning_rate": 6.701384970800436e-06,
+      "loss": 0.8145,
+      "step": 25047
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.8900786638259888,
+      "learning_rate": 6.701137043994192e-06,
+      "loss": 0.551,
+      "step": 25048
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.9808104038238525,
+      "learning_rate": 6.700889112457725e-06,
+      "loss": 0.6282,
+      "step": 25049
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.4046175479888916,
+      "learning_rate": 6.70064117619172e-06,
+      "loss": 0.6447,
+      "step": 25050
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.3989275693893433,
+      "learning_rate": 6.70039323519687e-06,
+      "loss": 0.7221,
+      "step": 25051
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.8854327201843262,
+      "learning_rate": 6.700145289473864e-06,
+      "loss": 0.6136,
+      "step": 25052
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.385221481323242,
+      "learning_rate": 6.6998973390233914e-06,
+      "loss": 0.5787,
+      "step": 25053
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.028404712677002,
+      "learning_rate": 6.699649383846139e-06,
+      "loss": 0.5869,
+      "step": 25054
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 3.2351200580596924,
+      "learning_rate": 6.6994014239428e-06,
+      "loss": 0.9732,
+      "step": 25055
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.687439203262329,
+      "learning_rate": 6.69915345931406e-06,
+      "loss": 0.7902,
+      "step": 25056
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.3104629516601562,
+      "learning_rate": 6.6989054899606135e-06,
+      "loss": 0.5395,
+      "step": 25057
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.73818302154541,
+      "learning_rate": 6.698657515883145e-06,
+      "loss": 0.724,
+      "step": 25058
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.5508322715759277,
+      "learning_rate": 6.698409537082349e-06,
+      "loss": 0.5177,
+      "step": 25059
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 3.0974457263946533,
+      "learning_rate": 6.69816155355891e-06,
+      "loss": 0.594,
+      "step": 25060
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.529656171798706,
+      "learning_rate": 6.697913565313521e-06,
+      "loss": 0.6404,
+      "step": 25061
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 3.8715877532958984,
+      "learning_rate": 6.69766557234687e-06,
+      "loss": 0.8675,
+      "step": 25062
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 4.032351016998291,
+      "learning_rate": 6.697417574659646e-06,
+      "loss": 0.7676,
+      "step": 25063
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.275665521621704,
+      "learning_rate": 6.697169572252541e-06,
+      "loss": 0.7129,
+      "step": 25064
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 4.74440860748291,
+      "learning_rate": 6.696921565126243e-06,
+      "loss": 0.6952,
+      "step": 25065
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.9921106100082397,
+      "learning_rate": 6.696673553281442e-06,
+      "loss": 0.533,
+      "step": 25066
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.907393455505371,
+      "learning_rate": 6.696425536718827e-06,
+      "loss": 0.5731,
+      "step": 25067
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.303581714630127,
+      "learning_rate": 6.6961775154390885e-06,
+      "loss": 0.5396,
+      "step": 25068
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.690147042274475,
+      "learning_rate": 6.695929489442914e-06,
+      "loss": 0.739,
+      "step": 25069
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.7339928150177,
+      "learning_rate": 6.6956814587309975e-06,
+      "loss": 0.8129,
+      "step": 25070
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.7889561653137207,
+      "learning_rate": 6.695433423304024e-06,
+      "loss": 0.7076,
+      "step": 25071
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.8568902015686035,
+      "learning_rate": 6.695185383162687e-06,
+      "loss": 0.5695,
+      "step": 25072
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 4.207023620605469,
+      "learning_rate": 6.6949373383076735e-06,
+      "loss": 0.5486,
+      "step": 25073
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.82696795463562,
+      "learning_rate": 6.694689288739674e-06,
+      "loss": 0.4341,
+      "step": 25074
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 3.447726249694824,
+      "learning_rate": 6.694441234459379e-06,
+      "loss": 0.4797,
+      "step": 25075
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 3.3947267532348633,
+      "learning_rate": 6.694193175467478e-06,
+      "loss": 0.4308,
+      "step": 25076
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.0655300617218018,
+      "learning_rate": 6.693945111764659e-06,
+      "loss": 0.7928,
+      "step": 25077
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 3.2409398555755615,
+      "learning_rate": 6.693697043351615e-06,
+      "loss": 0.6785,
+      "step": 25078
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.999724864959717,
+      "learning_rate": 6.693448970229033e-06,
+      "loss": 0.6029,
+      "step": 25079
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.206456422805786,
+      "learning_rate": 6.693200892397603e-06,
+      "loss": 0.6534,
+      "step": 25080
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 3.1904282569885254,
+      "learning_rate": 6.692952809858017e-06,
+      "loss": 0.69,
+      "step": 25081
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.5469863414764404,
+      "learning_rate": 6.6927047226109635e-06,
+      "loss": 0.7775,
+      "step": 25082
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.2176895141601562,
+      "learning_rate": 6.692456630657132e-06,
+      "loss": 0.6529,
+      "step": 25083
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.5143804550170898,
+      "learning_rate": 6.692208533997212e-06,
+      "loss": 0.7671,
+      "step": 25084
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.3189384937286377,
+      "learning_rate": 6.691960432631894e-06,
+      "loss": 0.7482,
+      "step": 25085
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 31.673656463623047,
+      "learning_rate": 6.691712326561868e-06,
+      "loss": 0.6794,
+      "step": 25086
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.2153825759887695,
+      "learning_rate": 6.691464215787824e-06,
+      "loss": 0.477,
+      "step": 25087
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.55947744846344,
+      "learning_rate": 6.691216100310451e-06,
+      "loss": 0.7265,
+      "step": 25088
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.1697311401367188,
+      "learning_rate": 6.690967980130442e-06,
+      "loss": 0.6614,
+      "step": 25089
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.2561182975769043,
+      "learning_rate": 6.690719855248481e-06,
+      "loss": 0.5459,
+      "step": 25090
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.9226834774017334,
+      "learning_rate": 6.690471725665263e-06,
+      "loss": 0.8978,
+      "step": 25091
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.982231616973877,
+      "learning_rate": 6.690223591381477e-06,
+      "loss": 0.6269,
+      "step": 25092
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.00334095954895,
+      "learning_rate": 6.689975452397812e-06,
+      "loss": 0.8266,
+      "step": 25093
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.8581576347351074,
+      "learning_rate": 6.6897273087149585e-06,
+      "loss": 0.7394,
+      "step": 25094
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.9936001300811768,
+      "learning_rate": 6.6894791603336055e-06,
+      "loss": 0.4528,
+      "step": 25095
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.0065386295318604,
+      "learning_rate": 6.689231007254444e-06,
+      "loss": 0.4822,
+      "step": 25096
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.6959810256958008,
+      "learning_rate": 6.688982849478165e-06,
+      "loss": 0.4846,
+      "step": 25097
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.8893465995788574,
+      "learning_rate": 6.688734687005457e-06,
+      "loss": 0.6789,
+      "step": 25098
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.2063050270080566,
+      "learning_rate": 6.6884865198370095e-06,
+      "loss": 0.6026,
+      "step": 25099
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.434695243835449,
+      "learning_rate": 6.688238347973514e-06,
+      "loss": 0.5611,
+      "step": 25100
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.0605058670043945,
+      "learning_rate": 6.6879901714156604e-06,
+      "loss": 0.6901,
+      "step": 25101
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.8301594257354736,
+      "learning_rate": 6.6877419901641385e-06,
+      "loss": 0.7684,
+      "step": 25102
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.0311977863311768,
+      "learning_rate": 6.687493804219638e-06,
+      "loss": 0.4981,
+      "step": 25103
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.235642194747925,
+      "learning_rate": 6.687245613582849e-06,
+      "loss": 0.5824,
+      "step": 25104
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.9593390226364136,
+      "learning_rate": 6.6869974182544625e-06,
+      "loss": 0.6447,
+      "step": 25105
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.745750904083252,
+      "learning_rate": 6.686749218235169e-06,
+      "loss": 0.6928,
+      "step": 25106
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 3.170728921890259,
+      "learning_rate": 6.686501013525657e-06,
+      "loss": 0.8092,
+      "step": 25107
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.886129379272461,
+      "learning_rate": 6.686252804126617e-06,
+      "loss": 0.5643,
+      "step": 25108
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.953432321548462,
+      "learning_rate": 6.686004590038741e-06,
+      "loss": 0.8213,
+      "step": 25109
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 4.00826358795166,
+      "learning_rate": 6.685756371262715e-06,
+      "loss": 0.6995,
+      "step": 25110
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.3716189861297607,
+      "learning_rate": 6.685508147799235e-06,
+      "loss": 0.6977,
+      "step": 25111
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.9473073482513428,
+      "learning_rate": 6.685259919648987e-06,
+      "loss": 0.5108,
+      "step": 25112
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.9196715354919434,
+      "learning_rate": 6.685011686812663e-06,
+      "loss": 0.6697,
+      "step": 25113
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.077988862991333,
+      "learning_rate": 6.684763449290952e-06,
+      "loss": 0.5727,
+      "step": 25114
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.031501293182373,
+      "learning_rate": 6.684515207084546e-06,
+      "loss": 0.6678,
+      "step": 25115
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 3.1936683654785156,
+      "learning_rate": 6.6842669601941324e-06,
+      "loss": 0.6328,
+      "step": 25116
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 3.6417574882507324,
+      "learning_rate": 6.684018708620404e-06,
+      "loss": 0.5558,
+      "step": 25117
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.7754430770874023,
+      "learning_rate": 6.683770452364052e-06,
+      "loss": 0.7661,
+      "step": 25118
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.653618812561035,
+      "learning_rate": 6.683522191425763e-06,
+      "loss": 0.6655,
+      "step": 25119
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 3.236952543258667,
+      "learning_rate": 6.683273925806231e-06,
+      "loss": 0.5455,
+      "step": 25120
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.999753475189209,
+      "learning_rate": 6.6830256555061435e-06,
+      "loss": 0.9396,
+      "step": 25121
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.014972686767578,
+      "learning_rate": 6.682777380526193e-06,
+      "loss": 0.8538,
+      "step": 25122
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.3370754718780518,
+      "learning_rate": 6.682529100867068e-06,
+      "loss": 0.7577,
+      "step": 25123
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.020869731903076,
+      "learning_rate": 6.682280816529462e-06,
+      "loss": 0.6306,
+      "step": 25124
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 3.003432273864746,
+      "learning_rate": 6.682032527514062e-06,
+      "loss": 0.8424,
+      "step": 25125
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.9594008922576904,
+      "learning_rate": 6.6817842338215584e-06,
+      "loss": 0.6941,
+      "step": 25126
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.4595487117767334,
+      "learning_rate": 6.681535935452646e-06,
+      "loss": 0.7823,
+      "step": 25127
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.8153421878814697,
+      "learning_rate": 6.681287632408009e-06,
+      "loss": 0.5056,
+      "step": 25128
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.9964349269866943,
+      "learning_rate": 6.681039324688342e-06,
+      "loss": 0.6804,
+      "step": 25129
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.6359305381774902,
+      "learning_rate": 6.680791012294334e-06,
+      "loss": 0.5803,
+      "step": 25130
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 3.8645474910736084,
+      "learning_rate": 6.680542695226676e-06,
+      "loss": 0.9946,
+      "step": 25131
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 6.02371883392334,
+      "learning_rate": 6.680294373486059e-06,
+      "loss": 0.7704,
+      "step": 25132
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.749260902404785,
+      "learning_rate": 6.680046047073172e-06,
+      "loss": 0.7005,
+      "step": 25133
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.8927313089370728,
+      "learning_rate": 6.679797715988707e-06,
+      "loss": 0.7244,
+      "step": 25134
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 3.311354160308838,
+      "learning_rate": 6.679549380233354e-06,
+      "loss": 0.6194,
+      "step": 25135
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.969963788986206,
+      "learning_rate": 6.679301039807803e-06,
+      "loss": 0.4592,
+      "step": 25136
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.6807031631469727,
+      "learning_rate": 6.679052694712744e-06,
+      "loss": 0.8572,
+      "step": 25137
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.7201318740844727,
+      "learning_rate": 6.67880434494887e-06,
+      "loss": 0.7897,
+      "step": 25138
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.9997570514678955,
+      "learning_rate": 6.678555990516869e-06,
+      "loss": 0.6734,
+      "step": 25139
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.0932540893554688,
+      "learning_rate": 6.678307631417433e-06,
+      "loss": 0.6814,
+      "step": 25140
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.6058088541030884,
+      "learning_rate": 6.678059267651251e-06,
+      "loss": 0.6992,
+      "step": 25141
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 3.038233995437622,
+      "learning_rate": 6.677810899219016e-06,
+      "loss": 0.7038,
+      "step": 25142
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.604037046432495,
+      "learning_rate": 6.677562526121417e-06,
+      "loss": 0.7683,
+      "step": 25143
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.4507548809051514,
+      "learning_rate": 6.677314148359147e-06,
+      "loss": 0.6746,
+      "step": 25144
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 5.313934803009033,
+      "learning_rate": 6.677065765932892e-06,
+      "loss": 0.7033,
+      "step": 25145
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.2133986949920654,
+      "learning_rate": 6.676817378843347e-06,
+      "loss": 0.6271,
+      "step": 25146
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.3562309741973877,
+      "learning_rate": 6.6765689870912e-06,
+      "loss": 0.6583,
+      "step": 25147
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.209502696990967,
+      "learning_rate": 6.676320590677143e-06,
+      "loss": 0.5311,
+      "step": 25148
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.2548670768737793,
+      "learning_rate": 6.676072189601866e-06,
+      "loss": 0.7204,
+      "step": 25149
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.6326388120651245,
+      "learning_rate": 6.675823783866061e-06,
+      "loss": 0.6653,
+      "step": 25150
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.5350027084350586,
+      "learning_rate": 6.675575373470416e-06,
+      "loss": 0.629,
+      "step": 25151
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.6345171928405762,
+      "learning_rate": 6.6753269584156255e-06,
+      "loss": 0.528,
+      "step": 25152
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.942606210708618,
+      "learning_rate": 6.675078538702377e-06,
+      "loss": 0.4816,
+      "step": 25153
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.012014865875244,
+      "learning_rate": 6.674830114331364e-06,
+      "loss": 0.7215,
+      "step": 25154
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.7328057289123535,
+      "learning_rate": 6.674581685303275e-06,
+      "loss": 0.8444,
+      "step": 25155
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.034879207611084,
+      "learning_rate": 6.674333251618802e-06,
+      "loss": 0.5374,
+      "step": 25156
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.352109432220459,
+      "learning_rate": 6.674084813278637e-06,
+      "loss": 0.5808,
+      "step": 25157
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.6583222150802612,
+      "learning_rate": 6.673836370283466e-06,
+      "loss": 0.7303,
+      "step": 25158
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 3.0716569423675537,
+      "learning_rate": 6.673587922633985e-06,
+      "loss": 0.414,
+      "step": 25159
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.9329938888549805,
+      "learning_rate": 6.673339470330883e-06,
+      "loss": 0.486,
+      "step": 25160
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.8153249025344849,
+      "learning_rate": 6.673091013374849e-06,
+      "loss": 0.3881,
+      "step": 25161
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.7095966339111328,
+      "learning_rate": 6.672842551766577e-06,
+      "loss": 0.5801,
+      "step": 25162
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.5563442707061768,
+      "learning_rate": 6.672594085506756e-06,
+      "loss": 0.5065,
+      "step": 25163
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.9315626621246338,
+      "learning_rate": 6.672345614596078e-06,
+      "loss": 0.7802,
+      "step": 25164
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 3.624802827835083,
+      "learning_rate": 6.672097139035234e-06,
+      "loss": 0.787,
+      "step": 25165
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.9956870079040527,
+      "learning_rate": 6.671848658824913e-06,
+      "loss": 0.5664,
+      "step": 25166
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.6067073345184326,
+      "learning_rate": 6.6716001739658085e-06,
+      "loss": 0.6336,
+      "step": 25167
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.738356351852417,
+      "learning_rate": 6.671351684458609e-06,
+      "loss": 0.599,
+      "step": 25168
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.4992177486419678,
+      "learning_rate": 6.671103190304007e-06,
+      "loss": 0.6072,
+      "step": 25169
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.98350191116333,
+      "learning_rate": 6.670854691502693e-06,
+      "loss": 0.7281,
+      "step": 25170
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.962510108947754,
+      "learning_rate": 6.670606188055357e-06,
+      "loss": 0.9219,
+      "step": 25171
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 3.8282389640808105,
+      "learning_rate": 6.670357679962692e-06,
+      "loss": 0.6583,
+      "step": 25172
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.6635457277297974,
+      "learning_rate": 6.6701091672253885e-06,
+      "loss": 0.6336,
+      "step": 25173
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.1022708415985107,
+      "learning_rate": 6.669860649844136e-06,
+      "loss": 0.4905,
+      "step": 25174
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.7670642137527466,
+      "learning_rate": 6.6696121278196265e-06,
+      "loss": 0.6383,
+      "step": 25175
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.1550259590148926,
+      "learning_rate": 6.669363601152552e-06,
+      "loss": 0.5294,
+      "step": 25176
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 4.2174296379089355,
+      "learning_rate": 6.669115069843604e-06,
+      "loss": 0.8288,
+      "step": 25177
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 3.5560202598571777,
+      "learning_rate": 6.668866533893469e-06,
+      "loss": 0.8254,
+      "step": 25178
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.525940179824829,
+      "learning_rate": 6.668617993302843e-06,
+      "loss": 0.7685,
+      "step": 25179
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.9998308420181274,
+      "learning_rate": 6.668369448072415e-06,
+      "loss": 0.8377,
+      "step": 25180
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.3642985820770264,
+      "learning_rate": 6.668120898202877e-06,
+      "loss": 0.5484,
+      "step": 25181
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 4.53744649887085,
+      "learning_rate": 6.66787234369492e-06,
+      "loss": 0.6409,
+      "step": 25182
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.8274214267730713,
+      "learning_rate": 6.667623784549233e-06,
+      "loss": 0.744,
+      "step": 25183
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.1216013431549072,
+      "learning_rate": 6.667375220766511e-06,
+      "loss": 0.7106,
+      "step": 25184
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 3.2199456691741943,
+      "learning_rate": 6.667126652347442e-06,
+      "loss": 0.5532,
+      "step": 25185
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.5966613292694092,
+      "learning_rate": 6.666878079292719e-06,
+      "loss": 0.7056,
+      "step": 25186
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.7928545475006104,
+      "learning_rate": 6.666629501603032e-06,
+      "loss": 0.4906,
+      "step": 25187
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 4.689874649047852,
+      "learning_rate": 6.666380919279074e-06,
+      "loss": 0.9556,
+      "step": 25188
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.211350202560425,
+      "learning_rate": 6.666132332321533e-06,
+      "loss": 0.7269,
+      "step": 25189
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.8820583820343018,
+      "learning_rate": 6.665883740731104e-06,
+      "loss": 0.4079,
+      "step": 25190
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 3.2764670848846436,
+      "learning_rate": 6.6656351445084745e-06,
+      "loss": 0.7528,
+      "step": 25191
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.053048610687256,
+      "learning_rate": 6.665386543654339e-06,
+      "loss": 0.5761,
+      "step": 25192
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.6920664310455322,
+      "learning_rate": 6.6651379381693855e-06,
+      "loss": 0.5253,
+      "step": 25193
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.9507988691329956,
+      "learning_rate": 6.664889328054309e-06,
+      "loss": 0.592,
+      "step": 25194
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.6652040481567383,
+      "learning_rate": 6.664640713309799e-06,
+      "loss": 0.648,
+      "step": 25195
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.7950536012649536,
+      "learning_rate": 6.664392093936547e-06,
+      "loss": 0.6494,
+      "step": 25196
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.756075143814087,
+      "learning_rate": 6.664143469935243e-06,
+      "loss": 0.4761,
+      "step": 25197
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.6955556869506836,
+      "learning_rate": 6.663894841306579e-06,
+      "loss": 0.7306,
+      "step": 25198
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.7961002588272095,
+      "learning_rate": 6.663646208051249e-06,
+      "loss": 0.8402,
+      "step": 25199
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 5.42650032043457,
+      "learning_rate": 6.66339757016994e-06,
+      "loss": 0.5489,
+      "step": 25200
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.352806806564331,
+      "learning_rate": 6.663148927663346e-06,
+      "loss": 0.5833,
+      "step": 25201
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 4.476074695587158,
+      "learning_rate": 6.662900280532157e-06,
+      "loss": 0.9086,
+      "step": 25202
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.5902942419052124,
+      "learning_rate": 6.662651628777067e-06,
+      "loss": 0.8481,
+      "step": 25203
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.1397576332092285,
+      "learning_rate": 6.662402972398765e-06,
+      "loss": 0.6201,
+      "step": 25204
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.4367806911468506,
+      "learning_rate": 6.662154311397944e-06,
+      "loss": 0.6097,
+      "step": 25205
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.8728740215301514,
+      "learning_rate": 6.661905645775293e-06,
+      "loss": 0.5499,
+      "step": 25206
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.7786649465560913,
+      "learning_rate": 6.661656975531505e-06,
+      "loss": 0.5957,
+      "step": 25207
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.7797352075576782,
+      "learning_rate": 6.6614083006672726e-06,
+      "loss": 0.6306,
+      "step": 25208
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.8461222648620605,
+      "learning_rate": 6.661159621183285e-06,
+      "loss": 0.4943,
+      "step": 25209
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.7223129272460938,
+      "learning_rate": 6.660910937080236e-06,
+      "loss": 0.4708,
+      "step": 25210
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 3.2051451206207275,
+      "learning_rate": 6.660662248358815e-06,
+      "loss": 0.8867,
+      "step": 25211
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.382434844970703,
+      "learning_rate": 6.660413555019714e-06,
+      "loss": 0.543,
+      "step": 25212
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.883471131324768,
+      "learning_rate": 6.660164857063625e-06,
+      "loss": 0.7605,
+      "step": 25213
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.0109055042266846,
+      "learning_rate": 6.65991615449124e-06,
+      "loss": 0.4246,
+      "step": 25214
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.9816737174987793,
+      "learning_rate": 6.65966744730325e-06,
+      "loss": 0.6455,
+      "step": 25215
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.6105934381484985,
+      "learning_rate": 6.659418735500346e-06,
+      "loss": 0.5664,
+      "step": 25216
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.3304672241210938,
+      "learning_rate": 6.659170019083221e-06,
+      "loss": 0.4169,
+      "step": 25217
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 3.1980206966400146,
+      "learning_rate": 6.658921298052565e-06,
+      "loss": 0.7689,
+      "step": 25218
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.8249927759170532,
+      "learning_rate": 6.658672572409071e-06,
+      "loss": 0.6312,
+      "step": 25219
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.0648179054260254,
+      "learning_rate": 6.658423842153429e-06,
+      "loss": 0.587,
+      "step": 25220
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.2675793170928955,
+      "learning_rate": 6.658175107286332e-06,
+      "loss": 0.6188,
+      "step": 25221
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.3584303855895996,
+      "learning_rate": 6.657926367808471e-06,
+      "loss": 0.3805,
+      "step": 25222
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 3.033658742904663,
+      "learning_rate": 6.6576776237205375e-06,
+      "loss": 0.7605,
+      "step": 25223
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 4.780362129211426,
+      "learning_rate": 6.657428875023224e-06,
+      "loss": 0.5879,
+      "step": 25224
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.6258599758148193,
+      "learning_rate": 6.6571801217172215e-06,
+      "loss": 0.2065,
+      "step": 25225
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.6477270126342773,
+      "learning_rate": 6.656931363803223e-06,
+      "loss": 0.5856,
+      "step": 25226
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 4.0843095779418945,
+      "learning_rate": 6.656682601281919e-06,
+      "loss": 0.6487,
+      "step": 25227
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.7134913206100464,
+      "learning_rate": 6.656433834153999e-06,
+      "loss": 0.5168,
+      "step": 25228
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 3.779649257659912,
+      "learning_rate": 6.65618506242016e-06,
+      "loss": 0.7882,
+      "step": 25229
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.8260161876678467,
+      "learning_rate": 6.655936286081088e-06,
+      "loss": 0.5329,
+      "step": 25230
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.921792984008789,
+      "learning_rate": 6.655687505137479e-06,
+      "loss": 0.6846,
+      "step": 25231
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.528684616088867,
+      "learning_rate": 6.655438719590023e-06,
+      "loss": 0.7207,
+      "step": 25232
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 3.12541127204895,
+      "learning_rate": 6.6551899294394115e-06,
+      "loss": 0.7103,
+      "step": 25233
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.073890447616577,
+      "learning_rate": 6.654941134686338e-06,
+      "loss": 0.8001,
+      "step": 25234
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.7882039546966553,
+      "learning_rate": 6.654692335331492e-06,
+      "loss": 0.4891,
+      "step": 25235
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.6399333477020264,
+      "learning_rate": 6.6544435313755674e-06,
+      "loss": 0.7039,
+      "step": 25236
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.1344592571258545,
+      "learning_rate": 6.654194722819255e-06,
+      "loss": 0.6996,
+      "step": 25237
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 3.54655385017395,
+      "learning_rate": 6.653945909663247e-06,
+      "loss": 0.7643,
+      "step": 25238
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.924113154411316,
+      "learning_rate": 6.653697091908234e-06,
+      "loss": 0.5968,
+      "step": 25239
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.5682666301727295,
+      "learning_rate": 6.653448269554909e-06,
+      "loss": 0.5811,
+      "step": 25240
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.4118945598602295,
+      "learning_rate": 6.653199442603964e-06,
+      "loss": 0.6161,
+      "step": 25241
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.7049787044525146,
+      "learning_rate": 6.6529506110560915e-06,
+      "loss": 0.4001,
+      "step": 25242
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.6747969388961792,
+      "learning_rate": 6.6527017749119815e-06,
+      "loss": 0.4968,
+      "step": 25243
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 3.3098528385162354,
+      "learning_rate": 6.652452934172328e-06,
+      "loss": 0.5924,
+      "step": 25244
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.9633405208587646,
+      "learning_rate": 6.652204088837821e-06,
+      "loss": 0.4183,
+      "step": 25245
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.9537270069122314,
+      "learning_rate": 6.651955238909153e-06,
+      "loss": 0.5908,
+      "step": 25246
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.3993358612060547,
+      "learning_rate": 6.651706384387017e-06,
+      "loss": 0.5413,
+      "step": 25247
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.7732491493225098,
+      "learning_rate": 6.651457525272105e-06,
+      "loss": 0.8878,
+      "step": 25248
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.7159435749053955,
+      "learning_rate": 6.651208661565107e-06,
+      "loss": 0.6493,
+      "step": 25249
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.621685028076172,
+      "learning_rate": 6.650959793266717e-06,
+      "loss": 0.5507,
+      "step": 25250
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 3.2670962810516357,
+      "learning_rate": 6.650710920377625e-06,
+      "loss": 0.531,
+      "step": 25251
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.092662811279297,
+      "learning_rate": 6.650462042898525e-06,
+      "loss": 0.5799,
+      "step": 25252
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.0652546882629395,
+      "learning_rate": 6.650213160830109e-06,
+      "loss": 0.7384,
+      "step": 25253
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.7921110391616821,
+      "learning_rate": 6.649964274173069e-06,
+      "loss": 0.4988,
+      "step": 25254
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.3022077083587646,
+      "learning_rate": 6.649715382928095e-06,
+      "loss": 0.4073,
+      "step": 25255
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 3.3989269733428955,
+      "learning_rate": 6.649466487095881e-06,
+      "loss": 0.638,
+      "step": 25256
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 3.590203285217285,
+      "learning_rate": 6.649217586677119e-06,
+      "loss": 0.4591,
+      "step": 25257
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.3558070659637451,
+      "learning_rate": 6.6489686816725005e-06,
+      "loss": 0.7319,
+      "step": 25258
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.115682601928711,
+      "learning_rate": 6.6487197720827175e-06,
+      "loss": 0.5805,
+      "step": 25259
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 3.4810891151428223,
+      "learning_rate": 6.648470857908464e-06,
+      "loss": 0.6098,
+      "step": 25260
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.9029953479766846,
+      "learning_rate": 6.648221939150429e-06,
+      "loss": 0.7118,
+      "step": 25261
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.578102707862854,
+      "learning_rate": 6.647973015809306e-06,
+      "loss": 0.6615,
+      "step": 25262
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 3.9878265857696533,
+      "learning_rate": 6.647724087885788e-06,
+      "loss": 0.5942,
+      "step": 25263
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.7322325706481934,
+      "learning_rate": 6.647475155380565e-06,
+      "loss": 0.4486,
+      "step": 25264
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.101491689682007,
+      "learning_rate": 6.647226218294333e-06,
+      "loss": 0.5427,
+      "step": 25265
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.8793187141418457,
+      "learning_rate": 6.64697727662778e-06,
+      "loss": 0.5222,
+      "step": 25266
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.837407112121582,
+      "learning_rate": 6.646728330381602e-06,
+      "loss": 0.8182,
+      "step": 25267
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 4.3420562744140625,
+      "learning_rate": 6.646479379556489e-06,
+      "loss": 0.4538,
+      "step": 25268
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.541809558868408,
+      "learning_rate": 6.646230424153133e-06,
+      "loss": 0.4874,
+      "step": 25269
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.4800209999084473,
+      "learning_rate": 6.645981464172227e-06,
+      "loss": 0.7064,
+      "step": 25270
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.7768871784210205,
+      "learning_rate": 6.645732499614463e-06,
+      "loss": 0.5608,
+      "step": 25271
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.2673585414886475,
+      "learning_rate": 6.645483530480534e-06,
+      "loss": 0.9082,
+      "step": 25272
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 4.601502895355225,
+      "learning_rate": 6.645234556771132e-06,
+      "loss": 0.8205,
+      "step": 25273
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 11.042618751525879,
+      "learning_rate": 6.644985578486947e-06,
+      "loss": 0.6843,
+      "step": 25274
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.347498655319214,
+      "learning_rate": 6.6447365956286754e-06,
+      "loss": 0.666,
+      "step": 25275
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 3.4536097049713135,
+      "learning_rate": 6.644487608197008e-06,
+      "loss": 0.5797,
+      "step": 25276
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.509779214859009,
+      "learning_rate": 6.644238616192635e-06,
+      "loss": 1.0377,
+      "step": 25277
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 4.615306854248047,
+      "learning_rate": 6.643989619616251e-06,
+      "loss": 0.3506,
+      "step": 25278
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.0534520149230957,
+      "learning_rate": 6.6437406184685475e-06,
+      "loss": 0.7211,
+      "step": 25279
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.8792363405227661,
+      "learning_rate": 6.643491612750217e-06,
+      "loss": 0.6823,
+      "step": 25280
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.8205783367156982,
+      "learning_rate": 6.643242602461953e-06,
+      "loss": 0.6317,
+      "step": 25281
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.1756739616394043,
+      "learning_rate": 6.642993587604445e-06,
+      "loss": 0.7161,
+      "step": 25282
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.820080041885376,
+      "learning_rate": 6.6427445681783875e-06,
+      "loss": 0.8179,
+      "step": 25283
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.1004414558410645,
+      "learning_rate": 6.642495544184474e-06,
+      "loss": 0.9038,
+      "step": 25284
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.4306286573410034,
+      "learning_rate": 6.642246515623396e-06,
+      "loss": 0.3097,
+      "step": 25285
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.090531349182129,
+      "learning_rate": 6.641997482495845e-06,
+      "loss": 0.6027,
+      "step": 25286
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.954659104347229,
+      "learning_rate": 6.641748444802514e-06,
+      "loss": 0.7895,
+      "step": 25287
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.623103141784668,
+      "learning_rate": 6.641499402544096e-06,
+      "loss": 0.5562,
+      "step": 25288
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.3650524616241455,
+      "learning_rate": 6.641250355721282e-06,
+      "loss": 0.5917,
+      "step": 25289
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.5598268508911133,
+      "learning_rate": 6.641001304334766e-06,
+      "loss": 0.8695,
+      "step": 25290
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 3.5471441745758057,
+      "learning_rate": 6.64075224838524e-06,
+      "loss": 0.6195,
+      "step": 25291
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.6081838607788086,
+      "learning_rate": 6.6405031878733986e-06,
+      "loss": 0.8341,
+      "step": 25292
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.2488319873809814,
+      "learning_rate": 6.6402541227999294e-06,
+      "loss": 0.8867,
+      "step": 25293
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.4662933349609375,
+      "learning_rate": 6.640005053165529e-06,
+      "loss": 0.7126,
+      "step": 25294
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.6736035346984863,
+      "learning_rate": 6.63975597897089e-06,
+      "loss": 0.5886,
+      "step": 25295
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.441084861755371,
+      "learning_rate": 6.639506900216704e-06,
+      "loss": 0.6293,
+      "step": 25296
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.2863810062408447,
+      "learning_rate": 6.639257816903662e-06,
+      "loss": 0.6881,
+      "step": 25297
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 3.030836343765259,
+      "learning_rate": 6.639008729032459e-06,
+      "loss": 0.6211,
+      "step": 25298
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 4.575911998748779,
+      "learning_rate": 6.638759636603786e-06,
+      "loss": 0.8665,
+      "step": 25299
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.8345104455947876,
+      "learning_rate": 6.638510539618337e-06,
+      "loss": 0.516,
+      "step": 25300
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 4.808453559875488,
+      "learning_rate": 6.638261438076803e-06,
+      "loss": 0.6838,
+      "step": 25301
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.3311338424682617,
+      "learning_rate": 6.638012331979877e-06,
+      "loss": 0.7806,
+      "step": 25302
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.824530601501465,
+      "learning_rate": 6.6377632213282554e-06,
+      "loss": 0.6719,
+      "step": 25303
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.973419189453125,
+      "learning_rate": 6.6375141061226255e-06,
+      "loss": 0.6757,
+      "step": 25304
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.185546636581421,
+      "learning_rate": 6.637264986363683e-06,
+      "loss": 0.6012,
+      "step": 25305
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 4.437071323394775,
+      "learning_rate": 6.63701586205212e-06,
+      "loss": 0.6922,
+      "step": 25306
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.3328988552093506,
+      "learning_rate": 6.6367667331886275e-06,
+      "loss": 0.6533,
+      "step": 25307
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.180490732192993,
+      "learning_rate": 6.636517599773902e-06,
+      "loss": 0.8359,
+      "step": 25308
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.347891092300415,
+      "learning_rate": 6.636268461808634e-06,
+      "loss": 0.5163,
+      "step": 25309
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 4.456543922424316,
+      "learning_rate": 6.6360193192935155e-06,
+      "loss": 0.7457,
+      "step": 25310
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.8092291355133057,
+      "learning_rate": 6.63577017222924e-06,
+      "loss": 0.5594,
+      "step": 25311
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.8365106582641602,
+      "learning_rate": 6.635521020616501e-06,
+      "loss": 0.7382,
+      "step": 25312
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.84698748588562,
+      "learning_rate": 6.63527186445599e-06,
+      "loss": 0.7229,
+      "step": 25313
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.308150053024292,
+      "learning_rate": 6.635022703748402e-06,
+      "loss": 0.6113,
+      "step": 25314
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 3.307926893234253,
+      "learning_rate": 6.634773538494427e-06,
+      "loss": 0.6616,
+      "step": 25315
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.80644690990448,
+      "learning_rate": 6.634524368694759e-06,
+      "loss": 0.7865,
+      "step": 25316
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.536083698272705,
+      "learning_rate": 6.634275194350093e-06,
+      "loss": 0.6274,
+      "step": 25317
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 3.035876750946045,
+      "learning_rate": 6.634026015461118e-06,
+      "loss": 0.474,
+      "step": 25318
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 3.0442113876342773,
+      "learning_rate": 6.633776832028529e-06,
+      "loss": 0.6159,
+      "step": 25319
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.3725782632827759,
+      "learning_rate": 6.633527644053019e-06,
+      "loss": 0.623,
+      "step": 25320
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.134592294692993,
+      "learning_rate": 6.633278451535281e-06,
+      "loss": 0.6099,
+      "step": 25321
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.8521523475646973,
+      "learning_rate": 6.6330292544760066e-06,
+      "loss": 0.7236,
+      "step": 25322
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.2288715839385986,
+      "learning_rate": 6.632780052875891e-06,
+      "loss": 0.4811,
+      "step": 25323
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 6.043630123138428,
+      "learning_rate": 6.6325308467356234e-06,
+      "loss": 0.6544,
+      "step": 25324
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.2738864421844482,
+      "learning_rate": 6.632281636055901e-06,
+      "loss": 0.7282,
+      "step": 25325
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.060851573944092,
+      "learning_rate": 6.632032420837413e-06,
+      "loss": 0.5238,
+      "step": 25326
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.316338539123535,
+      "learning_rate": 6.631783201080857e-06,
+      "loss": 0.7423,
+      "step": 25327
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.742626428604126,
+      "learning_rate": 6.631533976786921e-06,
+      "loss": 0.7842,
+      "step": 25328
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.187955617904663,
+      "learning_rate": 6.631284747956301e-06,
+      "loss": 0.5848,
+      "step": 25329
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 3.027898073196411,
+      "learning_rate": 6.6310355145896895e-06,
+      "loss": 0.5563,
+      "step": 25330
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.3457757234573364,
+      "learning_rate": 6.630786276687779e-06,
+      "loss": 0.3908,
+      "step": 25331
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.611294984817505,
+      "learning_rate": 6.630537034251262e-06,
+      "loss": 0.5086,
+      "step": 25332
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.7710696458816528,
+      "learning_rate": 6.6302877872808335e-06,
+      "loss": 0.3939,
+      "step": 25333
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.80894136428833,
+      "learning_rate": 6.630038535777184e-06,
+      "loss": 0.5493,
+      "step": 25334
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.945442795753479,
+      "learning_rate": 6.629789279741008e-06,
+      "loss": 0.5853,
+      "step": 25335
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.6978917121887207,
+      "learning_rate": 6.629540019173e-06,
+      "loss": 0.4579,
+      "step": 25336
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.5397114753723145,
+      "learning_rate": 6.629290754073851e-06,
+      "loss": 0.7802,
+      "step": 25337
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.747192144393921,
+      "learning_rate": 6.629041484444254e-06,
+      "loss": 0.5255,
+      "step": 25338
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.353747606277466,
+      "learning_rate": 6.628792210284905e-06,
+      "loss": 0.7313,
+      "step": 25339
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.612898349761963,
+      "learning_rate": 6.6285429315964934e-06,
+      "loss": 0.7731,
+      "step": 25340
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 3.1281635761260986,
+      "learning_rate": 6.628293648379713e-06,
+      "loss": 1.0015,
+      "step": 25341
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.2442712783813477,
+      "learning_rate": 6.628044360635259e-06,
+      "loss": 0.7958,
+      "step": 25342
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 3.2977261543273926,
+      "learning_rate": 6.627795068363825e-06,
+      "loss": 0.5133,
+      "step": 25343
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.0674939155578613,
+      "learning_rate": 6.6275457715661e-06,
+      "loss": 0.2777,
+      "step": 25344
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.665940761566162,
+      "learning_rate": 6.627296470242781e-06,
+      "loss": 0.8396,
+      "step": 25345
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.5867640972137451,
+      "learning_rate": 6.627047164394561e-06,
+      "loss": 0.7134,
+      "step": 25346
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.964386224746704,
+      "learning_rate": 6.626797854022131e-06,
+      "loss": 0.6922,
+      "step": 25347
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.087589740753174,
+      "learning_rate": 6.6265485391261876e-06,
+      "loss": 0.7197,
+      "step": 25348
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 3.601039171218872,
+      "learning_rate": 6.62629921970742e-06,
+      "loss": 0.3318,
+      "step": 25349
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.5622148513793945,
+      "learning_rate": 6.626049895766524e-06,
+      "loss": 0.655,
+      "step": 25350
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.6785671710968018,
+      "learning_rate": 6.625800567304193e-06,
+      "loss": 0.4763,
+      "step": 25351
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.8445528745651245,
+      "learning_rate": 6.625551234321119e-06,
+      "loss": 0.5846,
+      "step": 25352
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.032592535018921,
+      "learning_rate": 6.6253018968179975e-06,
+      "loss": 0.7122,
+      "step": 25353
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.110255718231201,
+      "learning_rate": 6.625052554795517e-06,
+      "loss": 0.872,
+      "step": 25354
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.883165121078491,
+      "learning_rate": 6.624803208254377e-06,
+      "loss": 0.6555,
+      "step": 25355
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.6456868648529053,
+      "learning_rate": 6.624553857195266e-06,
+      "loss": 0.7862,
+      "step": 25356
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.363835573196411,
+      "learning_rate": 6.624304501618881e-06,
+      "loss": 0.5426,
+      "step": 25357
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.2470033168792725,
+      "learning_rate": 6.624055141525914e-06,
+      "loss": 0.7266,
+      "step": 25358
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 3.305128335952759,
+      "learning_rate": 6.623805776917056e-06,
+      "loss": 0.7048,
+      "step": 25359
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.7642128467559814,
+      "learning_rate": 6.623556407793004e-06,
+      "loss": 0.7301,
+      "step": 25360
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.917770504951477,
+      "learning_rate": 6.623307034154449e-06,
+      "loss": 0.5862,
+      "step": 25361
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 3.0397329330444336,
+      "learning_rate": 6.623057656002085e-06,
+      "loss": 0.6226,
+      "step": 25362
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.3192334175109863,
+      "learning_rate": 6.622808273336606e-06,
+      "loss": 0.4182,
+      "step": 25363
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.9846417903900146,
+      "learning_rate": 6.622558886158705e-06,
+      "loss": 0.5389,
+      "step": 25364
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 3.3640263080596924,
+      "learning_rate": 6.622309494469076e-06,
+      "loss": 0.7422,
+      "step": 25365
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.7713412046432495,
+      "learning_rate": 6.6220600982684124e-06,
+      "loss": 0.6255,
+      "step": 25366
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.7103333473205566,
+      "learning_rate": 6.621810697557407e-06,
+      "loss": 0.6069,
+      "step": 25367
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.8019449710845947,
+      "learning_rate": 6.621561292336752e-06,
+      "loss": 0.6095,
+      "step": 25368
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.0288820266723633,
+      "learning_rate": 6.621311882607144e-06,
+      "loss": 0.7617,
+      "step": 25369
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.8060784339904785,
+      "learning_rate": 6.621062468369275e-06,
+      "loss": 0.4307,
+      "step": 25370
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.8556824922561646,
+      "learning_rate": 6.620813049623838e-06,
+      "loss": 0.5697,
+      "step": 25371
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 3.0772626399993896,
+      "learning_rate": 6.620563626371528e-06,
+      "loss": 0.6095,
+      "step": 25372
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 3.0739810466766357,
+      "learning_rate": 6.620314198613037e-06,
+      "loss": 0.5719,
+      "step": 25373
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.161198139190674,
+      "learning_rate": 6.6200647663490596e-06,
+      "loss": 0.6945,
+      "step": 25374
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 3.002561092376709,
+      "learning_rate": 6.6198153295802884e-06,
+      "loss": 0.6931,
+      "step": 25375
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.1874496936798096,
+      "learning_rate": 6.619565888307418e-06,
+      "loss": 0.4674,
+      "step": 25376
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 3.0029029846191406,
+      "learning_rate": 6.619316442531141e-06,
+      "loss": 0.8261,
+      "step": 25377
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.634406328201294,
+      "learning_rate": 6.619066992252152e-06,
+      "loss": 0.6763,
+      "step": 25378
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.9066296815872192,
+      "learning_rate": 6.618817537471144e-06,
+      "loss": 0.5683,
+      "step": 25379
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.8247485160827637,
+      "learning_rate": 6.618568078188811e-06,
+      "loss": 0.6507,
+      "step": 25380
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 3.2122325897216797,
+      "learning_rate": 6.618318614405847e-06,
+      "loss": 0.7384,
+      "step": 25381
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 4.187242031097412,
+      "learning_rate": 6.618069146122945e-06,
+      "loss": 0.7919,
+      "step": 25382
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.3389604091644287,
+      "learning_rate": 6.617819673340799e-06,
+      "loss": 0.5034,
+      "step": 25383
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.044001579284668,
+      "learning_rate": 6.617570196060101e-06,
+      "loss": 0.7735,
+      "step": 25384
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.5312542915344238,
+      "learning_rate": 6.6173207142815475e-06,
+      "loss": 0.4131,
+      "step": 25385
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.9543660879135132,
+      "learning_rate": 6.61707122800583e-06,
+      "loss": 0.7551,
+      "step": 25386
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 4.340054035186768,
+      "learning_rate": 6.616821737233645e-06,
+      "loss": 0.5443,
+      "step": 25387
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.407144069671631,
+      "learning_rate": 6.616572241965683e-06,
+      "loss": 0.5983,
+      "step": 25388
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.7571029663085938,
+      "learning_rate": 6.61632274220264e-06,
+      "loss": 0.6576,
+      "step": 25389
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.8383090496063232,
+      "learning_rate": 6.616073237945209e-06,
+      "loss": 0.8164,
+      "step": 25390
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.319547414779663,
+      "learning_rate": 6.615823729194083e-06,
+      "loss": 0.8013,
+      "step": 25391
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.0392589569091797,
+      "learning_rate": 6.615574215949957e-06,
+      "loss": 0.5731,
+      "step": 25392
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.9456307888031006,
+      "learning_rate": 6.615324698213525e-06,
+      "loss": 0.8221,
+      "step": 25393
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.024582624435425,
+      "learning_rate": 6.615075175985478e-06,
+      "loss": 0.5728,
+      "step": 25394
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.936762571334839,
+      "learning_rate": 6.614825649266512e-06,
+      "loss": 0.5602,
+      "step": 25395
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.6760473251342773,
+      "learning_rate": 6.614576118057323e-06,
+      "loss": 0.7991,
+      "step": 25396
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 3.4536290168762207,
+      "learning_rate": 6.614326582358601e-06,
+      "loss": 0.7025,
+      "step": 25397
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.175861120223999,
+      "learning_rate": 6.614077042171043e-06,
+      "loss": 0.6616,
+      "step": 25398
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.861222743988037,
+      "learning_rate": 6.61382749749534e-06,
+      "loss": 0.6028,
+      "step": 25399
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.449385166168213,
+      "learning_rate": 6.613577948332188e-06,
+      "loss": 0.645,
+      "step": 25400
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.546931266784668,
+      "learning_rate": 6.61332839468228e-06,
+      "loss": 0.6103,
+      "step": 25401
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.808764100074768,
+      "learning_rate": 6.61307883654631e-06,
+      "loss": 0.6849,
+      "step": 25402
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.2162811756134033,
+      "learning_rate": 6.612829273924972e-06,
+      "loss": 0.6659,
+      "step": 25403
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.2652390003204346,
+      "learning_rate": 6.612579706818961e-06,
+      "loss": 0.6624,
+      "step": 25404
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.670969009399414,
+      "learning_rate": 6.612330135228968e-06,
+      "loss": 0.8693,
+      "step": 25405
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.602562665939331,
+      "learning_rate": 6.612080559155691e-06,
+      "loss": 0.7457,
+      "step": 25406
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.595949649810791,
+      "learning_rate": 6.61183097859982e-06,
+      "loss": 0.6333,
+      "step": 25407
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.125807046890259,
+      "learning_rate": 6.611581393562052e-06,
+      "loss": 0.7153,
+      "step": 25408
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.992696762084961,
+      "learning_rate": 6.6113318040430795e-06,
+      "loss": 0.5046,
+      "step": 25409
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.0055477619171143,
+      "learning_rate": 6.6110822100435965e-06,
+      "loss": 0.7847,
+      "step": 25410
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.2046828269958496,
+      "learning_rate": 6.610832611564298e-06,
+      "loss": 0.5588,
+      "step": 25411
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.9988840818405151,
+      "learning_rate": 6.610583008605878e-06,
+      "loss": 0.6985,
+      "step": 25412
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.8253400325775146,
+      "learning_rate": 6.610333401169029e-06,
+      "loss": 0.8056,
+      "step": 25413
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.753765821456909,
+      "learning_rate": 6.610083789254445e-06,
+      "loss": 0.4309,
+      "step": 25414
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.4192390441894531,
+      "learning_rate": 6.609834172862822e-06,
+      "loss": 0.4781,
+      "step": 25415
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.3618931770324707,
+      "learning_rate": 6.609584551994854e-06,
+      "loss": 0.5381,
+      "step": 25416
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.868192195892334,
+      "learning_rate": 6.609334926651234e-06,
+      "loss": 0.54,
+      "step": 25417
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.1454827785491943,
+      "learning_rate": 6.609085296832657e-06,
+      "loss": 0.5502,
+      "step": 25418
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.8275885581970215,
+      "learning_rate": 6.608835662539815e-06,
+      "loss": 0.7405,
+      "step": 25419
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.657738447189331,
+      "learning_rate": 6.608586023773406e-06,
+      "loss": 0.5279,
+      "step": 25420
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.937598466873169,
+      "learning_rate": 6.60833638053412e-06,
+      "loss": 0.5057,
+      "step": 25421
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 4.3658928871154785,
+      "learning_rate": 6.608086732822654e-06,
+      "loss": 1.0448,
+      "step": 25422
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.779284954071045,
+      "learning_rate": 6.607837080639701e-06,
+      "loss": 0.7929,
+      "step": 25423
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 5.1484479904174805,
+      "learning_rate": 6.607587423985953e-06,
+      "loss": 0.5987,
+      "step": 25424
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 3.4877450466156006,
+      "learning_rate": 6.607337762862109e-06,
+      "loss": 0.7234,
+      "step": 25425
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.8986682891845703,
+      "learning_rate": 6.607088097268861e-06,
+      "loss": 0.5113,
+      "step": 25426
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 6.065760612487793,
+      "learning_rate": 6.606838427206903e-06,
+      "loss": 0.5678,
+      "step": 25427
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.7468597888946533,
+      "learning_rate": 6.606588752676928e-06,
+      "loss": 0.8836,
+      "step": 25428
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.6072299480438232,
+      "learning_rate": 6.6063390736796325e-06,
+      "loss": 0.5727,
+      "step": 25429
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 4.688535213470459,
+      "learning_rate": 6.606089390215709e-06,
+      "loss": 0.7148,
+      "step": 25430
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.452486753463745,
+      "learning_rate": 6.605839702285853e-06,
+      "loss": 0.6919,
+      "step": 25431
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 3.4664909839630127,
+      "learning_rate": 6.605590009890759e-06,
+      "loss": 0.6677,
+      "step": 25432
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.1231751441955566,
+      "learning_rate": 6.60534031303112e-06,
+      "loss": 0.7834,
+      "step": 25433
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 3.05790638923645,
+      "learning_rate": 6.605090611707632e-06,
+      "loss": 0.8975,
+      "step": 25434
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.2996609210968018,
+      "learning_rate": 6.604840905920986e-06,
+      "loss": 0.5693,
+      "step": 25435
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 3.4298250675201416,
+      "learning_rate": 6.60459119567188e-06,
+      "loss": 0.5665,
+      "step": 25436
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.2004756927490234,
+      "learning_rate": 6.604341480961007e-06,
+      "loss": 0.7087,
+      "step": 25437
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 3.1179966926574707,
+      "learning_rate": 6.604091761789062e-06,
+      "loss": 0.7084,
+      "step": 25438
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.8337011337280273,
+      "learning_rate": 6.603842038156738e-06,
+      "loss": 0.6542,
+      "step": 25439
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 4.331559658050537,
+      "learning_rate": 6.60359231006473e-06,
+      "loss": 0.5384,
+      "step": 25440
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.4924144744873047,
+      "learning_rate": 6.603342577513733e-06,
+      "loss": 0.4262,
+      "step": 25441
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.629488229751587,
+      "learning_rate": 6.60309284050444e-06,
+      "loss": 0.5534,
+      "step": 25442
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.1924567222595215,
+      "learning_rate": 6.602843099037548e-06,
+      "loss": 0.6005,
+      "step": 25443
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 3.608536720275879,
+      "learning_rate": 6.602593353113749e-06,
+      "loss": 0.6883,
+      "step": 25444
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 3.3655624389648438,
+      "learning_rate": 6.602343602733737e-06,
+      "loss": 0.8147,
+      "step": 25445
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.2635347843170166,
+      "learning_rate": 6.602093847898207e-06,
+      "loss": 0.6338,
+      "step": 25446
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.8474286794662476,
+      "learning_rate": 6.601844088607857e-06,
+      "loss": 0.7282,
+      "step": 25447
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.3661699295043945,
+      "learning_rate": 6.601594324863377e-06,
+      "loss": 0.4642,
+      "step": 25448
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.632698655128479,
+      "learning_rate": 6.601344556665465e-06,
+      "loss": 0.4645,
+      "step": 25449
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.956339716911316,
+      "learning_rate": 6.601094784014812e-06,
+      "loss": 0.612,
+      "step": 25450
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 3.472334146499634,
+      "learning_rate": 6.600845006912114e-06,
+      "loss": 0.726,
+      "step": 25451
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.7527687549591064,
+      "learning_rate": 6.600595225358066e-06,
+      "loss": 0.5499,
+      "step": 25452
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 3.1196482181549072,
+      "learning_rate": 6.600345439353363e-06,
+      "loss": 0.6153,
+      "step": 25453
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 4.377305507659912,
+      "learning_rate": 6.600095648898699e-06,
+      "loss": 0.7407,
+      "step": 25454
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.9557372331619263,
+      "learning_rate": 6.5998458539947675e-06,
+      "loss": 0.7993,
+      "step": 25455
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.9578006267547607,
+      "learning_rate": 6.599596054642263e-06,
+      "loss": 0.7496,
+      "step": 25456
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.2949202060699463,
+      "learning_rate": 6.5993462508418825e-06,
+      "loss": 0.6212,
+      "step": 25457
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 3.0981154441833496,
+      "learning_rate": 6.5990964425943195e-06,
+      "loss": 0.8662,
+      "step": 25458
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.3717150688171387,
+      "learning_rate": 6.598846629900267e-06,
+      "loss": 0.8016,
+      "step": 25459
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.632046937942505,
+      "learning_rate": 6.598596812760423e-06,
+      "loss": 0.8494,
+      "step": 25460
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 3.6011147499084473,
+      "learning_rate": 6.598346991175478e-06,
+      "loss": 0.5612,
+      "step": 25461
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.2246274948120117,
+      "learning_rate": 6.59809716514613e-06,
+      "loss": 0.7369,
+      "step": 25462
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.5109461545944214,
+      "learning_rate": 6.597847334673073e-06,
+      "loss": 0.7057,
+      "step": 25463
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 6.578826427459717,
+      "learning_rate": 6.597597499757001e-06,
+      "loss": 0.7064,
+      "step": 25464
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.5663435459136963,
+      "learning_rate": 6.597347660398608e-06,
+      "loss": 0.633,
+      "step": 25465
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.5487076044082642,
+      "learning_rate": 6.597097816598588e-06,
+      "loss": 0.4558,
+      "step": 25466
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.4495320320129395,
+      "learning_rate": 6.596847968357639e-06,
+      "loss": 0.6146,
+      "step": 25467
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.0143790245056152,
+      "learning_rate": 6.596598115676454e-06,
+      "loss": 0.6182,
+      "step": 25468
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.555553674697876,
+      "learning_rate": 6.596348258555727e-06,
+      "loss": 0.6107,
+      "step": 25469
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 4.524078845977783,
+      "learning_rate": 6.596098396996155e-06,
+      "loss": 0.587,
+      "step": 25470
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 3.251790761947632,
+      "learning_rate": 6.59584853099843e-06,
+      "loss": 0.5654,
+      "step": 25471
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 4.250102519989014,
+      "learning_rate": 6.59559866056325e-06,
+      "loss": 0.4328,
+      "step": 25472
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.0528879165649414,
+      "learning_rate": 6.5953487856913055e-06,
+      "loss": 0.7624,
+      "step": 25473
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.9371932744979858,
+      "learning_rate": 6.595098906383295e-06,
+      "loss": 0.5803,
+      "step": 25474
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.8103060722351074,
+      "learning_rate": 6.594849022639911e-06,
+      "loss": 0.4581,
+      "step": 25475
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.169232130050659,
+      "learning_rate": 6.594599134461849e-06,
+      "loss": 0.7164,
+      "step": 25476
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 3.5798470973968506,
+      "learning_rate": 6.594349241849804e-06,
+      "loss": 0.6796,
+      "step": 25477
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.4608662128448486,
+      "learning_rate": 6.594099344804472e-06,
+      "loss": 0.6405,
+      "step": 25478
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.8236515522003174,
+      "learning_rate": 6.5938494433265475e-06,
+      "loss": 0.6566,
+      "step": 25479
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.7005103826522827,
+      "learning_rate": 6.5935995374167235e-06,
+      "loss": 0.4454,
+      "step": 25480
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.6893141269683838,
+      "learning_rate": 6.593349627075698e-06,
+      "loss": 0.6148,
+      "step": 25481
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.981851577758789,
+      "learning_rate": 6.593099712304163e-06,
+      "loss": 0.834,
+      "step": 25482
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 3.932339906692505,
+      "learning_rate": 6.592849793102813e-06,
+      "loss": 0.7176,
+      "step": 25483
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.031423807144165,
+      "learning_rate": 6.592599869472347e-06,
+      "loss": 0.3276,
+      "step": 25484
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 5.263908863067627,
+      "learning_rate": 6.5923499414134564e-06,
+      "loss": 0.6691,
+      "step": 25485
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 4.466301918029785,
+      "learning_rate": 6.592100008926835e-06,
+      "loss": 0.8068,
+      "step": 25486
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.6851872205734253,
+      "learning_rate": 6.591850072013182e-06,
+      "loss": 0.5352,
+      "step": 25487
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.8139657974243164,
+      "learning_rate": 6.59160013067319e-06,
+      "loss": 0.7009,
+      "step": 25488
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.9408159255981445,
+      "learning_rate": 6.591350184907554e-06,
+      "loss": 0.7195,
+      "step": 25489
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.660642623901367,
+      "learning_rate": 6.59110023471697e-06,
+      "loss": 0.7542,
+      "step": 25490
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.419683814048767,
+      "learning_rate": 6.590850280102133e-06,
+      "loss": 0.3907,
+      "step": 25491
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.8370447158813477,
+      "learning_rate": 6.590600321063734e-06,
+      "loss": 0.6416,
+      "step": 25492
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 3.134633779525757,
+      "learning_rate": 6.590350357602474e-06,
+      "loss": 0.5879,
+      "step": 25493
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.792182207107544,
+      "learning_rate": 6.590100389719045e-06,
+      "loss": 0.7592,
+      "step": 25494
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.200340986251831,
+      "learning_rate": 6.589850417414142e-06,
+      "loss": 0.4914,
+      "step": 25495
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.943544626235962,
+      "learning_rate": 6.589600440688462e-06,
+      "loss": 0.6402,
+      "step": 25496
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.80063533782959,
+      "learning_rate": 6.589350459542697e-06,
+      "loss": 0.8727,
+      "step": 25497
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.532041311264038,
+      "learning_rate": 6.5891004739775434e-06,
+      "loss": 0.6119,
+      "step": 25498
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 3.006453275680542,
+      "learning_rate": 6.588850483993697e-06,
+      "loss": 0.8637,
+      "step": 25499
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.6894307136535645,
+      "learning_rate": 6.588600489591854e-06,
+      "loss": 0.3621,
+      "step": 25500
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 3.0595273971557617,
+      "learning_rate": 6.588350490772707e-06,
+      "loss": 0.879,
+      "step": 25501
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.525794267654419,
+      "learning_rate": 6.588100487536952e-06,
+      "loss": 0.686,
+      "step": 25502
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.838141441345215,
+      "learning_rate": 6.587850479885286e-06,
+      "loss": 0.679,
+      "step": 25503
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.399444341659546,
+      "learning_rate": 6.587600467818403e-06,
+      "loss": 0.6742,
+      "step": 25504
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.324320077896118,
+      "learning_rate": 6.587350451336996e-06,
+      "loss": 0.7804,
+      "step": 25505
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.8815577030181885,
+      "learning_rate": 6.587100430441762e-06,
+      "loss": 0.5899,
+      "step": 25506
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.4752357006073,
+      "learning_rate": 6.586850405133396e-06,
+      "loss": 0.9988,
+      "step": 25507
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 3.157766342163086,
+      "learning_rate": 6.586600375412595e-06,
+      "loss": 0.4572,
+      "step": 25508
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.68723201751709,
+      "learning_rate": 6.586350341280052e-06,
+      "loss": 0.6395,
+      "step": 25509
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.396040916442871,
+      "learning_rate": 6.586100302736462e-06,
+      "loss": 0.6823,
+      "step": 25510
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.4998629093170166,
+      "learning_rate": 6.585850259782524e-06,
+      "loss": 0.8257,
+      "step": 25511
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.0951905250549316,
+      "learning_rate": 6.585600212418928e-06,
+      "loss": 0.5124,
+      "step": 25512
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.5918958187103271,
+      "learning_rate": 6.5853501606463735e-06,
+      "loss": 0.4258,
+      "step": 25513
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 4.123767852783203,
+      "learning_rate": 6.585100104465552e-06,
+      "loss": 0.9395,
+      "step": 25514
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 3.1807448863983154,
+      "learning_rate": 6.584850043877162e-06,
+      "loss": 0.6216,
+      "step": 25515
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.0225987434387207,
+      "learning_rate": 6.584599978881898e-06,
+      "loss": 0.7133,
+      "step": 25516
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 3.4479520320892334,
+      "learning_rate": 6.584349909480454e-06,
+      "loss": 0.7845,
+      "step": 25517
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.9159188270568848,
+      "learning_rate": 6.584099835673527e-06,
+      "loss": 0.3815,
+      "step": 25518
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.0152528285980225,
+      "learning_rate": 6.583849757461812e-06,
+      "loss": 0.6359,
+      "step": 25519
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.0888402462005615,
+      "learning_rate": 6.583599674846004e-06,
+      "loss": 0.7187,
+      "step": 25520
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.434091567993164,
+      "learning_rate": 6.5833495878267996e-06,
+      "loss": 0.7506,
+      "step": 25521
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.7878787517547607,
+      "learning_rate": 6.583099496404892e-06,
+      "loss": 0.5691,
+      "step": 25522
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.7612597942352295,
+      "learning_rate": 6.5828494005809786e-06,
+      "loss": 0.5041,
+      "step": 25523
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.7233965396881104,
+      "learning_rate": 6.582599300355753e-06,
+      "loss": 0.6758,
+      "step": 25524
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.089784622192383,
+      "learning_rate": 6.5823491957299115e-06,
+      "loss": 0.706,
+      "step": 25525
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.928527355194092,
+      "learning_rate": 6.58209908670415e-06,
+      "loss": 0.7397,
+      "step": 25526
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.543709397315979,
+      "learning_rate": 6.581848973279163e-06,
+      "loss": 0.8503,
+      "step": 25527
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.7817587852478027,
+      "learning_rate": 6.581598855455647e-06,
+      "loss": 0.6082,
+      "step": 25528
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 3.389052152633667,
+      "learning_rate": 6.581348733234298e-06,
+      "loss": 0.6731,
+      "step": 25529
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 3.3497774600982666,
+      "learning_rate": 6.58109860661581e-06,
+      "loss": 0.5223,
+      "step": 25530
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.199251174926758,
+      "learning_rate": 6.580848475600879e-06,
+      "loss": 0.5919,
+      "step": 25531
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.1179494857788086,
+      "learning_rate": 6.580598340190201e-06,
+      "loss": 0.3494,
+      "step": 25532
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.8573130369186401,
+      "learning_rate": 6.58034820038447e-06,
+      "loss": 0.6443,
+      "step": 25533
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.6381919384002686,
+      "learning_rate": 6.580098056184383e-06,
+      "loss": 0.7286,
+      "step": 25534
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 4.145601272583008,
+      "learning_rate": 6.579847907590636e-06,
+      "loss": 0.6657,
+      "step": 25535
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.6975067853927612,
+      "learning_rate": 6.579597754603923e-06,
+      "loss": 0.6655,
+      "step": 25536
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 4.347354888916016,
+      "learning_rate": 6.57934759722494e-06,
+      "loss": 0.525,
+      "step": 25537
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.5688729286193848,
+      "learning_rate": 6.579097435454384e-06,
+      "loss": 0.5338,
+      "step": 25538
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.3731529712677,
+      "learning_rate": 6.57884726929295e-06,
+      "loss": 0.7688,
+      "step": 25539
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.9408283233642578,
+      "learning_rate": 6.578597098741332e-06,
+      "loss": 0.6114,
+      "step": 25540
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 8.53222942352295,
+      "learning_rate": 6.578346923800227e-06,
+      "loss": 0.7501,
+      "step": 25541
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.9389725923538208,
+      "learning_rate": 6.578096744470332e-06,
+      "loss": 0.6288,
+      "step": 25542
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.4357056617736816,
+      "learning_rate": 6.577846560752339e-06,
+      "loss": 0.5284,
+      "step": 25543
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 4.318861961364746,
+      "learning_rate": 6.577596372646946e-06,
+      "loss": 0.8407,
+      "step": 25544
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.6995540857315063,
+      "learning_rate": 6.57734618015485e-06,
+      "loss": 0.6799,
+      "step": 25545
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.9615659713745117,
+      "learning_rate": 6.577095983276745e-06,
+      "loss": 0.5529,
+      "step": 25546
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.8778544664382935,
+      "learning_rate": 6.576845782013324e-06,
+      "loss": 0.427,
+      "step": 25547
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.860491394996643,
+      "learning_rate": 6.576595576365288e-06,
+      "loss": 0.7841,
+      "step": 25548
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.6595253944396973,
+      "learning_rate": 6.57634536633333e-06,
+      "loss": 0.6874,
+      "step": 25549
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.1559081077575684,
+      "learning_rate": 6.576095151918147e-06,
+      "loss": 0.6056,
+      "step": 25550
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.6174120903015137,
+      "learning_rate": 6.575844933120432e-06,
+      "loss": 0.4674,
+      "step": 25551
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 3.8261406421661377,
+      "learning_rate": 6.575594709940885e-06,
+      "loss": 0.5412,
+      "step": 25552
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.218909978866577,
+      "learning_rate": 6.575344482380197e-06,
+      "loss": 0.6129,
+      "step": 25553
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.728975534439087,
+      "learning_rate": 6.5750942504390655e-06,
+      "loss": 0.6601,
+      "step": 25554
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 3.3453922271728516,
+      "learning_rate": 6.574844014118189e-06,
+      "loss": 0.4021,
+      "step": 25555
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.121422290802002,
+      "learning_rate": 6.5745937734182605e-06,
+      "loss": 0.5224,
+      "step": 25556
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.336026191711426,
+      "learning_rate": 6.574343528339977e-06,
+      "loss": 0.5237,
+      "step": 25557
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.589719772338867,
+      "learning_rate": 6.574093278884033e-06,
+      "loss": 0.5598,
+      "step": 25558
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.448737621307373,
+      "learning_rate": 6.573843025051125e-06,
+      "loss": 0.7645,
+      "step": 25559
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.7630746364593506,
+      "learning_rate": 6.573592766841949e-06,
+      "loss": 0.5943,
+      "step": 25560
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.4957047700881958,
+      "learning_rate": 6.573342504257203e-06,
+      "loss": 0.4915,
+      "step": 25561
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.2565925121307373,
+      "learning_rate": 6.57309223729758e-06,
+      "loss": 0.5647,
+      "step": 25562
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.6894006729125977,
+      "learning_rate": 6.572841965963777e-06,
+      "loss": 0.6161,
+      "step": 25563
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 4.505602836608887,
+      "learning_rate": 6.572591690256489e-06,
+      "loss": 0.657,
+      "step": 25564
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.5500123500823975,
+      "learning_rate": 6.572341410176413e-06,
+      "loss": 0.6682,
+      "step": 25565
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.904805898666382,
+      "learning_rate": 6.5720911257242445e-06,
+      "loss": 0.59,
+      "step": 25566
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 3.1724348068237305,
+      "learning_rate": 6.5718408369006804e-06,
+      "loss": 0.5274,
+      "step": 25567
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 4.652194499969482,
+      "learning_rate": 6.571590543706413e-06,
+      "loss": 0.528,
+      "step": 25568
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.9098044633865356,
+      "learning_rate": 6.571340246142144e-06,
+      "loss": 0.5909,
+      "step": 25569
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.13852596282959,
+      "learning_rate": 6.571089944208564e-06,
+      "loss": 0.5764,
+      "step": 25570
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 3.0240087509155273,
+      "learning_rate": 6.570839637906373e-06,
+      "loss": 0.6225,
+      "step": 25571
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.069150447845459,
+      "learning_rate": 6.570589327236266e-06,
+      "loss": 0.455,
+      "step": 25572
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.102797508239746,
+      "learning_rate": 6.570339012198938e-06,
+      "loss": 0.8886,
+      "step": 25573
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.4740490913391113,
+      "learning_rate": 6.5700886927950845e-06,
+      "loss": 0.5875,
+      "step": 25574
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 4.251688480377197,
+      "learning_rate": 6.569838369025403e-06,
+      "loss": 0.6398,
+      "step": 25575
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.873598098754883,
+      "learning_rate": 6.569588040890589e-06,
+      "loss": 0.4838,
+      "step": 25576
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.806318521499634,
+      "learning_rate": 6.569337708391339e-06,
+      "loss": 0.5445,
+      "step": 25577
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.4696781635284424,
+      "learning_rate": 6.569087371528347e-06,
+      "loss": 0.8252,
+      "step": 25578
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 4.548944473266602,
+      "learning_rate": 6.568837030302313e-06,
+      "loss": 0.6825,
+      "step": 25579
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 3.70245623588562,
+      "learning_rate": 6.56858668471393e-06,
+      "loss": 0.6804,
+      "step": 25580
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.7836380004882812,
+      "learning_rate": 6.568336334763895e-06,
+      "loss": 0.5827,
+      "step": 25581
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.86517596244812,
+      "learning_rate": 6.568085980452905e-06,
+      "loss": 0.679,
+      "step": 25582
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.6407798528671265,
+      "learning_rate": 6.567835621781655e-06,
+      "loss": 0.6774,
+      "step": 25583
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.4122111797332764,
+      "learning_rate": 6.5675852587508415e-06,
+      "loss": 0.6754,
+      "step": 25584
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.585064649581909,
+      "learning_rate": 6.56733489136116e-06,
+      "loss": 0.597,
+      "step": 25585
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.926013946533203,
+      "learning_rate": 6.567084519613307e-06,
+      "loss": 0.7198,
+      "step": 25586
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.434565544128418,
+      "learning_rate": 6.5668341435079795e-06,
+      "loss": 0.553,
+      "step": 25587
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.066911458969116,
+      "learning_rate": 6.566583763045873e-06,
+      "loss": 0.7627,
+      "step": 25588
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.527297019958496,
+      "learning_rate": 6.566333378227684e-06,
+      "loss": 0.501,
+      "step": 25589
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.9468001127243042,
+      "learning_rate": 6.566082989054109e-06,
+      "loss": 0.6312,
+      "step": 25590
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.7743303775787354,
+      "learning_rate": 6.565832595525844e-06,
+      "loss": 0.6588,
+      "step": 25591
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.3729748725891113,
+      "learning_rate": 6.565582197643585e-06,
+      "loss": 0.8357,
+      "step": 25592
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.369575023651123,
+      "learning_rate": 6.5653317954080275e-06,
+      "loss": 0.7754,
+      "step": 25593
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.7339928150177002,
+      "learning_rate": 6.565081388819869e-06,
+      "loss": 0.6894,
+      "step": 25594
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 3.7882578372955322,
+      "learning_rate": 6.564830977879807e-06,
+      "loss": 0.805,
+      "step": 25595
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 3.6774768829345703,
+      "learning_rate": 6.564580562588535e-06,
+      "loss": 0.4033,
+      "step": 25596
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.9664762020111084,
+      "learning_rate": 6.564330142946751e-06,
+      "loss": 0.6967,
+      "step": 25597
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.8427220582962036,
+      "learning_rate": 6.56407971895515e-06,
+      "loss": 0.5584,
+      "step": 25598
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.106403350830078,
+      "learning_rate": 6.56382929061443e-06,
+      "loss": 0.8718,
+      "step": 25599
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.230560541152954,
+      "learning_rate": 6.563578857925286e-06,
+      "loss": 0.7274,
+      "step": 25600
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 3.3784873485565186,
+      "learning_rate": 6.563328420888415e-06,
+      "loss": 0.8948,
+      "step": 25601
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.8142666816711426,
+      "learning_rate": 6.563077979504513e-06,
+      "loss": 0.8495,
+      "step": 25602
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 3.8159518241882324,
+      "learning_rate": 6.562827533774278e-06,
+      "loss": 0.5665,
+      "step": 25603
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 3.0034074783325195,
+      "learning_rate": 6.562577083698404e-06,
+      "loss": 0.4755,
+      "step": 25604
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.8212058544158936,
+      "learning_rate": 6.5623266292775886e-06,
+      "loss": 0.9441,
+      "step": 25605
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.9298923015594482,
+      "learning_rate": 6.5620761705125295e-06,
+      "loss": 0.7532,
+      "step": 25606
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.58298659324646,
+      "learning_rate": 6.561825707403919e-06,
+      "loss": 0.5826,
+      "step": 25607
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.839235782623291,
+      "learning_rate": 6.561575239952458e-06,
+      "loss": 0.6157,
+      "step": 25608
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.3908486366271973,
+      "learning_rate": 6.561324768158841e-06,
+      "loss": 0.8533,
+      "step": 25609
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.0666544437408447,
+      "learning_rate": 6.5610742920237644e-06,
+      "loss": 0.4074,
+      "step": 25610
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.0805206298828125,
+      "learning_rate": 6.560823811547925e-06,
+      "loss": 0.5321,
+      "step": 25611
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.6263954639434814,
+      "learning_rate": 6.560573326732021e-06,
+      "loss": 0.4121,
+      "step": 25612
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.292173385620117,
+      "learning_rate": 6.560322837576745e-06,
+      "loss": 0.7249,
+      "step": 25613
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.3064382076263428,
+      "learning_rate": 6.560072344082797e-06,
+      "loss": 0.6343,
+      "step": 25614
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.4152445793151855,
+      "learning_rate": 6.559821846250871e-06,
+      "loss": 0.6697,
+      "step": 25615
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 3.31699538230896,
+      "learning_rate": 6.559571344081666e-06,
+      "loss": 0.8825,
+      "step": 25616
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.8771398067474365,
+      "learning_rate": 6.559320837575876e-06,
+      "loss": 0.5779,
+      "step": 25617
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 4.710187911987305,
+      "learning_rate": 6.5590703267342e-06,
+      "loss": 0.5422,
+      "step": 25618
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.5389299392700195,
+      "learning_rate": 6.558819811557333e-06,
+      "loss": 0.7675,
+      "step": 25619
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.7274829149246216,
+      "learning_rate": 6.558569292045972e-06,
+      "loss": 0.5521,
+      "step": 25620
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.5000202655792236,
+      "learning_rate": 6.558318768200813e-06,
+      "loss": 0.5359,
+      "step": 25621
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.8833110332489014,
+      "learning_rate": 6.558068240022555e-06,
+      "loss": 0.5186,
+      "step": 25622
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.3894619941711426,
+      "learning_rate": 6.557817707511891e-06,
+      "loss": 0.4711,
+      "step": 25623
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.263009548187256,
+      "learning_rate": 6.557567170669521e-06,
+      "loss": 0.6253,
+      "step": 25624
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.7514488697052002,
+      "learning_rate": 6.55731662949614e-06,
+      "loss": 0.5229,
+      "step": 25625
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.1483919620513916,
+      "learning_rate": 6.557066083992445e-06,
+      "loss": 0.4651,
+      "step": 25626
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.795230507850647,
+      "learning_rate": 6.556815534159131e-06,
+      "loss": 0.6489,
+      "step": 25627
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.5011487007141113,
+      "learning_rate": 6.556564979996898e-06,
+      "loss": 0.793,
+      "step": 25628
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 7.149918556213379,
+      "learning_rate": 6.556314421506439e-06,
+      "loss": 0.7779,
+      "step": 25629
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.930475950241089,
+      "learning_rate": 6.556063858688454e-06,
+      "loss": 0.4969,
+      "step": 25630
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.265557289123535,
+      "learning_rate": 6.555813291543638e-06,
+      "loss": 0.5277,
+      "step": 25631
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.7241792678833008,
+      "learning_rate": 6.555562720072688e-06,
+      "loss": 0.669,
+      "step": 25632
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.3272690773010254,
+      "learning_rate": 6.555312144276301e-06,
+      "loss": 0.9657,
+      "step": 25633
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 3.2222065925598145,
+      "learning_rate": 6.555061564155174e-06,
+      "loss": 0.6994,
+      "step": 25634
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.3891372680664062,
+      "learning_rate": 6.554810979710002e-06,
+      "loss": 0.7394,
+      "step": 25635
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 3.098055124282837,
+      "learning_rate": 6.554560390941484e-06,
+      "loss": 0.7641,
+      "step": 25636
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.0142245292663574,
+      "learning_rate": 6.554309797850317e-06,
+      "loss": 0.5395,
+      "step": 25637
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 3.716265916824341,
+      "learning_rate": 6.554059200437195e-06,
+      "loss": 0.7942,
+      "step": 25638
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.9476650953292847,
+      "learning_rate": 6.553808598702817e-06,
+      "loss": 0.6752,
+      "step": 25639
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 5.2657246589660645,
+      "learning_rate": 6.553557992647879e-06,
+      "loss": 1.0232,
+      "step": 25640
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.6898727416992188,
+      "learning_rate": 6.553307382273079e-06,
+      "loss": 0.6745,
+      "step": 25641
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.1092073917388916,
+      "learning_rate": 6.553056767579113e-06,
+      "loss": 0.5534,
+      "step": 25642
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.0744783878326416,
+      "learning_rate": 6.552806148566677e-06,
+      "loss": 0.5996,
+      "step": 25643
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 3.874765396118164,
+      "learning_rate": 6.5525555252364705e-06,
+      "loss": 0.5463,
+      "step": 25644
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.307328939437866,
+      "learning_rate": 6.552304897589187e-06,
+      "loss": 0.6345,
+      "step": 25645
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.404876470565796,
+      "learning_rate": 6.552054265625526e-06,
+      "loss": 0.7111,
+      "step": 25646
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.7775523662567139,
+      "learning_rate": 6.551803629346183e-06,
+      "loss": 0.7277,
+      "step": 25647
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.1081607341766357,
+      "learning_rate": 6.5515529887518555e-06,
+      "loss": 0.8549,
+      "step": 25648
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.5642831325531006,
+      "learning_rate": 6.55130234384324e-06,
+      "loss": 0.7407,
+      "step": 25649
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 3.175076484680176,
+      "learning_rate": 6.551051694621034e-06,
+      "loss": 0.7669,
+      "step": 25650
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.9671268463134766,
+      "learning_rate": 6.550801041085935e-06,
+      "loss": 0.9207,
+      "step": 25651
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.9469115734100342,
+      "learning_rate": 6.5505503832386385e-06,
+      "loss": 0.5805,
+      "step": 25652
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.7960195541381836,
+      "learning_rate": 6.550299721079842e-06,
+      "loss": 0.7181,
+      "step": 25653
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.1074600219726562,
+      "learning_rate": 6.5500490546102434e-06,
+      "loss": 0.7638,
+      "step": 25654
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.9246807098388672,
+      "learning_rate": 6.549798383830539e-06,
+      "loss": 0.7507,
+      "step": 25655
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.015662908554077,
+      "learning_rate": 6.549547708741426e-06,
+      "loss": 0.7717,
+      "step": 25656
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.3135735988616943,
+      "learning_rate": 6.549297029343602e-06,
+      "loss": 0.5257,
+      "step": 25657
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.6455271244049072,
+      "learning_rate": 6.549046345637762e-06,
+      "loss": 0.7217,
+      "step": 25658
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.8293673992156982,
+      "learning_rate": 6.5487956576246035e-06,
+      "loss": 0.6028,
+      "step": 25659
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.781722068786621,
+      "learning_rate": 6.548544965304826e-06,
+      "loss": 0.6114,
+      "step": 25660
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.6754770278930664,
+      "learning_rate": 6.548294268679125e-06,
+      "loss": 0.4077,
+      "step": 25661
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.096818208694458,
+      "learning_rate": 6.548043567748198e-06,
+      "loss": 0.7783,
+      "step": 25662
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 3.185990571975708,
+      "learning_rate": 6.54779286251274e-06,
+      "loss": 0.5225,
+      "step": 25663
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.7527060508728027,
+      "learning_rate": 6.547542152973452e-06,
+      "loss": 0.7077,
+      "step": 25664
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.43373966217041,
+      "learning_rate": 6.547291439131028e-06,
+      "loss": 0.4653,
+      "step": 25665
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.5829654932022095,
+      "learning_rate": 6.547040720986166e-06,
+      "loss": 0.5072,
+      "step": 25666
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 3.438385009765625,
+      "learning_rate": 6.546789998539564e-06,
+      "loss": 0.7374,
+      "step": 25667
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.614384889602661,
+      "learning_rate": 6.546539271791919e-06,
+      "loss": 0.4885,
+      "step": 25668
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.1712214946746826,
+      "learning_rate": 6.546288540743924e-06,
+      "loss": 0.5174,
+      "step": 25669
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 3.2714462280273438,
+      "learning_rate": 6.546037805396284e-06,
+      "loss": 0.5291,
+      "step": 25670
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.9970576763153076,
+      "learning_rate": 6.54578706574969e-06,
+      "loss": 0.6213,
+      "step": 25671
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.8520052433013916,
+      "learning_rate": 6.545536321804842e-06,
+      "loss": 0.8188,
+      "step": 25672
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.5357751846313477,
+      "learning_rate": 6.545285573562436e-06,
+      "loss": 0.6023,
+      "step": 25673
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.450920343399048,
+      "learning_rate": 6.54503482102317e-06,
+      "loss": 0.4891,
+      "step": 25674
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 3.4599685668945312,
+      "learning_rate": 6.54478406418774e-06,
+      "loss": 0.8989,
+      "step": 25675
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.120162010192871,
+      "learning_rate": 6.544533303056846e-06,
+      "loss": 0.4844,
+      "step": 25676
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.2363505363464355,
+      "learning_rate": 6.544282537631182e-06,
+      "loss": 0.4551,
+      "step": 25677
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.7287200689315796,
+      "learning_rate": 6.5440317679114465e-06,
+      "loss": 0.4122,
+      "step": 25678
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.9960482120513916,
+      "learning_rate": 6.543780993898338e-06,
+      "loss": 0.6009,
+      "step": 25679
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 4.676633834838867,
+      "learning_rate": 6.543530215592553e-06,
+      "loss": 0.554,
+      "step": 25680
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.7043092250823975,
+      "learning_rate": 6.543279432994788e-06,
+      "loss": 0.6414,
+      "step": 25681
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 3.592355251312256,
+      "learning_rate": 6.543028646105742e-06,
+      "loss": 0.4707,
+      "step": 25682
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 3.2403481006622314,
+      "learning_rate": 6.542777854926109e-06,
+      "loss": 0.7666,
+      "step": 25683
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.0090911388397217,
+      "learning_rate": 6.54252705945659e-06,
+      "loss": 0.7396,
+      "step": 25684
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 3.326308488845825,
+      "learning_rate": 6.542276259697882e-06,
+      "loss": 0.875,
+      "step": 25685
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.8456306457519531,
+      "learning_rate": 6.542025455650681e-06,
+      "loss": 0.6229,
+      "step": 25686
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.8851191997528076,
+      "learning_rate": 6.541774647315685e-06,
+      "loss": 0.6645,
+      "step": 25687
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.7759311199188232,
+      "learning_rate": 6.541523834693592e-06,
+      "loss": 0.501,
+      "step": 25688
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.790208101272583,
+      "learning_rate": 6.541273017785097e-06,
+      "loss": 0.4595,
+      "step": 25689
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.705251693725586,
+      "learning_rate": 6.5410221965909e-06,
+      "loss": 0.6679,
+      "step": 25690
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.3062479496002197,
+      "learning_rate": 6.540771371111697e-06,
+      "loss": 0.6992,
+      "step": 25691
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.6698333024978638,
+      "learning_rate": 6.540520541348187e-06,
+      "loss": 0.5286,
+      "step": 25692
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.416072130203247,
+      "learning_rate": 6.540269707301065e-06,
+      "loss": 0.5739,
+      "step": 25693
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.4452736377716064,
+      "learning_rate": 6.540018868971032e-06,
+      "loss": 0.6142,
+      "step": 25694
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.780543565750122,
+      "learning_rate": 6.5397680263587815e-06,
+      "loss": 0.7222,
+      "step": 25695
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.266281843185425,
+      "learning_rate": 6.539517179465015e-06,
+      "loss": 0.434,
+      "step": 25696
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.5260825157165527,
+      "learning_rate": 6.539266328290426e-06,
+      "loss": 0.6805,
+      "step": 25697
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.07079815864563,
+      "learning_rate": 6.539015472835715e-06,
+      "loss": 0.6628,
+      "step": 25698
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.3037641048431396,
+      "learning_rate": 6.538764613101578e-06,
+      "loss": 0.5217,
+      "step": 25699
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.5555670261383057,
+      "learning_rate": 6.538513749088713e-06,
+      "loss": 0.9256,
+      "step": 25700
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 3.316293716430664,
+      "learning_rate": 6.5382628807978175e-06,
+      "loss": 0.6992,
+      "step": 25701
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.8584439754486084,
+      "learning_rate": 6.538012008229589e-06,
+      "loss": 0.6657,
+      "step": 25702
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 3.75152325630188,
+      "learning_rate": 6.537761131384727e-06,
+      "loss": 0.5528,
+      "step": 25703
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.0236785411834717,
+      "learning_rate": 6.5375102502639255e-06,
+      "loss": 0.872,
+      "step": 25704
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.891116142272949,
+      "learning_rate": 6.537259364867885e-06,
+      "loss": 0.5021,
+      "step": 25705
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.7551802396774292,
+      "learning_rate": 6.537008475197302e-06,
+      "loss": 0.583,
+      "step": 25706
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.889073133468628,
+      "learning_rate": 6.536757581252872e-06,
+      "loss": 0.6679,
+      "step": 25707
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.6957257986068726,
+      "learning_rate": 6.536506683035297e-06,
+      "loss": 0.6455,
+      "step": 25708
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 3.883633613586426,
+      "learning_rate": 6.536255780545273e-06,
+      "loss": 0.7208,
+      "step": 25709
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 4.2416090965271,
+      "learning_rate": 6.536004873783496e-06,
+      "loss": 0.5417,
+      "step": 25710
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.9619531631469727,
+      "learning_rate": 6.535753962750665e-06,
+      "loss": 0.7364,
+      "step": 25711
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.9185205698013306,
+      "learning_rate": 6.535503047447478e-06,
+      "loss": 0.4045,
+      "step": 25712
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.689976453781128,
+      "learning_rate": 6.535252127874631e-06,
+      "loss": 0.6749,
+      "step": 25713
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.9336223602294922,
+      "learning_rate": 6.535001204032825e-06,
+      "loss": 0.6049,
+      "step": 25714
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 4.3253350257873535,
+      "learning_rate": 6.534750275922754e-06,
+      "loss": 0.7694,
+      "step": 25715
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.9410529136657715,
+      "learning_rate": 6.534499343545119e-06,
+      "loss": 0.6221,
+      "step": 25716
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.0167925357818604,
+      "learning_rate": 6.534248406900615e-06,
+      "loss": 0.7764,
+      "step": 25717
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.6519129276275635,
+      "learning_rate": 6.533997465989941e-06,
+      "loss": 0.6398,
+      "step": 25718
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.720842719078064,
+      "learning_rate": 6.533746520813795e-06,
+      "loss": 0.737,
+      "step": 25719
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.7241053581237793,
+      "learning_rate": 6.5334955713728736e-06,
+      "loss": 0.7802,
+      "step": 25720
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 4.979592800140381,
+      "learning_rate": 6.533244617667875e-06,
+      "loss": 0.74,
+      "step": 25721
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 3.123044490814209,
+      "learning_rate": 6.5329936596995e-06,
+      "loss": 0.5103,
+      "step": 25722
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.3658154010772705,
+      "learning_rate": 6.532742697468443e-06,
+      "loss": 0.7456,
+      "step": 25723
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 3.7690107822418213,
+      "learning_rate": 6.5324917309754025e-06,
+      "loss": 0.5117,
+      "step": 25724
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.283644676208496,
+      "learning_rate": 6.532240760221076e-06,
+      "loss": 0.5842,
+      "step": 25725
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.202784776687622,
+      "learning_rate": 6.531989785206162e-06,
+      "loss": 0.6087,
+      "step": 25726
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.907435894012451,
+      "learning_rate": 6.531738805931359e-06,
+      "loss": 0.5734,
+      "step": 25727
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 3.2647533416748047,
+      "learning_rate": 6.5314878223973645e-06,
+      "loss": 0.8577,
+      "step": 25728
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.3687353134155273,
+      "learning_rate": 6.5312368346048745e-06,
+      "loss": 0.7308,
+      "step": 25729
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.0227060317993164,
+      "learning_rate": 6.530985842554589e-06,
+      "loss": 0.3977,
+      "step": 25730
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.508531093597412,
+      "learning_rate": 6.530734846247205e-06,
+      "loss": 0.9406,
+      "step": 25731
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.9714882373809814,
+      "learning_rate": 6.530483845683422e-06,
+      "loss": 0.7888,
+      "step": 25732
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.525860548019409,
+      "learning_rate": 6.530232840863936e-06,
+      "loss": 0.3821,
+      "step": 25733
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 3.9431722164154053,
+      "learning_rate": 6.529981831789447e-06,
+      "loss": 0.3897,
+      "step": 25734
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.9550297260284424,
+      "learning_rate": 6.52973081846065e-06,
+      "loss": 0.6185,
+      "step": 25735
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.4142236709594727,
+      "learning_rate": 6.529479800878246e-06,
+      "loss": 0.5357,
+      "step": 25736
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.617002487182617,
+      "learning_rate": 6.529228779042931e-06,
+      "loss": 0.6789,
+      "step": 25737
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.1578590869903564,
+      "learning_rate": 6.5289777529554035e-06,
+      "loss": 0.63,
+      "step": 25738
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 4.930082321166992,
+      "learning_rate": 6.5287267226163605e-06,
+      "loss": 0.5759,
+      "step": 25739
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.5409231185913086,
+      "learning_rate": 6.528475688026503e-06,
+      "loss": 0.5563,
+      "step": 25740
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.4101808071136475,
+      "learning_rate": 6.528224649186526e-06,
+      "loss": 0.7012,
+      "step": 25741
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 4.769445896148682,
+      "learning_rate": 6.52797360609713e-06,
+      "loss": 0.5208,
+      "step": 25742
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.8476805686950684,
+      "learning_rate": 6.527722558759011e-06,
+      "loss": 0.7407,
+      "step": 25743
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 4.007918357849121,
+      "learning_rate": 6.5274715071728666e-06,
+      "loss": 0.8019,
+      "step": 25744
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.324389934539795,
+      "learning_rate": 6.527220451339397e-06,
+      "loss": 0.7938,
+      "step": 25745
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.4039371013641357,
+      "learning_rate": 6.526969391259301e-06,
+      "loss": 0.4429,
+      "step": 25746
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.8470180034637451,
+      "learning_rate": 6.526718326933275e-06,
+      "loss": 0.6332,
+      "step": 25747
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 3.666778802871704,
+      "learning_rate": 6.526467258362016e-06,
+      "loss": 0.5612,
+      "step": 25748
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.3547353744506836,
+      "learning_rate": 6.526216185546224e-06,
+      "loss": 0.4656,
+      "step": 25749
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.2667319774627686,
+      "learning_rate": 6.525965108486596e-06,
+      "loss": 0.5386,
+      "step": 25750
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.8413125276565552,
+      "learning_rate": 6.525714027183829e-06,
+      "loss": 0.6409,
+      "step": 25751
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 3.1178700923919678,
+      "learning_rate": 6.525462941638625e-06,
+      "loss": 0.8332,
+      "step": 25752
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.7095608711242676,
+      "learning_rate": 6.52521185185168e-06,
+      "loss": 0.7579,
+      "step": 25753
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.4671413898468018,
+      "learning_rate": 6.524960757823693e-06,
+      "loss": 0.5141,
+      "step": 25754
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.551550030708313,
+      "learning_rate": 6.52470965955536e-06,
+      "loss": 0.6102,
+      "step": 25755
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 3.125934362411499,
+      "learning_rate": 6.524458557047381e-06,
+      "loss": 0.7207,
+      "step": 25756
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.168358564376831,
+      "learning_rate": 6.5242074503004535e-06,
+      "loss": 0.7392,
+      "step": 25757
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.885707974433899,
+      "learning_rate": 6.523956339315276e-06,
+      "loss": 0.6519,
+      "step": 25758
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 3.8524885177612305,
+      "learning_rate": 6.523705224092548e-06,
+      "loss": 0.6634,
+      "step": 25759
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.446786403656006,
+      "learning_rate": 6.5234541046329655e-06,
+      "loss": 0.5818,
+      "step": 25760
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 3.3325955867767334,
+      "learning_rate": 6.523202980937227e-06,
+      "loss": 0.5896,
+      "step": 25761
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.8272147178649902,
+      "learning_rate": 6.522951853006033e-06,
+      "loss": 0.8259,
+      "step": 25762
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.6775460243225098,
+      "learning_rate": 6.522700720840081e-06,
+      "loss": 0.6826,
+      "step": 25763
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.3637592792510986,
+      "learning_rate": 6.522449584440067e-06,
+      "loss": 0.5694,
+      "step": 25764
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.679317593574524,
+      "learning_rate": 6.522198443806692e-06,
+      "loss": 0.5594,
+      "step": 25765
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.2355000972747803,
+      "learning_rate": 6.5219472989406526e-06,
+      "loss": 0.767,
+      "step": 25766
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 4.594425678253174,
+      "learning_rate": 6.521696149842649e-06,
+      "loss": 0.5664,
+      "step": 25767
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 3.518798589706421,
+      "learning_rate": 6.521444996513378e-06,
+      "loss": 0.6018,
+      "step": 25768
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 3.1136057376861572,
+      "learning_rate": 6.521193838953539e-06,
+      "loss": 0.6882,
+      "step": 25769
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.0039336681365967,
+      "learning_rate": 6.520942677163828e-06,
+      "loss": 0.7371,
+      "step": 25770
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.5168728828430176,
+      "learning_rate": 6.520691511144946e-06,
+      "loss": 0.5903,
+      "step": 25771
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.903200626373291,
+      "learning_rate": 6.520440340897591e-06,
+      "loss": 0.7142,
+      "step": 25772
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 3.6975109577178955,
+      "learning_rate": 6.520189166422462e-06,
+      "loss": 0.6069,
+      "step": 25773
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 3.2406504154205322,
+      "learning_rate": 6.519937987720254e-06,
+      "loss": 0.9166,
+      "step": 25774
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.398785352706909,
+      "learning_rate": 6.51968680479167e-06,
+      "loss": 0.8625,
+      "step": 25775
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.4655535221099854,
+      "learning_rate": 6.5194356176374056e-06,
+      "loss": 0.7203,
+      "step": 25776
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.839600682258606,
+      "learning_rate": 6.51918442625816e-06,
+      "loss": 0.5724,
+      "step": 25777
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.997149109840393,
+      "learning_rate": 6.51893323065463e-06,
+      "loss": 0.6448,
+      "step": 25778
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.8266438245773315,
+      "learning_rate": 6.5186820308275175e-06,
+      "loss": 0.564,
+      "step": 25779
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.735410690307617,
+      "learning_rate": 6.518430826777519e-06,
+      "loss": 0.9978,
+      "step": 25780
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.886508584022522,
+      "learning_rate": 6.5181796185053315e-06,
+      "loss": 0.7654,
+      "step": 25781
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.7050981521606445,
+      "learning_rate": 6.5179284060116575e-06,
+      "loss": 0.9825,
+      "step": 25782
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 5.8196845054626465,
+      "learning_rate": 6.517677189297192e-06,
+      "loss": 1.1283,
+      "step": 25783
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.689345121383667,
+      "learning_rate": 6.517425968362635e-06,
+      "loss": 0.5885,
+      "step": 25784
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.7861671447753906,
+      "learning_rate": 6.517174743208685e-06,
+      "loss": 0.5341,
+      "step": 25785
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.1439056396484375,
+      "learning_rate": 6.516923513836041e-06,
+      "loss": 0.7232,
+      "step": 25786
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 5.374731063842773,
+      "learning_rate": 6.5166722802454e-06,
+      "loss": 0.5443,
+      "step": 25787
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 3.2899038791656494,
+      "learning_rate": 6.516421042437462e-06,
+      "loss": 0.6601,
+      "step": 25788
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.464001417160034,
+      "learning_rate": 6.5161698004129244e-06,
+      "loss": 0.5689,
+      "step": 25789
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.049372911453247,
+      "learning_rate": 6.515918554172488e-06,
+      "loss": 0.7268,
+      "step": 25790
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.534862756729126,
+      "learning_rate": 6.515667303716848e-06,
+      "loss": 0.595,
+      "step": 25791
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 4.116041660308838,
+      "learning_rate": 6.515416049046706e-06,
+      "loss": 0.4747,
+      "step": 25792
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.8190736770629883,
+      "learning_rate": 6.51516479016276e-06,
+      "loss": 0.6172,
+      "step": 25793
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.951452374458313,
+      "learning_rate": 6.514913527065708e-06,
+      "loss": 0.6596,
+      "step": 25794
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.481483221054077,
+      "learning_rate": 6.514662259756249e-06,
+      "loss": 0.5818,
+      "step": 25795
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.1076903343200684,
+      "learning_rate": 6.5144109882350824e-06,
+      "loss": 0.712,
+      "step": 25796
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.1081459522247314,
+      "learning_rate": 6.514159712502905e-06,
+      "loss": 0.7226,
+      "step": 25797
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 3.1300454139709473,
+      "learning_rate": 6.513908432560417e-06,
+      "loss": 0.8326,
+      "step": 25798
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.7089040279388428,
+      "learning_rate": 6.5136571484083164e-06,
+      "loss": 0.5963,
+      "step": 25799
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 3.1089425086975098,
+      "learning_rate": 6.513405860047302e-06,
+      "loss": 0.4094,
+      "step": 25800
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.732232689857483,
+      "learning_rate": 6.513154567478075e-06,
+      "loss": 0.5125,
+      "step": 25801
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 4.067490100860596,
+      "learning_rate": 6.51290327070133e-06,
+      "loss": 0.5432,
+      "step": 25802
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.78696608543396,
+      "learning_rate": 6.512651969717768e-06,
+      "loss": 0.4496,
+      "step": 25803
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.1406702995300293,
+      "learning_rate": 6.512400664528089e-06,
+      "loss": 0.6916,
+      "step": 25804
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.265157699584961,
+      "learning_rate": 6.512149355132989e-06,
+      "loss": 0.5945,
+      "step": 25805
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 3.3808176517486572,
+      "learning_rate": 6.511898041533167e-06,
+      "loss": 0.7047,
+      "step": 25806
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.8067479133605957,
+      "learning_rate": 6.511646723729325e-06,
+      "loss": 0.6543,
+      "step": 25807
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.5724785327911377,
+      "learning_rate": 6.511395401722158e-06,
+      "loss": 0.5596,
+      "step": 25808
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.7407031059265137,
+      "learning_rate": 6.511144075512368e-06,
+      "loss": 0.7004,
+      "step": 25809
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 3.3787970542907715,
+      "learning_rate": 6.510892745100652e-06,
+      "loss": 0.7539,
+      "step": 25810
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 3.88962459564209,
+      "learning_rate": 6.510641410487709e-06,
+      "loss": 0.7642,
+      "step": 25811
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.6312878131866455,
+      "learning_rate": 6.510390071674237e-06,
+      "loss": 0.6314,
+      "step": 25812
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.969389796257019,
+      "learning_rate": 6.5101387286609386e-06,
+      "loss": 0.6872,
+      "step": 25813
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.5493630170822144,
+      "learning_rate": 6.509887381448509e-06,
+      "loss": 0.4148,
+      "step": 25814
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.9212836027145386,
+      "learning_rate": 6.509636030037647e-06,
+      "loss": 0.5634,
+      "step": 25815
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 4.996562957763672,
+      "learning_rate": 6.509384674429054e-06,
+      "loss": 0.4281,
+      "step": 25816
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 3.679485559463501,
+      "learning_rate": 6.5091333146234285e-06,
+      "loss": 0.6115,
+      "step": 25817
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.3252737522125244,
+      "learning_rate": 6.508881950621468e-06,
+      "loss": 0.8552,
+      "step": 25818
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.2175893783569336,
+      "learning_rate": 6.508630582423871e-06,
+      "loss": 0.6113,
+      "step": 25819
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.7850754261016846,
+      "learning_rate": 6.5083792100313385e-06,
+      "loss": 0.6247,
+      "step": 25820
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.6390341520309448,
+      "learning_rate": 6.508127833444568e-06,
+      "loss": 0.4147,
+      "step": 25821
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.1427950859069824,
+      "learning_rate": 6.507876452664258e-06,
+      "loss": 0.4871,
+      "step": 25822
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 3.1071503162384033,
+      "learning_rate": 6.507625067691109e-06,
+      "loss": 0.7029,
+      "step": 25823
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.7915492057800293,
+      "learning_rate": 6.5073736785258204e-06,
+      "loss": 0.8335,
+      "step": 25824
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.039409637451172,
+      "learning_rate": 6.50712228516909e-06,
+      "loss": 0.3876,
+      "step": 25825
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.628680944442749,
+      "learning_rate": 6.506870887621617e-06,
+      "loss": 0.4494,
+      "step": 25826
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 4.493323802947998,
+      "learning_rate": 6.5066194858841e-06,
+      "loss": 0.6805,
+      "step": 25827
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.9142814874649048,
+      "learning_rate": 6.5063680799572396e-06,
+      "loss": 0.6579,
+      "step": 25828
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.2413196563720703,
+      "learning_rate": 6.5061166698417336e-06,
+      "loss": 0.7102,
+      "step": 25829
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 4.04716682434082,
+      "learning_rate": 6.505865255538281e-06,
+      "loss": 0.7817,
+      "step": 25830
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.9501042366027832,
+      "learning_rate": 6.5056138370475805e-06,
+      "loss": 0.6765,
+      "step": 25831
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.6398983001708984,
+      "learning_rate": 6.505362414370332e-06,
+      "loss": 0.8001,
+      "step": 25832
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 3.850513219833374,
+      "learning_rate": 6.505110987507235e-06,
+      "loss": 0.6958,
+      "step": 25833
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.9699628353118896,
+      "learning_rate": 6.504859556458989e-06,
+      "loss": 0.7743,
+      "step": 25834
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.821364164352417,
+      "learning_rate": 6.504608121226292e-06,
+      "loss": 0.5569,
+      "step": 25835
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.4165196418762207,
+      "learning_rate": 6.504356681809844e-06,
+      "loss": 0.6023,
+      "step": 25836
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.4350292682647705,
+      "learning_rate": 6.504105238210342e-06,
+      "loss": 0.5687,
+      "step": 25837
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.7885717153549194,
+      "learning_rate": 6.5038537904284885e-06,
+      "loss": 0.2709,
+      "step": 25838
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.75651478767395,
+      "learning_rate": 6.503602338464979e-06,
+      "loss": 0.7707,
+      "step": 25839
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.416818857192993,
+      "learning_rate": 6.503350882320516e-06,
+      "loss": 0.6821,
+      "step": 25840
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 3.60372257232666,
+      "learning_rate": 6.503099421995797e-06,
+      "loss": 0.5529,
+      "step": 25841
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.9983714818954468,
+      "learning_rate": 6.5028479574915205e-06,
+      "loss": 0.6415,
+      "step": 25842
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.022587299346924,
+      "learning_rate": 6.502596488808388e-06,
+      "loss": 0.5851,
+      "step": 25843
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.606167793273926,
+      "learning_rate": 6.502345015947098e-06,
+      "loss": 0.7821,
+      "step": 25844
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.295504331588745,
+      "learning_rate": 6.50209353890835e-06,
+      "loss": 0.568,
+      "step": 25845
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.2429232597351074,
+      "learning_rate": 6.501842057692841e-06,
+      "loss": 0.8196,
+      "step": 25846
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.8021836280822754,
+      "learning_rate": 6.501590572301272e-06,
+      "loss": 0.6601,
+      "step": 25847
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.5354349613189697,
+      "learning_rate": 6.501339082734343e-06,
+      "loss": 0.4839,
+      "step": 25848
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 3.8048055171966553,
+      "learning_rate": 6.501087588992752e-06,
+      "loss": 0.5334,
+      "step": 25849
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 3.2614905834198,
+      "learning_rate": 6.500836091077198e-06,
+      "loss": 0.4977,
+      "step": 25850
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 3.0479328632354736,
+      "learning_rate": 6.500584588988381e-06,
+      "loss": 0.8298,
+      "step": 25851
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 3.2561161518096924,
+      "learning_rate": 6.5003330827270005e-06,
+      "loss": 0.6565,
+      "step": 25852
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.1046979427337646,
+      "learning_rate": 6.500081572293756e-06,
+      "loss": 0.6274,
+      "step": 25853
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.800485372543335,
+      "learning_rate": 6.499830057689347e-06,
+      "loss": 0.5671,
+      "step": 25854
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.873070001602173,
+      "learning_rate": 6.499578538914473e-06,
+      "loss": 0.9336,
+      "step": 25855
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.469341516494751,
+      "learning_rate": 6.499327015969832e-06,
+      "loss": 0.5653,
+      "step": 25856
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 3.229071855545044,
+      "learning_rate": 6.4990754888561235e-06,
+      "loss": 0.6497,
+      "step": 25857
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.667276620864868,
+      "learning_rate": 6.498823957574049e-06,
+      "loss": 0.7078,
+      "step": 25858
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.9801335334777832,
+      "learning_rate": 6.498572422124306e-06,
+      "loss": 0.7807,
+      "step": 25859
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.857114315032959,
+      "learning_rate": 6.498320882507595e-06,
+      "loss": 0.5542,
+      "step": 25860
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 3.0349690914154053,
+      "learning_rate": 6.498069338724614e-06,
+      "loss": 0.6973,
+      "step": 25861
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.864808440208435,
+      "learning_rate": 6.497817790776063e-06,
+      "loss": 0.8283,
+      "step": 25862
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.904974937438965,
+      "learning_rate": 6.497566238662642e-06,
+      "loss": 0.7452,
+      "step": 25863
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.088029146194458,
+      "learning_rate": 6.497314682385052e-06,
+      "loss": 0.7395,
+      "step": 25864
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.152055501937866,
+      "learning_rate": 6.49706312194399e-06,
+      "loss": 0.7536,
+      "step": 25865
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 3.673912525177002,
+      "learning_rate": 6.4968115573401546e-06,
+      "loss": 0.9878,
+      "step": 25866
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.5830156803131104,
+      "learning_rate": 6.496559988574249e-06,
+      "loss": 0.5458,
+      "step": 25867
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.456301689147949,
+      "learning_rate": 6.496308415646969e-06,
+      "loss": 0.8209,
+      "step": 25868
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 3.4647045135498047,
+      "learning_rate": 6.496056838559017e-06,
+      "loss": 0.7392,
+      "step": 25869
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 3.778475284576416,
+      "learning_rate": 6.495805257311091e-06,
+      "loss": 0.5139,
+      "step": 25870
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.7325305938720703,
+      "learning_rate": 6.495553671903891e-06,
+      "loss": 0.3428,
+      "step": 25871
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 3.3357527256011963,
+      "learning_rate": 6.495302082338118e-06,
+      "loss": 0.5742,
+      "step": 25872
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.7882071733474731,
+      "learning_rate": 6.495050488614467e-06,
+      "loss": 0.8636,
+      "step": 25873
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.913390874862671,
+      "learning_rate": 6.494798890733642e-06,
+      "loss": 0.347,
+      "step": 25874
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.2133002281188965,
+      "learning_rate": 6.494547288696341e-06,
+      "loss": 0.5068,
+      "step": 25875
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 3.5566909313201904,
+      "learning_rate": 6.494295682503265e-06,
+      "loss": 0.7194,
+      "step": 25876
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.0746989250183105,
+      "learning_rate": 6.494044072155113e-06,
+      "loss": 0.7163,
+      "step": 25877
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.687751531600952,
+      "learning_rate": 6.493792457652581e-06,
+      "loss": 0.5177,
+      "step": 25878
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.001875400543213,
+      "learning_rate": 6.493540838996375e-06,
+      "loss": 0.7442,
+      "step": 25879
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.135115385055542,
+      "learning_rate": 6.49328921618719e-06,
+      "loss": 0.7266,
+      "step": 25880
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.118511199951172,
+      "learning_rate": 6.493037589225727e-06,
+      "loss": 0.5485,
+      "step": 25881
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.741362452507019,
+      "learning_rate": 6.492785958112686e-06,
+      "loss": 0.4435,
+      "step": 25882
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.7698452472686768,
+      "learning_rate": 6.492534322848766e-06,
+      "loss": 0.5039,
+      "step": 25883
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.413428783416748,
+      "learning_rate": 6.492282683434667e-06,
+      "loss": 0.6304,
+      "step": 25884
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.8869754076004028,
+      "learning_rate": 6.492031039871091e-06,
+      "loss": 0.605,
+      "step": 25885
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.6151039600372314,
+      "learning_rate": 6.491779392158733e-06,
+      "loss": 0.5672,
+      "step": 25886
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 3.202885389328003,
+      "learning_rate": 6.491527740298297e-06,
+      "loss": 0.6967,
+      "step": 25887
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.1120548248291016,
+      "learning_rate": 6.491276084290481e-06,
+      "loss": 0.8999,
+      "step": 25888
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.370582103729248,
+      "learning_rate": 6.491024424135984e-06,
+      "loss": 0.473,
+      "step": 25889
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.2207212448120117,
+      "learning_rate": 6.490772759835507e-06,
+      "loss": 0.6645,
+      "step": 25890
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.007885217666626,
+      "learning_rate": 6.49052109138975e-06,
+      "loss": 0.7269,
+      "step": 25891
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.1196110248565674,
+      "learning_rate": 6.490269418799412e-06,
+      "loss": 0.6213,
+      "step": 25892
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.9118969440460205,
+      "learning_rate": 6.490017742065191e-06,
+      "loss": 0.7601,
+      "step": 25893
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 5.962871551513672,
+      "learning_rate": 6.489766061187791e-06,
+      "loss": 0.8197,
+      "step": 25894
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.0569677352905273,
+      "learning_rate": 6.489514376167909e-06,
+      "loss": 0.6617,
+      "step": 25895
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.5319807529449463,
+      "learning_rate": 6.489262687006246e-06,
+      "loss": 0.8146,
+      "step": 25896
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.5618896484375,
+      "learning_rate": 6.489010993703502e-06,
+      "loss": 0.8482,
+      "step": 25897
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.760258197784424,
+      "learning_rate": 6.488759296260374e-06,
+      "loss": 0.6681,
+      "step": 25898
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.671170711517334,
+      "learning_rate": 6.488507594677565e-06,
+      "loss": 0.6314,
+      "step": 25899
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 3.5014266967773438,
+      "learning_rate": 6.488255888955775e-06,
+      "loss": 0.6254,
+      "step": 25900
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.3127284049987793,
+      "learning_rate": 6.4880041790957024e-06,
+      "loss": 0.4907,
+      "step": 25901
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 4.524931907653809,
+      "learning_rate": 6.487752465098047e-06,
+      "loss": 0.645,
+      "step": 25902
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.3078380823135376,
+      "learning_rate": 6.48750074696351e-06,
+      "loss": 0.4541,
+      "step": 25903
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.10009503364563,
+      "learning_rate": 6.48724902469279e-06,
+      "loss": 0.4045,
+      "step": 25904
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.7866547107696533,
+      "learning_rate": 6.486997298286588e-06,
+      "loss": 0.7577,
+      "step": 25905
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.2692737579345703,
+      "learning_rate": 6.486745567745604e-06,
+      "loss": 0.6657,
+      "step": 25906
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.4994935989379883,
+      "learning_rate": 6.486493833070537e-06,
+      "loss": 0.5024,
+      "step": 25907
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.5285515785217285,
+      "learning_rate": 6.4862420942620875e-06,
+      "loss": 0.7165,
+      "step": 25908
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.0920937061309814,
+      "learning_rate": 6.485990351320956e-06,
+      "loss": 0.5896,
+      "step": 25909
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.1850533485412598,
+      "learning_rate": 6.485738604247842e-06,
+      "loss": 0.5714,
+      "step": 25910
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 3.4880788326263428,
+      "learning_rate": 6.485486853043445e-06,
+      "loss": 0.7211,
+      "step": 25911
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.1650354862213135,
+      "learning_rate": 6.485235097708466e-06,
+      "loss": 0.799,
+      "step": 25912
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.9751687049865723,
+      "learning_rate": 6.4849833382436025e-06,
+      "loss": 0.6502,
+      "step": 25913
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 6.867035388946533,
+      "learning_rate": 6.4847315746495586e-06,
+      "loss": 0.5153,
+      "step": 25914
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.2728536128997803,
+      "learning_rate": 6.484479806927032e-06,
+      "loss": 0.6555,
+      "step": 25915
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.049182891845703,
+      "learning_rate": 6.484228035076723e-06,
+      "loss": 0.5699,
+      "step": 25916
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.947646975517273,
+      "learning_rate": 6.483976259099333e-06,
+      "loss": 0.5614,
+      "step": 25917
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.1247711181640625,
+      "learning_rate": 6.48372447899556e-06,
+      "loss": 0.7201,
+      "step": 25918
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.8384296894073486,
+      "learning_rate": 6.483472694766105e-06,
+      "loss": 0.4589,
+      "step": 25919
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 4.580520153045654,
+      "learning_rate": 6.483220906411668e-06,
+      "loss": 0.5394,
+      "step": 25920
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.9046764373779297,
+      "learning_rate": 6.48296911393295e-06,
+      "loss": 0.4573,
+      "step": 25921
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.667465329170227,
+      "learning_rate": 6.48271731733065e-06,
+      "loss": 0.5295,
+      "step": 25922
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 3.0599424839019775,
+      "learning_rate": 6.482465516605468e-06,
+      "loss": 0.7258,
+      "step": 25923
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.213515520095825,
+      "learning_rate": 6.482213711758105e-06,
+      "loss": 0.8172,
+      "step": 25924
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.7165603637695312,
+      "learning_rate": 6.481961902789261e-06,
+      "loss": 0.6056,
+      "step": 25925
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.678440570831299,
+      "learning_rate": 6.481710089699636e-06,
+      "loss": 0.4756,
+      "step": 25926
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.099630117416382,
+      "learning_rate": 6.481458272489931e-06,
+      "loss": 0.8221,
+      "step": 25927
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.2068240642547607,
+      "learning_rate": 6.481206451160845e-06,
+      "loss": 0.635,
+      "step": 25928
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.5609986782073975,
+      "learning_rate": 6.480954625713079e-06,
+      "loss": 0.7395,
+      "step": 25929
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 4.624282360076904,
+      "learning_rate": 6.480702796147332e-06,
+      "loss": 0.5533,
+      "step": 25930
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.942234754562378,
+      "learning_rate": 6.480450962464306e-06,
+      "loss": 0.5424,
+      "step": 25931
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.5347158908843994,
+      "learning_rate": 6.4801991246647015e-06,
+      "loss": 0.6082,
+      "step": 25932
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.2483439445495605,
+      "learning_rate": 6.479947282749216e-06,
+      "loss": 0.4939,
+      "step": 25933
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.6246541738510132,
+      "learning_rate": 6.479695436718552e-06,
+      "loss": 0.5917,
+      "step": 25934
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.9556330442428589,
+      "learning_rate": 6.479443586573409e-06,
+      "loss": 0.5201,
+      "step": 25935
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.208545684814453,
+      "learning_rate": 6.479191732314488e-06,
+      "loss": 0.6135,
+      "step": 25936
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.738926649093628,
+      "learning_rate": 6.478939873942489e-06,
+      "loss": 0.6662,
+      "step": 25937
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.8555891513824463,
+      "learning_rate": 6.478688011458112e-06,
+      "loss": 0.5219,
+      "step": 25938
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.1522927284240723,
+      "learning_rate": 6.478436144862058e-06,
+      "loss": 0.9002,
+      "step": 25939
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.7214417457580566,
+      "learning_rate": 6.478184274155027e-06,
+      "loss": 0.6004,
+      "step": 25940
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.4272021055221558,
+      "learning_rate": 6.477932399337719e-06,
+      "loss": 0.581,
+      "step": 25941
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.8635352849960327,
+      "learning_rate": 6.477680520410835e-06,
+      "loss": 0.7795,
+      "step": 25942
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.4791767597198486,
+      "learning_rate": 6.477428637375075e-06,
+      "loss": 0.7799,
+      "step": 25943
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 4.528265476226807,
+      "learning_rate": 6.477176750231138e-06,
+      "loss": 1.0665,
+      "step": 25944
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.081937789916992,
+      "learning_rate": 6.476924858979727e-06,
+      "loss": 0.7969,
+      "step": 25945
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.9890319108963013,
+      "learning_rate": 6.4766729636215405e-06,
+      "loss": 0.6114,
+      "step": 25946
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 3.558505058288574,
+      "learning_rate": 6.476421064157281e-06,
+      "loss": 0.7165,
+      "step": 25947
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.9159440994262695,
+      "learning_rate": 6.476169160587647e-06,
+      "loss": 0.6522,
+      "step": 25948
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.3207592964172363,
+      "learning_rate": 6.47591725291334e-06,
+      "loss": 0.5014,
+      "step": 25949
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.7825825214385986,
+      "learning_rate": 6.475665341135059e-06,
+      "loss": 0.4933,
+      "step": 25950
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.2295784950256348,
+      "learning_rate": 6.4754134252535066e-06,
+      "loss": 0.7793,
+      "step": 25951
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.797396421432495,
+      "learning_rate": 6.475161505269382e-06,
+      "loss": 0.5939,
+      "step": 25952
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.946515440940857,
+      "learning_rate": 6.474909581183385e-06,
+      "loss": 0.7388,
+      "step": 25953
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 3.6024222373962402,
+      "learning_rate": 6.474657652996217e-06,
+      "loss": 0.5863,
+      "step": 25954
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.172830820083618,
+      "learning_rate": 6.4744057207085784e-06,
+      "loss": 0.5089,
+      "step": 25955
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.7620776891708374,
+      "learning_rate": 6.47415378432117e-06,
+      "loss": 0.7719,
+      "step": 25956
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.3460965156555176,
+      "learning_rate": 6.4739018438346925e-06,
+      "loss": 0.5475,
+      "step": 25957
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.9936202764511108,
+      "learning_rate": 6.473649899249847e-06,
+      "loss": 0.6703,
+      "step": 25958
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 3.5098490715026855,
+      "learning_rate": 6.473397950567332e-06,
+      "loss": 0.5292,
+      "step": 25959
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 3.2329931259155273,
+      "learning_rate": 6.473145997787849e-06,
+      "loss": 0.6818,
+      "step": 25960
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.7238940000534058,
+      "learning_rate": 6.4728940409120985e-06,
+      "loss": 0.5028,
+      "step": 25961
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.429447889328003,
+      "learning_rate": 6.472642079940783e-06,
+      "loss": 0.4989,
+      "step": 25962
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.0162694454193115,
+      "learning_rate": 6.4723901148746e-06,
+      "loss": 0.5969,
+      "step": 25963
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.9219287633895874,
+      "learning_rate": 6.472138145714253e-06,
+      "loss": 0.6637,
+      "step": 25964
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.6265910863876343,
+      "learning_rate": 6.4718861724604395e-06,
+      "loss": 0.6336,
+      "step": 25965
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.138747453689575,
+      "learning_rate": 6.471634195113864e-06,
+      "loss": 0.7235,
+      "step": 25966
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.222928047180176,
+      "learning_rate": 6.4713822136752235e-06,
+      "loss": 0.715,
+      "step": 25967
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.6927056312561035,
+      "learning_rate": 6.471130228145222e-06,
+      "loss": 0.6373,
+      "step": 25968
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.6649413108825684,
+      "learning_rate": 6.4708782385245576e-06,
+      "loss": 0.6176,
+      "step": 25969
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 3.6325435638427734,
+      "learning_rate": 6.470626244813931e-06,
+      "loss": 0.5501,
+      "step": 25970
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 3.996241331100464,
+      "learning_rate": 6.470374247014045e-06,
+      "loss": 0.6165,
+      "step": 25971
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.8312833309173584,
+      "learning_rate": 6.470122245125598e-06,
+      "loss": 0.4855,
+      "step": 25972
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.5229878425598145,
+      "learning_rate": 6.469870239149293e-06,
+      "loss": 0.5705,
+      "step": 25973
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 3.4129271507263184,
+      "learning_rate": 6.469618229085828e-06,
+      "loss": 0.6248,
+      "step": 25974
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 4.055616855621338,
+      "learning_rate": 6.469366214935905e-06,
+      "loss": 0.7128,
+      "step": 25975
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.9453208446502686,
+      "learning_rate": 6.469114196700226e-06,
+      "loss": 0.6045,
+      "step": 25976
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.3725168704986572,
+      "learning_rate": 6.468862174379491e-06,
+      "loss": 0.4917,
+      "step": 25977
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.087076425552368,
+      "learning_rate": 6.4686101479744e-06,
+      "loss": 0.6612,
+      "step": 25978
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 3.0247952938079834,
+      "learning_rate": 6.468358117485655e-06,
+      "loss": 0.5849,
+      "step": 25979
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.949851393699646,
+      "learning_rate": 6.468106082913955e-06,
+      "loss": 0.7836,
+      "step": 25980
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.5680670738220215,
+      "learning_rate": 6.467854044260004e-06,
+      "loss": 0.4111,
+      "step": 25981
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 3.4269614219665527,
+      "learning_rate": 6.4676020015244986e-06,
+      "loss": 0.5378,
+      "step": 25982
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.964502215385437,
+      "learning_rate": 6.467349954708142e-06,
+      "loss": 0.925,
+      "step": 25983
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 3.078343629837036,
+      "learning_rate": 6.4670979038116354e-06,
+      "loss": 0.8112,
+      "step": 25984
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.9263888597488403,
+      "learning_rate": 6.466845848835679e-06,
+      "loss": 0.5145,
+      "step": 25985
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.853026032447815,
+      "learning_rate": 6.466593789780974e-06,
+      "loss": 0.4769,
+      "step": 25986
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.6617844104766846,
+      "learning_rate": 6.466341726648221e-06,
+      "loss": 0.4977,
+      "step": 25987
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.220024347305298,
+      "learning_rate": 6.4660896594381215e-06,
+      "loss": 0.6619,
+      "step": 25988
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.6276142597198486,
+      "learning_rate": 6.4658375881513745e-06,
+      "loss": 0.5903,
+      "step": 25989
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 3.118429183959961,
+      "learning_rate": 6.465585512788684e-06,
+      "loss": 0.8666,
+      "step": 25990
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.7848498821258545,
+      "learning_rate": 6.465333433350746e-06,
+      "loss": 0.4408,
+      "step": 25991
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.9369313716888428,
+      "learning_rate": 6.465081349838268e-06,
+      "loss": 0.8576,
+      "step": 25992
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.7023541927337646,
+      "learning_rate": 6.464829262251947e-06,
+      "loss": 0.6867,
+      "step": 25993
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.635751724243164,
+      "learning_rate": 6.464577170592484e-06,
+      "loss": 0.9249,
+      "step": 25994
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 3.892268180847168,
+      "learning_rate": 6.46432507486058e-06,
+      "loss": 0.8177,
+      "step": 25995
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.3246476650238037,
+      "learning_rate": 6.464072975056939e-06,
+      "loss": 0.4266,
+      "step": 25996
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.393233060836792,
+      "learning_rate": 6.463820871182256e-06,
+      "loss": 0.6981,
+      "step": 25997
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.142146587371826,
+      "learning_rate": 6.463568763237237e-06,
+      "loss": 0.6439,
+      "step": 25998
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.00335955619812,
+      "learning_rate": 6.463316651222582e-06,
+      "loss": 0.7496,
+      "step": 25999
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.4482223987579346,
+      "learning_rate": 6.463064535138991e-06,
+      "loss": 0.9285,
+      "step": 26000
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.6677868366241455,
+      "learning_rate": 6.462812414987167e-06,
+      "loss": 0.6079,
+      "step": 26001
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.123016119003296,
+      "learning_rate": 6.462560290767808e-06,
+      "loss": 0.4694,
+      "step": 26002
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.951970100402832,
+      "learning_rate": 6.4623081624816175e-06,
+      "loss": 0.8424,
+      "step": 26003
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.4388062953948975,
+      "learning_rate": 6.462056030129296e-06,
+      "loss": 0.5295,
+      "step": 26004
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 4.867737293243408,
+      "learning_rate": 6.4618038937115426e-06,
+      "loss": 0.8738,
+      "step": 26005
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.7941986322402954,
+      "learning_rate": 6.461551753229061e-06,
+      "loss": 0.318,
+      "step": 26006
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.1379647254943848,
+      "learning_rate": 6.4612996086825516e-06,
+      "loss": 0.6552,
+      "step": 26007
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.157905101776123,
+      "learning_rate": 6.461047460072717e-06,
+      "loss": 0.4283,
+      "step": 26008
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.676251769065857,
+      "learning_rate": 6.460795307400256e-06,
+      "loss": 0.8832,
+      "step": 26009
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.910064220428467,
+      "learning_rate": 6.46054315066587e-06,
+      "loss": 0.734,
+      "step": 26010
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.9199061393737793,
+      "learning_rate": 6.46029098987026e-06,
+      "loss": 0.545,
+      "step": 26011
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.516519546508789,
+      "learning_rate": 6.460038825014129e-06,
+      "loss": 0.769,
+      "step": 26012
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.8285956382751465,
+      "learning_rate": 6.459786656098177e-06,
+      "loss": 0.6738,
+      "step": 26013
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 3.7233970165252686,
+      "learning_rate": 6.459534483123105e-06,
+      "loss": 0.7187,
+      "step": 26014
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.3604183197021484,
+      "learning_rate": 6.459282306089611e-06,
+      "loss": 0.8,
+      "step": 26015
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.7491252422332764,
+      "learning_rate": 6.459030124998404e-06,
+      "loss": 0.527,
+      "step": 26016
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.7037928104400635,
+      "learning_rate": 6.458777939850179e-06,
+      "loss": 0.7312,
+      "step": 26017
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.673187255859375,
+      "learning_rate": 6.458525750645639e-06,
+      "loss": 0.7716,
+      "step": 26018
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.8391029834747314,
+      "learning_rate": 6.458273557385486e-06,
+      "loss": 0.5196,
+      "step": 26019
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.9828734397888184,
+      "learning_rate": 6.45802136007042e-06,
+      "loss": 0.5251,
+      "step": 26020
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.7781858444213867,
+      "learning_rate": 6.457769158701144e-06,
+      "loss": 0.6049,
+      "step": 26021
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.8026454448699951,
+      "learning_rate": 6.457516953278357e-06,
+      "loss": 0.3628,
+      "step": 26022
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.1117310523986816,
+      "learning_rate": 6.45726474380276e-06,
+      "loss": 0.3507,
+      "step": 26023
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.390015125274658,
+      "learning_rate": 6.4570125302750575e-06,
+      "loss": 0.7242,
+      "step": 26024
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 3.4843783378601074,
+      "learning_rate": 6.4567603126959475e-06,
+      "loss": 0.4321,
+      "step": 26025
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.4881867170333862,
+      "learning_rate": 6.4565080910661335e-06,
+      "loss": 0.8317,
+      "step": 26026
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.202273368835449,
+      "learning_rate": 6.456255865386317e-06,
+      "loss": 0.5021,
+      "step": 26027
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.754495859146118,
+      "learning_rate": 6.456003635657198e-06,
+      "loss": 0.7956,
+      "step": 26028
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.2703068256378174,
+      "learning_rate": 6.455751401879478e-06,
+      "loss": 0.6542,
+      "step": 26029
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.403339147567749,
+      "learning_rate": 6.455499164053859e-06,
+      "loss": 0.6895,
+      "step": 26030
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.5548160076141357,
+      "learning_rate": 6.4552469221810425e-06,
+      "loss": 0.4713,
+      "step": 26031
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.2768466472625732,
+      "learning_rate": 6.454994676261728e-06,
+      "loss": 0.7543,
+      "step": 26032
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.477163076400757,
+      "learning_rate": 6.45474242629662e-06,
+      "loss": 0.7008,
+      "step": 26033
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.8368589878082275,
+      "learning_rate": 6.4544901722864175e-06,
+      "loss": 0.8898,
+      "step": 26034
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.910301923751831,
+      "learning_rate": 6.454237914231821e-06,
+      "loss": 0.9165,
+      "step": 26035
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.9674187898635864,
+      "learning_rate": 6.453985652133535e-06,
+      "loss": 0.4292,
+      "step": 26036
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.7488261461257935,
+      "learning_rate": 6.45373338599226e-06,
+      "loss": 0.5261,
+      "step": 26037
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.319399833679199,
+      "learning_rate": 6.453481115808697e-06,
+      "loss": 0.5954,
+      "step": 26038
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.750727415084839,
+      "learning_rate": 6.4532288415835475e-06,
+      "loss": 0.5811,
+      "step": 26039
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.2089476585388184,
+      "learning_rate": 6.452976563317513e-06,
+      "loss": 0.3669,
+      "step": 26040
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 4.082499027252197,
+      "learning_rate": 6.4527242810112956e-06,
+      "loss": 0.798,
+      "step": 26041
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.0863583087921143,
+      "learning_rate": 6.4524719946655946e-06,
+      "loss": 0.6912,
+      "step": 26042
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.3614323139190674,
+      "learning_rate": 6.452219704281115e-06,
+      "loss": 0.6758,
+      "step": 26043
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.1299219131469727,
+      "learning_rate": 6.451967409858555e-06,
+      "loss": 1.0167,
+      "step": 26044
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 4.467588901519775,
+      "learning_rate": 6.451715111398616e-06,
+      "loss": 0.766,
+      "step": 26045
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.7736119031906128,
+      "learning_rate": 6.4514628089020034e-06,
+      "loss": 0.796,
+      "step": 26046
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 3.016275644302368,
+      "learning_rate": 6.451210502369416e-06,
+      "loss": 0.765,
+      "step": 26047
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 5.751634120941162,
+      "learning_rate": 6.450958191801556e-06,
+      "loss": 0.6037,
+      "step": 26048
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.0244040489196777,
+      "learning_rate": 6.450705877199126e-06,
+      "loss": 0.8248,
+      "step": 26049
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.483046054840088,
+      "learning_rate": 6.450453558562824e-06,
+      "loss": 0.8514,
+      "step": 26050
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.9368826150894165,
+      "learning_rate": 6.450201235893356e-06,
+      "loss": 0.4113,
+      "step": 26051
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.9608347415924072,
+      "learning_rate": 6.44994890919142e-06,
+      "loss": 0.6806,
+      "step": 26052
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.2362340688705444,
+      "learning_rate": 6.449696578457719e-06,
+      "loss": 0.5029,
+      "step": 26053
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.1780612468719482,
+      "learning_rate": 6.449444243692956e-06,
+      "loss": 0.8521,
+      "step": 26054
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.802205801010132,
+      "learning_rate": 6.4491919048978315e-06,
+      "loss": 0.7015,
+      "step": 26055
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.5714988708496094,
+      "learning_rate": 6.448939562073048e-06,
+      "loss": 0.4951,
+      "step": 26056
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 3.2501914501190186,
+      "learning_rate": 6.448687215219305e-06,
+      "loss": 0.5018,
+      "step": 26057
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.8000743389129639,
+      "learning_rate": 6.4484348643373065e-06,
+      "loss": 0.4489,
+      "step": 26058
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.9133801460266113,
+      "learning_rate": 6.448182509427752e-06,
+      "loss": 0.5428,
+      "step": 26059
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.6346420049667358,
+      "learning_rate": 6.447930150491345e-06,
+      "loss": 0.6082,
+      "step": 26060
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.464613437652588,
+      "learning_rate": 6.447677787528787e-06,
+      "loss": 0.7927,
+      "step": 26061
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 4.851053237915039,
+      "learning_rate": 6.447425420540779e-06,
+      "loss": 0.625,
+      "step": 26062
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.1397461891174316,
+      "learning_rate": 6.447173049528024e-06,
+      "loss": 0.5748,
+      "step": 26063
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.880463719367981,
+      "learning_rate": 6.446920674491222e-06,
+      "loss": 0.7066,
+      "step": 26064
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.8325626850128174,
+      "learning_rate": 6.446668295431076e-06,
+      "loss": 0.5078,
+      "step": 26065
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.093280076980591,
+      "learning_rate": 6.446415912348285e-06,
+      "loss": 0.8058,
+      "step": 26066
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.2656733989715576,
+      "learning_rate": 6.446163525243556e-06,
+      "loss": 0.6449,
+      "step": 26067
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.8926942348480225,
+      "learning_rate": 6.445911134117587e-06,
+      "loss": 0.4695,
+      "step": 26068
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 3.701531410217285,
+      "learning_rate": 6.445658738971081e-06,
+      "loss": 0.7428,
+      "step": 26069
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.056255340576172,
+      "learning_rate": 6.445406339804741e-06,
+      "loss": 0.9016,
+      "step": 26070
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.8273804187774658,
+      "learning_rate": 6.445153936619266e-06,
+      "loss": 0.513,
+      "step": 26071
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.879471778869629,
+      "learning_rate": 6.444901529415359e-06,
+      "loss": 0.6011,
+      "step": 26072
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.3228998184204102,
+      "learning_rate": 6.444649118193723e-06,
+      "loss": 0.8851,
+      "step": 26073
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.9926737546920776,
+      "learning_rate": 6.444396702955059e-06,
+      "loss": 0.6573,
+      "step": 26074
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 4.112290859222412,
+      "learning_rate": 6.444144283700069e-06,
+      "loss": 0.7693,
+      "step": 26075
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.971361756324768,
+      "learning_rate": 6.443891860429453e-06,
+      "loss": 0.7324,
+      "step": 26076
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.2497646808624268,
+      "learning_rate": 6.443639433143916e-06,
+      "loss": 0.5591,
+      "step": 26077
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 3.800130605697632,
+      "learning_rate": 6.443387001844158e-06,
+      "loss": 0.7239,
+      "step": 26078
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 6.385242462158203,
+      "learning_rate": 6.443134566530882e-06,
+      "loss": 0.5057,
+      "step": 26079
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 3.386073112487793,
+      "learning_rate": 6.4428821272047895e-06,
+      "loss": 0.7625,
+      "step": 26080
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 3.8045034408569336,
+      "learning_rate": 6.442629683866582e-06,
+      "loss": 0.7963,
+      "step": 26081
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.2611777782440186,
+      "learning_rate": 6.4423772365169615e-06,
+      "loss": 0.3472,
+      "step": 26082
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.7070069313049316,
+      "learning_rate": 6.44212478515663e-06,
+      "loss": 0.6582,
+      "step": 26083
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 3.3020668029785156,
+      "learning_rate": 6.4418723297862905e-06,
+      "loss": 0.6315,
+      "step": 26084
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 3.6052980422973633,
+      "learning_rate": 6.441619870406645e-06,
+      "loss": 0.8361,
+      "step": 26085
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.485762119293213,
+      "learning_rate": 6.4413674070183915e-06,
+      "loss": 0.5602,
+      "step": 26086
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.2903380393981934,
+      "learning_rate": 6.441114939622238e-06,
+      "loss": 0.6074,
+      "step": 26087
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.9320292472839355,
+      "learning_rate": 6.440862468218883e-06,
+      "loss": 0.7279,
+      "step": 26088
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.0326521396636963,
+      "learning_rate": 6.44060999280903e-06,
+      "loss": 0.442,
+      "step": 26089
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.183685779571533,
+      "learning_rate": 6.440357513393378e-06,
+      "loss": 0.5957,
+      "step": 26090
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.9319874048233032,
+      "learning_rate": 6.440105029972634e-06,
+      "loss": 0.8112,
+      "step": 26091
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.638904094696045,
+      "learning_rate": 6.439852542547496e-06,
+      "loss": 0.9049,
+      "step": 26092
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 3.444810628890991,
+      "learning_rate": 6.439600051118669e-06,
+      "loss": 0.8907,
+      "step": 26093
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.5591650009155273,
+      "learning_rate": 6.4393475556868515e-06,
+      "loss": 0.762,
+      "step": 26094
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.8690872192382812,
+      "learning_rate": 6.439095056252748e-06,
+      "loss": 0.9413,
+      "step": 26095
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.642216444015503,
+      "learning_rate": 6.438842552817059e-06,
+      "loss": 0.4924,
+      "step": 26096
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.4480103254318237,
+      "learning_rate": 6.4385900453804904e-06,
+      "loss": 0.5612,
+      "step": 26097
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.3215436935424805,
+      "learning_rate": 6.438337533943741e-06,
+      "loss": 0.5269,
+      "step": 26098
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.8620800971984863,
+      "learning_rate": 6.438085018507514e-06,
+      "loss": 0.712,
+      "step": 26099
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.3405191898345947,
+      "learning_rate": 6.43783249907251e-06,
+      "loss": 0.7048,
+      "step": 26100
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.503929376602173,
+      "learning_rate": 6.437579975639434e-06,
+      "loss": 0.8963,
+      "step": 26101
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.187274932861328,
+      "learning_rate": 6.437327448208986e-06,
+      "loss": 0.5322,
+      "step": 26102
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.1626639366149902,
+      "learning_rate": 6.437074916781869e-06,
+      "loss": 0.9418,
+      "step": 26103
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 5.264601230621338,
+      "learning_rate": 6.436822381358785e-06,
+      "loss": 0.6495,
+      "step": 26104
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.2832846641540527,
+      "learning_rate": 6.436569841940436e-06,
+      "loss": 0.5667,
+      "step": 26105
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.754287004470825,
+      "learning_rate": 6.436317298527522e-06,
+      "loss": 0.7242,
+      "step": 26106
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.7083213329315186,
+      "learning_rate": 6.436064751120751e-06,
+      "loss": 0.7328,
+      "step": 26107
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.6006693840026855,
+      "learning_rate": 6.435812199720821e-06,
+      "loss": 0.5838,
+      "step": 26108
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.077211380004883,
+      "learning_rate": 6.435559644328435e-06,
+      "loss": 0.498,
+      "step": 26109
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.184603691101074,
+      "learning_rate": 6.435307084944296e-06,
+      "loss": 0.7658,
+      "step": 26110
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.033285617828369,
+      "learning_rate": 6.435054521569104e-06,
+      "loss": 0.8027,
+      "step": 26111
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.2342000007629395,
+      "learning_rate": 6.434801954203565e-06,
+      "loss": 0.8847,
+      "step": 26112
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.5288875102996826,
+      "learning_rate": 6.434549382848379e-06,
+      "loss": 0.8487,
+      "step": 26113
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.199317455291748,
+      "learning_rate": 6.434296807504247e-06,
+      "loss": 0.4673,
+      "step": 26114
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.3722901344299316,
+      "learning_rate": 6.434044228171873e-06,
+      "loss": 0.3968,
+      "step": 26115
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.5846807956695557,
+      "learning_rate": 6.43379164485196e-06,
+      "loss": 0.5903,
+      "step": 26116
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.175180435180664,
+      "learning_rate": 6.43353905754521e-06,
+      "loss": 0.5371,
+      "step": 26117
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.807579755783081,
+      "learning_rate": 6.433286466252324e-06,
+      "loss": 0.8394,
+      "step": 26118
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 8.102144241333008,
+      "learning_rate": 6.433033870974006e-06,
+      "loss": 0.4375,
+      "step": 26119
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.5631980895996094,
+      "learning_rate": 6.432781271710957e-06,
+      "loss": 0.5487,
+      "step": 26120
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.467965841293335,
+      "learning_rate": 6.43252866846388e-06,
+      "loss": 0.7815,
+      "step": 26121
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.3984745740890503,
+      "learning_rate": 6.4322760612334775e-06,
+      "loss": 0.5431,
+      "step": 26122
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.026508331298828,
+      "learning_rate": 6.432023450020451e-06,
+      "loss": 0.6344,
+      "step": 26123
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.6088979244232178,
+      "learning_rate": 6.431770834825506e-06,
+      "loss": 0.9468,
+      "step": 26124
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.8372387886047363,
+      "learning_rate": 6.431518215649341e-06,
+      "loss": 0.7454,
+      "step": 26125
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.9812748432159424,
+      "learning_rate": 6.431265592492662e-06,
+      "loss": 0.6615,
+      "step": 26126
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.486297607421875,
+      "learning_rate": 6.431012965356166e-06,
+      "loss": 0.6876,
+      "step": 26127
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.8798209428787231,
+      "learning_rate": 6.430760334240561e-06,
+      "loss": 0.5175,
+      "step": 26128
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.1707847118377686,
+      "learning_rate": 6.4305076991465475e-06,
+      "loss": 0.804,
+      "step": 26129
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.229008674621582,
+      "learning_rate": 6.430255060074828e-06,
+      "loss": 0.4776,
+      "step": 26130
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.5061755180358887,
+      "learning_rate": 6.430002417026106e-06,
+      "loss": 0.7171,
+      "step": 26131
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.629859447479248,
+      "learning_rate": 6.429749770001081e-06,
+      "loss": 0.7991,
+      "step": 26132
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.886035680770874,
+      "learning_rate": 6.4294971190004585e-06,
+      "loss": 0.5208,
+      "step": 26133
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 4.200827121734619,
+      "learning_rate": 6.429244464024941e-06,
+      "loss": 0.5713,
+      "step": 26134
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.4726868867874146,
+      "learning_rate": 6.428991805075229e-06,
+      "loss": 0.8104,
+      "step": 26135
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.5920464992523193,
+      "learning_rate": 6.428739142152026e-06,
+      "loss": 0.3811,
+      "step": 26136
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.819404125213623,
+      "learning_rate": 6.428486475256034e-06,
+      "loss": 0.5568,
+      "step": 26137
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.9783464670181274,
+      "learning_rate": 6.428233804387956e-06,
+      "loss": 0.4328,
+      "step": 26138
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.0740699768066406,
+      "learning_rate": 6.427981129548498e-06,
+      "loss": 0.7741,
+      "step": 26139
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.035407543182373,
+      "learning_rate": 6.427728450738356e-06,
+      "loss": 0.6682,
+      "step": 26140
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 4.368335723876953,
+      "learning_rate": 6.427475767958238e-06,
+      "loss": 0.8032,
+      "step": 26141
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.0845091342926025,
+      "learning_rate": 6.427223081208845e-06,
+      "loss": 0.6443,
+      "step": 26142
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.2110462188720703,
+      "learning_rate": 6.426970390490877e-06,
+      "loss": 0.7006,
+      "step": 26143
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.685042381286621,
+      "learning_rate": 6.426717695805041e-06,
+      "loss": 0.5752,
+      "step": 26144
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.957103967666626,
+      "learning_rate": 6.426464997152036e-06,
+      "loss": 0.4504,
+      "step": 26145
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.573683738708496,
+      "learning_rate": 6.426212294532566e-06,
+      "loss": 0.6317,
+      "step": 26146
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.030458688735962,
+      "learning_rate": 6.425959587947334e-06,
+      "loss": 0.6194,
+      "step": 26147
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 4.624210834503174,
+      "learning_rate": 6.425706877397045e-06,
+      "loss": 0.8002,
+      "step": 26148
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 4.586579322814941,
+      "learning_rate": 6.425454162882397e-06,
+      "loss": 0.5917,
+      "step": 26149
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.536373257637024,
+      "learning_rate": 6.4252014444040945e-06,
+      "loss": 0.6666,
+      "step": 26150
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.9496382474899292,
+      "learning_rate": 6.424948721962843e-06,
+      "loss": 0.7349,
+      "step": 26151
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.2879745960235596,
+      "learning_rate": 6.424695995559341e-06,
+      "loss": 0.8289,
+      "step": 26152
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.186906576156616,
+      "learning_rate": 6.424443265194294e-06,
+      "loss": 0.6006,
+      "step": 26153
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.235940933227539,
+      "learning_rate": 6.424190530868404e-06,
+      "loss": 0.5159,
+      "step": 26154
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.0114572048187256,
+      "learning_rate": 6.423937792582373e-06,
+      "loss": 0.74,
+      "step": 26155
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.5877065658569336,
+      "learning_rate": 6.4236850503369054e-06,
+      "loss": 0.6004,
+      "step": 26156
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.274362564086914,
+      "learning_rate": 6.423432304132701e-06,
+      "loss": 0.8165,
+      "step": 26157
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.042607307434082,
+      "learning_rate": 6.4231795539704656e-06,
+      "loss": 0.691,
+      "step": 26158
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.023869037628174,
+      "learning_rate": 6.422926799850902e-06,
+      "loss": 0.5667,
+      "step": 26159
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.9847601652145386,
+      "learning_rate": 6.42267404177471e-06,
+      "loss": 0.4181,
+      "step": 26160
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.993234634399414,
+      "learning_rate": 6.422421279742597e-06,
+      "loss": 0.8024,
+      "step": 26161
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.10776424407959,
+      "learning_rate": 6.422168513755262e-06,
+      "loss": 0.5088,
+      "step": 26162
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.3609950542449951,
+      "learning_rate": 6.421915743813409e-06,
+      "loss": 0.6522,
+      "step": 26163
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.8081891536712646,
+      "learning_rate": 6.421662969917741e-06,
+      "loss": 0.4944,
+      "step": 26164
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.501079559326172,
+      "learning_rate": 6.421410192068961e-06,
+      "loss": 0.5731,
+      "step": 26165
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.802898645401001,
+      "learning_rate": 6.421157410267771e-06,
+      "loss": 0.6567,
+      "step": 26166
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.1627750396728516,
+      "learning_rate": 6.4209046245148745e-06,
+      "loss": 0.6079,
+      "step": 26167
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.0545411109924316,
+      "learning_rate": 6.420651834810976e-06,
+      "loss": 0.2152,
+      "step": 26168
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.9240379333496094,
+      "learning_rate": 6.420399041156776e-06,
+      "loss": 0.5346,
+      "step": 26169
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 4.55134916305542,
+      "learning_rate": 6.4201462435529786e-06,
+      "loss": 0.6339,
+      "step": 26170
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.284196615219116,
+      "learning_rate": 6.419893442000287e-06,
+      "loss": 0.3705,
+      "step": 26171
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.4103691577911377,
+      "learning_rate": 6.419640636499402e-06,
+      "loss": 0.4461,
+      "step": 26172
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 5.170942306518555,
+      "learning_rate": 6.41938782705103e-06,
+      "loss": 0.8235,
+      "step": 26173
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.0317935943603516,
+      "learning_rate": 6.419135013655872e-06,
+      "loss": 0.5633,
+      "step": 26174
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.6498358249664307,
+      "learning_rate": 6.418882196314631e-06,
+      "loss": 0.7919,
+      "step": 26175
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.145939588546753,
+      "learning_rate": 6.418629375028008e-06,
+      "loss": 0.5557,
+      "step": 26176
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.0334999561309814,
+      "learning_rate": 6.41837654979671e-06,
+      "loss": 0.837,
+      "step": 26177
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.8966028690338135,
+      "learning_rate": 6.4181237206214385e-06,
+      "loss": 0.6703,
+      "step": 26178
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.4237701892852783,
+      "learning_rate": 6.417870887502896e-06,
+      "loss": 0.638,
+      "step": 26179
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.009799003601074,
+      "learning_rate": 6.4176180504417855e-06,
+      "loss": 0.9738,
+      "step": 26180
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.617424726486206,
+      "learning_rate": 6.4173652094388106e-06,
+      "loss": 0.7152,
+      "step": 26181
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.053560495376587,
+      "learning_rate": 6.417112364494672e-06,
+      "loss": 0.6215,
+      "step": 26182
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.4897081851959229,
+      "learning_rate": 6.4168595156100765e-06,
+      "loss": 0.5518,
+      "step": 26183
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.813767433166504,
+      "learning_rate": 6.416606662785726e-06,
+      "loss": 0.7367,
+      "step": 26184
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.451629161834717,
+      "learning_rate": 6.4163538060223215e-06,
+      "loss": 1.0437,
+      "step": 26185
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.7401857376098633,
+      "learning_rate": 6.416100945320569e-06,
+      "loss": 0.439,
+      "step": 26186
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.1769943237304688,
+      "learning_rate": 6.415848080681171e-06,
+      "loss": 0.5126,
+      "step": 26187
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.7393134832382202,
+      "learning_rate": 6.415595212104827e-06,
+      "loss": 0.6416,
+      "step": 26188
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.485529899597168,
+      "learning_rate": 6.415342339592244e-06,
+      "loss": 0.9792,
+      "step": 26189
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.9698582887649536,
+      "learning_rate": 6.415089463144125e-06,
+      "loss": 0.8406,
+      "step": 26190
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 5.856765270233154,
+      "learning_rate": 6.4148365827611725e-06,
+      "loss": 0.5946,
+      "step": 26191
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.1076412200927734,
+      "learning_rate": 6.414583698444089e-06,
+      "loss": 0.6308,
+      "step": 26192
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.871124029159546,
+      "learning_rate": 6.414330810193579e-06,
+      "loss": 0.6132,
+      "step": 26193
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.861508369445801,
+      "learning_rate": 6.414077918010344e-06,
+      "loss": 0.6323,
+      "step": 26194
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.029911756515503,
+      "learning_rate": 6.413825021895088e-06,
+      "loss": 0.6496,
+      "step": 26195
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.9892051219940186,
+      "learning_rate": 6.413572121848515e-06,
+      "loss": 0.6993,
+      "step": 26196
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.023573398590088,
+      "learning_rate": 6.4133192178713265e-06,
+      "loss": 0.6696,
+      "step": 26197
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.5878522396087646,
+      "learning_rate": 6.413066309964227e-06,
+      "loss": 0.5723,
+      "step": 26198
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.0357019901275635,
+      "learning_rate": 6.4128133981279184e-06,
+      "loss": 0.4765,
+      "step": 26199
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 4.741719722747803,
+      "learning_rate": 6.412560482363106e-06,
+      "loss": 0.6984,
+      "step": 26200
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.7884305715560913,
+      "learning_rate": 6.412307562670492e-06,
+      "loss": 0.5953,
+      "step": 26201
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.1871213912963867,
+      "learning_rate": 6.41205463905078e-06,
+      "loss": 0.523,
+      "step": 26202
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.862891674041748,
+      "learning_rate": 6.411801711504673e-06,
+      "loss": 0.8479,
+      "step": 26203
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.276146650314331,
+      "learning_rate": 6.411548780032875e-06,
+      "loss": 0.5736,
+      "step": 26204
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.1204779148101807,
+      "learning_rate": 6.4112958446360875e-06,
+      "loss": 0.6393,
+      "step": 26205
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.444150686264038,
+      "learning_rate": 6.411042905315016e-06,
+      "loss": 0.592,
+      "step": 26206
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.150791645050049,
+      "learning_rate": 6.410789962070362e-06,
+      "loss": 0.7448,
+      "step": 26207
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 4.425879955291748,
+      "learning_rate": 6.4105370149028275e-06,
+      "loss": 0.552,
+      "step": 26208
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.0112226009368896,
+      "learning_rate": 6.410284063813122e-06,
+      "loss": 0.5231,
+      "step": 26209
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.5039634704589844,
+      "learning_rate": 6.410031108801941e-06,
+      "loss": 0.4296,
+      "step": 26210
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.1851937770843506,
+      "learning_rate": 6.409778149869994e-06,
+      "loss": 0.6361,
+      "step": 26211
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.2424654960632324,
+      "learning_rate": 6.409525187017982e-06,
+      "loss": 0.8244,
+      "step": 26212
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.6653834581375122,
+      "learning_rate": 6.409272220246609e-06,
+      "loss": 0.7786,
+      "step": 26213
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.1538233757019043,
+      "learning_rate": 6.409019249556576e-06,
+      "loss": 0.5771,
+      "step": 26214
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.82378089427948,
+      "learning_rate": 6.40876627494859e-06,
+      "loss": 0.6598,
+      "step": 26215
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.850356101989746,
+      "learning_rate": 6.408513296423352e-06,
+      "loss": 0.5673,
+      "step": 26216
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.6910736560821533,
+      "learning_rate": 6.408260313981567e-06,
+      "loss": 0.5772,
+      "step": 26217
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.8628406524658203,
+      "learning_rate": 6.408007327623935e-06,
+      "loss": 0.6371,
+      "step": 26218
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.4178197383880615,
+      "learning_rate": 6.407754337351164e-06,
+      "loss": 0.4999,
+      "step": 26219
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.112548589706421,
+      "learning_rate": 6.4075013431639555e-06,
+      "loss": 0.7203,
+      "step": 26220
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.0655298233032227,
+      "learning_rate": 6.407248345063013e-06,
+      "loss": 0.7945,
+      "step": 26221
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.9636223316192627,
+      "learning_rate": 6.40699534304904e-06,
+      "loss": 0.7927,
+      "step": 26222
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.8647894859313965,
+      "learning_rate": 6.406742337122741e-06,
+      "loss": 0.8662,
+      "step": 26223
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.189098834991455,
+      "learning_rate": 6.406489327284817e-06,
+      "loss": 0.5753,
+      "step": 26224
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.0044307708740234,
+      "learning_rate": 6.406236313535974e-06,
+      "loss": 0.398,
+      "step": 26225
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.10768985748291,
+      "learning_rate": 6.405983295876914e-06,
+      "loss": 0.8177,
+      "step": 26226
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.9906599521636963,
+      "learning_rate": 6.405730274308342e-06,
+      "loss": 0.4818,
+      "step": 26227
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.8334790468215942,
+      "learning_rate": 6.405477248830959e-06,
+      "loss": 0.8537,
+      "step": 26228
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.2886767387390137,
+      "learning_rate": 6.4052242194454715e-06,
+      "loss": 0.6268,
+      "step": 26229
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.7193751335144043,
+      "learning_rate": 6.404971186152582e-06,
+      "loss": 0.7199,
+      "step": 26230
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.5722546577453613,
+      "learning_rate": 6.404718148952993e-06,
+      "loss": 0.3111,
+      "step": 26231
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.2177116870880127,
+      "learning_rate": 6.404465107847411e-06,
+      "loss": 0.8062,
+      "step": 26232
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.5868762731552124,
+      "learning_rate": 6.404212062836536e-06,
+      "loss": 0.5515,
+      "step": 26233
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.6864876747131348,
+      "learning_rate": 6.403959013921074e-06,
+      "loss": 0.697,
+      "step": 26234
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.6466808319091797,
+      "learning_rate": 6.403705961101727e-06,
+      "loss": 0.6407,
+      "step": 26235
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.1595873832702637,
+      "learning_rate": 6.403452904379201e-06,
+      "loss": 0.4952,
+      "step": 26236
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.529148817062378,
+      "learning_rate": 6.403199843754197e-06,
+      "loss": 0.7568,
+      "step": 26237
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.8248183727264404,
+      "learning_rate": 6.4029467792274196e-06,
+      "loss": 0.3726,
+      "step": 26238
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.822539806365967,
+      "learning_rate": 6.402693710799572e-06,
+      "loss": 0.5894,
+      "step": 26239
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.7545111179351807,
+      "learning_rate": 6.402440638471361e-06,
+      "loss": 0.5978,
+      "step": 26240
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.7549965381622314,
+      "learning_rate": 6.402187562243487e-06,
+      "loss": 0.8616,
+      "step": 26241
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.6154792308807373,
+      "learning_rate": 6.401934482116655e-06,
+      "loss": 0.8373,
+      "step": 26242
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.8424367904663086,
+      "learning_rate": 6.401681398091568e-06,
+      "loss": 0.8147,
+      "step": 26243
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.4694745540618896,
+      "learning_rate": 6.401428310168929e-06,
+      "loss": 0.5592,
+      "step": 26244
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.831851840019226,
+      "learning_rate": 6.401175218349445e-06,
+      "loss": 0.7335,
+      "step": 26245
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.189544439315796,
+      "learning_rate": 6.400922122633816e-06,
+      "loss": 0.5692,
+      "step": 26246
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.7647379636764526,
+      "learning_rate": 6.4006690230227474e-06,
+      "loss": 0.4347,
+      "step": 26247
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.6702818870544434,
+      "learning_rate": 6.400415919516945e-06,
+      "loss": 0.759,
+      "step": 26248
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.91499924659729,
+      "learning_rate": 6.400162812117108e-06,
+      "loss": 0.6313,
+      "step": 26249
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.380305290222168,
+      "learning_rate": 6.399909700823943e-06,
+      "loss": 0.743,
+      "step": 26250
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.8964600563049316,
+      "learning_rate": 6.399656585638154e-06,
+      "loss": 0.5176,
+      "step": 26251
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 4.824275016784668,
+      "learning_rate": 6.399403466560445e-06,
+      "loss": 0.4906,
+      "step": 26252
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.4011571407318115,
+      "learning_rate": 6.399150343591519e-06,
+      "loss": 0.5169,
+      "step": 26253
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.8987975120544434,
+      "learning_rate": 6.39889721673208e-06,
+      "loss": 0.7706,
+      "step": 26254
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.062516689300537,
+      "learning_rate": 6.398644085982832e-06,
+      "loss": 0.56,
+      "step": 26255
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.149428606033325,
+      "learning_rate": 6.39839095134448e-06,
+      "loss": 0.5712,
+      "step": 26256
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.997285842895508,
+      "learning_rate": 6.398137812817724e-06,
+      "loss": 0.7488,
+      "step": 26257
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 7.917980194091797,
+      "learning_rate": 6.397884670403271e-06,
+      "loss": 0.7142,
+      "step": 26258
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.812621831893921,
+      "learning_rate": 6.397631524101825e-06,
+      "loss": 0.5565,
+      "step": 26259
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.2339060306549072,
+      "learning_rate": 6.397378373914089e-06,
+      "loss": 0.6602,
+      "step": 26260
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.9932334423065186,
+      "learning_rate": 6.3971252198407685e-06,
+      "loss": 0.9906,
+      "step": 26261
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.430596351623535,
+      "learning_rate": 6.396872061882565e-06,
+      "loss": 0.3049,
+      "step": 26262
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.243462085723877,
+      "learning_rate": 6.3966189000401835e-06,
+      "loss": 0.699,
+      "step": 26263
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.3196089267730713,
+      "learning_rate": 6.3963657343143296e-06,
+      "loss": 0.7033,
+      "step": 26264
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.9566360712051392,
+      "learning_rate": 6.396112564705703e-06,
+      "loss": 0.4271,
+      "step": 26265
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.742672085762024,
+      "learning_rate": 6.395859391215012e-06,
+      "loss": 0.6372,
+      "step": 26266
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 8.294462203979492,
+      "learning_rate": 6.395606213842959e-06,
+      "loss": 0.6207,
+      "step": 26267
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.103102922439575,
+      "learning_rate": 6.395353032590248e-06,
+      "loss": 0.5437,
+      "step": 26268
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.7091516256332397,
+      "learning_rate": 6.395099847457582e-06,
+      "loss": 0.4365,
+      "step": 26269
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.938780426979065,
+      "learning_rate": 6.394846658445666e-06,
+      "loss": 0.4448,
+      "step": 26270
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.029881000518799,
+      "learning_rate": 6.394593465555205e-06,
+      "loss": 0.5543,
+      "step": 26271
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.9821219444274902,
+      "learning_rate": 6.3943402687869e-06,
+      "loss": 0.7336,
+      "step": 26272
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.8386496305465698,
+      "learning_rate": 6.394087068141459e-06,
+      "loss": 0.4737,
+      "step": 26273
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.000786066055298,
+      "learning_rate": 6.393833863619584e-06,
+      "loss": 0.432,
+      "step": 26274
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.092536449432373,
+      "learning_rate": 6.393580655221979e-06,
+      "loss": 0.699,
+      "step": 26275
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.1621005535125732,
+      "learning_rate": 6.393327442949348e-06,
+      "loss": 0.3128,
+      "step": 26276
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 6.5380330085754395,
+      "learning_rate": 6.393074226802396e-06,
+      "loss": 0.7375,
+      "step": 26277
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.01356840133667,
+      "learning_rate": 6.392821006781825e-06,
+      "loss": 0.5426,
+      "step": 26278
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.089141368865967,
+      "learning_rate": 6.39256778288834e-06,
+      "loss": 0.7383,
+      "step": 26279
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.8200987577438354,
+      "learning_rate": 6.392314555122647e-06,
+      "loss": 0.752,
+      "step": 26280
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.9889307022094727,
+      "learning_rate": 6.39206132348545e-06,
+      "loss": 0.8524,
+      "step": 26281
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.6336212158203125,
+      "learning_rate": 6.39180808797745e-06,
+      "loss": 0.5333,
+      "step": 26282
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.6586029529571533,
+      "learning_rate": 6.391554848599354e-06,
+      "loss": 0.599,
+      "step": 26283
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.5878896713256836,
+      "learning_rate": 6.391301605351865e-06,
+      "loss": 0.6656,
+      "step": 26284
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.5335729122161865,
+      "learning_rate": 6.3910483582356875e-06,
+      "loss": 0.7778,
+      "step": 26285
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.47027325630188,
+      "learning_rate": 6.3907951072515264e-06,
+      "loss": 0.7089,
+      "step": 26286
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.5538190603256226,
+      "learning_rate": 6.390541852400084e-06,
+      "loss": 0.7071,
+      "step": 26287
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.615468144416809,
+      "learning_rate": 6.390288593682066e-06,
+      "loss": 0.5504,
+      "step": 26288
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.035440683364868,
+      "learning_rate": 6.390035331098175e-06,
+      "loss": 0.7485,
+      "step": 26289
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.9938620328903198,
+      "learning_rate": 6.389782064649117e-06,
+      "loss": 0.4928,
+      "step": 26290
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.7309235334396362,
+      "learning_rate": 6.3895287943355975e-06,
+      "loss": 0.5419,
+      "step": 26291
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.5598018169403076,
+      "learning_rate": 6.389275520158318e-06,
+      "loss": 0.7625,
+      "step": 26292
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.9514784812927246,
+      "learning_rate": 6.3890222421179825e-06,
+      "loss": 0.7919,
+      "step": 26293
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.708320379257202,
+      "learning_rate": 6.388768960215298e-06,
+      "loss": 0.2912,
+      "step": 26294
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.308328151702881,
+      "learning_rate": 6.388515674450967e-06,
+      "loss": 0.534,
+      "step": 26295
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.0205609798431396,
+      "learning_rate": 6.388262384825694e-06,
+      "loss": 0.6166,
+      "step": 26296
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.8263168334960938,
+      "learning_rate": 6.3880090913401825e-06,
+      "loss": 0.6179,
+      "step": 26297
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.143857479095459,
+      "learning_rate": 6.3877557939951385e-06,
+      "loss": 0.6909,
+      "step": 26298
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.075185537338257,
+      "learning_rate": 6.3875024927912655e-06,
+      "loss": 0.8507,
+      "step": 26299
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.432159662246704,
+      "learning_rate": 6.387249187729268e-06,
+      "loss": 0.6657,
+      "step": 26300
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.6415265798568726,
+      "learning_rate": 6.386995878809849e-06,
+      "loss": 0.5745,
+      "step": 26301
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.611783266067505,
+      "learning_rate": 6.386742566033714e-06,
+      "loss": 0.7152,
+      "step": 26302
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.0024309158325195,
+      "learning_rate": 6.38648924940157e-06,
+      "loss": 0.6922,
+      "step": 26303
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.876570701599121,
+      "learning_rate": 6.386235928914116e-06,
+      "loss": 0.7044,
+      "step": 26304
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.5741220712661743,
+      "learning_rate": 6.385982604572061e-06,
+      "loss": 0.4585,
+      "step": 26305
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.351060390472412,
+      "learning_rate": 6.385729276376106e-06,
+      "loss": 0.6078,
+      "step": 26306
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.4098122119903564,
+      "learning_rate": 6.385475944326957e-06,
+      "loss": 0.596,
+      "step": 26307
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.4818856716156006,
+      "learning_rate": 6.38522260842532e-06,
+      "loss": 0.6968,
+      "step": 26308
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.5227991342544556,
+      "learning_rate": 6.384969268671897e-06,
+      "loss": 0.7518,
+      "step": 26309
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.2300474643707275,
+      "learning_rate": 6.384715925067393e-06,
+      "loss": 0.6932,
+      "step": 26310
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.721062660217285,
+      "learning_rate": 6.384462577612512e-06,
+      "loss": 0.4352,
+      "step": 26311
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.9622215032577515,
+      "learning_rate": 6.384209226307962e-06,
+      "loss": 0.342,
+      "step": 26312
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.4462478160858154,
+      "learning_rate": 6.383955871154441e-06,
+      "loss": 0.9919,
+      "step": 26313
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.870224952697754,
+      "learning_rate": 6.383702512152661e-06,
+      "loss": 0.7652,
+      "step": 26314
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.2820918560028076,
+      "learning_rate": 6.38344914930332e-06,
+      "loss": 0.5892,
+      "step": 26315
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.5525693893432617,
+      "learning_rate": 6.3831957826071256e-06,
+      "loss": 0.9425,
+      "step": 26316
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 4.038574695587158,
+      "learning_rate": 6.382942412064783e-06,
+      "loss": 0.951,
+      "step": 26317
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.4148651361465454,
+      "learning_rate": 6.382689037676994e-06,
+      "loss": 0.4475,
+      "step": 26318
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.3969082832336426,
+      "learning_rate": 6.382435659444467e-06,
+      "loss": 0.8376,
+      "step": 26319
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.8251899480819702,
+      "learning_rate": 6.382182277367901e-06,
+      "loss": 0.6144,
+      "step": 26320
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.3509349822998047,
+      "learning_rate": 6.381928891448006e-06,
+      "loss": 0.4622,
+      "step": 26321
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.9964147806167603,
+      "learning_rate": 6.3816755016854845e-06,
+      "loss": 0.6227,
+      "step": 26322
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.110651969909668,
+      "learning_rate": 6.381422108081041e-06,
+      "loss": 0.6142,
+      "step": 26323
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.029797315597534,
+      "learning_rate": 6.38116871063538e-06,
+      "loss": 0.7153,
+      "step": 26324
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.9713175296783447,
+      "learning_rate": 6.380915309349206e-06,
+      "loss": 0.5634,
+      "step": 26325
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.553008556365967,
+      "learning_rate": 6.380661904223224e-06,
+      "loss": 0.5446,
+      "step": 26326
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.877845287322998,
+      "learning_rate": 6.380408495258139e-06,
+      "loss": 0.6771,
+      "step": 26327
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.105860710144043,
+      "learning_rate": 6.380155082454654e-06,
+      "loss": 0.5981,
+      "step": 26328
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.0888478755950928,
+      "learning_rate": 6.379901665813475e-06,
+      "loss": 0.6477,
+      "step": 26329
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.194521188735962,
+      "learning_rate": 6.379648245335306e-06,
+      "loss": 0.6172,
+      "step": 26330
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.5820956230163574,
+      "learning_rate": 6.379394821020852e-06,
+      "loss": 0.5062,
+      "step": 26331
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.221342086791992,
+      "learning_rate": 6.379141392870819e-06,
+      "loss": 0.7421,
+      "step": 26332
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.4442436695098877,
+      "learning_rate": 6.37888796088591e-06,
+      "loss": 0.5431,
+      "step": 26333
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 4.4956841468811035,
+      "learning_rate": 6.37863452506683e-06,
+      "loss": 0.644,
+      "step": 26334
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.9202877283096313,
+      "learning_rate": 6.3783810854142835e-06,
+      "loss": 0.4058,
+      "step": 26335
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.5937507152557373,
+      "learning_rate": 6.378127641928976e-06,
+      "loss": 0.6495,
+      "step": 26336
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.711857557296753,
+      "learning_rate": 6.377874194611612e-06,
+      "loss": 0.7062,
+      "step": 26337
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.8420816659927368,
+      "learning_rate": 6.377620743462896e-06,
+      "loss": 0.6954,
+      "step": 26338
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.8078598976135254,
+      "learning_rate": 6.377367288483532e-06,
+      "loss": 0.9719,
+      "step": 26339
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.9532560110092163,
+      "learning_rate": 6.377113829674225e-06,
+      "loss": 0.5753,
+      "step": 26340
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.8940658569335938,
+      "learning_rate": 6.376860367035681e-06,
+      "loss": 0.5195,
+      "step": 26341
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.7791438102722168,
+      "learning_rate": 6.376606900568605e-06,
+      "loss": 0.7096,
+      "step": 26342
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.7915323972702026,
+      "learning_rate": 6.3763534302737e-06,
+      "loss": 0.5307,
+      "step": 26343
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 6.471492290496826,
+      "learning_rate": 6.376099956151672e-06,
+      "loss": 0.7536,
+      "step": 26344
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.1008901596069336,
+      "learning_rate": 6.3758464782032255e-06,
+      "loss": 0.7308,
+      "step": 26345
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.6300323009490967,
+      "learning_rate": 6.375592996429065e-06,
+      "loss": 0.6049,
+      "step": 26346
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.686716079711914,
+      "learning_rate": 6.375339510829897e-06,
+      "loss": 0.6603,
+      "step": 26347
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.340618133544922,
+      "learning_rate": 6.3750860214064235e-06,
+      "loss": 0.4152,
+      "step": 26348
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.5056729316711426,
+      "learning_rate": 6.374832528159351e-06,
+      "loss": 0.6698,
+      "step": 26349
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.4401395320892334,
+      "learning_rate": 6.3745790310893855e-06,
+      "loss": 0.7384,
+      "step": 26350
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.545647144317627,
+      "learning_rate": 6.374325530197229e-06,
+      "loss": 0.6842,
+      "step": 26351
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.9856008291244507,
+      "learning_rate": 6.374072025483588e-06,
+      "loss": 0.5836,
+      "step": 26352
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.3273606300354004,
+      "learning_rate": 6.3738185169491696e-06,
+      "loss": 0.4875,
+      "step": 26353
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 6.988214492797852,
+      "learning_rate": 6.373565004594675e-06,
+      "loss": 0.5241,
+      "step": 26354
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.7559733390808105,
+      "learning_rate": 6.373311488420812e-06,
+      "loss": 0.7916,
+      "step": 26355
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.9949886798858643,
+      "learning_rate": 6.373057968428282e-06,
+      "loss": 0.7538,
+      "step": 26356
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.2133331298828125,
+      "learning_rate": 6.372804444617794e-06,
+      "loss": 0.5123,
+      "step": 26357
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.854151487350464,
+      "learning_rate": 6.37255091699005e-06,
+      "loss": 0.7284,
+      "step": 26358
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.4713187217712402,
+      "learning_rate": 6.372297385545757e-06,
+      "loss": 0.7359,
+      "step": 26359
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.9249038696289062,
+      "learning_rate": 6.372043850285618e-06,
+      "loss": 0.4091,
+      "step": 26360
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.217268466949463,
+      "learning_rate": 6.371790311210339e-06,
+      "loss": 0.4742,
+      "step": 26361
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.1344170570373535,
+      "learning_rate": 6.371536768320625e-06,
+      "loss": 0.5285,
+      "step": 26362
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.0788700580596924,
+      "learning_rate": 6.371283221617182e-06,
+      "loss": 0.4972,
+      "step": 26363
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.39719820022583,
+      "learning_rate": 6.371029671100715e-06,
+      "loss": 0.7266,
+      "step": 26364
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.32614803314209,
+      "learning_rate": 6.370776116771926e-06,
+      "loss": 0.6091,
+      "step": 26365
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.7495734691619873,
+      "learning_rate": 6.3705225586315225e-06,
+      "loss": 0.8147,
+      "step": 26366
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 4.631075382232666,
+      "learning_rate": 6.3702689966802085e-06,
+      "loss": 0.6834,
+      "step": 26367
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.177710771560669,
+      "learning_rate": 6.370015430918691e-06,
+      "loss": 0.6339,
+      "step": 26368
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 4.037644386291504,
+      "learning_rate": 6.369761861347674e-06,
+      "loss": 0.7197,
+      "step": 26369
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.8746129274368286,
+      "learning_rate": 6.369508287967863e-06,
+      "loss": 0.7626,
+      "step": 26370
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 22.008285522460938,
+      "learning_rate": 6.369254710779961e-06,
+      "loss": 0.6297,
+      "step": 26371
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.5274733304977417,
+      "learning_rate": 6.369001129784676e-06,
+      "loss": 0.7985,
+      "step": 26372
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.7893218994140625,
+      "learning_rate": 6.368747544982708e-06,
+      "loss": 0.5586,
+      "step": 26373
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.9613710641860962,
+      "learning_rate": 6.3684939563747695e-06,
+      "loss": 0.7141,
+      "step": 26374
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 5.320328235626221,
+      "learning_rate": 6.3682403639615605e-06,
+      "loss": 0.5468,
+      "step": 26375
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.0658018589019775,
+      "learning_rate": 6.367986767743788e-06,
+      "loss": 0.6847,
+      "step": 26376
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.9289743900299072,
+      "learning_rate": 6.367733167722158e-06,
+      "loss": 0.66,
+      "step": 26377
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.396653652191162,
+      "learning_rate": 6.367479563897373e-06,
+      "loss": 0.526,
+      "step": 26378
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.214890480041504,
+      "learning_rate": 6.36722595627014e-06,
+      "loss": 0.7583,
+      "step": 26379
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.8069210052490234,
+      "learning_rate": 6.366972344841164e-06,
+      "loss": 0.5878,
+      "step": 26380
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.680449962615967,
+      "learning_rate": 6.366718729611148e-06,
+      "loss": 0.7272,
+      "step": 26381
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.9697372913360596,
+      "learning_rate": 6.366465110580802e-06,
+      "loss": 0.4876,
+      "step": 26382
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.978543758392334,
+      "learning_rate": 6.366211487750826e-06,
+      "loss": 0.7114,
+      "step": 26383
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.3956680297851562,
+      "learning_rate": 6.36595786112193e-06,
+      "loss": 0.5138,
+      "step": 26384
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.4247539043426514,
+      "learning_rate": 6.365704230694815e-06,
+      "loss": 0.6323,
+      "step": 26385
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.7052927017211914,
+      "learning_rate": 6.36545059647019e-06,
+      "loss": 0.8282,
+      "step": 26386
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.7132292985916138,
+      "learning_rate": 6.365196958448757e-06,
+      "loss": 0.8032,
+      "step": 26387
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.1922733783721924,
+      "learning_rate": 6.364943316631223e-06,
+      "loss": 0.7118,
+      "step": 26388
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.630483627319336,
+      "learning_rate": 6.364689671018294e-06,
+      "loss": 0.4554,
+      "step": 26389
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.4815633296966553,
+      "learning_rate": 6.364436021610672e-06,
+      "loss": 0.6647,
+      "step": 26390
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.417257070541382,
+      "learning_rate": 6.364182368409066e-06,
+      "loss": 0.8506,
+      "step": 26391
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.581862211227417,
+      "learning_rate": 6.363928711414179e-06,
+      "loss": 0.6764,
+      "step": 26392
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.167586326599121,
+      "learning_rate": 6.3636750506267185e-06,
+      "loss": 0.4576,
+      "step": 26393
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.2931277751922607,
+      "learning_rate": 6.363421386047388e-06,
+      "loss": 0.7096,
+      "step": 26394
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.860201120376587,
+      "learning_rate": 6.363167717676893e-06,
+      "loss": 0.6665,
+      "step": 26395
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.3767189979553223,
+      "learning_rate": 6.36291404551594e-06,
+      "loss": 0.6909,
+      "step": 26396
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.846840739250183,
+      "learning_rate": 6.3626603695652325e-06,
+      "loss": 0.5755,
+      "step": 26397
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.570854425430298,
+      "learning_rate": 6.362406689825478e-06,
+      "loss": 0.315,
+      "step": 26398
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.70530366897583,
+      "learning_rate": 6.36215300629738e-06,
+      "loss": 0.6954,
+      "step": 26399
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.5346648693084717,
+      "learning_rate": 6.361899318981645e-06,
+      "loss": 0.8919,
+      "step": 26400
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.940661072731018,
+      "learning_rate": 6.361645627878979e-06,
+      "loss": 0.6448,
+      "step": 26401
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.4819655418395996,
+      "learning_rate": 6.3613919329900855e-06,
+      "loss": 0.6299,
+      "step": 26402
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.868621349334717,
+      "learning_rate": 6.361138234315671e-06,
+      "loss": 0.5995,
+      "step": 26403
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.9921611547470093,
+      "learning_rate": 6.360884531856441e-06,
+      "loss": 0.8542,
+      "step": 26404
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.4330880641937256,
+      "learning_rate": 6.360630825613102e-06,
+      "loss": 0.7588,
+      "step": 26405
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.765097141265869,
+      "learning_rate": 6.360377115586358e-06,
+      "loss": 0.4019,
+      "step": 26406
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.9186978340148926,
+      "learning_rate": 6.360123401776914e-06,
+      "loss": 0.837,
+      "step": 26407
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.4564085006713867,
+      "learning_rate": 6.359869684185478e-06,
+      "loss": 0.6691,
+      "step": 26408
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.713658332824707,
+      "learning_rate": 6.359615962812753e-06,
+      "loss": 0.7488,
+      "step": 26409
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.121758460998535,
+      "learning_rate": 6.359362237659445e-06,
+      "loss": 0.7581,
+      "step": 26410
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.5751832723617554,
+      "learning_rate": 6.359108508726259e-06,
+      "loss": 0.6943,
+      "step": 26411
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.8920961618423462,
+      "learning_rate": 6.358854776013902e-06,
+      "loss": 0.864,
+      "step": 26412
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.597820997238159,
+      "learning_rate": 6.35860103952308e-06,
+      "loss": 0.64,
+      "step": 26413
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.8086674213409424,
+      "learning_rate": 6.358347299254496e-06,
+      "loss": 0.4706,
+      "step": 26414
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.945073127746582,
+      "learning_rate": 6.358093555208858e-06,
+      "loss": 0.6976,
+      "step": 26415
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.8664215803146362,
+      "learning_rate": 6.35783980738687e-06,
+      "loss": 0.6191,
+      "step": 26416
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 4.049529552459717,
+      "learning_rate": 6.357586055789239e-06,
+      "loss": 0.9698,
+      "step": 26417
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.369900703430176,
+      "learning_rate": 6.357332300416669e-06,
+      "loss": 0.6695,
+      "step": 26418
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.53152334690094,
+      "learning_rate": 6.357078541269867e-06,
+      "loss": 0.5078,
+      "step": 26419
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.1839349269866943,
+      "learning_rate": 6.356824778349537e-06,
+      "loss": 0.6243,
+      "step": 26420
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.8062663078308105,
+      "learning_rate": 6.356571011656386e-06,
+      "loss": 0.5467,
+      "step": 26421
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.1375598907470703,
+      "learning_rate": 6.356317241191119e-06,
+      "loss": 0.508,
+      "step": 26422
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.2869958877563477,
+      "learning_rate": 6.356063466954442e-06,
+      "loss": 0.515,
+      "step": 26423
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.218886137008667,
+      "learning_rate": 6.35580968894706e-06,
+      "loss": 0.8092,
+      "step": 26424
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.7953879833221436,
+      "learning_rate": 6.35555590716968e-06,
+      "loss": 0.7711,
+      "step": 26425
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.661541223526001,
+      "learning_rate": 6.355302121623006e-06,
+      "loss": 0.7184,
+      "step": 26426
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.8371193408966064,
+      "learning_rate": 6.355048332307746e-06,
+      "loss": 0.5399,
+      "step": 26427
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 8.524480819702148,
+      "learning_rate": 6.354794539224602e-06,
+      "loss": 0.5824,
+      "step": 26428
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.9474358558654785,
+      "learning_rate": 6.354540742374283e-06,
+      "loss": 0.7032,
+      "step": 26429
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.8230563402175903,
+      "learning_rate": 6.354286941757493e-06,
+      "loss": 0.7343,
+      "step": 26430
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.0455946922302246,
+      "learning_rate": 6.35403313737494e-06,
+      "loss": 0.606,
+      "step": 26431
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.253929853439331,
+      "learning_rate": 6.3537793292273275e-06,
+      "loss": 0.6781,
+      "step": 26432
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.0483570098876953,
+      "learning_rate": 6.353525517315362e-06,
+      "loss": 0.6573,
+      "step": 26433
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.174121618270874,
+      "learning_rate": 6.353271701639748e-06,
+      "loss": 0.7705,
+      "step": 26434
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.0651938915252686,
+      "learning_rate": 6.353017882201193e-06,
+      "loss": 0.701,
+      "step": 26435
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.202385902404785,
+      "learning_rate": 6.352764059000401e-06,
+      "loss": 0.7988,
+      "step": 26436
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 5.407050609588623,
+      "learning_rate": 6.3525102320380805e-06,
+      "loss": 0.7217,
+      "step": 26437
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.7812509536743164,
+      "learning_rate": 6.352256401314936e-06,
+      "loss": 0.6472,
+      "step": 26438
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.27510142326355,
+      "learning_rate": 6.352002566831671e-06,
+      "loss": 0.4721,
+      "step": 26439
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 4.023129940032959,
+      "learning_rate": 6.351748728588995e-06,
+      "loss": 0.6941,
+      "step": 26440
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.440530300140381,
+      "learning_rate": 6.351494886587612e-06,
+      "loss": 0.5551,
+      "step": 26441
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.4304141998291016,
+      "learning_rate": 6.351241040828226e-06,
+      "loss": 0.7924,
+      "step": 26442
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.1806626319885254,
+      "learning_rate": 6.350987191311548e-06,
+      "loss": 0.7246,
+      "step": 26443
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.4377474784851074,
+      "learning_rate": 6.350733338038279e-06,
+      "loss": 0.5029,
+      "step": 26444
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.0235843658447266,
+      "learning_rate": 6.350479481009126e-06,
+      "loss": 0.595,
+      "step": 26445
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 4.251864433288574,
+      "learning_rate": 6.350225620224797e-06,
+      "loss": 0.5965,
+      "step": 26446
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.185896158218384,
+      "learning_rate": 6.349971755685995e-06,
+      "loss": 0.7919,
+      "step": 26447
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.5841829776763916,
+      "learning_rate": 6.349717887393429e-06,
+      "loss": 0.7478,
+      "step": 26448
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.9315330982208252,
+      "learning_rate": 6.349464015347801e-06,
+      "loss": 0.5197,
+      "step": 26449
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.6891634464263916,
+      "learning_rate": 6.349210139549821e-06,
+      "loss": 0.7253,
+      "step": 26450
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.5954556465148926,
+      "learning_rate": 6.348956260000192e-06,
+      "loss": 0.5115,
+      "step": 26451
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.4346578121185303,
+      "learning_rate": 6.3487023766996204e-06,
+      "loss": 0.6637,
+      "step": 26452
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.8639400005340576,
+      "learning_rate": 6.348448489648813e-06,
+      "loss": 0.737,
+      "step": 26453
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.10247802734375,
+      "learning_rate": 6.348194598848476e-06,
+      "loss": 0.6301,
+      "step": 26454
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.1474649906158447,
+      "learning_rate": 6.347940704299316e-06,
+      "loss": 0.5729,
+      "step": 26455
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.6115827560424805,
+      "learning_rate": 6.3476868060020354e-06,
+      "loss": 0.6654,
+      "step": 26456
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.6898040771484375,
+      "learning_rate": 6.347432903957345e-06,
+      "loss": 0.7213,
+      "step": 26457
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.174482822418213,
+      "learning_rate": 6.347178998165947e-06,
+      "loss": 0.6733,
+      "step": 26458
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.1832211017608643,
+      "learning_rate": 6.34692508862855e-06,
+      "loss": 0.535,
+      "step": 26459
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.6663076877593994,
+      "learning_rate": 6.346671175345858e-06,
+      "loss": 0.5809,
+      "step": 26460
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.0155253410339355,
+      "learning_rate": 6.346417258318578e-06,
+      "loss": 0.49,
+      "step": 26461
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.5809166431427,
+      "learning_rate": 6.346163337547415e-06,
+      "loss": 0.5395,
+      "step": 26462
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.2423079013824463,
+      "learning_rate": 6.345909413033078e-06,
+      "loss": 0.7209,
+      "step": 26463
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.3612842559814453,
+      "learning_rate": 6.34565548477627e-06,
+      "loss": 0.6197,
+      "step": 26464
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.8787765502929688,
+      "learning_rate": 6.345401552777698e-06,
+      "loss": 0.525,
+      "step": 26465
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.4973063468933105,
+      "learning_rate": 6.345147617038069e-06,
+      "loss": 0.8735,
+      "step": 26466
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.0169718265533447,
+      "learning_rate": 6.344893677558089e-06,
+      "loss": 0.6125,
+      "step": 26467
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.2563388347625732,
+      "learning_rate": 6.344639734338461e-06,
+      "loss": 0.5895,
+      "step": 26468
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.7035611867904663,
+      "learning_rate": 6.344385787379897e-06,
+      "loss": 0.5456,
+      "step": 26469
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.160396933555603,
+      "learning_rate": 6.344131836683097e-06,
+      "loss": 0.5653,
+      "step": 26470
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.7457149028778076,
+      "learning_rate": 6.3438778822487714e-06,
+      "loss": 0.7212,
+      "step": 26471
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.709585189819336,
+      "learning_rate": 6.343623924077622e-06,
+      "loss": 0.5872,
+      "step": 26472
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.759921669960022,
+      "learning_rate": 6.3433699621703606e-06,
+      "loss": 0.7142,
+      "step": 26473
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.1557905673980713,
+      "learning_rate": 6.34311599652769e-06,
+      "loss": 0.6778,
+      "step": 26474
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.5201468467712402,
+      "learning_rate": 6.342862027150317e-06,
+      "loss": 0.7393,
+      "step": 26475
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.5653533935546875,
+      "learning_rate": 6.342608054038948e-06,
+      "loss": 0.5903,
+      "step": 26476
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.877209186553955,
+      "learning_rate": 6.34235407719429e-06,
+      "loss": 0.9212,
+      "step": 26477
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.8077274560928345,
+      "learning_rate": 6.342100096617045e-06,
+      "loss": 0.5161,
+      "step": 26478
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.1805100440979004,
+      "learning_rate": 6.341846112307926e-06,
+      "loss": 0.4409,
+      "step": 26479
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.7090344429016113,
+      "learning_rate": 6.341592124267634e-06,
+      "loss": 0.545,
+      "step": 26480
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.5777932405471802,
+      "learning_rate": 6.341338132496876e-06,
+      "loss": 0.637,
+      "step": 26481
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.7048990726470947,
+      "learning_rate": 6.341084136996361e-06,
+      "loss": 0.7014,
+      "step": 26482
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.7311480045318604,
+      "learning_rate": 6.34083013776679e-06,
+      "loss": 0.5399,
+      "step": 26483
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.909562349319458,
+      "learning_rate": 6.340576134808876e-06,
+      "loss": 0.8119,
+      "step": 26484
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.8122847080230713,
+      "learning_rate": 6.340322128123321e-06,
+      "loss": 0.6196,
+      "step": 26485
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.9329174757003784,
+      "learning_rate": 6.340068117710832e-06,
+      "loss": 0.5124,
+      "step": 26486
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.2869746685028076,
+      "learning_rate": 6.339814103572115e-06,
+      "loss": 0.7036,
+      "step": 26487
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.742795705795288,
+      "learning_rate": 6.339560085707878e-06,
+      "loss": 0.4335,
+      "step": 26488
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.7769172191619873,
+      "learning_rate": 6.3393060641188264e-06,
+      "loss": 0.6926,
+      "step": 26489
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.7619808912277222,
+      "learning_rate": 6.3390520388056654e-06,
+      "loss": 0.8839,
+      "step": 26490
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.094301700592041,
+      "learning_rate": 6.338798009769103e-06,
+      "loss": 0.5638,
+      "step": 26491
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.908463954925537,
+      "learning_rate": 6.3385439770098435e-06,
+      "loss": 0.6992,
+      "step": 26492
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.5594043731689453,
+      "learning_rate": 6.3382899405285954e-06,
+      "loss": 0.7139,
+      "step": 26493
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.4704184532165527,
+      "learning_rate": 6.338035900326065e-06,
+      "loss": 0.6072,
+      "step": 26494
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.6692442893981934,
+      "learning_rate": 6.337781856402958e-06,
+      "loss": 0.5669,
+      "step": 26495
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.5417850017547607,
+      "learning_rate": 6.33752780875998e-06,
+      "loss": 0.5315,
+      "step": 26496
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.2529666423797607,
+      "learning_rate": 6.3372737573978385e-06,
+      "loss": 0.7252,
+      "step": 26497
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.808546543121338,
+      "learning_rate": 6.337019702317239e-06,
+      "loss": 0.5653,
+      "step": 26498
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.2723426818847656,
+      "learning_rate": 6.33676564351889e-06,
+      "loss": 0.6263,
+      "step": 26499
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.348255157470703,
+      "learning_rate": 6.336511581003495e-06,
+      "loss": 0.5212,
+      "step": 26500
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.7314279079437256,
+      "learning_rate": 6.336257514771763e-06,
+      "loss": 0.7713,
+      "step": 26501
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.4254677295684814,
+      "learning_rate": 6.3360034448244e-06,
+      "loss": 0.5314,
+      "step": 26502
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.5924298763275146,
+      "learning_rate": 6.33574937116211e-06,
+      "loss": 0.6671,
+      "step": 26503
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.8472983837127686,
+      "learning_rate": 6.335495293785603e-06,
+      "loss": 0.7432,
+      "step": 26504
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.961158037185669,
+      "learning_rate": 6.335241212695583e-06,
+      "loss": 0.8019,
+      "step": 26505
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.488032102584839,
+      "learning_rate": 6.334987127892757e-06,
+      "loss": 0.8538,
+      "step": 26506
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.1976802349090576,
+      "learning_rate": 6.334733039377834e-06,
+      "loss": 0.5657,
+      "step": 26507
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.1323509216308594,
+      "learning_rate": 6.3344789471515165e-06,
+      "loss": 0.5667,
+      "step": 26508
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.4559147357940674,
+      "learning_rate": 6.334224851214514e-06,
+      "loss": 0.9159,
+      "step": 26509
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.8961172103881836,
+      "learning_rate": 6.333970751567531e-06,
+      "loss": 0.8298,
+      "step": 26510
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.0260121822357178,
+      "learning_rate": 6.333716648211276e-06,
+      "loss": 0.5727,
+      "step": 26511
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.203507423400879,
+      "learning_rate": 6.3334625411464545e-06,
+      "loss": 0.6757,
+      "step": 26512
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.7381051778793335,
+      "learning_rate": 6.333208430373772e-06,
+      "loss": 0.6736,
+      "step": 26513
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.691427230834961,
+      "learning_rate": 6.332954315893936e-06,
+      "loss": 0.7497,
+      "step": 26514
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.9074569940567017,
+      "learning_rate": 6.3327001977076555e-06,
+      "loss": 0.6005,
+      "step": 26515
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.1441354751586914,
+      "learning_rate": 6.332446075815634e-06,
+      "loss": 0.8136,
+      "step": 26516
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 4.162297248840332,
+      "learning_rate": 6.332191950218579e-06,
+      "loss": 0.9117,
+      "step": 26517
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.8083757162094116,
+      "learning_rate": 6.331937820917198e-06,
+      "loss": 0.7266,
+      "step": 26518
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.7758395671844482,
+      "learning_rate": 6.331683687912196e-06,
+      "loss": 0.4811,
+      "step": 26519
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.2169387340545654,
+      "learning_rate": 6.331429551204281e-06,
+      "loss": 0.7663,
+      "step": 26520
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.7030484676361084,
+      "learning_rate": 6.3311754107941595e-06,
+      "loss": 0.5651,
+      "step": 26521
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.526130199432373,
+      "learning_rate": 6.330921266682537e-06,
+      "loss": 0.7034,
+      "step": 26522
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.806452512741089,
+      "learning_rate": 6.330667118870121e-06,
+      "loss": 0.6532,
+      "step": 26523
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.4097518920898438,
+      "learning_rate": 6.330412967357618e-06,
+      "loss": 0.6464,
+      "step": 26524
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.268489122390747,
+      "learning_rate": 6.3301588121457355e-06,
+      "loss": 0.6564,
+      "step": 26525
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.976529121398926,
+      "learning_rate": 6.3299046532351795e-06,
+      "loss": 0.7935,
+      "step": 26526
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.4948127269744873,
+      "learning_rate": 6.329650490626657e-06,
+      "loss": 0.7128,
+      "step": 26527
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.060706377029419,
+      "learning_rate": 6.329396324320875e-06,
+      "loss": 0.5922,
+      "step": 26528
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.5319719314575195,
+      "learning_rate": 6.32914215431854e-06,
+      "loss": 0.8245,
+      "step": 26529
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.751542091369629,
+      "learning_rate": 6.328887980620357e-06,
+      "loss": 0.5954,
+      "step": 26530
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 5.468276023864746,
+      "learning_rate": 6.328633803227036e-06,
+      "loss": 0.7098,
+      "step": 26531
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 4.457767486572266,
+      "learning_rate": 6.328379622139282e-06,
+      "loss": 0.6163,
+      "step": 26532
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.303184986114502,
+      "learning_rate": 6.3281254373578e-06,
+      "loss": 0.556,
+      "step": 26533
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.4712324142456055,
+      "learning_rate": 6.3278712488833e-06,
+      "loss": 0.6447,
+      "step": 26534
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.336843252182007,
+      "learning_rate": 6.327617056716488e-06,
+      "loss": 0.3954,
+      "step": 26535
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.5195693969726562,
+      "learning_rate": 6.327362860858069e-06,
+      "loss": 0.6447,
+      "step": 26536
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.9591853618621826,
+      "learning_rate": 6.327108661308753e-06,
+      "loss": 0.6906,
+      "step": 26537
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.2150275707244873,
+      "learning_rate": 6.326854458069243e-06,
+      "loss": 0.5536,
+      "step": 26538
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.2008440494537354,
+      "learning_rate": 6.326600251140249e-06,
+      "loss": 0.6488,
+      "step": 26539
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.187980890274048,
+      "learning_rate": 6.326346040522477e-06,
+      "loss": 0.2908,
+      "step": 26540
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.1362380981445312,
+      "learning_rate": 6.326091826216633e-06,
+      "loss": 0.4557,
+      "step": 26541
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.2152905464172363,
+      "learning_rate": 6.325837608223424e-06,
+      "loss": 0.6996,
+      "step": 26542
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.5395348072052,
+      "learning_rate": 6.325583386543558e-06,
+      "loss": 0.4959,
+      "step": 26543
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 4.140082836151123,
+      "learning_rate": 6.32532916117774e-06,
+      "loss": 0.76,
+      "step": 26544
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.2281458377838135,
+      "learning_rate": 6.325074932126678e-06,
+      "loss": 0.7438,
+      "step": 26545
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.3358654975891113,
+      "learning_rate": 6.32482069939108e-06,
+      "loss": 0.8261,
+      "step": 26546
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.217733383178711,
+      "learning_rate": 6.324566462971652e-06,
+      "loss": 0.8583,
+      "step": 26547
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.8557984828948975,
+      "learning_rate": 6.3243122228691e-06,
+      "loss": 0.5333,
+      "step": 26548
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 7.917334079742432,
+      "learning_rate": 6.324057979084133e-06,
+      "loss": 0.8837,
+      "step": 26549
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.866202712059021,
+      "learning_rate": 6.323803731617456e-06,
+      "loss": 0.5465,
+      "step": 26550
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.153818130493164,
+      "learning_rate": 6.323549480469778e-06,
+      "loss": 0.6664,
+      "step": 26551
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.3635804653167725,
+      "learning_rate": 6.323295225641804e-06,
+      "loss": 0.5138,
+      "step": 26552
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.960581660270691,
+      "learning_rate": 6.32304096713424e-06,
+      "loss": 0.7468,
+      "step": 26553
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.5512213706970215,
+      "learning_rate": 6.322786704947795e-06,
+      "loss": 0.9745,
+      "step": 26554
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.6346933841705322,
+      "learning_rate": 6.3225324390831774e-06,
+      "loss": 1.0064,
+      "step": 26555
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.0084733963012695,
+      "learning_rate": 6.322278169541092e-06,
+      "loss": 0.7581,
+      "step": 26556
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.823857307434082,
+      "learning_rate": 6.322023896322246e-06,
+      "loss": 0.6979,
+      "step": 26557
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.319366693496704,
+      "learning_rate": 6.321769619427346e-06,
+      "loss": 0.8929,
+      "step": 26558
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.028047800064087,
+      "learning_rate": 6.321515338857101e-06,
+      "loss": 0.71,
+      "step": 26559
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 4.24907112121582,
+      "learning_rate": 6.321261054612216e-06,
+      "loss": 0.4795,
+      "step": 26560
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.620206356048584,
+      "learning_rate": 6.3210067666934e-06,
+      "loss": 0.6962,
+      "step": 26561
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.7436320781707764,
+      "learning_rate": 6.3207524751013595e-06,
+      "loss": 0.6858,
+      "step": 26562
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 5.661412239074707,
+      "learning_rate": 6.3204981798368e-06,
+      "loss": 0.5019,
+      "step": 26563
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.332818031311035,
+      "learning_rate": 6.32024388090043e-06,
+      "loss": 0.5503,
+      "step": 26564
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 4.551425457000732,
+      "learning_rate": 6.319989578292955e-06,
+      "loss": 0.9595,
+      "step": 26565
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 7.030488014221191,
+      "learning_rate": 6.319735272015085e-06,
+      "loss": 0.4209,
+      "step": 26566
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.231692314147949,
+      "learning_rate": 6.319480962067526e-06,
+      "loss": 0.7633,
+      "step": 26567
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.9323222637176514,
+      "learning_rate": 6.319226648450985e-06,
+      "loss": 0.8213,
+      "step": 26568
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.55483341217041,
+      "learning_rate": 6.318972331166166e-06,
+      "loss": 0.6162,
+      "step": 26569
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.064229726791382,
+      "learning_rate": 6.318718010213783e-06,
+      "loss": 0.6101,
+      "step": 26570
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.2114789485931396,
+      "learning_rate": 6.318463685594537e-06,
+      "loss": 0.7726,
+      "step": 26571
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.5959455966949463,
+      "learning_rate": 6.318209357309137e-06,
+      "loss": 0.5696,
+      "step": 26572
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.974365472793579,
+      "learning_rate": 6.3179550253582925e-06,
+      "loss": 0.7129,
+      "step": 26573
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.4965230226516724,
+      "learning_rate": 6.317700689742706e-06,
+      "loss": 0.5691,
+      "step": 26574
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.884701132774353,
+      "learning_rate": 6.317446350463089e-06,
+      "loss": 0.4451,
+      "step": 26575
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.409182071685791,
+      "learning_rate": 6.3171920075201474e-06,
+      "loss": 0.6351,
+      "step": 26576
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 4.58384370803833,
+      "learning_rate": 6.316937660914588e-06,
+      "loss": 0.7495,
+      "step": 26577
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.3445446491241455,
+      "learning_rate": 6.3166833106471194e-06,
+      "loss": 0.6665,
+      "step": 26578
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.825364351272583,
+      "learning_rate": 6.316428956718446e-06,
+      "loss": 0.6219,
+      "step": 26579
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.5501198768615723,
+      "learning_rate": 6.316174599129279e-06,
+      "loss": 0.5648,
+      "step": 26580
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.1377758979797363,
+      "learning_rate": 6.315920237880322e-06,
+      "loss": 0.7129,
+      "step": 26581
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.9417933225631714,
+      "learning_rate": 6.315665872972283e-06,
+      "loss": 0.5144,
+      "step": 26582
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.8103809356689453,
+      "learning_rate": 6.315411504405872e-06,
+      "loss": 0.7082,
+      "step": 26583
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.631669759750366,
+      "learning_rate": 6.315157132181792e-06,
+      "loss": 0.5183,
+      "step": 26584
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.237070083618164,
+      "learning_rate": 6.314902756300754e-06,
+      "loss": 0.6056,
+      "step": 26585
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.5676136016845703,
+      "learning_rate": 6.314648376763464e-06,
+      "loss": 0.462,
+      "step": 26586
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.5220680236816406,
+      "learning_rate": 6.31439399357063e-06,
+      "loss": 0.6052,
+      "step": 26587
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.6749236583709717,
+      "learning_rate": 6.314139606722958e-06,
+      "loss": 0.4367,
+      "step": 26588
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.193526029586792,
+      "learning_rate": 6.3138852162211565e-06,
+      "loss": 0.6986,
+      "step": 26589
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.8331809043884277,
+      "learning_rate": 6.3136308220659316e-06,
+      "loss": 0.7602,
+      "step": 26590
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.1753346920013428,
+      "learning_rate": 6.313376424257992e-06,
+      "loss": 0.6018,
+      "step": 26591
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.663811445236206,
+      "learning_rate": 6.313122022798045e-06,
+      "loss": 0.6024,
+      "step": 26592
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.4944136142730713,
+      "learning_rate": 6.312867617686796e-06,
+      "loss": 0.5693,
+      "step": 26593
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.3475050926208496,
+      "learning_rate": 6.312613208924955e-06,
+      "loss": 0.4823,
+      "step": 26594
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.9824326038360596,
+      "learning_rate": 6.312358796513228e-06,
+      "loss": 0.5314,
+      "step": 26595
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.046037197113037,
+      "learning_rate": 6.312104380452324e-06,
+      "loss": 0.4887,
+      "step": 26596
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.849156379699707,
+      "learning_rate": 6.311849960742948e-06,
+      "loss": 0.8901,
+      "step": 26597
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.027482748031616,
+      "learning_rate": 6.311595537385809e-06,
+      "loss": 0.7628,
+      "step": 26598
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.1295080184936523,
+      "learning_rate": 6.311341110381615e-06,
+      "loss": 0.5971,
+      "step": 26599
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.2511096000671387,
+      "learning_rate": 6.311086679731071e-06,
+      "loss": 0.6867,
+      "step": 26600
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.8876452445983887,
+      "learning_rate": 6.310832245434888e-06,
+      "loss": 0.8527,
+      "step": 26601
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.473724126815796,
+      "learning_rate": 6.310577807493769e-06,
+      "loss": 0.7444,
+      "step": 26602
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.3042452335357666,
+      "learning_rate": 6.310323365908427e-06,
+      "loss": 0.638,
+      "step": 26603
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.60072922706604,
+      "learning_rate": 6.310068920679565e-06,
+      "loss": 0.6322,
+      "step": 26604
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.7966530323028564,
+      "learning_rate": 6.309814471807891e-06,
+      "loss": 0.5305,
+      "step": 26605
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.844188928604126,
+      "learning_rate": 6.309560019294116e-06,
+      "loss": 0.5425,
+      "step": 26606
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.2487430572509766,
+      "learning_rate": 6.309305563138943e-06,
+      "loss": 0.5313,
+      "step": 26607
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.308685302734375,
+      "learning_rate": 6.3090511033430836e-06,
+      "loss": 0.6322,
+      "step": 26608
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.8951451778411865,
+      "learning_rate": 6.308796639907242e-06,
+      "loss": 0.6497,
+      "step": 26609
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.993668794631958,
+      "learning_rate": 6.3085421728321285e-06,
+      "loss": 0.5598,
+      "step": 26610
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.151010513305664,
+      "learning_rate": 6.30828770211845e-06,
+      "loss": 0.7347,
+      "step": 26611
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.7981173992156982,
+      "learning_rate": 6.308033227766911e-06,
+      "loss": 0.6282,
+      "step": 26612
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.657387375831604,
+      "learning_rate": 6.307778749778224e-06,
+      "loss": 0.6495,
+      "step": 26613
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 4.4092841148376465,
+      "learning_rate": 6.307524268153092e-06,
+      "loss": 0.6642,
+      "step": 26614
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.9933326244354248,
+      "learning_rate": 6.307269782892224e-06,
+      "loss": 0.6545,
+      "step": 26615
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.2550411224365234,
+      "learning_rate": 6.307015293996331e-06,
+      "loss": 0.4775,
+      "step": 26616
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.6646065711975098,
+      "learning_rate": 6.306760801466119e-06,
+      "loss": 0.72,
+      "step": 26617
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.448996067047119,
+      "learning_rate": 6.3065063053022925e-06,
+      "loss": 0.7842,
+      "step": 26618
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.7412540912628174,
+      "learning_rate": 6.306251805505562e-06,
+      "loss": 0.7211,
+      "step": 26619
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 6.2253899574279785,
+      "learning_rate": 6.3059973020766345e-06,
+      "loss": 0.5801,
+      "step": 26620
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 6.822980880737305,
+      "learning_rate": 6.305742795016217e-06,
+      "loss": 0.482,
+      "step": 26621
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.363767385482788,
+      "learning_rate": 6.305488284325019e-06,
+      "loss": 0.6527,
+      "step": 26622
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.301394462585449,
+      "learning_rate": 6.305233770003747e-06,
+      "loss": 0.4881,
+      "step": 26623
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.997690439224243,
+      "learning_rate": 6.304979252053109e-06,
+      "loss": 0.8349,
+      "step": 26624
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.8553775548934937,
+      "learning_rate": 6.304724730473812e-06,
+      "loss": 0.3144,
+      "step": 26625
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.689791440963745,
+      "learning_rate": 6.304470205266564e-06,
+      "loss": 0.8372,
+      "step": 26626
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 4.156919956207275,
+      "learning_rate": 6.304215676432074e-06,
+      "loss": 0.7531,
+      "step": 26627
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.535299062728882,
+      "learning_rate": 6.303961143971048e-06,
+      "loss": 0.8118,
+      "step": 26628
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.847012758255005,
+      "learning_rate": 6.303706607884194e-06,
+      "loss": 0.3617,
+      "step": 26629
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.2338883876800537,
+      "learning_rate": 6.303452068172222e-06,
+      "loss": 0.6524,
+      "step": 26630
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.4117591381073,
+      "learning_rate": 6.303197524835838e-06,
+      "loss": 0.6945,
+      "step": 26631
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 5.55844783782959,
+      "learning_rate": 6.302942977875749e-06,
+      "loss": 0.716,
+      "step": 26632
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.194977283477783,
+      "learning_rate": 6.302688427292664e-06,
+      "loss": 0.5897,
+      "step": 26633
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.0694637298583984,
+      "learning_rate": 6.3024338730872905e-06,
+      "loss": 0.459,
+      "step": 26634
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.3133352994918823,
+      "learning_rate": 6.302179315260335e-06,
+      "loss": 0.5866,
+      "step": 26635
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.9550148248672485,
+      "learning_rate": 6.301924753812507e-06,
+      "loss": 0.7977,
+      "step": 26636
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.6017799377441406,
+      "learning_rate": 6.301670188744516e-06,
+      "loss": 0.5774,
+      "step": 26637
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.055854082107544,
+      "learning_rate": 6.301415620057066e-06,
+      "loss": 0.7718,
+      "step": 26638
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.001967191696167,
+      "learning_rate": 6.301161047750868e-06,
+      "loss": 0.3327,
+      "step": 26639
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.7085744142532349,
+      "learning_rate": 6.3009064718266275e-06,
+      "loss": 0.4448,
+      "step": 26640
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.9763569831848145,
+      "learning_rate": 6.300651892285054e-06,
+      "loss": 0.512,
+      "step": 26641
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.298048496246338,
+      "learning_rate": 6.300397309126854e-06,
+      "loss": 0.6346,
+      "step": 26642
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.484103202819824,
+      "learning_rate": 6.300142722352738e-06,
+      "loss": 0.4894,
+      "step": 26643
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.7214335203170776,
+      "learning_rate": 6.29988813196341e-06,
+      "loss": 0.8141,
+      "step": 26644
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.2108561992645264,
+      "learning_rate": 6.29963353795958e-06,
+      "loss": 0.7331,
+      "step": 26645
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 5.347782135009766,
+      "learning_rate": 6.299378940341957e-06,
+      "loss": 1.0447,
+      "step": 26646
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.329228401184082,
+      "learning_rate": 6.299124339111249e-06,
+      "loss": 0.8721,
+      "step": 26647
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.647148847579956,
+      "learning_rate": 6.298869734268162e-06,
+      "loss": 0.4637,
+      "step": 26648
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.729398488998413,
+      "learning_rate": 6.298615125813403e-06,
+      "loss": 0.6638,
+      "step": 26649
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.2292873859405518,
+      "learning_rate": 6.298360513747684e-06,
+      "loss": 0.7319,
+      "step": 26650
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.661839246749878,
+      "learning_rate": 6.2981058980717106e-06,
+      "loss": 0.6442,
+      "step": 26651
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 5.028087139129639,
+      "learning_rate": 6.297851278786191e-06,
+      "loss": 0.8551,
+      "step": 26652
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.704716205596924,
+      "learning_rate": 6.297596655891832e-06,
+      "loss": 0.6367,
+      "step": 26653
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.45200514793396,
+      "learning_rate": 6.2973420293893435e-06,
+      "loss": 0.7392,
+      "step": 26654
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.2346668243408203,
+      "learning_rate": 6.297087399279433e-06,
+      "loss": 0.7286,
+      "step": 26655
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.258873462677002,
+      "learning_rate": 6.296832765562808e-06,
+      "loss": 0.6621,
+      "step": 26656
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.6246633529663086,
+      "learning_rate": 6.296578128240177e-06,
+      "loss": 0.7691,
+      "step": 26657
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.1941418647766113,
+      "learning_rate": 6.296323487312248e-06,
+      "loss": 0.6444,
+      "step": 26658
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.137186288833618,
+      "learning_rate": 6.2960688427797286e-06,
+      "loss": 0.8776,
+      "step": 26659
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.4378578662872314,
+      "learning_rate": 6.295814194643328e-06,
+      "loss": 0.6285,
+      "step": 26660
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.7147605419158936,
+      "learning_rate": 6.295559542903753e-06,
+      "loss": 0.9098,
+      "step": 26661
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.114189624786377,
+      "learning_rate": 6.295304887561713e-06,
+      "loss": 0.603,
+      "step": 26662
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.4465856552124023,
+      "learning_rate": 6.295050228617915e-06,
+      "loss": 0.5759,
+      "step": 26663
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.883461833000183,
+      "learning_rate": 6.294795566073067e-06,
+      "loss": 0.83,
+      "step": 26664
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.0924930572509766,
+      "learning_rate": 6.294540899927878e-06,
+      "loss": 0.5447,
+      "step": 26665
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.9506512880325317,
+      "learning_rate": 6.294286230183054e-06,
+      "loss": 0.542,
+      "step": 26666
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.2373080253601074,
+      "learning_rate": 6.294031556839307e-06,
+      "loss": 0.6127,
+      "step": 26667
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.9668618440628052,
+      "learning_rate": 6.293776879897342e-06,
+      "loss": 0.3851,
+      "step": 26668
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.719045639038086,
+      "learning_rate": 6.293522199357869e-06,
+      "loss": 0.7307,
+      "step": 26669
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.2383499145507812,
+      "learning_rate": 6.293267515221595e-06,
+      "loss": 0.6784,
+      "step": 26670
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.0522196292877197,
+      "learning_rate": 6.293012827489229e-06,
+      "loss": 0.5143,
+      "step": 26671
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.9819594621658325,
+      "learning_rate": 6.292758136161477e-06,
+      "loss": 0.5661,
+      "step": 26672
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.7409911155700684,
+      "learning_rate": 6.292503441239051e-06,
+      "loss": 0.804,
+      "step": 26673
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.248542547225952,
+      "learning_rate": 6.292248742722655e-06,
+      "loss": 0.4699,
+      "step": 26674
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.4766671657562256,
+      "learning_rate": 6.291994040613001e-06,
+      "loss": 0.929,
+      "step": 26675
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.8651931285858154,
+      "learning_rate": 6.291739334910795e-06,
+      "loss": 0.9217,
+      "step": 26676
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.3440475463867188,
+      "learning_rate": 6.291484625616745e-06,
+      "loss": 0.644,
+      "step": 26677
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.895601511001587,
+      "learning_rate": 6.291229912731561e-06,
+      "loss": 0.5921,
+      "step": 26678
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.5816562175750732,
+      "learning_rate": 6.29097519625595e-06,
+      "loss": 0.9403,
+      "step": 26679
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.203800678253174,
+      "learning_rate": 6.290720476190621e-06,
+      "loss": 0.4897,
+      "step": 26680
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.331103563308716,
+      "learning_rate": 6.290465752536282e-06,
+      "loss": 0.7572,
+      "step": 26681
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.8932358026504517,
+      "learning_rate": 6.29021102529364e-06,
+      "loss": 0.7695,
+      "step": 26682
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.6388769149780273,
+      "learning_rate": 6.289956294463405e-06,
+      "loss": 0.7602,
+      "step": 26683
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.9633489847183228,
+      "learning_rate": 6.289701560046285e-06,
+      "loss": 0.5911,
+      "step": 26684
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.1749179363250732,
+      "learning_rate": 6.289446822042988e-06,
+      "loss": 0.662,
+      "step": 26685
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 4.334047317504883,
+      "learning_rate": 6.289192080454223e-06,
+      "loss": 0.6448,
+      "step": 26686
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.7807247638702393,
+      "learning_rate": 6.288937335280697e-06,
+      "loss": 0.4568,
+      "step": 26687
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.1014482975006104,
+      "learning_rate": 6.288682586523118e-06,
+      "loss": 0.6952,
+      "step": 26688
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.2182366847991943,
+      "learning_rate": 6.288427834182198e-06,
+      "loss": 0.4946,
+      "step": 26689
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.292541265487671,
+      "learning_rate": 6.288173078258642e-06,
+      "loss": 0.6006,
+      "step": 26690
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.0450987815856934,
+      "learning_rate": 6.287918318753159e-06,
+      "loss": 0.4542,
+      "step": 26691
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.5212507247924805,
+      "learning_rate": 6.2876635556664565e-06,
+      "loss": 0.6447,
+      "step": 26692
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.1784608364105225,
+      "learning_rate": 6.287408788999245e-06,
+      "loss": 0.5115,
+      "step": 26693
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 4.951125144958496,
+      "learning_rate": 6.287154018752233e-06,
+      "loss": 0.6038,
+      "step": 26694
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.203484058380127,
+      "learning_rate": 6.2868992449261265e-06,
+      "loss": 0.4782,
+      "step": 26695
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.5686547756195068,
+      "learning_rate": 6.286644467521636e-06,
+      "loss": 0.7454,
+      "step": 26696
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.7362900972366333,
+      "learning_rate": 6.286389686539468e-06,
+      "loss": 0.4655,
+      "step": 26697
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 4.659455299377441,
+      "learning_rate": 6.286134901980333e-06,
+      "loss": 0.5645,
+      "step": 26698
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.12682843208313,
+      "learning_rate": 6.285880113844939e-06,
+      "loss": 0.6673,
+      "step": 26699
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.2720718383789062,
+      "learning_rate": 6.2856253221339944e-06,
+      "loss": 0.5714,
+      "step": 26700
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.8059998750686646,
+      "learning_rate": 6.285370526848207e-06,
+      "loss": 0.534,
+      "step": 26701
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.429266929626465,
+      "learning_rate": 6.285115727988284e-06,
+      "loss": 0.4468,
+      "step": 26702
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.652458429336548,
+      "learning_rate": 6.284860925554939e-06,
+      "loss": 0.5034,
+      "step": 26703
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.4430087804794312,
+      "learning_rate": 6.284606119548875e-06,
+      "loss": 0.6039,
+      "step": 26704
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.4580233097076416,
+      "learning_rate": 6.284351309970802e-06,
+      "loss": 0.6502,
+      "step": 26705
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.7900404930114746,
+      "learning_rate": 6.284096496821429e-06,
+      "loss": 0.5428,
+      "step": 26706
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.651550054550171,
+      "learning_rate": 6.2838416801014655e-06,
+      "loss": 0.7375,
+      "step": 26707
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.7535896301269531,
+      "learning_rate": 6.28358685981162e-06,
+      "loss": 0.7658,
+      "step": 26708
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.4074597358703613,
+      "learning_rate": 6.283332035952599e-06,
+      "loss": 0.7877,
+      "step": 26709
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.74761962890625,
+      "learning_rate": 6.283077208525113e-06,
+      "loss": 0.6174,
+      "step": 26710
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.942737340927124,
+      "learning_rate": 6.28282237752987e-06,
+      "loss": 0.6022,
+      "step": 26711
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.6159021854400635,
+      "learning_rate": 6.282567542967579e-06,
+      "loss": 0.4337,
+      "step": 26712
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.6140635013580322,
+      "learning_rate": 6.2823127048389475e-06,
+      "loss": 0.6096,
+      "step": 26713
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.807358980178833,
+      "learning_rate": 6.282057863144685e-06,
+      "loss": 0.772,
+      "step": 26714
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.7529845237731934,
+      "learning_rate": 6.281803017885499e-06,
+      "loss": 0.6434,
+      "step": 26715
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.8861756324768066,
+      "learning_rate": 6.281548169062099e-06,
+      "loss": 0.6738,
+      "step": 26716
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.5492618083953857,
+      "learning_rate": 6.281293316675195e-06,
+      "loss": 0.5573,
+      "step": 26717
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.739684581756592,
+      "learning_rate": 6.281038460725493e-06,
+      "loss": 0.6624,
+      "step": 26718
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.191504955291748,
+      "learning_rate": 6.280783601213703e-06,
+      "loss": 0.6443,
+      "step": 26719
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.0582897663116455,
+      "learning_rate": 6.280528738140535e-06,
+      "loss": 0.752,
+      "step": 26720
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.864478349685669,
+      "learning_rate": 6.280273871506695e-06,
+      "loss": 0.5496,
+      "step": 26721
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 4.289148330688477,
+      "learning_rate": 6.280019001312894e-06,
+      "loss": 0.6571,
+      "step": 26722
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.032702684402466,
+      "learning_rate": 6.2797641275598375e-06,
+      "loss": 0.7553,
+      "step": 26723
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.6179498434066772,
+      "learning_rate": 6.279509250248239e-06,
+      "loss": 0.5681,
+      "step": 26724
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.937766194343567,
+      "learning_rate": 6.279254369378803e-06,
+      "loss": 0.7839,
+      "step": 26725
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.7252233028411865,
+      "learning_rate": 6.278999484952239e-06,
+      "loss": 0.9842,
+      "step": 26726
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.5457239151000977,
+      "learning_rate": 6.278744596969257e-06,
+      "loss": 0.5612,
+      "step": 26727
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.6532604694366455,
+      "learning_rate": 6.2784897054305675e-06,
+      "loss": 0.6558,
+      "step": 26728
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.1798148155212402,
+      "learning_rate": 6.278234810336875e-06,
+      "loss": 0.5691,
+      "step": 26729
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.458587408065796,
+      "learning_rate": 6.277979911688891e-06,
+      "loss": 0.6424,
+      "step": 26730
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.98172926902771,
+      "learning_rate": 6.277725009487323e-06,
+      "loss": 0.3837,
+      "step": 26731
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.852684497833252,
+      "learning_rate": 6.277470103732882e-06,
+      "loss": 0.5619,
+      "step": 26732
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.473322868347168,
+      "learning_rate": 6.277215194426274e-06,
+      "loss": 0.7908,
+      "step": 26733
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.6123480796813965,
+      "learning_rate": 6.27696028156821e-06,
+      "loss": 0.5642,
+      "step": 26734
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.9306658506393433,
+      "learning_rate": 6.276705365159397e-06,
+      "loss": 0.6378,
+      "step": 26735
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.39814829826355,
+      "learning_rate": 6.276450445200544e-06,
+      "loss": 0.3893,
+      "step": 26736
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.983916759490967,
+      "learning_rate": 6.27619552169236e-06,
+      "loss": 0.6763,
+      "step": 26737
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.6903061866760254,
+      "learning_rate": 6.275940594635556e-06,
+      "loss": 0.745,
+      "step": 26738
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.8630307912826538,
+      "learning_rate": 6.275685664030838e-06,
+      "loss": 0.6908,
+      "step": 26739
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.8324652910232544,
+      "learning_rate": 6.275430729878917e-06,
+      "loss": 0.7675,
+      "step": 26740
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.3700451850891113,
+      "learning_rate": 6.275175792180501e-06,
+      "loss": 0.5709,
+      "step": 26741
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.2184882164001465,
+      "learning_rate": 6.274920850936298e-06,
+      "loss": 0.7343,
+      "step": 26742
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.207766056060791,
+      "learning_rate": 6.274665906147017e-06,
+      "loss": 0.978,
+      "step": 26743
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.8179947137832642,
+      "learning_rate": 6.274410957813368e-06,
+      "loss": 0.5512,
+      "step": 26744
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.7601969242095947,
+      "learning_rate": 6.274156005936061e-06,
+      "loss": 0.8416,
+      "step": 26745
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.5558719635009766,
+      "learning_rate": 6.273901050515803e-06,
+      "loss": 0.5392,
+      "step": 26746
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.08378267288208,
+      "learning_rate": 6.273646091553302e-06,
+      "loss": 0.5988,
+      "step": 26747
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.8796571493148804,
+      "learning_rate": 6.273391129049271e-06,
+      "loss": 0.5403,
+      "step": 26748
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.316563844680786,
+      "learning_rate": 6.273136163004414e-06,
+      "loss": 0.8192,
+      "step": 26749
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.5925073623657227,
+      "learning_rate": 6.272881193419443e-06,
+      "loss": 0.6881,
+      "step": 26750
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.1303584575653076,
+      "learning_rate": 6.272626220295066e-06,
+      "loss": 0.6994,
+      "step": 26751
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.6393659114837646,
+      "learning_rate": 6.272371243631992e-06,
+      "loss": 0.5532,
+      "step": 26752
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.3555521965026855,
+      "learning_rate": 6.27211626343093e-06,
+      "loss": 0.5186,
+      "step": 26753
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.826066732406616,
+      "learning_rate": 6.271861279692591e-06,
+      "loss": 0.6355,
+      "step": 26754
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.1476480960845947,
+      "learning_rate": 6.271606292417681e-06,
+      "loss": 0.3978,
+      "step": 26755
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.2048139572143555,
+      "learning_rate": 6.27135130160691e-06,
+      "loss": 0.641,
+      "step": 26756
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.2916581630706787,
+      "learning_rate": 6.271096307260986e-06,
+      "loss": 0.8025,
+      "step": 26757
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 5.902559280395508,
+      "learning_rate": 6.270841309380622e-06,
+      "loss": 1.1536,
+      "step": 26758
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.601529121398926,
+      "learning_rate": 6.270586307966524e-06,
+      "loss": 0.5345,
+      "step": 26759
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.6975557804107666,
+      "learning_rate": 6.2703313030194e-06,
+      "loss": 0.7142,
+      "step": 26760
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.5586128234863281,
+      "learning_rate": 6.270076294539962e-06,
+      "loss": 0.6537,
+      "step": 26761
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 4.6699700355529785,
+      "learning_rate": 6.2698212825289174e-06,
+      "loss": 0.8322,
+      "step": 26762
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.7012457847595215,
+      "learning_rate": 6.269566266986975e-06,
+      "loss": 0.8544,
+      "step": 26763
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.2696521282196045,
+      "learning_rate": 6.269311247914844e-06,
+      "loss": 0.551,
+      "step": 26764
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.9139459133148193,
+      "learning_rate": 6.269056225313235e-06,
+      "loss": 0.806,
+      "step": 26765
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.4348032474517822,
+      "learning_rate": 6.268801199182855e-06,
+      "loss": 0.5802,
+      "step": 26766
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.420775890350342,
+      "learning_rate": 6.2685461695244145e-06,
+      "loss": 0.6111,
+      "step": 26767
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.9226421117782593,
+      "learning_rate": 6.268291136338622e-06,
+      "loss": 0.9115,
+      "step": 26768
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.9835305213928223,
+      "learning_rate": 6.268036099626187e-06,
+      "loss": 0.3907,
+      "step": 26769
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.143195390701294,
+      "learning_rate": 6.267781059387819e-06,
+      "loss": 0.7464,
+      "step": 26770
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.7115572690963745,
+      "learning_rate": 6.267526015624227e-06,
+      "loss": 0.6133,
+      "step": 26771
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 4.960007667541504,
+      "learning_rate": 6.26727096833612e-06,
+      "loss": 0.8115,
+      "step": 26772
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.856825828552246,
+      "learning_rate": 6.267015917524206e-06,
+      "loss": 0.5893,
+      "step": 26773
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.256312131881714,
+      "learning_rate": 6.266760863189197e-06,
+      "loss": 0.5287,
+      "step": 26774
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 4.132086277008057,
+      "learning_rate": 6.2665058053318e-06,
+      "loss": 0.7344,
+      "step": 26775
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.2840468883514404,
+      "learning_rate": 6.266250743952724e-06,
+      "loss": 0.7672,
+      "step": 26776
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.966317057609558,
+      "learning_rate": 6.265995679052679e-06,
+      "loss": 0.7594,
+      "step": 26777
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.9424943923950195,
+      "learning_rate": 6.265740610632376e-06,
+      "loss": 0.493,
+      "step": 26778
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.016212224960327,
+      "learning_rate": 6.265485538692521e-06,
+      "loss": 0.7926,
+      "step": 26779
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.3987858295440674,
+      "learning_rate": 6.265230463233825e-06,
+      "loss": 0.584,
+      "step": 26780
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.1474764347076416,
+      "learning_rate": 6.264975384256997e-06,
+      "loss": 0.7439,
+      "step": 26781
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.667153835296631,
+      "learning_rate": 6.2647203017627474e-06,
+      "loss": 0.5151,
+      "step": 26782
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.3050007820129395,
+      "learning_rate": 6.264465215751784e-06,
+      "loss": 0.6289,
+      "step": 26783
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.120888113975525,
+      "learning_rate": 6.264210126224816e-06,
+      "loss": 0.525,
+      "step": 26784
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.0672481060028076,
+      "learning_rate": 6.263955033182554e-06,
+      "loss": 0.6407,
+      "step": 26785
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.1337783336639404,
+      "learning_rate": 6.263699936625705e-06,
+      "loss": 0.6184,
+      "step": 26786
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.3105247020721436,
+      "learning_rate": 6.2634448365549815e-06,
+      "loss": 0.596,
+      "step": 26787
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.1426827907562256,
+      "learning_rate": 6.263189732971089e-06,
+      "loss": 0.5936,
+      "step": 26788
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.487839937210083,
+      "learning_rate": 6.262934625874741e-06,
+      "loss": 0.9727,
+      "step": 26789
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.893023729324341,
+      "learning_rate": 6.262679515266644e-06,
+      "loss": 0.4533,
+      "step": 26790
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.8986445665359497,
+      "learning_rate": 6.2624244011475085e-06,
+      "loss": 0.4952,
+      "step": 26791
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.259305238723755,
+      "learning_rate": 6.262169283518045e-06,
+      "loss": 1.0104,
+      "step": 26792
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.4282126426696777,
+      "learning_rate": 6.2619141623789605e-06,
+      "loss": 0.4086,
+      "step": 26793
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.870868682861328,
+      "learning_rate": 6.261659037730965e-06,
+      "loss": 0.6708,
+      "step": 26794
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.2411417961120605,
+      "learning_rate": 6.261403909574768e-06,
+      "loss": 0.4237,
+      "step": 26795
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 5.510100841522217,
+      "learning_rate": 6.261148777911082e-06,
+      "loss": 0.8994,
+      "step": 26796
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.2088377475738525,
+      "learning_rate": 6.260893642740611e-06,
+      "loss": 0.6571,
+      "step": 26797
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.578925132751465,
+      "learning_rate": 6.260638504064066e-06,
+      "loss": 0.854,
+      "step": 26798
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.9051568508148193,
+      "learning_rate": 6.26038336188216e-06,
+      "loss": 0.5968,
+      "step": 26799
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.65743887424469,
+      "learning_rate": 6.260128216195599e-06,
+      "loss": 0.6505,
+      "step": 26800
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.251000165939331,
+      "learning_rate": 6.259873067005095e-06,
+      "loss": 0.7242,
+      "step": 26801
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.6136656999588013,
+      "learning_rate": 6.259617914311353e-06,
+      "loss": 0.4053,
+      "step": 26802
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.9857137203216553,
+      "learning_rate": 6.2593627581150884e-06,
+      "loss": 0.6043,
+      "step": 26803
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.6610147953033447,
+      "learning_rate": 6.259107598417007e-06,
+      "loss": 0.7079,
+      "step": 26804
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 4.65814208984375,
+      "learning_rate": 6.258852435217817e-06,
+      "loss": 0.7123,
+      "step": 26805
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.9974839687347412,
+      "learning_rate": 6.258597268518232e-06,
+      "loss": 0.3985,
+      "step": 26806
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.875870943069458,
+      "learning_rate": 6.2583420983189595e-06,
+      "loss": 0.7138,
+      "step": 26807
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.503607988357544,
+      "learning_rate": 6.258086924620707e-06,
+      "loss": 0.4611,
+      "step": 26808
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.981990933418274,
+      "learning_rate": 6.257831747424189e-06,
+      "loss": 0.7511,
+      "step": 26809
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.9410628080368042,
+      "learning_rate": 6.25757656673011e-06,
+      "loss": 0.7211,
+      "step": 26810
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.0663931369781494,
+      "learning_rate": 6.257321382539182e-06,
+      "loss": 0.7017,
+      "step": 26811
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.1176888942718506,
+      "learning_rate": 6.257066194852115e-06,
+      "loss": 0.5079,
+      "step": 26812
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.3858327865600586,
+      "learning_rate": 6.256811003669617e-06,
+      "loss": 0.7065,
+      "step": 26813
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.4104682207107544,
+      "learning_rate": 6.256555808992399e-06,
+      "loss": 0.6605,
+      "step": 26814
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 5.468951225280762,
+      "learning_rate": 6.25630061082117e-06,
+      "loss": 0.6767,
+      "step": 26815
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.108048915863037,
+      "learning_rate": 6.256045409156639e-06,
+      "loss": 0.888,
+      "step": 26816
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.6350724697113037,
+      "learning_rate": 6.255790203999518e-06,
+      "loss": 0.5435,
+      "step": 26817
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.8648570775985718,
+      "learning_rate": 6.255534995350512e-06,
+      "loss": 0.3274,
+      "step": 26818
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.388944149017334,
+      "learning_rate": 6.255279783210335e-06,
+      "loss": 0.7664,
+      "step": 26819
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.9911890029907227,
+      "learning_rate": 6.255024567579696e-06,
+      "loss": 0.5282,
+      "step": 26820
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.277679681777954,
+      "learning_rate": 6.254769348459303e-06,
+      "loss": 0.6301,
+      "step": 26821
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.2901411056518555,
+      "learning_rate": 6.254514125849867e-06,
+      "loss": 0.7591,
+      "step": 26822
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.8363139629364014,
+      "learning_rate": 6.254258899752097e-06,
+      "loss": 0.7831,
+      "step": 26823
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 4.255459308624268,
+      "learning_rate": 6.254003670166702e-06,
+      "loss": 0.8605,
+      "step": 26824
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.472177743911743,
+      "learning_rate": 6.2537484370943936e-06,
+      "loss": 0.7677,
+      "step": 26825
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.4564435482025146,
+      "learning_rate": 6.253493200535881e-06,
+      "loss": 0.7003,
+      "step": 26826
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.315277338027954,
+      "learning_rate": 6.253237960491871e-06,
+      "loss": 0.6102,
+      "step": 26827
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.5521764755249023,
+      "learning_rate": 6.252982716963076e-06,
+      "loss": 1.0184,
+      "step": 26828
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.542090654373169,
+      "learning_rate": 6.252727469950207e-06,
+      "loss": 0.7641,
+      "step": 26829
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.208404541015625,
+      "learning_rate": 6.252472219453972e-06,
+      "loss": 0.8545,
+      "step": 26830
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.852141857147217,
+      "learning_rate": 6.252216965475081e-06,
+      "loss": 0.3515,
+      "step": 26831
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 14.001016616821289,
+      "learning_rate": 6.251961708014244e-06,
+      "loss": 0.7241,
+      "step": 26832
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.4519858360290527,
+      "learning_rate": 6.251706447072169e-06,
+      "loss": 0.5568,
+      "step": 26833
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.4982370138168335,
+      "learning_rate": 6.251451182649568e-06,
+      "loss": 0.5877,
+      "step": 26834
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.748009204864502,
+      "learning_rate": 6.2511959147471514e-06,
+      "loss": 0.794,
+      "step": 26835
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.541015863418579,
+      "learning_rate": 6.250940643365626e-06,
+      "loss": 0.688,
+      "step": 26836
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.356870651245117,
+      "learning_rate": 6.2506853685057045e-06,
+      "loss": 0.8328,
+      "step": 26837
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.3667397499084473,
+      "learning_rate": 6.250430090168094e-06,
+      "loss": 0.759,
+      "step": 26838
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.7283670902252197,
+      "learning_rate": 6.250174808353507e-06,
+      "loss": 0.578,
+      "step": 26839
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.7015960216522217,
+      "learning_rate": 6.249919523062651e-06,
+      "loss": 0.7754,
+      "step": 26840
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.179966688156128,
+      "learning_rate": 6.249664234296238e-06,
+      "loss": 0.8096,
+      "step": 26841
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.961562156677246,
+      "learning_rate": 6.249408942054976e-06,
+      "loss": 0.6133,
+      "step": 26842
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.0896973609924316,
+      "learning_rate": 6.249153646339577e-06,
+      "loss": 0.6529,
+      "step": 26843
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.799443006515503,
+      "learning_rate": 6.2488983471507495e-06,
+      "loss": 0.6085,
+      "step": 26844
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.598841428756714,
+      "learning_rate": 6.2486430444892035e-06,
+      "loss": 0.7561,
+      "step": 26845
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 4.417217254638672,
+      "learning_rate": 6.248387738355648e-06,
+      "loss": 0.6291,
+      "step": 26846
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.0520176887512207,
+      "learning_rate": 6.248132428750795e-06,
+      "loss": 0.9145,
+      "step": 26847
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.01129150390625,
+      "learning_rate": 6.247877115675353e-06,
+      "loss": 0.7634,
+      "step": 26848
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.3788042068481445,
+      "learning_rate": 6.2476217991300315e-06,
+      "loss": 0.5561,
+      "step": 26849
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.0914559364318848,
+      "learning_rate": 6.247366479115541e-06,
+      "loss": 0.4962,
+      "step": 26850
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.746586561203003,
+      "learning_rate": 6.247111155632593e-06,
+      "loss": 0.7366,
+      "step": 26851
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.3060874938964844,
+      "learning_rate": 6.246855828681895e-06,
+      "loss": 0.5484,
+      "step": 26852
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.5710337162017822,
+      "learning_rate": 6.246600498264158e-06,
+      "loss": 0.6779,
+      "step": 26853
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 4.395318984985352,
+      "learning_rate": 6.246345164380092e-06,
+      "loss": 0.4503,
+      "step": 26854
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.4082672595977783,
+      "learning_rate": 6.246089827030408e-06,
+      "loss": 0.6434,
+      "step": 26855
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.1076338291168213,
+      "learning_rate": 6.245834486215815e-06,
+      "loss": 0.5911,
+      "step": 26856
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.2191758155822754,
+      "learning_rate": 6.245579141937021e-06,
+      "loss": 0.5099,
+      "step": 26857
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.4162063598632812,
+      "learning_rate": 6.2453237941947395e-06,
+      "loss": 0.6879,
+      "step": 26858
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.7179827690124512,
+      "learning_rate": 6.2450684429896795e-06,
+      "loss": 0.6518,
+      "step": 26859
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.231778621673584,
+      "learning_rate": 6.244813088322549e-06,
+      "loss": 0.6442,
+      "step": 26860
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.473613977432251,
+      "learning_rate": 6.24455773019406e-06,
+      "loss": 0.3322,
+      "step": 26861
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.9109854698181152,
+      "learning_rate": 6.244302368604923e-06,
+      "loss": 0.6548,
+      "step": 26862
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.430724620819092,
+      "learning_rate": 6.244047003555848e-06,
+      "loss": 0.8351,
+      "step": 26863
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.479074001312256,
+      "learning_rate": 6.243791635047543e-06,
+      "loss": 0.7266,
+      "step": 26864
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.209540367126465,
+      "learning_rate": 6.243536263080719e-06,
+      "loss": 0.4801,
+      "step": 26865
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 4.032709121704102,
+      "learning_rate": 6.243280887656088e-06,
+      "loss": 0.7714,
+      "step": 26866
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.1146650314331055,
+      "learning_rate": 6.243025508774357e-06,
+      "loss": 0.6591,
+      "step": 26867
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.655848979949951,
+      "learning_rate": 6.242770126436238e-06,
+      "loss": 0.7521,
+      "step": 26868
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 9.816300392150879,
+      "learning_rate": 6.242514740642441e-06,
+      "loss": 0.3953,
+      "step": 26869
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 4.4515228271484375,
+      "learning_rate": 6.242259351393677e-06,
+      "loss": 1.1204,
+      "step": 26870
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 4.230494976043701,
+      "learning_rate": 6.2420039586906544e-06,
+      "loss": 0.7111,
+      "step": 26871
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.8896005153656006,
+      "learning_rate": 6.241748562534084e-06,
+      "loss": 0.4986,
+      "step": 26872
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.5127954483032227,
+      "learning_rate": 6.241493162924674e-06,
+      "loss": 0.4556,
+      "step": 26873
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 4.039949893951416,
+      "learning_rate": 6.24123775986314e-06,
+      "loss": 0.8775,
+      "step": 26874
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.0472679138183594,
+      "learning_rate": 6.240982353350187e-06,
+      "loss": 0.6562,
+      "step": 26875
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.6683461666107178,
+      "learning_rate": 6.2407269433865285e-06,
+      "loss": 0.7397,
+      "step": 26876
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.8967833518981934,
+      "learning_rate": 6.240471529972872e-06,
+      "loss": 0.5145,
+      "step": 26877
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.1959638595581055,
+      "learning_rate": 6.240216113109928e-06,
+      "loss": 0.4977,
+      "step": 26878
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.7123093605041504,
+      "learning_rate": 6.239960692798408e-06,
+      "loss": 0.6393,
+      "step": 26879
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.945995569229126,
+      "learning_rate": 6.239705269039023e-06,
+      "loss": 0.6529,
+      "step": 26880
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.5372743606567383,
+      "learning_rate": 6.23944984183248e-06,
+      "loss": 0.5596,
+      "step": 26881
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.028789520263672,
+      "learning_rate": 6.2391944111794936e-06,
+      "loss": 0.5775,
+      "step": 26882
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.05090594291687,
+      "learning_rate": 6.2389389770807705e-06,
+      "loss": 0.7726,
+      "step": 26883
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.192970037460327,
+      "learning_rate": 6.238683539537023e-06,
+      "loss": 0.8585,
+      "step": 26884
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.285773277282715,
+      "learning_rate": 6.23842809854896e-06,
+      "loss": 0.4536,
+      "step": 26885
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.14640474319458,
+      "learning_rate": 6.238172654117291e-06,
+      "loss": 0.6322,
+      "step": 26886
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.1558876037597656,
+      "learning_rate": 6.237917206242731e-06,
+      "loss": 0.8456,
+      "step": 26887
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.883009672164917,
+      "learning_rate": 6.2376617549259844e-06,
+      "loss": 0.8301,
+      "step": 26888
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.2025978565216064,
+      "learning_rate": 6.237406300167764e-06,
+      "loss": 0.6922,
+      "step": 26889
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.5479116439819336,
+      "learning_rate": 6.237150841968781e-06,
+      "loss": 0.5316,
+      "step": 26890
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.0017457008361816,
+      "learning_rate": 6.236895380329745e-06,
+      "loss": 0.4279,
+      "step": 26891
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.2810962200164795,
+      "learning_rate": 6.236639915251367e-06,
+      "loss": 0.761,
+      "step": 26892
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.7767342329025269,
+      "learning_rate": 6.236384446734356e-06,
+      "loss": 0.5731,
+      "step": 26893
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.728085517883301,
+      "learning_rate": 6.2361289747794225e-06,
+      "loss": 0.7656,
+      "step": 26894
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.8406791687011719,
+      "learning_rate": 6.235873499387278e-06,
+      "loss": 0.6617,
+      "step": 26895
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.3851828575134277,
+      "learning_rate": 6.235618020558632e-06,
+      "loss": 0.8376,
+      "step": 26896
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.3071038722991943,
+      "learning_rate": 6.2353625382941965e-06,
+      "loss": 0.8163,
+      "step": 26897
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.4204394817352295,
+      "learning_rate": 6.2351070525946785e-06,
+      "loss": 0.7801,
+      "step": 26898
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.9595375061035156,
+      "learning_rate": 6.234851563460791e-06,
+      "loss": 0.7698,
+      "step": 26899
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 4.037227630615234,
+      "learning_rate": 6.2345960708932444e-06,
+      "loss": 1.0179,
+      "step": 26900
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 4.5548577308654785,
+      "learning_rate": 6.234340574892748e-06,
+      "loss": 0.6981,
+      "step": 26901
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.915825843811035,
+      "learning_rate": 6.234085075460014e-06,
+      "loss": 0.4806,
+      "step": 26902
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.28939151763916,
+      "learning_rate": 6.23382957259575e-06,
+      "loss": 0.7284,
+      "step": 26903
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.7543888092041016,
+      "learning_rate": 6.233574066300669e-06,
+      "loss": 0.8079,
+      "step": 26904
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.2751829624176025,
+      "learning_rate": 6.233318556575482e-06,
+      "loss": 0.5465,
+      "step": 26905
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.8656630516052246,
+      "learning_rate": 6.233063043420896e-06,
+      "loss": 0.5245,
+      "step": 26906
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.4638497829437256,
+      "learning_rate": 6.232807526837625e-06,
+      "loss": 0.6853,
+      "step": 26907
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.327537775039673,
+      "learning_rate": 6.232552006826378e-06,
+      "loss": 0.5377,
+      "step": 26908
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.285923719406128,
+      "learning_rate": 6.232296483387865e-06,
+      "loss": 0.6424,
+      "step": 26909
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.1081221103668213,
+      "learning_rate": 6.232040956522796e-06,
+      "loss": 0.8314,
+      "step": 26910
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.725780963897705,
+      "learning_rate": 6.231785426231884e-06,
+      "loss": 0.7715,
+      "step": 26911
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.9563357830047607,
+      "learning_rate": 6.231529892515839e-06,
+      "loss": 0.7958,
+      "step": 26912
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.473273277282715,
+      "learning_rate": 6.231274355375369e-06,
+      "loss": 0.6993,
+      "step": 26913
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 4.727413177490234,
+      "learning_rate": 6.231018814811187e-06,
+      "loss": 0.5618,
+      "step": 26914
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.7755353450775146,
+      "learning_rate": 6.230763270824003e-06,
+      "loss": 0.8472,
+      "step": 26915
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.8868076801300049,
+      "learning_rate": 6.230507723414527e-06,
+      "loss": 0.4465,
+      "step": 26916
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.6451098918914795,
+      "learning_rate": 6.230252172583471e-06,
+      "loss": 0.6984,
+      "step": 26917
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.4573330879211426,
+      "learning_rate": 6.229996618331544e-06,
+      "loss": 0.7663,
+      "step": 26918
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.0340118408203125,
+      "learning_rate": 6.229741060659457e-06,
+      "loss": 0.55,
+      "step": 26919
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.8694703578948975,
+      "learning_rate": 6.22948549956792e-06,
+      "loss": 0.5052,
+      "step": 26920
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.4817501306533813,
+      "learning_rate": 6.229229935057644e-06,
+      "loss": 0.3637,
+      "step": 26921
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.9908835887908936,
+      "learning_rate": 6.228974367129343e-06,
+      "loss": 0.8289,
+      "step": 26922
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.9549800157546997,
+      "learning_rate": 6.228718795783721e-06,
+      "loss": 0.5343,
+      "step": 26923
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.439997673034668,
+      "learning_rate": 6.228463221021494e-06,
+      "loss": 0.6176,
+      "step": 26924
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.147610664367676,
+      "learning_rate": 6.228207642843372e-06,
+      "loss": 0.6202,
+      "step": 26925
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.29563045501709,
+      "learning_rate": 6.2279520612500644e-06,
+      "loss": 0.7294,
+      "step": 26926
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.7812176942825317,
+      "learning_rate": 6.227696476242282e-06,
+      "loss": 0.5228,
+      "step": 26927
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.7469664812088013,
+      "learning_rate": 6.227440887820735e-06,
+      "loss": 0.568,
+      "step": 26928
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.1085243225097656,
+      "learning_rate": 6.2271852959861335e-06,
+      "loss": 0.7303,
+      "step": 26929
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.5157285928726196,
+      "learning_rate": 6.226929700739191e-06,
+      "loss": 0.6631,
+      "step": 26930
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.926445722579956,
+      "learning_rate": 6.226674102080617e-06,
+      "loss": 0.6258,
+      "step": 26931
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.2600913047790527,
+      "learning_rate": 6.22641850001112e-06,
+      "loss": 0.7278,
+      "step": 26932
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.4070217609405518,
+      "learning_rate": 6.226162894531414e-06,
+      "loss": 0.6006,
+      "step": 26933
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.224827289581299,
+      "learning_rate": 6.225907285642208e-06,
+      "loss": 0.3999,
+      "step": 26934
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.273425579071045,
+      "learning_rate": 6.225651673344213e-06,
+      "loss": 0.5814,
+      "step": 26935
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.7149181365966797,
+      "learning_rate": 6.22539605763814e-06,
+      "loss": 0.5357,
+      "step": 26936
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.2968131303787231,
+      "learning_rate": 6.225140438524699e-06,
+      "loss": 0.6166,
+      "step": 26937
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.6153953075408936,
+      "learning_rate": 6.224884816004603e-06,
+      "loss": 0.4357,
+      "step": 26938
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.752579927444458,
+      "learning_rate": 6.22462919007856e-06,
+      "loss": 0.7527,
+      "step": 26939
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.878965377807617,
+      "learning_rate": 6.224373560747281e-06,
+      "loss": 0.6825,
+      "step": 26940
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.6743128299713135,
+      "learning_rate": 6.224117928011479e-06,
+      "loss": 0.5684,
+      "step": 26941
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.3030433654785156,
+      "learning_rate": 6.223862291871864e-06,
+      "loss": 0.6206,
+      "step": 26942
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.6107213497161865,
+      "learning_rate": 6.223606652329146e-06,
+      "loss": 0.6169,
+      "step": 26943
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.382072687149048,
+      "learning_rate": 6.2233510093840355e-06,
+      "loss": 0.4552,
+      "step": 26944
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.290104389190674,
+      "learning_rate": 6.223095363037246e-06,
+      "loss": 0.8057,
+      "step": 26945
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.737710952758789,
+      "learning_rate": 6.222839713289485e-06,
+      "loss": 0.4874,
+      "step": 26946
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.1949689388275146,
+      "learning_rate": 6.222584060141465e-06,
+      "loss": 0.6708,
+      "step": 26947
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.779360055923462,
+      "learning_rate": 6.222328403593897e-06,
+      "loss": 0.7622,
+      "step": 26948
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.9604170322418213,
+      "learning_rate": 6.2220727436474915e-06,
+      "loss": 0.7632,
+      "step": 26949
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.236095666885376,
+      "learning_rate": 6.221817080302959e-06,
+      "loss": 0.6442,
+      "step": 26950
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.6926472187042236,
+      "learning_rate": 6.2215614135610115e-06,
+      "loss": 0.609,
+      "step": 26951
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.5267932415008545,
+      "learning_rate": 6.221305743422359e-06,
+      "loss": 0.5484,
+      "step": 26952
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.625962018966675,
+      "learning_rate": 6.221050069887713e-06,
+      "loss": 0.7522,
+      "step": 26953
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.736842632293701,
+      "learning_rate": 6.220794392957784e-06,
+      "loss": 0.3848,
+      "step": 26954
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.1506168842315674,
+      "learning_rate": 6.220538712633284e-06,
+      "loss": 0.5748,
+      "step": 26955
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.392897367477417,
+      "learning_rate": 6.220283028914922e-06,
+      "loss": 0.4516,
+      "step": 26956
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.5193793773651123,
+      "learning_rate": 6.2200273418034105e-06,
+      "loss": 0.6471,
+      "step": 26957
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.0405964851379395,
+      "learning_rate": 6.21977165129946e-06,
+      "loss": 0.4255,
+      "step": 26958
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.8592967987060547,
+      "learning_rate": 6.219515957403781e-06,
+      "loss": 0.5103,
+      "step": 26959
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 4.199376106262207,
+      "learning_rate": 6.219260260117085e-06,
+      "loss": 0.8938,
+      "step": 26960
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.8650482892990112,
+      "learning_rate": 6.219004559440083e-06,
+      "loss": 0.6916,
+      "step": 26961
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.335390567779541,
+      "learning_rate": 6.218748855373487e-06,
+      "loss": 0.7776,
+      "step": 26962
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.3757522106170654,
+      "learning_rate": 6.2184931479180055e-06,
+      "loss": 0.4548,
+      "step": 26963
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.5230531692504883,
+      "learning_rate": 6.218237437074353e-06,
+      "loss": 0.7269,
+      "step": 26964
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.44279408454895,
+      "learning_rate": 6.217981722843236e-06,
+      "loss": 0.8361,
+      "step": 26965
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.8693113327026367,
+      "learning_rate": 6.21772600522537e-06,
+      "loss": 0.9096,
+      "step": 26966
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.0133378505706787,
+      "learning_rate": 6.217470284221464e-06,
+      "loss": 0.4469,
+      "step": 26967
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 4.323156833648682,
+      "learning_rate": 6.217214559832229e-06,
+      "loss": 0.4778,
+      "step": 26968
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.8830307722091675,
+      "learning_rate": 6.2169588320583755e-06,
+      "loss": 0.6034,
+      "step": 26969
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.030888080596924,
+      "learning_rate": 6.216703100900616e-06,
+      "loss": 0.8038,
+      "step": 26970
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.5375845432281494,
+      "learning_rate": 6.21644736635966e-06,
+      "loss": 0.3216,
+      "step": 26971
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.9187628030776978,
+      "learning_rate": 6.21619162843622e-06,
+      "loss": 0.4014,
+      "step": 26972
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.703611135482788,
+      "learning_rate": 6.2159358871310075e-06,
+      "loss": 0.4234,
+      "step": 26973
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.137741804122925,
+      "learning_rate": 6.215680142444732e-06,
+      "loss": 0.4519,
+      "step": 26974
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.3741633892059326,
+      "learning_rate": 6.215424394378107e-06,
+      "loss": 0.7117,
+      "step": 26975
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.7166876792907715,
+      "learning_rate": 6.21516864293184e-06,
+      "loss": 0.6351,
+      "step": 26976
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.977485179901123,
+      "learning_rate": 6.214912888106645e-06,
+      "loss": 0.5611,
+      "step": 26977
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.8697965145111084,
+      "learning_rate": 6.214657129903233e-06,
+      "loss": 0.5759,
+      "step": 26978
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.9385311603546143,
+      "learning_rate": 6.214401368322314e-06,
+      "loss": 0.6718,
+      "step": 26979
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.497472047805786,
+      "learning_rate": 6.2141456033646e-06,
+      "loss": 0.6037,
+      "step": 26980
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.7789267301559448,
+      "learning_rate": 6.213889835030801e-06,
+      "loss": 0.5058,
+      "step": 26981
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.6301522254943848,
+      "learning_rate": 6.21363406332163e-06,
+      "loss": 0.6084,
+      "step": 26982
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.0321803092956543,
+      "learning_rate": 6.2133782882377966e-06,
+      "loss": 0.601,
+      "step": 26983
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.97794771194458,
+      "learning_rate": 6.213122509780013e-06,
+      "loss": 0.546,
+      "step": 26984
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.149460792541504,
+      "learning_rate": 6.212866727948991e-06,
+      "loss": 0.7656,
+      "step": 26985
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.01969575881958,
+      "learning_rate": 6.21261094274544e-06,
+      "loss": 0.4555,
+      "step": 26986
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.277132272720337,
+      "learning_rate": 6.212355154170073e-06,
+      "loss": 0.6956,
+      "step": 26987
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.2136199474334717,
+      "learning_rate": 6.2120993622236e-06,
+      "loss": 0.5841,
+      "step": 26988
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.4386816024780273,
+      "learning_rate": 6.211843566906732e-06,
+      "loss": 0.5023,
+      "step": 26989
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.206230401992798,
+      "learning_rate": 6.211587768220182e-06,
+      "loss": 0.6581,
+      "step": 26990
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.068354845046997,
+      "learning_rate": 6.21133196616466e-06,
+      "loss": 0.8264,
+      "step": 26991
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.9782103300094604,
+      "learning_rate": 6.211076160740877e-06,
+      "loss": 0.6237,
+      "step": 26992
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.6946916580200195,
+      "learning_rate": 6.210820351949546e-06,
+      "loss": 0.5658,
+      "step": 26993
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.7514169216156006,
+      "learning_rate": 6.2105645397913775e-06,
+      "loss": 0.5204,
+      "step": 26994
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.3165924549102783,
+      "learning_rate": 6.210308724267081e-06,
+      "loss": 0.6974,
+      "step": 26995
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.1098196506500244,
+      "learning_rate": 6.210052905377368e-06,
+      "loss": 0.8531,
+      "step": 26996
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 4.107371807098389,
+      "learning_rate": 6.209797083122955e-06,
+      "loss": 0.676,
+      "step": 26997
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.911758542060852,
+      "learning_rate": 6.209541257504547e-06,
+      "loss": 0.6601,
+      "step": 26998
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.8243080377578735,
+      "learning_rate": 6.209285428522859e-06,
+      "loss": 0.672,
+      "step": 26999
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.6037826538085938,
+      "learning_rate": 6.209029596178601e-06,
+      "loss": 0.6031,
+      "step": 27000
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.6275012493133545,
+      "learning_rate": 6.208773760472484e-06,
+      "loss": 0.7445,
+      "step": 27001
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.6085007190704346,
+      "learning_rate": 6.20851792140522e-06,
+      "loss": 0.7149,
+      "step": 27002
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.415502905845642,
+      "learning_rate": 6.20826207897752e-06,
+      "loss": 0.6221,
+      "step": 27003
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.3393123149871826,
+      "learning_rate": 6.208006233190097e-06,
+      "loss": 0.5918,
+      "step": 27004
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.3696696758270264,
+      "learning_rate": 6.20775038404366e-06,
+      "loss": 0.6258,
+      "step": 27005
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.2156965732574463,
+      "learning_rate": 6.207494531538923e-06,
+      "loss": 0.6737,
+      "step": 27006
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.993857979774475,
+      "learning_rate": 6.207238675676596e-06,
+      "loss": 0.5997,
+      "step": 27007
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.2756636142730713,
+      "learning_rate": 6.20698281645739e-06,
+      "loss": 0.8378,
+      "step": 27008
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.2534754276275635,
+      "learning_rate": 6.206726953882016e-06,
+      "loss": 0.6517,
+      "step": 27009
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 4.785937786102295,
+      "learning_rate": 6.206471087951188e-06,
+      "loss": 0.3526,
+      "step": 27010
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 4.759292125701904,
+      "learning_rate": 6.2062152186656145e-06,
+      "loss": 0.9072,
+      "step": 27011
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.0797016620635986,
+      "learning_rate": 6.205959346026008e-06,
+      "loss": 0.7845,
+      "step": 27012
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.938225030899048,
+      "learning_rate": 6.205703470033082e-06,
+      "loss": 0.5757,
+      "step": 27013
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.6773927211761475,
+      "learning_rate": 6.205447590687546e-06,
+      "loss": 0.7224,
+      "step": 27014
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.051945924758911,
+      "learning_rate": 6.20519170799011e-06,
+      "loss": 0.6575,
+      "step": 27015
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.8748629093170166,
+      "learning_rate": 6.204935821941489e-06,
+      "loss": 0.4541,
+      "step": 27016
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.9885330200195312,
+      "learning_rate": 6.204679932542393e-06,
+      "loss": 0.8792,
+      "step": 27017
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.4091172218322754,
+      "learning_rate": 6.2044240397935326e-06,
+      "loss": 0.6007,
+      "step": 27018
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.2032437324523926,
+      "learning_rate": 6.2041681436956216e-06,
+      "loss": 0.5796,
+      "step": 27019
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.0137364864349365,
+      "learning_rate": 6.203912244249368e-06,
+      "loss": 0.8147,
+      "step": 27020
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.2838447093963623,
+      "learning_rate": 6.203656341455486e-06,
+      "loss": 0.5658,
+      "step": 27021
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.9009945392608643,
+      "learning_rate": 6.203400435314687e-06,
+      "loss": 0.58,
+      "step": 27022
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.4940474033355713,
+      "learning_rate": 6.203144525827682e-06,
+      "loss": 0.5113,
+      "step": 27023
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.1968464851379395,
+      "learning_rate": 6.202888612995183e-06,
+      "loss": 0.6151,
+      "step": 27024
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.587719440460205,
+      "learning_rate": 6.2026326968179015e-06,
+      "loss": 0.7599,
+      "step": 27025
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.3620034456253052,
+      "learning_rate": 6.20237677729655e-06,
+      "loss": 0.504,
+      "step": 27026
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 4.049697399139404,
+      "learning_rate": 6.202120854431838e-06,
+      "loss": 0.6463,
+      "step": 27027
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.747266173362732,
+      "learning_rate": 6.201864928224478e-06,
+      "loss": 0.4657,
+      "step": 27028
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.8520166873931885,
+      "learning_rate": 6.201608998675182e-06,
+      "loss": 0.7587,
+      "step": 27029
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 5.5176167488098145,
+      "learning_rate": 6.2013530657846625e-06,
+      "loss": 0.5487,
+      "step": 27030
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.0303094387054443,
+      "learning_rate": 6.201097129553628e-06,
+      "loss": 0.419,
+      "step": 27031
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.7458534240722656,
+      "learning_rate": 6.200841189982794e-06,
+      "loss": 0.684,
+      "step": 27032
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.8216568231582642,
+      "learning_rate": 6.200585247072869e-06,
+      "loss": 0.315,
+      "step": 27033
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.397611618041992,
+      "learning_rate": 6.200329300824568e-06,
+      "loss": 0.4333,
+      "step": 27034
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.410670280456543,
+      "learning_rate": 6.2000733512386004e-06,
+      "loss": 0.7793,
+      "step": 27035
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.7637550830841064,
+      "learning_rate": 6.1998173983156775e-06,
+      "loss": 0.8263,
+      "step": 27036
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.9755685329437256,
+      "learning_rate": 6.199561442056512e-06,
+      "loss": 0.5659,
+      "step": 27037
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.571183919906616,
+      "learning_rate": 6.199305482461817e-06,
+      "loss": 0.4647,
+      "step": 27038
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.0663764476776123,
+      "learning_rate": 6.199049519532301e-06,
+      "loss": 0.4741,
+      "step": 27039
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.863807439804077,
+      "learning_rate": 6.1987935532686785e-06,
+      "loss": 0.817,
+      "step": 27040
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.6733932495117188,
+      "learning_rate": 6.1985375836716595e-06,
+      "loss": 0.5151,
+      "step": 27041
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.68234920501709,
+      "learning_rate": 6.198281610741956e-06,
+      "loss": 0.6725,
+      "step": 27042
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 10.330212593078613,
+      "learning_rate": 6.1980256344802815e-06,
+      "loss": 0.4965,
+      "step": 27043
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.7060800790786743,
+      "learning_rate": 6.197769654887345e-06,
+      "loss": 0.8142,
+      "step": 27044
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.013011932373047,
+      "learning_rate": 6.197513671963862e-06,
+      "loss": 0.3977,
+      "step": 27045
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.6780649423599243,
+      "learning_rate": 6.19725768571054e-06,
+      "loss": 0.7636,
+      "step": 27046
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.2450461387634277,
+      "learning_rate": 6.197001696128095e-06,
+      "loss": 0.5211,
+      "step": 27047
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.2660584449768066,
+      "learning_rate": 6.196745703217235e-06,
+      "loss": 0.794,
+      "step": 27048
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.2512807846069336,
+      "learning_rate": 6.196489706978674e-06,
+      "loss": 0.6424,
+      "step": 27049
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.7265887260437012,
+      "learning_rate": 6.196233707413124e-06,
+      "loss": 0.544,
+      "step": 27050
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 4.159212589263916,
+      "learning_rate": 6.1959777045212955e-06,
+      "loss": 0.6849,
+      "step": 27051
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.6277830600738525,
+      "learning_rate": 6.195721698303899e-06,
+      "loss": 0.7054,
+      "step": 27052
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.8392868041992188,
+      "learning_rate": 6.1954656887616515e-06,
+      "loss": 0.5862,
+      "step": 27053
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.7115917205810547,
+      "learning_rate": 6.1952096758952605e-06,
+      "loss": 0.7577,
+      "step": 27054
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.8987014293670654,
+      "learning_rate": 6.194953659705439e-06,
+      "loss": 0.6956,
+      "step": 27055
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.153120279312134,
+      "learning_rate": 6.194697640192899e-06,
+      "loss": 0.7322,
+      "step": 27056
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.568584680557251,
+      "learning_rate": 6.194441617358353e-06,
+      "loss": 0.6086,
+      "step": 27057
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 5.94217586517334,
+      "learning_rate": 6.194185591202512e-06,
+      "loss": 0.689,
+      "step": 27058
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.4363949298858643,
+      "learning_rate": 6.193929561726088e-06,
+      "loss": 0.7554,
+      "step": 27059
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 4.222168445587158,
+      "learning_rate": 6.193673528929793e-06,
+      "loss": 0.705,
+      "step": 27060
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.2286581993103027,
+      "learning_rate": 6.193417492814339e-06,
+      "loss": 0.6209,
+      "step": 27061
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.4191701412200928,
+      "learning_rate": 6.19316145338044e-06,
+      "loss": 0.662,
+      "step": 27062
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.2360117435455322,
+      "learning_rate": 6.192905410628804e-06,
+      "loss": 0.6133,
+      "step": 27063
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.280700206756592,
+      "learning_rate": 6.1926493645601435e-06,
+      "loss": 0.9053,
+      "step": 27064
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 4.133857727050781,
+      "learning_rate": 6.192393315175175e-06,
+      "loss": 0.4052,
+      "step": 27065
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.696706771850586,
+      "learning_rate": 6.1921372624746055e-06,
+      "loss": 0.7564,
+      "step": 27066
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 39.42601776123047,
+      "learning_rate": 6.191881206459149e-06,
+      "loss": 0.8919,
+      "step": 27067
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.161844491958618,
+      "learning_rate": 6.191625147129519e-06,
+      "loss": 1.0234,
+      "step": 27068
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.820070743560791,
+      "learning_rate": 6.1913690844864246e-06,
+      "loss": 0.37,
+      "step": 27069
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 4.087552547454834,
+      "learning_rate": 6.191113018530579e-06,
+      "loss": 0.8742,
+      "step": 27070
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 5.179637908935547,
+      "learning_rate": 6.1908569492626945e-06,
+      "loss": 0.6214,
+      "step": 27071
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 4.165667533874512,
+      "learning_rate": 6.190600876683483e-06,
+      "loss": 0.5009,
+      "step": 27072
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.492932081222534,
+      "learning_rate": 6.190344800793655e-06,
+      "loss": 0.6686,
+      "step": 27073
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.4033586978912354,
+      "learning_rate": 6.190088721593925e-06,
+      "loss": 0.5289,
+      "step": 27074
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.036160707473755,
+      "learning_rate": 6.189832639085005e-06,
+      "loss": 0.7051,
+      "step": 27075
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.984994411468506,
+      "learning_rate": 6.1895765532676065e-06,
+      "loss": 0.6362,
+      "step": 27076
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.5386736392974854,
+      "learning_rate": 6.1893204641424395e-06,
+      "loss": 0.4427,
+      "step": 27077
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.3324403762817383,
+      "learning_rate": 6.1890643717102195e-06,
+      "loss": 0.4966,
+      "step": 27078
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.984663486480713,
+      "learning_rate": 6.1888082759716565e-06,
+      "loss": 0.9056,
+      "step": 27079
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.01945161819458,
+      "learning_rate": 6.188552176927463e-06,
+      "loss": 0.6975,
+      "step": 27080
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.5485403537750244,
+      "learning_rate": 6.18829607457835e-06,
+      "loss": 0.3085,
+      "step": 27081
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 4.233279705047607,
+      "learning_rate": 6.18803996892503e-06,
+      "loss": 0.6487,
+      "step": 27082
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.9799715280532837,
+      "learning_rate": 6.187783859968218e-06,
+      "loss": 0.8498,
+      "step": 27083
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.310945749282837,
+      "learning_rate": 6.187527747708624e-06,
+      "loss": 0.9146,
+      "step": 27084
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.410968780517578,
+      "learning_rate": 6.18727163214696e-06,
+      "loss": 0.9949,
+      "step": 27085
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.9335196018218994,
+      "learning_rate": 6.187015513283937e-06,
+      "loss": 0.294,
+      "step": 27086
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 4.00086784362793,
+      "learning_rate": 6.18675939112027e-06,
+      "loss": 0.6071,
+      "step": 27087
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.70819091796875,
+      "learning_rate": 6.186503265656669e-06,
+      "loss": 0.5517,
+      "step": 27088
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 4.562689304351807,
+      "learning_rate": 6.186247136893847e-06,
+      "loss": 1.0353,
+      "step": 27089
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.572162389755249,
+      "learning_rate": 6.185991004832517e-06,
+      "loss": 0.5889,
+      "step": 27090
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.2894015312194824,
+      "learning_rate": 6.18573486947339e-06,
+      "loss": 0.6244,
+      "step": 27091
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.7978625297546387,
+      "learning_rate": 6.185478730817177e-06,
+      "loss": 0.5878,
+      "step": 27092
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.5007636547088623,
+      "learning_rate": 6.185222588864593e-06,
+      "loss": 0.7008,
+      "step": 27093
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 4.966455459594727,
+      "learning_rate": 6.184966443616348e-06,
+      "loss": 0.9529,
+      "step": 27094
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.3819539546966553,
+      "learning_rate": 6.1847102950731566e-06,
+      "loss": 0.5119,
+      "step": 27095
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 6.118348121643066,
+      "learning_rate": 6.1844541432357284e-06,
+      "loss": 0.7795,
+      "step": 27096
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.1464762687683105,
+      "learning_rate": 6.184197988104778e-06,
+      "loss": 0.4406,
+      "step": 27097
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.9636638164520264,
+      "learning_rate": 6.183941829681016e-06,
+      "loss": 0.3563,
+      "step": 27098
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.9938454627990723,
+      "learning_rate": 6.1836856679651555e-06,
+      "loss": 0.3926,
+      "step": 27099
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.554727792739868,
+      "learning_rate": 6.183429502957908e-06,
+      "loss": 0.6177,
+      "step": 27100
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.70708966255188,
+      "learning_rate": 6.183173334659987e-06,
+      "loss": 0.5899,
+      "step": 27101
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.5635364055633545,
+      "learning_rate": 6.182917163072106e-06,
+      "loss": 0.7872,
+      "step": 27102
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.8826284408569336,
+      "learning_rate": 6.182660988194971e-06,
+      "loss": 0.5224,
+      "step": 27103
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 4.399783611297607,
+      "learning_rate": 6.182404810029302e-06,
+      "loss": 0.7993,
+      "step": 27104
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.3764867782592773,
+      "learning_rate": 6.182148628575808e-06,
+      "loss": 0.6206,
+      "step": 27105
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.3097784519195557,
+      "learning_rate": 6.1818924438352e-06,
+      "loss": 0.8611,
+      "step": 27106
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.0693657398223877,
+      "learning_rate": 6.181636255808193e-06,
+      "loss": 0.3359,
+      "step": 27107
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.476701259613037,
+      "learning_rate": 6.181380064495499e-06,
+      "loss": 0.5594,
+      "step": 27108
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.6826109886169434,
+      "learning_rate": 6.18112386989783e-06,
+      "loss": 0.7759,
+      "step": 27109
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.4856317043304443,
+      "learning_rate": 6.1808676720158954e-06,
+      "loss": 0.7057,
+      "step": 27110
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.8543777465820312,
+      "learning_rate": 6.180611470850412e-06,
+      "loss": 0.6939,
+      "step": 27111
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.3954012393951416,
+      "learning_rate": 6.180355266402089e-06,
+      "loss": 0.4647,
+      "step": 27112
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 4.1634345054626465,
+      "learning_rate": 6.1800990586716405e-06,
+      "loss": 0.9079,
+      "step": 27113
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.0759952068328857,
+      "learning_rate": 6.179842847659779e-06,
+      "loss": 0.5178,
+      "step": 27114
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.328953742980957,
+      "learning_rate": 6.179586633367217e-06,
+      "loss": 0.3736,
+      "step": 27115
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.0503995418548584,
+      "learning_rate": 6.179330415794666e-06,
+      "loss": 0.7008,
+      "step": 27116
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.549910306930542,
+      "learning_rate": 6.17907419494284e-06,
+      "loss": 0.518,
+      "step": 27117
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.844923496246338,
+      "learning_rate": 6.178817970812449e-06,
+      "loss": 0.7925,
+      "step": 27118
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.229917287826538,
+      "learning_rate": 6.178561743404207e-06,
+      "loss": 0.6568,
+      "step": 27119
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.9811341762542725,
+      "learning_rate": 6.1783055127188255e-06,
+      "loss": 0.5606,
+      "step": 27120
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 5.28331184387207,
+      "learning_rate": 6.178049278757019e-06,
+      "loss": 0.584,
+      "step": 27121
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.2780203819274902,
+      "learning_rate": 6.177793041519499e-06,
+      "loss": 0.5638,
+      "step": 27122
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.003004312515259,
+      "learning_rate": 6.177536801006977e-06,
+      "loss": 0.5574,
+      "step": 27123
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.2859740257263184,
+      "learning_rate": 6.1772805572201665e-06,
+      "loss": 0.519,
+      "step": 27124
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.9875950813293457,
+      "learning_rate": 6.177024310159779e-06,
+      "loss": 0.5731,
+      "step": 27125
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.4349253177642822,
+      "learning_rate": 6.176768059826529e-06,
+      "loss": 0.7446,
+      "step": 27126
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.209357261657715,
+      "learning_rate": 6.176511806221127e-06,
+      "loss": 0.4675,
+      "step": 27127
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.0576488971710205,
+      "learning_rate": 6.176255549344288e-06,
+      "loss": 0.568,
+      "step": 27128
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.4117279052734375,
+      "learning_rate": 6.175999289196721e-06,
+      "loss": 0.5634,
+      "step": 27129
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.8165490627288818,
+      "learning_rate": 6.175743025779142e-06,
+      "loss": 0.5046,
+      "step": 27130
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.2378292083740234,
+      "learning_rate": 6.175486759092262e-06,
+      "loss": 0.5729,
+      "step": 27131
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.1018476486206055,
+      "learning_rate": 6.175230489136792e-06,
+      "loss": 0.6242,
+      "step": 27132
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.721256971359253,
+      "learning_rate": 6.1749742159134474e-06,
+      "loss": 0.7413,
+      "step": 27133
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.013505220413208,
+      "learning_rate": 6.174717939422939e-06,
+      "loss": 0.6303,
+      "step": 27134
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.8435704708099365,
+      "learning_rate": 6.174461659665981e-06,
+      "loss": 0.5789,
+      "step": 27135
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.166139602661133,
+      "learning_rate": 6.174205376643285e-06,
+      "loss": 0.8658,
+      "step": 27136
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.7960543632507324,
+      "learning_rate": 6.1739490903555635e-06,
+      "loss": 0.7432,
+      "step": 27137
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.6864705085754395,
+      "learning_rate": 6.173692800803529e-06,
+      "loss": 0.7571,
+      "step": 27138
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.6109395027160645,
+      "learning_rate": 6.173436507987896e-06,
+      "loss": 0.4255,
+      "step": 27139
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.0621330738067627,
+      "learning_rate": 6.1731802119093745e-06,
+      "loss": 0.7781,
+      "step": 27140
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.4051082134246826,
+      "learning_rate": 6.172923912568678e-06,
+      "loss": 0.3342,
+      "step": 27141
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.9714959859848022,
+      "learning_rate": 6.1726676099665196e-06,
+      "loss": 0.6578,
+      "step": 27142
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 4.216745376586914,
+      "learning_rate": 6.1724113041036115e-06,
+      "loss": 0.7341,
+      "step": 27143
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.3516743183135986,
+      "learning_rate": 6.172154994980667e-06,
+      "loss": 0.9651,
+      "step": 27144
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.858386754989624,
+      "learning_rate": 6.171898682598399e-06,
+      "loss": 0.5135,
+      "step": 27145
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.536818027496338,
+      "learning_rate": 6.17164236695752e-06,
+      "loss": 0.5859,
+      "step": 27146
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.0355453491210938,
+      "learning_rate": 6.171386048058743e-06,
+      "loss": 0.6952,
+      "step": 27147
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.2115087509155273,
+      "learning_rate": 6.171129725902779e-06,
+      "loss": 0.5831,
+      "step": 27148
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.572139263153076,
+      "learning_rate": 6.170873400490343e-06,
+      "loss": 0.7054,
+      "step": 27149
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.2907960414886475,
+      "learning_rate": 6.170617071822145e-06,
+      "loss": 0.6435,
+      "step": 27150
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.36008882522583,
+      "learning_rate": 6.1703607398989e-06,
+      "loss": 0.5973,
+      "step": 27151
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.893864393234253,
+      "learning_rate": 6.170104404721322e-06,
+      "loss": 0.74,
+      "step": 27152
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.4263103008270264,
+      "learning_rate": 6.169848066290121e-06,
+      "loss": 0.4583,
+      "step": 27153
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.9587417840957642,
+      "learning_rate": 6.169591724606009e-06,
+      "loss": 0.8396,
+      "step": 27154
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 4.576152801513672,
+      "learning_rate": 6.169335379669702e-06,
+      "loss": 0.5904,
+      "step": 27155
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.5675950050354004,
+      "learning_rate": 6.1690790314819116e-06,
+      "loss": 0.8879,
+      "step": 27156
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.9671998023986816,
+      "learning_rate": 6.16882268004335e-06,
+      "loss": 0.4595,
+      "step": 27157
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.2141847610473633,
+      "learning_rate": 6.16856632535473e-06,
+      "loss": 0.6482,
+      "step": 27158
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.7225669622421265,
+      "learning_rate": 6.168309967416766e-06,
+      "loss": 0.5231,
+      "step": 27159
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.7212069034576416,
+      "learning_rate": 6.168053606230168e-06,
+      "loss": 0.6615,
+      "step": 27160
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.401520848274231,
+      "learning_rate": 6.167797241795652e-06,
+      "loss": 0.3561,
+      "step": 27161
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.0581023693084717,
+      "learning_rate": 6.167540874113929e-06,
+      "loss": 0.7987,
+      "step": 27162
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.4456329345703125,
+      "learning_rate": 6.167284503185712e-06,
+      "loss": 0.5093,
+      "step": 27163
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.205880880355835,
+      "learning_rate": 6.167028129011713e-06,
+      "loss": 0.4895,
+      "step": 27164
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.7304141521453857,
+      "learning_rate": 6.1667717515926475e-06,
+      "loss": 0.7747,
+      "step": 27165
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.8332929611206055,
+      "learning_rate": 6.1665153709292254e-06,
+      "loss": 0.5854,
+      "step": 27166
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.4404854774475098,
+      "learning_rate": 6.166258987022162e-06,
+      "loss": 0.4345,
+      "step": 27167
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.949561595916748,
+      "learning_rate": 6.166002599872169e-06,
+      "loss": 0.6456,
+      "step": 27168
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.6925396919250488,
+      "learning_rate": 6.1657462094799605e-06,
+      "loss": 0.5158,
+      "step": 27169
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.555727481842041,
+      "learning_rate": 6.165489815846248e-06,
+      "loss": 0.564,
+      "step": 27170
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.0246732234954834,
+      "learning_rate": 6.165233418971745e-06,
+      "loss": 0.5673,
+      "step": 27171
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.110095262527466,
+      "learning_rate": 6.164977018857163e-06,
+      "loss": 0.6038,
+      "step": 27172
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 4.277245044708252,
+      "learning_rate": 6.1647206155032176e-06,
+      "loss": 0.5735,
+      "step": 27173
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.8820269107818604,
+      "learning_rate": 6.16446420891062e-06,
+      "loss": 0.7618,
+      "step": 27174
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 4.550244331359863,
+      "learning_rate": 6.164207799080084e-06,
+      "loss": 0.6998,
+      "step": 27175
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 4.734322547912598,
+      "learning_rate": 6.163951386012322e-06,
+      "loss": 0.5711,
+      "step": 27176
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.455693483352661,
+      "learning_rate": 6.163694969708047e-06,
+      "loss": 0.6308,
+      "step": 27177
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 4.4744553565979,
+      "learning_rate": 6.163438550167973e-06,
+      "loss": 0.635,
+      "step": 27178
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.140754461288452,
+      "learning_rate": 6.163182127392811e-06,
+      "loss": 0.6811,
+      "step": 27179
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 4.154538154602051,
+      "learning_rate": 6.162925701383276e-06,
+      "loss": 0.9506,
+      "step": 27180
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.955064296722412,
+      "learning_rate": 6.16266927214008e-06,
+      "loss": 0.7106,
+      "step": 27181
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.9629695415496826,
+      "learning_rate": 6.162412839663937e-06,
+      "loss": 0.5622,
+      "step": 27182
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 4.333662033081055,
+      "learning_rate": 6.1621564039555595e-06,
+      "loss": 0.8468,
+      "step": 27183
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.264108657836914,
+      "learning_rate": 6.16189996501566e-06,
+      "loss": 0.4457,
+      "step": 27184
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.78438401222229,
+      "learning_rate": 6.1616435228449515e-06,
+      "loss": 0.6904,
+      "step": 27185
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.514169216156006,
+      "learning_rate": 6.161387077444147e-06,
+      "loss": 0.6175,
+      "step": 27186
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 4.264625549316406,
+      "learning_rate": 6.1611306288139604e-06,
+      "loss": 0.4069,
+      "step": 27187
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.030614137649536,
+      "learning_rate": 6.160874176955105e-06,
+      "loss": 0.5186,
+      "step": 27188
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.5464165210723877,
+      "learning_rate": 6.160617721868294e-06,
+      "loss": 0.6369,
+      "step": 27189
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.259572982788086,
+      "learning_rate": 6.160361263554238e-06,
+      "loss": 0.7148,
+      "step": 27190
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.7556606531143188,
+      "learning_rate": 6.160104802013654e-06,
+      "loss": 0.7396,
+      "step": 27191
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.5307884216308594,
+      "learning_rate": 6.1598483372472525e-06,
+      "loss": 0.5988,
+      "step": 27192
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 6.822695255279541,
+      "learning_rate": 6.1595918692557475e-06,
+      "loss": 0.9682,
+      "step": 27193
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.646877408027649,
+      "learning_rate": 6.159335398039851e-06,
+      "loss": 0.446,
+      "step": 27194
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.6950865983963013,
+      "learning_rate": 6.159078923600277e-06,
+      "loss": 0.6924,
+      "step": 27195
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.2663626670837402,
+      "learning_rate": 6.15882244593774e-06,
+      "loss": 0.6919,
+      "step": 27196
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.1146254539489746,
+      "learning_rate": 6.158565965052951e-06,
+      "loss": 0.5529,
+      "step": 27197
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.060030460357666,
+      "learning_rate": 6.158309480946625e-06,
+      "loss": 0.4259,
+      "step": 27198
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.2628438472747803,
+      "learning_rate": 6.158052993619473e-06,
+      "loss": 0.7409,
+      "step": 27199
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.9290995597839355,
+      "learning_rate": 6.157796503072211e-06,
+      "loss": 0.5952,
+      "step": 27200
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.4592294692993164,
+      "learning_rate": 6.157540009305549e-06,
+      "loss": 0.4244,
+      "step": 27201
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.083043098449707,
+      "learning_rate": 6.157283512320202e-06,
+      "loss": 0.7148,
+      "step": 27202
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.548762559890747,
+      "learning_rate": 6.1570270121168844e-06,
+      "loss": 0.7245,
+      "step": 27203
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.4534592628479004,
+      "learning_rate": 6.156770508696306e-06,
+      "loss": 0.8459,
+      "step": 27204
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.031893014907837,
+      "learning_rate": 6.156514002059184e-06,
+      "loss": 0.6914,
+      "step": 27205
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.603907346725464,
+      "learning_rate": 6.156257492206229e-06,
+      "loss": 0.5986,
+      "step": 27206
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 4.752491474151611,
+      "learning_rate": 6.156000979138155e-06,
+      "loss": 0.7618,
+      "step": 27207
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.5215401649475098,
+      "learning_rate": 6.155744462855676e-06,
+      "loss": 0.8117,
+      "step": 27208
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.6223669052124023,
+      "learning_rate": 6.155487943359504e-06,
+      "loss": 0.7599,
+      "step": 27209
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.350876569747925,
+      "learning_rate": 6.155231420650354e-06,
+      "loss": 0.851,
+      "step": 27210
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.026566743850708,
+      "learning_rate": 6.154974894728936e-06,
+      "loss": 0.4967,
+      "step": 27211
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.190372943878174,
+      "learning_rate": 6.1547183655959674e-06,
+      "loss": 0.5759,
+      "step": 27212
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.5404560565948486,
+      "learning_rate": 6.154461833252159e-06,
+      "loss": 0.6154,
+      "step": 27213
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.893740177154541,
+      "learning_rate": 6.154205297698225e-06,
+      "loss": 0.9057,
+      "step": 27214
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 5.336728096008301,
+      "learning_rate": 6.153948758934877e-06,
+      "loss": 0.7335,
+      "step": 27215
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.402540683746338,
+      "learning_rate": 6.1536922169628314e-06,
+      "loss": 0.7199,
+      "step": 27216
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.998449683189392,
+      "learning_rate": 6.1534356717828e-06,
+      "loss": 0.6157,
+      "step": 27217
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.049745798110962,
+      "learning_rate": 6.153179123395495e-06,
+      "loss": 0.5873,
+      "step": 27218
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.5883822441101074,
+      "learning_rate": 6.152922571801632e-06,
+      "loss": 0.2824,
+      "step": 27219
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.1072165966033936,
+      "learning_rate": 6.152666017001921e-06,
+      "loss": 0.5573,
+      "step": 27220
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.7674225568771362,
+      "learning_rate": 6.1524094589970806e-06,
+      "loss": 0.5111,
+      "step": 27221
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.575490951538086,
+      "learning_rate": 6.15215289778782e-06,
+      "loss": 0.8359,
+      "step": 27222
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 4.282196044921875,
+      "learning_rate": 6.151896333374852e-06,
+      "loss": 0.7849,
+      "step": 27223
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.2337965965270996,
+      "learning_rate": 6.151639765758895e-06,
+      "loss": 0.753,
+      "step": 27224
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.637270927429199,
+      "learning_rate": 6.151383194940656e-06,
+      "loss": 0.6531,
+      "step": 27225
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.8747379779815674,
+      "learning_rate": 6.151126620920853e-06,
+      "loss": 0.6402,
+      "step": 27226
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.5105175971984863,
+      "learning_rate": 6.150870043700199e-06,
+      "loss": 0.7405,
+      "step": 27227
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.2446460723876953,
+      "learning_rate": 6.150613463279406e-06,
+      "loss": 0.7436,
+      "step": 27228
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.8634402751922607,
+      "learning_rate": 6.150356879659187e-06,
+      "loss": 0.7546,
+      "step": 27229
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.894036054611206,
+      "learning_rate": 6.150100292840259e-06,
+      "loss": 0.5041,
+      "step": 27230
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.4506781101226807,
+      "learning_rate": 6.149843702823331e-06,
+      "loss": 0.3975,
+      "step": 27231
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 4.4410481452941895,
+      "learning_rate": 6.1495871096091184e-06,
+      "loss": 0.6212,
+      "step": 27232
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.9400296211242676,
+      "learning_rate": 6.149330513198335e-06,
+      "loss": 0.7268,
+      "step": 27233
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.4086270332336426,
+      "learning_rate": 6.149073913591695e-06,
+      "loss": 0.8132,
+      "step": 27234
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.1167383193969727,
+      "learning_rate": 6.148817310789907e-06,
+      "loss": 0.5628,
+      "step": 27235
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.073345184326172,
+      "learning_rate": 6.148560704793692e-06,
+      "loss": 0.6047,
+      "step": 27236
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.2902469635009766,
+      "learning_rate": 6.14830409560376e-06,
+      "loss": 0.6482,
+      "step": 27237
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.085451602935791,
+      "learning_rate": 6.148047483220822e-06,
+      "loss": 0.651,
+      "step": 27238
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.0673325061798096,
+      "learning_rate": 6.1477908676455965e-06,
+      "loss": 0.632,
+      "step": 27239
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.5115160942077637,
+      "learning_rate": 6.147534248878795e-06,
+      "loss": 0.7159,
+      "step": 27240
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.401754856109619,
+      "learning_rate": 6.147277626921128e-06,
+      "loss": 0.7561,
+      "step": 27241
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.1277215480804443,
+      "learning_rate": 6.147021001773313e-06,
+      "loss": 0.7296,
+      "step": 27242
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.708531618118286,
+      "learning_rate": 6.146764373436062e-06,
+      "loss": 0.5832,
+      "step": 27243
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.210418939590454,
+      "learning_rate": 6.146507741910089e-06,
+      "loss": 0.9056,
+      "step": 27244
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.0244245529174805,
+      "learning_rate": 6.146251107196108e-06,
+      "loss": 0.5501,
+      "step": 27245
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.264620304107666,
+      "learning_rate": 6.145994469294833e-06,
+      "loss": 0.7655,
+      "step": 27246
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.03646183013916,
+      "learning_rate": 6.145737828206975e-06,
+      "loss": 0.7579,
+      "step": 27247
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 4.499457836151123,
+      "learning_rate": 6.14548118393325e-06,
+      "loss": 0.7207,
+      "step": 27248
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.164389133453369,
+      "learning_rate": 6.145224536474369e-06,
+      "loss": 0.4502,
+      "step": 27249
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.6918935775756836,
+      "learning_rate": 6.144967885831049e-06,
+      "loss": 0.6469,
+      "step": 27250
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.3753604888916016,
+      "learning_rate": 6.144711232004003e-06,
+      "loss": 0.6399,
+      "step": 27251
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.5184154510498047,
+      "learning_rate": 6.144454574993943e-06,
+      "loss": 0.7788,
+      "step": 27252
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.9947636127471924,
+      "learning_rate": 6.144197914801584e-06,
+      "loss": 0.5587,
+      "step": 27253
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.8984007835388184,
+      "learning_rate": 6.143941251427639e-06,
+      "loss": 0.5234,
+      "step": 27254
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 7.381682395935059,
+      "learning_rate": 6.143684584872822e-06,
+      "loss": 0.6894,
+      "step": 27255
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.486138105392456,
+      "learning_rate": 6.143427915137846e-06,
+      "loss": 0.4278,
+      "step": 27256
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 11.771151542663574,
+      "learning_rate": 6.1431712422234265e-06,
+      "loss": 0.502,
+      "step": 27257
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.130720853805542,
+      "learning_rate": 6.142914566130276e-06,
+      "loss": 0.5125,
+      "step": 27258
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.333374261856079,
+      "learning_rate": 6.142657886859107e-06,
+      "loss": 0.533,
+      "step": 27259
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.359639883041382,
+      "learning_rate": 6.142401204410635e-06,
+      "loss": 0.8874,
+      "step": 27260
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.2023842334747314,
+      "learning_rate": 6.142144518785575e-06,
+      "loss": 0.713,
+      "step": 27261
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.971223831176758,
+      "learning_rate": 6.141887829984636e-06,
+      "loss": 0.7014,
+      "step": 27262
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.0798933506011963,
+      "learning_rate": 6.141631138008537e-06,
+      "loss": 0.5272,
+      "step": 27263
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.072134017944336,
+      "learning_rate": 6.141374442857988e-06,
+      "loss": 0.7417,
+      "step": 27264
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.890488862991333,
+      "learning_rate": 6.141117744533703e-06,
+      "loss": 0.6058,
+      "step": 27265
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 6.02951717376709,
+      "learning_rate": 6.140861043036399e-06,
+      "loss": 0.4043,
+      "step": 27266
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.3194184303283691,
+      "learning_rate": 6.140604338366788e-06,
+      "loss": 0.593,
+      "step": 27267
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.9826884269714355,
+      "learning_rate": 6.140347630525584e-06,
+      "loss": 0.7703,
+      "step": 27268
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.3269524574279785,
+      "learning_rate": 6.1400909195135e-06,
+      "loss": 0.9623,
+      "step": 27269
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 6.520365238189697,
+      "learning_rate": 6.139834205331249e-06,
+      "loss": 0.6314,
+      "step": 27270
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.781820297241211,
+      "learning_rate": 6.139577487979547e-06,
+      "loss": 0.378,
+      "step": 27271
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 5.242080211639404,
+      "learning_rate": 6.139320767459107e-06,
+      "loss": 0.8089,
+      "step": 27272
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 4.938631534576416,
+      "learning_rate": 6.139064043770643e-06,
+      "loss": 0.6723,
+      "step": 27273
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.29319167137146,
+      "learning_rate": 6.138807316914868e-06,
+      "loss": 0.7996,
+      "step": 27274
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.2914440631866455,
+      "learning_rate": 6.138550586892497e-06,
+      "loss": 0.6583,
+      "step": 27275
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.85313081741333,
+      "learning_rate": 6.138293853704242e-06,
+      "loss": 0.3975,
+      "step": 27276
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 5.692899703979492,
+      "learning_rate": 6.138037117350819e-06,
+      "loss": 0.4879,
+      "step": 27277
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.1609559059143066,
+      "learning_rate": 6.137780377832941e-06,
+      "loss": 0.6374,
+      "step": 27278
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.156456708908081,
+      "learning_rate": 6.137523635151322e-06,
+      "loss": 0.3398,
+      "step": 27279
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.872265338897705,
+      "learning_rate": 6.137266889306677e-06,
+      "loss": 0.8461,
+      "step": 27280
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.9496026039123535,
+      "learning_rate": 6.137010140299717e-06,
+      "loss": 0.6259,
+      "step": 27281
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 4.343191623687744,
+      "learning_rate": 6.136753388131159e-06,
+      "loss": 0.4925,
+      "step": 27282
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.9110655784606934,
+      "learning_rate": 6.136496632801716e-06,
+      "loss": 0.6727,
+      "step": 27283
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.399683952331543,
+      "learning_rate": 6.136239874312101e-06,
+      "loss": 0.6994,
+      "step": 27284
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.455514430999756,
+      "learning_rate": 6.1359831126630285e-06,
+      "loss": 0.6748,
+      "step": 27285
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.1588845252990723,
+      "learning_rate": 6.135726347855212e-06,
+      "loss": 0.454,
+      "step": 27286
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.438969373703003,
+      "learning_rate": 6.135469579889366e-06,
+      "loss": 0.5784,
+      "step": 27287
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.7123632431030273,
+      "learning_rate": 6.135212808766206e-06,
+      "loss": 0.7328,
+      "step": 27288
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.0328245162963867,
+      "learning_rate": 6.1349560344864435e-06,
+      "loss": 0.5834,
+      "step": 27289
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.242514133453369,
+      "learning_rate": 6.134699257050792e-06,
+      "loss": 0.7041,
+      "step": 27290
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.7187955379486084,
+      "learning_rate": 6.13444247645997e-06,
+      "loss": 0.7217,
+      "step": 27291
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.5481295585632324,
+      "learning_rate": 6.134185692714686e-06,
+      "loss": 0.4734,
+      "step": 27292
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.7709057331085205,
+      "learning_rate": 6.133928905815657e-06,
+      "loss": 0.7791,
+      "step": 27293
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 4.836090564727783,
+      "learning_rate": 6.133672115763597e-06,
+      "loss": 0.6702,
+      "step": 27294
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 4.064436435699463,
+      "learning_rate": 6.1334153225592195e-06,
+      "loss": 0.5326,
+      "step": 27295
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.9121060371398926,
+      "learning_rate": 6.133158526203237e-06,
+      "loss": 0.6372,
+      "step": 27296
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.752798080444336,
+      "learning_rate": 6.132901726696366e-06,
+      "loss": 0.5747,
+      "step": 27297
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.030184268951416,
+      "learning_rate": 6.13264492403932e-06,
+      "loss": 0.6569,
+      "step": 27298
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 4.43228816986084,
+      "learning_rate": 6.132388118232814e-06,
+      "loss": 0.8222,
+      "step": 27299
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.3141188621520996,
+      "learning_rate": 6.132131309277559e-06,
+      "loss": 0.5937,
+      "step": 27300
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.5402350425720215,
+      "learning_rate": 6.131874497174272e-06,
+      "loss": 0.7329,
+      "step": 27301
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.9300830364227295,
+      "learning_rate": 6.1316176819236665e-06,
+      "loss": 0.5762,
+      "step": 27302
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.6529037952423096,
+      "learning_rate": 6.131360863526454e-06,
+      "loss": 0.672,
+      "step": 27303
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 5.855959415435791,
+      "learning_rate": 6.131104041983354e-06,
+      "loss": 0.6819,
+      "step": 27304
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.474714517593384,
+      "learning_rate": 6.130847217295073e-06,
+      "loss": 0.6584,
+      "step": 27305
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.492868185043335,
+      "learning_rate": 6.130590389462332e-06,
+      "loss": 0.7551,
+      "step": 27306
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.7490469217300415,
+      "learning_rate": 6.130333558485844e-06,
+      "loss": 0.4549,
+      "step": 27307
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.7531118392944336,
+      "learning_rate": 6.1300767243663195e-06,
+      "loss": 0.844,
+      "step": 27308
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 7.183650493621826,
+      "learning_rate": 6.1298198871044756e-06,
+      "loss": 0.5112,
+      "step": 27309
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 4.451757907867432,
+      "learning_rate": 6.129563046701026e-06,
+      "loss": 0.8707,
+      "step": 27310
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.3917081356048584,
+      "learning_rate": 6.129306203156684e-06,
+      "loss": 0.6038,
+      "step": 27311
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 4.05960750579834,
+      "learning_rate": 6.129049356472165e-06,
+      "loss": 0.7185,
+      "step": 27312
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.846074342727661,
+      "learning_rate": 6.1287925066481846e-06,
+      "loss": 0.7284,
+      "step": 27313
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.252408742904663,
+      "learning_rate": 6.128535653685453e-06,
+      "loss": 0.5631,
+      "step": 27314
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.9922828674316406,
+      "learning_rate": 6.128278797584687e-06,
+      "loss": 0.1943,
+      "step": 27315
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.156388521194458,
+      "learning_rate": 6.1280219383466e-06,
+      "loss": 0.5448,
+      "step": 27316
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.7577595710754395,
+      "learning_rate": 6.127765075971906e-06,
+      "loss": 0.5976,
+      "step": 27317
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.0801174640655518,
+      "learning_rate": 6.127508210461321e-06,
+      "loss": 0.4814,
+      "step": 27318
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.1348330974578857,
+      "learning_rate": 6.127251341815557e-06,
+      "loss": 0.953,
+      "step": 27319
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.8241987228393555,
+      "learning_rate": 6.126994470035331e-06,
+      "loss": 0.6752,
+      "step": 27320
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.6846084594726562,
+      "learning_rate": 6.126737595121355e-06,
+      "loss": 0.5361,
+      "step": 27321
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.579923629760742,
+      "learning_rate": 6.126480717074343e-06,
+      "loss": 0.6077,
+      "step": 27322
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.6429550647735596,
+      "learning_rate": 6.126223835895011e-06,
+      "loss": 0.8191,
+      "step": 27323
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 4.12472677230835,
+      "learning_rate": 6.125966951584073e-06,
+      "loss": 0.5548,
+      "step": 27324
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.416679859161377,
+      "learning_rate": 6.12571006414224e-06,
+      "loss": 0.3891,
+      "step": 27325
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.618419885635376,
+      "learning_rate": 6.1254531735702305e-06,
+      "loss": 0.6968,
+      "step": 27326
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.6086101531982422,
+      "learning_rate": 6.1251962798687585e-06,
+      "loss": 0.4726,
+      "step": 27327
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.3413162231445312,
+      "learning_rate": 6.124939383038535e-06,
+      "loss": 0.5088,
+      "step": 27328
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.1168570518493652,
+      "learning_rate": 6.124682483080278e-06,
+      "loss": 0.4568,
+      "step": 27329
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 6.284075736999512,
+      "learning_rate": 6.1244255799947e-06,
+      "loss": 0.6229,
+      "step": 27330
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.6345467567443848,
+      "learning_rate": 6.124168673782516e-06,
+      "loss": 0.7605,
+      "step": 27331
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 12.074620246887207,
+      "learning_rate": 6.12391176444444e-06,
+      "loss": 0.5204,
+      "step": 27332
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.8406636714935303,
+      "learning_rate": 6.123654851981185e-06,
+      "loss": 0.8088,
+      "step": 27333
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.8801522254943848,
+      "learning_rate": 6.1233979363934695e-06,
+      "loss": 0.5665,
+      "step": 27334
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 5.3301801681518555,
+      "learning_rate": 6.123141017682003e-06,
+      "loss": 0.6579,
+      "step": 27335
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.8039610385894775,
+      "learning_rate": 6.122884095847503e-06,
+      "loss": 0.5786,
+      "step": 27336
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.76138973236084,
+      "learning_rate": 6.122627170890681e-06,
+      "loss": 0.6129,
+      "step": 27337
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.444814443588257,
+      "learning_rate": 6.122370242812256e-06,
+      "loss": 0.7918,
+      "step": 27338
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.4723474979400635,
+      "learning_rate": 6.122113311612938e-06,
+      "loss": 0.5177,
+      "step": 27339
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.5016034841537476,
+      "learning_rate": 6.121856377293445e-06,
+      "loss": 0.5457,
+      "step": 27340
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.1425118446350098,
+      "learning_rate": 6.121599439854489e-06,
+      "loss": 0.9031,
+      "step": 27341
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.110614061355591,
+      "learning_rate": 6.121342499296784e-06,
+      "loss": 0.5094,
+      "step": 27342
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.8198468685150146,
+      "learning_rate": 6.121085555621047e-06,
+      "loss": 0.6484,
+      "step": 27343
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.3722240924835205,
+      "learning_rate": 6.12082860882799e-06,
+      "loss": 0.4006,
+      "step": 27344
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 4.881547451019287,
+      "learning_rate": 6.120571658918329e-06,
+      "loss": 0.4581,
+      "step": 27345
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 4.269969463348389,
+      "learning_rate": 6.120314705892778e-06,
+      "loss": 0.7184,
+      "step": 27346
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.297299861907959,
+      "learning_rate": 6.120057749752051e-06,
+      "loss": 0.8356,
+      "step": 27347
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.713103175163269,
+      "learning_rate": 6.119800790496863e-06,
+      "loss": 0.5712,
+      "step": 27348
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.6598551273345947,
+      "learning_rate": 6.11954382812793e-06,
+      "loss": 0.4124,
+      "step": 27349
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 5.455626964569092,
+      "learning_rate": 6.119286862645963e-06,
+      "loss": 0.7329,
+      "step": 27350
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.106599807739258,
+      "learning_rate": 6.119029894051679e-06,
+      "loss": 0.6444,
+      "step": 27351
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.9894392490386963,
+      "learning_rate": 6.118772922345792e-06,
+      "loss": 0.5158,
+      "step": 27352
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.00136399269104,
+      "learning_rate": 6.118515947529018e-06,
+      "loss": 0.658,
+      "step": 27353
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.625713586807251,
+      "learning_rate": 6.11825896960207e-06,
+      "loss": 0.4608,
+      "step": 27354
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.9690074920654297,
+      "learning_rate": 6.118001988565662e-06,
+      "loss": 0.8492,
+      "step": 27355
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.1575145721435547,
+      "learning_rate": 6.117745004420509e-06,
+      "loss": 0.6895,
+      "step": 27356
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.6269292831420898,
+      "learning_rate": 6.117488017167325e-06,
+      "loss": 0.969,
+      "step": 27357
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.7249767780303955,
+      "learning_rate": 6.117231026806827e-06,
+      "loss": 0.4682,
+      "step": 27358
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.9284117221832275,
+      "learning_rate": 6.116974033339727e-06,
+      "loss": 0.6836,
+      "step": 27359
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.8814058303833008,
+      "learning_rate": 6.116717036766742e-06,
+      "loss": 0.5928,
+      "step": 27360
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.5316879749298096,
+      "learning_rate": 6.116460037088584e-06,
+      "loss": 0.7115,
+      "step": 27361
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 4.999596118927002,
+      "learning_rate": 6.11620303430597e-06,
+      "loss": 0.67,
+      "step": 27362
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.8811917304992676,
+      "learning_rate": 6.115946028419612e-06,
+      "loss": 0.5424,
+      "step": 27363
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.6738369464874268,
+      "learning_rate": 6.115689019430227e-06,
+      "loss": 0.7456,
+      "step": 27364
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.454254150390625,
+      "learning_rate": 6.11543200733853e-06,
+      "loss": 0.4904,
+      "step": 27365
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 9.551643371582031,
+      "learning_rate": 6.115174992145232e-06,
+      "loss": 0.6249,
+      "step": 27366
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.1914384365081787,
+      "learning_rate": 6.114917973851051e-06,
+      "loss": 0.7903,
+      "step": 27367
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 5.909374713897705,
+      "learning_rate": 6.1146609524567014e-06,
+      "loss": 0.8111,
+      "step": 27368
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.283369541168213,
+      "learning_rate": 6.1144039279628975e-06,
+      "loss": 0.587,
+      "step": 27369
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 4.149546146392822,
+      "learning_rate": 6.114146900370352e-06,
+      "loss": 0.6194,
+      "step": 27370
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.3105099201202393,
+      "learning_rate": 6.113889869679783e-06,
+      "loss": 0.4729,
+      "step": 27371
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 4.663582801818848,
+      "learning_rate": 6.113632835891903e-06,
+      "loss": 0.7017,
+      "step": 27372
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.1208934783935547,
+      "learning_rate": 6.113375799007427e-06,
+      "loss": 0.5861,
+      "step": 27373
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.8491675853729248,
+      "learning_rate": 6.11311875902707e-06,
+      "loss": 0.5533,
+      "step": 27374
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 4.198029041290283,
+      "learning_rate": 6.112861715951547e-06,
+      "loss": 0.7378,
+      "step": 27375
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.4164419174194336,
+      "learning_rate": 6.112604669781572e-06,
+      "loss": 0.6969,
+      "step": 27376
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.807065725326538,
+      "learning_rate": 6.112347620517861e-06,
+      "loss": 0.713,
+      "step": 27377
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.744570255279541,
+      "learning_rate": 6.112090568161128e-06,
+      "loss": 0.6952,
+      "step": 27378
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.477302312850952,
+      "learning_rate": 6.111833512712086e-06,
+      "loss": 0.7261,
+      "step": 27379
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.147634983062744,
+      "learning_rate": 6.111576454171454e-06,
+      "loss": 0.4186,
+      "step": 27380
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.0462825298309326,
+      "learning_rate": 6.111319392539943e-06,
+      "loss": 0.6385,
+      "step": 27381
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.408961057662964,
+      "learning_rate": 6.111062327818269e-06,
+      "loss": 0.8927,
+      "step": 27382
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.1334829330444336,
+      "learning_rate": 6.110805260007147e-06,
+      "loss": 0.7865,
+      "step": 27383
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.1358542442321777,
+      "learning_rate": 6.110548189107292e-06,
+      "loss": 0.8073,
+      "step": 27384
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.44303822517395,
+      "learning_rate": 6.110291115119419e-06,
+      "loss": 0.5613,
+      "step": 27385
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.9439144134521484,
+      "learning_rate": 6.110034038044241e-06,
+      "loss": 0.5391,
+      "step": 27386
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.5208988189697266,
+      "learning_rate": 6.109776957882474e-06,
+      "loss": 0.7217,
+      "step": 27387
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 4.265333652496338,
+      "learning_rate": 6.109519874634834e-06,
+      "loss": 0.651,
+      "step": 27388
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.6766581535339355,
+      "learning_rate": 6.109262788302035e-06,
+      "loss": 0.7426,
+      "step": 27389
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.1707520484924316,
+      "learning_rate": 6.1090056988847916e-06,
+      "loss": 0.7055,
+      "step": 27390
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.812446355819702,
+      "learning_rate": 6.1087486063838186e-06,
+      "loss": 0.759,
+      "step": 27391
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.3377723693847656,
+      "learning_rate": 6.108491510799831e-06,
+      "loss": 0.952,
+      "step": 27392
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 5.496927738189697,
+      "learning_rate": 6.108234412133545e-06,
+      "loss": 0.3861,
+      "step": 27393
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.488827109336853,
+      "learning_rate": 6.107977310385672e-06,
+      "loss": 0.5521,
+      "step": 27394
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.7997612953186035,
+      "learning_rate": 6.10772020555693e-06,
+      "loss": 0.6636,
+      "step": 27395
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.435770034790039,
+      "learning_rate": 6.107463097648034e-06,
+      "loss": 1.0628,
+      "step": 27396
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.1056151390075684,
+      "learning_rate": 6.107205986659698e-06,
+      "loss": 0.8262,
+      "step": 27397
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.9797009229660034,
+      "learning_rate": 6.106948872592635e-06,
+      "loss": 0.4397,
+      "step": 27398
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.388976812362671,
+      "learning_rate": 6.106691755447562e-06,
+      "loss": 0.7351,
+      "step": 27399
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.880079746246338,
+      "learning_rate": 6.106434635225196e-06,
+      "loss": 0.3432,
+      "step": 27400
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 5.574230670928955,
+      "learning_rate": 6.106177511926249e-06,
+      "loss": 0.4594,
+      "step": 27401
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.3331007957458496,
+      "learning_rate": 6.105920385551436e-06,
+      "loss": 0.679,
+      "step": 27402
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.378513813018799,
+      "learning_rate": 6.105663256101473e-06,
+      "loss": 0.7811,
+      "step": 27403
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.154289722442627,
+      "learning_rate": 6.1054061235770745e-06,
+      "loss": 0.7451,
+      "step": 27404
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.0933125019073486,
+      "learning_rate": 6.105148987978956e-06,
+      "loss": 0.6253,
+      "step": 27405
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.9290993213653564,
+      "learning_rate": 6.104891849307832e-06,
+      "loss": 0.5795,
+      "step": 27406
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.1557114124298096,
+      "learning_rate": 6.1046347075644185e-06,
+      "loss": 0.5157,
+      "step": 27407
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.43088698387146,
+      "learning_rate": 6.1043775627494286e-06,
+      "loss": 0.6046,
+      "step": 27408
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.298557996749878,
+      "learning_rate": 6.104120414863578e-06,
+      "loss": 0.294,
+      "step": 27409
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.7611911296844482,
+      "learning_rate": 6.1038632639075835e-06,
+      "loss": 0.6088,
+      "step": 27410
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.3017001152038574,
+      "learning_rate": 6.103606109882157e-06,
+      "loss": 0.8529,
+      "step": 27411
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.9621050357818604,
+      "learning_rate": 6.103348952788018e-06,
+      "loss": 0.5724,
+      "step": 27412
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.0006465911865234,
+      "learning_rate": 6.103091792625877e-06,
+      "loss": 0.7333,
+      "step": 27413
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.1828770637512207,
+      "learning_rate": 6.102834629396452e-06,
+      "loss": 0.7117,
+      "step": 27414
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.97019624710083,
+      "learning_rate": 6.1025774631004555e-06,
+      "loss": 0.8472,
+      "step": 27415
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.5945864915847778,
+      "learning_rate": 6.102320293738605e-06,
+      "loss": 0.7293,
+      "step": 27416
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.407949686050415,
+      "learning_rate": 6.102063121311616e-06,
+      "loss": 0.6686,
+      "step": 27417
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.0878474712371826,
+      "learning_rate": 6.101805945820199e-06,
+      "loss": 0.7846,
+      "step": 27418
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.968156099319458,
+      "learning_rate": 6.101548767265074e-06,
+      "loss": 0.5749,
+      "step": 27419
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.8185291290283203,
+      "learning_rate": 6.101291585646956e-06,
+      "loss": 0.5874,
+      "step": 27420
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.102013349533081,
+      "learning_rate": 6.101034400966557e-06,
+      "loss": 0.7817,
+      "step": 27421
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.6121188402175903,
+      "learning_rate": 6.1007772132245945e-06,
+      "loss": 0.7018,
+      "step": 27422
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.7466375827789307,
+      "learning_rate": 6.100520022421783e-06,
+      "loss": 0.5256,
+      "step": 27423
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.265078067779541,
+      "learning_rate": 6.100262828558837e-06,
+      "loss": 0.9834,
+      "step": 27424
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.5462660789489746,
+      "learning_rate": 6.100005631636473e-06,
+      "loss": 0.7163,
+      "step": 27425
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.5005414485931396,
+      "learning_rate": 6.0997484316554056e-06,
+      "loss": 0.6501,
+      "step": 27426
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 4.07559061050415,
+      "learning_rate": 6.099491228616349e-06,
+      "loss": 0.6739,
+      "step": 27427
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 18.156503677368164,
+      "learning_rate": 6.099234022520019e-06,
+      "loss": 0.6813,
+      "step": 27428
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 7.773051738739014,
+      "learning_rate": 6.09897681336713e-06,
+      "loss": 0.3627,
+      "step": 27429
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.4660648107528687,
+      "learning_rate": 6.0987196011584e-06,
+      "loss": 0.453,
+      "step": 27430
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.056232452392578,
+      "learning_rate": 6.0984623858945426e-06,
+      "loss": 0.673,
+      "step": 27431
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.869097948074341,
+      "learning_rate": 6.098205167576272e-06,
+      "loss": 0.5469,
+      "step": 27432
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 6.591504096984863,
+      "learning_rate": 6.097947946204305e-06,
+      "loss": 0.7564,
+      "step": 27433
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.143629789352417,
+      "learning_rate": 6.097690721779353e-06,
+      "loss": 0.7611,
+      "step": 27434
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.930365800857544,
+      "learning_rate": 6.0974334943021375e-06,
+      "loss": 0.8717,
+      "step": 27435
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.828916549682617,
+      "learning_rate": 6.09717626377337e-06,
+      "loss": 1.0196,
+      "step": 27436
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.6289050579071045,
+      "learning_rate": 6.096919030193765e-06,
+      "loss": 0.4506,
+      "step": 27437
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.336111307144165,
+      "learning_rate": 6.09666179356404e-06,
+      "loss": 0.5589,
+      "step": 27438
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.520852565765381,
+      "learning_rate": 6.09640455388491e-06,
+      "loss": 0.9463,
+      "step": 27439
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.9199355840682983,
+      "learning_rate": 6.096147311157089e-06,
+      "loss": 0.7071,
+      "step": 27440
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.489987373352051,
+      "learning_rate": 6.095890065381292e-06,
+      "loss": 0.3553,
+      "step": 27441
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.989098072052002,
+      "learning_rate": 6.095632816558236e-06,
+      "loss": 0.6552,
+      "step": 27442
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.8867123126983643,
+      "learning_rate": 6.095375564688637e-06,
+      "loss": 0.7861,
+      "step": 27443
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 4.753630638122559,
+      "learning_rate": 6.095118309773208e-06,
+      "loss": 0.8759,
+      "step": 27444
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.7796587944030762,
+      "learning_rate": 6.094861051812666e-06,
+      "loss": 0.4259,
+      "step": 27445
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.9735857248306274,
+      "learning_rate": 6.094603790807725e-06,
+      "loss": 0.6682,
+      "step": 27446
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.4131019115448,
+      "learning_rate": 6.0943465267591e-06,
+      "loss": 0.852,
+      "step": 27447
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.573481321334839,
+      "learning_rate": 6.094089259667508e-06,
+      "loss": 0.54,
+      "step": 27448
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.557671308517456,
+      "learning_rate": 6.093831989533663e-06,
+      "loss": 0.7537,
+      "step": 27449
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.4198837280273438,
+      "learning_rate": 6.09357471635828e-06,
+      "loss": 0.7199,
+      "step": 27450
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.3383963108062744,
+      "learning_rate": 6.093317440142078e-06,
+      "loss": 0.7133,
+      "step": 27451
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.0875115394592285,
+      "learning_rate": 6.093060160885769e-06,
+      "loss": 0.9421,
+      "step": 27452
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.3482778072357178,
+      "learning_rate": 6.0928028785900686e-06,
+      "loss": 0.7057,
+      "step": 27453
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.0003976821899414,
+      "learning_rate": 6.092545593255692e-06,
+      "loss": 0.8043,
+      "step": 27454
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.5834505558013916,
+      "learning_rate": 6.092288304883357e-06,
+      "loss": 0.516,
+      "step": 27455
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.0737526416778564,
+      "learning_rate": 6.092031013473777e-06,
+      "loss": 0.5782,
+      "step": 27456
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.7767415046691895,
+      "learning_rate": 6.091773719027667e-06,
+      "loss": 0.4412,
+      "step": 27457
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.504284143447876,
+      "learning_rate": 6.091516421545744e-06,
+      "loss": 0.6226,
+      "step": 27458
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.8072059154510498,
+      "learning_rate": 6.091259121028722e-06,
+      "loss": 0.7529,
+      "step": 27459
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.8244367837905884,
+      "learning_rate": 6.091001817477317e-06,
+      "loss": 0.606,
+      "step": 27460
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.1462109088897705,
+      "learning_rate": 6.090744510892246e-06,
+      "loss": 0.5969,
+      "step": 27461
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.8522181510925293,
+      "learning_rate": 6.090487201274222e-06,
+      "loss": 0.6555,
+      "step": 27462
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.0818140506744385,
+      "learning_rate": 6.090229888623962e-06,
+      "loss": 0.7113,
+      "step": 27463
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.0680394172668457,
+      "learning_rate": 6.089972572942182e-06,
+      "loss": 0.8722,
+      "step": 27464
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.810490369796753,
+      "learning_rate": 6.089715254229596e-06,
+      "loss": 0.8009,
+      "step": 27465
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.00154185295105,
+      "learning_rate": 6.08945793248692e-06,
+      "loss": 0.5819,
+      "step": 27466
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.8144729137420654,
+      "learning_rate": 6.089200607714869e-06,
+      "loss": 0.5633,
+      "step": 27467
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.122270345687866,
+      "learning_rate": 6.088943279914159e-06,
+      "loss": 0.9294,
+      "step": 27468
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.992814779281616,
+      "learning_rate": 6.088685949085506e-06,
+      "loss": 0.5007,
+      "step": 27469
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.0665767192840576,
+      "learning_rate": 6.088428615229626e-06,
+      "loss": 0.8366,
+      "step": 27470
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.0798676013946533,
+      "learning_rate": 6.0881712783472325e-06,
+      "loss": 0.8412,
+      "step": 27471
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.6802845001220703,
+      "learning_rate": 6.087913938439043e-06,
+      "loss": 0.7843,
+      "step": 27472
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.069220542907715,
+      "learning_rate": 6.087656595505773e-06,
+      "loss": 0.4997,
+      "step": 27473
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 6.635715007781982,
+      "learning_rate": 6.087399249548137e-06,
+      "loss": 0.7055,
+      "step": 27474
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.6764113903045654,
+      "learning_rate": 6.087141900566851e-06,
+      "loss": 0.7089,
+      "step": 27475
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.4824094772338867,
+      "learning_rate": 6.08688454856263e-06,
+      "loss": 0.5736,
+      "step": 27476
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.208523988723755,
+      "learning_rate": 6.086627193536191e-06,
+      "loss": 0.479,
+      "step": 27477
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.9557386636734009,
+      "learning_rate": 6.086369835488248e-06,
+      "loss": 0.6782,
+      "step": 27478
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.013312578201294,
+      "learning_rate": 6.086112474419518e-06,
+      "loss": 0.5039,
+      "step": 27479
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.2585697174072266,
+      "learning_rate": 6.085855110330716e-06,
+      "loss": 0.7182,
+      "step": 27480
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 4.110748291015625,
+      "learning_rate": 6.085597743222558e-06,
+      "loss": 0.5198,
+      "step": 27481
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.623929023742676,
+      "learning_rate": 6.085340373095759e-06,
+      "loss": 0.6225,
+      "step": 27482
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.315192222595215,
+      "learning_rate": 6.085082999951035e-06,
+      "loss": 0.4772,
+      "step": 27483
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.4674663543701172,
+      "learning_rate": 6.084825623789103e-06,
+      "loss": 0.693,
+      "step": 27484
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.1304523944854736,
+      "learning_rate": 6.084568244610676e-06,
+      "loss": 0.6119,
+      "step": 27485
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.9718035459518433,
+      "learning_rate": 6.08431086241647e-06,
+      "loss": 0.5937,
+      "step": 27486
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.2218477725982666,
+      "learning_rate": 6.084053477207203e-06,
+      "loss": 0.6141,
+      "step": 27487
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.556826591491699,
+      "learning_rate": 6.08379608898359e-06,
+      "loss": 0.704,
+      "step": 27488
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.3991650342941284,
+      "learning_rate": 6.083538697746344e-06,
+      "loss": 0.8251,
+      "step": 27489
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.1490824222564697,
+      "learning_rate": 6.083281303496185e-06,
+      "loss": 0.6051,
+      "step": 27490
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.473093032836914,
+      "learning_rate": 6.0830239062338244e-06,
+      "loss": 0.8253,
+      "step": 27491
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.6800116300582886,
+      "learning_rate": 6.082766505959982e-06,
+      "loss": 0.7348,
+      "step": 27492
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.564713954925537,
+      "learning_rate": 6.08250910267537e-06,
+      "loss": 0.548,
+      "step": 27493
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.3235177993774414,
+      "learning_rate": 6.082251696380708e-06,
+      "loss": 0.4647,
+      "step": 27494
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.3152592182159424,
+      "learning_rate": 6.081994287076707e-06,
+      "loss": 0.7722,
+      "step": 27495
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.0614418983459473,
+      "learning_rate": 6.081736874764085e-06,
+      "loss": 0.5841,
+      "step": 27496
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.9119014739990234,
+      "learning_rate": 6.081479459443559e-06,
+      "loss": 0.3317,
+      "step": 27497
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.175065279006958,
+      "learning_rate": 6.081222041115844e-06,
+      "loss": 0.6706,
+      "step": 27498
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.4806246757507324,
+      "learning_rate": 6.080964619781655e-06,
+      "loss": 0.3848,
+      "step": 27499
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.558209180831909,
+      "learning_rate": 6.08070719544171e-06,
+      "loss": 0.7671,
+      "step": 27500
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.6709492206573486,
+      "learning_rate": 6.08044976809672e-06,
+      "loss": 0.532,
+      "step": 27501
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.8355040550231934,
+      "learning_rate": 6.080192337747406e-06,
+      "loss": 0.688,
+      "step": 27502
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.118459701538086,
+      "learning_rate": 6.079934904394482e-06,
+      "loss": 0.5445,
+      "step": 27503
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.933866262435913,
+      "learning_rate": 6.079677468038662e-06,
+      "loss": 0.7722,
+      "step": 27504
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.7782833576202393,
+      "learning_rate": 6.079420028680666e-06,
+      "loss": 0.5797,
+      "step": 27505
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.594470739364624,
+      "learning_rate": 6.079162586321206e-06,
+      "loss": 0.7023,
+      "step": 27506
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.3794026374816895,
+      "learning_rate": 6.078905140960999e-06,
+      "loss": 0.7197,
+      "step": 27507
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 4.514443397521973,
+      "learning_rate": 6.078647692600761e-06,
+      "loss": 0.4783,
+      "step": 27508
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.3731727600097656,
+      "learning_rate": 6.078390241241207e-06,
+      "loss": 0.6501,
+      "step": 27509
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.2524781227111816,
+      "learning_rate": 6.078132786883054e-06,
+      "loss": 0.4436,
+      "step": 27510
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.5323704481124878,
+      "learning_rate": 6.077875329527018e-06,
+      "loss": 0.4298,
+      "step": 27511
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.2887136936187744,
+      "learning_rate": 6.077617869173815e-06,
+      "loss": 0.6145,
+      "step": 27512
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.7707960605621338,
+      "learning_rate": 6.0773604058241595e-06,
+      "loss": 0.5446,
+      "step": 27513
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.6758387088775635,
+      "learning_rate": 6.07710293947877e-06,
+      "loss": 0.8582,
+      "step": 27514
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.3213768005371094,
+      "learning_rate": 6.0768454701383586e-06,
+      "loss": 0.5075,
+      "step": 27515
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.156332492828369,
+      "learning_rate": 6.0765879978036445e-06,
+      "loss": 0.5684,
+      "step": 27516
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.72601318359375,
+      "learning_rate": 6.076330522475342e-06,
+      "loss": 0.5282,
+      "step": 27517
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.7476238012313843,
+      "learning_rate": 6.076073044154167e-06,
+      "loss": 0.5476,
+      "step": 27518
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.0634355545043945,
+      "learning_rate": 6.075815562840838e-06,
+      "loss": 0.5389,
+      "step": 27519
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 5.481206893920898,
+      "learning_rate": 6.075558078536066e-06,
+      "loss": 0.6036,
+      "step": 27520
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.9861568212509155,
+      "learning_rate": 6.075300591240572e-06,
+      "loss": 0.8387,
+      "step": 27521
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 5.03562068939209,
+      "learning_rate": 6.075043100955069e-06,
+      "loss": 0.6282,
+      "step": 27522
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.9849885702133179,
+      "learning_rate": 6.074785607680273e-06,
+      "loss": 0.8574,
+      "step": 27523
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.271082639694214,
+      "learning_rate": 6.074528111416903e-06,
+      "loss": 0.6724,
+      "step": 27524
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.9308772087097168,
+      "learning_rate": 6.074270612165672e-06,
+      "loss": 0.6354,
+      "step": 27525
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.1764891147613525,
+      "learning_rate": 6.074013109927297e-06,
+      "loss": 0.4702,
+      "step": 27526
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.197479248046875,
+      "learning_rate": 6.073755604702494e-06,
+      "loss": 0.7324,
+      "step": 27527
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.284104585647583,
+      "learning_rate": 6.073498096491979e-06,
+      "loss": 0.7266,
+      "step": 27528
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.968569755554199,
+      "learning_rate": 6.073240585296467e-06,
+      "loss": 0.9097,
+      "step": 27529
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 4.346066951751709,
+      "learning_rate": 6.072983071116675e-06,
+      "loss": 0.4074,
+      "step": 27530
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.5803911685943604,
+      "learning_rate": 6.072725553953319e-06,
+      "loss": 0.4583,
+      "step": 27531
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 4.2427825927734375,
+      "learning_rate": 6.072468033807115e-06,
+      "loss": 0.4554,
+      "step": 27532
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.0216336250305176,
+      "learning_rate": 6.072210510678781e-06,
+      "loss": 0.6993,
+      "step": 27533
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.2337234020233154,
+      "learning_rate": 6.071952984569029e-06,
+      "loss": 0.6813,
+      "step": 27534
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.4919726848602295,
+      "learning_rate": 6.071695455478579e-06,
+      "loss": 0.6077,
+      "step": 27535
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.7956373691558838,
+      "learning_rate": 6.071437923408144e-06,
+      "loss": 0.5863,
+      "step": 27536
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.491339683532715,
+      "learning_rate": 6.071180388358443e-06,
+      "loss": 0.7192,
+      "step": 27537
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.715707540512085,
+      "learning_rate": 6.07092285033019e-06,
+      "loss": 0.7345,
+      "step": 27538
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.6010849475860596,
+      "learning_rate": 6.070665309324102e-06,
+      "loss": 0.7068,
+      "step": 27539
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.8487930297851562,
+      "learning_rate": 6.070407765340893e-06,
+      "loss": 0.7308,
+      "step": 27540
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.7238850593566895,
+      "learning_rate": 6.070150218381282e-06,
+      "loss": 0.6266,
+      "step": 27541
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.0320115089416504,
+      "learning_rate": 6.069892668445985e-06,
+      "loss": 0.6795,
+      "step": 27542
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.3421051502227783,
+      "learning_rate": 6.069635115535717e-06,
+      "loss": 0.6883,
+      "step": 27543
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 4.243109703063965,
+      "learning_rate": 6.069377559651194e-06,
+      "loss": 0.6518,
+      "step": 27544
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.320284843444824,
+      "learning_rate": 6.069120000793133e-06,
+      "loss": 0.6549,
+      "step": 27545
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.4373857975006104,
+      "learning_rate": 6.06886243896225e-06,
+      "loss": 0.7797,
+      "step": 27546
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 4.099469184875488,
+      "learning_rate": 6.06860487415926e-06,
+      "loss": 0.6524,
+      "step": 27547
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.034280776977539,
+      "learning_rate": 6.068347306384881e-06,
+      "loss": 0.7448,
+      "step": 27548
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.6633251905441284,
+      "learning_rate": 6.0680897356398285e-06,
+      "loss": 0.6582,
+      "step": 27549
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.108133554458618,
+      "learning_rate": 6.067832161924817e-06,
+      "loss": 0.571,
+      "step": 27550
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.8843681812286377,
+      "learning_rate": 6.067574585240566e-06,
+      "loss": 0.628,
+      "step": 27551
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.9594075679779053,
+      "learning_rate": 6.0673170055877884e-06,
+      "loss": 0.5804,
+      "step": 27552
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.759671688079834,
+      "learning_rate": 6.067059422967203e-06,
+      "loss": 0.5264,
+      "step": 27553
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.7752492427825928,
+      "learning_rate": 6.066801837379526e-06,
+      "loss": 0.5539,
+      "step": 27554
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.086684465408325,
+      "learning_rate": 6.066544248825471e-06,
+      "loss": 0.5479,
+      "step": 27555
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.7970200777053833,
+      "learning_rate": 6.066286657305758e-06,
+      "loss": 0.5038,
+      "step": 27556
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 4.737143039703369,
+      "learning_rate": 6.066029062821098e-06,
+      "loss": 0.4828,
+      "step": 27557
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.6282825469970703,
+      "learning_rate": 6.065771465372213e-06,
+      "loss": 0.4221,
+      "step": 27558
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.6158862113952637,
+      "learning_rate": 6.065513864959816e-06,
+      "loss": 0.8839,
+      "step": 27559
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.3848233222961426,
+      "learning_rate": 6.065256261584625e-06,
+      "loss": 0.5262,
+      "step": 27560
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.5360805988311768,
+      "learning_rate": 6.064998655247355e-06,
+      "loss": 0.5354,
+      "step": 27561
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.869190216064453,
+      "learning_rate": 6.064741045948723e-06,
+      "loss": 0.7738,
+      "step": 27562
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.7205772399902344,
+      "learning_rate": 6.064483433689443e-06,
+      "loss": 0.5288,
+      "step": 27563
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.8817815780639648,
+      "learning_rate": 6.064225818470235e-06,
+      "loss": 0.8998,
+      "step": 27564
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.183366537094116,
+      "learning_rate": 6.0639682002918144e-06,
+      "loss": 0.4182,
+      "step": 27565
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.041565418243408,
+      "learning_rate": 6.063710579154896e-06,
+      "loss": 0.7972,
+      "step": 27566
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.5265023708343506,
+      "learning_rate": 6.063452955060196e-06,
+      "loss": 0.8938,
+      "step": 27567
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.8061400651931763,
+      "learning_rate": 6.063195328008433e-06,
+      "loss": 0.6823,
+      "step": 27568
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.090691089630127,
+      "learning_rate": 6.062937698000322e-06,
+      "loss": 0.6353,
+      "step": 27569
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.1348159313201904,
+      "learning_rate": 6.062680065036578e-06,
+      "loss": 0.5697,
+      "step": 27570
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.8968297243118286,
+      "learning_rate": 6.06242242911792e-06,
+      "loss": 0.7699,
+      "step": 27571
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.3517143726348877,
+      "learning_rate": 6.062164790245064e-06,
+      "loss": 0.5447,
+      "step": 27572
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.4232938289642334,
+      "learning_rate": 6.061907148418724e-06,
+      "loss": 0.4595,
+      "step": 27573
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.186265707015991,
+      "learning_rate": 6.061649503639619e-06,
+      "loss": 0.5357,
+      "step": 27574
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.0686328411102295,
+      "learning_rate": 6.061391855908465e-06,
+      "loss": 0.8163,
+      "step": 27575
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.0444655418395996,
+      "learning_rate": 6.061134205225977e-06,
+      "loss": 0.8093,
+      "step": 27576
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.7077572345733643,
+      "learning_rate": 6.060876551592872e-06,
+      "loss": 0.7638,
+      "step": 27577
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 4.564427852630615,
+      "learning_rate": 6.060618895009868e-06,
+      "loss": 0.6233,
+      "step": 27578
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.3273324966430664,
+      "learning_rate": 6.060361235477679e-06,
+      "loss": 0.6904,
+      "step": 27579
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 12.953280448913574,
+      "learning_rate": 6.060103572997023e-06,
+      "loss": 0.6586,
+      "step": 27580
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.3716516494750977,
+      "learning_rate": 6.0598459075686155e-06,
+      "loss": 0.5558,
+      "step": 27581
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.3833465576171875,
+      "learning_rate": 6.059588239193174e-06,
+      "loss": 0.5886,
+      "step": 27582
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.8159946203231812,
+      "learning_rate": 6.059330567871414e-06,
+      "loss": 0.7157,
+      "step": 27583
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.4584646224975586,
+      "learning_rate": 6.059072893604055e-06,
+      "loss": 0.5501,
+      "step": 27584
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.1220537424087524,
+      "learning_rate": 6.058815216391808e-06,
+      "loss": 0.5149,
+      "step": 27585
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.314424753189087,
+      "learning_rate": 6.058557536235395e-06,
+      "loss": 0.5193,
+      "step": 27586
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.2449052333831787,
+      "learning_rate": 6.058299853135527e-06,
+      "loss": 0.6232,
+      "step": 27587
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.045473575592041,
+      "learning_rate": 6.058042167092926e-06,
+      "loss": 0.7875,
+      "step": 27588
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 5.347228050231934,
+      "learning_rate": 6.057784478108305e-06,
+      "loss": 0.7387,
+      "step": 27589
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.3388798236846924,
+      "learning_rate": 6.057526786182383e-06,
+      "loss": 0.7683,
+      "step": 27590
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 4.337011814117432,
+      "learning_rate": 6.057269091315872e-06,
+      "loss": 0.9198,
+      "step": 27591
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.800666332244873,
+      "learning_rate": 6.0570113935094945e-06,
+      "loss": 0.8969,
+      "step": 27592
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.042485475540161,
+      "learning_rate": 6.056753692763964e-06,
+      "loss": 0.5686,
+      "step": 27593
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.119300365447998,
+      "learning_rate": 6.0564959890799964e-06,
+      "loss": 0.7193,
+      "step": 27594
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 4.374243259429932,
+      "learning_rate": 6.056238282458311e-06,
+      "loss": 0.5457,
+      "step": 27595
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.415342092514038,
+      "learning_rate": 6.055980572899621e-06,
+      "loss": 0.6354,
+      "step": 27596
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.968614935874939,
+      "learning_rate": 6.055722860404644e-06,
+      "loss": 0.6787,
+      "step": 27597
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.257124900817871,
+      "learning_rate": 6.055465144974099e-06,
+      "loss": 0.8575,
+      "step": 27598
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.136669397354126,
+      "learning_rate": 6.0552074266087004e-06,
+      "loss": 0.6613,
+      "step": 27599
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.6816354990005493,
+      "learning_rate": 6.054949705309166e-06,
+      "loss": 0.6618,
+      "step": 27600
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.157374143600464,
+      "learning_rate": 6.054691981076209e-06,
+      "loss": 0.4045,
+      "step": 27601
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.7217133045196533,
+      "learning_rate": 6.054434253910551e-06,
+      "loss": 0.5141,
+      "step": 27602
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.6109775304794312,
+      "learning_rate": 6.054176523812904e-06,
+      "loss": 0.551,
+      "step": 27603
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.11283802986145,
+      "learning_rate": 6.0539187907839895e-06,
+      "loss": 0.6592,
+      "step": 27604
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.061429262161255,
+      "learning_rate": 6.053661054824521e-06,
+      "loss": 0.6055,
+      "step": 27605
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.63727068901062,
+      "learning_rate": 6.053403315935216e-06,
+      "loss": 0.8135,
+      "step": 27606
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.141845464706421,
+      "learning_rate": 6.05314557411679e-06,
+      "loss": 0.4789,
+      "step": 27607
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.2060422897338867,
+      "learning_rate": 6.052887829369961e-06,
+      "loss": 0.4686,
+      "step": 27608
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.9656051397323608,
+      "learning_rate": 6.052630081695447e-06,
+      "loss": 0.5685,
+      "step": 27609
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.3192660808563232,
+      "learning_rate": 6.052372331093961e-06,
+      "loss": 0.7573,
+      "step": 27610
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.4163615703582764,
+      "learning_rate": 6.052114577566222e-06,
+      "loss": 0.5091,
+      "step": 27611
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.7060588598251343,
+      "learning_rate": 6.051856821112947e-06,
+      "loss": 0.4993,
+      "step": 27612
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.637179374694824,
+      "learning_rate": 6.0515990617348516e-06,
+      "loss": 0.7108,
+      "step": 27613
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.136220932006836,
+      "learning_rate": 6.051341299432653e-06,
+      "loss": 0.6866,
+      "step": 27614
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.1707162857055664,
+      "learning_rate": 6.051083534207069e-06,
+      "loss": 0.6407,
+      "step": 27615
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.4294917583465576,
+      "learning_rate": 6.0508257660588165e-06,
+      "loss": 0.3612,
+      "step": 27616
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 4.981433868408203,
+      "learning_rate": 6.05056799498861e-06,
+      "loss": 0.5366,
+      "step": 27617
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.307927370071411,
+      "learning_rate": 6.050310220997167e-06,
+      "loss": 0.7352,
+      "step": 27618
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.8559601306915283,
+      "learning_rate": 6.050052444085204e-06,
+      "loss": 0.8034,
+      "step": 27619
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.0550520420074463,
+      "learning_rate": 6.04979466425344e-06,
+      "loss": 0.6606,
+      "step": 27620
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.401885747909546,
+      "learning_rate": 6.04953688150259e-06,
+      "loss": 0.7124,
+      "step": 27621
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.9077584743499756,
+      "learning_rate": 6.049279095833371e-06,
+      "loss": 0.5551,
+      "step": 27622
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.0923244953155518,
+      "learning_rate": 6.0490213072465e-06,
+      "loss": 0.6994,
+      "step": 27623
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.4888107776641846,
+      "learning_rate": 6.048763515742694e-06,
+      "loss": 0.8095,
+      "step": 27624
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.39762806892395,
+      "learning_rate": 6.048505721322668e-06,
+      "loss": 0.3649,
+      "step": 27625
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.0385942459106445,
+      "learning_rate": 6.048247923987142e-06,
+      "loss": 0.5998,
+      "step": 27626
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.392984628677368,
+      "learning_rate": 6.04799012373683e-06,
+      "loss": 0.432,
+      "step": 27627
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.3723511695861816,
+      "learning_rate": 6.047732320572451e-06,
+      "loss": 0.4696,
+      "step": 27628
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.7908971309661865,
+      "learning_rate": 6.047474514494721e-06,
+      "loss": 0.6733,
+      "step": 27629
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.4704442024230957,
+      "learning_rate": 6.047216705504356e-06,
+      "loss": 0.8462,
+      "step": 27630
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.6766350269317627,
+      "learning_rate": 6.046958893602074e-06,
+      "loss": 0.6908,
+      "step": 27631
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.265075206756592,
+      "learning_rate": 6.046701078788592e-06,
+      "loss": 0.5394,
+      "step": 27632
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.9109864234924316,
+      "learning_rate": 6.046443261064625e-06,
+      "loss": 0.356,
+      "step": 27633
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.1630828380584717,
+      "learning_rate": 6.046185440430893e-06,
+      "loss": 0.9002,
+      "step": 27634
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.9398959875106812,
+      "learning_rate": 6.045927616888111e-06,
+      "loss": 0.4333,
+      "step": 27635
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.2934513092041016,
+      "learning_rate": 6.045669790436996e-06,
+      "loss": 0.7411,
+      "step": 27636
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.4286060333251953,
+      "learning_rate": 6.0454119610782636e-06,
+      "loss": 0.7584,
+      "step": 27637
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.5007587671279907,
+      "learning_rate": 6.045154128812634e-06,
+      "loss": 0.5709,
+      "step": 27638
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 5.77984094619751,
+      "learning_rate": 6.0448962936408214e-06,
+      "loss": 1.1308,
+      "step": 27639
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.0883853435516357,
+      "learning_rate": 6.044638455563544e-06,
+      "loss": 0.584,
+      "step": 27640
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.6907520294189453,
+      "learning_rate": 6.044380614581517e-06,
+      "loss": 0.6148,
+      "step": 27641
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.132122039794922,
+      "learning_rate": 6.0441227706954595e-06,
+      "loss": 0.5366,
+      "step": 27642
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.043286085128784,
+      "learning_rate": 6.043864923906089e-06,
+      "loss": 0.7199,
+      "step": 27643
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.8447856903076172,
+      "learning_rate": 6.04360707421412e-06,
+      "loss": 0.689,
+      "step": 27644
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.0016932487487793,
+      "learning_rate": 6.043349221620272e-06,
+      "loss": 0.5411,
+      "step": 27645
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 6.083364486694336,
+      "learning_rate": 6.043091366125258e-06,
+      "loss": 0.8819,
+      "step": 27646
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.114014148712158,
+      "learning_rate": 6.0428335077298e-06,
+      "loss": 0.6356,
+      "step": 27647
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.500274658203125,
+      "learning_rate": 6.042575646434612e-06,
+      "loss": 0.6848,
+      "step": 27648
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.491142988204956,
+      "learning_rate": 6.042317782240413e-06,
+      "loss": 0.5109,
+      "step": 27649
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.7349863052368164,
+      "learning_rate": 6.042059915147916e-06,
+      "loss": 0.6869,
+      "step": 27650
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.6156368255615234,
+      "learning_rate": 6.041802045157843e-06,
+      "loss": 0.5693,
+      "step": 27651
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.144151210784912,
+      "learning_rate": 6.041544172270907e-06,
+      "loss": 0.6488,
+      "step": 27652
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 5.094791412353516,
+      "learning_rate": 6.041286296487827e-06,
+      "loss": 0.6503,
+      "step": 27653
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.496119976043701,
+      "learning_rate": 6.04102841780932e-06,
+      "loss": 0.6811,
+      "step": 27654
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.207296371459961,
+      "learning_rate": 6.040770536236105e-06,
+      "loss": 0.6742,
+      "step": 27655
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.1736948490142822,
+      "learning_rate": 6.040512651768895e-06,
+      "loss": 0.6514,
+      "step": 27656
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.8862714767456055,
+      "learning_rate": 6.04025476440841e-06,
+      "loss": 0.5124,
+      "step": 27657
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.5700441598892212,
+      "learning_rate": 6.0399968741553646e-06,
+      "loss": 0.4274,
+      "step": 27658
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.6893858909606934,
+      "learning_rate": 6.03973898101048e-06,
+      "loss": 0.4853,
+      "step": 27659
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.7076579332351685,
+      "learning_rate": 6.039481084974469e-06,
+      "loss": 0.7279,
+      "step": 27660
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.6404547691345215,
+      "learning_rate": 6.039223186048051e-06,
+      "loss": 0.8205,
+      "step": 27661
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.183335065841675,
+      "learning_rate": 6.038965284231942e-06,
+      "loss": 0.591,
+      "step": 27662
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.4935953617095947,
+      "learning_rate": 6.0387073795268605e-06,
+      "loss": 0.7004,
+      "step": 27663
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.410633087158203,
+      "learning_rate": 6.038449471933523e-06,
+      "loss": 0.5594,
+      "step": 27664
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 10.950356483459473,
+      "learning_rate": 6.038191561452646e-06,
+      "loss": 0.494,
+      "step": 27665
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.072385787963867,
+      "learning_rate": 6.037933648084948e-06,
+      "loss": 0.501,
+      "step": 27666
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 4.169102668762207,
+      "learning_rate": 6.037675731831145e-06,
+      "loss": 0.6446,
+      "step": 27667
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 4.337927341461182,
+      "learning_rate": 6.037417812691955e-06,
+      "loss": 0.512,
+      "step": 27668
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.7004096508026123,
+      "learning_rate": 6.0371598906680925e-06,
+      "loss": 0.6608,
+      "step": 27669
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.6217056512832642,
+      "learning_rate": 6.036901965760279e-06,
+      "loss": 0.7292,
+      "step": 27670
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.208723783493042,
+      "learning_rate": 6.03664403796923e-06,
+      "loss": 0.6058,
+      "step": 27671
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.2638885974884033,
+      "learning_rate": 6.036386107295659e-06,
+      "loss": 0.6945,
+      "step": 27672
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.779517650604248,
+      "learning_rate": 6.03612817374029e-06,
+      "loss": 0.5432,
+      "step": 27673
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.8468858003616333,
+      "learning_rate": 6.0358702373038346e-06,
+      "loss": 0.6262,
+      "step": 27674
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.0827527046203613,
+      "learning_rate": 6.035612297987014e-06,
+      "loss": 0.9229,
+      "step": 27675
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 4.27360725402832,
+      "learning_rate": 6.035354355790541e-06,
+      "loss": 0.4782,
+      "step": 27676
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.9128648042678833,
+      "learning_rate": 6.0350964107151376e-06,
+      "loss": 0.603,
+      "step": 27677
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.927258849143982,
+      "learning_rate": 6.034838462761518e-06,
+      "loss": 0.5344,
+      "step": 27678
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.025777578353882,
+      "learning_rate": 6.034580511930401e-06,
+      "loss": 0.687,
+      "step": 27679
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.5870909690856934,
+      "learning_rate": 6.034322558222503e-06,
+      "loss": 0.5575,
+      "step": 27680
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 4.208167552947998,
+      "learning_rate": 6.034064601638539e-06,
+      "loss": 0.6636,
+      "step": 27681
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.4533159732818604,
+      "learning_rate": 6.03380664217923e-06,
+      "loss": 0.7873,
+      "step": 27682
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.7557610273361206,
+      "learning_rate": 6.033548679845294e-06,
+      "loss": 0.6461,
+      "step": 27683
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.8329719305038452,
+      "learning_rate": 6.033290714637445e-06,
+      "loss": 0.5999,
+      "step": 27684
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.405893325805664,
+      "learning_rate": 6.033032746556402e-06,
+      "loss": 0.5081,
+      "step": 27685
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.9303607940673828,
+      "learning_rate": 6.03277477560288e-06,
+      "loss": 0.4415,
+      "step": 27686
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.673882484436035,
+      "learning_rate": 6.0325168017776005e-06,
+      "loss": 0.3911,
+      "step": 27687
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.918987512588501,
+      "learning_rate": 6.032258825081278e-06,
+      "loss": 0.2315,
+      "step": 27688
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.585614562034607,
+      "learning_rate": 6.03200084551463e-06,
+      "loss": 0.7402,
+      "step": 27689
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.3985114097595215,
+      "learning_rate": 6.031742863078374e-06,
+      "loss": 0.7146,
+      "step": 27690
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.9645205736160278,
+      "learning_rate": 6.031484877773228e-06,
+      "loss": 0.7585,
+      "step": 27691
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.5262885093688965,
+      "learning_rate": 6.031226889599908e-06,
+      "loss": 0.4928,
+      "step": 27692
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.3103251457214355,
+      "learning_rate": 6.030968898559133e-06,
+      "loss": 0.4664,
+      "step": 27693
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.215934991836548,
+      "learning_rate": 6.030710904651621e-06,
+      "loss": 0.5628,
+      "step": 27694
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.434159755706787,
+      "learning_rate": 6.030452907878087e-06,
+      "loss": 0.7463,
+      "step": 27695
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.911992311477661,
+      "learning_rate": 6.03019490823925e-06,
+      "loss": 0.7753,
+      "step": 27696
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.8849976062774658,
+      "learning_rate": 6.029936905735826e-06,
+      "loss": 0.8648,
+      "step": 27697
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 5.27945613861084,
+      "learning_rate": 6.029678900368534e-06,
+      "loss": 0.7177,
+      "step": 27698
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.4269046783447266,
+      "learning_rate": 6.02942089213809e-06,
+      "loss": 0.6018,
+      "step": 27699
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.2421860694885254,
+      "learning_rate": 6.029162881045213e-06,
+      "loss": 0.5066,
+      "step": 27700
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.7927930355072021,
+      "learning_rate": 6.028904867090619e-06,
+      "loss": 0.8338,
+      "step": 27701
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.785146951675415,
+      "learning_rate": 6.028646850275026e-06,
+      "loss": 0.6668,
+      "step": 27702
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.390206813812256,
+      "learning_rate": 6.028388830599151e-06,
+      "loss": 0.6064,
+      "step": 27703
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 4.014309883117676,
+      "learning_rate": 6.028130808063712e-06,
+      "loss": 0.7947,
+      "step": 27704
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.7065677642822266,
+      "learning_rate": 6.027872782669428e-06,
+      "loss": 0.7302,
+      "step": 27705
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.7627060413360596,
+      "learning_rate": 6.0276147544170134e-06,
+      "loss": 0.6566,
+      "step": 27706
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.392810344696045,
+      "learning_rate": 6.027356723307187e-06,
+      "loss": 0.8412,
+      "step": 27707
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.90736722946167,
+      "learning_rate": 6.027098689340668e-06,
+      "loss": 0.7063,
+      "step": 27708
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.7843499183654785,
+      "learning_rate": 6.026840652518171e-06,
+      "loss": 0.7152,
+      "step": 27709
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.37001371383667,
+      "learning_rate": 6.026582612840415e-06,
+      "loss": 0.6433,
+      "step": 27710
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.637082815170288,
+      "learning_rate": 6.026324570308118e-06,
+      "loss": 0.8073,
+      "step": 27711
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.652719259262085,
+      "learning_rate": 6.026066524921996e-06,
+      "loss": 0.7162,
+      "step": 27712
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.6170921325683594,
+      "learning_rate": 6.025808476682766e-06,
+      "loss": 0.6215,
+      "step": 27713
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.511928081512451,
+      "learning_rate": 6.025550425591149e-06,
+      "loss": 0.3999,
+      "step": 27714
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.62235689163208,
+      "learning_rate": 6.025292371647859e-06,
+      "loss": 0.8125,
+      "step": 27715
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.982979655265808,
+      "learning_rate": 6.025034314853616e-06,
+      "loss": 0.6033,
+      "step": 27716
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.00974178314209,
+      "learning_rate": 6.024776255209137e-06,
+      "loss": 0.6229,
+      "step": 27717
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.4540975093841553,
+      "learning_rate": 6.0245181927151385e-06,
+      "loss": 0.7015,
+      "step": 27718
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.541746139526367,
+      "learning_rate": 6.024260127372339e-06,
+      "loss": 0.5997,
+      "step": 27719
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.05607533454895,
+      "learning_rate": 6.024002059181455e-06,
+      "loss": 0.6114,
+      "step": 27720
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.255449056625366,
+      "learning_rate": 6.023743988143206e-06,
+      "loss": 0.8013,
+      "step": 27721
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.981767177581787,
+      "learning_rate": 6.023485914258308e-06,
+      "loss": 0.6925,
+      "step": 27722
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.589611768722534,
+      "learning_rate": 6.023227837527477e-06,
+      "loss": 0.568,
+      "step": 27723
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.68898344039917,
+      "learning_rate": 6.022969757951436e-06,
+      "loss": 0.7366,
+      "step": 27724
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.2452192306518555,
+      "learning_rate": 6.0227116755308974e-06,
+      "loss": 0.7249,
+      "step": 27725
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.7206029891967773,
+      "learning_rate": 6.0224535902665815e-06,
+      "loss": 0.7556,
+      "step": 27726
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.6042442321777344,
+      "learning_rate": 6.022195502159205e-06,
+      "loss": 0.5098,
+      "step": 27727
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.6306309700012207,
+      "learning_rate": 6.021937411209487e-06,
+      "loss": 0.623,
+      "step": 27728
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.7562836408615112,
+      "learning_rate": 6.021679317418142e-06,
+      "loss": 0.5521,
+      "step": 27729
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.0325253009796143,
+      "learning_rate": 6.02142122078589e-06,
+      "loss": 0.3048,
+      "step": 27730
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.6847126483917236,
+      "learning_rate": 6.021163121313448e-06,
+      "loss": 0.6164,
+      "step": 27731
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.292656898498535,
+      "learning_rate": 6.020905019001535e-06,
+      "loss": 0.6919,
+      "step": 27732
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 4.136567115783691,
+      "learning_rate": 6.020646913850865e-06,
+      "loss": 0.6925,
+      "step": 27733
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.92802095413208,
+      "learning_rate": 6.020388805862161e-06,
+      "loss": 0.6912,
+      "step": 27734
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.7392771244049072,
+      "learning_rate": 6.020130695036137e-06,
+      "loss": 0.5447,
+      "step": 27735
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.739170789718628,
+      "learning_rate": 6.019872581373511e-06,
+      "loss": 0.5556,
+      "step": 27736
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.9544546604156494,
+      "learning_rate": 6.0196144648750036e-06,
+      "loss": 0.4815,
+      "step": 27737
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.2970519065856934,
+      "learning_rate": 6.0193563455413284e-06,
+      "loss": 0.7189,
+      "step": 27738
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.536429524421692,
+      "learning_rate": 6.019098223373205e-06,
+      "loss": 0.5606,
+      "step": 27739
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.1934969425201416,
+      "learning_rate": 6.018840098371352e-06,
+      "loss": 0.6833,
+      "step": 27740
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.534299373626709,
+      "learning_rate": 6.018581970536487e-06,
+      "loss": 0.4962,
+      "step": 27741
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.121335744857788,
+      "learning_rate": 6.0183238398693245e-06,
+      "loss": 0.8168,
+      "step": 27742
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.9394992589950562,
+      "learning_rate": 6.018065706370587e-06,
+      "loss": 0.8695,
+      "step": 27743
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.266519546508789,
+      "learning_rate": 6.01780757004099e-06,
+      "loss": 0.6875,
+      "step": 27744
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.185019016265869,
+      "learning_rate": 6.017549430881251e-06,
+      "loss": 0.5238,
+      "step": 27745
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.5577292442321777,
+      "learning_rate": 6.017291288892089e-06,
+      "loss": 0.8152,
+      "step": 27746
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.3623547554016113,
+      "learning_rate": 6.0170331440742205e-06,
+      "loss": 0.5396,
+      "step": 27747
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.6560745239257812,
+      "learning_rate": 6.016774996428362e-06,
+      "loss": 0.2834,
+      "step": 27748
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 4.264473915100098,
+      "learning_rate": 6.016516845955236e-06,
+      "loss": 0.6744,
+      "step": 27749
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.087594509124756,
+      "learning_rate": 6.016258692655555e-06,
+      "loss": 0.606,
+      "step": 27750
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.314146041870117,
+      "learning_rate": 6.016000536530041e-06,
+      "loss": 0.4735,
+      "step": 27751
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.0676653385162354,
+      "learning_rate": 6.01574237757941e-06,
+      "loss": 0.9229,
+      "step": 27752
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.212714195251465,
+      "learning_rate": 6.01548421580438e-06,
+      "loss": 0.7256,
+      "step": 27753
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.9528088569641113,
+      "learning_rate": 6.015226051205667e-06,
+      "loss": 0.6203,
+      "step": 27754
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.3506622314453125,
+      "learning_rate": 6.014967883783993e-06,
+      "loss": 0.7568,
+      "step": 27755
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.83695912361145,
+      "learning_rate": 6.014709713540073e-06,
+      "loss": 0.5367,
+      "step": 27756
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.3255441188812256,
+      "learning_rate": 6.014451540474624e-06,
+      "loss": 0.4685,
+      "step": 27757
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.2916555404663086,
+      "learning_rate": 6.014193364588367e-06,
+      "loss": 0.7264,
+      "step": 27758
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.9842941761016846,
+      "learning_rate": 6.013935185882017e-06,
+      "loss": 0.6911,
+      "step": 27759
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.8231194019317627,
+      "learning_rate": 6.013677004356294e-06,
+      "loss": 0.7215,
+      "step": 27760
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.3270983695983887,
+      "learning_rate": 6.013418820011915e-06,
+      "loss": 0.512,
+      "step": 27761
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 4.763181686401367,
+      "learning_rate": 6.0131606328495976e-06,
+      "loss": 0.7477,
+      "step": 27762
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.6124510765075684,
+      "learning_rate": 6.01290244287006e-06,
+      "loss": 0.6794,
+      "step": 27763
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.1868371963500977,
+      "learning_rate": 6.012644250074019e-06,
+      "loss": 0.3097,
+      "step": 27764
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.5532044172286987,
+      "learning_rate": 6.012386054462195e-06,
+      "loss": 0.7204,
+      "step": 27765
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.1678967475891113,
+      "learning_rate": 6.012127856035303e-06,
+      "loss": 0.8174,
+      "step": 27766
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.5592069625854492,
+      "learning_rate": 6.011869654794065e-06,
+      "loss": 0.4496,
+      "step": 27767
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.899372100830078,
+      "learning_rate": 6.0116114507391955e-06,
+      "loss": 0.8689,
+      "step": 27768
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.545408248901367,
+      "learning_rate": 6.011353243871414e-06,
+      "loss": 0.6636,
+      "step": 27769
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.7357678413391113,
+      "learning_rate": 6.011095034191437e-06,
+      "loss": 0.396,
+      "step": 27770
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.8384459018707275,
+      "learning_rate": 6.010836821699984e-06,
+      "loss": 0.5772,
+      "step": 27771
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.1708712577819824,
+      "learning_rate": 6.010578606397772e-06,
+      "loss": 0.3987,
+      "step": 27772
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.3101539611816406,
+      "learning_rate": 6.010320388285521e-06,
+      "loss": 0.4491,
+      "step": 27773
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.8830994367599487,
+      "learning_rate": 6.010062167363945e-06,
+      "loss": 0.5583,
+      "step": 27774
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.0578229427337646,
+      "learning_rate": 6.009803943633766e-06,
+      "loss": 0.6157,
+      "step": 27775
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.058258295059204,
+      "learning_rate": 6.009545717095701e-06,
+      "loss": 0.4337,
+      "step": 27776
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 4.069552421569824,
+      "learning_rate": 6.009287487750467e-06,
+      "loss": 0.4152,
+      "step": 27777
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.0091819763183594,
+      "learning_rate": 6.0090292555987815e-06,
+      "loss": 0.8019,
+      "step": 27778
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.483025550842285,
+      "learning_rate": 6.008771020641365e-06,
+      "loss": 0.7271,
+      "step": 27779
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 7.207122325897217,
+      "learning_rate": 6.008512782878935e-06,
+      "loss": 0.6417,
+      "step": 27780
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.379147529602051,
+      "learning_rate": 6.008254542312206e-06,
+      "loss": 0.7431,
+      "step": 27781
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 6.220123291015625,
+      "learning_rate": 6.007996298941901e-06,
+      "loss": 0.3773,
+      "step": 27782
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.6841404438018799,
+      "learning_rate": 6.007738052768735e-06,
+      "loss": 0.6784,
+      "step": 27783
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.252668857574463,
+      "learning_rate": 6.007479803793427e-06,
+      "loss": 0.8627,
+      "step": 27784
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.523143768310547,
+      "learning_rate": 6.007221552016695e-06,
+      "loss": 0.4684,
+      "step": 27785
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.9098660945892334,
+      "learning_rate": 6.006963297439258e-06,
+      "loss": 0.6489,
+      "step": 27786
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.380781650543213,
+      "learning_rate": 6.0067050400618335e-06,
+      "loss": 0.7016,
+      "step": 27787
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.365114450454712,
+      "learning_rate": 6.006446779885138e-06,
+      "loss": 0.5254,
+      "step": 27788
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.225405216217041,
+      "learning_rate": 6.006188516909892e-06,
+      "loss": 0.406,
+      "step": 27789
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.1999785900115967,
+      "learning_rate": 6.005930251136812e-06,
+      "loss": 0.845,
+      "step": 27790
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.2417843341827393,
+      "learning_rate": 6.005671982566618e-06,
+      "loss": 0.7107,
+      "step": 27791
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.5962371826171875,
+      "learning_rate": 6.0054137112000256e-06,
+      "loss": 0.6777,
+      "step": 27792
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.802762269973755,
+      "learning_rate": 6.005155437037755e-06,
+      "loss": 0.6184,
+      "step": 27793
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.532046318054199,
+      "learning_rate": 6.004897160080523e-06,
+      "loss": 0.5228,
+      "step": 27794
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 6.257410526275635,
+      "learning_rate": 6.004638880329049e-06,
+      "loss": 0.7484,
+      "step": 27795
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.7331554889678955,
+      "learning_rate": 6.00438059778405e-06,
+      "loss": 0.592,
+      "step": 27796
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.6431756019592285,
+      "learning_rate": 6.004122312446247e-06,
+      "loss": 0.8391,
+      "step": 27797
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.570504903793335,
+      "learning_rate": 6.0038640243163534e-06,
+      "loss": 0.8767,
+      "step": 27798
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.7335567474365234,
+      "learning_rate": 6.003605733395092e-06,
+      "loss": 0.6282,
+      "step": 27799
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.0633976459503174,
+      "learning_rate": 6.003347439683177e-06,
+      "loss": 0.6685,
+      "step": 27800
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.377993583679199,
+      "learning_rate": 6.00308914318133e-06,
+      "loss": 0.4881,
+      "step": 27801
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.4915727376937866,
+      "learning_rate": 6.002830843890269e-06,
+      "loss": 0.6466,
+      "step": 27802
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.127206325531006,
+      "learning_rate": 6.002572541810708e-06,
+      "loss": 0.7674,
+      "step": 27803
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.5768096446990967,
+      "learning_rate": 6.00231423694337e-06,
+      "loss": 0.5536,
+      "step": 27804
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.654672384262085,
+      "learning_rate": 6.0020559292889714e-06,
+      "loss": 0.6507,
+      "step": 27805
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.5910661220550537,
+      "learning_rate": 6.001797618848231e-06,
+      "loss": 0.6094,
+      "step": 27806
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.1379716396331787,
+      "learning_rate": 6.001539305621866e-06,
+      "loss": 0.6002,
+      "step": 27807
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.6632463932037354,
+      "learning_rate": 6.001280989610596e-06,
+      "loss": 0.4992,
+      "step": 27808
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.802168607711792,
+      "learning_rate": 6.001022670815139e-06,
+      "loss": 0.6345,
+      "step": 27809
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.094573974609375,
+      "learning_rate": 6.000764349236211e-06,
+      "loss": 0.7977,
+      "step": 27810
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.0810976028442383,
+      "learning_rate": 6.000506024874534e-06,
+      "loss": 0.6953,
+      "step": 27811
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.019052028656006,
+      "learning_rate": 6.0002476977308245e-06,
+      "loss": 0.5571,
+      "step": 27812
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.1966044902801514,
+      "learning_rate": 5.9999893678058005e-06,
+      "loss": 0.7881,
+      "step": 27813
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.769589424133301,
+      "learning_rate": 5.999731035100181e-06,
+      "loss": 0.6813,
+      "step": 27814
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.0031492710113525,
+      "learning_rate": 5.999472699614683e-06,
+      "loss": 0.8256,
+      "step": 27815
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.961754083633423,
+      "learning_rate": 5.999214361350025e-06,
+      "loss": 0.6245,
+      "step": 27816
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.033273458480835,
+      "learning_rate": 5.998956020306929e-06,
+      "loss": 0.6027,
+      "step": 27817
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.265493869781494,
+      "learning_rate": 5.998697676486109e-06,
+      "loss": 0.5279,
+      "step": 27818
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.427048921585083,
+      "learning_rate": 5.998439329888286e-06,
+      "loss": 0.5538,
+      "step": 27819
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.542914390563965,
+      "learning_rate": 5.998180980514176e-06,
+      "loss": 0.8414,
+      "step": 27820
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.8674765825271606,
+      "learning_rate": 5.997922628364499e-06,
+      "loss": 0.5324,
+      "step": 27821
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.7694039344787598,
+      "learning_rate": 5.997664273439974e-06,
+      "loss": 0.7736,
+      "step": 27822
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.00974178314209,
+      "learning_rate": 5.997405915741316e-06,
+      "loss": 0.9584,
+      "step": 27823
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 4.969989776611328,
+      "learning_rate": 5.9971475552692484e-06,
+      "loss": 0.5431,
+      "step": 27824
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.3902835845947266,
+      "learning_rate": 5.996889192024484e-06,
+      "loss": 0.6997,
+      "step": 27825
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.514145612716675,
+      "learning_rate": 5.996630826007746e-06,
+      "loss": 0.408,
+      "step": 27826
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 4.031291484832764,
+      "learning_rate": 5.996372457219751e-06,
+      "loss": 0.8759,
+      "step": 27827
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.5846997499465942,
+      "learning_rate": 5.996114085661218e-06,
+      "loss": 0.6464,
+      "step": 27828
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.1767826080322266,
+      "learning_rate": 5.995855711332864e-06,
+      "loss": 0.776,
+      "step": 27829
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.233035087585449,
+      "learning_rate": 5.995597334235409e-06,
+      "loss": 0.6866,
+      "step": 27830
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.166025161743164,
+      "learning_rate": 5.99533895436957e-06,
+      "loss": 0.8714,
+      "step": 27831
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.6507034301757812,
+      "learning_rate": 5.995080571736067e-06,
+      "loss": 0.7096,
+      "step": 27832
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.727355480194092,
+      "learning_rate": 5.994822186335619e-06,
+      "loss": 0.5184,
+      "step": 27833
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.130105972290039,
+      "learning_rate": 5.994563798168941e-06,
+      "loss": 0.7147,
+      "step": 27834
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.7023701667785645,
+      "learning_rate": 5.994305407236753e-06,
+      "loss": 0.5108,
+      "step": 27835
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.9928568601608276,
+      "learning_rate": 5.994047013539776e-06,
+      "loss": 0.5726,
+      "step": 27836
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.8563923835754395,
+      "learning_rate": 5.9937886170787256e-06,
+      "loss": 0.5404,
+      "step": 27837
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.638347864151001,
+      "learning_rate": 5.993530217854322e-06,
+      "loss": 0.8834,
+      "step": 27838
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.655705213546753,
+      "learning_rate": 5.993271815867282e-06,
+      "loss": 0.7718,
+      "step": 27839
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.0697505474090576,
+      "learning_rate": 5.993013411118326e-06,
+      "loss": 0.6869,
+      "step": 27840
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.4514787197113037,
+      "learning_rate": 5.992755003608171e-06,
+      "loss": 0.5011,
+      "step": 27841
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.514500856399536,
+      "learning_rate": 5.992496593337538e-06,
+      "loss": 0.5921,
+      "step": 27842
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.8550816774368286,
+      "learning_rate": 5.992238180307143e-06,
+      "loss": 0.3817,
+      "step": 27843
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 8.570405006408691,
+      "learning_rate": 5.991979764517705e-06,
+      "loss": 0.7528,
+      "step": 27844
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 7.806558609008789,
+      "learning_rate": 5.991721345969941e-06,
+      "loss": 0.6047,
+      "step": 27845
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.031421422958374,
+      "learning_rate": 5.9914629246645735e-06,
+      "loss": 0.6403,
+      "step": 27846
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.981382369995117,
+      "learning_rate": 5.991204500602319e-06,
+      "loss": 0.6158,
+      "step": 27847
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.9773552417755127,
+      "learning_rate": 5.990946073783896e-06,
+      "loss": 0.6227,
+      "step": 27848
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.902278423309326,
+      "learning_rate": 5.990687644210022e-06,
+      "loss": 0.5604,
+      "step": 27849
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.5816670656204224,
+      "learning_rate": 5.990429211881419e-06,
+      "loss": 0.5847,
+      "step": 27850
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 4.014002323150635,
+      "learning_rate": 5.990170776798803e-06,
+      "loss": 0.941,
+      "step": 27851
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.9343512058258057,
+      "learning_rate": 5.989912338962891e-06,
+      "loss": 0.6881,
+      "step": 27852
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.9706432819366455,
+      "learning_rate": 5.989653898374405e-06,
+      "loss": 0.4649,
+      "step": 27853
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.6100808382034302,
+      "learning_rate": 5.9893954550340625e-06,
+      "loss": 0.5043,
+      "step": 27854
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.2625179290771484,
+      "learning_rate": 5.98913700894258e-06,
+      "loss": 0.6149,
+      "step": 27855
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 5.020245552062988,
+      "learning_rate": 5.9888785601006806e-06,
+      "loss": 0.6926,
+      "step": 27856
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.81043541431427,
+      "learning_rate": 5.988620108509079e-06,
+      "loss": 0.7918,
+      "step": 27857
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.5718743801116943,
+      "learning_rate": 5.988361654168495e-06,
+      "loss": 0.6439,
+      "step": 27858
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.479221820831299,
+      "learning_rate": 5.988103197079649e-06,
+      "loss": 0.6128,
+      "step": 27859
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.1593363285064697,
+      "learning_rate": 5.9878447372432565e-06,
+      "loss": 0.6505,
+      "step": 27860
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.0164995193481445,
+      "learning_rate": 5.987586274660039e-06,
+      "loss": 0.6453,
+      "step": 27861
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.1092312335968018,
+      "learning_rate": 5.987327809330713e-06,
+      "loss": 0.503,
+      "step": 27862
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.8309036493301392,
+      "learning_rate": 5.987069341255999e-06,
+      "loss": 0.7925,
+      "step": 27863
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.539839029312134,
+      "learning_rate": 5.986810870436615e-06,
+      "loss": 0.564,
+      "step": 27864
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.8728888034820557,
+      "learning_rate": 5.9865523968732775e-06,
+      "loss": 0.8685,
+      "step": 27865
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.8321541547775269,
+      "learning_rate": 5.986293920566709e-06,
+      "loss": 0.3823,
+      "step": 27866
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.9280503988265991,
+      "learning_rate": 5.986035441517627e-06,
+      "loss": 0.5553,
+      "step": 27867
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.913607597351074,
+      "learning_rate": 5.98577695972675e-06,
+      "loss": 0.7797,
+      "step": 27868
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.014082431793213,
+      "learning_rate": 5.985518475194796e-06,
+      "loss": 0.6753,
+      "step": 27869
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.4923479557037354,
+      "learning_rate": 5.985259987922485e-06,
+      "loss": 0.7643,
+      "step": 27870
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.319976568222046,
+      "learning_rate": 5.985001497910534e-06,
+      "loss": 0.4886,
+      "step": 27871
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.8518826961517334,
+      "learning_rate": 5.984743005159662e-06,
+      "loss": 0.6429,
+      "step": 27872
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.6087168455123901,
+      "learning_rate": 5.9844845096705906e-06,
+      "loss": 0.6867,
+      "step": 27873
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.2088303565979004,
+      "learning_rate": 5.984226011444036e-06,
+      "loss": 0.5603,
+      "step": 27874
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.6011271476745605,
+      "learning_rate": 5.983967510480717e-06,
+      "loss": 0.59,
+      "step": 27875
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.0120465755462646,
+      "learning_rate": 5.983709006781353e-06,
+      "loss": 0.6765,
+      "step": 27876
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.450883150100708,
+      "learning_rate": 5.983450500346663e-06,
+      "loss": 0.5768,
+      "step": 27877
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.3800771236419678,
+      "learning_rate": 5.983191991177365e-06,
+      "loss": 0.324,
+      "step": 27878
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.8065598011016846,
+      "learning_rate": 5.98293347927418e-06,
+      "loss": 0.8573,
+      "step": 27879
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.773800849914551,
+      "learning_rate": 5.982674964637823e-06,
+      "loss": 0.6258,
+      "step": 27880
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.060093879699707,
+      "learning_rate": 5.982416447269017e-06,
+      "loss": 0.8522,
+      "step": 27881
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.1583383083343506,
+      "learning_rate": 5.982157927168478e-06,
+      "loss": 0.5396,
+      "step": 27882
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.1283304691314697,
+      "learning_rate": 5.981899404336926e-06,
+      "loss": 0.6783,
+      "step": 27883
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.58503794670105,
+      "learning_rate": 5.98164087877508e-06,
+      "loss": 0.7699,
+      "step": 27884
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.7184720039367676,
+      "learning_rate": 5.981382350483658e-06,
+      "loss": 0.8023,
+      "step": 27885
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.750581741333008,
+      "learning_rate": 5.981123819463378e-06,
+      "loss": 0.7225,
+      "step": 27886
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 4.3819899559021,
+      "learning_rate": 5.980865285714962e-06,
+      "loss": 0.6713,
+      "step": 27887
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.515049457550049,
+      "learning_rate": 5.9806067492391265e-06,
+      "loss": 0.7879,
+      "step": 27888
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 15.592601776123047,
+      "learning_rate": 5.9803482100365905e-06,
+      "loss": 0.5565,
+      "step": 27889
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.6549220085144043,
+      "learning_rate": 5.9800896681080746e-06,
+      "loss": 0.5124,
+      "step": 27890
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 4.570529937744141,
+      "learning_rate": 5.979831123454296e-06,
+      "loss": 0.8164,
+      "step": 27891
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.9520670175552368,
+      "learning_rate": 5.979572576075974e-06,
+      "loss": 0.7153,
+      "step": 27892
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.6869328022003174,
+      "learning_rate": 5.979314025973827e-06,
+      "loss": 0.8847,
+      "step": 27893
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 13.201737403869629,
+      "learning_rate": 5.979055473148576e-06,
+      "loss": 0.7084,
+      "step": 27894
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.7178406715393066,
+      "learning_rate": 5.978796917600938e-06,
+      "loss": 0.3785,
+      "step": 27895
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.880781888961792,
+      "learning_rate": 5.978538359331631e-06,
+      "loss": 0.5689,
+      "step": 27896
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.74231219291687,
+      "learning_rate": 5.978279798341375e-06,
+      "loss": 0.7069,
+      "step": 27897
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 4.04589319229126,
+      "learning_rate": 5.97802123463089e-06,
+      "loss": 0.7248,
+      "step": 27898
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.4651031494140625,
+      "learning_rate": 5.977762668200897e-06,
+      "loss": 0.7846,
+      "step": 27899
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.359861373901367,
+      "learning_rate": 5.977504099052109e-06,
+      "loss": 0.7653,
+      "step": 27900
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.176374912261963,
+      "learning_rate": 5.97724552718525e-06,
+      "loss": 0.4902,
+      "step": 27901
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.729940891265869,
+      "learning_rate": 5.976986952601038e-06,
+      "loss": 0.6877,
+      "step": 27902
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.7580004930496216,
+      "learning_rate": 5.97672837530019e-06,
+      "loss": 0.4648,
+      "step": 27903
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.36417293548584,
+      "learning_rate": 5.976469795283426e-06,
+      "loss": 0.5655,
+      "step": 27904
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 4.029008865356445,
+      "learning_rate": 5.976211212551466e-06,
+      "loss": 0.8476,
+      "step": 27905
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.9181675910949707,
+      "learning_rate": 5.975952627105027e-06,
+      "loss": 0.7395,
+      "step": 27906
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.567634105682373,
+      "learning_rate": 5.9756940389448305e-06,
+      "loss": 0.5488,
+      "step": 27907
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.59157395362854,
+      "learning_rate": 5.975435448071595e-06,
+      "loss": 0.5347,
+      "step": 27908
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.7179362773895264,
+      "learning_rate": 5.975176854486038e-06,
+      "loss": 0.8117,
+      "step": 27909
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.0121991634368896,
+      "learning_rate": 5.97491825818888e-06,
+      "loss": 0.9178,
+      "step": 27910
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.6974793672561646,
+      "learning_rate": 5.974659659180839e-06,
+      "loss": 0.7738,
+      "step": 27911
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.171754837036133,
+      "learning_rate": 5.974401057462636e-06,
+      "loss": 0.8,
+      "step": 27912
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.547763705253601,
+      "learning_rate": 5.9741424530349865e-06,
+      "loss": 0.6268,
+      "step": 27913
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.7047407627105713,
+      "learning_rate": 5.9738838458986124e-06,
+      "loss": 0.6773,
+      "step": 27914
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.5828962326049805,
+      "learning_rate": 5.973625236054233e-06,
+      "loss": 0.5957,
+      "step": 27915
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.0545265674591064,
+      "learning_rate": 5.973366623502566e-06,
+      "loss": 0.5805,
+      "step": 27916
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.4484200477600098,
+      "learning_rate": 5.973108008244331e-06,
+      "loss": 0.4193,
+      "step": 27917
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.2348194122314453,
+      "learning_rate": 5.972849390280246e-06,
+      "loss": 0.9844,
+      "step": 27918
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.6901872158050537,
+      "learning_rate": 5.972590769611033e-06,
+      "loss": 0.881,
+      "step": 27919
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.7516801357269287,
+      "learning_rate": 5.97233214623741e-06,
+      "loss": 0.8372,
+      "step": 27920
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.3374812602996826,
+      "learning_rate": 5.972073520160094e-06,
+      "loss": 0.5795,
+      "step": 27921
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.328897476196289,
+      "learning_rate": 5.971814891379805e-06,
+      "loss": 0.8127,
+      "step": 27922
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.7639809846878052,
+      "learning_rate": 5.971556259897266e-06,
+      "loss": 0.6478,
+      "step": 27923
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.8752843141555786,
+      "learning_rate": 5.971297625713191e-06,
+      "loss": 0.783,
+      "step": 27924
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.429201364517212,
+      "learning_rate": 5.9710389888283005e-06,
+      "loss": 0.5298,
+      "step": 27925
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.797598361968994,
+      "learning_rate": 5.970780349243315e-06,
+      "loss": 0.3316,
+      "step": 27926
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.4222285747528076,
+      "learning_rate": 5.970521706958953e-06,
+      "loss": 0.8756,
+      "step": 27927
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.1261680126190186,
+      "learning_rate": 5.970263061975933e-06,
+      "loss": 0.4328,
+      "step": 27928
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.9644840955734253,
+      "learning_rate": 5.970004414294976e-06,
+      "loss": 0.5629,
+      "step": 27929
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.524775505065918,
+      "learning_rate": 5.9697457639167995e-06,
+      "loss": 0.774,
+      "step": 27930
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.9225404262542725,
+      "learning_rate": 5.969487110842124e-06,
+      "loss": 0.6414,
+      "step": 27931
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.446402072906494,
+      "learning_rate": 5.969228455071667e-06,
+      "loss": 0.5499,
+      "step": 27932
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 5.493910789489746,
+      "learning_rate": 5.968969796606149e-06,
+      "loss": 0.6488,
+      "step": 27933
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.4083611965179443,
+      "learning_rate": 5.968711135446289e-06,
+      "loss": 0.8318,
+      "step": 27934
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.7869808673858643,
+      "learning_rate": 5.968452471592807e-06,
+      "loss": 0.886,
+      "step": 27935
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.0603039264678955,
+      "learning_rate": 5.96819380504642e-06,
+      "loss": 0.6249,
+      "step": 27936
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.86643123626709,
+      "learning_rate": 5.967935135807851e-06,
+      "loss": 0.5716,
+      "step": 27937
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.8790675401687622,
+      "learning_rate": 5.967676463877815e-06,
+      "loss": 0.5715,
+      "step": 27938
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.2830257415771484,
+      "learning_rate": 5.967417789257033e-06,
+      "loss": 0.4577,
+      "step": 27939
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.0346195697784424,
+      "learning_rate": 5.967159111946225e-06,
+      "loss": 0.6237,
+      "step": 27940
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.3831489086151123,
+      "learning_rate": 5.966900431946111e-06,
+      "loss": 0.5978,
+      "step": 27941
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.9520184993743896,
+      "learning_rate": 5.966641749257409e-06,
+      "loss": 0.4459,
+      "step": 27942
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 4.022247314453125,
+      "learning_rate": 5.9663830638808374e-06,
+      "loss": 0.4956,
+      "step": 27943
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.015949249267578,
+      "learning_rate": 5.966124375817117e-06,
+      "loss": 0.6783,
+      "step": 27944
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.8025484085083008,
+      "learning_rate": 5.965865685066967e-06,
+      "loss": 0.548,
+      "step": 27945
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.7406344413757324,
+      "learning_rate": 5.965606991631105e-06,
+      "loss": 0.6881,
+      "step": 27946
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.1916322708129883,
+      "learning_rate": 5.9653482955102525e-06,
+      "loss": 0.8252,
+      "step": 27947
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.4157652854919434,
+      "learning_rate": 5.9650895967051285e-06,
+      "loss": 0.573,
+      "step": 27948
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 4.19114875793457,
+      "learning_rate": 5.964830895216452e-06,
+      "loss": 0.7164,
+      "step": 27949
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.934065818786621,
+      "learning_rate": 5.96457219104494e-06,
+      "loss": 0.6809,
+      "step": 27950
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.2335500717163086,
+      "learning_rate": 5.964313484191317e-06,
+      "loss": 0.5448,
+      "step": 27951
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.7688722610473633,
+      "learning_rate": 5.964054774656298e-06,
+      "loss": 0.568,
+      "step": 27952
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.770951271057129,
+      "learning_rate": 5.9637960624406035e-06,
+      "loss": 0.415,
+      "step": 27953
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.2633304595947266,
+      "learning_rate": 5.9635373475449545e-06,
+      "loss": 0.6539,
+      "step": 27954
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.4450104236602783,
+      "learning_rate": 5.963278629970068e-06,
+      "loss": 0.6034,
+      "step": 27955
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.53524112701416,
+      "learning_rate": 5.963019909716665e-06,
+      "loss": 0.7853,
+      "step": 27956
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.0649890899658203,
+      "learning_rate": 5.962761186785463e-06,
+      "loss": 0.5564,
+      "step": 27957
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.4903886318206787,
+      "learning_rate": 5.9625024611771845e-06,
+      "loss": 0.7205,
+      "step": 27958
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.3434178829193115,
+      "learning_rate": 5.962243732892547e-06,
+      "loss": 0.7837,
+      "step": 27959
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.4437742233276367,
+      "learning_rate": 5.961985001932269e-06,
+      "loss": 0.5141,
+      "step": 27960
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.5748980045318604,
+      "learning_rate": 5.961726268297072e-06,
+      "loss": 0.6877,
+      "step": 27961
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.04278302192688,
+      "learning_rate": 5.961467531987675e-06,
+      "loss": 0.643,
+      "step": 27962
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.232696533203125,
+      "learning_rate": 5.961208793004796e-06,
+      "loss": 0.7906,
+      "step": 27963
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.265805244445801,
+      "learning_rate": 5.960950051349156e-06,
+      "loss": 0.4683,
+      "step": 27964
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.8728344440460205,
+      "learning_rate": 5.960691307021473e-06,
+      "loss": 0.5197,
+      "step": 27965
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 4.674010276794434,
+      "learning_rate": 5.9604325600224685e-06,
+      "loss": 0.9599,
+      "step": 27966
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.598719358444214,
+      "learning_rate": 5.960173810352859e-06,
+      "loss": 0.7916,
+      "step": 27967
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.3581252098083496,
+      "learning_rate": 5.959915058013368e-06,
+      "loss": 0.8088,
+      "step": 27968
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.8056302070617676,
+      "learning_rate": 5.9596563030047115e-06,
+      "loss": 0.601,
+      "step": 27969
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 4.711788654327393,
+      "learning_rate": 5.959397545327611e-06,
+      "loss": 0.6634,
+      "step": 27970
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.3304684162139893,
+      "learning_rate": 5.9591387849827856e-06,
+      "loss": 0.7988,
+      "step": 27971
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.665670156478882,
+      "learning_rate": 5.958880021970954e-06,
+      "loss": 0.8364,
+      "step": 27972
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.527240514755249,
+      "learning_rate": 5.958621256292837e-06,
+      "loss": 0.5985,
+      "step": 27973
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 4.313438415527344,
+      "learning_rate": 5.958362487949152e-06,
+      "loss": 0.6192,
+      "step": 27974
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.552074432373047,
+      "learning_rate": 5.958103716940621e-06,
+      "loss": 0.5978,
+      "step": 27975
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.467292308807373,
+      "learning_rate": 5.957844943267962e-06,
+      "loss": 0.5608,
+      "step": 27976
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.5137040615081787,
+      "learning_rate": 5.957586166931895e-06,
+      "loss": 0.4657,
+      "step": 27977
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.831315279006958,
+      "learning_rate": 5.95732738793314e-06,
+      "loss": 0.7489,
+      "step": 27978
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.412808418273926,
+      "learning_rate": 5.957068606272416e-06,
+      "loss": 0.771,
+      "step": 27979
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.1071114540100098,
+      "learning_rate": 5.956809821950442e-06,
+      "loss": 0.5944,
+      "step": 27980
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.185549020767212,
+      "learning_rate": 5.956551034967939e-06,
+      "loss": 0.6772,
+      "step": 27981
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 4.104035377502441,
+      "learning_rate": 5.956292245325626e-06,
+      "loss": 0.7111,
+      "step": 27982
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.164444923400879,
+      "learning_rate": 5.956033453024223e-06,
+      "loss": 0.6688,
+      "step": 27983
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.4536871910095215,
+      "learning_rate": 5.955774658064448e-06,
+      "loss": 0.8687,
+      "step": 27984
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.216498613357544,
+      "learning_rate": 5.955515860447023e-06,
+      "loss": 0.4305,
+      "step": 27985
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 6.058712005615234,
+      "learning_rate": 5.955257060172665e-06,
+      "loss": 0.7689,
+      "step": 27986
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.0598363876342773,
+      "learning_rate": 5.954998257242095e-06,
+      "loss": 0.8713,
+      "step": 27987
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.6937570571899414,
+      "learning_rate": 5.954739451656033e-06,
+      "loss": 0.7739,
+      "step": 27988
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.3324010372161865,
+      "learning_rate": 5.954480643415198e-06,
+      "loss": 0.7595,
+      "step": 27989
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.721953868865967,
+      "learning_rate": 5.95422183252031e-06,
+      "loss": 0.771,
+      "step": 27990
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.0055158138275146,
+      "learning_rate": 5.95396301897209e-06,
+      "loss": 0.5828,
+      "step": 27991
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.1469709873199463,
+      "learning_rate": 5.953704202771254e-06,
+      "loss": 0.8248,
+      "step": 27992
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.83872652053833,
+      "learning_rate": 5.9534453839185255e-06,
+      "loss": 0.4794,
+      "step": 27993
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.651857614517212,
+      "learning_rate": 5.953186562414622e-06,
+      "loss": 0.5563,
+      "step": 27994
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.2745859622955322,
+      "learning_rate": 5.9529277382602626e-06,
+      "loss": 0.7988,
+      "step": 27995
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.214317798614502,
+      "learning_rate": 5.95266891145617e-06,
+      "loss": 0.3031,
+      "step": 27996
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.4491233825683594,
+      "learning_rate": 5.95241008200306e-06,
+      "loss": 0.6195,
+      "step": 27997
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.5318775177001953,
+      "learning_rate": 5.9521512499016564e-06,
+      "loss": 0.7011,
+      "step": 27998
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.853623628616333,
+      "learning_rate": 5.951892415152676e-06,
+      "loss": 0.8765,
+      "step": 27999
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.3625495433807373,
+      "learning_rate": 5.951633577756839e-06,
+      "loss": 0.5454,
+      "step": 28000
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.573164224624634,
+      "learning_rate": 5.951374737714866e-06,
+      "loss": 0.7553,
+      "step": 28001
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.5710983276367188,
+      "learning_rate": 5.951115895027476e-06,
+      "loss": 0.4835,
+      "step": 28002
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.928741931915283,
+      "learning_rate": 5.950857049695389e-06,
+      "loss": 0.6359,
+      "step": 28003
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 4.2690629959106445,
+      "learning_rate": 5.950598201719326e-06,
+      "loss": 0.7724,
+      "step": 28004
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.7437796592712402,
+      "learning_rate": 5.950339351100003e-06,
+      "loss": 0.6794,
+      "step": 28005
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.0656168460845947,
+      "learning_rate": 5.950080497838144e-06,
+      "loss": 0.9247,
+      "step": 28006
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.7335426807403564,
+      "learning_rate": 5.9498216419344665e-06,
+      "loss": 0.6259,
+      "step": 28007
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.176414966583252,
+      "learning_rate": 5.949562783389691e-06,
+      "loss": 0.6936,
+      "step": 28008
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.0397427082061768,
+      "learning_rate": 5.949303922204537e-06,
+      "loss": 0.6564,
+      "step": 28009
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.336834192276001,
+      "learning_rate": 5.949045058379725e-06,
+      "loss": 0.7797,
+      "step": 28010
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.6561996936798096,
+      "learning_rate": 5.948786191915973e-06,
+      "loss": 0.5903,
+      "step": 28011
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.6088366508483887,
+      "learning_rate": 5.9485273228140026e-06,
+      "loss": 0.7008,
+      "step": 28012
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.699801206588745,
+      "learning_rate": 5.948268451074533e-06,
+      "loss": 0.6194,
+      "step": 28013
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.6342215538024902,
+      "learning_rate": 5.948009576698285e-06,
+      "loss": 0.5686,
+      "step": 28014
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.630906343460083,
+      "learning_rate": 5.9477506996859764e-06,
+      "loss": 0.4882,
+      "step": 28015
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.0095980167388916,
+      "learning_rate": 5.947491820038329e-06,
+      "loss": 0.5254,
+      "step": 28016
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.6807079315185547,
+      "learning_rate": 5.947232937756062e-06,
+      "loss": 0.6694,
+      "step": 28017
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.03098464012146,
+      "learning_rate": 5.946974052839894e-06,
+      "loss": 0.607,
+      "step": 28018
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.760115146636963,
+      "learning_rate": 5.946715165290545e-06,
+      "loss": 0.7632,
+      "step": 28019
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.596649408340454,
+      "learning_rate": 5.9464562751087375e-06,
+      "loss": 0.8237,
+      "step": 28020
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.8333897590637207,
+      "learning_rate": 5.946197382295189e-06,
+      "loss": 0.7777,
+      "step": 28021
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.2677462100982666,
+      "learning_rate": 5.945938486850621e-06,
+      "loss": 0.5148,
+      "step": 28022
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 6.762393951416016,
+      "learning_rate": 5.9456795887757515e-06,
+      "loss": 0.6313,
+      "step": 28023
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.561829090118408,
+      "learning_rate": 5.945420688071302e-06,
+      "loss": 0.8534,
+      "step": 28024
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.205411911010742,
+      "learning_rate": 5.945161784737991e-06,
+      "loss": 0.7832,
+      "step": 28025
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.5965805053710938,
+      "learning_rate": 5.94490287877654e-06,
+      "loss": 0.8167,
+      "step": 28026
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.2379817962646484,
+      "learning_rate": 5.944643970187667e-06,
+      "loss": 0.8567,
+      "step": 28027
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.077803134918213,
+      "learning_rate": 5.944385058972092e-06,
+      "loss": 0.7213,
+      "step": 28028
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.5647153854370117,
+      "learning_rate": 5.944126145130538e-06,
+      "loss": 0.3818,
+      "step": 28029
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.7506179809570312,
+      "learning_rate": 5.943867228663722e-06,
+      "loss": 0.788,
+      "step": 28030
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.2382566928863525,
+      "learning_rate": 5.943608309572365e-06,
+      "loss": 0.7488,
+      "step": 28031
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.1169140338897705,
+      "learning_rate": 5.943349387857187e-06,
+      "loss": 0.7719,
+      "step": 28032
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.641101598739624,
+      "learning_rate": 5.943090463518908e-06,
+      "loss": 0.5099,
+      "step": 28033
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.4932618141174316,
+      "learning_rate": 5.9428315365582465e-06,
+      "loss": 0.3884,
+      "step": 28034
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 5.023479461669922,
+      "learning_rate": 5.942572606975925e-06,
+      "loss": 0.75,
+      "step": 28035
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.3418121337890625,
+      "learning_rate": 5.942313674772662e-06,
+      "loss": 0.594,
+      "step": 28036
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.6372907161712646,
+      "learning_rate": 5.942054739949177e-06,
+      "loss": 0.5833,
+      "step": 28037
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.3686232566833496,
+      "learning_rate": 5.94179580250619e-06,
+      "loss": 0.6734,
+      "step": 28038
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 4.197732448577881,
+      "learning_rate": 5.941536862444423e-06,
+      "loss": 0.845,
+      "step": 28039
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.4350392818450928,
+      "learning_rate": 5.941277919764594e-06,
+      "loss": 0.4286,
+      "step": 28040
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.624020576477051,
+      "learning_rate": 5.941018974467424e-06,
+      "loss": 0.5693,
+      "step": 28041
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 4.552500247955322,
+      "learning_rate": 5.940760026553634e-06,
+      "loss": 0.5769,
+      "step": 28042
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.409639596939087,
+      "learning_rate": 5.940501076023942e-06,
+      "loss": 0.6557,
+      "step": 28043
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 4.929016590118408,
+      "learning_rate": 5.940242122879067e-06,
+      "loss": 0.8017,
+      "step": 28044
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.598027467727661,
+      "learning_rate": 5.939983167119733e-06,
+      "loss": 0.7676,
+      "step": 28045
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 5.37863826751709,
+      "learning_rate": 5.939724208746657e-06,
+      "loss": 0.506,
+      "step": 28046
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.907308578491211,
+      "learning_rate": 5.9394652477605605e-06,
+      "loss": 0.6612,
+      "step": 28047
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.1615047454833984,
+      "learning_rate": 5.939206284162161e-06,
+      "loss": 0.7606,
+      "step": 28048
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.1248655319213867,
+      "learning_rate": 5.938947317952183e-06,
+      "loss": 0.8526,
+      "step": 28049
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.7996281385421753,
+      "learning_rate": 5.938688349131344e-06,
+      "loss": 0.6893,
+      "step": 28050
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.0169715881347656,
+      "learning_rate": 5.938429377700364e-06,
+      "loss": 0.5519,
+      "step": 28051
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.396364688873291,
+      "learning_rate": 5.9381704036599645e-06,
+      "loss": 0.5251,
+      "step": 28052
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 7.878626823425293,
+      "learning_rate": 5.937911427010863e-06,
+      "loss": 0.4893,
+      "step": 28053
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.8853962421417236,
+      "learning_rate": 5.937652447753781e-06,
+      "loss": 0.5393,
+      "step": 28054
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.373568296432495,
+      "learning_rate": 5.937393465889441e-06,
+      "loss": 0.7728,
+      "step": 28055
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.229395627975464,
+      "learning_rate": 5.937134481418558e-06,
+      "loss": 0.6647,
+      "step": 28056
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.8101141452789307,
+      "learning_rate": 5.936875494341856e-06,
+      "loss": 0.7978,
+      "step": 28057
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.1775717735290527,
+      "learning_rate": 5.936616504660055e-06,
+      "loss": 0.6074,
+      "step": 28058
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.0219318866729736,
+      "learning_rate": 5.936357512373874e-06,
+      "loss": 0.5967,
+      "step": 28059
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.409855842590332,
+      "learning_rate": 5.936098517484034e-06,
+      "loss": 0.3708,
+      "step": 28060
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 4.842263698577881,
+      "learning_rate": 5.935839519991255e-06,
+      "loss": 0.7141,
+      "step": 28061
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.4224371910095215,
+      "learning_rate": 5.935580519896255e-06,
+      "loss": 0.5768,
+      "step": 28062
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.2359721660614014,
+      "learning_rate": 5.935321517199757e-06,
+      "loss": 0.6092,
+      "step": 28063
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.7062253952026367,
+      "learning_rate": 5.935062511902481e-06,
+      "loss": 0.5125,
+      "step": 28064
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.741029739379883,
+      "learning_rate": 5.934803504005145e-06,
+      "loss": 0.6654,
+      "step": 28065
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 4.132022380828857,
+      "learning_rate": 5.934544493508471e-06,
+      "loss": 0.5847,
+      "step": 28066
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.749848484992981,
+      "learning_rate": 5.93428548041318e-06,
+      "loss": 0.4902,
+      "step": 28067
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 4.178291320800781,
+      "learning_rate": 5.93402646471999e-06,
+      "loss": 0.5749,
+      "step": 28068
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 4.731532096862793,
+      "learning_rate": 5.9337674464296224e-06,
+      "loss": 0.6476,
+      "step": 28069
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.2515902519226074,
+      "learning_rate": 5.933508425542798e-06,
+      "loss": 0.8389,
+      "step": 28070
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.2961413860321045,
+      "learning_rate": 5.933249402060236e-06,
+      "loss": 0.8302,
+      "step": 28071
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.06925106048584,
+      "learning_rate": 5.932990375982657e-06,
+      "loss": 0.6161,
+      "step": 28072
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.153453826904297,
+      "learning_rate": 5.932731347310782e-06,
+      "loss": 0.5789,
+      "step": 28073
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.3025882244110107,
+      "learning_rate": 5.932472316045329e-06,
+      "loss": 0.8639,
+      "step": 28074
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.5149831771850586,
+      "learning_rate": 5.932213282187022e-06,
+      "loss": 0.4847,
+      "step": 28075
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.6666460037231445,
+      "learning_rate": 5.931954245736577e-06,
+      "loss": 0.3677,
+      "step": 28076
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.8133320808410645,
+      "learning_rate": 5.931695206694718e-06,
+      "loss": 0.7812,
+      "step": 28077
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.232795476913452,
+      "learning_rate": 5.931436165062162e-06,
+      "loss": 0.6267,
+      "step": 28078
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.107600450515747,
+      "learning_rate": 5.931177120839631e-06,
+      "loss": 0.4655,
+      "step": 28079
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.5570242404937744,
+      "learning_rate": 5.930918074027846e-06,
+      "loss": 0.478,
+      "step": 28080
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.0931038856506348,
+      "learning_rate": 5.930659024627526e-06,
+      "loss": 0.5078,
+      "step": 28081
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.6079931259155273,
+      "learning_rate": 5.930399972639393e-06,
+      "loss": 0.5196,
+      "step": 28082
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.3423070907592773,
+      "learning_rate": 5.930140918064166e-06,
+      "loss": 0.6256,
+      "step": 28083
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.1666364669799805,
+      "learning_rate": 5.929881860902564e-06,
+      "loss": 0.7049,
+      "step": 28084
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.7570033073425293,
+      "learning_rate": 5.929622801155311e-06,
+      "loss": 0.7145,
+      "step": 28085
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.194873332977295,
+      "learning_rate": 5.929363738823124e-06,
+      "loss": 0.8014,
+      "step": 28086
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 4.258387088775635,
+      "learning_rate": 5.929104673906725e-06,
+      "loss": 0.5739,
+      "step": 28087
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 5.836826324462891,
+      "learning_rate": 5.9288456064068345e-06,
+      "loss": 0.5804,
+      "step": 28088
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.714268445968628,
+      "learning_rate": 5.92858653632417e-06,
+      "loss": 0.643,
+      "step": 28089
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.544322967529297,
+      "learning_rate": 5.9283274636594565e-06,
+      "loss": 0.5824,
+      "step": 28090
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.867981433868408,
+      "learning_rate": 5.9280683884134115e-06,
+      "loss": 0.4853,
+      "step": 28091
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.9833099842071533,
+      "learning_rate": 5.927809310586756e-06,
+      "loss": 0.7659,
+      "step": 28092
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.17881441116333,
+      "learning_rate": 5.92755023018021e-06,
+      "loss": 0.8518,
+      "step": 28093
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 5.482243537902832,
+      "learning_rate": 5.927291147194495e-06,
+      "loss": 0.5386,
+      "step": 28094
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.5567879676818848,
+      "learning_rate": 5.92703206163033e-06,
+      "loss": 0.4924,
+      "step": 28095
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.9871058464050293,
+      "learning_rate": 5.926772973488437e-06,
+      "loss": 0.4364,
+      "step": 28096
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.5804100036621094,
+      "learning_rate": 5.926513882769535e-06,
+      "loss": 0.6278,
+      "step": 28097
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 5.5626983642578125,
+      "learning_rate": 5.926254789474345e-06,
+      "loss": 0.8599,
+      "step": 28098
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.4841959476470947,
+      "learning_rate": 5.9259956936035856e-06,
+      "loss": 0.881,
+      "step": 28099
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 8.266518592834473,
+      "learning_rate": 5.9257365951579815e-06,
+      "loss": 0.973,
+      "step": 28100
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.106574296951294,
+      "learning_rate": 5.9254774941382495e-06,
+      "loss": 0.5937,
+      "step": 28101
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.6361489295959473,
+      "learning_rate": 5.925218390545112e-06,
+      "loss": 0.5396,
+      "step": 28102
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.1178715229034424,
+      "learning_rate": 5.924959284379289e-06,
+      "loss": 0.5552,
+      "step": 28103
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.0547313690185547,
+      "learning_rate": 5.9247001756415e-06,
+      "loss": 0.3961,
+      "step": 28104
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.3540596961975098,
+      "learning_rate": 5.924441064332467e-06,
+      "loss": 0.6411,
+      "step": 28105
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.8732495307922363,
+      "learning_rate": 5.9241819504529084e-06,
+      "loss": 0.8423,
+      "step": 28106
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.7219769954681396,
+      "learning_rate": 5.923922834003547e-06,
+      "loss": 0.6924,
+      "step": 28107
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.5993983745574951,
+      "learning_rate": 5.923663714985102e-06,
+      "loss": 0.4644,
+      "step": 28108
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.021029472351074,
+      "learning_rate": 5.923404593398294e-06,
+      "loss": 0.7647,
+      "step": 28109
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.4358654022216797,
+      "learning_rate": 5.923145469243844e-06,
+      "loss": 0.6882,
+      "step": 28110
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.7118775844573975,
+      "learning_rate": 5.922886342522472e-06,
+      "loss": 0.6842,
+      "step": 28111
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.0939972400665283,
+      "learning_rate": 5.922627213234899e-06,
+      "loss": 0.3905,
+      "step": 28112
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.5707154273986816,
+      "learning_rate": 5.922368081381847e-06,
+      "loss": 0.6602,
+      "step": 28113
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.3560292720794678,
+      "learning_rate": 5.922108946964032e-06,
+      "loss": 0.5752,
+      "step": 28114
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.159388542175293,
+      "learning_rate": 5.921849809982178e-06,
+      "loss": 0.4068,
+      "step": 28115
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 4.210239887237549,
+      "learning_rate": 5.921590670437008e-06,
+      "loss": 0.5778,
+      "step": 28116
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.442350149154663,
+      "learning_rate": 5.921331528329236e-06,
+      "loss": 0.5549,
+      "step": 28117
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.886423110961914,
+      "learning_rate": 5.921072383659587e-06,
+      "loss": 0.6011,
+      "step": 28118
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.0187220573425293,
+      "learning_rate": 5.920813236428782e-06,
+      "loss": 0.4223,
+      "step": 28119
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 5.412357807159424,
+      "learning_rate": 5.920554086637539e-06,
+      "loss": 0.6248,
+      "step": 28120
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.5039267539978027,
+      "learning_rate": 5.92029493428658e-06,
+      "loss": 0.4507,
+      "step": 28121
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.147165060043335,
+      "learning_rate": 5.920035779376627e-06,
+      "loss": 0.7236,
+      "step": 28122
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.4980502128601074,
+      "learning_rate": 5.919776621908397e-06,
+      "loss": 0.4578,
+      "step": 28123
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.430267572402954,
+      "learning_rate": 5.919517461882614e-06,
+      "loss": 0.5007,
+      "step": 28124
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.9417669773101807,
+      "learning_rate": 5.919258299299997e-06,
+      "loss": 0.3653,
+      "step": 28125
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.0356228351593018,
+      "learning_rate": 5.9189991341612675e-06,
+      "loss": 0.8288,
+      "step": 28126
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.6814892292022705,
+      "learning_rate": 5.918739966467146e-06,
+      "loss": 0.84,
+      "step": 28127
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.1599533557891846,
+      "learning_rate": 5.918480796218352e-06,
+      "loss": 0.8585,
+      "step": 28128
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.2039482593536377,
+      "learning_rate": 5.9182216234156074e-06,
+      "loss": 0.6713,
+      "step": 28129
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 5.284038543701172,
+      "learning_rate": 5.917962448059632e-06,
+      "loss": 0.7869,
+      "step": 28130
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.679443359375,
+      "learning_rate": 5.9177032701511474e-06,
+      "loss": 0.5964,
+      "step": 28131
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.600947856903076,
+      "learning_rate": 5.9174440896908736e-06,
+      "loss": 0.858,
+      "step": 28132
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.8083298206329346,
+      "learning_rate": 5.917184906679532e-06,
+      "loss": 0.483,
+      "step": 28133
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.1161813735961914,
+      "learning_rate": 5.916925721117842e-06,
+      "loss": 0.5312,
+      "step": 28134
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.927481174468994,
+      "learning_rate": 5.916666533006526e-06,
+      "loss": 0.8638,
+      "step": 28135
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.578296422958374,
+      "learning_rate": 5.916407342346303e-06,
+      "loss": 0.5488,
+      "step": 28136
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.0672450065612793,
+      "learning_rate": 5.9161481491378945e-06,
+      "loss": 0.7809,
+      "step": 28137
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.3770222663879395,
+      "learning_rate": 5.915888953382022e-06,
+      "loss": 0.7714,
+      "step": 28138
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.0092780590057373,
+      "learning_rate": 5.915629755079405e-06,
+      "loss": 0.5124,
+      "step": 28139
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.598088264465332,
+      "learning_rate": 5.915370554230763e-06,
+      "loss": 0.7306,
+      "step": 28140
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.5149552822113037,
+      "learning_rate": 5.91511135083682e-06,
+      "loss": 0.5402,
+      "step": 28141
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.344276189804077,
+      "learning_rate": 5.914852144898295e-06,
+      "loss": 0.3295,
+      "step": 28142
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.5431714057922363,
+      "learning_rate": 5.914592936415909e-06,
+      "loss": 0.8279,
+      "step": 28143
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.6429898738861084,
+      "learning_rate": 5.914333725390382e-06,
+      "loss": 0.8284,
+      "step": 28144
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.0379483699798584,
+      "learning_rate": 5.914074511822436e-06,
+      "loss": 0.4022,
+      "step": 28145
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.412662982940674,
+      "learning_rate": 5.913815295712791e-06,
+      "loss": 0.5247,
+      "step": 28146
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.5460151433944702,
+      "learning_rate": 5.913556077062169e-06,
+      "loss": 0.6176,
+      "step": 28147
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.654539108276367,
+      "learning_rate": 5.9132968558712885e-06,
+      "loss": 0.5783,
+      "step": 28148
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.7727922201156616,
+      "learning_rate": 5.913037632140872e-06,
+      "loss": 0.5562,
+      "step": 28149
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.7086029052734375,
+      "learning_rate": 5.91277840587164e-06,
+      "loss": 0.4775,
+      "step": 28150
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.912712812423706,
+      "learning_rate": 5.912519177064313e-06,
+      "loss": 0.7218,
+      "step": 28151
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.239312171936035,
+      "learning_rate": 5.912259945719612e-06,
+      "loss": 0.489,
+      "step": 28152
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.883068799972534,
+      "learning_rate": 5.912000711838257e-06,
+      "loss": 0.3721,
+      "step": 28153
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.120521068572998,
+      "learning_rate": 5.9117414754209714e-06,
+      "loss": 0.8007,
+      "step": 28154
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.5822298526763916,
+      "learning_rate": 5.911482236468473e-06,
+      "loss": 0.515,
+      "step": 28155
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.706712245941162,
+      "learning_rate": 5.911222994981485e-06,
+      "loss": 0.7071,
+      "step": 28156
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.6707136631011963,
+      "learning_rate": 5.910963750960727e-06,
+      "loss": 0.8991,
+      "step": 28157
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.1366939544677734,
+      "learning_rate": 5.910704504406919e-06,
+      "loss": 0.4985,
+      "step": 28158
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.1880338191986084,
+      "learning_rate": 5.910445255320784e-06,
+      "loss": 0.6835,
+      "step": 28159
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.5090184211730957,
+      "learning_rate": 5.91018600370304e-06,
+      "loss": 0.6348,
+      "step": 28160
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.004218816757202,
+      "learning_rate": 5.909926749554411e-06,
+      "loss": 0.5716,
+      "step": 28161
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.305143117904663,
+      "learning_rate": 5.909667492875617e-06,
+      "loss": 0.6552,
+      "step": 28162
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 4.79878568649292,
+      "learning_rate": 5.909408233667378e-06,
+      "loss": 0.4702,
+      "step": 28163
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.7258734703063965,
+      "learning_rate": 5.909148971930416e-06,
+      "loss": 0.6968,
+      "step": 28164
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.8406174182891846,
+      "learning_rate": 5.908889707665451e-06,
+      "loss": 0.5096,
+      "step": 28165
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.3974573612213135,
+      "learning_rate": 5.9086304408732045e-06,
+      "loss": 0.6534,
+      "step": 28166
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.008248805999756,
+      "learning_rate": 5.9083711715543955e-06,
+      "loss": 0.7986,
+      "step": 28167
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.613154888153076,
+      "learning_rate": 5.908111899709749e-06,
+      "loss": 0.4134,
+      "step": 28168
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.512390613555908,
+      "learning_rate": 5.9078526253399814e-06,
+      "loss": 0.5255,
+      "step": 28169
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 10.938352584838867,
+      "learning_rate": 5.907593348445816e-06,
+      "loss": 0.8676,
+      "step": 28170
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.5469517707824707,
+      "learning_rate": 5.907334069027973e-06,
+      "loss": 0.4816,
+      "step": 28171
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.7110540866851807,
+      "learning_rate": 5.907074787087176e-06,
+      "loss": 0.7272,
+      "step": 28172
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.2681591510772705,
+      "learning_rate": 5.906815502624143e-06,
+      "loss": 0.6022,
+      "step": 28173
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.4950709342956543,
+      "learning_rate": 5.906556215639595e-06,
+      "loss": 0.4671,
+      "step": 28174
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.295722484588623,
+      "learning_rate": 5.906296926134254e-06,
+      "loss": 0.5277,
+      "step": 28175
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 4.998647212982178,
+      "learning_rate": 5.906037634108841e-06,
+      "loss": 0.7987,
+      "step": 28176
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.46667218208313,
+      "learning_rate": 5.905778339564076e-06,
+      "loss": 0.6024,
+      "step": 28177
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.6699109077453613,
+      "learning_rate": 5.905519042500683e-06,
+      "loss": 0.826,
+      "step": 28178
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.1574296951293945,
+      "learning_rate": 5.90525974291938e-06,
+      "loss": 0.5542,
+      "step": 28179
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.682802200317383,
+      "learning_rate": 5.905000440820886e-06,
+      "loss": 0.7657,
+      "step": 28180
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.3154146671295166,
+      "learning_rate": 5.904741136205927e-06,
+      "loss": 0.7225,
+      "step": 28181
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.0624446868896484,
+      "learning_rate": 5.904481829075222e-06,
+      "loss": 0.8265,
+      "step": 28182
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.82403826713562,
+      "learning_rate": 5.904222519429491e-06,
+      "loss": 0.7752,
+      "step": 28183
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.28311288356781,
+      "learning_rate": 5.903963207269456e-06,
+      "loss": 0.5456,
+      "step": 28184
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 4.997801303863525,
+      "learning_rate": 5.903703892595839e-06,
+      "loss": 0.6093,
+      "step": 28185
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.9640237092971802,
+      "learning_rate": 5.903444575409358e-06,
+      "loss": 0.7036,
+      "step": 28186
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.876056432723999,
+      "learning_rate": 5.903185255710737e-06,
+      "loss": 0.7502,
+      "step": 28187
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 4.078547477722168,
+      "learning_rate": 5.902925933500697e-06,
+      "loss": 0.7124,
+      "step": 28188
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.291109800338745,
+      "learning_rate": 5.902666608779959e-06,
+      "loss": 0.5608,
+      "step": 28189
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.335012674331665,
+      "learning_rate": 5.902407281549242e-06,
+      "loss": 0.6329,
+      "step": 28190
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.008626937866211,
+      "learning_rate": 5.902147951809269e-06,
+      "loss": 0.724,
+      "step": 28191
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.2659637928009033,
+      "learning_rate": 5.901888619560758e-06,
+      "loss": 0.4538,
+      "step": 28192
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 4.295670032501221,
+      "learning_rate": 5.901629284804434e-06,
+      "loss": 0.7022,
+      "step": 28193
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.2463083267211914,
+      "learning_rate": 5.901369947541018e-06,
+      "loss": 0.7328,
+      "step": 28194
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.960582971572876,
+      "learning_rate": 5.90111060777123e-06,
+      "loss": 0.509,
+      "step": 28195
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.8973453044891357,
+      "learning_rate": 5.90085126549579e-06,
+      "loss": 0.7486,
+      "step": 28196
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.0475943088531494,
+      "learning_rate": 5.9005919207154195e-06,
+      "loss": 0.3719,
+      "step": 28197
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.496572971343994,
+      "learning_rate": 5.900332573430841e-06,
+      "loss": 0.9097,
+      "step": 28198
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.974119782447815,
+      "learning_rate": 5.900073223642775e-06,
+      "loss": 0.6541,
+      "step": 28199
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.570376992225647,
+      "learning_rate": 5.899813871351943e-06,
+      "loss": 0.4969,
+      "step": 28200
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.861858367919922,
+      "learning_rate": 5.899554516559064e-06,
+      "loss": 0.6478,
+      "step": 28201
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 4.850593090057373,
+      "learning_rate": 5.899295159264862e-06,
+      "loss": 0.7392,
+      "step": 28202
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.7624907493591309,
+      "learning_rate": 5.899035799470056e-06,
+      "loss": 0.5715,
+      "step": 28203
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.2285399436950684,
+      "learning_rate": 5.89877643717537e-06,
+      "loss": 0.7109,
+      "step": 28204
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.0712006092071533,
+      "learning_rate": 5.898517072381524e-06,
+      "loss": 0.6709,
+      "step": 28205
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.068138837814331,
+      "learning_rate": 5.898257705089237e-06,
+      "loss": 0.5922,
+      "step": 28206
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.4537150859832764,
+      "learning_rate": 5.897998335299231e-06,
+      "loss": 0.799,
+      "step": 28207
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.8007051944732666,
+      "learning_rate": 5.89773896301223e-06,
+      "loss": 0.8363,
+      "step": 28208
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.211448907852173,
+      "learning_rate": 5.897479588228951e-06,
+      "loss": 0.6765,
+      "step": 28209
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.7933807373046875,
+      "learning_rate": 5.89722021095012e-06,
+      "loss": 0.4186,
+      "step": 28210
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.4480929374694824,
+      "learning_rate": 5.896960831176453e-06,
+      "loss": 0.8728,
+      "step": 28211
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.110267162322998,
+      "learning_rate": 5.896701448908676e-06,
+      "loss": 0.7402,
+      "step": 28212
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.1005566120147705,
+      "learning_rate": 5.8964420641475075e-06,
+      "loss": 0.9095,
+      "step": 28213
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.261826753616333,
+      "learning_rate": 5.8961826768936695e-06,
+      "loss": 0.6704,
+      "step": 28214
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.6163554191589355,
+      "learning_rate": 5.8959232871478835e-06,
+      "loss": 0.3997,
+      "step": 28215
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.150806188583374,
+      "learning_rate": 5.89566389491087e-06,
+      "loss": 0.6063,
+      "step": 28216
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.4607441425323486,
+      "learning_rate": 5.8954045001833514e-06,
+      "loss": 0.7418,
+      "step": 28217
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.2231571674346924,
+      "learning_rate": 5.895145102966048e-06,
+      "loss": 0.8893,
+      "step": 28218
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.0727031230926514,
+      "learning_rate": 5.89488570325968e-06,
+      "loss": 0.7837,
+      "step": 28219
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.0098536014556885,
+      "learning_rate": 5.894626301064973e-06,
+      "loss": 0.5453,
+      "step": 28220
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.8506393432617188,
+      "learning_rate": 5.894366896382643e-06,
+      "loss": 0.8683,
+      "step": 28221
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.0656938552856445,
+      "learning_rate": 5.894107489213415e-06,
+      "loss": 0.8554,
+      "step": 28222
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.855543851852417,
+      "learning_rate": 5.893848079558007e-06,
+      "loss": 0.631,
+      "step": 28223
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.3530220985412598,
+      "learning_rate": 5.893588667417144e-06,
+      "loss": 0.6192,
+      "step": 28224
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.517188549041748,
+      "learning_rate": 5.893329252791545e-06,
+      "loss": 0.5311,
+      "step": 28225
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.015110492706299,
+      "learning_rate": 5.893069835681933e-06,
+      "loss": 0.5855,
+      "step": 28226
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.9620400667190552,
+      "learning_rate": 5.892810416089028e-06,
+      "loss": 0.6107,
+      "step": 28227
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.9717631340026855,
+      "learning_rate": 5.892550994013552e-06,
+      "loss": 0.7019,
+      "step": 28228
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.8431000709533691,
+      "learning_rate": 5.892291569456225e-06,
+      "loss": 0.7886,
+      "step": 28229
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 4.704517841339111,
+      "learning_rate": 5.892032142417771e-06,
+      "loss": 0.726,
+      "step": 28230
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.7573554515838623,
+      "learning_rate": 5.891772712898906e-06,
+      "loss": 0.5527,
+      "step": 28231
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.3400557041168213,
+      "learning_rate": 5.891513280900358e-06,
+      "loss": 0.6813,
+      "step": 28232
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.8880460262298584,
+      "learning_rate": 5.891253846422846e-06,
+      "loss": 0.577,
+      "step": 28233
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.57588791847229,
+      "learning_rate": 5.89099440946709e-06,
+      "loss": 0.7446,
+      "step": 28234
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.5967845916748047,
+      "learning_rate": 5.890734970033813e-06,
+      "loss": 0.6028,
+      "step": 28235
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 4.390178203582764,
+      "learning_rate": 5.890475528123736e-06,
+      "loss": 0.768,
+      "step": 28236
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.1518349647521973,
+      "learning_rate": 5.8902160837375785e-06,
+      "loss": 0.5278,
+      "step": 28237
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.937077522277832,
+      "learning_rate": 5.889956636876065e-06,
+      "loss": 0.5417,
+      "step": 28238
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.7371833324432373,
+      "learning_rate": 5.8896971875399155e-06,
+      "loss": 0.5584,
+      "step": 28239
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.0697171688079834,
+      "learning_rate": 5.889437735729852e-06,
+      "loss": 0.4641,
+      "step": 28240
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.6371662616729736,
+      "learning_rate": 5.889178281446593e-06,
+      "loss": 0.5838,
+      "step": 28241
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.4406296014785767,
+      "learning_rate": 5.888918824690863e-06,
+      "loss": 0.5015,
+      "step": 28242
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.8483730554580688,
+      "learning_rate": 5.888659365463384e-06,
+      "loss": 0.5047,
+      "step": 28243
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.9286317825317383,
+      "learning_rate": 5.888399903764875e-06,
+      "loss": 0.5658,
+      "step": 28244
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.105689287185669,
+      "learning_rate": 5.88814043959606e-06,
+      "loss": 0.5048,
+      "step": 28245
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.9184937477111816,
+      "learning_rate": 5.887880972957658e-06,
+      "loss": 0.5021,
+      "step": 28246
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.6375184059143066,
+      "learning_rate": 5.887621503850391e-06,
+      "loss": 0.6028,
+      "step": 28247
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.1834182739257812,
+      "learning_rate": 5.887362032274982e-06,
+      "loss": 0.7518,
+      "step": 28248
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.031174898147583,
+      "learning_rate": 5.8871025582321515e-06,
+      "loss": 0.6118,
+      "step": 28249
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.3290674686431885,
+      "learning_rate": 5.886843081722622e-06,
+      "loss": 0.3429,
+      "step": 28250
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.891937255859375,
+      "learning_rate": 5.886583602747112e-06,
+      "loss": 0.8016,
+      "step": 28251
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.5166906118392944,
+      "learning_rate": 5.886324121306347e-06,
+      "loss": 0.6685,
+      "step": 28252
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.7675387859344482,
+      "learning_rate": 5.886064637401045e-06,
+      "loss": 0.6305,
+      "step": 28253
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.1490683555603027,
+      "learning_rate": 5.885805151031932e-06,
+      "loss": 0.7007,
+      "step": 28254
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.7673304080963135,
+      "learning_rate": 5.885545662199724e-06,
+      "loss": 0.7369,
+      "step": 28255
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.1182913780212402,
+      "learning_rate": 5.885286170905147e-06,
+      "loss": 0.4347,
+      "step": 28256
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.722928285598755,
+      "learning_rate": 5.88502667714892e-06,
+      "loss": 0.5767,
+      "step": 28257
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.3710254430770874,
+      "learning_rate": 5.884767180931765e-06,
+      "loss": 0.6011,
+      "step": 28258
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.4705084562301636,
+      "learning_rate": 5.8845076822544045e-06,
+      "loss": 0.464,
+      "step": 28259
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.662656307220459,
+      "learning_rate": 5.8842481811175585e-06,
+      "loss": 0.5928,
+      "step": 28260
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.9240950345993042,
+      "learning_rate": 5.883988677521951e-06,
+      "loss": 0.6201,
+      "step": 28261
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.34286367893219,
+      "learning_rate": 5.883729171468301e-06,
+      "loss": 0.6585,
+      "step": 28262
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.5524842739105225,
+      "learning_rate": 5.883469662957332e-06,
+      "loss": 0.6226,
+      "step": 28263
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.7139496803283691,
+      "learning_rate": 5.883210151989765e-06,
+      "loss": 0.4941,
+      "step": 28264
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.9092257022857666,
+      "learning_rate": 5.8829506385663204e-06,
+      "loss": 0.6189,
+      "step": 28265
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.5804054737091064,
+      "learning_rate": 5.8826911226877215e-06,
+      "loss": 0.577,
+      "step": 28266
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.155956745147705,
+      "learning_rate": 5.882431604354689e-06,
+      "loss": 0.5412,
+      "step": 28267
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.649689197540283,
+      "learning_rate": 5.882172083567946e-06,
+      "loss": 0.8814,
+      "step": 28268
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.7137539386749268,
+      "learning_rate": 5.881912560328211e-06,
+      "loss": 0.6795,
+      "step": 28269
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.0850112438201904,
+      "learning_rate": 5.881653034636208e-06,
+      "loss": 0.6607,
+      "step": 28270
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.5906789302825928,
+      "learning_rate": 5.881393506492659e-06,
+      "loss": 0.8079,
+      "step": 28271
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.3816590309143066,
+      "learning_rate": 5.881133975898283e-06,
+      "loss": 0.6462,
+      "step": 28272
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.8037383556365967,
+      "learning_rate": 5.880874442853806e-06,
+      "loss": 0.7435,
+      "step": 28273
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.866621255874634,
+      "learning_rate": 5.880614907359945e-06,
+      "loss": 0.5676,
+      "step": 28274
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.761313796043396,
+      "learning_rate": 5.880355369417425e-06,
+      "loss": 0.6062,
+      "step": 28275
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.4021522998809814,
+      "learning_rate": 5.880095829026967e-06,
+      "loss": 0.7269,
+      "step": 28276
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.803661823272705,
+      "learning_rate": 5.879836286189291e-06,
+      "loss": 0.4564,
+      "step": 28277
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 4.151736736297607,
+      "learning_rate": 5.879576740905121e-06,
+      "loss": 0.754,
+      "step": 28278
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.0957303047180176,
+      "learning_rate": 5.879317193175177e-06,
+      "loss": 0.5409,
+      "step": 28279
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.827012062072754,
+      "learning_rate": 5.879057643000181e-06,
+      "loss": 0.7572,
+      "step": 28280
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.2105236053466797,
+      "learning_rate": 5.8787980903808565e-06,
+      "loss": 0.8837,
+      "step": 28281
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.586946725845337,
+      "learning_rate": 5.878538535317921e-06,
+      "loss": 0.7331,
+      "step": 28282
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.4152848720550537,
+      "learning_rate": 5.878278977812101e-06,
+      "loss": 0.4477,
+      "step": 28283
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.4086930751800537,
+      "learning_rate": 5.878019417864115e-06,
+      "loss": 0.7464,
+      "step": 28284
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.328052520751953,
+      "learning_rate": 5.877759855474687e-06,
+      "loss": 0.7157,
+      "step": 28285
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.8074843883514404,
+      "learning_rate": 5.877500290644537e-06,
+      "loss": 0.5737,
+      "step": 28286
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.6030960083007812,
+      "learning_rate": 5.8772407233743876e-06,
+      "loss": 0.7161,
+      "step": 28287
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.682304859161377,
+      "learning_rate": 5.87698115366496e-06,
+      "loss": 0.6192,
+      "step": 28288
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.7254090309143066,
+      "learning_rate": 5.8767215815169775e-06,
+      "loss": 0.6463,
+      "step": 28289
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.429438591003418,
+      "learning_rate": 5.876462006931159e-06,
+      "loss": 0.7102,
+      "step": 28290
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.1397030353546143,
+      "learning_rate": 5.87620242990823e-06,
+      "loss": 0.7225,
+      "step": 28291
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.764052391052246,
+      "learning_rate": 5.875942850448908e-06,
+      "loss": 0.6495,
+      "step": 28292
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.695363998413086,
+      "learning_rate": 5.875683268553919e-06,
+      "loss": 0.3928,
+      "step": 28293
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.0468931198120117,
+      "learning_rate": 5.875423684223982e-06,
+      "loss": 0.5523,
+      "step": 28294
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 4.411375522613525,
+      "learning_rate": 5.87516409745982e-06,
+      "loss": 0.7705,
+      "step": 28295
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 7.445242404937744,
+      "learning_rate": 5.8749045082621534e-06,
+      "loss": 0.6506,
+      "step": 28296
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 6.128384113311768,
+      "learning_rate": 5.8746449166317065e-06,
+      "loss": 0.649,
+      "step": 28297
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.6156127452850342,
+      "learning_rate": 5.874385322569199e-06,
+      "loss": 0.4838,
+      "step": 28298
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.7970454692840576,
+      "learning_rate": 5.874125726075355e-06,
+      "loss": 0.5222,
+      "step": 28299
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.1409764289855957,
+      "learning_rate": 5.873866127150893e-06,
+      "loss": 0.479,
+      "step": 28300
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.841691017150879,
+      "learning_rate": 5.873606525796537e-06,
+      "loss": 0.5468,
+      "step": 28301
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.167248487472534,
+      "learning_rate": 5.873346922013008e-06,
+      "loss": 0.6251,
+      "step": 28302
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 4.674651622772217,
+      "learning_rate": 5.873087315801029e-06,
+      "loss": 0.6552,
+      "step": 28303
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.9745475053787231,
+      "learning_rate": 5.872827707161321e-06,
+      "loss": 0.5901,
+      "step": 28304
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.9387927055358887,
+      "learning_rate": 5.872568096094607e-06,
+      "loss": 0.7342,
+      "step": 28305
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.6631646156311035,
+      "learning_rate": 5.872308482601607e-06,
+      "loss": 0.7138,
+      "step": 28306
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 5.958928108215332,
+      "learning_rate": 5.872048866683045e-06,
+      "loss": 0.7036,
+      "step": 28307
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.1730730533599854,
+      "learning_rate": 5.871789248339641e-06,
+      "loss": 0.4979,
+      "step": 28308
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 4.000099182128906,
+      "learning_rate": 5.871529627572117e-06,
+      "loss": 0.753,
+      "step": 28309
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.155243396759033,
+      "learning_rate": 5.871270004381196e-06,
+      "loss": 0.7028,
+      "step": 28310
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.6470723152160645,
+      "learning_rate": 5.871010378767599e-06,
+      "loss": 0.6691,
+      "step": 28311
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.7881804704666138,
+      "learning_rate": 5.870750750732051e-06,
+      "loss": 0.6676,
+      "step": 28312
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.46873140335083,
+      "learning_rate": 5.870491120275269e-06,
+      "loss": 0.7606,
+      "step": 28313
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.5376880168914795,
+      "learning_rate": 5.870231487397978e-06,
+      "loss": 0.4239,
+      "step": 28314
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.6948838233947754,
+      "learning_rate": 5.869971852100898e-06,
+      "loss": 0.8909,
+      "step": 28315
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.5410206317901611,
+      "learning_rate": 5.869712214384754e-06,
+      "loss": 0.4944,
+      "step": 28316
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.641587257385254,
+      "learning_rate": 5.869452574250265e-06,
+      "loss": 0.6491,
+      "step": 28317
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.6268935203552246,
+      "learning_rate": 5.869192931698155e-06,
+      "loss": 0.3882,
+      "step": 28318
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.127018928527832,
+      "learning_rate": 5.868933286729145e-06,
+      "loss": 0.5144,
+      "step": 28319
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.0266976356506348,
+      "learning_rate": 5.8686736393439555e-06,
+      "loss": 0.7997,
+      "step": 28320
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.547235369682312,
+      "learning_rate": 5.868413989543313e-06,
+      "loss": 0.7019,
+      "step": 28321
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.8724987506866455,
+      "learning_rate": 5.868154337327934e-06,
+      "loss": 0.5418,
+      "step": 28322
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.705640196800232,
+      "learning_rate": 5.867894682698543e-06,
+      "loss": 0.6279,
+      "step": 28323
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.3296713829040527,
+      "learning_rate": 5.8676350256558625e-06,
+      "loss": 0.7128,
+      "step": 28324
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 4.165096759796143,
+      "learning_rate": 5.8673753662006135e-06,
+      "loss": 0.6552,
+      "step": 28325
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.1884961128234863,
+      "learning_rate": 5.86711570433352e-06,
+      "loss": 0.7252,
+      "step": 28326
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.8845266103744507,
+      "learning_rate": 5.866856040055302e-06,
+      "loss": 0.7926,
+      "step": 28327
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.165067672729492,
+      "learning_rate": 5.8665963733666816e-06,
+      "loss": 0.6025,
+      "step": 28328
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.839158296585083,
+      "learning_rate": 5.866336704268383e-06,
+      "loss": 0.6731,
+      "step": 28329
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.9484260082244873,
+      "learning_rate": 5.866077032761124e-06,
+      "loss": 0.7352,
+      "step": 28330
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.6755099296569824,
+      "learning_rate": 5.865817358845631e-06,
+      "loss": 0.5598,
+      "step": 28331
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.43434739112854,
+      "learning_rate": 5.865557682522623e-06,
+      "loss": 0.4439,
+      "step": 28332
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.799283266067505,
+      "learning_rate": 5.865298003792823e-06,
+      "loss": 0.3307,
+      "step": 28333
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.6080167293548584,
+      "learning_rate": 5.865038322656955e-06,
+      "loss": 0.712,
+      "step": 28334
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.6113650798797607,
+      "learning_rate": 5.864778639115739e-06,
+      "loss": 0.7249,
+      "step": 28335
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.9196301698684692,
+      "learning_rate": 5.864518953169898e-06,
+      "loss": 0.6564,
+      "step": 28336
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.216613531112671,
+      "learning_rate": 5.864259264820153e-06,
+      "loss": 0.4397,
+      "step": 28337
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.9441734552383423,
+      "learning_rate": 5.863999574067226e-06,
+      "loss": 0.5363,
+      "step": 28338
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.365039587020874,
+      "learning_rate": 5.863739880911842e-06,
+      "loss": 0.7667,
+      "step": 28339
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.232630491256714,
+      "learning_rate": 5.86348018535472e-06,
+      "loss": 0.836,
+      "step": 28340
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.230551242828369,
+      "learning_rate": 5.863220487396581e-06,
+      "loss": 0.7851,
+      "step": 28341
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.8539929389953613,
+      "learning_rate": 5.862960787038152e-06,
+      "loss": 0.6541,
+      "step": 28342
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.842347502708435,
+      "learning_rate": 5.86270108428015e-06,
+      "loss": 0.6858,
+      "step": 28343
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.424945831298828,
+      "learning_rate": 5.8624413791233e-06,
+      "loss": 0.7757,
+      "step": 28344
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.4811046123504639,
+      "learning_rate": 5.862181671568326e-06,
+      "loss": 0.572,
+      "step": 28345
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.270693778991699,
+      "learning_rate": 5.861921961615945e-06,
+      "loss": 0.5637,
+      "step": 28346
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.9544087648391724,
+      "learning_rate": 5.861662249266883e-06,
+      "loss": 0.7817,
+      "step": 28347
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.6616902351379395,
+      "learning_rate": 5.861402534521862e-06,
+      "loss": 0.5828,
+      "step": 28348
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.434229850769043,
+      "learning_rate": 5.861142817381602e-06,
+      "loss": 0.7461,
+      "step": 28349
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.3865745067596436,
+      "learning_rate": 5.860883097846826e-06,
+      "loss": 0.6651,
+      "step": 28350
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.8952546119689941,
+      "learning_rate": 5.8606233759182574e-06,
+      "loss": 0.5616,
+      "step": 28351
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.9839123487472534,
+      "learning_rate": 5.860363651596618e-06,
+      "loss": 0.5003,
+      "step": 28352
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.254795551300049,
+      "learning_rate": 5.860103924882627e-06,
+      "loss": 0.6162,
+      "step": 28353
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.782937526702881,
+      "learning_rate": 5.859844195777013e-06,
+      "loss": 0.577,
+      "step": 28354
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.7272707223892212,
+      "learning_rate": 5.859584464280493e-06,
+      "loss": 0.7657,
+      "step": 28355
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.786017656326294,
+      "learning_rate": 5.8593247303937895e-06,
+      "loss": 0.832,
+      "step": 28356
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 4.90329122543335,
+      "learning_rate": 5.859064994117626e-06,
+      "loss": 0.6704,
+      "step": 28357
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.051474094390869,
+      "learning_rate": 5.858805255452725e-06,
+      "loss": 0.5663,
+      "step": 28358
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.926238775253296,
+      "learning_rate": 5.858545514399809e-06,
+      "loss": 0.8188,
+      "step": 28359
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.6039180755615234,
+      "learning_rate": 5.858285770959599e-06,
+      "loss": 0.6224,
+      "step": 28360
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.0937111377716064,
+      "learning_rate": 5.858026025132818e-06,
+      "loss": 0.6768,
+      "step": 28361
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.656590223312378,
+      "learning_rate": 5.857766276920188e-06,
+      "loss": 0.4819,
+      "step": 28362
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.7503148317337036,
+      "learning_rate": 5.8575065263224294e-06,
+      "loss": 0.4904,
+      "step": 28363
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.7511842250823975,
+      "learning_rate": 5.857246773340268e-06,
+      "loss": 0.8088,
+      "step": 28364
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 4.622251510620117,
+      "learning_rate": 5.856987017974425e-06,
+      "loss": 0.4538,
+      "step": 28365
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.697838544845581,
+      "learning_rate": 5.856727260225622e-06,
+      "loss": 0.7307,
+      "step": 28366
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.40059494972229,
+      "learning_rate": 5.856467500094581e-06,
+      "loss": 0.701,
+      "step": 28367
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.0759313106536865,
+      "learning_rate": 5.856207737582025e-06,
+      "loss": 0.5839,
+      "step": 28368
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.0527184009552,
+      "learning_rate": 5.855947972688676e-06,
+      "loss": 0.675,
+      "step": 28369
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.5204718112945557,
+      "learning_rate": 5.855688205415256e-06,
+      "loss": 0.6047,
+      "step": 28370
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.380570411682129,
+      "learning_rate": 5.855428435762487e-06,
+      "loss": 0.662,
+      "step": 28371
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 6.620061874389648,
+      "learning_rate": 5.855168663731093e-06,
+      "loss": 0.6505,
+      "step": 28372
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.2770347595214844,
+      "learning_rate": 5.8549088893217954e-06,
+      "loss": 1.0091,
+      "step": 28373
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.7004520893096924,
+      "learning_rate": 5.8546491125353164e-06,
+      "loss": 0.4905,
+      "step": 28374
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.723818302154541,
+      "learning_rate": 5.854389333372378e-06,
+      "loss": 0.792,
+      "step": 28375
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.727696657180786,
+      "learning_rate": 5.854129551833703e-06,
+      "loss": 0.7611,
+      "step": 28376
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.3245294094085693,
+      "learning_rate": 5.8538697679200126e-06,
+      "loss": 0.8261,
+      "step": 28377
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 13.355100631713867,
+      "learning_rate": 5.853609981632031e-06,
+      "loss": 0.7003,
+      "step": 28378
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 4.036264419555664,
+      "learning_rate": 5.8533501929704815e-06,
+      "loss": 0.7319,
+      "step": 28379
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.2169196605682373,
+      "learning_rate": 5.853090401936083e-06,
+      "loss": 0.6728,
+      "step": 28380
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 5.100291728973389,
+      "learning_rate": 5.852830608529561e-06,
+      "loss": 0.5555,
+      "step": 28381
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.7934865951538086,
+      "learning_rate": 5.852570812751636e-06,
+      "loss": 0.631,
+      "step": 28382
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.2051520347595215,
+      "learning_rate": 5.852311014603029e-06,
+      "loss": 0.6131,
+      "step": 28383
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.0605628490448,
+      "learning_rate": 5.852051214084466e-06,
+      "loss": 0.6261,
+      "step": 28384
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.7439823150634766,
+      "learning_rate": 5.851791411196668e-06,
+      "loss": 0.568,
+      "step": 28385
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.1139023303985596,
+      "learning_rate": 5.851531605940357e-06,
+      "loss": 0.554,
+      "step": 28386
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.967583417892456,
+      "learning_rate": 5.851271798316256e-06,
+      "loss": 0.6066,
+      "step": 28387
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.049600839614868,
+      "learning_rate": 5.8510119883250856e-06,
+      "loss": 0.6705,
+      "step": 28388
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.183821201324463,
+      "learning_rate": 5.8507521759675715e-06,
+      "loss": 0.58,
+      "step": 28389
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.460764169692993,
+      "learning_rate": 5.850492361244433e-06,
+      "loss": 0.68,
+      "step": 28390
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.2696828842163086,
+      "learning_rate": 5.850232544156394e-06,
+      "loss": 0.7403,
+      "step": 28391
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.849003553390503,
+      "learning_rate": 5.849972724704177e-06,
+      "loss": 0.6391,
+      "step": 28392
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.384674310684204,
+      "learning_rate": 5.849712902888505e-06,
+      "loss": 0.4419,
+      "step": 28393
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.198749303817749,
+      "learning_rate": 5.849453078710098e-06,
+      "loss": 0.7111,
+      "step": 28394
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.111375570297241,
+      "learning_rate": 5.849193252169681e-06,
+      "loss": 0.7174,
+      "step": 28395
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.213700532913208,
+      "learning_rate": 5.8489334232679765e-06,
+      "loss": 0.6723,
+      "step": 28396
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.6414177417755127,
+      "learning_rate": 5.848673592005706e-06,
+      "loss": 0.6951,
+      "step": 28397
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.1910552978515625,
+      "learning_rate": 5.848413758383591e-06,
+      "loss": 0.7471,
+      "step": 28398
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.465717315673828,
+      "learning_rate": 5.8481539224023564e-06,
+      "loss": 0.7803,
+      "step": 28399
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.9246302843093872,
+      "learning_rate": 5.847894084062723e-06,
+      "loss": 0.6512,
+      "step": 28400
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.8590658903121948,
+      "learning_rate": 5.847634243365415e-06,
+      "loss": 0.6044,
+      "step": 28401
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.4479036331176758,
+      "learning_rate": 5.847374400311151e-06,
+      "loss": 0.8243,
+      "step": 28402
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.6243071556091309,
+      "learning_rate": 5.847114554900658e-06,
+      "loss": 0.5809,
+      "step": 28403
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.0935964584350586,
+      "learning_rate": 5.8468547071346545e-06,
+      "loss": 0.7992,
+      "step": 28404
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.457632303237915,
+      "learning_rate": 5.846594857013869e-06,
+      "loss": 0.3771,
+      "step": 28405
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.9238226413726807,
+      "learning_rate": 5.846335004539018e-06,
+      "loss": 0.4976,
+      "step": 28406
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.3885538578033447,
+      "learning_rate": 5.846075149710827e-06,
+      "loss": 0.5775,
+      "step": 28407
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.0449721813201904,
+      "learning_rate": 5.845815292530018e-06,
+      "loss": 0.6007,
+      "step": 28408
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.9424718618392944,
+      "learning_rate": 5.845555432997314e-06,
+      "loss": 0.8154,
+      "step": 28409
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.1378986835479736,
+      "learning_rate": 5.845295571113436e-06,
+      "loss": 0.7217,
+      "step": 28410
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.605128765106201,
+      "learning_rate": 5.845035706879109e-06,
+      "loss": 0.7972,
+      "step": 28411
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.7631263732910156,
+      "learning_rate": 5.844775840295053e-06,
+      "loss": 0.6115,
+      "step": 28412
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.069927930831909,
+      "learning_rate": 5.844515971361992e-06,
+      "loss": 0.7158,
+      "step": 28413
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.245826005935669,
+      "learning_rate": 5.844256100080648e-06,
+      "loss": 0.6191,
+      "step": 28414
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.602856159210205,
+      "learning_rate": 5.843996226451745e-06,
+      "loss": 0.3643,
+      "step": 28415
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.004500389099121,
+      "learning_rate": 5.843736350476005e-06,
+      "loss": 0.6337,
+      "step": 28416
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.458523988723755,
+      "learning_rate": 5.843476472154149e-06,
+      "loss": 0.7731,
+      "step": 28417
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.0568947792053223,
+      "learning_rate": 5.8432165914869024e-06,
+      "loss": 0.6631,
+      "step": 28418
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.2806856632232666,
+      "learning_rate": 5.842956708474986e-06,
+      "loss": 0.5995,
+      "step": 28419
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.081364393234253,
+      "learning_rate": 5.842696823119122e-06,
+      "loss": 0.7546,
+      "step": 28420
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.061026096343994,
+      "learning_rate": 5.842436935420035e-06,
+      "loss": 0.4599,
+      "step": 28421
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.0313854217529297,
+      "learning_rate": 5.842177045378445e-06,
+      "loss": 0.8012,
+      "step": 28422
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.1125173568725586,
+      "learning_rate": 5.841917152995077e-06,
+      "loss": 0.5663,
+      "step": 28423
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.0184760093688965,
+      "learning_rate": 5.8416572582706525e-06,
+      "loss": 0.5934,
+      "step": 28424
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.6069459915161133,
+      "learning_rate": 5.841397361205894e-06,
+      "loss": 0.5966,
+      "step": 28425
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.8383647203445435,
+      "learning_rate": 5.8411374618015245e-06,
+      "loss": 0.6458,
+      "step": 28426
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.7743964195251465,
+      "learning_rate": 5.840877560058268e-06,
+      "loss": 0.6298,
+      "step": 28427
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.890499472618103,
+      "learning_rate": 5.840617655976846e-06,
+      "loss": 0.6941,
+      "step": 28428
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.489107131958008,
+      "learning_rate": 5.84035774955798e-06,
+      "loss": 0.6481,
+      "step": 28429
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.8429834842681885,
+      "learning_rate": 5.840097840802394e-06,
+      "loss": 0.2884,
+      "step": 28430
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.0640869140625,
+      "learning_rate": 5.8398379297108125e-06,
+      "loss": 0.6937,
+      "step": 28431
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.260617733001709,
+      "learning_rate": 5.839578016283956e-06,
+      "loss": 0.8273,
+      "step": 28432
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.43496835231781,
+      "learning_rate": 5.839318100522544e-06,
+      "loss": 0.5033,
+      "step": 28433
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.4704068899154663,
+      "learning_rate": 5.839058182427306e-06,
+      "loss": 0.5398,
+      "step": 28434
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.5422654151916504,
+      "learning_rate": 5.838798261998961e-06,
+      "loss": 0.4059,
+      "step": 28435
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.561793088912964,
+      "learning_rate": 5.838538339238233e-06,
+      "loss": 0.5014,
+      "step": 28436
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.9596902132034302,
+      "learning_rate": 5.8382784141458425e-06,
+      "loss": 0.5694,
+      "step": 28437
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.0733108520507812,
+      "learning_rate": 5.838018486722514e-06,
+      "loss": 0.4757,
+      "step": 28438
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.34797739982605,
+      "learning_rate": 5.8377585569689695e-06,
+      "loss": 0.4567,
+      "step": 28439
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.6559789180755615,
+      "learning_rate": 5.8374986248859335e-06,
+      "loss": 0.7364,
+      "step": 28440
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.7332446575164795,
+      "learning_rate": 5.837238690474128e-06,
+      "loss": 0.4167,
+      "step": 28441
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.1301443576812744,
+      "learning_rate": 5.836978753734274e-06,
+      "loss": 0.7772,
+      "step": 28442
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.2870841026306152,
+      "learning_rate": 5.836718814667096e-06,
+      "loss": 0.7847,
+      "step": 28443
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.3170876502990723,
+      "learning_rate": 5.836458873273317e-06,
+      "loss": 0.7284,
+      "step": 28444
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.5345768928527832,
+      "learning_rate": 5.836198929553658e-06,
+      "loss": 0.6257,
+      "step": 28445
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.7551987171173096,
+      "learning_rate": 5.835938983508844e-06,
+      "loss": 1.0718,
+      "step": 28446
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.8630839586257935,
+      "learning_rate": 5.835679035139596e-06,
+      "loss": 0.5859,
+      "step": 28447
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.0546510219573975,
+      "learning_rate": 5.835419084446638e-06,
+      "loss": 0.8616,
+      "step": 28448
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.5591485500335693,
+      "learning_rate": 5.835159131430694e-06,
+      "loss": 0.982,
+      "step": 28449
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.206322431564331,
+      "learning_rate": 5.834899176092484e-06,
+      "loss": 0.6399,
+      "step": 28450
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.600363850593567,
+      "learning_rate": 5.834639218432733e-06,
+      "loss": 0.6732,
+      "step": 28451
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 5.4296369552612305,
+      "learning_rate": 5.834379258452162e-06,
+      "loss": 0.8352,
+      "step": 28452
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.4020423889160156,
+      "learning_rate": 5.834119296151495e-06,
+      "loss": 0.5948,
+      "step": 28453
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.321554183959961,
+      "learning_rate": 5.833859331531455e-06,
+      "loss": 0.5874,
+      "step": 28454
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 6.514938831329346,
+      "learning_rate": 5.833599364592763e-06,
+      "loss": 0.5925,
+      "step": 28455
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.368455648422241,
+      "learning_rate": 5.8333393953361454e-06,
+      "loss": 0.4787,
+      "step": 28456
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.886540412902832,
+      "learning_rate": 5.8330794237623226e-06,
+      "loss": 0.7367,
+      "step": 28457
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.67358136177063,
+      "learning_rate": 5.832819449872018e-06,
+      "loss": 0.6109,
+      "step": 28458
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.930074453353882,
+      "learning_rate": 5.832559473665955e-06,
+      "loss": 0.8103,
+      "step": 28459
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 4.064944744110107,
+      "learning_rate": 5.832299495144855e-06,
+      "loss": 0.6806,
+      "step": 28460
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.1104300022125244,
+      "learning_rate": 5.8320395143094425e-06,
+      "loss": 0.6591,
+      "step": 28461
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.499600648880005,
+      "learning_rate": 5.83177953116044e-06,
+      "loss": 0.8345,
+      "step": 28462
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.249183177947998,
+      "learning_rate": 5.831519545698569e-06,
+      "loss": 0.4414,
+      "step": 28463
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 4.413694381713867,
+      "learning_rate": 5.831259557924556e-06,
+      "loss": 0.5723,
+      "step": 28464
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.6188724040985107,
+      "learning_rate": 5.830999567839119e-06,
+      "loss": 0.4751,
+      "step": 28465
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.4165210723876953,
+      "learning_rate": 5.830739575442984e-06,
+      "loss": 0.7325,
+      "step": 28466
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.9852067232131958,
+      "learning_rate": 5.830479580736874e-06,
+      "loss": 0.7962,
+      "step": 28467
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.390383243560791,
+      "learning_rate": 5.830219583721511e-06,
+      "loss": 0.7327,
+      "step": 28468
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.0055031776428223,
+      "learning_rate": 5.82995958439762e-06,
+      "loss": 0.3721,
+      "step": 28469
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.212203025817871,
+      "learning_rate": 5.829699582765921e-06,
+      "loss": 0.6974,
+      "step": 28470
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.757850408554077,
+      "learning_rate": 5.8294395788271385e-06,
+      "loss": 0.6376,
+      "step": 28471
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.714653491973877,
+      "learning_rate": 5.8291795725819955e-06,
+      "loss": 0.7742,
+      "step": 28472
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.7893552780151367,
+      "learning_rate": 5.828919564031214e-06,
+      "loss": 0.6197,
+      "step": 28473
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.6983013153076172,
+      "learning_rate": 5.828659553175518e-06,
+      "loss": 0.4579,
+      "step": 28474
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.2501329183578491,
+      "learning_rate": 5.828399540015629e-06,
+      "loss": 0.6283,
+      "step": 28475
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.0893704891204834,
+      "learning_rate": 5.828139524552272e-06,
+      "loss": 0.6815,
+      "step": 28476
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.7555420398712158,
+      "learning_rate": 5.827879506786171e-06,
+      "loss": 0.6591,
+      "step": 28477
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.778977394104004,
+      "learning_rate": 5.8276194867180456e-06,
+      "loss": 0.8245,
+      "step": 28478
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 5.05515718460083,
+      "learning_rate": 5.827359464348621e-06,
+      "loss": 0.7959,
+      "step": 28479
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.9307817220687866,
+      "learning_rate": 5.827099439678618e-06,
+      "loss": 0.6691,
+      "step": 28480
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.7718225717544556,
+      "learning_rate": 5.826839412708763e-06,
+      "loss": 0.5545,
+      "step": 28481
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.9213106632232666,
+      "learning_rate": 5.8265793834397775e-06,
+      "loss": 0.3199,
+      "step": 28482
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.515392541885376,
+      "learning_rate": 5.826319351872385e-06,
+      "loss": 0.8444,
+      "step": 28483
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.676173210144043,
+      "learning_rate": 5.8260593180073065e-06,
+      "loss": 0.5888,
+      "step": 28484
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.310798406600952,
+      "learning_rate": 5.825799281845266e-06,
+      "loss": 0.6962,
+      "step": 28485
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.2344424724578857,
+      "learning_rate": 5.825539243386987e-06,
+      "loss": 0.8027,
+      "step": 28486
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.4531328678131104,
+      "learning_rate": 5.825279202633194e-06,
+      "loss": 0.7458,
+      "step": 28487
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.104013442993164,
+      "learning_rate": 5.8250191595846084e-06,
+      "loss": 0.8079,
+      "step": 28488
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.639909029006958,
+      "learning_rate": 5.8247591142419535e-06,
+      "loss": 0.7882,
+      "step": 28489
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.7537753582000732,
+      "learning_rate": 5.824499066605952e-06,
+      "loss": 0.6932,
+      "step": 28490
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.1143410205841064,
+      "learning_rate": 5.8242390166773275e-06,
+      "loss": 0.581,
+      "step": 28491
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.5835883617401123,
+      "learning_rate": 5.823978964456803e-06,
+      "loss": 0.5376,
+      "step": 28492
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.7310571670532227,
+      "learning_rate": 5.823718909945103e-06,
+      "loss": 0.7496,
+      "step": 28493
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.020737886428833,
+      "learning_rate": 5.823458853142947e-06,
+      "loss": 0.7327,
+      "step": 28494
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.8471834659576416,
+      "learning_rate": 5.823198794051062e-06,
+      "loss": 0.5384,
+      "step": 28495
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.1570932865142822,
+      "learning_rate": 5.822938732670169e-06,
+      "loss": 0.5772,
+      "step": 28496
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.082348585128784,
+      "learning_rate": 5.822678669000992e-06,
+      "loss": 0.6131,
+      "step": 28497
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.6702169179916382,
+      "learning_rate": 5.822418603044254e-06,
+      "loss": 0.6776,
+      "step": 28498
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.333129644393921,
+      "learning_rate": 5.822158534800677e-06,
+      "loss": 0.6401,
+      "step": 28499
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.2200522422790527,
+      "learning_rate": 5.821898464270985e-06,
+      "loss": 0.6211,
+      "step": 28500
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.8992737531661987,
+      "learning_rate": 5.821638391455903e-06,
+      "loss": 0.7786,
+      "step": 28501
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.2623268365859985,
+      "learning_rate": 5.821378316356151e-06,
+      "loss": 0.4776,
+      "step": 28502
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.8820788860321045,
+      "learning_rate": 5.821118238972454e-06,
+      "loss": 0.6524,
+      "step": 28503
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.9158499240875244,
+      "learning_rate": 5.820858159305535e-06,
+      "loss": 0.7894,
+      "step": 28504
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.6742210388183594,
+      "learning_rate": 5.820598077356117e-06,
+      "loss": 0.5217,
+      "step": 28505
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.57698392868042,
+      "learning_rate": 5.820337993124921e-06,
+      "loss": 0.5752,
+      "step": 28506
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.7665321826934814,
+      "learning_rate": 5.820077906612675e-06,
+      "loss": 0.7473,
+      "step": 28507
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.3665287494659424,
+      "learning_rate": 5.819817817820099e-06,
+      "loss": 0.7989,
+      "step": 28508
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.1719815731048584,
+      "learning_rate": 5.819557726747916e-06,
+      "loss": 0.735,
+      "step": 28509
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.944205641746521,
+      "learning_rate": 5.819297633396851e-06,
+      "loss": 0.8433,
+      "step": 28510
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.9481438398361206,
+      "learning_rate": 5.819037537767626e-06,
+      "loss": 0.5017,
+      "step": 28511
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.445653200149536,
+      "learning_rate": 5.818777439860964e-06,
+      "loss": 0.8246,
+      "step": 28512
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.7012221813201904,
+      "learning_rate": 5.818517339677589e-06,
+      "loss": 0.4946,
+      "step": 28513
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 4.431755542755127,
+      "learning_rate": 5.818257237218223e-06,
+      "loss": 0.8267,
+      "step": 28514
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.734921932220459,
+      "learning_rate": 5.817997132483592e-06,
+      "loss": 0.7106,
+      "step": 28515
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.0862221717834473,
+      "learning_rate": 5.817737025474415e-06,
+      "loss": 0.4725,
+      "step": 28516
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.8555092811584473,
+      "learning_rate": 5.817476916191419e-06,
+      "loss": 0.7106,
+      "step": 28517
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 5.373317718505859,
+      "learning_rate": 5.817216804635326e-06,
+      "loss": 0.8223,
+      "step": 28518
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.361360549926758,
+      "learning_rate": 5.816956690806859e-06,
+      "loss": 0.5121,
+      "step": 28519
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.7306275367736816,
+      "learning_rate": 5.816696574706742e-06,
+      "loss": 0.7058,
+      "step": 28520
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.9396668672561646,
+      "learning_rate": 5.816436456335697e-06,
+      "loss": 0.6667,
+      "step": 28521
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.953558325767517,
+      "learning_rate": 5.816176335694449e-06,
+      "loss": 0.6031,
+      "step": 28522
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 11.970355033874512,
+      "learning_rate": 5.8159162127837185e-06,
+      "loss": 0.6336,
+      "step": 28523
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.520869731903076,
+      "learning_rate": 5.815656087604233e-06,
+      "loss": 0.5087,
+      "step": 28524
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.570794105529785,
+      "learning_rate": 5.815395960156712e-06,
+      "loss": 0.8424,
+      "step": 28525
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.4712979793548584,
+      "learning_rate": 5.81513583044188e-06,
+      "loss": 0.6634,
+      "step": 28526
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.9493211507797241,
+      "learning_rate": 5.814875698460461e-06,
+      "loss": 0.8237,
+      "step": 28527
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.455066680908203,
+      "learning_rate": 5.814615564213178e-06,
+      "loss": 0.6561,
+      "step": 28528
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.6648430824279785,
+      "learning_rate": 5.814355427700755e-06,
+      "loss": 0.7756,
+      "step": 28529
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.5665717124938965,
+      "learning_rate": 5.8140952889239145e-06,
+      "loss": 0.6102,
+      "step": 28530
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.4740283489227295,
+      "learning_rate": 5.81383514788338e-06,
+      "loss": 0.6099,
+      "step": 28531
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.7688868045806885,
+      "learning_rate": 5.813575004579875e-06,
+      "loss": 0.6218,
+      "step": 28532
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.149289131164551,
+      "learning_rate": 5.8133148590141206e-06,
+      "loss": 0.8221,
+      "step": 28533
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.8924729824066162,
+      "learning_rate": 5.8130547111868445e-06,
+      "loss": 0.7213,
+      "step": 28534
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.7247036695480347,
+      "learning_rate": 5.812794561098767e-06,
+      "loss": 0.4792,
+      "step": 28535
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.5998828411102295,
+      "learning_rate": 5.812534408750611e-06,
+      "loss": 0.6699,
+      "step": 28536
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.229980707168579,
+      "learning_rate": 5.812274254143103e-06,
+      "loss": 0.5962,
+      "step": 28537
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.3941643238067627,
+      "learning_rate": 5.812014097276965e-06,
+      "loss": 0.5985,
+      "step": 28538
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.7331814765930176,
+      "learning_rate": 5.811753938152919e-06,
+      "loss": 0.6445,
+      "step": 28539
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.241990089416504,
+      "learning_rate": 5.811493776771688e-06,
+      "loss": 0.6954,
+      "step": 28540
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 5.798827171325684,
+      "learning_rate": 5.811233613134e-06,
+      "loss": 0.837,
+      "step": 28541
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.591458559036255,
+      "learning_rate": 5.810973447240573e-06,
+      "loss": 0.7988,
+      "step": 28542
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.4281535148620605,
+      "learning_rate": 5.810713279092133e-06,
+      "loss": 0.6556,
+      "step": 28543
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.2856132984161377,
+      "learning_rate": 5.810453108689404e-06,
+      "loss": 0.5083,
+      "step": 28544
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.854532480239868,
+      "learning_rate": 5.810192936033108e-06,
+      "loss": 0.5116,
+      "step": 28545
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.4754834175109863,
+      "learning_rate": 5.8099327611239664e-06,
+      "loss": 0.774,
+      "step": 28546
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.961061954498291,
+      "learning_rate": 5.809672583962709e-06,
+      "loss": 0.5881,
+      "step": 28547
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.264979839324951,
+      "learning_rate": 5.809412404550053e-06,
+      "loss": 0.5358,
+      "step": 28548
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.7059029340744019,
+      "learning_rate": 5.809152222886726e-06,
+      "loss": 0.5012,
+      "step": 28549
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.133824586868286,
+      "learning_rate": 5.808892038973448e-06,
+      "loss": 0.7414,
+      "step": 28550
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 4.353518009185791,
+      "learning_rate": 5.8086318528109465e-06,
+      "loss": 0.8104,
+      "step": 28551
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.403362989425659,
+      "learning_rate": 5.808371664399941e-06,
+      "loss": 0.5027,
+      "step": 28552
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.0152735710144043,
+      "learning_rate": 5.808111473741158e-06,
+      "loss": 0.6255,
+      "step": 28553
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 4.755702972412109,
+      "learning_rate": 5.807851280835318e-06,
+      "loss": 0.6672,
+      "step": 28554
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.6927998065948486,
+      "learning_rate": 5.807591085683149e-06,
+      "loss": 0.5376,
+      "step": 28555
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.4350225925445557,
+      "learning_rate": 5.807330888285368e-06,
+      "loss": 0.6263,
+      "step": 28556
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.7442561388015747,
+      "learning_rate": 5.807070688642704e-06,
+      "loss": 0.5875,
+      "step": 28557
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.547476291656494,
+      "learning_rate": 5.80681048675588e-06,
+      "loss": 0.8159,
+      "step": 28558
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.8513787984848022,
+      "learning_rate": 5.806550282625616e-06,
+      "loss": 0.5196,
+      "step": 28559
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.9875454902648926,
+      "learning_rate": 5.80629007625264e-06,
+      "loss": 0.6569,
+      "step": 28560
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.0518417358398438,
+      "learning_rate": 5.806029867637672e-06,
+      "loss": 0.7248,
+      "step": 28561
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.6620484590530396,
+      "learning_rate": 5.805769656781436e-06,
+      "loss": 0.6325,
+      "step": 28562
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.3818471431732178,
+      "learning_rate": 5.805509443684658e-06,
+      "loss": 0.7444,
+      "step": 28563
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 3.47998309135437,
+      "learning_rate": 5.80524922834806e-06,
+      "loss": 0.7809,
+      "step": 28564
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.8483293056488037,
+      "learning_rate": 5.804989010772365e-06,
+      "loss": 0.7312,
+      "step": 28565
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.8431708812713623,
+      "learning_rate": 5.804728790958298e-06,
+      "loss": 0.6274,
+      "step": 28566
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 4.138050079345703,
+      "learning_rate": 5.804468568906581e-06,
+      "loss": 0.8665,
+      "step": 28567
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.591524362564087,
+      "learning_rate": 5.804208344617937e-06,
+      "loss": 0.6112,
+      "step": 28568
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.1746551990509033,
+      "learning_rate": 5.803948118093091e-06,
+      "loss": 0.6544,
+      "step": 28569
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 3.326355218887329,
+      "learning_rate": 5.803687889332768e-06,
+      "loss": 0.6734,
+      "step": 28570
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.3685803413391113,
+      "learning_rate": 5.80342765833769e-06,
+      "loss": 0.6603,
+      "step": 28571
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.735572576522827,
+      "learning_rate": 5.80316742510858e-06,
+      "loss": 0.7767,
+      "step": 28572
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.7430170774459839,
+      "learning_rate": 5.802907189646163e-06,
+      "loss": 0.5941,
+      "step": 28573
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 3.672227621078491,
+      "learning_rate": 5.802646951951161e-06,
+      "loss": 0.7614,
+      "step": 28574
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.7503933906555176,
+      "learning_rate": 5.8023867120242995e-06,
+      "loss": 0.7155,
+      "step": 28575
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.2187798023223877,
+      "learning_rate": 5.8021264698663005e-06,
+      "loss": 0.6491,
+      "step": 28576
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.040957450866699,
+      "learning_rate": 5.801866225477888e-06,
+      "loss": 0.5391,
+      "step": 28577
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.0123767852783203,
+      "learning_rate": 5.801605978859786e-06,
+      "loss": 0.6417,
+      "step": 28578
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.891301393508911,
+      "learning_rate": 5.8013457300127185e-06,
+      "loss": 0.7426,
+      "step": 28579
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.019512176513672,
+      "learning_rate": 5.801085478937408e-06,
+      "loss": 0.7119,
+      "step": 28580
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 7.9863104820251465,
+      "learning_rate": 5.80082522563458e-06,
+      "loss": 0.8258,
+      "step": 28581
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.8583463430404663,
+      "learning_rate": 5.800564970104956e-06,
+      "loss": 0.613,
+      "step": 28582
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.7550872564315796,
+      "learning_rate": 5.800304712349262e-06,
+      "loss": 0.4294,
+      "step": 28583
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.8936922550201416,
+      "learning_rate": 5.800044452368219e-06,
+      "loss": 0.8565,
+      "step": 28584
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.756882667541504,
+      "learning_rate": 5.799784190162553e-06,
+      "loss": 0.576,
+      "step": 28585
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.978986144065857,
+      "learning_rate": 5.799523925732986e-06,
+      "loss": 0.447,
+      "step": 28586
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 3.5750162601470947,
+      "learning_rate": 5.799263659080242e-06,
+      "loss": 0.7666,
+      "step": 28587
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.469491481781006,
+      "learning_rate": 5.799003390205046e-06,
+      "loss": 0.8506,
+      "step": 28588
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.228280544281006,
+      "learning_rate": 5.7987431191081215e-06,
+      "loss": 0.4849,
+      "step": 28589
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.1351945400238037,
+      "learning_rate": 5.798482845790191e-06,
+      "loss": 0.4288,
+      "step": 28590
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 3.379237651824951,
+      "learning_rate": 5.798222570251979e-06,
+      "loss": 0.6323,
+      "step": 28591
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.0450632572174072,
+      "learning_rate": 5.797962292494209e-06,
+      "loss": 0.621,
+      "step": 28592
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.7976361513137817,
+      "learning_rate": 5.7977020125176046e-06,
+      "loss": 0.6366,
+      "step": 28593
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.5328569412231445,
+      "learning_rate": 5.79744173032289e-06,
+      "loss": 0.4532,
+      "step": 28594
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.1465930938720703,
+      "learning_rate": 5.797181445910789e-06,
+      "loss": 0.5597,
+      "step": 28595
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 3.495668888092041,
+      "learning_rate": 5.796921159282024e-06,
+      "loss": 0.6317,
+      "step": 28596
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.1820180416107178,
+      "learning_rate": 5.796660870437319e-06,
+      "loss": 0.6847,
+      "step": 28597
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.6203243732452393,
+      "learning_rate": 5.7964005793774e-06,
+      "loss": 0.5984,
+      "step": 28598
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.7267367839813232,
+      "learning_rate": 5.796140286102989e-06,
+      "loss": 0.6313,
+      "step": 28599
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.5782697200775146,
+      "learning_rate": 5.795879990614811e-06,
+      "loss": 0.6213,
+      "step": 28600
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.9182926416397095,
+      "learning_rate": 5.795619692913588e-06,
+      "loss": 0.7715,
+      "step": 28601
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.883681058883667,
+      "learning_rate": 5.795359393000045e-06,
+      "loss": 0.7192,
+      "step": 28602
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.8988962173461914,
+      "learning_rate": 5.795099090874906e-06,
+      "loss": 0.4811,
+      "step": 28603
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.6298903226852417,
+      "learning_rate": 5.794838786538893e-06,
+      "loss": 0.7507,
+      "step": 28604
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.7433066368103027,
+      "learning_rate": 5.794578479992732e-06,
+      "loss": 0.377,
+      "step": 28605
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 4.241661071777344,
+      "learning_rate": 5.794318171237145e-06,
+      "loss": 0.8636,
+      "step": 28606
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.8443775177001953,
+      "learning_rate": 5.794057860272857e-06,
+      "loss": 0.5988,
+      "step": 28607
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 3.276449680328369,
+      "learning_rate": 5.793797547100593e-06,
+      "loss": 0.763,
+      "step": 28608
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.101226568222046,
+      "learning_rate": 5.793537231721074e-06,
+      "loss": 0.5736,
+      "step": 28609
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.6500797271728516,
+      "learning_rate": 5.7932769141350255e-06,
+      "loss": 0.6922,
+      "step": 28610
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 11.67391586303711,
+      "learning_rate": 5.7930165943431704e-06,
+      "loss": 0.7746,
+      "step": 28611
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.7078094482421875,
+      "learning_rate": 5.792756272346234e-06,
+      "loss": 0.8609,
+      "step": 28612
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.124131917953491,
+      "learning_rate": 5.792495948144939e-06,
+      "loss": 0.5622,
+      "step": 28613
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.6213788986206055,
+      "learning_rate": 5.792235621740011e-06,
+      "loss": 0.7412,
+      "step": 28614
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.4559872150421143,
+      "learning_rate": 5.791975293132172e-06,
+      "loss": 0.7293,
+      "step": 28615
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 3.240961790084839,
+      "learning_rate": 5.791714962322146e-06,
+      "loss": 0.6744,
+      "step": 28616
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.3452630043029785,
+      "learning_rate": 5.791454629310657e-06,
+      "loss": 0.5042,
+      "step": 28617
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.8440186977386475,
+      "learning_rate": 5.791194294098428e-06,
+      "loss": 0.4317,
+      "step": 28618
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 4.46062707901001,
+      "learning_rate": 5.7909339566861855e-06,
+      "loss": 0.5855,
+      "step": 28619
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.875528573989868,
+      "learning_rate": 5.790673617074651e-06,
+      "loss": 0.4226,
+      "step": 28620
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.5526378154754639,
+      "learning_rate": 5.79041327526455e-06,
+      "loss": 0.5235,
+      "step": 28621
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.473112940788269,
+      "learning_rate": 5.790152931256607e-06,
+      "loss": 0.7326,
+      "step": 28622
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 7.303375720977783,
+      "learning_rate": 5.789892585051543e-06,
+      "loss": 0.5902,
+      "step": 28623
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.171266794204712,
+      "learning_rate": 5.7896322366500834e-06,
+      "loss": 0.5432,
+      "step": 28624
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.3617217540740967,
+      "learning_rate": 5.789371886052954e-06,
+      "loss": 0.5829,
+      "step": 28625
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.7861496210098267,
+      "learning_rate": 5.789111533260876e-06,
+      "loss": 0.6595,
+      "step": 28626
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.192931890487671,
+      "learning_rate": 5.7888511782745745e-06,
+      "loss": 0.7896,
+      "step": 28627
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 3.2356910705566406,
+      "learning_rate": 5.788590821094774e-06,
+      "loss": 0.7249,
+      "step": 28628
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.6703373193740845,
+      "learning_rate": 5.788330461722196e-06,
+      "loss": 0.5297,
+      "step": 28629
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.34804368019104,
+      "learning_rate": 5.788070100157568e-06,
+      "loss": 0.6637,
+      "step": 28630
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.69311785697937,
+      "learning_rate": 5.787809736401612e-06,
+      "loss": 0.6324,
+      "step": 28631
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.009653091430664,
+      "learning_rate": 5.787549370455053e-06,
+      "loss": 0.6107,
+      "step": 28632
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.741654634475708,
+      "learning_rate": 5.787289002318613e-06,
+      "loss": 0.6126,
+      "step": 28633
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.162665367126465,
+      "learning_rate": 5.787028631993018e-06,
+      "loss": 0.7181,
+      "step": 28634
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.955960750579834,
+      "learning_rate": 5.786768259478991e-06,
+      "loss": 0.728,
+      "step": 28635
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.76081120967865,
+      "learning_rate": 5.786507884777256e-06,
+      "loss": 0.4298,
+      "step": 28636
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 4.725941181182861,
+      "learning_rate": 5.786247507888537e-06,
+      "loss": 0.5234,
+      "step": 28637
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.4468579292297363,
+      "learning_rate": 5.785987128813558e-06,
+      "loss": 0.7771,
+      "step": 28638
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.5565900802612305,
+      "learning_rate": 5.785726747553043e-06,
+      "loss": 0.4443,
+      "step": 28639
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.0091235637664795,
+      "learning_rate": 5.785466364107718e-06,
+      "loss": 0.6038,
+      "step": 28640
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.7043323516845703,
+      "learning_rate": 5.785205978478305e-06,
+      "loss": 0.7115,
+      "step": 28641
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 6.447793006896973,
+      "learning_rate": 5.784945590665527e-06,
+      "loss": 0.7604,
+      "step": 28642
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.6194705963134766,
+      "learning_rate": 5.78468520067011e-06,
+      "loss": 0.6354,
+      "step": 28643
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.6783292293548584,
+      "learning_rate": 5.784424808492777e-06,
+      "loss": 0.5548,
+      "step": 28644
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 4.647036075592041,
+      "learning_rate": 5.784164414134253e-06,
+      "loss": 0.523,
+      "step": 28645
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.4135515689849854,
+      "learning_rate": 5.783904017595262e-06,
+      "loss": 0.637,
+      "step": 28646
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.5862605571746826,
+      "learning_rate": 5.783643618876526e-06,
+      "loss": 0.882,
+      "step": 28647
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.4008393287658691,
+      "learning_rate": 5.783383217978771e-06,
+      "loss": 0.6921,
+      "step": 28648
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.138667106628418,
+      "learning_rate": 5.783122814902722e-06,
+      "loss": 0.6513,
+      "step": 28649
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.5656075477600098,
+      "learning_rate": 5.7828624096491005e-06,
+      "loss": 0.7052,
+      "step": 28650
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.2453949451446533,
+      "learning_rate": 5.782602002218633e-06,
+      "loss": 0.6831,
+      "step": 28651
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.498121500015259,
+      "learning_rate": 5.782341592612042e-06,
+      "loss": 0.6175,
+      "step": 28652
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 3.125225782394409,
+      "learning_rate": 5.7820811808300525e-06,
+      "loss": 0.5904,
+      "step": 28653
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.839383840560913,
+      "learning_rate": 5.781820766873387e-06,
+      "loss": 0.6071,
+      "step": 28654
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.4583239555358887,
+      "learning_rate": 5.781560350742772e-06,
+      "loss": 0.7272,
+      "step": 28655
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 3.293275833129883,
+      "learning_rate": 5.781299932438931e-06,
+      "loss": 0.7038,
+      "step": 28656
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.852008581161499,
+      "learning_rate": 5.781039511962586e-06,
+      "loss": 0.7615,
+      "step": 28657
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.3204874992370605,
+      "learning_rate": 5.780779089314463e-06,
+      "loss": 0.5954,
+      "step": 28658
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.653970956802368,
+      "learning_rate": 5.780518664495286e-06,
+      "loss": 0.7124,
+      "step": 28659
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 3.5750930309295654,
+      "learning_rate": 5.780258237505779e-06,
+      "loss": 0.7674,
+      "step": 28660
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 5.303805351257324,
+      "learning_rate": 5.779997808346667e-06,
+      "loss": 0.5089,
+      "step": 28661
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.8700735569000244,
+      "learning_rate": 5.779737377018672e-06,
+      "loss": 0.6544,
+      "step": 28662
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 3.7218997478485107,
+      "learning_rate": 5.7794769435225204e-06,
+      "loss": 0.7703,
+      "step": 28663
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.8742640018463135,
+      "learning_rate": 5.779216507858936e-06,
+      "loss": 0.8173,
+      "step": 28664
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 3.0994105339050293,
+      "learning_rate": 5.778956070028641e-06,
+      "loss": 0.4566,
+      "step": 28665
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.882875919342041,
+      "learning_rate": 5.778695630032362e-06,
+      "loss": 0.7996,
+      "step": 28666
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 3.1586220264434814,
+      "learning_rate": 5.778435187870821e-06,
+      "loss": 0.604,
+      "step": 28667
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 3.1102428436279297,
+      "learning_rate": 5.778174743544743e-06,
+      "loss": 0.7765,
+      "step": 28668
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.6850907802581787,
+      "learning_rate": 5.777914297054854e-06,
+      "loss": 0.6046,
+      "step": 28669
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 3.7164366245269775,
+      "learning_rate": 5.7776538484018765e-06,
+      "loss": 0.5344,
+      "step": 28670
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.3955111503601074,
+      "learning_rate": 5.777393397586535e-06,
+      "loss": 0.6904,
+      "step": 28671
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.284127712249756,
+      "learning_rate": 5.777132944609554e-06,
+      "loss": 0.4975,
+      "step": 28672
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 3.4116129875183105,
+      "learning_rate": 5.776872489471657e-06,
+      "loss": 0.4706,
+      "step": 28673
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.807083010673523,
+      "learning_rate": 5.7766120321735675e-06,
+      "loss": 0.7195,
+      "step": 28674
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 3.096649169921875,
+      "learning_rate": 5.776351572716013e-06,
+      "loss": 0.6643,
+      "step": 28675
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.90206778049469,
+      "learning_rate": 5.776091111099713e-06,
+      "loss": 0.784,
+      "step": 28676
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 4.362464427947998,
+      "learning_rate": 5.775830647325397e-06,
+      "loss": 0.6075,
+      "step": 28677
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 4.414953231811523,
+      "learning_rate": 5.775570181393785e-06,
+      "loss": 0.7687,
+      "step": 28678
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 3.501854181289673,
+      "learning_rate": 5.775309713305602e-06,
+      "loss": 0.8095,
+      "step": 28679
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.244382381439209,
+      "learning_rate": 5.775049243061575e-06,
+      "loss": 0.7282,
+      "step": 28680
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.6754560470581055,
+      "learning_rate": 5.774788770662425e-06,
+      "loss": 0.5416,
+      "step": 28681
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.8433215618133545,
+      "learning_rate": 5.774528296108879e-06,
+      "loss": 0.7152,
+      "step": 28682
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.747746229171753,
+      "learning_rate": 5.77426781940166e-06,
+      "loss": 0.6895,
+      "step": 28683
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.6069328784942627,
+      "learning_rate": 5.774007340541491e-06,
+      "loss": 0.5043,
+      "step": 28684
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.5195822715759277,
+      "learning_rate": 5.773746859529097e-06,
+      "loss": 0.4473,
+      "step": 28685
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.0546011924743652,
+      "learning_rate": 5.773486376365205e-06,
+      "loss": 0.4203,
+      "step": 28686
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 3.306182622909546,
+      "learning_rate": 5.773225891050536e-06,
+      "loss": 0.394,
+      "step": 28687
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.1486942768096924,
+      "learning_rate": 5.772965403585815e-06,
+      "loss": 0.5708,
+      "step": 28688
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.7349817752838135,
+      "learning_rate": 5.772704913971767e-06,
+      "loss": 0.5496,
+      "step": 28689
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 3.155799388885498,
+      "learning_rate": 5.772444422209117e-06,
+      "loss": 0.8949,
+      "step": 28690
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 3.1729562282562256,
+      "learning_rate": 5.772183928298587e-06,
+      "loss": 0.6055,
+      "step": 28691
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.866440534591675,
+      "learning_rate": 5.771923432240904e-06,
+      "loss": 0.5443,
+      "step": 28692
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.0731451511383057,
+      "learning_rate": 5.771662934036791e-06,
+      "loss": 0.5881,
+      "step": 28693
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 3.3512017726898193,
+      "learning_rate": 5.771402433686973e-06,
+      "loss": 0.8552,
+      "step": 28694
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.2068841457366943,
+      "learning_rate": 5.7711419311921726e-06,
+      "loss": 0.6191,
+      "step": 28695
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.948977470397949,
+      "learning_rate": 5.770881426553115e-06,
+      "loss": 0.4592,
+      "step": 28696
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.7423532009124756,
+      "learning_rate": 5.770620919770527e-06,
+      "loss": 0.5476,
+      "step": 28697
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.1211445331573486,
+      "learning_rate": 5.770360410845128e-06,
+      "loss": 0.6378,
+      "step": 28698
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.5828053951263428,
+      "learning_rate": 5.770099899777647e-06,
+      "loss": 0.5306,
+      "step": 28699
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.7543463706970215,
+      "learning_rate": 5.769839386568806e-06,
+      "loss": 0.62,
+      "step": 28700
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.9722506999969482,
+      "learning_rate": 5.7695788712193315e-06,
+      "loss": 0.7417,
+      "step": 28701
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 3.6257500648498535,
+      "learning_rate": 5.769318353729945e-06,
+      "loss": 0.5074,
+      "step": 28702
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 3.2953104972839355,
+      "learning_rate": 5.7690578341013734e-06,
+      "loss": 0.4618,
+      "step": 28703
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 3.2850089073181152,
+      "learning_rate": 5.76879731233434e-06,
+      "loss": 0.4385,
+      "step": 28704
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 3.013772964477539,
+      "learning_rate": 5.7685367884295685e-06,
+      "loss": 0.5939,
+      "step": 28705
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 3.9254419803619385,
+      "learning_rate": 5.768276262387784e-06,
+      "loss": 0.7438,
+      "step": 28706
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.028714179992676,
+      "learning_rate": 5.768015734209712e-06,
+      "loss": 0.4986,
+      "step": 28707
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.922276258468628,
+      "learning_rate": 5.767755203896075e-06,
+      "loss": 0.7367,
+      "step": 28708
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 3.8627665042877197,
+      "learning_rate": 5.767494671447598e-06,
+      "loss": 0.6815,
+      "step": 28709
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 3.369109869003296,
+      "learning_rate": 5.767234136865007e-06,
+      "loss": 0.3953,
+      "step": 28710
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 3.58626389503479,
+      "learning_rate": 5.766973600149025e-06,
+      "loss": 0.7644,
+      "step": 28711
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.0982565879821777,
+      "learning_rate": 5.7667130613003766e-06,
+      "loss": 0.6913,
+      "step": 28712
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.0211191177368164,
+      "learning_rate": 5.766452520319787e-06,
+      "loss": 0.4356,
+      "step": 28713
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.6524064540863037,
+      "learning_rate": 5.766191977207979e-06,
+      "loss": 0.8076,
+      "step": 28714
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.6816725730895996,
+      "learning_rate": 5.765931431965679e-06,
+      "loss": 0.5512,
+      "step": 28715
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.101280450820923,
+      "learning_rate": 5.76567088459361e-06,
+      "loss": 0.6633,
+      "step": 28716
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.684291362762451,
+      "learning_rate": 5.7654103350924975e-06,
+      "loss": 0.5473,
+      "step": 28717
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.6430535316467285,
+      "learning_rate": 5.765149783463065e-06,
+      "loss": 0.7195,
+      "step": 28718
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.639995574951172,
+      "learning_rate": 5.764889229706038e-06,
+      "loss": 0.7276,
+      "step": 28719
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.9290804862976074,
+      "learning_rate": 5.76462867382214e-06,
+      "loss": 0.68,
+      "step": 28720
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.5472546815872192,
+      "learning_rate": 5.764368115812097e-06,
+      "loss": 0.5302,
+      "step": 28721
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.6713311672210693,
+      "learning_rate": 5.764107555676633e-06,
+      "loss": 0.6452,
+      "step": 28722
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 3.9062283039093018,
+      "learning_rate": 5.763846993416471e-06,
+      "loss": 0.5508,
+      "step": 28723
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.2707860469818115,
+      "learning_rate": 5.763586429032337e-06,
+      "loss": 0.6991,
+      "step": 28724
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 4.1426239013671875,
+      "learning_rate": 5.763325862524956e-06,
+      "loss": 0.5366,
+      "step": 28725
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.0484700202941895,
+      "learning_rate": 5.7630652938950505e-06,
+      "loss": 0.5729,
+      "step": 28726
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.690044641494751,
+      "learning_rate": 5.762804723143347e-06,
+      "loss": 0.7492,
+      "step": 28727
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.791401743888855,
+      "learning_rate": 5.762544150270569e-06,
+      "loss": 0.4602,
+      "step": 28728
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.999549627304077,
+      "learning_rate": 5.762283575277441e-06,
+      "loss": 0.7628,
+      "step": 28729
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.8695776462554932,
+      "learning_rate": 5.762022998164688e-06,
+      "loss": 0.7061,
+      "step": 28730
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 3.221447706222534,
+      "learning_rate": 5.761762418933034e-06,
+      "loss": 0.7366,
+      "step": 28731
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.7246155738830566,
+      "learning_rate": 5.7615018375832055e-06,
+      "loss": 0.8666,
+      "step": 28732
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.947503685951233,
+      "learning_rate": 5.761241254115926e-06,
+      "loss": 0.367,
+      "step": 28733
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.0716941356658936,
+      "learning_rate": 5.760980668531918e-06,
+      "loss": 0.634,
+      "step": 28734
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 3.127664804458618,
+      "learning_rate": 5.76072008083191e-06,
+      "loss": 0.5149,
+      "step": 28735
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.5167078971862793,
+      "learning_rate": 5.760459491016622e-06,
+      "loss": 0.5229,
+      "step": 28736
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.563206195831299,
+      "learning_rate": 5.760198899086783e-06,
+      "loss": 0.6605,
+      "step": 28737
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.9900145530700684,
+      "learning_rate": 5.759938305043115e-06,
+      "loss": 0.7559,
+      "step": 28738
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.6965694427490234,
+      "learning_rate": 5.759677708886342e-06,
+      "loss": 0.7212,
+      "step": 28739
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 3.5881805419921875,
+      "learning_rate": 5.7594171106171904e-06,
+      "loss": 0.617,
+      "step": 28740
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 3.292128562927246,
+      "learning_rate": 5.759156510236385e-06,
+      "loss": 0.5558,
+      "step": 28741
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 3.4436142444610596,
+      "learning_rate": 5.758895907744649e-06,
+      "loss": 0.7273,
+      "step": 28742
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 3.187128782272339,
+      "learning_rate": 5.758635303142709e-06,
+      "loss": 0.8372,
+      "step": 28743
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.079068422317505,
+      "learning_rate": 5.758374696431288e-06,
+      "loss": 0.6511,
+      "step": 28744
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 4.597455024719238,
+      "learning_rate": 5.7581140876111105e-06,
+      "loss": 0.7298,
+      "step": 28745
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.5269039869308472,
+      "learning_rate": 5.757853476682902e-06,
+      "loss": 0.4978,
+      "step": 28746
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.689750909805298,
+      "learning_rate": 5.757592863647385e-06,
+      "loss": 0.6126,
+      "step": 28747
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.6685576438903809,
+      "learning_rate": 5.75733224850529e-06,
+      "loss": 0.6521,
+      "step": 28748
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.3561935424804688,
+      "learning_rate": 5.757071631257335e-06,
+      "loss": 0.6052,
+      "step": 28749
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 3.0122716426849365,
+      "learning_rate": 5.756811011904247e-06,
+      "loss": 0.5652,
+      "step": 28750
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.5794737339019775,
+      "learning_rate": 5.756550390446754e-06,
+      "loss": 0.3154,
+      "step": 28751
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.9066216945648193,
+      "learning_rate": 5.756289766885575e-06,
+      "loss": 0.5226,
+      "step": 28752
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 3.315476655960083,
+      "learning_rate": 5.7560291412214375e-06,
+      "loss": 0.5681,
+      "step": 28753
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.6510396003723145,
+      "learning_rate": 5.755768513455066e-06,
+      "loss": 0.5366,
+      "step": 28754
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.679765224456787,
+      "learning_rate": 5.755507883587188e-06,
+      "loss": 0.7323,
+      "step": 28755
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 3.6486570835113525,
+      "learning_rate": 5.755247251618523e-06,
+      "loss": 0.575,
+      "step": 28756
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.0345277786254883,
+      "learning_rate": 5.754986617549801e-06,
+      "loss": 0.7788,
+      "step": 28757
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.1420655250549316,
+      "learning_rate": 5.754725981381741e-06,
+      "loss": 0.7385,
+      "step": 28758
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.9275002479553223,
+      "learning_rate": 5.7544653431150735e-06,
+      "loss": 0.6531,
+      "step": 28759
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 3.046806812286377,
+      "learning_rate": 5.754204702750518e-06,
+      "loss": 0.7709,
+      "step": 28760
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.442735433578491,
+      "learning_rate": 5.753944060288803e-06,
+      "loss": 0.8129,
+      "step": 28761
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.2060258388519287,
+      "learning_rate": 5.753683415730651e-06,
+      "loss": 0.4154,
+      "step": 28762
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 3.667125701904297,
+      "learning_rate": 5.75342276907679e-06,
+      "loss": 0.3692,
+      "step": 28763
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.273038625717163,
+      "learning_rate": 5.753162120327941e-06,
+      "loss": 0.8339,
+      "step": 28764
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 3.934243679046631,
+      "learning_rate": 5.75290146948483e-06,
+      "loss": 0.7497,
+      "step": 28765
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.469545602798462,
+      "learning_rate": 5.752640816548183e-06,
+      "loss": 0.3766,
+      "step": 28766
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.963853597640991,
+      "learning_rate": 5.7523801615187235e-06,
+      "loss": 0.5882,
+      "step": 28767
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 3.80851674079895,
+      "learning_rate": 5.752119504397175e-06,
+      "loss": 0.8595,
+      "step": 28768
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.3425683975219727,
+      "learning_rate": 5.751858845184266e-06,
+      "loss": 0.63,
+      "step": 28769
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 3.162505626678467,
+      "learning_rate": 5.751598183880718e-06,
+      "loss": 0.7416,
+      "step": 28770
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.984424591064453,
+      "learning_rate": 5.751337520487258e-06,
+      "loss": 0.4554,
+      "step": 28771
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.7987349033355713,
+      "learning_rate": 5.751076855004609e-06,
+      "loss": 0.5516,
+      "step": 28772
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 3.2324275970458984,
+      "learning_rate": 5.7508161874334975e-06,
+      "loss": 0.6212,
+      "step": 28773
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 6.23443078994751,
+      "learning_rate": 5.750555517774647e-06,
+      "loss": 0.7043,
+      "step": 28774
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.088118553161621,
+      "learning_rate": 5.750294846028783e-06,
+      "loss": 0.6382,
+      "step": 28775
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 3.130728006362915,
+      "learning_rate": 5.75003417219663e-06,
+      "loss": 0.4639,
+      "step": 28776
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.280322790145874,
+      "learning_rate": 5.7497734962789135e-06,
+      "loss": 0.7427,
+      "step": 28777
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 3.0449185371398926,
+      "learning_rate": 5.7495128182763575e-06,
+      "loss": 0.5604,
+      "step": 28778
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.852468729019165,
+      "learning_rate": 5.749252138189686e-06,
+      "loss": 0.6247,
+      "step": 28779
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 4.463912010192871,
+      "learning_rate": 5.748991456019625e-06,
+      "loss": 0.6808,
+      "step": 28780
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 3.5551161766052246,
+      "learning_rate": 5.748730771766899e-06,
+      "loss": 0.5372,
+      "step": 28781
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.8603428602218628,
+      "learning_rate": 5.748470085432235e-06,
+      "loss": 0.5324,
+      "step": 28782
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.8731722831726074,
+      "learning_rate": 5.748209397016356e-06,
+      "loss": 0.7874,
+      "step": 28783
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.960548162460327,
+      "learning_rate": 5.747948706519986e-06,
+      "loss": 0.5121,
+      "step": 28784
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.215596914291382,
+      "learning_rate": 5.747688013943851e-06,
+      "loss": 0.6185,
+      "step": 28785
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.6931724548339844,
+      "learning_rate": 5.7474273192886755e-06,
+      "loss": 0.5416,
+      "step": 28786
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 3.771289110183716,
+      "learning_rate": 5.747166622555184e-06,
+      "loss": 0.838,
+      "step": 28787
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.6807299852371216,
+      "learning_rate": 5.746905923744104e-06,
+      "loss": 0.5629,
+      "step": 28788
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.364983081817627,
+      "learning_rate": 5.746645222856157e-06,
+      "loss": 0.7445,
+      "step": 28789
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.2917613983154297,
+      "learning_rate": 5.74638451989207e-06,
+      "loss": 0.721,
+      "step": 28790
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 4.620317459106445,
+      "learning_rate": 5.746123814852567e-06,
+      "loss": 0.7877,
+      "step": 28791
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.6383088827133179,
+      "learning_rate": 5.7458631077383735e-06,
+      "loss": 0.4442,
+      "step": 28792
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.205049514770508,
+      "learning_rate": 5.745602398550213e-06,
+      "loss": 0.6142,
+      "step": 28793
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.7958767414093018,
+      "learning_rate": 5.745341687288812e-06,
+      "loss": 0.6588,
+      "step": 28794
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.4728717803955078,
+      "learning_rate": 5.7450809739548956e-06,
+      "loss": 0.5294,
+      "step": 28795
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.703251600265503,
+      "learning_rate": 5.744820258549187e-06,
+      "loss": 0.509,
+      "step": 28796
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 3.369438648223877,
+      "learning_rate": 5.744559541072414e-06,
+      "loss": 0.7919,
+      "step": 28797
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.7814433574676514,
+      "learning_rate": 5.744298821525299e-06,
+      "loss": 0.5478,
+      "step": 28798
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.433520555496216,
+      "learning_rate": 5.744038099908568e-06,
+      "loss": 0.6814,
+      "step": 28799
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.9864752292633057,
+      "learning_rate": 5.743777376222945e-06,
+      "loss": 0.6638,
+      "step": 28800
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.7736029624938965,
+      "learning_rate": 5.743516650469156e-06,
+      "loss": 0.5987,
+      "step": 28801
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.397979497909546,
+      "learning_rate": 5.743255922647926e-06,
+      "loss": 0.6662,
+      "step": 28802
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.5448198318481445,
+      "learning_rate": 5.742995192759979e-06,
+      "loss": 0.5981,
+      "step": 28803
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.312946081161499,
+      "learning_rate": 5.742734460806042e-06,
+      "loss": 0.6152,
+      "step": 28804
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.01265811920166,
+      "learning_rate": 5.742473726786838e-06,
+      "loss": 0.5386,
+      "step": 28805
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.3036139011383057,
+      "learning_rate": 5.742212990703092e-06,
+      "loss": 0.6806,
+      "step": 28806
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.7103239297866821,
+      "learning_rate": 5.741952252555532e-06,
+      "loss": 0.6856,
+      "step": 28807
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.2609570026397705,
+      "learning_rate": 5.741691512344878e-06,
+      "loss": 0.8944,
+      "step": 28808
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.208876609802246,
+      "learning_rate": 5.741430770071858e-06,
+      "loss": 0.6227,
+      "step": 28809
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.8958580493927,
+      "learning_rate": 5.741170025737198e-06,
+      "loss": 0.5539,
+      "step": 28810
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.5903429985046387,
+      "learning_rate": 5.740909279341622e-06,
+      "loss": 0.9314,
+      "step": 28811
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.078408718109131,
+      "learning_rate": 5.7406485308858535e-06,
+      "loss": 0.4966,
+      "step": 28812
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.8240680694580078,
+      "learning_rate": 5.7403877803706195e-06,
+      "loss": 0.4908,
+      "step": 28813
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.4679372310638428,
+      "learning_rate": 5.740127027796646e-06,
+      "loss": 0.6859,
+      "step": 28814
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.373467206954956,
+      "learning_rate": 5.739866273164654e-06,
+      "loss": 0.5369,
+      "step": 28815
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.3122453689575195,
+      "learning_rate": 5.739605516475372e-06,
+      "loss": 0.6549,
+      "step": 28816
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.6393383741378784,
+      "learning_rate": 5.739344757729525e-06,
+      "loss": 0.4957,
+      "step": 28817
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 3.135528326034546,
+      "learning_rate": 5.739083996927836e-06,
+      "loss": 0.6017,
+      "step": 28818
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.9201050996780396,
+      "learning_rate": 5.7388232340710316e-06,
+      "loss": 0.5472,
+      "step": 28819
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.1213910579681396,
+      "learning_rate": 5.738562469159837e-06,
+      "loss": 0.6873,
+      "step": 28820
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.994805097579956,
+      "learning_rate": 5.738301702194976e-06,
+      "loss": 0.5471,
+      "step": 28821
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.923053741455078,
+      "learning_rate": 5.738040933177175e-06,
+      "loss": 0.5892,
+      "step": 28822
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.9251060485839844,
+      "learning_rate": 5.737780162107159e-06,
+      "loss": 0.8455,
+      "step": 28823
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.6570218801498413,
+      "learning_rate": 5.737519388985653e-06,
+      "loss": 0.7869,
+      "step": 28824
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 5.761864185333252,
+      "learning_rate": 5.737258613813381e-06,
+      "loss": 0.7228,
+      "step": 28825
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.2283203601837158,
+      "learning_rate": 5.736997836591069e-06,
+      "loss": 0.4002,
+      "step": 28826
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 3.074934959411621,
+      "learning_rate": 5.736737057319444e-06,
+      "loss": 0.7348,
+      "step": 28827
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.1037397384643555,
+      "learning_rate": 5.736476275999227e-06,
+      "loss": 0.9694,
+      "step": 28828
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.665731430053711,
+      "learning_rate": 5.736215492631146e-06,
+      "loss": 0.6973,
+      "step": 28829
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.3294897079467773,
+      "learning_rate": 5.735954707215926e-06,
+      "loss": 0.4867,
+      "step": 28830
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 4.399470806121826,
+      "learning_rate": 5.73569391975429e-06,
+      "loss": 0.6249,
+      "step": 28831
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.4463930130004883,
+      "learning_rate": 5.735433130246966e-06,
+      "loss": 0.5937,
+      "step": 28832
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.5377557277679443,
+      "learning_rate": 5.735172338694678e-06,
+      "loss": 0.398,
+      "step": 28833
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.9280333518981934,
+      "learning_rate": 5.734911545098152e-06,
+      "loss": 0.6358,
+      "step": 28834
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.0459938049316406,
+      "learning_rate": 5.7346507494581104e-06,
+      "loss": 0.6113,
+      "step": 28835
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.0986626148223877,
+      "learning_rate": 5.734389951775282e-06,
+      "loss": 0.4853,
+      "step": 28836
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.23323130607605,
+      "learning_rate": 5.73412915205039e-06,
+      "loss": 0.7172,
+      "step": 28837
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.0545172691345215,
+      "learning_rate": 5.733868350284159e-06,
+      "loss": 0.5646,
+      "step": 28838
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.3700549602508545,
+      "learning_rate": 5.7336075464773154e-06,
+      "loss": 0.3534,
+      "step": 28839
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 4.251424789428711,
+      "learning_rate": 5.733346740630584e-06,
+      "loss": 0.8158,
+      "step": 28840
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.8556973934173584,
+      "learning_rate": 5.73308593274469e-06,
+      "loss": 0.8076,
+      "step": 28841
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.6213173866271973,
+      "learning_rate": 5.732825122820358e-06,
+      "loss": 0.5906,
+      "step": 28842
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.4223928451538086,
+      "learning_rate": 5.732564310858315e-06,
+      "loss": 0.5691,
+      "step": 28843
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.7342216968536377,
+      "learning_rate": 5.732303496859285e-06,
+      "loss": 0.5364,
+      "step": 28844
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 3.443427562713623,
+      "learning_rate": 5.732042680823993e-06,
+      "loss": 0.5437,
+      "step": 28845
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 3.9132347106933594,
+      "learning_rate": 5.731781862753164e-06,
+      "loss": 0.4717,
+      "step": 28846
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.7814366817474365,
+      "learning_rate": 5.731521042647525e-06,
+      "loss": 0.6776,
+      "step": 28847
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.471012592315674,
+      "learning_rate": 5.731260220507799e-06,
+      "loss": 0.5799,
+      "step": 28848
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.39188814163208,
+      "learning_rate": 5.730999396334713e-06,
+      "loss": 0.5612,
+      "step": 28849
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 3.8869051933288574,
+      "learning_rate": 5.730738570128991e-06,
+      "loss": 0.512,
+      "step": 28850
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.546895980834961,
+      "learning_rate": 5.730477741891358e-06,
+      "loss": 0.1894,
+      "step": 28851
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 3.467811107635498,
+      "learning_rate": 5.73021691162254e-06,
+      "loss": 0.8622,
+      "step": 28852
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 3.4183974266052246,
+      "learning_rate": 5.729956079323265e-06,
+      "loss": 0.637,
+      "step": 28853
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.8639612197875977,
+      "learning_rate": 5.729695244994254e-06,
+      "loss": 0.5181,
+      "step": 28854
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.9332163333892822,
+      "learning_rate": 5.729434408636233e-06,
+      "loss": 0.7396,
+      "step": 28855
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.1558175086975098,
+      "learning_rate": 5.729173570249929e-06,
+      "loss": 0.7814,
+      "step": 28856
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.371868133544922,
+      "learning_rate": 5.728912729836066e-06,
+      "loss": 0.4848,
+      "step": 28857
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.381830930709839,
+      "learning_rate": 5.72865188739537e-06,
+      "loss": 0.6785,
+      "step": 28858
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.315553903579712,
+      "learning_rate": 5.728391042928566e-06,
+      "loss": 0.6525,
+      "step": 28859
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.4483816623687744,
+      "learning_rate": 5.72813019643638e-06,
+      "loss": 0.6808,
+      "step": 28860
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.991034984588623,
+      "learning_rate": 5.727869347919536e-06,
+      "loss": 0.5394,
+      "step": 28861
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.5264272689819336,
+      "learning_rate": 5.727608497378759e-06,
+      "loss": 0.4821,
+      "step": 28862
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.1052298545837402,
+      "learning_rate": 5.727347644814777e-06,
+      "loss": 0.4922,
+      "step": 28863
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.659964919090271,
+      "learning_rate": 5.727086790228313e-06,
+      "loss": 0.6615,
+      "step": 28864
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.125044107437134,
+      "learning_rate": 5.726825933620094e-06,
+      "loss": 0.7332,
+      "step": 28865
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.9295402765274048,
+      "learning_rate": 5.726565074990843e-06,
+      "loss": 0.6906,
+      "step": 28866
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.005556106567383,
+      "learning_rate": 5.726304214341287e-06,
+      "loss": 0.7665,
+      "step": 28867
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.753309488296509,
+      "learning_rate": 5.726043351672151e-06,
+      "loss": 0.5454,
+      "step": 28868
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.4610812664031982,
+      "learning_rate": 5.725782486984161e-06,
+      "loss": 0.6459,
+      "step": 28869
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.6090688705444336,
+      "learning_rate": 5.725521620278041e-06,
+      "loss": 0.3809,
+      "step": 28870
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.192181348800659,
+      "learning_rate": 5.725260751554517e-06,
+      "loss": 0.7084,
+      "step": 28871
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.886154294013977,
+      "learning_rate": 5.724999880814316e-06,
+      "loss": 0.5357,
+      "step": 28872
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.5364856719970703,
+      "learning_rate": 5.724739008058161e-06,
+      "loss": 0.4043,
+      "step": 28873
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.9068105220794678,
+      "learning_rate": 5.724478133286779e-06,
+      "loss": 0.7168,
+      "step": 28874
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.7791435718536377,
+      "learning_rate": 5.724217256500895e-06,
+      "loss": 0.6573,
+      "step": 28875
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.214492082595825,
+      "learning_rate": 5.723956377701232e-06,
+      "loss": 0.8282,
+      "step": 28876
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 4.107304573059082,
+      "learning_rate": 5.723695496888519e-06,
+      "loss": 0.6151,
+      "step": 28877
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.343144655227661,
+      "learning_rate": 5.723434614063479e-06,
+      "loss": 0.6313,
+      "step": 28878
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 4.5283708572387695,
+      "learning_rate": 5.723173729226838e-06,
+      "loss": 0.5419,
+      "step": 28879
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.565396308898926,
+      "learning_rate": 5.722912842379325e-06,
+      "loss": 0.6548,
+      "step": 28880
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.9558488130569458,
+      "learning_rate": 5.72265195352166e-06,
+      "loss": 0.7018,
+      "step": 28881
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.206932783126831,
+      "learning_rate": 5.722391062654569e-06,
+      "loss": 0.495,
+      "step": 28882
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.7627538442611694,
+      "learning_rate": 5.72213016977878e-06,
+      "loss": 0.5478,
+      "step": 28883
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.0529184341430664,
+      "learning_rate": 5.7218692748950186e-06,
+      "loss": 0.5631,
+      "step": 28884
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.9300410747528076,
+      "learning_rate": 5.721608378004009e-06,
+      "loss": 0.7841,
+      "step": 28885
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 3.723639488220215,
+      "learning_rate": 5.721347479106477e-06,
+      "loss": 0.431,
+      "step": 28886
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.2900478839874268,
+      "learning_rate": 5.721086578203148e-06,
+      "loss": 0.5643,
+      "step": 28887
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.599528193473816,
+      "learning_rate": 5.720825675294747e-06,
+      "loss": 0.3589,
+      "step": 28888
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.2327561378479004,
+      "learning_rate": 5.720564770381999e-06,
+      "loss": 0.6025,
+      "step": 28889
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.197472095489502,
+      "learning_rate": 5.720303863465631e-06,
+      "loss": 0.5078,
+      "step": 28890
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.9620476961135864,
+      "learning_rate": 5.720042954546368e-06,
+      "loss": 0.55,
+      "step": 28891
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.3897745609283447,
+      "learning_rate": 5.719782043624934e-06,
+      "loss": 0.7216,
+      "step": 28892
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.505256175994873,
+      "learning_rate": 5.719521130702057e-06,
+      "loss": 0.5242,
+      "step": 28893
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 6.56123161315918,
+      "learning_rate": 5.719260215778461e-06,
+      "loss": 0.6549,
+      "step": 28894
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.358455181121826,
+      "learning_rate": 5.718999298854873e-06,
+      "loss": 0.4787,
+      "step": 28895
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.5832581520080566,
+      "learning_rate": 5.718738379932015e-06,
+      "loss": 0.7781,
+      "step": 28896
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.7383910417556763,
+      "learning_rate": 5.718477459010616e-06,
+      "loss": 0.7458,
+      "step": 28897
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.908751130104065,
+      "learning_rate": 5.718216536091401e-06,
+      "loss": 0.9307,
+      "step": 28898
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.0597798824310303,
+      "learning_rate": 5.717955611175095e-06,
+      "loss": 0.808,
+      "step": 28899
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.121347665786743,
+      "learning_rate": 5.717694684262421e-06,
+      "loss": 0.5073,
+      "step": 28900
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.926551103591919,
+      "learning_rate": 5.717433755354109e-06,
+      "loss": 0.6282,
+      "step": 28901
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.9147599935531616,
+      "learning_rate": 5.7171728244508805e-06,
+      "loss": 0.8179,
+      "step": 28902
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.321128487586975,
+      "learning_rate": 5.716911891553465e-06,
+      "loss": 0.3055,
+      "step": 28903
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.9198657274246216,
+      "learning_rate": 5.716650956662586e-06,
+      "loss": 0.5541,
+      "step": 28904
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 3.0795695781707764,
+      "learning_rate": 5.716390019778969e-06,
+      "loss": 0.5855,
+      "step": 28905
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.7957148551940918,
+      "learning_rate": 5.71612908090334e-06,
+      "loss": 0.8124,
+      "step": 28906
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.2785756587982178,
+      "learning_rate": 5.715868140036424e-06,
+      "loss": 0.4711,
+      "step": 28907
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.0191941261291504,
+      "learning_rate": 5.7156071971789464e-06,
+      "loss": 0.6211,
+      "step": 28908
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.973396062850952,
+      "learning_rate": 5.715346252331633e-06,
+      "loss": 0.4823,
+      "step": 28909
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.9822911024093628,
+      "learning_rate": 5.71508530549521e-06,
+      "loss": 0.4992,
+      "step": 28910
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.2074151039123535,
+      "learning_rate": 5.7148243566704034e-06,
+      "loss": 0.5281,
+      "step": 28911
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.5936450958251953,
+      "learning_rate": 5.7145634058579355e-06,
+      "loss": 0.6047,
+      "step": 28912
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.9662213325500488,
+      "learning_rate": 5.7143024530585376e-06,
+      "loss": 0.6766,
+      "step": 28913
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.248749017715454,
+      "learning_rate": 5.714041498272931e-06,
+      "loss": 0.7739,
+      "step": 28914
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 3.15022873878479,
+      "learning_rate": 5.713780541501842e-06,
+      "loss": 0.4278,
+      "step": 28915
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.770569086074829,
+      "learning_rate": 5.713519582745998e-06,
+      "loss": 0.7217,
+      "step": 28916
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.348405361175537,
+      "learning_rate": 5.7132586220061215e-06,
+      "loss": 0.7887,
+      "step": 28917
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.302743434906006,
+      "learning_rate": 5.7129976592829415e-06,
+      "loss": 0.7462,
+      "step": 28918
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 3.8618903160095215,
+      "learning_rate": 5.712736694577181e-06,
+      "loss": 0.6821,
+      "step": 28919
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 3.040334463119507,
+      "learning_rate": 5.712475727889568e-06,
+      "loss": 0.4581,
+      "step": 28920
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 3.6153459548950195,
+      "learning_rate": 5.7122147592208266e-06,
+      "loss": 0.5309,
+      "step": 28921
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.5578532218933105,
+      "learning_rate": 5.711953788571682e-06,
+      "loss": 0.682,
+      "step": 28922
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 7.356600284576416,
+      "learning_rate": 5.71169281594286e-06,
+      "loss": 0.6803,
+      "step": 28923
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.049750804901123,
+      "learning_rate": 5.711431841335088e-06,
+      "loss": 0.4426,
+      "step": 28924
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 3.5138304233551025,
+      "learning_rate": 5.71117086474909e-06,
+      "loss": 0.6297,
+      "step": 28925
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.280745267868042,
+      "learning_rate": 5.710909886185593e-06,
+      "loss": 0.659,
+      "step": 28926
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 3.5232670307159424,
+      "learning_rate": 5.710648905645322e-06,
+      "loss": 0.5876,
+      "step": 28927
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.877211570739746,
+      "learning_rate": 5.710387923129001e-06,
+      "loss": 0.7857,
+      "step": 28928
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.421555757522583,
+      "learning_rate": 5.710126938637359e-06,
+      "loss": 0.4245,
+      "step": 28929
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.906705141067505,
+      "learning_rate": 5.70986595217112e-06,
+      "loss": 0.6959,
+      "step": 28930
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.403823137283325,
+      "learning_rate": 5.709604963731009e-06,
+      "loss": 0.7403,
+      "step": 28931
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 4.431974411010742,
+      "learning_rate": 5.709343973317752e-06,
+      "loss": 0.6981,
+      "step": 28932
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.8882863521575928,
+      "learning_rate": 5.709082980932076e-06,
+      "loss": 0.5963,
+      "step": 28933
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.768599510192871,
+      "learning_rate": 5.708821986574705e-06,
+      "loss": 0.6678,
+      "step": 28934
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.536247968673706,
+      "learning_rate": 5.708560990246367e-06,
+      "loss": 0.7741,
+      "step": 28935
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.2831199169158936,
+      "learning_rate": 5.7082999919477855e-06,
+      "loss": 0.7099,
+      "step": 28936
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.9795233011245728,
+      "learning_rate": 5.708038991679687e-06,
+      "loss": 0.6613,
+      "step": 28937
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.3091325759887695,
+      "learning_rate": 5.707777989442796e-06,
+      "loss": 0.6238,
+      "step": 28938
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.9276423454284668,
+      "learning_rate": 5.707516985237841e-06,
+      "loss": 0.6351,
+      "step": 28939
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.93259596824646,
+      "learning_rate": 5.707255979065547e-06,
+      "loss": 0.633,
+      "step": 28940
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 3.5101828575134277,
+      "learning_rate": 5.706994970926638e-06,
+      "loss": 0.5989,
+      "step": 28941
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.399437665939331,
+      "learning_rate": 5.706733960821841e-06,
+      "loss": 0.7939,
+      "step": 28942
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.437969923019409,
+      "learning_rate": 5.706472948751882e-06,
+      "loss": 0.5338,
+      "step": 28943
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.5125887393951416,
+      "learning_rate": 5.706211934717485e-06,
+      "loss": 0.4547,
+      "step": 28944
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.9314215183258057,
+      "learning_rate": 5.7059509187193784e-06,
+      "loss": 0.5644,
+      "step": 28945
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 4.2689642906188965,
+      "learning_rate": 5.705689900758286e-06,
+      "loss": 0.4597,
+      "step": 28946
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.4848055839538574,
+      "learning_rate": 5.705428880834936e-06,
+      "loss": 0.6625,
+      "step": 28947
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.2108486890792847,
+      "learning_rate": 5.705167858950052e-06,
+      "loss": 0.7311,
+      "step": 28948
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.5752627849578857,
+      "learning_rate": 5.704906835104359e-06,
+      "loss": 0.4717,
+      "step": 28949
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.2884953022003174,
+      "learning_rate": 5.704645809298586e-06,
+      "loss": 0.4232,
+      "step": 28950
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 4.519185543060303,
+      "learning_rate": 5.704384781533455e-06,
+      "loss": 0.7056,
+      "step": 28951
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.7554763555526733,
+      "learning_rate": 5.7041237518096945e-06,
+      "loss": 0.6857,
+      "step": 28952
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 4.1373796463012695,
+      "learning_rate": 5.703862720128029e-06,
+      "loss": 0.4906,
+      "step": 28953
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.799675226211548,
+      "learning_rate": 5.703601686489186e-06,
+      "loss": 0.6106,
+      "step": 28954
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.8071396350860596,
+      "learning_rate": 5.703340650893891e-06,
+      "loss": 0.6407,
+      "step": 28955
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 3.3857295513153076,
+      "learning_rate": 5.703079613342868e-06,
+      "loss": 0.5185,
+      "step": 28956
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 3.5011250972747803,
+      "learning_rate": 5.702818573836845e-06,
+      "loss": 0.8654,
+      "step": 28957
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.097564458847046,
+      "learning_rate": 5.702557532376546e-06,
+      "loss": 0.5654,
+      "step": 28958
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 4.0307159423828125,
+      "learning_rate": 5.702296488962697e-06,
+      "loss": 0.5653,
+      "step": 28959
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.5927914381027222,
+      "learning_rate": 5.7020354435960255e-06,
+      "loss": 0.4928,
+      "step": 28960
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.6310243606567383,
+      "learning_rate": 5.701774396277257e-06,
+      "loss": 0.5592,
+      "step": 28961
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.1442346572875977,
+      "learning_rate": 5.701513347007116e-06,
+      "loss": 0.6661,
+      "step": 28962
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 3.0712058544158936,
+      "learning_rate": 5.701252295786328e-06,
+      "loss": 0.5688,
+      "step": 28963
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.148695468902588,
+      "learning_rate": 5.700991242615621e-06,
+      "loss": 0.6112,
+      "step": 28964
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.8340892791748047,
+      "learning_rate": 5.70073018749572e-06,
+      "loss": 0.7356,
+      "step": 28965
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.568790078163147,
+      "learning_rate": 5.700469130427352e-06,
+      "loss": 0.6649,
+      "step": 28966
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.6007195711135864,
+      "learning_rate": 5.700208071411241e-06,
+      "loss": 0.5952,
+      "step": 28967
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 5.355129241943359,
+      "learning_rate": 5.699947010448112e-06,
+      "loss": 0.4227,
+      "step": 28968
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 3.406496524810791,
+      "learning_rate": 5.699685947538695e-06,
+      "loss": 0.4772,
+      "step": 28969
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.765275001525879,
+      "learning_rate": 5.6994248826837126e-06,
+      "loss": 0.7124,
+      "step": 28970
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.0090949535369873,
+      "learning_rate": 5.699163815883892e-06,
+      "loss": 0.5969,
+      "step": 28971
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 3.2007102966308594,
+      "learning_rate": 5.698902747139958e-06,
+      "loss": 0.4634,
+      "step": 28972
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 8.213455200195312,
+      "learning_rate": 5.6986416764526366e-06,
+      "loss": 0.5988,
+      "step": 28973
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.4437201023101807,
+      "learning_rate": 5.698380603822655e-06,
+      "loss": 0.6072,
+      "step": 28974
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.205263376235962,
+      "learning_rate": 5.698119529250738e-06,
+      "loss": 0.6337,
+      "step": 28975
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.1595427989959717,
+      "learning_rate": 5.697858452737615e-06,
+      "loss": 0.741,
+      "step": 28976
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.8801778554916382,
+      "learning_rate": 5.697597374284006e-06,
+      "loss": 0.6509,
+      "step": 28977
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 4.068122386932373,
+      "learning_rate": 5.6973362938906415e-06,
+      "loss": 0.7021,
+      "step": 28978
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.7170368432998657,
+      "learning_rate": 5.697075211558246e-06,
+      "loss": 0.4989,
+      "step": 28979
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 3.993006706237793,
+      "learning_rate": 5.696814127287544e-06,
+      "loss": 0.8057,
+      "step": 28980
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.8953379392623901,
+      "learning_rate": 5.6965530410792645e-06,
+      "loss": 0.555,
+      "step": 28981
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.5549184083938599,
+      "learning_rate": 5.696291952934132e-06,
+      "loss": 0.4576,
+      "step": 28982
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 3.0901858806610107,
+      "learning_rate": 5.696030862852871e-06,
+      "loss": 0.4943,
+      "step": 28983
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.9163391590118408,
+      "learning_rate": 5.6957697708362095e-06,
+      "loss": 0.583,
+      "step": 28984
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.8945584297180176,
+      "learning_rate": 5.695508676884873e-06,
+      "loss": 0.7169,
+      "step": 28985
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.9122705459594727,
+      "learning_rate": 5.695247580999588e-06,
+      "loss": 0.5808,
+      "step": 28986
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 4.078526020050049,
+      "learning_rate": 5.6949864831810794e-06,
+      "loss": 0.7373,
+      "step": 28987
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.981194257736206,
+      "learning_rate": 5.694725383430074e-06,
+      "loss": 0.5523,
+      "step": 28988
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.6820287704467773,
+      "learning_rate": 5.694464281747297e-06,
+      "loss": 0.8611,
+      "step": 28989
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.0170345306396484,
+      "learning_rate": 5.694203178133476e-06,
+      "loss": 0.4619,
+      "step": 28990
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.1836817264556885,
+      "learning_rate": 5.693942072589336e-06,
+      "loss": 0.5409,
+      "step": 28991
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 3.394801616668701,
+      "learning_rate": 5.693680965115602e-06,
+      "loss": 0.4577,
+      "step": 28992
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 4.7607221603393555,
+      "learning_rate": 5.693419855713002e-06,
+      "loss": 0.7585,
+      "step": 28993
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 3.633151054382324,
+      "learning_rate": 5.693158744382259e-06,
+      "loss": 0.8448,
+      "step": 28994
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.3810598850250244,
+      "learning_rate": 5.692897631124104e-06,
+      "loss": 0.512,
+      "step": 28995
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 4.690384387969971,
+      "learning_rate": 5.692636515939259e-06,
+      "loss": 0.8961,
+      "step": 28996
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.7422560453414917,
+      "learning_rate": 5.692375398828451e-06,
+      "loss": 0.5394,
+      "step": 28997
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 5.237585067749023,
+      "learning_rate": 5.692114279792407e-06,
+      "loss": 0.561,
+      "step": 28998
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.603540301322937,
+      "learning_rate": 5.691853158831853e-06,
+      "loss": 0.6377,
+      "step": 28999
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.481729507446289,
+      "learning_rate": 5.691592035947513e-06,
+      "loss": 0.7342,
+      "step": 29000
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 4.342411041259766,
+      "learning_rate": 5.6913309111401145e-06,
+      "loss": 0.72,
+      "step": 29001
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.4264273643493652,
+      "learning_rate": 5.691069784410386e-06,
+      "loss": 0.4885,
+      "step": 29002
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.5745255947113037,
+      "learning_rate": 5.6908086557590495e-06,
+      "loss": 0.6606,
+      "step": 29003
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.9574215412139893,
+      "learning_rate": 5.690547525186834e-06,
+      "loss": 0.7141,
+      "step": 29004
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.342552900314331,
+      "learning_rate": 5.690286392694462e-06,
+      "loss": 0.5937,
+      "step": 29005
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.2649450302124023,
+      "learning_rate": 5.690025258282664e-06,
+      "loss": 0.5967,
+      "step": 29006
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 4.543116092681885,
+      "learning_rate": 5.689764121952165e-06,
+      "loss": 0.6808,
+      "step": 29007
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.9380147457122803,
+      "learning_rate": 5.689502983703689e-06,
+      "loss": 0.7241,
+      "step": 29008
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.776987075805664,
+      "learning_rate": 5.689241843537965e-06,
+      "loss": 0.7273,
+      "step": 29009
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.9789929389953613,
+      "learning_rate": 5.688980701455717e-06,
+      "loss": 0.5436,
+      "step": 29010
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 3.1064460277557373,
+      "learning_rate": 5.688719557457671e-06,
+      "loss": 0.5508,
+      "step": 29011
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.368022918701172,
+      "learning_rate": 5.688458411544555e-06,
+      "loss": 0.5105,
+      "step": 29012
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.746584892272949,
+      "learning_rate": 5.688197263717094e-06,
+      "loss": 0.6963,
+      "step": 29013
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.1899566650390625,
+      "learning_rate": 5.687936113976013e-06,
+      "loss": 0.767,
+      "step": 29014
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.3113608360290527,
+      "learning_rate": 5.68767496232204e-06,
+      "loss": 0.7252,
+      "step": 29015
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.371523380279541,
+      "learning_rate": 5.687413808755899e-06,
+      "loss": 0.8709,
+      "step": 29016
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.0719149112701416,
+      "learning_rate": 5.68715265327832e-06,
+      "loss": 0.8451,
+      "step": 29017
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.3307061195373535,
+      "learning_rate": 5.6868914958900265e-06,
+      "loss": 0.7208,
+      "step": 29018
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.1547160148620605,
+      "learning_rate": 5.6866303365917455e-06,
+      "loss": 0.6607,
+      "step": 29019
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 3.7931604385375977,
+      "learning_rate": 5.686369175384202e-06,
+      "loss": 0.5388,
+      "step": 29020
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.6565048694610596,
+      "learning_rate": 5.6861080122681235e-06,
+      "loss": 0.7586,
+      "step": 29021
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 3.2313156127929688,
+      "learning_rate": 5.685846847244234e-06,
+      "loss": 0.6451,
+      "step": 29022
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.263873815536499,
+      "learning_rate": 5.685585680313263e-06,
+      "loss": 0.5955,
+      "step": 29023
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.1309213638305664,
+      "learning_rate": 5.6853245114759335e-06,
+      "loss": 0.5146,
+      "step": 29024
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.8701202869415283,
+      "learning_rate": 5.6850633407329745e-06,
+      "loss": 0.7212,
+      "step": 29025
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 3.3012773990631104,
+      "learning_rate": 5.68480216808511e-06,
+      "loss": 0.491,
+      "step": 29026
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 5.064642906188965,
+      "learning_rate": 5.684540993533069e-06,
+      "loss": 0.7444,
+      "step": 29027
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.26735258102417,
+      "learning_rate": 5.6842798170775746e-06,
+      "loss": 0.7021,
+      "step": 29028
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 3.5846076011657715,
+      "learning_rate": 5.684018638719356e-06,
+      "loss": 0.3912,
+      "step": 29029
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.5327091217041016,
+      "learning_rate": 5.683757458459136e-06,
+      "loss": 0.4682,
+      "step": 29030
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 3.6569876670837402,
+      "learning_rate": 5.6834962762976435e-06,
+      "loss": 0.7449,
+      "step": 29031
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 3.4532594680786133,
+      "learning_rate": 5.683235092235603e-06,
+      "loss": 0.6846,
+      "step": 29032
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.408982753753662,
+      "learning_rate": 5.682973906273742e-06,
+      "loss": 0.5073,
+      "step": 29033
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 3.4327523708343506,
+      "learning_rate": 5.682712718412786e-06,
+      "loss": 0.7255,
+      "step": 29034
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 3.5307438373565674,
+      "learning_rate": 5.682451528653462e-06,
+      "loss": 0.9563,
+      "step": 29035
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.423168897628784,
+      "learning_rate": 5.682190336996497e-06,
+      "loss": 0.5904,
+      "step": 29036
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.5827891826629639,
+      "learning_rate": 5.681929143442615e-06,
+      "loss": 0.8153,
+      "step": 29037
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.8121274709701538,
+      "learning_rate": 5.681667947992544e-06,
+      "loss": 0.6495,
+      "step": 29038
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.5804643630981445,
+      "learning_rate": 5.681406750647011e-06,
+      "loss": 0.4937,
+      "step": 29039
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.9099409580230713,
+      "learning_rate": 5.68114555140674e-06,
+      "loss": 0.5069,
+      "step": 29040
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.712869644165039,
+      "learning_rate": 5.680884350272458e-06,
+      "loss": 0.5072,
+      "step": 29041
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.088571071624756,
+      "learning_rate": 5.680623147244891e-06,
+      "loss": 0.6203,
+      "step": 29042
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.720136880874634,
+      "learning_rate": 5.680361942324768e-06,
+      "loss": 0.9874,
+      "step": 29043
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.0267810821533203,
+      "learning_rate": 5.680100735512812e-06,
+      "loss": 0.5127,
+      "step": 29044
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 5.637770175933838,
+      "learning_rate": 5.67983952680975e-06,
+      "loss": 0.6076,
+      "step": 29045
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.7347981929779053,
+      "learning_rate": 5.67957831621631e-06,
+      "loss": 0.795,
+      "step": 29046
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.3614771366119385,
+      "learning_rate": 5.679317103733217e-06,
+      "loss": 0.7527,
+      "step": 29047
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 3.205745220184326,
+      "learning_rate": 5.679055889361198e-06,
+      "loss": 0.4696,
+      "step": 29048
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.364654779434204,
+      "learning_rate": 5.678794673100978e-06,
+      "loss": 0.7895,
+      "step": 29049
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.3452394008636475,
+      "learning_rate": 5.678533454953286e-06,
+      "loss": 0.6211,
+      "step": 29050
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.5351076126098633,
+      "learning_rate": 5.678272234918845e-06,
+      "loss": 0.5854,
+      "step": 29051
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 3.8097848892211914,
+      "learning_rate": 5.6780110129983836e-06,
+      "loss": 0.6516,
+      "step": 29052
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.084357738494873,
+      "learning_rate": 5.677749789192628e-06,
+      "loss": 0.6917,
+      "step": 29053
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.925546646118164,
+      "learning_rate": 5.677488563502304e-06,
+      "loss": 0.6123,
+      "step": 29054
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 3.3218724727630615,
+      "learning_rate": 5.677227335928136e-06,
+      "loss": 0.6589,
+      "step": 29055
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 10.837174415588379,
+      "learning_rate": 5.676966106470855e-06,
+      "loss": 0.5781,
+      "step": 29056
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.839972734451294,
+      "learning_rate": 5.676704875131184e-06,
+      "loss": 0.9038,
+      "step": 29057
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 4.601522445678711,
+      "learning_rate": 5.67644364190985e-06,
+      "loss": 0.6925,
+      "step": 29058
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.7414649724960327,
+      "learning_rate": 5.67618240680758e-06,
+      "loss": 0.6414,
+      "step": 29059
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 3.9640777111053467,
+      "learning_rate": 5.6759211698251e-06,
+      "loss": 0.5258,
+      "step": 29060
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.8698680400848389,
+      "learning_rate": 5.675659930963136e-06,
+      "loss": 0.4883,
+      "step": 29061
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.281179666519165,
+      "learning_rate": 5.675398690222416e-06,
+      "loss": 0.573,
+      "step": 29062
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 10.425180435180664,
+      "learning_rate": 5.6751374476036635e-06,
+      "loss": 0.4921,
+      "step": 29063
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.5171254873275757,
+      "learning_rate": 5.674876203107608e-06,
+      "loss": 0.4807,
+      "step": 29064
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.392648935317993,
+      "learning_rate": 5.674614956734974e-06,
+      "loss": 0.5764,
+      "step": 29065
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 3.085637331008911,
+      "learning_rate": 5.67435370848649e-06,
+      "loss": 0.786,
+      "step": 29066
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 3.3583874702453613,
+      "learning_rate": 5.674092458362877e-06,
+      "loss": 0.3868,
+      "step": 29067
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 3.1408798694610596,
+      "learning_rate": 5.673831206364869e-06,
+      "loss": 0.6858,
+      "step": 29068
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.4880247116088867,
+      "learning_rate": 5.673569952493189e-06,
+      "loss": 0.7812,
+      "step": 29069
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 3.3773128986358643,
+      "learning_rate": 5.673308696748562e-06,
+      "loss": 1.0104,
+      "step": 29070
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 3.770923376083374,
+      "learning_rate": 5.673047439131716e-06,
+      "loss": 0.5622,
+      "step": 29071
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.561758518218994,
+      "learning_rate": 5.672786179643377e-06,
+      "loss": 0.6472,
+      "step": 29072
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.2049968242645264,
+      "learning_rate": 5.672524918284272e-06,
+      "loss": 0.538,
+      "step": 29073
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.7102313041687012,
+      "learning_rate": 5.672263655055128e-06,
+      "loss": 0.6499,
+      "step": 29074
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.1695449352264404,
+      "learning_rate": 5.672002389956668e-06,
+      "loss": 0.5732,
+      "step": 29075
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.0948140621185303,
+      "learning_rate": 5.671741122989623e-06,
+      "loss": 0.615,
+      "step": 29076
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 3.918099880218506,
+      "learning_rate": 5.671479854154718e-06,
+      "loss": 0.5385,
+      "step": 29077
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.917876720428467,
+      "learning_rate": 5.6712185834526775e-06,
+      "loss": 0.811,
+      "step": 29078
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.0715529918670654,
+      "learning_rate": 5.670957310884231e-06,
+      "loss": 0.7813,
+      "step": 29079
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.360797166824341,
+      "learning_rate": 5.6706960364501025e-06,
+      "loss": 0.7279,
+      "step": 29080
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.239842653274536,
+      "learning_rate": 5.67043476015102e-06,
+      "loss": 0.6311,
+      "step": 29081
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.251509189605713,
+      "learning_rate": 5.670173481987711e-06,
+      "loss": 0.4949,
+      "step": 29082
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.7233529090881348,
+      "learning_rate": 5.669912201960899e-06,
+      "loss": 0.7457,
+      "step": 29083
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 3.4218573570251465,
+      "learning_rate": 5.6696509200713125e-06,
+      "loss": 0.6394,
+      "step": 29084
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 3.2501890659332275,
+      "learning_rate": 5.6693896363196766e-06,
+      "loss": 0.8539,
+      "step": 29085
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 3.879251480102539,
+      "learning_rate": 5.66912835070672e-06,
+      "loss": 0.821,
+      "step": 29086
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.337308168411255,
+      "learning_rate": 5.6688670632331675e-06,
+      "loss": 0.3863,
+      "step": 29087
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.515735149383545,
+      "learning_rate": 5.668605773899747e-06,
+      "loss": 0.6681,
+      "step": 29088
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.8253085613250732,
+      "learning_rate": 5.668344482707184e-06,
+      "loss": 0.7417,
+      "step": 29089
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.2509350776672363,
+      "learning_rate": 5.668083189656205e-06,
+      "loss": 0.8396,
+      "step": 29090
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 3.9690911769866943,
+      "learning_rate": 5.667821894747538e-06,
+      "loss": 0.9732,
+      "step": 29091
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 3.331800937652588,
+      "learning_rate": 5.667560597981907e-06,
+      "loss": 0.5805,
+      "step": 29092
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.8031444549560547,
+      "learning_rate": 5.6672992993600416e-06,
+      "loss": 0.3716,
+      "step": 29093
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 3.4187729358673096,
+      "learning_rate": 5.667037998882666e-06,
+      "loss": 0.6467,
+      "step": 29094
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.469331979751587,
+      "learning_rate": 5.666776696550507e-06,
+      "loss": 0.5586,
+      "step": 29095
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 4.383641242980957,
+      "learning_rate": 5.666515392364293e-06,
+      "loss": 0.4164,
+      "step": 29096
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 3.920686960220337,
+      "learning_rate": 5.666254086324748e-06,
+      "loss": 0.7603,
+      "step": 29097
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.194201111793518,
+      "learning_rate": 5.6659927784326e-06,
+      "loss": 0.3986,
+      "step": 29098
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.8397340774536133,
+      "learning_rate": 5.665731468688578e-06,
+      "loss": 0.5311,
+      "step": 29099
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 3.540294885635376,
+      "learning_rate": 5.665470157093404e-06,
+      "loss": 0.4585,
+      "step": 29100
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 4.095560073852539,
+      "learning_rate": 5.665208843647807e-06,
+      "loss": 0.6947,
+      "step": 29101
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.001175880432129,
+      "learning_rate": 5.664947528352515e-06,
+      "loss": 0.7859,
+      "step": 29102
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.4285036325454712,
+      "learning_rate": 5.664686211208251e-06,
+      "loss": 0.4639,
+      "step": 29103
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 3.597748279571533,
+      "learning_rate": 5.664424892215744e-06,
+      "loss": 0.6396,
+      "step": 29104
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.2387592792510986,
+      "learning_rate": 5.664163571375722e-06,
+      "loss": 0.6216,
+      "step": 29105
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.5899176597595215,
+      "learning_rate": 5.663902248688906e-06,
+      "loss": 0.7231,
+      "step": 29106
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.8745012283325195,
+      "learning_rate": 5.6636409241560295e-06,
+      "loss": 0.7288,
+      "step": 29107
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.0052614212036133,
+      "learning_rate": 5.663379597777816e-06,
+      "loss": 0.6384,
+      "step": 29108
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 3.4536306858062744,
+      "learning_rate": 5.663118269554992e-06,
+      "loss": 0.676,
+      "step": 29109
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.9957194328308105,
+      "learning_rate": 5.662856939488285e-06,
+      "loss": 0.6965,
+      "step": 29110
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.124067544937134,
+      "learning_rate": 5.662595607578421e-06,
+      "loss": 0.55,
+      "step": 29111
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 4.326457977294922,
+      "learning_rate": 5.662334273826127e-06,
+      "loss": 0.6132,
+      "step": 29112
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.551563262939453,
+      "learning_rate": 5.66207293823213e-06,
+      "loss": 0.8754,
+      "step": 29113
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.4176199436187744,
+      "learning_rate": 5.661811600797155e-06,
+      "loss": 0.6165,
+      "step": 29114
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 6.350000858306885,
+      "learning_rate": 5.66155026152193e-06,
+      "loss": 0.8374,
+      "step": 29115
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.2286641597747803,
+      "learning_rate": 5.6612889204071805e-06,
+      "loss": 0.7227,
+      "step": 29116
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.9969470500946045,
+      "learning_rate": 5.661027577453635e-06,
+      "loss": 0.635,
+      "step": 29117
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.4701380729675293,
+      "learning_rate": 5.66076623266202e-06,
+      "loss": 0.4297,
+      "step": 29118
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.8170415163040161,
+      "learning_rate": 5.660504886033063e-06,
+      "loss": 0.4635,
+      "step": 29119
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.794297695159912,
+      "learning_rate": 5.660243537567487e-06,
+      "loss": 0.8352,
+      "step": 29120
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.167696237564087,
+      "learning_rate": 5.6599821872660224e-06,
+      "loss": 0.6766,
+      "step": 29121
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.7961335182189941,
+      "learning_rate": 5.659720835129394e-06,
+      "loss": 0.7068,
+      "step": 29122
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.163756847381592,
+      "learning_rate": 5.659459481158328e-06,
+      "loss": 0.5057,
+      "step": 29123
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 12.187287330627441,
+      "learning_rate": 5.659198125353553e-06,
+      "loss": 0.566,
+      "step": 29124
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 4.6562604904174805,
+      "learning_rate": 5.6589367677157956e-06,
+      "loss": 0.73,
+      "step": 29125
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.3811798095703125,
+      "learning_rate": 5.658675408245782e-06,
+      "loss": 0.6546,
+      "step": 29126
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 3.428727626800537,
+      "learning_rate": 5.658414046944239e-06,
+      "loss": 0.7105,
+      "step": 29127
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 3.8518881797790527,
+      "learning_rate": 5.658152683811893e-06,
+      "loss": 0.4381,
+      "step": 29128
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.9217941761016846,
+      "learning_rate": 5.657891318849469e-06,
+      "loss": 0.5727,
+      "step": 29129
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.8268909454345703,
+      "learning_rate": 5.657629952057697e-06,
+      "loss": 0.6205,
+      "step": 29130
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.2862350940704346,
+      "learning_rate": 5.657368583437304e-06,
+      "loss": 0.7843,
+      "step": 29131
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.4131367206573486,
+      "learning_rate": 5.657107212989013e-06,
+      "loss": 0.644,
+      "step": 29132
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.407137155532837,
+      "learning_rate": 5.656845840713554e-06,
+      "loss": 0.7004,
+      "step": 29133
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.843680500984192,
+      "learning_rate": 5.656584466611653e-06,
+      "loss": 0.7414,
+      "step": 29134
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.1068460941314697,
+      "learning_rate": 5.656323090684037e-06,
+      "loss": 0.3725,
+      "step": 29135
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.553316593170166,
+      "learning_rate": 5.656061712931431e-06,
+      "loss": 0.5269,
+      "step": 29136
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.521958112716675,
+      "learning_rate": 5.655800333354563e-06,
+      "loss": 0.6648,
+      "step": 29137
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.553514003753662,
+      "learning_rate": 5.655538951954161e-06,
+      "loss": 0.587,
+      "step": 29138
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 3.2510550022125244,
+      "learning_rate": 5.65527756873095e-06,
+      "loss": 0.6416,
+      "step": 29139
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.7111071348190308,
+      "learning_rate": 5.655016183685659e-06,
+      "loss": 0.777,
+      "step": 29140
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.0068836212158203,
+      "learning_rate": 5.654754796819012e-06,
+      "loss": 0.6432,
+      "step": 29141
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 3.8936607837677,
+      "learning_rate": 5.654493408131738e-06,
+      "loss": 0.4433,
+      "step": 29142
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 6.233859062194824,
+      "learning_rate": 5.654232017624563e-06,
+      "loss": 0.5444,
+      "step": 29143
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 4.284278869628906,
+      "learning_rate": 5.653970625298213e-06,
+      "loss": 0.8131,
+      "step": 29144
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.5969130992889404,
+      "learning_rate": 5.653709231153417e-06,
+      "loss": 0.515,
+      "step": 29145
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.2728445529937744,
+      "learning_rate": 5.653447835190898e-06,
+      "loss": 0.5284,
+      "step": 29146
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.1384170055389404,
+      "learning_rate": 5.653186437411388e-06,
+      "loss": 0.6304,
+      "step": 29147
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.7000672817230225,
+      "learning_rate": 5.6529250378156095e-06,
+      "loss": 0.3934,
+      "step": 29148
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.507279396057129,
+      "learning_rate": 5.652663636404292e-06,
+      "loss": 0.8692,
+      "step": 29149
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 6.392444133758545,
+      "learning_rate": 5.652402233178162e-06,
+      "loss": 0.5743,
+      "step": 29150
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.753806233406067,
+      "learning_rate": 5.652140828137944e-06,
+      "loss": 0.625,
+      "step": 29151
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.666531562805176,
+      "learning_rate": 5.6518794212843674e-06,
+      "loss": 0.9105,
+      "step": 29152
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 3.44732403755188,
+      "learning_rate": 5.6516180126181595e-06,
+      "loss": 0.6562,
+      "step": 29153
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 4.228507995605469,
+      "learning_rate": 5.651356602140045e-06,
+      "loss": 0.4665,
+      "step": 29154
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.3289947509765625,
+      "learning_rate": 5.651095189850752e-06,
+      "loss": 0.507,
+      "step": 29155
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 3.503276824951172,
+      "learning_rate": 5.650833775751006e-06,
+      "loss": 0.658,
+      "step": 29156
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.6250393390655518,
+      "learning_rate": 5.6505723598415365e-06,
+      "loss": 0.7934,
+      "step": 29157
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.1320559978485107,
+      "learning_rate": 5.650310942123069e-06,
+      "loss": 0.6555,
+      "step": 29158
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 3.5814051628112793,
+      "learning_rate": 5.650049522596329e-06,
+      "loss": 0.7189,
+      "step": 29159
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.18200421333313,
+      "learning_rate": 5.649788101262046e-06,
+      "loss": 0.4946,
+      "step": 29160
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.1793429851531982,
+      "learning_rate": 5.649526678120946e-06,
+      "loss": 0.7867,
+      "step": 29161
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 3.268547296524048,
+      "learning_rate": 5.649265253173755e-06,
+      "loss": 0.7146,
+      "step": 29162
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.8563549518585205,
+      "learning_rate": 5.649003826421201e-06,
+      "loss": 0.549,
+      "step": 29163
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.492238998413086,
+      "learning_rate": 5.64874239786401e-06,
+      "loss": 0.6684,
+      "step": 29164
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 5.074670314788818,
+      "learning_rate": 5.648480967502909e-06,
+      "loss": 0.6393,
+      "step": 29165
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.938643455505371,
+      "learning_rate": 5.6482195353386275e-06,
+      "loss": 0.6223,
+      "step": 29166
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 15.152512550354004,
+      "learning_rate": 5.647958101371887e-06,
+      "loss": 0.5794,
+      "step": 29167
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.9825407266616821,
+      "learning_rate": 5.64769666560342e-06,
+      "loss": 0.6458,
+      "step": 29168
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 5.668345928192139,
+      "learning_rate": 5.647435228033951e-06,
+      "loss": 0.5406,
+      "step": 29169
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.2433249950408936,
+      "learning_rate": 5.647173788664206e-06,
+      "loss": 0.7097,
+      "step": 29170
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.433082342147827,
+      "learning_rate": 5.646912347494914e-06,
+      "loss": 0.5859,
+      "step": 29171
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.068920612335205,
+      "learning_rate": 5.646650904526801e-06,
+      "loss": 0.8969,
+      "step": 29172
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.0984866619110107,
+      "learning_rate": 5.646389459760595e-06,
+      "loss": 0.4043,
+      "step": 29173
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 3.0053210258483887,
+      "learning_rate": 5.646128013197021e-06,
+      "loss": 0.7591,
+      "step": 29174
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.3153932094573975,
+      "learning_rate": 5.645866564836807e-06,
+      "loss": 0.557,
+      "step": 29175
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.8059566020965576,
+      "learning_rate": 5.645605114680681e-06,
+      "loss": 0.5572,
+      "step": 29176
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.8491028547286987,
+      "learning_rate": 5.645343662729367e-06,
+      "loss": 0.9212,
+      "step": 29177
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.5538225173950195,
+      "learning_rate": 5.645082208983595e-06,
+      "loss": 0.6109,
+      "step": 29178
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.9531848430633545,
+      "learning_rate": 5.644820753444092e-06,
+      "loss": 0.8757,
+      "step": 29179
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 3.9184324741363525,
+      "learning_rate": 5.644559296111584e-06,
+      "loss": 0.5582,
+      "step": 29180
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 4.564438343048096,
+      "learning_rate": 5.6442978369867964e-06,
+      "loss": 0.7276,
+      "step": 29181
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.3732450008392334,
+      "learning_rate": 5.64403637607046e-06,
+      "loss": 0.9081,
+      "step": 29182
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.386967420578003,
+      "learning_rate": 5.643774913363298e-06,
+      "loss": 0.6086,
+      "step": 29183
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.0366718769073486,
+      "learning_rate": 5.64351344886604e-06,
+      "loss": 0.9029,
+      "step": 29184
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 3.909360647201538,
+      "learning_rate": 5.643251982579411e-06,
+      "loss": 0.8105,
+      "step": 29185
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 3.910309076309204,
+      "learning_rate": 5.64299051450414e-06,
+      "loss": 0.6228,
+      "step": 29186
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 4.970645427703857,
+      "learning_rate": 5.642729044640953e-06,
+      "loss": 0.634,
+      "step": 29187
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.8774702548980713,
+      "learning_rate": 5.642467572990579e-06,
+      "loss": 0.5452,
+      "step": 29188
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.773301124572754,
+      "learning_rate": 5.642206099553742e-06,
+      "loss": 0.6216,
+      "step": 29189
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.8828606605529785,
+      "learning_rate": 5.641944624331171e-06,
+      "loss": 0.7845,
+      "step": 29190
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.848921060562134,
+      "learning_rate": 5.64168314732359e-06,
+      "loss": 1.0374,
+      "step": 29191
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.5278701782226562,
+      "learning_rate": 5.64142166853173e-06,
+      "loss": 0.5311,
+      "step": 29192
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 5.335574150085449,
+      "learning_rate": 5.641160187956317e-06,
+      "loss": 0.7265,
+      "step": 29193
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.7151849269866943,
+      "learning_rate": 5.640898705598078e-06,
+      "loss": 0.5768,
+      "step": 29194
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 7.662405967712402,
+      "learning_rate": 5.64063722145774e-06,
+      "loss": 0.8326,
+      "step": 29195
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.1777422428131104,
+      "learning_rate": 5.64037573553603e-06,
+      "loss": 0.5352,
+      "step": 29196
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.9869887828826904,
+      "learning_rate": 5.640114247833674e-06,
+      "loss": 0.7943,
+      "step": 29197
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.210171699523926,
+      "learning_rate": 5.639852758351399e-06,
+      "loss": 0.4881,
+      "step": 29198
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.249516010284424,
+      "learning_rate": 5.639591267089935e-06,
+      "loss": 0.5864,
+      "step": 29199
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 3.5791070461273193,
+      "learning_rate": 5.639329774050007e-06,
+      "loss": 0.578,
+      "step": 29200
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.0209872722625732,
+      "learning_rate": 5.6390682792323426e-06,
+      "loss": 0.7314,
+      "step": 29201
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.2090961933135986,
+      "learning_rate": 5.638806782637669e-06,
+      "loss": 0.5674,
+      "step": 29202
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.4902303218841553,
+      "learning_rate": 5.638545284266712e-06,
+      "loss": 0.7995,
+      "step": 29203
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 5.311319828033447,
+      "learning_rate": 5.638283784120201e-06,
+      "loss": 0.59,
+      "step": 29204
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.683696746826172,
+      "learning_rate": 5.638022282198861e-06,
+      "loss": 0.7237,
+      "step": 29205
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.0794272422790527,
+      "learning_rate": 5.637760778503421e-06,
+      "loss": 0.697,
+      "step": 29206
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.411726951599121,
+      "learning_rate": 5.637499273034605e-06,
+      "loss": 0.5025,
+      "step": 29207
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 3.4181880950927734,
+      "learning_rate": 5.6372377657931435e-06,
+      "loss": 0.6433,
+      "step": 29208
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 4.05644416809082,
+      "learning_rate": 5.636976256779762e-06,
+      "loss": 0.534,
+      "step": 29209
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.2357447147369385,
+      "learning_rate": 5.63671474599519e-06,
+      "loss": 0.8013,
+      "step": 29210
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.502195119857788,
+      "learning_rate": 5.636453233440151e-06,
+      "loss": 0.4257,
+      "step": 29211
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 3.7431142330169678,
+      "learning_rate": 5.636191719115375e-06,
+      "loss": 0.7194,
+      "step": 29212
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.774795413017273,
+      "learning_rate": 5.6359302030215876e-06,
+      "loss": 0.6212,
+      "step": 29213
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.9937403202056885,
+      "learning_rate": 5.635668685159517e-06,
+      "loss": 0.6467,
+      "step": 29214
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.6946752071380615,
+      "learning_rate": 5.63540716552989e-06,
+      "loss": 0.5558,
+      "step": 29215
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 3.2544443607330322,
+      "learning_rate": 5.635145644133433e-06,
+      "loss": 0.7196,
+      "step": 29216
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.20698618888855,
+      "learning_rate": 5.6348841209708734e-06,
+      "loss": 0.4193,
+      "step": 29217
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.4628379344940186,
+      "learning_rate": 5.63462259604294e-06,
+      "loss": 0.8003,
+      "step": 29218
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.507688045501709,
+      "learning_rate": 5.634361069350359e-06,
+      "loss": 0.6766,
+      "step": 29219
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.9585940837860107,
+      "learning_rate": 5.634099540893857e-06,
+      "loss": 0.563,
+      "step": 29220
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 3.2149410247802734,
+      "learning_rate": 5.633838010674162e-06,
+      "loss": 0.7626,
+      "step": 29221
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 4.514256954193115,
+      "learning_rate": 5.633576478692001e-06,
+      "loss": 0.4178,
+      "step": 29222
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.8984097242355347,
+      "learning_rate": 5.633314944948101e-06,
+      "loss": 0.6139,
+      "step": 29223
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.914024829864502,
+      "learning_rate": 5.633053409443189e-06,
+      "loss": 0.4947,
+      "step": 29224
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 8.218805313110352,
+      "learning_rate": 5.632791872177993e-06,
+      "loss": 0.5924,
+      "step": 29225
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.7350573539733887,
+      "learning_rate": 5.63253033315324e-06,
+      "loss": 0.5971,
+      "step": 29226
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.1354494094848633,
+      "learning_rate": 5.632268792369658e-06,
+      "loss": 0.5777,
+      "step": 29227
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.7228457927703857,
+      "learning_rate": 5.63200724982797e-06,
+      "loss": 0.3929,
+      "step": 29228
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.2277774810791016,
+      "learning_rate": 5.631745705528909e-06,
+      "loss": 0.8204,
+      "step": 29229
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.8976116180419922,
+      "learning_rate": 5.631484159473199e-06,
+      "loss": 0.5407,
+      "step": 29230
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.1186211109161377,
+      "learning_rate": 5.6312226116615695e-06,
+      "loss": 0.8529,
+      "step": 29231
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 4.290999412536621,
+      "learning_rate": 5.630961062094746e-06,
+      "loss": 0.9946,
+      "step": 29232
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 3.619542121887207,
+      "learning_rate": 5.6306995107734566e-06,
+      "loss": 0.7205,
+      "step": 29233
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.938122034072876,
+      "learning_rate": 5.630437957698427e-06,
+      "loss": 0.695,
+      "step": 29234
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.95697021484375,
+      "learning_rate": 5.630176402870386e-06,
+      "loss": 0.4728,
+      "step": 29235
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 3.3420228958129883,
+      "learning_rate": 5.62991484629006e-06,
+      "loss": 0.6611,
+      "step": 29236
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.8492345809936523,
+      "learning_rate": 5.629653287958178e-06,
+      "loss": 0.7404,
+      "step": 29237
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.6810824871063232,
+      "learning_rate": 5.629391727875464e-06,
+      "loss": 0.4903,
+      "step": 29238
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.3678665161132812,
+      "learning_rate": 5.629130166042649e-06,
+      "loss": 0.4399,
+      "step": 29239
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.8089499473571777,
+      "learning_rate": 5.628868602460458e-06,
+      "loss": 0.5051,
+      "step": 29240
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.379985809326172,
+      "learning_rate": 5.6286070371296195e-06,
+      "loss": 0.5409,
+      "step": 29241
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.542231559753418,
+      "learning_rate": 5.62834547005086e-06,
+      "loss": 0.5817,
+      "step": 29242
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 3.27021861076355,
+      "learning_rate": 5.628083901224908e-06,
+      "loss": 0.4746,
+      "step": 29243
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.2007744312286377,
+      "learning_rate": 5.627822330652489e-06,
+      "loss": 0.8287,
+      "step": 29244
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.7752888202667236,
+      "learning_rate": 5.627560758334332e-06,
+      "loss": 0.7702,
+      "step": 29245
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.9953023195266724,
+      "learning_rate": 5.627299184271163e-06,
+      "loss": 0.4546,
+      "step": 29246
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.3573906421661377,
+      "learning_rate": 5.627037608463709e-06,
+      "loss": 0.7104,
+      "step": 29247
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.3953027725219727,
+      "learning_rate": 5.6267760309127e-06,
+      "loss": 0.535,
+      "step": 29248
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 3.068319320678711,
+      "learning_rate": 5.626514451618861e-06,
+      "loss": 0.374,
+      "step": 29249
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.180447578430176,
+      "learning_rate": 5.62625287058292e-06,
+      "loss": 0.5507,
+      "step": 29250
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.852983236312866,
+      "learning_rate": 5.625991287805605e-06,
+      "loss": 0.6216,
+      "step": 29251
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 3.4885592460632324,
+      "learning_rate": 5.6257297032876414e-06,
+      "loss": 0.5936,
+      "step": 29252
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 4.330453395843506,
+      "learning_rate": 5.625468117029758e-06,
+      "loss": 0.4387,
+      "step": 29253
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.8682109117507935,
+      "learning_rate": 5.625206529032683e-06,
+      "loss": 0.6425,
+      "step": 29254
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 4.552891731262207,
+      "learning_rate": 5.624944939297142e-06,
+      "loss": 0.6987,
+      "step": 29255
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.617431640625,
+      "learning_rate": 5.624683347823865e-06,
+      "loss": 0.799,
+      "step": 29256
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.775186061859131,
+      "learning_rate": 5.624421754613576e-06,
+      "loss": 0.7336,
+      "step": 29257
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 3.7097458839416504,
+      "learning_rate": 5.624160159667004e-06,
+      "loss": 0.609,
+      "step": 29258
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 3.0291571617126465,
+      "learning_rate": 5.623898562984877e-06,
+      "loss": 0.7324,
+      "step": 29259
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.9878954887390137,
+      "learning_rate": 5.623636964567921e-06,
+      "loss": 0.4578,
+      "step": 29260
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.6792049407958984,
+      "learning_rate": 5.623375364416866e-06,
+      "loss": 0.8247,
+      "step": 29261
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.122084617614746,
+      "learning_rate": 5.623113762532436e-06,
+      "loss": 0.7012,
+      "step": 29262
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.3172123432159424,
+      "learning_rate": 5.622852158915361e-06,
+      "loss": 0.4349,
+      "step": 29263
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.9232081174850464,
+      "learning_rate": 5.622590553566367e-06,
+      "loss": 0.7717,
+      "step": 29264
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.366954803466797,
+      "learning_rate": 5.622328946486182e-06,
+      "loss": 0.3511,
+      "step": 29265
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.4160890579223633,
+      "learning_rate": 5.622067337675533e-06,
+      "loss": 0.6768,
+      "step": 29266
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 3.4108493328094482,
+      "learning_rate": 5.621805727135149e-06,
+      "loss": 0.7108,
+      "step": 29267
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 9.487455368041992,
+      "learning_rate": 5.621544114865753e-06,
+      "loss": 0.5136,
+      "step": 29268
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.7551286220550537,
+      "learning_rate": 5.6212825008680795e-06,
+      "loss": 0.6904,
+      "step": 29269
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.6757960319519043,
+      "learning_rate": 5.621020885142851e-06,
+      "loss": 0.7795,
+      "step": 29270
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 6.1708502769470215,
+      "learning_rate": 5.620759267690795e-06,
+      "loss": 0.8833,
+      "step": 29271
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.022658586502075,
+      "learning_rate": 5.62049764851264e-06,
+      "loss": 0.7366,
+      "step": 29272
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 3.004173755645752,
+      "learning_rate": 5.620236027609115e-06,
+      "loss": 0.6994,
+      "step": 29273
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.8671565055847168,
+      "learning_rate": 5.619974404980945e-06,
+      "loss": 0.5839,
+      "step": 29274
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.7746763229370117,
+      "learning_rate": 5.619712780628858e-06,
+      "loss": 0.6498,
+      "step": 29275
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 5.113682746887207,
+      "learning_rate": 5.619451154553582e-06,
+      "loss": 0.5741,
+      "step": 29276
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.7010204792022705,
+      "learning_rate": 5.619189526755846e-06,
+      "loss": 0.7289,
+      "step": 29277
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.464632987976074,
+      "learning_rate": 5.618927897236373e-06,
+      "loss": 0.5747,
+      "step": 29278
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.3913559913635254,
+      "learning_rate": 5.618666265995896e-06,
+      "loss": 0.6163,
+      "step": 29279
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 6.9549055099487305,
+      "learning_rate": 5.618404633035138e-06,
+      "loss": 0.6752,
+      "step": 29280
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.5902594327926636,
+      "learning_rate": 5.61814299835483e-06,
+      "loss": 0.3329,
+      "step": 29281
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.059601068496704,
+      "learning_rate": 5.617881361955697e-06,
+      "loss": 0.7158,
+      "step": 29282
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.2858667373657227,
+      "learning_rate": 5.617619723838468e-06,
+      "loss": 0.438,
+      "step": 29283
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.2706000804901123,
+      "learning_rate": 5.617358084003869e-06,
+      "loss": 0.7317,
+      "step": 29284
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.3449416160583496,
+      "learning_rate": 5.617096442452629e-06,
+      "loss": 0.5142,
+      "step": 29285
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 3.2473251819610596,
+      "learning_rate": 5.6168347991854755e-06,
+      "loss": 0.6038,
+      "step": 29286
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 5.39579439163208,
+      "learning_rate": 5.616573154203134e-06,
+      "loss": 0.6096,
+      "step": 29287
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.935003399848938,
+      "learning_rate": 5.6163115075063345e-06,
+      "loss": 0.5432,
+      "step": 29288
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.7688071727752686,
+      "learning_rate": 5.6160498590958035e-06,
+      "loss": 0.6293,
+      "step": 29289
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.8264861106872559,
+      "learning_rate": 5.6157882089722685e-06,
+      "loss": 0.515,
+      "step": 29290
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.8430544137954712,
+      "learning_rate": 5.615526557136457e-06,
+      "loss": 0.5866,
+      "step": 29291
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.9302247762680054,
+      "learning_rate": 5.6152649035890975e-06,
+      "loss": 0.8797,
+      "step": 29292
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 3.1676788330078125,
+      "learning_rate": 5.615003248330916e-06,
+      "loss": 0.677,
+      "step": 29293
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.1732382774353027,
+      "learning_rate": 5.6147415913626415e-06,
+      "loss": 0.6708,
+      "step": 29294
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 3.085700273513794,
+      "learning_rate": 5.6144799326850005e-06,
+      "loss": 0.4569,
+      "step": 29295
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 3.7431910037994385,
+      "learning_rate": 5.614218272298721e-06,
+      "loss": 0.6666,
+      "step": 29296
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 3.1650567054748535,
+      "learning_rate": 5.61395661020453e-06,
+      "loss": 0.7038,
+      "step": 29297
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.65487539768219,
+      "learning_rate": 5.6136949464031566e-06,
+      "loss": 0.5557,
+      "step": 29298
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 3.749055862426758,
+      "learning_rate": 5.6134332808953265e-06,
+      "loss": 0.9358,
+      "step": 29299
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.7364485263824463,
+      "learning_rate": 5.613171613681769e-06,
+      "loss": 0.7144,
+      "step": 29300
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.1934118270874023,
+      "learning_rate": 5.61290994476321e-06,
+      "loss": 0.5109,
+      "step": 29301
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 3.0165719985961914,
+      "learning_rate": 5.6126482741403795e-06,
+      "loss": 0.7416,
+      "step": 29302
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.210409164428711,
+      "learning_rate": 5.6123866018140036e-06,
+      "loss": 0.7895,
+      "step": 29303
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.5774261951446533,
+      "learning_rate": 5.6121249277848086e-06,
+      "loss": 0.6989,
+      "step": 29304
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.5911718606948853,
+      "learning_rate": 5.611863252053525e-06,
+      "loss": 0.6203,
+      "step": 29305
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.4301695823669434,
+      "learning_rate": 5.611601574620878e-06,
+      "loss": 0.698,
+      "step": 29306
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.7980577945709229,
+      "learning_rate": 5.6113398954875966e-06,
+      "loss": 0.7497,
+      "step": 29307
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.8135199546813965,
+      "learning_rate": 5.611078214654406e-06,
+      "loss": 0.6596,
+      "step": 29308
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.114962577819824,
+      "learning_rate": 5.610816532122039e-06,
+      "loss": 0.6875,
+      "step": 29309
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.175267457962036,
+      "learning_rate": 5.610554847891218e-06,
+      "loss": 0.7735,
+      "step": 29310
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.9755125045776367,
+      "learning_rate": 5.610293161962674e-06,
+      "loss": 0.8828,
+      "step": 29311
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 3.4190235137939453,
+      "learning_rate": 5.610031474337132e-06,
+      "loss": 0.6438,
+      "step": 29312
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.41690194606781,
+      "learning_rate": 5.609769785015321e-06,
+      "loss": 0.481,
+      "step": 29313
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 3.0930233001708984,
+      "learning_rate": 5.609508093997968e-06,
+      "loss": 0.8238,
+      "step": 29314
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.343529462814331,
+      "learning_rate": 5.609246401285803e-06,
+      "loss": 0.8901,
+      "step": 29315
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.9431979656219482,
+      "learning_rate": 5.6089847068795524e-06,
+      "loss": 0.8729,
+      "step": 29316
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 3.8193109035491943,
+      "learning_rate": 5.608723010779942e-06,
+      "loss": 0.2489,
+      "step": 29317
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 8.193083763122559,
+      "learning_rate": 5.608461312987702e-06,
+      "loss": 0.5091,
+      "step": 29318
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 6.452747344970703,
+      "learning_rate": 5.608199613503558e-06,
+      "loss": 0.5837,
+      "step": 29319
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.3868167400360107,
+      "learning_rate": 5.607937912328238e-06,
+      "loss": 0.6461,
+      "step": 29320
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.70668888092041,
+      "learning_rate": 5.607676209462472e-06,
+      "loss": 0.5928,
+      "step": 29321
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.6945571899414062,
+      "learning_rate": 5.607414504906986e-06,
+      "loss": 0.3871,
+      "step": 29322
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.208761692047119,
+      "learning_rate": 5.607152798662507e-06,
+      "loss": 0.5787,
+      "step": 29323
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.1783268451690674,
+      "learning_rate": 5.606891090729765e-06,
+      "loss": 0.7478,
+      "step": 29324
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 3.258466958999634,
+      "learning_rate": 5.606629381109485e-06,
+      "loss": 0.5683,
+      "step": 29325
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 10.411436080932617,
+      "learning_rate": 5.606367669802396e-06,
+      "loss": 0.9868,
+      "step": 29326
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.840686798095703,
+      "learning_rate": 5.606105956809226e-06,
+      "loss": 0.6898,
+      "step": 29327
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.7960494756698608,
+      "learning_rate": 5.605844242130702e-06,
+      "loss": 0.3963,
+      "step": 29328
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.736793279647827,
+      "learning_rate": 5.605582525767553e-06,
+      "loss": 0.8215,
+      "step": 29329
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.5184755325317383,
+      "learning_rate": 5.605320807720505e-06,
+      "loss": 0.5687,
+      "step": 29330
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.9660379886627197,
+      "learning_rate": 5.605059087990285e-06,
+      "loss": 0.4969,
+      "step": 29331
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.5921149253845215,
+      "learning_rate": 5.604797366577626e-06,
+      "loss": 0.5783,
+      "step": 29332
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 3.0836238861083984,
+      "learning_rate": 5.60453564348325e-06,
+      "loss": 0.7441,
+      "step": 29333
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 4.143434047698975,
+      "learning_rate": 5.604273918707886e-06,
+      "loss": 0.5832,
+      "step": 29334
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.952580451965332,
+      "learning_rate": 5.604012192252264e-06,
+      "loss": 0.6983,
+      "step": 29335
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.1756815910339355,
+      "learning_rate": 5.60375046411711e-06,
+      "loss": 0.6126,
+      "step": 29336
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.54015851020813,
+      "learning_rate": 5.603488734303153e-06,
+      "loss": 0.576,
+      "step": 29337
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.0322134494781494,
+      "learning_rate": 5.603227002811119e-06,
+      "loss": 0.7219,
+      "step": 29338
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 3.111443281173706,
+      "learning_rate": 5.6029652696417355e-06,
+      "loss": 0.9541,
+      "step": 29339
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.8074400424957275,
+      "learning_rate": 5.6027035347957325e-06,
+      "loss": 0.7257,
+      "step": 29340
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.4342644214630127,
+      "learning_rate": 5.602441798273837e-06,
+      "loss": 0.4601,
+      "step": 29341
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.730318546295166,
+      "learning_rate": 5.602180060076777e-06,
+      "loss": 0.5886,
+      "step": 29342
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.0025875568389893,
+      "learning_rate": 5.601918320205279e-06,
+      "loss": 0.5541,
+      "step": 29343
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.0451951026916504,
+      "learning_rate": 5.601656578660073e-06,
+      "loss": 0.6377,
+      "step": 29344
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 4.575465679168701,
+      "learning_rate": 5.601394835441885e-06,
+      "loss": 0.4017,
+      "step": 29345
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.0929219722747803,
+      "learning_rate": 5.601133090551441e-06,
+      "loss": 0.7208,
+      "step": 29346
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 3.1014981269836426,
+      "learning_rate": 5.600871343989473e-06,
+      "loss": 0.5618,
+      "step": 29347
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.0372958183288574,
+      "learning_rate": 5.6006095957567074e-06,
+      "loss": 0.6193,
+      "step": 29348
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.9704265594482422,
+      "learning_rate": 5.60034784585387e-06,
+      "loss": 0.575,
+      "step": 29349
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.159066915512085,
+      "learning_rate": 5.600086094281691e-06,
+      "loss": 0.6683,
+      "step": 29350
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.297212839126587,
+      "learning_rate": 5.599824341040897e-06,
+      "loss": 0.6916,
+      "step": 29351
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 3.631554126739502,
+      "learning_rate": 5.599562586132217e-06,
+      "loss": 0.6085,
+      "step": 29352
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.621577739715576,
+      "learning_rate": 5.599300829556376e-06,
+      "loss": 0.6514,
+      "step": 29353
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.4833505153656006,
+      "learning_rate": 5.599039071314106e-06,
+      "loss": 0.8885,
+      "step": 29354
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.3666014671325684,
+      "learning_rate": 5.598777311406133e-06,
+      "loss": 0.4799,
+      "step": 29355
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 3.697610378265381,
+      "learning_rate": 5.598515549833183e-06,
+      "loss": 0.4852,
+      "step": 29356
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.2191083431243896,
+      "learning_rate": 5.598253786595986e-06,
+      "loss": 0.5479,
+      "step": 29357
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.0487475395202637,
+      "learning_rate": 5.59799202169527e-06,
+      "loss": 0.8042,
+      "step": 29358
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.8923068046569824,
+      "learning_rate": 5.597730255131761e-06,
+      "loss": 0.4045,
+      "step": 29359
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.3471617698669434,
+      "learning_rate": 5.597468486906189e-06,
+      "loss": 0.7068,
+      "step": 29360
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.2331762313842773,
+      "learning_rate": 5.59720671701928e-06,
+      "loss": 0.4504,
+      "step": 29361
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.2566213607788086,
+      "learning_rate": 5.596944945471764e-06,
+      "loss": 0.4263,
+      "step": 29362
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.532278060913086,
+      "learning_rate": 5.596683172264368e-06,
+      "loss": 0.5951,
+      "step": 29363
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.0674357414245605,
+      "learning_rate": 5.596421397397819e-06,
+      "loss": 0.705,
+      "step": 29364
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.2337288856506348,
+      "learning_rate": 5.596159620872845e-06,
+      "loss": 0.5884,
+      "step": 29365
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 4.340366840362549,
+      "learning_rate": 5.595897842690176e-06,
+      "loss": 0.9048,
+      "step": 29366
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.7220754623413086,
+      "learning_rate": 5.595636062850537e-06,
+      "loss": 0.5858,
+      "step": 29367
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 3.1968793869018555,
+      "learning_rate": 5.595374281354658e-06,
+      "loss": 0.5085,
+      "step": 29368
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.9765379428863525,
+      "learning_rate": 5.5951124982032644e-06,
+      "loss": 0.8333,
+      "step": 29369
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.34878671169281,
+      "learning_rate": 5.594850713397086e-06,
+      "loss": 0.4413,
+      "step": 29370
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 5.428833484649658,
+      "learning_rate": 5.594588926936853e-06,
+      "loss": 0.7772,
+      "step": 29371
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.8813300132751465,
+      "learning_rate": 5.594327138823289e-06,
+      "loss": 0.5135,
+      "step": 29372
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.3062875270843506,
+      "learning_rate": 5.594065349057124e-06,
+      "loss": 0.6671,
+      "step": 29373
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.8600162267684937,
+      "learning_rate": 5.593803557639087e-06,
+      "loss": 0.504,
+      "step": 29374
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 3.0633955001831055,
+      "learning_rate": 5.593541764569904e-06,
+      "loss": 0.6502,
+      "step": 29375
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.359543800354004,
+      "learning_rate": 5.5932799698503024e-06,
+      "loss": 0.4281,
+      "step": 29376
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.802124500274658,
+      "learning_rate": 5.593018173481013e-06,
+      "loss": 0.6476,
+      "step": 29377
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 4.246407985687256,
+      "learning_rate": 5.592756375462762e-06,
+      "loss": 0.6624,
+      "step": 29378
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.320805549621582,
+      "learning_rate": 5.592494575796278e-06,
+      "loss": 0.5848,
+      "step": 29379
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.8458516597747803,
+      "learning_rate": 5.592232774482288e-06,
+      "loss": 0.5714,
+      "step": 29380
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.8297027349472046,
+      "learning_rate": 5.5919709715215206e-06,
+      "loss": 0.6479,
+      "step": 29381
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.949385404586792,
+      "learning_rate": 5.591709166914703e-06,
+      "loss": 0.5268,
+      "step": 29382
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.3642685413360596,
+      "learning_rate": 5.591447360662564e-06,
+      "loss": 0.7816,
+      "step": 29383
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 7.382896423339844,
+      "learning_rate": 5.591185552765832e-06,
+      "loss": 0.3804,
+      "step": 29384
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.6270434856414795,
+      "learning_rate": 5.590923743225235e-06,
+      "loss": 0.6456,
+      "step": 29385
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.9706016778945923,
+      "learning_rate": 5.5906619320415005e-06,
+      "loss": 0.5068,
+      "step": 29386
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.928777813911438,
+      "learning_rate": 5.5904001192153556e-06,
+      "loss": 0.3584,
+      "step": 29387
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.688603401184082,
+      "learning_rate": 5.590138304747529e-06,
+      "loss": 0.6537,
+      "step": 29388
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.613312244415283,
+      "learning_rate": 5.589876488638749e-06,
+      "loss": 0.6962,
+      "step": 29389
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.20180606842041,
+      "learning_rate": 5.589614670889743e-06,
+      "loss": 0.4003,
+      "step": 29390
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.962343215942383,
+      "learning_rate": 5.58935285150124e-06,
+      "loss": 0.3761,
+      "step": 29391
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 3.1763570308685303,
+      "learning_rate": 5.589091030473968e-06,
+      "loss": 0.8002,
+      "step": 29392
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.2751026153564453,
+      "learning_rate": 5.588829207808653e-06,
+      "loss": 0.671,
+      "step": 29393
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.6789509057998657,
+      "learning_rate": 5.588567383506027e-06,
+      "loss": 0.4527,
+      "step": 29394
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 3.361553192138672,
+      "learning_rate": 5.588305557566814e-06,
+      "loss": 0.7939,
+      "step": 29395
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.8527801036834717,
+      "learning_rate": 5.5880437299917445e-06,
+      "loss": 0.4463,
+      "step": 29396
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.6896347999572754,
+      "learning_rate": 5.587781900781545e-06,
+      "loss": 0.657,
+      "step": 29397
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.7814645767211914,
+      "learning_rate": 5.587520069936945e-06,
+      "loss": 0.4689,
+      "step": 29398
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 12.2978515625,
+      "learning_rate": 5.587258237458671e-06,
+      "loss": 0.5318,
+      "step": 29399
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 3.324876546859741,
+      "learning_rate": 5.586996403347451e-06,
+      "loss": 0.719,
+      "step": 29400
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.6356375217437744,
+      "learning_rate": 5.586734567604014e-06,
+      "loss": 0.5236,
+      "step": 29401
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.457337737083435,
+      "learning_rate": 5.586472730229089e-06,
+      "loss": 0.5496,
+      "step": 29402
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.9846588373184204,
+      "learning_rate": 5.586210891223404e-06,
+      "loss": 0.2991,
+      "step": 29403
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.998467206954956,
+      "learning_rate": 5.585949050587684e-06,
+      "loss": 0.5384,
+      "step": 29404
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.9169632196426392,
+      "learning_rate": 5.585687208322661e-06,
+      "loss": 0.4844,
+      "step": 29405
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 3.090758800506592,
+      "learning_rate": 5.58542536442906e-06,
+      "loss": 0.5137,
+      "step": 29406
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.1309030055999756,
+      "learning_rate": 5.5851635189076105e-06,
+      "loss": 0.8413,
+      "step": 29407
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 4.489197731018066,
+      "learning_rate": 5.584901671759041e-06,
+      "loss": 0.4102,
+      "step": 29408
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.632135272026062,
+      "learning_rate": 5.584639822984079e-06,
+      "loss": 0.479,
+      "step": 29409
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 3.3700168132781982,
+      "learning_rate": 5.584377972583453e-06,
+      "loss": 0.8793,
+      "step": 29410
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.8187825679779053,
+      "learning_rate": 5.5841161205578886e-06,
+      "loss": 0.6669,
+      "step": 29411
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 4.483738899230957,
+      "learning_rate": 5.5838542669081175e-06,
+      "loss": 0.6934,
+      "step": 29412
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.314587116241455,
+      "learning_rate": 5.583592411634867e-06,
+      "loss": 0.7173,
+      "step": 29413
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.9981269836425781,
+      "learning_rate": 5.5833305547388635e-06,
+      "loss": 0.5311,
+      "step": 29414
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 3.1205668449401855,
+      "learning_rate": 5.583068696220838e-06,
+      "loss": 0.8872,
+      "step": 29415
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.9282218217849731,
+      "learning_rate": 5.582806836081516e-06,
+      "loss": 0.7176,
+      "step": 29416
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.934438705444336,
+      "learning_rate": 5.582544974321626e-06,
+      "loss": 0.735,
+      "step": 29417
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.057610511779785,
+      "learning_rate": 5.582283110941897e-06,
+      "loss": 0.8874,
+      "step": 29418
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.60221529006958,
+      "learning_rate": 5.582021245943058e-06,
+      "loss": 0.4861,
+      "step": 29419
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.689785361289978,
+      "learning_rate": 5.581759379325834e-06,
+      "loss": 0.7852,
+      "step": 29420
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 6.418756484985352,
+      "learning_rate": 5.581497511090955e-06,
+      "loss": 0.7995,
+      "step": 29421
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.7883641719818115,
+      "learning_rate": 5.58123564123915e-06,
+      "loss": 0.4685,
+      "step": 29422
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.802053689956665,
+      "learning_rate": 5.580973769771148e-06,
+      "loss": 0.5828,
+      "step": 29423
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.2386107444763184,
+      "learning_rate": 5.580711896687673e-06,
+      "loss": 0.7154,
+      "step": 29424
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.700962781906128,
+      "learning_rate": 5.580450021989457e-06,
+      "loss": 0.6931,
+      "step": 29425
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 3.0120842456817627,
+      "learning_rate": 5.580188145677227e-06,
+      "loss": 0.6137,
+      "step": 29426
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 3.2406563758850098,
+      "learning_rate": 5.579926267751711e-06,
+      "loss": 0.6018,
+      "step": 29427
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.016235113143921,
+      "learning_rate": 5.579664388213637e-06,
+      "loss": 0.5919,
+      "step": 29428
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 3.3763647079467773,
+      "learning_rate": 5.5794025070637335e-06,
+      "loss": 0.8535,
+      "step": 29429
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.4977335929870605,
+      "learning_rate": 5.579140624302729e-06,
+      "loss": 0.5781,
+      "step": 29430
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.6918089389801025,
+      "learning_rate": 5.57887873993135e-06,
+      "loss": 1.0068,
+      "step": 29431
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 3.1076810359954834,
+      "learning_rate": 5.578616853950328e-06,
+      "loss": 0.6273,
+      "step": 29432
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.7489358186721802,
+      "learning_rate": 5.578354966360389e-06,
+      "loss": 0.5742,
+      "step": 29433
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.5693258047103882,
+      "learning_rate": 5.578093077162261e-06,
+      "loss": 0.8151,
+      "step": 29434
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 3.7548060417175293,
+      "learning_rate": 5.5778311863566725e-06,
+      "loss": 0.6733,
+      "step": 29435
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 4.1486310958862305,
+      "learning_rate": 5.577569293944353e-06,
+      "loss": 0.7358,
+      "step": 29436
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.719879627227783,
+      "learning_rate": 5.5773073999260275e-06,
+      "loss": 0.4855,
+      "step": 29437
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.2367446422576904,
+      "learning_rate": 5.577045504302429e-06,
+      "loss": 0.5574,
+      "step": 29438
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.606013536453247,
+      "learning_rate": 5.576783607074281e-06,
+      "loss": 0.7059,
+      "step": 29439
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 4.4515509605407715,
+      "learning_rate": 5.576521708242315e-06,
+      "loss": 0.7812,
+      "step": 29440
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 4.628631591796875,
+      "learning_rate": 5.576259807807259e-06,
+      "loss": 0.6643,
+      "step": 29441
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 3.953051805496216,
+      "learning_rate": 5.575997905769838e-06,
+      "loss": 0.7814,
+      "step": 29442
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.973494052886963,
+      "learning_rate": 5.5757360021307825e-06,
+      "loss": 0.5984,
+      "step": 29443
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.5611772537231445,
+      "learning_rate": 5.575474096890823e-06,
+      "loss": 0.583,
+      "step": 29444
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.8809947967529297,
+      "learning_rate": 5.575212190050685e-06,
+      "loss": 0.5921,
+      "step": 29445
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 5.923041343688965,
+      "learning_rate": 5.5749502816110965e-06,
+      "loss": 0.6998,
+      "step": 29446
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.388953447341919,
+      "learning_rate": 5.574688371572787e-06,
+      "loss": 0.585,
+      "step": 29447
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 3.9291512966156006,
+      "learning_rate": 5.574426459936485e-06,
+      "loss": 1.082,
+      "step": 29448
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.7817360162734985,
+      "learning_rate": 5.574164546702918e-06,
+      "loss": 0.5605,
+      "step": 29449
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.9383583068847656,
+      "learning_rate": 5.573902631872815e-06,
+      "loss": 0.6367,
+      "step": 29450
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.554954767227173,
+      "learning_rate": 5.573640715446902e-06,
+      "loss": 0.6626,
+      "step": 29451
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 3.9785561561584473,
+      "learning_rate": 5.57337879742591e-06,
+      "loss": 0.3569,
+      "step": 29452
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 3.449172258377075,
+      "learning_rate": 5.5731168778105675e-06,
+      "loss": 0.6134,
+      "step": 29453
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 3.516265630722046,
+      "learning_rate": 5.572854956601601e-06,
+      "loss": 0.6118,
+      "step": 29454
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.20685076713562,
+      "learning_rate": 5.572593033799739e-06,
+      "loss": 0.8404,
+      "step": 29455
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.01465106010437,
+      "learning_rate": 5.572331109405711e-06,
+      "loss": 0.5584,
+      "step": 29456
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.427907705307007,
+      "learning_rate": 5.572069183420244e-06,
+      "loss": 0.3778,
+      "step": 29457
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.517629623413086,
+      "learning_rate": 5.571807255844066e-06,
+      "loss": 0.577,
+      "step": 29458
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.5825002193450928,
+      "learning_rate": 5.571545326677907e-06,
+      "loss": 0.5138,
+      "step": 29459
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.9102399349212646,
+      "learning_rate": 5.5712833959224955e-06,
+      "loss": 0.3905,
+      "step": 29460
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.319746255874634,
+      "learning_rate": 5.5710214635785575e-06,
+      "loss": 0.4096,
+      "step": 29461
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.4559569358825684,
+      "learning_rate": 5.570759529646823e-06,
+      "loss": 0.8048,
+      "step": 29462
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.0839998722076416,
+      "learning_rate": 5.57049759412802e-06,
+      "loss": 0.659,
+      "step": 29463
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.4264549016952515,
+      "learning_rate": 5.570235657022878e-06,
+      "loss": 0.6068,
+      "step": 29464
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 6.468077182769775,
+      "learning_rate": 5.569973718332123e-06,
+      "loss": 0.5861,
+      "step": 29465
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 3.787545919418335,
+      "learning_rate": 5.569711778056486e-06,
+      "loss": 0.706,
+      "step": 29466
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.087604522705078,
+      "learning_rate": 5.569449836196693e-06,
+      "loss": 0.7152,
+      "step": 29467
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 7.170905113220215,
+      "learning_rate": 5.569187892753474e-06,
+      "loss": 0.8033,
+      "step": 29468
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 4.027805805206299,
+      "learning_rate": 5.568925947727556e-06,
+      "loss": 0.5892,
+      "step": 29469
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.4405906200408936,
+      "learning_rate": 5.568664001119668e-06,
+      "loss": 0.9164,
+      "step": 29470
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.166799783706665,
+      "learning_rate": 5.5684020529305395e-06,
+      "loss": 0.5752,
+      "step": 29471
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.923225998878479,
+      "learning_rate": 5.568140103160895e-06,
+      "loss": 0.5954,
+      "step": 29472
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 3.4010722637176514,
+      "learning_rate": 5.567878151811469e-06,
+      "loss": 0.3267,
+      "step": 29473
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 3.116222620010376,
+      "learning_rate": 5.567616198882985e-06,
+      "loss": 0.5904,
+      "step": 29474
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.2959665060043335,
+      "learning_rate": 5.567354244376173e-06,
+      "loss": 0.3272,
+      "step": 29475
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 4.183252811431885,
+      "learning_rate": 5.567092288291762e-06,
+      "loss": 0.7168,
+      "step": 29476
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 3.474699020385742,
+      "learning_rate": 5.566830330630479e-06,
+      "loss": 0.8698,
+      "step": 29477
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.4348411560058594,
+      "learning_rate": 5.566568371393054e-06,
+      "loss": 0.7855,
+      "step": 29478
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.0001015663146973,
+      "learning_rate": 5.566306410580215e-06,
+      "loss": 0.5638,
+      "step": 29479
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.8852952718734741,
+      "learning_rate": 5.566044448192689e-06,
+      "loss": 0.724,
+      "step": 29480
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.9411879777908325,
+      "learning_rate": 5.565782484231206e-06,
+      "loss": 0.506,
+      "step": 29481
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.301888942718506,
+      "learning_rate": 5.565520518696492e-06,
+      "loss": 0.4352,
+      "step": 29482
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 3.508776903152466,
+      "learning_rate": 5.565258551589279e-06,
+      "loss": 0.6151,
+      "step": 29483
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.770521402359009,
+      "learning_rate": 5.564996582910293e-06,
+      "loss": 0.3539,
+      "step": 29484
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.305786371231079,
+      "learning_rate": 5.564734612660264e-06,
+      "loss": 0.5573,
+      "step": 29485
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.01530122756958,
+      "learning_rate": 5.564472640839919e-06,
+      "loss": 0.7417,
+      "step": 29486
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 3.091329574584961,
+      "learning_rate": 5.564210667449989e-06,
+      "loss": 0.5175,
+      "step": 29487
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.6448330879211426,
+      "learning_rate": 5.563948692491198e-06,
+      "loss": 0.5596,
+      "step": 29488
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.48285174369812,
+      "learning_rate": 5.563686715964278e-06,
+      "loss": 0.4718,
+      "step": 29489
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 4.060225963592529,
+      "learning_rate": 5.563424737869956e-06,
+      "loss": 0.7206,
+      "step": 29490
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.9040167331695557,
+      "learning_rate": 5.563162758208961e-06,
+      "loss": 0.6596,
+      "step": 29491
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.939127802848816,
+      "learning_rate": 5.562900776982021e-06,
+      "loss": 0.6878,
+      "step": 29492
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.1478941440582275,
+      "learning_rate": 5.562638794189865e-06,
+      "loss": 0.6201,
+      "step": 29493
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.192824363708496,
+      "learning_rate": 5.562376809833221e-06,
+      "loss": 0.4888,
+      "step": 29494
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.519083023071289,
+      "learning_rate": 5.562114823912818e-06,
+      "loss": 0.621,
+      "step": 29495
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.136030912399292,
+      "learning_rate": 5.561852836429385e-06,
+      "loss": 0.4244,
+      "step": 29496
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.1280574798583984,
+      "learning_rate": 5.56159084738365e-06,
+      "loss": 0.6021,
+      "step": 29497
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.1476807594299316,
+      "learning_rate": 5.5613288567763405e-06,
+      "loss": 0.4744,
+      "step": 29498
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.219780921936035,
+      "learning_rate": 5.561066864608184e-06,
+      "loss": 0.8783,
+      "step": 29499
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.5684683322906494,
+      "learning_rate": 5.5608048708799125e-06,
+      "loss": 0.4297,
+      "step": 29500
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.0482630729675293,
+      "learning_rate": 5.560542875592253e-06,
+      "loss": 0.4745,
+      "step": 29501
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.2763428688049316,
+      "learning_rate": 5.560280878745933e-06,
+      "loss": 0.4422,
+      "step": 29502
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 3.2409064769744873,
+      "learning_rate": 5.560018880341682e-06,
+      "loss": 0.6585,
+      "step": 29503
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.282834768295288,
+      "learning_rate": 5.55975688038023e-06,
+      "loss": 0.6368,
+      "step": 29504
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.6538116931915283,
+      "learning_rate": 5.5594948788623e-06,
+      "loss": 0.6028,
+      "step": 29505
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 4.651047229766846,
+      "learning_rate": 5.5592328757886275e-06,
+      "loss": 0.7049,
+      "step": 29506
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.775700569152832,
+      "learning_rate": 5.558970871159936e-06,
+      "loss": 0.6991,
+      "step": 29507
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.5642545223236084,
+      "learning_rate": 5.5587088649769574e-06,
+      "loss": 0.7158,
+      "step": 29508
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.9712681770324707,
+      "learning_rate": 5.5584468572404185e-06,
+      "loss": 0.9173,
+      "step": 29509
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 3.223595380783081,
+      "learning_rate": 5.558184847951047e-06,
+      "loss": 0.7265,
+      "step": 29510
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 3.7759313583374023,
+      "learning_rate": 5.557922837109574e-06,
+      "loss": 0.808,
+      "step": 29511
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.8556946516036987,
+      "learning_rate": 5.557660824716725e-06,
+      "loss": 0.5087,
+      "step": 29512
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.8647289276123047,
+      "learning_rate": 5.557398810773231e-06,
+      "loss": 0.6443,
+      "step": 29513
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.1156578063964844,
+      "learning_rate": 5.557136795279819e-06,
+      "loss": 0.6873,
+      "step": 29514
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.097888946533203,
+      "learning_rate": 5.55687477823722e-06,
+      "loss": 0.6802,
+      "step": 29515
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.4757566452026367,
+      "learning_rate": 5.55661275964616e-06,
+      "loss": 0.3589,
+      "step": 29516
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.9370989799499512,
+      "learning_rate": 5.556350739507368e-06,
+      "loss": 0.6305,
+      "step": 29517
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.9950218200683594,
+      "learning_rate": 5.556088717821574e-06,
+      "loss": 0.6447,
+      "step": 29518
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.5305991172790527,
+      "learning_rate": 5.555826694589504e-06,
+      "loss": 0.525,
+      "step": 29519
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.7132236957550049,
+      "learning_rate": 5.555564669811891e-06,
+      "loss": 0.5394,
+      "step": 29520
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.1099228858947754,
+      "learning_rate": 5.5553026434894575e-06,
+      "loss": 0.5146,
+      "step": 29521
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 3.3164591789245605,
+      "learning_rate": 5.555040615622937e-06,
+      "loss": 0.6332,
+      "step": 29522
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.915034532546997,
+      "learning_rate": 5.554778586213056e-06,
+      "loss": 0.6913,
+      "step": 29523
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 3.217904806137085,
+      "learning_rate": 5.554516555260543e-06,
+      "loss": 0.7442,
+      "step": 29524
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.4019415378570557,
+      "learning_rate": 5.554254522766128e-06,
+      "loss": 0.6043,
+      "step": 29525
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.4482781887054443,
+      "learning_rate": 5.55399248873054e-06,
+      "loss": 0.69,
+      "step": 29526
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.380589723587036,
+      "learning_rate": 5.553730453154504e-06,
+      "loss": 0.7726,
+      "step": 29527
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 3.0202300548553467,
+      "learning_rate": 5.5534684160387535e-06,
+      "loss": 0.6917,
+      "step": 29528
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.0354862213134766,
+      "learning_rate": 5.553206377384013e-06,
+      "loss": 0.6066,
+      "step": 29529
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.179694414138794,
+      "learning_rate": 5.552944337191013e-06,
+      "loss": 0.7791,
+      "step": 29530
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.3965578079223633,
+      "learning_rate": 5.552682295460482e-06,
+      "loss": 0.5342,
+      "step": 29531
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 3.786816358566284,
+      "learning_rate": 5.5524202521931495e-06,
+      "loss": 0.4485,
+      "step": 29532
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.9541970491409302,
+      "learning_rate": 5.552158207389741e-06,
+      "loss": 0.5406,
+      "step": 29533
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.8283730745315552,
+      "learning_rate": 5.551896161050989e-06,
+      "loss": 0.5777,
+      "step": 29534
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.8956193923950195,
+      "learning_rate": 5.551634113177619e-06,
+      "loss": 0.6303,
+      "step": 29535
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 8.950512886047363,
+      "learning_rate": 5.551372063770364e-06,
+      "loss": 0.821,
+      "step": 29536
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.910428762435913,
+      "learning_rate": 5.551110012829948e-06,
+      "loss": 0.702,
+      "step": 29537
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 4.2162675857543945,
+      "learning_rate": 5.550847960357102e-06,
+      "loss": 0.5324,
+      "step": 29538
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.769059181213379,
+      "learning_rate": 5.550585906352555e-06,
+      "loss": 0.7656,
+      "step": 29539
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.1553001403808594,
+      "learning_rate": 5.5503238508170335e-06,
+      "loss": 0.5224,
+      "step": 29540
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.5900315046310425,
+      "learning_rate": 5.550061793751268e-06,
+      "loss": 0.6237,
+      "step": 29541
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.4481306076049805,
+      "learning_rate": 5.549799735155986e-06,
+      "loss": 0.5559,
+      "step": 29542
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.6525089740753174,
+      "learning_rate": 5.5495376750319174e-06,
+      "loss": 0.7545,
+      "step": 29543
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.4238760471343994,
+      "learning_rate": 5.54927561337979e-06,
+      "loss": 0.7186,
+      "step": 29544
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 3.920955181121826,
+      "learning_rate": 5.549013550200334e-06,
+      "loss": 0.6222,
+      "step": 29545
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.1600399017333984,
+      "learning_rate": 5.548751485494276e-06,
+      "loss": 0.4627,
+      "step": 29546
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.914318561553955,
+      "learning_rate": 5.548489419262347e-06,
+      "loss": 0.6889,
+      "step": 29547
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.0615453720092773,
+      "learning_rate": 5.548227351505273e-06,
+      "loss": 0.5812,
+      "step": 29548
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.665785789489746,
+      "learning_rate": 5.547965282223784e-06,
+      "loss": 0.7368,
+      "step": 29549
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.8034253120422363,
+      "learning_rate": 5.547703211418611e-06,
+      "loss": 0.947,
+      "step": 29550
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.178814172744751,
+      "learning_rate": 5.547441139090478e-06,
+      "loss": 0.5871,
+      "step": 29551
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.7363827228546143,
+      "learning_rate": 5.547179065240117e-06,
+      "loss": 0.647,
+      "step": 29552
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 3.158149242401123,
+      "learning_rate": 5.546916989868255e-06,
+      "loss": 0.6594,
+      "step": 29553
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.3784894943237305,
+      "learning_rate": 5.546654912975623e-06,
+      "loss": 0.8921,
+      "step": 29554
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.1087632179260254,
+      "learning_rate": 5.546392834562948e-06,
+      "loss": 0.2833,
+      "step": 29555
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 3.176879405975342,
+      "learning_rate": 5.5461307546309605e-06,
+      "loss": 0.6377,
+      "step": 29556
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.8951668739318848,
+      "learning_rate": 5.5458686731803866e-06,
+      "loss": 0.5814,
+      "step": 29557
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.9588007926940918,
+      "learning_rate": 5.545606590211957e-06,
+      "loss": 0.2734,
+      "step": 29558
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.7469509840011597,
+      "learning_rate": 5.5453445057264e-06,
+      "loss": 0.5846,
+      "step": 29559
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.837429165840149,
+      "learning_rate": 5.545082419724444e-06,
+      "loss": 0.63,
+      "step": 29560
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 3.2026751041412354,
+      "learning_rate": 5.544820332206817e-06,
+      "loss": 0.8552,
+      "step": 29561
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.5734095573425293,
+      "learning_rate": 5.5445582431742496e-06,
+      "loss": 0.374,
+      "step": 29562
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 3.833413600921631,
+      "learning_rate": 5.544296152627469e-06,
+      "loss": 0.7472,
+      "step": 29563
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 3.301398515701294,
+      "learning_rate": 5.544034060567206e-06,
+      "loss": 0.7421,
+      "step": 29564
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.943551778793335,
+      "learning_rate": 5.543771966994188e-06,
+      "loss": 0.6882,
+      "step": 29565
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 4.536456108093262,
+      "learning_rate": 5.543509871909144e-06,
+      "loss": 0.6604,
+      "step": 29566
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 4.076638698577881,
+      "learning_rate": 5.543247775312801e-06,
+      "loss": 0.3767,
+      "step": 29567
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.954529047012329,
+      "learning_rate": 5.54298567720589e-06,
+      "loss": 0.4994,
+      "step": 29568
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 4.181187629699707,
+      "learning_rate": 5.54272357758914e-06,
+      "loss": 0.6873,
+      "step": 29569
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.85792875289917,
+      "learning_rate": 5.542461476463279e-06,
+      "loss": 0.6072,
+      "step": 29570
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.0271310806274414,
+      "learning_rate": 5.542199373829035e-06,
+      "loss": 0.8399,
+      "step": 29571
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.492330312728882,
+      "learning_rate": 5.541937269687138e-06,
+      "loss": 0.6512,
+      "step": 29572
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 3.5209405422210693,
+      "learning_rate": 5.541675164038317e-06,
+      "loss": 0.6425,
+      "step": 29573
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 3.3003692626953125,
+      "learning_rate": 5.5414130568833e-06,
+      "loss": 0.7704,
+      "step": 29574
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 3.4141464233398438,
+      "learning_rate": 5.541150948222816e-06,
+      "loss": 0.722,
+      "step": 29575
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 3.218140125274658,
+      "learning_rate": 5.540888838057594e-06,
+      "loss": 0.6907,
+      "step": 29576
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.0634543895721436,
+      "learning_rate": 5.540626726388363e-06,
+      "loss": 0.6117,
+      "step": 29577
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.783017635345459,
+      "learning_rate": 5.540364613215852e-06,
+      "loss": 0.637,
+      "step": 29578
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.4364137649536133,
+      "learning_rate": 5.540102498540789e-06,
+      "loss": 0.8346,
+      "step": 29579
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 5.517111301422119,
+      "learning_rate": 5.539840382363903e-06,
+      "loss": 0.6441,
+      "step": 29580
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.0655157566070557,
+      "learning_rate": 5.539578264685923e-06,
+      "loss": 0.6022,
+      "step": 29581
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 3.2101876735687256,
+      "learning_rate": 5.539316145507579e-06,
+      "loss": 0.8238,
+      "step": 29582
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 3.6017472743988037,
+      "learning_rate": 5.539054024829597e-06,
+      "loss": 0.7129,
+      "step": 29583
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.233150005340576,
+      "learning_rate": 5.538791902652709e-06,
+      "loss": 0.5448,
+      "step": 29584
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.810765027999878,
+      "learning_rate": 5.538529778977643e-06,
+      "loss": 0.6694,
+      "step": 29585
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 4.239930152893066,
+      "learning_rate": 5.538267653805126e-06,
+      "loss": 0.8588,
+      "step": 29586
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 3.03609299659729,
+      "learning_rate": 5.538005527135889e-06,
+      "loss": 0.7162,
+      "step": 29587
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.6662373542785645,
+      "learning_rate": 5.537743398970661e-06,
+      "loss": 0.525,
+      "step": 29588
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.6501004695892334,
+      "learning_rate": 5.537481269310169e-06,
+      "loss": 0.4662,
+      "step": 29589
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.8645904064178467,
+      "learning_rate": 5.537219138155143e-06,
+      "loss": 0.6876,
+      "step": 29590
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.3587236404418945,
+      "learning_rate": 5.5369570055063126e-06,
+      "loss": 0.4333,
+      "step": 29591
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.3887014389038086,
+      "learning_rate": 5.536694871364405e-06,
+      "loss": 0.7932,
+      "step": 29592
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 3.3202803134918213,
+      "learning_rate": 5.536432735730151e-06,
+      "loss": 0.7093,
+      "step": 29593
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.2129011154174805,
+      "learning_rate": 5.536170598604277e-06,
+      "loss": 0.728,
+      "step": 29594
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.3434319496154785,
+      "learning_rate": 5.5359084599875135e-06,
+      "loss": 0.5815,
+      "step": 29595
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.7179254293441772,
+      "learning_rate": 5.535646319880591e-06,
+      "loss": 0.7194,
+      "step": 29596
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.460841655731201,
+      "learning_rate": 5.535384178284236e-06,
+      "loss": 0.6063,
+      "step": 29597
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.0121352672576904,
+      "learning_rate": 5.535122035199177e-06,
+      "loss": 0.8024,
+      "step": 29598
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 6.059523105621338,
+      "learning_rate": 5.5348598906261455e-06,
+      "loss": 0.573,
+      "step": 29599
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.9166135787963867,
+      "learning_rate": 5.5345977445658695e-06,
+      "loss": 0.4519,
+      "step": 29600
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 5.204462051391602,
+      "learning_rate": 5.534335597019076e-06,
+      "loss": 0.588,
+      "step": 29601
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.374991774559021,
+      "learning_rate": 5.534073447986497e-06,
+      "loss": 0.54,
+      "step": 29602
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.3359551429748535,
+      "learning_rate": 5.533811297468858e-06,
+      "loss": 0.586,
+      "step": 29603
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 3.017307996749878,
+      "learning_rate": 5.53354914546689e-06,
+      "loss": 0.7118,
+      "step": 29604
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 3.153848171234131,
+      "learning_rate": 5.5332869919813215e-06,
+      "loss": 0.8685,
+      "step": 29605
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.871748685836792,
+      "learning_rate": 5.533024837012882e-06,
+      "loss": 0.8519,
+      "step": 29606
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.2487237453460693,
+      "learning_rate": 5.532762680562301e-06,
+      "loss": 0.5734,
+      "step": 29607
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 4.036352157592773,
+      "learning_rate": 5.532500522630306e-06,
+      "loss": 0.9119,
+      "step": 29608
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.975044846534729,
+      "learning_rate": 5.532238363217627e-06,
+      "loss": 0.6176,
+      "step": 29609
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.4703845977783203,
+      "learning_rate": 5.5319762023249926e-06,
+      "loss": 0.651,
+      "step": 29610
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.464104652404785,
+      "learning_rate": 5.53171403995313e-06,
+      "loss": 0.5697,
+      "step": 29611
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 4.313408374786377,
+      "learning_rate": 5.531451876102771e-06,
+      "loss": 0.6854,
+      "step": 29612
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.686063289642334,
+      "learning_rate": 5.531189710774644e-06,
+      "loss": 0.7775,
+      "step": 29613
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 3.5743863582611084,
+      "learning_rate": 5.530927543969475e-06,
+      "loss": 0.5083,
+      "step": 29614
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.9458566904067993,
+      "learning_rate": 5.530665375687998e-06,
+      "loss": 0.632,
+      "step": 29615
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.1044936180114746,
+      "learning_rate": 5.5304032059309384e-06,
+      "loss": 0.635,
+      "step": 29616
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.724144220352173,
+      "learning_rate": 5.530141034699026e-06,
+      "loss": 0.677,
+      "step": 29617
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.9628660678863525,
+      "learning_rate": 5.529878861992991e-06,
+      "loss": 0.7639,
+      "step": 29618
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.7139358520507812,
+      "learning_rate": 5.52961668781356e-06,
+      "loss": 0.5584,
+      "step": 29619
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.913991928100586,
+      "learning_rate": 5.529354512161464e-06,
+      "loss": 0.6371,
+      "step": 29620
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.122562885284424,
+      "learning_rate": 5.529092335037433e-06,
+      "loss": 0.7421,
+      "step": 29621
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 3.5728611946105957,
+      "learning_rate": 5.528830156442192e-06,
+      "loss": 0.7049,
+      "step": 29622
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.683844804763794,
+      "learning_rate": 5.528567976376472e-06,
+      "loss": 0.7019,
+      "step": 29623
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.329481601715088,
+      "learning_rate": 5.528305794841004e-06,
+      "loss": 0.7208,
+      "step": 29624
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.520761728286743,
+      "learning_rate": 5.5280436118365155e-06,
+      "loss": 0.8504,
+      "step": 29625
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.336822748184204,
+      "learning_rate": 5.527781427363736e-06,
+      "loss": 0.7043,
+      "step": 29626
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.2189252376556396,
+      "learning_rate": 5.527519241423393e-06,
+      "loss": 0.5814,
+      "step": 29627
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.0272738933563232,
+      "learning_rate": 5.527257054016216e-06,
+      "loss": 0.8393,
+      "step": 29628
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.7416906356811523,
+      "learning_rate": 5.526994865142935e-06,
+      "loss": 0.8182,
+      "step": 29629
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 5.161158084869385,
+      "learning_rate": 5.52673267480428e-06,
+      "loss": 0.7724,
+      "step": 29630
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.9731858968734741,
+      "learning_rate": 5.5264704830009784e-06,
+      "loss": 0.7484,
+      "step": 29631
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 3.1291604042053223,
+      "learning_rate": 5.526208289733759e-06,
+      "loss": 0.8042,
+      "step": 29632
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.592536449432373,
+      "learning_rate": 5.525946095003352e-06,
+      "loss": 0.6993,
+      "step": 29633
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.308035373687744,
+      "learning_rate": 5.5256838988104845e-06,
+      "loss": 0.7883,
+      "step": 29634
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 7.864814758300781,
+      "learning_rate": 5.525421701155888e-06,
+      "loss": 0.7187,
+      "step": 29635
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.781591534614563,
+      "learning_rate": 5.5251595020402916e-06,
+      "loss": 0.5383,
+      "step": 29636
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.7692863941192627,
+      "learning_rate": 5.524897301464423e-06,
+      "loss": 0.6572,
+      "step": 29637
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 3.232795476913452,
+      "learning_rate": 5.52463509942901e-06,
+      "loss": 0.5394,
+      "step": 29638
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.286010503768921,
+      "learning_rate": 5.524372895934784e-06,
+      "loss": 0.3611,
+      "step": 29639
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.8056631088256836,
+      "learning_rate": 5.524110690982475e-06,
+      "loss": 0.5689,
+      "step": 29640
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 3.2551491260528564,
+      "learning_rate": 5.523848484572809e-06,
+      "loss": 0.5366,
+      "step": 29641
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.8119430541992188,
+      "learning_rate": 5.5235862767065175e-06,
+      "loss": 0.6242,
+      "step": 29642
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 3.2675726413726807,
+      "learning_rate": 5.523324067384328e-06,
+      "loss": 0.4632,
+      "step": 29643
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.831291675567627,
+      "learning_rate": 5.52306185660697e-06,
+      "loss": 0.5916,
+      "step": 29644
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.969592332839966,
+      "learning_rate": 5.522799644375173e-06,
+      "loss": 0.7808,
+      "step": 29645
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 3.6623477935791016,
+      "learning_rate": 5.522537430689667e-06,
+      "loss": 0.6088,
+      "step": 29646
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.65297794342041,
+      "learning_rate": 5.52227521555118e-06,
+      "loss": 0.4637,
+      "step": 29647
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 3.410217761993408,
+      "learning_rate": 5.52201299896044e-06,
+      "loss": 0.4727,
+      "step": 29648
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.8003660440444946,
+      "learning_rate": 5.521750780918178e-06,
+      "loss": 0.4034,
+      "step": 29649
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 4.664442539215088,
+      "learning_rate": 5.521488561425123e-06,
+      "loss": 0.6373,
+      "step": 29650
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.618088960647583,
+      "learning_rate": 5.521226340482004e-06,
+      "loss": 0.7514,
+      "step": 29651
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 3.210435628890991,
+      "learning_rate": 5.520964118089549e-06,
+      "loss": 0.8889,
+      "step": 29652
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 6.119719505310059,
+      "learning_rate": 5.520701894248488e-06,
+      "loss": 0.8779,
+      "step": 29653
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.7157132625579834,
+      "learning_rate": 5.520439668959551e-06,
+      "loss": 0.5428,
+      "step": 29654
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 3.8511571884155273,
+      "learning_rate": 5.5201774422234645e-06,
+      "loss": 0.5119,
+      "step": 29655
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.7301737070083618,
+      "learning_rate": 5.519915214040961e-06,
+      "loss": 0.7921,
+      "step": 29656
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 6.427343845367432,
+      "learning_rate": 5.519652984412766e-06,
+      "loss": 0.7124,
+      "step": 29657
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.5294444561004639,
+      "learning_rate": 5.519390753339613e-06,
+      "loss": 0.5009,
+      "step": 29658
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.291804075241089,
+      "learning_rate": 5.519128520822228e-06,
+      "loss": 0.7181,
+      "step": 29659
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.8724430799484253,
+      "learning_rate": 5.5188662868613416e-06,
+      "loss": 0.638,
+      "step": 29660
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 5.407505989074707,
+      "learning_rate": 5.518604051457682e-06,
+      "loss": 0.7223,
+      "step": 29661
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 3.8105645179748535,
+      "learning_rate": 5.518341814611979e-06,
+      "loss": 0.8534,
+      "step": 29662
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.9848036766052246,
+      "learning_rate": 5.518079576324962e-06,
+      "loss": 0.5283,
+      "step": 29663
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.260842800140381,
+      "learning_rate": 5.517817336597358e-06,
+      "loss": 0.7965,
+      "step": 29664
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.5928303003311157,
+      "learning_rate": 5.517555095429899e-06,
+      "loss": 0.5558,
+      "step": 29665
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.6003942489624023,
+      "learning_rate": 5.517292852823313e-06,
+      "loss": 0.5634,
+      "step": 29666
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.5474226474761963,
+      "learning_rate": 5.51703060877833e-06,
+      "loss": 0.6964,
+      "step": 29667
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 3.957019805908203,
+      "learning_rate": 5.516768363295678e-06,
+      "loss": 0.5337,
+      "step": 29668
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.9092373847961426,
+      "learning_rate": 5.516506116376087e-06,
+      "loss": 0.8123,
+      "step": 29669
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 4.426114559173584,
+      "learning_rate": 5.516243868020286e-06,
+      "loss": 0.5853,
+      "step": 29670
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.341683864593506,
+      "learning_rate": 5.515981618229004e-06,
+      "loss": 0.6175,
+      "step": 29671
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 5.970010280609131,
+      "learning_rate": 5.515719367002971e-06,
+      "loss": 0.6597,
+      "step": 29672
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.483804941177368,
+      "learning_rate": 5.5154571143429145e-06,
+      "loss": 0.8206,
+      "step": 29673
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.536225438117981,
+      "learning_rate": 5.515194860249566e-06,
+      "loss": 0.8038,
+      "step": 29674
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.7193883657455444,
+      "learning_rate": 5.514932604723653e-06,
+      "loss": 0.6853,
+      "step": 29675
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.3233461380004883,
+      "learning_rate": 5.514670347765905e-06,
+      "loss": 0.6416,
+      "step": 29676
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 4.475750923156738,
+      "learning_rate": 5.514408089377053e-06,
+      "loss": 0.5686,
+      "step": 29677
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.6625685691833496,
+      "learning_rate": 5.514145829557823e-06,
+      "loss": 0.6409,
+      "step": 29678
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 3.5336339473724365,
+      "learning_rate": 5.513883568308946e-06,
+      "loss": 0.7521,
+      "step": 29679
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 61.76371765136719,
+      "learning_rate": 5.513621305631153e-06,
+      "loss": 0.5156,
+      "step": 29680
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.3856773376464844,
+      "learning_rate": 5.51335904152517e-06,
+      "loss": 0.5739,
+      "step": 29681
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 3.684731960296631,
+      "learning_rate": 5.5130967759917295e-06,
+      "loss": 0.7117,
+      "step": 29682
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 4.192325592041016,
+      "learning_rate": 5.5128345090315584e-06,
+      "loss": 0.7623,
+      "step": 29683
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.17427921295166,
+      "learning_rate": 5.512572240645385e-06,
+      "loss": 0.5892,
+      "step": 29684
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.1575803756713867,
+      "learning_rate": 5.512309970833942e-06,
+      "loss": 0.4383,
+      "step": 29685
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.9310705661773682,
+      "learning_rate": 5.512047699597956e-06,
+      "loss": 0.4776,
+      "step": 29686
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.9251363277435303,
+      "learning_rate": 5.511785426938157e-06,
+      "loss": 0.5858,
+      "step": 29687
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.235375165939331,
+      "learning_rate": 5.511523152855276e-06,
+      "loss": 0.7446,
+      "step": 29688
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 3.963611125946045,
+      "learning_rate": 5.51126087735004e-06,
+      "loss": 0.712,
+      "step": 29689
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 3.3073670864105225,
+      "learning_rate": 5.510998600423177e-06,
+      "loss": 0.5578,
+      "step": 29690
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.6516306400299072,
+      "learning_rate": 5.5107363220754205e-06,
+      "loss": 0.4413,
+      "step": 29691
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 4.734988689422607,
+      "learning_rate": 5.510474042307498e-06,
+      "loss": 0.6446,
+      "step": 29692
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.5140624046325684,
+      "learning_rate": 5.510211761120138e-06,
+      "loss": 0.4943,
+      "step": 29693
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.9834513664245605,
+      "learning_rate": 5.50994947851407e-06,
+      "loss": 0.5362,
+      "step": 29694
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.8741737604141235,
+      "learning_rate": 5.509687194490023e-06,
+      "loss": 0.7184,
+      "step": 29695
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.2271203994750977,
+      "learning_rate": 5.509424909048726e-06,
+      "loss": 0.6209,
+      "step": 29696
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.6850242614746094,
+      "learning_rate": 5.509162622190911e-06,
+      "loss": 0.5821,
+      "step": 29697
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 3.9366955757141113,
+      "learning_rate": 5.5089003339173055e-06,
+      "loss": 0.6506,
+      "step": 29698
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.747056484222412,
+      "learning_rate": 5.508638044228638e-06,
+      "loss": 0.9739,
+      "step": 29699
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.7787787914276123,
+      "learning_rate": 5.508375753125639e-06,
+      "loss": 0.6224,
+      "step": 29700
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 3.165198802947998,
+      "learning_rate": 5.508113460609037e-06,
+      "loss": 0.5252,
+      "step": 29701
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.8038464784622192,
+      "learning_rate": 5.507851166679562e-06,
+      "loss": 0.5558,
+      "step": 29702
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.1232857704162598,
+      "learning_rate": 5.507588871337944e-06,
+      "loss": 0.8433,
+      "step": 29703
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.2207202911376953,
+      "learning_rate": 5.50732657458491e-06,
+      "loss": 0.5843,
+      "step": 29704
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 4.014288425445557,
+      "learning_rate": 5.507064276421191e-06,
+      "loss": 0.6198,
+      "step": 29705
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.5606772899627686,
+      "learning_rate": 5.5068019768475165e-06,
+      "loss": 0.6261,
+      "step": 29706
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 3.0748910903930664,
+      "learning_rate": 5.506539675864616e-06,
+      "loss": 0.8459,
+      "step": 29707
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.948092222213745,
+      "learning_rate": 5.506277373473219e-06,
+      "loss": 0.5996,
+      "step": 29708
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.7308235168457031,
+      "learning_rate": 5.506015069674053e-06,
+      "loss": 0.6177,
+      "step": 29709
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.8054805994033813,
+      "learning_rate": 5.50575276446785e-06,
+      "loss": 0.7582,
+      "step": 29710
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 3.457129955291748,
+      "learning_rate": 5.505490457855337e-06,
+      "loss": 0.7448,
+      "step": 29711
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.029480457305908,
+      "learning_rate": 5.505228149837245e-06,
+      "loss": 0.5761,
+      "step": 29712
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.535189628601074,
+      "learning_rate": 5.504965840414303e-06,
+      "loss": 0.7526,
+      "step": 29713
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.955303192138672,
+      "learning_rate": 5.504703529587239e-06,
+      "loss": 0.6513,
+      "step": 29714
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.6675268411636353,
+      "learning_rate": 5.504441217356784e-06,
+      "loss": 0.6076,
+      "step": 29715
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 3.3158960342407227,
+      "learning_rate": 5.504178903723667e-06,
+      "loss": 0.6528,
+      "step": 29716
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.9259473085403442,
+      "learning_rate": 5.503916588688617e-06,
+      "loss": 0.7975,
+      "step": 29717
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.7790533304214478,
+      "learning_rate": 5.503654272252365e-06,
+      "loss": 0.5497,
+      "step": 29718
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.2105448246002197,
+      "learning_rate": 5.503391954415638e-06,
+      "loss": 0.7387,
+      "step": 29719
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.7444401979446411,
+      "learning_rate": 5.503129635179166e-06,
+      "loss": 0.7143,
+      "step": 29720
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.951764702796936,
+      "learning_rate": 5.502867314543681e-06,
+      "loss": 0.7387,
+      "step": 29721
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.1400203704833984,
+      "learning_rate": 5.502604992509909e-06,
+      "loss": 0.7895,
+      "step": 29722
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.8891422748565674,
+      "learning_rate": 5.5023426690785804e-06,
+      "loss": 0.7408,
+      "step": 29723
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.625216007232666,
+      "learning_rate": 5.502080344250427e-06,
+      "loss": 0.7924,
+      "step": 29724
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.307664632797241,
+      "learning_rate": 5.501818018026174e-06,
+      "loss": 0.7598,
+      "step": 29725
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.0539333820343018,
+      "learning_rate": 5.501555690406553e-06,
+      "loss": 0.8657,
+      "step": 29726
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.484795093536377,
+      "learning_rate": 5.501293361392295e-06,
+      "loss": 0.6702,
+      "step": 29727
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 3.4505417346954346,
+      "learning_rate": 5.501031030984127e-06,
+      "loss": 0.6215,
+      "step": 29728
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.4961538314819336,
+      "learning_rate": 5.50076869918278e-06,
+      "loss": 0.8256,
+      "step": 29729
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.2267813682556152,
+      "learning_rate": 5.500506365988982e-06,
+      "loss": 0.602,
+      "step": 29730
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.152420997619629,
+      "learning_rate": 5.500244031403464e-06,
+      "loss": 0.6918,
+      "step": 29731
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.9743678569793701,
+      "learning_rate": 5.4999816954269555e-06,
+      "loss": 0.582,
+      "step": 29732
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.0699658393859863,
+      "learning_rate": 5.4997193580601836e-06,
+      "loss": 0.8296,
+      "step": 29733
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.3263888359069824,
+      "learning_rate": 5.499457019303881e-06,
+      "loss": 0.5908,
+      "step": 29734
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.496594190597534,
+      "learning_rate": 5.499194679158774e-06,
+      "loss": 0.5977,
+      "step": 29735
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.8946340084075928,
+      "learning_rate": 5.498932337625595e-06,
+      "loss": 0.7323,
+      "step": 29736
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.3756155967712402,
+      "learning_rate": 5.49866999470507e-06,
+      "loss": 0.659,
+      "step": 29737
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.671797513961792,
+      "learning_rate": 5.498407650397932e-06,
+      "loss": 0.9213,
+      "step": 29738
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 4.25623893737793,
+      "learning_rate": 5.498145304704909e-06,
+      "loss": 0.704,
+      "step": 29739
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.212899684906006,
+      "learning_rate": 5.497882957626731e-06,
+      "loss": 0.7538,
+      "step": 29740
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 8.665586471557617,
+      "learning_rate": 5.497620609164126e-06,
+      "loss": 0.7614,
+      "step": 29741
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.0285820960998535,
+      "learning_rate": 5.497358259317826e-06,
+      "loss": 0.2738,
+      "step": 29742
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 4.706019401550293,
+      "learning_rate": 5.4970959080885575e-06,
+      "loss": 0.6114,
+      "step": 29743
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.7943053245544434,
+      "learning_rate": 5.496833555477051e-06,
+      "loss": 0.5038,
+      "step": 29744
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.2552380561828613,
+      "learning_rate": 5.496571201484038e-06,
+      "loss": 0.578,
+      "step": 29745
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.9625035524368286,
+      "learning_rate": 5.496308846110245e-06,
+      "loss": 0.6277,
+      "step": 29746
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.450009822845459,
+      "learning_rate": 5.496046489356404e-06,
+      "loss": 0.5817,
+      "step": 29747
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.3417205810546875,
+      "learning_rate": 5.495784131223244e-06,
+      "loss": 0.6503,
+      "step": 29748
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 3.282538652420044,
+      "learning_rate": 5.495521771711494e-06,
+      "loss": 0.6197,
+      "step": 29749
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 5.1872687339782715,
+      "learning_rate": 5.4952594108218825e-06,
+      "loss": 0.7618,
+      "step": 29750
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.7075763940811157,
+      "learning_rate": 5.4949970485551405e-06,
+      "loss": 0.6462,
+      "step": 29751
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.8450870513916016,
+      "learning_rate": 5.494734684911996e-06,
+      "loss": 0.805,
+      "step": 29752
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.7256101369857788,
+      "learning_rate": 5.4944723198931816e-06,
+      "loss": 0.9093,
+      "step": 29753
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.78574800491333,
+      "learning_rate": 5.494209953499426e-06,
+      "loss": 0.8428,
+      "step": 29754
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 4.272730827331543,
+      "learning_rate": 5.493947585731455e-06,
+      "loss": 0.536,
+      "step": 29755
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.238884449005127,
+      "learning_rate": 5.493685216590003e-06,
+      "loss": 0.6071,
+      "step": 29756
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 3.1517958641052246,
+      "learning_rate": 5.493422846075796e-06,
+      "loss": 0.5694,
+      "step": 29757
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.418164014816284,
+      "learning_rate": 5.493160474189563e-06,
+      "loss": 0.5615,
+      "step": 29758
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.561269998550415,
+      "learning_rate": 5.4928981009320385e-06,
+      "loss": 0.502,
+      "step": 29759
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.8079845905303955,
+      "learning_rate": 5.4926357263039475e-06,
+      "loss": 0.4124,
+      "step": 29760
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.801184892654419,
+      "learning_rate": 5.492373350306021e-06,
+      "loss": 0.4491,
+      "step": 29761
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.734907627105713,
+      "learning_rate": 5.49211097293899e-06,
+      "loss": 0.6366,
+      "step": 29762
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 3.7671470642089844,
+      "learning_rate": 5.491848594203582e-06,
+      "loss": 0.7093,
+      "step": 29763
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.642350435256958,
+      "learning_rate": 5.491586214100527e-06,
+      "loss": 0.5979,
+      "step": 29764
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.9918925762176514,
+      "learning_rate": 5.491323832630556e-06,
+      "loss": 0.6744,
+      "step": 29765
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.0797736644744873,
+      "learning_rate": 5.491061449794396e-06,
+      "loss": 0.6359,
+      "step": 29766
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 3.159069061279297,
+      "learning_rate": 5.490799065592779e-06,
+      "loss": 0.5627,
+      "step": 29767
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.366874933242798,
+      "learning_rate": 5.490536680026433e-06,
+      "loss": 0.4785,
+      "step": 29768
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.7402777671813965,
+      "learning_rate": 5.490274293096089e-06,
+      "loss": 0.4936,
+      "step": 29769
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.114863395690918,
+      "learning_rate": 5.4900119048024756e-06,
+      "loss": 0.7057,
+      "step": 29770
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.2823705673217773,
+      "learning_rate": 5.4897495151463235e-06,
+      "loss": 0.561,
+      "step": 29771
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 3.1442999839782715,
+      "learning_rate": 5.48948712412836e-06,
+      "loss": 0.6618,
+      "step": 29772
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.552342176437378,
+      "learning_rate": 5.489224731749317e-06,
+      "loss": 0.6099,
+      "step": 29773
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.638575315475464,
+      "learning_rate": 5.488962338009922e-06,
+      "loss": 0.4553,
+      "step": 29774
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 3.5711476802825928,
+      "learning_rate": 5.488699942910908e-06,
+      "loss": 0.819,
+      "step": 29775
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.6355414390563965,
+      "learning_rate": 5.488437546453001e-06,
+      "loss": 0.6259,
+      "step": 29776
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.8296208381652832,
+      "learning_rate": 5.488175148636932e-06,
+      "loss": 0.8395,
+      "step": 29777
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.954565167427063,
+      "learning_rate": 5.487912749463431e-06,
+      "loss": 0.5712,
+      "step": 29778
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.537924289703369,
+      "learning_rate": 5.487650348933228e-06,
+      "loss": 0.7407,
+      "step": 29779
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 3.4350435733795166,
+      "learning_rate": 5.4873879470470524e-06,
+      "loss": 0.5724,
+      "step": 29780
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.079810857772827,
+      "learning_rate": 5.487125543805633e-06,
+      "loss": 0.4998,
+      "step": 29781
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.450685977935791,
+      "learning_rate": 5.486863139209699e-06,
+      "loss": 0.6625,
+      "step": 29782
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.82145094871521,
+      "learning_rate": 5.4866007332599825e-06,
+      "loss": 0.6955,
+      "step": 29783
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.869905710220337,
+      "learning_rate": 5.486338325957211e-06,
+      "loss": 0.4094,
+      "step": 29784
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.2700226306915283,
+      "learning_rate": 5.486075917302115e-06,
+      "loss": 0.5254,
+      "step": 29785
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 3.231829881668091,
+      "learning_rate": 5.485813507295425e-06,
+      "loss": 0.9436,
+      "step": 29786
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.292522668838501,
+      "learning_rate": 5.485551095937867e-06,
+      "loss": 0.664,
+      "step": 29787
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.247800588607788,
+      "learning_rate": 5.485288683230175e-06,
+      "loss": 0.4732,
+      "step": 29788
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.008760929107666,
+      "learning_rate": 5.485026269173076e-06,
+      "loss": 0.6681,
+      "step": 29789
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.1110377311706543,
+      "learning_rate": 5.484763853767302e-06,
+      "loss": 0.4135,
+      "step": 29790
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.2144665718078613,
+      "learning_rate": 5.484501437013581e-06,
+      "loss": 0.6309,
+      "step": 29791
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.8701342344284058,
+      "learning_rate": 5.484239018912642e-06,
+      "loss": 0.6249,
+      "step": 29792
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.547060251235962,
+      "learning_rate": 5.4839765994652174e-06,
+      "loss": 0.5264,
+      "step": 29793
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.272094488143921,
+      "learning_rate": 5.4837141786720335e-06,
+      "loss": 0.5366,
+      "step": 29794
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.3117101192474365,
+      "learning_rate": 5.483451756533823e-06,
+      "loss": 0.6917,
+      "step": 29795
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.9413946866989136,
+      "learning_rate": 5.483189333051313e-06,
+      "loss": 0.7336,
+      "step": 29796
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.5307679176330566,
+      "learning_rate": 5.482926908225235e-06,
+      "loss": 0.6129,
+      "step": 29797
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.760812759399414,
+      "learning_rate": 5.482664482056318e-06,
+      "loss": 0.4883,
+      "step": 29798
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.189779758453369,
+      "learning_rate": 5.482402054545291e-06,
+      "loss": 0.6111,
+      "step": 29799
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.7280545234680176,
+      "learning_rate": 5.482139625692887e-06,
+      "loss": 0.7016,
+      "step": 29800
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.0785834789276123,
+      "learning_rate": 5.481877195499832e-06,
+      "loss": 0.658,
+      "step": 29801
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.770101547241211,
+      "learning_rate": 5.481614763966857e-06,
+      "loss": 0.5811,
+      "step": 29802
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.315434217453003,
+      "learning_rate": 5.481352331094691e-06,
+      "loss": 0.6306,
+      "step": 29803
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.8937783241271973,
+      "learning_rate": 5.481089896884067e-06,
+      "loss": 0.544,
+      "step": 29804
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.0886292457580566,
+      "learning_rate": 5.480827461335709e-06,
+      "loss": 0.7474,
+      "step": 29805
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 4.142237186431885,
+      "learning_rate": 5.480565024450353e-06,
+      "loss": 0.5487,
+      "step": 29806
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.930332899093628,
+      "learning_rate": 5.480302586228723e-06,
+      "loss": 0.66,
+      "step": 29807
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.054199695587158,
+      "learning_rate": 5.480040146671553e-06,
+      "loss": 0.6695,
+      "step": 29808
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.5243587493896484,
+      "learning_rate": 5.47977770577957e-06,
+      "loss": 0.5359,
+      "step": 29809
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 4.703972339630127,
+      "learning_rate": 5.479515263553505e-06,
+      "loss": 0.8467,
+      "step": 29810
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.0604403018951416,
+      "learning_rate": 5.47925281999409e-06,
+      "loss": 0.5833,
+      "step": 29811
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.6151679754257202,
+      "learning_rate": 5.47899037510205e-06,
+      "loss": 0.5009,
+      "step": 29812
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.2998926639556885,
+      "learning_rate": 5.478727928878117e-06,
+      "loss": 0.5295,
+      "step": 29813
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.7116174697875977,
+      "learning_rate": 5.478465481323022e-06,
+      "loss": 0.5793,
+      "step": 29814
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.684454917907715,
+      "learning_rate": 5.4782030324374945e-06,
+      "loss": 0.64,
+      "step": 29815
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.721306562423706,
+      "learning_rate": 5.477940582222263e-06,
+      "loss": 0.6629,
+      "step": 29816
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.1549293994903564,
+      "learning_rate": 5.477678130678057e-06,
+      "loss": 0.4055,
+      "step": 29817
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.793687343597412,
+      "learning_rate": 5.477415677805607e-06,
+      "loss": 0.4747,
+      "step": 29818
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.774498701095581,
+      "learning_rate": 5.477153223605644e-06,
+      "loss": 0.459,
+      "step": 29819
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.1205251216888428,
+      "learning_rate": 5.476890768078896e-06,
+      "loss": 0.5926,
+      "step": 29820
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.160064697265625,
+      "learning_rate": 5.4766283112260935e-06,
+      "loss": 0.6545,
+      "step": 29821
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.829524040222168,
+      "learning_rate": 5.476365853047966e-06,
+      "loss": 0.613,
+      "step": 29822
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.9047789573669434,
+      "learning_rate": 5.4761033935452435e-06,
+      "loss": 0.5491,
+      "step": 29823
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.8893897533416748,
+      "learning_rate": 5.475840932718656e-06,
+      "loss": 0.6924,
+      "step": 29824
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.6388180255889893,
+      "learning_rate": 5.475578470568933e-06,
+      "loss": 0.6147,
+      "step": 29825
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.763914108276367,
+      "learning_rate": 5.475316007096805e-06,
+      "loss": 0.5171,
+      "step": 29826
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.76825213432312,
+      "learning_rate": 5.4750535423029995e-06,
+      "loss": 0.8216,
+      "step": 29827
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.4371554851531982,
+      "learning_rate": 5.47479107618825e-06,
+      "loss": 0.5466,
+      "step": 29828
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.831719398498535,
+      "learning_rate": 5.474528608753284e-06,
+      "loss": 0.7345,
+      "step": 29829
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.490206718444824,
+      "learning_rate": 5.474266139998832e-06,
+      "loss": 0.6752,
+      "step": 29830
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.652705669403076,
+      "learning_rate": 5.4740036699256225e-06,
+      "loss": 0.7051,
+      "step": 29831
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.990869402885437,
+      "learning_rate": 5.473741198534387e-06,
+      "loss": 0.4849,
+      "step": 29832
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.47894549369812,
+      "learning_rate": 5.473478725825855e-06,
+      "loss": 0.6548,
+      "step": 29833
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.200145721435547,
+      "learning_rate": 5.473216251800755e-06,
+      "loss": 0.6829,
+      "step": 29834
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.3399159908294678,
+      "learning_rate": 5.472953776459819e-06,
+      "loss": 0.4461,
+      "step": 29835
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 16.91690444946289,
+      "learning_rate": 5.472691299803775e-06,
+      "loss": 0.643,
+      "step": 29836
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.6634180545806885,
+      "learning_rate": 5.472428821833353e-06,
+      "loss": 0.8068,
+      "step": 29837
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.9797956943511963,
+      "learning_rate": 5.472166342549283e-06,
+      "loss": 0.6062,
+      "step": 29838
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.4329428672790527,
+      "learning_rate": 5.471903861952297e-06,
+      "loss": 0.6911,
+      "step": 29839
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 4.264002323150635,
+      "learning_rate": 5.4716413800431225e-06,
+      "loss": 0.6082,
+      "step": 29840
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.149958372116089,
+      "learning_rate": 5.4713788968224905e-06,
+      "loss": 0.6361,
+      "step": 29841
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.7374093532562256,
+      "learning_rate": 5.47111641229113e-06,
+      "loss": 0.6942,
+      "step": 29842
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.5027029514312744,
+      "learning_rate": 5.4708539264497715e-06,
+      "loss": 0.9554,
+      "step": 29843
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 4.22763729095459,
+      "learning_rate": 5.470591439299145e-06,
+      "loss": 0.7227,
+      "step": 29844
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.1665682792663574,
+      "learning_rate": 5.470328950839978e-06,
+      "loss": 0.5748,
+      "step": 29845
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.0555002689361572,
+      "learning_rate": 5.470066461073005e-06,
+      "loss": 0.516,
+      "step": 29846
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.958876609802246,
+      "learning_rate": 5.469803969998952e-06,
+      "loss": 0.8376,
+      "step": 29847
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.4335038661956787,
+      "learning_rate": 5.469541477618549e-06,
+      "loss": 0.584,
+      "step": 29848
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.351722478866577,
+      "learning_rate": 5.469278983932527e-06,
+      "loss": 0.6254,
+      "step": 29849
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.9763007164001465,
+      "learning_rate": 5.469016488941618e-06,
+      "loss": 0.4358,
+      "step": 29850
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.88510000705719,
+      "learning_rate": 5.46875399264655e-06,
+      "loss": 0.5003,
+      "step": 29851
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.1659886837005615,
+      "learning_rate": 5.468491495048051e-06,
+      "loss": 0.653,
+      "step": 29852
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.0325214862823486,
+      "learning_rate": 5.468228996146853e-06,
+      "loss": 0.6809,
+      "step": 29853
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.913983106613159,
+      "learning_rate": 5.467966495943686e-06,
+      "loss": 0.729,
+      "step": 29854
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.5961427688598633,
+      "learning_rate": 5.46770399443928e-06,
+      "loss": 0.6413,
+      "step": 29855
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.1748385429382324,
+      "learning_rate": 5.4674414916343635e-06,
+      "loss": 0.4642,
+      "step": 29856
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.538774013519287,
+      "learning_rate": 5.4671789875296675e-06,
+      "loss": 1.0209,
+      "step": 29857
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.6913706064224243,
+      "learning_rate": 5.466916482125921e-06,
+      "loss": 0.5116,
+      "step": 29858
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.017993450164795,
+      "learning_rate": 5.466653975423855e-06,
+      "loss": 0.4413,
+      "step": 29859
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.1228315830230713,
+      "learning_rate": 5.4663914674242e-06,
+      "loss": 0.5246,
+      "step": 29860
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.7881278991699219,
+      "learning_rate": 5.4661289581276846e-06,
+      "loss": 0.6771,
+      "step": 29861
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.6840691566467285,
+      "learning_rate": 5.46586644753504e-06,
+      "loss": 0.6639,
+      "step": 29862
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.114485025405884,
+      "learning_rate": 5.465603935646994e-06,
+      "loss": 0.4667,
+      "step": 29863
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.262676477432251,
+      "learning_rate": 5.465341422464279e-06,
+      "loss": 0.6442,
+      "step": 29864
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.253758668899536,
+      "learning_rate": 5.465078907987623e-06,
+      "loss": 0.7851,
+      "step": 29865
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.009582042694092,
+      "learning_rate": 5.464816392217757e-06,
+      "loss": 0.4381,
+      "step": 29866
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.159876823425293,
+      "learning_rate": 5.464553875155411e-06,
+      "loss": 0.8028,
+      "step": 29867
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.9925084114074707,
+      "learning_rate": 5.464291356801313e-06,
+      "loss": 0.6391,
+      "step": 29868
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.323443651199341,
+      "learning_rate": 5.464028837156197e-06,
+      "loss": 0.8716,
+      "step": 29869
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.0180575847625732,
+      "learning_rate": 5.46376631622079e-06,
+      "loss": 0.7919,
+      "step": 29870
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.7475547790527344,
+      "learning_rate": 5.463503793995823e-06,
+      "loss": 0.4403,
+      "step": 29871
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.899543046951294,
+      "learning_rate": 5.463241270482024e-06,
+      "loss": 0.5869,
+      "step": 29872
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 4.1981611251831055,
+      "learning_rate": 5.4629787456801254e-06,
+      "loss": 0.6047,
+      "step": 29873
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.897512435913086,
+      "learning_rate": 5.462716219590857e-06,
+      "loss": 0.5112,
+      "step": 29874
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.4289140701293945,
+      "learning_rate": 5.462453692214946e-06,
+      "loss": 0.7435,
+      "step": 29875
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.3373045921325684,
+      "learning_rate": 5.462191163553126e-06,
+      "loss": 0.4481,
+      "step": 29876
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.1474146842956543,
+      "learning_rate": 5.461928633606126e-06,
+      "loss": 0.6664,
+      "step": 29877
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.7891533374786377,
+      "learning_rate": 5.4616661023746755e-06,
+      "loss": 0.6468,
+      "step": 29878
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.9786570072174072,
+      "learning_rate": 5.461403569859504e-06,
+      "loss": 0.6056,
+      "step": 29879
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.8238558769226074,
+      "learning_rate": 5.4611410360613424e-06,
+      "loss": 0.7475,
+      "step": 29880
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.197908401489258,
+      "learning_rate": 5.460878500980919e-06,
+      "loss": 0.7009,
+      "step": 29881
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.651628017425537,
+      "learning_rate": 5.460615964618966e-06,
+      "loss": 0.8326,
+      "step": 29882
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.209768772125244,
+      "learning_rate": 5.460353426976214e-06,
+      "loss": 0.7625,
+      "step": 29883
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.7616167068481445,
+      "learning_rate": 5.46009088805339e-06,
+      "loss": 0.6686,
+      "step": 29884
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.640812635421753,
+      "learning_rate": 5.459828347851226e-06,
+      "loss": 0.6507,
+      "step": 29885
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.6457908153533936,
+      "learning_rate": 5.459565806370452e-06,
+      "loss": 0.566,
+      "step": 29886
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.3421099185943604,
+      "learning_rate": 5.459303263611797e-06,
+      "loss": 0.7282,
+      "step": 29887
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 5.584699630737305,
+      "learning_rate": 5.459040719575991e-06,
+      "loss": 0.7514,
+      "step": 29888
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.345973014831543,
+      "learning_rate": 5.458778174263766e-06,
+      "loss": 0.6509,
+      "step": 29889
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.8001940250396729,
+      "learning_rate": 5.45851562767585e-06,
+      "loss": 0.7148,
+      "step": 29890
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.3178043365478516,
+      "learning_rate": 5.458253079812974e-06,
+      "loss": 0.5983,
+      "step": 29891
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.0489695072174072,
+      "learning_rate": 5.457990530675868e-06,
+      "loss": 0.6942,
+      "step": 29892
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.076216220855713,
+      "learning_rate": 5.4577279802652625e-06,
+      "loss": 0.6589,
+      "step": 29893
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.1365044116973877,
+      "learning_rate": 5.4574654285818864e-06,
+      "loss": 0.7484,
+      "step": 29894
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.427506923675537,
+      "learning_rate": 5.45720287562647e-06,
+      "loss": 0.7952,
+      "step": 29895
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.6988142728805542,
+      "learning_rate": 5.456940321399744e-06,
+      "loss": 0.721,
+      "step": 29896
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.962067723274231,
+      "learning_rate": 5.4566777659024386e-06,
+      "loss": 0.5226,
+      "step": 29897
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.799800395965576,
+      "learning_rate": 5.4564152091352826e-06,
+      "loss": 0.6872,
+      "step": 29898
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.9786815643310547,
+      "learning_rate": 5.456152651099007e-06,
+      "loss": 0.6675,
+      "step": 29899
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.1054770946502686,
+      "learning_rate": 5.455890091794341e-06,
+      "loss": 0.5624,
+      "step": 29900
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.2658135890960693,
+      "learning_rate": 5.4556275312220174e-06,
+      "loss": 0.6919,
+      "step": 29901
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.8530247211456299,
+      "learning_rate": 5.455364969382764e-06,
+      "loss": 0.7938,
+      "step": 29902
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.667039155960083,
+      "learning_rate": 5.455102406277309e-06,
+      "loss": 0.6639,
+      "step": 29903
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 7.642103672027588,
+      "learning_rate": 5.454839841906388e-06,
+      "loss": 0.5512,
+      "step": 29904
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.731695532798767,
+      "learning_rate": 5.454577276270725e-06,
+      "loss": 0.9047,
+      "step": 29905
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.9365122318267822,
+      "learning_rate": 5.454314709371053e-06,
+      "loss": 0.8052,
+      "step": 29906
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.626613140106201,
+      "learning_rate": 5.454052141208104e-06,
+      "loss": 0.4717,
+      "step": 29907
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.568246364593506,
+      "learning_rate": 5.453789571782605e-06,
+      "loss": 0.6096,
+      "step": 29908
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.9072257280349731,
+      "learning_rate": 5.4535270010952865e-06,
+      "loss": 0.494,
+      "step": 29909
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.5009263753890991,
+      "learning_rate": 5.453264429146879e-06,
+      "loss": 0.5429,
+      "step": 29910
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.105801582336426,
+      "learning_rate": 5.453001855938115e-06,
+      "loss": 0.6584,
+      "step": 29911
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.5066816806793213,
+      "learning_rate": 5.452739281469721e-06,
+      "loss": 0.5778,
+      "step": 29912
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.6108357906341553,
+      "learning_rate": 5.45247670574243e-06,
+      "loss": 0.637,
+      "step": 29913
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.955983877182007,
+      "learning_rate": 5.452214128756969e-06,
+      "loss": 0.5337,
+      "step": 29914
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.0286412239074707,
+      "learning_rate": 5.4519515505140705e-06,
+      "loss": 0.7627,
+      "step": 29915
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.9631108045578003,
+      "learning_rate": 5.4516889710144645e-06,
+      "loss": 0.6281,
+      "step": 29916
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.7666642665863037,
+      "learning_rate": 5.4514263902588815e-06,
+      "loss": 0.6631,
+      "step": 29917
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.797891616821289,
+      "learning_rate": 5.45116380824805e-06,
+      "loss": 0.698,
+      "step": 29918
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.3855338096618652,
+      "learning_rate": 5.4509012249827e-06,
+      "loss": 0.836,
+      "step": 29919
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.1148183345794678,
+      "learning_rate": 5.450638640463563e-06,
+      "loss": 0.6003,
+      "step": 29920
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.1910815238952637,
+      "learning_rate": 5.450376054691369e-06,
+      "loss": 0.869,
+      "step": 29921
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.9509477615356445,
+      "learning_rate": 5.450113467666847e-06,
+      "loss": 0.5461,
+      "step": 29922
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.093989372253418,
+      "learning_rate": 5.44985087939073e-06,
+      "loss": 0.4037,
+      "step": 29923
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 4.108108997344971,
+      "learning_rate": 5.449588289863745e-06,
+      "loss": 0.6424,
+      "step": 29924
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.865940809249878,
+      "learning_rate": 5.449325699086625e-06,
+      "loss": 0.6291,
+      "step": 29925
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.493955135345459,
+      "learning_rate": 5.449063107060097e-06,
+      "loss": 0.7521,
+      "step": 29926
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.3931891918182373,
+      "learning_rate": 5.448800513784893e-06,
+      "loss": 0.2826,
+      "step": 29927
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.847660541534424,
+      "learning_rate": 5.448537919261743e-06,
+      "loss": 0.5113,
+      "step": 29928
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.028007745742798,
+      "learning_rate": 5.448275323491375e-06,
+      "loss": 0.8129,
+      "step": 29929
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 4.7169671058654785,
+      "learning_rate": 5.448012726474524e-06,
+      "loss": 0.8986,
+      "step": 29930
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.3692221641540527,
+      "learning_rate": 5.447750128211918e-06,
+      "loss": 0.5989,
+      "step": 29931
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.7062950134277344,
+      "learning_rate": 5.447487528704284e-06,
+      "loss": 0.7101,
+      "step": 29932
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.6793160438537598,
+      "learning_rate": 5.447224927952356e-06,
+      "loss": 0.578,
+      "step": 29933
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.1530866622924805,
+      "learning_rate": 5.446962325956864e-06,
+      "loss": 0.7834,
+      "step": 29934
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.9023706912994385,
+      "learning_rate": 5.446699722718536e-06,
+      "loss": 0.6764,
+      "step": 29935
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.6557188034057617,
+      "learning_rate": 5.446437118238104e-06,
+      "loss": 0.7667,
+      "step": 29936
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.6975339651107788,
+      "learning_rate": 5.446174512516296e-06,
+      "loss": 0.7058,
+      "step": 29937
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.034142255783081,
+      "learning_rate": 5.445911905553845e-06,
+      "loss": 0.8563,
+      "step": 29938
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.262779712677002,
+      "learning_rate": 5.445649297351481e-06,
+      "loss": 0.7341,
+      "step": 29939
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.941974639892578,
+      "learning_rate": 5.445386687909931e-06,
+      "loss": 0.7635,
+      "step": 29940
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.076810359954834,
+      "learning_rate": 5.445124077229929e-06,
+      "loss": 0.953,
+      "step": 29941
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.213116407394409,
+      "learning_rate": 5.444861465312204e-06,
+      "loss": 0.7136,
+      "step": 29942
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.1102216243743896,
+      "learning_rate": 5.4445988521574854e-06,
+      "loss": 0.5078,
+      "step": 29943
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.7023862600326538,
+      "learning_rate": 5.444336237766504e-06,
+      "loss": 0.6659,
+      "step": 29944
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.817857027053833,
+      "learning_rate": 5.44407362213999e-06,
+      "loss": 0.5486,
+      "step": 29945
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 6.454439163208008,
+      "learning_rate": 5.443811005278674e-06,
+      "loss": 0.3536,
+      "step": 29946
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 4.074324131011963,
+      "learning_rate": 5.443548387183285e-06,
+      "loss": 0.5163,
+      "step": 29947
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.7389391660690308,
+      "learning_rate": 5.443285767854554e-06,
+      "loss": 0.6078,
+      "step": 29948
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.4455816745758057,
+      "learning_rate": 5.443023147293212e-06,
+      "loss": 0.7432,
+      "step": 29949
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.600210666656494,
+      "learning_rate": 5.442760525499988e-06,
+      "loss": 0.5484,
+      "step": 29950
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.7300693988800049,
+      "learning_rate": 5.442497902475614e-06,
+      "loss": 0.4967,
+      "step": 29951
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.0864052772521973,
+      "learning_rate": 5.442235278220818e-06,
+      "loss": 0.7619,
+      "step": 29952
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.0977940559387207,
+      "learning_rate": 5.441972652736332e-06,
+      "loss": 0.6844,
+      "step": 29953
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.076720952987671,
+      "learning_rate": 5.441710026022885e-06,
+      "loss": 0.6321,
+      "step": 29954
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.0123660564422607,
+      "learning_rate": 5.441447398081209e-06,
+      "loss": 0.6961,
+      "step": 29955
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.9151365756988525,
+      "learning_rate": 5.441184768912032e-06,
+      "loss": 0.6795,
+      "step": 29956
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.5083839893341064,
+      "learning_rate": 5.440922138516086e-06,
+      "loss": 0.5961,
+      "step": 29957
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.761932611465454,
+      "learning_rate": 5.4406595068941014e-06,
+      "loss": 0.4521,
+      "step": 29958
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.795318841934204,
+      "learning_rate": 5.440396874046807e-06,
+      "loss": 0.6572,
+      "step": 29959
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.8162211179733276,
+      "learning_rate": 5.440134239974932e-06,
+      "loss": 0.5551,
+      "step": 29960
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.385648727416992,
+      "learning_rate": 5.439871604679211e-06,
+      "loss": 0.4858,
+      "step": 29961
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.8864128589630127,
+      "learning_rate": 5.439608968160371e-06,
+      "loss": 0.6765,
+      "step": 29962
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.510307550430298,
+      "learning_rate": 5.439346330419144e-06,
+      "loss": 0.7226,
+      "step": 29963
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.885080099105835,
+      "learning_rate": 5.439083691456258e-06,
+      "loss": 0.7303,
+      "step": 29964
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.2855844497680664,
+      "learning_rate": 5.438821051272446e-06,
+      "loss": 0.571,
+      "step": 29965
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.1567018032073975,
+      "learning_rate": 5.438558409868438e-06,
+      "loss": 0.7361,
+      "step": 29966
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.0719082355499268,
+      "learning_rate": 5.438295767244961e-06,
+      "loss": 0.6637,
+      "step": 29967
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.575484037399292,
+      "learning_rate": 5.438033123402748e-06,
+      "loss": 0.5699,
+      "step": 29968
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.336733818054199,
+      "learning_rate": 5.43777047834253e-06,
+      "loss": 0.6313,
+      "step": 29969
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.0654470920562744,
+      "learning_rate": 5.437507832065036e-06,
+      "loss": 0.6953,
+      "step": 29970
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.4722237586975098,
+      "learning_rate": 5.4372451845709966e-06,
+      "loss": 0.7195,
+      "step": 29971
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.8773959875106812,
+      "learning_rate": 5.436982535861142e-06,
+      "loss": 0.542,
+      "step": 29972
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.866184949874878,
+      "learning_rate": 5.436719885936203e-06,
+      "loss": 0.5697,
+      "step": 29973
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.580242872238159,
+      "learning_rate": 5.436457234796908e-06,
+      "loss": 0.7072,
+      "step": 29974
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.225391387939453,
+      "learning_rate": 5.436194582443991e-06,
+      "loss": 0.6602,
+      "step": 29975
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.777858018875122,
+      "learning_rate": 5.43593192887818e-06,
+      "loss": 0.7938,
+      "step": 29976
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.462747812271118,
+      "learning_rate": 5.4356692741002045e-06,
+      "loss": 0.5131,
+      "step": 29977
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.3952853679656982,
+      "learning_rate": 5.435406618110797e-06,
+      "loss": 0.91,
+      "step": 29978
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.8710637092590332,
+      "learning_rate": 5.435143960910686e-06,
+      "loss": 0.6845,
+      "step": 29979
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.5286920070648193,
+      "learning_rate": 5.434881302500602e-06,
+      "loss": 0.6824,
+      "step": 29980
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.5107357501983643,
+      "learning_rate": 5.434618642881278e-06,
+      "loss": 0.8337,
+      "step": 29981
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.1689558029174805,
+      "learning_rate": 5.434355982053441e-06,
+      "loss": 0.5792,
+      "step": 29982
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.8485023975372314,
+      "learning_rate": 5.434093320017822e-06,
+      "loss": 0.5492,
+      "step": 29983
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.9294790029525757,
+      "learning_rate": 5.433830656775154e-06,
+      "loss": 0.5813,
+      "step": 29984
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.746877908706665,
+      "learning_rate": 5.433567992326165e-06,
+      "loss": 0.9623,
+      "step": 29985
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.4599790573120117,
+      "learning_rate": 5.4333053266715855e-06,
+      "loss": 0.7383,
+      "step": 29986
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.1758711338043213,
+      "learning_rate": 5.433042659812146e-06,
+      "loss": 0.6627,
+      "step": 29987
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.6365042924880981,
+      "learning_rate": 5.432779991748578e-06,
+      "loss": 0.8739,
+      "step": 29988
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.905656337738037,
+      "learning_rate": 5.432517322481609e-06,
+      "loss": 0.5013,
+      "step": 29989
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.257732391357422,
+      "learning_rate": 5.432254652011973e-06,
+      "loss": 0.7451,
+      "step": 29990
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.789566993713379,
+      "learning_rate": 5.431991980340398e-06,
+      "loss": 0.5183,
+      "step": 29991
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 4.1973395347595215,
+      "learning_rate": 5.431729307467614e-06,
+      "loss": 0.33,
+      "step": 29992
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.319751262664795,
+      "learning_rate": 5.431466633394356e-06,
+      "loss": 0.7876,
+      "step": 29993
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.253512382507324,
+      "learning_rate": 5.431203958121348e-06,
+      "loss": 0.6732,
+      "step": 29994
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.9795029163360596,
+      "learning_rate": 5.4309412816493245e-06,
+      "loss": 0.4604,
+      "step": 29995
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.8642266988754272,
+      "learning_rate": 5.430678603979015e-06,
+      "loss": 0.6038,
+      "step": 29996
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.7473361492156982,
+      "learning_rate": 5.430415925111149e-06,
+      "loss": 0.4438,
+      "step": 29997
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.2081680297851562,
+      "learning_rate": 5.430153245046458e-06,
+      "loss": 0.7163,
+      "step": 29998
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.6072306632995605,
+      "learning_rate": 5.4298905637856706e-06,
+      "loss": 0.6748,
+      "step": 29999
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.502854108810425,
+      "learning_rate": 5.429627881329519e-06,
+      "loss": 0.5037,
+      "step": 30000
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.3536837100982666,
+      "learning_rate": 5.429365197678734e-06,
+      "loss": 0.6744,
+      "step": 30001
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.8302674293518066,
+      "learning_rate": 5.4291025128340454e-06,
+      "loss": 0.6033,
+      "step": 30002
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.914604902267456,
+      "learning_rate": 5.428839826796183e-06,
+      "loss": 0.598,
+      "step": 30003
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.7271696329116821,
+      "learning_rate": 5.4285771395658775e-06,
+      "loss": 0.6852,
+      "step": 30004
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 4.956901550292969,
+      "learning_rate": 5.4283144511438594e-06,
+      "loss": 0.8435,
+      "step": 30005
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.359255790710449,
+      "learning_rate": 5.428051761530859e-06,
+      "loss": 0.4923,
+      "step": 30006
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.518678665161133,
+      "learning_rate": 5.427789070727608e-06,
+      "loss": 0.8121,
+      "step": 30007
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.433645248413086,
+      "learning_rate": 5.4275263787348365e-06,
+      "loss": 0.5816,
+      "step": 30008
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.460726261138916,
+      "learning_rate": 5.4272636855532734e-06,
+      "loss": 0.8425,
+      "step": 30009
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 5.5896735191345215,
+      "learning_rate": 5.427000991183648e-06,
+      "loss": 0.8768,
+      "step": 30010
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.8341118097305298,
+      "learning_rate": 5.426738295626696e-06,
+      "loss": 0.4738,
+      "step": 30011
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.268113136291504,
+      "learning_rate": 5.426475598883143e-06,
+      "loss": 0.5859,
+      "step": 30012
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.6105332374572754,
+      "learning_rate": 5.4262129009537214e-06,
+      "loss": 0.6826,
+      "step": 30013
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.5353047847747803,
+      "learning_rate": 5.425950201839162e-06,
+      "loss": 0.5859,
+      "step": 30014
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.041717052459717,
+      "learning_rate": 5.425687501540194e-06,
+      "loss": 0.6804,
+      "step": 30015
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.3301305770874023,
+      "learning_rate": 5.425424800057548e-06,
+      "loss": 0.3938,
+      "step": 30016
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.127584457397461,
+      "learning_rate": 5.425162097391956e-06,
+      "loss": 0.7717,
+      "step": 30017
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.364867687225342,
+      "learning_rate": 5.424899393544148e-06,
+      "loss": 0.7585,
+      "step": 30018
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.1573777198791504,
+      "learning_rate": 5.424636688514853e-06,
+      "loss": 0.8875,
+      "step": 30019
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.1037137508392334,
+      "learning_rate": 5.424373982304802e-06,
+      "loss": 0.7666,
+      "step": 30020
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.5338568687438965,
+      "learning_rate": 5.4241112749147264e-06,
+      "loss": 0.5858,
+      "step": 30021
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.7774510383605957,
+      "learning_rate": 5.423848566345356e-06,
+      "loss": 0.4611,
+      "step": 30022
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.923489570617676,
+      "learning_rate": 5.423585856597422e-06,
+      "loss": 0.6783,
+      "step": 30023
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.007567882537842,
+      "learning_rate": 5.4233231456716554e-06,
+      "loss": 0.4156,
+      "step": 30024
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.8528366088867188,
+      "learning_rate": 5.423060433568784e-06,
+      "loss": 0.5339,
+      "step": 30025
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.080333709716797,
+      "learning_rate": 5.422797720289541e-06,
+      "loss": 0.7359,
+      "step": 30026
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.7635804414749146,
+      "learning_rate": 5.422535005834655e-06,
+      "loss": 0.6333,
+      "step": 30027
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.9084564447402954,
+      "learning_rate": 5.422272290204859e-06,
+      "loss": 0.8614,
+      "step": 30028
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.2802557945251465,
+      "learning_rate": 5.42200957340088e-06,
+      "loss": 0.7842,
+      "step": 30029
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 5.813424587249756,
+      "learning_rate": 5.421746855423451e-06,
+      "loss": 0.5241,
+      "step": 30030
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.9811208248138428,
+      "learning_rate": 5.421484136273302e-06,
+      "loss": 0.548,
+      "step": 30031
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.8611721992492676,
+      "learning_rate": 5.421221415951163e-06,
+      "loss": 0.7706,
+      "step": 30032
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.8587000370025635,
+      "learning_rate": 5.420958694457766e-06,
+      "loss": 0.5008,
+      "step": 30033
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.222414493560791,
+      "learning_rate": 5.420695971793841e-06,
+      "loss": 0.5356,
+      "step": 30034
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.6828992366790771,
+      "learning_rate": 5.420433247960118e-06,
+      "loss": 0.6385,
+      "step": 30035
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.7566523551940918,
+      "learning_rate": 5.4201705229573274e-06,
+      "loss": 0.7092,
+      "step": 30036
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.031856060028076,
+      "learning_rate": 5.419907796786199e-06,
+      "loss": 0.5329,
+      "step": 30037
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.4827420711517334,
+      "learning_rate": 5.419645069447466e-06,
+      "loss": 0.8157,
+      "step": 30038
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.5786471366882324,
+      "learning_rate": 5.419382340941857e-06,
+      "loss": 0.4621,
+      "step": 30039
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.4973673820495605,
+      "learning_rate": 5.4191196112701015e-06,
+      "loss": 0.7309,
+      "step": 30040
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.3979032039642334,
+      "learning_rate": 5.4188568804329314e-06,
+      "loss": 0.3981,
+      "step": 30041
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.4480957984924316,
+      "learning_rate": 5.4185941484310776e-06,
+      "loss": 0.4207,
+      "step": 30042
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.995986819267273,
+      "learning_rate": 5.41833141526527e-06,
+      "loss": 0.5612,
+      "step": 30043
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.1143298149108887,
+      "learning_rate": 5.4180686809362405e-06,
+      "loss": 0.7302,
+      "step": 30044
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.209280252456665,
+      "learning_rate": 5.4178059454447184e-06,
+      "loss": 0.6658,
+      "step": 30045
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.3901138305664062,
+      "learning_rate": 5.417543208791434e-06,
+      "loss": 0.9489,
+      "step": 30046
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.9933964014053345,
+      "learning_rate": 5.417280470977119e-06,
+      "loss": 0.7333,
+      "step": 30047
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.2122862339019775,
+      "learning_rate": 5.417017732002503e-06,
+      "loss": 0.8106,
+      "step": 30048
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.002141237258911,
+      "learning_rate": 5.416754991868316e-06,
+      "loss": 0.8195,
+      "step": 30049
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.9532028436660767,
+      "learning_rate": 5.416492250575291e-06,
+      "loss": 0.4722,
+      "step": 30050
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.325176239013672,
+      "learning_rate": 5.416229508124155e-06,
+      "loss": 0.599,
+      "step": 30051
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.3214731216430664,
+      "learning_rate": 5.415966764515642e-06,
+      "loss": 0.6204,
+      "step": 30052
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.4736114740371704,
+      "learning_rate": 5.415704019750482e-06,
+      "loss": 0.5881,
+      "step": 30053
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.753610134124756,
+      "learning_rate": 5.415441273829404e-06,
+      "loss": 0.7769,
+      "step": 30054
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.924734115600586,
+      "learning_rate": 5.415178526753139e-06,
+      "loss": 0.6357,
+      "step": 30055
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.470273494720459,
+      "learning_rate": 5.4149157785224185e-06,
+      "loss": 0.7081,
+      "step": 30056
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.94818913936615,
+      "learning_rate": 5.414653029137973e-06,
+      "loss": 0.6783,
+      "step": 30057
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.78350031375885,
+      "learning_rate": 5.4143902786005335e-06,
+      "loss": 0.5994,
+      "step": 30058
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.9109630584716797,
+      "learning_rate": 5.414127526910829e-06,
+      "loss": 0.6988,
+      "step": 30059
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.245011568069458,
+      "learning_rate": 5.413864774069589e-06,
+      "loss": 0.6821,
+      "step": 30060
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.2511847019195557,
+      "learning_rate": 5.413602020077548e-06,
+      "loss": 0.3074,
+      "step": 30061
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.857544183731079,
+      "learning_rate": 5.413339264935434e-06,
+      "loss": 0.5599,
+      "step": 30062
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.7401971817016602,
+      "learning_rate": 5.4130765086439795e-06,
+      "loss": 0.6809,
+      "step": 30063
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.2520523071289062,
+      "learning_rate": 5.4128137512039135e-06,
+      "loss": 0.718,
+      "step": 30064
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.7797536849975586,
+      "learning_rate": 5.412550992615967e-06,
+      "loss": 0.8538,
+      "step": 30065
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.0176289081573486,
+      "learning_rate": 5.41228823288087e-06,
+      "loss": 0.6435,
+      "step": 30066
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.7378900051116943,
+      "learning_rate": 5.412025471999355e-06,
+      "loss": 0.8691,
+      "step": 30067
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 4.761762619018555,
+      "learning_rate": 5.411762709972151e-06,
+      "loss": 0.7803,
+      "step": 30068
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.9632219076156616,
+      "learning_rate": 5.41149994679999e-06,
+      "loss": 0.7203,
+      "step": 30069
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.977184534072876,
+      "learning_rate": 5.411237182483601e-06,
+      "loss": 0.7113,
+      "step": 30070
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.7830491065979004,
+      "learning_rate": 5.4109744170237154e-06,
+      "loss": 0.885,
+      "step": 30071
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.0379133224487305,
+      "learning_rate": 5.410711650421063e-06,
+      "loss": 0.2603,
+      "step": 30072
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.62015962600708,
+      "learning_rate": 5.410448882676377e-06,
+      "loss": 0.4243,
+      "step": 30073
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.8987691402435303,
+      "learning_rate": 5.410186113790385e-06,
+      "loss": 0.6587,
+      "step": 30074
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.496814250946045,
+      "learning_rate": 5.409923343763821e-06,
+      "loss": 0.5343,
+      "step": 30075
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.3487160205841064,
+      "learning_rate": 5.409660572597413e-06,
+      "loss": 0.6441,
+      "step": 30076
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.0955021381378174,
+      "learning_rate": 5.409397800291891e-06,
+      "loss": 0.4561,
+      "step": 30077
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.575565814971924,
+      "learning_rate": 5.4091350268479895e-06,
+      "loss": 0.7568,
+      "step": 30078
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 4.155429840087891,
+      "learning_rate": 5.408872252266435e-06,
+      "loss": 0.923,
+      "step": 30079
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.3890397548675537,
+      "learning_rate": 5.40860947654796e-06,
+      "loss": 0.6388,
+      "step": 30080
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 5.102009296417236,
+      "learning_rate": 5.408346699693295e-06,
+      "loss": 0.4997,
+      "step": 30081
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.47226881980896,
+      "learning_rate": 5.408083921703171e-06,
+      "loss": 0.5472,
+      "step": 30082
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.5193943977355957,
+      "learning_rate": 5.407821142578319e-06,
+      "loss": 0.7873,
+      "step": 30083
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.6936943531036377,
+      "learning_rate": 5.4075583623194695e-06,
+      "loss": 0.6674,
+      "step": 30084
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.03936767578125,
+      "learning_rate": 5.407295580927352e-06,
+      "loss": 0.6635,
+      "step": 30085
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.7815749645233154,
+      "learning_rate": 5.407032798402698e-06,
+      "loss": 0.9619,
+      "step": 30086
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.7536530494689941,
+      "learning_rate": 5.406770014746239e-06,
+      "loss": 0.4605,
+      "step": 30087
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.102756977081299,
+      "learning_rate": 5.406507229958704e-06,
+      "loss": 0.6948,
+      "step": 30088
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.492539405822754,
+      "learning_rate": 5.406244444040826e-06,
+      "loss": 0.6944,
+      "step": 30089
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.9931401014328003,
+      "learning_rate": 5.405981656993333e-06,
+      "loss": 0.5156,
+      "step": 30090
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.7505812644958496,
+      "learning_rate": 5.405718868816958e-06,
+      "loss": 0.6352,
+      "step": 30091
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.1114981174468994,
+      "learning_rate": 5.40545607951243e-06,
+      "loss": 0.7517,
+      "step": 30092
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.544196128845215,
+      "learning_rate": 5.405193289080482e-06,
+      "loss": 0.7858,
+      "step": 30093
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.5475780963897705,
+      "learning_rate": 5.404930497521842e-06,
+      "loss": 0.537,
+      "step": 30094
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.266073226928711,
+      "learning_rate": 5.404667704837242e-06,
+      "loss": 0.6107,
+      "step": 30095
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.2369511127471924,
+      "learning_rate": 5.404404911027413e-06,
+      "loss": 0.4288,
+      "step": 30096
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.0197770595550537,
+      "learning_rate": 5.404142116093085e-06,
+      "loss": 0.6592,
+      "step": 30097
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.710620403289795,
+      "learning_rate": 5.4038793200349905e-06,
+      "loss": 0.6331,
+      "step": 30098
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 4.054183006286621,
+      "learning_rate": 5.403616522853858e-06,
+      "loss": 0.6953,
+      "step": 30099
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.2288172245025635,
+      "learning_rate": 5.40335372455042e-06,
+      "loss": 0.6524,
+      "step": 30100
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.9065734148025513,
+      "learning_rate": 5.403090925125405e-06,
+      "loss": 0.8068,
+      "step": 30101
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.617355465888977,
+      "learning_rate": 5.402828124579544e-06,
+      "loss": 0.7371,
+      "step": 30102
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.9530028104782104,
+      "learning_rate": 5.402565322913571e-06,
+      "loss": 0.5848,
+      "step": 30103
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 5.1112470626831055,
+      "learning_rate": 5.402302520128214e-06,
+      "loss": 0.6512,
+      "step": 30104
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.309795618057251,
+      "learning_rate": 5.402039716224205e-06,
+      "loss": 0.5537,
+      "step": 30105
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 4.6465325355529785,
+      "learning_rate": 5.401776911202274e-06,
+      "loss": 0.4865,
+      "step": 30106
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.8259143829345703,
+      "learning_rate": 5.401514105063153e-06,
+      "loss": 0.4431,
+      "step": 30107
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.5954288244247437,
+      "learning_rate": 5.401251297807569e-06,
+      "loss": 0.3838,
+      "step": 30108
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.4709970951080322,
+      "learning_rate": 5.400988489436257e-06,
+      "loss": 0.7816,
+      "step": 30109
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.917480707168579,
+      "learning_rate": 5.400725679949946e-06,
+      "loss": 0.7874,
+      "step": 30110
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.62376070022583,
+      "learning_rate": 5.400462869349367e-06,
+      "loss": 0.7504,
+      "step": 30111
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 4.779388904571533,
+      "learning_rate": 5.400200057635251e-06,
+      "loss": 0.6929,
+      "step": 30112
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 4.445816993713379,
+      "learning_rate": 5.399937244808328e-06,
+      "loss": 1.1576,
+      "step": 30113
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.1300225257873535,
+      "learning_rate": 5.39967443086933e-06,
+      "loss": 0.6443,
+      "step": 30114
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.354957103729248,
+      "learning_rate": 5.399411615818987e-06,
+      "loss": 0.5591,
+      "step": 30115
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.5906150341033936,
+      "learning_rate": 5.39914879965803e-06,
+      "loss": 0.4846,
+      "step": 30116
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.8047229051589966,
+      "learning_rate": 5.39888598238719e-06,
+      "loss": 0.6175,
+      "step": 30117
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.5382962226867676,
+      "learning_rate": 5.3986231640071964e-06,
+      "loss": 0.6006,
+      "step": 30118
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.136073350906372,
+      "learning_rate": 5.398360344518782e-06,
+      "loss": 0.7884,
+      "step": 30119
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.859896659851074,
+      "learning_rate": 5.398097523922676e-06,
+      "loss": 0.8887,
+      "step": 30120
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.2197790145874023,
+      "learning_rate": 5.397834702219611e-06,
+      "loss": 0.7713,
+      "step": 30121
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.2532026767730713,
+      "learning_rate": 5.3975718794103146e-06,
+      "loss": 0.7642,
+      "step": 30122
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.3005099296569824,
+      "learning_rate": 5.397309055495522e-06,
+      "loss": 0.733,
+      "step": 30123
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.4339914321899414,
+      "learning_rate": 5.397046230475961e-06,
+      "loss": 0.4678,
+      "step": 30124
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.805081605911255,
+      "learning_rate": 5.396783404352364e-06,
+      "loss": 0.8594,
+      "step": 30125
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.8888269662857056,
+      "learning_rate": 5.3965205771254595e-06,
+      "loss": 0.6076,
+      "step": 30126
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.9508311748504639,
+      "learning_rate": 5.39625774879598e-06,
+      "loss": 0.594,
+      "step": 30127
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.1184287071228027,
+      "learning_rate": 5.395994919364656e-06,
+      "loss": 0.5949,
+      "step": 30128
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.267770767211914,
+      "learning_rate": 5.395732088832219e-06,
+      "loss": 0.7248,
+      "step": 30129
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.8050379753112793,
+      "learning_rate": 5.395469257199399e-06,
+      "loss": 0.6448,
+      "step": 30130
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.072129011154175,
+      "learning_rate": 5.395206424466928e-06,
+      "loss": 0.691,
+      "step": 30131
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.502364158630371,
+      "learning_rate": 5.394943590635536e-06,
+      "loss": 0.5938,
+      "step": 30132
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.473193407058716,
+      "learning_rate": 5.394680755705954e-06,
+      "loss": 0.7504,
+      "step": 30133
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.9981876611709595,
+      "learning_rate": 5.39441791967891e-06,
+      "loss": 0.7438,
+      "step": 30134
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.221575975418091,
+      "learning_rate": 5.394155082555141e-06,
+      "loss": 0.4154,
+      "step": 30135
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.359027862548828,
+      "learning_rate": 5.393892244335371e-06,
+      "loss": 0.6094,
+      "step": 30136
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.613303542137146,
+      "learning_rate": 5.3936294050203374e-06,
+      "loss": 0.5475,
+      "step": 30137
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.6599764823913574,
+      "learning_rate": 5.393366564610766e-06,
+      "loss": 0.4931,
+      "step": 30138
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.347835063934326,
+      "learning_rate": 5.39310372310739e-06,
+      "loss": 0.6025,
+      "step": 30139
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.8799636363983154,
+      "learning_rate": 5.39284088051094e-06,
+      "loss": 0.5071,
+      "step": 30140
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.035580635070801,
+      "learning_rate": 5.392578036822147e-06,
+      "loss": 0.5146,
+      "step": 30141
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.608430862426758,
+      "learning_rate": 5.392315192041742e-06,
+      "loss": 0.7183,
+      "step": 30142
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.22241473197937,
+      "learning_rate": 5.392052346170454e-06,
+      "loss": 0.7053,
+      "step": 30143
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.7962727546691895,
+      "learning_rate": 5.391789499209015e-06,
+      "loss": 0.7501,
+      "step": 30144
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.8917932510375977,
+      "learning_rate": 5.3915266511581565e-06,
+      "loss": 0.5055,
+      "step": 30145
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.5061285495758057,
+      "learning_rate": 5.39126380201861e-06,
+      "loss": 0.6677,
+      "step": 30146
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.9637433290481567,
+      "learning_rate": 5.391000951791105e-06,
+      "loss": 0.8358,
+      "step": 30147
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.837144613265991,
+      "learning_rate": 5.3907381004763735e-06,
+      "loss": 0.5394,
+      "step": 30148
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.9199256896972656,
+      "learning_rate": 5.390475248075144e-06,
+      "loss": 0.4922,
+      "step": 30149
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.66550874710083,
+      "learning_rate": 5.390212394588151e-06,
+      "loss": 0.3647,
+      "step": 30150
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.278623342514038,
+      "learning_rate": 5.389949540016123e-06,
+      "loss": 0.5243,
+      "step": 30151
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.486687660217285,
+      "learning_rate": 5.389686684359791e-06,
+      "loss": 0.8825,
+      "step": 30152
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.8886384963989258,
+      "learning_rate": 5.3894238276198856e-06,
+      "loss": 0.6396,
+      "step": 30153
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.341374397277832,
+      "learning_rate": 5.3891609697971395e-06,
+      "loss": 0.5958,
+      "step": 30154
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.549945831298828,
+      "learning_rate": 5.388898110892283e-06,
+      "loss": 0.7103,
+      "step": 30155
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 4.79037618637085,
+      "learning_rate": 5.388635250906045e-06,
+      "loss": 0.6969,
+      "step": 30156
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.8783187866210938,
+      "learning_rate": 5.3883723898391586e-06,
+      "loss": 0.441,
+      "step": 30157
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 4.348104476928711,
+      "learning_rate": 5.388109527692354e-06,
+      "loss": 0.6216,
+      "step": 30158
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.744290590286255,
+      "learning_rate": 5.387846664466363e-06,
+      "loss": 0.866,
+      "step": 30159
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.5456995964050293,
+      "learning_rate": 5.387583800161915e-06,
+      "loss": 0.6086,
+      "step": 30160
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.7833510637283325,
+      "learning_rate": 5.387320934779742e-06,
+      "loss": 0.7072,
+      "step": 30161
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 24.485074996948242,
+      "learning_rate": 5.3870580683205744e-06,
+      "loss": 0.7807,
+      "step": 30162
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.7364966869354248,
+      "learning_rate": 5.386795200785142e-06,
+      "loss": 0.7473,
+      "step": 30163
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.3599915504455566,
+      "learning_rate": 5.386532332174178e-06,
+      "loss": 0.6872,
+      "step": 30164
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.5860092639923096,
+      "learning_rate": 5.386269462488413e-06,
+      "loss": 0.8631,
+      "step": 30165
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.4105663299560547,
+      "learning_rate": 5.3860065917285776e-06,
+      "loss": 0.6605,
+      "step": 30166
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.731255054473877,
+      "learning_rate": 5.385743719895401e-06,
+      "loss": 0.3965,
+      "step": 30167
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.4677029848098755,
+      "learning_rate": 5.385480846989617e-06,
+      "loss": 0.377,
+      "step": 30168
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.7355141639709473,
+      "learning_rate": 5.385217973011955e-06,
+      "loss": 0.7713,
+      "step": 30169
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.2309937477111816,
+      "learning_rate": 5.384955097963146e-06,
+      "loss": 0.7981,
+      "step": 30170
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.0568010807037354,
+      "learning_rate": 5.38469222184392e-06,
+      "loss": 0.5144,
+      "step": 30171
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 4.870428562164307,
+      "learning_rate": 5.384429344655011e-06,
+      "loss": 0.7405,
+      "step": 30172
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.7795913219451904,
+      "learning_rate": 5.384166466397145e-06,
+      "loss": 0.5808,
+      "step": 30173
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.1132240295410156,
+      "learning_rate": 5.383903587071059e-06,
+      "loss": 0.3836,
+      "step": 30174
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.199373245239258,
+      "learning_rate": 5.383640706677479e-06,
+      "loss": 0.7405,
+      "step": 30175
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.413038969039917,
+      "learning_rate": 5.383377825217139e-06,
+      "loss": 0.6893,
+      "step": 30176
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.7452114820480347,
+      "learning_rate": 5.383114942690769e-06,
+      "loss": 0.7514,
+      "step": 30177
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.805541753768921,
+      "learning_rate": 5.3828520590991e-06,
+      "loss": 0.5241,
+      "step": 30178
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.077352285385132,
+      "learning_rate": 5.382589174442862e-06,
+      "loss": 0.6236,
+      "step": 30179
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.637629985809326,
+      "learning_rate": 5.3823262887227865e-06,
+      "loss": 0.446,
+      "step": 30180
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.903449296951294,
+      "learning_rate": 5.382063401939605e-06,
+      "loss": 0.6765,
+      "step": 30181
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 4.590291976928711,
+      "learning_rate": 5.38180051409405e-06,
+      "loss": 0.7554,
+      "step": 30182
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.016446113586426,
+      "learning_rate": 5.3815376251868475e-06,
+      "loss": 0.6303,
+      "step": 30183
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.4447360038757324,
+      "learning_rate": 5.381274735218734e-06,
+      "loss": 0.623,
+      "step": 30184
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.5051019191741943,
+      "learning_rate": 5.381011844190439e-06,
+      "loss": 0.504,
+      "step": 30185
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.6544251441955566,
+      "learning_rate": 5.380748952102692e-06,
+      "loss": 0.7798,
+      "step": 30186
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.0057241916656494,
+      "learning_rate": 5.380486058956224e-06,
+      "loss": 0.6016,
+      "step": 30187
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.5638551712036133,
+      "learning_rate": 5.3802231647517675e-06,
+      "loss": 0.7321,
+      "step": 30188
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.6909253597259521,
+      "learning_rate": 5.379960269490053e-06,
+      "loss": 0.2773,
+      "step": 30189
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.1429405212402344,
+      "learning_rate": 5.3796973731718105e-06,
+      "loss": 0.5209,
+      "step": 30190
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.862321138381958,
+      "learning_rate": 5.379434475797772e-06,
+      "loss": 0.6809,
+      "step": 30191
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.0422892570495605,
+      "learning_rate": 5.379171577368669e-06,
+      "loss": 0.618,
+      "step": 30192
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.6783576011657715,
+      "learning_rate": 5.378908677885233e-06,
+      "loss": 0.782,
+      "step": 30193
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.170912504196167,
+      "learning_rate": 5.378645777348192e-06,
+      "loss": 0.4776,
+      "step": 30194
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.2362608909606934,
+      "learning_rate": 5.378382875758279e-06,
+      "loss": 0.7913,
+      "step": 30195
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.3312528133392334,
+      "learning_rate": 5.378119973116226e-06,
+      "loss": 0.6693,
+      "step": 30196
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.5088775157928467,
+      "learning_rate": 5.3778570694227616e-06,
+      "loss": 0.4906,
+      "step": 30197
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.920923948287964,
+      "learning_rate": 5.37759416467862e-06,
+      "loss": 0.7609,
+      "step": 30198
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.81231689453125,
+      "learning_rate": 5.377331258884529e-06,
+      "loss": 0.6397,
+      "step": 30199
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.435917615890503,
+      "learning_rate": 5.377068352041223e-06,
+      "loss": 0.6892,
+      "step": 30200
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.8620052337646484,
+      "learning_rate": 5.37680544414943e-06,
+      "loss": 0.4895,
+      "step": 30201
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.310807466506958,
+      "learning_rate": 5.376542535209882e-06,
+      "loss": 0.4987,
+      "step": 30202
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.298598051071167,
+      "learning_rate": 5.376279625223311e-06,
+      "loss": 1.0063,
+      "step": 30203
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.6194510459899902,
+      "learning_rate": 5.376016714190445e-06,
+      "loss": 0.782,
+      "step": 30204
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.2255825996398926,
+      "learning_rate": 5.3757538021120194e-06,
+      "loss": 0.4631,
+      "step": 30205
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.9601143598556519,
+      "learning_rate": 5.375490888988762e-06,
+      "loss": 0.5164,
+      "step": 30206
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.320266008377075,
+      "learning_rate": 5.375227974821407e-06,
+      "loss": 0.5036,
+      "step": 30207
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.1575286388397217,
+      "learning_rate": 5.3749650596106826e-06,
+      "loss": 0.57,
+      "step": 30208
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.2415146827697754,
+      "learning_rate": 5.374702143357321e-06,
+      "loss": 0.5813,
+      "step": 30209
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.85274338722229,
+      "learning_rate": 5.374439226062054e-06,
+      "loss": 0.6725,
+      "step": 30210
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.73968505859375,
+      "learning_rate": 5.374176307725611e-06,
+      "loss": 0.6981,
+      "step": 30211
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.5425283908843994,
+      "learning_rate": 5.373913388348723e-06,
+      "loss": 0.5952,
+      "step": 30212
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 5.891293048858643,
+      "learning_rate": 5.373650467932122e-06,
+      "loss": 0.4005,
+      "step": 30213
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.765934944152832,
+      "learning_rate": 5.373387546476539e-06,
+      "loss": 0.7911,
+      "step": 30214
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.326310157775879,
+      "learning_rate": 5.373124623982706e-06,
+      "loss": 0.8666,
+      "step": 30215
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.014909267425537,
+      "learning_rate": 5.372861700451353e-06,
+      "loss": 0.4788,
+      "step": 30216
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.8279675245285034,
+      "learning_rate": 5.372598775883211e-06,
+      "loss": 0.6263,
+      "step": 30217
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.003953218460083,
+      "learning_rate": 5.372335850279011e-06,
+      "loss": 0.7097,
+      "step": 30218
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.1092023849487305,
+      "learning_rate": 5.372072923639485e-06,
+      "loss": 0.6819,
+      "step": 30219
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.6700010299682617,
+      "learning_rate": 5.371809995965365e-06,
+      "loss": 0.6348,
+      "step": 30220
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.0777394771575928,
+      "learning_rate": 5.3715470672573785e-06,
+      "loss": 0.6862,
+      "step": 30221
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.5872514247894287,
+      "learning_rate": 5.371284137516259e-06,
+      "loss": 0.7253,
+      "step": 30222
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.7846163511276245,
+      "learning_rate": 5.3710212067427394e-06,
+      "loss": 0.6088,
+      "step": 30223
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.1482245922088623,
+      "learning_rate": 5.370758274937546e-06,
+      "loss": 0.4796,
+      "step": 30224
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.0045969486236572,
+      "learning_rate": 5.3704953421014135e-06,
+      "loss": 0.4839,
+      "step": 30225
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 4.824685096740723,
+      "learning_rate": 5.370232408235073e-06,
+      "loss": 0.776,
+      "step": 30226
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.6490081548690796,
+      "learning_rate": 5.369969473339256e-06,
+      "loss": 0.4449,
+      "step": 30227
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 4.091923236846924,
+      "learning_rate": 5.36970653741469e-06,
+      "loss": 0.3605,
+      "step": 30228
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.642019033432007,
+      "learning_rate": 5.369443600462111e-06,
+      "loss": 0.6691,
+      "step": 30229
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.8844588994979858,
+      "learning_rate": 5.369180662482246e-06,
+      "loss": 0.6981,
+      "step": 30230
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.797344446182251,
+      "learning_rate": 5.36891772347583e-06,
+      "loss": 0.5294,
+      "step": 30231
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.5141377449035645,
+      "learning_rate": 5.36865478344359e-06,
+      "loss": 0.5735,
+      "step": 30232
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.6305747032165527,
+      "learning_rate": 5.368391842386259e-06,
+      "loss": 0.9005,
+      "step": 30233
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.2146623134613037,
+      "learning_rate": 5.368128900304568e-06,
+      "loss": 0.2474,
+      "step": 30234
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.2112131118774414,
+      "learning_rate": 5.36786595719925e-06,
+      "loss": 0.5771,
+      "step": 30235
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.7952253818511963,
+      "learning_rate": 5.367603013071033e-06,
+      "loss": 0.5787,
+      "step": 30236
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.432619094848633,
+      "learning_rate": 5.367340067920651e-06,
+      "loss": 0.4157,
+      "step": 30237
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.155315637588501,
+      "learning_rate": 5.367077121748834e-06,
+      "loss": 0.5325,
+      "step": 30238
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 4.210883140563965,
+      "learning_rate": 5.366814174556313e-06,
+      "loss": 0.971,
+      "step": 30239
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.549030065536499,
+      "learning_rate": 5.366551226343818e-06,
+      "loss": 0.5783,
+      "step": 30240
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.8233811855316162,
+      "learning_rate": 5.366288277112083e-06,
+      "loss": 0.8038,
+      "step": 30241
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.4022955894470215,
+      "learning_rate": 5.366025326861837e-06,
+      "loss": 0.5649,
+      "step": 30242
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.103994846343994,
+      "learning_rate": 5.365762375593811e-06,
+      "loss": 0.4857,
+      "step": 30243
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.6504251956939697,
+      "learning_rate": 5.365499423308736e-06,
+      "loss": 0.7847,
+      "step": 30244
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 4.0027666091918945,
+      "learning_rate": 5.365236470007345e-06,
+      "loss": 0.5519,
+      "step": 30245
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.361924648284912,
+      "learning_rate": 5.364973515690368e-06,
+      "loss": 0.5211,
+      "step": 30246
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 5.72017240524292,
+      "learning_rate": 5.364710560358537e-06,
+      "loss": 0.6844,
+      "step": 30247
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.9189796447753906,
+      "learning_rate": 5.364447604012583e-06,
+      "loss": 0.8561,
+      "step": 30248
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.76728093624115,
+      "learning_rate": 5.364184646653236e-06,
+      "loss": 0.7221,
+      "step": 30249
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.5328240394592285,
+      "learning_rate": 5.363921688281228e-06,
+      "loss": 0.683,
+      "step": 30250
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 4.621147632598877,
+      "learning_rate": 5.3636587288972885e-06,
+      "loss": 0.6513,
+      "step": 30251
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.3597073554992676,
+      "learning_rate": 5.363395768502152e-06,
+      "loss": 0.6118,
+      "step": 30252
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.0683796405792236,
+      "learning_rate": 5.363132807096549e-06,
+      "loss": 0.8184,
+      "step": 30253
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.4060964584350586,
+      "learning_rate": 5.362869844681208e-06,
+      "loss": 0.6423,
+      "step": 30254
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.24489688873291,
+      "learning_rate": 5.362606881256862e-06,
+      "loss": 0.4738,
+      "step": 30255
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.260126829147339,
+      "learning_rate": 5.362343916824243e-06,
+      "loss": 0.4725,
+      "step": 30256
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.701772689819336,
+      "learning_rate": 5.36208095138408e-06,
+      "loss": 0.7306,
+      "step": 30257
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.497246742248535,
+      "learning_rate": 5.3618179849371065e-06,
+      "loss": 0.4408,
+      "step": 30258
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.129312753677368,
+      "learning_rate": 5.3615550174840525e-06,
+      "loss": 0.8335,
+      "step": 30259
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.8282058238983154,
+      "learning_rate": 5.361292049025651e-06,
+      "loss": 0.8123,
+      "step": 30260
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.311438798904419,
+      "learning_rate": 5.36102907956263e-06,
+      "loss": 0.4473,
+      "step": 30261
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.500471353530884,
+      "learning_rate": 5.360766109095722e-06,
+      "loss": 0.5235,
+      "step": 30262
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.1826372146606445,
+      "learning_rate": 5.360503137625659e-06,
+      "loss": 0.377,
+      "step": 30263
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.2097129821777344,
+      "learning_rate": 5.360240165153172e-06,
+      "loss": 0.6739,
+      "step": 30264
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.1586194038391113,
+      "learning_rate": 5.359977191678991e-06,
+      "loss": 0.6202,
+      "step": 30265
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.328824996948242,
+      "learning_rate": 5.359714217203849e-06,
+      "loss": 0.5183,
+      "step": 30266
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.3579535484313965,
+      "learning_rate": 5.359451241728477e-06,
+      "loss": 0.8338,
+      "step": 30267
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.233778953552246,
+      "learning_rate": 5.359188265253606e-06,
+      "loss": 0.6713,
+      "step": 30268
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.556938886642456,
+      "learning_rate": 5.358925287779967e-06,
+      "loss": 0.6439,
+      "step": 30269
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.381695032119751,
+      "learning_rate": 5.358662309308292e-06,
+      "loss": 0.5722,
+      "step": 30270
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.8365025520324707,
+      "learning_rate": 5.358399329839309e-06,
+      "loss": 0.5949,
+      "step": 30271
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.711435556411743,
+      "learning_rate": 5.358136349373753e-06,
+      "loss": 0.8673,
+      "step": 30272
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.7567334175109863,
+      "learning_rate": 5.357873367912354e-06,
+      "loss": 0.6534,
+      "step": 30273
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.281209707260132,
+      "learning_rate": 5.3576103854558435e-06,
+      "loss": 0.4038,
+      "step": 30274
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.5963208675384521,
+      "learning_rate": 5.357347402004951e-06,
+      "loss": 0.5871,
+      "step": 30275
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.7718850374221802,
+      "learning_rate": 5.357084417560411e-06,
+      "loss": 0.7041,
+      "step": 30276
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.0448594093322754,
+      "learning_rate": 5.356821432122953e-06,
+      "loss": 0.4099,
+      "step": 30277
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.974191665649414,
+      "learning_rate": 5.356558445693307e-06,
+      "loss": 0.4607,
+      "step": 30278
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.8857295513153076,
+      "learning_rate": 5.356295458272207e-06,
+      "loss": 0.6309,
+      "step": 30279
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.8806912899017334,
+      "learning_rate": 5.356032469860381e-06,
+      "loss": 0.747,
+      "step": 30280
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 4.150951385498047,
+      "learning_rate": 5.355769480458564e-06,
+      "loss": 0.4669,
+      "step": 30281
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.84979248046875,
+      "learning_rate": 5.355506490067485e-06,
+      "loss": 0.5164,
+      "step": 30282
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.4952384233474731,
+      "learning_rate": 5.355243498687875e-06,
+      "loss": 0.4344,
+      "step": 30283
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.1858632564544678,
+      "learning_rate": 5.354980506320465e-06,
+      "loss": 0.4507,
+      "step": 30284
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.8154112100601196,
+      "learning_rate": 5.3547175129659875e-06,
+      "loss": 0.5549,
+      "step": 30285
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.8011293411254883,
+      "learning_rate": 5.354454518625175e-06,
+      "loss": 0.5576,
+      "step": 30286
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.3319904804229736,
+      "learning_rate": 5.354191523298756e-06,
+      "loss": 0.6122,
+      "step": 30287
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.4525402784347534,
+      "learning_rate": 5.353928526987464e-06,
+      "loss": 0.5707,
+      "step": 30288
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.5992223024368286,
+      "learning_rate": 5.353665529692028e-06,
+      "loss": 0.5875,
+      "step": 30289
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.670955181121826,
+      "learning_rate": 5.353402531413182e-06,
+      "loss": 0.7199,
+      "step": 30290
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 6.052705764770508,
+      "learning_rate": 5.353139532151656e-06,
+      "loss": 0.9038,
+      "step": 30291
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.7649126052856445,
+      "learning_rate": 5.35287653190818e-06,
+      "loss": 0.6308,
+      "step": 30292
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.0074820518493652,
+      "learning_rate": 5.352613530683488e-06,
+      "loss": 0.4142,
+      "step": 30293
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.9538588523864746,
+      "learning_rate": 5.352350528478309e-06,
+      "loss": 0.6023,
+      "step": 30294
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.7008434534072876,
+      "learning_rate": 5.3520875252933744e-06,
+      "loss": 0.7025,
+      "step": 30295
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.1242570877075195,
+      "learning_rate": 5.351824521129417e-06,
+      "loss": 0.6351,
+      "step": 30296
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.641425371170044,
+      "learning_rate": 5.351561515987168e-06,
+      "loss": 0.7108,
+      "step": 30297
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 4.0049214363098145,
+      "learning_rate": 5.351298509867358e-06,
+      "loss": 0.8896,
+      "step": 30298
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.610130548477173,
+      "learning_rate": 5.351035502770717e-06,
+      "loss": 0.8703,
+      "step": 30299
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.460998296737671,
+      "learning_rate": 5.35077249469798e-06,
+      "loss": 0.6921,
+      "step": 30300
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.566729784011841,
+      "learning_rate": 5.350509485649875e-06,
+      "loss": 0.7332,
+      "step": 30301
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.9774401187896729,
+      "learning_rate": 5.350246475627134e-06,
+      "loss": 0.4421,
+      "step": 30302
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.7394479513168335,
+      "learning_rate": 5.3499834646304896e-06,
+      "loss": 0.6355,
+      "step": 30303
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.5657036304473877,
+      "learning_rate": 5.349720452660672e-06,
+      "loss": 0.5073,
+      "step": 30304
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.6649093627929688,
+      "learning_rate": 5.3494574397184115e-06,
+      "loss": 0.5508,
+      "step": 30305
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.8387508392333984,
+      "learning_rate": 5.349194425804442e-06,
+      "loss": 0.6558,
+      "step": 30306
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.099452257156372,
+      "learning_rate": 5.348931410919493e-06,
+      "loss": 0.7362,
+      "step": 30307
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.6230273246765137,
+      "learning_rate": 5.348668395064298e-06,
+      "loss": 0.5088,
+      "step": 30308
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.5487887859344482,
+      "learning_rate": 5.348405378239586e-06,
+      "loss": 0.7055,
+      "step": 30309
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.5789546966552734,
+      "learning_rate": 5.348142360446089e-06,
+      "loss": 0.745,
+      "step": 30310
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.41195011138916,
+      "learning_rate": 5.3478793416845386e-06,
+      "loss": 0.7922,
+      "step": 30311
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 7.705288410186768,
+      "learning_rate": 5.347616321955666e-06,
+      "loss": 0.6656,
+      "step": 30312
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.548717737197876,
+      "learning_rate": 5.347353301260202e-06,
+      "loss": 0.6994,
+      "step": 30313
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.8658106327056885,
+      "learning_rate": 5.3470902795988786e-06,
+      "loss": 0.3618,
+      "step": 30314
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.7698917388916016,
+      "learning_rate": 5.346827256972428e-06,
+      "loss": 0.6054,
+      "step": 30315
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.9202591180801392,
+      "learning_rate": 5.346564233381581e-06,
+      "loss": 0.4032,
+      "step": 30316
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.503957509994507,
+      "learning_rate": 5.3463012088270685e-06,
+      "loss": 0.5226,
+      "step": 30317
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.7810606956481934,
+      "learning_rate": 5.3460381833096205e-06,
+      "loss": 0.6379,
+      "step": 30318
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.2916054725646973,
+      "learning_rate": 5.345775156829971e-06,
+      "loss": 0.4938,
+      "step": 30319
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.8271028995513916,
+      "learning_rate": 5.34551212938885e-06,
+      "loss": 0.6816,
+      "step": 30320
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.236767530441284,
+      "learning_rate": 5.34524910098699e-06,
+      "loss": 0.8419,
+      "step": 30321
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.9045426845550537,
+      "learning_rate": 5.344986071625123e-06,
+      "loss": 0.8042,
+      "step": 30322
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.312812566757202,
+      "learning_rate": 5.344723041303976e-06,
+      "loss": 0.689,
+      "step": 30323
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.258423328399658,
+      "learning_rate": 5.344460010024285e-06,
+      "loss": 0.5351,
+      "step": 30324
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.051196575164795,
+      "learning_rate": 5.34419697778678e-06,
+      "loss": 0.5955,
+      "step": 30325
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.5603203773498535,
+      "learning_rate": 5.3439339445921905e-06,
+      "loss": 0.8133,
+      "step": 30326
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 10.215462684631348,
+      "learning_rate": 5.34367091044125e-06,
+      "loss": 0.382,
+      "step": 30327
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.7254141569137573,
+      "learning_rate": 5.343407875334689e-06,
+      "loss": 0.8011,
+      "step": 30328
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.4907917976379395,
+      "learning_rate": 5.343144839273241e-06,
+      "loss": 0.5687,
+      "step": 30329
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.0351123809814453,
+      "learning_rate": 5.342881802257636e-06,
+      "loss": 0.7023,
+      "step": 30330
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.237464666366577,
+      "learning_rate": 5.342618764288604e-06,
+      "loss": 0.5368,
+      "step": 30331
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.7860913276672363,
+      "learning_rate": 5.342355725366878e-06,
+      "loss": 0.6589,
+      "step": 30332
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 4.1171464920043945,
+      "learning_rate": 5.342092685493189e-06,
+      "loss": 0.6542,
+      "step": 30333
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.171502113342285,
+      "learning_rate": 5.341829644668268e-06,
+      "loss": 0.6858,
+      "step": 30334
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.0568132400512695,
+      "learning_rate": 5.341566602892847e-06,
+      "loss": 0.7796,
+      "step": 30335
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.201535224914551,
+      "learning_rate": 5.341303560167657e-06,
+      "loss": 0.3672,
+      "step": 30336
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.8761780261993408,
+      "learning_rate": 5.341040516493429e-06,
+      "loss": 0.6012,
+      "step": 30337
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.1302974224090576,
+      "learning_rate": 5.340777471870896e-06,
+      "loss": 0.8095,
+      "step": 30338
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.0559017658233643,
+      "learning_rate": 5.340514426300789e-06,
+      "loss": 0.5112,
+      "step": 30339
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.0974807739257812,
+      "learning_rate": 5.340251379783839e-06,
+      "loss": 0.6514,
+      "step": 30340
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.579080104827881,
+      "learning_rate": 5.339988332320776e-06,
+      "loss": 0.7779,
+      "step": 30341
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 4.545013904571533,
+      "learning_rate": 5.3397252839123335e-06,
+      "loss": 0.686,
+      "step": 30342
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.055158853530884,
+      "learning_rate": 5.339462234559243e-06,
+      "loss": 0.5784,
+      "step": 30343
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.4173734188079834,
+      "learning_rate": 5.339199184262234e-06,
+      "loss": 0.8236,
+      "step": 30344
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.551645040512085,
+      "learning_rate": 5.33893613302204e-06,
+      "loss": 0.5583,
+      "step": 30345
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.758531332015991,
+      "learning_rate": 5.33867308083939e-06,
+      "loss": 0.6215,
+      "step": 30346
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.38996958732605,
+      "learning_rate": 5.338410027715018e-06,
+      "loss": 0.4975,
+      "step": 30347
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.513066530227661,
+      "learning_rate": 5.338146973649655e-06,
+      "loss": 0.4423,
+      "step": 30348
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.7862179279327393,
+      "learning_rate": 5.337883918644033e-06,
+      "loss": 0.6456,
+      "step": 30349
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.0573320388793945,
+      "learning_rate": 5.33762086269888e-06,
+      "loss": 0.521,
+      "step": 30350
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 4.283066272735596,
+      "learning_rate": 5.337357805814931e-06,
+      "loss": 0.5589,
+      "step": 30351
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.630185604095459,
+      "learning_rate": 5.337094747992917e-06,
+      "loss": 0.4869,
+      "step": 30352
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.530410051345825,
+      "learning_rate": 5.336831689233568e-06,
+      "loss": 0.7484,
+      "step": 30353
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.3391485214233398,
+      "learning_rate": 5.336568629537616e-06,
+      "loss": 0.6797,
+      "step": 30354
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.3224380016326904,
+      "learning_rate": 5.336305568905793e-06,
+      "loss": 0.6617,
+      "step": 30355
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.205526828765869,
+      "learning_rate": 5.336042507338829e-06,
+      "loss": 0.6173,
+      "step": 30356
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.8543307781219482,
+      "learning_rate": 5.3357794448374574e-06,
+      "loss": 0.4123,
+      "step": 30357
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.1728742122650146,
+      "learning_rate": 5.33551638140241e-06,
+      "loss": 0.8076,
+      "step": 30358
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.5315117835998535,
+      "learning_rate": 5.335253317034417e-06,
+      "loss": 0.7827,
+      "step": 30359
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.9712449312210083,
+      "learning_rate": 5.33499025173421e-06,
+      "loss": 0.6868,
+      "step": 30360
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.732893943786621,
+      "learning_rate": 5.334727185502521e-06,
+      "loss": 0.6496,
+      "step": 30361
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.1270973682403564,
+      "learning_rate": 5.33446411834008e-06,
+      "loss": 0.6104,
+      "step": 30362
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.3218984603881836,
+      "learning_rate": 5.334201050247621e-06,
+      "loss": 0.4072,
+      "step": 30363
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.675803542137146,
+      "learning_rate": 5.333937981225873e-06,
+      "loss": 0.5169,
+      "step": 30364
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.4407122135162354,
+      "learning_rate": 5.333674911275568e-06,
+      "loss": 0.6528,
+      "step": 30365
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.0769073963165283,
+      "learning_rate": 5.333411840397437e-06,
+      "loss": 0.7866,
+      "step": 30366
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.9411535263061523,
+      "learning_rate": 5.3331487685922155e-06,
+      "loss": 0.5963,
+      "step": 30367
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.2924070358276367,
+      "learning_rate": 5.332885695860631e-06,
+      "loss": 0.7583,
+      "step": 30368
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.639768362045288,
+      "learning_rate": 5.332622622203416e-06,
+      "loss": 0.768,
+      "step": 30369
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.356663703918457,
+      "learning_rate": 5.332359547621302e-06,
+      "loss": 0.8985,
+      "step": 30370
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.070211887359619,
+      "learning_rate": 5.33209647211502e-06,
+      "loss": 0.6087,
+      "step": 30371
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.7339630126953125,
+      "learning_rate": 5.331833395685304e-06,
+      "loss": 0.2713,
+      "step": 30372
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.9291303157806396,
+      "learning_rate": 5.331570318332882e-06,
+      "loss": 0.8099,
+      "step": 30373
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.7814446687698364,
+      "learning_rate": 5.331307240058488e-06,
+      "loss": 0.4439,
+      "step": 30374
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.384739875793457,
+      "learning_rate": 5.331044160862852e-06,
+      "loss": 0.6662,
+      "step": 30375
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.2752490043640137,
+      "learning_rate": 5.330781080746705e-06,
+      "loss": 0.7522,
+      "step": 30376
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.8392893075942993,
+      "learning_rate": 5.330517999710782e-06,
+      "loss": 0.6567,
+      "step": 30377
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.4911861419677734,
+      "learning_rate": 5.330254917755812e-06,
+      "loss": 0.5595,
+      "step": 30378
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.2188429832458496,
+      "learning_rate": 5.329991834882527e-06,
+      "loss": 0.6468,
+      "step": 30379
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.636375904083252,
+      "learning_rate": 5.3297287510916555e-06,
+      "loss": 0.6583,
+      "step": 30380
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.1124513149261475,
+      "learning_rate": 5.329465666383934e-06,
+      "loss": 0.7989,
+      "step": 30381
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.3801090717315674,
+      "learning_rate": 5.329202580760092e-06,
+      "loss": 0.8881,
+      "step": 30382
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.608105182647705,
+      "learning_rate": 5.328939494220861e-06,
+      "loss": 0.5898,
+      "step": 30383
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.5982505083084106,
+      "learning_rate": 5.328676406766972e-06,
+      "loss": 0.7071,
+      "step": 30384
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 4.091827392578125,
+      "learning_rate": 5.328413318399157e-06,
+      "loss": 0.5514,
+      "step": 30385
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.079587697982788,
+      "learning_rate": 5.328150229118148e-06,
+      "loss": 0.5942,
+      "step": 30386
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.3169167041778564,
+      "learning_rate": 5.327887138924675e-06,
+      "loss": 0.4022,
+      "step": 30387
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.2566912174224854,
+      "learning_rate": 5.3276240478194715e-06,
+      "loss": 0.7649,
+      "step": 30388
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 5.552763938903809,
+      "learning_rate": 5.327360955803269e-06,
+      "loss": 0.4428,
+      "step": 30389
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.087977647781372,
+      "learning_rate": 5.327097862876798e-06,
+      "loss": 0.5814,
+      "step": 30390
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.6309025287628174,
+      "learning_rate": 5.32683476904079e-06,
+      "loss": 0.5147,
+      "step": 30391
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.1008291244506836,
+      "learning_rate": 5.326571674295977e-06,
+      "loss": 0.4648,
+      "step": 30392
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.0414326190948486,
+      "learning_rate": 5.326308578643091e-06,
+      "loss": 0.8041,
+      "step": 30393
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.234977960586548,
+      "learning_rate": 5.3260454820828635e-06,
+      "loss": 0.4942,
+      "step": 30394
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.0927939414978027,
+      "learning_rate": 5.325782384616025e-06,
+      "loss": 0.6184,
+      "step": 30395
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.4142115116119385,
+      "learning_rate": 5.325519286243307e-06,
+      "loss": 0.4516,
+      "step": 30396
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.8352792263031006,
+      "learning_rate": 5.325256186965442e-06,
+      "loss": 0.5948,
+      "step": 30397
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 9.38884449005127,
+      "learning_rate": 5.324993086783161e-06,
+      "loss": 0.4536,
+      "step": 30398
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.6248753070831299,
+      "learning_rate": 5.324729985697198e-06,
+      "loss": 0.5473,
+      "step": 30399
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.6432154178619385,
+      "learning_rate": 5.324466883708281e-06,
+      "loss": 0.6616,
+      "step": 30400
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.108640432357788,
+      "learning_rate": 5.324203780817143e-06,
+      "loss": 0.7131,
+      "step": 30401
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.438567876815796,
+      "learning_rate": 5.323940677024517e-06,
+      "loss": 0.7202,
+      "step": 30402
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.136674165725708,
+      "learning_rate": 5.323677572331132e-06,
+      "loss": 0.4734,
+      "step": 30403
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.367152690887451,
+      "learning_rate": 5.323414466737722e-06,
+      "loss": 0.7439,
+      "step": 30404
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.7161648273468018,
+      "learning_rate": 5.323151360245017e-06,
+      "loss": 0.4209,
+      "step": 30405
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.5224359035491943,
+      "learning_rate": 5.3228882528537495e-06,
+      "loss": 0.7202,
+      "step": 30406
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.4824275970458984,
+      "learning_rate": 5.322625144564649e-06,
+      "loss": 0.3872,
+      "step": 30407
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.1105241775512695,
+      "learning_rate": 5.32236203537845e-06,
+      "loss": 0.8891,
+      "step": 30408
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 4.198816776275635,
+      "learning_rate": 5.3220989252958835e-06,
+      "loss": 0.9896,
+      "step": 30409
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 5.216126441955566,
+      "learning_rate": 5.32183581431768e-06,
+      "loss": 0.6799,
+      "step": 30410
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.120378017425537,
+      "learning_rate": 5.3215727024445715e-06,
+      "loss": 0.6091,
+      "step": 30411
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.3972513675689697,
+      "learning_rate": 5.32130958967729e-06,
+      "loss": 0.5568,
+      "step": 30412
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 4.209053039550781,
+      "learning_rate": 5.321046476016568e-06,
+      "loss": 0.479,
+      "step": 30413
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.129408359527588,
+      "learning_rate": 5.320783361463134e-06,
+      "loss": 0.7691,
+      "step": 30414
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.487613320350647,
+      "learning_rate": 5.3205202460177226e-06,
+      "loss": 0.705,
+      "step": 30415
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.6101598739624023,
+      "learning_rate": 5.320257129681066e-06,
+      "loss": 0.6301,
+      "step": 30416
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.5691003799438477,
+      "learning_rate": 5.319994012453891e-06,
+      "loss": 0.6599,
+      "step": 30417
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.315751314163208,
+      "learning_rate": 5.319730894336934e-06,
+      "loss": 0.5447,
+      "step": 30418
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 4.169788360595703,
+      "learning_rate": 5.319467775330927e-06,
+      "loss": 0.5239,
+      "step": 30419
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.362933874130249,
+      "learning_rate": 5.3192046554365975e-06,
+      "loss": 0.7999,
+      "step": 30420
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.4855921268463135,
+      "learning_rate": 5.31894153465468e-06,
+      "loss": 0.611,
+      "step": 30421
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.1417999267578125,
+      "learning_rate": 5.318678412985906e-06,
+      "loss": 0.7578,
+      "step": 30422
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 5.084993839263916,
+      "learning_rate": 5.318415290431007e-06,
+      "loss": 0.6886,
+      "step": 30423
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.6413142681121826,
+      "learning_rate": 5.318152166990713e-06,
+      "loss": 0.6233,
+      "step": 30424
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.18038272857666,
+      "learning_rate": 5.317889042665758e-06,
+      "loss": 0.7075,
+      "step": 30425
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.624605655670166,
+      "learning_rate": 5.317625917456873e-06,
+      "loss": 0.7283,
+      "step": 30426
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.0434486865997314,
+      "learning_rate": 5.317362791364788e-06,
+      "loss": 0.4805,
+      "step": 30427
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.7975016832351685,
+      "learning_rate": 5.317099664390237e-06,
+      "loss": 0.6913,
+      "step": 30428
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.341866970062256,
+      "learning_rate": 5.31683653653395e-06,
+      "loss": 0.7813,
+      "step": 30429
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.86614990234375,
+      "learning_rate": 5.316573407796661e-06,
+      "loss": 0.408,
+      "step": 30430
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.4071035385131836,
+      "learning_rate": 5.316310278179098e-06,
+      "loss": 0.6161,
+      "step": 30431
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.9234548807144165,
+      "learning_rate": 5.316047147681995e-06,
+      "loss": 0.8,
+      "step": 30432
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.827918291091919,
+      "learning_rate": 5.3157840163060834e-06,
+      "loss": 0.5988,
+      "step": 30433
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.5975538492202759,
+      "learning_rate": 5.315520884052095e-06,
+      "loss": 0.5027,
+      "step": 30434
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.2881829738616943,
+      "learning_rate": 5.315257750920761e-06,
+      "loss": 0.7788,
+      "step": 30435
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 4.323757171630859,
+      "learning_rate": 5.314994616912813e-06,
+      "loss": 0.7138,
+      "step": 30436
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.9314152002334595,
+      "learning_rate": 5.314731482028983e-06,
+      "loss": 0.6516,
+      "step": 30437
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.040371894836426,
+      "learning_rate": 5.3144683462700035e-06,
+      "loss": 0.5938,
+      "step": 30438
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.1273622512817383,
+      "learning_rate": 5.314205209636605e-06,
+      "loss": 0.5012,
+      "step": 30439
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.764036178588867,
+      "learning_rate": 5.313942072129518e-06,
+      "loss": 0.6934,
+      "step": 30440
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 4.8951826095581055,
+      "learning_rate": 5.313678933749478e-06,
+      "loss": 0.5761,
+      "step": 30441
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.1204166412353516,
+      "learning_rate": 5.313415794497212e-06,
+      "loss": 0.7564,
+      "step": 30442
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.078514814376831,
+      "learning_rate": 5.313152654373456e-06,
+      "loss": 0.7013,
+      "step": 30443
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.5281929969787598,
+      "learning_rate": 5.31288951337894e-06,
+      "loss": 0.739,
+      "step": 30444
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.9241726398468018,
+      "learning_rate": 5.3126263715143944e-06,
+      "loss": 0.5797,
+      "step": 30445
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.7810717821121216,
+      "learning_rate": 5.312363228780553e-06,
+      "loss": 0.4785,
+      "step": 30446
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.2343814373016357,
+      "learning_rate": 5.312100085178144e-06,
+      "loss": 0.559,
+      "step": 30447
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.1605613231658936,
+      "learning_rate": 5.311836940707904e-06,
+      "loss": 0.5215,
+      "step": 30448
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.7494940757751465,
+      "learning_rate": 5.311573795370562e-06,
+      "loss": 0.6687,
+      "step": 30449
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.144195318222046,
+      "learning_rate": 5.311310649166849e-06,
+      "loss": 0.6362,
+      "step": 30450
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 4.93067741394043,
+      "learning_rate": 5.311047502097498e-06,
+      "loss": 0.814,
+      "step": 30451
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.8201910257339478,
+      "learning_rate": 5.310784354163241e-06,
+      "loss": 0.5881,
+      "step": 30452
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.1778957843780518,
+      "learning_rate": 5.3105212053648094e-06,
+      "loss": 0.7933,
+      "step": 30453
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.0098304748535156,
+      "learning_rate": 5.310258055702934e-06,
+      "loss": 0.5011,
+      "step": 30454
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.0570411682128906,
+      "learning_rate": 5.3099949051783475e-06,
+      "loss": 0.5727,
+      "step": 30455
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.285346031188965,
+      "learning_rate": 5.3097317537917815e-06,
+      "loss": 0.777,
+      "step": 30456
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 5.206923007965088,
+      "learning_rate": 5.309468601543967e-06,
+      "loss": 0.8618,
+      "step": 30457
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.269190549850464,
+      "learning_rate": 5.309205448435636e-06,
+      "loss": 0.5654,
+      "step": 30458
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.0374577045440674,
+      "learning_rate": 5.30894229446752e-06,
+      "loss": 0.6455,
+      "step": 30459
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.9070420265197754,
+      "learning_rate": 5.308679139640352e-06,
+      "loss": 0.4743,
+      "step": 30460
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.861581802368164,
+      "learning_rate": 5.3084159839548636e-06,
+      "loss": 0.7322,
+      "step": 30461
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.754317283630371,
+      "learning_rate": 5.3081528274117854e-06,
+      "loss": 0.6143,
+      "step": 30462
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.0916616916656494,
+      "learning_rate": 5.30788967001185e-06,
+      "loss": 0.4754,
+      "step": 30463
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.1328635215759277,
+      "learning_rate": 5.307626511755788e-06,
+      "loss": 0.7303,
+      "step": 30464
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.9792839288711548,
+      "learning_rate": 5.307363352644332e-06,
+      "loss": 0.6464,
+      "step": 30465
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.894467830657959,
+      "learning_rate": 5.307100192678214e-06,
+      "loss": 0.7888,
+      "step": 30466
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.3265480995178223,
+      "learning_rate": 5.306837031858166e-06,
+      "loss": 0.6867,
+      "step": 30467
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.5750913619995117,
+      "learning_rate": 5.306573870184917e-06,
+      "loss": 0.5322,
+      "step": 30468
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.114504098892212,
+      "learning_rate": 5.306310707659202e-06,
+      "loss": 0.945,
+      "step": 30469
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.6510413885116577,
+      "learning_rate": 5.306047544281753e-06,
+      "loss": 0.7338,
+      "step": 30470
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.65324330329895,
+      "learning_rate": 5.305784380053299e-06,
+      "loss": 0.5513,
+      "step": 30471
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.624269962310791,
+      "learning_rate": 5.305521214974574e-06,
+      "loss": 0.7099,
+      "step": 30472
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.277663230895996,
+      "learning_rate": 5.3052580490463094e-06,
+      "loss": 0.8042,
+      "step": 30473
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.8893444538116455,
+      "learning_rate": 5.304994882269235e-06,
+      "loss": 0.6133,
+      "step": 30474
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.197711706161499,
+      "learning_rate": 5.3047317146440846e-06,
+      "loss": 0.3835,
+      "step": 30475
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.6324191093444824,
+      "learning_rate": 5.304468546171591e-06,
+      "loss": 0.6031,
+      "step": 30476
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.8986167907714844,
+      "learning_rate": 5.304205376852483e-06,
+      "loss": 0.4933,
+      "step": 30477
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.795170307159424,
+      "learning_rate": 5.303942206687493e-06,
+      "loss": 0.9091,
+      "step": 30478
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.950867176055908,
+      "learning_rate": 5.303679035677354e-06,
+      "loss": 0.8008,
+      "step": 30479
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.922558307647705,
+      "learning_rate": 5.303415863822797e-06,
+      "loss": 0.6807,
+      "step": 30480
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.3738725185394287,
+      "learning_rate": 5.303152691124555e-06,
+      "loss": 0.4561,
+      "step": 30481
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.7109558582305908,
+      "learning_rate": 5.30288951758336e-06,
+      "loss": 0.6025,
+      "step": 30482
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.5862864255905151,
+      "learning_rate": 5.3026263431999415e-06,
+      "loss": 0.48,
+      "step": 30483
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.1045591831207275,
+      "learning_rate": 5.302363167975033e-06,
+      "loss": 0.5456,
+      "step": 30484
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.164041042327881,
+      "learning_rate": 5.302099991909365e-06,
+      "loss": 0.6226,
+      "step": 30485
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.316758155822754,
+      "learning_rate": 5.301836815003671e-06,
+      "loss": 0.7099,
+      "step": 30486
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.716381788253784,
+      "learning_rate": 5.3015736372586815e-06,
+      "loss": 0.684,
+      "step": 30487
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.9136109352111816,
+      "learning_rate": 5.301310458675128e-06,
+      "loss": 0.8214,
+      "step": 30488
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 5.773531436920166,
+      "learning_rate": 5.301047279253743e-06,
+      "loss": 0.4167,
+      "step": 30489
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.025926113128662,
+      "learning_rate": 5.3007840989952585e-06,
+      "loss": 0.6561,
+      "step": 30490
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.631821393966675,
+      "learning_rate": 5.300520917900407e-06,
+      "loss": 0.4278,
+      "step": 30491
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.2105422019958496,
+      "learning_rate": 5.300257735969919e-06,
+      "loss": 0.6828,
+      "step": 30492
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.102849245071411,
+      "learning_rate": 5.299994553204527e-06,
+      "loss": 0.6662,
+      "step": 30493
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.71506929397583,
+      "learning_rate": 5.299731369604962e-06,
+      "loss": 0.5787,
+      "step": 30494
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.26184344291687,
+      "learning_rate": 5.299468185171956e-06,
+      "loss": 0.7687,
+      "step": 30495
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.1974518299102783,
+      "learning_rate": 5.299204999906242e-06,
+      "loss": 0.4034,
+      "step": 30496
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.2723915576934814,
+      "learning_rate": 5.2989418138085516e-06,
+      "loss": 0.5466,
+      "step": 30497
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.569446086883545,
+      "learning_rate": 5.298678626879613e-06,
+      "loss": 0.6013,
+      "step": 30498
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.7195117473602295,
+      "learning_rate": 5.298415439120164e-06,
+      "loss": 0.6973,
+      "step": 30499
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.740004301071167,
+      "learning_rate": 5.298152250530933e-06,
+      "loss": 0.7463,
+      "step": 30500
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 4.120452404022217,
+      "learning_rate": 5.297889061112651e-06,
+      "loss": 0.5871,
+      "step": 30501
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.501749038696289,
+      "learning_rate": 5.297625870866052e-06,
+      "loss": 0.6868,
+      "step": 30502
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.565042495727539,
+      "learning_rate": 5.297362679791867e-06,
+      "loss": 0.729,
+      "step": 30503
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.589107036590576,
+      "learning_rate": 5.297099487890826e-06,
+      "loss": 0.6041,
+      "step": 30504
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.501946449279785,
+      "learning_rate": 5.2968362951636644e-06,
+      "loss": 0.7777,
+      "step": 30505
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.1095755100250244,
+      "learning_rate": 5.296573101611112e-06,
+      "loss": 0.6034,
+      "step": 30506
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.026729106903076,
+      "learning_rate": 5.296309907233902e-06,
+      "loss": 0.5449,
+      "step": 30507
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.8346285820007324,
+      "learning_rate": 5.296046712032764e-06,
+      "loss": 0.7211,
+      "step": 30508
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.7991485595703125,
+      "learning_rate": 5.2957835160084315e-06,
+      "loss": 0.7154,
+      "step": 30509
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.784992218017578,
+      "learning_rate": 5.295520319161634e-06,
+      "loss": 0.4947,
+      "step": 30510
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.2629218101501465,
+      "learning_rate": 5.295257121493107e-06,
+      "loss": 0.47,
+      "step": 30511
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.231818199157715,
+      "learning_rate": 5.29499392300358e-06,
+      "loss": 0.4449,
+      "step": 30512
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.4285480976104736,
+      "learning_rate": 5.294730723693786e-06,
+      "loss": 0.6472,
+      "step": 30513
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.8202083110809326,
+      "learning_rate": 5.294467523564456e-06,
+      "loss": 0.5455,
+      "step": 30514
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.3557796478271484,
+      "learning_rate": 5.294204322616322e-06,
+      "loss": 0.8249,
+      "step": 30515
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.460330843925476,
+      "learning_rate": 5.293941120850116e-06,
+      "loss": 0.4232,
+      "step": 30516
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.9690775871276855,
+      "learning_rate": 5.29367791826657e-06,
+      "loss": 0.7487,
+      "step": 30517
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.6126813888549805,
+      "learning_rate": 5.293414714866416e-06,
+      "loss": 0.4444,
+      "step": 30518
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.3396224975585938,
+      "learning_rate": 5.293151510650384e-06,
+      "loss": 0.6767,
+      "step": 30519
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.4528310298919678,
+      "learning_rate": 5.292888305619208e-06,
+      "loss": 0.4267,
+      "step": 30520
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.283517837524414,
+      "learning_rate": 5.29262509977362e-06,
+      "loss": 0.6489,
+      "step": 30521
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.5801475048065186,
+      "learning_rate": 5.292361893114352e-06,
+      "loss": 0.6785,
+      "step": 30522
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.582248330116272,
+      "learning_rate": 5.292098685642134e-06,
+      "loss": 0.5607,
+      "step": 30523
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.9950076341629028,
+      "learning_rate": 5.2918354773577e-06,
+      "loss": 0.3306,
+      "step": 30524
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.4901387691497803,
+      "learning_rate": 5.29157226826178e-06,
+      "loss": 0.6388,
+      "step": 30525
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.8188421726226807,
+      "learning_rate": 5.291309058355108e-06,
+      "loss": 0.6821,
+      "step": 30526
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 4.639069080352783,
+      "learning_rate": 5.291045847638413e-06,
+      "loss": 0.82,
+      "step": 30527
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.5812947750091553,
+      "learning_rate": 5.2907826361124296e-06,
+      "loss": 0.7783,
+      "step": 30528
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.237053632736206,
+      "learning_rate": 5.2905194237778865e-06,
+      "loss": 0.442,
+      "step": 30529
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.608300805091858,
+      "learning_rate": 5.29025621063552e-06,
+      "loss": 0.6403,
+      "step": 30530
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.96071195602417,
+      "learning_rate": 5.289992996686058e-06,
+      "loss": 0.5208,
+      "step": 30531
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 4.6734771728515625,
+      "learning_rate": 5.289729781930235e-06,
+      "loss": 0.6351,
+      "step": 30532
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.9461334943771362,
+      "learning_rate": 5.289466566368783e-06,
+      "loss": 0.7501,
+      "step": 30533
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.376089096069336,
+      "learning_rate": 5.289203350002432e-06,
+      "loss": 0.6596,
+      "step": 30534
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.7794904708862305,
+      "learning_rate": 5.288940132831915e-06,
+      "loss": 0.6245,
+      "step": 30535
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.3887038230895996,
+      "learning_rate": 5.288676914857963e-06,
+      "loss": 0.8166,
+      "step": 30536
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.8755335807800293,
+      "learning_rate": 5.288413696081309e-06,
+      "loss": 0.5988,
+      "step": 30537
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.599212408065796,
+      "learning_rate": 5.288150476502686e-06,
+      "loss": 0.6689,
+      "step": 30538
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.4642210006713867,
+      "learning_rate": 5.2878872561228225e-06,
+      "loss": 0.7368,
+      "step": 30539
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.376589059829712,
+      "learning_rate": 5.287624034942453e-06,
+      "loss": 0.7314,
+      "step": 30540
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.5094974040985107,
+      "learning_rate": 5.287360812962309e-06,
+      "loss": 0.6946,
+      "step": 30541
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.063014507293701,
+      "learning_rate": 5.287097590183122e-06,
+      "loss": 0.5461,
+      "step": 30542
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.670631170272827,
+      "learning_rate": 5.286834366605625e-06,
+      "loss": 0.6558,
+      "step": 30543
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.972710371017456,
+      "learning_rate": 5.286571142230548e-06,
+      "loss": 0.7343,
+      "step": 30544
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.6207895278930664,
+      "learning_rate": 5.286307917058624e-06,
+      "loss": 0.4693,
+      "step": 30545
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.9886364936828613,
+      "learning_rate": 5.286044691090586e-06,
+      "loss": 0.6353,
+      "step": 30546
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.147343873977661,
+      "learning_rate": 5.285781464327165e-06,
+      "loss": 0.8001,
+      "step": 30547
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.6849467754364014,
+      "learning_rate": 5.285518236769092e-06,
+      "loss": 0.5205,
+      "step": 30548
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.4498064517974854,
+      "learning_rate": 5.285255008417098e-06,
+      "loss": 0.6478,
+      "step": 30549
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.1938445568084717,
+      "learning_rate": 5.284991779271919e-06,
+      "loss": 0.6496,
+      "step": 30550
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.170794725418091,
+      "learning_rate": 5.284728549334284e-06,
+      "loss": 0.6748,
+      "step": 30551
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.230086326599121,
+      "learning_rate": 5.284465318604925e-06,
+      "loss": 0.9581,
+      "step": 30552
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.14642596244812,
+      "learning_rate": 5.284202087084575e-06,
+      "loss": 0.8847,
+      "step": 30553
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.489161968231201,
+      "learning_rate": 5.283938854773967e-06,
+      "loss": 0.4343,
+      "step": 30554
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 4.1993889808654785,
+      "learning_rate": 5.283675621673829e-06,
+      "loss": 0.9491,
+      "step": 30555
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.0438809394836426,
+      "learning_rate": 5.2834123877848966e-06,
+      "loss": 0.6535,
+      "step": 30556
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 5.322375774383545,
+      "learning_rate": 5.2831491531079e-06,
+      "loss": 0.8306,
+      "step": 30557
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.1643850803375244,
+      "learning_rate": 5.282885917643572e-06,
+      "loss": 0.7489,
+      "step": 30558
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 5.783599853515625,
+      "learning_rate": 5.282622681392644e-06,
+      "loss": 0.8824,
+      "step": 30559
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 4.67957067489624,
+      "learning_rate": 5.282359444355848e-06,
+      "loss": 0.8776,
+      "step": 30560
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.8441379070281982,
+      "learning_rate": 5.282096206533917e-06,
+      "loss": 0.5664,
+      "step": 30561
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 4.115435600280762,
+      "learning_rate": 5.281832967927581e-06,
+      "loss": 0.6205,
+      "step": 30562
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.0609922409057617,
+      "learning_rate": 5.281569728537574e-06,
+      "loss": 0.7972,
+      "step": 30563
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.795928716659546,
+      "learning_rate": 5.281306488364627e-06,
+      "loss": 0.7528,
+      "step": 30564
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.6717941761016846,
+      "learning_rate": 5.281043247409472e-06,
+      "loss": 0.61,
+      "step": 30565
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.9699499607086182,
+      "learning_rate": 5.2807800056728395e-06,
+      "loss": 0.6919,
+      "step": 30566
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.7698019742965698,
+      "learning_rate": 5.280516763155466e-06,
+      "loss": 0.8097,
+      "step": 30567
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.977351427078247,
+      "learning_rate": 5.280253519858079e-06,
+      "loss": 0.7572,
+      "step": 30568
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 4.437828063964844,
+      "learning_rate": 5.2799902757814114e-06,
+      "loss": 0.7095,
+      "step": 30569
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.8033552169799805,
+      "learning_rate": 5.279727030926196e-06,
+      "loss": 0.5884,
+      "step": 30570
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.320721387863159,
+      "learning_rate": 5.279463785293163e-06,
+      "loss": 0.6408,
+      "step": 30571
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 5.735476493835449,
+      "learning_rate": 5.279200538883048e-06,
+      "loss": 0.5355,
+      "step": 30572
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.773111343383789,
+      "learning_rate": 5.278937291696581e-06,
+      "loss": 0.7786,
+      "step": 30573
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.850247621536255,
+      "learning_rate": 5.278674043734493e-06,
+      "loss": 0.7914,
+      "step": 30574
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.6841779947280884,
+      "learning_rate": 5.278410794997517e-06,
+      "loss": 0.5318,
+      "step": 30575
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.474273681640625,
+      "learning_rate": 5.278147545486385e-06,
+      "loss": 0.5261,
+      "step": 30576
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.452528953552246,
+      "learning_rate": 5.2778842952018295e-06,
+      "loss": 0.4855,
+      "step": 30577
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.9735488891601562,
+      "learning_rate": 5.2776210441445815e-06,
+      "loss": 0.7076,
+      "step": 30578
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.978520154953003,
+      "learning_rate": 5.277357792315373e-06,
+      "loss": 0.6923,
+      "step": 30579
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.8692071437835693,
+      "learning_rate": 5.277094539714935e-06,
+      "loss": 0.7715,
+      "step": 30580
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.2269678115844727,
+      "learning_rate": 5.276831286344003e-06,
+      "loss": 0.6526,
+      "step": 30581
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.1713454723358154,
+      "learning_rate": 5.276568032203306e-06,
+      "loss": 0.9459,
+      "step": 30582
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.119145631790161,
+      "learning_rate": 5.276304777293578e-06,
+      "loss": 0.4371,
+      "step": 30583
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.4053443670272827,
+      "learning_rate": 5.276041521615549e-06,
+      "loss": 0.6943,
+      "step": 30584
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.178879499435425,
+      "learning_rate": 5.275778265169953e-06,
+      "loss": 0.8525,
+      "step": 30585
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 4.662625789642334,
+      "learning_rate": 5.275515007957519e-06,
+      "loss": 0.8069,
+      "step": 30586
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.1961512565612793,
+      "learning_rate": 5.275251749978982e-06,
+      "loss": 0.4648,
+      "step": 30587
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.2313103675842285,
+      "learning_rate": 5.274988491235074e-06,
+      "loss": 0.5529,
+      "step": 30588
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.4755349159240723,
+      "learning_rate": 5.274725231726525e-06,
+      "loss": 0.7139,
+      "step": 30589
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.2923359870910645,
+      "learning_rate": 5.274461971454067e-06,
+      "loss": 0.4605,
+      "step": 30590
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.3500523567199707,
+      "learning_rate": 5.274198710418434e-06,
+      "loss": 0.7146,
+      "step": 30591
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.667910575866699,
+      "learning_rate": 5.273935448620358e-06,
+      "loss": 0.6392,
+      "step": 30592
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.0709481239318848,
+      "learning_rate": 5.2736721860605685e-06,
+      "loss": 0.7588,
+      "step": 30593
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.1555824279785156,
+      "learning_rate": 5.273408922739801e-06,
+      "loss": 0.6129,
+      "step": 30594
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.3608458042144775,
+      "learning_rate": 5.273145658658785e-06,
+      "loss": 0.566,
+      "step": 30595
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.0108561515808105,
+      "learning_rate": 5.272882393818252e-06,
+      "loss": 0.5621,
+      "step": 30596
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.722683906555176,
+      "learning_rate": 5.272619128218938e-06,
+      "loss": 0.7703,
+      "step": 30597
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.569708824157715,
+      "learning_rate": 5.27235586186157e-06,
+      "loss": 0.7263,
+      "step": 30598
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.6771317720413208,
+      "learning_rate": 5.272092594746882e-06,
+      "loss": 0.5713,
+      "step": 30599
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 5.1079583168029785,
+      "learning_rate": 5.271829326875607e-06,
+      "loss": 0.706,
+      "step": 30600
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.447605609893799,
+      "learning_rate": 5.271566058248477e-06,
+      "loss": 0.4203,
+      "step": 30601
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 7.036708354949951,
+      "learning_rate": 5.2713027888662235e-06,
+      "loss": 0.5132,
+      "step": 30602
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.5709503889083862,
+      "learning_rate": 5.2710395187295796e-06,
+      "loss": 0.5248,
+      "step": 30603
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.2785422801971436,
+      "learning_rate": 5.2707762478392745e-06,
+      "loss": 0.6293,
+      "step": 30604
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.6402932405471802,
+      "learning_rate": 5.270512976196043e-06,
+      "loss": 0.4979,
+      "step": 30605
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.118140697479248,
+      "learning_rate": 5.270249703800616e-06,
+      "loss": 0.4769,
+      "step": 30606
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.556222677230835,
+      "learning_rate": 5.269986430653726e-06,
+      "loss": 0.4975,
+      "step": 30607
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.932743787765503,
+      "learning_rate": 5.269723156756106e-06,
+      "loss": 0.6575,
+      "step": 30608
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.1729416847229004,
+      "learning_rate": 5.2694598821084855e-06,
+      "loss": 0.5955,
+      "step": 30609
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.501551628112793,
+      "learning_rate": 5.269196606711597e-06,
+      "loss": 0.5165,
+      "step": 30610
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.5077109336853027,
+      "learning_rate": 5.268933330566175e-06,
+      "loss": 0.8215,
+      "step": 30611
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.3134684562683105,
+      "learning_rate": 5.26867005367295e-06,
+      "loss": 0.5801,
+      "step": 30612
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.69042706489563,
+      "learning_rate": 5.268406776032654e-06,
+      "loss": 0.6237,
+      "step": 30613
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.8289551734924316,
+      "learning_rate": 5.26814349764602e-06,
+      "loss": 0.3941,
+      "step": 30614
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.19706654548645,
+      "learning_rate": 5.267880218513779e-06,
+      "loss": 0.7849,
+      "step": 30615
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.8688167333602905,
+      "learning_rate": 5.267616938636663e-06,
+      "loss": 0.6684,
+      "step": 30616
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.6544740200042725,
+      "learning_rate": 5.267353658015405e-06,
+      "loss": 0.7639,
+      "step": 30617
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.228983163833618,
+      "learning_rate": 5.267090376650736e-06,
+      "loss": 0.696,
+      "step": 30618
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.4447181224823,
+      "learning_rate": 5.26682709454339e-06,
+      "loss": 0.5672,
+      "step": 30619
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.7072157859802246,
+      "learning_rate": 5.266563811694095e-06,
+      "loss": 0.706,
+      "step": 30620
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.8311299085617065,
+      "learning_rate": 5.266300528103588e-06,
+      "loss": 0.7507,
+      "step": 30621
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.9810870885849,
+      "learning_rate": 5.2660372437726005e-06,
+      "loss": 0.5154,
+      "step": 30622
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.6097328662872314,
+      "learning_rate": 5.265773958701861e-06,
+      "loss": 0.6401,
+      "step": 30623
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.7497310638427734,
+      "learning_rate": 5.265510672892105e-06,
+      "loss": 0.5947,
+      "step": 30624
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.602438449859619,
+      "learning_rate": 5.265247386344062e-06,
+      "loss": 0.5796,
+      "step": 30625
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.8604915142059326,
+      "learning_rate": 5.264984099058465e-06,
+      "loss": 0.7422,
+      "step": 30626
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.844841718673706,
+      "learning_rate": 5.264720811036047e-06,
+      "loss": 0.5835,
+      "step": 30627
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 4.229579448699951,
+      "learning_rate": 5.26445752227754e-06,
+      "loss": 0.6911,
+      "step": 30628
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.9977980852127075,
+      "learning_rate": 5.264194232783676e-06,
+      "loss": 0.801,
+      "step": 30629
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.626864194869995,
+      "learning_rate": 5.263930942555186e-06,
+      "loss": 0.626,
+      "step": 30630
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.919306755065918,
+      "learning_rate": 5.263667651592804e-06,
+      "loss": 0.6129,
+      "step": 30631
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.404662609100342,
+      "learning_rate": 5.263404359897261e-06,
+      "loss": 0.8815,
+      "step": 30632
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.9155635833740234,
+      "learning_rate": 5.263141067469286e-06,
+      "loss": 0.5084,
+      "step": 30633
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.1011316776275635,
+      "learning_rate": 5.262877774309617e-06,
+      "loss": 0.7282,
+      "step": 30634
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.9445743560791016,
+      "learning_rate": 5.262614480418983e-06,
+      "loss": 0.7633,
+      "step": 30635
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.3842439651489258,
+      "learning_rate": 5.262351185798118e-06,
+      "loss": 0.6774,
+      "step": 30636
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.307563066482544,
+      "learning_rate": 5.262087890447752e-06,
+      "loss": 0.9685,
+      "step": 30637
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.860063076019287,
+      "learning_rate": 5.261824594368616e-06,
+      "loss": 0.5599,
+      "step": 30638
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.3757376670837402,
+      "learning_rate": 5.261561297561445e-06,
+      "loss": 0.8153,
+      "step": 30639
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.4398751258850098,
+      "learning_rate": 5.261298000026971e-06,
+      "loss": 0.6214,
+      "step": 30640
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.8005499839782715,
+      "learning_rate": 5.2610347017659224e-06,
+      "loss": 0.5063,
+      "step": 30641
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.838937282562256,
+      "learning_rate": 5.260771402779036e-06,
+      "loss": 0.3843,
+      "step": 30642
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.258816719055176,
+      "learning_rate": 5.260508103067043e-06,
+      "loss": 0.4995,
+      "step": 30643
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.583514451980591,
+      "learning_rate": 5.260244802630672e-06,
+      "loss": 0.6574,
+      "step": 30644
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.669785261154175,
+      "learning_rate": 5.259981501470659e-06,
+      "loss": 0.6294,
+      "step": 30645
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.2605769634246826,
+      "learning_rate": 5.259718199587735e-06,
+      "loss": 0.4483,
+      "step": 30646
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.318655014038086,
+      "learning_rate": 5.259454896982632e-06,
+      "loss": 0.648,
+      "step": 30647
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.017850637435913,
+      "learning_rate": 5.259191593656082e-06,
+      "loss": 0.5474,
+      "step": 30648
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.043820858001709,
+      "learning_rate": 5.258928289608818e-06,
+      "loss": 0.4279,
+      "step": 30649
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.638169050216675,
+      "learning_rate": 5.258664984841571e-06,
+      "loss": 0.6813,
+      "step": 30650
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.3719706535339355,
+      "learning_rate": 5.258401679355072e-06,
+      "loss": 0.7433,
+      "step": 30651
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.439863920211792,
+      "learning_rate": 5.258138373150056e-06,
+      "loss": 0.6505,
+      "step": 30652
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 5.0567121505737305,
+      "learning_rate": 5.257875066227254e-06,
+      "loss": 1.0211,
+      "step": 30653
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.590087413787842,
+      "learning_rate": 5.257611758587398e-06,
+      "loss": 0.7527,
+      "step": 30654
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.005587100982666,
+      "learning_rate": 5.25734845023122e-06,
+      "loss": 0.5701,
+      "step": 30655
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 4.31454610824585,
+      "learning_rate": 5.2570851411594525e-06,
+      "loss": 0.6811,
+      "step": 30656
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.0779075622558594,
+      "learning_rate": 5.256821831372828e-06,
+      "loss": 0.5829,
+      "step": 30657
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 4.823254108428955,
+      "learning_rate": 5.256558520872078e-06,
+      "loss": 0.7742,
+      "step": 30658
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.6546043157577515,
+      "learning_rate": 5.256295209657934e-06,
+      "loss": 0.5259,
+      "step": 30659
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.6674156188964844,
+      "learning_rate": 5.25603189773113e-06,
+      "loss": 0.7355,
+      "step": 30660
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.8976120948791504,
+      "learning_rate": 5.255768585092396e-06,
+      "loss": 0.6866,
+      "step": 30661
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.219501495361328,
+      "learning_rate": 5.255505271742466e-06,
+      "loss": 0.7744,
+      "step": 30662
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.321579933166504,
+      "learning_rate": 5.255241957682072e-06,
+      "loss": 0.4859,
+      "step": 30663
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.2404356002807617,
+      "learning_rate": 5.254978642911945e-06,
+      "loss": 0.8402,
+      "step": 30664
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.9252580404281616,
+      "learning_rate": 5.254715327432819e-06,
+      "loss": 0.3723,
+      "step": 30665
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.670328140258789,
+      "learning_rate": 5.254452011245424e-06,
+      "loss": 0.8146,
+      "step": 30666
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.0057590007781982,
+      "learning_rate": 5.254188694350495e-06,
+      "loss": 0.625,
+      "step": 30667
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.3147544860839844,
+      "learning_rate": 5.2539253767487605e-06,
+      "loss": 0.4535,
+      "step": 30668
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.4245779514312744,
+      "learning_rate": 5.253662058440956e-06,
+      "loss": 0.6845,
+      "step": 30669
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.5922917127609253,
+      "learning_rate": 5.253398739427812e-06,
+      "loss": 0.5259,
+      "step": 30670
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.0211355686187744,
+      "learning_rate": 5.25313541971006e-06,
+      "loss": 0.5731,
+      "step": 30671
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.4738801717758179,
+      "learning_rate": 5.252872099288434e-06,
+      "loss": 0.4584,
+      "step": 30672
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.3957605361938477,
+      "learning_rate": 5.252608778163665e-06,
+      "loss": 0.6332,
+      "step": 30673
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.1982407569885254,
+      "learning_rate": 5.252345456336486e-06,
+      "loss": 0.7654,
+      "step": 30674
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.251525640487671,
+      "learning_rate": 5.2520821338076285e-06,
+      "loss": 0.7653,
+      "step": 30675
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.082721710205078,
+      "learning_rate": 5.251818810577827e-06,
+      "loss": 0.869,
+      "step": 30676
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.356065034866333,
+      "learning_rate": 5.2515554866478095e-06,
+      "loss": 0.6889,
+      "step": 30677
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.2229650020599365,
+      "learning_rate": 5.251292162018311e-06,
+      "loss": 0.8705,
+      "step": 30678
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.1493101119995117,
+      "learning_rate": 5.251028836690063e-06,
+      "loss": 0.5212,
+      "step": 30679
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.4948673248291016,
+      "learning_rate": 5.250765510663798e-06,
+      "loss": 0.6898,
+      "step": 30680
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.9829535484313965,
+      "learning_rate": 5.250502183940247e-06,
+      "loss": 0.7743,
+      "step": 30681
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.186673641204834,
+      "learning_rate": 5.250238856520145e-06,
+      "loss": 0.6951,
+      "step": 30682
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.6346709728240967,
+      "learning_rate": 5.249975528404222e-06,
+      "loss": 0.5323,
+      "step": 30683
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.631568431854248,
+      "learning_rate": 5.24971219959321e-06,
+      "loss": 0.5524,
+      "step": 30684
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 4.125629425048828,
+      "learning_rate": 5.249448870087842e-06,
+      "loss": 0.6765,
+      "step": 30685
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.7007126808166504,
+      "learning_rate": 5.249185539888852e-06,
+      "loss": 0.6401,
+      "step": 30686
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.4873850345611572,
+      "learning_rate": 5.248922208996968e-06,
+      "loss": 0.5546,
+      "step": 30687
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.719780445098877,
+      "learning_rate": 5.248658877412925e-06,
+      "loss": 0.5181,
+      "step": 30688
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.8089120388031006,
+      "learning_rate": 5.248395545137455e-06,
+      "loss": 0.7289,
+      "step": 30689
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.723620891571045,
+      "learning_rate": 5.2481322121712906e-06,
+      "loss": 0.7117,
+      "step": 30690
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.082031011581421,
+      "learning_rate": 5.247868878515163e-06,
+      "loss": 0.6411,
+      "step": 30691
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.1293842792510986,
+      "learning_rate": 5.247605544169805e-06,
+      "loss": 0.5087,
+      "step": 30692
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.98309063911438,
+      "learning_rate": 5.2473422091359495e-06,
+      "loss": 0.748,
+      "step": 30693
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 5.823720932006836,
+      "learning_rate": 5.247078873414327e-06,
+      "loss": 0.599,
+      "step": 30694
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.181685447692871,
+      "learning_rate": 5.24681553700567e-06,
+      "loss": 0.7009,
+      "step": 30695
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 4.075458526611328,
+      "learning_rate": 5.246552199910713e-06,
+      "loss": 0.7255,
+      "step": 30696
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.1999192237854004,
+      "learning_rate": 5.246288862130185e-06,
+      "loss": 0.5544,
+      "step": 30697
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.7700681686401367,
+      "learning_rate": 5.2460255236648215e-06,
+      "loss": 0.6282,
+      "step": 30698
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.1697590351104736,
+      "learning_rate": 5.245762184515354e-06,
+      "loss": 0.7487,
+      "step": 30699
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.6010277271270752,
+      "learning_rate": 5.2454988446825125e-06,
+      "loss": 0.6568,
+      "step": 30700
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.1166610717773438,
+      "learning_rate": 5.245235504167031e-06,
+      "loss": 0.8249,
+      "step": 30701
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.1302242279052734,
+      "learning_rate": 5.24497216296964e-06,
+      "loss": 0.526,
+      "step": 30702
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 4.204526901245117,
+      "learning_rate": 5.244708821091074e-06,
+      "loss": 0.8401,
+      "step": 30703
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.601208448410034,
+      "learning_rate": 5.244445478532066e-06,
+      "loss": 0.6237,
+      "step": 30704
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.910804271697998,
+      "learning_rate": 5.244182135293345e-06,
+      "loss": 0.6236,
+      "step": 30705
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.2521965503692627,
+      "learning_rate": 5.243918791375646e-06,
+      "loss": 0.6718,
+      "step": 30706
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.8517329692840576,
+      "learning_rate": 5.2436554467797e-06,
+      "loss": 0.5836,
+      "step": 30707
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.2784900665283203,
+      "learning_rate": 5.243392101506239e-06,
+      "loss": 0.6715,
+      "step": 30708
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.7653863430023193,
+      "learning_rate": 5.243128755555996e-06,
+      "loss": 0.5406,
+      "step": 30709
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.9385933876037598,
+      "learning_rate": 5.2428654089297035e-06,
+      "loss": 0.6325,
+      "step": 30710
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.3162723779678345,
+      "learning_rate": 5.242602061628093e-06,
+      "loss": 0.6545,
+      "step": 30711
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.1852951049804688,
+      "learning_rate": 5.242338713651895e-06,
+      "loss": 0.4017,
+      "step": 30712
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.680509567260742,
+      "learning_rate": 5.242075365001846e-06,
+      "loss": 0.6245,
+      "step": 30713
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.6287879943847656,
+      "learning_rate": 5.241812015678677e-06,
+      "loss": 0.4984,
+      "step": 30714
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.5140745639801025,
+      "learning_rate": 5.241548665683118e-06,
+      "loss": 0.7335,
+      "step": 30715
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.469835877418518,
+      "learning_rate": 5.2412853150159025e-06,
+      "loss": 0.3963,
+      "step": 30716
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.333965539932251,
+      "learning_rate": 5.241021963677764e-06,
+      "loss": 0.8302,
+      "step": 30717
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 5.896119594573975,
+      "learning_rate": 5.240758611669432e-06,
+      "loss": 0.755,
+      "step": 30718
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.1282799243927,
+      "learning_rate": 5.240495258991642e-06,
+      "loss": 0.6934,
+      "step": 30719
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.1275997161865234,
+      "learning_rate": 5.240231905645124e-06,
+      "loss": 0.8682,
+      "step": 30720
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 5.0870561599731445,
+      "learning_rate": 5.239968551630612e-06,
+      "loss": 0.6484,
+      "step": 30721
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.5882079601287842,
+      "learning_rate": 5.239705196948835e-06,
+      "loss": 0.4832,
+      "step": 30722
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.2616076469421387,
+      "learning_rate": 5.23944184160053e-06,
+      "loss": 0.5898,
+      "step": 30723
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.866734504699707,
+      "learning_rate": 5.239178485586426e-06,
+      "loss": 0.6169,
+      "step": 30724
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.979248523712158,
+      "learning_rate": 5.238915128907256e-06,
+      "loss": 0.524,
+      "step": 30725
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.4417879581451416,
+      "learning_rate": 5.238651771563753e-06,
+      "loss": 0.7655,
+      "step": 30726
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.150754451751709,
+      "learning_rate": 5.23838841355665e-06,
+      "loss": 0.4264,
+      "step": 30727
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.6369917392730713,
+      "learning_rate": 5.238125054886676e-06,
+      "loss": 0.753,
+      "step": 30728
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.354818820953369,
+      "learning_rate": 5.237861695554567e-06,
+      "loss": 0.6742,
+      "step": 30729
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 7.925278663635254,
+      "learning_rate": 5.237598335561053e-06,
+      "loss": 0.5995,
+      "step": 30730
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 4.204823017120361,
+      "learning_rate": 5.237334974906867e-06,
+      "loss": 0.6593,
+      "step": 30731
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.123852252960205,
+      "learning_rate": 5.237071613592741e-06,
+      "loss": 0.4308,
+      "step": 30732
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.954632520675659,
+      "learning_rate": 5.236808251619407e-06,
+      "loss": 0.6213,
+      "step": 30733
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.846725344657898,
+      "learning_rate": 5.2365448889876006e-06,
+      "loss": 0.4665,
+      "step": 30734
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.869434118270874,
+      "learning_rate": 5.236281525698049e-06,
+      "loss": 0.5823,
+      "step": 30735
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.6428046226501465,
+      "learning_rate": 5.236018161751488e-06,
+      "loss": 0.5901,
+      "step": 30736
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.8871524333953857,
+      "learning_rate": 5.2357547971486485e-06,
+      "loss": 0.7328,
+      "step": 30737
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.7160346508026123,
+      "learning_rate": 5.235491431890264e-06,
+      "loss": 0.7471,
+      "step": 30738
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.6763134002685547,
+      "learning_rate": 5.235228065977066e-06,
+      "loss": 0.5221,
+      "step": 30739
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.0257627964019775,
+      "learning_rate": 5.234964699409787e-06,
+      "loss": 0.7433,
+      "step": 30740
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.068692445755005,
+      "learning_rate": 5.234701332189158e-06,
+      "loss": 0.6051,
+      "step": 30741
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.0605275630950928,
+      "learning_rate": 5.234437964315912e-06,
+      "loss": 0.3555,
+      "step": 30742
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.12041974067688,
+      "learning_rate": 5.234174595790783e-06,
+      "loss": 0.5069,
+      "step": 30743
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.736907720565796,
+      "learning_rate": 5.233911226614503e-06,
+      "loss": 0.5071,
+      "step": 30744
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.576103687286377,
+      "learning_rate": 5.233647856787803e-06,
+      "loss": 0.8206,
+      "step": 30745
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.731048107147217,
+      "learning_rate": 5.233384486311415e-06,
+      "loss": 0.5334,
+      "step": 30746
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 4.19002628326416,
+      "learning_rate": 5.233121115186073e-06,
+      "loss": 0.6371,
+      "step": 30747
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.2833218574523926,
+      "learning_rate": 5.2328577434125085e-06,
+      "loss": 0.5408,
+      "step": 30748
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.115652322769165,
+      "learning_rate": 5.232594370991453e-06,
+      "loss": 0.867,
+      "step": 30749
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.7596359252929688,
+      "learning_rate": 5.232330997923641e-06,
+      "loss": 0.4189,
+      "step": 30750
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.0547969341278076,
+      "learning_rate": 5.232067624209801e-06,
+      "loss": 0.6153,
+      "step": 30751
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.3636746406555176,
+      "learning_rate": 5.23180424985067e-06,
+      "loss": 0.7388,
+      "step": 30752
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.8056905269622803,
+      "learning_rate": 5.2315408748469775e-06,
+      "loss": 0.642,
+      "step": 30753
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.2145655155181885,
+      "learning_rate": 5.231277499199458e-06,
+      "loss": 0.2614,
+      "step": 30754
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.9356037378311157,
+      "learning_rate": 5.231014122908841e-06,
+      "loss": 0.865,
+      "step": 30755
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.424899101257324,
+      "learning_rate": 5.2307507459758585e-06,
+      "loss": 0.7161,
+      "step": 30756
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.1558992862701416,
+      "learning_rate": 5.230487368401248e-06,
+      "loss": 0.7031,
+      "step": 30757
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.7050061225891113,
+      "learning_rate": 5.230223990185737e-06,
+      "loss": 0.7005,
+      "step": 30758
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.373694658279419,
+      "learning_rate": 5.229960611330059e-06,
+      "loss": 0.6285,
+      "step": 30759
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.6736599206924438,
+      "learning_rate": 5.229697231834948e-06,
+      "loss": 0.6846,
+      "step": 30760
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.9392321109771729,
+      "learning_rate": 5.229433851701134e-06,
+      "loss": 0.7557,
+      "step": 30761
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.026536703109741,
+      "learning_rate": 5.2291704709293515e-06,
+      "loss": 0.6446,
+      "step": 30762
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.7063112258911133,
+      "learning_rate": 5.22890708952033e-06,
+      "loss": 0.6215,
+      "step": 30763
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.8149163722991943,
+      "learning_rate": 5.228643707474804e-06,
+      "loss": 0.7354,
+      "step": 30764
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.0936639308929443,
+      "learning_rate": 5.228380324793506e-06,
+      "loss": 0.3723,
+      "step": 30765
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.437628984451294,
+      "learning_rate": 5.228116941477168e-06,
+      "loss": 0.6471,
+      "step": 30766
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.0604913234710693,
+      "learning_rate": 5.227853557526522e-06,
+      "loss": 0.821,
+      "step": 30767
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.192847490310669,
+      "learning_rate": 5.2275901729423005e-06,
+      "loss": 0.7529,
+      "step": 30768
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.308487892150879,
+      "learning_rate": 5.227326787725236e-06,
+      "loss": 0.4657,
+      "step": 30769
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.494544267654419,
+      "learning_rate": 5.2270634018760625e-06,
+      "loss": 0.6705,
+      "step": 30770
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.3063979148864746,
+      "learning_rate": 5.226800015395509e-06,
+      "loss": 0.6576,
+      "step": 30771
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.7298905849456787,
+      "learning_rate": 5.22653662828431e-06,
+      "loss": 0.6542,
+      "step": 30772
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 5.201772689819336,
+      "learning_rate": 5.226273240543197e-06,
+      "loss": 0.7725,
+      "step": 30773
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.0605897903442383,
+      "learning_rate": 5.226009852172903e-06,
+      "loss": 0.6552,
+      "step": 30774
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.04938006401062,
+      "learning_rate": 5.2257464631741604e-06,
+      "loss": 0.6303,
+      "step": 30775
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.4135022163391113,
+      "learning_rate": 5.225483073547703e-06,
+      "loss": 0.7021,
+      "step": 30776
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.5356898307800293,
+      "learning_rate": 5.225219683294259e-06,
+      "loss": 0.8167,
+      "step": 30777
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.5086337327957153,
+      "learning_rate": 5.224956292414566e-06,
+      "loss": 0.5713,
+      "step": 30778
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.0565850734710693,
+      "learning_rate": 5.224692900909352e-06,
+      "loss": 0.7448,
+      "step": 30779
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.70013427734375,
+      "learning_rate": 5.224429508779352e-06,
+      "loss": 0.846,
+      "step": 30780
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.8574998378753662,
+      "learning_rate": 5.224166116025298e-06,
+      "loss": 0.6363,
+      "step": 30781
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.24379825592041,
+      "learning_rate": 5.223902722647921e-06,
+      "loss": 0.7559,
+      "step": 30782
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 4.571788311004639,
+      "learning_rate": 5.223639328647954e-06,
+      "loss": 0.4981,
+      "step": 30783
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.3075051307678223,
+      "learning_rate": 5.22337593402613e-06,
+      "loss": 0.5557,
+      "step": 30784
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.972909927368164,
+      "learning_rate": 5.223112538783182e-06,
+      "loss": 0.559,
+      "step": 30785
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.045847177505493,
+      "learning_rate": 5.222849142919841e-06,
+      "loss": 0.7998,
+      "step": 30786
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.313305616378784,
+      "learning_rate": 5.2225857464368415e-06,
+      "loss": 0.4338,
+      "step": 30787
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.944192886352539,
+      "learning_rate": 5.222322349334913e-06,
+      "loss": 0.3776,
+      "step": 30788
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.9986345767974854,
+      "learning_rate": 5.22205895161479e-06,
+      "loss": 0.3011,
+      "step": 30789
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.1283316612243652,
+      "learning_rate": 5.221795553277203e-06,
+      "loss": 0.7921,
+      "step": 30790
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.878760576248169,
+      "learning_rate": 5.221532154322888e-06,
+      "loss": 0.6398,
+      "step": 30791
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.5491936206817627,
+      "learning_rate": 5.221268754752573e-06,
+      "loss": 0.7282,
+      "step": 30792
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.4685635566711426,
+      "learning_rate": 5.221005354566992e-06,
+      "loss": 0.7098,
+      "step": 30793
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.8598964214324951,
+      "learning_rate": 5.220741953766879e-06,
+      "loss": 0.5906,
+      "step": 30794
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.2947959899902344,
+      "learning_rate": 5.220478552352966e-06,
+      "loss": 0.7926,
+      "step": 30795
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.375197172164917,
+      "learning_rate": 5.2202151503259846e-06,
+      "loss": 0.6995,
+      "step": 30796
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 4.192079067230225,
+      "learning_rate": 5.219951747686667e-06,
+      "loss": 0.5288,
+      "step": 30797
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.6419482231140137,
+      "learning_rate": 5.2196883444357445e-06,
+      "loss": 0.5885,
+      "step": 30798
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.9304530620574951,
+      "learning_rate": 5.2194249405739535e-06,
+      "loss": 0.8198,
+      "step": 30799
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.2978503704071045,
+      "learning_rate": 5.219161536102023e-06,
+      "loss": 0.5729,
+      "step": 30800
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.9957228899002075,
+      "learning_rate": 5.218898131020686e-06,
+      "loss": 0.7333,
+      "step": 30801
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.0640981197357178,
+      "learning_rate": 5.218634725330676e-06,
+      "loss": 0.7596,
+      "step": 30802
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.7270538806915283,
+      "learning_rate": 5.218371319032722e-06,
+      "loss": 0.6381,
+      "step": 30803
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.259587049484253,
+      "learning_rate": 5.218107912127562e-06,
+      "loss": 0.5757,
+      "step": 30804
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.8754745721817017,
+      "learning_rate": 5.217844504615925e-06,
+      "loss": 0.8237,
+      "step": 30805
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.0863146781921387,
+      "learning_rate": 5.217581096498544e-06,
+      "loss": 0.5678,
+      "step": 30806
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 5.898994445800781,
+      "learning_rate": 5.217317687776152e-06,
+      "loss": 0.7782,
+      "step": 30807
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.0546326637268066,
+      "learning_rate": 5.2170542784494805e-06,
+      "loss": 0.8059,
+      "step": 30808
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.075329542160034,
+      "learning_rate": 5.216790868519263e-06,
+      "loss": 0.6264,
+      "step": 30809
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 4.108133792877197,
+      "learning_rate": 5.21652745798623e-06,
+      "loss": 0.5608,
+      "step": 30810
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 4.383813858032227,
+      "learning_rate": 5.216264046851116e-06,
+      "loss": 0.6098,
+      "step": 30811
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 4.475008964538574,
+      "learning_rate": 5.2160006351146506e-06,
+      "loss": 0.4976,
+      "step": 30812
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.546255111694336,
+      "learning_rate": 5.215737222777571e-06,
+      "loss": 0.7391,
+      "step": 30813
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.519011974334717,
+      "learning_rate": 5.215473809840607e-06,
+      "loss": 0.4817,
+      "step": 30814
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 4.816680431365967,
+      "learning_rate": 5.215210396304489e-06,
+      "loss": 0.4641,
+      "step": 30815
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.4906315803527832,
+      "learning_rate": 5.214946982169954e-06,
+      "loss": 0.5941,
+      "step": 30816
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.9285733699798584,
+      "learning_rate": 5.21468356743773e-06,
+      "loss": 0.5154,
+      "step": 30817
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.5992772579193115,
+      "learning_rate": 5.214420152108551e-06,
+      "loss": 0.3282,
+      "step": 30818
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.1080963611602783,
+      "learning_rate": 5.214156736183151e-06,
+      "loss": 0.7296,
+      "step": 30819
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.7354286909103394,
+      "learning_rate": 5.213893319662262e-06,
+      "loss": 0.581,
+      "step": 30820
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.7499293088912964,
+      "learning_rate": 5.2136299025466155e-06,
+      "loss": 0.8014,
+      "step": 30821
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.9951601028442383,
+      "learning_rate": 5.213366484836944e-06,
+      "loss": 0.3356,
+      "step": 30822
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.3851571083068848,
+      "learning_rate": 5.21310306653398e-06,
+      "loss": 0.6894,
+      "step": 30823
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.6478993892669678,
+      "learning_rate": 5.212839647638456e-06,
+      "loss": 0.4275,
+      "step": 30824
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.1363557577133179,
+      "learning_rate": 5.212576228151105e-06,
+      "loss": 0.5762,
+      "step": 30825
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.9189448356628418,
+      "learning_rate": 5.212312808072659e-06,
+      "loss": 0.6964,
+      "step": 30826
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.0235307216644287,
+      "learning_rate": 5.212049387403852e-06,
+      "loss": 0.6915,
+      "step": 30827
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.811974048614502,
+      "learning_rate": 5.211785966145413e-06,
+      "loss": 0.5175,
+      "step": 30828
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.020565986633301,
+      "learning_rate": 5.211522544298077e-06,
+      "loss": 0.6542,
+      "step": 30829
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.5745089054107666,
+      "learning_rate": 5.2112591218625775e-06,
+      "loss": 0.6223,
+      "step": 30830
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.7500356435775757,
+      "learning_rate": 5.210995698839645e-06,
+      "loss": 0.6335,
+      "step": 30831
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.881012201309204,
+      "learning_rate": 5.210732275230011e-06,
+      "loss": 0.5945,
+      "step": 30832
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.2019202709198,
+      "learning_rate": 5.210468851034412e-06,
+      "loss": 0.7549,
+      "step": 30833
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.5213418006896973,
+      "learning_rate": 5.210205426253575e-06,
+      "loss": 0.8408,
+      "step": 30834
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.2240118980407715,
+      "learning_rate": 5.2099420008882375e-06,
+      "loss": 0.7346,
+      "step": 30835
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.4722766876220703,
+      "learning_rate": 5.209678574939129e-06,
+      "loss": 0.7898,
+      "step": 30836
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.041219711303711,
+      "learning_rate": 5.209415148406983e-06,
+      "loss": 0.5903,
+      "step": 30837
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 5.137691497802734,
+      "learning_rate": 5.209151721292533e-06,
+      "loss": 0.7067,
+      "step": 30838
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.9182019233703613,
+      "learning_rate": 5.20888829359651e-06,
+      "loss": 0.634,
+      "step": 30839
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.8771474361419678,
+      "learning_rate": 5.208624865319647e-06,
+      "loss": 0.4367,
+      "step": 30840
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.0309526920318604,
+      "learning_rate": 5.208361436462676e-06,
+      "loss": 0.5374,
+      "step": 30841
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.56888484954834,
+      "learning_rate": 5.20809800702633e-06,
+      "loss": 0.8554,
+      "step": 30842
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.9376894235610962,
+      "learning_rate": 5.207834577011342e-06,
+      "loss": 0.7746,
+      "step": 30843
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.5798189640045166,
+      "learning_rate": 5.207571146418441e-06,
+      "loss": 0.5958,
+      "step": 30844
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.512545347213745,
+      "learning_rate": 5.2073077152483655e-06,
+      "loss": 0.7321,
+      "step": 30845
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.9054856300354004,
+      "learning_rate": 5.207044283501843e-06,
+      "loss": 0.6649,
+      "step": 30846
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.4886362552642822,
+      "learning_rate": 5.206780851179609e-06,
+      "loss": 0.3524,
+      "step": 30847
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.857362747192383,
+      "learning_rate": 5.206517418282395e-06,
+      "loss": 0.8426,
+      "step": 30848
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 5.19559383392334,
+      "learning_rate": 5.206253984810934e-06,
+      "loss": 0.6713,
+      "step": 30849
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.0468838214874268,
+      "learning_rate": 5.205990550765957e-06,
+      "loss": 0.8533,
+      "step": 30850
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.1159934997558594,
+      "learning_rate": 5.205727116148198e-06,
+      "loss": 0.4821,
+      "step": 30851
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.781635046005249,
+      "learning_rate": 5.205463680958388e-06,
+      "loss": 0.7242,
+      "step": 30852
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.102672576904297,
+      "learning_rate": 5.205200245197262e-06,
+      "loss": 0.5432,
+      "step": 30853
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.1954784393310547,
+      "learning_rate": 5.204936808865549e-06,
+      "loss": 0.3364,
+      "step": 30854
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.91930890083313,
+      "learning_rate": 5.204673371963984e-06,
+      "loss": 0.7453,
+      "step": 30855
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.216750383377075,
+      "learning_rate": 5.2044099344933e-06,
+      "loss": 0.6383,
+      "step": 30856
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.166198253631592,
+      "learning_rate": 5.204146496454227e-06,
+      "loss": 0.7435,
+      "step": 30857
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 4.201423168182373,
+      "learning_rate": 5.203883057847501e-06,
+      "loss": 0.9096,
+      "step": 30858
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.467911958694458,
+      "learning_rate": 5.203619618673852e-06,
+      "loss": 0.7179,
+      "step": 30859
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 4.061522483825684,
+      "learning_rate": 5.203356178934012e-06,
+      "loss": 0.3441,
+      "step": 30860
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 6.536722183227539,
+      "learning_rate": 5.203092738628717e-06,
+      "loss": 0.6234,
+      "step": 30861
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.58133864402771,
+      "learning_rate": 5.202829297758696e-06,
+      "loss": 0.7395,
+      "step": 30862
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.7885076999664307,
+      "learning_rate": 5.202565856324682e-06,
+      "loss": 0.7308,
+      "step": 30863
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.961496353149414,
+      "learning_rate": 5.2023024143274085e-06,
+      "loss": 0.7162,
+      "step": 30864
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.857004165649414,
+      "learning_rate": 5.202038971767607e-06,
+      "loss": 0.4565,
+      "step": 30865
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.586058259010315,
+      "learning_rate": 5.201775528646012e-06,
+      "loss": 0.5636,
+      "step": 30866
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.7918260097503662,
+      "learning_rate": 5.201512084963355e-06,
+      "loss": 0.7552,
+      "step": 30867
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.5700541734695435,
+      "learning_rate": 5.201248640720367e-06,
+      "loss": 0.5108,
+      "step": 30868
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.138282299041748,
+      "learning_rate": 5.2009851959177835e-06,
+      "loss": 0.3652,
+      "step": 30869
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.8859002590179443,
+      "learning_rate": 5.200721750556334e-06,
+      "loss": 0.6908,
+      "step": 30870
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.500040054321289,
+      "learning_rate": 5.2004583046367544e-06,
+      "loss": 0.6556,
+      "step": 30871
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.0596330165863037,
+      "learning_rate": 5.200194858159775e-06,
+      "loss": 0.8372,
+      "step": 30872
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.2071664333343506,
+      "learning_rate": 5.199931411126128e-06,
+      "loss": 0.6693,
+      "step": 30873
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.04176664352417,
+      "learning_rate": 5.199667963536546e-06,
+      "loss": 1.0243,
+      "step": 30874
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.573805093765259,
+      "learning_rate": 5.199404515391761e-06,
+      "loss": 0.7528,
+      "step": 30875
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.718658685684204,
+      "learning_rate": 5.19914106669251e-06,
+      "loss": 0.3763,
+      "step": 30876
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.303338050842285,
+      "learning_rate": 5.19887761743952e-06,
+      "loss": 0.7243,
+      "step": 30877
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 4.001837253570557,
+      "learning_rate": 5.198614167633528e-06,
+      "loss": 0.7414,
+      "step": 30878
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.597707986831665,
+      "learning_rate": 5.198350717275262e-06,
+      "loss": 0.6478,
+      "step": 30879
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 4.987223148345947,
+      "learning_rate": 5.198087266365458e-06,
+      "loss": 1.0366,
+      "step": 30880
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.4232919216156006,
+      "learning_rate": 5.197823814904848e-06,
+      "loss": 0.3792,
+      "step": 30881
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.0045182704925537,
+      "learning_rate": 5.197560362894163e-06,
+      "loss": 0.696,
+      "step": 30882
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.1836268901824951,
+      "learning_rate": 5.197296910334138e-06,
+      "loss": 0.3178,
+      "step": 30883
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.1865503787994385,
+      "learning_rate": 5.197033457225503e-06,
+      "loss": 0.6432,
+      "step": 30884
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.6027677059173584,
+      "learning_rate": 5.196770003568992e-06,
+      "loss": 0.6082,
+      "step": 30885
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.9244277477264404,
+      "learning_rate": 5.196506549365336e-06,
+      "loss": 0.7071,
+      "step": 30886
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 4.723567008972168,
+      "learning_rate": 5.196243094615272e-06,
+      "loss": 0.5459,
+      "step": 30887
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.8701441287994385,
+      "learning_rate": 5.195979639319527e-06,
+      "loss": 0.5376,
+      "step": 30888
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.4316952228546143,
+      "learning_rate": 5.195716183478837e-06,
+      "loss": 0.6348,
+      "step": 30889
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.437629222869873,
+      "learning_rate": 5.195452727093935e-06,
+      "loss": 0.8795,
+      "step": 30890
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.158662796020508,
+      "learning_rate": 5.19518927016555e-06,
+      "loss": 0.6648,
+      "step": 30891
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.0475058555603027,
+      "learning_rate": 5.194925812694418e-06,
+      "loss": 0.6172,
+      "step": 30892
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.7206742763519287,
+      "learning_rate": 5.19466235468127e-06,
+      "loss": 0.6435,
+      "step": 30893
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.3956093788146973,
+      "learning_rate": 5.194398896126839e-06,
+      "loss": 0.6969,
+      "step": 30894
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.8518459796905518,
+      "learning_rate": 5.1941354370318566e-06,
+      "loss": 0.5973,
+      "step": 30895
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.2547593116760254,
+      "learning_rate": 5.193871977397056e-06,
+      "loss": 0.6845,
+      "step": 30896
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 4.950571060180664,
+      "learning_rate": 5.193608517223171e-06,
+      "loss": 0.7274,
+      "step": 30897
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.779608964920044,
+      "learning_rate": 5.1933450565109335e-06,
+      "loss": 0.4512,
+      "step": 30898
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 4.238165378570557,
+      "learning_rate": 5.193081595261076e-06,
+      "loss": 0.5873,
+      "step": 30899
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.8998706340789795,
+      "learning_rate": 5.192818133474331e-06,
+      "loss": 0.5291,
+      "step": 30900
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.3063292503356934,
+      "learning_rate": 5.1925546711514305e-06,
+      "loss": 0.4882,
+      "step": 30901
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.5885586738586426,
+      "learning_rate": 5.192291208293108e-06,
+      "loss": 0.7173,
+      "step": 30902
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.3533505201339722,
+      "learning_rate": 5.192027744900095e-06,
+      "loss": 0.4431,
+      "step": 30903
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.1336913108825684,
+      "learning_rate": 5.191764280973125e-06,
+      "loss": 0.6907,
+      "step": 30904
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.4859870672225952,
+      "learning_rate": 5.191500816512929e-06,
+      "loss": 0.3615,
+      "step": 30905
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.77783203125,
+      "learning_rate": 5.1912373515202426e-06,
+      "loss": 0.773,
+      "step": 30906
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.9623022079467773,
+      "learning_rate": 5.190973885995797e-06,
+      "loss": 0.6863,
+      "step": 30907
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.1807847023010254,
+      "learning_rate": 5.190710419940323e-06,
+      "loss": 0.817,
+      "step": 30908
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.1835694313049316,
+      "learning_rate": 5.190446953354557e-06,
+      "loss": 0.7194,
+      "step": 30909
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 4.037649154663086,
+      "learning_rate": 5.190183486239229e-06,
+      "loss": 0.4788,
+      "step": 30910
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.523662567138672,
+      "learning_rate": 5.189920018595071e-06,
+      "loss": 0.5753,
+      "step": 30911
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.4309654235839844,
+      "learning_rate": 5.189656550422817e-06,
+      "loss": 0.6845,
+      "step": 30912
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.9983545541763306,
+      "learning_rate": 5.189393081723199e-06,
+      "loss": 0.426,
+      "step": 30913
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.652907133102417,
+      "learning_rate": 5.189129612496949e-06,
+      "loss": 0.6586,
+      "step": 30914
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.6065338850021362,
+      "learning_rate": 5.188866142744801e-06,
+      "loss": 0.7529,
+      "step": 30915
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.777700901031494,
+      "learning_rate": 5.188602672467486e-06,
+      "loss": 0.5093,
+      "step": 30916
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.283588171005249,
+      "learning_rate": 5.188339201665739e-06,
+      "loss": 0.5915,
+      "step": 30917
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.1667606830596924,
+      "learning_rate": 5.1880757303402914e-06,
+      "loss": 0.4115,
+      "step": 30918
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.8993549346923828,
+      "learning_rate": 5.187812258491875e-06,
+      "loss": 0.4657,
+      "step": 30919
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.2217161655426025,
+      "learning_rate": 5.187548786121224e-06,
+      "loss": 0.4814,
+      "step": 30920
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.8509613275527954,
+      "learning_rate": 5.18728531322907e-06,
+      "loss": 0.7125,
+      "step": 30921
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.8491321802139282,
+      "learning_rate": 5.187021839816146e-06,
+      "loss": 0.5493,
+      "step": 30922
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.5882292985916138,
+      "learning_rate": 5.186758365883182e-06,
+      "loss": 0.8166,
+      "step": 30923
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.5890138149261475,
+      "learning_rate": 5.186494891430916e-06,
+      "loss": 0.6638,
+      "step": 30924
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.1471269130706787,
+      "learning_rate": 5.186231416460074e-06,
+      "loss": 0.7101,
+      "step": 30925
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.8930327892303467,
+      "learning_rate": 5.185967940971394e-06,
+      "loss": 0.6577,
+      "step": 30926
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.136843681335449,
+      "learning_rate": 5.185704464965608e-06,
+      "loss": 0.5717,
+      "step": 30927
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.1044726371765137,
+      "learning_rate": 5.185440988443445e-06,
+      "loss": 0.6001,
+      "step": 30928
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 5.96146821975708,
+      "learning_rate": 5.1851775114056426e-06,
+      "loss": 0.512,
+      "step": 30929
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.0971624851226807,
+      "learning_rate": 5.1849140338529305e-06,
+      "loss": 0.7577,
+      "step": 30930
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.9860594272613525,
+      "learning_rate": 5.184650555786041e-06,
+      "loss": 0.6939,
+      "step": 30931
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.4015772342681885,
+      "learning_rate": 5.184387077205707e-06,
+      "loss": 0.4362,
+      "step": 30932
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.6908156871795654,
+      "learning_rate": 5.1841235981126625e-06,
+      "loss": 0.5684,
+      "step": 30933
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.456057071685791,
+      "learning_rate": 5.183860118507639e-06,
+      "loss": 0.802,
+      "step": 30934
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.845276117324829,
+      "learning_rate": 5.1835966383913696e-06,
+      "loss": 0.6945,
+      "step": 30935
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.544220209121704,
+      "learning_rate": 5.183333157764585e-06,
+      "loss": 1.0216,
+      "step": 30936
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.8507882356643677,
+      "learning_rate": 5.183069676628021e-06,
+      "loss": 0.2986,
+      "step": 30937
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.091621160507202,
+      "learning_rate": 5.182806194982408e-06,
+      "loss": 0.4608,
+      "step": 30938
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 4.025825500488281,
+      "learning_rate": 5.182542712828481e-06,
+      "loss": 0.646,
+      "step": 30939
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.680154323577881,
+      "learning_rate": 5.18227923016697e-06,
+      "loss": 0.6029,
+      "step": 30940
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.4018311500549316,
+      "learning_rate": 5.18201574699861e-06,
+      "loss": 0.5769,
+      "step": 30941
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.1581690311431885,
+      "learning_rate": 5.18175226332413e-06,
+      "loss": 0.6233,
+      "step": 30942
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.6399736404418945,
+      "learning_rate": 5.181488779144267e-06,
+      "loss": 0.5925,
+      "step": 30943
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.3487484455108643,
+      "learning_rate": 5.181225294459751e-06,
+      "loss": 0.7299,
+      "step": 30944
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.3045856952667236,
+      "learning_rate": 5.1809618092713155e-06,
+      "loss": 0.5892,
+      "step": 30945
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.311110496520996,
+      "learning_rate": 5.1806983235796934e-06,
+      "loss": 0.9689,
+      "step": 30946
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.9203929901123047,
+      "learning_rate": 5.180434837385615e-06,
+      "loss": 0.3991,
+      "step": 30947
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.361846685409546,
+      "learning_rate": 5.1801713506898164e-06,
+      "loss": 0.6698,
+      "step": 30948
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.450169086456299,
+      "learning_rate": 5.17990786349303e-06,
+      "loss": 0.7296,
+      "step": 30949
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.865326166152954,
+      "learning_rate": 5.179644375795986e-06,
+      "loss": 0.6816,
+      "step": 30950
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.265862464904785,
+      "learning_rate": 5.179380887599418e-06,
+      "loss": 0.6583,
+      "step": 30951
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.126951217651367,
+      "learning_rate": 5.17911739890406e-06,
+      "loss": 0.4151,
+      "step": 30952
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.071772575378418,
+      "learning_rate": 5.178853909710643e-06,
+      "loss": 0.4404,
+      "step": 30953
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.5443902015686035,
+      "learning_rate": 5.1785904200199e-06,
+      "loss": 0.6097,
+      "step": 30954
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.7880351543426514,
+      "learning_rate": 5.1783269298325646e-06,
+      "loss": 0.5912,
+      "step": 30955
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.5072238445281982,
+      "learning_rate": 5.178063439149367e-06,
+      "loss": 0.56,
+      "step": 30956
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.6687140464782715,
+      "learning_rate": 5.1777999479710425e-06,
+      "loss": 0.6396,
+      "step": 30957
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 4.005619525909424,
+      "learning_rate": 5.177536456298324e-06,
+      "loss": 0.3552,
+      "step": 30958
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 4.56036376953125,
+      "learning_rate": 5.177272964131943e-06,
+      "loss": 0.4697,
+      "step": 30959
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.2294654846191406,
+      "learning_rate": 5.1770094714726316e-06,
+      "loss": 0.74,
+      "step": 30960
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.731261134147644,
+      "learning_rate": 5.176745978321124e-06,
+      "loss": 0.6319,
+      "step": 30961
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 4.05061674118042,
+      "learning_rate": 5.176482484678151e-06,
+      "loss": 0.8381,
+      "step": 30962
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.694258451461792,
+      "learning_rate": 5.176218990544448e-06,
+      "loss": 0.6792,
+      "step": 30963
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.912994861602783,
+      "learning_rate": 5.175955495920744e-06,
+      "loss": 0.6521,
+      "step": 30964
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.3746299743652344,
+      "learning_rate": 5.175692000807775e-06,
+      "loss": 0.5639,
+      "step": 30965
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.2158889770507812,
+      "learning_rate": 5.175428505206271e-06,
+      "loss": 0.5993,
+      "step": 30966
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.234830617904663,
+      "learning_rate": 5.175165009116967e-06,
+      "loss": 0.5871,
+      "step": 30967
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.037013053894043,
+      "learning_rate": 5.174901512540596e-06,
+      "loss": 0.8757,
+      "step": 30968
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.6781041622161865,
+      "learning_rate": 5.1746380154778874e-06,
+      "loss": 0.5648,
+      "step": 30969
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.575929641723633,
+      "learning_rate": 5.174374517929577e-06,
+      "loss": 0.7546,
+      "step": 30970
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.0508759021759033,
+      "learning_rate": 5.174111019896395e-06,
+      "loss": 0.6416,
+      "step": 30971
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 4.353960990905762,
+      "learning_rate": 5.173847521379077e-06,
+      "loss": 0.8961,
+      "step": 30972
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.5521039962768555,
+      "learning_rate": 5.173584022378354e-06,
+      "loss": 0.566,
+      "step": 30973
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.3392672538757324,
+      "learning_rate": 5.173320522894958e-06,
+      "loss": 0.8521,
+      "step": 30974
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.009213447570801,
+      "learning_rate": 5.1730570229296236e-06,
+      "loss": 0.5068,
+      "step": 30975
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.357944965362549,
+      "learning_rate": 5.1727935224830805e-06,
+      "loss": 0.4777,
+      "step": 30976
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.6147570610046387,
+      "learning_rate": 5.172530021556065e-06,
+      "loss": 0.62,
+      "step": 30977
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.753983497619629,
+      "learning_rate": 5.172266520149308e-06,
+      "loss": 0.8295,
+      "step": 30978
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.1888052225112915,
+      "learning_rate": 5.172003018263543e-06,
+      "loss": 0.4746,
+      "step": 30979
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 4.508598804473877,
+      "learning_rate": 5.171739515899501e-06,
+      "loss": 0.9433,
+      "step": 30980
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.3808577060699463,
+      "learning_rate": 5.1714760130579155e-06,
+      "loss": 0.8826,
+      "step": 30981
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.330139398574829,
+      "learning_rate": 5.171212509739521e-06,
+      "loss": 0.5076,
+      "step": 30982
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.4422144889831543,
+      "learning_rate": 5.170949005945047e-06,
+      "loss": 0.6027,
+      "step": 30983
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.201652765274048,
+      "learning_rate": 5.1706855016752285e-06,
+      "loss": 0.7189,
+      "step": 30984
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.8160455226898193,
+      "learning_rate": 5.170421996930798e-06,
+      "loss": 0.9245,
+      "step": 30985
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.7964069843292236,
+      "learning_rate": 5.1701584917124855e-06,
+      "loss": 0.5754,
+      "step": 30986
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.256457567214966,
+      "learning_rate": 5.169894986021028e-06,
+      "loss": 0.5792,
+      "step": 30987
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.100627899169922,
+      "learning_rate": 5.169631479857156e-06,
+      "loss": 0.3189,
+      "step": 30988
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.0457141399383545,
+      "learning_rate": 5.169367973221603e-06,
+      "loss": 0.6459,
+      "step": 30989
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.5679657459259033,
+      "learning_rate": 5.1691044661151e-06,
+      "loss": 0.7017,
+      "step": 30990
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.6021413803100586,
+      "learning_rate": 5.168840958538381e-06,
+      "loss": 0.6575,
+      "step": 30991
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.80551815032959,
+      "learning_rate": 5.16857745049218e-06,
+      "loss": 0.4808,
+      "step": 30992
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.3308513164520264,
+      "learning_rate": 5.168313941977227e-06,
+      "loss": 0.7446,
+      "step": 30993
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.3681910037994385,
+      "learning_rate": 5.168050432994256e-06,
+      "loss": 0.5082,
+      "step": 30994
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 4.512857437133789,
+      "learning_rate": 5.167786923543999e-06,
+      "loss": 0.7172,
+      "step": 30995
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.1387827396392822,
+      "learning_rate": 5.167523413627191e-06,
+      "loss": 0.414,
+      "step": 30996
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.2014265060424805,
+      "learning_rate": 5.167259903244562e-06,
+      "loss": 0.494,
+      "step": 30997
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.432004928588867,
+      "learning_rate": 5.166996392396845e-06,
+      "loss": 0.8742,
+      "step": 30998
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.0515377521514893,
+      "learning_rate": 5.166732881084775e-06,
+      "loss": 0.4916,
+      "step": 30999
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.320192575454712,
+      "learning_rate": 5.166469369309083e-06,
+      "loss": 0.747,
+      "step": 31000
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.2040722370147705,
+      "learning_rate": 5.166205857070503e-06,
+      "loss": 0.5703,
+      "step": 31001
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.7201597690582275,
+      "learning_rate": 5.165942344369766e-06,
+      "loss": 0.7162,
+      "step": 31002
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.3957912921905518,
+      "learning_rate": 5.165678831207605e-06,
+      "loss": 0.5417,
+      "step": 31003
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.2241313457489014,
+      "learning_rate": 5.1654153175847535e-06,
+      "loss": 0.7132,
+      "step": 31004
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.9103331565856934,
+      "learning_rate": 5.165151803501943e-06,
+      "loss": 0.5624,
+      "step": 31005
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.9134557247161865,
+      "learning_rate": 5.164888288959909e-06,
+      "loss": 0.6519,
+      "step": 31006
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.661466598510742,
+      "learning_rate": 5.164624773959381e-06,
+      "loss": 0.4936,
+      "step": 31007
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.9755771160125732,
+      "learning_rate": 5.164361258501095e-06,
+      "loss": 0.6689,
+      "step": 31008
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.5908775329589844,
+      "learning_rate": 5.16409774258578e-06,
+      "loss": 0.4038,
+      "step": 31009
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.1937942504882812,
+      "learning_rate": 5.1638342262141704e-06,
+      "loss": 0.7578,
+      "step": 31010
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.3536083698272705,
+      "learning_rate": 5.163570709387001e-06,
+      "loss": 0.6325,
+      "step": 31011
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 4.6969122886657715,
+      "learning_rate": 5.1633071921050025e-06,
+      "loss": 0.6173,
+      "step": 31012
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.308826446533203,
+      "learning_rate": 5.163043674368908e-06,
+      "loss": 0.5056,
+      "step": 31013
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.6290363073349,
+      "learning_rate": 5.1627801561794485e-06,
+      "loss": 0.4428,
+      "step": 31014
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.066812515258789,
+      "learning_rate": 5.16251663753736e-06,
+      "loss": 0.6334,
+      "step": 31015
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.085573673248291,
+      "learning_rate": 5.1622531184433736e-06,
+      "loss": 0.5887,
+      "step": 31016
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.9927656650543213,
+      "learning_rate": 5.16198959889822e-06,
+      "loss": 0.6541,
+      "step": 31017
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.308591365814209,
+      "learning_rate": 5.161726078902635e-06,
+      "loss": 0.4805,
+      "step": 31018
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.7654783725738525,
+      "learning_rate": 5.161462558457351e-06,
+      "loss": 0.6506,
+      "step": 31019
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.1675913333892822,
+      "learning_rate": 5.1611990375631e-06,
+      "loss": 0.7311,
+      "step": 31020
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.463090419769287,
+      "learning_rate": 5.160935516220615e-06,
+      "loss": 0.4294,
+      "step": 31021
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.7746007442474365,
+      "learning_rate": 5.160671994430628e-06,
+      "loss": 0.452,
+      "step": 31022
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.9230304956436157,
+      "learning_rate": 5.160408472193873e-06,
+      "loss": 0.7301,
+      "step": 31023
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.438019037246704,
+      "learning_rate": 5.160144949511082e-06,
+      "loss": 0.6251,
+      "step": 31024
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 4.6338372230529785,
+      "learning_rate": 5.159881426382987e-06,
+      "loss": 0.5472,
+      "step": 31025
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.2561676502227783,
+      "learning_rate": 5.159617902810323e-06,
+      "loss": 0.6525,
+      "step": 31026
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.0832366943359375,
+      "learning_rate": 5.15935437879382e-06,
+      "loss": 0.7375,
+      "step": 31027
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.854492425918579,
+      "learning_rate": 5.159090854334213e-06,
+      "loss": 0.9858,
+      "step": 31028
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.8612629175186157,
+      "learning_rate": 5.158827329432234e-06,
+      "loss": 0.6217,
+      "step": 31029
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 3.472383975982666,
+      "learning_rate": 5.158563804088616e-06,
+      "loss": 0.8629,
+      "step": 31030
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.3494231700897217,
+      "learning_rate": 5.158300278304091e-06,
+      "loss": 0.7526,
+      "step": 31031
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.131730556488037,
+      "learning_rate": 5.158036752079393e-06,
+      "loss": 0.7712,
+      "step": 31032
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.716156244277954,
+      "learning_rate": 5.157773225415253e-06,
+      "loss": 0.6257,
+      "step": 31033
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.024913787841797,
+      "learning_rate": 5.157509698312405e-06,
+      "loss": 0.811,
+      "step": 31034
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.2128777503967285,
+      "learning_rate": 5.1572461707715826e-06,
+      "loss": 0.5892,
+      "step": 31035
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.4695228338241577,
+      "learning_rate": 5.156982642793517e-06,
+      "loss": 0.5533,
+      "step": 31036
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.7170195579528809,
+      "learning_rate": 5.1567191143789394e-06,
+      "loss": 0.7493,
+      "step": 31037
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.40893816947937,
+      "learning_rate": 5.156455585528586e-06,
+      "loss": 0.6004,
+      "step": 31038
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 3.333728075027466,
+      "learning_rate": 5.156192056243189e-06,
+      "loss": 0.5769,
+      "step": 31039
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.977203130722046,
+      "learning_rate": 5.1559285265234795e-06,
+      "loss": 0.7388,
+      "step": 31040
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.7005717754364014,
+      "learning_rate": 5.1556649963701925e-06,
+      "loss": 0.6173,
+      "step": 31041
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.5093181133270264,
+      "learning_rate": 5.155401465784058e-06,
+      "loss": 0.7352,
+      "step": 31042
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.6383376121520996,
+      "learning_rate": 5.155137934765811e-06,
+      "loss": 0.4895,
+      "step": 31043
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.124669313430786,
+      "learning_rate": 5.154874403316184e-06,
+      "loss": 0.4248,
+      "step": 31044
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 3.5235207080841064,
+      "learning_rate": 5.154610871435908e-06,
+      "loss": 0.7207,
+      "step": 31045
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 4.658172607421875,
+      "learning_rate": 5.154347339125717e-06,
+      "loss": 0.46,
+      "step": 31046
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 3.7427725791931152,
+      "learning_rate": 5.154083806386344e-06,
+      "loss": 0.6082,
+      "step": 31047
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 3.1149420738220215,
+      "learning_rate": 5.1538202732185225e-06,
+      "loss": 0.7264,
+      "step": 31048
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.0792019367218018,
+      "learning_rate": 5.153556739622983e-06,
+      "loss": 0.4623,
+      "step": 31049
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 4.321531295776367,
+      "learning_rate": 5.153293205600462e-06,
+      "loss": 0.7019,
+      "step": 31050
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.661407709121704,
+      "learning_rate": 5.153029671151689e-06,
+      "loss": 0.7397,
+      "step": 31051
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.393765449523926,
+      "learning_rate": 5.152766136277397e-06,
+      "loss": 0.8171,
+      "step": 31052
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.8125402927398682,
+      "learning_rate": 5.152502600978321e-06,
+      "loss": 0.7631,
+      "step": 31053
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.203174352645874,
+      "learning_rate": 5.1522390652551925e-06,
+      "loss": 0.6648,
+      "step": 31054
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.7103722095489502,
+      "learning_rate": 5.1519755291087425e-06,
+      "loss": 0.6326,
+      "step": 31055
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.6180450916290283,
+      "learning_rate": 5.151711992539706e-06,
+      "loss": 0.8814,
+      "step": 31056
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.91679048538208,
+      "learning_rate": 5.151448455548817e-06,
+      "loss": 0.573,
+      "step": 31057
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.571483850479126,
+      "learning_rate": 5.151184918136803e-06,
+      "loss": 0.499,
+      "step": 31058
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.8966097831726074,
+      "learning_rate": 5.1509213803044035e-06,
+      "loss": 0.7701,
+      "step": 31059
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.9545159339904785,
+      "learning_rate": 5.1506578420523465e-06,
+      "loss": 0.7399,
+      "step": 31060
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.44403076171875,
+      "learning_rate": 5.150394303381369e-06,
+      "loss": 0.595,
+      "step": 31061
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.160757064819336,
+      "learning_rate": 5.150130764292199e-06,
+      "loss": 0.5785,
+      "step": 31062
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.616408109664917,
+      "learning_rate": 5.1498672247855715e-06,
+      "loss": 0.913,
+      "step": 31063
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.028064727783203,
+      "learning_rate": 5.14960368486222e-06,
+      "loss": 0.7402,
+      "step": 31064
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.6727325916290283,
+      "learning_rate": 5.149340144522877e-06,
+      "loss": 0.808,
+      "step": 31065
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.111520767211914,
+      "learning_rate": 5.149076603768274e-06,
+      "loss": 0.8059,
+      "step": 31066
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.511395215988159,
+      "learning_rate": 5.148813062599145e-06,
+      "loss": 0.6905,
+      "step": 31067
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 4.392109394073486,
+      "learning_rate": 5.148549521016224e-06,
+      "loss": 0.6909,
+      "step": 31068
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.4368557929992676,
+      "learning_rate": 5.148285979020241e-06,
+      "loss": 0.4637,
+      "step": 31069
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 3.285853147506714,
+      "learning_rate": 5.1480224366119315e-06,
+      "loss": 0.7184,
+      "step": 31070
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 3.7622230052948,
+      "learning_rate": 5.147758893792026e-06,
+      "loss": 0.5238,
+      "step": 31071
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.3608717918395996,
+      "learning_rate": 5.147495350561258e-06,
+      "loss": 0.6611,
+      "step": 31072
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 3.5167527198791504,
+      "learning_rate": 5.147231806920362e-06,
+      "loss": 0.6346,
+      "step": 31073
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.975400447845459,
+      "learning_rate": 5.146968262870069e-06,
+      "loss": 0.8161,
+      "step": 31074
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.460794687271118,
+      "learning_rate": 5.146704718411112e-06,
+      "loss": 0.6905,
+      "step": 31075
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.9649572372436523,
+      "learning_rate": 5.146441173544224e-06,
+      "loss": 0.7448,
+      "step": 31076
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 4.15146541595459,
+      "learning_rate": 5.146177628270139e-06,
+      "loss": 0.5558,
+      "step": 31077
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.631002426147461,
+      "learning_rate": 5.145914082589587e-06,
+      "loss": 0.5376,
+      "step": 31078
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.9110664129257202,
+      "learning_rate": 5.1456505365033034e-06,
+      "loss": 0.6349,
+      "step": 31079
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 5.075586795806885,
+      "learning_rate": 5.1453869900120214e-06,
+      "loss": 0.8967,
+      "step": 31080
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 3.5259132385253906,
+      "learning_rate": 5.145123443116471e-06,
+      "loss": 0.6669,
+      "step": 31081
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 3.3434832096099854,
+      "learning_rate": 5.144859895817388e-06,
+      "loss": 0.6892,
+      "step": 31082
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.169863700866699,
+      "learning_rate": 5.144596348115503e-06,
+      "loss": 0.7519,
+      "step": 31083
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.6311755180358887,
+      "learning_rate": 5.14433280001155e-06,
+      "loss": 0.6743,
+      "step": 31084
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.4831109046936035,
+      "learning_rate": 5.144069251506261e-06,
+      "loss": 0.483,
+      "step": 31085
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 3.248178482055664,
+      "learning_rate": 5.1438057026003705e-06,
+      "loss": 0.5448,
+      "step": 31086
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 3.2353827953338623,
+      "learning_rate": 5.143542153294611e-06,
+      "loss": 0.8547,
+      "step": 31087
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 3.6217808723449707,
+      "learning_rate": 5.143278603589712e-06,
+      "loss": 0.5533,
+      "step": 31088
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.1882777214050293,
+      "learning_rate": 5.1430150534864095e-06,
+      "loss": 0.6842,
+      "step": 31089
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.9751734733581543,
+      "learning_rate": 5.142751502985437e-06,
+      "loss": 0.8275,
+      "step": 31090
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 3.4234890937805176,
+      "learning_rate": 5.1424879520875246e-06,
+      "loss": 0.4939,
+      "step": 31091
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.824876308441162,
+      "learning_rate": 5.142224400793407e-06,
+      "loss": 0.413,
+      "step": 31092
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 4.033320426940918,
+      "learning_rate": 5.141960849103817e-06,
+      "loss": 0.9032,
+      "step": 31093
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.3408560752868652,
+      "learning_rate": 5.141697297019488e-06,
+      "loss": 0.5607,
+      "step": 31094
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 3.018094539642334,
+      "learning_rate": 5.141433744541151e-06,
+      "loss": 0.8157,
+      "step": 31095
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.6279038190841675,
+      "learning_rate": 5.14117019166954e-06,
+      "loss": 0.6692,
+      "step": 31096
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.092419147491455,
+      "learning_rate": 5.140906638405387e-06,
+      "loss": 0.5923,
+      "step": 31097
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.125462532043457,
+      "learning_rate": 5.140643084749425e-06,
+      "loss": 0.4917,
+      "step": 31098
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 3.0940089225769043,
+      "learning_rate": 5.140379530702388e-06,
+      "loss": 0.7078,
+      "step": 31099
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.818467617034912,
+      "learning_rate": 5.1401159762650075e-06,
+      "loss": 0.44,
+      "step": 31100
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.4118242263793945,
+      "learning_rate": 5.139852421438018e-06,
+      "loss": 0.5706,
+      "step": 31101
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.1768810749053955,
+      "learning_rate": 5.1395888662221504e-06,
+      "loss": 0.6204,
+      "step": 31102
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.186204195022583,
+      "learning_rate": 5.13932531061814e-06,
+      "loss": 0.4233,
+      "step": 31103
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.536825656890869,
+      "learning_rate": 5.139061754626716e-06,
+      "loss": 0.5912,
+      "step": 31104
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.98919677734375,
+      "learning_rate": 5.138798198248615e-06,
+      "loss": 0.9351,
+      "step": 31105
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.6737558841705322,
+      "learning_rate": 5.138534641484567e-06,
+      "loss": 0.7092,
+      "step": 31106
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.979439377784729,
+      "learning_rate": 5.138271084335307e-06,
+      "loss": 0.5967,
+      "step": 31107
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.5250096321105957,
+      "learning_rate": 5.138007526801566e-06,
+      "loss": 0.5388,
+      "step": 31108
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.513223648071289,
+      "learning_rate": 5.137743968884078e-06,
+      "loss": 0.5428,
+      "step": 31109
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.1711881160736084,
+      "learning_rate": 5.137480410583576e-06,
+      "loss": 0.7614,
+      "step": 31110
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.0832679271698,
+      "learning_rate": 5.137216851900792e-06,
+      "loss": 0.7091,
+      "step": 31111
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.6459078788757324,
+      "learning_rate": 5.136953292836461e-06,
+      "loss": 0.5727,
+      "step": 31112
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.451667547225952,
+      "learning_rate": 5.1366897333913125e-06,
+      "loss": 0.5915,
+      "step": 31113
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.282130479812622,
+      "learning_rate": 5.136426173566081e-06,
+      "loss": 0.7349,
+      "step": 31114
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 3.1897971630096436,
+      "learning_rate": 5.136162613361501e-06,
+      "loss": 0.7548,
+      "step": 31115
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 5.734198093414307,
+      "learning_rate": 5.135899052778303e-06,
+      "loss": 0.6745,
+      "step": 31116
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.111781120300293,
+      "learning_rate": 5.135635491817219e-06,
+      "loss": 0.6489,
+      "step": 31117
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 4.372147560119629,
+      "learning_rate": 5.135371930478985e-06,
+      "loss": 0.6765,
+      "step": 31118
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 3.049730062484741,
+      "learning_rate": 5.135108368764331e-06,
+      "loss": 0.5876,
+      "step": 31119
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.6273562908172607,
+      "learning_rate": 5.134844806673993e-06,
+      "loss": 0.7733,
+      "step": 31120
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.796970009803772,
+      "learning_rate": 5.134581244208701e-06,
+      "loss": 0.9109,
+      "step": 31121
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.8679455518722534,
+      "learning_rate": 5.13431768136919e-06,
+      "loss": 0.6267,
+      "step": 31122
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 3.155590534210205,
+      "learning_rate": 5.134054118156192e-06,
+      "loss": 0.6724,
+      "step": 31123
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.9722380638122559,
+      "learning_rate": 5.1337905545704395e-06,
+      "loss": 0.7215,
+      "step": 31124
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.6411292552948,
+      "learning_rate": 5.133526990612665e-06,
+      "loss": 0.7866,
+      "step": 31125
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.9359866380691528,
+      "learning_rate": 5.1332634262836024e-06,
+      "loss": 0.4267,
+      "step": 31126
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.3403539657592773,
+      "learning_rate": 5.132999861583983e-06,
+      "loss": 0.5073,
+      "step": 31127
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.973656177520752,
+      "learning_rate": 5.132736296514541e-06,
+      "loss": 0.882,
+      "step": 31128
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.76955246925354,
+      "learning_rate": 5.13247273107601e-06,
+      "loss": 0.7383,
+      "step": 31129
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 3.8632805347442627,
+      "learning_rate": 5.132209165269122e-06,
+      "loss": 0.8172,
+      "step": 31130
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.9943770170211792,
+      "learning_rate": 5.131945599094608e-06,
+      "loss": 0.8181,
+      "step": 31131
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.647495746612549,
+      "learning_rate": 5.131682032553206e-06,
+      "loss": 0.5562,
+      "step": 31132
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.3787598609924316,
+      "learning_rate": 5.1314184656456425e-06,
+      "loss": 0.6429,
+      "step": 31133
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.9343897104263306,
+      "learning_rate": 5.1311548983726544e-06,
+      "loss": 0.4476,
+      "step": 31134
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 3.094236135482788,
+      "learning_rate": 5.130891330734975e-06,
+      "loss": 0.5634,
+      "step": 31135
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.712451457977295,
+      "learning_rate": 5.130627762733334e-06,
+      "loss": 0.5324,
+      "step": 31136
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.6543002128601074,
+      "learning_rate": 5.130364194368468e-06,
+      "loss": 0.6302,
+      "step": 31137
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.7609729766845703,
+      "learning_rate": 5.130100625641107e-06,
+      "loss": 0.7872,
+      "step": 31138
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 3.1985421180725098,
+      "learning_rate": 5.129837056551984e-06,
+      "loss": 0.7506,
+      "step": 31139
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.7619447708129883,
+      "learning_rate": 5.129573487101833e-06,
+      "loss": 0.599,
+      "step": 31140
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 4.4870476722717285,
+      "learning_rate": 5.129309917291388e-06,
+      "loss": 0.5676,
+      "step": 31141
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.8092620372772217,
+      "learning_rate": 5.129046347121381e-06,
+      "loss": 0.4065,
+      "step": 31142
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 3.7336645126342773,
+      "learning_rate": 5.128782776592542e-06,
+      "loss": 0.884,
+      "step": 31143
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.0385401248931885,
+      "learning_rate": 5.128519205705609e-06,
+      "loss": 0.7606,
+      "step": 31144
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.611210584640503,
+      "learning_rate": 5.128255634461311e-06,
+      "loss": 0.6195,
+      "step": 31145
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 4.186048984527588,
+      "learning_rate": 5.127992062860382e-06,
+      "loss": 0.6051,
+      "step": 31146
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.983912944793701,
+      "learning_rate": 5.127728490903555e-06,
+      "loss": 0.9272,
+      "step": 31147
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.7755151987075806,
+      "learning_rate": 5.127464918591564e-06,
+      "loss": 0.7736,
+      "step": 31148
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.586252212524414,
+      "learning_rate": 5.127201345925138e-06,
+      "loss": 0.5766,
+      "step": 31149
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.6577765941619873,
+      "learning_rate": 5.126937772905015e-06,
+      "loss": 0.6036,
+      "step": 31150
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 3.752805233001709,
+      "learning_rate": 5.1266741995319245e-06,
+      "loss": 0.6818,
+      "step": 31151
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.96502685546875,
+      "learning_rate": 5.126410625806601e-06,
+      "loss": 0.685,
+      "step": 31152
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.1207170486450195,
+      "learning_rate": 5.126147051729779e-06,
+      "loss": 0.5414,
+      "step": 31153
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.342653274536133,
+      "learning_rate": 5.125883477302186e-06,
+      "loss": 0.6058,
+      "step": 31154
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.107079029083252,
+      "learning_rate": 5.12561990252456e-06,
+      "loss": 0.5136,
+      "step": 31155
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.0406157970428467,
+      "learning_rate": 5.125356327397632e-06,
+      "loss": 0.4713,
+      "step": 31156
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.0872764587402344,
+      "learning_rate": 5.125092751922135e-06,
+      "loss": 0.8326,
+      "step": 31157
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.3862688541412354,
+      "learning_rate": 5.124829176098801e-06,
+      "loss": 0.6708,
+      "step": 31158
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.7688021659851074,
+      "learning_rate": 5.124565599928364e-06,
+      "loss": 0.5991,
+      "step": 31159
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 3.4922080039978027,
+      "learning_rate": 5.124302023411557e-06,
+      "loss": 0.7311,
+      "step": 31160
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.3355958461761475,
+      "learning_rate": 5.124038446549113e-06,
+      "loss": 0.5898,
+      "step": 31161
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 3.1202242374420166,
+      "learning_rate": 5.123774869341764e-06,
+      "loss": 0.5841,
+      "step": 31162
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.177018880844116,
+      "learning_rate": 5.123511291790244e-06,
+      "loss": 0.6507,
+      "step": 31163
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.848576307296753,
+      "learning_rate": 5.123247713895286e-06,
+      "loss": 0.7477,
+      "step": 31164
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 3.4986307621002197,
+      "learning_rate": 5.122984135657622e-06,
+      "loss": 0.6807,
+      "step": 31165
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.9383773803710938,
+      "learning_rate": 5.122720557077984e-06,
+      "loss": 0.7069,
+      "step": 31166
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.642551064491272,
+      "learning_rate": 5.122456978157107e-06,
+      "loss": 0.8874,
+      "step": 31167
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.933746814727783,
+      "learning_rate": 5.122193398895723e-06,
+      "loss": 0.5707,
+      "step": 31168
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 7.02437686920166,
+      "learning_rate": 5.121929819294563e-06,
+      "loss": 0.6247,
+      "step": 31169
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.9632103443145752,
+      "learning_rate": 5.1216662393543635e-06,
+      "loss": 0.6604,
+      "step": 31170
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.3656773567199707,
+      "learning_rate": 5.121402659075857e-06,
+      "loss": 0.642,
+      "step": 31171
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 3.2240090370178223,
+      "learning_rate": 5.121139078459773e-06,
+      "loss": 0.7211,
+      "step": 31172
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.6102168560028076,
+      "learning_rate": 5.120875497506848e-06,
+      "loss": 0.3379,
+      "step": 31173
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.893115997314453,
+      "learning_rate": 5.120611916217814e-06,
+      "loss": 0.4884,
+      "step": 31174
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.6346691846847534,
+      "learning_rate": 5.120348334593403e-06,
+      "loss": 0.3537,
+      "step": 31175
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.9787774085998535,
+      "learning_rate": 5.120084752634348e-06,
+      "loss": 0.7917,
+      "step": 31176
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.331496000289917,
+      "learning_rate": 5.119821170341381e-06,
+      "loss": 0.6858,
+      "step": 31177
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.755772590637207,
+      "learning_rate": 5.1195575877152385e-06,
+      "loss": 0.5738,
+      "step": 31178
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.155982494354248,
+      "learning_rate": 5.1192940047566496e-06,
+      "loss": 0.7486,
+      "step": 31179
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.9448091983795166,
+      "learning_rate": 5.119030421466349e-06,
+      "loss": 0.6878,
+      "step": 31180
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 3.6821303367614746,
+      "learning_rate": 5.118766837845069e-06,
+      "loss": 0.7286,
+      "step": 31181
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 3.798050880432129,
+      "learning_rate": 5.118503253893544e-06,
+      "loss": 0.614,
+      "step": 31182
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.8995654582977295,
+      "learning_rate": 5.118239669612506e-06,
+      "loss": 0.5836,
+      "step": 31183
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.7681725025177,
+      "learning_rate": 5.117976085002687e-06,
+      "loss": 0.8448,
+      "step": 31184
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.7585325241088867,
+      "learning_rate": 5.117712500064821e-06,
+      "loss": 0.7192,
+      "step": 31185
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.6709682941436768,
+      "learning_rate": 5.1174489147996394e-06,
+      "loss": 0.78,
+      "step": 31186
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.6081793308258057,
+      "learning_rate": 5.117185329207879e-06,
+      "loss": 0.8629,
+      "step": 31187
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.808135986328125,
+      "learning_rate": 5.116921743290267e-06,
+      "loss": 0.6007,
+      "step": 31188
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.937028169631958,
+      "learning_rate": 5.116658157047542e-06,
+      "loss": 0.5396,
+      "step": 31189
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 3.3376100063323975,
+      "learning_rate": 5.116394570480433e-06,
+      "loss": 0.7846,
+      "step": 31190
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 3.240095376968384,
+      "learning_rate": 5.116130983589674e-06,
+      "loss": 0.4171,
+      "step": 31191
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 3.354616403579712,
+      "learning_rate": 5.115867396376e-06,
+      "loss": 0.7019,
+      "step": 31192
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.038586378097534,
+      "learning_rate": 5.115603808840141e-06,
+      "loss": 0.9717,
+      "step": 31193
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.9521557092666626,
+      "learning_rate": 5.11534022098283e-06,
+      "loss": 0.7455,
+      "step": 31194
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.786480188369751,
+      "learning_rate": 5.115076632804802e-06,
+      "loss": 0.578,
+      "step": 31195
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 4.9673380851745605,
+      "learning_rate": 5.114813044306789e-06,
+      "loss": 0.7258,
+      "step": 31196
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.9544992446899414,
+      "learning_rate": 5.114549455489525e-06,
+      "loss": 0.5087,
+      "step": 31197
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 3.5342676639556885,
+      "learning_rate": 5.114285866353741e-06,
+      "loss": 0.5761,
+      "step": 31198
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.233616590499878,
+      "learning_rate": 5.11402227690017e-06,
+      "loss": 0.7929,
+      "step": 31199
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.161139965057373,
+      "learning_rate": 5.113758687129547e-06,
+      "loss": 0.6161,
+      "step": 31200
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.519329309463501,
+      "learning_rate": 5.113495097042602e-06,
+      "loss": 0.6006,
+      "step": 31201
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.0718417167663574,
+      "learning_rate": 5.113231506640072e-06,
+      "loss": 0.6208,
+      "step": 31202
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.9051430225372314,
+      "learning_rate": 5.112967915922686e-06,
+      "loss": 0.5056,
+      "step": 31203
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.7657721042633057,
+      "learning_rate": 5.11270432489118e-06,
+      "loss": 0.5266,
+      "step": 31204
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.5075531005859375,
+      "learning_rate": 5.112440733546284e-06,
+      "loss": 0.6249,
+      "step": 31205
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.692143201828003,
+      "learning_rate": 5.112177141888733e-06,
+      "loss": 0.6534,
+      "step": 31206
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.0782575607299805,
+      "learning_rate": 5.1119135499192595e-06,
+      "loss": 0.5305,
+      "step": 31207
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.905741810798645,
+      "learning_rate": 5.111649957638597e-06,
+      "loss": 0.7324,
+      "step": 31208
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.91279673576355,
+      "learning_rate": 5.111386365047477e-06,
+      "loss": 0.8093,
+      "step": 31209
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.6204419136047363,
+      "learning_rate": 5.1111227721466325e-06,
+      "loss": 0.3973,
+      "step": 31210
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.8264325857162476,
+      "learning_rate": 5.110859178936798e-06,
+      "loss": 0.5889,
+      "step": 31211
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.8363131284713745,
+      "learning_rate": 5.110595585418706e-06,
+      "loss": 0.8265,
+      "step": 31212
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 3.0759806632995605,
+      "learning_rate": 5.110331991593088e-06,
+      "loss": 0.7622,
+      "step": 31213
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.7426505088806152,
+      "learning_rate": 5.110068397460679e-06,
+      "loss": 0.6415,
+      "step": 31214
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 3.691092014312744,
+      "learning_rate": 5.109804803022211e-06,
+      "loss": 0.848,
+      "step": 31215
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.488746404647827,
+      "learning_rate": 5.1095412082784175e-06,
+      "loss": 0.7291,
+      "step": 31216
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.4366880655288696,
+      "learning_rate": 5.10927761323003e-06,
+      "loss": 0.6297,
+      "step": 31217
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.7069896459579468,
+      "learning_rate": 5.109014017877784e-06,
+      "loss": 0.7792,
+      "step": 31218
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 3.0405068397521973,
+      "learning_rate": 5.108750422222408e-06,
+      "loss": 0.6293,
+      "step": 31219
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 3.9912657737731934,
+      "learning_rate": 5.108486826264639e-06,
+      "loss": 0.9193,
+      "step": 31220
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.5914649963378906,
+      "learning_rate": 5.1082232300052095e-06,
+      "loss": 0.7794,
+      "step": 31221
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 3.9055724143981934,
+      "learning_rate": 5.107959633444851e-06,
+      "loss": 0.7797,
+      "step": 31222
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.681901216506958,
+      "learning_rate": 5.1076960365842976e-06,
+      "loss": 0.6278,
+      "step": 31223
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 6.316961765289307,
+      "learning_rate": 5.107432439424282e-06,
+      "loss": 0.7028,
+      "step": 31224
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.691499948501587,
+      "learning_rate": 5.107168841965537e-06,
+      "loss": 0.6706,
+      "step": 31225
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 3.902315616607666,
+      "learning_rate": 5.1069052442087955e-06,
+      "loss": 0.7708,
+      "step": 31226
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 3.4075539112091064,
+      "learning_rate": 5.106641646154791e-06,
+      "loss": 0.7201,
+      "step": 31227
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.2541587352752686,
+      "learning_rate": 5.106378047804256e-06,
+      "loss": 0.5566,
+      "step": 31228
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.5676599740982056,
+      "learning_rate": 5.106114449157923e-06,
+      "loss": 0.4393,
+      "step": 31229
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 3.158863067626953,
+      "learning_rate": 5.1058508502165235e-06,
+      "loss": 0.4863,
+      "step": 31230
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.4272791147232056,
+      "learning_rate": 5.105587250980795e-06,
+      "loss": 0.5583,
+      "step": 31231
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 5.369319915771484,
+      "learning_rate": 5.105323651451467e-06,
+      "loss": 0.3492,
+      "step": 31232
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 3.0231149196624756,
+      "learning_rate": 5.105060051629275e-06,
+      "loss": 0.7242,
+      "step": 31233
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.978677988052368,
+      "learning_rate": 5.104796451514948e-06,
+      "loss": 0.6393,
+      "step": 31234
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 6.0154852867126465,
+      "learning_rate": 5.104532851109223e-06,
+      "loss": 0.5601,
+      "step": 31235
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 3.4502713680267334,
+      "learning_rate": 5.104269250412831e-06,
+      "loss": 0.8782,
+      "step": 31236
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.9850679636001587,
+      "learning_rate": 5.104005649426504e-06,
+      "loss": 0.676,
+      "step": 31237
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.405360460281372,
+      "learning_rate": 5.103742048150978e-06,
+      "loss": 0.6225,
+      "step": 31238
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.560506820678711,
+      "learning_rate": 5.103478446586984e-06,
+      "loss": 0.7334,
+      "step": 31239
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.6940054893493652,
+      "learning_rate": 5.103214844735255e-06,
+      "loss": 0.3974,
+      "step": 31240
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.137615919113159,
+      "learning_rate": 5.102951242596522e-06,
+      "loss": 0.5395,
+      "step": 31241
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.040471315383911,
+      "learning_rate": 5.102687640171523e-06,
+      "loss": 0.6308,
+      "step": 31242
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 3.2070841789245605,
+      "learning_rate": 5.102424037460987e-06,
+      "loss": 0.6781,
+      "step": 31243
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 4.3477020263671875,
+      "learning_rate": 5.102160434465648e-06,
+      "loss": 0.7122,
+      "step": 31244
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.757237434387207,
+      "learning_rate": 5.101896831186239e-06,
+      "loss": 0.7072,
+      "step": 31245
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 3.3231399059295654,
+      "learning_rate": 5.101633227623495e-06,
+      "loss": 0.744,
+      "step": 31246
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.248492956161499,
+      "learning_rate": 5.101369623778145e-06,
+      "loss": 0.516,
+      "step": 31247
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.838178873062134,
+      "learning_rate": 5.101106019650925e-06,
+      "loss": 0.7308,
+      "step": 31248
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 3.2339234352111816,
+      "learning_rate": 5.100842415242566e-06,
+      "loss": 0.6093,
+      "step": 31249
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.699361562728882,
+      "learning_rate": 5.100578810553802e-06,
+      "loss": 0.8621,
+      "step": 31250
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 3.615999460220337,
+      "learning_rate": 5.100315205585367e-06,
+      "loss": 0.9303,
+      "step": 31251
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.4686238765716553,
+      "learning_rate": 5.100051600337992e-06,
+      "loss": 0.5847,
+      "step": 31252
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.7340433597564697,
+      "learning_rate": 5.0997879948124115e-06,
+      "loss": 0.5634,
+      "step": 31253
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 4.359885215759277,
+      "learning_rate": 5.099524389009359e-06,
+      "loss": 0.6328,
+      "step": 31254
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.277331590652466,
+      "learning_rate": 5.099260782929565e-06,
+      "loss": 0.6985,
+      "step": 31255
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.7637455463409424,
+      "learning_rate": 5.098997176573764e-06,
+      "loss": 0.5598,
+      "step": 31256
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.5543601512908936,
+      "learning_rate": 5.09873356994269e-06,
+      "loss": 0.5939,
+      "step": 31257
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 12.635486602783203,
+      "learning_rate": 5.098469963037074e-06,
+      "loss": 0.6822,
+      "step": 31258
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.6639597415924072,
+      "learning_rate": 5.09820635585765e-06,
+      "loss": 0.6575,
+      "step": 31259
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 3.3203017711639404,
+      "learning_rate": 5.0979427484051515e-06,
+      "loss": 0.9379,
+      "step": 31260
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 3.3763210773468018,
+      "learning_rate": 5.09767914068031e-06,
+      "loss": 0.5305,
+      "step": 31261
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.8484432697296143,
+      "learning_rate": 5.097415532683859e-06,
+      "loss": 0.7849,
+      "step": 31262
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 3.442100763320923,
+      "learning_rate": 5.097151924416533e-06,
+      "loss": 0.4329,
+      "step": 31263
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.918313980102539,
+      "learning_rate": 5.096888315879065e-06,
+      "loss": 0.8772,
+      "step": 31264
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.121026039123535,
+      "learning_rate": 5.096624707072185e-06,
+      "loss": 0.7225,
+      "step": 31265
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.7351276874542236,
+      "learning_rate": 5.096361097996629e-06,
+      "loss": 0.5373,
+      "step": 31266
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 4.095761775970459,
+      "learning_rate": 5.096097488653129e-06,
+      "loss": 0.7111,
+      "step": 31267
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 3.0820705890655518,
+      "learning_rate": 5.095833879042417e-06,
+      "loss": 0.8765,
+      "step": 31268
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 3.2437241077423096,
+      "learning_rate": 5.095570269165228e-06,
+      "loss": 0.5361,
+      "step": 31269
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.7161470651626587,
+      "learning_rate": 5.095306659022292e-06,
+      "loss": 0.7271,
+      "step": 31270
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.282349109649658,
+      "learning_rate": 5.095043048614344e-06,
+      "loss": 0.7908,
+      "step": 31271
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.903372287750244,
+      "learning_rate": 5.094779437942118e-06,
+      "loss": 0.5954,
+      "step": 31272
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.471367359161377,
+      "learning_rate": 5.094515827006347e-06,
+      "loss": 0.521,
+      "step": 31273
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.650603771209717,
+      "learning_rate": 5.094252215807762e-06,
+      "loss": 0.5718,
+      "step": 31274
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 5.387938022613525,
+      "learning_rate": 5.093988604347095e-06,
+      "loss": 0.5406,
+      "step": 31275
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 4.186703205108643,
+      "learning_rate": 5.093724992625083e-06,
+      "loss": 0.7016,
+      "step": 31276
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.4609763622283936,
+      "learning_rate": 5.093461380642456e-06,
+      "loss": 0.5267,
+      "step": 31277
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.659363031387329,
+      "learning_rate": 5.093197768399949e-06,
+      "loss": 0.3924,
+      "step": 31278
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.6920053958892822,
+      "learning_rate": 5.092934155898292e-06,
+      "loss": 0.555,
+      "step": 31279
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.00403094291687,
+      "learning_rate": 5.092670543138221e-06,
+      "loss": 0.9627,
+      "step": 31280
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.4095606803894043,
+      "learning_rate": 5.0924069301204685e-06,
+      "loss": 0.6329,
+      "step": 31281
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 3.6423070430755615,
+      "learning_rate": 5.0921433168457666e-06,
+      "loss": 0.538,
+      "step": 31282
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.4386656284332275,
+      "learning_rate": 5.091879703314848e-06,
+      "loss": 0.7203,
+      "step": 31283
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.4833385944366455,
+      "learning_rate": 5.091616089528447e-06,
+      "loss": 0.8068,
+      "step": 31284
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.0919137001037598,
+      "learning_rate": 5.091352475487297e-06,
+      "loss": 0.7293,
+      "step": 31285
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.9965749979019165,
+      "learning_rate": 5.091088861192129e-06,
+      "loss": 0.9627,
+      "step": 31286
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.3367607593536377,
+      "learning_rate": 5.0908252466436755e-06,
+      "loss": 0.6614,
+      "step": 31287
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.643226385116577,
+      "learning_rate": 5.0905616318426735e-06,
+      "loss": 0.6706,
+      "step": 31288
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.8438568115234375,
+      "learning_rate": 5.090298016789852e-06,
+      "loss": 0.4898,
+      "step": 31289
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 3.0270073413848877,
+      "learning_rate": 5.090034401485946e-06,
+      "loss": 0.7288,
+      "step": 31290
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 5.935880184173584,
+      "learning_rate": 5.089770785931687e-06,
+      "loss": 0.7447,
+      "step": 31291
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.5096917152404785,
+      "learning_rate": 5.08950717012781e-06,
+      "loss": 0.6161,
+      "step": 31292
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.1302382946014404,
+      "learning_rate": 5.089243554075048e-06,
+      "loss": 0.7181,
+      "step": 31293
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.3002145290374756,
+      "learning_rate": 5.0889799377741325e-06,
+      "loss": 0.3827,
+      "step": 31294
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.3345947265625,
+      "learning_rate": 5.0887163212257964e-06,
+      "loss": 0.5286,
+      "step": 31295
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 3.473008394241333,
+      "learning_rate": 5.088452704430774e-06,
+      "loss": 0.5275,
+      "step": 31296
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.902764916419983,
+      "learning_rate": 5.0881890873897975e-06,
+      "loss": 0.4233,
+      "step": 31297
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.540210723876953,
+      "learning_rate": 5.087925470103601e-06,
+      "loss": 0.5418,
+      "step": 31298
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.7030415534973145,
+      "learning_rate": 5.087661852572916e-06,
+      "loss": 0.4101,
+      "step": 31299
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 3.4940850734710693,
+      "learning_rate": 5.087398234798476e-06,
+      "loss": 0.5911,
+      "step": 31300
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.234732151031494,
+      "learning_rate": 5.087134616781015e-06,
+      "loss": 0.6384,
+      "step": 31301
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 3.3586320877075195,
+      "learning_rate": 5.086870998521264e-06,
+      "loss": 0.7241,
+      "step": 31302
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.541909098625183,
+      "learning_rate": 5.086607380019958e-06,
+      "loss": 0.5625,
+      "step": 31303
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.5013484954833984,
+      "learning_rate": 5.08634376127783e-06,
+      "loss": 0.4711,
+      "step": 31304
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.571342945098877,
+      "learning_rate": 5.086080142295612e-06,
+      "loss": 0.6059,
+      "step": 31305
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.668212413787842,
+      "learning_rate": 5.085816523074036e-06,
+      "loss": 0.8635,
+      "step": 31306
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 3.624368906021118,
+      "learning_rate": 5.085552903613839e-06,
+      "loss": 0.6022,
+      "step": 31307
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.1384525299072266,
+      "learning_rate": 5.08528928391575e-06,
+      "loss": 0.5999,
+      "step": 31308
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.4195548295974731,
+      "learning_rate": 5.0850256639805025e-06,
+      "loss": 0.4377,
+      "step": 31309
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.24255633354187,
+      "learning_rate": 5.084762043808832e-06,
+      "loss": 0.4618,
+      "step": 31310
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.3311684131622314,
+      "learning_rate": 5.084498423401469e-06,
+      "loss": 0.5065,
+      "step": 31311
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 3.4228386878967285,
+      "learning_rate": 5.084234802759147e-06,
+      "loss": 0.4784,
+      "step": 31312
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 3.553895950317383,
+      "learning_rate": 5.083971181882601e-06,
+      "loss": 0.6107,
+      "step": 31313
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.4887313842773438,
+      "learning_rate": 5.083707560772563e-06,
+      "loss": 0.7977,
+      "step": 31314
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.705514430999756,
+      "learning_rate": 5.083443939429764e-06,
+      "loss": 0.9497,
+      "step": 31315
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 3.8054797649383545,
+      "learning_rate": 5.0831803178549395e-06,
+      "loss": 0.5617,
+      "step": 31316
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 3.5601255893707275,
+      "learning_rate": 5.082916696048822e-06,
+      "loss": 0.8036,
+      "step": 31317
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.7855453491210938,
+      "learning_rate": 5.0826530740121425e-06,
+      "loss": 0.6507,
+      "step": 31318
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.9123769998550415,
+      "learning_rate": 5.082389451745637e-06,
+      "loss": 0.5775,
+      "step": 31319
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.590494394302368,
+      "learning_rate": 5.082125829250038e-06,
+      "loss": 0.4446,
+      "step": 31320
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.668438196182251,
+      "learning_rate": 5.081862206526077e-06,
+      "loss": 0.4533,
+      "step": 31321
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.028611421585083,
+      "learning_rate": 5.0815985835744885e-06,
+      "loss": 0.5871,
+      "step": 31322
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 3.095494508743286,
+      "learning_rate": 5.081334960396004e-06,
+      "loss": 0.5642,
+      "step": 31323
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.7411983013153076,
+      "learning_rate": 5.081071336991358e-06,
+      "loss": 0.4698,
+      "step": 31324
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.882041573524475,
+      "learning_rate": 5.080807713361283e-06,
+      "loss": 0.7114,
+      "step": 31325
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 6.672456741333008,
+      "learning_rate": 5.080544089506511e-06,
+      "loss": 0.5861,
+      "step": 31326
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.6552096605300903,
+      "learning_rate": 5.080280465427778e-06,
+      "loss": 0.6289,
+      "step": 31327
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 3.151071071624756,
+      "learning_rate": 5.080016841125814e-06,
+      "loss": 0.46,
+      "step": 31328
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.229398250579834,
+      "learning_rate": 5.079753216601353e-06,
+      "loss": 0.5553,
+      "step": 31329
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 4.99466609954834,
+      "learning_rate": 5.079489591855129e-06,
+      "loss": 0.779,
+      "step": 31330
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.4486093521118164,
+      "learning_rate": 5.079225966887874e-06,
+      "loss": 0.7517,
+      "step": 31331
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 5.613972187042236,
+      "learning_rate": 5.07896234170032e-06,
+      "loss": 0.5971,
+      "step": 31332
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 4.046144008636475,
+      "learning_rate": 5.078698716293202e-06,
+      "loss": 0.7598,
+      "step": 31333
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.2815144062042236,
+      "learning_rate": 5.078435090667253e-06,
+      "loss": 0.7037,
+      "step": 31334
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.947857618331909,
+      "learning_rate": 5.078171464823205e-06,
+      "loss": 0.3674,
+      "step": 31335
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.5458617210388184,
+      "learning_rate": 5.077907838761792e-06,
+      "loss": 0.5415,
+      "step": 31336
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 3.0176446437835693,
+      "learning_rate": 5.077644212483746e-06,
+      "loss": 0.3911,
+      "step": 31337
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 3.3946995735168457,
+      "learning_rate": 5.0773805859897996e-06,
+      "loss": 0.5956,
+      "step": 31338
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.9265109300613403,
+      "learning_rate": 5.0771169592806875e-06,
+      "loss": 0.5974,
+      "step": 31339
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 4.243945598602295,
+      "learning_rate": 5.076853332357142e-06,
+      "loss": 0.6297,
+      "step": 31340
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.867896795272827,
+      "learning_rate": 5.0765897052198974e-06,
+      "loss": 0.6478,
+      "step": 31341
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.897239089012146,
+      "learning_rate": 5.076326077869683e-06,
+      "loss": 0.6913,
+      "step": 31342
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 4.378599166870117,
+      "learning_rate": 5.076062450307236e-06,
+      "loss": 0.5701,
+      "step": 31343
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.0534701347351074,
+      "learning_rate": 5.075798822533288e-06,
+      "loss": 0.4942,
+      "step": 31344
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 6.965412139892578,
+      "learning_rate": 5.075535194548572e-06,
+      "loss": 0.6178,
+      "step": 31345
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.69692063331604,
+      "learning_rate": 5.07527156635382e-06,
+      "loss": 0.4913,
+      "step": 31346
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 5.488050937652588,
+      "learning_rate": 5.075007937949766e-06,
+      "loss": 0.7544,
+      "step": 31347
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 3.1144073009490967,
+      "learning_rate": 5.074744309337144e-06,
+      "loss": 0.7555,
+      "step": 31348
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 3.0859951972961426,
+      "learning_rate": 5.0744806805166855e-06,
+      "loss": 0.6593,
+      "step": 31349
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.0988497734069824,
+      "learning_rate": 5.074217051489124e-06,
+      "loss": 0.6596,
+      "step": 31350
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.899824380874634,
+      "learning_rate": 5.073953422255193e-06,
+      "loss": 0.6037,
+      "step": 31351
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.8803253173828125,
+      "learning_rate": 5.073689792815625e-06,
+      "loss": 0.5607,
+      "step": 31352
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.942614793777466,
+      "learning_rate": 5.073426163171152e-06,
+      "loss": 0.421,
+      "step": 31353
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.816804885864258,
+      "learning_rate": 5.073162533322511e-06,
+      "loss": 0.4506,
+      "step": 31354
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.7124717235565186,
+      "learning_rate": 5.0728989032704315e-06,
+      "loss": 0.533,
+      "step": 31355
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.3927903175354004,
+      "learning_rate": 5.0726352730156454e-06,
+      "loss": 0.6683,
+      "step": 31356
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.4776337146759033,
+      "learning_rate": 5.072371642558891e-06,
+      "loss": 0.4609,
+      "step": 31357
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.3219408988952637,
+      "learning_rate": 5.072108011900896e-06,
+      "loss": 0.689,
+      "step": 31358
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.0853209495544434,
+      "learning_rate": 5.071844381042397e-06,
+      "loss": 0.9693,
+      "step": 31359
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.2653021812438965,
+      "learning_rate": 5.071580749984124e-06,
+      "loss": 0.4181,
+      "step": 31360
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.616249918937683,
+      "learning_rate": 5.071317118726813e-06,
+      "loss": 0.4166,
+      "step": 31361
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.9370245933532715,
+      "learning_rate": 5.071053487271196e-06,
+      "loss": 0.5596,
+      "step": 31362
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.7564897537231445,
+      "learning_rate": 5.0707898556180045e-06,
+      "loss": 0.5071,
+      "step": 31363
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.7901729345321655,
+      "learning_rate": 5.070526223767974e-06,
+      "loss": 0.7118,
+      "step": 31364
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.8547919988632202,
+      "learning_rate": 5.070262591721836e-06,
+      "loss": 0.592,
+      "step": 31365
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 3.6033055782318115,
+      "learning_rate": 5.069998959480324e-06,
+      "loss": 0.4314,
+      "step": 31366
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.3463852405548096,
+      "learning_rate": 5.069735327044172e-06,
+      "loss": 0.6809,
+      "step": 31367
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.163630485534668,
+      "learning_rate": 5.069471694414112e-06,
+      "loss": 0.5015,
+      "step": 31368
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.526188611984253,
+      "learning_rate": 5.069208061590877e-06,
+      "loss": 0.5753,
+      "step": 31369
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 3.5417253971099854,
+      "learning_rate": 5.0689444285752e-06,
+      "loss": 0.7545,
+      "step": 31370
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 3.605394124984741,
+      "learning_rate": 5.068680795367815e-06,
+      "loss": 0.4649,
+      "step": 31371
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.486554741859436,
+      "learning_rate": 5.068417161969454e-06,
+      "loss": 0.5837,
+      "step": 31372
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.84365177154541,
+      "learning_rate": 5.068153528380849e-06,
+      "loss": 0.6649,
+      "step": 31373
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 3.030343532562256,
+      "learning_rate": 5.067889894602737e-06,
+      "loss": 0.6591,
+      "step": 31374
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.094322681427002,
+      "learning_rate": 5.067626260635848e-06,
+      "loss": 0.3608,
+      "step": 31375
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.2405097484588623,
+      "learning_rate": 5.067362626480915e-06,
+      "loss": 0.6473,
+      "step": 31376
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.31415057182312,
+      "learning_rate": 5.067098992138673e-06,
+      "loss": 0.7305,
+      "step": 31377
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.0922610759735107,
+      "learning_rate": 5.0668353576098525e-06,
+      "loss": 0.699,
+      "step": 31378
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.3374223709106445,
+      "learning_rate": 5.066571722895187e-06,
+      "loss": 0.586,
+      "step": 31379
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 4.130370616912842,
+      "learning_rate": 5.066308087995412e-06,
+      "loss": 0.5557,
+      "step": 31380
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.135186195373535,
+      "learning_rate": 5.066044452911259e-06,
+      "loss": 0.3167,
+      "step": 31381
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.666214942932129,
+      "learning_rate": 5.065780817643462e-06,
+      "loss": 0.7112,
+      "step": 31382
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.7964165210723877,
+      "learning_rate": 5.0655171821927516e-06,
+      "loss": 0.6055,
+      "step": 31383
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.46018648147583,
+      "learning_rate": 5.065253546559863e-06,
+      "loss": 0.8638,
+      "step": 31384
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 4.086038112640381,
+      "learning_rate": 5.064989910745528e-06,
+      "loss": 0.6201,
+      "step": 31385
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 3.0112557411193848,
+      "learning_rate": 5.064726274750481e-06,
+      "loss": 1.0072,
+      "step": 31386
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 3.7990458011627197,
+      "learning_rate": 5.064462638575456e-06,
+      "loss": 0.7138,
+      "step": 31387
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.967008113861084,
+      "learning_rate": 5.064199002221183e-06,
+      "loss": 0.7881,
+      "step": 31388
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.640540361404419,
+      "learning_rate": 5.063935365688397e-06,
+      "loss": 0.6292,
+      "step": 31389
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 3.1717660427093506,
+      "learning_rate": 5.06367172897783e-06,
+      "loss": 0.5865,
+      "step": 31390
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.6623058319091797,
+      "learning_rate": 5.063408092090217e-06,
+      "loss": 0.5698,
+      "step": 31391
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.358710765838623,
+      "learning_rate": 5.06314445502629e-06,
+      "loss": 0.6531,
+      "step": 31392
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.851674199104309,
+      "learning_rate": 5.06288081778678e-06,
+      "loss": 0.5609,
+      "step": 31393
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 3.1553707122802734,
+      "learning_rate": 5.062617180372422e-06,
+      "loss": 0.3546,
+      "step": 31394
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.764824390411377,
+      "learning_rate": 5.062353542783951e-06,
+      "loss": 0.6884,
+      "step": 31395
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 5.2723388671875,
+      "learning_rate": 5.062089905022097e-06,
+      "loss": 0.6674,
+      "step": 31396
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 4.1241841316223145,
+      "learning_rate": 5.061826267087595e-06,
+      "loss": 0.4493,
+      "step": 31397
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.365004062652588,
+      "learning_rate": 5.061562628981177e-06,
+      "loss": 0.5291,
+      "step": 31398
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.9198710918426514,
+      "learning_rate": 5.061298990703577e-06,
+      "loss": 0.7089,
+      "step": 31399
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.500661849975586,
+      "learning_rate": 5.061035352255527e-06,
+      "loss": 0.5446,
+      "step": 31400
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.4192187786102295,
+      "learning_rate": 5.06077171363776e-06,
+      "loss": 0.5265,
+      "step": 31401
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.719163179397583,
+      "learning_rate": 5.06050807485101e-06,
+      "loss": 0.5847,
+      "step": 31402
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.1896262168884277,
+      "learning_rate": 5.060244435896009e-06,
+      "loss": 0.6932,
+      "step": 31403
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 3.6501920223236084,
+      "learning_rate": 5.059980796773492e-06,
+      "loss": 0.6379,
+      "step": 31404
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.7969971895217896,
+      "learning_rate": 5.05971715748419e-06,
+      "loss": 0.5966,
+      "step": 31405
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 3.9729857444763184,
+      "learning_rate": 5.059453518028837e-06,
+      "loss": 0.5,
+      "step": 31406
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 3.2909905910491943,
+      "learning_rate": 5.059189878408166e-06,
+      "loss": 0.7484,
+      "step": 31407
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.0530200004577637,
+      "learning_rate": 5.058926238622911e-06,
+      "loss": 0.6328,
+      "step": 31408
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.187145233154297,
+      "learning_rate": 5.058662598673803e-06,
+      "loss": 0.65,
+      "step": 31409
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.6459641456604004,
+      "learning_rate": 5.058398958561578e-06,
+      "loss": 0.4501,
+      "step": 31410
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.658477544784546,
+      "learning_rate": 5.058135318286967e-06,
+      "loss": 0.6566,
+      "step": 31411
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.7112183570861816,
+      "learning_rate": 5.057871677850702e-06,
+      "loss": 0.332,
+      "step": 31412
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.068958044052124,
+      "learning_rate": 5.057608037253518e-06,
+      "loss": 0.8119,
+      "step": 31413
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 4.397372245788574,
+      "learning_rate": 5.057344396496147e-06,
+      "loss": 0.4737,
+      "step": 31414
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.423781394958496,
+      "learning_rate": 5.057080755579325e-06,
+      "loss": 0.5239,
+      "step": 31415
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.697091579437256,
+      "learning_rate": 5.056817114503782e-06,
+      "loss": 0.6987,
+      "step": 31416
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 3.0618906021118164,
+      "learning_rate": 5.05655347327025e-06,
+      "loss": 0.5571,
+      "step": 31417
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.5227550268173218,
+      "learning_rate": 5.056289831879466e-06,
+      "loss": 0.6921,
+      "step": 31418
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.448315143585205,
+      "learning_rate": 5.0560261903321605e-06,
+      "loss": 0.6155,
+      "step": 31419
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.426520824432373,
+      "learning_rate": 5.055762548629068e-06,
+      "loss": 0.7569,
+      "step": 31420
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 3.93853759765625,
+      "learning_rate": 5.05549890677092e-06,
+      "loss": 0.7538,
+      "step": 31421
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.3228089809417725,
+      "learning_rate": 5.055235264758449e-06,
+      "loss": 0.5445,
+      "step": 31422
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.21055006980896,
+      "learning_rate": 5.054971622592391e-06,
+      "loss": 0.565,
+      "step": 31423
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.7824676036834717,
+      "learning_rate": 5.054707980273475e-06,
+      "loss": 0.7395,
+      "step": 31424
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 4.372676372528076,
+      "learning_rate": 5.05444433780244e-06,
+      "loss": 1.0359,
+      "step": 31425
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.0411598682403564,
+      "learning_rate": 5.054180695180015e-06,
+      "loss": 0.6789,
+      "step": 31426
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.833923578262329,
+      "learning_rate": 5.053917052406933e-06,
+      "loss": 0.6636,
+      "step": 31427
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.9901784658432007,
+      "learning_rate": 5.053653409483927e-06,
+      "loss": 0.6356,
+      "step": 31428
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.9161386489868164,
+      "learning_rate": 5.053389766411733e-06,
+      "loss": 0.834,
+      "step": 31429
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.8534297943115234,
+      "learning_rate": 5.053126123191081e-06,
+      "loss": 0.6211,
+      "step": 31430
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.0240700244903564,
+      "learning_rate": 5.052862479822705e-06,
+      "loss": 0.6652,
+      "step": 31431
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.558002233505249,
+      "learning_rate": 5.052598836307338e-06,
+      "loss": 0.5749,
+      "step": 31432
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 3.8066036701202393,
+      "learning_rate": 5.052335192645714e-06,
+      "loss": 0.6569,
+      "step": 31433
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 5.095291614532471,
+      "learning_rate": 5.052071548838564e-06,
+      "loss": 0.6517,
+      "step": 31434
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.8428153991699219,
+      "learning_rate": 5.0518079048866245e-06,
+      "loss": 0.5942,
+      "step": 31435
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.9595447778701782,
+      "learning_rate": 5.051544260790626e-06,
+      "loss": 0.6328,
+      "step": 31436
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.7057130336761475,
+      "learning_rate": 5.051280616551302e-06,
+      "loss": 0.6527,
+      "step": 31437
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.3735947608947754,
+      "learning_rate": 5.051016972169386e-06,
+      "loss": 0.6273,
+      "step": 31438
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.876593589782715,
+      "learning_rate": 5.050753327645611e-06,
+      "loss": 0.6297,
+      "step": 31439
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.9621614217758179,
+      "learning_rate": 5.050489682980709e-06,
+      "loss": 0.4478,
+      "step": 31440
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 3.7365548610687256,
+      "learning_rate": 5.050226038175414e-06,
+      "loss": 0.393,
+      "step": 31441
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.0371592044830322,
+      "learning_rate": 5.049962393230461e-06,
+      "loss": 0.5755,
+      "step": 31442
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.311607599258423,
+      "learning_rate": 5.049698748146581e-06,
+      "loss": 0.6127,
+      "step": 31443
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.041370153427124,
+      "learning_rate": 5.049435102924506e-06,
+      "loss": 0.7189,
+      "step": 31444
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 3.1830005645751953,
+      "learning_rate": 5.049171457564972e-06,
+      "loss": 0.5447,
+      "step": 31445
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.4942426681518555,
+      "learning_rate": 5.04890781206871e-06,
+      "loss": 0.5272,
+      "step": 31446
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 3.93963623046875,
+      "learning_rate": 5.048644166436453e-06,
+      "loss": 0.9445,
+      "step": 31447
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.144428253173828,
+      "learning_rate": 5.048380520668935e-06,
+      "loss": 0.6108,
+      "step": 31448
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.968672752380371,
+      "learning_rate": 5.048116874766889e-06,
+      "loss": 0.5867,
+      "step": 31449
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 3.317734956741333,
+      "learning_rate": 5.047853228731049e-06,
+      "loss": 0.8185,
+      "step": 31450
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.6760430335998535,
+      "learning_rate": 5.0475895825621456e-06,
+      "loss": 0.7042,
+      "step": 31451
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.163820505142212,
+      "learning_rate": 5.047325936260915e-06,
+      "loss": 0.4517,
+      "step": 31452
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 3.0756113529205322,
+      "learning_rate": 5.047062289828088e-06,
+      "loss": 0.6098,
+      "step": 31453
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.974635601043701,
+      "learning_rate": 5.046798643264397e-06,
+      "loss": 0.4803,
+      "step": 31454
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.7631468772888184,
+      "learning_rate": 5.046534996570578e-06,
+      "loss": 0.603,
+      "step": 31455
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.760347843170166,
+      "learning_rate": 5.046271349747362e-06,
+      "loss": 0.7752,
+      "step": 31456
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 4.254865646362305,
+      "learning_rate": 5.046007702795483e-06,
+      "loss": 0.8755,
+      "step": 31457
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.735264778137207,
+      "learning_rate": 5.045744055715674e-06,
+      "loss": 0.6612,
+      "step": 31458
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.0381455421447754,
+      "learning_rate": 5.045480408508668e-06,
+      "loss": 0.7745,
+      "step": 31459
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.792142629623413,
+      "learning_rate": 5.045216761175199e-06,
+      "loss": 0.4261,
+      "step": 31460
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.145331621170044,
+      "learning_rate": 5.044953113715998e-06,
+      "loss": 0.4487,
+      "step": 31461
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.9538956880569458,
+      "learning_rate": 5.044689466131799e-06,
+      "loss": 0.655,
+      "step": 31462
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.8287757635116577,
+      "learning_rate": 5.044425818423337e-06,
+      "loss": 0.5577,
+      "step": 31463
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.9643330574035645,
+      "learning_rate": 5.0441621705913405e-06,
+      "loss": 0.5185,
+      "step": 31464
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 3.4983720779418945,
+      "learning_rate": 5.0438985226365466e-06,
+      "loss": 0.4626,
+      "step": 31465
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.8363869190216064,
+      "learning_rate": 5.0436348745596876e-06,
+      "loss": 0.4868,
+      "step": 31466
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.4552695751190186,
+      "learning_rate": 5.043371226361496e-06,
+      "loss": 0.4619,
+      "step": 31467
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 7.239287376403809,
+      "learning_rate": 5.043107578042706e-06,
+      "loss": 0.7258,
+      "step": 31468
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.9857277870178223,
+      "learning_rate": 5.042843929604051e-06,
+      "loss": 0.6916,
+      "step": 31469
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.2813925743103027,
+      "learning_rate": 5.042580281046262e-06,
+      "loss": 0.6512,
+      "step": 31470
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.8561794757843018,
+      "learning_rate": 5.042316632370073e-06,
+      "loss": 0.6588,
+      "step": 31471
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 4.441668510437012,
+      "learning_rate": 5.042052983576218e-06,
+      "loss": 0.6098,
+      "step": 31472
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 5.895129680633545,
+      "learning_rate": 5.041789334665428e-06,
+      "loss": 0.4862,
+      "step": 31473
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 4.281916618347168,
+      "learning_rate": 5.041525685638437e-06,
+      "loss": 0.7231,
+      "step": 31474
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.2874969244003296,
+      "learning_rate": 5.04126203649598e-06,
+      "loss": 0.3602,
+      "step": 31475
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.284675121307373,
+      "learning_rate": 5.0409983872387895e-06,
+      "loss": 0.4423,
+      "step": 31476
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.445612668991089,
+      "learning_rate": 5.040734737867597e-06,
+      "loss": 0.6483,
+      "step": 31477
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.060370445251465,
+      "learning_rate": 5.040471088383136e-06,
+      "loss": 0.518,
+      "step": 31478
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.3180599212646484,
+      "learning_rate": 5.040207438786141e-06,
+      "loss": 0.7786,
+      "step": 31479
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.961592435836792,
+      "learning_rate": 5.0399437890773436e-06,
+      "loss": 0.7933,
+      "step": 31480
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.2365610599517822,
+      "learning_rate": 5.039680139257478e-06,
+      "loss": 0.6565,
+      "step": 31481
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.0688233375549316,
+      "learning_rate": 5.039416489327277e-06,
+      "loss": 0.4702,
+      "step": 31482
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 18.882553100585938,
+      "learning_rate": 5.0391528392874725e-06,
+      "loss": 0.5848,
+      "step": 31483
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.7277733087539673,
+      "learning_rate": 5.038889189138799e-06,
+      "loss": 0.5672,
+      "step": 31484
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.5828232765197754,
+      "learning_rate": 5.038625538881988e-06,
+      "loss": 0.5847,
+      "step": 31485
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 3.53291916847229,
+      "learning_rate": 5.038361888517776e-06,
+      "loss": 0.5445,
+      "step": 31486
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 3.154764175415039,
+      "learning_rate": 5.0380982380468924e-06,
+      "loss": 0.6502,
+      "step": 31487
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.8789591789245605,
+      "learning_rate": 5.037834587470073e-06,
+      "loss": 0.8399,
+      "step": 31488
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.7288472652435303,
+      "learning_rate": 5.037570936788048e-06,
+      "loss": 0.5651,
+      "step": 31489
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 3.112332344055176,
+      "learning_rate": 5.037307286001555e-06,
+      "loss": 0.5037,
+      "step": 31490
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 3.4382288455963135,
+      "learning_rate": 5.037043635111322e-06,
+      "loss": 0.7225,
+      "step": 31491
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.404999256134033,
+      "learning_rate": 5.036779984118087e-06,
+      "loss": 0.6378,
+      "step": 31492
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 7.113224029541016,
+      "learning_rate": 5.036516333022578e-06,
+      "loss": 0.6669,
+      "step": 31493
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 4.592696666717529,
+      "learning_rate": 5.036252681825533e-06,
+      "loss": 0.9024,
+      "step": 31494
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 4.506311893463135,
+      "learning_rate": 5.03598903052768e-06,
+      "loss": 0.5501,
+      "step": 31495
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.9385862350463867,
+      "learning_rate": 5.035725379129758e-06,
+      "loss": 0.5931,
+      "step": 31496
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 3.4846346378326416,
+      "learning_rate": 5.035461727632496e-06,
+      "loss": 0.7264,
+      "step": 31497
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.431501865386963,
+      "learning_rate": 5.035198076036628e-06,
+      "loss": 0.7707,
+      "step": 31498
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 3.1105849742889404,
+      "learning_rate": 5.034934424342888e-06,
+      "loss": 0.4477,
+      "step": 31499
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 3.4775359630584717,
+      "learning_rate": 5.034670772552007e-06,
+      "loss": 0.5978,
+      "step": 31500
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.6958420276641846,
+      "learning_rate": 5.034407120664722e-06,
+      "loss": 0.6221,
+      "step": 31501
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.762216091156006,
+      "learning_rate": 5.034143468681761e-06,
+      "loss": 0.5021,
+      "step": 31502
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.3130691051483154,
+      "learning_rate": 5.033879816603861e-06,
+      "loss": 0.5384,
+      "step": 31503
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.623699903488159,
+      "learning_rate": 5.033616164431754e-06,
+      "loss": 0.7219,
+      "step": 31504
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 3.9117114543914795,
+      "learning_rate": 5.033352512166172e-06,
+      "loss": 0.5478,
+      "step": 31505
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.6545488834381104,
+      "learning_rate": 5.0330888598078505e-06,
+      "loss": 0.6397,
+      "step": 31506
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 3.5506908893585205,
+      "learning_rate": 5.0328252073575216e-06,
+      "loss": 0.647,
+      "step": 31507
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.772035598754883,
+      "learning_rate": 5.032561554815917e-06,
+      "loss": 0.8936,
+      "step": 31508
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.967048168182373,
+      "learning_rate": 5.03229790218377e-06,
+      "loss": 0.5384,
+      "step": 31509
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 3.011776924133301,
+      "learning_rate": 5.032034249461815e-06,
+      "loss": 0.7462,
+      "step": 31510
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.813323497772217,
+      "learning_rate": 5.031770596650787e-06,
+      "loss": 0.5371,
+      "step": 31511
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 3.415822982788086,
+      "learning_rate": 5.031506943751415e-06,
+      "loss": 0.6296,
+      "step": 31512
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 3.0361297130584717,
+      "learning_rate": 5.031243290764435e-06,
+      "loss": 0.5209,
+      "step": 31513
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 3.1551880836486816,
+      "learning_rate": 5.030979637690579e-06,
+      "loss": 0.9496,
+      "step": 31514
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.9247499704360962,
+      "learning_rate": 5.030715984530578e-06,
+      "loss": 0.7373,
+      "step": 31515
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.393132209777832,
+      "learning_rate": 5.030452331285171e-06,
+      "loss": 0.5375,
+      "step": 31516
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.235661268234253,
+      "learning_rate": 5.030188677955085e-06,
+      "loss": 0.9139,
+      "step": 31517
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.3139383792877197,
+      "learning_rate": 5.029925024541056e-06,
+      "loss": 0.5488,
+      "step": 31518
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 4.843389987945557,
+      "learning_rate": 5.029661371043817e-06,
+      "loss": 0.6292,
+      "step": 31519
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.5753161907196045,
+      "learning_rate": 5.029397717464102e-06,
+      "loss": 0.7566,
+      "step": 31520
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.164177656173706,
+      "learning_rate": 5.029134063802641e-06,
+      "loss": 0.5002,
+      "step": 31521
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 5.760721683502197,
+      "learning_rate": 5.02887041006017e-06,
+      "loss": 1.0045,
+      "step": 31522
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.9348037242889404,
+      "learning_rate": 5.028606756237422e-06,
+      "loss": 0.5525,
+      "step": 31523
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.6951403617858887,
+      "learning_rate": 5.028343102335128e-06,
+      "loss": 0.8822,
+      "step": 31524
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.624234914779663,
+      "learning_rate": 5.028079448354022e-06,
+      "loss": 0.6159,
+      "step": 31525
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.4044947624206543,
+      "learning_rate": 5.027815794294838e-06,
+      "loss": 0.542,
+      "step": 31526
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.3779256343841553,
+      "learning_rate": 5.027552140158309e-06,
+      "loss": 0.6594,
+      "step": 31527
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 4.357349872589111,
+      "learning_rate": 5.027288485945169e-06,
+      "loss": 0.43,
+      "step": 31528
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.5795068740844727,
+      "learning_rate": 5.0270248316561484e-06,
+      "loss": 0.653,
+      "step": 31529
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.3472484350204468,
+      "learning_rate": 5.026761177291982e-06,
+      "loss": 0.614,
+      "step": 31530
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 4.245285987854004,
+      "learning_rate": 5.026497522853404e-06,
+      "loss": 0.6514,
+      "step": 31531
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.02247953414917,
+      "learning_rate": 5.026233868341146e-06,
+      "loss": 0.358,
+      "step": 31532
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.5318098068237305,
+      "learning_rate": 5.0259702137559406e-06,
+      "loss": 0.7022,
+      "step": 31533
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 3.672434091567993,
+      "learning_rate": 5.025706559098522e-06,
+      "loss": 0.661,
+      "step": 31534
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.8738583326339722,
+      "learning_rate": 5.0254429043696214e-06,
+      "loss": 0.6773,
+      "step": 31535
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 3.803953170776367,
+      "learning_rate": 5.025179249569977e-06,
+      "loss": 0.6899,
+      "step": 31536
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 3.101557970046997,
+      "learning_rate": 5.024915594700317e-06,
+      "loss": 0.5315,
+      "step": 31537
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.4690299034118652,
+      "learning_rate": 5.024651939761376e-06,
+      "loss": 0.4837,
+      "step": 31538
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.497352361679077,
+      "learning_rate": 5.024388284753887e-06,
+      "loss": 0.7059,
+      "step": 31539
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.625553607940674,
+      "learning_rate": 5.024124629678584e-06,
+      "loss": 0.5192,
+      "step": 31540
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.9244544506072998,
+      "learning_rate": 5.0238609745361985e-06,
+      "loss": 0.611,
+      "step": 31541
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.6010576486587524,
+      "learning_rate": 5.023597319327464e-06,
+      "loss": 0.7259,
+      "step": 31542
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.6996915340423584,
+      "learning_rate": 5.023333664053116e-06,
+      "loss": 0.4128,
+      "step": 31543
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 3.9211883544921875,
+      "learning_rate": 5.023070008713885e-06,
+      "loss": 0.682,
+      "step": 31544
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.2025208473205566,
+      "learning_rate": 5.022806353310505e-06,
+      "loss": 0.4209,
+      "step": 31545
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.5300350189208984,
+      "learning_rate": 5.022542697843708e-06,
+      "loss": 0.3382,
+      "step": 31546
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.0516202449798584,
+      "learning_rate": 5.02227904231423e-06,
+      "loss": 0.7582,
+      "step": 31547
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.406747817993164,
+      "learning_rate": 5.022015386722801e-06,
+      "loss": 0.87,
+      "step": 31548
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.8784558773040771,
+      "learning_rate": 5.021751731070157e-06,
+      "loss": 0.6311,
+      "step": 31549
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.9359697103500366,
+      "learning_rate": 5.021488075357028e-06,
+      "loss": 0.6893,
+      "step": 31550
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.0030674934387207,
+      "learning_rate": 5.0212244195841496e-06,
+      "loss": 0.7006,
+      "step": 31551
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.217691421508789,
+      "learning_rate": 5.020960763752254e-06,
+      "loss": 0.4373,
+      "step": 31552
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.9751274585723877,
+      "learning_rate": 5.020697107862074e-06,
+      "loss": 0.357,
+      "step": 31553
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.8516559600830078,
+      "learning_rate": 5.020433451914344e-06,
+      "loss": 0.6431,
+      "step": 31554
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.4645519256591797,
+      "learning_rate": 5.020169795909795e-06,
+      "loss": 0.4088,
+      "step": 31555
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.9428364038467407,
+      "learning_rate": 5.019906139849161e-06,
+      "loss": 0.8232,
+      "step": 31556
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.173356294631958,
+      "learning_rate": 5.019642483733177e-06,
+      "loss": 0.553,
+      "step": 31557
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.3129541873931885,
+      "learning_rate": 5.019378827562573e-06,
+      "loss": 0.7089,
+      "step": 31558
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.6674108505249023,
+      "learning_rate": 5.019115171338085e-06,
+      "loss": 0.8189,
+      "step": 31559
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 3.8885631561279297,
+      "learning_rate": 5.018851515060444e-06,
+      "loss": 0.775,
+      "step": 31560
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.4296510219573975,
+      "learning_rate": 5.018587858730386e-06,
+      "loss": 0.6055,
+      "step": 31561
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.9616800546646118,
+      "learning_rate": 5.01832420234864e-06,
+      "loss": 0.4184,
+      "step": 31562
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.5711417198181152,
+      "learning_rate": 5.018060545915942e-06,
+      "loss": 0.7363,
+      "step": 31563
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 3.527348279953003,
+      "learning_rate": 5.017796889433024e-06,
+      "loss": 0.664,
+      "step": 31564
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.1536731719970703,
+      "learning_rate": 5.01753323290062e-06,
+      "loss": 0.6545,
+      "step": 31565
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.076606512069702,
+      "learning_rate": 5.017269576319463e-06,
+      "loss": 0.6188,
+      "step": 31566
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 3.533522367477417,
+      "learning_rate": 5.017005919690284e-06,
+      "loss": 0.621,
+      "step": 31567
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.187063455581665,
+      "learning_rate": 5.01674226301382e-06,
+      "loss": 0.4219,
+      "step": 31568
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.9530842304229736,
+      "learning_rate": 5.016478606290801e-06,
+      "loss": 0.6409,
+      "step": 31569
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.8297600746154785,
+      "learning_rate": 5.016214949521961e-06,
+      "loss": 0.7306,
+      "step": 31570
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.893052339553833,
+      "learning_rate": 5.015951292708033e-06,
+      "loss": 0.5166,
+      "step": 31571
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.2307801246643066,
+      "learning_rate": 5.0156876358497505e-06,
+      "loss": 0.7426,
+      "step": 31572
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.6864076852798462,
+      "learning_rate": 5.0154239789478475e-06,
+      "loss": 0.4951,
+      "step": 31573
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.3026318550109863,
+      "learning_rate": 5.015160322003057e-06,
+      "loss": 0.6264,
+      "step": 31574
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.5168943405151367,
+      "learning_rate": 5.014896665016109e-06,
+      "loss": 0.7297,
+      "step": 31575
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.9458236694335938,
+      "learning_rate": 5.014633007987739e-06,
+      "loss": 0.5567,
+      "step": 31576
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.6107159852981567,
+      "learning_rate": 5.0143693509186816e-06,
+      "loss": 0.6804,
+      "step": 31577
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.6013898849487305,
+      "learning_rate": 5.014105693809669e-06,
+      "loss": 0.6772,
+      "step": 31578
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.8839155435562134,
+      "learning_rate": 5.0138420366614326e-06,
+      "loss": 0.4146,
+      "step": 31579
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.000880718231201,
+      "learning_rate": 5.013578379474706e-06,
+      "loss": 0.7469,
+      "step": 31580
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.6689820289611816,
+      "learning_rate": 5.013314722250224e-06,
+      "loss": 0.5247,
+      "step": 31581
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 4.558499336242676,
+      "learning_rate": 5.0130510649887185e-06,
+      "loss": 0.5728,
+      "step": 31582
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.7298638820648193,
+      "learning_rate": 5.012787407690923e-06,
+      "loss": 0.6733,
+      "step": 31583
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.3784642219543457,
+      "learning_rate": 5.012523750357572e-06,
+      "loss": 0.6339,
+      "step": 31584
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 3.5181820392608643,
+      "learning_rate": 5.012260092989395e-06,
+      "loss": 0.5729,
+      "step": 31585
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.9053987264633179,
+      "learning_rate": 5.011996435587126e-06,
+      "loss": 0.7192,
+      "step": 31586
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 4.154245376586914,
+      "learning_rate": 5.011732778151503e-06,
+      "loss": 0.724,
+      "step": 31587
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.4284842014312744,
+      "learning_rate": 5.011469120683253e-06,
+      "loss": 0.8981,
+      "step": 31588
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.317836046218872,
+      "learning_rate": 5.011205463183113e-06,
+      "loss": 0.6211,
+      "step": 31589
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.026432991027832,
+      "learning_rate": 5.010941805651814e-06,
+      "loss": 0.7886,
+      "step": 31590
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 4.352902889251709,
+      "learning_rate": 5.010678148090091e-06,
+      "loss": 0.4608,
+      "step": 31591
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.1540138721466064,
+      "learning_rate": 5.010414490498675e-06,
+      "loss": 0.6968,
+      "step": 31592
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 3.44462513923645,
+      "learning_rate": 5.0101508328783014e-06,
+      "loss": 0.6323,
+      "step": 31593
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.912733554840088,
+      "learning_rate": 5.009887175229701e-06,
+      "loss": 0.5132,
+      "step": 31594
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.172556161880493,
+      "learning_rate": 5.009623517553608e-06,
+      "loss": 0.5329,
+      "step": 31595
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.8861865997314453,
+      "learning_rate": 5.009359859850756e-06,
+      "loss": 0.6457,
+      "step": 31596
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.974275827407837,
+      "learning_rate": 5.009096202121877e-06,
+      "loss": 0.9498,
+      "step": 31597
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.293863296508789,
+      "learning_rate": 5.008832544367705e-06,
+      "loss": 0.37,
+      "step": 31598
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.288721799850464,
+      "learning_rate": 5.008568886588975e-06,
+      "loss": 0.4783,
+      "step": 31599
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.5088438987731934,
+      "learning_rate": 5.008305228786417e-06,
+      "loss": 0.6321,
+      "step": 31600
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.7388923168182373,
+      "learning_rate": 5.008041570960764e-06,
+      "loss": 0.6314,
+      "step": 31601
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.248255968093872,
+      "learning_rate": 5.0077779131127515e-06,
+      "loss": 0.6213,
+      "step": 31602
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.2385926246643066,
+      "learning_rate": 5.007514255243111e-06,
+      "loss": 0.655,
+      "step": 31603
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 3.799104928970337,
+      "learning_rate": 5.007250597352577e-06,
+      "loss": 1.0924,
+      "step": 31604
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.1812708377838135,
+      "learning_rate": 5.006986939441882e-06,
+      "loss": 0.6443,
+      "step": 31605
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 5.412332057952881,
+      "learning_rate": 5.006723281511759e-06,
+      "loss": 0.8419,
+      "step": 31606
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.033374309539795,
+      "learning_rate": 5.006459623562938e-06,
+      "loss": 0.6503,
+      "step": 31607
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.9199821949005127,
+      "learning_rate": 5.006195965596158e-06,
+      "loss": 0.6948,
+      "step": 31608
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 5.039068222045898,
+      "learning_rate": 5.005932307612149e-06,
+      "loss": 0.592,
+      "step": 31609
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.8400299549102783,
+      "learning_rate": 5.0056686496116445e-06,
+      "loss": 0.7334,
+      "step": 31610
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.3803259134292603,
+      "learning_rate": 5.005404991595378e-06,
+      "loss": 0.4616,
+      "step": 31611
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.5502127408981323,
+      "learning_rate": 5.005141333564082e-06,
+      "loss": 0.5846,
+      "step": 31612
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.2319743633270264,
+      "learning_rate": 5.004877675518489e-06,
+      "loss": 0.7564,
+      "step": 31613
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.8050246238708496,
+      "learning_rate": 5.004614017459334e-06,
+      "loss": 0.6526,
+      "step": 31614
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.7163891792297363,
+      "learning_rate": 5.004350359387349e-06,
+      "loss": 0.446,
+      "step": 31615
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.188560724258423,
+      "learning_rate": 5.004086701303267e-06,
+      "loss": 0.4965,
+      "step": 31616
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.0432066917419434,
+      "learning_rate": 5.0038230432078205e-06,
+      "loss": 0.5845,
+      "step": 31617
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.092606782913208,
+      "learning_rate": 5.003559385101745e-06,
+      "loss": 0.9778,
+      "step": 31618
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.7849977016448975,
+      "learning_rate": 5.003295726985772e-06,
+      "loss": 0.4761,
+      "step": 31619
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.956519603729248,
+      "learning_rate": 5.003032068860635e-06,
+      "loss": 0.2575,
+      "step": 31620
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.795687198638916,
+      "learning_rate": 5.002768410727066e-06,
+      "loss": 0.637,
+      "step": 31621
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.8852334022521973,
+      "learning_rate": 5.0025047525858e-06,
+      "loss": 0.761,
+      "step": 31622
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.5225391387939453,
+      "learning_rate": 5.002241094437569e-06,
+      "loss": 0.4734,
+      "step": 31623
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.9061532020568848,
+      "learning_rate": 5.001977436283105e-06,
+      "loss": 0.6569,
+      "step": 31624
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 4.354029655456543,
+      "learning_rate": 5.001713778123145e-06,
+      "loss": 0.7188,
+      "step": 31625
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.6222485303878784,
+      "learning_rate": 5.001450119958416e-06,
+      "loss": 0.5045,
+      "step": 31626
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 3.0623185634613037,
+      "learning_rate": 5.001186461789658e-06,
+      "loss": 0.6068,
+      "step": 31627
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.716871976852417,
+      "learning_rate": 5.0009228036176e-06,
+      "loss": 0.6484,
+      "step": 31628
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 4.5993123054504395,
+      "learning_rate": 5.000659145442977e-06,
+      "loss": 0.6254,
+      "step": 31629
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.30635666847229,
+      "learning_rate": 5.00039548726652e-06,
+      "loss": 0.6058,
+      "step": 31630
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 5.686979293823242,
+      "learning_rate": 5.000131829088963e-06,
+      "loss": 0.5265,
+      "step": 31631
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.991259813308716,
+      "learning_rate": 4.999868170911039e-06,
+      "loss": 0.7626,
+      "step": 31632
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.0770323276519775,
+      "learning_rate": 4.999604512733482e-06,
+      "loss": 0.534,
+      "step": 31633
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.9008030891418457,
+      "learning_rate": 4.999340854557026e-06,
+      "loss": 0.6703,
+      "step": 31634
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 3.1255459785461426,
+      "learning_rate": 4.9990771963824015e-06,
+      "loss": 0.7766,
+      "step": 31635
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.8643639087677002,
+      "learning_rate": 4.998813538210344e-06,
+      "loss": 0.7487,
+      "step": 31636
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 3.4680604934692383,
+      "learning_rate": 4.998549880041584e-06,
+      "loss": 0.5567,
+      "step": 31637
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.926591396331787,
+      "learning_rate": 4.998286221876858e-06,
+      "loss": 0.4903,
+      "step": 31638
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.0823233127593994,
+      "learning_rate": 4.9980225637168965e-06,
+      "loss": 0.3739,
+      "step": 31639
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.6085734367370605,
+      "learning_rate": 4.997758905562433e-06,
+      "loss": 0.5,
+      "step": 31640
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 3.2724151611328125,
+      "learning_rate": 4.997495247414203e-06,
+      "loss": 0.581,
+      "step": 31641
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.5769402980804443,
+      "learning_rate": 4.9972315892729355e-06,
+      "loss": 0.35,
+      "step": 31642
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.7361785173416138,
+      "learning_rate": 4.996967931139367e-06,
+      "loss": 0.5586,
+      "step": 31643
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.1927943229675293,
+      "learning_rate": 4.996704273014228e-06,
+      "loss": 0.689,
+      "step": 31644
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 3.0086569786071777,
+      "learning_rate": 4.996440614898255e-06,
+      "loss": 0.6522,
+      "step": 31645
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.483513355255127,
+      "learning_rate": 4.9961769567921795e-06,
+      "loss": 0.6485,
+      "step": 31646
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 5.143411159515381,
+      "learning_rate": 4.995913298696734e-06,
+      "loss": 0.6382,
+      "step": 31647
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.897516965866089,
+      "learning_rate": 4.995649640612653e-06,
+      "loss": 0.8347,
+      "step": 31648
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.8856496810913086,
+      "learning_rate": 4.995385982540667e-06,
+      "loss": 0.7243,
+      "step": 31649
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.5140382051467896,
+      "learning_rate": 4.995122324481512e-06,
+      "loss": 0.6125,
+      "step": 31650
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.1914141178131104,
+      "learning_rate": 4.99485866643592e-06,
+      "loss": 0.7233,
+      "step": 31651
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.158843755722046,
+      "learning_rate": 4.9945950084046235e-06,
+      "loss": 0.7425,
+      "step": 31652
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.736832618713379,
+      "learning_rate": 4.994331350388356e-06,
+      "loss": 0.6962,
+      "step": 31653
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.003990888595581,
+      "learning_rate": 4.9940676923878524e-06,
+      "loss": 0.7301,
+      "step": 31654
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.6662601232528687,
+      "learning_rate": 4.993804034403843e-06,
+      "loss": 0.5177,
+      "step": 31655
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.9797619581222534,
+      "learning_rate": 4.993540376437063e-06,
+      "loss": 0.4508,
+      "step": 31656
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.240675926208496,
+      "learning_rate": 4.993276718488244e-06,
+      "loss": 0.6107,
+      "step": 31657
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 3.513235092163086,
+      "learning_rate": 4.993013060558121e-06,
+      "loss": 0.5581,
+      "step": 31658
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.5422921180725098,
+      "learning_rate": 4.992749402647425e-06,
+      "loss": 0.5539,
+      "step": 31659
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 3.104624032974243,
+      "learning_rate": 4.992485744756891e-06,
+      "loss": 0.5892,
+      "step": 31660
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.600388288497925,
+      "learning_rate": 4.992222086887251e-06,
+      "loss": 0.6094,
+      "step": 31661
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 4.869024753570557,
+      "learning_rate": 4.9919584290392385e-06,
+      "loss": 0.5861,
+      "step": 31662
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.7057578563690186,
+      "learning_rate": 4.991694771213584e-06,
+      "loss": 0.9747,
+      "step": 31663
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.2492001056671143,
+      "learning_rate": 4.991431113411026e-06,
+      "loss": 0.6566,
+      "step": 31664
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.3789029121398926,
+      "learning_rate": 4.991167455632295e-06,
+      "loss": 0.6561,
+      "step": 31665
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.9119576215744019,
+      "learning_rate": 4.990903797878123e-06,
+      "loss": 0.7848,
+      "step": 31666
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.4756064414978027,
+      "learning_rate": 4.990640140149245e-06,
+      "loss": 0.776,
+      "step": 31667
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 4.152053356170654,
+      "learning_rate": 4.990376482446393e-06,
+      "loss": 0.5607,
+      "step": 31668
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.0941455364227295,
+      "learning_rate": 4.990112824770301e-06,
+      "loss": 0.7479,
+      "step": 31669
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 4.277801990509033,
+      "learning_rate": 4.989849167121701e-06,
+      "loss": 0.6883,
+      "step": 31670
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.509437084197998,
+      "learning_rate": 4.989585509501326e-06,
+      "loss": 0.5026,
+      "step": 31671
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.3923254013061523,
+      "learning_rate": 4.98932185190991e-06,
+      "loss": 0.8138,
+      "step": 31672
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.074687957763672,
+      "learning_rate": 4.989058194348187e-06,
+      "loss": 0.5982,
+      "step": 31673
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 4.194581985473633,
+      "learning_rate": 4.988794536816888e-06,
+      "loss": 0.8521,
+      "step": 31674
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.266206979751587,
+      "learning_rate": 4.9885308793167485e-06,
+      "loss": 0.6056,
+      "step": 31675
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.2280449867248535,
+      "learning_rate": 4.988267221848499e-06,
+      "loss": 0.7479,
+      "step": 31676
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.570138692855835,
+      "learning_rate": 4.9880035644128745e-06,
+      "loss": 0.5545,
+      "step": 31677
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.788728713989258,
+      "learning_rate": 4.987739907010607e-06,
+      "loss": 0.6026,
+      "step": 31678
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 5.918752193450928,
+      "learning_rate": 4.987476249642431e-06,
+      "loss": 0.6504,
+      "step": 31679
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 3.139971971511841,
+      "learning_rate": 4.987212592309078e-06,
+      "loss": 0.4336,
+      "step": 31680
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.7597672939300537,
+      "learning_rate": 4.986948935011283e-06,
+      "loss": 0.86,
+      "step": 31681
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.5897457599639893,
+      "learning_rate": 4.986685277749777e-06,
+      "loss": 0.6442,
+      "step": 31682
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.1521496772766113,
+      "learning_rate": 4.986421620525294e-06,
+      "loss": 0.7327,
+      "step": 31683
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.8032121658325195,
+      "learning_rate": 4.986157963338568e-06,
+      "loss": 0.5731,
+      "step": 31684
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.95501708984375,
+      "learning_rate": 4.985894306190332e-06,
+      "loss": 1.1463,
+      "step": 31685
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.9413269758224487,
+      "learning_rate": 4.985630649081318e-06,
+      "loss": 0.7269,
+      "step": 31686
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 3.297315835952759,
+      "learning_rate": 4.985366992012261e-06,
+      "loss": 0.5361,
+      "step": 31687
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.418442726135254,
+      "learning_rate": 4.985103334983892e-06,
+      "loss": 0.7027,
+      "step": 31688
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.5717544555664062,
+      "learning_rate": 4.984839677996946e-06,
+      "loss": 0.5145,
+      "step": 31689
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.0567660331726074,
+      "learning_rate": 4.984576021052154e-06,
+      "loss": 0.5256,
+      "step": 31690
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.3208091259002686,
+      "learning_rate": 4.98431236415025e-06,
+      "loss": 0.5655,
+      "step": 31691
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 3.6624176502227783,
+      "learning_rate": 4.984048707291969e-06,
+      "loss": 0.8003,
+      "step": 31692
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 3.1668825149536133,
+      "learning_rate": 4.983785050478041e-06,
+      "loss": 0.5527,
+      "step": 31693
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 3.1446495056152344,
+      "learning_rate": 4.983521393709201e-06,
+      "loss": 0.8007,
+      "step": 31694
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 3.329352378845215,
+      "learning_rate": 4.983257736986182e-06,
+      "loss": 0.9281,
+      "step": 31695
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 4.260315895080566,
+      "learning_rate": 4.982994080309717e-06,
+      "loss": 0.4235,
+      "step": 31696
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.444878578186035,
+      "learning_rate": 4.98273042368054e-06,
+      "loss": 0.6013,
+      "step": 31697
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.743543863296509,
+      "learning_rate": 4.982466767099382e-06,
+      "loss": 0.6755,
+      "step": 31698
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.6373581886291504,
+      "learning_rate": 4.9822031105669776e-06,
+      "loss": 0.638,
+      "step": 31699
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.8174723386764526,
+      "learning_rate": 4.98193945408406e-06,
+      "loss": 0.6452,
+      "step": 31700
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.5077602863311768,
+      "learning_rate": 4.981675797651363e-06,
+      "loss": 0.7751,
+      "step": 31701
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.2956759929656982,
+      "learning_rate": 4.981412141269617e-06,
+      "loss": 0.5434,
+      "step": 31702
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.3425214290618896,
+      "learning_rate": 4.981148484939557e-06,
+      "loss": 0.9098,
+      "step": 31703
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.0222301483154297,
+      "learning_rate": 4.980884828661916e-06,
+      "loss": 0.7066,
+      "step": 31704
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.0851104259490967,
+      "learning_rate": 4.980621172437427e-06,
+      "loss": 0.5183,
+      "step": 31705
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.5248310565948486,
+      "learning_rate": 4.980357516266824e-06,
+      "loss": 0.553,
+      "step": 31706
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 3.2769343852996826,
+      "learning_rate": 4.980093860150839e-06,
+      "loss": 0.7404,
+      "step": 31707
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.148138999938965,
+      "learning_rate": 4.9798302040902065e-06,
+      "loss": 0.6236,
+      "step": 31708
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 3.7950236797332764,
+      "learning_rate": 4.979566548085658e-06,
+      "loss": 0.9074,
+      "step": 31709
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 3.5024611949920654,
+      "learning_rate": 4.979302892137927e-06,
+      "loss": 0.5859,
+      "step": 31710
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.5460617542266846,
+      "learning_rate": 4.9790392362477475e-06,
+      "loss": 0.6925,
+      "step": 31711
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 3.1870503425598145,
+      "learning_rate": 4.978775580415851e-06,
+      "loss": 0.7959,
+      "step": 31712
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.8002897500991821,
+      "learning_rate": 4.978511924642973e-06,
+      "loss": 0.7584,
+      "step": 31713
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.695387601852417,
+      "learning_rate": 4.978248268929844e-06,
+      "loss": 0.8451,
+      "step": 31714
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.1672849655151367,
+      "learning_rate": 4.9779846132772e-06,
+      "loss": 0.5581,
+      "step": 31715
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.4344985485076904,
+      "learning_rate": 4.977720957685772e-06,
+      "loss": 0.6656,
+      "step": 31716
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.9744484424591064,
+      "learning_rate": 4.9774573021562925e-06,
+      "loss": 0.7095,
+      "step": 31717
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.6124980449676514,
+      "learning_rate": 4.977193646689497e-06,
+      "loss": 0.6533,
+      "step": 31718
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 3.0436089038848877,
+      "learning_rate": 4.9769299912861165e-06,
+      "loss": 0.389,
+      "step": 31719
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 3.6694557666778564,
+      "learning_rate": 4.976666335946887e-06,
+      "loss": 0.8234,
+      "step": 31720
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.262711763381958,
+      "learning_rate": 4.976402680672537e-06,
+      "loss": 0.5809,
+      "step": 31721
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.959040880203247,
+      "learning_rate": 4.976139025463804e-06,
+      "loss": 0.6163,
+      "step": 31722
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.936812400817871,
+      "learning_rate": 4.975875370321419e-06,
+      "loss": 0.5652,
+      "step": 31723
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 5.450789451599121,
+      "learning_rate": 4.9756117152461136e-06,
+      "loss": 0.7771,
+      "step": 31724
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.777197003364563,
+      "learning_rate": 4.975348060238625e-06,
+      "loss": 0.5443,
+      "step": 31725
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.704079270362854,
+      "learning_rate": 4.975084405299684e-06,
+      "loss": 0.5741,
+      "step": 31726
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 5.548101425170898,
+      "learning_rate": 4.974820750430024e-06,
+      "loss": 0.9259,
+      "step": 31727
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.0144877433776855,
+      "learning_rate": 4.974557095630378e-06,
+      "loss": 0.6632,
+      "step": 31728
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 3.1164119243621826,
+      "learning_rate": 4.974293440901479e-06,
+      "loss": 0.8137,
+      "step": 31729
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.9888885021209717,
+      "learning_rate": 4.97402978624406e-06,
+      "loss": 0.617,
+      "step": 31730
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.103868246078491,
+      "learning_rate": 4.973766131658856e-06,
+      "loss": 0.5465,
+      "step": 31731
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.64846134185791,
+      "learning_rate": 4.973502477146597e-06,
+      "loss": 0.9088,
+      "step": 31732
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.59037446975708,
+      "learning_rate": 4.973238822708018e-06,
+      "loss": 0.4543,
+      "step": 31733
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.7348333597183228,
+      "learning_rate": 4.972975168343852e-06,
+      "loss": 0.4339,
+      "step": 31734
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.9058550596237183,
+      "learning_rate": 4.9727115140548326e-06,
+      "loss": 0.5533,
+      "step": 31735
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.784717321395874,
+      "learning_rate": 4.972447859841692e-06,
+      "loss": 0.4095,
+      "step": 31736
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.6067919731140137,
+      "learning_rate": 4.972184205705163e-06,
+      "loss": 0.5498,
+      "step": 31737
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.6323952674865723,
+      "learning_rate": 4.97192055164598e-06,
+      "loss": 0.6155,
+      "step": 31738
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.933651328086853,
+      "learning_rate": 4.971656897664874e-06,
+      "loss": 0.7281,
+      "step": 31739
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.7931668758392334,
+      "learning_rate": 4.971393243762581e-06,
+      "loss": 0.8057,
+      "step": 31740
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.4361064434051514,
+      "learning_rate": 4.971129589939832e-06,
+      "loss": 0.4357,
+      "step": 31741
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 7.776578903198242,
+      "learning_rate": 4.9708659361973614e-06,
+      "loss": 0.746,
+      "step": 31742
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.3083354234695435,
+      "learning_rate": 4.9706022825359016e-06,
+      "loss": 0.564,
+      "step": 31743
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.5932233333587646,
+      "learning_rate": 4.970338628956183e-06,
+      "loss": 0.7425,
+      "step": 31744
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 4.099177360534668,
+      "learning_rate": 4.970074975458944e-06,
+      "loss": 0.6238,
+      "step": 31745
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.7359552383422852,
+      "learning_rate": 4.969811322044916e-06,
+      "loss": 0.4225,
+      "step": 31746
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 3.8061892986297607,
+      "learning_rate": 4.96954766871483e-06,
+      "loss": 0.7594,
+      "step": 31747
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.8045908212661743,
+      "learning_rate": 4.9692840154694216e-06,
+      "loss": 0.4923,
+      "step": 31748
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.49926495552063,
+      "learning_rate": 4.969020362309422e-06,
+      "loss": 0.6543,
+      "step": 31749
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.4954464435577393,
+      "learning_rate": 4.968756709235567e-06,
+      "loss": 0.6713,
+      "step": 31750
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.2295501232147217,
+      "learning_rate": 4.9684930562485866e-06,
+      "loss": 0.6961,
+      "step": 31751
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.041822671890259,
+      "learning_rate": 4.968229403349215e-06,
+      "loss": 0.8624,
+      "step": 31752
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.643254280090332,
+      "learning_rate": 4.9679657505381854e-06,
+      "loss": 0.6196,
+      "step": 31753
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 3.2218105792999268,
+      "learning_rate": 4.967702097816232e-06,
+      "loss": 0.5422,
+      "step": 31754
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.0597808361053467,
+      "learning_rate": 4.9674384451840845e-06,
+      "loss": 0.6215,
+      "step": 31755
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.240615129470825,
+      "learning_rate": 4.967174792642481e-06,
+      "loss": 0.8366,
+      "step": 31756
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.7965545654296875,
+      "learning_rate": 4.966911140192151e-06,
+      "loss": 0.7305,
+      "step": 31757
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.5644373893737793,
+      "learning_rate": 4.9666474878338286e-06,
+      "loss": 0.615,
+      "step": 31758
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 3.0138161182403564,
+      "learning_rate": 4.9663838355682485e-06,
+      "loss": 0.7438,
+      "step": 31759
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.9955027103424072,
+      "learning_rate": 4.966120183396141e-06,
+      "loss": 0.6179,
+      "step": 31760
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.928896427154541,
+      "learning_rate": 4.9658565313182414e-06,
+      "loss": 0.6716,
+      "step": 31761
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.1973252296447754,
+      "learning_rate": 4.9655928793352815e-06,
+      "loss": 0.5098,
+      "step": 31762
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.7012609243392944,
+      "learning_rate": 4.965329227447996e-06,
+      "loss": 0.5038,
+      "step": 31763
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.1995508670806885,
+      "learning_rate": 4.965065575657114e-06,
+      "loss": 0.4772,
+      "step": 31764
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.4011149406433105,
+      "learning_rate": 4.9648019239633735e-06,
+      "loss": 0.8398,
+      "step": 31765
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.1848294734954834,
+      "learning_rate": 4.964538272367506e-06,
+      "loss": 0.7258,
+      "step": 31766
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.2190544605255127,
+      "learning_rate": 4.964274620870242e-06,
+      "loss": 0.693,
+      "step": 31767
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.6012115478515625,
+      "learning_rate": 4.96401096947232e-06,
+      "loss": 0.6825,
+      "step": 31768
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 3.549506187438965,
+      "learning_rate": 4.963747318174468e-06,
+      "loss": 0.6441,
+      "step": 31769
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.650727391242981,
+      "learning_rate": 4.9634836669774225e-06,
+      "loss": 0.3575,
+      "step": 31770
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.254087209701538,
+      "learning_rate": 4.963220015881915e-06,
+      "loss": 0.5314,
+      "step": 31771
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 7.3943681716918945,
+      "learning_rate": 4.9629563648886784e-06,
+      "loss": 0.8078,
+      "step": 31772
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.9692922830581665,
+      "learning_rate": 4.962692713998447e-06,
+      "loss": 0.6994,
+      "step": 31773
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.7940729856491089,
+      "learning_rate": 4.962429063211952e-06,
+      "loss": 0.3481,
+      "step": 31774
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.7535429000854492,
+      "learning_rate": 4.962165412529928e-06,
+      "loss": 0.5377,
+      "step": 31775
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 5.908168792724609,
+      "learning_rate": 4.961901761953109e-06,
+      "loss": 0.6732,
+      "step": 31776
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.41054630279541,
+      "learning_rate": 4.961638111482226e-06,
+      "loss": 0.7883,
+      "step": 31777
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.4699225425720215,
+      "learning_rate": 4.961374461118013e-06,
+      "loss": 0.3887,
+      "step": 31778
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.916410446166992,
+      "learning_rate": 4.961110810861204e-06,
+      "loss": 0.7041,
+      "step": 31779
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.473665475845337,
+      "learning_rate": 4.96084716071253e-06,
+      "loss": 0.4954,
+      "step": 31780
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.6365835666656494,
+      "learning_rate": 4.960583510672726e-06,
+      "loss": 0.6034,
+      "step": 31781
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.6193511486053467,
+      "learning_rate": 4.9603198607425246e-06,
+      "loss": 0.6317,
+      "step": 31782
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.692734956741333,
+      "learning_rate": 4.960056210922659e-06,
+      "loss": 0.579,
+      "step": 31783
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 3.1099259853363037,
+      "learning_rate": 4.959792561213861e-06,
+      "loss": 0.4987,
+      "step": 31784
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.7260173559188843,
+      "learning_rate": 4.959528911616864e-06,
+      "loss": 0.3841,
+      "step": 31785
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 6.997459888458252,
+      "learning_rate": 4.959265262132404e-06,
+      "loss": 0.6242,
+      "step": 31786
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 4.8936052322387695,
+      "learning_rate": 4.959001612761211e-06,
+      "loss": 0.3632,
+      "step": 31787
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.451082468032837,
+      "learning_rate": 4.9587379635040196e-06,
+      "loss": 0.4151,
+      "step": 31788
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.0382063388824463,
+      "learning_rate": 4.958474314361563e-06,
+      "loss": 0.7518,
+      "step": 31789
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.824986219406128,
+      "learning_rate": 4.958210665334573e-06,
+      "loss": 0.6924,
+      "step": 31790
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.73203706741333,
+      "learning_rate": 4.957947016423784e-06,
+      "loss": 0.703,
+      "step": 31791
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 3.5351622104644775,
+      "learning_rate": 4.9576833676299286e-06,
+      "loss": 0.5163,
+      "step": 31792
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.8932101726531982,
+      "learning_rate": 4.9574197189537395e-06,
+      "loss": 0.5029,
+      "step": 31793
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.858419179916382,
+      "learning_rate": 4.957156070395951e-06,
+      "loss": 0.5824,
+      "step": 31794
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.8994855880737305,
+      "learning_rate": 4.956892421957295e-06,
+      "loss": 0.5271,
+      "step": 31795
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 3.7036471366882324,
+      "learning_rate": 4.956628773638505e-06,
+      "loss": 0.7953,
+      "step": 31796
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 4.934227466583252,
+      "learning_rate": 4.956365125440313e-06,
+      "loss": 0.5289,
+      "step": 31797
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.32120680809021,
+      "learning_rate": 4.956101477363455e-06,
+      "loss": 0.6685,
+      "step": 31798
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.4172003269195557,
+      "learning_rate": 4.955837829408662e-06,
+      "loss": 0.6659,
+      "step": 31799
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.0150959491729736,
+      "learning_rate": 4.955574181576667e-06,
+      "loss": 0.4243,
+      "step": 31800
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 3.0032811164855957,
+      "learning_rate": 4.9553105338682035e-06,
+      "loss": 0.75,
+      "step": 31801
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.801745891571045,
+      "learning_rate": 4.955046886284004e-06,
+      "loss": 0.5209,
+      "step": 31802
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.3790085315704346,
+      "learning_rate": 4.954783238824803e-06,
+      "loss": 0.5079,
+      "step": 31803
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.281006097793579,
+      "learning_rate": 4.9545195914913335e-06,
+      "loss": 0.6312,
+      "step": 31804
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 3.883530855178833,
+      "learning_rate": 4.954255944284325e-06,
+      "loss": 0.74,
+      "step": 31805
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 16.29667854309082,
+      "learning_rate": 4.953992297204516e-06,
+      "loss": 0.5478,
+      "step": 31806
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.3009774684906006,
+      "learning_rate": 4.953728650252638e-06,
+      "loss": 0.849,
+      "step": 31807
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.0090296268463135,
+      "learning_rate": 4.953465003429422e-06,
+      "loss": 0.5539,
+      "step": 31808
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.0803186893463135,
+      "learning_rate": 4.953201356735603e-06,
+      "loss": 0.7766,
+      "step": 31809
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 3.3113512992858887,
+      "learning_rate": 4.952937710171913e-06,
+      "loss": 0.5928,
+      "step": 31810
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.9661405086517334,
+      "learning_rate": 4.952674063739087e-06,
+      "loss": 0.3452,
+      "step": 31811
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.687631368637085,
+      "learning_rate": 4.952410417437855e-06,
+      "loss": 0.8232,
+      "step": 31812
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.6298115253448486,
+      "learning_rate": 4.952146771268952e-06,
+      "loss": 0.546,
+      "step": 31813
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.141508102416992,
+      "learning_rate": 4.951883125233112e-06,
+      "loss": 0.9019,
+      "step": 31814
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.2442309856414795,
+      "learning_rate": 4.951619479331066e-06,
+      "loss": 0.5432,
+      "step": 31815
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.8853121995925903,
+      "learning_rate": 4.9513558335635496e-06,
+      "loss": 0.5483,
+      "step": 31816
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 4.16092586517334,
+      "learning_rate": 4.951092187931292e-06,
+      "loss": 0.8473,
+      "step": 31817
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 7.184257507324219,
+      "learning_rate": 4.950828542435029e-06,
+      "loss": 0.7572,
+      "step": 31818
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 3.252901315689087,
+      "learning_rate": 4.950564897075495e-06,
+      "loss": 0.5298,
+      "step": 31819
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 4.627556800842285,
+      "learning_rate": 4.950301251853421e-06,
+      "loss": 0.4899,
+      "step": 31820
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.5251598358154297,
+      "learning_rate": 4.950037606769541e-06,
+      "loss": 0.6586,
+      "step": 31821
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.5479180812835693,
+      "learning_rate": 4.949773961824587e-06,
+      "loss": 0.6743,
+      "step": 31822
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.4778642654418945,
+      "learning_rate": 4.949510317019293e-06,
+      "loss": 0.546,
+      "step": 31823
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 4.048463821411133,
+      "learning_rate": 4.949246672354392e-06,
+      "loss": 0.3994,
+      "step": 31824
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 5.183289051055908,
+      "learning_rate": 4.948983027830616e-06,
+      "loss": 0.7275,
+      "step": 31825
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.8920937776565552,
+      "learning_rate": 4.9487193834487e-06,
+      "loss": 0.6468,
+      "step": 31826
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.392073154449463,
+      "learning_rate": 4.948455739209375e-06,
+      "loss": 0.7065,
+      "step": 31827
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.9151816368103027,
+      "learning_rate": 4.948192095113377e-06,
+      "loss": 0.4833,
+      "step": 31828
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 3.3153653144836426,
+      "learning_rate": 4.947928451161436e-06,
+      "loss": 0.4949,
+      "step": 31829
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.9921523332595825,
+      "learning_rate": 4.947664807354287e-06,
+      "loss": 0.7265,
+      "step": 31830
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.7863821983337402,
+      "learning_rate": 4.947401163692663e-06,
+      "loss": 0.759,
+      "step": 31831
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.1932711601257324,
+      "learning_rate": 4.947137520177296e-06,
+      "loss": 0.7018,
+      "step": 31832
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.5140936374664307,
+      "learning_rate": 4.946873876808921e-06,
+      "loss": 0.7002,
+      "step": 31833
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.7043261528015137,
+      "learning_rate": 4.946610233588269e-06,
+      "loss": 0.5796,
+      "step": 31834
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.09065842628479,
+      "learning_rate": 4.946346590516074e-06,
+      "loss": 0.7488,
+      "step": 31835
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.0730111598968506,
+      "learning_rate": 4.94608294759307e-06,
+      "loss": 0.6588,
+      "step": 31836
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 4.859510898590088,
+      "learning_rate": 4.945819304819988e-06,
+      "loss": 0.3543,
+      "step": 31837
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 4.720347881317139,
+      "learning_rate": 4.945555662197562e-06,
+      "loss": 0.8344,
+      "step": 31838
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 6.013689994812012,
+      "learning_rate": 4.945292019726525e-06,
+      "loss": 0.4479,
+      "step": 31839
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.871367931365967,
+      "learning_rate": 4.945028377407612e-06,
+      "loss": 0.6847,
+      "step": 31840
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.1093978881835938,
+      "learning_rate": 4.944764735241554e-06,
+      "loss": 0.5908,
+      "step": 31841
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.4340620040893555,
+      "learning_rate": 4.944501093229084e-06,
+      "loss": 0.6487,
+      "step": 31842
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.1706788539886475,
+      "learning_rate": 4.944237451370936e-06,
+      "loss": 0.6082,
+      "step": 31843
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.714693546295166,
+      "learning_rate": 4.943973809667842e-06,
+      "loss": 0.524,
+      "step": 31844
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.5467469692230225,
+      "learning_rate": 4.943710168120536e-06,
+      "loss": 0.5572,
+      "step": 31845
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.8884785175323486,
+      "learning_rate": 4.94344652672975e-06,
+      "loss": 0.6765,
+      "step": 31846
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.7199029922485352,
+      "learning_rate": 4.943182885496219e-06,
+      "loss": 0.4424,
+      "step": 31847
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 3.058784008026123,
+      "learning_rate": 4.942919244420676e-06,
+      "loss": 0.5446,
+      "step": 31848
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.932830572128296,
+      "learning_rate": 4.942655603503853e-06,
+      "loss": 0.4934,
+      "step": 31849
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 4.359203338623047,
+      "learning_rate": 4.942391962746483e-06,
+      "loss": 0.5882,
+      "step": 31850
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.3796956539154053,
+      "learning_rate": 4.942128322149299e-06,
+      "loss": 0.5962,
+      "step": 31851
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 3.1984643936157227,
+      "learning_rate": 4.941864681713035e-06,
+      "loss": 0.735,
+      "step": 31852
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.918235182762146,
+      "learning_rate": 4.9416010414384235e-06,
+      "loss": 0.6707,
+      "step": 31853
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 3.2409632205963135,
+      "learning_rate": 4.9413374013261975e-06,
+      "loss": 0.5322,
+      "step": 31854
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 3.0103588104248047,
+      "learning_rate": 4.941073761377091e-06,
+      "loss": 0.6756,
+      "step": 31855
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.832663893699646,
+      "learning_rate": 4.940810121591835e-06,
+      "loss": 0.6223,
+      "step": 31856
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.4638429880142212,
+      "learning_rate": 4.940546481971165e-06,
+      "loss": 0.6786,
+      "step": 31857
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.2637016773223877,
+      "learning_rate": 4.940282842515812e-06,
+      "loss": 0.4229,
+      "step": 31858
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.6825690269470215,
+      "learning_rate": 4.940019203226511e-06,
+      "loss": 0.6777,
+      "step": 31859
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 3.6208457946777344,
+      "learning_rate": 4.939755564103993e-06,
+      "loss": 0.9815,
+      "step": 31860
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.785557746887207,
+      "learning_rate": 4.939491925148992e-06,
+      "loss": 0.5383,
+      "step": 31861
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.9397711753845215,
+      "learning_rate": 4.9392282863622425e-06,
+      "loss": 0.4539,
+      "step": 31862
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.2604897022247314,
+      "learning_rate": 4.938964647744476e-06,
+      "loss": 0.7228,
+      "step": 31863
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.8562906980514526,
+      "learning_rate": 4.938701009296426e-06,
+      "loss": 0.581,
+      "step": 31864
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.0213496685028076,
+      "learning_rate": 4.9384373710188255e-06,
+      "loss": 0.4886,
+      "step": 31865
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.2870707511901855,
+      "learning_rate": 4.938173732912405e-06,
+      "loss": 0.5795,
+      "step": 31866
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 4.371947288513184,
+      "learning_rate": 4.937910094977903e-06,
+      "loss": 0.6141,
+      "step": 31867
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.0582048892974854,
+      "learning_rate": 4.9376464572160495e-06,
+      "loss": 0.5523,
+      "step": 31868
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.9614884853363037,
+      "learning_rate": 4.937382819627578e-06,
+      "loss": 0.5192,
+      "step": 31869
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.36907696723938,
+      "learning_rate": 4.937119182213221e-06,
+      "loss": 0.6957,
+      "step": 31870
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.8922981023788452,
+      "learning_rate": 4.9368555449737124e-06,
+      "loss": 0.6332,
+      "step": 31871
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.1396429538726807,
+      "learning_rate": 4.936591907909784e-06,
+      "loss": 0.5036,
+      "step": 31872
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.7228853702545166,
+      "learning_rate": 4.936328271022171e-06,
+      "loss": 0.5617,
+      "step": 31873
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.9753811359405518,
+      "learning_rate": 4.936064634311605e-06,
+      "loss": 0.9434,
+      "step": 31874
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.676382064819336,
+      "learning_rate": 4.935800997778818e-06,
+      "loss": 0.5147,
+      "step": 31875
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.444800615310669,
+      "learning_rate": 4.935537361424546e-06,
+      "loss": 0.6076,
+      "step": 31876
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.42805814743042,
+      "learning_rate": 4.93527372524952e-06,
+      "loss": 0.4984,
+      "step": 31877
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.5194640159606934,
+      "learning_rate": 4.935010089254473e-06,
+      "loss": 0.1964,
+      "step": 31878
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.045764207839966,
+      "learning_rate": 4.934746453440138e-06,
+      "loss": 0.7462,
+      "step": 31879
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.90798020362854,
+      "learning_rate": 4.934482817807249e-06,
+      "loss": 0.6285,
+      "step": 31880
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.607299327850342,
+      "learning_rate": 4.93421918235654e-06,
+      "loss": 0.6735,
+      "step": 31881
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.6060447692871094,
+      "learning_rate": 4.933955547088742e-06,
+      "loss": 0.7881,
+      "step": 31882
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 4.415400505065918,
+      "learning_rate": 4.9336919120045886e-06,
+      "loss": 0.7379,
+      "step": 31883
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.9533523321151733,
+      "learning_rate": 4.9334282771048144e-06,
+      "loss": 0.3886,
+      "step": 31884
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 3.7515363693237305,
+      "learning_rate": 4.93316464239015e-06,
+      "loss": 0.6461,
+      "step": 31885
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 3.4278006553649902,
+      "learning_rate": 4.932901007861329e-06,
+      "loss": 0.4734,
+      "step": 31886
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 3.3098762035369873,
+      "learning_rate": 4.932637373519086e-06,
+      "loss": 0.6034,
+      "step": 31887
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.8265490531921387,
+      "learning_rate": 4.932373739364154e-06,
+      "loss": 0.7361,
+      "step": 31888
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.031684398651123,
+      "learning_rate": 4.932110105397265e-06,
+      "loss": 0.7628,
+      "step": 31889
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.080402374267578,
+      "learning_rate": 4.931846471619151e-06,
+      "loss": 0.5752,
+      "step": 31890
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.7376999855041504,
+      "learning_rate": 4.9315828380305475e-06,
+      "loss": 0.7292,
+      "step": 31891
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 4.678151607513428,
+      "learning_rate": 4.931319204632187e-06,
+      "loss": 0.9117,
+      "step": 31892
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 3.3550591468811035,
+      "learning_rate": 4.931055571424801e-06,
+      "loss": 0.5091,
+      "step": 31893
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.4454503059387207,
+      "learning_rate": 4.930791938409125e-06,
+      "loss": 0.6313,
+      "step": 31894
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.682382345199585,
+      "learning_rate": 4.9305283055858894e-06,
+      "loss": 0.6274,
+      "step": 31895
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.735614538192749,
+      "learning_rate": 4.930264672955829e-06,
+      "loss": 0.6697,
+      "step": 31896
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.7189257144927979,
+      "learning_rate": 4.9300010405196765e-06,
+      "loss": 0.4927,
+      "step": 31897
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.031127691268921,
+      "learning_rate": 4.929737408278165e-06,
+      "loss": 0.576,
+      "step": 31898
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.8644330501556396,
+      "learning_rate": 4.9294737762320275e-06,
+      "loss": 0.4842,
+      "step": 31899
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.414172410964966,
+      "learning_rate": 4.929210144381997e-06,
+      "loss": 0.613,
+      "step": 31900
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.1856002807617188,
+      "learning_rate": 4.928946512728806e-06,
+      "loss": 0.4176,
+      "step": 31901
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.7677202224731445,
+      "learning_rate": 4.928682881273189e-06,
+      "loss": 0.7353,
+      "step": 31902
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 3.487123489379883,
+      "learning_rate": 4.928419250015878e-06,
+      "loss": 0.5539,
+      "step": 31903
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 3.0434837341308594,
+      "learning_rate": 4.928155618957606e-06,
+      "loss": 0.6941,
+      "step": 31904
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.3387107849121094,
+      "learning_rate": 4.927891988099106e-06,
+      "loss": 0.5017,
+      "step": 31905
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.9832892417907715,
+      "learning_rate": 4.927628357441112e-06,
+      "loss": 0.9437,
+      "step": 31906
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.891994833946228,
+      "learning_rate": 4.927364726984354e-06,
+      "loss": 0.6316,
+      "step": 31907
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 3.1267905235290527,
+      "learning_rate": 4.927101096729569e-06,
+      "loss": 0.9083,
+      "step": 31908
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 6.206677436828613,
+      "learning_rate": 4.92683746667749e-06,
+      "loss": 0.8246,
+      "step": 31909
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.3642784357070923,
+      "learning_rate": 4.926573836828847e-06,
+      "loss": 0.3303,
+      "step": 31910
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.9543980360031128,
+      "learning_rate": 4.926310207184376e-06,
+      "loss": 0.5961,
+      "step": 31911
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.999667763710022,
+      "learning_rate": 4.926046577744808e-06,
+      "loss": 0.8723,
+      "step": 31912
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.027212381362915,
+      "learning_rate": 4.925782948510877e-06,
+      "loss": 0.52,
+      "step": 31913
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.064635992050171,
+      "learning_rate": 4.925519319483315e-06,
+      "loss": 0.3302,
+      "step": 31914
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.3846349716186523,
+      "learning_rate": 4.9252556906628576e-06,
+      "loss": 0.7328,
+      "step": 31915
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.341881275177002,
+      "learning_rate": 4.9249920620502355e-06,
+      "loss": 0.544,
+      "step": 31916
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 3.3044800758361816,
+      "learning_rate": 4.924728433646182e-06,
+      "loss": 0.6747,
+      "step": 31917
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.7317395210266113,
+      "learning_rate": 4.92446480545143e-06,
+      "loss": 0.4812,
+      "step": 31918
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 4.66473913192749,
+      "learning_rate": 4.924201177466714e-06,
+      "loss": 0.5664,
+      "step": 31919
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.847956895828247,
+      "learning_rate": 4.9239375496927655e-06,
+      "loss": 0.7762,
+      "step": 31920
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.5350489616394043,
+      "learning_rate": 4.923673922130318e-06,
+      "loss": 0.6284,
+      "step": 31921
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.873513102531433,
+      "learning_rate": 4.923410294780105e-06,
+      "loss": 0.7488,
+      "step": 31922
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 3.9879159927368164,
+      "learning_rate": 4.923146667642859e-06,
+      "loss": 0.4552,
+      "step": 31923
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 3.1989946365356445,
+      "learning_rate": 4.922883040719314e-06,
+      "loss": 0.5605,
+      "step": 31924
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.3485476970672607,
+      "learning_rate": 4.922619414010202e-06,
+      "loss": 0.4753,
+      "step": 31925
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.913889169692993,
+      "learning_rate": 4.922355787516257e-06,
+      "loss": 0.6049,
+      "step": 31926
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 3.170593023300171,
+      "learning_rate": 4.922092161238209e-06,
+      "loss": 0.8635,
+      "step": 31927
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 4.9172797203063965,
+      "learning_rate": 4.921828535176795e-06,
+      "loss": 0.6539,
+      "step": 31928
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.831773519515991,
+      "learning_rate": 4.921564909332748e-06,
+      "loss": 0.7601,
+      "step": 31929
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 3.6303718090057373,
+      "learning_rate": 4.921301283706798e-06,
+      "loss": 0.6617,
+      "step": 31930
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 3.5687167644500732,
+      "learning_rate": 4.92103765829968e-06,
+      "loss": 0.6503,
+      "step": 31931
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 4.409082889556885,
+      "learning_rate": 4.920774033112128e-06,
+      "loss": 0.9827,
+      "step": 31932
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.028836488723755,
+      "learning_rate": 4.9205104081448725e-06,
+      "loss": 0.4494,
+      "step": 31933
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 3.2678394317626953,
+      "learning_rate": 4.9202467833986475e-06,
+      "loss": 0.6746,
+      "step": 31934
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.213622570037842,
+      "learning_rate": 4.9199831588741876e-06,
+      "loss": 0.8062,
+      "step": 31935
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.0366508960723877,
+      "learning_rate": 4.919719534572224e-06,
+      "loss": 0.6681,
+      "step": 31936
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 4.792669773101807,
+      "learning_rate": 4.9194559104934895e-06,
+      "loss": 0.5716,
+      "step": 31937
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.4032044410705566,
+      "learning_rate": 4.919192286638718e-06,
+      "loss": 0.6446,
+      "step": 31938
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 3.635052442550659,
+      "learning_rate": 4.918928663008644e-06,
+      "loss": 0.5978,
+      "step": 31939
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.49196457862854,
+      "learning_rate": 4.918665039603998e-06,
+      "loss": 0.8006,
+      "step": 31940
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.721277952194214,
+      "learning_rate": 4.918401416425513e-06,
+      "loss": 0.7329,
+      "step": 31941
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.718768358230591,
+      "learning_rate": 4.918137793473924e-06,
+      "loss": 0.7053,
+      "step": 31942
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.048593044281006,
+      "learning_rate": 4.917874170749963e-06,
+      "loss": 0.656,
+      "step": 31943
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.78005313873291,
+      "learning_rate": 4.917610548254364e-06,
+      "loss": 0.9812,
+      "step": 31944
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.298712730407715,
+      "learning_rate": 4.917346925987858e-06,
+      "loss": 0.5632,
+      "step": 31945
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 3.8172173500061035,
+      "learning_rate": 4.917083303951181e-06,
+      "loss": 0.6864,
+      "step": 31946
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.2383222579956055,
+      "learning_rate": 4.916819682145061e-06,
+      "loss": 0.4401,
+      "step": 31947
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.821133852005005,
+      "learning_rate": 4.916556060570237e-06,
+      "loss": 0.8116,
+      "step": 31948
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.373871326446533,
+      "learning_rate": 4.916292439227439e-06,
+      "loss": 0.5479,
+      "step": 31949
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 5.266529083251953,
+      "learning_rate": 4.916028818117401e-06,
+      "loss": 0.6184,
+      "step": 31950
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.8646348714828491,
+      "learning_rate": 4.915765197240854e-06,
+      "loss": 0.6266,
+      "step": 31951
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.0096547603607178,
+      "learning_rate": 4.915501576598532e-06,
+      "loss": 0.4142,
+      "step": 31952
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.0821657180786133,
+      "learning_rate": 4.915237956191169e-06,
+      "loss": 0.8526,
+      "step": 31953
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.2322022914886475,
+      "learning_rate": 4.914974336019498e-06,
+      "loss": 0.6495,
+      "step": 31954
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.783578872680664,
+      "learning_rate": 4.914710716084252e-06,
+      "loss": 0.568,
+      "step": 31955
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.3068320751190186,
+      "learning_rate": 4.914447096386163e-06,
+      "loss": 0.4941,
+      "step": 31956
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.1221044063568115,
+      "learning_rate": 4.914183476925965e-06,
+      "loss": 0.4504,
+      "step": 31957
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.2861952781677246,
+      "learning_rate": 4.91391985770439e-06,
+      "loss": 0.4291,
+      "step": 31958
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 3.809311866760254,
+      "learning_rate": 4.913656238722172e-06,
+      "loss": 0.5019,
+      "step": 31959
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.7961939573287964,
+      "learning_rate": 4.913392619980043e-06,
+      "loss": 0.4447,
+      "step": 31960
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 3.1584484577178955,
+      "learning_rate": 4.913129001478737e-06,
+      "loss": 0.6226,
+      "step": 31961
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 3.106193780899048,
+      "learning_rate": 4.912865383218988e-06,
+      "loss": 0.5537,
+      "step": 31962
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 12.773919105529785,
+      "learning_rate": 4.912601765201526e-06,
+      "loss": 0.7294,
+      "step": 31963
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.9011763334274292,
+      "learning_rate": 4.912338147427087e-06,
+      "loss": 0.6844,
+      "step": 31964
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 4.192582130432129,
+      "learning_rate": 4.912074529896402e-06,
+      "loss": 0.7134,
+      "step": 31965
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 4.9434051513671875,
+      "learning_rate": 4.911810912610205e-06,
+      "loss": 0.6299,
+      "step": 31966
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 3.7871692180633545,
+      "learning_rate": 4.911547295569229e-06,
+      "loss": 0.7654,
+      "step": 31967
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.631086587905884,
+      "learning_rate": 4.911283678774204e-06,
+      "loss": 0.6556,
+      "step": 31968
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 4.430957317352295,
+      "learning_rate": 4.911020062225868e-06,
+      "loss": 0.4778,
+      "step": 31969
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 4.085979461669922,
+      "learning_rate": 4.910756445924953e-06,
+      "loss": 0.7809,
+      "step": 31970
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 4.3161301612854,
+      "learning_rate": 4.91049282987219e-06,
+      "loss": 0.5561,
+      "step": 31971
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.9382038116455078,
+      "learning_rate": 4.9102292140683135e-06,
+      "loss": 0.7698,
+      "step": 31972
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 3.163949966430664,
+      "learning_rate": 4.909965598514055e-06,
+      "loss": 0.6585,
+      "step": 31973
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.827564239501953,
+      "learning_rate": 4.90970198321015e-06,
+      "loss": 0.5124,
+      "step": 31974
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.8553918600082397,
+      "learning_rate": 4.909438368157328e-06,
+      "loss": 0.618,
+      "step": 31975
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.176542043685913,
+      "learning_rate": 4.909174753356325e-06,
+      "loss": 0.6499,
+      "step": 31976
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 3.394230604171753,
+      "learning_rate": 4.908911138807873e-06,
+      "loss": 0.734,
+      "step": 31977
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.9436500072479248,
+      "learning_rate": 4.9086475245127056e-06,
+      "loss": 0.7645,
+      "step": 31978
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.8223862648010254,
+      "learning_rate": 4.908383910471554e-06,
+      "loss": 0.5143,
+      "step": 31979
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.169903516769409,
+      "learning_rate": 4.9081202966851535e-06,
+      "loss": 0.6693,
+      "step": 31980
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.5891568660736084,
+      "learning_rate": 4.907856683154235e-06,
+      "loss": 0.3984,
+      "step": 31981
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.1752989292144775,
+      "learning_rate": 4.907593069879533e-06,
+      "loss": 0.57,
+      "step": 31982
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.9277188777923584,
+      "learning_rate": 4.90732945686178e-06,
+      "loss": 0.4565,
+      "step": 31983
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.238363265991211,
+      "learning_rate": 4.907065844101709e-06,
+      "loss": 0.726,
+      "step": 31984
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.3211960792541504,
+      "learning_rate": 4.906802231600054e-06,
+      "loss": 0.6959,
+      "step": 31985
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.7515119314193726,
+      "learning_rate": 4.906538619357546e-06,
+      "loss": 0.739,
+      "step": 31986
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.213434934616089,
+      "learning_rate": 4.906275007374919e-06,
+      "loss": 0.5359,
+      "step": 31987
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.8664530515670776,
+      "learning_rate": 4.906011395652905e-06,
+      "loss": 0.5188,
+      "step": 31988
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 3.267383098602295,
+      "learning_rate": 4.905747784192239e-06,
+      "loss": 0.8371,
+      "step": 31989
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.281286597251892,
+      "learning_rate": 4.905484172993654e-06,
+      "loss": 0.4576,
+      "step": 31990
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 3.6938774585723877,
+      "learning_rate": 4.905220562057882e-06,
+      "loss": 0.5354,
+      "step": 31991
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 4.097996711730957,
+      "learning_rate": 4.904956951385656e-06,
+      "loss": 0.6244,
+      "step": 31992
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.5932531356811523,
+      "learning_rate": 4.90469334097771e-06,
+      "loss": 0.8423,
+      "step": 31993
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 3.502586841583252,
+      "learning_rate": 4.904429730834775e-06,
+      "loss": 0.8981,
+      "step": 31994
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.9074193239212036,
+      "learning_rate": 4.904166120957585e-06,
+      "loss": 0.4758,
+      "step": 31995
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.353823661804199,
+      "learning_rate": 4.903902511346873e-06,
+      "loss": 0.5744,
+      "step": 31996
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.6488889455795288,
+      "learning_rate": 4.903638902003373e-06,
+      "loss": 0.5716,
+      "step": 31997
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 7.105127334594727,
+      "learning_rate": 4.903375292927816e-06,
+      "loss": 0.6007,
+      "step": 31998
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.265660285949707,
+      "learning_rate": 4.903111684120937e-06,
+      "loss": 0.4943,
+      "step": 31999
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 8.2998046875,
+      "learning_rate": 4.9028480755834675e-06,
+      "loss": 0.5903,
+      "step": 32000
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.4924514293670654,
+      "learning_rate": 4.9025844673161415e-06,
+      "loss": 0.7326,
+      "step": 32001
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.627925157546997,
+      "learning_rate": 4.902320859319691e-06,
+      "loss": 0.8549,
+      "step": 32002
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.8249266147613525,
+      "learning_rate": 4.90205725159485e-06,
+      "loss": 0.7004,
+      "step": 32003
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.2204525470733643,
+      "learning_rate": 4.901793644142351e-06,
+      "loss": 0.6513,
+      "step": 32004
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 3.8552770614624023,
+      "learning_rate": 4.901530036962928e-06,
+      "loss": 0.5429,
+      "step": 32005
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.462853193283081,
+      "learning_rate": 4.901266430057312e-06,
+      "loss": 0.4342,
+      "step": 32006
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.889183521270752,
+      "learning_rate": 4.901002823426237e-06,
+      "loss": 0.7979,
+      "step": 32007
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.4870991706848145,
+      "learning_rate": 4.900739217070436e-06,
+      "loss": 0.682,
+      "step": 32008
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.7713702917099,
+      "learning_rate": 4.9004756109906425e-06,
+      "loss": 0.4352,
+      "step": 32009
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.180342435836792,
+      "learning_rate": 4.900212005187589e-06,
+      "loss": 0.6437,
+      "step": 32010
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.620922803878784,
+      "learning_rate": 4.899948399662009e-06,
+      "loss": 0.7732,
+      "step": 32011
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.8717070817947388,
+      "learning_rate": 4.899684794414635e-06,
+      "loss": 0.5539,
+      "step": 32012
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.0250730514526367,
+      "learning_rate": 4.8994211894462e-06,
+      "loss": 0.7747,
+      "step": 32013
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 5.336828708648682,
+      "learning_rate": 4.899157584757435e-06,
+      "loss": 0.52,
+      "step": 32014
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 4.283689498901367,
+      "learning_rate": 4.898893980349077e-06,
+      "loss": 0.7717,
+      "step": 32015
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.7446510791778564,
+      "learning_rate": 4.898630376221857e-06,
+      "loss": 0.6795,
+      "step": 32016
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.8039817810058594,
+      "learning_rate": 4.8983667723765076e-06,
+      "loss": 0.6778,
+      "step": 32017
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.9376217126846313,
+      "learning_rate": 4.898103168813762e-06,
+      "loss": 0.6808,
+      "step": 32018
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.185441017150879,
+      "learning_rate": 4.897839565534353e-06,
+      "loss": 0.7548,
+      "step": 32019
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.8372668027877808,
+      "learning_rate": 4.897575962539015e-06,
+      "loss": 0.7032,
+      "step": 32020
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.3537347316741943,
+      "learning_rate": 4.897312359828479e-06,
+      "loss": 0.7122,
+      "step": 32021
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.1471521854400635,
+      "learning_rate": 4.8970487574034795e-06,
+      "loss": 0.912,
+      "step": 32022
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.283498525619507,
+      "learning_rate": 4.896785155264748e-06,
+      "loss": 0.491,
+      "step": 32023
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.218756914138794,
+      "learning_rate": 4.8965215534130185e-06,
+      "loss": 0.7968,
+      "step": 32024
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 3.644205331802368,
+      "learning_rate": 4.896257951849025e-06,
+      "loss": 0.6083,
+      "step": 32025
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.5445663928985596,
+      "learning_rate": 4.8959943505734975e-06,
+      "loss": 0.5095,
+      "step": 32026
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 4.2275190353393555,
+      "learning_rate": 4.895730749587172e-06,
+      "loss": 0.5654,
+      "step": 32027
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.7566243410110474,
+      "learning_rate": 4.8954671488907795e-06,
+      "loss": 0.4564,
+      "step": 32028
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 4.397303104400635,
+      "learning_rate": 4.895203548485052e-06,
+      "loss": 0.6724,
+      "step": 32029
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 3.3567628860473633,
+      "learning_rate": 4.894939948370726e-06,
+      "loss": 0.5804,
+      "step": 32030
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.544569492340088,
+      "learning_rate": 4.894676348548533e-06,
+      "loss": 0.5658,
+      "step": 32031
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.9305613040924072,
+      "learning_rate": 4.894412749019205e-06,
+      "loss": 0.5574,
+      "step": 32032
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.212735414505005,
+      "learning_rate": 4.894149149783476e-06,
+      "loss": 0.6485,
+      "step": 32033
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.9555394649505615,
+      "learning_rate": 4.893885550842079e-06,
+      "loss": 0.7016,
+      "step": 32034
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.9470590353012085,
+      "learning_rate": 4.893621952195746e-06,
+      "loss": 0.6811,
+      "step": 32035
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.066302537918091,
+      "learning_rate": 4.893358353845211e-06,
+      "loss": 0.5085,
+      "step": 32036
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 7.9348063468933105,
+      "learning_rate": 4.893094755791205e-06,
+      "loss": 0.4154,
+      "step": 32037
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.58943510055542,
+      "learning_rate": 4.892831158034465e-06,
+      "loss": 0.6125,
+      "step": 32038
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.4612855911254883,
+      "learning_rate": 4.8925675605757194e-06,
+      "loss": 0.6616,
+      "step": 32039
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 3.2497942447662354,
+      "learning_rate": 4.892303963415704e-06,
+      "loss": 0.3816,
+      "step": 32040
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.4422881603240967,
+      "learning_rate": 4.8920403665551505e-06,
+      "loss": 0.561,
+      "step": 32041
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 3.3682055473327637,
+      "learning_rate": 4.891776769994792e-06,
+      "loss": 0.8591,
+      "step": 32042
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.890946865081787,
+      "learning_rate": 4.891513173735363e-06,
+      "loss": 0.6441,
+      "step": 32043
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 3.861114740371704,
+      "learning_rate": 4.891249577777594e-06,
+      "loss": 0.9187,
+      "step": 32044
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 7.999619483947754,
+      "learning_rate": 4.89098598212222e-06,
+      "loss": 0.8681,
+      "step": 32045
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.782534122467041,
+      "learning_rate": 4.890722386769972e-06,
+      "loss": 0.3722,
+      "step": 32046
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 3.030317783355713,
+      "learning_rate": 4.890458791721585e-06,
+      "loss": 0.4632,
+      "step": 32047
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 3.5569393634796143,
+      "learning_rate": 4.890195196977791e-06,
+      "loss": 0.7038,
+      "step": 32048
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 5.87838888168335,
+      "learning_rate": 4.889931602539321e-06,
+      "loss": 0.5764,
+      "step": 32049
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.0678505897521973,
+      "learning_rate": 4.889668008406911e-06,
+      "loss": 0.7822,
+      "step": 32050
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.9490355253219604,
+      "learning_rate": 4.8894044145812944e-06,
+      "loss": 0.6381,
+      "step": 32051
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 9.147781372070312,
+      "learning_rate": 4.889140821063202e-06,
+      "loss": 0.7581,
+      "step": 32052
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 3.2503111362457275,
+      "learning_rate": 4.888877227853368e-06,
+      "loss": 0.5549,
+      "step": 32053
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.1165547370910645,
+      "learning_rate": 4.888613634952524e-06,
+      "loss": 0.5173,
+      "step": 32054
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 3.286433696746826,
+      "learning_rate": 4.888350042361405e-06,
+      "loss": 0.611,
+      "step": 32055
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.6877737045288086,
+      "learning_rate": 4.888086450080741e-06,
+      "loss": 0.6147,
+      "step": 32056
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.0855772495269775,
+      "learning_rate": 4.887822858111268e-06,
+      "loss": 0.6391,
+      "step": 32057
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.4959752559661865,
+      "learning_rate": 4.887559266453717e-06,
+      "loss": 0.6578,
+      "step": 32058
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.1770100593566895,
+      "learning_rate": 4.887295675108822e-06,
+      "loss": 0.5336,
+      "step": 32059
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 3.836235761642456,
+      "learning_rate": 4.887032084077315e-06,
+      "loss": 0.6818,
+      "step": 32060
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 3.674410343170166,
+      "learning_rate": 4.88676849335993e-06,
+      "loss": 0.4758,
+      "step": 32061
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.1034204959869385,
+      "learning_rate": 4.886504902957399e-06,
+      "loss": 0.5616,
+      "step": 32062
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 4.720527172088623,
+      "learning_rate": 4.886241312870455e-06,
+      "loss": 0.6211,
+      "step": 32063
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.780323028564453,
+      "learning_rate": 4.8859777230998305e-06,
+      "loss": 0.4504,
+      "step": 32064
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.0227527618408203,
+      "learning_rate": 4.885714133646261e-06,
+      "loss": 0.4876,
+      "step": 32065
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.849054217338562,
+      "learning_rate": 4.885450544510477e-06,
+      "loss": 0.4095,
+      "step": 32066
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 5.210023403167725,
+      "learning_rate": 4.885186955693212e-06,
+      "loss": 0.6141,
+      "step": 32067
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.792051315307617,
+      "learning_rate": 4.8849233671952e-06,
+      "loss": 0.4704,
+      "step": 32068
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.439844846725464,
+      "learning_rate": 4.884659779017171e-06,
+      "loss": 0.5135,
+      "step": 32069
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 3.409823417663574,
+      "learning_rate": 4.884396191159861e-06,
+      "loss": 0.692,
+      "step": 32070
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 4.204355239868164,
+      "learning_rate": 4.8841326036240026e-06,
+      "loss": 0.529,
+      "step": 32071
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.974177360534668,
+      "learning_rate": 4.883869016410327e-06,
+      "loss": 0.4278,
+      "step": 32072
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.2116758823394775,
+      "learning_rate": 4.8836054295195686e-06,
+      "loss": 0.6651,
+      "step": 32073
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.995579719543457,
+      "learning_rate": 4.88334184295246e-06,
+      "loss": 0.531,
+      "step": 32074
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 3.8703086376190186,
+      "learning_rate": 4.883078256709735e-06,
+      "loss": 0.8975,
+      "step": 32075
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 3.8153421878814697,
+      "learning_rate": 4.882814670792123e-06,
+      "loss": 0.5981,
+      "step": 32076
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 3.4906721115112305,
+      "learning_rate": 4.882551085200361e-06,
+      "loss": 0.7491,
+      "step": 32077
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.662019729614258,
+      "learning_rate": 4.8822874999351814e-06,
+      "loss": 0.4242,
+      "step": 32078
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.382927179336548,
+      "learning_rate": 4.8820239149973145e-06,
+      "loss": 0.789,
+      "step": 32079
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.1312384605407715,
+      "learning_rate": 4.881760330387496e-06,
+      "loss": 0.7172,
+      "step": 32080
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.0196340084075928,
+      "learning_rate": 4.881496746106458e-06,
+      "loss": 0.6262,
+      "step": 32081
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.749422073364258,
+      "learning_rate": 4.881233162154932e-06,
+      "loss": 0.6364,
+      "step": 32082
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.6047251224517822,
+      "learning_rate": 4.880969578533653e-06,
+      "loss": 0.6051,
+      "step": 32083
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.8853373527526855,
+      "learning_rate": 4.880705995243353e-06,
+      "loss": 0.8636,
+      "step": 32084
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.6619137525558472,
+      "learning_rate": 4.880442412284764e-06,
+      "loss": 0.5021,
+      "step": 32085
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.3754749298095703,
+      "learning_rate": 4.88017882965862e-06,
+      "loss": 0.4395,
+      "step": 32086
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.40444016456604,
+      "learning_rate": 4.879915247365655e-06,
+      "loss": 0.5473,
+      "step": 32087
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.085083246231079,
+      "learning_rate": 4.8796516654066e-06,
+      "loss": 0.7952,
+      "step": 32088
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.3832569122314453,
+      "learning_rate": 4.879388083782189e-06,
+      "loss": 0.47,
+      "step": 32089
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.362297534942627,
+      "learning_rate": 4.879124502493152e-06,
+      "loss": 0.7694,
+      "step": 32090
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.303292989730835,
+      "learning_rate": 4.878860921540227e-06,
+      "loss": 0.5916,
+      "step": 32091
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 3.21073317527771,
+      "learning_rate": 4.878597340924144e-06,
+      "loss": 0.7218,
+      "step": 32092
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 5.455931663513184,
+      "learning_rate": 4.878333760645636e-06,
+      "loss": 0.8093,
+      "step": 32093
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.6467583179473877,
+      "learning_rate": 4.878070180705437e-06,
+      "loss": 0.5767,
+      "step": 32094
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 3.0539345741271973,
+      "learning_rate": 4.877806601104279e-06,
+      "loss": 0.4377,
+      "step": 32095
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.7109798192977905,
+      "learning_rate": 4.877543021842894e-06,
+      "loss": 0.6525,
+      "step": 32096
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.3599345684051514,
+      "learning_rate": 4.877279442922017e-06,
+      "loss": 0.6457,
+      "step": 32097
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 5.637377738952637,
+      "learning_rate": 4.87701586434238e-06,
+      "loss": 0.5082,
+      "step": 32098
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.8042222261428833,
+      "learning_rate": 4.876752286104716e-06,
+      "loss": 0.4217,
+      "step": 32099
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.956355571746826,
+      "learning_rate": 4.876488708209757e-06,
+      "loss": 0.7856,
+      "step": 32100
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.643076181411743,
+      "learning_rate": 4.876225130658237e-06,
+      "loss": 0.6612,
+      "step": 32101
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.428114891052246,
+      "learning_rate": 4.8759615534508885e-06,
+      "loss": 0.3936,
+      "step": 32102
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.6725656986236572,
+      "learning_rate": 4.875697976588444e-06,
+      "loss": 0.5018,
+      "step": 32103
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.72358775138855,
+      "learning_rate": 4.875434400071637e-06,
+      "loss": 0.6412,
+      "step": 32104
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.739283323287964,
+      "learning_rate": 4.875170823901202e-06,
+      "loss": 0.3693,
+      "step": 32105
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.0186612606048584,
+      "learning_rate": 4.874907248077868e-06,
+      "loss": 0.6729,
+      "step": 32106
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 3.153074264526367,
+      "learning_rate": 4.874643672602371e-06,
+      "loss": 0.6478,
+      "step": 32107
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.785243034362793,
+      "learning_rate": 4.8743800974754425e-06,
+      "loss": 0.6563,
+      "step": 32108
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 3.121880292892456,
+      "learning_rate": 4.874116522697816e-06,
+      "loss": 0.775,
+      "step": 32109
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.443652391433716,
+      "learning_rate": 4.873852948270222e-06,
+      "loss": 0.7266,
+      "step": 32110
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.8265063762664795,
+      "learning_rate": 4.873589374193399e-06,
+      "loss": 0.5934,
+      "step": 32111
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.820539712905884,
+      "learning_rate": 4.8733258004680755e-06,
+      "loss": 0.4852,
+      "step": 32112
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.544980049133301,
+      "learning_rate": 4.873062227094986e-06,
+      "loss": 0.5995,
+      "step": 32113
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.7935372591018677,
+      "learning_rate": 4.872798654074862e-06,
+      "loss": 0.4338,
+      "step": 32114
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.3327332735061646,
+      "learning_rate": 4.8725350814084385e-06,
+      "loss": 0.3682,
+      "step": 32115
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.9091507196426392,
+      "learning_rate": 4.872271509096446e-06,
+      "loss": 0.6754,
+      "step": 32116
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.6337108612060547,
+      "learning_rate": 4.872007937139619e-06,
+      "loss": 0.6459,
+      "step": 32117
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.285722255706787,
+      "learning_rate": 4.871744365538691e-06,
+      "loss": 0.6979,
+      "step": 32118
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.7757047414779663,
+      "learning_rate": 4.871480794294393e-06,
+      "loss": 0.4623,
+      "step": 32119
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 3.4635062217712402,
+      "learning_rate": 4.871217223407459e-06,
+      "loss": 0.4403,
+      "step": 32120
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.99334454536438,
+      "learning_rate": 4.870953652878621e-06,
+      "loss": 0.6669,
+      "step": 32121
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.9975188970565796,
+      "learning_rate": 4.870690082708613e-06,
+      "loss": 0.6463,
+      "step": 32122
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.8845876455307007,
+      "learning_rate": 4.870426512898168e-06,
+      "loss": 0.5741,
+      "step": 32123
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.8224457502365112,
+      "learning_rate": 4.870162943448018e-06,
+      "loss": 0.6495,
+      "step": 32124
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.1746087074279785,
+      "learning_rate": 4.869899374358896e-06,
+      "loss": 0.6167,
+      "step": 32125
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.0905399322509766,
+      "learning_rate": 4.869635805631534e-06,
+      "loss": 0.6971,
+      "step": 32126
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.8520654439926147,
+      "learning_rate": 4.869372237266667e-06,
+      "loss": 0.516,
+      "step": 32127
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.64516282081604,
+      "learning_rate": 4.869108669265027e-06,
+      "loss": 0.3379,
+      "step": 32128
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.9598811864852905,
+      "learning_rate": 4.868845101627347e-06,
+      "loss": 0.7683,
+      "step": 32129
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 3.017397880554199,
+      "learning_rate": 4.868581534354358e-06,
+      "loss": 0.7277,
+      "step": 32130
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.5820045471191406,
+      "learning_rate": 4.8683179674467964e-06,
+      "loss": 0.5321,
+      "step": 32131
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.907436728477478,
+      "learning_rate": 4.8680544009053925e-06,
+      "loss": 0.7241,
+      "step": 32132
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.5685787200927734,
+      "learning_rate": 4.86779083473088e-06,
+      "loss": 0.2141,
+      "step": 32133
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 3.0304503440856934,
+      "learning_rate": 4.867527268923992e-06,
+      "loss": 0.6018,
+      "step": 32134
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.544975757598877,
+      "learning_rate": 4.86726370348546e-06,
+      "loss": 0.7196,
+      "step": 32135
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.9459164142608643,
+      "learning_rate": 4.867000138416019e-06,
+      "loss": 0.5326,
+      "step": 32136
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.2950401306152344,
+      "learning_rate": 4.866736573716399e-06,
+      "loss": 0.4531,
+      "step": 32137
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.720612049102783,
+      "learning_rate": 4.866473009387337e-06,
+      "loss": 0.6159,
+      "step": 32138
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.167745351791382,
+      "learning_rate": 4.866209445429563e-06,
+      "loss": 0.5383,
+      "step": 32139
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.9661647081375122,
+      "learning_rate": 4.86594588184381e-06,
+      "loss": 0.6942,
+      "step": 32140
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 3.413665533065796,
+      "learning_rate": 4.865682318630811e-06,
+      "loss": 0.5995,
+      "step": 32141
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.6941897869110107,
+      "learning_rate": 4.8654187557912994e-06,
+      "loss": 0.5938,
+      "step": 32142
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 3.1549458503723145,
+      "learning_rate": 4.865155193326009e-06,
+      "loss": 0.6638,
+      "step": 32143
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.936912178993225,
+      "learning_rate": 4.86489163123567e-06,
+      "loss": 0.6067,
+      "step": 32144
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.9697906970977783,
+      "learning_rate": 4.864628069521018e-06,
+      "loss": 0.8704,
+      "step": 32145
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.0487639904022217,
+      "learning_rate": 4.864364508182783e-06,
+      "loss": 0.6278,
+      "step": 32146
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 3.7485549449920654,
+      "learning_rate": 4.864100947221701e-06,
+      "loss": 0.6516,
+      "step": 32147
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.2807083129882812,
+      "learning_rate": 4.8638373866385026e-06,
+      "loss": 0.8039,
+      "step": 32148
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 4.233466148376465,
+      "learning_rate": 4.8635738264339215e-06,
+      "loss": 0.3185,
+      "step": 32149
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.729665756225586,
+      "learning_rate": 4.86331026660869e-06,
+      "loss": 0.8764,
+      "step": 32150
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.9855930805206299,
+      "learning_rate": 4.86304670716354e-06,
+      "loss": 0.5865,
+      "step": 32151
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 6.289918899536133,
+      "learning_rate": 4.862783148099208e-06,
+      "loss": 0.6599,
+      "step": 32152
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.1305503845214844,
+      "learning_rate": 4.862519589416424e-06,
+      "loss": 0.601,
+      "step": 32153
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.02132248878479,
+      "learning_rate": 4.862256031115923e-06,
+      "loss": 0.4863,
+      "step": 32154
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 3.6024625301361084,
+      "learning_rate": 4.861992473198435e-06,
+      "loss": 0.7456,
+      "step": 32155
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.0288643836975098,
+      "learning_rate": 4.861728915664694e-06,
+      "loss": 0.7137,
+      "step": 32156
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.105271100997925,
+      "learning_rate": 4.861465358515434e-06,
+      "loss": 0.5982,
+      "step": 32157
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.9162654876708984,
+      "learning_rate": 4.861201801751386e-06,
+      "loss": 0.6095,
+      "step": 32158
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.1761863231658936,
+      "learning_rate": 4.860938245373285e-06,
+      "loss": 0.7573,
+      "step": 32159
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.2209603786468506,
+      "learning_rate": 4.860674689381862e-06,
+      "loss": 0.5184,
+      "step": 32160
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.7376195192337036,
+      "learning_rate": 4.86041113377785e-06,
+      "loss": 0.3694,
+      "step": 32161
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 3.9789340496063232,
+      "learning_rate": 4.860147578561983e-06,
+      "loss": 0.8194,
+      "step": 32162
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 3.907139539718628,
+      "learning_rate": 4.859884023734993e-06,
+      "loss": 0.7629,
+      "step": 32163
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 6.013734817504883,
+      "learning_rate": 4.859620469297613e-06,
+      "loss": 0.585,
+      "step": 32164
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.4040539264678955,
+      "learning_rate": 4.859356915250576e-06,
+      "loss": 0.6901,
+      "step": 32165
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 3.8869290351867676,
+      "learning_rate": 4.859093361594616e-06,
+      "loss": 0.6791,
+      "step": 32166
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.9935845136642456,
+      "learning_rate": 4.858829808330463e-06,
+      "loss": 0.7063,
+      "step": 32167
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 3.4054806232452393,
+      "learning_rate": 4.858566255458852e-06,
+      "loss": 0.7099,
+      "step": 32168
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 3.723527193069458,
+      "learning_rate": 4.858302702980515e-06,
+      "loss": 0.7115,
+      "step": 32169
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.7661938667297363,
+      "learning_rate": 4.858039150896184e-06,
+      "loss": 0.5666,
+      "step": 32170
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.424522638320923,
+      "learning_rate": 4.857775599206593e-06,
+      "loss": 0.6539,
+      "step": 32171
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 3.023317813873291,
+      "learning_rate": 4.857512047912475e-06,
+      "loss": 0.5555,
+      "step": 32172
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.1853444576263428,
+      "learning_rate": 4.857248497014564e-06,
+      "loss": 0.3794,
+      "step": 32173
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 10.523093223571777,
+      "learning_rate": 4.8569849465135904e-06,
+      "loss": 0.4895,
+      "step": 32174
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 3.553347110748291,
+      "learning_rate": 4.856721396410289e-06,
+      "loss": 0.4409,
+      "step": 32175
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.875485420227051,
+      "learning_rate": 4.856457846705391e-06,
+      "loss": 0.5415,
+      "step": 32176
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 6.674830436706543,
+      "learning_rate": 4.85619429739963e-06,
+      "loss": 0.5972,
+      "step": 32177
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 5.331669807434082,
+      "learning_rate": 4.8559307484937394e-06,
+      "loss": 0.6395,
+      "step": 32178
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.988565444946289,
+      "learning_rate": 4.855667199988451e-06,
+      "loss": 0.6328,
+      "step": 32179
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.0568013191223145,
+      "learning_rate": 4.855403651884498e-06,
+      "loss": 0.8029,
+      "step": 32180
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.263697862625122,
+      "learning_rate": 4.855140104182614e-06,
+      "loss": 0.6248,
+      "step": 32181
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.3458189964294434,
+      "learning_rate": 4.854876556883531e-06,
+      "loss": 0.8191,
+      "step": 32182
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.1474618911743164,
+      "learning_rate": 4.854613009987981e-06,
+      "loss": 0.8254,
+      "step": 32183
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.4424569606781006,
+      "learning_rate": 4.854349463496698e-06,
+      "loss": 0.6452,
+      "step": 32184
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.759469985961914,
+      "learning_rate": 4.854085917410415e-06,
+      "loss": 0.6235,
+      "step": 32185
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.5173957347869873,
+      "learning_rate": 4.853822371729864e-06,
+      "loss": 0.5247,
+      "step": 32186
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 4.147523880004883,
+      "learning_rate": 4.853558826455777e-06,
+      "loss": 0.7461,
+      "step": 32187
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.6542518138885498,
+      "learning_rate": 4.853295281588889e-06,
+      "loss": 0.5069,
+      "step": 32188
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.995433807373047,
+      "learning_rate": 4.8530317371299325e-06,
+      "loss": 0.574,
+      "step": 32189
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.692657947540283,
+      "learning_rate": 4.85276819307964e-06,
+      "loss": 0.765,
+      "step": 32190
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.601599931716919,
+      "learning_rate": 4.852504649438743e-06,
+      "loss": 0.8827,
+      "step": 32191
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 4.0608625411987305,
+      "learning_rate": 4.852241106207975e-06,
+      "loss": 0.592,
+      "step": 32192
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 3.1200835704803467,
+      "learning_rate": 4.85197756338807e-06,
+      "loss": 0.7755,
+      "step": 32193
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 3.1383421421051025,
+      "learning_rate": 4.85171402097976e-06,
+      "loss": 0.5128,
+      "step": 32194
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.577897071838379,
+      "learning_rate": 4.851450478983777e-06,
+      "loss": 0.4517,
+      "step": 32195
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.9372668266296387,
+      "learning_rate": 4.8511869374008556e-06,
+      "loss": 0.5574,
+      "step": 32196
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 3.4360737800598145,
+      "learning_rate": 4.850923396231727e-06,
+      "loss": 0.5288,
+      "step": 32197
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.267866611480713,
+      "learning_rate": 4.850659855477125e-06,
+      "loss": 0.7688,
+      "step": 32198
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 4.080955505371094,
+      "learning_rate": 4.8503963151377805e-06,
+      "loss": 0.5989,
+      "step": 32199
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.9131494760513306,
+      "learning_rate": 4.850132775214429e-06,
+      "loss": 0.5182,
+      "step": 32200
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.6043918132781982,
+      "learning_rate": 4.849869235707802e-06,
+      "loss": 0.6152,
+      "step": 32201
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 3.094224452972412,
+      "learning_rate": 4.8496056966186335e-06,
+      "loss": 0.7216,
+      "step": 32202
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.339104175567627,
+      "learning_rate": 4.849342157947654e-06,
+      "loss": 0.5091,
+      "step": 32203
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.454741954803467,
+      "learning_rate": 4.849078619695598e-06,
+      "loss": 0.4736,
+      "step": 32204
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.9198862314224243,
+      "learning_rate": 4.848815081863198e-06,
+      "loss": 0.5704,
+      "step": 32205
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.0199217796325684,
+      "learning_rate": 4.848551544451186e-06,
+      "loss": 0.5959,
+      "step": 32206
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.396578788757324,
+      "learning_rate": 4.8482880074602955e-06,
+      "loss": 0.742,
+      "step": 32207
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.7783228158950806,
+      "learning_rate": 4.84802447089126e-06,
+      "loss": 0.5324,
+      "step": 32208
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.0282492637634277,
+      "learning_rate": 4.847760934744811e-06,
+      "loss": 0.6243,
+      "step": 32209
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 3.1976616382598877,
+      "learning_rate": 4.847497399021682e-06,
+      "loss": 0.5561,
+      "step": 32210
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 3.4947049617767334,
+      "learning_rate": 4.847233863722604e-06,
+      "loss": 0.7774,
+      "step": 32211
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 3.334836006164551,
+      "learning_rate": 4.846970328848311e-06,
+      "loss": 0.9713,
+      "step": 32212
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 3.417064666748047,
+      "learning_rate": 4.846706794399538e-06,
+      "loss": 0.7453,
+      "step": 32213
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 3.5679337978363037,
+      "learning_rate": 4.846443260377017e-06,
+      "loss": 0.8559,
+      "step": 32214
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.271287441253662,
+      "learning_rate": 4.846179726781478e-06,
+      "loss": 0.4814,
+      "step": 32215
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.544687271118164,
+      "learning_rate": 4.845916193613656e-06,
+      "loss": 0.4331,
+      "step": 32216
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 4.0844292640686035,
+      "learning_rate": 4.8456526608742835e-06,
+      "loss": 0.4747,
+      "step": 32217
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.9101359844207764,
+      "learning_rate": 4.845389128564093e-06,
+      "loss": 0.7161,
+      "step": 32218
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.976719617843628,
+      "learning_rate": 4.845125596683819e-06,
+      "loss": 0.669,
+      "step": 32219
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.2678182125091553,
+      "learning_rate": 4.84486206523419e-06,
+      "loss": 0.5274,
+      "step": 32220
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.5675151348114014,
+      "learning_rate": 4.8445985342159436e-06,
+      "loss": 0.58,
+      "step": 32221
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 3.5365447998046875,
+      "learning_rate": 4.84433500362981e-06,
+      "loss": 0.8001,
+      "step": 32222
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.3803634643554688,
+      "learning_rate": 4.844071473476522e-06,
+      "loss": 0.7047,
+      "step": 32223
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 3.632246732711792,
+      "learning_rate": 4.843807943756813e-06,
+      "loss": 0.8487,
+      "step": 32224
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.8565263748168945,
+      "learning_rate": 4.843544414471415e-06,
+      "loss": 0.6231,
+      "step": 32225
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.5546042919158936,
+      "learning_rate": 4.843280885621062e-06,
+      "loss": 0.5633,
+      "step": 32226
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.6013760566711426,
+      "learning_rate": 4.8430173572064865e-06,
+      "loss": 0.6875,
+      "step": 32227
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.7958917617797852,
+      "learning_rate": 4.842753829228421e-06,
+      "loss": 0.5032,
+      "step": 32228
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.2966020107269287,
+      "learning_rate": 4.842490301687597e-06,
+      "loss": 0.6759,
+      "step": 32229
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.4416120052337646,
+      "learning_rate": 4.842226774584749e-06,
+      "loss": 0.6014,
+      "step": 32230
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.8876378536224365,
+      "learning_rate": 4.841963247920609e-06,
+      "loss": 0.4893,
+      "step": 32231
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.375725746154785,
+      "learning_rate": 4.841699721695909e-06,
+      "loss": 0.3506,
+      "step": 32232
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.9710285663604736,
+      "learning_rate": 4.841436195911385e-06,
+      "loss": 0.5797,
+      "step": 32233
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.72222638130188,
+      "learning_rate": 4.841172670567766e-06,
+      "loss": 0.7589,
+      "step": 32234
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.946075439453125,
+      "learning_rate": 4.840909145665787e-06,
+      "loss": 0.6655,
+      "step": 32235
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 4.1199951171875,
+      "learning_rate": 4.84064562120618e-06,
+      "loss": 0.5936,
+      "step": 32236
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.945704698562622,
+      "learning_rate": 4.840382097189678e-06,
+      "loss": 0.7073,
+      "step": 32237
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.9056260585784912,
+      "learning_rate": 4.8401185736170135e-06,
+      "loss": 0.3599,
+      "step": 32238
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 3.122523546218872,
+      "learning_rate": 4.8398550504889195e-06,
+      "loss": 0.6799,
+      "step": 32239
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.8192663192749023,
+      "learning_rate": 4.8395915278061285e-06,
+      "loss": 0.6528,
+      "step": 32240
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.1299080848693848,
+      "learning_rate": 4.839328005569373e-06,
+      "loss": 0.5679,
+      "step": 32241
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.6589689254760742,
+      "learning_rate": 4.839064483779386e-06,
+      "loss": 0.7255,
+      "step": 32242
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.7500747442245483,
+      "learning_rate": 4.838800962436901e-06,
+      "loss": 0.4467,
+      "step": 32243
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 3.0173964500427246,
+      "learning_rate": 4.838537441542651e-06,
+      "loss": 0.6561,
+      "step": 32244
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.182594060897827,
+      "learning_rate": 4.838273921097366e-06,
+      "loss": 0.5657,
+      "step": 32245
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 3.4110050201416016,
+      "learning_rate": 4.838010401101782e-06,
+      "loss": 0.7981,
+      "step": 32246
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.989472508430481,
+      "learning_rate": 4.83774688155663e-06,
+      "loss": 0.9636,
+      "step": 32247
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.36445951461792,
+      "learning_rate": 4.837483362462643e-06,
+      "loss": 0.6444,
+      "step": 32248
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.2600302696228027,
+      "learning_rate": 4.837219843820552e-06,
+      "loss": 0.7397,
+      "step": 32249
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.6690908670425415,
+      "learning_rate": 4.836956325631094e-06,
+      "loss": 0.5996,
+      "step": 32250
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.041809320449829,
+      "learning_rate": 4.836692807894999e-06,
+      "loss": 0.9117,
+      "step": 32251
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.7785526514053345,
+      "learning_rate": 4.8364292906129996e-06,
+      "loss": 0.6039,
+      "step": 32252
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.707308053970337,
+      "learning_rate": 4.8361657737858295e-06,
+      "loss": 0.8493,
+      "step": 32253
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 5.537076950073242,
+      "learning_rate": 4.835902257414221e-06,
+      "loss": 0.6909,
+      "step": 32254
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.844775676727295,
+      "learning_rate": 4.835638741498907e-06,
+      "loss": 0.5438,
+      "step": 32255
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 3.030968189239502,
+      "learning_rate": 4.8353752260406195e-06,
+      "loss": 0.7577,
+      "step": 32256
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 4.400347709655762,
+      "learning_rate": 4.835111711040092e-06,
+      "loss": 0.5613,
+      "step": 32257
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.059589385986328,
+      "learning_rate": 4.8348481964980585e-06,
+      "loss": 0.826,
+      "step": 32258
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.149435520172119,
+      "learning_rate": 4.834584682415249e-06,
+      "loss": 0.5017,
+      "step": 32259
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 3.702885389328003,
+      "learning_rate": 4.834321168792398e-06,
+      "loss": 0.8616,
+      "step": 32260
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.058475971221924,
+      "learning_rate": 4.834057655630236e-06,
+      "loss": 0.58,
+      "step": 32261
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.4673359394073486,
+      "learning_rate": 4.833794142929499e-06,
+      "loss": 0.5076,
+      "step": 32262
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.8822810649871826,
+      "learning_rate": 4.833530630690918e-06,
+      "loss": 0.8601,
+      "step": 32263
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.7277259826660156,
+      "learning_rate": 4.833267118915226e-06,
+      "loss": 0.5221,
+      "step": 32264
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.595292806625366,
+      "learning_rate": 4.833003607603156e-06,
+      "loss": 0.9913,
+      "step": 32265
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.585718870162964,
+      "learning_rate": 4.832740096755441e-06,
+      "loss": 0.6311,
+      "step": 32266
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 3.7598698139190674,
+      "learning_rate": 4.832476586372812e-06,
+      "loss": 0.6405,
+      "step": 32267
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.873847007751465,
+      "learning_rate": 4.832213076456003e-06,
+      "loss": 0.634,
+      "step": 32268
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 4.412145137786865,
+      "learning_rate": 4.831949567005747e-06,
+      "loss": 0.41,
+      "step": 32269
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.4235782623291016,
+      "learning_rate": 4.831686058022776e-06,
+      "loss": 0.5324,
+      "step": 32270
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.5745913982391357,
+      "learning_rate": 4.831422549507823e-06,
+      "loss": 0.4487,
+      "step": 32271
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.3487210273742676,
+      "learning_rate": 4.831159041461621e-06,
+      "loss": 0.6045,
+      "step": 32272
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.8989322185516357,
+      "learning_rate": 4.8308955338849e-06,
+      "loss": 0.9864,
+      "step": 32273
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.658713698387146,
+      "learning_rate": 4.830632026778398e-06,
+      "loss": 0.6678,
+      "step": 32274
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.9333175420761108,
+      "learning_rate": 4.830368520142844e-06,
+      "loss": 0.6343,
+      "step": 32275
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.011784791946411,
+      "learning_rate": 4.830105013978973e-06,
+      "loss": 0.8247,
+      "step": 32276
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.7735822200775146,
+      "learning_rate": 4.8298415082875144e-06,
+      "loss": 0.7099,
+      "step": 32277
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.6350831985473633,
+      "learning_rate": 4.829578003069204e-06,
+      "loss": 0.5258,
+      "step": 32278
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 5.931538105010986,
+      "learning_rate": 4.829314498324773e-06,
+      "loss": 0.6024,
+      "step": 32279
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.9999728202819824,
+      "learning_rate": 4.829050994054955e-06,
+      "loss": 0.5595,
+      "step": 32280
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.7955477237701416,
+      "learning_rate": 4.828787490260482e-06,
+      "loss": 0.6225,
+      "step": 32281
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.142632484436035,
+      "learning_rate": 4.828523986942086e-06,
+      "loss": 0.4195,
+      "step": 32282
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 4.063715934753418,
+      "learning_rate": 4.828260484100501e-06,
+      "loss": 0.5667,
+      "step": 32283
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.401637077331543,
+      "learning_rate": 4.827996981736459e-06,
+      "loss": 0.5799,
+      "step": 32284
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.4728271961212158,
+      "learning_rate": 4.827733479850693e-06,
+      "loss": 0.6239,
+      "step": 32285
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.7654508352279663,
+      "learning_rate": 4.827469978443937e-06,
+      "loss": 0.6642,
+      "step": 32286
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.9573829174041748,
+      "learning_rate": 4.827206477516921e-06,
+      "loss": 0.6477,
+      "step": 32287
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 6.956393241882324,
+      "learning_rate": 4.826942977070379e-06,
+      "loss": 0.6236,
+      "step": 32288
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.926218271255493,
+      "learning_rate": 4.826679477105044e-06,
+      "loss": 0.5554,
+      "step": 32289
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.890881299972534,
+      "learning_rate": 4.826415977621649e-06,
+      "loss": 0.5765,
+      "step": 32290
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 3.5003621578216553,
+      "learning_rate": 4.826152478620925e-06,
+      "loss": 0.5924,
+      "step": 32291
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 3.9760549068450928,
+      "learning_rate": 4.825888980103607e-06,
+      "loss": 0.5098,
+      "step": 32292
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.2616941928863525,
+      "learning_rate": 4.825625482070424e-06,
+      "loss": 0.5874,
+      "step": 32293
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.6917715072631836,
+      "learning_rate": 4.825361984522113e-06,
+      "loss": 0.491,
+      "step": 32294
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 4.698920726776123,
+      "learning_rate": 4.825098487459405e-06,
+      "loss": 0.7235,
+      "step": 32295
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 3.076974630355835,
+      "learning_rate": 4.8248349908830336e-06,
+      "loss": 0.6417,
+      "step": 32296
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 3.4566357135772705,
+      "learning_rate": 4.82457149479373e-06,
+      "loss": 0.4711,
+      "step": 32297
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.780380964279175,
+      "learning_rate": 4.824307999192226e-06,
+      "loss": 0.7811,
+      "step": 32298
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 3.8676159381866455,
+      "learning_rate": 4.824044504079256e-06,
+      "loss": 0.4726,
+      "step": 32299
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.087681293487549,
+      "learning_rate": 4.823781009455554e-06,
+      "loss": 0.4701,
+      "step": 32300
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.6470304727554321,
+      "learning_rate": 4.82351751532185e-06,
+      "loss": 0.5218,
+      "step": 32301
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 4.1836347579956055,
+      "learning_rate": 4.8232540216788774e-06,
+      "loss": 0.9076,
+      "step": 32302
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.9973050355911255,
+      "learning_rate": 4.822990528527369e-06,
+      "loss": 0.824,
+      "step": 32303
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.112279176712036,
+      "learning_rate": 4.822727035868058e-06,
+      "loss": 0.7137,
+      "step": 32304
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.6178761720657349,
+      "learning_rate": 4.822463543701677e-06,
+      "loss": 0.6338,
+      "step": 32305
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.078639030456543,
+      "learning_rate": 4.822200052028958e-06,
+      "loss": 0.7639,
+      "step": 32306
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.9836288690567017,
+      "learning_rate": 4.821936560850634e-06,
+      "loss": 0.4263,
+      "step": 32307
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.772642135620117,
+      "learning_rate": 4.821673070167438e-06,
+      "loss": 0.7931,
+      "step": 32308
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.9677791595458984,
+      "learning_rate": 4.821409579980103e-06,
+      "loss": 0.4795,
+      "step": 32309
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.7445051670074463,
+      "learning_rate": 4.82114609028936e-06,
+      "loss": 0.5943,
+      "step": 32310
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.2413692474365234,
+      "learning_rate": 4.820882601095942e-06,
+      "loss": 0.5854,
+      "step": 32311
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 5.590824604034424,
+      "learning_rate": 4.820619112400583e-06,
+      "loss": 0.7316,
+      "step": 32312
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.356630802154541,
+      "learning_rate": 4.820355624204014e-06,
+      "loss": 0.737,
+      "step": 32313
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.088343381881714,
+      "learning_rate": 4.8200921365069705e-06,
+      "loss": 0.5747,
+      "step": 32314
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.2280988693237305,
+      "learning_rate": 4.8198286493101835e-06,
+      "loss": 0.6867,
+      "step": 32315
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.815918207168579,
+      "learning_rate": 4.819565162614384e-06,
+      "loss": 0.6029,
+      "step": 32316
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.6401535272598267,
+      "learning_rate": 4.819301676420308e-06,
+      "loss": 0.3695,
+      "step": 32317
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 3.3599987030029297,
+      "learning_rate": 4.819038190728686e-06,
+      "loss": 0.7508,
+      "step": 32318
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 5.171523571014404,
+      "learning_rate": 4.818774705540251e-06,
+      "loss": 0.5223,
+      "step": 32319
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 3.7887892723083496,
+      "learning_rate": 4.818511220855735e-06,
+      "loss": 0.498,
+      "step": 32320
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 3.0560340881347656,
+      "learning_rate": 4.8182477366758716e-06,
+      "loss": 0.686,
+      "step": 32321
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.8986777067184448,
+      "learning_rate": 4.817984253001392e-06,
+      "loss": 0.5999,
+      "step": 32322
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.732285499572754,
+      "learning_rate": 4.8177207698330305e-06,
+      "loss": 0.7588,
+      "step": 32323
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.0248398780822754,
+      "learning_rate": 4.8174572871715206e-06,
+      "loss": 0.462,
+      "step": 32324
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 3.242151975631714,
+      "learning_rate": 4.8171938050175925e-06,
+      "loss": 0.4592,
+      "step": 32325
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.575556993484497,
+      "learning_rate": 4.816930323371981e-06,
+      "loss": 0.5153,
+      "step": 32326
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.063382387161255,
+      "learning_rate": 4.8166668422354165e-06,
+      "loss": 0.465,
+      "step": 32327
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.6669784784317017,
+      "learning_rate": 4.816403361608634e-06,
+      "loss": 0.7441,
+      "step": 32328
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 3.1809117794036865,
+      "learning_rate": 4.816139881492364e-06,
+      "loss": 0.864,
+      "step": 32329
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.8186596632003784,
+      "learning_rate": 4.815876401887341e-06,
+      "loss": 0.6911,
+      "step": 32330
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.206157922744751,
+      "learning_rate": 4.815612922794295e-06,
+      "loss": 0.3515,
+      "step": 32331
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.6840217113494873,
+      "learning_rate": 4.815349444213962e-06,
+      "loss": 0.4802,
+      "step": 32332
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.8988587856292725,
+      "learning_rate": 4.815085966147073e-06,
+      "loss": 0.6176,
+      "step": 32333
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.7315672636032104,
+      "learning_rate": 4.814822488594358e-06,
+      "loss": 0.6441,
+      "step": 32334
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 3.5879101753234863,
+      "learning_rate": 4.814559011556554e-06,
+      "loss": 0.8136,
+      "step": 32335
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.9799031019210815,
+      "learning_rate": 4.814295535034393e-06,
+      "loss": 0.3246,
+      "step": 32336
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.9726338386535645,
+      "learning_rate": 4.814032059028606e-06,
+      "loss": 0.5189,
+      "step": 32337
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.2952725887298584,
+      "learning_rate": 4.813768583539926e-06,
+      "loss": 0.7013,
+      "step": 32338
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 4.378575801849365,
+      "learning_rate": 4.813505108569087e-06,
+      "loss": 0.4441,
+      "step": 32339
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.531792640686035,
+      "learning_rate": 4.8132416341168185e-06,
+      "loss": 0.5941,
+      "step": 32340
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.7152822017669678,
+      "learning_rate": 4.812978160183857e-06,
+      "loss": 0.5742,
+      "step": 32341
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.952870488166809,
+      "learning_rate": 4.812714686770932e-06,
+      "loss": 0.7193,
+      "step": 32342
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 3.5344078540802,
+      "learning_rate": 4.812451213878777e-06,
+      "loss": 0.6231,
+      "step": 32343
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.9596000909805298,
+      "learning_rate": 4.8121877415081255e-06,
+      "loss": 0.6993,
+      "step": 32344
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 4.18306827545166,
+      "learning_rate": 4.811924269659709e-06,
+      "loss": 0.745,
+      "step": 32345
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 3.1251397132873535,
+      "learning_rate": 4.8116607983342616e-06,
+      "loss": 0.9889,
+      "step": 32346
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.6117136478424072,
+      "learning_rate": 4.811397327532515e-06,
+      "loss": 0.6513,
+      "step": 32347
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 3.010934591293335,
+      "learning_rate": 4.811133857255201e-06,
+      "loss": 0.8049,
+      "step": 32348
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.5113701820373535,
+      "learning_rate": 4.810870387503053e-06,
+      "loss": 0.6576,
+      "step": 32349
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 5.539472579956055,
+      "learning_rate": 4.810606918276804e-06,
+      "loss": 0.5799,
+      "step": 32350
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.7050508260726929,
+      "learning_rate": 4.810343449577186e-06,
+      "loss": 0.6573,
+      "step": 32351
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 3.566819667816162,
+      "learning_rate": 4.810079981404932e-06,
+      "loss": 0.5472,
+      "step": 32352
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 3.777756690979004,
+      "learning_rate": 4.809816513760775e-06,
+      "loss": 0.4676,
+      "step": 32353
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.3651201725006104,
+      "learning_rate": 4.809553046645443e-06,
+      "loss": 0.6949,
+      "step": 32354
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 3.219083070755005,
+      "learning_rate": 4.809289580059676e-06,
+      "loss": 0.7293,
+      "step": 32355
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.1965675354003906,
+      "learning_rate": 4.8090261140042035e-06,
+      "loss": 0.6926,
+      "step": 32356
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.8684237003326416,
+      "learning_rate": 4.808762648479758e-06,
+      "loss": 0.76,
+      "step": 32357
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.0632169246673584,
+      "learning_rate": 4.808499183487071e-06,
+      "loss": 0.6061,
+      "step": 32358
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.294490098953247,
+      "learning_rate": 4.808235719026876e-06,
+      "loss": 0.3516,
+      "step": 32359
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.9986127614974976,
+      "learning_rate": 4.807972255099906e-06,
+      "loss": 0.7866,
+      "step": 32360
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 3.349414825439453,
+      "learning_rate": 4.807708791706895e-06,
+      "loss": 0.5535,
+      "step": 32361
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.5022915601730347,
+      "learning_rate": 4.807445328848571e-06,
+      "loss": 0.4,
+      "step": 32362
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 5.592823505401611,
+      "learning_rate": 4.807181866525671e-06,
+      "loss": 0.9564,
+      "step": 32363
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.8270680904388428,
+      "learning_rate": 4.806918404738925e-06,
+      "loss": 0.5434,
+      "step": 32364
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.928945779800415,
+      "learning_rate": 4.806654943489067e-06,
+      "loss": 0.5709,
+      "step": 32365
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.9941043853759766,
+      "learning_rate": 4.80639148277683e-06,
+      "loss": 0.8455,
+      "step": 32366
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.957161545753479,
+      "learning_rate": 4.806128022602945e-06,
+      "loss": 0.5342,
+      "step": 32367
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.373149871826172,
+      "learning_rate": 4.805864562968145e-06,
+      "loss": 0.6118,
+      "step": 32368
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.9958082437515259,
+      "learning_rate": 4.8056011038731636e-06,
+      "loss": 0.439,
+      "step": 32369
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.0388436317443848,
+      "learning_rate": 4.805337645318733e-06,
+      "loss": 0.7131,
+      "step": 32370
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.2749593257904053,
+      "learning_rate": 4.805074187305585e-06,
+      "loss": 0.7486,
+      "step": 32371
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.398444175720215,
+      "learning_rate": 4.804810729834452e-06,
+      "loss": 0.8663,
+      "step": 32372
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.9860203266143799,
+      "learning_rate": 4.804547272906067e-06,
+      "loss": 0.672,
+      "step": 32373
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.7084450721740723,
+      "learning_rate": 4.804283816521163e-06,
+      "loss": 0.6109,
+      "step": 32374
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.7458715438842773,
+      "learning_rate": 4.8040203606804735e-06,
+      "loss": 0.796,
+      "step": 32375
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.1495158672332764,
+      "learning_rate": 4.803756905384729e-06,
+      "loss": 0.5931,
+      "step": 32376
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.7190005779266357,
+      "learning_rate": 4.803493450634664e-06,
+      "loss": 0.7732,
+      "step": 32377
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.4948525428771973,
+      "learning_rate": 4.803229996431009e-06,
+      "loss": 0.3985,
+      "step": 32378
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.6208062171936035,
+      "learning_rate": 4.8029665427744985e-06,
+      "loss": 0.4604,
+      "step": 32379
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.8304152488708496,
+      "learning_rate": 4.802703089665864e-06,
+      "loss": 0.6738,
+      "step": 32380
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 4.19019889831543,
+      "learning_rate": 4.802439637105838e-06,
+      "loss": 0.8327,
+      "step": 32381
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.4297029972076416,
+      "learning_rate": 4.802176185095155e-06,
+      "loss": 0.7625,
+      "step": 32382
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.808209180831909,
+      "learning_rate": 4.801912733634544e-06,
+      "loss": 0.738,
+      "step": 32383
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 6.295312404632568,
+      "learning_rate": 4.801649282724739e-06,
+      "loss": 0.7951,
+      "step": 32384
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 3.1714301109313965,
+      "learning_rate": 4.8013858323664745e-06,
+      "loss": 0.6982,
+      "step": 32385
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.416081190109253,
+      "learning_rate": 4.801122382560481e-06,
+      "loss": 0.7062,
+      "step": 32386
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.495666265487671,
+      "learning_rate": 4.800858933307492e-06,
+      "loss": 0.7228,
+      "step": 32387
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.8776359558105469,
+      "learning_rate": 4.8005954846082395e-06,
+      "loss": 0.5391,
+      "step": 32388
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.6201351881027222,
+      "learning_rate": 4.800332036463457e-06,
+      "loss": 0.7646,
+      "step": 32389
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.4976062774658203,
+      "learning_rate": 4.800068588873875e-06,
+      "loss": 0.5452,
+      "step": 32390
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.7732666730880737,
+      "learning_rate": 4.799805141840228e-06,
+      "loss": 0.5631,
+      "step": 32391
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.7859432697296143,
+      "learning_rate": 4.799541695363248e-06,
+      "loss": 0.8852,
+      "step": 32392
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.8974595069885254,
+      "learning_rate": 4.7992782494436675e-06,
+      "loss": 0.737,
+      "step": 32393
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.9053492546081543,
+      "learning_rate": 4.799014804082219e-06,
+      "loss": 0.699,
+      "step": 32394
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.8161983489990234,
+      "learning_rate": 4.7987513592796335e-06,
+      "loss": 0.5858,
+      "step": 32395
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 3.0589723587036133,
+      "learning_rate": 4.798487915036646e-06,
+      "loss": 0.5453,
+      "step": 32396
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.0145297050476074,
+      "learning_rate": 4.798224471353988e-06,
+      "loss": 0.5622,
+      "step": 32397
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.6678595542907715,
+      "learning_rate": 4.797961028232393e-06,
+      "loss": 0.5992,
+      "step": 32398
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.9540737867355347,
+      "learning_rate": 4.797697585672592e-06,
+      "loss": 0.7484,
+      "step": 32399
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.1092307567596436,
+      "learning_rate": 4.79743414367532e-06,
+      "loss": 0.6734,
+      "step": 32400
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 6.961771488189697,
+      "learning_rate": 4.797170702241306e-06,
+      "loss": 0.6542,
+      "step": 32401
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.4916558265686035,
+      "learning_rate": 4.796907261371285e-06,
+      "loss": 0.7742,
+      "step": 32402
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.9745888710021973,
+      "learning_rate": 4.7966438210659885e-06,
+      "loss": 0.767,
+      "step": 32403
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 3.9602560997009277,
+      "learning_rate": 4.7963803813261494e-06,
+      "loss": 0.4035,
+      "step": 32404
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.204115629196167,
+      "learning_rate": 4.7961169421525e-06,
+      "loss": 0.8253,
+      "step": 32405
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.392073154449463,
+      "learning_rate": 4.7958535035457735e-06,
+      "loss": 0.6268,
+      "step": 32406
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 5.5574421882629395,
+      "learning_rate": 4.795590065506702e-06,
+      "loss": 0.5853,
+      "step": 32407
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.222317934036255,
+      "learning_rate": 4.795326628036018e-06,
+      "loss": 0.6082,
+      "step": 32408
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.969486117362976,
+      "learning_rate": 4.795063191134453e-06,
+      "loss": 0.6107,
+      "step": 32409
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.9718297719955444,
+      "learning_rate": 4.794799754802742e-06,
+      "loss": 0.535,
+      "step": 32410
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.0071372985839844,
+      "learning_rate": 4.7945363190416144e-06,
+      "loss": 0.727,
+      "step": 32411
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 3.627866506576538,
+      "learning_rate": 4.794272883851805e-06,
+      "loss": 0.8151,
+      "step": 32412
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.621711254119873,
+      "learning_rate": 4.794009449234045e-06,
+      "loss": 0.6906,
+      "step": 32413
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.015456199645996,
+      "learning_rate": 4.793746015189069e-06,
+      "loss": 0.7735,
+      "step": 32414
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 3.393711805343628,
+      "learning_rate": 4.793482581717605e-06,
+      "loss": 0.9443,
+      "step": 32415
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.7225757837295532,
+      "learning_rate": 4.793219148820391e-06,
+      "loss": 0.6748,
+      "step": 32416
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.7986564636230469,
+      "learning_rate": 4.792955716498157e-06,
+      "loss": 0.5224,
+      "step": 32417
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 3.1371285915374756,
+      "learning_rate": 4.792692284751635e-06,
+      "loss": 0.5217,
+      "step": 32418
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 3.864300489425659,
+      "learning_rate": 4.79242885358156e-06,
+      "loss": 0.8026,
+      "step": 32419
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.8806959390640259,
+      "learning_rate": 4.792165422988661e-06,
+      "loss": 0.7216,
+      "step": 32420
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.2963991165161133,
+      "learning_rate": 4.791901992973672e-06,
+      "loss": 0.3725,
+      "step": 32421
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.617781400680542,
+      "learning_rate": 4.791638563537326e-06,
+      "loss": 0.6721,
+      "step": 32422
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.9646098613739014,
+      "learning_rate": 4.791375134680355e-06,
+      "loss": 0.6994,
+      "step": 32423
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 3.007457971572876,
+      "learning_rate": 4.791111706403491e-06,
+      "loss": 0.8436,
+      "step": 32424
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.1749496459960938,
+      "learning_rate": 4.790848278707468e-06,
+      "loss": 0.6403,
+      "step": 32425
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.341327667236328,
+      "learning_rate": 4.790584851593018e-06,
+      "loss": 0.7874,
+      "step": 32426
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.891208291053772,
+      "learning_rate": 4.790321425060872e-06,
+      "loss": 0.6952,
+      "step": 32427
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.095839023590088,
+      "learning_rate": 4.790057999111764e-06,
+      "loss": 0.5924,
+      "step": 32428
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.53766131401062,
+      "learning_rate": 4.789794573746427e-06,
+      "loss": 0.5972,
+      "step": 32429
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 4.1316304206848145,
+      "learning_rate": 4.789531148965591e-06,
+      "loss": 0.5876,
+      "step": 32430
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 3.060966968536377,
+      "learning_rate": 4.7892677247699905e-06,
+      "loss": 0.5104,
+      "step": 32431
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 3.4277725219726562,
+      "learning_rate": 4.789004301160357e-06,
+      "loss": 0.6642,
+      "step": 32432
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.8415294885635376,
+      "learning_rate": 4.788740878137425e-06,
+      "loss": 0.6795,
+      "step": 32433
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 7.842648983001709,
+      "learning_rate": 4.7884774557019235e-06,
+      "loss": 0.7058,
+      "step": 32434
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.271268129348755,
+      "learning_rate": 4.7882140338545865e-06,
+      "loss": 0.7911,
+      "step": 32435
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.189181089401245,
+      "learning_rate": 4.787950612596149e-06,
+      "loss": 0.5924,
+      "step": 32436
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.6688334941864014,
+      "learning_rate": 4.787687191927341e-06,
+      "loss": 0.638,
+      "step": 32437
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.1439828872680664,
+      "learning_rate": 4.787423771848896e-06,
+      "loss": 0.6478,
+      "step": 32438
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 4.825998783111572,
+      "learning_rate": 4.787160352361545e-06,
+      "loss": 0.6958,
+      "step": 32439
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.720921516418457,
+      "learning_rate": 4.786896933466021e-06,
+      "loss": 0.6819,
+      "step": 32440
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.941796064376831,
+      "learning_rate": 4.786633515163057e-06,
+      "loss": 0.6811,
+      "step": 32441
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.141075849533081,
+      "learning_rate": 4.786370097453386e-06,
+      "loss": 0.4901,
+      "step": 32442
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.249986171722412,
+      "learning_rate": 4.7861066803377395e-06,
+      "loss": 0.7118,
+      "step": 32443
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.808739185333252,
+      "learning_rate": 4.78584326381685e-06,
+      "loss": 0.609,
+      "step": 32444
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.439394474029541,
+      "learning_rate": 4.78557984789145e-06,
+      "loss": 0.4549,
+      "step": 32445
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.292536735534668,
+      "learning_rate": 4.785316432562271e-06,
+      "loss": 0.5943,
+      "step": 32446
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 4.884596824645996,
+      "learning_rate": 4.785053017830048e-06,
+      "loss": 0.9662,
+      "step": 32447
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 3.1050808429718018,
+      "learning_rate": 4.784789603695512e-06,
+      "loss": 0.7664,
+      "step": 32448
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 3.864428997039795,
+      "learning_rate": 4.7845261901593955e-06,
+      "loss": 0.8968,
+      "step": 32449
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.8737016916275024,
+      "learning_rate": 4.78426277722243e-06,
+      "loss": 0.6104,
+      "step": 32450
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.6234688758850098,
+      "learning_rate": 4.78399936488535e-06,
+      "loss": 0.7731,
+      "step": 32451
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.6842060089111328,
+      "learning_rate": 4.783735953148887e-06,
+      "loss": 0.6008,
+      "step": 32452
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.405869483947754,
+      "learning_rate": 4.7834725420137725e-06,
+      "loss": 0.7649,
+      "step": 32453
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.8298379182815552,
+      "learning_rate": 4.783209131480741e-06,
+      "loss": 0.7327,
+      "step": 32454
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.0293235778808594,
+      "learning_rate": 4.782945721550522e-06,
+      "loss": 0.6099,
+      "step": 32455
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.305859327316284,
+      "learning_rate": 4.78268231222385e-06,
+      "loss": 0.6992,
+      "step": 32456
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 3.2690114974975586,
+      "learning_rate": 4.782418903501456e-06,
+      "loss": 0.6746,
+      "step": 32457
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.028195858001709,
+      "learning_rate": 4.782155495384076e-06,
+      "loss": 0.618,
+      "step": 32458
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.8613253831863403,
+      "learning_rate": 4.7818920878724385e-06,
+      "loss": 0.5008,
+      "step": 32459
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.715092182159424,
+      "learning_rate": 4.781628680967278e-06,
+      "loss": 0.2753,
+      "step": 32460
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.073758125305176,
+      "learning_rate": 4.781365274669326e-06,
+      "loss": 0.6661,
+      "step": 32461
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.5498985052108765,
+      "learning_rate": 4.781101868979315e-06,
+      "loss": 0.6737,
+      "step": 32462
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.7959927320480347,
+      "learning_rate": 4.780838463897979e-06,
+      "loss": 0.491,
+      "step": 32463
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 6.538157939910889,
+      "learning_rate": 4.780575059426048e-06,
+      "loss": 0.6786,
+      "step": 32464
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.30515193939209,
+      "learning_rate": 4.780311655564256e-06,
+      "loss": 0.68,
+      "step": 32465
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.4150192737579346,
+      "learning_rate": 4.780048252313335e-06,
+      "loss": 0.7144,
+      "step": 32466
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.7229803800582886,
+      "learning_rate": 4.779784849674017e-06,
+      "loss": 0.5028,
+      "step": 32467
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 8.064190864562988,
+      "learning_rate": 4.779521447647035e-06,
+      "loss": 0.5698,
+      "step": 32468
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.4391167163848877,
+      "learning_rate": 4.779258046233122e-06,
+      "loss": 0.6187,
+      "step": 32469
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.9263389110565186,
+      "learning_rate": 4.778994645433009e-06,
+      "loss": 0.8286,
+      "step": 32470
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.1474204063415527,
+      "learning_rate": 4.77873124524743e-06,
+      "loss": 0.632,
+      "step": 32471
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.1159520149230957,
+      "learning_rate": 4.778467845677115e-06,
+      "loss": 0.682,
+      "step": 32472
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.2871577739715576,
+      "learning_rate": 4.778204446722798e-06,
+      "loss": 0.671,
+      "step": 32473
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.390242099761963,
+      "learning_rate": 4.777941048385213e-06,
+      "loss": 0.4724,
+      "step": 32474
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 5.042768955230713,
+      "learning_rate": 4.7776776506650895e-06,
+      "loss": 0.9982,
+      "step": 32475
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.8293473720550537,
+      "learning_rate": 4.777414253563159e-06,
+      "loss": 0.7735,
+      "step": 32476
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.8840444087982178,
+      "learning_rate": 4.777150857080159e-06,
+      "loss": 0.5917,
+      "step": 32477
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.395892381668091,
+      "learning_rate": 4.7768874612168185e-06,
+      "loss": 0.803,
+      "step": 32478
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.9148141145706177,
+      "learning_rate": 4.77662406597387e-06,
+      "loss": 0.6076,
+      "step": 32479
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 3.787081241607666,
+      "learning_rate": 4.776360671352046e-06,
+      "loss": 0.7187,
+      "step": 32480
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.150312662124634,
+      "learning_rate": 4.776097277352081e-06,
+      "loss": 0.6374,
+      "step": 32481
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.6522130966186523,
+      "learning_rate": 4.775833883974704e-06,
+      "loss": 0.4612,
+      "step": 32482
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.2033257484436035,
+      "learning_rate": 4.77557049122065e-06,
+      "loss": 0.7533,
+      "step": 32483
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.169703960418701,
+      "learning_rate": 4.775307099090649e-06,
+      "loss": 0.7424,
+      "step": 32484
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.4791064262390137,
+      "learning_rate": 4.775043707585436e-06,
+      "loss": 0.6465,
+      "step": 32485
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 3.1255455017089844,
+      "learning_rate": 4.7747803167057425e-06,
+      "loss": 0.6372,
+      "step": 32486
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.2497901916503906,
+      "learning_rate": 4.774516926452299e-06,
+      "loss": 0.4888,
+      "step": 32487
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.9808852672576904,
+      "learning_rate": 4.774253536825841e-06,
+      "loss": 0.5757,
+      "step": 32488
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.1385085582733154,
+      "learning_rate": 4.773990147827098e-06,
+      "loss": 0.7771,
+      "step": 32489
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.4324545860290527,
+      "learning_rate": 4.773726759456805e-06,
+      "loss": 0.483,
+      "step": 32490
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.3162944316864014,
+      "learning_rate": 4.773463371715692e-06,
+      "loss": 0.6563,
+      "step": 32491
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 3.5827205181121826,
+      "learning_rate": 4.773199984604494e-06,
+      "loss": 0.682,
+      "step": 32492
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.0595998764038086,
+      "learning_rate": 4.772936598123941e-06,
+      "loss": 0.5586,
+      "step": 32493
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.775061845779419,
+      "learning_rate": 4.7726732122747655e-06,
+      "loss": 0.7685,
+      "step": 32494
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.7007770538330078,
+      "learning_rate": 4.772409827057701e-06,
+      "loss": 0.5056,
+      "step": 32495
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.901859760284424,
+      "learning_rate": 4.772146442473478e-06,
+      "loss": 0.5862,
+      "step": 32496
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.855964183807373,
+      "learning_rate": 4.771883058522832e-06,
+      "loss": 0.3251,
+      "step": 32497
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.5379765033721924,
+      "learning_rate": 4.7716196752064944e-06,
+      "loss": 0.6737,
+      "step": 32498
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 3.2000300884246826,
+      "learning_rate": 4.771356292525197e-06,
+      "loss": 0.7841,
+      "step": 32499
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.168565511703491,
+      "learning_rate": 4.771092910479671e-06,
+      "loss": 0.62,
+      "step": 32500
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 3.5293381214141846,
+      "learning_rate": 4.770829529070651e-06,
+      "loss": 0.4067,
+      "step": 32501
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.8811157941818237,
+      "learning_rate": 4.770566148298867e-06,
+      "loss": 0.6625,
+      "step": 32502
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.4513752460479736,
+      "learning_rate": 4.7703027681650534e-06,
+      "loss": 0.7218,
+      "step": 32503
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.588099956512451,
+      "learning_rate": 4.770039388669942e-06,
+      "loss": 0.5046,
+      "step": 32504
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 3.219348192214966,
+      "learning_rate": 4.769776009814264e-06,
+      "loss": 0.809,
+      "step": 32505
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 4.916172027587891,
+      "learning_rate": 4.769512631598754e-06,
+      "loss": 0.5928,
+      "step": 32506
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 3.062993049621582,
+      "learning_rate": 4.769249254024142e-06,
+      "loss": 0.5976,
+      "step": 32507
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.6022720336914062,
+      "learning_rate": 4.768985877091161e-06,
+      "loss": 0.8568,
+      "step": 32508
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.9202499389648438,
+      "learning_rate": 4.768722500800544e-06,
+      "loss": 0.6016,
+      "step": 32509
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.3741116523742676,
+      "learning_rate": 4.768459125153023e-06,
+      "loss": 0.8378,
+      "step": 32510
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.3134872913360596,
+      "learning_rate": 4.7681957501493315e-06,
+      "loss": 0.5261,
+      "step": 32511
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.8268663883209229,
+      "learning_rate": 4.7679323757902e-06,
+      "loss": 0.7001,
+      "step": 32512
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.381326913833618,
+      "learning_rate": 4.767669002076362e-06,
+      "loss": 0.7009,
+      "step": 32513
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.0736494064331055,
+      "learning_rate": 4.767405629008549e-06,
+      "loss": 0.6101,
+      "step": 32514
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.8505947589874268,
+      "learning_rate": 4.767142256587495e-06,
+      "loss": 0.9102,
+      "step": 32515
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.011122465133667,
+      "learning_rate": 4.766878884813929e-06,
+      "loss": 0.7249,
+      "step": 32516
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 3.1298348903656006,
+      "learning_rate": 4.766615513688587e-06,
+      "loss": 0.6777,
+      "step": 32517
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.0754144191741943,
+      "learning_rate": 4.766352143212199e-06,
+      "loss": 0.7774,
+      "step": 32518
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 3.089860677719116,
+      "learning_rate": 4.766088773385498e-06,
+      "loss": 0.6567,
+      "step": 32519
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.0806241035461426,
+      "learning_rate": 4.765825404209217e-06,
+      "loss": 0.6175,
+      "step": 32520
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.6064510345458984,
+      "learning_rate": 4.7655620356840886e-06,
+      "loss": 0.6457,
+      "step": 32521
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.2419216632843018,
+      "learning_rate": 4.765298667810843e-06,
+      "loss": 0.5919,
+      "step": 32522
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.2071468830108643,
+      "learning_rate": 4.765035300590215e-06,
+      "loss": 0.7103,
+      "step": 32523
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 3.4318110942840576,
+      "learning_rate": 4.764771934022936e-06,
+      "loss": 0.6942,
+      "step": 32524
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 3.099210023880005,
+      "learning_rate": 4.764508568109738e-06,
+      "loss": 0.7144,
+      "step": 32525
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 4.174825191497803,
+      "learning_rate": 4.764245202851352e-06,
+      "loss": 0.5416,
+      "step": 32526
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.3536198139190674,
+      "learning_rate": 4.763981838248513e-06,
+      "loss": 0.7594,
+      "step": 32527
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.073934316635132,
+      "learning_rate": 4.763718474301952e-06,
+      "loss": 0.7062,
+      "step": 32528
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.4385457038879395,
+      "learning_rate": 4.763455111012402e-06,
+      "loss": 0.7066,
+      "step": 32529
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 3.9114675521850586,
+      "learning_rate": 4.763191748380594e-06,
+      "loss": 1.0627,
+      "step": 32530
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.8678207397460938,
+      "learning_rate": 4.762928386407261e-06,
+      "loss": 0.5534,
+      "step": 32531
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.539738655090332,
+      "learning_rate": 4.762665025093135e-06,
+      "loss": 0.9309,
+      "step": 32532
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.702017068862915,
+      "learning_rate": 4.76240166443895e-06,
+      "loss": 0.7812,
+      "step": 32533
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.311490774154663,
+      "learning_rate": 4.762138304445436e-06,
+      "loss": 0.6226,
+      "step": 32534
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.9759025573730469,
+      "learning_rate": 4.761874945113326e-06,
+      "loss": 0.6915,
+      "step": 32535
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 3.32513165473938,
+      "learning_rate": 4.761611586443353e-06,
+      "loss": 0.6245,
+      "step": 32536
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.035338878631592,
+      "learning_rate": 4.761348228436247e-06,
+      "loss": 0.6478,
+      "step": 32537
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.677823066711426,
+      "learning_rate": 4.761084871092744e-06,
+      "loss": 0.8365,
+      "step": 32538
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 3.6318211555480957,
+      "learning_rate": 4.760821514413574e-06,
+      "loss": 0.8357,
+      "step": 32539
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.596758484840393,
+      "learning_rate": 4.760558158399471e-06,
+      "loss": 0.7878,
+      "step": 32540
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.0613653659820557,
+      "learning_rate": 4.760294803051166e-06,
+      "loss": 0.4984,
+      "step": 32541
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 4.120770454406738,
+      "learning_rate": 4.76003144836939e-06,
+      "loss": 0.5814,
+      "step": 32542
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.6778026819229126,
+      "learning_rate": 4.7597680943548766e-06,
+      "loss": 0.6962,
+      "step": 32543
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.576756477355957,
+      "learning_rate": 4.7595047410083594e-06,
+      "loss": 0.5762,
+      "step": 32544
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.381955623626709,
+      "learning_rate": 4.759241388330569e-06,
+      "loss": 0.5429,
+      "step": 32545
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.4163525104522705,
+      "learning_rate": 4.758978036322238e-06,
+      "loss": 0.599,
+      "step": 32546
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.1747543811798096,
+      "learning_rate": 4.758714684984098e-06,
+      "loss": 0.5553,
+      "step": 32547
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.356426239013672,
+      "learning_rate": 4.758451334316884e-06,
+      "loss": 0.3865,
+      "step": 32548
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 4.011422634124756,
+      "learning_rate": 4.7581879843213246e-06,
+      "loss": 0.3393,
+      "step": 32549
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.4950170516967773,
+      "learning_rate": 4.757924634998155e-06,
+      "loss": 0.6986,
+      "step": 32550
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 5.124736309051514,
+      "learning_rate": 4.757661286348106e-06,
+      "loss": 0.7391,
+      "step": 32551
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.6454029083251953,
+      "learning_rate": 4.75739793837191e-06,
+      "loss": 0.2972,
+      "step": 32552
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.6860711574554443,
+      "learning_rate": 4.757134591070299e-06,
+      "loss": 0.5668,
+      "step": 32553
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.489018440246582,
+      "learning_rate": 4.756871244444006e-06,
+      "loss": 0.5952,
+      "step": 32554
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.2717695236206055,
+      "learning_rate": 4.7566078984937634e-06,
+      "loss": 0.3801,
+      "step": 32555
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 4.4191155433654785,
+      "learning_rate": 4.756344553220302e-06,
+      "loss": 0.8589,
+      "step": 32556
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.8542859554290771,
+      "learning_rate": 4.7560812086243544e-06,
+      "loss": 0.6437,
+      "step": 32557
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.7049496173858643,
+      "learning_rate": 4.755817864706655e-06,
+      "loss": 0.8149,
+      "step": 32558
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.2581193447113037,
+      "learning_rate": 4.755554521467934e-06,
+      "loss": 0.6591,
+      "step": 32559
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.9375674724578857,
+      "learning_rate": 4.755291178908926e-06,
+      "loss": 0.5721,
+      "step": 32560
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 3.2564635276794434,
+      "learning_rate": 4.7550278370303605e-06,
+      "loss": 0.9313,
+      "step": 32561
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.2959976196289062,
+      "learning_rate": 4.754764495832971e-06,
+      "loss": 0.5315,
+      "step": 32562
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.970959186553955,
+      "learning_rate": 4.754501155317489e-06,
+      "loss": 0.6322,
+      "step": 32563
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.562985897064209,
+      "learning_rate": 4.754237815484648e-06,
+      "loss": 0.5203,
+      "step": 32564
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.5582351684570312,
+      "learning_rate": 4.753974476335179e-06,
+      "loss": 0.7318,
+      "step": 32565
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.774108648300171,
+      "learning_rate": 4.7537111378698156e-06,
+      "loss": 0.7312,
+      "step": 32566
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.9344282150268555,
+      "learning_rate": 4.753447800089289e-06,
+      "loss": 0.6328,
+      "step": 32567
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.414222240447998,
+      "learning_rate": 4.753184462994331e-06,
+      "loss": 0.6318,
+      "step": 32568
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.611921787261963,
+      "learning_rate": 4.752921126585675e-06,
+      "loss": 0.7387,
+      "step": 32569
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 6.363185882568359,
+      "learning_rate": 4.752657790864052e-06,
+      "loss": 0.8254,
+      "step": 32570
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 3.4522149562835693,
+      "learning_rate": 4.752394455830196e-06,
+      "loss": 0.6629,
+      "step": 32571
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 3.0990495681762695,
+      "learning_rate": 4.752131121484838e-06,
+      "loss": 0.4368,
+      "step": 32572
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 3.201631784439087,
+      "learning_rate": 4.751867787828711e-06,
+      "loss": 0.8179,
+      "step": 32573
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 6.880754470825195,
+      "learning_rate": 4.751604454862546e-06,
+      "loss": 0.6682,
+      "step": 32574
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.937186360359192,
+      "learning_rate": 4.751341122587077e-06,
+      "loss": 0.6432,
+      "step": 32575
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.8995277881622314,
+      "learning_rate": 4.751077791003035e-06,
+      "loss": 0.7299,
+      "step": 32576
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.5484061241149902,
+      "learning_rate": 4.750814460111152e-06,
+      "loss": 0.4911,
+      "step": 32577
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.0105717182159424,
+      "learning_rate": 4.750551129912159e-06,
+      "loss": 0.6166,
+      "step": 32578
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.4273297786712646,
+      "learning_rate": 4.750287800406792e-06,
+      "loss": 0.4888,
+      "step": 32579
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.8655602931976318,
+      "learning_rate": 4.75002447159578e-06,
+      "loss": 0.9396,
+      "step": 32580
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.2127878665924072,
+      "learning_rate": 4.749761143479856e-06,
+      "loss": 0.628,
+      "step": 32581
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.9955353736877441,
+      "learning_rate": 4.749497816059754e-06,
+      "loss": 0.5201,
+      "step": 32582
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.4414777755737305,
+      "learning_rate": 4.749234489336203e-06,
+      "loss": 0.6133,
+      "step": 32583
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 5.684086799621582,
+      "learning_rate": 4.748971163309938e-06,
+      "loss": 0.8442,
+      "step": 32584
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.0875368118286133,
+      "learning_rate": 4.74870783798169e-06,
+      "loss": 0.4305,
+      "step": 32585
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 3.6554195880889893,
+      "learning_rate": 4.748444513352192e-06,
+      "loss": 0.6519,
+      "step": 32586
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.2515785694122314,
+      "learning_rate": 4.748181189422176e-06,
+      "loss": 0.7507,
+      "step": 32587
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.392656087875366,
+      "learning_rate": 4.747917866192372e-06,
+      "loss": 0.5851,
+      "step": 32588
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.8969074487686157,
+      "learning_rate": 4.747654543663515e-06,
+      "loss": 0.521,
+      "step": 32589
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 3.2819437980651855,
+      "learning_rate": 4.747391221836336e-06,
+      "loss": 0.6233,
+      "step": 32590
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.314492702484131,
+      "learning_rate": 4.747127900711568e-06,
+      "loss": 0.7479,
+      "step": 32591
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.1730024814605713,
+      "learning_rate": 4.746864580289941e-06,
+      "loss": 0.5803,
+      "step": 32592
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.5813591480255127,
+      "learning_rate": 4.746601260572191e-06,
+      "loss": 0.6535,
+      "step": 32593
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 4.101174354553223,
+      "learning_rate": 4.746337941559047e-06,
+      "loss": 0.5034,
+      "step": 32594
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.477534532546997,
+      "learning_rate": 4.746074623251242e-06,
+      "loss": 0.7908,
+      "step": 32595
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 3.6915948390960693,
+      "learning_rate": 4.745811305649508e-06,
+      "loss": 0.402,
+      "step": 32596
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.145526885986328,
+      "learning_rate": 4.745547988754577e-06,
+      "loss": 0.7787,
+      "step": 32597
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.8573302030563354,
+      "learning_rate": 4.7452846725671805e-06,
+      "loss": 0.4731,
+      "step": 32598
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 3.2785513401031494,
+      "learning_rate": 4.745021357088055e-06,
+      "loss": 0.9981,
+      "step": 32599
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.3044357299804688,
+      "learning_rate": 4.744758042317928e-06,
+      "loss": 0.4761,
+      "step": 32600
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.7418020963668823,
+      "learning_rate": 4.744494728257534e-06,
+      "loss": 0.3366,
+      "step": 32601
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.5902364253997803,
+      "learning_rate": 4.744231414907604e-06,
+      "loss": 0.7136,
+      "step": 32602
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 3.107131242752075,
+      "learning_rate": 4.743968102268871e-06,
+      "loss": 0.7505,
+      "step": 32603
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.719480037689209,
+      "learning_rate": 4.743704790342067e-06,
+      "loss": 0.5951,
+      "step": 32604
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.9590340852737427,
+      "learning_rate": 4.7434414791279236e-06,
+      "loss": 0.6863,
+      "step": 32605
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.3355872631073,
+      "learning_rate": 4.743178168627173e-06,
+      "loss": 0.4575,
+      "step": 32606
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 6.005536079406738,
+      "learning_rate": 4.742914858840548e-06,
+      "loss": 0.6634,
+      "step": 32607
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.096433162689209,
+      "learning_rate": 4.7426515497687805e-06,
+      "loss": 0.891,
+      "step": 32608
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 4.422102451324463,
+      "learning_rate": 4.742388241412603e-06,
+      "loss": 0.6098,
+      "step": 32609
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 4.260572910308838,
+      "learning_rate": 4.742124933772747e-06,
+      "loss": 1.0072,
+      "step": 32610
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.4920170307159424,
+      "learning_rate": 4.741861626849945e-06,
+      "loss": 0.5031,
+      "step": 32611
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.562183141708374,
+      "learning_rate": 4.7415983206449295e-06,
+      "loss": 0.5399,
+      "step": 32612
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.5915274620056152,
+      "learning_rate": 4.741335015158432e-06,
+      "loss": 0.5161,
+      "step": 32613
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.834010362625122,
+      "learning_rate": 4.741071710391185e-06,
+      "loss": 0.6331,
+      "step": 32614
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.5619165897369385,
+      "learning_rate": 4.74080840634392e-06,
+      "loss": 0.6854,
+      "step": 32615
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 3.0439586639404297,
+      "learning_rate": 4.74054510301737e-06,
+      "loss": 0.4643,
+      "step": 32616
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.428844451904297,
+      "learning_rate": 4.740281800412267e-06,
+      "loss": 0.4562,
+      "step": 32617
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 3.9361181259155273,
+      "learning_rate": 4.740018498529341e-06,
+      "loss": 0.5194,
+      "step": 32618
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.9006316661834717,
+      "learning_rate": 4.7397551973693275e-06,
+      "loss": 0.4556,
+      "step": 32619
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.099864959716797,
+      "learning_rate": 4.739491896932958e-06,
+      "loss": 0.7756,
+      "step": 32620
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 3.2502458095550537,
+      "learning_rate": 4.739228597220964e-06,
+      "loss": 0.6724,
+      "step": 32621
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.057805061340332,
+      "learning_rate": 4.7389652982340775e-06,
+      "loss": 0.5019,
+      "step": 32622
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 4.614099502563477,
+      "learning_rate": 4.738701999973031e-06,
+      "loss": 0.4949,
+      "step": 32623
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.482544422149658,
+      "learning_rate": 4.738438702438556e-06,
+      "loss": 0.7762,
+      "step": 32624
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.8653138875961304,
+      "learning_rate": 4.738175405631385e-06,
+      "loss": 0.6736,
+      "step": 32625
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.4486958980560303,
+      "learning_rate": 4.73791210955225e-06,
+      "loss": 0.631,
+      "step": 32626
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.089085817337036,
+      "learning_rate": 4.737648814201884e-06,
+      "loss": 0.3515,
+      "step": 32627
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.4763665199279785,
+      "learning_rate": 4.737385519581017e-06,
+      "loss": 0.4869,
+      "step": 32628
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 3.153353691101074,
+      "learning_rate": 4.737122225690384e-06,
+      "loss": 0.97,
+      "step": 32629
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 3.129101514816284,
+      "learning_rate": 4.7368589325307144e-06,
+      "loss": 0.6178,
+      "step": 32630
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.6406502723693848,
+      "learning_rate": 4.736595640102742e-06,
+      "loss": 0.6565,
+      "step": 32631
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.4409396648406982,
+      "learning_rate": 4.7363323484071974e-06,
+      "loss": 0.6293,
+      "step": 32632
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.5153355598449707,
+      "learning_rate": 4.736069057444815e-06,
+      "loss": 0.7513,
+      "step": 32633
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.1198108196258545,
+      "learning_rate": 4.735805767216325e-06,
+      "loss": 0.6686,
+      "step": 32634
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.938289999961853,
+      "learning_rate": 4.735542477722461e-06,
+      "loss": 0.6263,
+      "step": 32635
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.22491192817688,
+      "learning_rate": 4.7352791889639545e-06,
+      "loss": 0.3999,
+      "step": 32636
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.2819669246673584,
+      "learning_rate": 4.735015900941537e-06,
+      "loss": 0.5036,
+      "step": 32637
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.9720990657806396,
+      "learning_rate": 4.7347526136559405e-06,
+      "loss": 0.7832,
+      "step": 32638
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 4.053157329559326,
+      "learning_rate": 4.734489327107898e-06,
+      "loss": 0.8373,
+      "step": 32639
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.2833683490753174,
+      "learning_rate": 4.734226041298141e-06,
+      "loss": 0.6683,
+      "step": 32640
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.808568000793457,
+      "learning_rate": 4.733962756227401e-06,
+      "loss": 0.65,
+      "step": 32641
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.648561716079712,
+      "learning_rate": 4.733699471896413e-06,
+      "loss": 0.3545,
+      "step": 32642
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 4.45475959777832,
+      "learning_rate": 4.733436188305904e-06,
+      "loss": 0.5803,
+      "step": 32643
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.606731414794922,
+      "learning_rate": 4.733172905456612e-06,
+      "loss": 0.7085,
+      "step": 32644
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.430110216140747,
+      "learning_rate": 4.732909623349265e-06,
+      "loss": 0.7193,
+      "step": 32645
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.871328353881836,
+      "learning_rate": 4.732646341984597e-06,
+      "loss": 0.6185,
+      "step": 32646
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 3.5449278354644775,
+      "learning_rate": 4.732383061363339e-06,
+      "loss": 0.7535,
+      "step": 32647
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.7024426460266113,
+      "learning_rate": 4.732119781486223e-06,
+      "loss": 0.2599,
+      "step": 32648
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 4.87140417098999,
+      "learning_rate": 4.731856502353982e-06,
+      "loss": 0.8405,
+      "step": 32649
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.8546435832977295,
+      "learning_rate": 4.7315932239673474e-06,
+      "loss": 0.348,
+      "step": 32650
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.165559768676758,
+      "learning_rate": 4.731329946327052e-06,
+      "loss": 0.5479,
+      "step": 32651
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.058290958404541,
+      "learning_rate": 4.731066669433827e-06,
+      "loss": 0.3821,
+      "step": 32652
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.8841805458068848,
+      "learning_rate": 4.730803393288404e-06,
+      "loss": 0.6013,
+      "step": 32653
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.0735414028167725,
+      "learning_rate": 4.730540117891517e-06,
+      "loss": 0.6241,
+      "step": 32654
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.214484214782715,
+      "learning_rate": 4.7302768432438974e-06,
+      "loss": 0.6045,
+      "step": 32655
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.749853253364563,
+      "learning_rate": 4.730013569346276e-06,
+      "loss": 0.8462,
+      "step": 32656
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.4050333499908447,
+      "learning_rate": 4.729750296199387e-06,
+      "loss": 0.7643,
+      "step": 32657
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 3.2896323204040527,
+      "learning_rate": 4.729487023803959e-06,
+      "loss": 0.7709,
+      "step": 32658
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 3.356518507003784,
+      "learning_rate": 4.7292237521607255e-06,
+      "loss": 0.6905,
+      "step": 32659
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 3.502918004989624,
+      "learning_rate": 4.728960481270421e-06,
+      "loss": 0.5011,
+      "step": 32660
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.336909770965576,
+      "learning_rate": 4.7286972111337764e-06,
+      "loss": 0.5037,
+      "step": 32661
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 3.2365875244140625,
+      "learning_rate": 4.728433941751523e-06,
+      "loss": 0.4216,
+      "step": 32662
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.08705997467041,
+      "learning_rate": 4.7281706731243924e-06,
+      "loss": 0.5926,
+      "step": 32663
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 11.36307144165039,
+      "learning_rate": 4.727907405253119e-06,
+      "loss": 0.5297,
+      "step": 32664
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.0926554203033447,
+      "learning_rate": 4.7276441381384315e-06,
+      "loss": 0.5378,
+      "step": 32665
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.7243245840072632,
+      "learning_rate": 4.727380871781064e-06,
+      "loss": 0.4893,
+      "step": 32666
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.139300584793091,
+      "learning_rate": 4.7271176061817484e-06,
+      "loss": 0.6559,
+      "step": 32667
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.563415050506592,
+      "learning_rate": 4.726854341341217e-06,
+      "loss": 0.5328,
+      "step": 32668
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 3.751325845718384,
+      "learning_rate": 4.726591077260201e-06,
+      "loss": 0.6874,
+      "step": 32669
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.619387149810791,
+      "learning_rate": 4.726327813939432e-06,
+      "loss": 0.4467,
+      "step": 32670
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.1735992431640625,
+      "learning_rate": 4.726064551379644e-06,
+      "loss": 0.711,
+      "step": 32671
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.792236566543579,
+      "learning_rate": 4.7258012895815665e-06,
+      "loss": 0.5243,
+      "step": 32672
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 3.1344070434570312,
+      "learning_rate": 4.725538028545935e-06,
+      "loss": 0.65,
+      "step": 32673
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.6500556468963623,
+      "learning_rate": 4.7252747682734776e-06,
+      "loss": 0.7088,
+      "step": 32674
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 3.35251522064209,
+      "learning_rate": 4.725011508764929e-06,
+      "loss": 0.936,
+      "step": 32675
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.085433006286621,
+      "learning_rate": 4.724748250021019e-06,
+      "loss": 0.6252,
+      "step": 32676
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.6956536769866943,
+      "learning_rate": 4.724484992042483e-06,
+      "loss": 0.6896,
+      "step": 32677
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.094312906265259,
+      "learning_rate": 4.72422173483005e-06,
+      "loss": 0.6406,
+      "step": 32678
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 3.8740146160125732,
+      "learning_rate": 4.723958478384451e-06,
+      "loss": 0.6029,
+      "step": 32679
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.530756950378418,
+      "learning_rate": 4.723695222706423e-06,
+      "loss": 0.9548,
+      "step": 32680
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.5842500925064087,
+      "learning_rate": 4.723431967796694e-06,
+      "loss": 0.4293,
+      "step": 32681
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 3.543196678161621,
+      "learning_rate": 4.723168713655998e-06,
+      "loss": 0.641,
+      "step": 32682
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.6823694705963135,
+      "learning_rate": 4.722905460285065e-06,
+      "loss": 0.5102,
+      "step": 32683
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 3.1152074337005615,
+      "learning_rate": 4.7226422076846286e-06,
+      "loss": 0.6899,
+      "step": 32684
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.0422136783599854,
+      "learning_rate": 4.72237895585542e-06,
+      "loss": 0.7716,
+      "step": 32685
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.8023581504821777,
+      "learning_rate": 4.722115704798172e-06,
+      "loss": 0.5737,
+      "step": 32686
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 3.728132963180542,
+      "learning_rate": 4.7218524545136155e-06,
+      "loss": 0.7184,
+      "step": 32687
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.4086419343948364,
+      "learning_rate": 4.721589205002484e-06,
+      "loss": 0.4769,
+      "step": 32688
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.6081639528274536,
+      "learning_rate": 4.721325956265508e-06,
+      "loss": 0.4611,
+      "step": 32689
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.983519196510315,
+      "learning_rate": 4.721062708303421e-06,
+      "loss": 0.5782,
+      "step": 32690
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.8733561038970947,
+      "learning_rate": 4.7207994611169535e-06,
+      "loss": 0.4251,
+      "step": 32691
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.450094223022461,
+      "learning_rate": 4.720536214706838e-06,
+      "loss": 0.561,
+      "step": 32692
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 3.8425352573394775,
+      "learning_rate": 4.720272969073806e-06,
+      "loss": 0.9839,
+      "step": 32693
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 5.37490177154541,
+      "learning_rate": 4.72000972421859e-06,
+      "loss": 0.655,
+      "step": 32694
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 3.273681640625,
+      "learning_rate": 4.719746480141924e-06,
+      "loss": 0.7291,
+      "step": 32695
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.951241970062256,
+      "learning_rate": 4.719483236844537e-06,
+      "loss": 0.559,
+      "step": 32696
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.042740821838379,
+      "learning_rate": 4.719219994327161e-06,
+      "loss": 0.6875,
+      "step": 32697
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.7347699403762817,
+      "learning_rate": 4.71895675259053e-06,
+      "loss": 0.5142,
+      "step": 32698
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 4.097830772399902,
+      "learning_rate": 4.718693511635375e-06,
+      "loss": 0.8261,
+      "step": 32699
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.3256545066833496,
+      "learning_rate": 4.718430271462427e-06,
+      "loss": 0.534,
+      "step": 32700
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.299532175064087,
+      "learning_rate": 4.71816703207242e-06,
+      "loss": 0.5263,
+      "step": 32701
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.547379732131958,
+      "learning_rate": 4.717903793466084e-06,
+      "loss": 0.5238,
+      "step": 32702
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.22418212890625,
+      "learning_rate": 4.717640555644153e-06,
+      "loss": 0.3982,
+      "step": 32703
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.4725959300994873,
+      "learning_rate": 4.7173773186073565e-06,
+      "loss": 0.8059,
+      "step": 32704
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.4260311126708984,
+      "learning_rate": 4.717114082356429e-06,
+      "loss": 0.5098,
+      "step": 32705
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 3.298968553543091,
+      "learning_rate": 4.716850846892101e-06,
+      "loss": 0.3884,
+      "step": 32706
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.9056448936462402,
+      "learning_rate": 4.716587612215104e-06,
+      "loss": 0.6564,
+      "step": 32707
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.980433464050293,
+      "learning_rate": 4.716324378326172e-06,
+      "loss": 0.5698,
+      "step": 32708
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 4.166598796844482,
+      "learning_rate": 4.716061145226035e-06,
+      "loss": 0.7477,
+      "step": 32709
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.1694681644439697,
+      "learning_rate": 4.715797912915426e-06,
+      "loss": 0.6617,
+      "step": 32710
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 3.2517857551574707,
+      "learning_rate": 4.715534681395076e-06,
+      "loss": 0.3338,
+      "step": 32711
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.4647884368896484,
+      "learning_rate": 4.715271450665718e-06,
+      "loss": 0.7161,
+      "step": 32712
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.7545995712280273,
+      "learning_rate": 4.715008220728083e-06,
+      "loss": 0.6244,
+      "step": 32713
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.4700090885162354,
+      "learning_rate": 4.714744991582904e-06,
+      "loss": 0.7015,
+      "step": 32714
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.2265334129333496,
+      "learning_rate": 4.714481763230912e-06,
+      "loss": 0.4323,
+      "step": 32715
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.5864222049713135,
+      "learning_rate": 4.714218535672839e-06,
+      "loss": 0.5534,
+      "step": 32716
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 3.4961791038513184,
+      "learning_rate": 4.713955308909417e-06,
+      "loss": 0.5833,
+      "step": 32717
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.762697458267212,
+      "learning_rate": 4.713692082941378e-06,
+      "loss": 0.4643,
+      "step": 32718
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.8054883480072021,
+      "learning_rate": 4.7134288577694545e-06,
+      "loss": 0.5663,
+      "step": 32719
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.2896251678466797,
+      "learning_rate": 4.713165633394376e-06,
+      "loss": 0.5815,
+      "step": 32720
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.069692611694336,
+      "learning_rate": 4.7129024098168785e-06,
+      "loss": 0.8598,
+      "step": 32721
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.7542442083358765,
+      "learning_rate": 4.712639187037692e-06,
+      "loss": 0.6716,
+      "step": 32722
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.073591470718384,
+      "learning_rate": 4.712375965057548e-06,
+      "loss": 0.7286,
+      "step": 32723
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.785599946975708,
+      "learning_rate": 4.712112743877178e-06,
+      "loss": 0.8214,
+      "step": 32724
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.6685004234313965,
+      "learning_rate": 4.7118495234973164e-06,
+      "loss": 0.6586,
+      "step": 32725
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.344235897064209,
+      "learning_rate": 4.711586303918692e-06,
+      "loss": 0.5546,
+      "step": 32726
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.0339467525482178,
+      "learning_rate": 4.711323085142038e-06,
+      "loss": 0.6297,
+      "step": 32727
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.3668540716171265,
+      "learning_rate": 4.711059867168087e-06,
+      "loss": 0.6128,
+      "step": 32728
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.2851667404174805,
+      "learning_rate": 4.71079664999757e-06,
+      "loss": 0.8249,
+      "step": 32729
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.767122268676758,
+      "learning_rate": 4.710533433631219e-06,
+      "loss": 0.7572,
+      "step": 32730
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.8017305135726929,
+      "learning_rate": 4.710270218069766e-06,
+      "loss": 0.5364,
+      "step": 32731
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 4.284587860107422,
+      "learning_rate": 4.710007003313943e-06,
+      "loss": 0.6125,
+      "step": 32732
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.5112218856811523,
+      "learning_rate": 4.709743789364482e-06,
+      "loss": 0.6167,
+      "step": 32733
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 3.966465473175049,
+      "learning_rate": 4.709480576222115e-06,
+      "loss": 0.6026,
+      "step": 32734
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.257957696914673,
+      "learning_rate": 4.709217363887574e-06,
+      "loss": 0.7262,
+      "step": 32735
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.536637783050537,
+      "learning_rate": 4.70895415236159e-06,
+      "loss": 0.6268,
+      "step": 32736
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.367154598236084,
+      "learning_rate": 4.708690941644896e-06,
+      "loss": 0.3806,
+      "step": 32737
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.7280975580215454,
+      "learning_rate": 4.708427731738223e-06,
+      "loss": 0.4568,
+      "step": 32738
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.970104694366455,
+      "learning_rate": 4.7081645226423025e-06,
+      "loss": 0.6811,
+      "step": 32739
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.9387730360031128,
+      "learning_rate": 4.707901314357866e-06,
+      "loss": 0.6812,
+      "step": 32740
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.6901838779449463,
+      "learning_rate": 4.707638106885648e-06,
+      "loss": 0.4537,
+      "step": 32741
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.0326690673828125,
+      "learning_rate": 4.70737490022638e-06,
+      "loss": 0.6917,
+      "step": 32742
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.4698660373687744,
+      "learning_rate": 4.707111694380792e-06,
+      "loss": 0.5937,
+      "step": 32743
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.0457236766815186,
+      "learning_rate": 4.706848489349617e-06,
+      "loss": 0.445,
+      "step": 32744
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.6789698600769043,
+      "learning_rate": 4.706585285133585e-06,
+      "loss": 0.5865,
+      "step": 32745
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.160660743713379,
+      "learning_rate": 4.7063220817334315e-06,
+      "loss": 0.6709,
+      "step": 32746
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 3.240755796432495,
+      "learning_rate": 4.706058879149885e-06,
+      "loss": 0.4287,
+      "step": 32747
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.9171078205108643,
+      "learning_rate": 4.705795677383679e-06,
+      "loss": 0.6452,
+      "step": 32748
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.364473581314087,
+      "learning_rate": 4.705532476435546e-06,
+      "loss": 0.3637,
+      "step": 32749
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.939286947250366,
+      "learning_rate": 4.7052692763062156e-06,
+      "loss": 0.5655,
+      "step": 32750
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.8805750608444214,
+      "learning_rate": 4.705006076996421e-06,
+      "loss": 0.5386,
+      "step": 32751
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.3209106922149658,
+      "learning_rate": 4.704742878506895e-06,
+      "loss": 0.6166,
+      "step": 32752
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 5.146029949188232,
+      "learning_rate": 4.704479680838367e-06,
+      "loss": 0.7988,
+      "step": 32753
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.865497350692749,
+      "learning_rate": 4.70421648399157e-06,
+      "loss": 0.346,
+      "step": 32754
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.033573865890503,
+      "learning_rate": 4.703953287967238e-06,
+      "loss": 0.8091,
+      "step": 32755
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.0709595680236816,
+      "learning_rate": 4.7036900927661e-06,
+      "loss": 0.6294,
+      "step": 32756
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.248523712158203,
+      "learning_rate": 4.703426898388889e-06,
+      "loss": 0.6808,
+      "step": 32757
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.7662923336029053,
+      "learning_rate": 4.703163704836336e-06,
+      "loss": 0.6266,
+      "step": 32758
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.217221260070801,
+      "learning_rate": 4.702900512109175e-06,
+      "loss": 0.5537,
+      "step": 32759
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.4507291316986084,
+      "learning_rate": 4.702637320208136e-06,
+      "loss": 0.235,
+      "step": 32760
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.0416383743286133,
+      "learning_rate": 4.7023741291339495e-06,
+      "loss": 0.7207,
+      "step": 32761
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 3.1534714698791504,
+      "learning_rate": 4.702110938887351e-06,
+      "loss": 0.457,
+      "step": 32762
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.7533351182937622,
+      "learning_rate": 4.70184774946907e-06,
+      "loss": 0.5885,
+      "step": 32763
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.2166740894317627,
+      "learning_rate": 4.701584560879838e-06,
+      "loss": 0.7489,
+      "step": 32764
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.218262195587158,
+      "learning_rate": 4.7013213731203885e-06,
+      "loss": 0.4563,
+      "step": 32765
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.1342170238494873,
+      "learning_rate": 4.701058186191451e-06,
+      "loss": 0.7002,
+      "step": 32766
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 4.137709617614746,
+      "learning_rate": 4.700795000093759e-06,
+      "loss": 0.7019,
+      "step": 32767
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 5.368877410888672,
+      "learning_rate": 4.700531814828045e-06,
+      "loss": 0.6561,
+      "step": 32768
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.1950347423553467,
+      "learning_rate": 4.70026863039504e-06,
+      "loss": 0.8215,
+      "step": 32769
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.388287305831909,
+      "learning_rate": 4.7000054467954744e-06,
+      "loss": 0.7178,
+      "step": 32770
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.378840446472168,
+      "learning_rate": 4.6997422640300825e-06,
+      "loss": 0.7049,
+      "step": 32771
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.3666493892669678,
+      "learning_rate": 4.699479082099595e-06,
+      "loss": 0.6787,
+      "step": 32772
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.11677622795105,
+      "learning_rate": 4.699215901004742e-06,
+      "loss": 0.5767,
+      "step": 32773
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 3.1744279861450195,
+      "learning_rate": 4.698952720746258e-06,
+      "loss": 0.5294,
+      "step": 32774
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.547916889190674,
+      "learning_rate": 4.698689541324874e-06,
+      "loss": 0.683,
+      "step": 32775
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 3.034651041030884,
+      "learning_rate": 4.698426362741321e-06,
+      "loss": 0.4822,
+      "step": 32776
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.9900593757629395,
+      "learning_rate": 4.6981631849963314e-06,
+      "loss": 0.5424,
+      "step": 32777
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.391798973083496,
+      "learning_rate": 4.697900008090637e-06,
+      "loss": 0.6175,
+      "step": 32778
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.565577507019043,
+      "learning_rate": 4.69763683202497e-06,
+      "loss": 0.5448,
+      "step": 32779
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.6103999614715576,
+      "learning_rate": 4.697373656800061e-06,
+      "loss": 0.6381,
+      "step": 32780
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.4588450193405151,
+      "learning_rate": 4.69711048241664e-06,
+      "loss": 0.5672,
+      "step": 32781
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.8244409561157227,
+      "learning_rate": 4.696847308875445e-06,
+      "loss": 0.556,
+      "step": 32782
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 3.141292095184326,
+      "learning_rate": 4.696584136177203e-06,
+      "loss": 0.4237,
+      "step": 32783
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.6369911432266235,
+      "learning_rate": 4.696320964322646e-06,
+      "loss": 0.7586,
+      "step": 32784
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.539475202560425,
+      "learning_rate": 4.696057793312508e-06,
+      "loss": 0.5885,
+      "step": 32785
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.3987865447998047,
+      "learning_rate": 4.695794623147519e-06,
+      "loss": 0.6529,
+      "step": 32786
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.9378447532653809,
+      "learning_rate": 4.695531453828412e-06,
+      "loss": 0.7203,
+      "step": 32787
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.041630744934082,
+      "learning_rate": 4.695268285355916e-06,
+      "loss": 0.6074,
+      "step": 32788
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.6442914009094238,
+      "learning_rate": 4.6950051177307664e-06,
+      "loss": 0.6518,
+      "step": 32789
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.0707626342773438,
+      "learning_rate": 4.694741950953693e-06,
+      "loss": 0.6403,
+      "step": 32790
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 3.898695468902588,
+      "learning_rate": 4.694478785025427e-06,
+      "loss": 0.7141,
+      "step": 32791
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.193964958190918,
+      "learning_rate": 4.694215619946702e-06,
+      "loss": 0.6133,
+      "step": 32792
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.2055604457855225,
+      "learning_rate": 4.693952455718249e-06,
+      "loss": 0.7535,
+      "step": 32793
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.085484743118286,
+      "learning_rate": 4.693689292340799e-06,
+      "loss": 0.5756,
+      "step": 32794
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.186335802078247,
+      "learning_rate": 4.693426129815084e-06,
+      "loss": 0.5162,
+      "step": 32795
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.8274402618408203,
+      "learning_rate": 4.693162968141837e-06,
+      "loss": 0.4711,
+      "step": 32796
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.7672600746154785,
+      "learning_rate": 4.692899807321788e-06,
+      "loss": 0.503,
+      "step": 32797
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.1622469425201416,
+      "learning_rate": 4.69263664735567e-06,
+      "loss": 0.7763,
+      "step": 32798
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.9891471862792969,
+      "learning_rate": 4.6923734882442146e-06,
+      "loss": 0.7098,
+      "step": 32799
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 4.755631446838379,
+      "learning_rate": 4.692110329988153e-06,
+      "loss": 0.5486,
+      "step": 32800
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 3.087006092071533,
+      "learning_rate": 4.6918471725882145e-06,
+      "loss": 0.5074,
+      "step": 32801
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.7055158615112305,
+      "learning_rate": 4.691584016045136e-06,
+      "loss": 0.7022,
+      "step": 32802
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.151089668273926,
+      "learning_rate": 4.691320860359648e-06,
+      "loss": 0.5646,
+      "step": 32803
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.998888611793518,
+      "learning_rate": 4.6910577055324806e-06,
+      "loss": 0.6512,
+      "step": 32804
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.855164885520935,
+      "learning_rate": 4.690794551564365e-06,
+      "loss": 0.6007,
+      "step": 32805
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.1845195293426514,
+      "learning_rate": 4.690531398456034e-06,
+      "loss": 0.6183,
+      "step": 32806
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.94697904586792,
+      "learning_rate": 4.69026824620822e-06,
+      "loss": 0.7643,
+      "step": 32807
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.1099607944488525,
+      "learning_rate": 4.690005094821653e-06,
+      "loss": 0.6718,
+      "step": 32808
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.3241467475891113,
+      "learning_rate": 4.689741944297067e-06,
+      "loss": 0.5044,
+      "step": 32809
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.998600721359253,
+      "learning_rate": 4.689478794635192e-06,
+      "loss": 0.5497,
+      "step": 32810
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.918299913406372,
+      "learning_rate": 4.68921564583676e-06,
+      "loss": 0.5618,
+      "step": 32811
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.6272574663162231,
+      "learning_rate": 4.688952497902503e-06,
+      "loss": 0.4524,
+      "step": 32812
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.109215259552002,
+      "learning_rate": 4.688689350833153e-06,
+      "loss": 0.6331,
+      "step": 32813
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.4255011081695557,
+      "learning_rate": 4.688426204629441e-06,
+      "loss": 0.6816,
+      "step": 32814
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 3.3215346336364746,
+      "learning_rate": 4.688163059292098e-06,
+      "loss": 0.5926,
+      "step": 32815
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 4.020599365234375,
+      "learning_rate": 4.6878999148218565e-06,
+      "loss": 0.9697,
+      "step": 32816
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.3796274662017822,
+      "learning_rate": 4.68763677121945e-06,
+      "loss": 0.6245,
+      "step": 32817
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.888625979423523,
+      "learning_rate": 4.687373628485607e-06,
+      "loss": 0.3669,
+      "step": 32818
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.7579560279846191,
+      "learning_rate": 4.687110486621062e-06,
+      "loss": 0.4484,
+      "step": 32819
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.8590223789215088,
+      "learning_rate": 4.686847345626546e-06,
+      "loss": 0.7972,
+      "step": 32820
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.3671116828918457,
+      "learning_rate": 4.686584205502789e-06,
+      "loss": 0.7194,
+      "step": 32821
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 3.841012954711914,
+      "learning_rate": 4.686321066250523e-06,
+      "loss": 0.5546,
+      "step": 32822
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.510535955429077,
+      "learning_rate": 4.6860579278704825e-06,
+      "loss": 0.6443,
+      "step": 32823
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.5213096141815186,
+      "learning_rate": 4.6857947903633975e-06,
+      "loss": 0.5443,
+      "step": 32824
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.9499869346618652,
+      "learning_rate": 4.685531653729998e-06,
+      "loss": 0.571,
+      "step": 32825
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.070111036300659,
+      "learning_rate": 4.685268517971018e-06,
+      "loss": 0.6301,
+      "step": 32826
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.4201323986053467,
+      "learning_rate": 4.685005383087189e-06,
+      "loss": 0.4814,
+      "step": 32827
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.502577543258667,
+      "learning_rate": 4.68474224907924e-06,
+      "loss": 0.7521,
+      "step": 32828
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.8812601566314697,
+      "learning_rate": 4.6844791159479065e-06,
+      "loss": 0.4306,
+      "step": 32829
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.1238949298858643,
+      "learning_rate": 4.684215983693917e-06,
+      "loss": 0.7448,
+      "step": 32830
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.8302905559539795,
+      "learning_rate": 4.683952852318007e-06,
+      "loss": 0.6266,
+      "step": 32831
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.6993331909179688,
+      "learning_rate": 4.683689721820904e-06,
+      "loss": 0.7635,
+      "step": 32832
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.6063218116760254,
+      "learning_rate": 4.683426592203342e-06,
+      "loss": 0.787,
+      "step": 32833
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.7994405031204224,
+      "learning_rate": 4.683163463466051e-06,
+      "loss": 0.6333,
+      "step": 32834
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 4.364181995391846,
+      "learning_rate": 4.682900335609764e-06,
+      "loss": 1.1088,
+      "step": 32835
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.8629488945007324,
+      "learning_rate": 4.682637208635213e-06,
+      "loss": 0.9318,
+      "step": 32836
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.7647043466567993,
+      "learning_rate": 4.6823740825431294e-06,
+      "loss": 0.5727,
+      "step": 32837
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 5.482471466064453,
+      "learning_rate": 4.682110957334244e-06,
+      "loss": 0.5536,
+      "step": 32838
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 3.1749823093414307,
+      "learning_rate": 4.6818478330092885e-06,
+      "loss": 0.6826,
+      "step": 32839
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 3.539471387863159,
+      "learning_rate": 4.681584709568996e-06,
+      "loss": 0.4372,
+      "step": 32840
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.1622543334960938,
+      "learning_rate": 4.681321587014096e-06,
+      "loss": 0.7325,
+      "step": 32841
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.843239665031433,
+      "learning_rate": 4.6810584653453205e-06,
+      "loss": 0.549,
+      "step": 32842
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.1410698890686035,
+      "learning_rate": 4.680795344563403e-06,
+      "loss": 0.6325,
+      "step": 32843
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.2957968711853027,
+      "learning_rate": 4.680532224669074e-06,
+      "loss": 0.7662,
+      "step": 32844
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.747335195541382,
+      "learning_rate": 4.680269105663066e-06,
+      "loss": 0.6125,
+      "step": 32845
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.6489797830581665,
+      "learning_rate": 4.680005987546109e-06,
+      "loss": 0.5467,
+      "step": 32846
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.7714568376541138,
+      "learning_rate": 4.679742870318936e-06,
+      "loss": 0.33,
+      "step": 32847
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.987497329711914,
+      "learning_rate": 4.679479753982278e-06,
+      "loss": 0.5169,
+      "step": 32848
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.0240886211395264,
+      "learning_rate": 4.679216638536867e-06,
+      "loss": 0.7163,
+      "step": 32849
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 3.4109010696411133,
+      "learning_rate": 4.678953523983434e-06,
+      "loss": 0.3334,
+      "step": 32850
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.9475300312042236,
+      "learning_rate": 4.678690410322711e-06,
+      "loss": 0.6989,
+      "step": 32851
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.833620309829712,
+      "learning_rate": 4.678427297555429e-06,
+      "loss": 0.4494,
+      "step": 32852
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 3.0681283473968506,
+      "learning_rate": 4.678164185682321e-06,
+      "loss": 0.6982,
+      "step": 32853
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 3.2153167724609375,
+      "learning_rate": 4.677901074704118e-06,
+      "loss": 0.6017,
+      "step": 32854
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.0040154457092285,
+      "learning_rate": 4.677637964621551e-06,
+      "loss": 0.5643,
+      "step": 32855
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.5082128047943115,
+      "learning_rate": 4.677374855435352e-06,
+      "loss": 0.6734,
+      "step": 32856
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.495100498199463,
+      "learning_rate": 4.677111747146254e-06,
+      "loss": 0.5344,
+      "step": 32857
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.333188056945801,
+      "learning_rate": 4.676848639754986e-06,
+      "loss": 0.7681,
+      "step": 32858
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.4699487686157227,
+      "learning_rate": 4.6765855332622805e-06,
+      "loss": 0.7742,
+      "step": 32859
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.9622430801391602,
+      "learning_rate": 4.67632242766887e-06,
+      "loss": 0.7863,
+      "step": 32860
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 7.628264904022217,
+      "learning_rate": 4.676059322975486e-06,
+      "loss": 0.6224,
+      "step": 32861
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 4.3301262855529785,
+      "learning_rate": 4.675796219182857e-06,
+      "loss": 0.2447,
+      "step": 32862
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.808807611465454,
+      "learning_rate": 4.67553311629172e-06,
+      "loss": 0.6782,
+      "step": 32863
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.9595115184783936,
+      "learning_rate": 4.675270014302803e-06,
+      "loss": 0.4135,
+      "step": 32864
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.6055313348770142,
+      "learning_rate": 4.675006913216839e-06,
+      "loss": 0.6572,
+      "step": 32865
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.9107561111450195,
+      "learning_rate": 4.674743813034559e-06,
+      "loss": 0.6772,
+      "step": 32866
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.9022184610366821,
+      "learning_rate": 4.674480713756694e-06,
+      "loss": 0.7277,
+      "step": 32867
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 3.5322773456573486,
+      "learning_rate": 4.674217615383977e-06,
+      "loss": 0.5888,
+      "step": 32868
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.2237751483917236,
+      "learning_rate": 4.673954517917138e-06,
+      "loss": 0.7039,
+      "step": 32869
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.5524568557739258,
+      "learning_rate": 4.67369142135691e-06,
+      "loss": 0.6716,
+      "step": 32870
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.0728676319122314,
+      "learning_rate": 4.6734283257040235e-06,
+      "loss": 0.6721,
+      "step": 32871
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 3.2901313304901123,
+      "learning_rate": 4.6731652309592105e-06,
+      "loss": 0.5807,
+      "step": 32872
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 3.334216833114624,
+      "learning_rate": 4.672902137123203e-06,
+      "loss": 0.541,
+      "step": 32873
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.7638099193572998,
+      "learning_rate": 4.672639044196732e-06,
+      "loss": 0.6263,
+      "step": 32874
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.1114094257354736,
+      "learning_rate": 4.672375952180529e-06,
+      "loss": 0.5761,
+      "step": 32875
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.1152658462524414,
+      "learning_rate": 4.6721128610753255e-06,
+      "loss": 0.5665,
+      "step": 32876
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.3008873462677,
+      "learning_rate": 4.671849770881854e-06,
+      "loss": 0.6757,
+      "step": 32877
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.9610333442687988,
+      "learning_rate": 4.671586681600843e-06,
+      "loss": 0.3909,
+      "step": 32878
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 3.6575896739959717,
+      "learning_rate": 4.671323593233029e-06,
+      "loss": 0.6038,
+      "step": 32879
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.715756416320801,
+      "learning_rate": 4.671060505779141e-06,
+      "loss": 0.4383,
+      "step": 32880
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.5559489727020264,
+      "learning_rate": 4.67079741923991e-06,
+      "loss": 0.5022,
+      "step": 32881
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.828404426574707,
+      "learning_rate": 4.670534333616068e-06,
+      "loss": 0.5643,
+      "step": 32882
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.6459481716156006,
+      "learning_rate": 4.6702712489083445e-06,
+      "loss": 0.6843,
+      "step": 32883
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.124781370162964,
+      "learning_rate": 4.670008165117476e-06,
+      "loss": 0.4428,
+      "step": 32884
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.8846957683563232,
+      "learning_rate": 4.66974508224419e-06,
+      "loss": 0.7467,
+      "step": 32885
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.4289898872375488,
+      "learning_rate": 4.669482000289219e-06,
+      "loss": 0.5278,
+      "step": 32886
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.9639896154403687,
+      "learning_rate": 4.669218919253296e-06,
+      "loss": 0.6208,
+      "step": 32887
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.8527028560638428,
+      "learning_rate": 4.66895583913715e-06,
+      "loss": 0.7558,
+      "step": 32888
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.8435282707214355,
+      "learning_rate": 4.668692759941513e-06,
+      "loss": 0.5387,
+      "step": 32889
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 4.619948387145996,
+      "learning_rate": 4.668429681667119e-06,
+      "loss": 0.7472,
+      "step": 32890
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 4.382472991943359,
+      "learning_rate": 4.668166604314698e-06,
+      "loss": 0.7252,
+      "step": 32891
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.718219041824341,
+      "learning_rate": 4.6679035278849805e-06,
+      "loss": 0.9032,
+      "step": 32892
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.983997344970703,
+      "learning_rate": 4.6676404523787e-06,
+      "loss": 0.5497,
+      "step": 32893
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 3.092992067337036,
+      "learning_rate": 4.667377377796585e-06,
+      "loss": 0.4679,
+      "step": 32894
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.528428316116333,
+      "learning_rate": 4.667114304139371e-06,
+      "loss": 0.2866,
+      "step": 32895
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.105705738067627,
+      "learning_rate": 4.666851231407786e-06,
+      "loss": 0.3903,
+      "step": 32896
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 4.166171550750732,
+      "learning_rate": 4.666588159602563e-06,
+      "loss": 0.3944,
+      "step": 32897
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 3.8035056591033936,
+      "learning_rate": 4.666325088724434e-06,
+      "loss": 0.5052,
+      "step": 32898
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.936394453048706,
+      "learning_rate": 4.66606201877413e-06,
+      "loss": 0.742,
+      "step": 32899
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.543003797531128,
+      "learning_rate": 4.665798949752383e-06,
+      "loss": 0.5075,
+      "step": 32900
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.1107852458953857,
+      "learning_rate": 4.665535881659922e-06,
+      "loss": 0.7735,
+      "step": 32901
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 3.32675838470459,
+      "learning_rate": 4.665272814497482e-06,
+      "loss": 0.6534,
+      "step": 32902
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 3.0087437629699707,
+      "learning_rate": 4.66500974826579e-06,
+      "loss": 0.6999,
+      "step": 32903
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.6881262063980103,
+      "learning_rate": 4.664746682965583e-06,
+      "loss": 0.4708,
+      "step": 32904
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 3.50870680809021,
+      "learning_rate": 4.66448361859759e-06,
+      "loss": 0.6597,
+      "step": 32905
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.4398441314697266,
+      "learning_rate": 4.664220555162542e-06,
+      "loss": 0.4999,
+      "step": 32906
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.1988251209259033,
+      "learning_rate": 4.663957492661171e-06,
+      "loss": 0.5754,
+      "step": 32907
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 3.0259740352630615,
+      "learning_rate": 4.6636944310942084e-06,
+      "loss": 0.6,
+      "step": 32908
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.6497292518615723,
+      "learning_rate": 4.663431370462385e-06,
+      "loss": 0.2475,
+      "step": 32909
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 3.012077569961548,
+      "learning_rate": 4.663168310766434e-06,
+      "loss": 0.7259,
+      "step": 32910
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.504258394241333,
+      "learning_rate": 4.662905252007085e-06,
+      "loss": 0.5042,
+      "step": 32911
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.384697198867798,
+      "learning_rate": 4.66264219418507e-06,
+      "loss": 0.4796,
+      "step": 32912
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.711800456047058,
+      "learning_rate": 4.6623791373011215e-06,
+      "loss": 0.6626,
+      "step": 32913
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.601503372192383,
+      "learning_rate": 4.66211608135597e-06,
+      "loss": 0.7607,
+      "step": 32914
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.456777811050415,
+      "learning_rate": 4.661853026350346e-06,
+      "loss": 0.444,
+      "step": 32915
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.5866525173187256,
+      "learning_rate": 4.661589972284983e-06,
+      "loss": 0.6583,
+      "step": 32916
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.1083579063415527,
+      "learning_rate": 4.6613269191606115e-06,
+      "loss": 0.7001,
+      "step": 32917
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.7452399730682373,
+      "learning_rate": 4.661063866977963e-06,
+      "loss": 0.5677,
+      "step": 32918
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.979632616043091,
+      "learning_rate": 4.660800815737768e-06,
+      "loss": 0.5404,
+      "step": 32919
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 4.649864673614502,
+      "learning_rate": 4.6605377654407606e-06,
+      "loss": 0.5631,
+      "step": 32920
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 3.8419668674468994,
+      "learning_rate": 4.660274716087669e-06,
+      "loss": 0.7308,
+      "step": 32921
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 3.2000575065612793,
+      "learning_rate": 4.660011667679227e-06,
+      "loss": 0.9193,
+      "step": 32922
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.7378350496292114,
+      "learning_rate": 4.659748620216162e-06,
+      "loss": 0.6935,
+      "step": 32923
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 4.320656776428223,
+      "learning_rate": 4.659485573699211e-06,
+      "loss": 0.6812,
+      "step": 32924
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.910108804702759,
+      "learning_rate": 4.6592225281291045e-06,
+      "loss": 0.5176,
+      "step": 32925
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.340104579925537,
+      "learning_rate": 4.658959483506572e-06,
+      "loss": 0.5006,
+      "step": 32926
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.101147413253784,
+      "learning_rate": 4.658696439832344e-06,
+      "loss": 0.2914,
+      "step": 32927
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.192117929458618,
+      "learning_rate": 4.658433397107154e-06,
+      "loss": 0.571,
+      "step": 32928
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 3.1803953647613525,
+      "learning_rate": 4.6581703553317335e-06,
+      "loss": 0.5331,
+      "step": 32929
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.8240835666656494,
+      "learning_rate": 4.657907314506813e-06,
+      "loss": 0.6612,
+      "step": 32930
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.888165235519409,
+      "learning_rate": 4.657644274633124e-06,
+      "loss": 0.7266,
+      "step": 32931
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.7572972774505615,
+      "learning_rate": 4.657381235711398e-06,
+      "loss": 0.7773,
+      "step": 32932
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.302205801010132,
+      "learning_rate": 4.657118197742366e-06,
+      "loss": 0.6505,
+      "step": 32933
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 3.5648353099823,
+      "learning_rate": 4.6568551607267604e-06,
+      "loss": 0.5775,
+      "step": 32934
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.059244394302368,
+      "learning_rate": 4.656592124665312e-06,
+      "loss": 0.379,
+      "step": 32935
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 4.157079219818115,
+      "learning_rate": 4.656329089558752e-06,
+      "loss": 0.5544,
+      "step": 32936
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.0952394008636475,
+      "learning_rate": 4.656066055407812e-06,
+      "loss": 0.7069,
+      "step": 32937
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.9311599731445312,
+      "learning_rate": 4.655803022213223e-06,
+      "loss": 0.7395,
+      "step": 32938
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.398109197616577,
+      "learning_rate": 4.6555399899757166e-06,
+      "loss": 0.6317,
+      "step": 32939
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 7.1626434326171875,
+      "learning_rate": 4.655276958696025e-06,
+      "loss": 0.5941,
+      "step": 32940
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.1708381175994873,
+      "learning_rate": 4.65501392837488e-06,
+      "loss": 0.7889,
+      "step": 32941
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.8614120483398438,
+      "learning_rate": 4.654750899013011e-06,
+      "loss": 0.7413,
+      "step": 32942
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 5.362030029296875,
+      "learning_rate": 4.654487870611151e-06,
+      "loss": 0.6317,
+      "step": 32943
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.8225481510162354,
+      "learning_rate": 4.654224843170029e-06,
+      "loss": 0.678,
+      "step": 32944
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.2112505435943604,
+      "learning_rate": 4.65396181669038e-06,
+      "loss": 0.5553,
+      "step": 32945
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.027711868286133,
+      "learning_rate": 4.653698791172934e-06,
+      "loss": 0.4155,
+      "step": 32946
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.8007326126098633,
+      "learning_rate": 4.65343576661842e-06,
+      "loss": 0.8234,
+      "step": 32947
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.601344585418701,
+      "learning_rate": 4.653172743027573e-06,
+      "loss": 0.8144,
+      "step": 32948
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.1383144855499268,
+      "learning_rate": 4.652909720401122e-06,
+      "loss": 0.6121,
+      "step": 32949
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.5994179248809814,
+      "learning_rate": 4.6526466987398e-06,
+      "loss": 0.5446,
+      "step": 32950
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.9368677139282227,
+      "learning_rate": 4.652383678044335e-06,
+      "loss": 0.4864,
+      "step": 32951
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.7118436098098755,
+      "learning_rate": 4.652120658315462e-06,
+      "loss": 0.2998,
+      "step": 32952
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.614112377166748,
+      "learning_rate": 4.651857639553913e-06,
+      "loss": 0.6906,
+      "step": 32953
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.2371749877929688,
+      "learning_rate": 4.651594621760416e-06,
+      "loss": 0.6116,
+      "step": 32954
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.441258192062378,
+      "learning_rate": 4.6513316049357034e-06,
+      "loss": 0.5233,
+      "step": 32955
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.9743876457214355,
+      "learning_rate": 4.651068589080508e-06,
+      "loss": 0.5939,
+      "step": 32956
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 3.4749889373779297,
+      "learning_rate": 4.650805574195559e-06,
+      "loss": 0.6611,
+      "step": 32957
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.8073205947875977,
+      "learning_rate": 4.65054256028159e-06,
+      "loss": 0.5596,
+      "step": 32958
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.8084428310394287,
+      "learning_rate": 4.6502795473393305e-06,
+      "loss": 0.6374,
+      "step": 32959
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.3114521503448486,
+      "learning_rate": 4.650016535369513e-06,
+      "loss": 0.4539,
+      "step": 32960
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.006988525390625,
+      "learning_rate": 4.6497535243728685e-06,
+      "loss": 0.7883,
+      "step": 32961
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.9857137203216553,
+      "learning_rate": 4.649490514350128e-06,
+      "loss": 0.5298,
+      "step": 32962
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.3548388481140137,
+      "learning_rate": 4.649227505302023e-06,
+      "loss": 0.618,
+      "step": 32963
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.3062593936920166,
+      "learning_rate": 4.648964497229283e-06,
+      "loss": 0.6451,
+      "step": 32964
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.8572192192077637,
+      "learning_rate": 4.648701490132643e-06,
+      "loss": 0.427,
+      "step": 32965
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.041215419769287,
+      "learning_rate": 4.648438484012833e-06,
+      "loss": 0.5489,
+      "step": 32966
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 5.033681392669678,
+      "learning_rate": 4.648175478870584e-06,
+      "loss": 0.6976,
+      "step": 32967
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.43491530418396,
+      "learning_rate": 4.647912474706626e-06,
+      "loss": 0.5433,
+      "step": 32968
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.222318172454834,
+      "learning_rate": 4.6476494715216924e-06,
+      "loss": 0.4962,
+      "step": 32969
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.7282252311706543,
+      "learning_rate": 4.647386469316514e-06,
+      "loss": 0.6765,
+      "step": 32970
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.8873863220214844,
+      "learning_rate": 4.647123468091821e-06,
+      "loss": 0.7106,
+      "step": 32971
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.489210605621338,
+      "learning_rate": 4.646860467848346e-06,
+      "loss": 0.6669,
+      "step": 32972
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.4308303594589233,
+      "learning_rate": 4.64659746858682e-06,
+      "loss": 0.509,
+      "step": 32973
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.4607021808624268,
+      "learning_rate": 4.6463344703079725e-06,
+      "loss": 0.514,
+      "step": 32974
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.431173324584961,
+      "learning_rate": 4.646071473012537e-06,
+      "loss": 0.5236,
+      "step": 32975
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.8897007703781128,
+      "learning_rate": 4.645808476701245e-06,
+      "loss": 0.5612,
+      "step": 32976
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.9197022914886475,
+      "learning_rate": 4.645545481374827e-06,
+      "loss": 0.6303,
+      "step": 32977
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.0978269577026367,
+      "learning_rate": 4.645282487034013e-06,
+      "loss": 0.8364,
+      "step": 32978
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.3682799339294434,
+      "learning_rate": 4.645019493679537e-06,
+      "loss": 0.4804,
+      "step": 32979
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 3.044497013092041,
+      "learning_rate": 4.644756501312128e-06,
+      "loss": 0.5756,
+      "step": 32980
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.120317220687866,
+      "learning_rate": 4.644493509932518e-06,
+      "loss": 0.7588,
+      "step": 32981
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 3.546050548553467,
+      "learning_rate": 4.6442305195414385e-06,
+      "loss": 0.5054,
+      "step": 32982
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.7728281021118164,
+      "learning_rate": 4.643967530139621e-06,
+      "loss": 0.4242,
+      "step": 32983
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.4036672115325928,
+      "learning_rate": 4.643704541727794e-06,
+      "loss": 0.7182,
+      "step": 32984
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.568105459213257,
+      "learning_rate": 4.643441554306693e-06,
+      "loss": 0.6876,
+      "step": 32985
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.0502045154571533,
+      "learning_rate": 4.643178567877048e-06,
+      "loss": 0.9027,
+      "step": 32986
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 3.172696352005005,
+      "learning_rate": 4.64291558243959e-06,
+      "loss": 0.754,
+      "step": 32987
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.7215449810028076,
+      "learning_rate": 4.642652597995049e-06,
+      "loss": 0.768,
+      "step": 32988
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 3.332669258117676,
+      "learning_rate": 4.642389614544157e-06,
+      "loss": 0.8037,
+      "step": 32989
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.2513139247894287,
+      "learning_rate": 4.642126632087647e-06,
+      "loss": 0.6771,
+      "step": 32990
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 3.897587537765503,
+      "learning_rate": 4.641863650626248e-06,
+      "loss": 0.33,
+      "step": 32991
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 3.9613873958587646,
+      "learning_rate": 4.641600670160692e-06,
+      "loss": 0.4383,
+      "step": 32992
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.012721061706543,
+      "learning_rate": 4.641337690691711e-06,
+      "loss": 0.7238,
+      "step": 32993
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.7300801277160645,
+      "learning_rate": 4.641074712220035e-06,
+      "loss": 0.7484,
+      "step": 32994
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.14388370513916,
+      "learning_rate": 4.640811734746395e-06,
+      "loss": 0.5584,
+      "step": 32995
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.1325554847717285,
+      "learning_rate": 4.640548758271524e-06,
+      "loss": 0.7769,
+      "step": 32996
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.8246175050735474,
+      "learning_rate": 4.640285782796151e-06,
+      "loss": 0.6537,
+      "step": 32997
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.6642744541168213,
+      "learning_rate": 4.6400228083210104e-06,
+      "loss": 0.5904,
+      "step": 32998
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 3.2455244064331055,
+      "learning_rate": 4.63975983484683e-06,
+      "loss": 0.5264,
+      "step": 32999
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.7795336246490479,
+      "learning_rate": 4.639496862374344e-06,
+      "loss": 0.5535,
+      "step": 33000
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.856494665145874,
+      "learning_rate": 4.639233890904279e-06,
+      "loss": 0.7014,
+      "step": 33001
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.9376126527786255,
+      "learning_rate": 4.638970920437373e-06,
+      "loss": 0.8106,
+      "step": 33002
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 3.0547289848327637,
+      "learning_rate": 4.638707950974352e-06,
+      "loss": 0.764,
+      "step": 33003
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.0753777027130127,
+      "learning_rate": 4.638444982515949e-06,
+      "loss": 0.6922,
+      "step": 33004
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.957365870475769,
+      "learning_rate": 4.6381820150628935e-06,
+      "loss": 0.8763,
+      "step": 33005
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 3.6606078147888184,
+      "learning_rate": 4.637919048615921e-06,
+      "loss": 0.4752,
+      "step": 33006
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.3969614505767822,
+      "learning_rate": 4.637656083175758e-06,
+      "loss": 0.6473,
+      "step": 33007
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.8899333477020264,
+      "learning_rate": 4.637393118743139e-06,
+      "loss": 0.4874,
+      "step": 33008
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.857802152633667,
+      "learning_rate": 4.637130155318794e-06,
+      "loss": 0.4467,
+      "step": 33009
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.655305862426758,
+      "learning_rate": 4.636867192903453e-06,
+      "loss": 0.7683,
+      "step": 33010
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 3.1903421878814697,
+      "learning_rate": 4.636604231497849e-06,
+      "loss": 0.8861,
+      "step": 33011
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 4.111989974975586,
+      "learning_rate": 4.636341271102712e-06,
+      "loss": 0.5463,
+      "step": 33012
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.1835217475891113,
+      "learning_rate": 4.636078311718774e-06,
+      "loss": 0.5502,
+      "step": 33013
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 3.763158082962036,
+      "learning_rate": 4.635815353346765e-06,
+      "loss": 0.6699,
+      "step": 33014
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.9483072757720947,
+      "learning_rate": 4.635552395987419e-06,
+      "loss": 0.6626,
+      "step": 33015
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.7836064100265503,
+      "learning_rate": 4.635289439641465e-06,
+      "loss": 0.5443,
+      "step": 33016
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.8496299982070923,
+      "learning_rate": 4.635026484309633e-06,
+      "loss": 0.5638,
+      "step": 33017
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 3.016240119934082,
+      "learning_rate": 4.634763529992656e-06,
+      "loss": 0.6745,
+      "step": 33018
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 3.0189242362976074,
+      "learning_rate": 4.634500576691266e-06,
+      "loss": 0.6473,
+      "step": 33019
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.9618464708328247,
+      "learning_rate": 4.634237624406192e-06,
+      "loss": 0.6901,
+      "step": 33020
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.908265233039856,
+      "learning_rate": 4.633974673138166e-06,
+      "loss": 0.5317,
+      "step": 33021
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.7389912605285645,
+      "learning_rate": 4.63371172288792e-06,
+      "loss": 0.9367,
+      "step": 33022
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.7601040601730347,
+      "learning_rate": 4.633448773656184e-06,
+      "loss": 0.5218,
+      "step": 33023
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.781400680541992,
+      "learning_rate": 4.63318582544369e-06,
+      "loss": 0.6777,
+      "step": 33024
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 7.721311092376709,
+      "learning_rate": 4.632922878251167e-06,
+      "loss": 0.6451,
+      "step": 33025
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.9556905031204224,
+      "learning_rate": 4.632659932079349e-06,
+      "loss": 0.4394,
+      "step": 33026
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.5322265625,
+      "learning_rate": 4.632396986928967e-06,
+      "loss": 0.7449,
+      "step": 33027
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.650017738342285,
+      "learning_rate": 4.632134042800751e-06,
+      "loss": 0.5531,
+      "step": 33028
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.7777042388916016,
+      "learning_rate": 4.631871099695433e-06,
+      "loss": 0.5491,
+      "step": 33029
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.1548714637756348,
+      "learning_rate": 4.631608157613742e-06,
+      "loss": 0.6416,
+      "step": 33030
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 3.7485413551330566,
+      "learning_rate": 4.631345216556412e-06,
+      "loss": 0.8474,
+      "step": 33031
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.487205743789673,
+      "learning_rate": 4.631082276524173e-06,
+      "loss": 1.0704,
+      "step": 33032
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.8426108360290527,
+      "learning_rate": 4.630819337517755e-06,
+      "loss": 0.6956,
+      "step": 33033
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 3.0233330726623535,
+      "learning_rate": 4.6305563995378914e-06,
+      "loss": 0.5062,
+      "step": 33034
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.3437018394470215,
+      "learning_rate": 4.6302934625853105e-06,
+      "loss": 0.7376,
+      "step": 33035
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.6272155046463013,
+      "learning_rate": 4.6300305266607455e-06,
+      "loss": 0.6029,
+      "step": 33036
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.880873203277588,
+      "learning_rate": 4.6297675917649275e-06,
+      "loss": 0.5654,
+      "step": 33037
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.941394567489624,
+      "learning_rate": 4.629504657898587e-06,
+      "loss": 0.8453,
+      "step": 33038
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.23604154586792,
+      "learning_rate": 4.629241725062455e-06,
+      "loss": 0.7268,
+      "step": 33039
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.707307815551758,
+      "learning_rate": 4.628978793257264e-06,
+      "loss": 0.599,
+      "step": 33040
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 4.194460868835449,
+      "learning_rate": 4.628715862483743e-06,
+      "loss": 0.6932,
+      "step": 33041
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 6.877475261688232,
+      "learning_rate": 4.628452932742624e-06,
+      "loss": 0.6526,
+      "step": 33042
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.8979219198226929,
+      "learning_rate": 4.628190004034638e-06,
+      "loss": 0.7545,
+      "step": 33043
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.8662643432617188,
+      "learning_rate": 4.6279270763605164e-06,
+      "loss": 0.6115,
+      "step": 33044
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.902946949005127,
+      "learning_rate": 4.627664149720989e-06,
+      "loss": 0.4651,
+      "step": 33045
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 3.102768898010254,
+      "learning_rate": 4.627401224116789e-06,
+      "loss": 0.6591,
+      "step": 33046
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.3887584209442139,
+      "learning_rate": 4.627138299548647e-06,
+      "loss": 0.4751,
+      "step": 33047
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.8851510286331177,
+      "learning_rate": 4.626875376017295e-06,
+      "loss": 0.6924,
+      "step": 33048
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 3.035968780517578,
+      "learning_rate": 4.626612453523461e-06,
+      "loss": 0.6179,
+      "step": 33049
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 5.185608863830566,
+      "learning_rate": 4.626349532067879e-06,
+      "loss": 0.8408,
+      "step": 33050
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 3.1014785766601562,
+      "learning_rate": 4.626086611651279e-06,
+      "loss": 0.5608,
+      "step": 33051
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.3629038333892822,
+      "learning_rate": 4.625823692274392e-06,
+      "loss": 0.6908,
+      "step": 33052
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.011460542678833,
+      "learning_rate": 4.6255607739379485e-06,
+      "loss": 0.4725,
+      "step": 33053
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.6520459651947021,
+      "learning_rate": 4.625297856642681e-06,
+      "loss": 0.6791,
+      "step": 33054
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.8292988538742065,
+      "learning_rate": 4.625034940389318e-06,
+      "loss": 0.7398,
+      "step": 33055
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.9733388423919678,
+      "learning_rate": 4.624772025178594e-06,
+      "loss": 0.4873,
+      "step": 33056
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 3.7144203186035156,
+      "learning_rate": 4.6245091110112385e-06,
+      "loss": 0.721,
+      "step": 33057
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.2519712448120117,
+      "learning_rate": 4.624246197887982e-06,
+      "loss": 0.7759,
+      "step": 33058
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 3.845508337020874,
+      "learning_rate": 4.623983285809556e-06,
+      "loss": 0.8132,
+      "step": 33059
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.336334466934204,
+      "learning_rate": 4.623720374776693e-06,
+      "loss": 0.4673,
+      "step": 33060
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.636584997177124,
+      "learning_rate": 4.6234574647901205e-06,
+      "loss": 0.7037,
+      "step": 33061
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 4.125634670257568,
+      "learning_rate": 4.6231945558505734e-06,
+      "loss": 0.3806,
+      "step": 33062
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.2789957523345947,
+      "learning_rate": 4.622931647958779e-06,
+      "loss": 0.5848,
+      "step": 33063
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.9889253377914429,
+      "learning_rate": 4.6226687411154716e-06,
+      "loss": 0.5372,
+      "step": 33064
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.304852247238159,
+      "learning_rate": 4.622405835321382e-06,
+      "loss": 0.4661,
+      "step": 33065
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 3.2326388359069824,
+      "learning_rate": 4.622142930577238e-06,
+      "loss": 0.8052,
+      "step": 33066
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.522069215774536,
+      "learning_rate": 4.621880026883775e-06,
+      "loss": 0.6899,
+      "step": 33067
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 3.3001039028167725,
+      "learning_rate": 4.621617124241721e-06,
+      "loss": 0.442,
+      "step": 33068
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.585188388824463,
+      "learning_rate": 4.621354222651809e-06,
+      "loss": 0.6244,
+      "step": 33069
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.5620715618133545,
+      "learning_rate": 4.621091322114769e-06,
+      "loss": 0.5085,
+      "step": 33070
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.985927104949951,
+      "learning_rate": 4.620828422631332e-06,
+      "loss": 0.5916,
+      "step": 33071
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.6857786178588867,
+      "learning_rate": 4.6205655242022284e-06,
+      "loss": 0.6074,
+      "step": 33072
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 3.2235159873962402,
+      "learning_rate": 4.620302626828191e-06,
+      "loss": 0.8989,
+      "step": 33073
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 3.2863645553588867,
+      "learning_rate": 4.6200397305099474e-06,
+      "loss": 0.686,
+      "step": 33074
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 4.523669242858887,
+      "learning_rate": 4.619776835248233e-06,
+      "loss": 0.8789,
+      "step": 33075
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.8739204406738281,
+      "learning_rate": 4.619513941043776e-06,
+      "loss": 0.5987,
+      "step": 33076
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.5721735954284668,
+      "learning_rate": 4.619251047897309e-06,
+      "loss": 0.7729,
+      "step": 33077
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.5601906776428223,
+      "learning_rate": 4.618988155809563e-06,
+      "loss": 0.6357,
+      "step": 33078
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 3.7494537830352783,
+      "learning_rate": 4.618725264781267e-06,
+      "loss": 0.735,
+      "step": 33079
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.61913800239563,
+      "learning_rate": 4.618462374813153e-06,
+      "loss": 0.7453,
+      "step": 33080
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.7517868280410767,
+      "learning_rate": 4.618199485905953e-06,
+      "loss": 0.4553,
+      "step": 33081
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.6750223636627197,
+      "learning_rate": 4.617936598060397e-06,
+      "loss": 0.5857,
+      "step": 33082
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.2398693561553955,
+      "learning_rate": 4.617673711277216e-06,
+      "loss": 0.5299,
+      "step": 33083
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.242486000061035,
+      "learning_rate": 4.617410825557141e-06,
+      "loss": 0.6477,
+      "step": 33084
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 3.0014188289642334,
+      "learning_rate": 4.6171479409009035e-06,
+      "loss": 0.5974,
+      "step": 33085
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.7376186847686768,
+      "learning_rate": 4.6168850573092315e-06,
+      "loss": 0.5924,
+      "step": 33086
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.7888202667236328,
+      "learning_rate": 4.616622174782861e-06,
+      "loss": 0.5783,
+      "step": 33087
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 3.0452682971954346,
+      "learning_rate": 4.616359293322521e-06,
+      "loss": 0.6918,
+      "step": 33088
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 3.1188931465148926,
+      "learning_rate": 4.616096412928942e-06,
+      "loss": 0.9444,
+      "step": 33089
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 3.7842178344726562,
+      "learning_rate": 4.615833533602855e-06,
+      "loss": 0.7361,
+      "step": 33090
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.950273275375366,
+      "learning_rate": 4.615570655344991e-06,
+      "loss": 0.762,
+      "step": 33091
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.108095169067383,
+      "learning_rate": 4.615307778156081e-06,
+      "loss": 0.8339,
+      "step": 33092
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.228222370147705,
+      "learning_rate": 4.6150449020368556e-06,
+      "loss": 0.5464,
+      "step": 33093
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 3.276465654373169,
+      "learning_rate": 4.614782026988047e-06,
+      "loss": 0.4597,
+      "step": 33094
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.158352851867676,
+      "learning_rate": 4.6145191530103846e-06,
+      "loss": 0.9696,
+      "step": 33095
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 3.606475353240967,
+      "learning_rate": 4.6142562801046e-06,
+      "loss": 0.6585,
+      "step": 33096
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.3907415866851807,
+      "learning_rate": 4.613993408271424e-06,
+      "loss": 0.6794,
+      "step": 33097
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.9471561908721924,
+      "learning_rate": 4.613730537511588e-06,
+      "loss": 0.7122,
+      "step": 33098
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.9064514636993408,
+      "learning_rate": 4.613467667825823e-06,
+      "loss": 0.56,
+      "step": 33099
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 3.2587826251983643,
+      "learning_rate": 4.6132047992148595e-06,
+      "loss": 0.802,
+      "step": 33100
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.357490062713623,
+      "learning_rate": 4.612941931679428e-06,
+      "loss": 0.6873,
+      "step": 33101
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.599152088165283,
+      "learning_rate": 4.61267906522026e-06,
+      "loss": 0.7671,
+      "step": 33102
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.253319025039673,
+      "learning_rate": 4.6124161998380875e-06,
+      "loss": 0.6777,
+      "step": 33103
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.2291550636291504,
+      "learning_rate": 4.61215333553364e-06,
+      "loss": 0.472,
+      "step": 33104
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.977128505706787,
+      "learning_rate": 4.611890472307648e-06,
+      "loss": 0.6051,
+      "step": 33105
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.7031450271606445,
+      "learning_rate": 4.611627610160841e-06,
+      "loss": 0.696,
+      "step": 33106
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.2611711025238037,
+      "learning_rate": 4.611364749093956e-06,
+      "loss": 0.3493,
+      "step": 33107
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 3.168442487716675,
+      "learning_rate": 4.611101889107718e-06,
+      "loss": 0.8271,
+      "step": 33108
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.2384679317474365,
+      "learning_rate": 4.610839030202861e-06,
+      "loss": 0.7462,
+      "step": 33109
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.0058233737945557,
+      "learning_rate": 4.610576172380115e-06,
+      "loss": 0.6079,
+      "step": 33110
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.021358013153076,
+      "learning_rate": 4.61031331564021e-06,
+      "loss": 0.449,
+      "step": 33111
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.549049139022827,
+      "learning_rate": 4.610050459983879e-06,
+      "loss": 0.7629,
+      "step": 33112
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.8577460050582886,
+      "learning_rate": 4.6097876054118504e-06,
+      "loss": 0.4996,
+      "step": 33113
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 4.163241386413574,
+      "learning_rate": 4.609524751924857e-06,
+      "loss": 0.6718,
+      "step": 33114
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.756786584854126,
+      "learning_rate": 4.609261899523628e-06,
+      "loss": 0.7789,
+      "step": 33115
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.0430877208709717,
+      "learning_rate": 4.608999048208896e-06,
+      "loss": 0.6473,
+      "step": 33116
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.5613996982574463,
+      "learning_rate": 4.608736197981391e-06,
+      "loss": 0.5165,
+      "step": 33117
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.353724241256714,
+      "learning_rate": 4.608473348841844e-06,
+      "loss": 0.4537,
+      "step": 33118
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.7605259418487549,
+      "learning_rate": 4.608210500790987e-06,
+      "loss": 0.4844,
+      "step": 33119
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.6725523471832275,
+      "learning_rate": 4.607947653829549e-06,
+      "loss": 0.8651,
+      "step": 33120
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.887260913848877,
+      "learning_rate": 4.6076848079582615e-06,
+      "loss": 0.5013,
+      "step": 33121
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.6981905698776245,
+      "learning_rate": 4.607421963177855e-06,
+      "loss": 0.493,
+      "step": 33122
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.5010615587234497,
+      "learning_rate": 4.607159119489062e-06,
+      "loss": 0.5413,
+      "step": 33123
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.855555534362793,
+      "learning_rate": 4.606896276892612e-06,
+      "loss": 0.8036,
+      "step": 33124
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.987959623336792,
+      "learning_rate": 4.606633435389235e-06,
+      "loss": 0.5149,
+      "step": 33125
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.9920766353607178,
+      "learning_rate": 4.606370594979664e-06,
+      "loss": 0.4227,
+      "step": 33126
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.7009174823760986,
+      "learning_rate": 4.606107755664628e-06,
+      "loss": 0.511,
+      "step": 33127
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.457334518432617,
+      "learning_rate": 4.60584491744486e-06,
+      "loss": 0.6066,
+      "step": 33128
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.0080549716949463,
+      "learning_rate": 4.60558208032109e-06,
+      "loss": 0.669,
+      "step": 33129
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.8078105449676514,
+      "learning_rate": 4.605319244294048e-06,
+      "loss": 0.5403,
+      "step": 33130
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 3.920517921447754,
+      "learning_rate": 4.6050564093644655e-06,
+      "loss": 0.6907,
+      "step": 33131
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 4.46325159072876,
+      "learning_rate": 4.604793575533073e-06,
+      "loss": 0.6604,
+      "step": 33132
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.7613756656646729,
+      "learning_rate": 4.604530742800602e-06,
+      "loss": 0.4827,
+      "step": 33133
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.462486982345581,
+      "learning_rate": 4.604267911167782e-06,
+      "loss": 0.7458,
+      "step": 33134
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.7700039148330688,
+      "learning_rate": 4.604005080635345e-06,
+      "loss": 0.6889,
+      "step": 33135
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.939401388168335,
+      "learning_rate": 4.603742251204021e-06,
+      "loss": 0.4678,
+      "step": 33136
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.475703239440918,
+      "learning_rate": 4.603479422874542e-06,
+      "loss": 0.4108,
+      "step": 33137
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.1776955127716064,
+      "learning_rate": 4.603216595647638e-06,
+      "loss": 0.5196,
+      "step": 33138
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.9751986265182495,
+      "learning_rate": 4.602953769524041e-06,
+      "loss": 0.8053,
+      "step": 33139
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 3.4321107864379883,
+      "learning_rate": 4.6026909445044794e-06,
+      "loss": 0.6678,
+      "step": 33140
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 3.157881498336792,
+      "learning_rate": 4.602428120589686e-06,
+      "loss": 0.619,
+      "step": 33141
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.2477877140045166,
+      "learning_rate": 4.602165297780391e-06,
+      "loss": 0.657,
+      "step": 33142
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.34540057182312,
+      "learning_rate": 4.601902476077326e-06,
+      "loss": 0.6186,
+      "step": 33143
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.5535848140716553,
+      "learning_rate": 4.60163965548122e-06,
+      "loss": 0.4858,
+      "step": 33144
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.6093640327453613,
+      "learning_rate": 4.601376835992806e-06,
+      "loss": 0.652,
+      "step": 33145
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.431518793106079,
+      "learning_rate": 4.6011140176128125e-06,
+      "loss": 0.5928,
+      "step": 33146
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 3.285351037979126,
+      "learning_rate": 4.600851200341971e-06,
+      "loss": 0.5546,
+      "step": 33147
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.3676822185516357,
+      "learning_rate": 4.600588384181013e-06,
+      "loss": 0.5084,
+      "step": 33148
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.851853847503662,
+      "learning_rate": 4.600325569130671e-06,
+      "loss": 0.7673,
+      "step": 33149
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 4.567963600158691,
+      "learning_rate": 4.6000627551916724e-06,
+      "loss": 0.5552,
+      "step": 33150
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.614267110824585,
+      "learning_rate": 4.59979994236475e-06,
+      "loss": 0.824,
+      "step": 33151
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.4814139604568481,
+      "learning_rate": 4.5995371306506335e-06,
+      "loss": 0.467,
+      "step": 33152
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.2379255294799805,
+      "learning_rate": 4.599274320050055e-06,
+      "loss": 0.7425,
+      "step": 33153
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.006981611251831,
+      "learning_rate": 4.599011510563744e-06,
+      "loss": 0.4528,
+      "step": 33154
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.6019246578216553,
+      "learning_rate": 4.598748702192432e-06,
+      "loss": 0.8085,
+      "step": 33155
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 3.468235492706299,
+      "learning_rate": 4.59848589493685e-06,
+      "loss": 0.5611,
+      "step": 33156
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.0163185596466064,
+      "learning_rate": 4.598223088797727e-06,
+      "loss": 0.5726,
+      "step": 33157
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.8751471042633057,
+      "learning_rate": 4.597960283775796e-06,
+      "loss": 0.6327,
+      "step": 33158
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.4778571128845215,
+      "learning_rate": 4.5976974798717865e-06,
+      "loss": 0.7459,
+      "step": 33159
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.430100202560425,
+      "learning_rate": 4.59743467708643e-06,
+      "loss": 0.754,
+      "step": 33160
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.4988821744918823,
+      "learning_rate": 4.597171875420457e-06,
+      "loss": 0.599,
+      "step": 33161
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.106938123703003,
+      "learning_rate": 4.596909074874598e-06,
+      "loss": 0.618,
+      "step": 33162
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 3.265875816345215,
+      "learning_rate": 4.5966462754495835e-06,
+      "loss": 0.8613,
+      "step": 33163
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 3.405336380004883,
+      "learning_rate": 4.596383477146144e-06,
+      "loss": 0.7857,
+      "step": 33164
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.6778067350387573,
+      "learning_rate": 4.596120679965012e-06,
+      "loss": 0.6959,
+      "step": 33165
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.9504613876342773,
+      "learning_rate": 4.595857883906917e-06,
+      "loss": 0.6093,
+      "step": 33166
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.4970331192016602,
+      "learning_rate": 4.595595088972587e-06,
+      "loss": 0.5525,
+      "step": 33167
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.9083703756332397,
+      "learning_rate": 4.595332295162758e-06,
+      "loss": 0.7578,
+      "step": 33168
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 3.014519453048706,
+      "learning_rate": 4.5950695024781585e-06,
+      "loss": 0.8968,
+      "step": 33169
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 4.289487838745117,
+      "learning_rate": 4.594806710919519e-06,
+      "loss": 0.8014,
+      "step": 33170
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.7868916988372803,
+      "learning_rate": 4.59454392048757e-06,
+      "loss": 0.6698,
+      "step": 33171
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.1660306453704834,
+      "learning_rate": 4.594281131183043e-06,
+      "loss": 0.5169,
+      "step": 33172
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.184842109680176,
+      "learning_rate": 4.5940183430066675e-06,
+      "loss": 0.5169,
+      "step": 33173
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 3.994549512863159,
+      "learning_rate": 4.593755555959176e-06,
+      "loss": 0.6256,
+      "step": 33174
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 3.2894837856292725,
+      "learning_rate": 4.593492770041297e-06,
+      "loss": 0.8182,
+      "step": 33175
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 3.6162497997283936,
+      "learning_rate": 4.593229985253763e-06,
+      "loss": 0.6119,
+      "step": 33176
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.969421148300171,
+      "learning_rate": 4.592967201597303e-06,
+      "loss": 0.638,
+      "step": 33177
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.591237783432007,
+      "learning_rate": 4.59270441907265e-06,
+      "loss": 0.7078,
+      "step": 33178
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.19924259185791,
+      "learning_rate": 4.592441637680532e-06,
+      "loss": 0.5113,
+      "step": 33179
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.3061869144439697,
+      "learning_rate": 4.592178857421682e-06,
+      "loss": 0.6272,
+      "step": 33180
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.892770528793335,
+      "learning_rate": 4.59191607829683e-06,
+      "loss": 0.6931,
+      "step": 33181
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.7704808712005615,
+      "learning_rate": 4.591653300306706e-06,
+      "loss": 0.7346,
+      "step": 33182
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.281468629837036,
+      "learning_rate": 4.591390523452042e-06,
+      "loss": 0.747,
+      "step": 33183
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.382023334503174,
+      "learning_rate": 4.591127747733568e-06,
+      "loss": 0.4884,
+      "step": 33184
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.260798931121826,
+      "learning_rate": 4.590864973152014e-06,
+      "loss": 0.67,
+      "step": 33185
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 3.427579164505005,
+      "learning_rate": 4.59060219970811e-06,
+      "loss": 0.588,
+      "step": 33186
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.55181622505188,
+      "learning_rate": 4.5903394274025896e-06,
+      "loss": 0.836,
+      "step": 33187
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.0781352519989014,
+      "learning_rate": 4.59007665623618e-06,
+      "loss": 0.695,
+      "step": 33188
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.186840057373047,
+      "learning_rate": 4.589813886209615e-06,
+      "loss": 0.4037,
+      "step": 33189
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.82522714138031,
+      "learning_rate": 4.589551117323624e-06,
+      "loss": 0.4914,
+      "step": 33190
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 3.8447346687316895,
+      "learning_rate": 4.589288349578937e-06,
+      "loss": 0.4766,
+      "step": 33191
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.6706888675689697,
+      "learning_rate": 4.589025582976286e-06,
+      "loss": 0.8418,
+      "step": 33192
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.6353275775909424,
+      "learning_rate": 4.5887628175164005e-06,
+      "loss": 0.5856,
+      "step": 33193
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.5056214332580566,
+      "learning_rate": 4.588500053200012e-06,
+      "loss": 0.6714,
+      "step": 33194
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.5802429914474487,
+      "learning_rate": 4.58823729002785e-06,
+      "loss": 0.7432,
+      "step": 33195
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.5956497192382812,
+      "learning_rate": 4.587974528000646e-06,
+      "loss": 0.4023,
+      "step": 33196
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 3.32700514793396,
+      "learning_rate": 4.587711767119131e-06,
+      "loss": 0.5055,
+      "step": 33197
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 4.224265098571777,
+      "learning_rate": 4.587449007384034e-06,
+      "loss": 0.4463,
+      "step": 33198
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.6231002807617188,
+      "learning_rate": 4.587186248796088e-06,
+      "loss": 0.746,
+      "step": 33199
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.5983026027679443,
+      "learning_rate": 4.586923491356022e-06,
+      "loss": 0.6331,
+      "step": 33200
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.427476644515991,
+      "learning_rate": 4.586660735064567e-06,
+      "loss": 0.6739,
+      "step": 33201
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 3.9550020694732666,
+      "learning_rate": 4.586397979922453e-06,
+      "loss": 0.5084,
+      "step": 33202
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.0515995025634766,
+      "learning_rate": 4.5861352259304125e-06,
+      "loss": 0.3698,
+      "step": 33203
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.2396750450134277,
+      "learning_rate": 4.585872473089175e-06,
+      "loss": 0.6404,
+      "step": 33204
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 3.320596933364868,
+      "learning_rate": 4.58560972139947e-06,
+      "loss": 0.6718,
+      "step": 33205
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.1715121269226074,
+      "learning_rate": 4.5853469708620295e-06,
+      "loss": 0.6102,
+      "step": 33206
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.2268152236938477,
+      "learning_rate": 4.585084221477584e-06,
+      "loss": 0.4549,
+      "step": 33207
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.7535467147827148,
+      "learning_rate": 4.584821473246862e-06,
+      "loss": 0.6415,
+      "step": 33208
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.3949217796325684,
+      "learning_rate": 4.584558726170597e-06,
+      "loss": 0.6488,
+      "step": 33209
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 4.36454963684082,
+      "learning_rate": 4.584295980249519e-06,
+      "loss": 0.6663,
+      "step": 33210
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.5116147994995117,
+      "learning_rate": 4.584033235484358e-06,
+      "loss": 0.4763,
+      "step": 33211
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.417118787765503,
+      "learning_rate": 4.5837704918758455e-06,
+      "loss": 0.7171,
+      "step": 33212
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 4.717606544494629,
+      "learning_rate": 4.58350774942471e-06,
+      "loss": 0.5245,
+      "step": 33213
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 3.249683380126953,
+      "learning_rate": 4.583245008131685e-06,
+      "loss": 0.6442,
+      "step": 33214
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.7731572389602661,
+      "learning_rate": 4.582982267997499e-06,
+      "loss": 0.8637,
+      "step": 33215
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 3.3832483291625977,
+      "learning_rate": 4.582719529022883e-06,
+      "loss": 0.7019,
+      "step": 33216
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 3.596578598022461,
+      "learning_rate": 4.582456791208567e-06,
+      "loss": 0.7541,
+      "step": 33217
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 7.066343784332275,
+      "learning_rate": 4.582194054555283e-06,
+      "loss": 0.8068,
+      "step": 33218
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.118086338043213,
+      "learning_rate": 4.581931319063761e-06,
+      "loss": 0.5169,
+      "step": 33219
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.0722532272338867,
+      "learning_rate": 4.581668584734731e-06,
+      "loss": 0.4048,
+      "step": 33220
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.8977549076080322,
+      "learning_rate": 4.581405851568924e-06,
+      "loss": 0.7176,
+      "step": 33221
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.5744338035583496,
+      "learning_rate": 4.58114311956707e-06,
+      "loss": 0.675,
+      "step": 33222
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.425128698348999,
+      "learning_rate": 4.580880388729901e-06,
+      "loss": 0.6541,
+      "step": 33223
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.5859735012054443,
+      "learning_rate": 4.580617659058147e-06,
+      "loss": 0.6808,
+      "step": 33224
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.9700243473052979,
+      "learning_rate": 4.5803549305525365e-06,
+      "loss": 0.5368,
+      "step": 33225
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 3.3826987743377686,
+      "learning_rate": 4.580092203213803e-06,
+      "loss": 0.6248,
+      "step": 33226
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.784062385559082,
+      "learning_rate": 4.579829477042675e-06,
+      "loss": 0.7513,
+      "step": 33227
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 4.29623556137085,
+      "learning_rate": 4.579566752039882e-06,
+      "loss": 0.774,
+      "step": 33228
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.297982931137085,
+      "learning_rate": 4.579304028206159e-06,
+      "loss": 0.4614,
+      "step": 33229
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.114871025085449,
+      "learning_rate": 4.579041305542233e-06,
+      "loss": 0.4932,
+      "step": 33230
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.5125751495361328,
+      "learning_rate": 4.578778584048836e-06,
+      "loss": 0.7658,
+      "step": 33231
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.9367809295654297,
+      "learning_rate": 4.578515863726698e-06,
+      "loss": 0.6555,
+      "step": 33232
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.683479905128479,
+      "learning_rate": 4.57825314457655e-06,
+      "loss": 0.4235,
+      "step": 33233
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.024475574493408,
+      "learning_rate": 4.5779904265991215e-06,
+      "loss": 0.673,
+      "step": 33234
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.933228611946106,
+      "learning_rate": 4.5777277097951436e-06,
+      "loss": 0.4505,
+      "step": 33235
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.582758665084839,
+      "learning_rate": 4.577464994165346e-06,
+      "loss": 0.5872,
+      "step": 33236
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 3.058320999145508,
+      "learning_rate": 4.577202279710461e-06,
+      "loss": 0.4969,
+      "step": 33237
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 4.197457313537598,
+      "learning_rate": 4.576939566431218e-06,
+      "loss": 0.9325,
+      "step": 33238
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.886122703552246,
+      "learning_rate": 4.576676854328347e-06,
+      "loss": 0.6629,
+      "step": 33239
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.506484031677246,
+      "learning_rate": 4.57641414340258e-06,
+      "loss": 0.7602,
+      "step": 33240
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 3.02756404876709,
+      "learning_rate": 4.576151433654645e-06,
+      "loss": 0.5845,
+      "step": 33241
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 3.19563889503479,
+      "learning_rate": 4.575888725085275e-06,
+      "loss": 0.5191,
+      "step": 33242
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 3.0958292484283447,
+      "learning_rate": 4.5756260176952e-06,
+      "loss": 0.638,
+      "step": 33243
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 3.1365792751312256,
+      "learning_rate": 4.575363311485149e-06,
+      "loss": 0.73,
+      "step": 33244
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 5.250082015991211,
+      "learning_rate": 4.575100606455855e-06,
+      "loss": 0.6497,
+      "step": 33245
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.950864791870117,
+      "learning_rate": 4.5748379026080454e-06,
+      "loss": 0.8693,
+      "step": 33246
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.2150251865386963,
+      "learning_rate": 4.5745751999424535e-06,
+      "loss": 0.7291,
+      "step": 33247
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.771749973297119,
+      "learning_rate": 4.574312498459807e-06,
+      "loss": 0.489,
+      "step": 33248
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.854229688644409,
+      "learning_rate": 4.574049798160838e-06,
+      "loss": 0.5935,
+      "step": 33249
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 3.9003145694732666,
+      "learning_rate": 4.573787099046279e-06,
+      "loss": 0.9589,
+      "step": 33250
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.1082375049591064,
+      "learning_rate": 4.573524401116857e-06,
+      "loss": 0.6953,
+      "step": 33251
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.7155771255493164,
+      "learning_rate": 4.573261704373305e-06,
+      "loss": 0.8655,
+      "step": 33252
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.7086806297302246,
+      "learning_rate": 4.572999008816352e-06,
+      "loss": 0.6459,
+      "step": 33253
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 3.560939073562622,
+      "learning_rate": 4.572736314446728e-06,
+      "loss": 0.8015,
+      "step": 33254
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.9866857528686523,
+      "learning_rate": 4.572473621265165e-06,
+      "loss": 0.5728,
+      "step": 33255
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 3.943950653076172,
+      "learning_rate": 4.572210929272393e-06,
+      "loss": 0.458,
+      "step": 33256
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.795180559158325,
+      "learning_rate": 4.571948238469142e-06,
+      "loss": 0.7222,
+      "step": 33257
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.395460605621338,
+      "learning_rate": 4.571685548856141e-06,
+      "loss": 0.8526,
+      "step": 33258
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.4407260417938232,
+      "learning_rate": 4.571422860434124e-06,
+      "loss": 0.5092,
+      "step": 33259
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.9262219667434692,
+      "learning_rate": 4.571160173203818e-06,
+      "loss": 0.7925,
+      "step": 33260
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.4495991468429565,
+      "learning_rate": 4.570897487165956e-06,
+      "loss": 0.5461,
+      "step": 33261
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.518150806427002,
+      "learning_rate": 4.5706348023212675e-06,
+      "loss": 0.6551,
+      "step": 33262
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 3.9443576335906982,
+      "learning_rate": 4.570372118670482e-06,
+      "loss": 0.4023,
+      "step": 33263
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.1787314414978027,
+      "learning_rate": 4.570109436214331e-06,
+      "loss": 0.6566,
+      "step": 33264
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.103740692138672,
+      "learning_rate": 4.569846754953545e-06,
+      "loss": 0.4525,
+      "step": 33265
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.6478123664855957,
+      "learning_rate": 4.569584074888853e-06,
+      "loss": 0.9461,
+      "step": 33266
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.97855281829834,
+      "learning_rate": 4.5693213960209875e-06,
+      "loss": 0.8572,
+      "step": 33267
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.7892777919769287,
+      "learning_rate": 4.569058718350678e-06,
+      "loss": 0.6974,
+      "step": 33268
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.5766491889953613,
+      "learning_rate": 4.568796041878653e-06,
+      "loss": 0.4852,
+      "step": 33269
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 4.853504657745361,
+      "learning_rate": 4.568533366605646e-06,
+      "loss": 0.5803,
+      "step": 33270
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 3.5186500549316406,
+      "learning_rate": 4.568270692532385e-06,
+      "loss": 0.5104,
+      "step": 33271
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 4.151260852813721,
+      "learning_rate": 4.568008019659603e-06,
+      "loss": 0.629,
+      "step": 33272
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.150197982788086,
+      "learning_rate": 4.567745347988028e-06,
+      "loss": 0.5314,
+      "step": 33273
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.4340553283691406,
+      "learning_rate": 4.567482677518391e-06,
+      "loss": 0.5243,
+      "step": 33274
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.6699633598327637,
+      "learning_rate": 4.567220008251424e-06,
+      "loss": 0.6316,
+      "step": 33275
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.6259914636611938,
+      "learning_rate": 4.566957340187855e-06,
+      "loss": 0.5348,
+      "step": 33276
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.34841251373291,
+      "learning_rate": 4.566694673328416e-06,
+      "loss": 0.4245,
+      "step": 33277
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.723391532897949,
+      "learning_rate": 4.5664320076738365e-06,
+      "loss": 0.6321,
+      "step": 33278
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.0500261783599854,
+      "learning_rate": 4.566169343224848e-06,
+      "loss": 0.7435,
+      "step": 33279
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.9308738708496094,
+      "learning_rate": 4.565906679982179e-06,
+      "loss": 0.4948,
+      "step": 33280
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.9424028396606445,
+      "learning_rate": 4.565644017946561e-06,
+      "loss": 0.5371,
+      "step": 33281
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.62445068359375,
+      "learning_rate": 4.565381357118725e-06,
+      "loss": 0.5172,
+      "step": 33282
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 3.8066349029541016,
+      "learning_rate": 4.5651186974994e-06,
+      "loss": 0.7826,
+      "step": 33283
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.5806589126586914,
+      "learning_rate": 4.564856039089317e-06,
+      "loss": 0.8472,
+      "step": 33284
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.7756433486938477,
+      "learning_rate": 4.564593381889206e-06,
+      "loss": 0.6993,
+      "step": 33285
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.8659977912902832,
+      "learning_rate": 4.564330725899798e-06,
+      "loss": 0.7116,
+      "step": 33286
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.8264302015304565,
+      "learning_rate": 4.564068071121824e-06,
+      "loss": 0.6593,
+      "step": 33287
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.774625301361084,
+      "learning_rate": 4.563805417556012e-06,
+      "loss": 0.5956,
+      "step": 33288
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.328996419906616,
+      "learning_rate": 4.563542765203092e-06,
+      "loss": 0.5002,
+      "step": 33289
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 3.814473867416382,
+      "learning_rate": 4.563280114063798e-06,
+      "loss": 0.4991,
+      "step": 33290
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 3.0171754360198975,
+      "learning_rate": 4.563017464138859e-06,
+      "loss": 0.5864,
+      "step": 33291
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.7053701877593994,
+      "learning_rate": 4.562754815429004e-06,
+      "loss": 0.532,
+      "step": 33292
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 3.38012957572937,
+      "learning_rate": 4.562492167934965e-06,
+      "loss": 0.4794,
+      "step": 33293
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.560079574584961,
+      "learning_rate": 4.562229521657471e-06,
+      "loss": 0.5599,
+      "step": 33294
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.3420815467834473,
+      "learning_rate": 4.561966876597253e-06,
+      "loss": 0.5328,
+      "step": 33295
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.8409627676010132,
+      "learning_rate": 4.56170423275504e-06,
+      "loss": 0.603,
+      "step": 33296
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 3.9334771633148193,
+      "learning_rate": 4.561441590131565e-06,
+      "loss": 0.6167,
+      "step": 33297
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.590333342552185,
+      "learning_rate": 4.5611789487275555e-06,
+      "loss": 0.6381,
+      "step": 33298
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 5.975875377655029,
+      "learning_rate": 4.560916308543743e-06,
+      "loss": 0.7218,
+      "step": 33299
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.5463123321533203,
+      "learning_rate": 4.560653669580858e-06,
+      "loss": 0.8093,
+      "step": 33300
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.266005039215088,
+      "learning_rate": 4.56039103183963e-06,
+      "loss": 0.7352,
+      "step": 33301
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.6248533725738525,
+      "learning_rate": 4.56012839532079e-06,
+      "loss": 0.4379,
+      "step": 33302
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.7871510982513428,
+      "learning_rate": 4.559865760025069e-06,
+      "loss": 0.7041,
+      "step": 33303
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.057468891143799,
+      "learning_rate": 4.5596031259531955e-06,
+      "loss": 0.4766,
+      "step": 33304
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.7543754577636719,
+      "learning_rate": 4.559340493105902e-06,
+      "loss": 0.7897,
+      "step": 33305
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.7031803131103516,
+      "learning_rate": 4.559077861483916e-06,
+      "loss": 0.543,
+      "step": 33306
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.4052088260650635,
+      "learning_rate": 4.5588152310879705e-06,
+      "loss": 0.51,
+      "step": 33307
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.9472601413726807,
+      "learning_rate": 4.558552601918794e-06,
+      "loss": 0.8153,
+      "step": 33308
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.535104513168335,
+      "learning_rate": 4.558289973977117e-06,
+      "loss": 0.6836,
+      "step": 33309
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.005350112915039,
+      "learning_rate": 4.558027347263668e-06,
+      "loss": 0.5572,
+      "step": 33310
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.9343873262405396,
+      "learning_rate": 4.557764721779182e-06,
+      "loss": 0.4566,
+      "step": 33311
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.3046889305114746,
+      "learning_rate": 4.5575020975243865e-06,
+      "loss": 0.7177,
+      "step": 33312
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.923351526260376,
+      "learning_rate": 4.557239474500012e-06,
+      "loss": 0.6686,
+      "step": 33313
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.119438648223877,
+      "learning_rate": 4.556976852706789e-06,
+      "loss": 0.8329,
+      "step": 33314
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.033176898956299,
+      "learning_rate": 4.5567142321454465e-06,
+      "loss": 0.5839,
+      "step": 33315
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.4692578315734863,
+      "learning_rate": 4.5564516128167165e-06,
+      "loss": 0.7359,
+      "step": 33316
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.43154239654541,
+      "learning_rate": 4.5561889947213275e-06,
+      "loss": 0.452,
+      "step": 33317
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.6021504402160645,
+      "learning_rate": 4.555926377860011e-06,
+      "loss": 0.6831,
+      "step": 33318
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 3.307305097579956,
+      "learning_rate": 4.555663762233497e-06,
+      "loss": 0.3934,
+      "step": 33319
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 11.721996307373047,
+      "learning_rate": 4.555401147842516e-06,
+      "loss": 0.7606,
+      "step": 33320
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.543724298477173,
+      "learning_rate": 4.555138534687797e-06,
+      "loss": 0.9121,
+      "step": 33321
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.346208333969116,
+      "learning_rate": 4.5548759227700714e-06,
+      "loss": 0.7939,
+      "step": 33322
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.3909249305725098,
+      "learning_rate": 4.55461331209007e-06,
+      "loss": 0.7071,
+      "step": 33323
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 3.1020421981811523,
+      "learning_rate": 4.5543507026485215e-06,
+      "loss": 0.6734,
+      "step": 33324
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.1409268379211426,
+      "learning_rate": 4.554088094446156e-06,
+      "loss": 0.614,
+      "step": 33325
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.4383398294448853,
+      "learning_rate": 4.553825487483705e-06,
+      "loss": 0.6859,
+      "step": 33326
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.8467620611190796,
+      "learning_rate": 4.553562881761899e-06,
+      "loss": 0.7321,
+      "step": 33327
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.5482412576675415,
+      "learning_rate": 4.553300277281467e-06,
+      "loss": 0.6682,
+      "step": 33328
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.7656898498535156,
+      "learning_rate": 4.553037674043139e-06,
+      "loss": 0.6505,
+      "step": 33329
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 3.770883798599243,
+      "learning_rate": 4.552775072047645e-06,
+      "loss": 0.7834,
+      "step": 33330
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.5474510192871094,
+      "learning_rate": 4.552512471295717e-06,
+      "loss": 0.8223,
+      "step": 33331
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 3.692627429962158,
+      "learning_rate": 4.552249871788085e-06,
+      "loss": 0.6444,
+      "step": 33332
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.702146053314209,
+      "learning_rate": 4.551987273525476e-06,
+      "loss": 0.395,
+      "step": 33333
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 3.0082757472991943,
+      "learning_rate": 4.551724676508624e-06,
+      "loss": 0.6377,
+      "step": 33334
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.937523603439331,
+      "learning_rate": 4.551462080738259e-06,
+      "loss": 0.6572,
+      "step": 33335
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.189713478088379,
+      "learning_rate": 4.551199486215109e-06,
+      "loss": 0.5666,
+      "step": 33336
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.119537353515625,
+      "learning_rate": 4.5509368929399045e-06,
+      "loss": 0.7359,
+      "step": 33337
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.1973252296447754,
+      "learning_rate": 4.550674300913377e-06,
+      "loss": 0.6702,
+      "step": 33338
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.1500062942504883,
+      "learning_rate": 4.550411710136255e-06,
+      "loss": 0.4874,
+      "step": 33339
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.291999578475952,
+      "learning_rate": 4.550149120609271e-06,
+      "loss": 0.6526,
+      "step": 33340
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.38472580909729,
+      "learning_rate": 4.549886532333153e-06,
+      "loss": 0.4991,
+      "step": 33341
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 7.245355606079102,
+      "learning_rate": 4.549623945308632e-06,
+      "loss": 0.6734,
+      "step": 33342
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.9654685258865356,
+      "learning_rate": 4.549361359536438e-06,
+      "loss": 0.3617,
+      "step": 33343
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.204528570175171,
+      "learning_rate": 4.549098775017302e-06,
+      "loss": 0.8027,
+      "step": 33344
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.881640911102295,
+      "learning_rate": 4.548836191751953e-06,
+      "loss": 0.2923,
+      "step": 33345
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.31734037399292,
+      "learning_rate": 4.548573609741122e-06,
+      "loss": 0.4287,
+      "step": 33346
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.5696146488189697,
+      "learning_rate": 4.548311028985537e-06,
+      "loss": 0.4975,
+      "step": 33347
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.598183274269104,
+      "learning_rate": 4.548048449485931e-06,
+      "loss": 0.6118,
+      "step": 33348
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 3.101057767868042,
+      "learning_rate": 4.547785871243033e-06,
+      "loss": 0.7678,
+      "step": 33349
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 3.1571099758148193,
+      "learning_rate": 4.547523294257571e-06,
+      "loss": 0.7015,
+      "step": 33350
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.3134803771972656,
+      "learning_rate": 4.5472607185302796e-06,
+      "loss": 0.5876,
+      "step": 33351
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.800644874572754,
+      "learning_rate": 4.546998144061886e-06,
+      "loss": 0.48,
+      "step": 33352
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.5465443134307861,
+      "learning_rate": 4.54673557085312e-06,
+      "loss": 0.6467,
+      "step": 33353
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 4.605696678161621,
+      "learning_rate": 4.546472998904714e-06,
+      "loss": 0.6289,
+      "step": 33354
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.7386362552642822,
+      "learning_rate": 4.546210428217396e-06,
+      "loss": 0.9044,
+      "step": 33355
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.7945878505706787,
+      "learning_rate": 4.545947858791897e-06,
+      "loss": 0.5346,
+      "step": 33356
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.9173424243927,
+      "learning_rate": 4.5456852906289475e-06,
+      "loss": 0.8706,
+      "step": 33357
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.9402318000793457,
+      "learning_rate": 4.545422723729277e-06,
+      "loss": 0.6688,
+      "step": 33358
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.2191824913024902,
+      "learning_rate": 4.545160158093615e-06,
+      "loss": 0.4597,
+      "step": 33359
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.2663938999176025,
+      "learning_rate": 4.544897593722691e-06,
+      "loss": 0.539,
+      "step": 33360
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 3.2840936183929443,
+      "learning_rate": 4.544635030617238e-06,
+      "loss": 0.8581,
+      "step": 33361
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.9598586559295654,
+      "learning_rate": 4.544372468777984e-06,
+      "loss": 0.726,
+      "step": 33362
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.3771096467971802,
+      "learning_rate": 4.54410990820566e-06,
+      "loss": 0.3936,
+      "step": 33363
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.8071649074554443,
+      "learning_rate": 4.543847348900995e-06,
+      "loss": 0.7341,
+      "step": 33364
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 3.0521044731140137,
+      "learning_rate": 4.54358479086472e-06,
+      "loss": 0.8877,
+      "step": 33365
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 3.5647687911987305,
+      "learning_rate": 4.543322234097564e-06,
+      "loss": 0.5626,
+      "step": 33366
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.641427755355835,
+      "learning_rate": 4.543059678600258e-06,
+      "loss": 0.5597,
+      "step": 33367
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 3.5001208782196045,
+      "learning_rate": 4.542797124373532e-06,
+      "loss": 0.8529,
+      "step": 33368
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.5225958824157715,
+      "learning_rate": 4.542534571418116e-06,
+      "loss": 0.8153,
+      "step": 33369
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.9566762447357178,
+      "learning_rate": 4.54227201973474e-06,
+      "loss": 0.7238,
+      "step": 33370
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 3.8128199577331543,
+      "learning_rate": 4.542009469324133e-06,
+      "loss": 0.7063,
+      "step": 33371
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.0474448204040527,
+      "learning_rate": 4.541746920187027e-06,
+      "loss": 0.515,
+      "step": 33372
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 3.402362823486328,
+      "learning_rate": 4.54148437232415e-06,
+      "loss": 0.7586,
+      "step": 33373
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.7480533123016357,
+      "learning_rate": 4.541221825736235e-06,
+      "loss": 0.7738,
+      "step": 33374
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.7194273471832275,
+      "learning_rate": 4.54095928042401e-06,
+      "loss": 0.4164,
+      "step": 33375
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.5435950756072998,
+      "learning_rate": 4.540696736388205e-06,
+      "loss": 0.7534,
+      "step": 33376
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 3.5710949897766113,
+      "learning_rate": 4.54043419362955e-06,
+      "loss": 0.5549,
+      "step": 33377
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.561603307723999,
+      "learning_rate": 4.540171652148776e-06,
+      "loss": 0.5435,
+      "step": 33378
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.760321021080017,
+      "learning_rate": 4.539909111946612e-06,
+      "loss": 0.8214,
+      "step": 33379
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.144678831100464,
+      "learning_rate": 4.539646573023788e-06,
+      "loss": 0.4941,
+      "step": 33380
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.867753744125366,
+      "learning_rate": 4.539384035381035e-06,
+      "loss": 0.4959,
+      "step": 33381
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.418356418609619,
+      "learning_rate": 4.539121499019082e-06,
+      "loss": 0.6161,
+      "step": 33382
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 3.0015954971313477,
+      "learning_rate": 4.538858963938658e-06,
+      "loss": 0.8459,
+      "step": 33383
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.631279230117798,
+      "learning_rate": 4.538596430140497e-06,
+      "loss": 0.5437,
+      "step": 33384
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.6386303901672363,
+      "learning_rate": 4.538333897625326e-06,
+      "loss": 0.4644,
+      "step": 33385
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.603142738342285,
+      "learning_rate": 4.538071366393875e-06,
+      "loss": 0.625,
+      "step": 33386
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 3.286154270172119,
+      "learning_rate": 4.5378088364468745e-06,
+      "loss": 0.5869,
+      "step": 33387
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.8792296648025513,
+      "learning_rate": 4.537546307785055e-06,
+      "loss": 0.6765,
+      "step": 33388
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.1380386352539062,
+      "learning_rate": 4.5372837804091455e-06,
+      "loss": 0.4314,
+      "step": 33389
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 3.3095815181732178,
+      "learning_rate": 4.537021254319877e-06,
+      "loss": 0.4965,
+      "step": 33390
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 3.079238176345825,
+      "learning_rate": 4.536758729517977e-06,
+      "loss": 0.793,
+      "step": 33391
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.906646251678467,
+      "learning_rate": 4.53649620600418e-06,
+      "loss": 0.7921,
+      "step": 33392
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.157036066055298,
+      "learning_rate": 4.536233683779212e-06,
+      "loss": 0.8059,
+      "step": 33393
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.6056556701660156,
+      "learning_rate": 4.5359711628438045e-06,
+      "loss": 0.4774,
+      "step": 33394
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 3.0036332607269287,
+      "learning_rate": 4.5357086431986866e-06,
+      "loss": 0.8967,
+      "step": 33395
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.126744508743286,
+      "learning_rate": 4.53544612484459e-06,
+      "loss": 0.4591,
+      "step": 33396
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.7470453977584839,
+      "learning_rate": 4.535183607782244e-06,
+      "loss": 0.6159,
+      "step": 33397
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.7247092723846436,
+      "learning_rate": 4.534921092012379e-06,
+      "loss": 0.5785,
+      "step": 33398
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 3.336146116256714,
+      "learning_rate": 4.534658577535723e-06,
+      "loss": 0.6313,
+      "step": 33399
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.6023311614990234,
+      "learning_rate": 4.534396064353007e-06,
+      "loss": 0.5857,
+      "step": 33400
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.0013935565948486,
+      "learning_rate": 4.534133552464962e-06,
+      "loss": 0.645,
+      "step": 33401
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.8747973442077637,
+      "learning_rate": 4.533871041872316e-06,
+      "loss": 0.6521,
+      "step": 33402
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.855426073074341,
+      "learning_rate": 4.533608532575801e-06,
+      "loss": 0.4126,
+      "step": 33403
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.787219762802124,
+      "learning_rate": 4.533346024576146e-06,
+      "loss": 0.6671,
+      "step": 33404
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 3.113525152206421,
+      "learning_rate": 4.5330835178740805e-06,
+      "loss": 0.8632,
+      "step": 33405
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.066059112548828,
+      "learning_rate": 4.532821012470335e-06,
+      "loss": 0.4173,
+      "step": 33406
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.4046173095703125,
+      "learning_rate": 4.532558508365639e-06,
+      "loss": 0.6096,
+      "step": 33407
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 3.777158260345459,
+      "learning_rate": 4.532296005560723e-06,
+      "loss": 0.6525,
+      "step": 33408
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.8866338729858398,
+      "learning_rate": 4.532033504056317e-06,
+      "loss": 0.3428,
+      "step": 33409
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.7952462434768677,
+      "learning_rate": 4.531771003853149e-06,
+      "loss": 0.4493,
+      "step": 33410
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.7977983951568604,
+      "learning_rate": 4.53150850495195e-06,
+      "loss": 0.6568,
+      "step": 33411
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.6167668104171753,
+      "learning_rate": 4.531246007353451e-06,
+      "loss": 0.6886,
+      "step": 33412
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.190308094024658,
+      "learning_rate": 4.530983511058382e-06,
+      "loss": 0.5439,
+      "step": 33413
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.0942792892456055,
+      "learning_rate": 4.530721016067473e-06,
+      "loss": 0.829,
+      "step": 33414
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.776116132736206,
+      "learning_rate": 4.530458522381452e-06,
+      "loss": 0.5369,
+      "step": 33415
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.7220685482025146,
+      "learning_rate": 4.53019603000105e-06,
+      "loss": 0.442,
+      "step": 33416
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 4.580280780792236,
+      "learning_rate": 4.529933538926997e-06,
+      "loss": 0.6201,
+      "step": 33417
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 3.2764780521392822,
+      "learning_rate": 4.529671049160023e-06,
+      "loss": 0.5827,
+      "step": 33418
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.580173969268799,
+      "learning_rate": 4.529408560700857e-06,
+      "loss": 0.8029,
+      "step": 33419
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 3.3076045513153076,
+      "learning_rate": 4.52914607355023e-06,
+      "loss": 0.9162,
+      "step": 33420
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.144242525100708,
+      "learning_rate": 4.5288835877088715e-06,
+      "loss": 0.4032,
+      "step": 33421
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.7741098403930664,
+      "learning_rate": 4.52862110317751e-06,
+      "loss": 0.62,
+      "step": 33422
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.1644492149353027,
+      "learning_rate": 4.528358619956878e-06,
+      "loss": 0.471,
+      "step": 33423
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.23177170753479,
+      "learning_rate": 4.528096138047704e-06,
+      "loss": 0.6869,
+      "step": 33424
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 3.272029161453247,
+      "learning_rate": 4.527833657450718e-06,
+      "loss": 0.5729,
+      "step": 33425
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.6675630807876587,
+      "learning_rate": 4.5275711781666486e-06,
+      "loss": 0.7054,
+      "step": 33426
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.091278553009033,
+      "learning_rate": 4.5273087001962275e-06,
+      "loss": 0.7101,
+      "step": 33427
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.7028976678848267,
+      "learning_rate": 4.527046223540184e-06,
+      "loss": 0.643,
+      "step": 33428
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.6541919708251953,
+      "learning_rate": 4.526783748199248e-06,
+      "loss": 0.7602,
+      "step": 33429
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.8627040386199951,
+      "learning_rate": 4.526521274174148e-06,
+      "loss": 0.3663,
+      "step": 33430
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.9003632068634033,
+      "learning_rate": 4.526258801465616e-06,
+      "loss": 0.6868,
+      "step": 33431
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.2596654891967773,
+      "learning_rate": 4.525996330074378e-06,
+      "loss": 0.7364,
+      "step": 33432
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.3468869924545288,
+      "learning_rate": 4.525733860001169e-06,
+      "loss": 0.5347,
+      "step": 33433
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.1580135822296143,
+      "learning_rate": 4.525471391246717e-06,
+      "loss": 0.609,
+      "step": 33434
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.0454392433166504,
+      "learning_rate": 4.52520892381175e-06,
+      "loss": 0.7067,
+      "step": 33435
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 3.087475299835205,
+      "learning_rate": 4.5249464576970005e-06,
+      "loss": 0.631,
+      "step": 33436
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.802804708480835,
+      "learning_rate": 4.524683992903197e-06,
+      "loss": 0.4899,
+      "step": 33437
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.487234115600586,
+      "learning_rate": 4.524421529431068e-06,
+      "loss": 0.6275,
+      "step": 33438
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.561143398284912,
+      "learning_rate": 4.524159067281345e-06,
+      "loss": 0.7023,
+      "step": 33439
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 3.0375819206237793,
+      "learning_rate": 4.523896606454758e-06,
+      "loss": 0.4929,
+      "step": 33440
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.5543832778930664,
+      "learning_rate": 4.5236341469520356e-06,
+      "loss": 0.6975,
+      "step": 33441
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 3.4555976390838623,
+      "learning_rate": 4.523371688773909e-06,
+      "loss": 0.8984,
+      "step": 33442
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 4.671028137207031,
+      "learning_rate": 4.523109231921106e-06,
+      "loss": 0.7175,
+      "step": 33443
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.5594594478607178,
+      "learning_rate": 4.522846776394359e-06,
+      "loss": 0.5861,
+      "step": 33444
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.2778937816619873,
+      "learning_rate": 4.522584322194393e-06,
+      "loss": 0.726,
+      "step": 33445
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 3.5169003009796143,
+      "learning_rate": 4.522321869321944e-06,
+      "loss": 0.5613,
+      "step": 33446
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 3.701336622238159,
+      "learning_rate": 4.522059417777739e-06,
+      "loss": 1.0901,
+      "step": 33447
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 4.1420063972473145,
+      "learning_rate": 4.521796967562507e-06,
+      "loss": 0.8751,
+      "step": 33448
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.505159378051758,
+      "learning_rate": 4.521534518676979e-06,
+      "loss": 0.49,
+      "step": 33449
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.5883151292800903,
+      "learning_rate": 4.521272071121884e-06,
+      "loss": 0.461,
+      "step": 33450
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.0695266723632812,
+      "learning_rate": 4.521009624897952e-06,
+      "loss": 0.4915,
+      "step": 33451
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 3.8682408332824707,
+      "learning_rate": 4.5207471800059125e-06,
+      "loss": 0.6325,
+      "step": 33452
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.015052080154419,
+      "learning_rate": 4.5204847364464955e-06,
+      "loss": 0.6816,
+      "step": 33453
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 3.5956485271453857,
+      "learning_rate": 4.520222294220431e-06,
+      "loss": 0.652,
+      "step": 33454
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 4.416799068450928,
+      "learning_rate": 4.51995985332845e-06,
+      "loss": 0.2647,
+      "step": 33455
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.4801926612854004,
+      "learning_rate": 4.519697413771278e-06,
+      "loss": 0.5009,
+      "step": 33456
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.5639872550964355,
+      "learning_rate": 4.519434975549649e-06,
+      "loss": 0.7535,
+      "step": 33457
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.276711940765381,
+      "learning_rate": 4.5191725386642915e-06,
+      "loss": 0.6793,
+      "step": 33458
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.5537633895874023,
+      "learning_rate": 4.518910103115935e-06,
+      "loss": 0.4617,
+      "step": 33459
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.372497320175171,
+      "learning_rate": 4.51864766890531e-06,
+      "loss": 0.5623,
+      "step": 33460
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.3474860191345215,
+      "learning_rate": 4.5183852360331444e-06,
+      "loss": 0.7278,
+      "step": 33461
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.6019716262817383,
+      "learning_rate": 4.51812280450017e-06,
+      "loss": 0.7613,
+      "step": 33462
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.6748533248901367,
+      "learning_rate": 4.517860374307115e-06,
+      "loss": 0.5002,
+      "step": 33463
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.0832390785217285,
+      "learning_rate": 4.5175979454547095e-06,
+      "loss": 0.6479,
+      "step": 33464
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.945331335067749,
+      "learning_rate": 4.5173355179436836e-06,
+      "loss": 0.7603,
+      "step": 33465
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.1929829120635986,
+      "learning_rate": 4.517073091774768e-06,
+      "loss": 0.5968,
+      "step": 33466
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.9802242517471313,
+      "learning_rate": 4.51681066694869e-06,
+      "loss": 0.7165,
+      "step": 33467
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.934251070022583,
+      "learning_rate": 4.51654824346618e-06,
+      "loss": 0.7927,
+      "step": 33468
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.828147053718567,
+      "learning_rate": 4.516285821327969e-06,
+      "loss": 0.6143,
+      "step": 33469
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.4716150760650635,
+      "learning_rate": 4.516023400534786e-06,
+      "loss": 0.6526,
+      "step": 33470
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.881474256515503,
+      "learning_rate": 4.51576098108736e-06,
+      "loss": 0.8365,
+      "step": 33471
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.3225367069244385,
+      "learning_rate": 4.51549856298642e-06,
+      "loss": 0.3487,
+      "step": 33472
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.5304371118545532,
+      "learning_rate": 4.515236146232699e-06,
+      "loss": 0.6045,
+      "step": 33473
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.138962984085083,
+      "learning_rate": 4.514973730826924e-06,
+      "loss": 0.6754,
+      "step": 33474
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 4.203965663909912,
+      "learning_rate": 4.514711316769826e-06,
+      "loss": 0.6391,
+      "step": 33475
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 3.164937734603882,
+      "learning_rate": 4.514448904062134e-06,
+      "loss": 0.5841,
+      "step": 33476
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.8510913848876953,
+      "learning_rate": 4.514186492704577e-06,
+      "loss": 0.7012,
+      "step": 33477
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.9595954418182373,
+      "learning_rate": 4.513924082697886e-06,
+      "loss": 0.6378,
+      "step": 33478
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 3.1860015392303467,
+      "learning_rate": 4.51366167404279e-06,
+      "loss": 0.4847,
+      "step": 33479
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 4.155419826507568,
+      "learning_rate": 4.513399266740018e-06,
+      "loss": 0.7204,
+      "step": 33480
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.8222599029541016,
+      "learning_rate": 4.5131368607903016e-06,
+      "loss": 0.7389,
+      "step": 33481
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 3.2005162239074707,
+      "learning_rate": 4.512874456194368e-06,
+      "loss": 0.6089,
+      "step": 33482
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.7805407047271729,
+      "learning_rate": 4.512612052952949e-06,
+      "loss": 0.6127,
+      "step": 33483
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 4.844632625579834,
+      "learning_rate": 4.512349651066773e-06,
+      "loss": 0.6238,
+      "step": 33484
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 3.167776584625244,
+      "learning_rate": 4.51208725053657e-06,
+      "loss": 0.7877,
+      "step": 33485
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 3.0291900634765625,
+      "learning_rate": 4.51182485136307e-06,
+      "loss": 0.7785,
+      "step": 33486
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.8464527130126953,
+      "learning_rate": 4.511562453547001e-06,
+      "loss": 0.4339,
+      "step": 33487
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 3.151851177215576,
+      "learning_rate": 4.511300057089094e-06,
+      "loss": 0.7515,
+      "step": 33488
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.7103474140167236,
+      "learning_rate": 4.51103766199008e-06,
+      "loss": 0.6398,
+      "step": 33489
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.055654525756836,
+      "learning_rate": 4.510775268250686e-06,
+      "loss": 0.7972,
+      "step": 33490
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 3.4535698890686035,
+      "learning_rate": 4.510512875871642e-06,
+      "loss": 0.7535,
+      "step": 33491
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.1893327236175537,
+      "learning_rate": 4.51025048485368e-06,
+      "loss": 0.8038,
+      "step": 33492
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.5032436847686768,
+      "learning_rate": 4.509988095197525e-06,
+      "loss": 0.7027,
+      "step": 33493
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.5450226068496704,
+      "learning_rate": 4.509725706903911e-06,
+      "loss": 0.6492,
+      "step": 33494
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.6500643491744995,
+      "learning_rate": 4.509463319973567e-06,
+      "loss": 0.6745,
+      "step": 33495
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.8520846366882324,
+      "learning_rate": 4.5092009344072214e-06,
+      "loss": 0.6034,
+      "step": 33496
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.119032144546509,
+      "learning_rate": 4.508938550205605e-06,
+      "loss": 0.6884,
+      "step": 33497
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.3608514070510864,
+      "learning_rate": 4.5086761673694455e-06,
+      "loss": 0.4206,
+      "step": 33498
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.320286273956299,
+      "learning_rate": 4.508413785899474e-06,
+      "loss": 0.7375,
+      "step": 33499
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.289553165435791,
+      "learning_rate": 4.5081514057964194e-06,
+      "loss": 0.4654,
+      "step": 33500
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.382361650466919,
+      "learning_rate": 4.5078890270610115e-06,
+      "loss": 0.377,
+      "step": 33501
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 5.7924933433532715,
+      "learning_rate": 4.5076266496939794e-06,
+      "loss": 0.5994,
+      "step": 33502
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.833171010017395,
+      "learning_rate": 4.507364273696054e-06,
+      "loss": 0.7492,
+      "step": 33503
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.701672315597534,
+      "learning_rate": 4.507101899067964e-06,
+      "loss": 0.5901,
+      "step": 33504
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.9985183477401733,
+      "learning_rate": 4.506839525810438e-06,
+      "loss": 0.489,
+      "step": 33505
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.4989778995513916,
+      "learning_rate": 4.506577153924207e-06,
+      "loss": 0.5201,
+      "step": 33506
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.468007802963257,
+      "learning_rate": 4.50631478341e-06,
+      "loss": 0.4432,
+      "step": 33507
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.0681514739990234,
+      "learning_rate": 4.506052414268546e-06,
+      "loss": 0.5967,
+      "step": 33508
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.9261975288391113,
+      "learning_rate": 4.505790046500577e-06,
+      "loss": 0.4984,
+      "step": 33509
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.228989839553833,
+      "learning_rate": 4.505527680106819e-06,
+      "loss": 0.4584,
+      "step": 33510
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 6.0490875244140625,
+      "learning_rate": 4.5052653150880046e-06,
+      "loss": 0.6292,
+      "step": 33511
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.8098530769348145,
+      "learning_rate": 4.505002951444861e-06,
+      "loss": 0.6498,
+      "step": 33512
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.8994545936584473,
+      "learning_rate": 4.504740589178118e-06,
+      "loss": 0.8509,
+      "step": 33513
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.8261903524398804,
+      "learning_rate": 4.504478228288508e-06,
+      "loss": 0.7262,
+      "step": 33514
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 3.2633588314056396,
+      "learning_rate": 4.5042158687767576e-06,
+      "loss": 0.8098,
+      "step": 33515
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 4.326955318450928,
+      "learning_rate": 4.503953510643597e-06,
+      "loss": 0.6847,
+      "step": 33516
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.891667604446411,
+      "learning_rate": 4.503691153889756e-06,
+      "loss": 0.5684,
+      "step": 33517
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.624852418899536,
+      "learning_rate": 4.503428798515963e-06,
+      "loss": 0.6847,
+      "step": 33518
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.734018087387085,
+      "learning_rate": 4.50316644452295e-06,
+      "loss": 0.5274,
+      "step": 33519
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.438406229019165,
+      "learning_rate": 4.502904091911444e-06,
+      "loss": 0.5686,
+      "step": 33520
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 3.4983084201812744,
+      "learning_rate": 4.502641740682176e-06,
+      "loss": 0.6178,
+      "step": 33521
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.6325953006744385,
+      "learning_rate": 4.502379390835875e-06,
+      "loss": 0.5329,
+      "step": 33522
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.2868194580078125,
+      "learning_rate": 4.502117042373271e-06,
+      "loss": 0.7656,
+      "step": 33523
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.9881768226623535,
+      "learning_rate": 4.501854695295092e-06,
+      "loss": 0.4052,
+      "step": 33524
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.0207033157348633,
+      "learning_rate": 4.501592349602069e-06,
+      "loss": 0.6731,
+      "step": 33525
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.7731657028198242,
+      "learning_rate": 4.501330005294931e-06,
+      "loss": 0.718,
+      "step": 33526
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.5598535537719727,
+      "learning_rate": 4.501067662374408e-06,
+      "loss": 0.6709,
+      "step": 33527
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.7702784538269043,
+      "learning_rate": 4.500805320841227e-06,
+      "loss": 0.7684,
+      "step": 33528
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 3.9367170333862305,
+      "learning_rate": 4.500542980696121e-06,
+      "loss": 0.8702,
+      "step": 33529
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.8857109546661377,
+      "learning_rate": 4.500280641939818e-06,
+      "loss": 0.5295,
+      "step": 33530
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 3.2792413234710693,
+      "learning_rate": 4.500018304573047e-06,
+      "loss": 0.6321,
+      "step": 33531
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.3770363330841064,
+      "learning_rate": 4.499755968596538e-06,
+      "loss": 0.41,
+      "step": 33532
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.8289320468902588,
+      "learning_rate": 4.499493634011018e-06,
+      "loss": 0.6268,
+      "step": 33533
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.9879204034805298,
+      "learning_rate": 4.4992313008172204e-06,
+      "loss": 0.6877,
+      "step": 33534
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.62097430229187,
+      "learning_rate": 4.498968969015873e-06,
+      "loss": 0.4912,
+      "step": 33535
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.4837937355041504,
+      "learning_rate": 4.498706638607706e-06,
+      "loss": 0.6032,
+      "step": 33536
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.8822253942489624,
+      "learning_rate": 4.498444309593447e-06,
+      "loss": 0.5889,
+      "step": 33537
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.8618600368499756,
+      "learning_rate": 4.498181981973828e-06,
+      "loss": 0.7169,
+      "step": 33538
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.3466665744781494,
+      "learning_rate": 4.4979196557495755e-06,
+      "loss": 0.6286,
+      "step": 33539
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.3142149448394775,
+      "learning_rate": 4.49765733092142e-06,
+      "loss": 0.5986,
+      "step": 33540
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 3.6895363330841064,
+      "learning_rate": 4.497395007490093e-06,
+      "loss": 0.755,
+      "step": 33541
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 3.3267486095428467,
+      "learning_rate": 4.497132685456321e-06,
+      "loss": 0.6431,
+      "step": 33542
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.3823626041412354,
+      "learning_rate": 4.496870364820835e-06,
+      "loss": 0.5563,
+      "step": 33543
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.6984823942184448,
+      "learning_rate": 4.4966080455843635e-06,
+      "loss": 0.5854,
+      "step": 33544
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 3.382549285888672,
+      "learning_rate": 4.496345727747637e-06,
+      "loss": 0.6242,
+      "step": 33545
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.2818877696990967,
+      "learning_rate": 4.496083411311385e-06,
+      "loss": 0.5809,
+      "step": 33546
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.6154059171676636,
+      "learning_rate": 4.495821096276335e-06,
+      "loss": 0.6981,
+      "step": 33547
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.6040024757385254,
+      "learning_rate": 4.495558782643218e-06,
+      "loss": 0.4722,
+      "step": 33548
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.660776138305664,
+      "learning_rate": 4.4952964704127635e-06,
+      "loss": 0.607,
+      "step": 33549
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 3.5633299350738525,
+      "learning_rate": 4.4950341595857e-06,
+      "loss": 0.9324,
+      "step": 33550
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.2636072635650635,
+      "learning_rate": 4.494771850162757e-06,
+      "loss": 0.3338,
+      "step": 33551
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 4.19869327545166,
+      "learning_rate": 4.494509542144666e-06,
+      "loss": 0.6534,
+      "step": 33552
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.6960747241973877,
+      "learning_rate": 4.494247235532152e-06,
+      "loss": 0.7308,
+      "step": 33553
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.521061897277832,
+      "learning_rate": 4.493984930325947e-06,
+      "loss": 0.4006,
+      "step": 33554
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 3.0692784786224365,
+      "learning_rate": 4.493722626526781e-06,
+      "loss": 0.569,
+      "step": 33555
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.55393123626709,
+      "learning_rate": 4.493460324135384e-06,
+      "loss": 0.5689,
+      "step": 33556
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.007147789001465,
+      "learning_rate": 4.493198023152483e-06,
+      "loss": 0.53,
+      "step": 33557
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.4993908405303955,
+      "learning_rate": 4.492935723578809e-06,
+      "loss": 0.5952,
+      "step": 33558
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.6300621032714844,
+      "learning_rate": 4.492673425415091e-06,
+      "loss": 0.4944,
+      "step": 33559
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.9201748371124268,
+      "learning_rate": 4.492411128662057e-06,
+      "loss": 0.776,
+      "step": 33560
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.8543845415115356,
+      "learning_rate": 4.4921488333204395e-06,
+      "loss": 0.638,
+      "step": 33561
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 3.710190534591675,
+      "learning_rate": 4.491886539390965e-06,
+      "loss": 0.6877,
+      "step": 33562
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.174543619155884,
+      "learning_rate": 4.491624246874363e-06,
+      "loss": 0.458,
+      "step": 33563
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.7944579124450684,
+      "learning_rate": 4.491361955771364e-06,
+      "loss": 0.664,
+      "step": 33564
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 3.181757926940918,
+      "learning_rate": 4.491099666082697e-06,
+      "loss": 0.5801,
+      "step": 33565
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.189406156539917,
+      "learning_rate": 4.490837377809091e-06,
+      "loss": 0.6957,
+      "step": 33566
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.980941891670227,
+      "learning_rate": 4.4905750909512755e-06,
+      "loss": 0.6615,
+      "step": 33567
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.5754456520080566,
+      "learning_rate": 4.490312805509978e-06,
+      "loss": 0.6763,
+      "step": 33568
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.294546604156494,
+      "learning_rate": 4.490050521485932e-06,
+      "loss": 0.6032,
+      "step": 33569
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.7195134162902832,
+      "learning_rate": 4.489788238879864e-06,
+      "loss": 0.6118,
+      "step": 33570
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.478517532348633,
+      "learning_rate": 4.489525957692504e-06,
+      "loss": 0.5538,
+      "step": 33571
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 3.1409778594970703,
+      "learning_rate": 4.48926367792458e-06,
+      "loss": 0.8071,
+      "step": 33572
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.851040840148926,
+      "learning_rate": 4.489001399576824e-06,
+      "loss": 0.6211,
+      "step": 33573
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 3.6205854415893555,
+      "learning_rate": 4.4887391226499616e-06,
+      "loss": 0.679,
+      "step": 33574
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.9317288398742676,
+      "learning_rate": 4.488476847144726e-06,
+      "loss": 0.9467,
+      "step": 33575
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.8864407539367676,
+      "learning_rate": 4.4882145730618435e-06,
+      "loss": 0.6621,
+      "step": 33576
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.809549570083618,
+      "learning_rate": 4.487952300402045e-06,
+      "loss": 0.8703,
+      "step": 33577
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.0743448734283447,
+      "learning_rate": 4.4876900291660595e-06,
+      "loss": 0.732,
+      "step": 33578
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.771484613418579,
+      "learning_rate": 4.487427759354616e-06,
+      "loss": 0.6158,
+      "step": 33579
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.7438181638717651,
+      "learning_rate": 4.487165490968443e-06,
+      "loss": 0.3338,
+      "step": 33580
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.404128074645996,
+      "learning_rate": 4.486903224008272e-06,
+      "loss": 0.5327,
+      "step": 33581
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 4.5280303955078125,
+      "learning_rate": 4.486640958474831e-06,
+      "loss": 0.8272,
+      "step": 33582
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.057950496673584,
+      "learning_rate": 4.486378694368848e-06,
+      "loss": 0.6985,
+      "step": 33583
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.5033280849456787,
+      "learning_rate": 4.4861164316910545e-06,
+      "loss": 0.7972,
+      "step": 33584
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.102487564086914,
+      "learning_rate": 4.485854170442178e-06,
+      "loss": 0.4673,
+      "step": 33585
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.5673398971557617,
+      "learning_rate": 4.48559191062295e-06,
+      "loss": 0.7196,
+      "step": 33586
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.9248374700546265,
+      "learning_rate": 4.485329652234097e-06,
+      "loss": 0.6612,
+      "step": 33587
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.2881100177764893,
+      "learning_rate": 4.485067395276349e-06,
+      "loss": 0.8282,
+      "step": 33588
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.447131395339966,
+      "learning_rate": 4.484805139750436e-06,
+      "loss": 0.5099,
+      "step": 33589
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 3.4727489948272705,
+      "learning_rate": 4.484542885657087e-06,
+      "loss": 0.9265,
+      "step": 33590
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.9196906089782715,
+      "learning_rate": 4.484280632997032e-06,
+      "loss": 0.5662,
+      "step": 33591
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.140001058578491,
+      "learning_rate": 4.4840183817709984e-06,
+      "loss": 0.635,
+      "step": 33592
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.7419023513793945,
+      "learning_rate": 4.483756131979716e-06,
+      "loss": 0.7243,
+      "step": 33593
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.293820381164551,
+      "learning_rate": 4.483493883623914e-06,
+      "loss": 0.7697,
+      "step": 33594
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.6427441835403442,
+      "learning_rate": 4.483231636704322e-06,
+      "loss": 0.5627,
+      "step": 33595
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.9358813762664795,
+      "learning_rate": 4.482969391221671e-06,
+      "loss": 0.7754,
+      "step": 33596
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.8219820261001587,
+      "learning_rate": 4.482707147176688e-06,
+      "loss": 0.955,
+      "step": 33597
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.0635828971862793,
+      "learning_rate": 4.482444904570102e-06,
+      "loss": 0.5606,
+      "step": 33598
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.5159633159637451,
+      "learning_rate": 4.4821826634026426e-06,
+      "loss": 0.8004,
+      "step": 33599
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.683834195137024,
+      "learning_rate": 4.48192042367504e-06,
+      "loss": 0.6012,
+      "step": 33600
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.033151149749756,
+      "learning_rate": 4.4816581853880225e-06,
+      "loss": 0.5432,
+      "step": 33601
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.3926355838775635,
+      "learning_rate": 4.481395948542319e-06,
+      "loss": 0.7595,
+      "step": 33602
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.2181472778320312,
+      "learning_rate": 4.48113371313866e-06,
+      "loss": 0.4586,
+      "step": 33603
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 3.701000928878784,
+      "learning_rate": 4.480871479177773e-06,
+      "loss": 0.6548,
+      "step": 33604
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.8382179737091064,
+      "learning_rate": 4.480609246660388e-06,
+      "loss": 0.5142,
+      "step": 33605
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.1260926723480225,
+      "learning_rate": 4.4803470155872345e-06,
+      "loss": 0.593,
+      "step": 33606
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.590710401535034,
+      "learning_rate": 4.480084785959041e-06,
+      "loss": 0.5484,
+      "step": 33607
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.212949752807617,
+      "learning_rate": 4.479822557776537e-06,
+      "loss": 0.6814,
+      "step": 33608
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.9778320789337158,
+      "learning_rate": 4.479560331040451e-06,
+      "loss": 0.7907,
+      "step": 33609
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 3.426927089691162,
+      "learning_rate": 4.479298105751514e-06,
+      "loss": 0.8292,
+      "step": 33610
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.8218088150024414,
+      "learning_rate": 4.4790358819104535e-06,
+      "loss": 0.6835,
+      "step": 33611
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.1150641441345215,
+      "learning_rate": 4.478773659517998e-06,
+      "loss": 0.4218,
+      "step": 33612
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.8185880184173584,
+      "learning_rate": 4.478511438574879e-06,
+      "loss": 0.7955,
+      "step": 33613
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.35836124420166,
+      "learning_rate": 4.478249219081825e-06,
+      "loss": 0.8142,
+      "step": 33614
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.9547433853149414,
+      "learning_rate": 4.477987001039561e-06,
+      "loss": 0.4818,
+      "step": 33615
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.701107978820801,
+      "learning_rate": 4.477724784448821e-06,
+      "loss": 0.5102,
+      "step": 33616
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.1345789432525635,
+      "learning_rate": 4.477462569310334e-06,
+      "loss": 0.4522,
+      "step": 33617
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.35421621799469,
+      "learning_rate": 4.477200355624827e-06,
+      "loss": 0.463,
+      "step": 33618
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.051692485809326,
+      "learning_rate": 4.476938143393031e-06,
+      "loss": 0.4375,
+      "step": 33619
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.8044092655181885,
+      "learning_rate": 4.476675932615674e-06,
+      "loss": 0.616,
+      "step": 33620
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.8429479598999023,
+      "learning_rate": 4.476413723293484e-06,
+      "loss": 0.6129,
+      "step": 33621
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.1688170433044434,
+      "learning_rate": 4.476151515427193e-06,
+      "loss": 0.48,
+      "step": 33622
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.010658025741577,
+      "learning_rate": 4.475889309017527e-06,
+      "loss": 0.51,
+      "step": 33623
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.27836012840271,
+      "learning_rate": 4.4756271040652165e-06,
+      "loss": 0.4166,
+      "step": 33624
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.6796423196792603,
+      "learning_rate": 4.4753649005709905e-06,
+      "loss": 0.5674,
+      "step": 33625
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.2902846336364746,
+      "learning_rate": 4.47510269853558e-06,
+      "loss": 0.5601,
+      "step": 33626
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.9315983057022095,
+      "learning_rate": 4.47484049795971e-06,
+      "loss": 0.4166,
+      "step": 33627
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.055147647857666,
+      "learning_rate": 4.474578298844113e-06,
+      "loss": 0.6416,
+      "step": 33628
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.516889810562134,
+      "learning_rate": 4.474316101189517e-06,
+      "loss": 0.6923,
+      "step": 33629
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.9167407751083374,
+      "learning_rate": 4.47405390499665e-06,
+      "loss": 0.4688,
+      "step": 33630
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 3.144265651702881,
+      "learning_rate": 4.473791710266243e-06,
+      "loss": 0.478,
+      "step": 33631
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 3.2639975547790527,
+      "learning_rate": 4.473529516999023e-06,
+      "loss": 0.5429,
+      "step": 33632
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.4656412601470947,
+      "learning_rate": 4.473267325195721e-06,
+      "loss": 0.6894,
+      "step": 33633
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.6597683429718018,
+      "learning_rate": 4.473005134857066e-06,
+      "loss": 0.6368,
+      "step": 33634
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.6982547044754028,
+      "learning_rate": 4.4727429459837845e-06,
+      "loss": 0.6568,
+      "step": 33635
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.32338809967041,
+      "learning_rate": 4.472480758576608e-06,
+      "loss": 0.4296,
+      "step": 33636
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.3302478790283203,
+      "learning_rate": 4.472218572636266e-06,
+      "loss": 0.5437,
+      "step": 33637
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.6236202716827393,
+      "learning_rate": 4.471956388163486e-06,
+      "loss": 0.4892,
+      "step": 33638
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.3669071197509766,
+      "learning_rate": 4.471694205158997e-06,
+      "loss": 0.6175,
+      "step": 33639
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.097442150115967,
+      "learning_rate": 4.47143202362353e-06,
+      "loss": 0.8644,
+      "step": 33640
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.8792612552642822,
+      "learning_rate": 4.47116984355781e-06,
+      "loss": 0.6091,
+      "step": 33641
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.558187246322632,
+      "learning_rate": 4.47090766496257e-06,
+      "loss": 0.6885,
+      "step": 33642
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.540998101234436,
+      "learning_rate": 4.470645487838537e-06,
+      "loss": 0.7459,
+      "step": 33643
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.4964275360107422,
+      "learning_rate": 4.470383312186441e-06,
+      "loss": 0.6268,
+      "step": 33644
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 3.168442964553833,
+      "learning_rate": 4.470121138007011e-06,
+      "loss": 0.6022,
+      "step": 33645
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 4.15545129776001,
+      "learning_rate": 4.469858965300975e-06,
+      "loss": 0.6689,
+      "step": 33646
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.8954739570617676,
+      "learning_rate": 4.469596794069063e-06,
+      "loss": 0.5406,
+      "step": 33647
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 3.871504545211792,
+      "learning_rate": 4.469334624312004e-06,
+      "loss": 0.7011,
+      "step": 33648
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.414380669593811,
+      "learning_rate": 4.469072456030526e-06,
+      "loss": 0.5441,
+      "step": 33649
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 3.720869302749634,
+      "learning_rate": 4.468810289225359e-06,
+      "loss": 0.4931,
+      "step": 33650
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.773435115814209,
+      "learning_rate": 4.468548123897231e-06,
+      "loss": 0.5343,
+      "step": 33651
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.7059426307678223,
+      "learning_rate": 4.468285960046872e-06,
+      "loss": 0.5216,
+      "step": 33652
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.636154055595398,
+      "learning_rate": 4.468023797675011e-06,
+      "loss": 0.6046,
+      "step": 33653
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.737466335296631,
+      "learning_rate": 4.4677616367823755e-06,
+      "loss": 0.8504,
+      "step": 33654
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.955140471458435,
+      "learning_rate": 4.467499477369695e-06,
+      "loss": 0.7016,
+      "step": 33655
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 3.8526506423950195,
+      "learning_rate": 4.4672373194377e-06,
+      "loss": 0.924,
+      "step": 33656
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.569983959197998,
+      "learning_rate": 4.466975162987118e-06,
+      "loss": 0.4552,
+      "step": 33657
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.9498229026794434,
+      "learning_rate": 4.4667130080186785e-06,
+      "loss": 0.5629,
+      "step": 33658
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.9719624519348145,
+      "learning_rate": 4.466450854533111e-06,
+      "loss": 0.7852,
+      "step": 33659
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.728285312652588,
+      "learning_rate": 4.4661887025311434e-06,
+      "loss": 0.6509,
+      "step": 33660
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.133979558944702,
+      "learning_rate": 4.465926552013506e-06,
+      "loss": 0.4314,
+      "step": 33661
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.6970410346984863,
+      "learning_rate": 4.465664402980926e-06,
+      "loss": 0.7344,
+      "step": 33662
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 3.2904036045074463,
+      "learning_rate": 4.465402255434132e-06,
+      "loss": 0.7023,
+      "step": 33663
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.8049834966659546,
+      "learning_rate": 4.465140109373855e-06,
+      "loss": 0.6895,
+      "step": 33664
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.259185314178467,
+      "learning_rate": 4.464877964800824e-06,
+      "loss": 0.6011,
+      "step": 33665
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.452979326248169,
+      "learning_rate": 4.464615821715766e-06,
+      "loss": 0.6884,
+      "step": 33666
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 3.9293503761291504,
+      "learning_rate": 4.464353680119411e-06,
+      "loss": 0.9387,
+      "step": 33667
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.6238397359848022,
+      "learning_rate": 4.464091540012488e-06,
+      "loss": 0.6198,
+      "step": 33668
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.065591812133789,
+      "learning_rate": 4.463829401395725e-06,
+      "loss": 0.5369,
+      "step": 33669
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.341486930847168,
+      "learning_rate": 4.463567264269852e-06,
+      "loss": 0.8221,
+      "step": 33670
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 3.385601043701172,
+      "learning_rate": 4.4633051286355975e-06,
+      "loss": 0.4821,
+      "step": 33671
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.5760152339935303,
+      "learning_rate": 4.463042994493691e-06,
+      "loss": 0.5252,
+      "step": 33672
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.820571184158325,
+      "learning_rate": 4.462780861844859e-06,
+      "loss": 0.501,
+      "step": 33673
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 4.27077054977417,
+      "learning_rate": 4.462518730689834e-06,
+      "loss": 0.5426,
+      "step": 33674
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 3.9330193996429443,
+      "learning_rate": 4.462256601029342e-06,
+      "loss": 0.5729,
+      "step": 33675
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 3.978752851486206,
+      "learning_rate": 4.461994472864111e-06,
+      "loss": 0.7464,
+      "step": 33676
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.3057265281677246,
+      "learning_rate": 4.4617323461948746e-06,
+      "loss": 0.66,
+      "step": 33677
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.1174156665802,
+      "learning_rate": 4.461470221022358e-06,
+      "loss": 0.4781,
+      "step": 33678
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.2670886516571045,
+      "learning_rate": 4.461208097347292e-06,
+      "loss": 0.962,
+      "step": 33679
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.0110106468200684,
+      "learning_rate": 4.460945975170403e-06,
+      "loss": 0.5167,
+      "step": 33680
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 4.204893112182617,
+      "learning_rate": 4.460683854492422e-06,
+      "loss": 0.6486,
+      "step": 33681
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.868107795715332,
+      "learning_rate": 4.4604217353140776e-06,
+      "loss": 0.7391,
+      "step": 33682
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 3.502297878265381,
+      "learning_rate": 4.4601596176360985e-06,
+      "loss": 0.6979,
+      "step": 33683
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.033747673034668,
+      "learning_rate": 4.4598975014592126e-06,
+      "loss": 0.83,
+      "step": 33684
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.1642472743988037,
+      "learning_rate": 4.45963538678415e-06,
+      "loss": 0.5907,
+      "step": 33685
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 3.548044204711914,
+      "learning_rate": 4.459373273611638e-06,
+      "loss": 0.7268,
+      "step": 33686
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.522923469543457,
+      "learning_rate": 4.4591111619424075e-06,
+      "loss": 0.5772,
+      "step": 33687
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.87668776512146,
+      "learning_rate": 4.4588490517771855e-06,
+      "loss": 0.7961,
+      "step": 33688
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.9435676336288452,
+      "learning_rate": 4.458586943116701e-06,
+      "loss": 0.536,
+      "step": 33689
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.008126974105835,
+      "learning_rate": 4.4583248359616845e-06,
+      "loss": 0.5687,
+      "step": 33690
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.0789008140563965,
+      "learning_rate": 4.458062730312863e-06,
+      "loss": 0.6397,
+      "step": 33691
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 4.979875087738037,
+      "learning_rate": 4.457800626170965e-06,
+      "loss": 0.5865,
+      "step": 33692
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.806175947189331,
+      "learning_rate": 4.457538523536723e-06,
+      "loss": 0.5107,
+      "step": 33693
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.9757025241851807,
+      "learning_rate": 4.457276422410861e-06,
+      "loss": 0.8099,
+      "step": 33694
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.0969693660736084,
+      "learning_rate": 4.457014322794111e-06,
+      "loss": 0.6755,
+      "step": 33695
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.6943118572235107,
+      "learning_rate": 4.4567522246872e-06,
+      "loss": 0.6224,
+      "step": 33696
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.2228012084960938,
+      "learning_rate": 4.456490128090858e-06,
+      "loss": 0.6508,
+      "step": 33697
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.8887909650802612,
+      "learning_rate": 4.456228033005813e-06,
+      "loss": 0.8266,
+      "step": 33698
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 3.215834140777588,
+      "learning_rate": 4.4559659394327944e-06,
+      "loss": 0.8155,
+      "step": 33699
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 3.290800094604492,
+      "learning_rate": 4.4557038473725315e-06,
+      "loss": 0.8196,
+      "step": 33700
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.591987133026123,
+      "learning_rate": 4.455441756825752e-06,
+      "loss": 0.6538,
+      "step": 33701
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.6910557746887207,
+      "learning_rate": 4.455179667793185e-06,
+      "loss": 0.755,
+      "step": 33702
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.3862292766571045,
+      "learning_rate": 4.454917580275557e-06,
+      "loss": 0.7301,
+      "step": 33703
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.949573040008545,
+      "learning_rate": 4.454655494273601e-06,
+      "loss": 0.6253,
+      "step": 33704
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.8343135118484497,
+      "learning_rate": 4.454393409788044e-06,
+      "loss": 0.5731,
+      "step": 33705
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 5.383066177368164,
+      "learning_rate": 4.454131326819614e-06,
+      "loss": 0.3898,
+      "step": 33706
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.290940999984741,
+      "learning_rate": 4.453869245369041e-06,
+      "loss": 0.5646,
+      "step": 33707
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.725445508956909,
+      "learning_rate": 4.4536071654370526e-06,
+      "loss": 0.578,
+      "step": 33708
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.6668293476104736,
+      "learning_rate": 4.453345087024378e-06,
+      "loss": 0.588,
+      "step": 33709
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.0851097106933594,
+      "learning_rate": 4.453083010131746e-06,
+      "loss": 0.3718,
+      "step": 33710
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 3.120622158050537,
+      "learning_rate": 4.452820934759886e-06,
+      "loss": 0.631,
+      "step": 33711
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.3166399002075195,
+      "learning_rate": 4.452558860909525e-06,
+      "loss": 0.9836,
+      "step": 33712
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.1248462200164795,
+      "learning_rate": 4.452296788581393e-06,
+      "loss": 0.4045,
+      "step": 33713
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.847611427307129,
+      "learning_rate": 4.452034717776218e-06,
+      "loss": 0.4621,
+      "step": 33714
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.6789883375167847,
+      "learning_rate": 4.45177264849473e-06,
+      "loss": 0.6444,
+      "step": 33715
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.6803512573242188,
+      "learning_rate": 4.451510580737654e-06,
+      "loss": 0.4621,
+      "step": 33716
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 4.08566427230835,
+      "learning_rate": 4.451248514505725e-06,
+      "loss": 0.7776,
+      "step": 33717
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.048511505126953,
+      "learning_rate": 4.4509864497996666e-06,
+      "loss": 0.51,
+      "step": 33718
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.64280366897583,
+      "learning_rate": 4.450724386620211e-06,
+      "loss": 0.7148,
+      "step": 33719
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.226900100708008,
+      "learning_rate": 4.450462324968083e-06,
+      "loss": 0.6842,
+      "step": 33720
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 3.0492069721221924,
+      "learning_rate": 4.450200264844015e-06,
+      "loss": 0.7371,
+      "step": 33721
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.113530397415161,
+      "learning_rate": 4.449938206248734e-06,
+      "loss": 0.5041,
+      "step": 33722
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 3.1890439987182617,
+      "learning_rate": 4.449676149182968e-06,
+      "loss": 0.8672,
+      "step": 33723
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.4415791034698486,
+      "learning_rate": 4.449414093647447e-06,
+      "loss": 0.4431,
+      "step": 33724
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 3.0669779777526855,
+      "learning_rate": 4.4491520396429e-06,
+      "loss": 0.97,
+      "step": 33725
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.39298415184021,
+      "learning_rate": 4.448889987170053e-06,
+      "loss": 0.6193,
+      "step": 33726
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.7902857065200806,
+      "learning_rate": 4.448627936229637e-06,
+      "loss": 0.3725,
+      "step": 33727
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.123302936553955,
+      "learning_rate": 4.4483658868223815e-06,
+      "loss": 0.6989,
+      "step": 33728
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.158991575241089,
+      "learning_rate": 4.448103838949013e-06,
+      "loss": 0.7596,
+      "step": 33729
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 3.163292646408081,
+      "learning_rate": 4.44784179261026e-06,
+      "loss": 0.6033,
+      "step": 33730
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 3.658562660217285,
+      "learning_rate": 4.447579747806853e-06,
+      "loss": 0.611,
+      "step": 33731
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.7109246253967285,
+      "learning_rate": 4.447317704539521e-06,
+      "loss": 0.659,
+      "step": 33732
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.3181350231170654,
+      "learning_rate": 4.44705566280899e-06,
+      "loss": 0.7286,
+      "step": 33733
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 3.5834896564483643,
+      "learning_rate": 4.44679362261599e-06,
+      "loss": 0.7859,
+      "step": 33734
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.8942056894302368,
+      "learning_rate": 4.44653158396125e-06,
+      "loss": 0.4471,
+      "step": 33735
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.3667092323303223,
+      "learning_rate": 4.446269546845498e-06,
+      "loss": 0.699,
+      "step": 33736
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.4921793937683105,
+      "learning_rate": 4.446007511269461e-06,
+      "loss": 0.6481,
+      "step": 33737
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.2124269008636475,
+      "learning_rate": 4.445745477233872e-06,
+      "loss": 0.8348,
+      "step": 33738
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.239884853363037,
+      "learning_rate": 4.445483444739457e-06,
+      "loss": 0.5688,
+      "step": 33739
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.374065399169922,
+      "learning_rate": 4.445221413786945e-06,
+      "loss": 0.3615,
+      "step": 33740
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.8851646184921265,
+      "learning_rate": 4.444959384377064e-06,
+      "loss": 0.4093,
+      "step": 33741
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.2774763107299805,
+      "learning_rate": 4.444697356510543e-06,
+      "loss": 0.7085,
+      "step": 33742
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.191556453704834,
+      "learning_rate": 4.444435330188112e-06,
+      "loss": 0.6763,
+      "step": 33743
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.9581031799316406,
+      "learning_rate": 4.4441733054104965e-06,
+      "loss": 0.5888,
+      "step": 33744
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.6903769969940186,
+      "learning_rate": 4.443911282178428e-06,
+      "loss": 0.589,
+      "step": 33745
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.3004164695739746,
+      "learning_rate": 4.443649260492633e-06,
+      "loss": 0.5819,
+      "step": 33746
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.2057244777679443,
+      "learning_rate": 4.443387240353841e-06,
+      "loss": 0.633,
+      "step": 33747
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 3.1187973022460938,
+      "learning_rate": 4.443125221762782e-06,
+      "loss": 0.6629,
+      "step": 33748
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 3.0737996101379395,
+      "learning_rate": 4.4428632047201815e-06,
+      "loss": 0.472,
+      "step": 33749
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.286511182785034,
+      "learning_rate": 4.442601189226771e-06,
+      "loss": 0.4304,
+      "step": 33750
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 3.919055223464966,
+      "learning_rate": 4.442339175283277e-06,
+      "loss": 0.4924,
+      "step": 33751
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.6897720098495483,
+      "learning_rate": 4.442077162890428e-06,
+      "loss": 0.4826,
+      "step": 33752
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 3.184495210647583,
+      "learning_rate": 4.441815152048954e-06,
+      "loss": 0.4148,
+      "step": 33753
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 3.397730827331543,
+      "learning_rate": 4.441553142759583e-06,
+      "loss": 0.7193,
+      "step": 33754
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 3.0423903465270996,
+      "learning_rate": 4.441291135023044e-06,
+      "loss": 0.7818,
+      "step": 33755
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.810755491256714,
+      "learning_rate": 4.441029128840065e-06,
+      "loss": 0.5329,
+      "step": 33756
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.859928011894226,
+      "learning_rate": 4.440767124211373e-06,
+      "loss": 0.6713,
+      "step": 33757
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.324794054031372,
+      "learning_rate": 4.440505121137701e-06,
+      "loss": 0.7168,
+      "step": 33758
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.9831881523132324,
+      "learning_rate": 4.440243119619773e-06,
+      "loss": 0.6938,
+      "step": 33759
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 3.0066378116607666,
+      "learning_rate": 4.439981119658319e-06,
+      "loss": 0.5057,
+      "step": 33760
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.9325107336044312,
+      "learning_rate": 4.439719121254068e-06,
+      "loss": 0.53,
+      "step": 33761
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.9008913040161133,
+      "learning_rate": 4.439457124407749e-06,
+      "loss": 0.7399,
+      "step": 33762
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 3.360503911972046,
+      "learning_rate": 4.439195129120088e-06,
+      "loss": 0.8448,
+      "step": 33763
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.3105926513671875,
+      "learning_rate": 4.4389331353918175e-06,
+      "loss": 0.6845,
+      "step": 33764
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.7846627235412598,
+      "learning_rate": 4.438671143223661e-06,
+      "loss": 0.7526,
+      "step": 33765
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.753400206565857,
+      "learning_rate": 4.438409152616352e-06,
+      "loss": 0.7652,
+      "step": 33766
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 3.110630989074707,
+      "learning_rate": 4.4381471635706164e-06,
+      "loss": 0.6959,
+      "step": 33767
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 3.303194761276245,
+      "learning_rate": 4.4378851760871825e-06,
+      "loss": 0.6722,
+      "step": 33768
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.0690877437591553,
+      "learning_rate": 4.43762319016678e-06,
+      "loss": 0.5979,
+      "step": 33769
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.059349775314331,
+      "learning_rate": 4.437361205810137e-06,
+      "loss": 0.6337,
+      "step": 33770
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.0515124797821045,
+      "learning_rate": 4.437099223017981e-06,
+      "loss": 0.6263,
+      "step": 33771
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.8985353708267212,
+      "learning_rate": 4.436837241791042e-06,
+      "loss": 0.6259,
+      "step": 33772
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.9289933443069458,
+      "learning_rate": 4.436575262130047e-06,
+      "loss": 0.6292,
+      "step": 33773
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.272825002670288,
+      "learning_rate": 4.436313284035725e-06,
+      "loss": 0.5824,
+      "step": 33774
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 4.560248374938965,
+      "learning_rate": 4.4360513075088045e-06,
+      "loss": 0.7156,
+      "step": 33775
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.0588698387145996,
+      "learning_rate": 4.435789332550015e-06,
+      "loss": 0.4866,
+      "step": 33776
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.7193591594696045,
+      "learning_rate": 4.435527359160081e-06,
+      "loss": 0.7568,
+      "step": 33777
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 3.530080556869507,
+      "learning_rate": 4.435265387339736e-06,
+      "loss": 0.7405,
+      "step": 33778
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 4.4807658195495605,
+      "learning_rate": 4.4350034170897075e-06,
+      "loss": 0.8315,
+      "step": 33779
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.7899950742721558,
+      "learning_rate": 4.434741448410722e-06,
+      "loss": 0.7807,
+      "step": 33780
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.3493409156799316,
+      "learning_rate": 4.434479481303509e-06,
+      "loss": 0.7646,
+      "step": 33781
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.932134747505188,
+      "learning_rate": 4.434217515768796e-06,
+      "loss": 0.6565,
+      "step": 33782
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.1512222290039062,
+      "learning_rate": 4.433955551807313e-06,
+      "loss": 0.843,
+      "step": 33783
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 3.28798770904541,
+      "learning_rate": 4.433693589419787e-06,
+      "loss": 0.6835,
+      "step": 33784
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 3.1699881553649902,
+      "learning_rate": 4.433431628606947e-06,
+      "loss": 0.709,
+      "step": 33785
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.5390243530273438,
+      "learning_rate": 4.433169669369522e-06,
+      "loss": 0.5446,
+      "step": 33786
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.0839264392852783,
+      "learning_rate": 4.43290771170824e-06,
+      "loss": 0.5782,
+      "step": 33787
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.961310386657715,
+      "learning_rate": 4.4326457556238285e-06,
+      "loss": 0.3052,
+      "step": 33788
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.4923412799835205,
+      "learning_rate": 4.432383801117017e-06,
+      "loss": 0.5196,
+      "step": 33789
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.7643539905548096,
+      "learning_rate": 4.432121848188533e-06,
+      "loss": 0.5673,
+      "step": 33790
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.283116340637207,
+      "learning_rate": 4.431859896839106e-06,
+      "loss": 0.6654,
+      "step": 33791
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.355750799179077,
+      "learning_rate": 4.431597947069464e-06,
+      "loss": 0.7016,
+      "step": 33792
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.8004562854766846,
+      "learning_rate": 4.431335998880334e-06,
+      "loss": 0.4836,
+      "step": 33793
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.8973908424377441,
+      "learning_rate": 4.431074052272447e-06,
+      "loss": 0.6123,
+      "step": 33794
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 3.580904960632324,
+      "learning_rate": 4.430812107246529e-06,
+      "loss": 0.6138,
+      "step": 33795
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 19.808744430541992,
+      "learning_rate": 4.430550163803309e-06,
+      "loss": 0.715,
+      "step": 33796
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.3128926753997803,
+      "learning_rate": 4.430288221943516e-06,
+      "loss": 0.6498,
+      "step": 33797
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 3.5828704833984375,
+      "learning_rate": 4.430026281667876e-06,
+      "loss": 0.6912,
+      "step": 33798
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.8104106187820435,
+      "learning_rate": 4.429764342977122e-06,
+      "loss": 0.4703,
+      "step": 33799
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.527101755142212,
+      "learning_rate": 4.42950240587198e-06,
+      "loss": 0.6231,
+      "step": 33800
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.946864366531372,
+      "learning_rate": 4.4292404703531775e-06,
+      "loss": 0.8968,
+      "step": 33801
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.300598621368408,
+      "learning_rate": 4.428978536421443e-06,
+      "loss": 0.6161,
+      "step": 33802
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 3.119398593902588,
+      "learning_rate": 4.428716604077506e-06,
+      "loss": 0.801,
+      "step": 33803
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.4981763362884521,
+      "learning_rate": 4.428454673322094e-06,
+      "loss": 0.5452,
+      "step": 33804
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 3.3480377197265625,
+      "learning_rate": 4.428192744155935e-06,
+      "loss": 0.7867,
+      "step": 33805
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.3137199878692627,
+      "learning_rate": 4.427930816579758e-06,
+      "loss": 0.5579,
+      "step": 33806
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.163139820098877,
+      "learning_rate": 4.4276688905942914e-06,
+      "loss": 0.7566,
+      "step": 33807
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 5.324731349945068,
+      "learning_rate": 4.4274069662002625e-06,
+      "loss": 0.7309,
+      "step": 33808
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.31709623336792,
+      "learning_rate": 4.427145043398401e-06,
+      "loss": 0.7801,
+      "step": 33809
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.5023510456085205,
+      "learning_rate": 4.426883122189434e-06,
+      "loss": 0.3783,
+      "step": 33810
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.566011667251587,
+      "learning_rate": 4.426621202574091e-06,
+      "loss": 0.7067,
+      "step": 33811
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.0849294662475586,
+      "learning_rate": 4.4263592845531e-06,
+      "loss": 0.7562,
+      "step": 33812
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 5.420380115509033,
+      "learning_rate": 4.4260973681271876e-06,
+      "loss": 0.619,
+      "step": 33813
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.4498258829116821,
+      "learning_rate": 4.425835453297084e-06,
+      "loss": 0.4934,
+      "step": 33814
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 3.2139551639556885,
+      "learning_rate": 4.425573540063516e-06,
+      "loss": 0.8019,
+      "step": 33815
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 3.217271566390991,
+      "learning_rate": 4.425311628427214e-06,
+      "loss": 0.6301,
+      "step": 33816
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 3.045621871948242,
+      "learning_rate": 4.425049718388904e-06,
+      "loss": 0.5908,
+      "step": 33817
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.889512062072754,
+      "learning_rate": 4.424787809949316e-06,
+      "loss": 0.4677,
+      "step": 33818
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.0086519718170166,
+      "learning_rate": 4.424525903109178e-06,
+      "loss": 0.5932,
+      "step": 33819
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 3.914785385131836,
+      "learning_rate": 4.4242639978692175e-06,
+      "loss": 0.5476,
+      "step": 33820
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.973433017730713,
+      "learning_rate": 4.424002094230163e-06,
+      "loss": 0.7216,
+      "step": 33821
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.4936625957489014,
+      "learning_rate": 4.423740192192744e-06,
+      "loss": 0.4755,
+      "step": 33822
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.4201982021331787,
+      "learning_rate": 4.4234782917576864e-06,
+      "loss": 0.4582,
+      "step": 33823
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.7015856504440308,
+      "learning_rate": 4.42321639292572e-06,
+      "loss": 0.7888,
+      "step": 33824
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.9162706136703491,
+      "learning_rate": 4.422954495697574e-06,
+      "loss": 0.4357,
+      "step": 33825
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.4655404090881348,
+      "learning_rate": 4.422692600073974e-06,
+      "loss": 0.6505,
+      "step": 33826
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.066795825958252,
+      "learning_rate": 4.422430706055649e-06,
+      "loss": 0.7248,
+      "step": 33827
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.573413133621216,
+      "learning_rate": 4.422168813643328e-06,
+      "loss": 0.6257,
+      "step": 33828
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.1611974239349365,
+      "learning_rate": 4.4219069228377404e-06,
+      "loss": 0.668,
+      "step": 33829
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.616374135017395,
+      "learning_rate": 4.421645033639613e-06,
+      "loss": 0.4357,
+      "step": 33830
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.5640575885772705,
+      "learning_rate": 4.421383146049673e-06,
+      "loss": 0.6578,
+      "step": 33831
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 3.2456414699554443,
+      "learning_rate": 4.421121260068651e-06,
+      "loss": 0.9357,
+      "step": 33832
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.2313246726989746,
+      "learning_rate": 4.4208593756972725e-06,
+      "loss": 0.522,
+      "step": 33833
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 4.339475154876709,
+      "learning_rate": 4.420597492936268e-06,
+      "loss": 0.7031,
+      "step": 33834
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.174941301345825,
+      "learning_rate": 4.420335611786365e-06,
+      "loss": 0.8336,
+      "step": 33835
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.7297301292419434,
+      "learning_rate": 4.420073732248292e-06,
+      "loss": 0.4202,
+      "step": 33836
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.6230733394622803,
+      "learning_rate": 4.419811854322775e-06,
+      "loss": 0.7737,
+      "step": 33837
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.4427244663238525,
+      "learning_rate": 4.4195499780105435e-06,
+      "loss": 0.6883,
+      "step": 33838
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 5.619607448577881,
+      "learning_rate": 4.419288103312327e-06,
+      "loss": 0.7541,
+      "step": 33839
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.290395736694336,
+      "learning_rate": 4.419026230228853e-06,
+      "loss": 0.6655,
+      "step": 33840
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.80440092086792,
+      "learning_rate": 4.4187643587608495e-06,
+      "loss": 0.5641,
+      "step": 33841
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.6344997882843018,
+      "learning_rate": 4.418502488909045e-06,
+      "loss": 0.6444,
+      "step": 33842
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 5.038094520568848,
+      "learning_rate": 4.418240620674168e-06,
+      "loss": 0.571,
+      "step": 33843
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 3.8425285816192627,
+      "learning_rate": 4.417978754056944e-06,
+      "loss": 0.7235,
+      "step": 33844
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.4575538635253906,
+      "learning_rate": 4.417716889058104e-06,
+      "loss": 0.6007,
+      "step": 33845
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.1202006340026855,
+      "learning_rate": 4.4174550256783746e-06,
+      "loss": 0.6131,
+      "step": 33846
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.5870790481567383,
+      "learning_rate": 4.417193163918486e-06,
+      "loss": 0.8239,
+      "step": 33847
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.325489044189453,
+      "learning_rate": 4.416931303779163e-06,
+      "loss": 0.7627,
+      "step": 33848
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.6563448905944824,
+      "learning_rate": 4.416669445261137e-06,
+      "loss": 0.5797,
+      "step": 33849
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.711824655532837,
+      "learning_rate": 4.4164075883651345e-06,
+      "loss": 0.4878,
+      "step": 33850
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 3.3409311771392822,
+      "learning_rate": 4.416145733091883e-06,
+      "loss": 0.733,
+      "step": 33851
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.3660154342651367,
+      "learning_rate": 4.415883879442113e-06,
+      "loss": 0.7294,
+      "step": 33852
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.489875555038452,
+      "learning_rate": 4.41562202741655e-06,
+      "loss": 0.6695,
+      "step": 33853
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.1278493404388428,
+      "learning_rate": 4.415360177015924e-06,
+      "loss": 0.4716,
+      "step": 33854
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.5905441045761108,
+      "learning_rate": 4.4150983282409616e-06,
+      "loss": 0.5186,
+      "step": 33855
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.247952699661255,
+      "learning_rate": 4.414836481092392e-06,
+      "loss": 0.5422,
+      "step": 33856
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.8022501468658447,
+      "learning_rate": 4.414574635570943e-06,
+      "loss": 0.6962,
+      "step": 33857
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.7365179061889648,
+      "learning_rate": 4.414312791677342e-06,
+      "loss": 0.7579,
+      "step": 33858
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 3.314270496368408,
+      "learning_rate": 4.414050949412317e-06,
+      "loss": 0.554,
+      "step": 33859
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.424372434616089,
+      "learning_rate": 4.413789108776597e-06,
+      "loss": 0.8502,
+      "step": 33860
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 4.340287685394287,
+      "learning_rate": 4.413527269770911e-06,
+      "loss": 0.5304,
+      "step": 33861
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.0919859409332275,
+      "learning_rate": 4.413265432395986e-06,
+      "loss": 0.7227,
+      "step": 33862
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.0274360179901123,
+      "learning_rate": 4.41300359665255e-06,
+      "loss": 0.5056,
+      "step": 33863
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.6276814937591553,
+      "learning_rate": 4.412741762541331e-06,
+      "loss": 0.6252,
+      "step": 33864
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 3.075438976287842,
+      "learning_rate": 4.412479930063057e-06,
+      "loss": 0.8823,
+      "step": 33865
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.1635489463806152,
+      "learning_rate": 4.412218099218457e-06,
+      "loss": 0.4732,
+      "step": 33866
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.5398869514465332,
+      "learning_rate": 4.411956270008257e-06,
+      "loss": 0.5105,
+      "step": 33867
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.020927667617798,
+      "learning_rate": 4.411694442433187e-06,
+      "loss": 0.4891,
+      "step": 33868
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.0388989448547363,
+      "learning_rate": 4.411432616493975e-06,
+      "loss": 0.7243,
+      "step": 33869
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.793508291244507,
+      "learning_rate": 4.4111707921913474e-06,
+      "loss": 0.76,
+      "step": 33870
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 4.134313106536865,
+      "learning_rate": 4.410908969526034e-06,
+      "loss": 0.7182,
+      "step": 33871
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.4615213871002197,
+      "learning_rate": 4.410647148498761e-06,
+      "loss": 0.537,
+      "step": 33872
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.869684100151062,
+      "learning_rate": 4.410385329110258e-06,
+      "loss": 0.4009,
+      "step": 33873
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.8194358348846436,
+      "learning_rate": 4.410123511361253e-06,
+      "loss": 0.5989,
+      "step": 33874
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 3.9489049911499023,
+      "learning_rate": 4.409861695252474e-06,
+      "loss": 0.5755,
+      "step": 33875
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 4.134536266326904,
+      "learning_rate": 4.409599880784648e-06,
+      "loss": 0.8753,
+      "step": 33876
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 3.748704671859741,
+      "learning_rate": 4.409338067958502e-06,
+      "loss": 0.5678,
+      "step": 33877
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 3.073237895965576,
+      "learning_rate": 4.409076256774767e-06,
+      "loss": 0.5112,
+      "step": 33878
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.7703062295913696,
+      "learning_rate": 4.408814447234168e-06,
+      "loss": 0.764,
+      "step": 33879
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.0476415157318115,
+      "learning_rate": 4.408552639337436e-06,
+      "loss": 0.4213,
+      "step": 33880
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.5211753845214844,
+      "learning_rate": 4.408290833085298e-06,
+      "loss": 0.4195,
+      "step": 33881
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.7001535892486572,
+      "learning_rate": 4.408029028478481e-06,
+      "loss": 0.5539,
+      "step": 33882
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.9403977394104004,
+      "learning_rate": 4.407767225517714e-06,
+      "loss": 0.4157,
+      "step": 33883
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.8846628665924072,
+      "learning_rate": 4.407505424203724e-06,
+      "loss": 0.7752,
+      "step": 33884
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.4039578437805176,
+      "learning_rate": 4.40724362453724e-06,
+      "loss": 0.6999,
+      "step": 33885
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.68056058883667,
+      "learning_rate": 4.406981826518988e-06,
+      "loss": 0.6225,
+      "step": 33886
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.221829652786255,
+      "learning_rate": 4.406720030149699e-06,
+      "loss": 0.6405,
+      "step": 33887
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.956620693206787,
+      "learning_rate": 4.406458235430097e-06,
+      "loss": 0.6156,
+      "step": 33888
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.0880632400512695,
+      "learning_rate": 4.406196442360915e-06,
+      "loss": 0.6033,
+      "step": 33889
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.4613473415374756,
+      "learning_rate": 4.405934650942878e-06,
+      "loss": 0.5278,
+      "step": 33890
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.4169013500213623,
+      "learning_rate": 4.405672861176713e-06,
+      "loss": 0.6601,
+      "step": 33891
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 4.161378383636475,
+      "learning_rate": 4.40541107306315e-06,
+      "loss": 0.6933,
+      "step": 33892
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.083475112915039,
+      "learning_rate": 4.405149286602915e-06,
+      "loss": 0.8372,
+      "step": 33893
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.394585609436035,
+      "learning_rate": 4.404887501796738e-06,
+      "loss": 0.7756,
+      "step": 33894
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.797584295272827,
+      "learning_rate": 4.4046257186453456e-06,
+      "loss": 0.8731,
+      "step": 33895
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 3.2486865520477295,
+      "learning_rate": 4.404363937149466e-06,
+      "loss": 0.6456,
+      "step": 33896
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.1685757637023926,
+      "learning_rate": 4.404102157309828e-06,
+      "loss": 0.6834,
+      "step": 33897
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.653092384338379,
+      "learning_rate": 4.4038403791271576e-06,
+      "loss": 0.5895,
+      "step": 33898
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.9585254192352295,
+      "learning_rate": 4.4035786026021826e-06,
+      "loss": 0.5205,
+      "step": 33899
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.0460100173950195,
+      "learning_rate": 4.403316827735633e-06,
+      "loss": 0.2274,
+      "step": 33900
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 3.3098716735839844,
+      "learning_rate": 4.403055054528236e-06,
+      "loss": 0.5159,
+      "step": 33901
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.4400899410247803,
+      "learning_rate": 4.40279328298072e-06,
+      "loss": 0.7584,
+      "step": 33902
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.4009897708892822,
+      "learning_rate": 4.4025315130938115e-06,
+      "loss": 0.6688,
+      "step": 33903
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 3.4001686573028564,
+      "learning_rate": 4.40226974486824e-06,
+      "loss": 0.7957,
+      "step": 33904
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 4.35502815246582,
+      "learning_rate": 4.402007978304731e-06,
+      "loss": 0.7246,
+      "step": 33905
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.0799708366394043,
+      "learning_rate": 4.401746213404015e-06,
+      "loss": 0.9413,
+      "step": 33906
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.4511592388153076,
+      "learning_rate": 4.401484450166818e-06,
+      "loss": 0.6715,
+      "step": 33907
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 3.3764822483062744,
+      "learning_rate": 4.401222688593869e-06,
+      "loss": 0.723,
+      "step": 33908
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 3.607788562774658,
+      "learning_rate": 4.400960928685895e-06,
+      "loss": 0.5642,
+      "step": 33909
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.9098823070526123,
+      "learning_rate": 4.400699170443624e-06,
+      "loss": 0.5874,
+      "step": 33910
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 4.690861701965332,
+      "learning_rate": 4.400437413867786e-06,
+      "loss": 0.7462,
+      "step": 33911
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 3.1917271614074707,
+      "learning_rate": 4.400175658959105e-06,
+      "loss": 0.6443,
+      "step": 33912
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.7084741592407227,
+      "learning_rate": 4.399913905718311e-06,
+      "loss": 0.8641,
+      "step": 33913
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 3.0539839267730713,
+      "learning_rate": 4.399652154146132e-06,
+      "loss": 0.5484,
+      "step": 33914
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.9275740385055542,
+      "learning_rate": 4.399390404243296e-06,
+      "loss": 0.6842,
+      "step": 33915
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.1396589279174805,
+      "learning_rate": 4.399128656010529e-06,
+      "loss": 0.7736,
+      "step": 33916
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 3.11236572265625,
+      "learning_rate": 4.398866909448561e-06,
+      "loss": 0.5652,
+      "step": 33917
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.1118481159210205,
+      "learning_rate": 4.398605164558119e-06,
+      "loss": 0.6325,
+      "step": 33918
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 5.4952569007873535,
+      "learning_rate": 4.39834342133993e-06,
+      "loss": 0.7931,
+      "step": 33919
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.886673927307129,
+      "learning_rate": 4.398081679794721e-06,
+      "loss": 0.7421,
+      "step": 33920
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 3.8155100345611572,
+      "learning_rate": 4.397819939923224e-06,
+      "loss": 0.442,
+      "step": 33921
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.9620180130004883,
+      "learning_rate": 4.397558201726163e-06,
+      "loss": 0.4263,
+      "step": 33922
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 3.302304744720459,
+      "learning_rate": 4.3972964652042675e-06,
+      "loss": 0.7139,
+      "step": 33923
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.436984062194824,
+      "learning_rate": 4.3970347303582644e-06,
+      "loss": 0.4834,
+      "step": 33924
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.8210649490356445,
+      "learning_rate": 4.396772997188883e-06,
+      "loss": 0.7095,
+      "step": 33925
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 4.070226669311523,
+      "learning_rate": 4.396511265696849e-06,
+      "loss": 0.7443,
+      "step": 33926
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.8584797382354736,
+      "learning_rate": 4.396249535882891e-06,
+      "loss": 0.582,
+      "step": 33927
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.3810391426086426,
+      "learning_rate": 4.395987807747737e-06,
+      "loss": 0.5609,
+      "step": 33928
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.9095158576965332,
+      "learning_rate": 4.395726081292115e-06,
+      "loss": 0.7054,
+      "step": 33929
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.336724042892456,
+      "learning_rate": 4.395464356516752e-06,
+      "loss": 0.4982,
+      "step": 33930
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.5894298553466797,
+      "learning_rate": 4.395202633422376e-06,
+      "loss": 0.4479,
+      "step": 33931
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.2802324295043945,
+      "learning_rate": 4.394940912009716e-06,
+      "loss": 0.6157,
+      "step": 33932
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.8778868913650513,
+      "learning_rate": 4.3946791922794975e-06,
+      "loss": 0.7587,
+      "step": 33933
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.1926462650299072,
+      "learning_rate": 4.39441747423245e-06,
+      "loss": 0.6293,
+      "step": 33934
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.670686960220337,
+      "learning_rate": 4.3941557578693e-06,
+      "loss": 0.5644,
+      "step": 33935
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.0193591117858887,
+      "learning_rate": 4.3938940431907765e-06,
+      "loss": 0.8115,
+      "step": 33936
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 5.536797046661377,
+      "learning_rate": 4.393632330197607e-06,
+      "loss": 0.5077,
+      "step": 33937
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.124230146408081,
+      "learning_rate": 4.393370618890517e-06,
+      "loss": 0.6689,
+      "step": 33938
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 3.504690170288086,
+      "learning_rate": 4.393108909270237e-06,
+      "loss": 0.5778,
+      "step": 33939
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.6956301927566528,
+      "learning_rate": 4.392847201337493e-06,
+      "loss": 0.5964,
+      "step": 33940
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 4.054904460906982,
+      "learning_rate": 4.392585495093015e-06,
+      "loss": 0.6896,
+      "step": 33941
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 3.205770254135132,
+      "learning_rate": 4.392323790537528e-06,
+      "loss": 0.8746,
+      "step": 33942
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.797365427017212,
+      "learning_rate": 4.392062087671762e-06,
+      "loss": 0.6915,
+      "step": 33943
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.343275308609009,
+      "learning_rate": 4.391800386496443e-06,
+      "loss": 0.5674,
+      "step": 33944
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.3498423099517822,
+      "learning_rate": 4.3915386870123e-06,
+      "loss": 0.5071,
+      "step": 33945
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.074734926223755,
+      "learning_rate": 4.391276989220059e-06,
+      "loss": 0.6548,
+      "step": 33946
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.5034929513931274,
+      "learning_rate": 4.39101529312045e-06,
+      "loss": 0.4885,
+      "step": 33947
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.1557624340057373,
+      "learning_rate": 4.390753598714198e-06,
+      "loss": 0.6532,
+      "step": 33948
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.0708136558532715,
+      "learning_rate": 4.390491906002033e-06,
+      "loss": 0.6296,
+      "step": 33949
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.8997037410736084,
+      "learning_rate": 4.39023021498468e-06,
+      "loss": 0.5333,
+      "step": 33950
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.092703342437744,
+      "learning_rate": 4.38996852566287e-06,
+      "loss": 0.8417,
+      "step": 33951
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 3.4618687629699707,
+      "learning_rate": 4.389706838037328e-06,
+      "loss": 0.6907,
+      "step": 33952
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.8495378494262695,
+      "learning_rate": 4.389445152108783e-06,
+      "loss": 0.677,
+      "step": 33953
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 3.8729631900787354,
+      "learning_rate": 4.389183467877963e-06,
+      "loss": 0.9041,
+      "step": 33954
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.542494773864746,
+      "learning_rate": 4.388921785345595e-06,
+      "loss": 0.7035,
+      "step": 33955
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.079517364501953,
+      "learning_rate": 4.388660104512406e-06,
+      "loss": 0.8182,
+      "step": 33956
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.9719951152801514,
+      "learning_rate": 4.388398425379124e-06,
+      "loss": 0.2169,
+      "step": 33957
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.383173704147339,
+      "learning_rate": 4.388136747946478e-06,
+      "loss": 0.5049,
+      "step": 33958
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.713001012802124,
+      "learning_rate": 4.387875072215193e-06,
+      "loss": 0.4926,
+      "step": 33959
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 3.1096456050872803,
+      "learning_rate": 4.387613398185998e-06,
+      "loss": 0.6894,
+      "step": 33960
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 3.09517765045166,
+      "learning_rate": 4.387351725859621e-06,
+      "loss": 0.5434,
+      "step": 33961
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.61456561088562,
+      "learning_rate": 4.38709005523679e-06,
+      "loss": 0.6504,
+      "step": 33962
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.181889772415161,
+      "learning_rate": 4.3868283863182315e-06,
+      "loss": 0.7875,
+      "step": 33963
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.3824362754821777,
+      "learning_rate": 4.386566719104674e-06,
+      "loss": 0.6783,
+      "step": 33964
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.9450491666793823,
+      "learning_rate": 4.386305053596844e-06,
+      "loss": 0.5423,
+      "step": 33965
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 3.313962697982788,
+      "learning_rate": 4.38604338979547e-06,
+      "loss": 0.8184,
+      "step": 33966
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.9868590831756592,
+      "learning_rate": 4.3857817277012805e-06,
+      "loss": 0.5257,
+      "step": 33967
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.3741722106933594,
+      "learning_rate": 4.385520067315001e-06,
+      "loss": 0.6963,
+      "step": 33968
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 4.272433757781982,
+      "learning_rate": 4.38525840863736e-06,
+      "loss": 0.8402,
+      "step": 33969
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 4.998876571655273,
+      "learning_rate": 4.384996751669086e-06,
+      "loss": 0.7613,
+      "step": 33970
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.716735601425171,
+      "learning_rate": 4.384735096410904e-06,
+      "loss": 0.3553,
+      "step": 33971
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 4.226287841796875,
+      "learning_rate": 4.384473442863544e-06,
+      "loss": 0.5368,
+      "step": 33972
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.303760051727295,
+      "learning_rate": 4.384211791027733e-06,
+      "loss": 0.4586,
+      "step": 33973
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 5.302651882171631,
+      "learning_rate": 4.383950140904198e-06,
+      "loss": 0.5464,
+      "step": 33974
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.841212272644043,
+      "learning_rate": 4.383688492493667e-06,
+      "loss": 0.7536,
+      "step": 33975
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.5541300773620605,
+      "learning_rate": 4.383426845796868e-06,
+      "loss": 0.5545,
+      "step": 33976
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.5931706428527832,
+      "learning_rate": 4.383165200814528e-06,
+      "loss": 0.5389,
+      "step": 33977
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.48327374458313,
+      "learning_rate": 4.382903557547373e-06,
+      "loss": 0.5156,
+      "step": 33978
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.4483646154403687,
+      "learning_rate": 4.3826419159961335e-06,
+      "loss": 0.4351,
+      "step": 33979
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.0623345375061035,
+      "learning_rate": 4.382380276161535e-06,
+      "loss": 0.4775,
+      "step": 33980
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 3.498671770095825,
+      "learning_rate": 4.382118638044303e-06,
+      "loss": 0.7799,
+      "step": 33981
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.26102876663208,
+      "learning_rate": 4.38185700164517e-06,
+      "loss": 0.6203,
+      "step": 33982
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 4.4242072105407715,
+      "learning_rate": 4.381595366964862e-06,
+      "loss": 0.7178,
+      "step": 33983
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.8714467287063599,
+      "learning_rate": 4.381333734004105e-06,
+      "loss": 0.6844,
+      "step": 33984
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 4.084875106811523,
+      "learning_rate": 4.381072102763627e-06,
+      "loss": 0.6927,
+      "step": 33985
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 3.100029706954956,
+      "learning_rate": 4.380810473244156e-06,
+      "loss": 0.9204,
+      "step": 33986
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.2391936779022217,
+      "learning_rate": 4.380548845446418e-06,
+      "loss": 0.7333,
+      "step": 33987
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.202554225921631,
+      "learning_rate": 4.380287219371143e-06,
+      "loss": 0.6182,
+      "step": 33988
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.982535481452942,
+      "learning_rate": 4.380025595019057e-06,
+      "loss": 0.6625,
+      "step": 33989
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.464235544204712,
+      "learning_rate": 4.379763972390887e-06,
+      "loss": 0.7004,
+      "step": 33990
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.3837523460388184,
+      "learning_rate": 4.3795023514873615e-06,
+      "loss": 0.4253,
+      "step": 33991
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.021831750869751,
+      "learning_rate": 4.379240732309207e-06,
+      "loss": 0.5325,
+      "step": 33992
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.5399608612060547,
+      "learning_rate": 4.378979114857152e-06,
+      "loss": 0.595,
+      "step": 33993
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.504007339477539,
+      "learning_rate": 4.378717499131922e-06,
+      "loss": 0.4733,
+      "step": 33994
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 3.5274250507354736,
+      "learning_rate": 4.3784558851342476e-06,
+      "loss": 0.7125,
+      "step": 33995
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.854970932006836,
+      "learning_rate": 4.378194272864854e-06,
+      "loss": 0.4057,
+      "step": 33996
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.9311459064483643,
+      "learning_rate": 4.377932662324469e-06,
+      "loss": 0.9233,
+      "step": 33997
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.1216742992401123,
+      "learning_rate": 4.377671053513821e-06,
+      "loss": 0.5408,
+      "step": 33998
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.8575479984283447,
+      "learning_rate": 4.3774094464336355e-06,
+      "loss": 0.7254,
+      "step": 33999
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.6079115867614746,
+      "learning_rate": 4.377147841084641e-06,
+      "loss": 0.5599,
+      "step": 34000
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 4.0692877769470215,
+      "learning_rate": 4.376886237467564e-06,
+      "loss": 0.6339,
+      "step": 34001
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.9054484367370605,
+      "learning_rate": 4.376624635583135e-06,
+      "loss": 0.6755,
+      "step": 34002
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.761394500732422,
+      "learning_rate": 4.376363035432079e-06,
+      "loss": 0.69,
+      "step": 34003
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.9632922410964966,
+      "learning_rate": 4.376101437015124e-06,
+      "loss": 0.6643,
+      "step": 34004
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.9468234777450562,
+      "learning_rate": 4.375839840332997e-06,
+      "loss": 0.4466,
+      "step": 34005
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.7771518230438232,
+      "learning_rate": 4.375578245386426e-06,
+      "loss": 0.534,
+      "step": 34006
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.626032829284668,
+      "learning_rate": 4.375316652176137e-06,
+      "loss": 0.6705,
+      "step": 34007
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.8106701374053955,
+      "learning_rate": 4.3750550607028585e-06,
+      "loss": 0.7308,
+      "step": 34008
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.9582382440567017,
+      "learning_rate": 4.374793470967319e-06,
+      "loss": 0.6332,
+      "step": 34009
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 3.1262190341949463,
+      "learning_rate": 4.374531882970243e-06,
+      "loss": 0.5918,
+      "step": 34010
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.285642147064209,
+      "learning_rate": 4.374270296712361e-06,
+      "loss": 0.6538,
+      "step": 34011
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.9555693864822388,
+      "learning_rate": 4.374008712194397e-06,
+      "loss": 0.5592,
+      "step": 34012
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 4.983397006988525,
+      "learning_rate": 4.373747129417081e-06,
+      "loss": 0.7211,
+      "step": 34013
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.6263248920440674,
+      "learning_rate": 4.3734855483811405e-06,
+      "loss": 0.43,
+      "step": 34014
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.254124164581299,
+      "learning_rate": 4.373223969087302e-06,
+      "loss": 0.8092,
+      "step": 34015
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.531156539916992,
+      "learning_rate": 4.372962391536292e-06,
+      "loss": 0.6375,
+      "step": 34016
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.5476362705230713,
+      "learning_rate": 4.372700815728839e-06,
+      "loss": 0.6061,
+      "step": 34017
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.9112894535064697,
+      "learning_rate": 4.372439241665671e-06,
+      "loss": 0.5917,
+      "step": 34018
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.9608314037322998,
+      "learning_rate": 4.3721776693475135e-06,
+      "loss": 0.5036,
+      "step": 34019
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.585505962371826,
+      "learning_rate": 4.371916098775095e-06,
+      "loss": 0.6605,
+      "step": 34020
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.832924723625183,
+      "learning_rate": 4.371654529949141e-06,
+      "loss": 0.7576,
+      "step": 34021
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 3.7777624130249023,
+      "learning_rate": 4.371392962870381e-06,
+      "loss": 0.7962,
+      "step": 34022
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.9042553901672363,
+      "learning_rate": 4.371131397539542e-06,
+      "loss": 0.7105,
+      "step": 34023
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.1214818954467773,
+      "learning_rate": 4.3708698339573515e-06,
+      "loss": 0.6686,
+      "step": 34024
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 3.015184164047241,
+      "learning_rate": 4.370608272124536e-06,
+      "loss": 0.8895,
+      "step": 34025
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 3.7858898639678955,
+      "learning_rate": 4.3703467120418235e-06,
+      "loss": 0.383,
+      "step": 34026
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.9321436882019043,
+      "learning_rate": 4.370085153709941e-06,
+      "loss": 0.5779,
+      "step": 34027
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.1020963191986084,
+      "learning_rate": 4.3698235971296155e-06,
+      "loss": 0.4241,
+      "step": 34028
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 6.450773239135742,
+      "learning_rate": 4.369562042301574e-06,
+      "loss": 0.7911,
+      "step": 34029
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 3.433199167251587,
+      "learning_rate": 4.369300489226546e-06,
+      "loss": 0.2956,
+      "step": 34030
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.0184295177459717,
+      "learning_rate": 4.3690389379052556e-06,
+      "loss": 0.8722,
+      "step": 34031
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.8813767433166504,
+      "learning_rate": 4.368777388338431e-06,
+      "loss": 1.0016,
+      "step": 34032
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.1663427352905273,
+      "learning_rate": 4.368515840526802e-06,
+      "loss": 0.5409,
+      "step": 34033
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.289431095123291,
+      "learning_rate": 4.368254294471092e-06,
+      "loss": 0.5016,
+      "step": 34034
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.6986093521118164,
+      "learning_rate": 4.367992750172031e-06,
+      "loss": 0.5399,
+      "step": 34035
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.701244592666626,
+      "learning_rate": 4.367731207630346e-06,
+      "loss": 0.7371,
+      "step": 34036
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 3.3194994926452637,
+      "learning_rate": 4.367469666846763e-06,
+      "loss": 0.6111,
+      "step": 34037
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.2624077796936035,
+      "learning_rate": 4.367208127822009e-06,
+      "loss": 0.6633,
+      "step": 34038
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.4767067432403564,
+      "learning_rate": 4.3669465905568135e-06,
+      "loss": 0.8798,
+      "step": 34039
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 3.0087316036224365,
+      "learning_rate": 4.366685055051902e-06,
+      "loss": 0.482,
+      "step": 34040
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.829789638519287,
+      "learning_rate": 4.366423521308002e-06,
+      "loss": 0.5979,
+      "step": 34041
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.1379470825195312,
+      "learning_rate": 4.366161989325838e-06,
+      "loss": 0.6153,
+      "step": 34042
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 15.161561965942383,
+      "learning_rate": 4.3659004591061435e-06,
+      "loss": 0.8198,
+      "step": 34043
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.189234495162964,
+      "learning_rate": 4.365638930649642e-06,
+      "loss": 0.5696,
+      "step": 34044
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.221395254135132,
+      "learning_rate": 4.36537740395706e-06,
+      "loss": 0.5193,
+      "step": 34045
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 3.4481866359710693,
+      "learning_rate": 4.365115879029127e-06,
+      "loss": 0.666,
+      "step": 34046
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.7566255331039429,
+      "learning_rate": 4.364854355866568e-06,
+      "loss": 0.5429,
+      "step": 34047
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.370537281036377,
+      "learning_rate": 4.3645928344701115e-06,
+      "loss": 0.661,
+      "step": 34048
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.190352439880371,
+      "learning_rate": 4.364331314840484e-06,
+      "loss": 0.4888,
+      "step": 34049
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 3.3517813682556152,
+      "learning_rate": 4.364069796978413e-06,
+      "loss": 0.6015,
+      "step": 34050
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.6374033689498901,
+      "learning_rate": 4.363808280884626e-06,
+      "loss": 0.4532,
+      "step": 34051
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.8277603387832642,
+      "learning_rate": 4.36354676655985e-06,
+      "loss": 0.5031,
+      "step": 34052
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.1570310592651367,
+      "learning_rate": 4.363285254004812e-06,
+      "loss": 0.6666,
+      "step": 34053
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.6368328332901,
+      "learning_rate": 4.3630237432202386e-06,
+      "loss": 0.5872,
+      "step": 34054
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.0583178997039795,
+      "learning_rate": 4.362762234206858e-06,
+      "loss": 0.6436,
+      "step": 34055
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.455948829650879,
+      "learning_rate": 4.362500726965397e-06,
+      "loss": 0.303,
+      "step": 34056
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 3.5880138874053955,
+      "learning_rate": 4.362239221496583e-06,
+      "loss": 0.4761,
+      "step": 34057
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.3206326961517334,
+      "learning_rate": 4.361977717801141e-06,
+      "loss": 0.8142,
+      "step": 34058
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.8899431228637695,
+      "learning_rate": 4.361716215879802e-06,
+      "loss": 0.7715,
+      "step": 34059
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.5747721195220947,
+      "learning_rate": 4.361454715733291e-06,
+      "loss": 0.337,
+      "step": 34060
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.2120745182037354,
+      "learning_rate": 4.361193217362334e-06,
+      "loss": 0.5989,
+      "step": 34061
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.2419371604919434,
+      "learning_rate": 4.360931720767657e-06,
+      "loss": 0.7015,
+      "step": 34062
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 7.962041854858398,
+      "learning_rate": 4.360670225949993e-06,
+      "loss": 0.5442,
+      "step": 34063
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.5474872589111328,
+      "learning_rate": 4.3604087329100655e-06,
+      "loss": 0.6516,
+      "step": 34064
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.5650837421417236,
+      "learning_rate": 4.360147241648601e-06,
+      "loss": 0.8597,
+      "step": 34065
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.9209407567977905,
+      "learning_rate": 4.359885752166327e-06,
+      "loss": 0.796,
+      "step": 34066
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 3.3624539375305176,
+      "learning_rate": 4.359624264463972e-06,
+      "loss": 0.755,
+      "step": 34067
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.301953077316284,
+      "learning_rate": 4.359362778542262e-06,
+      "loss": 0.6246,
+      "step": 34068
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 12.024761199951172,
+      "learning_rate": 4.359101294401923e-06,
+      "loss": 0.5305,
+      "step": 34069
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.540623426437378,
+      "learning_rate": 4.358839812043684e-06,
+      "loss": 0.5013,
+      "step": 34070
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.8907381296157837,
+      "learning_rate": 4.3585783314682715e-06,
+      "loss": 0.4712,
+      "step": 34071
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.6216325759887695,
+      "learning_rate": 4.358316852676412e-06,
+      "loss": 0.7332,
+      "step": 34072
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.4038028717041016,
+      "learning_rate": 4.358055375668832e-06,
+      "loss": 0.7579,
+      "step": 34073
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.0270464420318604,
+      "learning_rate": 4.35779390044626e-06,
+      "loss": 0.5345,
+      "step": 34074
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.6894636154174805,
+      "learning_rate": 4.3575324270094234e-06,
+      "loss": 0.5164,
+      "step": 34075
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 5.4275431632995605,
+      "learning_rate": 4.3572709553590475e-06,
+      "loss": 0.8965,
+      "step": 34076
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.9421014785766602,
+      "learning_rate": 4.357009485495861e-06,
+      "loss": 0.4848,
+      "step": 34077
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.3860491514205933,
+      "learning_rate": 4.356748017420591e-06,
+      "loss": 0.5048,
+      "step": 34078
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 3.3264429569244385,
+      "learning_rate": 4.356486551133963e-06,
+      "loss": 0.4958,
+      "step": 34079
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.7832603454589844,
+      "learning_rate": 4.356225086636704e-06,
+      "loss": 0.6932,
+      "step": 34080
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 3.125415563583374,
+      "learning_rate": 4.3559636239295434e-06,
+      "loss": 0.446,
+      "step": 34081
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.7939327955245972,
+      "learning_rate": 4.355702163013204e-06,
+      "loss": 0.6223,
+      "step": 34082
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.2439675331115723,
+      "learning_rate": 4.355440703888418e-06,
+      "loss": 0.5517,
+      "step": 34083
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.804391860961914,
+      "learning_rate": 4.355179246555909e-06,
+      "loss": 0.4579,
+      "step": 34084
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.1237597465515137,
+      "learning_rate": 4.354917791016405e-06,
+      "loss": 0.5904,
+      "step": 34085
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 3.148522138595581,
+      "learning_rate": 4.354656337270633e-06,
+      "loss": 0.7981,
+      "step": 34086
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.7434282302856445,
+      "learning_rate": 4.3543948853193206e-06,
+      "loss": 0.722,
+      "step": 34087
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.782309889793396,
+      "learning_rate": 4.354133435163194e-06,
+      "loss": 0.7496,
+      "step": 34088
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.9506561756134033,
+      "learning_rate": 4.35387198680298e-06,
+      "loss": 0.7902,
+      "step": 34089
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.6175625324249268,
+      "learning_rate": 4.353610540239407e-06,
+      "loss": 0.6184,
+      "step": 34090
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.5039305686950684,
+      "learning_rate": 4.3533490954732e-06,
+      "loss": 0.6943,
+      "step": 34091
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.899259090423584,
+      "learning_rate": 4.353087652505087e-06,
+      "loss": 0.6943,
+      "step": 34092
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.589185118675232,
+      "learning_rate": 4.352826211335796e-06,
+      "loss": 0.4664,
+      "step": 34093
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.9460883140563965,
+      "learning_rate": 4.352564771966051e-06,
+      "loss": 0.6635,
+      "step": 34094
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.082923650741577,
+      "learning_rate": 4.352303334396582e-06,
+      "loss": 0.6898,
+      "step": 34095
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.8362497091293335,
+      "learning_rate": 4.352041898628114e-06,
+      "loss": 0.7469,
+      "step": 34096
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.469055414199829,
+      "learning_rate": 4.351780464661376e-06,
+      "loss": 0.7056,
+      "step": 34097
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.9100879430770874,
+      "learning_rate": 4.3515190324970925e-06,
+      "loss": 0.315,
+      "step": 34098
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.6820905208587646,
+      "learning_rate": 4.351257602135992e-06,
+      "loss": 0.6032,
+      "step": 34099
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.6334455013275146,
+      "learning_rate": 4.350996173578801e-06,
+      "loss": 0.5153,
+      "step": 34100
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.799029588699341,
+      "learning_rate": 4.350734746826247e-06,
+      "loss": 0.8009,
+      "step": 34101
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.2256951332092285,
+      "learning_rate": 4.350473321879056e-06,
+      "loss": 0.461,
+      "step": 34102
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 4.338130950927734,
+      "learning_rate": 4.350211898737954e-06,
+      "loss": 0.5657,
+      "step": 34103
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 4.7673773765563965,
+      "learning_rate": 4.34995047740367e-06,
+      "loss": 0.7029,
+      "step": 34104
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.5585246086120605,
+      "learning_rate": 4.349689057876932e-06,
+      "loss": 0.7157,
+      "step": 34105
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.0084874629974365,
+      "learning_rate": 4.349427640158464e-06,
+      "loss": 0.6128,
+      "step": 34106
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 4.264321804046631,
+      "learning_rate": 4.349166224248994e-06,
+      "loss": 0.7414,
+      "step": 34107
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.9945945739746094,
+      "learning_rate": 4.348904810149249e-06,
+      "loss": 0.8199,
+      "step": 34108
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.8926670551300049,
+      "learning_rate": 4.348643397859956e-06,
+      "loss": 0.6514,
+      "step": 34109
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.687756299972534,
+      "learning_rate": 4.348381987381842e-06,
+      "loss": 0.6993,
+      "step": 34110
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.415484666824341,
+      "learning_rate": 4.348120578715633e-06,
+      "loss": 0.6119,
+      "step": 34111
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.064601421356201,
+      "learning_rate": 4.347859171862057e-06,
+      "loss": 0.6747,
+      "step": 34112
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.422072172164917,
+      "learning_rate": 4.34759776682184e-06,
+      "loss": 0.7202,
+      "step": 34113
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 4.794036388397217,
+      "learning_rate": 4.3473363635957095e-06,
+      "loss": 0.9535,
+      "step": 34114
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.05461049079895,
+      "learning_rate": 4.347074962184391e-06,
+      "loss": 0.8721,
+      "step": 34115
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.2772960662841797,
+      "learning_rate": 4.3468135625886145e-06,
+      "loss": 0.6296,
+      "step": 34116
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.311675548553467,
+      "learning_rate": 4.346552164809103e-06,
+      "loss": 0.4072,
+      "step": 34117
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.6247174739837646,
+      "learning_rate": 4.346290768846586e-06,
+      "loss": 0.5025,
+      "step": 34118
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.471003293991089,
+      "learning_rate": 4.346029374701789e-06,
+      "loss": 0.8195,
+      "step": 34119
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.464763879776001,
+      "learning_rate": 4.34576798237544e-06,
+      "loss": 0.4792,
+      "step": 34120
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.0721755027770996,
+      "learning_rate": 4.345506591868265e-06,
+      "loss": 0.4936,
+      "step": 34121
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.8332726955413818,
+      "learning_rate": 4.34524520318099e-06,
+      "loss": 0.5611,
+      "step": 34122
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.6247763633728027,
+      "learning_rate": 4.344983816314341e-06,
+      "loss": 0.8656,
+      "step": 34123
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.788457155227661,
+      "learning_rate": 4.34472243126905e-06,
+      "loss": 0.6719,
+      "step": 34124
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.303542137145996,
+      "learning_rate": 4.3444610480458395e-06,
+      "loss": 0.6207,
+      "step": 34125
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.9988577365875244,
+      "learning_rate": 4.344199666645437e-06,
+      "loss": 0.8912,
+      "step": 34126
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.8041062355041504,
+      "learning_rate": 4.3439382870685695e-06,
+      "loss": 0.5671,
+      "step": 34127
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.7525089979171753,
+      "learning_rate": 4.343676909315965e-06,
+      "loss": 0.7596,
+      "step": 34128
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.214855432510376,
+      "learning_rate": 4.343415533388348e-06,
+      "loss": 0.5347,
+      "step": 34129
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.792379379272461,
+      "learning_rate": 4.343154159286447e-06,
+      "loss": 0.4673,
+      "step": 34130
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.0353236198425293,
+      "learning_rate": 4.342892787010988e-06,
+      "loss": 0.6524,
+      "step": 34131
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.7873802185058594,
+      "learning_rate": 4.3426314165626986e-06,
+      "loss": 0.5103,
+      "step": 34132
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.9894163608551025,
+      "learning_rate": 4.342370047942303e-06,
+      "loss": 0.4179,
+      "step": 34133
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.556776762008667,
+      "learning_rate": 4.342108681150533e-06,
+      "loss": 0.6464,
+      "step": 34134
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.229031562805176,
+      "learning_rate": 4.341847316188109e-06,
+      "loss": 0.6431,
+      "step": 34135
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.379055976867676,
+      "learning_rate": 4.341585953055762e-06,
+      "loss": 0.7234,
+      "step": 34136
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 5.470171928405762,
+      "learning_rate": 4.341324591754219e-06,
+      "loss": 0.392,
+      "step": 34137
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.4107351303100586,
+      "learning_rate": 4.341063232284205e-06,
+      "loss": 0.6474,
+      "step": 34138
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.9407575130462646,
+      "learning_rate": 4.340801874646448e-06,
+      "loss": 0.3975,
+      "step": 34139
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.7347111701965332,
+      "learning_rate": 4.340540518841673e-06,
+      "loss": 0.4612,
+      "step": 34140
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.8239538669586182,
+      "learning_rate": 4.340279164870609e-06,
+      "loss": 0.6605,
+      "step": 34141
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 5.990835189819336,
+      "learning_rate": 4.34001781273398e-06,
+      "loss": 0.4611,
+      "step": 34142
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.513948678970337,
+      "learning_rate": 4.339756462432515e-06,
+      "loss": 0.5647,
+      "step": 34143
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.2166407108306885,
+      "learning_rate": 4.33949511396694e-06,
+      "loss": 0.6905,
+      "step": 34144
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.4164369106292725,
+      "learning_rate": 4.339233767337981e-06,
+      "loss": 0.5544,
+      "step": 34145
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.087911605834961,
+      "learning_rate": 4.3389724225463655e-06,
+      "loss": 0.6589,
+      "step": 34146
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.606295347213745,
+      "learning_rate": 4.33871107959282e-06,
+      "loss": 0.8535,
+      "step": 34147
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.021289587020874,
+      "learning_rate": 4.338449738478072e-06,
+      "loss": 0.4842,
+      "step": 34148
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.2089943885803223,
+      "learning_rate": 4.338188399202847e-06,
+      "loss": 0.7473,
+      "step": 34149
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.2819273471832275,
+      "learning_rate": 4.337927061767873e-06,
+      "loss": 0.6196,
+      "step": 34150
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.39799165725708,
+      "learning_rate": 4.337665726173875e-06,
+      "loss": 0.7314,
+      "step": 34151
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.176138162612915,
+      "learning_rate": 4.33740439242158e-06,
+      "loss": 0.508,
+      "step": 34152
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.515854835510254,
+      "learning_rate": 4.337143060511716e-06,
+      "loss": 0.5792,
+      "step": 34153
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.0994999408721924,
+      "learning_rate": 4.336881730445009e-06,
+      "loss": 0.6165,
+      "step": 34154
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.497328281402588,
+      "learning_rate": 4.3366204022221854e-06,
+      "loss": 0.7457,
+      "step": 34155
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.023322820663452,
+      "learning_rate": 4.336359075843971e-06,
+      "loss": 0.6288,
+      "step": 34156
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.361590623855591,
+      "learning_rate": 4.336097751311095e-06,
+      "loss": 0.5562,
+      "step": 34157
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.421144962310791,
+      "learning_rate": 4.3358364286242816e-06,
+      "loss": 0.5603,
+      "step": 34158
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.4695630073547363,
+      "learning_rate": 4.335575107784258e-06,
+      "loss": 0.5216,
+      "step": 34159
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 5.034235000610352,
+      "learning_rate": 4.335313788791752e-06,
+      "loss": 0.5998,
+      "step": 34160
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.9682155847549438,
+      "learning_rate": 4.335052471647488e-06,
+      "loss": 0.5755,
+      "step": 34161
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 4.317229747772217,
+      "learning_rate": 4.334791156352195e-06,
+      "loss": 0.7906,
+      "step": 34162
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.0880048274993896,
+      "learning_rate": 4.334529842906598e-06,
+      "loss": 0.7318,
+      "step": 34163
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.580091953277588,
+      "learning_rate": 4.334268531311423e-06,
+      "loss": 0.6068,
+      "step": 34164
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.7202441692352295,
+      "learning_rate": 4.3340072215674e-06,
+      "loss": 0.5197,
+      "step": 34165
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.8128330707550049,
+      "learning_rate": 4.333745913675253e-06,
+      "loss": 0.5762,
+      "step": 34166
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.491466522216797,
+      "learning_rate": 4.333484607635708e-06,
+      "loss": 0.8634,
+      "step": 34167
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.4419867992401123,
+      "learning_rate": 4.333223303449494e-06,
+      "loss": 0.5166,
+      "step": 34168
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 4.5686540603637695,
+      "learning_rate": 4.332962001117336e-06,
+      "loss": 0.6359,
+      "step": 34169
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.9323785305023193,
+      "learning_rate": 4.33270070063996e-06,
+      "loss": 0.6117,
+      "step": 34170
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.8948122262954712,
+      "learning_rate": 4.332439402018094e-06,
+      "loss": 0.502,
+      "step": 34171
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.7929060459136963,
+      "learning_rate": 4.3321781052524636e-06,
+      "loss": 0.6892,
+      "step": 34172
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.727797269821167,
+      "learning_rate": 4.331916810343796e-06,
+      "loss": 0.3898,
+      "step": 34173
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.8096719980239868,
+      "learning_rate": 4.3316555172928175e-06,
+      "loss": 0.7126,
+      "step": 34174
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.8832156658172607,
+      "learning_rate": 4.3313942261002546e-06,
+      "loss": 0.2624,
+      "step": 34175
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.7804712057113647,
+      "learning_rate": 4.331132936766834e-06,
+      "loss": 0.3871,
+      "step": 34176
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.8238413333892822,
+      "learning_rate": 4.330871649293282e-06,
+      "loss": 0.5526,
+      "step": 34177
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.845707893371582,
+      "learning_rate": 4.330610363680325e-06,
+      "loss": 0.4775,
+      "step": 34178
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.4366838932037354,
+      "learning_rate": 4.330349079928691e-06,
+      "loss": 0.326,
+      "step": 34179
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.9362993240356445,
+      "learning_rate": 4.3300877980391035e-06,
+      "loss": 0.5868,
+      "step": 34180
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.8903000354766846,
+      "learning_rate": 4.3298265180122924e-06,
+      "loss": 0.5302,
+      "step": 34181
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.411684989929199,
+      "learning_rate": 4.329565239848982e-06,
+      "loss": 0.3979,
+      "step": 34182
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.641306757926941,
+      "learning_rate": 4.329303963549899e-06,
+      "loss": 0.3468,
+      "step": 34183
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.5662691593170166,
+      "learning_rate": 4.32904268911577e-06,
+      "loss": 0.7344,
+      "step": 34184
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.2544662952423096,
+      "learning_rate": 4.3287814165473225e-06,
+      "loss": 0.4559,
+      "step": 34185
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 4.163387298583984,
+      "learning_rate": 4.328520145845283e-06,
+      "loss": 0.7293,
+      "step": 34186
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 8.416402816772461,
+      "learning_rate": 4.328258877010377e-06,
+      "loss": 0.5862,
+      "step": 34187
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.463258743286133,
+      "learning_rate": 4.327997610043332e-06,
+      "loss": 0.6971,
+      "step": 34188
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.7057971954345703,
+      "learning_rate": 4.327736344944874e-06,
+      "loss": 0.7987,
+      "step": 34189
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.434063196182251,
+      "learning_rate": 4.327475081715729e-06,
+      "loss": 0.7092,
+      "step": 34190
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.1771388053894043,
+      "learning_rate": 4.327213820356624e-06,
+      "loss": 0.5799,
+      "step": 34191
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.4328830242156982,
+      "learning_rate": 4.326952560868285e-06,
+      "loss": 0.7865,
+      "step": 34192
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.941875696182251,
+      "learning_rate": 4.32669130325144e-06,
+      "loss": 0.66,
+      "step": 34193
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.8930364847183228,
+      "learning_rate": 4.326430047506813e-06,
+      "loss": 0.5864,
+      "step": 34194
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.4127185344696045,
+      "learning_rate": 4.326168793635132e-06,
+      "loss": 0.7067,
+      "step": 34195
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.7486120462417603,
+      "learning_rate": 4.3259075416371235e-06,
+      "loss": 0.4937,
+      "step": 34196
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.084433078765869,
+      "learning_rate": 4.325646291513512e-06,
+      "loss": 0.811,
+      "step": 34197
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.9394476413726807,
+      "learning_rate": 4.325385043265027e-06,
+      "loss": 0.4298,
+      "step": 34198
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.6773629188537598,
+      "learning_rate": 4.325123796892393e-06,
+      "loss": 0.4772,
+      "step": 34199
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.448936700820923,
+      "learning_rate": 4.324862552396338e-06,
+      "loss": 0.6607,
+      "step": 34200
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.8709337711334229,
+      "learning_rate": 4.324601309777586e-06,
+      "loss": 0.4106,
+      "step": 34201
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.3436341285705566,
+      "learning_rate": 4.324340069036866e-06,
+      "loss": 0.5037,
+      "step": 34202
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.0439159870147705,
+      "learning_rate": 4.3240788301749025e-06,
+      "loss": 0.5635,
+      "step": 34203
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.035111904144287,
+      "learning_rate": 4.323817593192421e-06,
+      "loss": 0.4968,
+      "step": 34204
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.6258009672164917,
+      "learning_rate": 4.323556358090151e-06,
+      "loss": 0.6564,
+      "step": 34205
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.9820512533187866,
+      "learning_rate": 4.323295124868817e-06,
+      "loss": 0.6889,
+      "step": 34206
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.4945693016052246,
+      "learning_rate": 4.323033893529147e-06,
+      "loss": 0.6143,
+      "step": 34207
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.7862162590026855,
+      "learning_rate": 4.322772664071864e-06,
+      "loss": 0.7521,
+      "step": 34208
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.8083276748657227,
+      "learning_rate": 4.322511436497698e-06,
+      "loss": 0.6437,
+      "step": 34209
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.359651565551758,
+      "learning_rate": 4.322250210807374e-06,
+      "loss": 0.5273,
+      "step": 34210
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.1142241954803467,
+      "learning_rate": 4.321988987001618e-06,
+      "loss": 0.551,
+      "step": 34211
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.155280351638794,
+      "learning_rate": 4.321727765081156e-06,
+      "loss": 0.9661,
+      "step": 34212
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.6962950229644775,
+      "learning_rate": 4.3214665450467165e-06,
+      "loss": 0.6802,
+      "step": 34213
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.049561023712158,
+      "learning_rate": 4.321205326899023e-06,
+      "loss": 0.6131,
+      "step": 34214
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.393983840942383,
+      "learning_rate": 4.320944110638804e-06,
+      "loss": 0.6419,
+      "step": 34215
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 4.4920148849487305,
+      "learning_rate": 4.320682896266785e-06,
+      "loss": 0.6311,
+      "step": 34216
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.398224115371704,
+      "learning_rate": 4.320421683783692e-06,
+      "loss": 0.5649,
+      "step": 34217
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.4081283807754517,
+      "learning_rate": 4.320160473190252e-06,
+      "loss": 0.5153,
+      "step": 34218
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.9215081930160522,
+      "learning_rate": 4.31989926448719e-06,
+      "loss": 0.7494,
+      "step": 34219
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.2103569507598877,
+      "learning_rate": 4.319638057675235e-06,
+      "loss": 0.7498,
+      "step": 34220
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.8543716669082642,
+      "learning_rate": 4.319376852755111e-06,
+      "loss": 0.7263,
+      "step": 34221
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.469141721725464,
+      "learning_rate": 4.319115649727545e-06,
+      "loss": 0.6977,
+      "step": 34222
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.9739066362380981,
+      "learning_rate": 4.318854448593263e-06,
+      "loss": 0.758,
+      "step": 34223
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.6286180019378662,
+      "learning_rate": 4.3185932493529925e-06,
+      "loss": 0.8637,
+      "step": 34224
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.2089285850524902,
+      "learning_rate": 4.318332052007456e-06,
+      "loss": 0.5912,
+      "step": 34225
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 4.048410415649414,
+      "learning_rate": 4.3180708565573845e-06,
+      "loss": 0.7241,
+      "step": 34226
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 4.042751789093018,
+      "learning_rate": 4.317809663003504e-06,
+      "loss": 0.7263,
+      "step": 34227
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.8444081544876099,
+      "learning_rate": 4.317548471346538e-06,
+      "loss": 0.6271,
+      "step": 34228
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.5595698356628418,
+      "learning_rate": 4.317287281587214e-06,
+      "loss": 0.4021,
+      "step": 34229
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.760946750640869,
+      "learning_rate": 4.317026093726259e-06,
+      "loss": 0.4938,
+      "step": 34230
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.4749693870544434,
+      "learning_rate": 4.316764907764398e-06,
+      "loss": 0.4744,
+      "step": 34231
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.9036188125610352,
+      "learning_rate": 4.316503723702358e-06,
+      "loss": 0.7405,
+      "step": 34232
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.885775089263916,
+      "learning_rate": 4.316242541540866e-06,
+      "loss": 0.6067,
+      "step": 34233
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.067394495010376,
+      "learning_rate": 4.315981361280646e-06,
+      "loss": 0.8884,
+      "step": 34234
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.1255829334259033,
+      "learning_rate": 4.315720182922426e-06,
+      "loss": 0.5573,
+      "step": 34235
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.2596161365509033,
+      "learning_rate": 4.315459006466932e-06,
+      "loss": 0.4804,
+      "step": 34236
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.1057631969451904,
+      "learning_rate": 4.3151978319148905e-06,
+      "loss": 0.6765,
+      "step": 34237
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.9912959337234497,
+      "learning_rate": 4.314936659267026e-06,
+      "loss": 0.5889,
+      "step": 34238
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.0650641918182373,
+      "learning_rate": 4.314675488524067e-06,
+      "loss": 0.5455,
+      "step": 34239
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.8945456743240356,
+      "learning_rate": 4.314414319686739e-06,
+      "loss": 0.5258,
+      "step": 34240
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.0872421264648438,
+      "learning_rate": 4.314153152755768e-06,
+      "loss": 0.798,
+      "step": 34241
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.1758387088775635,
+      "learning_rate": 4.31389198773188e-06,
+      "loss": 0.8057,
+      "step": 34242
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.87186598777771,
+      "learning_rate": 4.313630824615801e-06,
+      "loss": 0.5552,
+      "step": 34243
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.617998480796814,
+      "learning_rate": 4.313369663408257e-06,
+      "loss": 0.5723,
+      "step": 34244
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.211357831954956,
+      "learning_rate": 4.3131085041099734e-06,
+      "loss": 0.7357,
+      "step": 34245
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.925384521484375,
+      "learning_rate": 4.31284734672168e-06,
+      "loss": 0.8756,
+      "step": 34246
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.2824487686157227,
+      "learning_rate": 4.3125861912441e-06,
+      "loss": 0.4722,
+      "step": 34247
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.5665311813354492,
+      "learning_rate": 4.312325037677961e-06,
+      "loss": 0.7389,
+      "step": 34248
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.058783769607544,
+      "learning_rate": 4.312063886023988e-06,
+      "loss": 0.6682,
+      "step": 34249
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.274574041366577,
+      "learning_rate": 4.311802736282908e-06,
+      "loss": 0.5,
+      "step": 34250
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.2399017810821533,
+      "learning_rate": 4.311541588455447e-06,
+      "loss": 0.7078,
+      "step": 34251
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 4.529086112976074,
+      "learning_rate": 4.31128044254233e-06,
+      "loss": 0.6416,
+      "step": 34252
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.718855381011963,
+      "learning_rate": 4.311019298544285e-06,
+      "loss": 0.5162,
+      "step": 34253
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.676799774169922,
+      "learning_rate": 4.310758156462037e-06,
+      "loss": 0.4252,
+      "step": 34254
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.6076369285583496,
+      "learning_rate": 4.310497016296312e-06,
+      "loss": 0.4251,
+      "step": 34255
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.5506861209869385,
+      "learning_rate": 4.310235878047837e-06,
+      "loss": 0.7135,
+      "step": 34256
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.4977567195892334,
+      "learning_rate": 4.309974741717337e-06,
+      "loss": 0.6561,
+      "step": 34257
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.5285584926605225,
+      "learning_rate": 4.309713607305539e-06,
+      "loss": 0.5409,
+      "step": 34258
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.7329115867614746,
+      "learning_rate": 4.309452474813168e-06,
+      "loss": 0.8983,
+      "step": 34259
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.688239336013794,
+      "learning_rate": 4.309191344240952e-06,
+      "loss": 0.4978,
+      "step": 34260
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.7304110527038574,
+      "learning_rate": 4.308930215589617e-06,
+      "loss": 0.5963,
+      "step": 34261
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.7712490558624268,
+      "learning_rate": 4.308669088859886e-06,
+      "loss": 0.6517,
+      "step": 34262
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.4712016582489014,
+      "learning_rate": 4.308407964052489e-06,
+      "loss": 0.6747,
+      "step": 34263
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.891584873199463,
+      "learning_rate": 4.30814684116815e-06,
+      "loss": 0.6089,
+      "step": 34264
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.184927463531494,
+      "learning_rate": 4.307885720207594e-06,
+      "loss": 0.4998,
+      "step": 34265
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.217649459838867,
+      "learning_rate": 4.30762460117155e-06,
+      "loss": 0.7517,
+      "step": 34266
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.0107228755950928,
+      "learning_rate": 4.307363484060742e-06,
+      "loss": 0.5993,
+      "step": 34267
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 4.256267070770264,
+      "learning_rate": 4.307102368875898e-06,
+      "loss": 0.7203,
+      "step": 34268
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 4.258719444274902,
+      "learning_rate": 4.306841255617742e-06,
+      "loss": 0.7047,
+      "step": 34269
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.63027286529541,
+      "learning_rate": 4.306580144287e-06,
+      "loss": 0.5851,
+      "step": 34270
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.309710741043091,
+      "learning_rate": 4.306319034884399e-06,
+      "loss": 0.6071,
+      "step": 34271
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.062532424926758,
+      "learning_rate": 4.306057927410666e-06,
+      "loss": 0.5868,
+      "step": 34272
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 6.433218955993652,
+      "learning_rate": 4.3057968218665254e-06,
+      "loss": 0.634,
+      "step": 34273
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.9755918979644775,
+      "learning_rate": 4.305535718252704e-06,
+      "loss": 0.4744,
+      "step": 34274
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.723435878753662,
+      "learning_rate": 4.305274616569927e-06,
+      "loss": 0.5906,
+      "step": 34275
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.9218297004699707,
+      "learning_rate": 4.305013516818921e-06,
+      "loss": 0.6353,
+      "step": 34276
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.2574095726013184,
+      "learning_rate": 4.304752419000413e-06,
+      "loss": 0.507,
+      "step": 34277
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.943308115005493,
+      "learning_rate": 4.304491323115128e-06,
+      "loss": 0.5437,
+      "step": 34278
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.2511990070343018,
+      "learning_rate": 4.304230229163791e-06,
+      "loss": 0.5452,
+      "step": 34279
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.734098196029663,
+      "learning_rate": 4.3039691371471304e-06,
+      "loss": 0.7607,
+      "step": 34280
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.8511998653411865,
+      "learning_rate": 4.30370804706587e-06,
+      "loss": 0.6916,
+      "step": 34281
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.8513081073760986,
+      "learning_rate": 4.303446958920737e-06,
+      "loss": 0.5684,
+      "step": 34282
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.512638807296753,
+      "learning_rate": 4.303185872712457e-06,
+      "loss": 0.6055,
+      "step": 34283
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.417515516281128,
+      "learning_rate": 4.302924788441757e-06,
+      "loss": 0.8759,
+      "step": 34284
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.2378976345062256,
+      "learning_rate": 4.302663706109361e-06,
+      "loss": 0.7227,
+      "step": 34285
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.581979513168335,
+      "learning_rate": 4.302402625715994e-06,
+      "loss": 0.6797,
+      "step": 34286
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.7779804468154907,
+      "learning_rate": 4.302141547262386e-06,
+      "loss": 0.3735,
+      "step": 34287
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.6765024662017822,
+      "learning_rate": 4.3018804707492616e-06,
+      "loss": 0.8439,
+      "step": 34288
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.4163124561309814,
+      "learning_rate": 4.301619396177346e-06,
+      "loss": 0.5278,
+      "step": 34289
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.2163825035095215,
+      "learning_rate": 4.301358323547364e-06,
+      "loss": 0.526,
+      "step": 34290
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.9492738246917725,
+      "learning_rate": 4.301097252860043e-06,
+      "loss": 0.6858,
+      "step": 34291
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.701740264892578,
+      "learning_rate": 4.30083618411611e-06,
+      "loss": 0.8233,
+      "step": 34292
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.0865390300750732,
+      "learning_rate": 4.300575117316289e-06,
+      "loss": 0.3725,
+      "step": 34293
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.2587153911590576,
+      "learning_rate": 4.300314052461306e-06,
+      "loss": 0.6014,
+      "step": 34294
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.1347334384918213,
+      "learning_rate": 4.300052989551889e-06,
+      "loss": 0.7832,
+      "step": 34295
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.6147820949554443,
+      "learning_rate": 4.299791928588761e-06,
+      "loss": 0.5731,
+      "step": 34296
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.1443068981170654,
+      "learning_rate": 4.29953086957265e-06,
+      "loss": 0.6623,
+      "step": 34297
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.556018829345703,
+      "learning_rate": 4.299269812504281e-06,
+      "loss": 0.8253,
+      "step": 34298
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.300105571746826,
+      "learning_rate": 4.29900875738438e-06,
+      "loss": 0.6613,
+      "step": 34299
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.217717170715332,
+      "learning_rate": 4.298747704213674e-06,
+      "loss": 0.7818,
+      "step": 34300
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.4012389183044434,
+      "learning_rate": 4.298486652992887e-06,
+      "loss": 0.729,
+      "step": 34301
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.288956642150879,
+      "learning_rate": 4.298225603722746e-06,
+      "loss": 0.5232,
+      "step": 34302
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.8973886966705322,
+      "learning_rate": 4.297964556403977e-06,
+      "loss": 0.6945,
+      "step": 34303
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.929922103881836,
+      "learning_rate": 4.297703511037306e-06,
+      "loss": 0.6353,
+      "step": 34304
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.75320565700531,
+      "learning_rate": 4.2974424676234575e-06,
+      "loss": 0.3994,
+      "step": 34305
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.084195852279663,
+      "learning_rate": 4.297181426163156e-06,
+      "loss": 0.6478,
+      "step": 34306
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.8564016819000244,
+      "learning_rate": 4.296920386657133e-06,
+      "loss": 0.5679,
+      "step": 34307
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.67905855178833,
+      "learning_rate": 4.296659349106109e-06,
+      "loss": 0.6537,
+      "step": 34308
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.420823335647583,
+      "learning_rate": 4.2963983135108136e-06,
+      "loss": 0.8007,
+      "step": 34309
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.1263933181762695,
+      "learning_rate": 4.296137279871971e-06,
+      "loss": 0.4957,
+      "step": 34310
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.004473924636841,
+      "learning_rate": 4.295876248190306e-06,
+      "loss": 0.6017,
+      "step": 34311
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.036661148071289,
+      "learning_rate": 4.2956152184665465e-06,
+      "loss": 0.4983,
+      "step": 34312
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.924647331237793,
+      "learning_rate": 4.295354190701417e-06,
+      "loss": 0.4885,
+      "step": 34313
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.885749340057373,
+      "learning_rate": 4.295093164895642e-06,
+      "loss": 0.5211,
+      "step": 34314
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.8610457181930542,
+      "learning_rate": 4.29483214104995e-06,
+      "loss": 0.7017,
+      "step": 34315
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.625268578529358,
+      "learning_rate": 4.294571119165066e-06,
+      "loss": 0.5861,
+      "step": 34316
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.2397077083587646,
+      "learning_rate": 4.294310099241714e-06,
+      "loss": 0.4145,
+      "step": 34317
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 5.268494129180908,
+      "learning_rate": 4.294049081280623e-06,
+      "loss": 0.5221,
+      "step": 34318
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.876510739326477,
+      "learning_rate": 4.2937880652825165e-06,
+      "loss": 0.7262,
+      "step": 34319
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.6152147054672241,
+      "learning_rate": 4.29352705124812e-06,
+      "loss": 0.3624,
+      "step": 34320
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 4.10030460357666,
+      "learning_rate": 4.29326603917816e-06,
+      "loss": 0.712,
+      "step": 34321
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.3577699661254883,
+      "learning_rate": 4.293005029073363e-06,
+      "loss": 0.3366,
+      "step": 34322
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.613448143005371,
+      "learning_rate": 4.292744020934455e-06,
+      "loss": 0.3612,
+      "step": 34323
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.3093345165252686,
+      "learning_rate": 4.29248301476216e-06,
+      "loss": 0.7109,
+      "step": 34324
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.6891778707504272,
+      "learning_rate": 4.292222010557205e-06,
+      "loss": 0.5722,
+      "step": 34325
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.911561965942383,
+      "learning_rate": 4.2919610083203146e-06,
+      "loss": 0.6789,
+      "step": 34326
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.3144023418426514,
+      "learning_rate": 4.291700008052216e-06,
+      "loss": 0.5041,
+      "step": 34327
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.2435784339904785,
+      "learning_rate": 4.2914390097536345e-06,
+      "loss": 0.4624,
+      "step": 34328
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.5962867736816406,
+      "learning_rate": 4.2911780134252955e-06,
+      "loss": 0.6345,
+      "step": 34329
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.0165202617645264,
+      "learning_rate": 4.290917019067926e-06,
+      "loss": 0.7527,
+      "step": 34330
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.7555105686187744,
+      "learning_rate": 4.290656026682249e-06,
+      "loss": 0.5641,
+      "step": 34331
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.462322235107422,
+      "learning_rate": 4.290395036268992e-06,
+      "loss": 0.3668,
+      "step": 34332
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.1311023235321045,
+      "learning_rate": 4.290134047828881e-06,
+      "loss": 0.7017,
+      "step": 34333
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.9084235429763794,
+      "learning_rate": 4.289873061362642e-06,
+      "loss": 0.6646,
+      "step": 34334
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.79227614402771,
+      "learning_rate": 4.2896120768709995e-06,
+      "loss": 0.6265,
+      "step": 34335
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.5142347812652588,
+      "learning_rate": 4.28935109435468e-06,
+      "loss": 0.6742,
+      "step": 34336
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.722207546234131,
+      "learning_rate": 4.289090113814408e-06,
+      "loss": 0.4682,
+      "step": 34337
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.689331531524658,
+      "learning_rate": 4.288829135250911e-06,
+      "loss": 0.7677,
+      "step": 34338
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.2039313316345215,
+      "learning_rate": 4.288568158664914e-06,
+      "loss": 0.7551,
+      "step": 34339
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.6953864097595215,
+      "learning_rate": 4.288307184057141e-06,
+      "loss": 0.4476,
+      "step": 34340
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.140591859817505,
+      "learning_rate": 4.28804621142832e-06,
+      "loss": 0.6414,
+      "step": 34341
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.9143120050430298,
+      "learning_rate": 4.287785240779177e-06,
+      "loss": 0.5067,
+      "step": 34342
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.265237808227539,
+      "learning_rate": 4.2875242721104344e-06,
+      "loss": 0.5857,
+      "step": 34343
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.5856451988220215,
+      "learning_rate": 4.287263305422821e-06,
+      "loss": 0.443,
+      "step": 34344
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.0988571643829346,
+      "learning_rate": 4.287002340717061e-06,
+      "loss": 0.5293,
+      "step": 34345
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.0077693462371826,
+      "learning_rate": 4.28674137799388e-06,
+      "loss": 0.686,
+      "step": 34346
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.9160654544830322,
+      "learning_rate": 4.286480417254003e-06,
+      "loss": 0.6638,
+      "step": 34347
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.5008914470672607,
+      "learning_rate": 4.2862194584981585e-06,
+      "loss": 0.735,
+      "step": 34348
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.9636335372924805,
+      "learning_rate": 4.28595850172707e-06,
+      "loss": 0.5152,
+      "step": 34349
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.183539867401123,
+      "learning_rate": 4.285697546941463e-06,
+      "loss": 0.9229,
+      "step": 34350
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.8705154657363892,
+      "learning_rate": 4.2854365941420636e-06,
+      "loss": 0.4092,
+      "step": 34351
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.5318779945373535,
+      "learning_rate": 4.285175643329598e-06,
+      "loss": 0.6179,
+      "step": 34352
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.1696016788482666,
+      "learning_rate": 4.2849146945047905e-06,
+      "loss": 0.5589,
+      "step": 34353
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.219888210296631,
+      "learning_rate": 4.284653747668368e-06,
+      "loss": 0.5293,
+      "step": 34354
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.133882999420166,
+      "learning_rate": 4.284392802821056e-06,
+      "loss": 0.5763,
+      "step": 34355
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.6420619487762451,
+      "learning_rate": 4.284131859963579e-06,
+      "loss": 0.6734,
+      "step": 34356
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 4.241917133331299,
+      "learning_rate": 4.283870919096662e-06,
+      "loss": 0.5503,
+      "step": 34357
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.8822033405303955,
+      "learning_rate": 4.2836099802210326e-06,
+      "loss": 0.6122,
+      "step": 34358
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.503544569015503,
+      "learning_rate": 4.283349043337415e-06,
+      "loss": 0.4194,
+      "step": 34359
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.4745213985443115,
+      "learning_rate": 4.283088108446536e-06,
+      "loss": 0.5125,
+      "step": 34360
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.274601459503174,
+      "learning_rate": 4.28282717554912e-06,
+      "loss": 0.6575,
+      "step": 34361
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.319772720336914,
+      "learning_rate": 4.282566244645893e-06,
+      "loss": 0.8702,
+      "step": 34362
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.640603542327881,
+      "learning_rate": 4.2823053157375814e-06,
+      "loss": 0.6515,
+      "step": 34363
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.0249316692352295,
+      "learning_rate": 4.282044388824909e-06,
+      "loss": 0.5906,
+      "step": 34364
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.6800849437713623,
+      "learning_rate": 4.281783463908602e-06,
+      "loss": 0.8881,
+      "step": 34365
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.4622843265533447,
+      "learning_rate": 4.281522540989387e-06,
+      "loss": 0.5161,
+      "step": 34366
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.9506475925445557,
+      "learning_rate": 4.2812616200679855e-06,
+      "loss": 0.6065,
+      "step": 34367
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.667980909347534,
+      "learning_rate": 4.281000701145128e-06,
+      "loss": 0.5831,
+      "step": 34368
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.544006586074829,
+      "learning_rate": 4.28073978422154e-06,
+      "loss": 0.7566,
+      "step": 34369
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.894742488861084,
+      "learning_rate": 4.280478869297944e-06,
+      "loss": 0.7608,
+      "step": 34370
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.31683087348938,
+      "learning_rate": 4.280217956375066e-06,
+      "loss": 0.4869,
+      "step": 34371
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.4533958435058594,
+      "learning_rate": 4.279957045453633e-06,
+      "loss": 0.7352,
+      "step": 34372
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.0070784091949463,
+      "learning_rate": 4.2796961365343705e-06,
+      "loss": 0.6856,
+      "step": 34373
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.612194538116455,
+      "learning_rate": 4.279435229618003e-06,
+      "loss": 0.8518,
+      "step": 34374
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.3209071159362793,
+      "learning_rate": 4.279174324705255e-06,
+      "loss": 0.6749,
+      "step": 34375
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.4206273555755615,
+      "learning_rate": 4.278913421796854e-06,
+      "loss": 0.73,
+      "step": 34376
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.561901330947876,
+      "learning_rate": 4.278652520893525e-06,
+      "loss": 0.4987,
+      "step": 34377
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.240922212600708,
+      "learning_rate": 4.278391621995992e-06,
+      "loss": 0.5723,
+      "step": 34378
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.2405166625976562,
+      "learning_rate": 4.278130725104982e-06,
+      "loss": 0.7113,
+      "step": 34379
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.770298957824707,
+      "learning_rate": 4.277869830221221e-06,
+      "loss": 0.5691,
+      "step": 34380
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.7651269435882568,
+      "learning_rate": 4.277608937345433e-06,
+      "loss": 0.4943,
+      "step": 34381
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.8057613372802734,
+      "learning_rate": 4.277348046478342e-06,
+      "loss": 0.4025,
+      "step": 34382
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.5517354011535645,
+      "learning_rate": 4.277087157620678e-06,
+      "loss": 0.3836,
+      "step": 34383
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.567664861679077,
+      "learning_rate": 4.2768262707731624e-06,
+      "loss": 0.4502,
+      "step": 34384
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.8240177631378174,
+      "learning_rate": 4.276565385936523e-06,
+      "loss": 0.5652,
+      "step": 34385
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.806872844696045,
+      "learning_rate": 4.276304503111483e-06,
+      "loss": 0.7516,
+      "step": 34386
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.6018245220184326,
+      "learning_rate": 4.276043622298769e-06,
+      "loss": 0.7181,
+      "step": 34387
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.978466272354126,
+      "learning_rate": 4.275782743499107e-06,
+      "loss": 0.6082,
+      "step": 34388
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.202592372894287,
+      "learning_rate": 4.275521866713222e-06,
+      "loss": 0.6224,
+      "step": 34389
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.7473517656326294,
+      "learning_rate": 4.27526099194184e-06,
+      "loss": 0.5721,
+      "step": 34390
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.11830997467041,
+      "learning_rate": 4.2750001191856856e-06,
+      "loss": 0.589,
+      "step": 34391
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.7666834592819214,
+      "learning_rate": 4.274739248445483e-06,
+      "loss": 0.6064,
+      "step": 34392
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.2964963912963867,
+      "learning_rate": 4.27447837972196e-06,
+      "loss": 0.7388,
+      "step": 34393
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.589002847671509,
+      "learning_rate": 4.2742175130158395e-06,
+      "loss": 0.5086,
+      "step": 34394
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.705674171447754,
+      "learning_rate": 4.2739566483278495e-06,
+      "loss": 0.7945,
+      "step": 34395
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.4297642707824707,
+      "learning_rate": 4.273695785658714e-06,
+      "loss": 0.6615,
+      "step": 34396
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.039046287536621,
+      "learning_rate": 4.273434925009158e-06,
+      "loss": 0.6459,
+      "step": 34397
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.9765048027038574,
+      "learning_rate": 4.2731740663799084e-06,
+      "loss": 0.6729,
+      "step": 34398
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.598818063735962,
+      "learning_rate": 4.272913209771689e-06,
+      "loss": 0.7244,
+      "step": 34399
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.276703357696533,
+      "learning_rate": 4.272652355185225e-06,
+      "loss": 0.9073,
+      "step": 34400
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.364853858947754,
+      "learning_rate": 4.272391502621242e-06,
+      "loss": 0.7572,
+      "step": 34401
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.9281055927276611,
+      "learning_rate": 4.272130652080466e-06,
+      "loss": 0.8658,
+      "step": 34402
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.749065399169922,
+      "learning_rate": 4.271869803563623e-06,
+      "loss": 0.69,
+      "step": 34403
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.4556784629821777,
+      "learning_rate": 4.271608957071436e-06,
+      "loss": 0.4615,
+      "step": 34404
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.3242475986480713,
+      "learning_rate": 4.2713481126046326e-06,
+      "loss": 0.5309,
+      "step": 34405
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.562864303588867,
+      "learning_rate": 4.271087270163937e-06,
+      "loss": 0.527,
+      "step": 34406
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 9.526226043701172,
+      "learning_rate": 4.270826429750074e-06,
+      "loss": 0.6407,
+      "step": 34407
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.313013792037964,
+      "learning_rate": 4.270565591363768e-06,
+      "loss": 0.6823,
+      "step": 34408
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.929368257522583,
+      "learning_rate": 4.270304755005747e-06,
+      "loss": 0.482,
+      "step": 34409
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.809352397918701,
+      "learning_rate": 4.270043920676736e-06,
+      "loss": 0.7506,
+      "step": 34410
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.9268300533294678,
+      "learning_rate": 4.26978308837746e-06,
+      "loss": 0.7844,
+      "step": 34411
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.6786432266235352,
+      "learning_rate": 4.2695222581086426e-06,
+      "loss": 0.489,
+      "step": 34412
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.1899070739746094,
+      "learning_rate": 4.2692614298710104e-06,
+      "loss": 0.6097,
+      "step": 34413
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.068532705307007,
+      "learning_rate": 4.269000603665288e-06,
+      "loss": 0.3443,
+      "step": 34414
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 4.146085262298584,
+      "learning_rate": 4.268739779492202e-06,
+      "loss": 0.6398,
+      "step": 34415
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.4409775733947754,
+      "learning_rate": 4.268478957352476e-06,
+      "loss": 0.5707,
+      "step": 34416
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.397367477416992,
+      "learning_rate": 4.2682181372468365e-06,
+      "loss": 0.6197,
+      "step": 34417
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.7256438732147217,
+      "learning_rate": 4.267957319176009e-06,
+      "loss": 1.0938,
+      "step": 34418
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 4.469523906707764,
+      "learning_rate": 4.267696503140717e-06,
+      "loss": 0.5516,
+      "step": 34419
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.1715760231018066,
+      "learning_rate": 4.267435689141686e-06,
+      "loss": 0.3229,
+      "step": 34420
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.4776506423950195,
+      "learning_rate": 4.267174877179643e-06,
+      "loss": 0.7057,
+      "step": 34421
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.4481990337371826,
+      "learning_rate": 4.266914067255312e-06,
+      "loss": 0.7411,
+      "step": 34422
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.2318856716156006,
+      "learning_rate": 4.266653259369418e-06,
+      "loss": 0.7219,
+      "step": 34423
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.2673771381378174,
+      "learning_rate": 4.266392453522687e-06,
+      "loss": 0.562,
+      "step": 34424
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.9827715158462524,
+      "learning_rate": 4.266131649715843e-06,
+      "loss": 0.4305,
+      "step": 34425
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 5.0396952629089355,
+      "learning_rate": 4.265870847949613e-06,
+      "loss": 0.7076,
+      "step": 34426
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.7124767303466797,
+      "learning_rate": 4.265610048224721e-06,
+      "loss": 0.695,
+      "step": 34427
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.230642557144165,
+      "learning_rate": 4.2653492505418895e-06,
+      "loss": 0.8598,
+      "step": 34428
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.8130549192428589,
+      "learning_rate": 4.2650884549018486e-06,
+      "loss": 0.5336,
+      "step": 34429
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.6178609132766724,
+      "learning_rate": 4.264827661305322e-06,
+      "loss": 0.7003,
+      "step": 34430
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.5039772987365723,
+      "learning_rate": 4.264566869753034e-06,
+      "loss": 0.476,
+      "step": 34431
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 4.660905838012695,
+      "learning_rate": 4.264306080245711e-06,
+      "loss": 0.6461,
+      "step": 34432
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.7376999855041504,
+      "learning_rate": 4.264045292784076e-06,
+      "loss": 0.3163,
+      "step": 34433
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.6509544849395752,
+      "learning_rate": 4.263784507368855e-06,
+      "loss": 0.5513,
+      "step": 34434
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.2563071250915527,
+      "learning_rate": 4.263523724000774e-06,
+      "loss": 0.7311,
+      "step": 34435
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.9962515830993652,
+      "learning_rate": 4.263262942680558e-06,
+      "loss": 0.3117,
+      "step": 34436
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.7975852489471436,
+      "learning_rate": 4.263002163408932e-06,
+      "loss": 0.8274,
+      "step": 34437
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.7636902332305908,
+      "learning_rate": 4.26274138618662e-06,
+      "loss": 0.8298,
+      "step": 34438
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.5167295932769775,
+      "learning_rate": 4.262480611014349e-06,
+      "loss": 0.7238,
+      "step": 34439
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.6271660327911377,
+      "learning_rate": 4.2622198378928425e-06,
+      "loss": 0.5323,
+      "step": 34440
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.0648229122161865,
+      "learning_rate": 4.261959066822826e-06,
+      "loss": 0.5657,
+      "step": 34441
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.1212267875671387,
+      "learning_rate": 4.2616982978050255e-06,
+      "loss": 0.5674,
+      "step": 34442
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.32635760307312,
+      "learning_rate": 4.261437530840165e-06,
+      "loss": 0.6738,
+      "step": 34443
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.370870590209961,
+      "learning_rate": 4.261176765928969e-06,
+      "loss": 0.6361,
+      "step": 34444
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.7450947761535645,
+      "learning_rate": 4.260916003072165e-06,
+      "loss": 0.6241,
+      "step": 34445
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.6050915718078613,
+      "learning_rate": 4.260655242270477e-06,
+      "loss": 0.5453,
+      "step": 34446
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.157041549682617,
+      "learning_rate": 4.260394483524629e-06,
+      "loss": 0.5592,
+      "step": 34447
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.360330104827881,
+      "learning_rate": 4.260133726835346e-06,
+      "loss": 0.7014,
+      "step": 34448
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.6412088871002197,
+      "learning_rate": 4.259872972203356e-06,
+      "loss": 0.6348,
+      "step": 34449
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 4.337399482727051,
+      "learning_rate": 4.259612219629381e-06,
+      "loss": 0.7095,
+      "step": 34450
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.6256847381591797,
+      "learning_rate": 4.259351469114147e-06,
+      "loss": 0.493,
+      "step": 34451
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 6.51164436340332,
+      "learning_rate": 4.25909072065838e-06,
+      "loss": 0.7495,
+      "step": 34452
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.1028895378112793,
+      "learning_rate": 4.258829974262803e-06,
+      "loss": 0.6804,
+      "step": 34453
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.399479627609253,
+      "learning_rate": 4.258569229928143e-06,
+      "loss": 0.6018,
+      "step": 34454
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.7911661863327026,
+      "learning_rate": 4.258308487655123e-06,
+      "loss": 0.5147,
+      "step": 34455
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.9059345722198486,
+      "learning_rate": 4.258047747444471e-06,
+      "loss": 0.8849,
+      "step": 34456
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.1015424728393555,
+      "learning_rate": 4.257787009296909e-06,
+      "loss": 0.5158,
+      "step": 34457
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.223742961883545,
+      "learning_rate": 4.257526273213164e-06,
+      "loss": 0.4894,
+      "step": 34458
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.8574225902557373,
+      "learning_rate": 4.25726553919396e-06,
+      "loss": 0.4689,
+      "step": 34459
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.4953129291534424,
+      "learning_rate": 4.257004807240022e-06,
+      "loss": 0.5739,
+      "step": 34460
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.323075294494629,
+      "learning_rate": 4.256744077352076e-06,
+      "loss": 0.3379,
+      "step": 34461
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 4.244275093078613,
+      "learning_rate": 4.256483349530846e-06,
+      "loss": 0.6556,
+      "step": 34462
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.480604648590088,
+      "learning_rate": 4.2562226237770575e-06,
+      "loss": 0.6196,
+      "step": 34463
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.495560884475708,
+      "learning_rate": 4.2559619000914345e-06,
+      "loss": 0.7452,
+      "step": 34464
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.9378094673156738,
+      "learning_rate": 4.255701178474704e-06,
+      "loss": 0.6791,
+      "step": 34465
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.8085739612579346,
+      "learning_rate": 4.255440458927588e-06,
+      "loss": 0.7073,
+      "step": 34466
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.2885997295379639,
+      "learning_rate": 4.255179741450814e-06,
+      "loss": 0.6877,
+      "step": 34467
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.0747764110565186,
+      "learning_rate": 4.254919026045107e-06,
+      "loss": 0.3004,
+      "step": 34468
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.3428940773010254,
+      "learning_rate": 4.254658312711188e-06,
+      "loss": 0.7316,
+      "step": 34469
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.964277505874634,
+      "learning_rate": 4.254397601449788e-06,
+      "loss": 0.5593,
+      "step": 34470
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.226344347000122,
+      "learning_rate": 4.254136892261627e-06,
+      "loss": 0.7262,
+      "step": 34471
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.0895135402679443,
+      "learning_rate": 4.253876185147434e-06,
+      "loss": 0.6372,
+      "step": 34472
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 4.078242301940918,
+      "learning_rate": 4.253615480107931e-06,
+      "loss": 0.8319,
+      "step": 34473
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.838079810142517,
+      "learning_rate": 4.2533547771438435e-06,
+      "loss": 0.4952,
+      "step": 34474
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.6551634073257446,
+      "learning_rate": 4.253094076255897e-06,
+      "loss": 0.6912,
+      "step": 34475
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.331744432449341,
+      "learning_rate": 4.252833377444816e-06,
+      "loss": 0.5203,
+      "step": 34476
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.370781660079956,
+      "learning_rate": 4.252572680711325e-06,
+      "loss": 0.7198,
+      "step": 34477
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.5001678466796875,
+      "learning_rate": 4.252311986056151e-06,
+      "loss": 0.5186,
+      "step": 34478
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.5485851764678955,
+      "learning_rate": 4.252051293480016e-06,
+      "loss": 0.5632,
+      "step": 34479
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.5065319538116455,
+      "learning_rate": 4.251790602983647e-06,
+      "loss": 0.6695,
+      "step": 34480
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.652989149093628,
+      "learning_rate": 4.251529914567767e-06,
+      "loss": 0.6879,
+      "step": 34481
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.7331109046936035,
+      "learning_rate": 4.251269228233102e-06,
+      "loss": 0.4525,
+      "step": 34482
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.4699339866638184,
+      "learning_rate": 4.251008543980377e-06,
+      "loss": 0.6467,
+      "step": 34483
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.3444201946258545,
+      "learning_rate": 4.250747861810317e-06,
+      "loss": 0.6224,
+      "step": 34484
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.8942301273345947,
+      "learning_rate": 4.250487181723646e-06,
+      "loss": 0.4002,
+      "step": 34485
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 7.072000026702881,
+      "learning_rate": 4.25022650372109e-06,
+      "loss": 0.7337,
+      "step": 34486
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.167490243911743,
+      "learning_rate": 4.249965827803373e-06,
+      "loss": 0.6549,
+      "step": 34487
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.2107431888580322,
+      "learning_rate": 4.2497051539712195e-06,
+      "loss": 0.4648,
+      "step": 34488
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.555000066757202,
+      "learning_rate": 4.249444482225354e-06,
+      "loss": 0.7096,
+      "step": 34489
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.048818588256836,
+      "learning_rate": 4.249183812566503e-06,
+      "loss": 0.8048,
+      "step": 34490
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.1589314937591553,
+      "learning_rate": 4.248923144995391e-06,
+      "loss": 0.7413,
+      "step": 34491
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.709789991378784,
+      "learning_rate": 4.2486624795127425e-06,
+      "loss": 0.6578,
+      "step": 34492
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.4159491062164307,
+      "learning_rate": 4.248401816119282e-06,
+      "loss": 0.4453,
+      "step": 34493
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.4254696369171143,
+      "learning_rate": 4.248141154815735e-06,
+      "loss": 0.8029,
+      "step": 34494
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.608955979347229,
+      "learning_rate": 4.247880495602825e-06,
+      "loss": 0.6555,
+      "step": 34495
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.0360445976257324,
+      "learning_rate": 4.247619838481278e-06,
+      "loss": 0.5453,
+      "step": 34496
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.914923667907715,
+      "learning_rate": 4.247359183451819e-06,
+      "loss": 0.6444,
+      "step": 34497
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.570216178894043,
+      "learning_rate": 4.247098530515171e-06,
+      "loss": 0.5814,
+      "step": 34498
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.343268394470215,
+      "learning_rate": 4.246837879672061e-06,
+      "loss": 0.5997,
+      "step": 34499
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.3978171348571777,
+      "learning_rate": 4.2465772309232125e-06,
+      "loss": 0.5932,
+      "step": 34500
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.242403984069824,
+      "learning_rate": 4.24631658426935e-06,
+      "loss": 0.7556,
+      "step": 34501
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.6931854486465454,
+      "learning_rate": 4.246055939711199e-06,
+      "loss": 0.6738,
+      "step": 34502
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.8659334182739258,
+      "learning_rate": 4.2457952972494845e-06,
+      "loss": 0.5049,
+      "step": 34503
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.881164312362671,
+      "learning_rate": 4.24553465688493e-06,
+      "loss": 0.8872,
+      "step": 34504
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.662648916244507,
+      "learning_rate": 4.24527401861826e-06,
+      "loss": 0.7315,
+      "step": 34505
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 4.142088890075684,
+      "learning_rate": 4.245013382450202e-06,
+      "loss": 0.819,
+      "step": 34506
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.7085907459259033,
+      "learning_rate": 4.244752748381478e-06,
+      "loss": 0.6166,
+      "step": 34507
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 4.838042736053467,
+      "learning_rate": 4.244492116412814e-06,
+      "loss": 0.5051,
+      "step": 34508
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.1395576000213623,
+      "learning_rate": 4.244231486544933e-06,
+      "loss": 0.7862,
+      "step": 34509
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.3488821983337402,
+      "learning_rate": 4.2439708587785625e-06,
+      "loss": 0.5947,
+      "step": 34510
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.5828850269317627,
+      "learning_rate": 4.243710233114426e-06,
+      "loss": 0.793,
+      "step": 34511
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.099721670150757,
+      "learning_rate": 4.243449609553249e-06,
+      "loss": 0.6383,
+      "step": 34512
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.940629482269287,
+      "learning_rate": 4.2431889880957534e-06,
+      "loss": 0.9807,
+      "step": 34513
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.24796724319458,
+      "learning_rate": 4.242928368742667e-06,
+      "loss": 0.4918,
+      "step": 34514
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.2567789554595947,
+      "learning_rate": 4.242667751494713e-06,
+      "loss": 0.7153,
+      "step": 34515
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.101670026779175,
+      "learning_rate": 4.2424071363526155e-06,
+      "loss": 0.7157,
+      "step": 34516
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.079456090927124,
+      "learning_rate": 4.242146523317099e-06,
+      "loss": 0.6061,
+      "step": 34517
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.2353904247283936,
+      "learning_rate": 4.241885912388891e-06,
+      "loss": 0.6522,
+      "step": 34518
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.103339195251465,
+      "learning_rate": 4.241625303568714e-06,
+      "loss": 0.637,
+      "step": 34519
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.6817762851715088,
+      "learning_rate": 4.241364696857293e-06,
+      "loss": 0.5865,
+      "step": 34520
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 4.933681011199951,
+      "learning_rate": 4.2411040922553524e-06,
+      "loss": 0.4286,
+      "step": 34521
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.317905902862549,
+      "learning_rate": 4.240843489763616e-06,
+      "loss": 0.485,
+      "step": 34522
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.2237350940704346,
+      "learning_rate": 4.240582889382811e-06,
+      "loss": 0.505,
+      "step": 34523
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 4.1830267906188965,
+      "learning_rate": 4.2403222911136595e-06,
+      "loss": 0.662,
+      "step": 34524
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.3167264461517334,
+      "learning_rate": 4.2400616949568885e-06,
+      "loss": 0.7353,
+      "step": 34525
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.711984634399414,
+      "learning_rate": 4.23980110091322e-06,
+      "loss": 0.7147,
+      "step": 34526
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.8803753852844238,
+      "learning_rate": 4.2395405089833805e-06,
+      "loss": 0.6948,
+      "step": 34527
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.069526195526123,
+      "learning_rate": 4.239279919168094e-06,
+      "loss": 0.6943,
+      "step": 34528
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 5.958366394042969,
+      "learning_rate": 4.2390193314680834e-06,
+      "loss": 0.6305,
+      "step": 34529
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.8626317977905273,
+      "learning_rate": 4.238758745884075e-06,
+      "loss": 0.6425,
+      "step": 34530
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.32013201713562,
+      "learning_rate": 4.2384981624167945e-06,
+      "loss": 0.8155,
+      "step": 34531
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 8.245375633239746,
+      "learning_rate": 4.238237581066966e-06,
+      "loss": 0.5057,
+      "step": 34532
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.3056366443634033,
+      "learning_rate": 4.237977001835313e-06,
+      "loss": 0.7014,
+      "step": 34533
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.9610602855682373,
+      "learning_rate": 4.23771642472256e-06,
+      "loss": 0.5803,
+      "step": 34534
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.7336463928222656,
+      "learning_rate": 4.2374558497294325e-06,
+      "loss": 0.6907,
+      "step": 34535
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.762521505355835,
+      "learning_rate": 4.2371952768566545e-06,
+      "loss": 0.7728,
+      "step": 34536
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.8047263622283936,
+      "learning_rate": 4.236934706104951e-06,
+      "loss": 0.676,
+      "step": 34537
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.8402812480926514,
+      "learning_rate": 4.236674137475047e-06,
+      "loss": 0.7339,
+      "step": 34538
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.120903730392456,
+      "learning_rate": 4.2364135709676645e-06,
+      "loss": 0.4888,
+      "step": 34539
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.564765691757202,
+      "learning_rate": 4.2361530065835305e-06,
+      "loss": 0.4017,
+      "step": 34540
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.899446725845337,
+      "learning_rate": 4.235892444323369e-06,
+      "loss": 0.4413,
+      "step": 34541
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.815629243850708,
+      "learning_rate": 4.235631884187905e-06,
+      "loss": 0.6112,
+      "step": 34542
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.42975115776062,
+      "learning_rate": 4.235371326177861e-06,
+      "loss": 0.7367,
+      "step": 34543
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 4.903228759765625,
+      "learning_rate": 4.235110770293964e-06,
+      "loss": 0.6634,
+      "step": 34544
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.4156105518341064,
+      "learning_rate": 4.234850216536937e-06,
+      "loss": 0.518,
+      "step": 34545
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.641359329223633,
+      "learning_rate": 4.234589664907505e-06,
+      "loss": 0.6628,
+      "step": 34546
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.691969156265259,
+      "learning_rate": 4.234329115406393e-06,
+      "loss": 0.5878,
+      "step": 34547
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 4.665002346038818,
+      "learning_rate": 4.234068568034324e-06,
+      "loss": 0.4119,
+      "step": 34548
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 4.102531433105469,
+      "learning_rate": 4.233808022792024e-06,
+      "loss": 0.6181,
+      "step": 34549
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.2959492206573486,
+      "learning_rate": 4.233547479680214e-06,
+      "loss": 0.806,
+      "step": 34550
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.204226016998291,
+      "learning_rate": 4.233286938699623e-06,
+      "loss": 0.8028,
+      "step": 34551
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.3107056617736816,
+      "learning_rate": 4.2330263998509756e-06,
+      "loss": 0.5619,
+      "step": 34552
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.913501262664795,
+      "learning_rate": 4.232765863134993e-06,
+      "loss": 0.6539,
+      "step": 34553
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.075160503387451,
+      "learning_rate": 4.232505328552402e-06,
+      "loss": 0.7511,
+      "step": 34554
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.4009974002838135,
+      "learning_rate": 4.232244796103926e-06,
+      "loss": 0.5857,
+      "step": 34555
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.69405198097229,
+      "learning_rate": 4.23198426579029e-06,
+      "loss": 0.4504,
+      "step": 34556
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.3982796669006348,
+      "learning_rate": 4.2317237376122165e-06,
+      "loss": 0.5935,
+      "step": 34557
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.6006641387939453,
+      "learning_rate": 4.231463211570433e-06,
+      "loss": 0.7289,
+      "step": 34558
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.649763822555542,
+      "learning_rate": 4.231202687665662e-06,
+      "loss": 0.5641,
+      "step": 34559
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.2147216796875,
+      "learning_rate": 4.230942165898628e-06,
+      "loss": 0.7106,
+      "step": 34560
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 20.8273983001709,
+      "learning_rate": 4.230681646270056e-06,
+      "loss": 0.5735,
+      "step": 34561
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.9732224941253662,
+      "learning_rate": 4.230421128780671e-06,
+      "loss": 0.7855,
+      "step": 34562
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.0411744117736816,
+      "learning_rate": 4.230160613431195e-06,
+      "loss": 0.6065,
+      "step": 34563
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.9819915294647217,
+      "learning_rate": 4.229900100222355e-06,
+      "loss": 0.7012,
+      "step": 34564
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.872326374053955,
+      "learning_rate": 4.229639589154874e-06,
+      "loss": 0.7017,
+      "step": 34565
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.417833089828491,
+      "learning_rate": 4.2293790802294765e-06,
+      "loss": 0.6957,
+      "step": 34566
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.4215171337127686,
+      "learning_rate": 4.229118573446886e-06,
+      "loss": 0.8902,
+      "step": 34567
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.0415632724761963,
+      "learning_rate": 4.22885806880783e-06,
+      "loss": 0.6752,
+      "step": 34568
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.7216086387634277,
+      "learning_rate": 4.22859756631303e-06,
+      "loss": 0.5422,
+      "step": 34569
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.1090242862701416,
+      "learning_rate": 4.228337065963209e-06,
+      "loss": 0.4684,
+      "step": 34570
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.7760937213897705,
+      "learning_rate": 4.228076567759096e-06,
+      "loss": 0.5825,
+      "step": 34571
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.7890806198120117,
+      "learning_rate": 4.2278160717014135e-06,
+      "loss": 0.6457,
+      "step": 34572
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.6578235626220703,
+      "learning_rate": 4.227555577790884e-06,
+      "loss": 0.5576,
+      "step": 34573
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.883103609085083,
+      "learning_rate": 4.227295086028234e-06,
+      "loss": 0.7218,
+      "step": 34574
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.550717830657959,
+      "learning_rate": 4.2270345964141865e-06,
+      "loss": 0.8147,
+      "step": 34575
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.624122142791748,
+      "learning_rate": 4.226774108949466e-06,
+      "loss": 0.4994,
+      "step": 34576
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.0839459896087646,
+      "learning_rate": 4.226513623634797e-06,
+      "loss": 0.7874,
+      "step": 34577
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.6876602172851562,
+      "learning_rate": 4.226253140470904e-06,
+      "loss": 0.5458,
+      "step": 34578
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.9930577278137207,
+      "learning_rate": 4.22599265945851e-06,
+      "loss": 0.8174,
+      "step": 34579
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.674795389175415,
+      "learning_rate": 4.225732180598342e-06,
+      "loss": 0.5415,
+      "step": 34580
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.1780312061309814,
+      "learning_rate": 4.225471703891123e-06,
+      "loss": 0.7023,
+      "step": 34581
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.4105851650238037,
+      "learning_rate": 4.2252112293375754e-06,
+      "loss": 0.8024,
+      "step": 34582
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.5449609756469727,
+      "learning_rate": 4.224950756938427e-06,
+      "loss": 0.4995,
+      "step": 34583
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.090111255645752,
+      "learning_rate": 4.224690286694399e-06,
+      "loss": 0.4921,
+      "step": 34584
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.2963294982910156,
+      "learning_rate": 4.224429818606217e-06,
+      "loss": 0.7186,
+      "step": 34585
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.3686838150024414,
+      "learning_rate": 4.224169352674607e-06,
+      "loss": 0.8959,
+      "step": 34586
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.4656572341918945,
+      "learning_rate": 4.223908888900289e-06,
+      "loss": 0.3722,
+      "step": 34587
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.5081753730773926,
+      "learning_rate": 4.223648427283991e-06,
+      "loss": 0.6123,
+      "step": 34588
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.7144882678985596,
+      "learning_rate": 4.223387967826435e-06,
+      "loss": 0.5732,
+      "step": 34589
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.242827892303467,
+      "learning_rate": 4.2231275105283464e-06,
+      "loss": 0.8075,
+      "step": 34590
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.6779627799987793,
+      "learning_rate": 4.222867055390447e-06,
+      "loss": 0.8945,
+      "step": 34591
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.8897850513458252,
+      "learning_rate": 4.222606602413466e-06,
+      "loss": 0.4886,
+      "step": 34592
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.025063991546631,
+      "learning_rate": 4.222346151598124e-06,
+      "loss": 0.4722,
+      "step": 34593
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.6946356296539307,
+      "learning_rate": 4.222085702945146e-06,
+      "loss": 0.4393,
+      "step": 34594
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.1804516315460205,
+      "learning_rate": 4.221825256455257e-06,
+      "loss": 0.5765,
+      "step": 34595
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 4.506147861480713,
+      "learning_rate": 4.2215648121291795e-06,
+      "loss": 0.6314,
+      "step": 34596
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.6715009212493896,
+      "learning_rate": 4.22130436996764e-06,
+      "loss": 0.5678,
+      "step": 34597
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.751697063446045,
+      "learning_rate": 4.22104392997136e-06,
+      "loss": 0.5221,
+      "step": 34598
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.4816839694976807,
+      "learning_rate": 4.220783492141066e-06,
+      "loss": 0.3941,
+      "step": 34599
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 7.050546169281006,
+      "learning_rate": 4.22052305647748e-06,
+      "loss": 0.7809,
+      "step": 34600
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.9419612884521484,
+      "learning_rate": 4.2202626229813285e-06,
+      "loss": 0.6461,
+      "step": 34601
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 5.731898307800293,
+      "learning_rate": 4.2200021916533346e-06,
+      "loss": 0.6091,
+      "step": 34602
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.3796496391296387,
+      "learning_rate": 4.219741762494222e-06,
+      "loss": 0.7544,
+      "step": 34603
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.5718779563903809,
+      "learning_rate": 4.219481335504715e-06,
+      "loss": 0.5142,
+      "step": 34604
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.9332385063171387,
+      "learning_rate": 4.2192209106855385e-06,
+      "loss": 0.664,
+      "step": 34605
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.5388267040252686,
+      "learning_rate": 4.2189604880374156e-06,
+      "loss": 0.8717,
+      "step": 34606
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.504848003387451,
+      "learning_rate": 4.2187000675610715e-06,
+      "loss": 0.5826,
+      "step": 34607
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.357860803604126,
+      "learning_rate": 4.21843964925723e-06,
+      "loss": 0.8235,
+      "step": 34608
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.9215140342712402,
+      "learning_rate": 4.2181792331266146e-06,
+      "loss": 0.4698,
+      "step": 34609
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.0288615226745605,
+      "learning_rate": 4.217918819169951e-06,
+      "loss": 0.5148,
+      "step": 34610
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.0509796142578125,
+      "learning_rate": 4.217658407387959e-06,
+      "loss": 0.5568,
+      "step": 34611
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.492614507675171,
+      "learning_rate": 4.2173979977813675e-06,
+      "loss": 0.5834,
+      "step": 34612
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.1090757846832275,
+      "learning_rate": 4.2171375903508995e-06,
+      "loss": 0.8803,
+      "step": 34613
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.8162124156951904,
+      "learning_rate": 4.216877185097279e-06,
+      "loss": 0.6729,
+      "step": 34614
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 5.731942653656006,
+      "learning_rate": 4.216616782021229e-06,
+      "loss": 0.7195,
+      "step": 34615
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.113070011138916,
+      "learning_rate": 4.216356381123475e-06,
+      "loss": 0.8233,
+      "step": 34616
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.1835076808929443,
+      "learning_rate": 4.21609598240474e-06,
+      "loss": 0.5654,
+      "step": 34617
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 4.109225273132324,
+      "learning_rate": 4.215835585865748e-06,
+      "loss": 0.6529,
+      "step": 34618
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 4.864064693450928,
+      "learning_rate": 4.215575191507224e-06,
+      "loss": 0.6779,
+      "step": 34619
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.9375954866409302,
+      "learning_rate": 4.2153147993298915e-06,
+      "loss": 0.4408,
+      "step": 34620
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.986671209335327,
+      "learning_rate": 4.2150544093344745e-06,
+      "loss": 0.669,
+      "step": 34621
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.8857219219207764,
+      "learning_rate": 4.214794021521698e-06,
+      "loss": 0.7258,
+      "step": 34622
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.2718629837036133,
+      "learning_rate": 4.214533635892284e-06,
+      "loss": 0.4966,
+      "step": 34623
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 6.176692008972168,
+      "learning_rate": 4.214273252446958e-06,
+      "loss": 0.5064,
+      "step": 34624
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.4898264408111572,
+      "learning_rate": 4.214012871186443e-06,
+      "loss": 0.7124,
+      "step": 34625
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 4.204509258270264,
+      "learning_rate": 4.2137524921114655e-06,
+      "loss": 0.6537,
+      "step": 34626
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.219022750854492,
+      "learning_rate": 4.213492115222747e-06,
+      "loss": 0.6361,
+      "step": 34627
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.1819722652435303,
+      "learning_rate": 4.213231740521012e-06,
+      "loss": 0.5294,
+      "step": 34628
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 6.584855079650879,
+      "learning_rate": 4.212971368006984e-06,
+      "loss": 0.4865,
+      "step": 34629
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.1827352046966553,
+      "learning_rate": 4.212710997681388e-06,
+      "loss": 0.5888,
+      "step": 34630
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 4.062024116516113,
+      "learning_rate": 4.212450629544948e-06,
+      "loss": 0.8492,
+      "step": 34631
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.3090786933898926,
+      "learning_rate": 4.212190263598388e-06,
+      "loss": 0.8112,
+      "step": 34632
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.8346318006515503,
+      "learning_rate": 4.211929899842432e-06,
+      "loss": 0.6226,
+      "step": 34633
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.914172649383545,
+      "learning_rate": 4.211669538277804e-06,
+      "loss": 0.5849,
+      "step": 34634
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.815809488296509,
+      "learning_rate": 4.211409178905228e-06,
+      "loss": 0.6621,
+      "step": 34635
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 4.18337869644165,
+      "learning_rate": 4.211148821725427e-06,
+      "loss": 0.7344,
+      "step": 34636
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.312359094619751,
+      "learning_rate": 4.210888466739126e-06,
+      "loss": 0.5317,
+      "step": 34637
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 5.187565803527832,
+      "learning_rate": 4.210628113947048e-06,
+      "loss": 0.6967,
+      "step": 34638
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.5041284561157227,
+      "learning_rate": 4.210367763349918e-06,
+      "loss": 0.6382,
+      "step": 34639
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.735652208328247,
+      "learning_rate": 4.210107414948458e-06,
+      "loss": 0.6958,
+      "step": 34640
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.607869863510132,
+      "learning_rate": 4.209847068743394e-06,
+      "loss": 0.3566,
+      "step": 34641
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.891864061355591,
+      "learning_rate": 4.209586724735451e-06,
+      "loss": 0.5136,
+      "step": 34642
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.2106411457061768,
+      "learning_rate": 4.20932638292535e-06,
+      "loss": 0.6487,
+      "step": 34643
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.2800302505493164,
+      "learning_rate": 4.209066043313816e-06,
+      "loss": 0.4138,
+      "step": 34644
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.4451847076416016,
+      "learning_rate": 4.2088057059015735e-06,
+      "loss": 0.8226,
+      "step": 34645
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.923946499824524,
+      "learning_rate": 4.208545370689346e-06,
+      "loss": 0.6095,
+      "step": 34646
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.4706690311431885,
+      "learning_rate": 4.2082850376778574e-06,
+      "loss": 0.5857,
+      "step": 34647
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.805316209793091,
+      "learning_rate": 4.208024706867832e-06,
+      "loss": 0.6003,
+      "step": 34648
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.296186685562134,
+      "learning_rate": 4.207764378259992e-06,
+      "loss": 0.396,
+      "step": 34649
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.6929032802581787,
+      "learning_rate": 4.207504051855062e-06,
+      "loss": 0.7432,
+      "step": 34650
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.652992010116577,
+      "learning_rate": 4.207243727653768e-06,
+      "loss": 0.8982,
+      "step": 34651
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.9586334228515625,
+      "learning_rate": 4.2069834056568295e-06,
+      "loss": 0.6159,
+      "step": 34652
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.646677255630493,
+      "learning_rate": 4.206723085864975e-06,
+      "loss": 0.5919,
+      "step": 34653
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.930487871170044,
+      "learning_rate": 4.206462768278927e-06,
+      "loss": 0.411,
+      "step": 34654
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.2865054607391357,
+      "learning_rate": 4.206202452899408e-06,
+      "loss": 0.5789,
+      "step": 34655
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.8368475437164307,
+      "learning_rate": 4.205942139727143e-06,
+      "loss": 0.6342,
+      "step": 34656
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.1996591091156006,
+      "learning_rate": 4.205681828762856e-06,
+      "loss": 0.605,
+      "step": 34657
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.9753551483154297,
+      "learning_rate": 4.205421520007269e-06,
+      "loss": 0.8408,
+      "step": 34658
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.603443145751953,
+      "learning_rate": 4.2051612134611084e-06,
+      "loss": 0.5666,
+      "step": 34659
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.5451507568359375,
+      "learning_rate": 4.204900909125096e-06,
+      "loss": 0.4862,
+      "step": 34660
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.166863441467285,
+      "learning_rate": 4.204640606999956e-06,
+      "loss": 0.4259,
+      "step": 34661
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.8907885551452637,
+      "learning_rate": 4.204380307086413e-06,
+      "loss": 0.728,
+      "step": 34662
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.9381978511810303,
+      "learning_rate": 4.20412000938519e-06,
+      "loss": 0.753,
+      "step": 34663
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.348966598510742,
+      "learning_rate": 4.203859713897012e-06,
+      "loss": 0.5544,
+      "step": 34664
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.2829221487045288,
+      "learning_rate": 4.203599420622601e-06,
+      "loss": 0.4447,
+      "step": 34665
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.361055374145508,
+      "learning_rate": 4.203339129562682e-06,
+      "loss": 0.7274,
+      "step": 34666
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.9961289167404175,
+      "learning_rate": 4.203078840717978e-06,
+      "loss": 0.6056,
+      "step": 34667
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.844454050064087,
+      "learning_rate": 4.202818554089214e-06,
+      "loss": 0.6273,
+      "step": 34668
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.8610283136367798,
+      "learning_rate": 4.2025582696771126e-06,
+      "loss": 0.6452,
+      "step": 34669
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.691413402557373,
+      "learning_rate": 4.202297987482398e-06,
+      "loss": 0.571,
+      "step": 34670
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.1933250427246094,
+      "learning_rate": 4.202037707505794e-06,
+      "loss": 0.3901,
+      "step": 34671
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.6584839820861816,
+      "learning_rate": 4.201777429748022e-06,
+      "loss": 0.7936,
+      "step": 34672
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 4.193641185760498,
+      "learning_rate": 4.20151715420981e-06,
+      "loss": 0.6996,
+      "step": 34673
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.197237014770508,
+      "learning_rate": 4.201256880891879e-06,
+      "loss": 0.5129,
+      "step": 34674
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.2473599910736084,
+      "learning_rate": 4.200996609794954e-06,
+      "loss": 0.5875,
+      "step": 34675
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.4682607650756836,
+      "learning_rate": 4.200736340919758e-06,
+      "loss": 0.7506,
+      "step": 34676
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.5681480169296265,
+      "learning_rate": 4.200476074267015e-06,
+      "loss": 0.698,
+      "step": 34677
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.884618043899536,
+      "learning_rate": 4.200215809837449e-06,
+      "loss": 0.6547,
+      "step": 34678
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.1436820030212402,
+      "learning_rate": 4.199955547631782e-06,
+      "loss": 0.703,
+      "step": 34679
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.911604881286621,
+      "learning_rate": 4.19969528765074e-06,
+      "loss": 0.519,
+      "step": 34680
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.8029823303222656,
+      "learning_rate": 4.199435029895045e-06,
+      "loss": 0.786,
+      "step": 34681
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.2288670539855957,
+      "learning_rate": 4.1991747743654224e-06,
+      "loss": 0.5004,
+      "step": 34682
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.9701776504516602,
+      "learning_rate": 4.198914521062593e-06,
+      "loss": 0.7878,
+      "step": 34683
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.7037408351898193,
+      "learning_rate": 4.198654269987283e-06,
+      "loss": 0.751,
+      "step": 34684
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.0646770000457764,
+      "learning_rate": 4.198394021140216e-06,
+      "loss": 0.5501,
+      "step": 34685
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.654935359954834,
+      "learning_rate": 4.198133774522114e-06,
+      "loss": 0.4654,
+      "step": 34686
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.9940755367279053,
+      "learning_rate": 4.197873530133702e-06,
+      "loss": 0.6281,
+      "step": 34687
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.0154478549957275,
+      "learning_rate": 4.197613287975703e-06,
+      "loss": 0.5482,
+      "step": 34688
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.5265737771987915,
+      "learning_rate": 4.197353048048841e-06,
+      "loss": 0.7396,
+      "step": 34689
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.594710111618042,
+      "learning_rate": 4.197092810353839e-06,
+      "loss": 0.5593,
+      "step": 34690
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.4274606704711914,
+      "learning_rate": 4.196832574891421e-06,
+      "loss": 0.5501,
+      "step": 34691
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.529703378677368,
+      "learning_rate": 4.1965723416623104e-06,
+      "loss": 0.5662,
+      "step": 34692
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.082681179046631,
+      "learning_rate": 4.1963121106672325e-06,
+      "loss": 0.6909,
+      "step": 34693
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.061436891555786,
+      "learning_rate": 4.196051881906909e-06,
+      "loss": 0.4954,
+      "step": 34694
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.9451771974563599,
+      "learning_rate": 4.195791655382064e-06,
+      "loss": 0.6512,
+      "step": 34695
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 5.391259670257568,
+      "learning_rate": 4.195531431093421e-06,
+      "loss": 0.5508,
+      "step": 34696
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.989793062210083,
+      "learning_rate": 4.195271209041704e-06,
+      "loss": 0.7622,
+      "step": 34697
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.5413126945495605,
+      "learning_rate": 4.195010989227637e-06,
+      "loss": 0.6932,
+      "step": 34698
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.7858575582504272,
+      "learning_rate": 4.194750771651942e-06,
+      "loss": 0.7593,
+      "step": 34699
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.7030246257781982,
+      "learning_rate": 4.194490556315344e-06,
+      "loss": 0.8753,
+      "step": 34700
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.4289183616638184,
+      "learning_rate": 4.194230343218565e-06,
+      "loss": 0.8613,
+      "step": 34701
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.5891056060791016,
+      "learning_rate": 4.193970132362329e-06,
+      "loss": 0.6894,
+      "step": 34702
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 3.1899006366729736,
+      "learning_rate": 4.1937099237473624e-06,
+      "loss": 0.8163,
+      "step": 34703
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.4449970722198486,
+      "learning_rate": 4.1934497173743845e-06,
+      "loss": 0.6832,
+      "step": 34704
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.7762281894683838,
+      "learning_rate": 4.193189513244123e-06,
+      "loss": 0.4948,
+      "step": 34705
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.3741326332092285,
+      "learning_rate": 4.192929311357297e-06,
+      "loss": 0.7251,
+      "step": 34706
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 3.5573689937591553,
+      "learning_rate": 4.192669111714633e-06,
+      "loss": 0.5136,
+      "step": 34707
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 3.0872111320495605,
+      "learning_rate": 4.1924089143168545e-06,
+      "loss": 0.5682,
+      "step": 34708
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 3.129631519317627,
+      "learning_rate": 4.192148719164684e-06,
+      "loss": 0.5544,
+      "step": 34709
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.603374719619751,
+      "learning_rate": 4.1918885262588445e-06,
+      "loss": 0.5748,
+      "step": 34710
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 3.0632662773132324,
+      "learning_rate": 4.191628335600062e-06,
+      "loss": 0.6112,
+      "step": 34711
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.0403215885162354,
+      "learning_rate": 4.191368147189057e-06,
+      "loss": 0.6649,
+      "step": 34712
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.102973222732544,
+      "learning_rate": 4.191107961026553e-06,
+      "loss": 0.7155,
+      "step": 34713
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.068791389465332,
+      "learning_rate": 4.190847777113275e-06,
+      "loss": 0.7758,
+      "step": 34714
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.2095556259155273,
+      "learning_rate": 4.190587595449947e-06,
+      "loss": 0.8266,
+      "step": 34715
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.519625663757324,
+      "learning_rate": 4.190327416037292e-06,
+      "loss": 0.6645,
+      "step": 34716
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.6522711515426636,
+      "learning_rate": 4.1900672388760335e-06,
+      "loss": 0.7671,
+      "step": 34717
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.5399065017700195,
+      "learning_rate": 4.189807063966894e-06,
+      "loss": 0.7037,
+      "step": 34718
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 3.6594369411468506,
+      "learning_rate": 4.189546891310598e-06,
+      "loss": 0.7537,
+      "step": 34719
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 7.468273639678955,
+      "learning_rate": 4.189286720907868e-06,
+      "loss": 0.6449,
+      "step": 34720
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 3.231410503387451,
+      "learning_rate": 4.189026552759428e-06,
+      "loss": 0.8366,
+      "step": 34721
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.9134491682052612,
+      "learning_rate": 4.1887663868660025e-06,
+      "loss": 0.5453,
+      "step": 34722
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.9926767349243164,
+      "learning_rate": 4.188506223228313e-06,
+      "loss": 0.7951,
+      "step": 34723
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.166276454925537,
+      "learning_rate": 4.188246061847083e-06,
+      "loss": 0.5701,
+      "step": 34724
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 3.059425115585327,
+      "learning_rate": 4.187985902723038e-06,
+      "loss": 0.6111,
+      "step": 34725
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.9081811904907227,
+      "learning_rate": 4.187725745856899e-06,
+      "loss": 0.5712,
+      "step": 34726
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.8792948722839355,
+      "learning_rate": 4.18746559124939e-06,
+      "loss": 0.4494,
+      "step": 34727
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 3.181903123855591,
+      "learning_rate": 4.187205438901235e-06,
+      "loss": 0.6539,
+      "step": 34728
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.259413719177246,
+      "learning_rate": 4.186945288813158e-06,
+      "loss": 0.5039,
+      "step": 34729
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.0427024364471436,
+      "learning_rate": 4.186685140985882e-06,
+      "loss": 0.5786,
+      "step": 34730
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 3.416186809539795,
+      "learning_rate": 4.186424995420129e-06,
+      "loss": 0.6922,
+      "step": 34731
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.4293367862701416,
+      "learning_rate": 4.1861648521166235e-06,
+      "loss": 0.5555,
+      "step": 34732
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.119873285293579,
+      "learning_rate": 4.185904711076086e-06,
+      "loss": 0.6793,
+      "step": 34733
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.0235798358917236,
+      "learning_rate": 4.185644572299245e-06,
+      "loss": 0.5583,
+      "step": 34734
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.3364880084991455,
+      "learning_rate": 4.185384435786822e-06,
+      "loss": 0.8263,
+      "step": 34735
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.097921371459961,
+      "learning_rate": 4.1851243015395396e-06,
+      "loss": 0.6949,
+      "step": 34736
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.098729372024536,
+      "learning_rate": 4.1848641695581205e-06,
+      "loss": 0.6424,
+      "step": 34737
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 3.6746044158935547,
+      "learning_rate": 4.184604039843289e-06,
+      "loss": 0.7243,
+      "step": 34738
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.91153085231781,
+      "learning_rate": 4.1843439123957685e-06,
+      "loss": 0.7754,
+      "step": 34739
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.534205913543701,
+      "learning_rate": 4.184083787216282e-06,
+      "loss": 0.6247,
+      "step": 34740
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.0888240337371826,
+      "learning_rate": 4.183823664305553e-06,
+      "loss": 0.5658,
+      "step": 34741
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.03261661529541,
+      "learning_rate": 4.183563543664304e-06,
+      "loss": 0.7103,
+      "step": 34742
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.2825920581817627,
+      "learning_rate": 4.18330342529326e-06,
+      "loss": 0.682,
+      "step": 34743
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 5.408255100250244,
+      "learning_rate": 4.183043309193142e-06,
+      "loss": 0.6725,
+      "step": 34744
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.4357831478118896,
+      "learning_rate": 4.182783195364676e-06,
+      "loss": 0.5248,
+      "step": 34745
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.609184503555298,
+      "learning_rate": 4.182523083808582e-06,
+      "loss": 0.5908,
+      "step": 34746
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 3.6512627601623535,
+      "learning_rate": 4.182262974525587e-06,
+      "loss": 0.8396,
+      "step": 34747
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.8389971256256104,
+      "learning_rate": 4.182002867516411e-06,
+      "loss": 0.5674,
+      "step": 34748
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.9133458137512207,
+      "learning_rate": 4.181742762781778e-06,
+      "loss": 0.6598,
+      "step": 34749
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.085294485092163,
+      "learning_rate": 4.181482660322414e-06,
+      "loss": 0.5141,
+      "step": 34750
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 3.451209783554077,
+      "learning_rate": 4.181222560139038e-06,
+      "loss": 0.8302,
+      "step": 34751
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.119316577911377,
+      "learning_rate": 4.180962462232376e-06,
+      "loss": 0.5432,
+      "step": 34752
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 3.8874921798706055,
+      "learning_rate": 4.180702366603149e-06,
+      "loss": 0.6687,
+      "step": 34753
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.1274254322052,
+      "learning_rate": 4.180442273252084e-06,
+      "loss": 0.7565,
+      "step": 34754
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 3.2675721645355225,
+      "learning_rate": 4.1801821821799015e-06,
+      "loss": 0.8318,
+      "step": 34755
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.1708791255950928,
+      "learning_rate": 4.179922093387325e-06,
+      "loss": 0.4274,
+      "step": 34756
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.9387108087539673,
+      "learning_rate": 4.179662006875079e-06,
+      "loss": 0.7693,
+      "step": 34757
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 3.743666648864746,
+      "learning_rate": 4.179401922643885e-06,
+      "loss": 0.6691,
+      "step": 34758
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.62045419216156,
+      "learning_rate": 4.179141840694467e-06,
+      "loss": 0.5195,
+      "step": 34759
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.3626620769500732,
+      "learning_rate": 4.178881761027547e-06,
+      "loss": 0.5923,
+      "step": 34760
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.7653706073760986,
+      "learning_rate": 4.178621683643851e-06,
+      "loss": 0.6129,
+      "step": 34761
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.073989152908325,
+      "learning_rate": 4.178361608544099e-06,
+      "loss": 0.6909,
+      "step": 34762
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.7936480045318604,
+      "learning_rate": 4.178101535729016e-06,
+      "loss": 0.8158,
+      "step": 34763
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.3869941234588623,
+      "learning_rate": 4.177841465199324e-06,
+      "loss": 0.5152,
+      "step": 34764
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 3.3374361991882324,
+      "learning_rate": 4.177581396955747e-06,
+      "loss": 0.9226,
+      "step": 34765
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.6347283124923706,
+      "learning_rate": 4.177321330999009e-06,
+      "loss": 0.3755,
+      "step": 34766
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 3.294527292251587,
+      "learning_rate": 4.1770612673298325e-06,
+      "loss": 0.5503,
+      "step": 34767
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.5224922895431519,
+      "learning_rate": 4.176801205948939e-06,
+      "loss": 0.5589,
+      "step": 34768
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.9675086736679077,
+      "learning_rate": 4.176541146857055e-06,
+      "loss": 0.6591,
+      "step": 34769
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.7465686798095703,
+      "learning_rate": 4.1762810900549e-06,
+      "loss": 0.6973,
+      "step": 34770
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.9927289485931396,
+      "learning_rate": 4.1760210355431985e-06,
+      "loss": 0.6948,
+      "step": 34771
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.8177733421325684,
+      "learning_rate": 4.175760983322675e-06,
+      "loss": 0.6489,
+      "step": 34772
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.6270861625671387,
+      "learning_rate": 4.175500933394051e-06,
+      "loss": 0.8316,
+      "step": 34773
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 3.191032886505127,
+      "learning_rate": 4.175240885758048e-06,
+      "loss": 0.8272,
+      "step": 34774
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 3.537196397781372,
+      "learning_rate": 4.174980840415392e-06,
+      "loss": 0.6327,
+      "step": 34775
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.92664635181427,
+      "learning_rate": 4.174720797366807e-06,
+      "loss": 0.5807,
+      "step": 34776
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.109941005706787,
+      "learning_rate": 4.174460756613013e-06,
+      "loss": 0.8144,
+      "step": 34777
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.1396727561950684,
+      "learning_rate": 4.174200718154735e-06,
+      "loss": 0.93,
+      "step": 34778
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 3.735060930252075,
+      "learning_rate": 4.173940681992695e-06,
+      "loss": 0.8737,
+      "step": 34779
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.9335440397262573,
+      "learning_rate": 4.173680648127617e-06,
+      "loss": 0.5586,
+      "step": 34780
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.716794490814209,
+      "learning_rate": 4.173420616560223e-06,
+      "loss": 0.6897,
+      "step": 34781
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.5011869668960571,
+      "learning_rate": 4.1731605872912374e-06,
+      "loss": 0.5794,
+      "step": 34782
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.090078353881836,
+      "learning_rate": 4.172900560321383e-06,
+      "loss": 0.7646,
+      "step": 34783
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.3539509773254395,
+      "learning_rate": 4.172640535651381e-06,
+      "loss": 0.7658,
+      "step": 34784
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.687361240386963,
+      "learning_rate": 4.172380513281956e-06,
+      "loss": 0.72,
+      "step": 34785
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.2048099040985107,
+      "learning_rate": 4.172120493213831e-06,
+      "loss": 0.7301,
+      "step": 34786
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.0946156978607178,
+      "learning_rate": 4.171860475447729e-06,
+      "loss": 0.6838,
+      "step": 34787
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.8097879886627197,
+      "learning_rate": 4.171600459984373e-06,
+      "loss": 0.6942,
+      "step": 34788
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.281865119934082,
+      "learning_rate": 4.171340446824485e-06,
+      "loss": 0.7662,
+      "step": 34789
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.1135902404785156,
+      "learning_rate": 4.171080435968788e-06,
+      "loss": 0.6514,
+      "step": 34790
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.156611919403076,
+      "learning_rate": 4.170820427418008e-06,
+      "loss": 0.6572,
+      "step": 34791
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 3.1007256507873535,
+      "learning_rate": 4.170560421172865e-06,
+      "loss": 0.4829,
+      "step": 34792
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.616091012954712,
+      "learning_rate": 4.170300417234082e-06,
+      "loss": 0.4124,
+      "step": 34793
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.881342649459839,
+      "learning_rate": 4.170040415602381e-06,
+      "loss": 0.5891,
+      "step": 34794
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.398198127746582,
+      "learning_rate": 4.169780416278488e-06,
+      "loss": 0.8045,
+      "step": 34795
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 3.4406535625457764,
+      "learning_rate": 4.169520419263126e-06,
+      "loss": 0.6017,
+      "step": 34796
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.402346134185791,
+      "learning_rate": 4.169260424557016e-06,
+      "loss": 0.6528,
+      "step": 34797
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.2936840057373047,
+      "learning_rate": 4.169000432160882e-06,
+      "loss": 0.6966,
+      "step": 34798
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.7654889822006226,
+      "learning_rate": 4.168740442075446e-06,
+      "loss": 0.6305,
+      "step": 34799
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.092254400253296,
+      "learning_rate": 4.1684804543014315e-06,
+      "loss": 0.7371,
+      "step": 34800
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.375617504119873,
+      "learning_rate": 4.1682204688395616e-06,
+      "loss": 0.7992,
+      "step": 34801
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 3.3715906143188477,
+      "learning_rate": 4.167960485690558e-06,
+      "loss": 0.5585,
+      "step": 34802
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.157036304473877,
+      "learning_rate": 4.167700504855146e-06,
+      "loss": 0.488,
+      "step": 34803
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 3.758493185043335,
+      "learning_rate": 4.167440526334047e-06,
+      "loss": 0.6531,
+      "step": 34804
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.141117572784424,
+      "learning_rate": 4.167180550127983e-06,
+      "loss": 0.4889,
+      "step": 34805
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.8011906147003174,
+      "learning_rate": 4.166920576237678e-06,
+      "loss": 0.5519,
+      "step": 34806
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.0559144020080566,
+      "learning_rate": 4.166660604663855e-06,
+      "loss": 0.3479,
+      "step": 34807
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.229381799697876,
+      "learning_rate": 4.166400635407238e-06,
+      "loss": 0.6382,
+      "step": 34808
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.497549295425415,
+      "learning_rate": 4.166140668468548e-06,
+      "loss": 0.8027,
+      "step": 34809
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.046269655227661,
+      "learning_rate": 4.165880703848507e-06,
+      "loss": 0.5581,
+      "step": 34810
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.348578929901123,
+      "learning_rate": 4.165620741547841e-06,
+      "loss": 0.836,
+      "step": 34811
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.02207612991333,
+      "learning_rate": 4.1653607815672706e-06,
+      "loss": 0.5561,
+      "step": 34812
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.391815423965454,
+      "learning_rate": 4.165100823907518e-06,
+      "loss": 0.6462,
+      "step": 34813
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.9195441007614136,
+      "learning_rate": 4.164840868569307e-06,
+      "loss": 0.5485,
+      "step": 34814
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 3.0791547298431396,
+      "learning_rate": 4.164580915553361e-06,
+      "loss": 0.6583,
+      "step": 34815
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 4.082622528076172,
+      "learning_rate": 4.164320964860404e-06,
+      "loss": 0.4171,
+      "step": 34816
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.2761571407318115,
+      "learning_rate": 4.164061016491157e-06,
+      "loss": 0.5076,
+      "step": 34817
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.326693534851074,
+      "learning_rate": 4.163801070446343e-06,
+      "loss": 0.612,
+      "step": 34818
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.6851959228515625,
+      "learning_rate": 4.163541126726684e-06,
+      "loss": 0.4426,
+      "step": 34819
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.409883737564087,
+      "learning_rate": 4.163281185332905e-06,
+      "loss": 0.5202,
+      "step": 34820
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.6094815731048584,
+      "learning_rate": 4.163021246265728e-06,
+      "loss": 0.6213,
+      "step": 34821
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.0207152366638184,
+      "learning_rate": 4.162761309525874e-06,
+      "loss": 0.7142,
+      "step": 34822
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.4886155128479004,
+      "learning_rate": 4.162501375114068e-06,
+      "loss": 0.6461,
+      "step": 34823
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.2344577312469482,
+      "learning_rate": 4.162241443031032e-06,
+      "loss": 0.8189,
+      "step": 34824
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.3330070972442627,
+      "learning_rate": 4.1619815132774874e-06,
+      "loss": 0.5681,
+      "step": 34825
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.988959312438965,
+      "learning_rate": 4.161721585854159e-06,
+      "loss": 0.6084,
+      "step": 34826
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 3.003021717071533,
+      "learning_rate": 4.161461660761769e-06,
+      "loss": 0.72,
+      "step": 34827
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.9728816747665405,
+      "learning_rate": 4.16120173800104e-06,
+      "loss": 0.6752,
+      "step": 34828
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.4359374046325684,
+      "learning_rate": 4.160941817572696e-06,
+      "loss": 0.5854,
+      "step": 34829
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.0870182514190674,
+      "learning_rate": 4.160681899477457e-06,
+      "loss": 0.4065,
+      "step": 34830
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.44692325592041,
+      "learning_rate": 4.160421983716048e-06,
+      "loss": 0.6325,
+      "step": 34831
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.9244844913482666,
+      "learning_rate": 4.16016207028919e-06,
+      "loss": 0.6272,
+      "step": 34832
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 9.707650184631348,
+      "learning_rate": 4.159902159197607e-06,
+      "loss": 0.7527,
+      "step": 34833
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.0256080627441406,
+      "learning_rate": 4.159642250442022e-06,
+      "loss": 0.5204,
+      "step": 34834
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.043809175491333,
+      "learning_rate": 4.159382344023156e-06,
+      "loss": 0.8527,
+      "step": 34835
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.7329578399658203,
+      "learning_rate": 4.159122439941733e-06,
+      "loss": 0.8736,
+      "step": 34836
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.1294121742248535,
+      "learning_rate": 4.1588625381984754e-06,
+      "loss": 0.5812,
+      "step": 34837
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.7826862335205078,
+      "learning_rate": 4.158602638794107e-06,
+      "loss": 0.6102,
+      "step": 34838
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 3.0984036922454834,
+      "learning_rate": 4.158342741729348e-06,
+      "loss": 0.5531,
+      "step": 34839
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.7632858753204346,
+      "learning_rate": 4.158082847004925e-06,
+      "loss": 0.5319,
+      "step": 34840
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 3.6042120456695557,
+      "learning_rate": 4.1578229546215564e-06,
+      "loss": 0.9829,
+      "step": 34841
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.9472923278808594,
+      "learning_rate": 4.157563064579967e-06,
+      "loss": 0.5668,
+      "step": 34842
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 3.113450288772583,
+      "learning_rate": 4.157303176880879e-06,
+      "loss": 0.7749,
+      "step": 34843
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 3.1163671016693115,
+      "learning_rate": 4.157043291525016e-06,
+      "loss": 0.5088,
+      "step": 34844
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 3.024847984313965,
+      "learning_rate": 4.156783408513099e-06,
+      "loss": 0.7664,
+      "step": 34845
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.9522860050201416,
+      "learning_rate": 4.156523527845852e-06,
+      "loss": 0.4237,
+      "step": 34846
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.5541975498199463,
+      "learning_rate": 4.156263649523997e-06,
+      "loss": 0.5302,
+      "step": 34847
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.4450557231903076,
+      "learning_rate": 4.156003773548256e-06,
+      "loss": 0.6007,
+      "step": 34848
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 3.5234692096710205,
+      "learning_rate": 4.1557438999193534e-06,
+      "loss": 0.5066,
+      "step": 34849
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.690977096557617,
+      "learning_rate": 4.15548402863801e-06,
+      "loss": 0.8358,
+      "step": 34850
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.8396413326263428,
+      "learning_rate": 4.155224159704949e-06,
+      "loss": 0.5489,
+      "step": 34851
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.623387336730957,
+      "learning_rate": 4.154964293120894e-06,
+      "loss": 0.6505,
+      "step": 34852
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.3254692554473877,
+      "learning_rate": 4.154704428886566e-06,
+      "loss": 0.5707,
+      "step": 34853
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.5233078002929688,
+      "learning_rate": 4.154444567002689e-06,
+      "loss": 0.759,
+      "step": 34854
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.3204357624053955,
+      "learning_rate": 4.154184707469983e-06,
+      "loss": 0.4296,
+      "step": 34855
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.9467432498931885,
+      "learning_rate": 4.153924850289173e-06,
+      "loss": 0.8146,
+      "step": 34856
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.315830945968628,
+      "learning_rate": 4.153664995460982e-06,
+      "loss": 0.6335,
+      "step": 34857
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.689012050628662,
+      "learning_rate": 4.153405142986132e-06,
+      "loss": 0.4671,
+      "step": 34858
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.27492618560791,
+      "learning_rate": 4.153145292865345e-06,
+      "loss": 0.6724,
+      "step": 34859
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.6057426929473877,
+      "learning_rate": 4.1528854450993435e-06,
+      "loss": 0.6168,
+      "step": 34860
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.827543020248413,
+      "learning_rate": 4.15262559968885e-06,
+      "loss": 0.5656,
+      "step": 34861
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.6848909854888916,
+      "learning_rate": 4.152365756634588e-06,
+      "loss": 0.6506,
+      "step": 34862
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.6438918113708496,
+      "learning_rate": 4.152105915937278e-06,
+      "loss": 0.6009,
+      "step": 34863
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 3.6821279525756836,
+      "learning_rate": 4.151846077597645e-06,
+      "loss": 0.4298,
+      "step": 34864
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 3.641225814819336,
+      "learning_rate": 4.15158624161641e-06,
+      "loss": 0.4392,
+      "step": 34865
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.6249891519546509,
+      "learning_rate": 4.151326407994296e-06,
+      "loss": 0.6264,
+      "step": 34866
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.1337199211120605,
+      "learning_rate": 4.151066576732025e-06,
+      "loss": 0.7429,
+      "step": 34867
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.6399227380752563,
+      "learning_rate": 4.15080674783032e-06,
+      "loss": 0.7318,
+      "step": 34868
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 3.17094087600708,
+      "learning_rate": 4.1505469212899036e-06,
+      "loss": 0.7399,
+      "step": 34869
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.778061628341675,
+      "learning_rate": 4.1502870971114975e-06,
+      "loss": 0.4954,
+      "step": 34870
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 3.8421218395233154,
+      "learning_rate": 4.150027275295825e-06,
+      "loss": 0.6259,
+      "step": 34871
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.811133623123169,
+      "learning_rate": 4.149767455843608e-06,
+      "loss": 0.7303,
+      "step": 34872
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.0695319175720215,
+      "learning_rate": 4.149507638755569e-06,
+      "loss": 0.631,
+      "step": 34873
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.0130836963653564,
+      "learning_rate": 4.149247824032431e-06,
+      "loss": 0.5525,
+      "step": 34874
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.639394521713257,
+      "learning_rate": 4.148988011674914e-06,
+      "loss": 1.0121,
+      "step": 34875
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.8296000957489014,
+      "learning_rate": 4.148728201683745e-06,
+      "loss": 0.7827,
+      "step": 34876
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.306483507156372,
+      "learning_rate": 4.1484683940596435e-06,
+      "loss": 0.5052,
+      "step": 34877
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 3.638847827911377,
+      "learning_rate": 4.148208588803333e-06,
+      "loss": 0.5647,
+      "step": 34878
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.1502339839935303,
+      "learning_rate": 4.147948785915534e-06,
+      "loss": 0.8561,
+      "step": 34879
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.8248928785324097,
+      "learning_rate": 4.147688985396972e-06,
+      "loss": 0.6779,
+      "step": 34880
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.6411018371582031,
+      "learning_rate": 4.1474291872483666e-06,
+      "loss": 0.5895,
+      "step": 34881
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 6.706393718719482,
+      "learning_rate": 4.147169391470441e-06,
+      "loss": 0.6383,
+      "step": 34882
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 3.001781940460205,
+      "learning_rate": 4.146909598063918e-06,
+      "loss": 0.6045,
+      "step": 34883
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.1041741371154785,
+      "learning_rate": 4.14664980702952e-06,
+      "loss": 0.8076,
+      "step": 34884
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.027385950088501,
+      "learning_rate": 4.1463900183679696e-06,
+      "loss": 0.6822,
+      "step": 34885
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.621683120727539,
+      "learning_rate": 4.146130232079988e-06,
+      "loss": 0.5535,
+      "step": 34886
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 3.192939281463623,
+      "learning_rate": 4.145870448166298e-06,
+      "loss": 0.3466,
+      "step": 34887
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.736018419265747,
+      "learning_rate": 4.145610666627623e-06,
+      "loss": 0.6946,
+      "step": 34888
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.9399484395980835,
+      "learning_rate": 4.145350887464685e-06,
+      "loss": 0.5747,
+      "step": 34889
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.6570229530334473,
+      "learning_rate": 4.145091110678206e-06,
+      "loss": 0.4847,
+      "step": 34890
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.232452869415283,
+      "learning_rate": 4.144831336268908e-06,
+      "loss": 0.6173,
+      "step": 34891
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.468263626098633,
+      "learning_rate": 4.144571564237514e-06,
+      "loss": 0.7524,
+      "step": 34892
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.8690800666809082,
+      "learning_rate": 4.144311794584746e-06,
+      "loss": 0.8118,
+      "step": 34893
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.054537534713745,
+      "learning_rate": 4.144052027311326e-06,
+      "loss": 0.5796,
+      "step": 34894
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.994357943534851,
+      "learning_rate": 4.143792262417977e-06,
+      "loss": 0.3591,
+      "step": 34895
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.2144060134887695,
+      "learning_rate": 4.14353249990542e-06,
+      "loss": 0.5809,
+      "step": 34896
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.7879528999328613,
+      "learning_rate": 4.14327273977438e-06,
+      "loss": 0.4728,
+      "step": 34897
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.3017454147338867,
+      "learning_rate": 4.143012982025576e-06,
+      "loss": 0.6239,
+      "step": 34898
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.982731580734253,
+      "learning_rate": 4.142753226659732e-06,
+      "loss": 0.7296,
+      "step": 34899
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.889925718307495,
+      "learning_rate": 4.1424934736775705e-06,
+      "loss": 0.4644,
+      "step": 34900
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.4794814586639404,
+      "learning_rate": 4.142233723079814e-06,
+      "loss": 0.6498,
+      "step": 34901
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.142345666885376,
+      "learning_rate": 4.141973974867183e-06,
+      "loss": 0.778,
+      "step": 34902
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 3.592315435409546,
+      "learning_rate": 4.141714229040402e-06,
+      "loss": 0.6718,
+      "step": 34903
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.9335615634918213,
+      "learning_rate": 4.141454485600192e-06,
+      "loss": 0.5273,
+      "step": 34904
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.7144267559051514,
+      "learning_rate": 4.141194744547276e-06,
+      "loss": 0.7702,
+      "step": 34905
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.552638530731201,
+      "learning_rate": 4.140935005882375e-06,
+      "loss": 0.5006,
+      "step": 34906
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 4.007106304168701,
+      "learning_rate": 4.140675269606212e-06,
+      "loss": 0.5959,
+      "step": 34907
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.7815669775009155,
+      "learning_rate": 4.14041553571951e-06,
+      "loss": 0.536,
+      "step": 34908
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.7754266262054443,
+      "learning_rate": 4.14015580422299e-06,
+      "loss": 0.4663,
+      "step": 34909
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.3197853565216064,
+      "learning_rate": 4.1398960751173736e-06,
+      "loss": 0.7034,
+      "step": 34910
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.8895294666290283,
+      "learning_rate": 4.1396363484033845e-06,
+      "loss": 0.4535,
+      "step": 34911
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.200410842895508,
+      "learning_rate": 4.139376624081744e-06,
+      "loss": 0.7572,
+      "step": 34912
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.2518723011016846,
+      "learning_rate": 4.139116902153176e-06,
+      "loss": 0.7297,
+      "step": 34913
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.830249309539795,
+      "learning_rate": 4.138857182618401e-06,
+      "loss": 0.7186,
+      "step": 34914
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.9914031028747559,
+      "learning_rate": 4.138597465478142e-06,
+      "loss": 0.5702,
+      "step": 34915
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.2797420024871826,
+      "learning_rate": 4.138337750733117e-06,
+      "loss": 0.6884,
+      "step": 34916
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.550361156463623,
+      "learning_rate": 4.138078038384055e-06,
+      "loss": 0.6393,
+      "step": 34917
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 3.2784390449523926,
+      "learning_rate": 4.137818328431675e-06,
+      "loss": 0.7108,
+      "step": 34918
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.3845016956329346,
+      "learning_rate": 4.1375586208767e-06,
+      "loss": 0.6899,
+      "step": 34919
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 3.802732229232788,
+      "learning_rate": 4.137298915719851e-06,
+      "loss": 0.6419,
+      "step": 34920
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.9396209716796875,
+      "learning_rate": 4.13703921296185e-06,
+      "loss": 0.5508,
+      "step": 34921
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.1937077045440674,
+      "learning_rate": 4.13677951260342e-06,
+      "loss": 0.9038,
+      "step": 34922
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.9376869201660156,
+      "learning_rate": 4.136519814645283e-06,
+      "loss": 0.7781,
+      "step": 34923
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.2873151302337646,
+      "learning_rate": 4.13626011908816e-06,
+      "loss": 0.5501,
+      "step": 34924
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 3.350593328475952,
+      "learning_rate": 4.136000425932775e-06,
+      "loss": 0.9877,
+      "step": 34925
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.7970757484436035,
+      "learning_rate": 4.135740735179848e-06,
+      "loss": 0.5567,
+      "step": 34926
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.8295071125030518,
+      "learning_rate": 4.135481046830104e-06,
+      "loss": 0.7721,
+      "step": 34927
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.2510383129119873,
+      "learning_rate": 4.135221360884262e-06,
+      "loss": 0.6453,
+      "step": 34928
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 4.502597332000732,
+      "learning_rate": 4.134961677343046e-06,
+      "loss": 0.5222,
+      "step": 34929
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.7016348838806152,
+      "learning_rate": 4.134701996207178e-06,
+      "loss": 0.6806,
+      "step": 34930
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.3973898887634277,
+      "learning_rate": 4.134442317477379e-06,
+      "loss": 0.619,
+      "step": 34931
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 3.1775903701782227,
+      "learning_rate": 4.134182641154372e-06,
+      "loss": 0.6233,
+      "step": 34932
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.753100633621216,
+      "learning_rate": 4.133922967238878e-06,
+      "loss": 0.5665,
+      "step": 34933
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.7778480052947998,
+      "learning_rate": 4.13366329573162e-06,
+      "loss": 0.784,
+      "step": 34934
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.416008949279785,
+      "learning_rate": 4.13340362663332e-06,
+      "loss": 0.8298,
+      "step": 34935
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.270521402359009,
+      "learning_rate": 4.133143959944699e-06,
+      "loss": 0.8664,
+      "step": 34936
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.8088085651397705,
+      "learning_rate": 4.13288429566648e-06,
+      "loss": 0.4846,
+      "step": 34937
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.856123924255371,
+      "learning_rate": 4.132624633799386e-06,
+      "loss": 0.5588,
+      "step": 34938
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.817826747894287,
+      "learning_rate": 4.132364974344138e-06,
+      "loss": 0.4886,
+      "step": 34939
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.779908299446106,
+      "learning_rate": 4.132105317301458e-06,
+      "loss": 0.5967,
+      "step": 34940
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 3.558849573135376,
+      "learning_rate": 4.131845662672067e-06,
+      "loss": 0.659,
+      "step": 34941
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.7226060628890991,
+      "learning_rate": 4.131586010456689e-06,
+      "loss": 0.6179,
+      "step": 34942
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.4531450271606445,
+      "learning_rate": 4.131326360656045e-06,
+      "loss": 0.8409,
+      "step": 34943
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 3.0271639823913574,
+      "learning_rate": 4.131066713270857e-06,
+      "loss": 0.5592,
+      "step": 34944
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 3.292372941970825,
+      "learning_rate": 4.130807068301846e-06,
+      "loss": 0.5521,
+      "step": 34945
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.818710207939148,
+      "learning_rate": 4.130547425749736e-06,
+      "loss": 0.4032,
+      "step": 34946
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 4.400223731994629,
+      "learning_rate": 4.130287785615248e-06,
+      "loss": 0.9472,
+      "step": 34947
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.0347204208374023,
+      "learning_rate": 4.1300281478991034e-06,
+      "loss": 0.5954,
+      "step": 34948
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.7550201416015625,
+      "learning_rate": 4.1297685126020235e-06,
+      "loss": 0.7602,
+      "step": 34949
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 3.1999809741973877,
+      "learning_rate": 4.129508879724732e-06,
+      "loss": 0.5606,
+      "step": 34950
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.2285945415496826,
+      "learning_rate": 4.129249249267951e-06,
+      "loss": 0.8327,
+      "step": 34951
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.835695266723633,
+      "learning_rate": 4.128989621232401e-06,
+      "loss": 0.6033,
+      "step": 34952
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.793874502182007,
+      "learning_rate": 4.1287299956188055e-06,
+      "loss": 0.4691,
+      "step": 34953
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.5753226280212402,
+      "learning_rate": 4.128470372427885e-06,
+      "loss": 0.5822,
+      "step": 34954
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 3.351433277130127,
+      "learning_rate": 4.1282107516603616e-06,
+      "loss": 0.6597,
+      "step": 34955
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 4.0482306480407715,
+      "learning_rate": 4.127951133316958e-06,
+      "loss": 0.8362,
+      "step": 34956
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 4.5543293952941895,
+      "learning_rate": 4.127691517398394e-06,
+      "loss": 0.9237,
+      "step": 34957
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.3925204277038574,
+      "learning_rate": 4.127431903905394e-06,
+      "loss": 0.5027,
+      "step": 34958
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.8242862224578857,
+      "learning_rate": 4.1271722928386794e-06,
+      "loss": 0.4447,
+      "step": 34959
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.9330748319625854,
+      "learning_rate": 4.1269126841989715e-06,
+      "loss": 0.6714,
+      "step": 34960
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 3.7393600940704346,
+      "learning_rate": 4.126653077986992e-06,
+      "loss": 0.6211,
+      "step": 34961
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.4147229194641113,
+      "learning_rate": 4.126393474203464e-06,
+      "loss": 0.5477,
+      "step": 34962
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.2769932746887207,
+      "learning_rate": 4.126133872849108e-06,
+      "loss": 0.4652,
+      "step": 34963
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.3307340145111084,
+      "learning_rate": 4.125874273924647e-06,
+      "loss": 0.617,
+      "step": 34964
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 4.803825855255127,
+      "learning_rate": 4.125614677430802e-06,
+      "loss": 0.8834,
+      "step": 34965
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.6821913719177246,
+      "learning_rate": 4.125355083368295e-06,
+      "loss": 0.7718,
+      "step": 34966
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.190460443496704,
+      "learning_rate": 4.125095491737847e-06,
+      "loss": 0.6168,
+      "step": 34967
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 3.0667974948883057,
+      "learning_rate": 4.124835902540182e-06,
+      "loss": 0.8548,
+      "step": 34968
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.9975816011428833,
+      "learning_rate": 4.12457631577602e-06,
+      "loss": 0.5289,
+      "step": 34969
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 3.092255115509033,
+      "learning_rate": 4.124316731446083e-06,
+      "loss": 0.7256,
+      "step": 34970
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.2640998363494873,
+      "learning_rate": 4.1240571495510935e-06,
+      "loss": 0.6207,
+      "step": 34971
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 4.144618034362793,
+      "learning_rate": 4.123797570091772e-06,
+      "loss": 0.7139,
+      "step": 34972
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.2479586601257324,
+      "learning_rate": 4.123537993068842e-06,
+      "loss": 0.6438,
+      "step": 34973
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.6687345504760742,
+      "learning_rate": 4.123278418483026e-06,
+      "loss": 0.5894,
+      "step": 34974
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.3237102031707764,
+      "learning_rate": 4.1230188463350415e-06,
+      "loss": 0.8406,
+      "step": 34975
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.804896831512451,
+      "learning_rate": 4.122759276625615e-06,
+      "loss": 0.8436,
+      "step": 34976
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.3288333415985107,
+      "learning_rate": 4.122499709355463e-06,
+      "loss": 0.5065,
+      "step": 34977
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.067582368850708,
+      "learning_rate": 4.122240144525314e-06,
+      "loss": 0.6834,
+      "step": 34978
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.394624948501587,
+      "learning_rate": 4.121980582135885e-06,
+      "loss": 0.8225,
+      "step": 34979
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.0752816200256348,
+      "learning_rate": 4.1217210221879e-06,
+      "loss": 0.3792,
+      "step": 34980
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.731351613998413,
+      "learning_rate": 4.1214614646820794e-06,
+      "loss": 0.4973,
+      "step": 34981
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.0294601917266846,
+      "learning_rate": 4.121201909619146e-06,
+      "loss": 0.6771,
+      "step": 34982
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 3.758218765258789,
+      "learning_rate": 4.12094235699982e-06,
+      "loss": 0.8661,
+      "step": 34983
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.9208316802978516,
+      "learning_rate": 4.1206828068248245e-06,
+      "loss": 0.7436,
+      "step": 34984
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.3668036460876465,
+      "learning_rate": 4.1204232590948806e-06,
+      "loss": 0.7147,
+      "step": 34985
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.0646440982818604,
+      "learning_rate": 4.1201637138107096e-06,
+      "loss": 0.6576,
+      "step": 34986
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.1910037994384766,
+      "learning_rate": 4.119904170973035e-06,
+      "loss": 0.5838,
+      "step": 34987
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 3.1558825969696045,
+      "learning_rate": 4.119644630582576e-06,
+      "loss": 0.9037,
+      "step": 34988
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.315964698791504,
+      "learning_rate": 4.119385092640056e-06,
+      "loss": 0.3974,
+      "step": 34989
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.4700756072998047,
+      "learning_rate": 4.1191255571461966e-06,
+      "loss": 0.5353,
+      "step": 34990
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 4.008283615112305,
+      "learning_rate": 4.118866024101718e-06,
+      "loss": 0.7052,
+      "step": 34991
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.9452333450317383,
+      "learning_rate": 4.1186064935073435e-06,
+      "loss": 0.365,
+      "step": 34992
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 3.821166753768921,
+      "learning_rate": 4.118346965363794e-06,
+      "loss": 0.4895,
+      "step": 34993
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 3.115678548812866,
+      "learning_rate": 4.1180874396717914e-06,
+      "loss": 0.4756,
+      "step": 34994
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.574817657470703,
+      "learning_rate": 4.117827916432058e-06,
+      "loss": 0.7969,
+      "step": 34995
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.5792057514190674,
+      "learning_rate": 4.117568395645314e-06,
+      "loss": 0.7263,
+      "step": 34996
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.9135509729385376,
+      "learning_rate": 4.117308877312279e-06,
+      "loss": 0.3812,
+      "step": 34997
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 3.458000659942627,
+      "learning_rate": 4.11704936143368e-06,
+      "loss": 0.7673,
+      "step": 34998
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 3.752685070037842,
+      "learning_rate": 4.116789848010236e-06,
+      "loss": 0.5574,
+      "step": 34999
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.6024365425109863,
+      "learning_rate": 4.116530337042669e-06,
+      "loss": 0.7353,
+      "step": 35000
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.253516435623169,
+      "learning_rate": 4.1162708285317e-06,
+      "loss": 0.564,
+      "step": 35001
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.0711669921875,
+      "learning_rate": 4.11601132247805e-06,
+      "loss": 0.561,
+      "step": 35002
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.596853017807007,
+      "learning_rate": 4.115751818882442e-06,
+      "loss": 0.738,
+      "step": 35003
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.9594722986221313,
+      "learning_rate": 4.115492317745597e-06,
+      "loss": 0.6749,
+      "step": 35004
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 6.2142205238342285,
+      "learning_rate": 4.115232819068237e-06,
+      "loss": 0.6597,
+      "step": 35005
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.1532022953033447,
+      "learning_rate": 4.114973322851082e-06,
+      "loss": 0.7573,
+      "step": 35006
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.9979335069656372,
+      "learning_rate": 4.114713829094855e-06,
+      "loss": 0.6324,
+      "step": 35007
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.5371017456054688,
+      "learning_rate": 4.114454337800278e-06,
+      "loss": 0.6354,
+      "step": 35008
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.9291930198669434,
+      "learning_rate": 4.114194848968071e-06,
+      "loss": 0.6655,
+      "step": 35009
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.517957925796509,
+      "learning_rate": 4.1139353625989555e-06,
+      "loss": 0.7168,
+      "step": 35010
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.8249205350875854,
+      "learning_rate": 4.113675878693654e-06,
+      "loss": 0.4044,
+      "step": 35011
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.857263445854187,
+      "learning_rate": 4.113416397252889e-06,
+      "loss": 0.5449,
+      "step": 35012
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.6451914310455322,
+      "learning_rate": 4.11315691827738e-06,
+      "loss": 0.9169,
+      "step": 35013
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.8087430000305176,
+      "learning_rate": 4.11289744176785e-06,
+      "loss": 0.4588,
+      "step": 35014
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.73398756980896,
+      "learning_rate": 4.112637967725019e-06,
+      "loss": 0.6083,
+      "step": 35015
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.066446304321289,
+      "learning_rate": 4.11237849614961e-06,
+      "loss": 0.6734,
+      "step": 35016
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.8692291975021362,
+      "learning_rate": 4.112119027042345e-06,
+      "loss": 0.5701,
+      "step": 35017
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.6234760284423828,
+      "learning_rate": 4.111859560403943e-06,
+      "loss": 0.6637,
+      "step": 35018
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.7093548774719238,
+      "learning_rate": 4.111600096235126e-06,
+      "loss": 0.567,
+      "step": 35019
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 3.264828681945801,
+      "learning_rate": 4.111340634536618e-06,
+      "loss": 0.8292,
+      "step": 35020
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 3.662890672683716,
+      "learning_rate": 4.1110811753091386e-06,
+      "loss": 0.8291,
+      "step": 35021
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.8850533962249756,
+      "learning_rate": 4.110821718553408e-06,
+      "loss": 0.4993,
+      "step": 35022
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.8833603858947754,
+      "learning_rate": 4.1105622642701506e-06,
+      "loss": 0.7,
+      "step": 35023
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.7539143562316895,
+      "learning_rate": 4.110302812460086e-06,
+      "loss": 0.6211,
+      "step": 35024
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.3698995113372803,
+      "learning_rate": 4.110043363123936e-06,
+      "loss": 0.7752,
+      "step": 35025
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.2783122062683105,
+      "learning_rate": 4.109783916262422e-06,
+      "loss": 0.5606,
+      "step": 35026
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 3.566997528076172,
+      "learning_rate": 4.1095244718762655e-06,
+      "loss": 0.9452,
+      "step": 35027
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.4470176696777344,
+      "learning_rate": 4.109265029966188e-06,
+      "loss": 0.8623,
+      "step": 35028
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.9695292711257935,
+      "learning_rate": 4.109005590532911e-06,
+      "loss": 0.573,
+      "step": 35029
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.614870309829712,
+      "learning_rate": 4.108746153577155e-06,
+      "loss": 0.7395,
+      "step": 35030
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.8649462461471558,
+      "learning_rate": 4.108486719099643e-06,
+      "loss": 0.5248,
+      "step": 35031
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.4744107723236084,
+      "learning_rate": 4.1082272871010945e-06,
+      "loss": 0.3991,
+      "step": 35032
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.7782344818115234,
+      "learning_rate": 4.107967857582233e-06,
+      "loss": 0.5129,
+      "step": 35033
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.9966070652008057,
+      "learning_rate": 4.107708430543777e-06,
+      "loss": 0.4474,
+      "step": 35034
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.5879461765289307,
+      "learning_rate": 4.107449005986451e-06,
+      "loss": 0.6674,
+      "step": 35035
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.8888601064682007,
+      "learning_rate": 4.107189583910974e-06,
+      "loss": 0.7025,
+      "step": 35036
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 3.250011920928955,
+      "learning_rate": 4.1069301643180695e-06,
+      "loss": 0.5375,
+      "step": 35037
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.399806261062622,
+      "learning_rate": 4.106670747208455e-06,
+      "loss": 0.68,
+      "step": 35038
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.4710220098495483,
+      "learning_rate": 4.1064113325828554e-06,
+      "loss": 0.3819,
+      "step": 35039
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.84651780128479,
+      "learning_rate": 4.1061519204419935e-06,
+      "loss": 0.629,
+      "step": 35040
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.1669275760650635,
+      "learning_rate": 4.105892510786586e-06,
+      "loss": 0.561,
+      "step": 35041
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.576664924621582,
+      "learning_rate": 4.105633103617358e-06,
+      "loss": 0.6404,
+      "step": 35042
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.4802334308624268,
+      "learning_rate": 4.105373698935029e-06,
+      "loss": 0.692,
+      "step": 35043
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.6533637046813965,
+      "learning_rate": 4.10511429674032e-06,
+      "loss": 0.4328,
+      "step": 35044
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.4578917026519775,
+      "learning_rate": 4.104854897033954e-06,
+      "loss": 0.5091,
+      "step": 35045
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 3.8095154762268066,
+      "learning_rate": 4.10459549981665e-06,
+      "loss": 0.4856,
+      "step": 35046
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 4.04259729385376,
+      "learning_rate": 4.104336105089131e-06,
+      "loss": 0.6821,
+      "step": 35047
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 3.1822023391723633,
+      "learning_rate": 4.104076712852118e-06,
+      "loss": 0.6168,
+      "step": 35048
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 4.163971424102783,
+      "learning_rate": 4.103817323106331e-06,
+      "loss": 0.8877,
+      "step": 35049
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 3.382267713546753,
+      "learning_rate": 4.103557935852494e-06,
+      "loss": 0.6392,
+      "step": 35050
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.4608232975006104,
+      "learning_rate": 4.103298551091326e-06,
+      "loss": 0.36,
+      "step": 35051
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 3.0393455028533936,
+      "learning_rate": 4.103039168823548e-06,
+      "loss": 0.7442,
+      "step": 35052
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.401517629623413,
+      "learning_rate": 4.102779789049883e-06,
+      "loss": 0.6394,
+      "step": 35053
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.1852846145629883,
+      "learning_rate": 4.10252041177105e-06,
+      "loss": 0.6326,
+      "step": 35054
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.5575252771377563,
+      "learning_rate": 4.102261036987773e-06,
+      "loss": 0.5174,
+      "step": 35055
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.7344090938568115,
+      "learning_rate": 4.102001664700771e-06,
+      "loss": 0.7079,
+      "step": 35056
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.1347568035125732,
+      "learning_rate": 4.101742294910767e-06,
+      "loss": 0.6053,
+      "step": 35057
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.827277421951294,
+      "learning_rate": 4.101482927618477e-06,
+      "loss": 0.7373,
+      "step": 35058
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 4.917786121368408,
+      "learning_rate": 4.10122356282463e-06,
+      "loss": 0.6613,
+      "step": 35059
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.174384355545044,
+      "learning_rate": 4.1009642005299436e-06,
+      "loss": 0.7069,
+      "step": 35060
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.3140153884887695,
+      "learning_rate": 4.100704840735138e-06,
+      "loss": 0.6051,
+      "step": 35061
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 3.182692289352417,
+      "learning_rate": 4.100445483440937e-06,
+      "loss": 0.5564,
+      "step": 35062
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.1165196895599365,
+      "learning_rate": 4.100186128648059e-06,
+      "loss": 0.6229,
+      "step": 35063
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 3.736325740814209,
+      "learning_rate": 4.099926776357226e-06,
+      "loss": 0.85,
+      "step": 35064
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.352461338043213,
+      "learning_rate": 4.09966742656916e-06,
+      "loss": 0.4527,
+      "step": 35065
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 3.1073014736175537,
+      "learning_rate": 4.099408079284581e-06,
+      "loss": 0.6117,
+      "step": 35066
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.4593310356140137,
+      "learning_rate": 4.099148734504212e-06,
+      "loss": 0.9463,
+      "step": 35067
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.289186954498291,
+      "learning_rate": 4.098889392228772e-06,
+      "loss": 0.5601,
+      "step": 35068
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.7088470458984375,
+      "learning_rate": 4.098630052458984e-06,
+      "loss": 0.6858,
+      "step": 35069
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 8.263252258300781,
+      "learning_rate": 4.0983707151955665e-06,
+      "loss": 0.5461,
+      "step": 35070
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.1102254390716553,
+      "learning_rate": 4.098111380439243e-06,
+      "loss": 0.7988,
+      "step": 35071
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 4.304808139801025,
+      "learning_rate": 4.097852048190734e-06,
+      "loss": 1.0311,
+      "step": 35072
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.2975614070892334,
+      "learning_rate": 4.09759271845076e-06,
+      "loss": 0.5717,
+      "step": 35073
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 3.057701349258423,
+      "learning_rate": 4.097333391220043e-06,
+      "loss": 0.6603,
+      "step": 35074
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.0949034690856934,
+      "learning_rate": 4.097074066499304e-06,
+      "loss": 0.8013,
+      "step": 35075
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 3.0838711261749268,
+      "learning_rate": 4.096814744289264e-06,
+      "loss": 0.3816,
+      "step": 35076
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 4.31422758102417,
+      "learning_rate": 4.096555424590643e-06,
+      "loss": 0.9943,
+      "step": 35077
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.469388961791992,
+      "learning_rate": 4.096296107404164e-06,
+      "loss": 0.6526,
+      "step": 35078
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.7688276767730713,
+      "learning_rate": 4.096036792730545e-06,
+      "loss": 0.5849,
+      "step": 35079
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.126803398132324,
+      "learning_rate": 4.09577748057051e-06,
+      "loss": 0.5748,
+      "step": 35080
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.0521631240844727,
+      "learning_rate": 4.09551817092478e-06,
+      "loss": 0.5799,
+      "step": 35081
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.5963330268859863,
+      "learning_rate": 4.095258863794075e-06,
+      "loss": 0.5696,
+      "step": 35082
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.8044333457946777,
+      "learning_rate": 4.094999559179116e-06,
+      "loss": 0.6658,
+      "step": 35083
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 4.3793535232543945,
+      "learning_rate": 4.094740257080623e-06,
+      "loss": 0.6989,
+      "step": 35084
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 3.245461940765381,
+      "learning_rate": 4.094480957499319e-06,
+      "loss": 0.6796,
+      "step": 35085
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.5777682065963745,
+      "learning_rate": 4.094221660435925e-06,
+      "loss": 0.4789,
+      "step": 35086
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.9683698415756226,
+      "learning_rate": 4.09396236589116e-06,
+      "loss": 0.7168,
+      "step": 35087
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.6748952865600586,
+      "learning_rate": 4.093703073865747e-06,
+      "loss": 0.6129,
+      "step": 35088
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.618069648742676,
+      "learning_rate": 4.093443784360407e-06,
+      "loss": 0.8004,
+      "step": 35089
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 3.890019178390503,
+      "learning_rate": 4.093184497375859e-06,
+      "loss": 0.7343,
+      "step": 35090
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.1062073707580566,
+      "learning_rate": 4.092925212912825e-06,
+      "loss": 0.6908,
+      "step": 35091
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.9522321224212646,
+      "learning_rate": 4.0926659309720276e-06,
+      "loss": 0.6645,
+      "step": 35092
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.4250588417053223,
+      "learning_rate": 4.092406651554186e-06,
+      "loss": 0.6498,
+      "step": 35093
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.8244085311889648,
+      "learning_rate": 4.092147374660021e-06,
+      "loss": 0.266,
+      "step": 35094
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 3.75968861579895,
+      "learning_rate": 4.091888100290255e-06,
+      "loss": 0.5313,
+      "step": 35095
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 3.0559651851654053,
+      "learning_rate": 4.091628828445606e-06,
+      "loss": 0.8851,
+      "step": 35096
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.9169139862060547,
+      "learning_rate": 4.091369559126799e-06,
+      "loss": 0.5194,
+      "step": 35097
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 4.159146308898926,
+      "learning_rate": 4.091110292334551e-06,
+      "loss": 0.452,
+      "step": 35098
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.538914442062378,
+      "learning_rate": 4.0908510280695845e-06,
+      "loss": 0.5096,
+      "step": 35099
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 3.9394278526306152,
+      "learning_rate": 4.0905917663326225e-06,
+      "loss": 0.5806,
+      "step": 35100
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.367323875427246,
+      "learning_rate": 4.090332507124383e-06,
+      "loss": 0.7992,
+      "step": 35101
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.5169739723205566,
+      "learning_rate": 4.090073250445589e-06,
+      "loss": 0.7275,
+      "step": 35102
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 6.902306079864502,
+      "learning_rate": 4.08981399629696e-06,
+      "loss": 0.66,
+      "step": 35103
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.5520310401916504,
+      "learning_rate": 4.089554744679217e-06,
+      "loss": 0.6042,
+      "step": 35104
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.65739369392395,
+      "learning_rate": 4.089295495593082e-06,
+      "loss": 0.4315,
+      "step": 35105
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.9936140775680542,
+      "learning_rate": 4.089036249039275e-06,
+      "loss": 0.6251,
+      "step": 35106
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.6816565990448,
+      "learning_rate": 4.088777005018516e-06,
+      "loss": 0.8454,
+      "step": 35107
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 3.145648241043091,
+      "learning_rate": 4.0885177635315285e-06,
+      "loss": 0.6538,
+      "step": 35108
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 4.177059173583984,
+      "learning_rate": 4.08825852457903e-06,
+      "loss": 0.7779,
+      "step": 35109
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 4.053732872009277,
+      "learning_rate": 4.087999288161744e-06,
+      "loss": 0.4836,
+      "step": 35110
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.885597586631775,
+      "learning_rate": 4.08774005428039e-06,
+      "loss": 0.6416,
+      "step": 35111
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 3.147219181060791,
+      "learning_rate": 4.087480822935689e-06,
+      "loss": 0.3091,
+      "step": 35112
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 5.4349541664123535,
+      "learning_rate": 4.0872215941283625e-06,
+      "loss": 0.7766,
+      "step": 35113
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.9434661865234375,
+      "learning_rate": 4.08696236785913e-06,
+      "loss": 0.5602,
+      "step": 35114
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 3.1363422870635986,
+      "learning_rate": 4.086703144128714e-06,
+      "loss": 0.7426,
+      "step": 35115
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.8042160272598267,
+      "learning_rate": 4.0864439229378334e-06,
+      "loss": 0.4511,
+      "step": 35116
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.3555235862731934,
+      "learning_rate": 4.086184704287211e-06,
+      "loss": 0.601,
+      "step": 35117
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 3.4723105430603027,
+      "learning_rate": 4.0859254881775655e-06,
+      "loss": 0.6571,
+      "step": 35118
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.2797164916992188,
+      "learning_rate": 4.085666274609618e-06,
+      "loss": 0.5552,
+      "step": 35119
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 4.721323013305664,
+      "learning_rate": 4.085407063584092e-06,
+      "loss": 0.6816,
+      "step": 35120
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 3.210752487182617,
+      "learning_rate": 4.0851478551017055e-06,
+      "loss": 0.5206,
+      "step": 35121
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.9760024547576904,
+      "learning_rate": 4.08488864916318e-06,
+      "loss": 0.7264,
+      "step": 35122
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.042396068572998,
+      "learning_rate": 4.084629445769237e-06,
+      "loss": 0.673,
+      "step": 35123
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.7207540273666382,
+      "learning_rate": 4.084370244920597e-06,
+      "loss": 0.6746,
+      "step": 35124
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.3558156490325928,
+      "learning_rate": 4.084111046617979e-06,
+      "loss": 0.3666,
+      "step": 35125
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.9849519729614258,
+      "learning_rate": 4.083851850862106e-06,
+      "loss": 0.8117,
+      "step": 35126
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 3.3006575107574463,
+      "learning_rate": 4.083592657653699e-06,
+      "loss": 0.8487,
+      "step": 35127
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.442716121673584,
+      "learning_rate": 4.083333466993475e-06,
+      "loss": 0.5428,
+      "step": 35128
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.8023786544799805,
+      "learning_rate": 4.0830742788821586e-06,
+      "loss": 0.5217,
+      "step": 35129
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.931111454963684,
+      "learning_rate": 4.082815093320469e-06,
+      "loss": 0.6446,
+      "step": 35130
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 3.3364901542663574,
+      "learning_rate": 4.082555910309127e-06,
+      "loss": 0.963,
+      "step": 35131
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 3.3948190212249756,
+      "learning_rate": 4.082296729848853e-06,
+      "loss": 0.6656,
+      "step": 35132
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.9828802347183228,
+      "learning_rate": 4.08203755194037e-06,
+      "loss": 0.6674,
+      "step": 35133
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.708338975906372,
+      "learning_rate": 4.081778376584393e-06,
+      "loss": 0.5231,
+      "step": 35134
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 3.4414103031158447,
+      "learning_rate": 4.081519203781649e-06,
+      "loss": 0.6176,
+      "step": 35135
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 4.741391181945801,
+      "learning_rate": 4.081260033532855e-06,
+      "loss": 0.6435,
+      "step": 35136
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.9347658157348633,
+      "learning_rate": 4.081000865838733e-06,
+      "loss": 0.7364,
+      "step": 35137
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.2220826148986816,
+      "learning_rate": 4.080741700700004e-06,
+      "loss": 0.7199,
+      "step": 35138
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.1540889739990234,
+      "learning_rate": 4.080482538117387e-06,
+      "loss": 0.5927,
+      "step": 35139
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.4269237518310547,
+      "learning_rate": 4.0802233780916035e-06,
+      "loss": 0.6989,
+      "step": 35140
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.964616060256958,
+      "learning_rate": 4.079964220623375e-06,
+      "loss": 0.6419,
+      "step": 35141
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.438321113586426,
+      "learning_rate": 4.07970506571342e-06,
+      "loss": 0.7077,
+      "step": 35142
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.9389965534210205,
+      "learning_rate": 4.079445913362463e-06,
+      "loss": 0.4895,
+      "step": 35143
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.0117874145507812,
+      "learning_rate": 4.07918676357122e-06,
+      "loss": 0.5363,
+      "step": 35144
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 3.5490779876708984,
+      "learning_rate": 4.078927616340414e-06,
+      "loss": 0.723,
+      "step": 35145
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.0343377590179443,
+      "learning_rate": 4.078668471670765e-06,
+      "loss": 0.4923,
+      "step": 35146
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.049192190170288,
+      "learning_rate": 4.078409329562995e-06,
+      "loss": 0.5805,
+      "step": 35147
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 4.156976699829102,
+      "learning_rate": 4.0781501900178226e-06,
+      "loss": 0.4892,
+      "step": 35148
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.7532426118850708,
+      "learning_rate": 4.077891053035969e-06,
+      "loss": 0.5963,
+      "step": 35149
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.9477834701538086,
+      "learning_rate": 4.077631918618156e-06,
+      "loss": 0.7845,
+      "step": 35150
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 3.5711092948913574,
+      "learning_rate": 4.077372786765102e-06,
+      "loss": 0.5923,
+      "step": 35151
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.7614456415176392,
+      "learning_rate": 4.0771136574775294e-06,
+      "loss": 0.4716,
+      "step": 35152
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.347355842590332,
+      "learning_rate": 4.076854530756158e-06,
+      "loss": 0.605,
+      "step": 35153
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 3.7540597915649414,
+      "learning_rate": 4.076595406601707e-06,
+      "loss": 0.6302,
+      "step": 35154
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.8616969585418701,
+      "learning_rate": 4.0763362850149e-06,
+      "loss": 0.6177,
+      "step": 35155
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 3.373552083969116,
+      "learning_rate": 4.076077165996455e-06,
+      "loss": 0.7934,
+      "step": 35156
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.4048843383789062,
+      "learning_rate": 4.075818049547094e-06,
+      "loss": 0.6089,
+      "step": 35157
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 4.270305633544922,
+      "learning_rate": 4.075558935667536e-06,
+      "loss": 0.5849,
+      "step": 35158
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.5511746406555176,
+      "learning_rate": 4.0752998243585024e-06,
+      "loss": 0.5565,
+      "step": 35159
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.7731752395629883,
+      "learning_rate": 4.075040715620712e-06,
+      "loss": 0.5958,
+      "step": 35160
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.5312581062316895,
+      "learning_rate": 4.074781609454888e-06,
+      "loss": 0.6179,
+      "step": 35161
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.652137279510498,
+      "learning_rate": 4.0745225058617505e-06,
+      "loss": 0.476,
+      "step": 35162
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 5.203002452850342,
+      "learning_rate": 4.074263404842019e-06,
+      "loss": 0.3913,
+      "step": 35163
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.7829341888427734,
+      "learning_rate": 4.074004306396414e-06,
+      "loss": 0.6627,
+      "step": 35164
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 3.2584128379821777,
+      "learning_rate": 4.073745210525657e-06,
+      "loss": 0.6008,
+      "step": 35165
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 3.4800679683685303,
+      "learning_rate": 4.073486117230467e-06,
+      "loss": 0.8338,
+      "step": 35166
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.890143632888794,
+      "learning_rate": 4.0732270265115645e-06,
+      "loss": 0.6145,
+      "step": 35167
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.2348592281341553,
+      "learning_rate": 4.072967938369671e-06,
+      "loss": 0.6749,
+      "step": 35168
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.0096709728240967,
+      "learning_rate": 4.072708852805507e-06,
+      "loss": 0.5327,
+      "step": 35169
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.3892719745635986,
+      "learning_rate": 4.072449769819791e-06,
+      "loss": 0.6893,
+      "step": 35170
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.8395546674728394,
+      "learning_rate": 4.0721906894132455e-06,
+      "loss": 0.6119,
+      "step": 35171
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 3.618298053741455,
+      "learning_rate": 4.07193161158659e-06,
+      "loss": 0.9187,
+      "step": 35172
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 3.159533977508545,
+      "learning_rate": 4.071672536340545e-06,
+      "loss": 0.6937,
+      "step": 35173
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.238339900970459,
+      "learning_rate": 4.0714134636758306e-06,
+      "loss": 0.5719,
+      "step": 35174
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.9933871030807495,
+      "learning_rate": 4.071154393593168e-06,
+      "loss": 0.7023,
+      "step": 35175
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 3.913799285888672,
+      "learning_rate": 4.0708953260932774e-06,
+      "loss": 0.6247,
+      "step": 35176
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.032935619354248,
+      "learning_rate": 4.070636261176878e-06,
+      "loss": 0.6165,
+      "step": 35177
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.9805179834365845,
+      "learning_rate": 4.0703771988446914e-06,
+      "loss": 0.7511,
+      "step": 35178
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 4.431562423706055,
+      "learning_rate": 4.070118139097438e-06,
+      "loss": 0.619,
+      "step": 35179
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.3777236938476562,
+      "learning_rate": 4.069859081935835e-06,
+      "loss": 0.6977,
+      "step": 35180
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.2639667987823486,
+      "learning_rate": 4.0696000273606075e-06,
+      "loss": 0.577,
+      "step": 35181
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.021934747695923,
+      "learning_rate": 4.069340975372474e-06,
+      "loss": 0.519,
+      "step": 35182
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.420860767364502,
+      "learning_rate": 4.069081925972155e-06,
+      "loss": 0.5509,
+      "step": 35183
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.213308095932007,
+      "learning_rate": 4.068822879160369e-06,
+      "loss": 0.4435,
+      "step": 35184
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 3.0099916458129883,
+      "learning_rate": 4.068563834937839e-06,
+      "loss": 0.6873,
+      "step": 35185
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.0383079051971436,
+      "learning_rate": 4.068304793305285e-06,
+      "loss": 0.3863,
+      "step": 35186
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.1421070098876953,
+      "learning_rate": 4.068045754263424e-06,
+      "loss": 0.6872,
+      "step": 35187
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.604734420776367,
+      "learning_rate": 4.067786717812981e-06,
+      "loss": 0.7193,
+      "step": 35188
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.8793981075286865,
+      "learning_rate": 4.0675276839546715e-06,
+      "loss": 0.6782,
+      "step": 35189
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.53928279876709,
+      "learning_rate": 4.06726865268922e-06,
+      "loss": 0.7646,
+      "step": 35190
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.2551627159118652,
+      "learning_rate": 4.067009624017344e-06,
+      "loss": 0.632,
+      "step": 35191
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.9170254468917847,
+      "learning_rate": 4.066750597939765e-06,
+      "loss": 0.5182,
+      "step": 35192
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.235482692718506,
+      "learning_rate": 4.066491574457204e-06,
+      "loss": 0.6598,
+      "step": 35193
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 3.0359508991241455,
+      "learning_rate": 4.066232553570379e-06,
+      "loss": 0.507,
+      "step": 35194
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 3.511671543121338,
+      "learning_rate": 4.065973535280012e-06,
+      "loss": 0.6871,
+      "step": 35195
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.0274956226348877,
+      "learning_rate": 4.065714519586822e-06,
+      "loss": 0.6546,
+      "step": 35196
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.4865872859954834,
+      "learning_rate": 4.0654555064915294e-06,
+      "loss": 0.5653,
+      "step": 35197
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 3.4970245361328125,
+      "learning_rate": 4.0651964959948564e-06,
+      "loss": 0.6186,
+      "step": 35198
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.905241012573242,
+      "learning_rate": 4.064937488097521e-06,
+      "loss": 0.8349,
+      "step": 35199
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.288198471069336,
+      "learning_rate": 4.064678482800245e-06,
+      "loss": 0.5282,
+      "step": 35200
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.109405040740967,
+      "learning_rate": 4.064419480103746e-06,
+      "loss": 0.8731,
+      "step": 35201
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 3.397554636001587,
+      "learning_rate": 4.064160480008748e-06,
+      "loss": 0.8741,
+      "step": 35202
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.9358069896697998,
+      "learning_rate": 4.063901482515968e-06,
+      "loss": 0.7412,
+      "step": 35203
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.980215311050415,
+      "learning_rate": 4.063642487626127e-06,
+      "loss": 0.6035,
+      "step": 35204
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.9989798069000244,
+      "learning_rate": 4.063383495339946e-06,
+      "loss": 0.45,
+      "step": 35205
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.5462422370910645,
+      "learning_rate": 4.063124505658145e-06,
+      "loss": 0.4884,
+      "step": 35206
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 4.3287458419799805,
+      "learning_rate": 4.0628655185814425e-06,
+      "loss": 0.8211,
+      "step": 35207
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.2258355617523193,
+      "learning_rate": 4.062606534110562e-06,
+      "loss": 0.4488,
+      "step": 35208
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 3.0104899406433105,
+      "learning_rate": 4.0623475522462195e-06,
+      "loss": 0.782,
+      "step": 35209
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 3.1117241382598877,
+      "learning_rate": 4.062088572989139e-06,
+      "loss": 0.6495,
+      "step": 35210
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.8735363483428955,
+      "learning_rate": 4.061829596340038e-06,
+      "loss": 0.5131,
+      "step": 35211
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.783482551574707,
+      "learning_rate": 4.061570622299638e-06,
+      "loss": 0.8855,
+      "step": 35212
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 3.390068292617798,
+      "learning_rate": 4.061311650868658e-06,
+      "loss": 0.6207,
+      "step": 35213
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.483281135559082,
+      "learning_rate": 4.0610526820478185e-06,
+      "loss": 0.7937,
+      "step": 35214
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.000828742980957,
+      "learning_rate": 4.06079371583784e-06,
+      "loss": 0.5551,
+      "step": 35215
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 4.871030330657959,
+      "learning_rate": 4.060534752239442e-06,
+      "loss": 0.7328,
+      "step": 35216
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.0107903480529785,
+      "learning_rate": 4.060275791253345e-06,
+      "loss": 0.5086,
+      "step": 35217
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.27917218208313,
+      "learning_rate": 4.06001683288027e-06,
+      "loss": 0.4206,
+      "step": 35218
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.6787893772125244,
+      "learning_rate": 4.059757877120935e-06,
+      "loss": 0.544,
+      "step": 35219
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.6258528232574463,
+      "learning_rate": 4.059498923976062e-06,
+      "loss": 0.4339,
+      "step": 35220
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.5354888439178467,
+      "learning_rate": 4.059239973446367e-06,
+      "loss": 0.7264,
+      "step": 35221
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 3.4131181240081787,
+      "learning_rate": 4.058981025532576e-06,
+      "loss": 0.4841,
+      "step": 35222
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.5490288734436035,
+      "learning_rate": 4.058722080235406e-06,
+      "loss": 0.3274,
+      "step": 35223
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.22529673576355,
+      "learning_rate": 4.058463137555578e-06,
+      "loss": 0.5991,
+      "step": 35224
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 4.938087463378906,
+      "learning_rate": 4.05820419749381e-06,
+      "loss": 0.6908,
+      "step": 35225
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.371514320373535,
+      "learning_rate": 4.057945260050824e-06,
+      "loss": 0.6397,
+      "step": 35226
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 3.147179126739502,
+      "learning_rate": 4.05768632522734e-06,
+      "loss": 0.8291,
+      "step": 35227
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.921333074569702,
+      "learning_rate": 4.057427393024076e-06,
+      "loss": 0.7597,
+      "step": 35228
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.7898294925689697,
+      "learning_rate": 4.057168463441754e-06,
+      "loss": 0.5304,
+      "step": 35229
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.338233709335327,
+      "learning_rate": 4.056909536481094e-06,
+      "loss": 0.65,
+      "step": 35230
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.714717388153076,
+      "learning_rate": 4.0566506121428145e-06,
+      "loss": 0.5011,
+      "step": 35231
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.4896044731140137,
+      "learning_rate": 4.056391690427636e-06,
+      "loss": 0.6447,
+      "step": 35232
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.1159757375717163,
+      "learning_rate": 4.05613277133628e-06,
+      "loss": 0.4158,
+      "step": 35233
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 3.3221888542175293,
+      "learning_rate": 4.055873854869464e-06,
+      "loss": 0.6934,
+      "step": 35234
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 3.3460166454315186,
+      "learning_rate": 4.0556149410279085e-06,
+      "loss": 1.0705,
+      "step": 35235
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.7695930004119873,
+      "learning_rate": 4.055356029812336e-06,
+      "loss": 0.7205,
+      "step": 35236
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.8571863174438477,
+      "learning_rate": 4.055097121223463e-06,
+      "loss": 0.5075,
+      "step": 35237
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.200329303741455,
+      "learning_rate": 4.054838215262012e-06,
+      "loss": 0.5598,
+      "step": 35238
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.472778797149658,
+      "learning_rate": 4.054579311928701e-06,
+      "loss": 0.5503,
+      "step": 35239
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.711969256401062,
+      "learning_rate": 4.054320411224251e-06,
+      "loss": 0.5951,
+      "step": 35240
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.322303056716919,
+      "learning_rate": 4.05406151314938e-06,
+      "loss": 0.6405,
+      "step": 35241
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.8798117637634277,
+      "learning_rate": 4.053802617704811e-06,
+      "loss": 0.738,
+      "step": 35242
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.503688097000122,
+      "learning_rate": 4.0535437248912625e-06,
+      "loss": 0.5604,
+      "step": 35243
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 3.037158727645874,
+      "learning_rate": 4.053284834709454e-06,
+      "loss": 0.7921,
+      "step": 35244
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.5785083770751953,
+      "learning_rate": 4.053025947160107e-06,
+      "loss": 0.692,
+      "step": 35245
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.253655433654785,
+      "learning_rate": 4.052767062243939e-06,
+      "loss": 0.5651,
+      "step": 35246
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.1915557384490967,
+      "learning_rate": 4.052508179961672e-06,
+      "loss": 0.6236,
+      "step": 35247
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.5564095973968506,
+      "learning_rate": 4.052249300314024e-06,
+      "loss": 0.7324,
+      "step": 35248
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 3.109466075897217,
+      "learning_rate": 4.051990423301716e-06,
+      "loss": 0.5623,
+      "step": 35249
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.7736576795578003,
+      "learning_rate": 4.0517315489254675e-06,
+      "loss": 0.6229,
+      "step": 35250
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.4506571292877197,
+      "learning_rate": 4.051472677185998e-06,
+      "loss": 1.0454,
+      "step": 35251
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.161525011062622,
+      "learning_rate": 4.051213808084028e-06,
+      "loss": 0.5232,
+      "step": 35252
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 4.067620277404785,
+      "learning_rate": 4.050954941620277e-06,
+      "loss": 0.5445,
+      "step": 35253
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.4152629375457764,
+      "learning_rate": 4.050696077795465e-06,
+      "loss": 0.4655,
+      "step": 35254
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.990457057952881,
+      "learning_rate": 4.0504372166103115e-06,
+      "loss": 0.7027,
+      "step": 35255
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.6552882194519043,
+      "learning_rate": 4.050178358065536e-06,
+      "loss": 0.5412,
+      "step": 35256
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.786404848098755,
+      "learning_rate": 4.049919502161858e-06,
+      "loss": 0.6998,
+      "step": 35257
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.6722238063812256,
+      "learning_rate": 4.049660648899998e-06,
+      "loss": 0.368,
+      "step": 35258
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.992481231689453,
+      "learning_rate": 4.049401798280677e-06,
+      "loss": 0.7601,
+      "step": 35259
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.43623948097229,
+      "learning_rate": 4.049142950304613e-06,
+      "loss": 0.5335,
+      "step": 35260
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.9004571437835693,
+      "learning_rate": 4.048884104972526e-06,
+      "loss": 0.4878,
+      "step": 35261
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.8637702465057373,
+      "learning_rate": 4.048625262285135e-06,
+      "loss": 0.7516,
+      "step": 35262
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.0497429370880127,
+      "learning_rate": 4.048366422243162e-06,
+      "loss": 0.5225,
+      "step": 35263
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.7669520378112793,
+      "learning_rate": 4.048107584847326e-06,
+      "loss": 0.8044,
+      "step": 35264
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 5.525844573974609,
+      "learning_rate": 4.047848750098345e-06,
+      "loss": 0.3981,
+      "step": 35265
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 3.084594964981079,
+      "learning_rate": 4.047589917996941e-06,
+      "loss": 0.6257,
+      "step": 35266
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.6713697910308838,
+      "learning_rate": 4.047331088543832e-06,
+      "loss": 0.3906,
+      "step": 35267
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 3.0656633377075195,
+      "learning_rate": 4.047072261739738e-06,
+      "loss": 0.7683,
+      "step": 35268
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.248964786529541,
+      "learning_rate": 4.04681343758538e-06,
+      "loss": 0.6094,
+      "step": 35269
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.225364923477173,
+      "learning_rate": 4.046554616081476e-06,
+      "loss": 0.6064,
+      "step": 35270
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 3.491852045059204,
+      "learning_rate": 4.046295797228747e-06,
+      "loss": 1.0618,
+      "step": 35271
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.2465364933013916,
+      "learning_rate": 4.046036981027912e-06,
+      "loss": 0.5294,
+      "step": 35272
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.814887762069702,
+      "learning_rate": 4.045778167479691e-06,
+      "loss": 0.6581,
+      "step": 35273
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.0739030838012695,
+      "learning_rate": 4.0455193565848025e-06,
+      "loss": 0.6044,
+      "step": 35274
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.873722791671753,
+      "learning_rate": 4.045260548343969e-06,
+      "loss": 0.6344,
+      "step": 35275
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.766068696975708,
+      "learning_rate": 4.045001742757907e-06,
+      "loss": 0.6831,
+      "step": 35276
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 3.751391887664795,
+      "learning_rate": 4.044742939827337e-06,
+      "loss": 0.9393,
+      "step": 35277
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.87641978263855,
+      "learning_rate": 4.04448413955298e-06,
+      "loss": 0.5014,
+      "step": 35278
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.9289203882217407,
+      "learning_rate": 4.044225341935554e-06,
+      "loss": 0.6356,
+      "step": 35279
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.1474738121032715,
+      "learning_rate": 4.0439665469757796e-06,
+      "loss": 0.6118,
+      "step": 35280
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.4479540586471558,
+      "learning_rate": 4.043707754674376e-06,
+      "loss": 0.5031,
+      "step": 35281
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 3.2980172634124756,
+      "learning_rate": 4.043448965032061e-06,
+      "loss": 0.7463,
+      "step": 35282
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.2281715869903564,
+      "learning_rate": 4.0431901780495585e-06,
+      "loss": 0.887,
+      "step": 35283
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 3.135141134262085,
+      "learning_rate": 4.042931393727585e-06,
+      "loss": 0.4634,
+      "step": 35284
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.9713138341903687,
+      "learning_rate": 4.042672612066861e-06,
+      "loss": 0.7005,
+      "step": 35285
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 3.745673894882202,
+      "learning_rate": 4.042413833068106e-06,
+      "loss": 0.4472,
+      "step": 35286
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.5601942539215088,
+      "learning_rate": 4.0421550567320385e-06,
+      "loss": 0.4677,
+      "step": 35287
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.0234806537628174,
+      "learning_rate": 4.04189628305938e-06,
+      "loss": 0.7005,
+      "step": 35288
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.13769268989563,
+      "learning_rate": 4.041637512050849e-06,
+      "loss": 0.5486,
+      "step": 35289
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 3.0935792922973633,
+      "learning_rate": 4.041378743707165e-06,
+      "loss": 0.7325,
+      "step": 35290
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.1443371772766113,
+      "learning_rate": 4.041119978029047e-06,
+      "loss": 0.7269,
+      "step": 35291
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.553660273551941,
+      "learning_rate": 4.040861215017216e-06,
+      "loss": 0.5588,
+      "step": 35292
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.356555461883545,
+      "learning_rate": 4.04060245467239e-06,
+      "loss": 0.9237,
+      "step": 35293
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.1830809116363525,
+      "learning_rate": 4.040343696995289e-06,
+      "loss": 0.5782,
+      "step": 35294
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.228870391845703,
+      "learning_rate": 4.040084941986634e-06,
+      "loss": 0.6689,
+      "step": 35295
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.8320837020874023,
+      "learning_rate": 4.0398261896471415e-06,
+      "loss": 0.559,
+      "step": 35296
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 3.468132495880127,
+      "learning_rate": 4.039567439977534e-06,
+      "loss": 0.3417,
+      "step": 35297
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.475919246673584,
+      "learning_rate": 4.039308692978529e-06,
+      "loss": 0.6698,
+      "step": 35298
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.6974725723266602,
+      "learning_rate": 4.039049948650847e-06,
+      "loss": 0.5647,
+      "step": 35299
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.006981372833252,
+      "learning_rate": 4.038791206995207e-06,
+      "loss": 0.5468,
+      "step": 35300
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.079515218734741,
+      "learning_rate": 4.0385324680123286e-06,
+      "loss": 0.6113,
+      "step": 35301
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.855491876602173,
+      "learning_rate": 4.0382737317029286e-06,
+      "loss": 0.7841,
+      "step": 35302
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.2390670776367188,
+      "learning_rate": 4.038014998067731e-06,
+      "loss": 0.5548,
+      "step": 35303
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.841089963912964,
+      "learning_rate": 4.037756267107454e-06,
+      "loss": 0.8202,
+      "step": 35304
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.5091986656188965,
+      "learning_rate": 4.037497538822816e-06,
+      "loss": 0.3127,
+      "step": 35305
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.383855104446411,
+      "learning_rate": 4.0372388132145375e-06,
+      "loss": 0.3433,
+      "step": 35306
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.5219783782958984,
+      "learning_rate": 4.036980090283337e-06,
+      "loss": 0.4892,
+      "step": 35307
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 3.2738735675811768,
+      "learning_rate": 4.036721370029933e-06,
+      "loss": 0.6572,
+      "step": 35308
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.6597511768341064,
+      "learning_rate": 4.036462652455048e-06,
+      "loss": 0.639,
+      "step": 35309
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.640817642211914,
+      "learning_rate": 4.036203937559397e-06,
+      "loss": 0.5763,
+      "step": 35310
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.037290573120117,
+      "learning_rate": 4.035945225343704e-06,
+      "loss": 0.5968,
+      "step": 35311
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.402339458465576,
+      "learning_rate": 4.035686515808685e-06,
+      "loss": 0.6452,
+      "step": 35312
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.623225450515747,
+      "learning_rate": 4.035427808955061e-06,
+      "loss": 0.5439,
+      "step": 35313
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.5934622287750244,
+      "learning_rate": 4.035169104783551e-06,
+      "loss": 0.4832,
+      "step": 35314
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.7392191886901855,
+      "learning_rate": 4.034910403294874e-06,
+      "loss": 0.8924,
+      "step": 35315
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.891270637512207,
+      "learning_rate": 4.034651704489749e-06,
+      "loss": 0.7858,
+      "step": 35316
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.4207713603973389,
+      "learning_rate": 4.034393008368897e-06,
+      "loss": 0.7495,
+      "step": 35317
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 6.609468936920166,
+      "learning_rate": 4.034134314933036e-06,
+      "loss": 0.667,
+      "step": 35318
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.9863070249557495,
+      "learning_rate": 4.033875624182885e-06,
+      "loss": 0.7039,
+      "step": 35319
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.112990140914917,
+      "learning_rate": 4.033616936119164e-06,
+      "loss": 0.688,
+      "step": 35320
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 3.664433240890503,
+      "learning_rate": 4.033358250742593e-06,
+      "loss": 0.7996,
+      "step": 35321
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.356764793395996,
+      "learning_rate": 4.0330995680538905e-06,
+      "loss": 0.498,
+      "step": 35322
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 7.465587615966797,
+      "learning_rate": 4.0328408880537746e-06,
+      "loss": 0.3695,
+      "step": 35323
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.224142551422119,
+      "learning_rate": 4.032582210742967e-06,
+      "loss": 0.3181,
+      "step": 35324
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 4.7944207191467285,
+      "learning_rate": 4.032323536122186e-06,
+      "loss": 0.939,
+      "step": 35325
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.878777027130127,
+      "learning_rate": 4.0320648641921515e-06,
+      "loss": 0.85,
+      "step": 35326
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.300570011138916,
+      "learning_rate": 4.031806194953581e-06,
+      "loss": 0.63,
+      "step": 35327
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.923606514930725,
+      "learning_rate": 4.031547528407195e-06,
+      "loss": 0.6758,
+      "step": 35328
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.7841943502426147,
+      "learning_rate": 4.031288864553712e-06,
+      "loss": 0.6435,
+      "step": 35329
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.88654363155365,
+      "learning_rate": 4.031030203393852e-06,
+      "loss": 0.5425,
+      "step": 35330
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.6483330726623535,
+      "learning_rate": 4.030771544928334e-06,
+      "loss": 0.5837,
+      "step": 35331
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.4991531372070312,
+      "learning_rate": 4.030512889157878e-06,
+      "loss": 0.5914,
+      "step": 35332
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 4.765017986297607,
+      "learning_rate": 4.030254236083202e-06,
+      "loss": 0.4029,
+      "step": 35333
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.114382266998291,
+      "learning_rate": 4.029995585705025e-06,
+      "loss": 0.4512,
+      "step": 35334
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.278036117553711,
+      "learning_rate": 4.029736938024068e-06,
+      "loss": 0.5588,
+      "step": 35335
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.0011675357818604,
+      "learning_rate": 4.029478293041049e-06,
+      "loss": 0.6111,
+      "step": 35336
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.2641143798828125,
+      "learning_rate": 4.029219650756687e-06,
+      "loss": 0.776,
+      "step": 35337
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.0869054794311523,
+      "learning_rate": 4.028961011171701e-06,
+      "loss": 0.554,
+      "step": 35338
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.0882112979888916,
+      "learning_rate": 4.028702374286812e-06,
+      "loss": 0.5328,
+      "step": 35339
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.993563175201416,
+      "learning_rate": 4.028443740102737e-06,
+      "loss": 0.7346,
+      "step": 35340
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.5772593021392822,
+      "learning_rate": 4.028185108620196e-06,
+      "loss": 0.536,
+      "step": 35341
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.9129278659820557,
+      "learning_rate": 4.027926479839909e-06,
+      "loss": 0.7058,
+      "step": 35342
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.4882121086120605,
+      "learning_rate": 4.027667853762591e-06,
+      "loss": 0.4893,
+      "step": 35343
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.059211254119873,
+      "learning_rate": 4.027409230388966e-06,
+      "loss": 0.7248,
+      "step": 35344
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 3.2482991218566895,
+      "learning_rate": 4.027150609719754e-06,
+      "loss": 0.7288,
+      "step": 35345
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.0741260051727295,
+      "learning_rate": 4.02689199175567e-06,
+      "loss": 0.694,
+      "step": 35346
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.4387476444244385,
+      "learning_rate": 4.026633376497435e-06,
+      "loss": 0.5681,
+      "step": 35347
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 3.633209705352783,
+      "learning_rate": 4.026374763945768e-06,
+      "loss": 0.5911,
+      "step": 35348
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.5857605934143066,
+      "learning_rate": 4.026116154101388e-06,
+      "loss": 0.7786,
+      "step": 35349
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.4346508979797363,
+      "learning_rate": 4.025857546965015e-06,
+      "loss": 0.5756,
+      "step": 35350
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 3.2616653442382812,
+      "learning_rate": 4.025598942537367e-06,
+      "loss": 0.5063,
+      "step": 35351
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.9457340240478516,
+      "learning_rate": 4.025340340819163e-06,
+      "loss": 0.4648,
+      "step": 35352
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.923182725906372,
+      "learning_rate": 4.025081741811122e-06,
+      "loss": 0.5366,
+      "step": 35353
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.3661956787109375,
+      "learning_rate": 4.024823145513964e-06,
+      "loss": 0.6241,
+      "step": 35354
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.1760165691375732,
+      "learning_rate": 4.024564551928407e-06,
+      "loss": 0.5982,
+      "step": 35355
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 3.02649188041687,
+      "learning_rate": 4.02430596105517e-06,
+      "loss": 0.9465,
+      "step": 35356
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.1865177154541016,
+      "learning_rate": 4.024047372894974e-06,
+      "loss": 0.5395,
+      "step": 35357
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 3.755781650543213,
+      "learning_rate": 4.023788787448536e-06,
+      "loss": 0.5831,
+      "step": 35358
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 4.865081310272217,
+      "learning_rate": 4.023530204716576e-06,
+      "loss": 0.5569,
+      "step": 35359
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.935636281967163,
+      "learning_rate": 4.023271624699812e-06,
+      "loss": 0.5156,
+      "step": 35360
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 3.042628765106201,
+      "learning_rate": 4.023013047398965e-06,
+      "loss": 0.7419,
+      "step": 35361
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 3.514906644821167,
+      "learning_rate": 4.0227544728147515e-06,
+      "loss": 0.914,
+      "step": 35362
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 3.0296406745910645,
+      "learning_rate": 4.02249590094789e-06,
+      "loss": 0.5809,
+      "step": 35363
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.951066255569458,
+      "learning_rate": 4.022237331799104e-06,
+      "loss": 0.622,
+      "step": 35364
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.3373281955718994,
+      "learning_rate": 4.021978765369109e-06,
+      "loss": 0.6161,
+      "step": 35365
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 6.2768635749816895,
+      "learning_rate": 4.021720201658625e-06,
+      "loss": 0.4769,
+      "step": 35366
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.8505940437316895,
+      "learning_rate": 4.02146164066837e-06,
+      "loss": 0.5598,
+      "step": 35367
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.8350045680999756,
+      "learning_rate": 4.021203082399065e-06,
+      "loss": 0.5215,
+      "step": 35368
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.7482773065567017,
+      "learning_rate": 4.0209445268514255e-06,
+      "loss": 0.6573,
+      "step": 35369
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.85697078704834,
+      "learning_rate": 4.020685974026174e-06,
+      "loss": 0.7119,
+      "step": 35370
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.501220703125,
+      "learning_rate": 4.020427423924028e-06,
+      "loss": 0.6863,
+      "step": 35371
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 3.3292202949523926,
+      "learning_rate": 4.0201688765457055e-06,
+      "loss": 0.8576,
+      "step": 35372
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 3.2659454345703125,
+      "learning_rate": 4.019910331891927e-06,
+      "loss": 0.8133,
+      "step": 35373
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 3.7163240909576416,
+      "learning_rate": 4.01965178996341e-06,
+      "loss": 0.6736,
+      "step": 35374
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.578824520111084,
+      "learning_rate": 4.019393250760874e-06,
+      "loss": 0.2988,
+      "step": 35375
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.0980093479156494,
+      "learning_rate": 4.01913471428504e-06,
+      "loss": 0.6062,
+      "step": 35376
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.8250555992126465,
+      "learning_rate": 4.018876180536623e-06,
+      "loss": 0.477,
+      "step": 35377
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 3.5405211448669434,
+      "learning_rate": 4.018617649516344e-06,
+      "loss": 0.6262,
+      "step": 35378
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.4276185035705566,
+      "learning_rate": 4.018359121224923e-06,
+      "loss": 0.5899,
+      "step": 35379
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.7172157764434814,
+      "learning_rate": 4.0181005956630754e-06,
+      "loss": 0.4193,
+      "step": 35380
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 5.786319255828857,
+      "learning_rate": 4.017842072831523e-06,
+      "loss": 0.5803,
+      "step": 35381
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 3.0233917236328125,
+      "learning_rate": 4.017583552730984e-06,
+      "loss": 0.8211,
+      "step": 35382
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.9464199542999268,
+      "learning_rate": 4.017325035362178e-06,
+      "loss": 0.9481,
+      "step": 35383
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 3.8930068016052246,
+      "learning_rate": 4.0170665207258206e-06,
+      "loss": 0.5711,
+      "step": 35384
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.3270814418792725,
+      "learning_rate": 4.016808008822635e-06,
+      "loss": 0.5658,
+      "step": 35385
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.1571671962738037,
+      "learning_rate": 4.016549499653337e-06,
+      "loss": 0.6779,
+      "step": 35386
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.9945719242095947,
+      "learning_rate": 4.016290993218648e-06,
+      "loss": 0.5337,
+      "step": 35387
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.6321778297424316,
+      "learning_rate": 4.016032489519284e-06,
+      "loss": 0.6545,
+      "step": 35388
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.0666937828063965,
+      "learning_rate": 4.015773988555966e-06,
+      "loss": 0.605,
+      "step": 35389
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.9137072563171387,
+      "learning_rate": 4.015515490329411e-06,
+      "loss": 0.5322,
+      "step": 35390
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 4.407649993896484,
+      "learning_rate": 4.015256994840339e-06,
+      "loss": 0.6531,
+      "step": 35391
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.3796005249023438,
+      "learning_rate": 4.014998502089467e-06,
+      "loss": 0.7639,
+      "step": 35392
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.6843955516815186,
+      "learning_rate": 4.014740012077517e-06,
+      "loss": 0.8306,
+      "step": 35393
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.7907668352127075,
+      "learning_rate": 4.0144815248052045e-06,
+      "loss": 0.3936,
+      "step": 35394
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.4312613010406494,
+      "learning_rate": 4.014223040273251e-06,
+      "loss": 0.2592,
+      "step": 35395
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.1913788318634033,
+      "learning_rate": 4.013964558482374e-06,
+      "loss": 0.5998,
+      "step": 35396
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.3650456666946411,
+      "learning_rate": 4.013706079433292e-06,
+      "loss": 0.81,
+      "step": 35397
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.1035094261169434,
+      "learning_rate": 4.013447603126723e-06,
+      "loss": 0.8276,
+      "step": 35398
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.3280603885650635,
+      "learning_rate": 4.013189129563388e-06,
+      "loss": 0.5425,
+      "step": 35399
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.8367438316345215,
+      "learning_rate": 4.0129306587440034e-06,
+      "loss": 0.3952,
+      "step": 35400
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.7488698959350586,
+      "learning_rate": 4.01267219066929e-06,
+      "loss": 0.7971,
+      "step": 35401
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.8838669061660767,
+      "learning_rate": 4.012413725339964e-06,
+      "loss": 0.49,
+      "step": 35402
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.212918281555176,
+      "learning_rate": 4.012155262756746e-06,
+      "loss": 0.4725,
+      "step": 35403
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 3.6211090087890625,
+      "learning_rate": 4.011896802920352e-06,
+      "loss": 0.8287,
+      "step": 35404
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.2203311920166016,
+      "learning_rate": 4.011638345831505e-06,
+      "loss": 0.4618,
+      "step": 35405
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.281308174133301,
+      "learning_rate": 4.011379891490921e-06,
+      "loss": 0.5143,
+      "step": 35406
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 3.2466208934783936,
+      "learning_rate": 4.01112143989932e-06,
+      "loss": 0.5518,
+      "step": 35407
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.6284193992614746,
+      "learning_rate": 4.01086299105742e-06,
+      "loss": 0.7036,
+      "step": 35408
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.9682655334472656,
+      "learning_rate": 4.010604544965938e-06,
+      "loss": 0.8604,
+      "step": 35409
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 3.367147207260132,
+      "learning_rate": 4.010346101625596e-06,
+      "loss": 0.5144,
+      "step": 35410
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.823136329650879,
+      "learning_rate": 4.01008766103711e-06,
+      "loss": 0.5693,
+      "step": 35411
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.7693796157836914,
+      "learning_rate": 4.009829223201199e-06,
+      "loss": 0.5587,
+      "step": 35412
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.8384147882461548,
+      "learning_rate": 4.009570788118582e-06,
+      "loss": 0.7841,
+      "step": 35413
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.1278791427612305,
+      "learning_rate": 4.009312355789978e-06,
+      "loss": 0.649,
+      "step": 35414
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.275322675704956,
+      "learning_rate": 4.0090539262161054e-06,
+      "loss": 0.6132,
+      "step": 35415
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.394629716873169,
+      "learning_rate": 4.008795499397682e-06,
+      "loss": 0.58,
+      "step": 35416
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.1786723136901855,
+      "learning_rate": 4.008537075335428e-06,
+      "loss": 0.5684,
+      "step": 35417
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.544971227645874,
+      "learning_rate": 4.00827865403006e-06,
+      "loss": 0.3529,
+      "step": 35418
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.055713415145874,
+      "learning_rate": 4.008020235482298e-06,
+      "loss": 0.7354,
+      "step": 35419
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.65995192527771,
+      "learning_rate": 4.00776181969286e-06,
+      "loss": 0.7626,
+      "step": 35420
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.5617926120758057,
+      "learning_rate": 4.0075034066624645e-06,
+      "loss": 0.4169,
+      "step": 35421
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.5184566974639893,
+      "learning_rate": 4.00724499639183e-06,
+      "loss": 0.6184,
+      "step": 35422
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.2408299446105957,
+      "learning_rate": 4.006986588881676e-06,
+      "loss": 0.734,
+      "step": 35423
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 3.9174563884735107,
+      "learning_rate": 4.006728184132718e-06,
+      "loss": 0.6406,
+      "step": 35424
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 3.6991090774536133,
+      "learning_rate": 4.006469782145679e-06,
+      "loss": 0.7172,
+      "step": 35425
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.5842783451080322,
+      "learning_rate": 4.006211382921275e-06,
+      "loss": 0.6164,
+      "step": 35426
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 4.017849445343018,
+      "learning_rate": 4.005952986460225e-06,
+      "loss": 0.6129,
+      "step": 35427
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 4.440933704376221,
+      "learning_rate": 4.005694592763247e-06,
+      "loss": 0.7239,
+      "step": 35428
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.2139554023742676,
+      "learning_rate": 4.0054362018310614e-06,
+      "loss": 0.6039,
+      "step": 35429
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.272301435470581,
+      "learning_rate": 4.005177813664384e-06,
+      "loss": 0.3986,
+      "step": 35430
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.0525498390197754,
+      "learning_rate": 4.004919428263934e-06,
+      "loss": 0.6301,
+      "step": 35431
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.141524076461792,
+      "learning_rate": 4.0046610456304305e-06,
+      "loss": 0.8009,
+      "step": 35432
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.9849849939346313,
+      "learning_rate": 4.004402665764592e-06,
+      "loss": 0.7874,
+      "step": 35433
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 3.545600175857544,
+      "learning_rate": 4.004144288667137e-06,
+      "loss": 0.7044,
+      "step": 35434
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.6318533420562744,
+      "learning_rate": 4.003885914338783e-06,
+      "loss": 0.6995,
+      "step": 35435
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 3.9630913734436035,
+      "learning_rate": 4.00362754278025e-06,
+      "loss": 0.7693,
+      "step": 35436
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.532898187637329,
+      "learning_rate": 4.0033691739922554e-06,
+      "loss": 0.379,
+      "step": 35437
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.9895949363708496,
+      "learning_rate": 4.0031108079755174e-06,
+      "loss": 0.4878,
+      "step": 35438
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.301273822784424,
+      "learning_rate": 4.002852444730755e-06,
+      "loss": 0.6921,
+      "step": 35439
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.3586976528167725,
+      "learning_rate": 4.002594084258686e-06,
+      "loss": 0.6263,
+      "step": 35440
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.8066935539245605,
+      "learning_rate": 4.002335726560029e-06,
+      "loss": 0.6958,
+      "step": 35441
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.1828958988189697,
+      "learning_rate": 4.002077371635503e-06,
+      "loss": 0.5361,
+      "step": 35442
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.0395047664642334,
+      "learning_rate": 4.001819019485825e-06,
+      "loss": 0.8372,
+      "step": 35443
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.8595025539398193,
+      "learning_rate": 4.001560670111716e-06,
+      "loss": 0.5911,
+      "step": 35444
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.2282307147979736,
+      "learning_rate": 4.001302323513892e-06,
+      "loss": 0.4895,
+      "step": 35445
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 3.8940999507904053,
+      "learning_rate": 4.001043979693072e-06,
+      "loss": 0.5108,
+      "step": 35446
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.856296420097351,
+      "learning_rate": 4.000785638649975e-06,
+      "loss": 0.5826,
+      "step": 35447
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.9453520774841309,
+      "learning_rate": 4.000527300385318e-06,
+      "loss": 0.3955,
+      "step": 35448
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.6597893238067627,
+      "learning_rate": 4.000268964899821e-06,
+      "loss": 0.4668,
+      "step": 35449
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.074481725692749,
+      "learning_rate": 4.000010632194201e-06,
+      "loss": 0.5663,
+      "step": 35450
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 3.138423204421997,
+      "learning_rate": 3.999752302269178e-06,
+      "loss": 0.8995,
+      "step": 35451
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.8524765968322754,
+      "learning_rate": 3.999493975125468e-06,
+      "loss": 0.7978,
+      "step": 35452
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.917785406112671,
+      "learning_rate": 3.99923565076379e-06,
+      "loss": 0.5424,
+      "step": 35453
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.320878505706787,
+      "learning_rate": 3.998977329184863e-06,
+      "loss": 0.7186,
+      "step": 35454
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 4.506786823272705,
+      "learning_rate": 3.998719010389406e-06,
+      "loss": 0.6416,
+      "step": 35455
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.5227367877960205,
+      "learning_rate": 3.998460694378136e-06,
+      "loss": 0.5086,
+      "step": 35456
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.319608211517334,
+      "learning_rate": 3.9982023811517715e-06,
+      "loss": 0.7449,
+      "step": 35457
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.7204830646514893,
+      "learning_rate": 3.99794407071103e-06,
+      "loss": 0.9376,
+      "step": 35458
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.9578068256378174,
+      "learning_rate": 3.997685763056632e-06,
+      "loss": 0.4881,
+      "step": 35459
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.068768262863159,
+      "learning_rate": 3.997427458189294e-06,
+      "loss": 0.4423,
+      "step": 35460
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.0156803131103516,
+      "learning_rate": 3.997169156109735e-06,
+      "loss": 0.5336,
+      "step": 35461
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 6.246429920196533,
+      "learning_rate": 3.996910856818672e-06,
+      "loss": 0.5994,
+      "step": 35462
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.2370071411132812,
+      "learning_rate": 3.996652560316825e-06,
+      "loss": 0.5056,
+      "step": 35463
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 3.718796968460083,
+      "learning_rate": 3.996394266604911e-06,
+      "loss": 0.6113,
+      "step": 35464
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 8.952288627624512,
+      "learning_rate": 3.996135975683647e-06,
+      "loss": 0.5902,
+      "step": 35465
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 3.767580270767212,
+      "learning_rate": 3.995877687553754e-06,
+      "loss": 0.5206,
+      "step": 35466
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 4.639126777648926,
+      "learning_rate": 3.9956194022159496e-06,
+      "loss": 0.9908,
+      "step": 35467
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 5.559257507324219,
+      "learning_rate": 3.995361119670951e-06,
+      "loss": 0.4523,
+      "step": 35468
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.214510440826416,
+      "learning_rate": 3.995102839919477e-06,
+      "loss": 0.6777,
+      "step": 35469
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.802384376525879,
+      "learning_rate": 3.9948445629622455e-06,
+      "loss": 0.577,
+      "step": 35470
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.3686487674713135,
+      "learning_rate": 3.994586288799975e-06,
+      "loss": 0.6834,
+      "step": 35471
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.720852017402649,
+      "learning_rate": 3.994328017433384e-06,
+      "loss": 0.2584,
+      "step": 35472
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.913878321647644,
+      "learning_rate": 3.994069748863189e-06,
+      "loss": 0.4671,
+      "step": 35473
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.8147032260894775,
+      "learning_rate": 3.993811483090109e-06,
+      "loss": 0.5447,
+      "step": 35474
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.90458083152771,
+      "learning_rate": 3.993553220114863e-06,
+      "loss": 0.6617,
+      "step": 35475
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.4160780906677246,
+      "learning_rate": 3.993294959938169e-06,
+      "loss": 0.6526,
+      "step": 35476
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.1568758487701416,
+      "learning_rate": 3.993036702560743e-06,
+      "loss": 0.7145,
+      "step": 35477
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.9470418691635132,
+      "learning_rate": 3.992778447983306e-06,
+      "loss": 0.4248,
+      "step": 35478
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.0772674083709717,
+      "learning_rate": 3.992520196206575e-06,
+      "loss": 0.2922,
+      "step": 35479
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.6828041076660156,
+      "learning_rate": 3.992261947231267e-06,
+      "loss": 0.7179,
+      "step": 35480
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 3.761963367462158,
+      "learning_rate": 3.992003701058101e-06,
+      "loss": 0.5794,
+      "step": 35481
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.733492136001587,
+      "learning_rate": 3.991745457687795e-06,
+      "loss": 0.6976,
+      "step": 35482
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 3.1675126552581787,
+      "learning_rate": 3.9914872171210685e-06,
+      "loss": 0.5068,
+      "step": 35483
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.303696632385254,
+      "learning_rate": 3.991228979358637e-06,
+      "loss": 0.5044,
+      "step": 35484
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.4171650409698486,
+      "learning_rate": 3.990970744401218e-06,
+      "loss": 0.8286,
+      "step": 35485
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.695521354675293,
+      "learning_rate": 3.990712512249534e-06,
+      "loss": 0.3732,
+      "step": 35486
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.417036533355713,
+      "learning_rate": 3.990454282904299e-06,
+      "loss": 0.6143,
+      "step": 35487
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.3965723514556885,
+      "learning_rate": 3.990196056366235e-06,
+      "loss": 0.7152,
+      "step": 35488
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.6498606204986572,
+      "learning_rate": 3.9899378326360545e-06,
+      "loss": 0.6689,
+      "step": 35489
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.210169553756714,
+      "learning_rate": 3.989679611714481e-06,
+      "loss": 0.7439,
+      "step": 35490
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 4.585049152374268,
+      "learning_rate": 3.989421393602229e-06,
+      "loss": 0.7044,
+      "step": 35491
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.400339126586914,
+      "learning_rate": 3.989163178300017e-06,
+      "loss": 0.6941,
+      "step": 35492
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.8171378374099731,
+      "learning_rate": 3.988904965808564e-06,
+      "loss": 0.6208,
+      "step": 35493
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 3.9188144207000732,
+      "learning_rate": 3.988646756128588e-06,
+      "loss": 0.6278,
+      "step": 35494
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.8446435928344727,
+      "learning_rate": 3.988388549260806e-06,
+      "loss": 0.4734,
+      "step": 35495
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.547300338745117,
+      "learning_rate": 3.988130345205936e-06,
+      "loss": 0.5437,
+      "step": 35496
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 3.1976187229156494,
+      "learning_rate": 3.9878721439646975e-06,
+      "loss": 0.4185,
+      "step": 35497
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.982898235321045,
+      "learning_rate": 3.987613945537807e-06,
+      "loss": 0.8066,
+      "step": 35498
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.0852997303009033,
+      "learning_rate": 3.987355749925983e-06,
+      "loss": 0.8065,
+      "step": 35499
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 3.153916358947754,
+      "learning_rate": 3.987097557129943e-06,
+      "loss": 0.4691,
+      "step": 35500
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.288167953491211,
+      "learning_rate": 3.986839367150405e-06,
+      "loss": 0.5007,
+      "step": 35501
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.338304042816162,
+      "learning_rate": 3.986581179988087e-06,
+      "loss": 0.6073,
+      "step": 35502
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.4256982803344727,
+      "learning_rate": 3.9863229956437086e-06,
+      "loss": 0.7418,
+      "step": 35503
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.981975793838501,
+      "learning_rate": 3.986064814117984e-06,
+      "loss": 0.6292,
+      "step": 35504
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.634488821029663,
+      "learning_rate": 3.985806635411635e-06,
+      "loss": 0.7269,
+      "step": 35505
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.6269640922546387,
+      "learning_rate": 3.9855484595253755e-06,
+      "loss": 0.447,
+      "step": 35506
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 3.052537202835083,
+      "learning_rate": 3.985290286459928e-06,
+      "loss": 0.599,
+      "step": 35507
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.6078276634216309,
+      "learning_rate": 3.985032116216008e-06,
+      "loss": 0.7979,
+      "step": 35508
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.435962677001953,
+      "learning_rate": 3.984773948794333e-06,
+      "loss": 0.6505,
+      "step": 35509
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.9401789903640747,
+      "learning_rate": 3.984515784195622e-06,
+      "loss": 0.687,
+      "step": 35510
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.882447600364685,
+      "learning_rate": 3.984257622420592e-06,
+      "loss": 0.4319,
+      "step": 35511
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.763213872909546,
+      "learning_rate": 3.98399946346996e-06,
+      "loss": 0.3942,
+      "step": 35512
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 3.1600799560546875,
+      "learning_rate": 3.983741307344446e-06,
+      "loss": 0.669,
+      "step": 35513
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.191418409347534,
+      "learning_rate": 3.983483154044766e-06,
+      "loss": 0.7227,
+      "step": 35514
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.7278552055358887,
+      "learning_rate": 3.983225003571639e-06,
+      "loss": 0.4906,
+      "step": 35515
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 3.397843837738037,
+      "learning_rate": 3.982966855925781e-06,
+      "loss": 0.6111,
+      "step": 35516
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.518465995788574,
+      "learning_rate": 3.982708711107913e-06,
+      "loss": 0.6383,
+      "step": 35517
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.9015758037567139,
+      "learning_rate": 3.982450569118751e-06,
+      "loss": 0.7313,
+      "step": 35518
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.109661340713501,
+      "learning_rate": 3.982192429959012e-06,
+      "loss": 0.6195,
+      "step": 35519
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.7134032249450684,
+      "learning_rate": 3.981934293629414e-06,
+      "loss": 0.6313,
+      "step": 35520
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 3.8354244232177734,
+      "learning_rate": 3.981676160130676e-06,
+      "loss": 0.6578,
+      "step": 35521
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.629671812057495,
+      "learning_rate": 3.981418029463516e-06,
+      "loss": 0.8798,
+      "step": 35522
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 3.057764768600464,
+      "learning_rate": 3.98115990162865e-06,
+      "loss": 0.6882,
+      "step": 35523
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 3.8010640144348145,
+      "learning_rate": 3.980901776626797e-06,
+      "loss": 0.9067,
+      "step": 35524
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.024662971496582,
+      "learning_rate": 3.980643654458674e-06,
+      "loss": 0.7216,
+      "step": 35525
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.249922275543213,
+      "learning_rate": 3.980385535124999e-06,
+      "loss": 0.5996,
+      "step": 35526
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 3.003040075302124,
+      "learning_rate": 3.980127418626489e-06,
+      "loss": 0.5992,
+      "step": 35527
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.0569329261779785,
+      "learning_rate": 3.979869304963864e-06,
+      "loss": 0.5638,
+      "step": 35528
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.9825618267059326,
+      "learning_rate": 3.97961119413784e-06,
+      "loss": 0.6282,
+      "step": 35529
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.7516106367111206,
+      "learning_rate": 3.979353086149134e-06,
+      "loss": 0.6123,
+      "step": 35530
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 3.263308048248291,
+      "learning_rate": 3.979094980998467e-06,
+      "loss": 0.5265,
+      "step": 35531
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.7757744789123535,
+      "learning_rate": 3.978836878686553e-06,
+      "loss": 0.7658,
+      "step": 35532
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.3109347820281982,
+      "learning_rate": 3.978578779214112e-06,
+      "loss": 0.4603,
+      "step": 35533
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.3045146465301514,
+      "learning_rate": 3.97832068258186e-06,
+      "loss": 0.6615,
+      "step": 35534
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 3.0078821182250977,
+      "learning_rate": 3.978062588790516e-06,
+      "loss": 0.7359,
+      "step": 35535
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.2662980556488037,
+      "learning_rate": 3.977804497840796e-06,
+      "loss": 0.7045,
+      "step": 35536
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.686603546142578,
+      "learning_rate": 3.97754640973342e-06,
+      "loss": 0.6819,
+      "step": 35537
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 6.579143047332764,
+      "learning_rate": 3.977288324469104e-06,
+      "loss": 0.4508,
+      "step": 35538
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 3.1028361320495605,
+      "learning_rate": 3.977030242048566e-06,
+      "loss": 0.6074,
+      "step": 35539
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.488741159439087,
+      "learning_rate": 3.9767721624725235e-06,
+      "loss": 0.6456,
+      "step": 35540
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.982844352722168,
+      "learning_rate": 3.976514085741694e-06,
+      "loss": 0.5402,
+      "step": 35541
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 3.824979305267334,
+      "learning_rate": 3.976256011856797e-06,
+      "loss": 0.4825,
+      "step": 35542
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.6408401727676392,
+      "learning_rate": 3.975997940818547e-06,
+      "loss": 0.6192,
+      "step": 35543
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 4.50908899307251,
+      "learning_rate": 3.975739872627664e-06,
+      "loss": 0.8501,
+      "step": 35544
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.0720086097717285,
+      "learning_rate": 3.975481807284864e-06,
+      "loss": 0.6567,
+      "step": 35545
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.189560651779175,
+      "learning_rate": 3.975223744790863e-06,
+      "loss": 0.5149,
+      "step": 35546
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 18.836315155029297,
+      "learning_rate": 3.974965685146383e-06,
+      "loss": 0.6327,
+      "step": 35547
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.8444573879241943,
+      "learning_rate": 3.974707628352141e-06,
+      "loss": 0.5862,
+      "step": 35548
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.0139522552490234,
+      "learning_rate": 3.974449574408852e-06,
+      "loss": 0.5132,
+      "step": 35549
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.1811306476593018,
+      "learning_rate": 3.974191523317234e-06,
+      "loss": 0.3413,
+      "step": 35550
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.979606866836548,
+      "learning_rate": 3.9739334750780056e-06,
+      "loss": 0.7754,
+      "step": 35551
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.5174636840820312,
+      "learning_rate": 3.973675429691884e-06,
+      "loss": 0.7929,
+      "step": 35552
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.7207562923431396,
+      "learning_rate": 3.973417387159586e-06,
+      "loss": 0.7756,
+      "step": 35553
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 3.810734272003174,
+      "learning_rate": 3.97315934748183e-06,
+      "loss": 0.4906,
+      "step": 35554
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 7.0522661209106445,
+      "learning_rate": 3.972901310659334e-06,
+      "loss": 0.793,
+      "step": 35555
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.8266180753707886,
+      "learning_rate": 3.972643276692813e-06,
+      "loss": 0.6167,
+      "step": 35556
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 3.081127643585205,
+      "learning_rate": 3.972385245582987e-06,
+      "loss": 0.5498,
+      "step": 35557
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.7963441610336304,
+      "learning_rate": 3.972127217330574e-06,
+      "loss": 0.6223,
+      "step": 35558
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.983502984046936,
+      "learning_rate": 3.971869191936289e-06,
+      "loss": 0.7732,
+      "step": 35559
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 3.5172502994537354,
+      "learning_rate": 3.97161116940085e-06,
+      "loss": 0.7413,
+      "step": 35560
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.8407235145568848,
+      "learning_rate": 3.9713531497249755e-06,
+      "loss": 0.7219,
+      "step": 35561
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 3.574791669845581,
+      "learning_rate": 3.971095132909383e-06,
+      "loss": 0.8076,
+      "step": 35562
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.7951397895812988,
+      "learning_rate": 3.970837118954789e-06,
+      "loss": 0.5504,
+      "step": 35563
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 3.862712860107422,
+      "learning_rate": 3.970579107861911e-06,
+      "loss": 0.6395,
+      "step": 35564
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 3.9768588542938232,
+      "learning_rate": 3.9703210996314685e-06,
+      "loss": 0.7303,
+      "step": 35565
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.8668551445007324,
+      "learning_rate": 3.970063094264175e-06,
+      "loss": 0.7548,
+      "step": 35566
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 4.833930015563965,
+      "learning_rate": 3.96980509176075e-06,
+      "loss": 0.6537,
+      "step": 35567
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.6949814558029175,
+      "learning_rate": 3.9695470921219135e-06,
+      "loss": 0.5183,
+      "step": 35568
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.4775145053863525,
+      "learning_rate": 3.96928909534838e-06,
+      "loss": 0.7385,
+      "step": 35569
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.8801209926605225,
+      "learning_rate": 3.969031101440867e-06,
+      "loss": 0.6074,
+      "step": 35570
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 3.988320827484131,
+      "learning_rate": 3.968773110400092e-06,
+      "loss": 0.7023,
+      "step": 35571
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.446214437484741,
+      "learning_rate": 3.968515122226773e-06,
+      "loss": 0.5737,
+      "step": 35572
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.7266113758087158,
+      "learning_rate": 3.968257136921627e-06,
+      "loss": 0.4617,
+      "step": 35573
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 4.136831283569336,
+      "learning_rate": 3.967999154485372e-06,
+      "loss": 0.4432,
+      "step": 35574
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.8187599182128906,
+      "learning_rate": 3.967741174918725e-06,
+      "loss": 0.7131,
+      "step": 35575
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.829144835472107,
+      "learning_rate": 3.967483198222401e-06,
+      "loss": 0.5317,
+      "step": 35576
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.103532075881958,
+      "learning_rate": 3.967225224397121e-06,
+      "loss": 0.6691,
+      "step": 35577
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.145256519317627,
+      "learning_rate": 3.9669672534436e-06,
+      "loss": 0.6303,
+      "step": 35578
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 3.515246868133545,
+      "learning_rate": 3.9667092853625565e-06,
+      "loss": 0.5995,
+      "step": 35579
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.4877381324768066,
+      "learning_rate": 3.966451320154707e-06,
+      "loss": 0.6494,
+      "step": 35580
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.3952620029449463,
+      "learning_rate": 3.9661933578207704e-06,
+      "loss": 0.3992,
+      "step": 35581
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.484144687652588,
+      "learning_rate": 3.965935398361462e-06,
+      "loss": 0.7952,
+      "step": 35582
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.8461875915527344,
+      "learning_rate": 3.9656774417775e-06,
+      "loss": 0.7751,
+      "step": 35583
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.9983381032943726,
+      "learning_rate": 3.965419488069602e-06,
+      "loss": 0.6635,
+      "step": 35584
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.880535840988159,
+      "learning_rate": 3.965161537238484e-06,
+      "loss": 0.6135,
+      "step": 35585
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.3911056518554688,
+      "learning_rate": 3.964903589284865e-06,
+      "loss": 0.6351,
+      "step": 35586
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.8271210193634033,
+      "learning_rate": 3.9646456442094595e-06,
+      "loss": 0.5315,
+      "step": 35587
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.4720741510391235,
+      "learning_rate": 3.964387702012989e-06,
+      "loss": 0.4695,
+      "step": 35588
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.5762417316436768,
+      "learning_rate": 3.964129762696165e-06,
+      "loss": 0.8419,
+      "step": 35589
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 3.000096082687378,
+      "learning_rate": 3.963871826259711e-06,
+      "loss": 0.8246,
+      "step": 35590
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.011477470397949,
+      "learning_rate": 3.963613892704341e-06,
+      "loss": 0.3786,
+      "step": 35591
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.961377501487732,
+      "learning_rate": 3.963355962030772e-06,
+      "loss": 0.7789,
+      "step": 35592
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.76119065284729,
+      "learning_rate": 3.963098034239722e-06,
+      "loss": 0.694,
+      "step": 35593
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 3.6298069953918457,
+      "learning_rate": 3.962840109331908e-06,
+      "loss": 0.6238,
+      "step": 35594
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.602720260620117,
+      "learning_rate": 3.962582187308047e-06,
+      "loss": 0.5626,
+      "step": 35595
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.0953643321990967,
+      "learning_rate": 3.9623242681688565e-06,
+      "loss": 0.621,
+      "step": 35596
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.482754945755005,
+      "learning_rate": 3.962066351915054e-06,
+      "loss": 0.5708,
+      "step": 35597
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 3.591658592224121,
+      "learning_rate": 3.9618084385473555e-06,
+      "loss": 0.6408,
+      "step": 35598
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 3.2246756553649902,
+      "learning_rate": 3.961550528066478e-06,
+      "loss": 0.5312,
+      "step": 35599
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.543135404586792,
+      "learning_rate": 3.96129262047314e-06,
+      "loss": 0.6779,
+      "step": 35600
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.3377809524536133,
+      "learning_rate": 3.9610347157680595e-06,
+      "loss": 0.4676,
+      "step": 35601
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.066551685333252,
+      "learning_rate": 3.960776813951951e-06,
+      "loss": 0.7017,
+      "step": 35602
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.0357155799865723,
+      "learning_rate": 3.960518915025533e-06,
+      "loss": 0.8451,
+      "step": 35603
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.806626081466675,
+      "learning_rate": 3.960261018989523e-06,
+      "loss": 0.6687,
+      "step": 35604
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.0008881092071533,
+      "learning_rate": 3.960003125844637e-06,
+      "loss": 0.654,
+      "step": 35605
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.025571584701538,
+      "learning_rate": 3.959745235591593e-06,
+      "loss": 0.2976,
+      "step": 35606
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.50818133354187,
+      "learning_rate": 3.9594873482311055e-06,
+      "loss": 0.7495,
+      "step": 35607
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.1780753135681152,
+      "learning_rate": 3.959229463763896e-06,
+      "loss": 0.7801,
+      "step": 35608
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.12027645111084,
+      "learning_rate": 3.958971582190679e-06,
+      "loss": 0.672,
+      "step": 35609
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.6145834922790527,
+      "learning_rate": 3.958713703512173e-06,
+      "loss": 0.6723,
+      "step": 35610
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.776124119758606,
+      "learning_rate": 3.958455827729094e-06,
+      "loss": 0.5932,
+      "step": 35611
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.460233211517334,
+      "learning_rate": 3.958197954842159e-06,
+      "loss": 0.612,
+      "step": 35612
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.1233856678009033,
+      "learning_rate": 3.9579400848520855e-06,
+      "loss": 0.4504,
+      "step": 35613
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.0074334144592285,
+      "learning_rate": 3.95768221775959e-06,
+      "loss": 0.4881,
+      "step": 35614
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 3.070364475250244,
+      "learning_rate": 3.9574243535653896e-06,
+      "loss": 0.7812,
+      "step": 35615
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 3.782569646835327,
+      "learning_rate": 3.957166492270201e-06,
+      "loss": 0.6427,
+      "step": 35616
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.949597120285034,
+      "learning_rate": 3.956908633874742e-06,
+      "loss": 0.4381,
+      "step": 35617
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 3.003370761871338,
+      "learning_rate": 3.9566507783797305e-06,
+      "loss": 0.7262,
+      "step": 35618
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.9176170825958252,
+      "learning_rate": 3.956392925785881e-06,
+      "loss": 0.5051,
+      "step": 35619
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.211500883102417,
+      "learning_rate": 3.956135076093913e-06,
+      "loss": 0.7198,
+      "step": 35620
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.9967777729034424,
+      "learning_rate": 3.955877229304541e-06,
+      "loss": 0.5867,
+      "step": 35621
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.913681983947754,
+      "learning_rate": 3.9556193854184845e-06,
+      "loss": 0.513,
+      "step": 35622
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.0135459899902344,
+      "learning_rate": 3.955361544436459e-06,
+      "loss": 0.6676,
+      "step": 35623
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.7152009010314941,
+      "learning_rate": 3.955103706359181e-06,
+      "loss": 0.7088,
+      "step": 35624
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.6691548824310303,
+      "learning_rate": 3.954845871187369e-06,
+      "loss": 0.8782,
+      "step": 35625
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.3184126615524292,
+      "learning_rate": 3.954588038921738e-06,
+      "loss": 0.4131,
+      "step": 35626
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 3.20399808883667,
+      "learning_rate": 3.954330209563007e-06,
+      "loss": 0.6141,
+      "step": 35627
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.710261821746826,
+      "learning_rate": 3.95407238311189e-06,
+      "loss": 0.6314,
+      "step": 35628
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.9233667850494385,
+      "learning_rate": 3.953814559569107e-06,
+      "loss": 0.6781,
+      "step": 35629
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.025203227996826,
+      "learning_rate": 3.953556738935374e-06,
+      "loss": 0.4183,
+      "step": 35630
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.187779664993286,
+      "learning_rate": 3.953298921211409e-06,
+      "loss": 0.7204,
+      "step": 35631
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.066005229949951,
+      "learning_rate": 3.953041106397927e-06,
+      "loss": 0.6528,
+      "step": 35632
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 3.809937000274658,
+      "learning_rate": 3.952783294495645e-06,
+      "loss": 0.5962,
+      "step": 35633
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 3.329364776611328,
+      "learning_rate": 3.9525254855052804e-06,
+      "loss": 0.7726,
+      "step": 35634
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.2787225246429443,
+      "learning_rate": 3.9522676794275505e-06,
+      "loss": 0.5121,
+      "step": 35635
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 3.0097169876098633,
+      "learning_rate": 3.952009876263171e-06,
+      "loss": 0.6326,
+      "step": 35636
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.7945603132247925,
+      "learning_rate": 3.95175207601286e-06,
+      "loss": 0.4159,
+      "step": 35637
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.996605396270752,
+      "learning_rate": 3.951494278677333e-06,
+      "loss": 0.6022,
+      "step": 35638
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.392132520675659,
+      "learning_rate": 3.951236484257308e-06,
+      "loss": 0.4563,
+      "step": 35639
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.7237703800201416,
+      "learning_rate": 3.950978692753501e-06,
+      "loss": 0.7503,
+      "step": 35640
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 3.438366413116455,
+      "learning_rate": 3.950720904166631e-06,
+      "loss": 0.5784,
+      "step": 35641
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.3231377601623535,
+      "learning_rate": 3.950463118497412e-06,
+      "loss": 0.563,
+      "step": 35642
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 3.088317632675171,
+      "learning_rate": 3.950205335746562e-06,
+      "loss": 0.5862,
+      "step": 35643
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.357374668121338,
+      "learning_rate": 3.949947555914797e-06,
+      "loss": 0.7384,
+      "step": 35644
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.235179901123047,
+      "learning_rate": 3.949689779002836e-06,
+      "loss": 0.8166,
+      "step": 35645
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 3.2695443630218506,
+      "learning_rate": 3.949432005011393e-06,
+      "loss": 0.9646,
+      "step": 35646
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.181190013885498,
+      "learning_rate": 3.949174233941187e-06,
+      "loss": 0.4167,
+      "step": 35647
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 3.173469066619873,
+      "learning_rate": 3.948916465792932e-06,
+      "loss": 0.6137,
+      "step": 35648
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 4.569045066833496,
+      "learning_rate": 3.9486587005673474e-06,
+      "loss": 0.3828,
+      "step": 35649
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.6581746339797974,
+      "learning_rate": 3.94840093826515e-06,
+      "loss": 0.6642,
+      "step": 35650
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.2345573902130127,
+      "learning_rate": 3.948143178887054e-06,
+      "loss": 0.6785,
+      "step": 35651
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.3203976154327393,
+      "learning_rate": 3.947885422433779e-06,
+      "loss": 0.6515,
+      "step": 35652
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 3.789212226867676,
+      "learning_rate": 3.947627668906041e-06,
+      "loss": 0.8444,
+      "step": 35653
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.734370470046997,
+      "learning_rate": 3.947369918304556e-06,
+      "loss": 0.5188,
+      "step": 35654
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 3.9513418674468994,
+      "learning_rate": 3.94711217063004e-06,
+      "loss": 0.5169,
+      "step": 35655
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.0416901111602783,
+      "learning_rate": 3.946854425883211e-06,
+      "loss": 0.549,
+      "step": 35656
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 9.3601713180542,
+      "learning_rate": 3.946596684064786e-06,
+      "loss": 0.517,
+      "step": 35657
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 3.1548497676849365,
+      "learning_rate": 3.946338945175481e-06,
+      "loss": 0.6565,
+      "step": 35658
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.2824625968933105,
+      "learning_rate": 3.946081209216011e-06,
+      "loss": 0.4578,
+      "step": 35659
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 3.191232442855835,
+      "learning_rate": 3.9458234761870965e-06,
+      "loss": 0.5854,
+      "step": 35660
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.9736236333847046,
+      "learning_rate": 3.945565746089452e-06,
+      "loss": 0.6004,
+      "step": 35661
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.0343222618103027,
+      "learning_rate": 3.945308018923792e-06,
+      "loss": 0.6668,
+      "step": 35662
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.1792478561401367,
+      "learning_rate": 3.9450502946908376e-06,
+      "loss": 0.6175,
+      "step": 35663
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.664337635040283,
+      "learning_rate": 3.944792573391302e-06,
+      "loss": 0.6015,
+      "step": 35664
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.1916813850402832,
+      "learning_rate": 3.944534855025903e-06,
+      "loss": 0.5618,
+      "step": 35665
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.8735334873199463,
+      "learning_rate": 3.944277139595357e-06,
+      "loss": 0.7277,
+      "step": 35666
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.596099615097046,
+      "learning_rate": 3.944019427100382e-06,
+      "loss": 0.4483,
+      "step": 35667
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.9741618633270264,
+      "learning_rate": 3.94376171754169e-06,
+      "loss": 0.7257,
+      "step": 35668
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.492992877960205,
+      "learning_rate": 3.9435040109200035e-06,
+      "loss": 0.6062,
+      "step": 35669
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.9143248796463013,
+      "learning_rate": 3.943246307236037e-06,
+      "loss": 0.5403,
+      "step": 35670
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.028702974319458,
+      "learning_rate": 3.9429886064905055e-06,
+      "loss": 0.6122,
+      "step": 35671
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.932539701461792,
+      "learning_rate": 3.942730908684128e-06,
+      "loss": 0.8633,
+      "step": 35672
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.4364373683929443,
+      "learning_rate": 3.942473213817619e-06,
+      "loss": 0.695,
+      "step": 35673
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.6178038120269775,
+      "learning_rate": 3.942215521891696e-06,
+      "loss": 0.4731,
+      "step": 35674
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.372262477874756,
+      "learning_rate": 3.941957832907075e-06,
+      "loss": 0.7389,
+      "step": 35675
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.473327159881592,
+      "learning_rate": 3.9417001468644736e-06,
+      "loss": 0.4788,
+      "step": 35676
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.9156196117401123,
+      "learning_rate": 3.941442463764608e-06,
+      "loss": 0.6746,
+      "step": 35677
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.583237409591675,
+      "learning_rate": 3.941184783608193e-06,
+      "loss": 0.6182,
+      "step": 35678
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.2927210330963135,
+      "learning_rate": 3.940927106395948e-06,
+      "loss": 0.5025,
+      "step": 35679
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.6111879348754883,
+      "learning_rate": 3.940669432128587e-06,
+      "loss": 0.7458,
+      "step": 35680
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.5420291423797607,
+      "learning_rate": 3.9404117608068276e-06,
+      "loss": 0.6876,
+      "step": 35681
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.262585401535034,
+      "learning_rate": 3.940154092431386e-06,
+      "loss": 0.8127,
+      "step": 35682
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.1969120502471924,
+      "learning_rate": 3.939896427002979e-06,
+      "loss": 0.6491,
+      "step": 35683
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.685234785079956,
+      "learning_rate": 3.939638764522324e-06,
+      "loss": 0.7271,
+      "step": 35684
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.9660322666168213,
+      "learning_rate": 3.939381104990135e-06,
+      "loss": 0.6317,
+      "step": 35685
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.3363842964172363,
+      "learning_rate": 3.93912344840713e-06,
+      "loss": 0.6339,
+      "step": 35686
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.4431512355804443,
+      "learning_rate": 3.938865794774025e-06,
+      "loss": 0.5579,
+      "step": 35687
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 3.698099374771118,
+      "learning_rate": 3.938608144091538e-06,
+      "loss": 0.7418,
+      "step": 35688
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.928764820098877,
+      "learning_rate": 3.938350496360381e-06,
+      "loss": 0.7273,
+      "step": 35689
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 3.71321702003479,
+      "learning_rate": 3.9380928515812756e-06,
+      "loss": 0.6269,
+      "step": 35690
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.4111647605895996,
+      "learning_rate": 3.937835209754937e-06,
+      "loss": 0.779,
+      "step": 35691
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.7622342109680176,
+      "learning_rate": 3.93757757088208e-06,
+      "loss": 0.3094,
+      "step": 35692
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.4212844371795654,
+      "learning_rate": 3.937319934963422e-06,
+      "loss": 0.6579,
+      "step": 35693
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.0239017009735107,
+      "learning_rate": 3.93706230199968e-06,
+      "loss": 0.3756,
+      "step": 35694
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.944287657737732,
+      "learning_rate": 3.936804671991568e-06,
+      "loss": 0.5985,
+      "step": 35695
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.563664436340332,
+      "learning_rate": 3.9365470449398045e-06,
+      "loss": 0.6984,
+      "step": 35696
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 3.649824380874634,
+      "learning_rate": 3.936289420845106e-06,
+      "loss": 0.5855,
+      "step": 35697
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 3.8860819339752197,
+      "learning_rate": 3.936031799708187e-06,
+      "loss": 0.9494,
+      "step": 35698
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.595273971557617,
+      "learning_rate": 3.9357741815297665e-06,
+      "loss": 0.5315,
+      "step": 35699
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.230551242828369,
+      "learning_rate": 3.935516566310558e-06,
+      "loss": 0.6696,
+      "step": 35700
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.3526244163513184,
+      "learning_rate": 3.935258954051279e-06,
+      "loss": 0.5859,
+      "step": 35701
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 3.4752228260040283,
+      "learning_rate": 3.935001344752647e-06,
+      "loss": 0.763,
+      "step": 35702
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 4.747034549713135,
+      "learning_rate": 3.934743738415376e-06,
+      "loss": 0.6689,
+      "step": 35703
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.7687795162200928,
+      "learning_rate": 3.9344861350401845e-06,
+      "loss": 0.645,
+      "step": 35704
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.166200637817383,
+      "learning_rate": 3.9342285346277885e-06,
+      "loss": 0.6239,
+      "step": 35705
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.0183773040771484,
+      "learning_rate": 3.933970937178903e-06,
+      "loss": 0.7883,
+      "step": 35706
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 3.268575429916382,
+      "learning_rate": 3.933713342694246e-06,
+      "loss": 0.6751,
+      "step": 35707
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.919811487197876,
+      "learning_rate": 3.933455751174531e-06,
+      "loss": 0.6526,
+      "step": 35708
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 3.117278575897217,
+      "learning_rate": 3.933198162620476e-06,
+      "loss": 0.6065,
+      "step": 35709
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.760796070098877,
+      "learning_rate": 3.932940577032798e-06,
+      "loss": 0.7156,
+      "step": 35710
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.5118343830108643,
+      "learning_rate": 3.932682994412212e-06,
+      "loss": 0.6248,
+      "step": 35711
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.7683029174804688,
+      "learning_rate": 3.932425414759437e-06,
+      "loss": 0.5734,
+      "step": 35712
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 6.671905994415283,
+      "learning_rate": 3.932167838075184e-06,
+      "loss": 0.76,
+      "step": 35713
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 4.104593276977539,
+      "learning_rate": 3.931910264360174e-06,
+      "loss": 0.7663,
+      "step": 35714
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.9710007905960083,
+      "learning_rate": 3.931652693615121e-06,
+      "loss": 0.6802,
+      "step": 35715
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.4734134674072266,
+      "learning_rate": 3.931395125840741e-06,
+      "loss": 0.6063,
+      "step": 35716
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.9479185342788696,
+      "learning_rate": 3.931137561037752e-06,
+      "loss": 0.6871,
+      "step": 35717
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.059147357940674,
+      "learning_rate": 3.930879999206868e-06,
+      "loss": 0.5263,
+      "step": 35718
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.0191400051116943,
+      "learning_rate": 3.9306224403488075e-06,
+      "loss": 0.6716,
+      "step": 35719
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.2903226613998413,
+      "learning_rate": 3.930364884464285e-06,
+      "loss": 0.4865,
+      "step": 35720
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 3.9811906814575195,
+      "learning_rate": 3.930107331554017e-06,
+      "loss": 0.8185,
+      "step": 35721
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.4609992504119873,
+      "learning_rate": 3.9298497816187196e-06,
+      "loss": 0.616,
+      "step": 35722
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.2283430099487305,
+      "learning_rate": 3.9295922346591086e-06,
+      "loss": 0.6816,
+      "step": 35723
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.789968490600586,
+      "learning_rate": 3.929334690675901e-06,
+      "loss": 0.6088,
+      "step": 35724
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.130493402481079,
+      "learning_rate": 3.929077149669813e-06,
+      "loss": 0.6625,
+      "step": 35725
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.9250305891036987,
+      "learning_rate": 3.928819611641559e-06,
+      "loss": 0.6824,
+      "step": 35726
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.0809524059295654,
+      "learning_rate": 3.9285620765918575e-06,
+      "loss": 0.6108,
+      "step": 35727
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.1342880725860596,
+      "learning_rate": 3.928304544521424e-06,
+      "loss": 0.6461,
+      "step": 35728
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.758060932159424,
+      "learning_rate": 3.928047015430971e-06,
+      "loss": 0.7906,
+      "step": 35729
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.496206283569336,
+      "learning_rate": 3.92778948932122e-06,
+      "loss": 0.571,
+      "step": 35730
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.340837001800537,
+      "learning_rate": 3.927531966192885e-06,
+      "loss": 0.6016,
+      "step": 35731
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.0339696407318115,
+      "learning_rate": 3.927274446046681e-06,
+      "loss": 0.3444,
+      "step": 35732
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.9414947032928467,
+      "learning_rate": 3.927016928883326e-06,
+      "loss": 0.6434,
+      "step": 35733
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 3.1806106567382812,
+      "learning_rate": 3.926759414703534e-06,
+      "loss": 0.5752,
+      "step": 35734
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 4.343175411224365,
+      "learning_rate": 3.9265019035080235e-06,
+      "loss": 0.5344,
+      "step": 35735
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.3342647552490234,
+      "learning_rate": 3.926244395297508e-06,
+      "loss": 0.7035,
+      "step": 35736
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.538240671157837,
+      "learning_rate": 3.925986890072704e-06,
+      "loss": 0.7256,
+      "step": 35737
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.163090467453003,
+      "learning_rate": 3.925729387834329e-06,
+      "loss": 0.6506,
+      "step": 35738
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.974443197250366,
+      "learning_rate": 3.925471888583098e-06,
+      "loss": 0.5175,
+      "step": 35739
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.489863872528076,
+      "learning_rate": 3.925214392319728e-06,
+      "loss": 0.4207,
+      "step": 35740
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.7785438299179077,
+      "learning_rate": 3.924956899044933e-06,
+      "loss": 0.7199,
+      "step": 35741
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 3.266932487487793,
+      "learning_rate": 3.92469940875943e-06,
+      "loss": 0.4586,
+      "step": 35742
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.5658798217773438,
+      "learning_rate": 3.924441921463936e-06,
+      "loss": 0.7308,
+      "step": 35743
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.0388565063476562,
+      "learning_rate": 3.924184437159165e-06,
+      "loss": 0.4411,
+      "step": 35744
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.731581926345825,
+      "learning_rate": 3.923926955845835e-06,
+      "loss": 0.6947,
+      "step": 35745
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 5.343479156494141,
+      "learning_rate": 3.923669477524661e-06,
+      "loss": 0.7529,
+      "step": 35746
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.9641362428665161,
+      "learning_rate": 3.923412002196358e-06,
+      "loss": 0.7351,
+      "step": 35747
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.551297187805176,
+      "learning_rate": 3.923154529861644e-06,
+      "loss": 0.6876,
+      "step": 35748
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.634434461593628,
+      "learning_rate": 3.922897060521234e-06,
+      "loss": 0.5231,
+      "step": 35749
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.2292957305908203,
+      "learning_rate": 3.9226395941758404e-06,
+      "loss": 0.7746,
+      "step": 35750
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.397216320037842,
+      "learning_rate": 3.922382130826186e-06,
+      "loss": 0.734,
+      "step": 35751
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.2357771396636963,
+      "learning_rate": 3.922124670472982e-06,
+      "loss": 0.49,
+      "step": 35752
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.8974487781524658,
+      "learning_rate": 3.9218672131169465e-06,
+      "loss": 0.6939,
+      "step": 35753
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.997763991355896,
+      "learning_rate": 3.921609758758794e-06,
+      "loss": 0.5511,
+      "step": 35754
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.26255464553833,
+      "learning_rate": 3.921352307399241e-06,
+      "loss": 0.8437,
+      "step": 35755
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 3.4049689769744873,
+      "learning_rate": 3.921094859039003e-06,
+      "loss": 0.8385,
+      "step": 35756
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.1817142963409424,
+      "learning_rate": 3.920837413678796e-06,
+      "loss": 0.7882,
+      "step": 35757
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.9164379835128784,
+      "learning_rate": 3.920579971319336e-06,
+      "loss": 0.7536,
+      "step": 35758
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.4241526126861572,
+      "learning_rate": 3.920322531961338e-06,
+      "loss": 0.6827,
+      "step": 35759
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 4.6415228843688965,
+      "learning_rate": 3.920065095605519e-06,
+      "loss": 0.5717,
+      "step": 35760
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.6073760986328125,
+      "learning_rate": 3.919807662252595e-06,
+      "loss": 0.9466,
+      "step": 35761
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.6146554946899414,
+      "learning_rate": 3.919550231903281e-06,
+      "loss": 0.6136,
+      "step": 35762
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.4697482585906982,
+      "learning_rate": 3.919292804558292e-06,
+      "loss": 0.5228,
+      "step": 35763
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.8047142028808594,
+      "learning_rate": 3.919035380218346e-06,
+      "loss": 0.4027,
+      "step": 35764
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.3173348903656006,
+      "learning_rate": 3.918777958884157e-06,
+      "loss": 0.57,
+      "step": 35765
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 3.118082046508789,
+      "learning_rate": 3.918520540556442e-06,
+      "loss": 0.4891,
+      "step": 35766
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.0844953060150146,
+      "learning_rate": 3.918263125235916e-06,
+      "loss": 0.506,
+      "step": 35767
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.4438676834106445,
+      "learning_rate": 3.9180057129232955e-06,
+      "loss": 0.6107,
+      "step": 35768
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.2595393657684326,
+      "learning_rate": 3.917748303619296e-06,
+      "loss": 0.5392,
+      "step": 35769
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.042536973953247,
+      "learning_rate": 3.917490897324631e-06,
+      "loss": 0.5368,
+      "step": 35770
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.023961305618286,
+      "learning_rate": 3.91723349404002e-06,
+      "loss": 0.4278,
+      "step": 35771
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.8546820878982544,
+      "learning_rate": 3.916976093766176e-06,
+      "loss": 0.6196,
+      "step": 35772
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 4.0212931632995605,
+      "learning_rate": 3.916718696503817e-06,
+      "loss": 0.7494,
+      "step": 35773
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.7911427021026611,
+      "learning_rate": 3.916461302253656e-06,
+      "loss": 0.6149,
+      "step": 35774
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 3.118758201599121,
+      "learning_rate": 3.9162039110164115e-06,
+      "loss": 0.7285,
+      "step": 35775
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.9257454872131348,
+      "learning_rate": 3.9159465227927975e-06,
+      "loss": 0.8378,
+      "step": 35776
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 3.316493272781372,
+      "learning_rate": 3.915689137583531e-06,
+      "loss": 0.7779,
+      "step": 35777
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.260223150253296,
+      "learning_rate": 3.915431755389326e-06,
+      "loss": 0.5503,
+      "step": 35778
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.106419563293457,
+      "learning_rate": 3.9151743762109e-06,
+      "loss": 0.6867,
+      "step": 35779
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.7293601036071777,
+      "learning_rate": 3.914917000048966e-06,
+      "loss": 0.378,
+      "step": 35780
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.8007203340530396,
+      "learning_rate": 3.914659626904242e-06,
+      "loss": 0.7175,
+      "step": 35781
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.2602922916412354,
+      "learning_rate": 3.914402256777444e-06,
+      "loss": 0.6011,
+      "step": 35782
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 3.8691301345825195,
+      "learning_rate": 3.914144889669286e-06,
+      "loss": 0.81,
+      "step": 35783
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 3.4077205657958984,
+      "learning_rate": 3.913887525580484e-06,
+      "loss": 0.6522,
+      "step": 35784
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.2788655757904053,
+      "learning_rate": 3.913630164511753e-06,
+      "loss": 0.5979,
+      "step": 35785
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.4002132415771484,
+      "learning_rate": 3.913372806463812e-06,
+      "loss": 0.841,
+      "step": 35786
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.8232054710388184,
+      "learning_rate": 3.913115451437372e-06,
+      "loss": 0.7039,
+      "step": 35787
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.1395111083984375,
+      "learning_rate": 3.912858099433152e-06,
+      "loss": 0.5124,
+      "step": 35788
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.0767879486083984,
+      "learning_rate": 3.912600750451866e-06,
+      "loss": 0.5672,
+      "step": 35789
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.7573132514953613,
+      "learning_rate": 3.912343404494228e-06,
+      "loss": 0.6133,
+      "step": 35790
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.368253469467163,
+      "learning_rate": 3.912086061560957e-06,
+      "loss": 0.7126,
+      "step": 35791
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.9079759120941162,
+      "learning_rate": 3.9118287216527675e-06,
+      "loss": 0.6989,
+      "step": 35792
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.7914822101593018,
+      "learning_rate": 3.911571384770375e-06,
+      "loss": 0.7179,
+      "step": 35793
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.808720827102661,
+      "learning_rate": 3.911314050914494e-06,
+      "loss": 0.5206,
+      "step": 35794
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.1249074935913086,
+      "learning_rate": 3.911056720085842e-06,
+      "loss": 0.5948,
+      "step": 35795
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.8046364784240723,
+      "learning_rate": 3.910799392285133e-06,
+      "loss": 0.4991,
+      "step": 35796
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.69820237159729,
+      "learning_rate": 3.910542067513082e-06,
+      "loss": 0.6775,
+      "step": 35797
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 9.09715461730957,
+      "learning_rate": 3.910284745770406e-06,
+      "loss": 0.6225,
+      "step": 35798
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.893369674682617,
+      "learning_rate": 3.91002742705782e-06,
+      "loss": 0.59,
+      "step": 35799
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 3.293405294418335,
+      "learning_rate": 3.9097701113760385e-06,
+      "loss": 0.6626,
+      "step": 35800
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.499488115310669,
+      "learning_rate": 3.909512798725779e-06,
+      "loss": 0.5392,
+      "step": 35801
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.7259225845336914,
+      "learning_rate": 3.909255489107756e-06,
+      "loss": 0.7637,
+      "step": 35802
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.3479843139648438,
+      "learning_rate": 3.908998182522684e-06,
+      "loss": 0.772,
+      "step": 35803
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.586068630218506,
+      "learning_rate": 3.90874087897128e-06,
+      "loss": 0.7825,
+      "step": 35804
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.0824949741363525,
+      "learning_rate": 3.908483578454258e-06,
+      "loss": 0.6436,
+      "step": 35805
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.8990285396575928,
+      "learning_rate": 3.908226280972335e-06,
+      "loss": 0.9132,
+      "step": 35806
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.7118995189666748,
+      "learning_rate": 3.907968986526226e-06,
+      "loss": 0.7797,
+      "step": 35807
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.8119194507598877,
+      "learning_rate": 3.907711695116645e-06,
+      "loss": 0.5312,
+      "step": 35808
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.998360514640808,
+      "learning_rate": 3.90745440674431e-06,
+      "loss": 0.532,
+      "step": 35809
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.4473803043365479,
+      "learning_rate": 3.907197121409935e-06,
+      "loss": 0.4836,
+      "step": 35810
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.2202601432800293,
+      "learning_rate": 3.906939839114232e-06,
+      "loss": 0.5297,
+      "step": 35811
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.914578914642334,
+      "learning_rate": 3.9066825598579226e-06,
+      "loss": 0.6502,
+      "step": 35812
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.1476352214813232,
+      "learning_rate": 3.90642528364172e-06,
+      "loss": 0.4781,
+      "step": 35813
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.5131163597106934,
+      "learning_rate": 3.906168010466338e-06,
+      "loss": 0.5113,
+      "step": 35814
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.4808590412139893,
+      "learning_rate": 3.905910740332493e-06,
+      "loss": 0.5384,
+      "step": 35815
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 4.028796195983887,
+      "learning_rate": 3.905653473240901e-06,
+      "loss": 0.6404,
+      "step": 35816
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.9285619258880615,
+      "learning_rate": 3.905396209192277e-06,
+      "loss": 0.7693,
+      "step": 35817
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 3.868126153945923,
+      "learning_rate": 3.905138948187336e-06,
+      "loss": 0.6631,
+      "step": 35818
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.604632616043091,
+      "learning_rate": 3.904881690226793e-06,
+      "loss": 0.6138,
+      "step": 35819
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 3.2650210857391357,
+      "learning_rate": 3.904624435311364e-06,
+      "loss": 0.5693,
+      "step": 35820
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 4.91909646987915,
+      "learning_rate": 3.9043671834417644e-06,
+      "loss": 1.0384,
+      "step": 35821
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.3526062965393066,
+      "learning_rate": 3.904109934618709e-06,
+      "loss": 0.4957,
+      "step": 35822
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 3.2085013389587402,
+      "learning_rate": 3.903852688842913e-06,
+      "loss": 0.7539,
+      "step": 35823
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.2818357944488525,
+      "learning_rate": 3.903595446115091e-06,
+      "loss": 0.7327,
+      "step": 35824
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.921196699142456,
+      "learning_rate": 3.9033382064359605e-06,
+      "loss": 0.4456,
+      "step": 35825
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 3.1336610317230225,
+      "learning_rate": 3.9030809698062354e-06,
+      "loss": 0.6959,
+      "step": 35826
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 3.3054938316345215,
+      "learning_rate": 3.902823736226632e-06,
+      "loss": 0.5704,
+      "step": 35827
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.821700930595398,
+      "learning_rate": 3.902566505697864e-06,
+      "loss": 0.6007,
+      "step": 35828
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.3076283931732178,
+      "learning_rate": 3.902309278220648e-06,
+      "loss": 0.4846,
+      "step": 35829
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.4182665348052979,
+      "learning_rate": 3.902052053795698e-06,
+      "loss": 0.3556,
+      "step": 35830
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.6045868396759033,
+      "learning_rate": 3.9017948324237295e-06,
+      "loss": 0.7098,
+      "step": 35831
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.2193105220794678,
+      "learning_rate": 3.901537614105459e-06,
+      "loss": 0.5411,
+      "step": 35832
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.1944708824157715,
+      "learning_rate": 3.901280398841601e-06,
+      "loss": 0.7638,
+      "step": 35833
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.5633389949798584,
+      "learning_rate": 3.9010231866328705e-06,
+      "loss": 0.7345,
+      "step": 35834
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.298834800720215,
+      "learning_rate": 3.900765977479984e-06,
+      "loss": 0.6778,
+      "step": 35835
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.6178343296051025,
+      "learning_rate": 3.9005087713836525e-06,
+      "loss": 0.7991,
+      "step": 35836
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.4068291187286377,
+      "learning_rate": 3.900251568344596e-06,
+      "loss": 0.5367,
+      "step": 35837
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.8675003051757812,
+      "learning_rate": 3.899994368363529e-06,
+      "loss": 0.5195,
+      "step": 35838
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.661104202270508,
+      "learning_rate": 3.899737171441164e-06,
+      "loss": 0.8515,
+      "step": 35839
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.0894575119018555,
+      "learning_rate": 3.899479977578219e-06,
+      "loss": 0.6425,
+      "step": 35840
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.174603223800659,
+      "learning_rate": 3.899222786775407e-06,
+      "loss": 0.4556,
+      "step": 35841
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.731760025024414,
+      "learning_rate": 3.898965599033444e-06,
+      "loss": 0.6509,
+      "step": 35842
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 3.5701518058776855,
+      "learning_rate": 3.898708414353047e-06,
+      "loss": 0.6798,
+      "step": 35843
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 3.1108365058898926,
+      "learning_rate": 3.898451232734927e-06,
+      "loss": 0.5082,
+      "step": 35844
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 3.448674201965332,
+      "learning_rate": 3.898194054179802e-06,
+      "loss": 0.7127,
+      "step": 35845
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.80046808719635,
+      "learning_rate": 3.897936878688388e-06,
+      "loss": 0.8061,
+      "step": 35846
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.4082982540130615,
+      "learning_rate": 3.897679706261397e-06,
+      "loss": 0.6406,
+      "step": 35847
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.7427304983139038,
+      "learning_rate": 3.897422536899547e-06,
+      "loss": 0.53,
+      "step": 35848
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 3.3206090927124023,
+      "learning_rate": 3.897165370603551e-06,
+      "loss": 0.8948,
+      "step": 35849
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.9294817447662354,
+      "learning_rate": 3.896908207374126e-06,
+      "loss": 0.6652,
+      "step": 35850
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 5.6239118576049805,
+      "learning_rate": 3.896651047211983e-06,
+      "loss": 0.3951,
+      "step": 35851
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.9755306243896484,
+      "learning_rate": 3.896393890117842e-06,
+      "loss": 0.7801,
+      "step": 35852
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.8166611194610596,
+      "learning_rate": 3.896136736092417e-06,
+      "loss": 0.4497,
+      "step": 35853
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.5119521617889404,
+      "learning_rate": 3.8958795851364225e-06,
+      "loss": 0.5522,
+      "step": 35854
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.6703293323516846,
+      "learning_rate": 3.895622437250572e-06,
+      "loss": 0.6612,
+      "step": 35855
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 5.447871208190918,
+      "learning_rate": 3.895365292435583e-06,
+      "loss": 0.7803,
+      "step": 35856
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.6612231731414795,
+      "learning_rate": 3.8951081506921685e-06,
+      "loss": 0.4889,
+      "step": 35857
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.9650304317474365,
+      "learning_rate": 3.894851012021045e-06,
+      "loss": 0.6704,
+      "step": 35858
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.7216970920562744,
+      "learning_rate": 3.894593876422926e-06,
+      "loss": 0.8256,
+      "step": 35859
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 3.869481325149536,
+      "learning_rate": 3.894336743898528e-06,
+      "loss": 0.4964,
+      "step": 35860
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.6819320917129517,
+      "learning_rate": 3.8940796144485654e-06,
+      "loss": 0.8074,
+      "step": 35861
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.1628966331481934,
+      "learning_rate": 3.893822488073753e-06,
+      "loss": 0.6714,
+      "step": 35862
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.5955111980438232,
+      "learning_rate": 3.893565364774806e-06,
+      "loss": 0.7435,
+      "step": 35863
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.8757312297821045,
+      "learning_rate": 3.893308244552439e-06,
+      "loss": 0.5898,
+      "step": 35864
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.6407463550567627,
+      "learning_rate": 3.893051127407366e-06,
+      "loss": 0.6999,
+      "step": 35865
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.928175687789917,
+      "learning_rate": 3.892794013340305e-06,
+      "loss": 0.4902,
+      "step": 35866
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.501840353012085,
+      "learning_rate": 3.892536902351969e-06,
+      "loss": 0.4199,
+      "step": 35867
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.597141742706299,
+      "learning_rate": 3.892279794443072e-06,
+      "loss": 0.7641,
+      "step": 35868
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 4.669424533843994,
+      "learning_rate": 3.892022689614331e-06,
+      "loss": 0.623,
+      "step": 35869
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.507458209991455,
+      "learning_rate": 3.891765587866458e-06,
+      "loss": 0.5973,
+      "step": 35870
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.7652183771133423,
+      "learning_rate": 3.891508489200171e-06,
+      "loss": 0.5428,
+      "step": 35871
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 5.2338080406188965,
+      "learning_rate": 3.891251393616181e-06,
+      "loss": 0.6616,
+      "step": 35872
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.1299474239349365,
+      "learning_rate": 3.890994301115209e-06,
+      "loss": 0.7275,
+      "step": 35873
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.7063491344451904,
+      "learning_rate": 3.890737211697966e-06,
+      "loss": 0.4039,
+      "step": 35874
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.025127649307251,
+      "learning_rate": 3.890480125365166e-06,
+      "loss": 0.5818,
+      "step": 35875
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.342402458190918,
+      "learning_rate": 3.890223042117526e-06,
+      "loss": 0.6258,
+      "step": 35876
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 5.445816516876221,
+      "learning_rate": 3.88996596195576e-06,
+      "loss": 0.6399,
+      "step": 35877
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 5.8759284019470215,
+      "learning_rate": 3.889708884880583e-06,
+      "loss": 0.5572,
+      "step": 35878
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.4789018630981445,
+      "learning_rate": 3.889451810892709e-06,
+      "loss": 0.645,
+      "step": 35879
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.929290294647217,
+      "learning_rate": 3.889194739992854e-06,
+      "loss": 0.7152,
+      "step": 35880
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.8997914791107178,
+      "learning_rate": 3.888937672181732e-06,
+      "loss": 0.5308,
+      "step": 35881
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.7753541469573975,
+      "learning_rate": 3.8886806074600585e-06,
+      "loss": 0.5971,
+      "step": 35882
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.0753440856933594,
+      "learning_rate": 3.888423545828548e-06,
+      "loss": 0.7534,
+      "step": 35883
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 3.071305751800537,
+      "learning_rate": 3.8881664872879145e-06,
+      "loss": 0.6212,
+      "step": 35884
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.3979244232177734,
+      "learning_rate": 3.887909431838874e-06,
+      "loss": 0.5985,
+      "step": 35885
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.9773348569869995,
+      "learning_rate": 3.88765237948214e-06,
+      "loss": 0.5945,
+      "step": 35886
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.3886098861694336,
+      "learning_rate": 3.887395330218429e-06,
+      "loss": 0.8289,
+      "step": 35887
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.415666103363037,
+      "learning_rate": 3.887138284048454e-06,
+      "loss": 0.4675,
+      "step": 35888
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.415175199508667,
+      "learning_rate": 3.886881240972931e-06,
+      "loss": 0.707,
+      "step": 35889
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.901803731918335,
+      "learning_rate": 3.886624200992575e-06,
+      "loss": 0.6399,
+      "step": 35890
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.6457056999206543,
+      "learning_rate": 3.886367164108099e-06,
+      "loss": 0.5557,
+      "step": 35891
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.6979506015777588,
+      "learning_rate": 3.886110130320218e-06,
+      "loss": 0.7238,
+      "step": 35892
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 4.7783203125,
+      "learning_rate": 3.885853099629649e-06,
+      "loss": 1.0364,
+      "step": 35893
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.9294480085372925,
+      "learning_rate": 3.885596072037104e-06,
+      "loss": 0.516,
+      "step": 35894
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.4823384284973145,
+      "learning_rate": 3.8853390475433e-06,
+      "loss": 0.6896,
+      "step": 35895
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.6158504486083984,
+      "learning_rate": 3.88508202614895e-06,
+      "loss": 0.7619,
+      "step": 35896
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.4429898262023926,
+      "learning_rate": 3.88482500785477e-06,
+      "loss": 0.5485,
+      "step": 35897
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.2596139907836914,
+      "learning_rate": 3.884567992661472e-06,
+      "loss": 0.7588,
+      "step": 35898
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 3.247281551361084,
+      "learning_rate": 3.8843109805697735e-06,
+      "loss": 0.4554,
+      "step": 35899
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.6506919860839844,
+      "learning_rate": 3.8840539715803885e-06,
+      "loss": 0.5866,
+      "step": 35900
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.3824496269226074,
+      "learning_rate": 3.8837969656940316e-06,
+      "loss": 0.705,
+      "step": 35901
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.39404559135437,
+      "learning_rate": 3.883539962911418e-06,
+      "loss": 0.418,
+      "step": 35902
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.6875193119049072,
+      "learning_rate": 3.88328296323326e-06,
+      "loss": 0.5296,
+      "step": 35903
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 4.351219654083252,
+      "learning_rate": 3.883025966660274e-06,
+      "loss": 0.7985,
+      "step": 35904
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.5726468563079834,
+      "learning_rate": 3.882768973193175e-06,
+      "loss": 0.5618,
+      "step": 35905
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.895740270614624,
+      "learning_rate": 3.882511982832676e-06,
+      "loss": 0.8365,
+      "step": 35906
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.2365896701812744,
+      "learning_rate": 3.882254995579494e-06,
+      "loss": 0.4933,
+      "step": 35907
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.044273853302002,
+      "learning_rate": 3.881998011434341e-06,
+      "loss": 0.4531,
+      "step": 35908
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.8489006757736206,
+      "learning_rate": 3.881741030397933e-06,
+      "loss": 0.8034,
+      "step": 35909
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 3.638810396194458,
+      "learning_rate": 3.881484052470984e-06,
+      "loss": 0.577,
+      "step": 35910
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 6.942093372344971,
+      "learning_rate": 3.881227077654209e-06,
+      "loss": 0.5143,
+      "step": 35911
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.786440372467041,
+      "learning_rate": 3.88097010594832e-06,
+      "loss": 0.4814,
+      "step": 35912
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 3.241629123687744,
+      "learning_rate": 3.880713137354037e-06,
+      "loss": 0.5971,
+      "step": 35913
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.7702996730804443,
+      "learning_rate": 3.880456171872071e-06,
+      "loss": 0.7738,
+      "step": 35914
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.202078104019165,
+      "learning_rate": 3.880199209503137e-06,
+      "loss": 0.4457,
+      "step": 35915
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.229543924331665,
+      "learning_rate": 3.87994225024795e-06,
+      "loss": 0.6377,
+      "step": 35916
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.4720516204833984,
+      "learning_rate": 3.879685294107223e-06,
+      "loss": 0.7061,
+      "step": 35917
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 3.084550142288208,
+      "learning_rate": 3.8794283410816725e-06,
+      "loss": 0.7201,
+      "step": 35918
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 3.2289345264434814,
+      "learning_rate": 3.879171391172011e-06,
+      "loss": 0.6231,
+      "step": 35919
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.9077955484390259,
+      "learning_rate": 3.878914444378955e-06,
+      "loss": 0.4345,
+      "step": 35920
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.1787140369415283,
+      "learning_rate": 3.878657500703217e-06,
+      "loss": 0.6264,
+      "step": 35921
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 3.0015830993652344,
+      "learning_rate": 3.878400560145513e-06,
+      "loss": 0.7397,
+      "step": 35922
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.635552167892456,
+      "learning_rate": 3.878143622706557e-06,
+      "loss": 0.6938,
+      "step": 35923
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.5316576957702637,
+      "learning_rate": 3.877886688387063e-06,
+      "loss": 0.6256,
+      "step": 35924
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.495548963546753,
+      "learning_rate": 3.877629757187746e-06,
+      "loss": 0.6613,
+      "step": 35925
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.4155690670013428,
+      "learning_rate": 3.87737282910932e-06,
+      "loss": 0.7288,
+      "step": 35926
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.2255868911743164,
+      "learning_rate": 3.8771159041525e-06,
+      "loss": 0.6408,
+      "step": 35927
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.7733187675476074,
+      "learning_rate": 3.876858982317999e-06,
+      "loss": 0.6345,
+      "step": 35928
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.4110841751098633,
+      "learning_rate": 3.876602063606534e-06,
+      "loss": 0.6466,
+      "step": 35929
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.434680938720703,
+      "learning_rate": 3.876345148018816e-06,
+      "loss": 0.6367,
+      "step": 35930
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.0942444801330566,
+      "learning_rate": 3.876088235555563e-06,
+      "loss": 0.6406,
+      "step": 35931
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.532677173614502,
+      "learning_rate": 3.8758313262174864e-06,
+      "loss": 0.6245,
+      "step": 35932
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 3.6291987895965576,
+      "learning_rate": 3.8755744200053005e-06,
+      "loss": 0.6478,
+      "step": 35933
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.927879810333252,
+      "learning_rate": 3.875317516919722e-06,
+      "loss": 0.5829,
+      "step": 35934
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.059238910675049,
+      "learning_rate": 3.875060616961464e-06,
+      "loss": 0.5568,
+      "step": 35935
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.3381364345550537,
+      "learning_rate": 3.874803720131242e-06,
+      "loss": 0.5626,
+      "step": 35936
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 4.842716217041016,
+      "learning_rate": 3.8745468264297694e-06,
+      "loss": 0.677,
+      "step": 35937
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.1094589233398438,
+      "learning_rate": 3.8742899358577605e-06,
+      "loss": 0.5359,
+      "step": 35938
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.5509016513824463,
+      "learning_rate": 3.87403304841593e-06,
+      "loss": 0.7508,
+      "step": 35939
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.4744863510131836,
+      "learning_rate": 3.87377616410499e-06,
+      "loss": 0.4542,
+      "step": 35940
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 4.207470417022705,
+      "learning_rate": 3.873519282925658e-06,
+      "loss": 0.9819,
+      "step": 35941
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 3.5921359062194824,
+      "learning_rate": 3.873262404878646e-06,
+      "loss": 0.949,
+      "step": 35942
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.649414539337158,
+      "learning_rate": 3.87300552996467e-06,
+      "loss": 0.6244,
+      "step": 35943
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.4365508556365967,
+      "learning_rate": 3.8727486581844434e-06,
+      "loss": 0.8588,
+      "step": 35944
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.1592180728912354,
+      "learning_rate": 3.87249178953868e-06,
+      "loss": 0.5647,
+      "step": 35945
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.853060007095337,
+      "learning_rate": 3.872234924028095e-06,
+      "loss": 0.5485,
+      "step": 35946
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.2498810291290283,
+      "learning_rate": 3.871978061653402e-06,
+      "loss": 0.7555,
+      "step": 35947
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.481632709503174,
+      "learning_rate": 3.871721202415314e-06,
+      "loss": 0.6837,
+      "step": 35948
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.4127719402313232,
+      "learning_rate": 3.871464346314548e-06,
+      "loss": 0.5967,
+      "step": 35949
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.231318235397339,
+      "learning_rate": 3.871207493351817e-06,
+      "loss": 0.699,
+      "step": 35950
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.2741012573242188,
+      "learning_rate": 3.870950643527836e-06,
+      "loss": 0.5089,
+      "step": 35951
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.8352980613708496,
+      "learning_rate": 3.870693796843317e-06,
+      "loss": 0.6721,
+      "step": 35952
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 3.580444097518921,
+      "learning_rate": 3.870436953298975e-06,
+      "loss": 0.6048,
+      "step": 35953
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.3491663932800293,
+      "learning_rate": 3.870180112895525e-06,
+      "loss": 0.6758,
+      "step": 35954
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.185364246368408,
+      "learning_rate": 3.869923275633682e-06,
+      "loss": 0.7662,
+      "step": 35955
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.084780693054199,
+      "learning_rate": 3.8696664415141584e-06,
+      "loss": 0.619,
+      "step": 35956
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.6524658203125,
+      "learning_rate": 3.869409610537669e-06,
+      "loss": 0.635,
+      "step": 35957
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 3.8047664165496826,
+      "learning_rate": 3.869152782704928e-06,
+      "loss": 0.6307,
+      "step": 35958
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.5882138013839722,
+      "learning_rate": 3.868895958016649e-06,
+      "loss": 0.4903,
+      "step": 35959
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.9878644943237305,
+      "learning_rate": 3.868639136473547e-06,
+      "loss": 0.7848,
+      "step": 35960
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.5064547061920166,
+      "learning_rate": 3.868382318076336e-06,
+      "loss": 0.8051,
+      "step": 35961
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 4.377518653869629,
+      "learning_rate": 3.868125502825729e-06,
+      "loss": 0.778,
+      "step": 35962
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.171158790588379,
+      "learning_rate": 3.8678686907224424e-06,
+      "loss": 0.5235,
+      "step": 35963
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 5.088051795959473,
+      "learning_rate": 3.8676118817671885e-06,
+      "loss": 0.4818,
+      "step": 35964
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.694417953491211,
+      "learning_rate": 3.867355075960681e-06,
+      "loss": 0.6338,
+      "step": 35965
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.8415855169296265,
+      "learning_rate": 3.867098273303635e-06,
+      "loss": 0.5457,
+      "step": 35966
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 3.6512622833251953,
+      "learning_rate": 3.866841473796764e-06,
+      "loss": 0.5539,
+      "step": 35967
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.264625310897827,
+      "learning_rate": 3.866584677440783e-06,
+      "loss": 0.481,
+      "step": 35968
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 3.0720999240875244,
+      "learning_rate": 3.866327884236406e-06,
+      "loss": 0.3124,
+      "step": 35969
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.9225292205810547,
+      "learning_rate": 3.866071094184345e-06,
+      "loss": 0.5995,
+      "step": 35970
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.6801915168762207,
+      "learning_rate": 3.865814307285317e-06,
+      "loss": 0.5941,
+      "step": 35971
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.598382592201233,
+      "learning_rate": 3.8655575235400335e-06,
+      "loss": 0.45,
+      "step": 35972
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.8723620176315308,
+      "learning_rate": 3.865300742949208e-06,
+      "loss": 0.5384,
+      "step": 35973
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.9870991706848145,
+      "learning_rate": 3.865043965513557e-06,
+      "loss": 0.7192,
+      "step": 35974
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.969667673110962,
+      "learning_rate": 3.864787191233795e-06,
+      "loss": 0.455,
+      "step": 35975
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 3.5887725353240967,
+      "learning_rate": 3.864530420110633e-06,
+      "loss": 0.7066,
+      "step": 35976
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 3.6940929889678955,
+      "learning_rate": 3.8642736521447886e-06,
+      "loss": 0.6491,
+      "step": 35977
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.5441030263900757,
+      "learning_rate": 3.864016887336972e-06,
+      "loss": 0.6071,
+      "step": 35978
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.692314863204956,
+      "learning_rate": 3.8637601256879e-06,
+      "loss": 0.743,
+      "step": 35979
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.69716215133667,
+      "learning_rate": 3.863503367198285e-06,
+      "loss": 0.5603,
+      "step": 35980
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 3.146416425704956,
+      "learning_rate": 3.863246611868842e-06,
+      "loss": 1.1179,
+      "step": 35981
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.9297913312911987,
+      "learning_rate": 3.8629898597002835e-06,
+      "loss": 0.3571,
+      "step": 35982
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 3.038597583770752,
+      "learning_rate": 3.862733110693325e-06,
+      "loss": 0.6127,
+      "step": 35983
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 4.426396369934082,
+      "learning_rate": 3.862476364848679e-06,
+      "loss": 0.7374,
+      "step": 35984
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.300452470779419,
+      "learning_rate": 3.862219622167061e-06,
+      "loss": 0.351,
+      "step": 35985
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.9245316982269287,
+      "learning_rate": 3.8619628826491824e-06,
+      "loss": 0.501,
+      "step": 35986
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.3130173683166504,
+      "learning_rate": 3.861706146295759e-06,
+      "loss": 0.4615,
+      "step": 35987
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.02363920211792,
+      "learning_rate": 3.861449413107506e-06,
+      "loss": 0.4562,
+      "step": 35988
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 3.223759651184082,
+      "learning_rate": 3.861192683085135e-06,
+      "loss": 0.8512,
+      "step": 35989
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.0080249309539795,
+      "learning_rate": 3.86093595622936e-06,
+      "loss": 0.7523,
+      "step": 35990
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.442638397216797,
+      "learning_rate": 3.860679232540895e-06,
+      "loss": 0.8177,
+      "step": 35991
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.8309053182601929,
+      "learning_rate": 3.860422512020455e-06,
+      "loss": 0.4535,
+      "step": 35992
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.3528668880462646,
+      "learning_rate": 3.860165794668753e-06,
+      "loss": 0.5702,
+      "step": 35993
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.0560081005096436,
+      "learning_rate": 3.859909080486501e-06,
+      "loss": 0.6035,
+      "step": 35994
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.2674102783203125,
+      "learning_rate": 3.859652369474417e-06,
+      "loss": 0.7301,
+      "step": 35995
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.3637094497680664,
+      "learning_rate": 3.8593956616332114e-06,
+      "loss": 0.6859,
+      "step": 35996
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.645488739013672,
+      "learning_rate": 3.8591389569636e-06,
+      "loss": 0.5292,
+      "step": 35997
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.0560710430145264,
+      "learning_rate": 3.858882255466296e-06,
+      "loss": 0.6615,
+      "step": 35998
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.393342971801758,
+      "learning_rate": 3.858625557142014e-06,
+      "loss": 0.8566,
+      "step": 35999
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.308171510696411,
+      "learning_rate": 3.858368861991465e-06,
+      "loss": 0.3975,
+      "step": 36000
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 4.257695198059082,
+      "learning_rate": 3.8581121700153654e-06,
+      "loss": 0.6591,
+      "step": 36001
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 3.970360040664673,
+      "learning_rate": 3.857855481214428e-06,
+      "loss": 0.8201,
+      "step": 36002
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.4546823501586914,
+      "learning_rate": 3.857598795589366e-06,
+      "loss": 0.7464,
+      "step": 36003
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 3.0243780612945557,
+      "learning_rate": 3.857342113140894e-06,
+      "loss": 0.4751,
+      "step": 36004
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.961784839630127,
+      "learning_rate": 3.8570854338697265e-06,
+      "loss": 0.4739,
+      "step": 36005
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.9855860471725464,
+      "learning_rate": 3.856828757776575e-06,
+      "loss": 0.5378,
+      "step": 36006
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.674375534057617,
+      "learning_rate": 3.8565720848621555e-06,
+      "loss": 0.6902,
+      "step": 36007
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.913672924041748,
+      "learning_rate": 3.85631541512718e-06,
+      "loss": 0.8352,
+      "step": 36008
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.185551166534424,
+      "learning_rate": 3.856058748572363e-06,
+      "loss": 0.5154,
+      "step": 36009
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.001309394836426,
+      "learning_rate": 3.855802085198417e-06,
+      "loss": 0.8736,
+      "step": 36010
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.3487045764923096,
+      "learning_rate": 3.8555454250060585e-06,
+      "loss": 0.6323,
+      "step": 36011
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.722203016281128,
+      "learning_rate": 3.855288767995998e-06,
+      "loss": 0.5071,
+      "step": 36012
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.1925463676452637,
+      "learning_rate": 3.8550321141689515e-06,
+      "loss": 0.6394,
+      "step": 36013
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.0760159492492676,
+      "learning_rate": 3.854775463525631e-06,
+      "loss": 0.5837,
+      "step": 36014
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.517543077468872,
+      "learning_rate": 3.854518816066752e-06,
+      "loss": 0.5322,
+      "step": 36015
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 3.1170647144317627,
+      "learning_rate": 3.854262171793027e-06,
+      "loss": 0.5092,
+      "step": 36016
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 3.166703701019287,
+      "learning_rate": 3.8540055307051696e-06,
+      "loss": 0.5586,
+      "step": 36017
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.3279623985290527,
+      "learning_rate": 3.853748892803893e-06,
+      "loss": 0.5814,
+      "step": 36018
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.7278873920440674,
+      "learning_rate": 3.853492258089912e-06,
+      "loss": 0.3017,
+      "step": 36019
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.920119047164917,
+      "learning_rate": 3.853235626563939e-06,
+      "loss": 0.4415,
+      "step": 36020
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.490065813064575,
+      "learning_rate": 3.852978998226688e-06,
+      "loss": 0.7834,
+      "step": 36021
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.806406021118164,
+      "learning_rate": 3.8527223730788725e-06,
+      "loss": 0.6979,
+      "step": 36022
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.210582971572876,
+      "learning_rate": 3.852465751121207e-06,
+      "loss": 0.4733,
+      "step": 36023
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.4355580806732178,
+      "learning_rate": 3.852209132354404e-06,
+      "loss": 0.7769,
+      "step": 36024
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.939204692840576,
+      "learning_rate": 3.851952516779179e-06,
+      "loss": 0.6151,
+      "step": 36025
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 3.9388842582702637,
+      "learning_rate": 3.851695904396242e-06,
+      "loss": 0.7253,
+      "step": 36026
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.9056463241577148,
+      "learning_rate": 3.851439295206309e-06,
+      "loss": 0.4655,
+      "step": 36027
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 3.5228488445281982,
+      "learning_rate": 3.851182689210094e-06,
+      "loss": 0.5361,
+      "step": 36028
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.218846321105957,
+      "learning_rate": 3.850926086408309e-06,
+      "loss": 0.6368,
+      "step": 36029
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 3.3832123279571533,
+      "learning_rate": 3.8506694868016674e-06,
+      "loss": 0.9496,
+      "step": 36030
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.5351498126983643,
+      "learning_rate": 3.850412890390884e-06,
+      "loss": 0.7929,
+      "step": 36031
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.267364978790283,
+      "learning_rate": 3.850156297176672e-06,
+      "loss": 0.536,
+      "step": 36032
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.2633535861968994,
+      "learning_rate": 3.8498997071597446e-06,
+      "loss": 0.519,
+      "step": 36033
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.7166662216186523,
+      "learning_rate": 3.849643120340812e-06,
+      "loss": 0.6759,
+      "step": 36034
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.6346380710601807,
+      "learning_rate": 3.849386536720595e-06,
+      "loss": 0.7629,
+      "step": 36035
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.9699437618255615,
+      "learning_rate": 3.849129956299801e-06,
+      "loss": 0.76,
+      "step": 36036
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.086820363998413,
+      "learning_rate": 3.8488733790791465e-06,
+      "loss": 0.7607,
+      "step": 36037
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.0420141220092773,
+      "learning_rate": 3.8486168050593446e-06,
+      "loss": 0.4346,
+      "step": 36038
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.508777618408203,
+      "learning_rate": 3.848360234241107e-06,
+      "loss": 0.6993,
+      "step": 36039
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 3.0663225650787354,
+      "learning_rate": 3.8481036666251484e-06,
+      "loss": 0.5214,
+      "step": 36040
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.1961827278137207,
+      "learning_rate": 3.847847102212182e-06,
+      "loss": 0.5113,
+      "step": 36041
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.8132240772247314,
+      "learning_rate": 3.847590541002921e-06,
+      "loss": 0.7492,
+      "step": 36042
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.987113118171692,
+      "learning_rate": 3.8473339829980795e-06,
+      "loss": 0.6002,
+      "step": 36043
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.5149462223052979,
+      "learning_rate": 3.84707742819837e-06,
+      "loss": 0.6635,
+      "step": 36044
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 3.0517375469207764,
+      "learning_rate": 3.846820876604507e-06,
+      "loss": 0.8667,
+      "step": 36045
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.180774688720703,
+      "learning_rate": 3.846564328217202e-06,
+      "loss": 0.6869,
+      "step": 36046
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.3323378562927246,
+      "learning_rate": 3.84630778303717e-06,
+      "loss": 0.5747,
+      "step": 36047
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.7241241931915283,
+      "learning_rate": 3.846051241065124e-06,
+      "loss": 0.7448,
+      "step": 36048
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.9537403583526611,
+      "learning_rate": 3.845794702301777e-06,
+      "loss": 0.4723,
+      "step": 36049
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.445112705230713,
+      "learning_rate": 3.845538166747843e-06,
+      "loss": 0.7401,
+      "step": 36050
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.0397746562957764,
+      "learning_rate": 3.845281634404035e-06,
+      "loss": 0.5554,
+      "step": 36051
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.4683828353881836,
+      "learning_rate": 3.845025105271066e-06,
+      "loss": 0.6327,
+      "step": 36052
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.9689773321151733,
+      "learning_rate": 3.844768579349649e-06,
+      "loss": 0.7053,
+      "step": 36053
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.18966007232666,
+      "learning_rate": 3.844512056640498e-06,
+      "loss": 0.6731,
+      "step": 36054
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.9282467365264893,
+      "learning_rate": 3.844255537144325e-06,
+      "loss": 0.6708,
+      "step": 36055
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.9663492441177368,
+      "learning_rate": 3.843999020861845e-06,
+      "loss": 0.5347,
+      "step": 36056
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.5771055221557617,
+      "learning_rate": 3.843742507793771e-06,
+      "loss": 0.3914,
+      "step": 36057
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.309196710586548,
+      "learning_rate": 3.843485997940817e-06,
+      "loss": 0.5015,
+      "step": 36058
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.0256965160369873,
+      "learning_rate": 3.843229491303695e-06,
+      "loss": 0.6803,
+      "step": 36059
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.8400490283966064,
+      "learning_rate": 3.842972987883117e-06,
+      "loss": 0.5621,
+      "step": 36060
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 3.478937864303589,
+      "learning_rate": 3.842716487679799e-06,
+      "loss": 0.7493,
+      "step": 36061
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.1415340900421143,
+      "learning_rate": 3.842459990694453e-06,
+      "loss": 0.5408,
+      "step": 36062
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 3.0425026416778564,
+      "learning_rate": 3.842203496927791e-06,
+      "loss": 0.8183,
+      "step": 36063
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 5.334286689758301,
+      "learning_rate": 3.841947006380528e-06,
+      "loss": 0.972,
+      "step": 36064
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.7188849449157715,
+      "learning_rate": 3.8416905190533775e-06,
+      "loss": 0.8098,
+      "step": 36065
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.363933563232422,
+      "learning_rate": 3.84143403494705e-06,
+      "loss": 0.676,
+      "step": 36066
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.7414499521255493,
+      "learning_rate": 3.8411775540622614e-06,
+      "loss": 0.3214,
+      "step": 36067
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.6300477981567383,
+      "learning_rate": 3.840921076399724e-06,
+      "loss": 0.594,
+      "step": 36068
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 4.185696125030518,
+      "learning_rate": 3.840664601960151e-06,
+      "loss": 0.5669,
+      "step": 36069
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.1921794414520264,
+      "learning_rate": 3.840408130744255e-06,
+      "loss": 0.568,
+      "step": 36070
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.4905730485916138,
+      "learning_rate": 3.840151662752748e-06,
+      "loss": 0.5588,
+      "step": 36071
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.9791783094406128,
+      "learning_rate": 3.839895197986347e-06,
+      "loss": 0.7352,
+      "step": 36072
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.9883739948272705,
+      "learning_rate": 3.8396387364457625e-06,
+      "loss": 0.3866,
+      "step": 36073
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.460608959197998,
+      "learning_rate": 3.839382278131708e-06,
+      "loss": 0.7279,
+      "step": 36074
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.6938750743865967,
+      "learning_rate": 3.839125823044895e-06,
+      "loss": 0.3971,
+      "step": 36075
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.558274745941162,
+      "learning_rate": 3.8388693711860395e-06,
+      "loss": 0.6995,
+      "step": 36076
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.9571659564971924,
+      "learning_rate": 3.8386129225558536e-06,
+      "loss": 0.7845,
+      "step": 36077
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.012859344482422,
+      "learning_rate": 3.83835647715505e-06,
+      "loss": 0.6275,
+      "step": 36078
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 3.217489004135132,
+      "learning_rate": 3.8381000349843426e-06,
+      "loss": 0.4191,
+      "step": 36079
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.118130683898926,
+      "learning_rate": 3.837843596044443e-06,
+      "loss": 0.342,
+      "step": 36080
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.2388176918029785,
+      "learning_rate": 3.837587160336065e-06,
+      "loss": 0.7019,
+      "step": 36081
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.1342341899871826,
+      "learning_rate": 3.837330727859921e-06,
+      "loss": 0.8962,
+      "step": 36082
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.1670241355895996,
+      "learning_rate": 3.8370742986167245e-06,
+      "loss": 0.3012,
+      "step": 36083
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 3.415010929107666,
+      "learning_rate": 3.8368178726071895e-06,
+      "loss": 0.7029,
+      "step": 36084
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 4.429166316986084,
+      "learning_rate": 3.836561449832029e-06,
+      "loss": 0.8925,
+      "step": 36085
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 3.382261276245117,
+      "learning_rate": 3.836305030291954e-06,
+      "loss": 0.782,
+      "step": 36086
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.855658769607544,
+      "learning_rate": 3.83604861398768e-06,
+      "loss": 0.4805,
+      "step": 36087
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 6.561038970947266,
+      "learning_rate": 3.835792200919918e-06,
+      "loss": 0.7217,
+      "step": 36088
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 3.430191993713379,
+      "learning_rate": 3.835535791089382e-06,
+      "loss": 0.4427,
+      "step": 36089
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.4981212615966797,
+      "learning_rate": 3.835279384496784e-06,
+      "loss": 0.5228,
+      "step": 36090
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.24436354637146,
+      "learning_rate": 3.8350229811428385e-06,
+      "loss": 0.5131,
+      "step": 36091
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.432018280029297,
+      "learning_rate": 3.834766581028258e-06,
+      "loss": 0.6757,
+      "step": 36092
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.8755367994308472,
+      "learning_rate": 3.834510184153754e-06,
+      "loss": 0.7383,
+      "step": 36093
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.1651225090026855,
+      "learning_rate": 3.834253790520042e-06,
+      "loss": 0.6063,
+      "step": 36094
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 3.4374747276306152,
+      "learning_rate": 3.833997400127831e-06,
+      "loss": 0.6866,
+      "step": 36095
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.9271981716156006,
+      "learning_rate": 3.833741012977838e-06,
+      "loss": 0.5874,
+      "step": 36096
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.4909961223602295,
+      "learning_rate": 3.8334846290707745e-06,
+      "loss": 0.6174,
+      "step": 36097
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 4.905289649963379,
+      "learning_rate": 3.833228248407353e-06,
+      "loss": 0.3996,
+      "step": 36098
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.0541739463806152,
+      "learning_rate": 3.832971870988288e-06,
+      "loss": 0.4156,
+      "step": 36099
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 3.419577121734619,
+      "learning_rate": 3.832715496814289e-06,
+      "loss": 0.4546,
+      "step": 36100
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 4.158425807952881,
+      "learning_rate": 3.832459125886072e-06,
+      "loss": 0.6561,
+      "step": 36101
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.96176278591156,
+      "learning_rate": 3.832202758204349e-06,
+      "loss": 0.776,
+      "step": 36102
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 3.7324018478393555,
+      "learning_rate": 3.831946393769832e-06,
+      "loss": 0.7554,
+      "step": 36103
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.687432289123535,
+      "learning_rate": 3.831690032583236e-06,
+      "loss": 0.7082,
+      "step": 36104
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 5.659268379211426,
+      "learning_rate": 3.831433674645272e-06,
+      "loss": 0.604,
+      "step": 36105
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.4958674907684326,
+      "learning_rate": 3.831177319956651e-06,
+      "loss": 0.6112,
+      "step": 36106
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.4466981887817383,
+      "learning_rate": 3.83092096851809e-06,
+      "loss": 0.9056,
+      "step": 36107
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.9327083826065063,
+      "learning_rate": 3.830664620330299e-06,
+      "loss": 0.605,
+      "step": 36108
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 5.326198577880859,
+      "learning_rate": 3.8304082753939926e-06,
+      "loss": 0.786,
+      "step": 36109
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 3.238987445831299,
+      "learning_rate": 3.8301519337098824e-06,
+      "loss": 0.8085,
+      "step": 36110
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.9440507888793945,
+      "learning_rate": 3.829895595278681e-06,
+      "loss": 0.7629,
+      "step": 36111
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 3.131002187728882,
+      "learning_rate": 3.829639260101102e-06,
+      "loss": 0.5364,
+      "step": 36112
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.4299726486206055,
+      "learning_rate": 3.8293829281778575e-06,
+      "loss": 0.6527,
+      "step": 36113
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 3.5412564277648926,
+      "learning_rate": 3.8291265995096605e-06,
+      "loss": 0.5349,
+      "step": 36114
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.8260852098464966,
+      "learning_rate": 3.8288702740972235e-06,
+      "loss": 0.6366,
+      "step": 36115
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.1314328908920288,
+      "learning_rate": 3.828613951941258e-06,
+      "loss": 0.2354,
+      "step": 36116
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 3.055150032043457,
+      "learning_rate": 3.828357633042481e-06,
+      "loss": 0.5287,
+      "step": 36117
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 6.120528221130371,
+      "learning_rate": 3.8281013174016014e-06,
+      "loss": 0.6616,
+      "step": 36118
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.9058657884597778,
+      "learning_rate": 3.827845005019332e-06,
+      "loss": 0.6819,
+      "step": 36119
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.2554361820220947,
+      "learning_rate": 3.827588695896389e-06,
+      "loss": 0.595,
+      "step": 36120
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.802480697631836,
+      "learning_rate": 3.827332390033481e-06,
+      "loss": 0.772,
+      "step": 36121
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.919746994972229,
+      "learning_rate": 3.827076087431323e-06,
+      "loss": 0.4265,
+      "step": 36122
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 3.031700849533081,
+      "learning_rate": 3.826819788090627e-06,
+      "loss": 0.7337,
+      "step": 36123
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.5906827449798584,
+      "learning_rate": 3.826563492012106e-06,
+      "loss": 0.4485,
+      "step": 36124
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.8046891689300537,
+      "learning_rate": 3.8263071991964715e-06,
+      "loss": 0.6211,
+      "step": 36125
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.8709616661071777,
+      "learning_rate": 3.826050909644437e-06,
+      "loss": 0.5624,
+      "step": 36126
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.7652544975280762,
+      "learning_rate": 3.8257946233567164e-06,
+      "loss": 0.5262,
+      "step": 36127
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.3143150806427,
+      "learning_rate": 3.8255383403340205e-06,
+      "loss": 0.6143,
+      "step": 36128
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.4049179553985596,
+      "learning_rate": 3.8252820605770626e-06,
+      "loss": 0.6268,
+      "step": 36129
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.8606847524642944,
+      "learning_rate": 3.825025784086554e-06,
+      "loss": 0.509,
+      "step": 36130
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 3.5513088703155518,
+      "learning_rate": 3.8247695108632095e-06,
+      "loss": 0.6679,
+      "step": 36131
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.6113128662109375,
+      "learning_rate": 3.824513240907742e-06,
+      "loss": 0.4941,
+      "step": 36132
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.142390012741089,
+      "learning_rate": 3.8242569742208596e-06,
+      "loss": 0.5584,
+      "step": 36133
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.9777079820632935,
+      "learning_rate": 3.82400071080328e-06,
+      "loss": 0.6946,
+      "step": 36134
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.4852231740951538,
+      "learning_rate": 3.823744450655714e-06,
+      "loss": 0.4336,
+      "step": 36135
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 3.331298828125,
+      "learning_rate": 3.823488193778873e-06,
+      "loss": 0.4543,
+      "step": 36136
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.467801094055176,
+      "learning_rate": 3.823231940173472e-06,
+      "loss": 0.5207,
+      "step": 36137
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.4367918968200684,
+      "learning_rate": 3.822975689840222e-06,
+      "loss": 0.5979,
+      "step": 36138
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.9064364433288574,
+      "learning_rate": 3.822719442779835e-06,
+      "loss": 0.7952,
+      "step": 36139
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.2768325805664062,
+      "learning_rate": 3.822463198993024e-06,
+      "loss": 0.6342,
+      "step": 36140
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 3.018636465072632,
+      "learning_rate": 3.822206958480503e-06,
+      "loss": 0.6589,
+      "step": 36141
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 3.267716407775879,
+      "learning_rate": 3.821950721242983e-06,
+      "loss": 0.5286,
+      "step": 36142
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.7318955659866333,
+      "learning_rate": 3.821694487281175e-06,
+      "loss": 0.6629,
+      "step": 36143
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.538015127182007,
+      "learning_rate": 3.821438256595796e-06,
+      "loss": 0.5837,
+      "step": 36144
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 3.6597883701324463,
+      "learning_rate": 3.8211820291875524e-06,
+      "loss": 0.765,
+      "step": 36145
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.3055241107940674,
+      "learning_rate": 3.820925805057162e-06,
+      "loss": 0.5117,
+      "step": 36146
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.132258176803589,
+      "learning_rate": 3.820669584205335e-06,
+      "loss": 0.754,
+      "step": 36147
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.8761472702026367,
+      "learning_rate": 3.820413366632784e-06,
+      "loss": 0.5278,
+      "step": 36148
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.2182276248931885,
+      "learning_rate": 3.820157152340222e-06,
+      "loss": 0.5293,
+      "step": 36149
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.3027963638305664,
+      "learning_rate": 3.81990094132836e-06,
+      "loss": 0.4958,
+      "step": 36150
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.6905510425567627,
+      "learning_rate": 3.819644733597912e-06,
+      "loss": 0.5334,
+      "step": 36151
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.2976453304290771,
+      "learning_rate": 3.819388529149591e-06,
+      "loss": 0.214,
+      "step": 36152
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 3.138606309890747,
+      "learning_rate": 3.819132327984107e-06,
+      "loss": 0.7803,
+      "step": 36153
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.6393927335739136,
+      "learning_rate": 3.818876130102174e-06,
+      "loss": 0.7876,
+      "step": 36154
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 4.730005741119385,
+      "learning_rate": 3.818619935504503e-06,
+      "loss": 0.6409,
+      "step": 36155
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 3.6806674003601074,
+      "learning_rate": 3.818363744191807e-06,
+      "loss": 0.687,
+      "step": 36156
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.741044759750366,
+      "learning_rate": 3.818107556164799e-06,
+      "loss": 0.5871,
+      "step": 36157
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 3.9725863933563232,
+      "learning_rate": 3.817851371424193e-06,
+      "loss": 0.8252,
+      "step": 36158
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.6824381351470947,
+      "learning_rate": 3.817595189970699e-06,
+      "loss": 0.4704,
+      "step": 36159
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.7609138488769531,
+      "learning_rate": 3.8173390118050295e-06,
+      "loss": 0.1821,
+      "step": 36160
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.178112745285034,
+      "learning_rate": 3.817082836927897e-06,
+      "loss": 0.5338,
+      "step": 36161
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 3.718850612640381,
+      "learning_rate": 3.8168266653400135e-06,
+      "loss": 0.715,
+      "step": 36162
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.3424012660980225,
+      "learning_rate": 3.8165704970420925e-06,
+      "loss": 0.7869,
+      "step": 36163
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.5933878421783447,
+      "learning_rate": 3.816314332034846e-06,
+      "loss": 0.864,
+      "step": 36164
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.222222089767456,
+      "learning_rate": 3.816058170318985e-06,
+      "loss": 0.7901,
+      "step": 36165
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 4.3495965003967285,
+      "learning_rate": 3.815802011895223e-06,
+      "loss": 0.8351,
+      "step": 36166
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 3.031637191772461,
+      "learning_rate": 3.815545856764273e-06,
+      "loss": 0.7693,
+      "step": 36167
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.9530760049819946,
+      "learning_rate": 3.815289704926846e-06,
+      "loss": 0.6683,
+      "step": 36168
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 4.360369682312012,
+      "learning_rate": 3.8150335563836535e-06,
+      "loss": 0.8756,
+      "step": 36169
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.615123748779297,
+      "learning_rate": 3.814777411135409e-06,
+      "loss": 0.8833,
+      "step": 36170
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.147902488708496,
+      "learning_rate": 3.814521269182825e-06,
+      "loss": 0.6035,
+      "step": 36171
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.0732038021087646,
+      "learning_rate": 3.814265130526613e-06,
+      "loss": 0.2742,
+      "step": 36172
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.8132623434066772,
+      "learning_rate": 3.814008995167486e-06,
+      "loss": 0.8799,
+      "step": 36173
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 3.246005058288574,
+      "learning_rate": 3.813752863106155e-06,
+      "loss": 0.5492,
+      "step": 36174
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 3.022765636444092,
+      "learning_rate": 3.813496734343333e-06,
+      "loss": 0.6842,
+      "step": 36175
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.983499050140381,
+      "learning_rate": 3.8132406088797323e-06,
+      "loss": 0.7363,
+      "step": 36176
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.4007186889648438,
+      "learning_rate": 3.8129844867160626e-06,
+      "loss": 0.6635,
+      "step": 36177
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.115041971206665,
+      "learning_rate": 3.812728367853041e-06,
+      "loss": 0.6685,
+      "step": 36178
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.527210235595703,
+      "learning_rate": 3.8124722522913764e-06,
+      "loss": 0.3561,
+      "step": 36179
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.7264299392700195,
+      "learning_rate": 3.8122161400317826e-06,
+      "loss": 0.6016,
+      "step": 36180
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.9471725225448608,
+      "learning_rate": 3.81196003107497e-06,
+      "loss": 0.5404,
+      "step": 36181
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.2646353244781494,
+      "learning_rate": 3.8117039254216513e-06,
+      "loss": 0.3028,
+      "step": 36182
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.4015188217163086,
+      "learning_rate": 3.8114478230725395e-06,
+      "loss": 0.5683,
+      "step": 36183
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.3730058670043945,
+      "learning_rate": 3.811191724028345e-06,
+      "loss": 0.6823,
+      "step": 36184
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 3.155302047729492,
+      "learning_rate": 3.810935628289782e-06,
+      "loss": 0.4731,
+      "step": 36185
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.9632307291030884,
+      "learning_rate": 3.810679535857561e-06,
+      "loss": 0.6119,
+      "step": 36186
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.470306634902954,
+      "learning_rate": 3.810423446732395e-06,
+      "loss": 0.3533,
+      "step": 36187
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.7175824642181396,
+      "learning_rate": 3.810167360914996e-06,
+      "loss": 0.406,
+      "step": 36188
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.388042688369751,
+      "learning_rate": 3.8099112784060754e-06,
+      "loss": 0.6497,
+      "step": 36189
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 3.3115909099578857,
+      "learning_rate": 3.8096551992063456e-06,
+      "loss": 0.7048,
+      "step": 36190
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.043625831604004,
+      "learning_rate": 3.8093991233165195e-06,
+      "loss": 0.4141,
+      "step": 36191
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.892606258392334,
+      "learning_rate": 3.8091430507373076e-06,
+      "loss": 0.6304,
+      "step": 36192
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 5.4804792404174805,
+      "learning_rate": 3.8088869814694233e-06,
+      "loss": 0.4356,
+      "step": 36193
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 4.732316017150879,
+      "learning_rate": 3.808630915513578e-06,
+      "loss": 0.5516,
+      "step": 36194
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 3.216686725616455,
+      "learning_rate": 3.8083748528704827e-06,
+      "loss": 0.6574,
+      "step": 36195
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.8046084642410278,
+      "learning_rate": 3.8081187935408516e-06,
+      "loss": 0.4801,
+      "step": 36196
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.997802734375,
+      "learning_rate": 3.8078627375253945e-06,
+      "loss": 0.7853,
+      "step": 36197
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.861638069152832,
+      "learning_rate": 3.807606684824826e-06,
+      "loss": 0.5076,
+      "step": 36198
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.954372763633728,
+      "learning_rate": 3.807350635439856e-06,
+      "loss": 0.6321,
+      "step": 36199
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.6800637245178223,
+      "learning_rate": 3.8070945893711975e-06,
+      "loss": 0.7895,
+      "step": 36200
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 4.1619744300842285,
+      "learning_rate": 3.8068385466195622e-06,
+      "loss": 0.5644,
+      "step": 36201
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 3.254183053970337,
+      "learning_rate": 3.8065825071856614e-06,
+      "loss": 0.5218,
+      "step": 36202
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.7796573638916016,
+      "learning_rate": 3.806326471070208e-06,
+      "loss": 0.5822,
+      "step": 36203
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 3.172041416168213,
+      "learning_rate": 3.806070438273914e-06,
+      "loss": 0.3928,
+      "step": 36204
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.5818676948547363,
+      "learning_rate": 3.8058144087974902e-06,
+      "loss": 0.7721,
+      "step": 36205
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.3230319023132324,
+      "learning_rate": 3.8055583826416483e-06,
+      "loss": 0.5997,
+      "step": 36206
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.687213659286499,
+      "learning_rate": 3.8053023598071017e-06,
+      "loss": 0.6118,
+      "step": 36207
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 4.076866626739502,
+      "learning_rate": 3.805046340294562e-06,
+      "loss": 0.5994,
+      "step": 36208
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.6067001819610596,
+      "learning_rate": 3.804790324104741e-06,
+      "loss": 0.6238,
+      "step": 36209
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 3.5701069831848145,
+      "learning_rate": 3.80453431123835e-06,
+      "loss": 0.7845,
+      "step": 36210
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.4052045345306396,
+      "learning_rate": 3.8042783016961015e-06,
+      "loss": 0.5458,
+      "step": 36211
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 3.1754472255706787,
+      "learning_rate": 3.8040222954787075e-06,
+      "loss": 0.9451,
+      "step": 36212
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.9698084592819214,
+      "learning_rate": 3.803766292586879e-06,
+      "loss": 0.7255,
+      "step": 36213
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.0834801197052,
+      "learning_rate": 3.8035102930213288e-06,
+      "loss": 0.7571,
+      "step": 36214
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.4033360481262207,
+      "learning_rate": 3.803254296782767e-06,
+      "loss": 0.5796,
+      "step": 36215
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.4671480655670166,
+      "learning_rate": 3.8029983038719075e-06,
+      "loss": 0.6385,
+      "step": 36216
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 3.394123077392578,
+      "learning_rate": 3.8027423142894604e-06,
+      "loss": 0.6165,
+      "step": 36217
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 3.1174676418304443,
+      "learning_rate": 3.8024863280361386e-06,
+      "loss": 0.5525,
+      "step": 36218
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.6321113109588623,
+      "learning_rate": 3.802230345112654e-06,
+      "loss": 0.5509,
+      "step": 36219
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 3.4402709007263184,
+      "learning_rate": 3.801974365519719e-06,
+      "loss": 0.6376,
+      "step": 36220
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 3.321690559387207,
+      "learning_rate": 3.801718389258044e-06,
+      "loss": 0.7211,
+      "step": 36221
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.7455790042877197,
+      "learning_rate": 3.8014624163283413e-06,
+      "loss": 0.655,
+      "step": 36222
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.9658405780792236,
+      "learning_rate": 3.801206446731323e-06,
+      "loss": 0.6829,
+      "step": 36223
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 6.068020343780518,
+      "learning_rate": 3.8009504804677e-06,
+      "loss": 0.6813,
+      "step": 36224
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.6087709665298462,
+      "learning_rate": 3.800694517538185e-06,
+      "loss": 0.5329,
+      "step": 36225
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 3.0670032501220703,
+      "learning_rate": 3.800438557943489e-06,
+      "loss": 0.6261,
+      "step": 36226
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 4.938014507293701,
+      "learning_rate": 3.800182601684324e-06,
+      "loss": 0.7697,
+      "step": 36227
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 4.3319597244262695,
+      "learning_rate": 3.7999266487614017e-06,
+      "loss": 0.9469,
+      "step": 36228
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.592162847518921,
+      "learning_rate": 3.799670699175434e-06,
+      "loss": 0.7049,
+      "step": 36229
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.606741189956665,
+      "learning_rate": 3.799414752927132e-06,
+      "loss": 0.651,
+      "step": 36230
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.2697179317474365,
+      "learning_rate": 3.799158810017208e-06,
+      "loss": 0.711,
+      "step": 36231
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.2350029945373535,
+      "learning_rate": 3.798902870446374e-06,
+      "loss": 0.5945,
+      "step": 36232
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.569638729095459,
+      "learning_rate": 3.798646934215341e-06,
+      "loss": 0.6313,
+      "step": 36233
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 3.7227704524993896,
+      "learning_rate": 3.7983910013248204e-06,
+      "loss": 0.6838,
+      "step": 36234
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.3055107593536377,
+      "learning_rate": 3.7981350717755243e-06,
+      "loss": 0.6877,
+      "step": 36235
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.9273301362991333,
+      "learning_rate": 3.7978791455681653e-06,
+      "loss": 0.6303,
+      "step": 36236
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 3.1414051055908203,
+      "learning_rate": 3.7976232227034524e-06,
+      "loss": 0.4519,
+      "step": 36237
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.2445638179779053,
+      "learning_rate": 3.7973673031820985e-06,
+      "loss": 0.5712,
+      "step": 36238
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.9504032135009766,
+      "learning_rate": 3.7971113870048166e-06,
+      "loss": 0.6157,
+      "step": 36239
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.6576881408691406,
+      "learning_rate": 3.796855474172318e-06,
+      "loss": 0.7204,
+      "step": 36240
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.1669867038726807,
+      "learning_rate": 3.796599564685313e-06,
+      "loss": 0.6858,
+      "step": 36241
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.7243406772613525,
+      "learning_rate": 3.796343658544514e-06,
+      "loss": 0.656,
+      "step": 36242
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 3.3848893642425537,
+      "learning_rate": 3.7960877557506327e-06,
+      "loss": 0.5091,
+      "step": 36243
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.594895124435425,
+      "learning_rate": 3.79583185630438e-06,
+      "loss": 0.804,
+      "step": 36244
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 3.304497003555298,
+      "learning_rate": 3.795575960206468e-06,
+      "loss": 0.867,
+      "step": 36245
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.732821226119995,
+      "learning_rate": 3.7953200674576087e-06,
+      "loss": 0.5942,
+      "step": 36246
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.6945934295654297,
+      "learning_rate": 3.795064178058512e-06,
+      "loss": 0.6474,
+      "step": 36247
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.6579959392547607,
+      "learning_rate": 3.7948082920098905e-06,
+      "loss": 0.4314,
+      "step": 36248
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.6110005378723145,
+      "learning_rate": 3.7945524093124566e-06,
+      "loss": 0.62,
+      "step": 36249
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.739461898803711,
+      "learning_rate": 3.7942965299669197e-06,
+      "loss": 0.4715,
+      "step": 36250
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.4445276260375977,
+      "learning_rate": 3.7940406539739926e-06,
+      "loss": 0.5247,
+      "step": 36251
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.744036316871643,
+      "learning_rate": 3.7937847813343876e-06,
+      "loss": 0.4373,
+      "step": 36252
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.771211862564087,
+      "learning_rate": 3.793528912048815e-06,
+      "loss": 0.495,
+      "step": 36253
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.92680287361145,
+      "learning_rate": 3.7932730461179858e-06,
+      "loss": 0.8699,
+      "step": 36254
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.4513070583343506,
+      "learning_rate": 3.793017183542613e-06,
+      "loss": 0.6173,
+      "step": 36255
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.184386730194092,
+      "learning_rate": 3.792761324323406e-06,
+      "loss": 0.7412,
+      "step": 36256
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.628286600112915,
+      "learning_rate": 3.7925054684610786e-06,
+      "loss": 0.8393,
+      "step": 36257
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.027081251144409,
+      "learning_rate": 3.792249615956339e-06,
+      "loss": 0.56,
+      "step": 36258
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.7771852016448975,
+      "learning_rate": 3.791993766809903e-06,
+      "loss": 0.5645,
+      "step": 36259
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.2013020515441895,
+      "learning_rate": 3.79173792102248e-06,
+      "loss": 0.6264,
+      "step": 36260
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.5734620094299316,
+      "learning_rate": 3.791482078594781e-06,
+      "loss": 0.6152,
+      "step": 36261
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 3.1551949977874756,
+      "learning_rate": 3.7912262395275175e-06,
+      "loss": 0.5301,
+      "step": 36262
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.272324562072754,
+      "learning_rate": 3.7909704038214014e-06,
+      "loss": 0.5898,
+      "step": 36263
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.9780373573303223,
+      "learning_rate": 3.7907145714771426e-06,
+      "loss": 0.8427,
+      "step": 36264
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.593480110168457,
+      "learning_rate": 3.790458742495454e-06,
+      "loss": 0.6247,
+      "step": 36265
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.034811019897461,
+      "learning_rate": 3.7902029168770473e-06,
+      "loss": 0.5272,
+      "step": 36266
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 4.554325103759766,
+      "learning_rate": 3.7899470946226326e-06,
+      "loss": 0.4166,
+      "step": 36267
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.9518420696258545,
+      "learning_rate": 3.789691275732921e-06,
+      "loss": 0.5489,
+      "step": 36268
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.9485056400299072,
+      "learning_rate": 3.7894354602086246e-06,
+      "loss": 0.5918,
+      "step": 36269
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.1965184211730957,
+      "learning_rate": 3.7891796480504556e-06,
+      "loss": 0.678,
+      "step": 36270
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.0945515632629395,
+      "learning_rate": 3.7889238392591243e-06,
+      "loss": 0.3642,
+      "step": 36271
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.072899580001831,
+      "learning_rate": 3.788668033835342e-06,
+      "loss": 0.4407,
+      "step": 36272
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 6.081783294677734,
+      "learning_rate": 3.78841223177982e-06,
+      "loss": 0.476,
+      "step": 36273
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.084902048110962,
+      "learning_rate": 3.7881564330932696e-06,
+      "loss": 0.55,
+      "step": 36274
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.1838173866271973,
+      "learning_rate": 3.7879006377764025e-06,
+      "loss": 0.5972,
+      "step": 36275
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.219367027282715,
+      "learning_rate": 3.7876448458299296e-06,
+      "loss": 0.6075,
+      "step": 36276
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.1800005435943604,
+      "learning_rate": 3.7873890572545626e-06,
+      "loss": 0.6152,
+      "step": 36277
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.600876808166504,
+      "learning_rate": 3.7871332720510106e-06,
+      "loss": 0.5318,
+      "step": 36278
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 5.049709796905518,
+      "learning_rate": 3.786877490219988e-06,
+      "loss": 0.6799,
+      "step": 36279
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 4.613811492919922,
+      "learning_rate": 3.786621711762204e-06,
+      "loss": 0.4793,
+      "step": 36280
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.6313896179199219,
+      "learning_rate": 3.7863659366783704e-06,
+      "loss": 0.4988,
+      "step": 36281
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 3.8993265628814697,
+      "learning_rate": 3.7861101649691993e-06,
+      "loss": 0.9134,
+      "step": 36282
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 4.355254650115967,
+      "learning_rate": 3.785854396635401e-06,
+      "loss": 0.5765,
+      "step": 36283
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.427157163619995,
+      "learning_rate": 3.785598631677687e-06,
+      "loss": 0.6115,
+      "step": 36284
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 3.686898946762085,
+      "learning_rate": 3.7853428700967683e-06,
+      "loss": 0.6618,
+      "step": 36285
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 3.104118824005127,
+      "learning_rate": 3.785087111893356e-06,
+      "loss": 0.9432,
+      "step": 36286
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.7703653573989868,
+      "learning_rate": 3.784831357068161e-06,
+      "loss": 0.6662,
+      "step": 36287
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.153684377670288,
+      "learning_rate": 3.784575605621895e-06,
+      "loss": 0.5732,
+      "step": 36288
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.0624430179595947,
+      "learning_rate": 3.7843198575552685e-06,
+      "loss": 0.5802,
+      "step": 36289
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.0706913471221924,
+      "learning_rate": 3.7840641128689937e-06,
+      "loss": 0.5963,
+      "step": 36290
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.492658853530884,
+      "learning_rate": 3.783808371563781e-06,
+      "loss": 0.6885,
+      "step": 36291
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.780444860458374,
+      "learning_rate": 3.7835526336403418e-06,
+      "loss": 0.7275,
+      "step": 36292
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.236898183822632,
+      "learning_rate": 3.7832968990993867e-06,
+      "loss": 0.5956,
+      "step": 36293
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.691391706466675,
+      "learning_rate": 3.7830411679416266e-06,
+      "loss": 0.661,
+      "step": 36294
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.16703724861145,
+      "learning_rate": 3.7827854401677745e-06,
+      "loss": 0.5114,
+      "step": 36295
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.551158905029297,
+      "learning_rate": 3.782529715778539e-06,
+      "loss": 0.5945,
+      "step": 36296
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 3.010545253753662,
+      "learning_rate": 3.7822739947746325e-06,
+      "loss": 0.9532,
+      "step": 36297
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.551688313484192,
+      "learning_rate": 3.782018277156766e-06,
+      "loss": 0.6521,
+      "step": 36298
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 3.1383774280548096,
+      "learning_rate": 3.781762562925648e-06,
+      "loss": 0.7866,
+      "step": 36299
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.715067982673645,
+      "learning_rate": 3.781506852081994e-06,
+      "loss": 0.5453,
+      "step": 36300
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.624399185180664,
+      "learning_rate": 3.7812511446265138e-06,
+      "loss": 0.6029,
+      "step": 36301
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.6362171173095703,
+      "learning_rate": 3.780995440559917e-06,
+      "loss": 0.58,
+      "step": 36302
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.0058579444885254,
+      "learning_rate": 3.7807397398829153e-06,
+      "loss": 0.6396,
+      "step": 36303
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.887436628341675,
+      "learning_rate": 3.7804840425962197e-06,
+      "loss": 0.3671,
+      "step": 36304
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 3.0086398124694824,
+      "learning_rate": 3.780228348700541e-06,
+      "loss": 0.5762,
+      "step": 36305
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.728372097015381,
+      "learning_rate": 3.7799726581965908e-06,
+      "loss": 0.7922,
+      "step": 36306
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.7670323848724365,
+      "learning_rate": 3.7797169710850788e-06,
+      "loss": 0.4274,
+      "step": 36307
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 3.699151039123535,
+      "learning_rate": 3.7794612873667177e-06,
+      "loss": 0.7819,
+      "step": 36308
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.1018640995025635,
+      "learning_rate": 3.779205607042217e-06,
+      "loss": 0.4776,
+      "step": 36309
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.3622725009918213,
+      "learning_rate": 3.7789499301122878e-06,
+      "loss": 0.6659,
+      "step": 36310
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.975994110107422,
+      "learning_rate": 3.778694256577642e-06,
+      "loss": 0.4607,
+      "step": 36311
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 4.297097682952881,
+      "learning_rate": 3.77843858643899e-06,
+      "loss": 0.8235,
+      "step": 36312
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.603790044784546,
+      "learning_rate": 3.7781829196970425e-06,
+      "loss": 0.6912,
+      "step": 36313
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.960792899131775,
+      "learning_rate": 3.7779272563525106e-06,
+      "loss": 0.5744,
+      "step": 36314
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 3.0684618949890137,
+      "learning_rate": 3.777671596406105e-06,
+      "loss": 0.4557,
+      "step": 36315
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 3.4439005851745605,
+      "learning_rate": 3.7774159398585373e-06,
+      "loss": 0.7719,
+      "step": 36316
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.1242921352386475,
+      "learning_rate": 3.7771602867105172e-06,
+      "loss": 0.4913,
+      "step": 36317
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.2055931091308594,
+      "learning_rate": 3.776904636962756e-06,
+      "loss": 0.5421,
+      "step": 36318
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.8188118934631348,
+      "learning_rate": 3.776648990615964e-06,
+      "loss": 0.6431,
+      "step": 36319
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.922858476638794,
+      "learning_rate": 3.776393347670855e-06,
+      "loss": 0.6153,
+      "step": 36320
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.6592246294021606,
+      "learning_rate": 3.776137708128137e-06,
+      "loss": 0.3856,
+      "step": 36321
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.524017572402954,
+      "learning_rate": 3.7758820719885213e-06,
+      "loss": 0.4733,
+      "step": 36322
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.7190388441085815,
+      "learning_rate": 3.7756264392527193e-06,
+      "loss": 0.3026,
+      "step": 36323
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.831874132156372,
+      "learning_rate": 3.7753708099214416e-06,
+      "loss": 0.8277,
+      "step": 36324
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.7346227169036865,
+      "learning_rate": 3.7751151839953986e-06,
+      "loss": 0.5164,
+      "step": 36325
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.9494953155517578,
+      "learning_rate": 3.774859561475302e-06,
+      "loss": 0.4537,
+      "step": 36326
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.3753039836883545,
+      "learning_rate": 3.7746039423618617e-06,
+      "loss": 0.7973,
+      "step": 36327
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.8060007095336914,
+      "learning_rate": 3.774348326655789e-06,
+      "loss": 0.8414,
+      "step": 36328
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 3.194275140762329,
+      "learning_rate": 3.774092714357794e-06,
+      "loss": 0.557,
+      "step": 36329
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.0865354537963867,
+      "learning_rate": 3.7738371054685875e-06,
+      "loss": 0.4717,
+      "step": 36330
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.7925121784210205,
+      "learning_rate": 3.773581499988881e-06,
+      "loss": 0.5335,
+      "step": 36331
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 3.208523988723755,
+      "learning_rate": 3.7733258979193854e-06,
+      "loss": 0.555,
+      "step": 36332
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.5060043334960938,
+      "learning_rate": 3.77307029926081e-06,
+      "loss": 0.5268,
+      "step": 36333
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.8851195573806763,
+      "learning_rate": 3.772814704013868e-06,
+      "loss": 0.5506,
+      "step": 36334
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 4.269241809844971,
+      "learning_rate": 3.7725591121792678e-06,
+      "loss": 0.6237,
+      "step": 36335
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.4356456995010376,
+      "learning_rate": 3.772303523757721e-06,
+      "loss": 0.3992,
+      "step": 36336
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 3.1433165073394775,
+      "learning_rate": 3.772047938749938e-06,
+      "loss": 0.6536,
+      "step": 36337
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 3.307384967803955,
+      "learning_rate": 3.77179235715663e-06,
+      "loss": 0.599,
+      "step": 36338
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.4512460231781006,
+      "learning_rate": 3.771536778978506e-06,
+      "loss": 0.559,
+      "step": 36339
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.4913547039031982,
+      "learning_rate": 3.7712812042162795e-06,
+      "loss": 0.5494,
+      "step": 36340
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.1760196685791016,
+      "learning_rate": 3.7710256328706586e-06,
+      "loss": 0.8307,
+      "step": 36341
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 3.7139523029327393,
+      "learning_rate": 3.7707700649423557e-06,
+      "loss": 0.6727,
+      "step": 36342
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.278118133544922,
+      "learning_rate": 3.770514500432081e-06,
+      "loss": 0.6123,
+      "step": 36343
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.3527824878692627,
+      "learning_rate": 3.770258939340545e-06,
+      "loss": 0.5782,
+      "step": 36344
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.670184850692749,
+      "learning_rate": 3.770003381668458e-06,
+      "loss": 0.6412,
+      "step": 36345
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.7703375816345215,
+      "learning_rate": 3.7697478274165306e-06,
+      "loss": 0.454,
+      "step": 36346
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.6179726123809814,
+      "learning_rate": 3.7694922765854743e-06,
+      "loss": 0.6919,
+      "step": 36347
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.4085662364959717,
+      "learning_rate": 3.7692367291759986e-06,
+      "loss": 0.8977,
+      "step": 36348
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.881883382797241,
+      "learning_rate": 3.7689811851888143e-06,
+      "loss": 0.8745,
+      "step": 36349
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 4.063307762145996,
+      "learning_rate": 3.7687256446246325e-06,
+      "loss": 0.7591,
+      "step": 36350
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 3.8483099937438965,
+      "learning_rate": 3.7684701074841633e-06,
+      "loss": 0.9065,
+      "step": 36351
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.9830279350280762,
+      "learning_rate": 3.7682145737681174e-06,
+      "loss": 0.5432,
+      "step": 36352
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 5.470386028289795,
+      "learning_rate": 3.767959043477205e-06,
+      "loss": 1.0339,
+      "step": 36353
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.071329355239868,
+      "learning_rate": 3.767703516612138e-06,
+      "loss": 0.7244,
+      "step": 36354
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 9.936063766479492,
+      "learning_rate": 3.7674479931736253e-06,
+      "loss": 0.3897,
+      "step": 36355
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.8149423599243164,
+      "learning_rate": 3.7671924731623777e-06,
+      "loss": 0.6164,
+      "step": 36356
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.076777935028076,
+      "learning_rate": 3.766936956579106e-06,
+      "loss": 0.7644,
+      "step": 36357
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 3.8376622200012207,
+      "learning_rate": 3.7666814434245215e-06,
+      "loss": 0.7693,
+      "step": 36358
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.359785318374634,
+      "learning_rate": 3.7664259336993333e-06,
+      "loss": 0.723,
+      "step": 36359
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 3.809561014175415,
+      "learning_rate": 3.76617042740425e-06,
+      "loss": 0.3584,
+      "step": 36360
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 3.6466729640960693,
+      "learning_rate": 3.7659149245399875e-06,
+      "loss": 0.714,
+      "step": 36361
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.110926628112793,
+      "learning_rate": 3.7656594251072525e-06,
+      "loss": 0.7652,
+      "step": 36362
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.191100597381592,
+      "learning_rate": 3.765403929106757e-06,
+      "loss": 0.5655,
+      "step": 36363
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 3.13201642036438,
+      "learning_rate": 3.76514843653921e-06,
+      "loss": 0.7756,
+      "step": 36364
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.8209173679351807,
+      "learning_rate": 3.764892947405323e-06,
+      "loss": 0.4817,
+      "step": 36365
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.42155122756958,
+      "learning_rate": 3.7646374617058056e-06,
+      "loss": 0.6621,
+      "step": 36366
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.7817392349243164,
+      "learning_rate": 3.764381979441369e-06,
+      "loss": 0.6652,
+      "step": 36367
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.8706309795379639,
+      "learning_rate": 3.764126500612723e-06,
+      "loss": 0.5756,
+      "step": 36368
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 3.0374903678894043,
+      "learning_rate": 3.763871025220579e-06,
+      "loss": 0.605,
+      "step": 36369
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.5161099433898926,
+      "learning_rate": 3.763615553265646e-06,
+      "loss": 0.5608,
+      "step": 36370
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.44447922706604,
+      "learning_rate": 3.763360084748635e-06,
+      "loss": 0.6457,
+      "step": 36371
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 3.182863473892212,
+      "learning_rate": 3.763104619670256e-06,
+      "loss": 0.6725,
+      "step": 36372
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.2080037593841553,
+      "learning_rate": 3.7628491580312204e-06,
+      "loss": 0.5637,
+      "step": 36373
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 3.430755615234375,
+      "learning_rate": 3.7625936998322377e-06,
+      "loss": 0.4028,
+      "step": 36374
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.525012969970703,
+      "learning_rate": 3.762338245074018e-06,
+      "loss": 0.6342,
+      "step": 36375
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.7012524604797363,
+      "learning_rate": 3.7620827937572725e-06,
+      "loss": 0.5982,
+      "step": 36376
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.1533846855163574,
+      "learning_rate": 3.7618273458827105e-06,
+      "loss": 0.4151,
+      "step": 36377
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.795283079147339,
+      "learning_rate": 3.761571901451043e-06,
+      "loss": 0.6717,
+      "step": 36378
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.7702982425689697,
+      "learning_rate": 3.7613164604629805e-06,
+      "loss": 0.4093,
+      "step": 36379
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.390728235244751,
+      "learning_rate": 3.76106102291923e-06,
+      "loss": 0.718,
+      "step": 36380
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 4.564661979675293,
+      "learning_rate": 3.7608055888205077e-06,
+      "loss": 0.5796,
+      "step": 36381
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 4.296501159667969,
+      "learning_rate": 3.7605501581675197e-06,
+      "loss": 0.3863,
+      "step": 36382
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.1304197311401367,
+      "learning_rate": 3.7602947309609783e-06,
+      "loss": 0.5784,
+      "step": 36383
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.230496644973755,
+      "learning_rate": 3.760039307201593e-06,
+      "loss": 0.6336,
+      "step": 36384
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.5659472942352295,
+      "learning_rate": 3.7597838868900734e-06,
+      "loss": 0.4893,
+      "step": 36385
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.787609100341797,
+      "learning_rate": 3.75952847002713e-06,
+      "loss": 0.738,
+      "step": 36386
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.1187663078308105,
+      "learning_rate": 3.759273056613474e-06,
+      "loss": 0.6709,
+      "step": 36387
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.7735815048217773,
+      "learning_rate": 3.7590176466498142e-06,
+      "loss": 0.4442,
+      "step": 36388
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.0642454624176025,
+      "learning_rate": 3.7587622401368616e-06,
+      "loss": 0.9177,
+      "step": 36389
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.4611871242523193,
+      "learning_rate": 3.7585068370753264e-06,
+      "loss": 0.7433,
+      "step": 36390
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.482593297958374,
+      "learning_rate": 3.7582514374659172e-06,
+      "loss": 0.5517,
+      "step": 36391
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.2913906574249268,
+      "learning_rate": 3.757996041309347e-06,
+      "loss": 0.7247,
+      "step": 36392
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 3.0044162273406982,
+      "learning_rate": 3.7577406486063245e-06,
+      "loss": 0.58,
+      "step": 36393
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.7650760412216187,
+      "learning_rate": 3.75748525935756e-06,
+      "loss": 0.5062,
+      "step": 36394
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 3.5025651454925537,
+      "learning_rate": 3.7572298735637636e-06,
+      "loss": 0.6231,
+      "step": 36395
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.7104140520095825,
+      "learning_rate": 3.7569744912256446e-06,
+      "loss": 0.4918,
+      "step": 36396
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 4.497755527496338,
+      "learning_rate": 3.7567191123439147e-06,
+      "loss": 0.7582,
+      "step": 36397
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.772649049758911,
+      "learning_rate": 3.756463736919283e-06,
+      "loss": 0.8241,
+      "step": 36398
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.7227011919021606,
+      "learning_rate": 3.7562083649524593e-06,
+      "loss": 0.6158,
+      "step": 36399
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 3.216050624847412,
+      "learning_rate": 3.755952996444154e-06,
+      "loss": 0.8245,
+      "step": 36400
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.4231374263763428,
+      "learning_rate": 3.7556976313950776e-06,
+      "loss": 0.7071,
+      "step": 36401
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.6065969467163086,
+      "learning_rate": 3.755442269805941e-06,
+      "loss": 0.5686,
+      "step": 36402
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.9617706537246704,
+      "learning_rate": 3.7551869116774516e-06,
+      "loss": 0.7414,
+      "step": 36403
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 3.3275229930877686,
+      "learning_rate": 3.754931557010322e-06,
+      "loss": 0.6011,
+      "step": 36404
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.9901072978973389,
+      "learning_rate": 3.754676205805261e-06,
+      "loss": 0.536,
+      "step": 36405
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 22.988645553588867,
+      "learning_rate": 3.754420858062979e-06,
+      "loss": 0.7698,
+      "step": 36406
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.7005670070648193,
+      "learning_rate": 3.754165513784187e-06,
+      "loss": 0.5194,
+      "step": 36407
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.6415499448776245,
+      "learning_rate": 3.7539101729695933e-06,
+      "loss": 0.4566,
+      "step": 36408
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.1464686393737793,
+      "learning_rate": 3.7536548356199086e-06,
+      "loss": 0.6183,
+      "step": 36409
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.167327404022217,
+      "learning_rate": 3.753399501735843e-06,
+      "loss": 0.5078,
+      "step": 36410
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.036417007446289,
+      "learning_rate": 3.7531441713181066e-06,
+      "loss": 0.5013,
+      "step": 36411
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 3.294532299041748,
+      "learning_rate": 3.752888844367409e-06,
+      "loss": 0.498,
+      "step": 36412
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.8029916286468506,
+      "learning_rate": 3.75263352088446e-06,
+      "loss": 0.6233,
+      "step": 36413
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.7866302728652954,
+      "learning_rate": 3.7523782008699706e-06,
+      "loss": 0.3294,
+      "step": 36414
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.1448700428009033,
+      "learning_rate": 3.752122884324649e-06,
+      "loss": 0.7147,
+      "step": 36415
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.8044098615646362,
+      "learning_rate": 3.7518675712492075e-06,
+      "loss": 0.4892,
+      "step": 36416
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.861740231513977,
+      "learning_rate": 3.751612261644354e-06,
+      "loss": 0.7178,
+      "step": 36417
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.5775744915008545,
+      "learning_rate": 3.7513569555107994e-06,
+      "loss": 0.5512,
+      "step": 36418
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.7655184268951416,
+      "learning_rate": 3.7511016528492526e-06,
+      "loss": 0.3718,
+      "step": 36419
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.9320796728134155,
+      "learning_rate": 3.7508463536604257e-06,
+      "loss": 0.5954,
+      "step": 36420
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.7223892211914062,
+      "learning_rate": 3.750591057945024e-06,
+      "loss": 0.6304,
+      "step": 36421
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.9766759872436523,
+      "learning_rate": 3.7503357657037628e-06,
+      "loss": 0.6307,
+      "step": 36422
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 3.87583589553833,
+      "learning_rate": 3.7500804769373493e-06,
+      "loss": 0.7714,
+      "step": 36423
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.5139517784118652,
+      "learning_rate": 3.7498251916464946e-06,
+      "loss": 0.436,
+      "step": 36424
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.3163557052612305,
+      "learning_rate": 3.749569909831907e-06,
+      "loss": 0.7648,
+      "step": 36425
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.0571320056915283,
+      "learning_rate": 3.7493146314942976e-06,
+      "loss": 0.573,
+      "step": 36426
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.4481654167175293,
+      "learning_rate": 3.7490593566343753e-06,
+      "loss": 0.668,
+      "step": 36427
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.976854681968689,
+      "learning_rate": 3.7488040852528507e-06,
+      "loss": 0.5572,
+      "step": 36428
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.982969045639038,
+      "learning_rate": 3.7485488173504325e-06,
+      "loss": 0.3945,
+      "step": 36429
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.5112743377685547,
+      "learning_rate": 3.748293552927832e-06,
+      "loss": 0.2708,
+      "step": 36430
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.789165496826172,
+      "learning_rate": 3.7480382919857583e-06,
+      "loss": 0.8528,
+      "step": 36431
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.058558702468872,
+      "learning_rate": 3.7477830345249207e-06,
+      "loss": 0.4992,
+      "step": 36432
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 3.0159637928009033,
+      "learning_rate": 3.7475277805460297e-06,
+      "loss": 0.3803,
+      "step": 36433
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.897651433944702,
+      "learning_rate": 3.7472725300497943e-06,
+      "loss": 0.6363,
+      "step": 36434
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.7229480743408203,
+      "learning_rate": 3.7470172830369243e-06,
+      "loss": 0.8654,
+      "step": 36435
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.9118635654449463,
+      "learning_rate": 3.746762039508131e-06,
+      "loss": 0.5689,
+      "step": 36436
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.715761661529541,
+      "learning_rate": 3.7465067994641226e-06,
+      "loss": 0.5795,
+      "step": 36437
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.3249804973602295,
+      "learning_rate": 3.746251562905609e-06,
+      "loss": 0.5162,
+      "step": 36438
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.8400859832763672,
+      "learning_rate": 3.7459963298333004e-06,
+      "loss": 0.6603,
+      "step": 36439
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.0410373210906982,
+      "learning_rate": 3.745741100247906e-06,
+      "loss": 0.6364,
+      "step": 36440
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.0849051475524902,
+      "learning_rate": 3.745485874150134e-06,
+      "loss": 0.6973,
+      "step": 36441
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.4400038719177246,
+      "learning_rate": 3.7452306515406976e-06,
+      "loss": 0.5282,
+      "step": 36442
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 3.5881314277648926,
+      "learning_rate": 3.744975432420305e-06,
+      "loss": 0.8758,
+      "step": 36443
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.320590019226074,
+      "learning_rate": 3.7447202167896646e-06,
+      "loss": 0.5396,
+      "step": 36444
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.8278119564056396,
+      "learning_rate": 3.744465004649488e-06,
+      "loss": 0.8975,
+      "step": 36445
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.2891128063201904,
+      "learning_rate": 3.744209796000484e-06,
+      "loss": 0.542,
+      "step": 36446
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.168140411376953,
+      "learning_rate": 3.7439545908433618e-06,
+      "loss": 0.5937,
+      "step": 36447
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.8954135179519653,
+      "learning_rate": 3.7436993891788307e-06,
+      "loss": 0.7466,
+      "step": 36448
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.696641206741333,
+      "learning_rate": 3.7434441910076025e-06,
+      "loss": 0.5916,
+      "step": 36449
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.2288265228271484,
+      "learning_rate": 3.7431889963303843e-06,
+      "loss": 0.4342,
+      "step": 36450
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 3.916620969772339,
+      "learning_rate": 3.7429338051478868e-06,
+      "loss": 0.7973,
+      "step": 36451
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.667949914932251,
+      "learning_rate": 3.7426786174608194e-06,
+      "loss": 0.6489,
+      "step": 36452
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.1312754154205322,
+      "learning_rate": 3.7424234332698907e-06,
+      "loss": 0.6831,
+      "step": 36453
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.336693048477173,
+      "learning_rate": 3.742168252575813e-06,
+      "loss": 0.6069,
+      "step": 36454
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.5230584144592285,
+      "learning_rate": 3.741913075379293e-06,
+      "loss": 0.6771,
+      "step": 36455
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.0775649547576904,
+      "learning_rate": 3.7416579016810426e-06,
+      "loss": 0.6296,
+      "step": 36456
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.3019165992736816,
+      "learning_rate": 3.7414027314817696e-06,
+      "loss": 0.7896,
+      "step": 36457
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.8668880462646484,
+      "learning_rate": 3.7411475647821843e-06,
+      "loss": 0.5954,
+      "step": 36458
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 3.104918956756592,
+      "learning_rate": 3.740892401582996e-06,
+      "loss": 0.4127,
+      "step": 36459
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.6473217010498047,
+      "learning_rate": 3.7406372418849145e-06,
+      "loss": 0.675,
+      "step": 36460
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.109041929244995,
+      "learning_rate": 3.7403820856886475e-06,
+      "loss": 0.5779,
+      "step": 36461
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.4611711502075195,
+      "learning_rate": 3.7401269329949074e-06,
+      "loss": 0.5381,
+      "step": 36462
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 3.1272404193878174,
+      "learning_rate": 3.7398717838044024e-06,
+      "loss": 0.8981,
+      "step": 36463
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.593260407447815,
+      "learning_rate": 3.7396166381178415e-06,
+      "loss": 0.5029,
+      "step": 36464
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.9561171531677246,
+      "learning_rate": 3.7393614959359336e-06,
+      "loss": 0.5317,
+      "step": 36465
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.2367968559265137,
+      "learning_rate": 3.739106357259391e-06,
+      "loss": 0.6079,
+      "step": 36466
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.651926875114441,
+      "learning_rate": 3.7388512220889204e-06,
+      "loss": 0.647,
+      "step": 36467
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.031784772872925,
+      "learning_rate": 3.7385960904252327e-06,
+      "loss": 0.5657,
+      "step": 36468
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 3.010650396347046,
+      "learning_rate": 3.7383409622690363e-06,
+      "loss": 0.7999,
+      "step": 36469
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.579366445541382,
+      "learning_rate": 3.7380858376210416e-06,
+      "loss": 0.492,
+      "step": 36470
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.4682514667510986,
+      "learning_rate": 3.737830716481957e-06,
+      "loss": 0.6885,
+      "step": 36471
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.107241153717041,
+      "learning_rate": 3.737575598852492e-06,
+      "loss": 0.5651,
+      "step": 36472
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.9106667041778564,
+      "learning_rate": 3.7373204847333573e-06,
+      "loss": 0.69,
+      "step": 36473
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.011918067932129,
+      "learning_rate": 3.7370653741252607e-06,
+      "loss": 0.6665,
+      "step": 36474
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 4.184152603149414,
+      "learning_rate": 3.7368102670289127e-06,
+      "loss": 0.5385,
+      "step": 36475
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.77284574508667,
+      "learning_rate": 3.7365551634450214e-06,
+      "loss": 0.6184,
+      "step": 36476
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.3366923332214355,
+      "learning_rate": 3.7363000633742975e-06,
+      "loss": 0.7273,
+      "step": 36477
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.3959765434265137,
+      "learning_rate": 3.7360449668174492e-06,
+      "loss": 0.4223,
+      "step": 36478
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.9370992183685303,
+      "learning_rate": 3.7357898737751862e-06,
+      "loss": 0.4323,
+      "step": 36479
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.451279401779175,
+      "learning_rate": 3.7355347842482192e-06,
+      "loss": 0.6119,
+      "step": 36480
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.8563148975372314,
+      "learning_rate": 3.735279698237255e-06,
+      "loss": 0.774,
+      "step": 36481
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.0391273498535156,
+      "learning_rate": 3.7350246157430028e-06,
+      "loss": 0.5725,
+      "step": 36482
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.3895819187164307,
+      "learning_rate": 3.734769536766175e-06,
+      "loss": 0.7207,
+      "step": 36483
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.4077651500701904,
+      "learning_rate": 3.734514461307479e-06,
+      "loss": 0.5584,
+      "step": 36484
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 3.301300525665283,
+      "learning_rate": 3.734259389367625e-06,
+      "loss": 0.5027,
+      "step": 36485
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.48703932762146,
+      "learning_rate": 3.734004320947321e-06,
+      "loss": 0.6015,
+      "step": 36486
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 3.568202257156372,
+      "learning_rate": 3.7337492560472766e-06,
+      "loss": 0.5414,
+      "step": 36487
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.2933850288391113,
+      "learning_rate": 3.7334941946682012e-06,
+      "loss": 0.7582,
+      "step": 36488
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.964043140411377,
+      "learning_rate": 3.7332391368108047e-06,
+      "loss": 0.6034,
+      "step": 36489
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.166119337081909,
+      "learning_rate": 3.7329840824757947e-06,
+      "loss": 0.471,
+      "step": 36490
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.4812963008880615,
+      "learning_rate": 3.7327290316638822e-06,
+      "loss": 0.648,
+      "step": 36491
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.9538238048553467,
+      "learning_rate": 3.7324739843757746e-06,
+      "loss": 0.8209,
+      "step": 36492
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.4110047817230225,
+      "learning_rate": 3.732218940612182e-06,
+      "loss": 0.922,
+      "step": 36493
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.979017734527588,
+      "learning_rate": 3.7319639003738146e-06,
+      "loss": 0.72,
+      "step": 36494
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.2290682792663574,
+      "learning_rate": 3.7317088636613797e-06,
+      "loss": 0.9281,
+      "step": 36495
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 3.2738842964172363,
+      "learning_rate": 3.7314538304755876e-06,
+      "loss": 0.8385,
+      "step": 36496
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 3.2704100608825684,
+      "learning_rate": 3.7311988008171473e-06,
+      "loss": 0.6845,
+      "step": 36497
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.497186303138733,
+      "learning_rate": 3.730943774686768e-06,
+      "loss": 0.3502,
+      "step": 36498
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.5990145206451416,
+      "learning_rate": 3.7306887520851586e-06,
+      "loss": 0.6941,
+      "step": 36499
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.9514527320861816,
+      "learning_rate": 3.730433733013028e-06,
+      "loss": 0.6443,
+      "step": 36500
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 3.2790966033935547,
+      "learning_rate": 3.730178717471086e-06,
+      "loss": 0.6372,
+      "step": 36501
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.240356206893921,
+      "learning_rate": 3.7299237054600384e-06,
+      "loss": 0.7584,
+      "step": 36502
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 3.2344536781311035,
+      "learning_rate": 3.7296686969805995e-06,
+      "loss": 0.6837,
+      "step": 36503
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.1710407733917236,
+      "learning_rate": 3.729413692033477e-06,
+      "loss": 0.5718,
+      "step": 36504
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 3.8989877700805664,
+      "learning_rate": 3.7291586906193777e-06,
+      "loss": 0.687,
+      "step": 36505
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.4105639457702637,
+      "learning_rate": 3.728903692739013e-06,
+      "loss": 0.692,
+      "step": 36506
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.670769453048706,
+      "learning_rate": 3.7286486983930906e-06,
+      "loss": 0.5896,
+      "step": 36507
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.2388527393341064,
+      "learning_rate": 3.7283937075823206e-06,
+      "loss": 0.7993,
+      "step": 36508
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 3.603363275527954,
+      "learning_rate": 3.7281387203074105e-06,
+      "loss": 0.6699,
+      "step": 36509
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.003244400024414,
+      "learning_rate": 3.72788373656907e-06,
+      "loss": 0.6956,
+      "step": 36510
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.9804892539978027,
+      "learning_rate": 3.727628756368009e-06,
+      "loss": 0.6218,
+      "step": 36511
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 3.200932502746582,
+      "learning_rate": 3.727373779704936e-06,
+      "loss": 0.6929,
+      "step": 36512
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.263559341430664,
+      "learning_rate": 3.7271188065805586e-06,
+      "loss": 0.5516,
+      "step": 36513
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 4.1943817138671875,
+      "learning_rate": 3.7268638369955877e-06,
+      "loss": 0.7106,
+      "step": 36514
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.4879024028778076,
+      "learning_rate": 3.7266088709507308e-06,
+      "loss": 0.5581,
+      "step": 36515
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.0224921703338623,
+      "learning_rate": 3.726353908446698e-06,
+      "loss": 0.5376,
+      "step": 36516
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.8185511827468872,
+      "learning_rate": 3.726098949484198e-06,
+      "loss": 0.6566,
+      "step": 36517
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.453750729560852,
+      "learning_rate": 3.72584399406394e-06,
+      "loss": 0.292,
+      "step": 36518
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.8035948276519775,
+      "learning_rate": 3.7255890421866325e-06,
+      "loss": 0.7166,
+      "step": 36519
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.4727206230163574,
+      "learning_rate": 3.7253340938529842e-06,
+      "loss": 0.5167,
+      "step": 36520
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.4924691915512085,
+      "learning_rate": 3.7250791490637047e-06,
+      "loss": 0.496,
+      "step": 36521
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.1962006092071533,
+      "learning_rate": 3.7248242078195007e-06,
+      "loss": 0.4809,
+      "step": 36522
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.13443922996521,
+      "learning_rate": 3.7245692701210846e-06,
+      "loss": 0.6921,
+      "step": 36523
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 3.393575429916382,
+      "learning_rate": 3.7243143359691637e-06,
+      "loss": 0.78,
+      "step": 36524
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.5743659734725952,
+      "learning_rate": 3.7240594053644457e-06,
+      "loss": 0.5023,
+      "step": 36525
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 3.2890501022338867,
+      "learning_rate": 3.72380447830764e-06,
+      "loss": 0.609,
+      "step": 36526
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.037919044494629,
+      "learning_rate": 3.723549554799457e-06,
+      "loss": 0.3636,
+      "step": 36527
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.152968645095825,
+      "learning_rate": 3.723294634840605e-06,
+      "loss": 0.5397,
+      "step": 36528
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.183478355407715,
+      "learning_rate": 3.7230397184317925e-06,
+      "loss": 0.6134,
+      "step": 36529
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.3403403759002686,
+      "learning_rate": 3.722784805573727e-06,
+      "loss": 0.424,
+      "step": 36530
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 3.554229259490967,
+      "learning_rate": 3.72252989626712e-06,
+      "loss": 0.7347,
+      "step": 36531
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.9292874336242676,
+      "learning_rate": 3.7222749905126777e-06,
+      "loss": 0.3946,
+      "step": 36532
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 3.9484150409698486,
+      "learning_rate": 3.7220200883111103e-06,
+      "loss": 0.7013,
+      "step": 36533
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.770266532897949,
+      "learning_rate": 3.7217651896631263e-06,
+      "loss": 0.7968,
+      "step": 36534
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.7621252536773682,
+      "learning_rate": 3.721510294569435e-06,
+      "loss": 0.5947,
+      "step": 36535
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.889939308166504,
+      "learning_rate": 3.7212554030307434e-06,
+      "loss": 0.6721,
+      "step": 36536
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.597732424736023,
+      "learning_rate": 3.7210005150477624e-06,
+      "loss": 0.5855,
+      "step": 36537
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 3.1432416439056396,
+      "learning_rate": 3.7207456306212e-06,
+      "loss": 0.444,
+      "step": 36538
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.206986665725708,
+      "learning_rate": 3.7204907497517638e-06,
+      "loss": 0.7496,
+      "step": 36539
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.631697416305542,
+      "learning_rate": 3.7202358724401646e-06,
+      "loss": 0.7139,
+      "step": 36540
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.8809332847595215,
+      "learning_rate": 3.7199809986871096e-06,
+      "loss": 0.3506,
+      "step": 36541
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.985762596130371,
+      "learning_rate": 3.719726128493308e-06,
+      "loss": 0.539,
+      "step": 36542
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.3898680210113525,
+      "learning_rate": 3.7194712618594657e-06,
+      "loss": 0.5345,
+      "step": 36543
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.5046396255493164,
+      "learning_rate": 3.719216398786297e-06,
+      "loss": 0.7293,
+      "step": 36544
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.5597366094589233,
+      "learning_rate": 3.718961539274507e-06,
+      "loss": 0.5516,
+      "step": 36545
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.067063093185425,
+      "learning_rate": 3.7187066833248065e-06,
+      "loss": 0.5999,
+      "step": 36546
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.221344232559204,
+      "learning_rate": 3.7184518309379014e-06,
+      "loss": 0.3759,
+      "step": 36547
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.9904251098632812,
+      "learning_rate": 3.718196982114502e-06,
+      "loss": 0.6197,
+      "step": 36548
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.474774122238159,
+      "learning_rate": 3.717942136855317e-06,
+      "loss": 0.5649,
+      "step": 36549
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.997542381286621,
+      "learning_rate": 3.717687295161054e-06,
+      "loss": 0.8692,
+      "step": 36550
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.7533408403396606,
+      "learning_rate": 3.717432457032423e-06,
+      "loss": 0.7088,
+      "step": 36551
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.2777910232543945,
+      "learning_rate": 3.7171776224701313e-06,
+      "loss": 0.5691,
+      "step": 36552
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.616455316543579,
+      "learning_rate": 3.716922791474888e-06,
+      "loss": 0.5872,
+      "step": 36553
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.7471930980682373,
+      "learning_rate": 3.7166679640474023e-06,
+      "loss": 0.6491,
+      "step": 36554
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.372401475906372,
+      "learning_rate": 3.7164131401883817e-06,
+      "loss": 0.4964,
+      "step": 36555
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.1972126960754395,
+      "learning_rate": 3.7161583198985353e-06,
+      "loss": 0.6988,
+      "step": 36556
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.385969877243042,
+      "learning_rate": 3.7159035031785723e-06,
+      "loss": 0.5268,
+      "step": 36557
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.004837989807129,
+      "learning_rate": 3.7156486900292e-06,
+      "loss": 0.4642,
+      "step": 36558
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.8933327198028564,
+      "learning_rate": 3.715393880451128e-06,
+      "loss": 0.5025,
+      "step": 36559
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.0425758361816406,
+      "learning_rate": 3.7151390744450644e-06,
+      "loss": 0.7828,
+      "step": 36560
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.4301931858062744,
+      "learning_rate": 3.7148842720117173e-06,
+      "loss": 0.7287,
+      "step": 36561
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.9782629013061523,
+      "learning_rate": 3.7146294731517964e-06,
+      "loss": 0.8186,
+      "step": 36562
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.663513660430908,
+      "learning_rate": 3.714374677866006e-06,
+      "loss": 0.5918,
+      "step": 36563
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.1778950691223145,
+      "learning_rate": 3.714119886155061e-06,
+      "loss": 0.3983,
+      "step": 36564
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.223715305328369,
+      "learning_rate": 3.7138650980196668e-06,
+      "loss": 0.7595,
+      "step": 36565
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.6487481594085693,
+      "learning_rate": 3.713610313460532e-06,
+      "loss": 0.6126,
+      "step": 36566
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.795433759689331,
+      "learning_rate": 3.7133555324783656e-06,
+      "loss": 0.6338,
+      "step": 36567
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 4.819710731506348,
+      "learning_rate": 3.713100755073874e-06,
+      "loss": 0.665,
+      "step": 36568
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.184492826461792,
+      "learning_rate": 3.7128459812477687e-06,
+      "loss": 0.48,
+      "step": 36569
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.7312066555023193,
+      "learning_rate": 3.712591211000756e-06,
+      "loss": 0.7971,
+      "step": 36570
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.0772626399993896,
+      "learning_rate": 3.7123364443335443e-06,
+      "loss": 0.6938,
+      "step": 36571
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.7718137502670288,
+      "learning_rate": 3.712081681246843e-06,
+      "loss": 0.3142,
+      "step": 36572
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.543123483657837,
+      "learning_rate": 3.71182692174136e-06,
+      "loss": 0.6468,
+      "step": 36573
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.9047825336456299,
+      "learning_rate": 3.711572165817804e-06,
+      "loss": 0.431,
+      "step": 36574
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.6260228157043457,
+      "learning_rate": 3.7113174134768824e-06,
+      "loss": 0.8368,
+      "step": 36575
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.8029204607009888,
+      "learning_rate": 3.711062664719304e-06,
+      "loss": 0.727,
+      "step": 36576
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.9769750833511353,
+      "learning_rate": 3.7108079195457786e-06,
+      "loss": 0.5968,
+      "step": 36577
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.731955051422119,
+      "learning_rate": 3.710553177957013e-06,
+      "loss": 0.5623,
+      "step": 36578
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 4.003232955932617,
+      "learning_rate": 3.7102984399537166e-06,
+      "loss": 0.5437,
+      "step": 36579
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.8565330505371094,
+      "learning_rate": 3.7100437055365967e-06,
+      "loss": 0.5191,
+      "step": 36580
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.1535589694976807,
+      "learning_rate": 3.709788974706362e-06,
+      "loss": 0.371,
+      "step": 36581
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.7666882276535034,
+      "learning_rate": 3.709534247463721e-06,
+      "loss": 0.5229,
+      "step": 36582
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.49019455909729,
+      "learning_rate": 3.7092795238093804e-06,
+      "loss": 0.6966,
+      "step": 36583
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 5.945919036865234,
+      "learning_rate": 3.7090248037440514e-06,
+      "loss": 0.5775,
+      "step": 36584
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 4.281700134277344,
+      "learning_rate": 3.708770087268441e-06,
+      "loss": 0.4931,
+      "step": 36585
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.5165188312530518,
+      "learning_rate": 3.7085153743832564e-06,
+      "loss": 0.3542,
+      "step": 36586
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.1612255573272705,
+      "learning_rate": 3.708260665089207e-06,
+      "loss": 0.8467,
+      "step": 36587
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 4.06171178817749,
+      "learning_rate": 3.708005959387e-06,
+      "loss": 0.7065,
+      "step": 36588
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.5576908588409424,
+      "learning_rate": 3.7077512572773457e-06,
+      "loss": 0.5813,
+      "step": 36589
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.8257880210876465,
+      "learning_rate": 3.7074965587609508e-06,
+      "loss": 0.6926,
+      "step": 36590
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.7386066913604736,
+      "learning_rate": 3.707241863838524e-06,
+      "loss": 0.6944,
+      "step": 36591
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.049010992050171,
+      "learning_rate": 3.706987172510773e-06,
+      "loss": 0.4453,
+      "step": 36592
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.102306604385376,
+      "learning_rate": 3.7067324847784065e-06,
+      "loss": 0.4436,
+      "step": 36593
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.292221784591675,
+      "learning_rate": 3.7064778006421322e-06,
+      "loss": 0.6878,
+      "step": 36594
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.076493740081787,
+      "learning_rate": 3.706223120102659e-06,
+      "loss": 0.8187,
+      "step": 36595
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.7106072902679443,
+      "learning_rate": 3.7059684431606946e-06,
+      "loss": 0.3906,
+      "step": 36596
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.1698200702667236,
+      "learning_rate": 3.7057137698169466e-06,
+      "loss": 0.7018,
+      "step": 36597
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.7538347244262695,
+      "learning_rate": 3.7054591000721242e-06,
+      "loss": 0.549,
+      "step": 36598
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.206246852874756,
+      "learning_rate": 3.7052044339269357e-06,
+      "loss": 0.7961,
+      "step": 36599
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.322610855102539,
+      "learning_rate": 3.7049497713820874e-06,
+      "loss": 0.4925,
+      "step": 36600
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.34132981300354,
+      "learning_rate": 3.7046951124382895e-06,
+      "loss": 0.51,
+      "step": 36601
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.28782057762146,
+      "learning_rate": 3.704440457096249e-06,
+      "loss": 0.5053,
+      "step": 36602
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.7370927333831787,
+      "learning_rate": 3.7041858053566743e-06,
+      "loss": 0.6294,
+      "step": 36603
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.8405470848083496,
+      "learning_rate": 3.7039311572202714e-06,
+      "loss": 0.4769,
+      "step": 36604
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.5590529441833496,
+      "learning_rate": 3.7036765126877528e-06,
+      "loss": 0.7529,
+      "step": 36605
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.4122726917266846,
+      "learning_rate": 3.7034218717598235e-06,
+      "loss": 0.7496,
+      "step": 36606
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.0106332302093506,
+      "learning_rate": 3.7031672344371926e-06,
+      "loss": 0.8555,
+      "step": 36607
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.626099109649658,
+      "learning_rate": 3.702912600720568e-06,
+      "loss": 0.673,
+      "step": 36608
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.4206010103225708,
+      "learning_rate": 3.7026579706106573e-06,
+      "loss": 0.3222,
+      "step": 36609
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.0561089515686035,
+      "learning_rate": 3.7024033441081684e-06,
+      "loss": 0.4625,
+      "step": 36610
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.8208664655685425,
+      "learning_rate": 3.7021487212138107e-06,
+      "loss": 0.3873,
+      "step": 36611
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.0483334064483643,
+      "learning_rate": 3.7018941019282907e-06,
+      "loss": 0.6949,
+      "step": 36612
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.5847342014312744,
+      "learning_rate": 3.7016394862523177e-06,
+      "loss": 0.6475,
+      "step": 36613
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.294520616531372,
+      "learning_rate": 3.7013848741865978e-06,
+      "loss": 0.7674,
+      "step": 36614
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.0566649436950684,
+      "learning_rate": 3.7011302657318403e-06,
+      "loss": 0.6095,
+      "step": 36615
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.263929605484009,
+      "learning_rate": 3.700875660888753e-06,
+      "loss": 0.7791,
+      "step": 36616
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.0351357460021973,
+      "learning_rate": 3.700621059658044e-06,
+      "loss": 0.6841,
+      "step": 36617
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.904222011566162,
+      "learning_rate": 3.700366462040421e-06,
+      "loss": 0.7565,
+      "step": 36618
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.9408174753189087,
+      "learning_rate": 3.700111868036592e-06,
+      "loss": 0.9986,
+      "step": 36619
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.3093748092651367,
+      "learning_rate": 3.699857277647265e-06,
+      "loss": 0.4948,
+      "step": 36620
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.3074541091918945,
+      "learning_rate": 3.699602690873148e-06,
+      "loss": 0.8838,
+      "step": 36621
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.3464090824127197,
+      "learning_rate": 3.6993481077149483e-06,
+      "loss": 0.4857,
+      "step": 36622
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.6982951164245605,
+      "learning_rate": 3.6990935281733755e-06,
+      "loss": 0.6702,
+      "step": 36623
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.9325954914093018,
+      "learning_rate": 3.698838952249133e-06,
+      "loss": 0.4884,
+      "step": 36624
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.8319759368896484,
+      "learning_rate": 3.6985843799429343e-06,
+      "loss": 0.8309,
+      "step": 36625
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.4920430183410645,
+      "learning_rate": 3.6983298112554843e-06,
+      "loss": 0.6554,
+      "step": 36626
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.099180221557617,
+      "learning_rate": 3.6980752461874924e-06,
+      "loss": 0.6421,
+      "step": 36627
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.2746078968048096,
+      "learning_rate": 3.6978206847396652e-06,
+      "loss": 0.7828,
+      "step": 36628
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.29384446144104,
+      "learning_rate": 3.697566126912711e-06,
+      "loss": 0.536,
+      "step": 36629
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.753643035888672,
+      "learning_rate": 3.697311572707337e-06,
+      "loss": 0.3413,
+      "step": 36630
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.6697819232940674,
+      "learning_rate": 3.6970570221242524e-06,
+      "loss": 0.516,
+      "step": 36631
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.746108293533325,
+      "learning_rate": 3.696802475164164e-06,
+      "loss": 0.5559,
+      "step": 36632
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.905957818031311,
+      "learning_rate": 3.696547931827779e-06,
+      "loss": 0.5986,
+      "step": 36633
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.966685175895691,
+      "learning_rate": 3.696293392115806e-06,
+      "loss": 0.6475,
+      "step": 36634
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.669961452484131,
+      "learning_rate": 3.6960388560289535e-06,
+      "loss": 0.6339,
+      "step": 36635
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.1221232414245605,
+      "learning_rate": 3.6957843235679282e-06,
+      "loss": 0.7192,
+      "step": 36636
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.368138074874878,
+      "learning_rate": 3.6955297947334377e-06,
+      "loss": 0.6876,
+      "step": 36637
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.0313870906829834,
+      "learning_rate": 3.6952752695261895e-06,
+      "loss": 0.6134,
+      "step": 36638
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.2471580505371094,
+      "learning_rate": 3.6950207479468924e-06,
+      "loss": 0.6747,
+      "step": 36639
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 4.141622543334961,
+      "learning_rate": 3.6947662299962543e-06,
+      "loss": 0.5529,
+      "step": 36640
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.3760735988616943,
+      "learning_rate": 3.694511715674982e-06,
+      "loss": 0.3681,
+      "step": 36641
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.451416254043579,
+      "learning_rate": 3.6942572049837844e-06,
+      "loss": 0.7045,
+      "step": 36642
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.4233760833740234,
+      "learning_rate": 3.694002697923368e-06,
+      "loss": 0.3324,
+      "step": 36643
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.9521167278289795,
+      "learning_rate": 3.6937481944944394e-06,
+      "loss": 0.6128,
+      "step": 36644
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.515641927719116,
+      "learning_rate": 3.693493694697709e-06,
+      "loss": 0.7368,
+      "step": 36645
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.201923131942749,
+      "learning_rate": 3.6932391985338833e-06,
+      "loss": 0.7527,
+      "step": 36646
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 5.090865612030029,
+      "learning_rate": 3.6929847060036696e-06,
+      "loss": 0.5939,
+      "step": 36647
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.2268097400665283,
+      "learning_rate": 3.692730217107776e-06,
+      "loss": 0.6288,
+      "step": 36648
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.5518381595611572,
+      "learning_rate": 3.69247573184691e-06,
+      "loss": 0.6333,
+      "step": 36649
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.849428653717041,
+      "learning_rate": 3.692221250221778e-06,
+      "loss": 0.7492,
+      "step": 36650
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.9498186111450195,
+      "learning_rate": 3.69196677223309e-06,
+      "loss": 0.3405,
+      "step": 36651
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.217240333557129,
+      "learning_rate": 3.6917122978815523e-06,
+      "loss": 0.3536,
+      "step": 36652
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.154559850692749,
+      "learning_rate": 3.6914578271678723e-06,
+      "loss": 0.6665,
+      "step": 36653
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.826503038406372,
+      "learning_rate": 3.691203360092759e-06,
+      "loss": 0.4216,
+      "step": 36654
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.9004825353622437,
+      "learning_rate": 3.690948896656918e-06,
+      "loss": 0.5096,
+      "step": 36655
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.944049596786499,
+      "learning_rate": 3.6906944368610575e-06,
+      "loss": 0.4804,
+      "step": 36656
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.953849196434021,
+      "learning_rate": 3.6904399807058855e-06,
+      "loss": 0.6469,
+      "step": 36657
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.9919345378875732,
+      "learning_rate": 3.6901855281921095e-06,
+      "loss": 0.5842,
+      "step": 36658
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.2568986415863037,
+      "learning_rate": 3.6899310793204374e-06,
+      "loss": 0.3964,
+      "step": 36659
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.0952510833740234,
+      "learning_rate": 3.689676634091576e-06,
+      "loss": 0.6752,
+      "step": 36660
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.6817359924316406,
+      "learning_rate": 3.6894221925062324e-06,
+      "loss": 0.5776,
+      "step": 36661
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.2824597358703613,
+      "learning_rate": 3.689167754565115e-06,
+      "loss": 0.5581,
+      "step": 36662
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 4.4541754722595215,
+      "learning_rate": 3.688913320268931e-06,
+      "loss": 0.8381,
+      "step": 36663
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.7791683673858643,
+      "learning_rate": 3.6886588896183876e-06,
+      "loss": 0.8228,
+      "step": 36664
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.557356119155884,
+      "learning_rate": 3.6884044626141912e-06,
+      "loss": 0.4741,
+      "step": 36665
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.400235414505005,
+      "learning_rate": 3.6881500392570523e-06,
+      "loss": 0.4378,
+      "step": 36666
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.047661066055298,
+      "learning_rate": 3.687895619547677e-06,
+      "loss": 0.6705,
+      "step": 36667
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.9875712394714355,
+      "learning_rate": 3.687641203486772e-06,
+      "loss": 0.7025,
+      "step": 36668
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.9470534324645996,
+      "learning_rate": 3.6873867910750455e-06,
+      "loss": 0.8073,
+      "step": 36669
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.312340259552002,
+      "learning_rate": 3.687132382313204e-06,
+      "loss": 0.632,
+      "step": 36670
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.9828553199768066,
+      "learning_rate": 3.686877977201957e-06,
+      "loss": 0.7323,
+      "step": 36671
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.9906394481658936,
+      "learning_rate": 3.6866235757420094e-06,
+      "loss": 0.6302,
+      "step": 36672
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.025949001312256,
+      "learning_rate": 3.6863691779340693e-06,
+      "loss": 0.6648,
+      "step": 36673
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.6686391830444336,
+      "learning_rate": 3.686114783778845e-06,
+      "loss": 0.9044,
+      "step": 36674
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.4485461711883545,
+      "learning_rate": 3.6858603932770432e-06,
+      "loss": 0.4398,
+      "step": 36675
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.5782856941223145,
+      "learning_rate": 3.685606006429372e-06,
+      "loss": 0.5462,
+      "step": 36676
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.262378454208374,
+      "learning_rate": 3.685351623236537e-06,
+      "loss": 0.5782,
+      "step": 36677
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.0769667625427246,
+      "learning_rate": 3.6850972436992472e-06,
+      "loss": 0.8102,
+      "step": 36678
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.3948163986206055,
+      "learning_rate": 3.6848428678182093e-06,
+      "loss": 0.7916,
+      "step": 36679
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.8915579319000244,
+      "learning_rate": 3.6845884955941306e-06,
+      "loss": 0.5583,
+      "step": 36680
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.562573194503784,
+      "learning_rate": 3.684334127027719e-06,
+      "loss": 0.7748,
+      "step": 36681
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.7894526720046997,
+      "learning_rate": 3.684079762119681e-06,
+      "loss": 0.5008,
+      "step": 36682
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.094937562942505,
+      "learning_rate": 3.683825400870724e-06,
+      "loss": 0.5241,
+      "step": 36683
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.4943130016326904,
+      "learning_rate": 3.6835710432815557e-06,
+      "loss": 0.7842,
+      "step": 36684
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.0532889366149902,
+      "learning_rate": 3.6833166893528814e-06,
+      "loss": 0.7722,
+      "step": 36685
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.7372630834579468,
+      "learning_rate": 3.6830623390854116e-06,
+      "loss": 0.4508,
+      "step": 36686
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.579133987426758,
+      "learning_rate": 3.6828079924798525e-06,
+      "loss": 0.7382,
+      "step": 36687
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.7343857288360596,
+      "learning_rate": 3.682553649536911e-06,
+      "loss": 0.6343,
+      "step": 36688
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.9113082885742188,
+      "learning_rate": 3.6822993102572944e-06,
+      "loss": 0.6529,
+      "step": 36689
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.4864978790283203,
+      "learning_rate": 3.6820449746417096e-06,
+      "loss": 0.6943,
+      "step": 36690
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.126675605773926,
+      "learning_rate": 3.6817906426908636e-06,
+      "loss": 0.7932,
+      "step": 36691
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.292611837387085,
+      "learning_rate": 3.681536314405465e-06,
+      "loss": 0.6367,
+      "step": 36692
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.172168254852295,
+      "learning_rate": 3.681281989786219e-06,
+      "loss": 0.6868,
+      "step": 36693
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.327993154525757,
+      "learning_rate": 3.681027668833834e-06,
+      "loss": 0.6771,
+      "step": 36694
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.349776029586792,
+      "learning_rate": 3.6807733515490175e-06,
+      "loss": 0.5374,
+      "step": 36695
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.658226490020752,
+      "learning_rate": 3.6805190379324758e-06,
+      "loss": 0.5651,
+      "step": 36696
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.209404945373535,
+      "learning_rate": 3.6802647279849157e-06,
+      "loss": 0.4274,
+      "step": 36697
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.7686611413955688,
+      "learning_rate": 3.680010421707046e-06,
+      "loss": 0.6145,
+      "step": 36698
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.316624402999878,
+      "learning_rate": 3.679756119099572e-06,
+      "loss": 0.5871,
+      "step": 36699
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 4.299418926239014,
+      "learning_rate": 3.6795018201632016e-06,
+      "loss": 0.5011,
+      "step": 36700
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.383086919784546,
+      "learning_rate": 3.6792475248986425e-06,
+      "loss": 0.8248,
+      "step": 36701
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.5571264028549194,
+      "learning_rate": 3.6789932333066015e-06,
+      "loss": 0.4079,
+      "step": 36702
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.0383365154266357,
+      "learning_rate": 3.6787389453877846e-06,
+      "loss": 0.7122,
+      "step": 36703
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.2885496616363525,
+      "learning_rate": 3.6784846611429006e-06,
+      "loss": 0.5004,
+      "step": 36704
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.846563696861267,
+      "learning_rate": 3.6782303805726543e-06,
+      "loss": 0.6563,
+      "step": 36705
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.3268558979034424,
+      "learning_rate": 3.677976103677755e-06,
+      "loss": 0.5655,
+      "step": 36706
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.249980926513672,
+      "learning_rate": 3.6777218304589095e-06,
+      "loss": 0.6904,
+      "step": 36707
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.805800199508667,
+      "learning_rate": 3.6774675609168242e-06,
+      "loss": 0.5934,
+      "step": 36708
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.4831154346466064,
+      "learning_rate": 3.6772132950522062e-06,
+      "loss": 0.8415,
+      "step": 36709
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.9487391710281372,
+      "learning_rate": 3.6769590328657623e-06,
+      "loss": 0.7261,
+      "step": 36710
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.3259758949279785,
+      "learning_rate": 3.6767047743581984e-06,
+      "loss": 0.5099,
+      "step": 36711
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.021803617477417,
+      "learning_rate": 3.6764505195302236e-06,
+      "loss": 0.5207,
+      "step": 36712
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.912840723991394,
+      "learning_rate": 3.6761962683825447e-06,
+      "loss": 0.6428,
+      "step": 36713
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.9397573471069336,
+      "learning_rate": 3.675942020915868e-06,
+      "loss": 0.5289,
+      "step": 36714
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.7497894763946533,
+      "learning_rate": 3.6756877771309007e-06,
+      "loss": 0.6481,
+      "step": 36715
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.9312493801116943,
+      "learning_rate": 3.675433537028349e-06,
+      "loss": 0.7578,
+      "step": 36716
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.5157477855682373,
+      "learning_rate": 3.675179300608921e-06,
+      "loss": 0.5856,
+      "step": 36717
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.155763864517212,
+      "learning_rate": 3.674925067873323e-06,
+      "loss": 0.4026,
+      "step": 36718
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.3944478034973145,
+      "learning_rate": 3.674670838822262e-06,
+      "loss": 0.432,
+      "step": 36719
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.572011947631836,
+      "learning_rate": 3.674416613456445e-06,
+      "loss": 0.5962,
+      "step": 36720
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.7278637886047363,
+      "learning_rate": 3.6741623917765784e-06,
+      "loss": 0.4581,
+      "step": 36721
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.337064027786255,
+      "learning_rate": 3.67390817378337e-06,
+      "loss": 0.8329,
+      "step": 36722
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.8278244733810425,
+      "learning_rate": 3.6736539594775254e-06,
+      "loss": 0.9004,
+      "step": 36723
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.6328938007354736,
+      "learning_rate": 3.6733997488597526e-06,
+      "loss": 0.811,
+      "step": 36724
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.901402235031128,
+      "learning_rate": 3.6731455419307587e-06,
+      "loss": 0.6245,
+      "step": 36725
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.120771646499634,
+      "learning_rate": 3.672891338691248e-06,
+      "loss": 0.7665,
+      "step": 36726
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.0050861835479736,
+      "learning_rate": 3.672637139141931e-06,
+      "loss": 0.855,
+      "step": 36727
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.369114398956299,
+      "learning_rate": 3.6723829432835124e-06,
+      "loss": 0.6211,
+      "step": 36728
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.618232488632202,
+      "learning_rate": 3.6721287511167e-06,
+      "loss": 0.4994,
+      "step": 36729
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.9637748003005981,
+      "learning_rate": 3.6718745626422003e-06,
+      "loss": 0.5868,
+      "step": 36730
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.336852788925171,
+      "learning_rate": 3.6716203778607196e-06,
+      "loss": 0.6864,
+      "step": 36731
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.9477860927581787,
+      "learning_rate": 3.6713661967729653e-06,
+      "loss": 0.6369,
+      "step": 36732
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.5966291427612305,
+      "learning_rate": 3.6711120193796435e-06,
+      "loss": 0.6519,
+      "step": 36733
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.042602300643921,
+      "learning_rate": 3.6708578456814615e-06,
+      "loss": 0.7852,
+      "step": 36734
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.570660352706909,
+      "learning_rate": 3.6706036756791264e-06,
+      "loss": 0.5498,
+      "step": 36735
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.4696857929229736,
+      "learning_rate": 3.670349509373344e-06,
+      "loss": 0.5857,
+      "step": 36736
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.082540988922119,
+      "learning_rate": 3.6700953467648213e-06,
+      "loss": 0.556,
+      "step": 36737
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.1517117023468018,
+      "learning_rate": 3.6698411878542658e-06,
+      "loss": 0.6535,
+      "step": 36738
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.1325831413269043,
+      "learning_rate": 3.6695870326423834e-06,
+      "loss": 0.5183,
+      "step": 36739
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.282773494720459,
+      "learning_rate": 3.669332881129881e-06,
+      "loss": 0.2764,
+      "step": 36740
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.5082286596298218,
+      "learning_rate": 3.6690787333174654e-06,
+      "loss": 0.3399,
+      "step": 36741
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.8821998834609985,
+      "learning_rate": 3.6688245892058435e-06,
+      "loss": 0.5262,
+      "step": 36742
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.941794991493225,
+      "learning_rate": 3.6685704487957212e-06,
+      "loss": 0.8787,
+      "step": 36743
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.8513283729553223,
+      "learning_rate": 3.6683163120878064e-06,
+      "loss": 0.5315,
+      "step": 36744
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.173633098602295,
+      "learning_rate": 3.668062179082805e-06,
+      "loss": 0.4159,
+      "step": 36745
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.239030599594116,
+      "learning_rate": 3.6678080497814206e-06,
+      "loss": 0.5953,
+      "step": 36746
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 4.248008728027344,
+      "learning_rate": 3.667553924184366e-06,
+      "loss": 0.4906,
+      "step": 36747
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.6504204273223877,
+      "learning_rate": 3.6672998022923445e-06,
+      "loss": 0.6807,
+      "step": 36748
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.9513957500457764,
+      "learning_rate": 3.6670456841060633e-06,
+      "loss": 0.7347,
+      "step": 36749
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.6058855056762695,
+      "learning_rate": 3.6667915696262292e-06,
+      "loss": 0.531,
+      "step": 36750
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.8357874155044556,
+      "learning_rate": 3.6665374588535476e-06,
+      "loss": 0.3552,
+      "step": 36751
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.5847604274749756,
+      "learning_rate": 3.666283351788725e-06,
+      "loss": 0.8338,
+      "step": 36752
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.411175489425659,
+      "learning_rate": 3.66602924843247e-06,
+      "loss": 0.6772,
+      "step": 36753
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.708224892616272,
+      "learning_rate": 3.6657751487854877e-06,
+      "loss": 0.6112,
+      "step": 36754
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.4204249382019043,
+      "learning_rate": 3.6655210528484848e-06,
+      "loss": 0.5732,
+      "step": 36755
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.3899989128112793,
+      "learning_rate": 3.665266960622168e-06,
+      "loss": 0.698,
+      "step": 36756
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.3979475498199463,
+      "learning_rate": 3.665012872107244e-06,
+      "loss": 0.6115,
+      "step": 36757
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 4.365051746368408,
+      "learning_rate": 3.6647587873044187e-06,
+      "loss": 0.7242,
+      "step": 36758
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.0843493938446045,
+      "learning_rate": 3.664504706214399e-06,
+      "loss": 0.6764,
+      "step": 36759
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.090512752532959,
+      "learning_rate": 3.6642506288378916e-06,
+      "loss": 0.7563,
+      "step": 36760
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.3084723949432373,
+      "learning_rate": 3.663996555175603e-06,
+      "loss": 1.0037,
+      "step": 36761
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.8296315670013428,
+      "learning_rate": 3.6637424852282377e-06,
+      "loss": 0.5608,
+      "step": 36762
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 4.686856746673584,
+      "learning_rate": 3.6634884189965058e-06,
+      "loss": 0.5833,
+      "step": 36763
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.9176405668258667,
+      "learning_rate": 3.663234356481112e-06,
+      "loss": 0.5911,
+      "step": 36764
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.701554775238037,
+      "learning_rate": 3.6629802976827622e-06,
+      "loss": 0.6123,
+      "step": 36765
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.5713917016983032,
+      "learning_rate": 3.6627262426021627e-06,
+      "loss": 0.5756,
+      "step": 36766
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.375948429107666,
+      "learning_rate": 3.662472191240021e-06,
+      "loss": 0.8225,
+      "step": 36767
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.779057741165161,
+      "learning_rate": 3.662218143597044e-06,
+      "loss": 0.8028,
+      "step": 36768
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.2322475910186768,
+      "learning_rate": 3.661964099673936e-06,
+      "loss": 0.4759,
+      "step": 36769
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.9167425632476807,
+      "learning_rate": 3.6617100594714054e-06,
+      "loss": 0.6153,
+      "step": 36770
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.555856227874756,
+      "learning_rate": 3.6614560229901573e-06,
+      "loss": 0.6709,
+      "step": 36771
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.741528272628784,
+      "learning_rate": 3.6612019902308995e-06,
+      "loss": 0.9077,
+      "step": 36772
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.8612794876098633,
+      "learning_rate": 3.6609479611943354e-06,
+      "loss": 0.5771,
+      "step": 36773
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.0876965522766113,
+      "learning_rate": 3.6606939358811757e-06,
+      "loss": 0.5767,
+      "step": 36774
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.374009132385254,
+      "learning_rate": 3.6604399142921233e-06,
+      "loss": 0.7016,
+      "step": 36775
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.8280245065689087,
+      "learning_rate": 3.6601858964278857e-06,
+      "loss": 0.5414,
+      "step": 36776
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.417506694793701,
+      "learning_rate": 3.6599318822891696e-06,
+      "loss": 0.561,
+      "step": 36777
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.3107903003692627,
+      "learning_rate": 3.6596778718766814e-06,
+      "loss": 0.5296,
+      "step": 36778
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.9813960790634155,
+      "learning_rate": 3.6594238651911263e-06,
+      "loss": 0.4788,
+      "step": 36779
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.3779683113098145,
+      "learning_rate": 3.6591698622332104e-06,
+      "loss": 0.4937,
+      "step": 36780
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.3311662673950195,
+      "learning_rate": 3.6589158630036425e-06,
+      "loss": 0.6412,
+      "step": 36781
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 4.3242902755737305,
+      "learning_rate": 3.658661867503126e-06,
+      "loss": 0.7182,
+      "step": 36782
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.440993309020996,
+      "learning_rate": 3.6584078757323695e-06,
+      "loss": 0.6833,
+      "step": 36783
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.820842742919922,
+      "learning_rate": 3.6581538876920773e-06,
+      "loss": 0.5242,
+      "step": 36784
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.103569269180298,
+      "learning_rate": 3.6578999033829566e-06,
+      "loss": 0.56,
+      "step": 36785
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.066519260406494,
+      "learning_rate": 3.6576459228057137e-06,
+      "loss": 0.6917,
+      "step": 36786
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 4.402455806732178,
+      "learning_rate": 3.657391945961052e-06,
+      "loss": 0.7285,
+      "step": 36787
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.9038480520248413,
+      "learning_rate": 3.6571379728496827e-06,
+      "loss": 0.6169,
+      "step": 36788
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.7265429496765137,
+      "learning_rate": 3.6568840034723096e-06,
+      "loss": 0.6055,
+      "step": 36789
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.0692501068115234,
+      "learning_rate": 3.6566300378296394e-06,
+      "loss": 0.5497,
+      "step": 36790
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 4.580939292907715,
+      "learning_rate": 3.6563760759223778e-06,
+      "loss": 0.4962,
+      "step": 36791
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.8367366790771484,
+      "learning_rate": 3.6561221177512307e-06,
+      "loss": 0.5454,
+      "step": 36792
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.599982261657715,
+      "learning_rate": 3.6558681633169045e-06,
+      "loss": 0.7069,
+      "step": 36793
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.7101680040359497,
+      "learning_rate": 3.6556142126201057e-06,
+      "loss": 0.5848,
+      "step": 36794
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.9951493740081787,
+      "learning_rate": 3.6553602656615394e-06,
+      "loss": 0.53,
+      "step": 36795
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.8518450260162354,
+      "learning_rate": 3.655106322441914e-06,
+      "loss": 0.5275,
+      "step": 36796
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.2850964069366455,
+      "learning_rate": 3.6548523829619328e-06,
+      "loss": 0.7149,
+      "step": 36797
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.5260157585144043,
+      "learning_rate": 3.6545984472223035e-06,
+      "loss": 0.4487,
+      "step": 36798
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.3478035926818848,
+      "learning_rate": 3.6543445152237324e-06,
+      "loss": 0.6406,
+      "step": 36799
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.7507758140563965,
+      "learning_rate": 3.654090586966924e-06,
+      "loss": 0.6894,
+      "step": 36800
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 4.080163478851318,
+      "learning_rate": 3.653836662452587e-06,
+      "loss": 0.6127,
+      "step": 36801
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.6985459327697754,
+      "learning_rate": 3.6535827416814245e-06,
+      "loss": 0.3649,
+      "step": 36802
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.790696620941162,
+      "learning_rate": 3.6533288246541447e-06,
+      "loss": 0.5955,
+      "step": 36803
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.9014132022857666,
+      "learning_rate": 3.653074911371453e-06,
+      "loss": 0.5162,
+      "step": 36804
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 4.449509143829346,
+      "learning_rate": 3.6528210018340554e-06,
+      "loss": 0.7178,
+      "step": 36805
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.980154275894165,
+      "learning_rate": 3.652567096042658e-06,
+      "loss": 0.9268,
+      "step": 36806
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.251441717147827,
+      "learning_rate": 3.6523131939979645e-06,
+      "loss": 0.5352,
+      "step": 36807
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.9087885618209839,
+      "learning_rate": 3.652059295700685e-06,
+      "loss": 0.4432,
+      "step": 36808
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.573816180229187,
+      "learning_rate": 3.651805401151524e-06,
+      "loss": 0.6002,
+      "step": 36809
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.8585076332092285,
+      "learning_rate": 3.6515515103511866e-06,
+      "loss": 0.3478,
+      "step": 36810
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 4.211245536804199,
+      "learning_rate": 3.6512976233003804e-06,
+      "loss": 0.6154,
+      "step": 36811
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 4.604774475097656,
+      "learning_rate": 3.6510437399998095e-06,
+      "loss": 0.5503,
+      "step": 36812
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.8422327041625977,
+      "learning_rate": 3.6507898604501806e-06,
+      "loss": 0.6443,
+      "step": 36813
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.549098253250122,
+      "learning_rate": 3.6505359846522002e-06,
+      "loss": 0.7045,
+      "step": 36814
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 4.369956016540527,
+      "learning_rate": 3.6502821126065733e-06,
+      "loss": 0.4848,
+      "step": 36815
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.939669370651245,
+      "learning_rate": 3.6500282443140063e-06,
+      "loss": 0.5157,
+      "step": 36816
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.6889065504074097,
+      "learning_rate": 3.649774379775205e-06,
+      "loss": 0.6646,
+      "step": 36817
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.365582227706909,
+      "learning_rate": 3.6495205189908757e-06,
+      "loss": 0.7099,
+      "step": 36818
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.350867509841919,
+      "learning_rate": 3.6492666619617233e-06,
+      "loss": 0.5628,
+      "step": 36819
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.2597591876983643,
+      "learning_rate": 3.649012808688455e-06,
+      "loss": 0.4202,
+      "step": 36820
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.5615360736846924,
+      "learning_rate": 3.6487589591717753e-06,
+      "loss": 0.8263,
+      "step": 36821
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.4540624618530273,
+      "learning_rate": 3.6485051134123904e-06,
+      "loss": 0.5842,
+      "step": 36822
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.0854456424713135,
+      "learning_rate": 3.6482512714110065e-06,
+      "loss": 0.5022,
+      "step": 36823
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.3997087478637695,
+      "learning_rate": 3.64799743316833e-06,
+      "loss": 0.5289,
+      "step": 36824
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.0259850025177,
+      "learning_rate": 3.647743598685066e-06,
+      "loss": 0.7613,
+      "step": 36825
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.4810683727264404,
+      "learning_rate": 3.6474897679619216e-06,
+      "loss": 0.6077,
+      "step": 36826
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.6495378017425537,
+      "learning_rate": 3.6472359409995996e-06,
+      "loss": 0.3396,
+      "step": 36827
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.7668838500976562,
+      "learning_rate": 3.6469821177988086e-06,
+      "loss": 0.5313,
+      "step": 36828
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.0093066692352295,
+      "learning_rate": 3.646728298360253e-06,
+      "loss": 0.7567,
+      "step": 36829
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.1067354679107666,
+      "learning_rate": 3.64647448268464e-06,
+      "loss": 0.5615,
+      "step": 36830
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.0694162845611572,
+      "learning_rate": 3.6462206707726738e-06,
+      "loss": 0.5154,
+      "step": 36831
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.116689443588257,
+      "learning_rate": 3.645966862625061e-06,
+      "loss": 0.4873,
+      "step": 36832
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.5250065326690674,
+      "learning_rate": 3.645713058242507e-06,
+      "loss": 0.2475,
+      "step": 36833
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.3224756717681885,
+      "learning_rate": 3.645459257625718e-06,
+      "loss": 0.8372,
+      "step": 36834
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 4.335112571716309,
+      "learning_rate": 3.6452054607753983e-06,
+      "loss": 0.6905,
+      "step": 36835
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.2799956798553467,
+      "learning_rate": 3.644951667692256e-06,
+      "loss": 0.9424,
+      "step": 36836
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.240292549133301,
+      "learning_rate": 3.644697878376995e-06,
+      "loss": 0.5217,
+      "step": 36837
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.3070063591003418,
+      "learning_rate": 3.6444440928303215e-06,
+      "loss": 0.6389,
+      "step": 36838
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.2314441204071045,
+      "learning_rate": 3.6441903110529407e-06,
+      "loss": 0.6737,
+      "step": 36839
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.175044536590576,
+      "learning_rate": 3.64393653304556e-06,
+      "loss": 0.5668,
+      "step": 36840
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.460632085800171,
+      "learning_rate": 3.643682758808883e-06,
+      "loss": 0.5701,
+      "step": 36841
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.3760626316070557,
+      "learning_rate": 3.6434289883436157e-06,
+      "loss": 0.7738,
+      "step": 36842
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.58107328414917,
+      "learning_rate": 3.6431752216504657e-06,
+      "loss": 0.7507,
+      "step": 36843
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 4.402501583099365,
+      "learning_rate": 3.642921458730136e-06,
+      "loss": 0.5298,
+      "step": 36844
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.3668971061706543,
+      "learning_rate": 3.642667699583333e-06,
+      "loss": 0.8043,
+      "step": 36845
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.3479363918304443,
+      "learning_rate": 3.6424139442107636e-06,
+      "loss": 0.7751,
+      "step": 36846
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.2906241416931152,
+      "learning_rate": 3.642160192613132e-06,
+      "loss": 0.5832,
+      "step": 36847
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.5213871002197266,
+      "learning_rate": 3.6419064447911422e-06,
+      "loss": 0.4617,
+      "step": 36848
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 6.62343692779541,
+      "learning_rate": 3.641652700745504e-06,
+      "loss": 0.609,
+      "step": 36849
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.3198702335357666,
+      "learning_rate": 3.6413989604769207e-06,
+      "loss": 0.5502,
+      "step": 36850
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.2013015747070312,
+      "learning_rate": 3.641145223986098e-06,
+      "loss": 0.6512,
+      "step": 36851
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.3137781620025635,
+      "learning_rate": 3.6408914912737415e-06,
+      "loss": 0.7169,
+      "step": 36852
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.1500072479248047,
+      "learning_rate": 3.6406377623405563e-06,
+      "loss": 0.6332,
+      "step": 36853
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.1607680320739746,
+      "learning_rate": 3.640384037187249e-06,
+      "loss": 0.661,
+      "step": 36854
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.859919786453247,
+      "learning_rate": 3.6401303158145235e-06,
+      "loss": 0.574,
+      "step": 36855
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.3926806449890137,
+      "learning_rate": 3.6398765982230866e-06,
+      "loss": 0.4569,
+      "step": 36856
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.7919775247573853,
+      "learning_rate": 3.6396228844136433e-06,
+      "loss": 0.6021,
+      "step": 36857
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 5.346597194671631,
+      "learning_rate": 3.6393691743868993e-06,
+      "loss": 0.828,
+      "step": 36858
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.575770616531372,
+      "learning_rate": 3.639115468143559e-06,
+      "loss": 0.9257,
+      "step": 36859
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.6590230464935303,
+      "learning_rate": 3.63886176568433e-06,
+      "loss": 0.8479,
+      "step": 36860
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.7490406036376953,
+      "learning_rate": 3.638608067009916e-06,
+      "loss": 0.5871,
+      "step": 36861
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.4643471240997314,
+      "learning_rate": 3.6383543721210235e-06,
+      "loss": 0.5338,
+      "step": 36862
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 4.170551776885986,
+      "learning_rate": 3.6381006810183567e-06,
+      "loss": 0.5295,
+      "step": 36863
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.219858407974243,
+      "learning_rate": 3.6378469937026217e-06,
+      "loss": 0.6498,
+      "step": 36864
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.1606810092926025,
+      "learning_rate": 3.637593310174524e-06,
+      "loss": 0.7088,
+      "step": 36865
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.8401631116867065,
+      "learning_rate": 3.6373396304347696e-06,
+      "loss": 0.4896,
+      "step": 36866
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.9713166952133179,
+      "learning_rate": 3.637085954484063e-06,
+      "loss": 0.6933,
+      "step": 36867
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.0664355754852295,
+      "learning_rate": 3.636832282323107e-06,
+      "loss": 0.5941,
+      "step": 36868
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.6774356365203857,
+      "learning_rate": 3.6365786139526125e-06,
+      "loss": 0.6048,
+      "step": 36869
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.2704386711120605,
+      "learning_rate": 3.6363249493732823e-06,
+      "loss": 0.7428,
+      "step": 36870
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.7656350135803223,
+      "learning_rate": 3.6360712885858214e-06,
+      "loss": 0.3526,
+      "step": 36871
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.566291093826294,
+      "learning_rate": 3.6358176315909348e-06,
+      "loss": 0.6221,
+      "step": 36872
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.214388608932495,
+      "learning_rate": 3.635563978389329e-06,
+      "loss": 0.5074,
+      "step": 36873
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.1707887649536133,
+      "learning_rate": 3.635310328981708e-06,
+      "loss": 0.5281,
+      "step": 36874
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.7321696281433105,
+      "learning_rate": 3.6350566833687785e-06,
+      "loss": 0.7,
+      "step": 36875
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.063812255859375,
+      "learning_rate": 3.634803041551244e-06,
+      "loss": 0.5259,
+      "step": 36876
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 14.142071723937988,
+      "learning_rate": 3.6345494035298118e-06,
+      "loss": 0.648,
+      "step": 36877
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.5902425050735474,
+      "learning_rate": 3.6342957693051863e-06,
+      "loss": 0.4662,
+      "step": 36878
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.0056424140930176,
+      "learning_rate": 3.6340421388780717e-06,
+      "loss": 0.7823,
+      "step": 36879
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.633347988128662,
+      "learning_rate": 3.6337885122491743e-06,
+      "loss": 0.351,
+      "step": 36880
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.0201637744903564,
+      "learning_rate": 3.6335348894192003e-06,
+      "loss": 0.5851,
+      "step": 36881
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.5608199834823608,
+      "learning_rate": 3.633281270388853e-06,
+      "loss": 0.4338,
+      "step": 36882
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 8.824407577514648,
+      "learning_rate": 3.633027655158839e-06,
+      "loss": 0.6319,
+      "step": 36883
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.099158525466919,
+      "learning_rate": 3.632774043729863e-06,
+      "loss": 0.6056,
+      "step": 36884
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.416950225830078,
+      "learning_rate": 3.632520436102629e-06,
+      "loss": 0.5862,
+      "step": 36885
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.0355913639068604,
+      "learning_rate": 3.632266832277844e-06,
+      "loss": 0.6924,
+      "step": 36886
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.677128314971924,
+      "learning_rate": 3.632013232256213e-06,
+      "loss": 0.6222,
+      "step": 36887
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.1493163108825684,
+      "learning_rate": 3.63175963603844e-06,
+      "loss": 0.2919,
+      "step": 36888
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 5.3464436531066895,
+      "learning_rate": 3.6315060436252313e-06,
+      "loss": 0.6996,
+      "step": 36889
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.539165735244751,
+      "learning_rate": 3.6312524550172916e-06,
+      "loss": 0.7484,
+      "step": 36890
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.9156467914581299,
+      "learning_rate": 3.6309988702153263e-06,
+      "loss": 0.5694,
+      "step": 36891
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.3328042030334473,
+      "learning_rate": 3.6307452892200406e-06,
+      "loss": 0.6344,
+      "step": 36892
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.8333418369293213,
+      "learning_rate": 3.630491712032139e-06,
+      "loss": 0.652,
+      "step": 36893
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.56809139251709,
+      "learning_rate": 3.6302381386523277e-06,
+      "loss": 0.6186,
+      "step": 36894
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.7931504249572754,
+      "learning_rate": 3.6299845690813103e-06,
+      "loss": 0.5507,
+      "step": 36895
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.4939351081848145,
+      "learning_rate": 3.6297310033197923e-06,
+      "loss": 0.6325,
+      "step": 36896
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.1296544075012207,
+      "learning_rate": 3.6294774413684788e-06,
+      "loss": 0.689,
+      "step": 36897
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.1413018703460693,
+      "learning_rate": 3.6292238832280757e-06,
+      "loss": 0.7169,
+      "step": 36898
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.259706974029541,
+      "learning_rate": 3.628970328899287e-06,
+      "loss": 0.4412,
+      "step": 36899
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.04935359954834,
+      "learning_rate": 3.6287167783828193e-06,
+      "loss": 0.7069,
+      "step": 36900
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.3915646076202393,
+      "learning_rate": 3.6284632316793757e-06,
+      "loss": 0.6289,
+      "step": 36901
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.6602182388305664,
+      "learning_rate": 3.6282096887896624e-06,
+      "loss": 0.6237,
+      "step": 36902
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.0569417476654053,
+      "learning_rate": 3.627956149714384e-06,
+      "loss": 0.5959,
+      "step": 36903
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.433500051498413,
+      "learning_rate": 3.627702614454246e-06,
+      "loss": 0.5621,
+      "step": 36904
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.1344099044799805,
+      "learning_rate": 3.627449083009952e-06,
+      "loss": 0.7259,
+      "step": 36905
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.795978307723999,
+      "learning_rate": 3.6271955553822086e-06,
+      "loss": 0.8582,
+      "step": 36906
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.9800184965133667,
+      "learning_rate": 3.6269420315717198e-06,
+      "loss": 0.5517,
+      "step": 36907
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.458186388015747,
+      "learning_rate": 3.6266885115791913e-06,
+      "loss": 0.5376,
+      "step": 36908
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.0441222190856934,
+      "learning_rate": 3.6264349954053253e-06,
+      "loss": 0.9906,
+      "step": 36909
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.142709493637085,
+      "learning_rate": 3.6261814830508313e-06,
+      "loss": 0.6114,
+      "step": 36910
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.6227707862854004,
+      "learning_rate": 3.625927974516411e-06,
+      "loss": 0.6109,
+      "step": 36911
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.797658920288086,
+      "learning_rate": 3.625674469802771e-06,
+      "loss": 0.8065,
+      "step": 36912
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.1846942901611328,
+      "learning_rate": 3.6254209689106158e-06,
+      "loss": 0.29,
+      "step": 36913
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.143793821334839,
+      "learning_rate": 3.625167471840649e-06,
+      "loss": 1.0341,
+      "step": 36914
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.3819777965545654,
+      "learning_rate": 3.6249139785935773e-06,
+      "loss": 0.5146,
+      "step": 36915
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.5274783372879028,
+      "learning_rate": 3.6246604891701045e-06,
+      "loss": 0.713,
+      "step": 36916
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.456427812576294,
+      "learning_rate": 3.6244070035709354e-06,
+      "loss": 0.702,
+      "step": 36917
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.662529230117798,
+      "learning_rate": 3.624153521796776e-06,
+      "loss": 0.5674,
+      "step": 36918
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.560380220413208,
+      "learning_rate": 3.6239000438483297e-06,
+      "loss": 0.7889,
+      "step": 36919
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.1345937252044678,
+      "learning_rate": 3.6236465697263013e-06,
+      "loss": 0.6398,
+      "step": 36920
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.3166027069091797,
+      "learning_rate": 3.6233930994313964e-06,
+      "loss": 0.6385,
+      "step": 36921
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.107750177383423,
+      "learning_rate": 3.6231396329643198e-06,
+      "loss": 0.3633,
+      "step": 36922
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.7077745199203491,
+      "learning_rate": 3.6228861703257766e-06,
+      "loss": 0.774,
+      "step": 36923
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.5475549697875977,
+      "learning_rate": 3.6226327115164707e-06,
+      "loss": 0.5403,
+      "step": 36924
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.817225694656372,
+      "learning_rate": 3.622379256537107e-06,
+      "loss": 0.6106,
+      "step": 36925
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.8409913778305054,
+      "learning_rate": 3.6221258053883908e-06,
+      "loss": 0.5312,
+      "step": 36926
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.9876468181610107,
+      "learning_rate": 3.621872358071026e-06,
+      "loss": 0.7228,
+      "step": 36927
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.957920789718628,
+      "learning_rate": 3.621618914585718e-06,
+      "loss": 0.5441,
+      "step": 36928
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.75090491771698,
+      "learning_rate": 3.6213654749331708e-06,
+      "loss": 0.8046,
+      "step": 36929
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 5.67184591293335,
+      "learning_rate": 3.6211120391140904e-06,
+      "loss": 0.6175,
+      "step": 36930
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.7191455364227295,
+      "learning_rate": 3.6208586071291812e-06,
+      "loss": 0.5284,
+      "step": 36931
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.5384387969970703,
+      "learning_rate": 3.6206051789791473e-06,
+      "loss": 0.712,
+      "step": 36932
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.7250638008117676,
+      "learning_rate": 3.6203517546646945e-06,
+      "loss": 0.5645,
+      "step": 36933
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.433288812637329,
+      "learning_rate": 3.620098334186526e-06,
+      "loss": 0.6955,
+      "step": 36934
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.0887699127197266,
+      "learning_rate": 3.6198449175453476e-06,
+      "loss": 0.3948,
+      "step": 36935
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.2105624675750732,
+      "learning_rate": 3.619591504741863e-06,
+      "loss": 0.525,
+      "step": 36936
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.241241455078125,
+      "learning_rate": 3.619338095776777e-06,
+      "loss": 0.4519,
+      "step": 36937
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.054798126220703,
+      "learning_rate": 3.619084690650795e-06,
+      "loss": 0.3119,
+      "step": 36938
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.405057907104492,
+      "learning_rate": 3.6188312893646217e-06,
+      "loss": 0.5255,
+      "step": 36939
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.4643664360046387,
+      "learning_rate": 3.6185778919189608e-06,
+      "loss": 0.5314,
+      "step": 36940
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.2980201244354248,
+      "learning_rate": 3.6183244983145168e-06,
+      "loss": 0.2906,
+      "step": 36941
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.432811737060547,
+      "learning_rate": 3.618071108551995e-06,
+      "loss": 0.651,
+      "step": 36942
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.9234340190887451,
+      "learning_rate": 3.6178177226321e-06,
+      "loss": 0.3949,
+      "step": 36943
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.859358787536621,
+      "learning_rate": 3.617564340555536e-06,
+      "loss": 0.7106,
+      "step": 36944
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.774761915206909,
+      "learning_rate": 3.6173109623230074e-06,
+      "loss": 0.5305,
+      "step": 36945
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.694464683532715,
+      "learning_rate": 3.6170575879352197e-06,
+      "loss": 0.5576,
+      "step": 36946
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.4561437368392944,
+      "learning_rate": 3.6168042173928753e-06,
+      "loss": 0.6421,
+      "step": 36947
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.953253149986267,
+      "learning_rate": 3.6165508506966812e-06,
+      "loss": 0.4458,
+      "step": 36948
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.8655762672424316,
+      "learning_rate": 3.61629748784734e-06,
+      "loss": 0.4944,
+      "step": 36949
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.075472831726074,
+      "learning_rate": 3.6160441288455584e-06,
+      "loss": 0.706,
+      "step": 36950
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.6416428089141846,
+      "learning_rate": 3.6157907736920395e-06,
+      "loss": 0.6189,
+      "step": 36951
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.784249782562256,
+      "learning_rate": 3.615537422387488e-06,
+      "loss": 0.6246,
+      "step": 36952
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.4586236476898193,
+      "learning_rate": 3.615284074932608e-06,
+      "loss": 0.5104,
+      "step": 36953
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.7359912395477295,
+      "learning_rate": 3.615030731328104e-06,
+      "loss": 0.4013,
+      "step": 36954
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.513087511062622,
+      "learning_rate": 3.614777391574682e-06,
+      "loss": 0.6891,
+      "step": 36955
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.647351026535034,
+      "learning_rate": 3.6145240556730443e-06,
+      "loss": 0.9261,
+      "step": 36956
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.6160475015640259,
+      "learning_rate": 3.614270723623896e-06,
+      "loss": 0.6429,
+      "step": 36957
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.3873589038848877,
+      "learning_rate": 3.6140173954279407e-06,
+      "loss": 0.5771,
+      "step": 36958
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.8822065591812134,
+      "learning_rate": 3.613764071085885e-06,
+      "loss": 0.5929,
+      "step": 36959
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.4092087745666504,
+      "learning_rate": 3.613510750598432e-06,
+      "loss": 0.4798,
+      "step": 36960
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.4370999336242676,
+      "learning_rate": 3.613257433966286e-06,
+      "loss": 0.7444,
+      "step": 36961
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.194350242614746,
+      "learning_rate": 3.613004121190152e-06,
+      "loss": 0.5895,
+      "step": 36962
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.8532333374023438,
+      "learning_rate": 3.6127508122707344e-06,
+      "loss": 0.5944,
+      "step": 36963
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.3766467571258545,
+      "learning_rate": 3.6124975072087365e-06,
+      "loss": 0.7742,
+      "step": 36964
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.0110738277435303,
+      "learning_rate": 3.6122442060048636e-06,
+      "loss": 0.7978,
+      "step": 36965
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.2258670330047607,
+      "learning_rate": 3.611990908659819e-06,
+      "loss": 0.5782,
+      "step": 36966
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.015801191329956,
+      "learning_rate": 3.611737615174309e-06,
+      "loss": 0.3151,
+      "step": 36967
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.74920392036438,
+      "learning_rate": 3.6114843255490356e-06,
+      "loss": 0.796,
+      "step": 36968
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.692868232727051,
+      "learning_rate": 3.6112310397847043e-06,
+      "loss": 0.6097,
+      "step": 36969
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.050417184829712,
+      "learning_rate": 3.610977757882017e-06,
+      "loss": 0.5029,
+      "step": 36970
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.9933066368103027,
+      "learning_rate": 3.610724479841683e-06,
+      "loss": 0.6448,
+      "step": 36971
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.2922425270080566,
+      "learning_rate": 3.610471205664403e-06,
+      "loss": 0.7406,
+      "step": 36972
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.325002431869507,
+      "learning_rate": 3.6102179353508827e-06,
+      "loss": 0.6129,
+      "step": 36973
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.1430301666259766,
+      "learning_rate": 3.6099646689018253e-06,
+      "loss": 0.5383,
+      "step": 36974
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 5.022866725921631,
+      "learning_rate": 3.6097114063179355e-06,
+      "loss": 0.8532,
+      "step": 36975
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.6883392333984375,
+      "learning_rate": 3.6094581475999173e-06,
+      "loss": 0.8039,
+      "step": 36976
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 4.664769649505615,
+      "learning_rate": 3.6092048927484756e-06,
+      "loss": 0.5971,
+      "step": 36977
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.9078289270401,
+      "learning_rate": 3.6089516417643133e-06,
+      "loss": 0.4351,
+      "step": 36978
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.5636680126190186,
+      "learning_rate": 3.6086983946481362e-06,
+      "loss": 0.7313,
+      "step": 36979
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.6182308197021484,
+      "learning_rate": 3.6084451514006475e-06,
+      "loss": 0.487,
+      "step": 36980
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.920660972595215,
+      "learning_rate": 3.6081919120225513e-06,
+      "loss": 0.6877,
+      "step": 36981
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.8117860555648804,
+      "learning_rate": 3.6079386765145517e-06,
+      "loss": 0.8387,
+      "step": 36982
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.857764720916748,
+      "learning_rate": 3.6076854448773537e-06,
+      "loss": 0.5634,
+      "step": 36983
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.4630203247070312,
+      "learning_rate": 3.6074322171116604e-06,
+      "loss": 0.4722,
+      "step": 36984
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.71574068069458,
+      "learning_rate": 3.6071789932181774e-06,
+      "loss": 0.7901,
+      "step": 36985
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 4.805689334869385,
+      "learning_rate": 3.6069257731976072e-06,
+      "loss": 0.7027,
+      "step": 36986
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.2295243740081787,
+      "learning_rate": 3.6066725570506543e-06,
+      "loss": 0.6023,
+      "step": 36987
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.906886339187622,
+      "learning_rate": 3.6064193447780237e-06,
+      "loss": 0.4994,
+      "step": 36988
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.0478529930114746,
+      "learning_rate": 3.6061661363804184e-06,
+      "loss": 0.4727,
+      "step": 36989
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.1846396923065186,
+      "learning_rate": 3.605912931858541e-06,
+      "loss": 0.6897,
+      "step": 36990
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.145864963531494,
+      "learning_rate": 3.6056597312130993e-06,
+      "loss": 0.4394,
+      "step": 36991
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 5.156406879425049,
+      "learning_rate": 3.6054065344447957e-06,
+      "loss": 0.4067,
+      "step": 36992
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.3319661617279053,
+      "learning_rate": 3.605153341554334e-06,
+      "loss": 0.5945,
+      "step": 36993
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.841567039489746,
+      "learning_rate": 3.604900152542419e-06,
+      "loss": 0.896,
+      "step": 36994
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.9786288738250732,
+      "learning_rate": 3.604646967409754e-06,
+      "loss": 0.31,
+      "step": 36995
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.809011697769165,
+      "learning_rate": 3.6043937861570423e-06,
+      "loss": 0.8349,
+      "step": 36996
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.272491455078125,
+      "learning_rate": 3.604140608784989e-06,
+      "loss": 0.3848,
+      "step": 36997
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.4421424865722656,
+      "learning_rate": 3.6038874352942975e-06,
+      "loss": 0.5687,
+      "step": 36998
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 4.16316556930542,
+      "learning_rate": 3.6036342656856725e-06,
+      "loss": 0.9488,
+      "step": 36999
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.7509331703186035,
+      "learning_rate": 3.6033810999598173e-06,
+      "loss": 0.7006,
+      "step": 37000
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.0265393257141113,
+      "learning_rate": 3.6031279381174367e-06,
+      "loss": 0.7411,
+      "step": 37001
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.8009229898452759,
+      "learning_rate": 3.6028747801592336e-06,
+      "loss": 0.7564,
+      "step": 37002
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.718522310256958,
+      "learning_rate": 3.6026216260859115e-06,
+      "loss": 0.5585,
+      "step": 37003
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.3745439052581787,
+      "learning_rate": 3.6023684758981765e-06,
+      "loss": 0.4623,
+      "step": 37004
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.213841199874878,
+      "learning_rate": 3.6021153295967303e-06,
+      "loss": 0.5444,
+      "step": 37005
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.7271721363067627,
+      "learning_rate": 3.601862187182278e-06,
+      "loss": 0.8728,
+      "step": 37006
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.7450950145721436,
+      "learning_rate": 3.601609048655523e-06,
+      "loss": 0.8064,
+      "step": 37007
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.4735069274902344,
+      "learning_rate": 3.601355914017169e-06,
+      "loss": 0.6567,
+      "step": 37008
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.519105911254883,
+      "learning_rate": 3.6011027832679213e-06,
+      "loss": 0.8262,
+      "step": 37009
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 4.128284454345703,
+      "learning_rate": 3.6008496564084804e-06,
+      "loss": 0.4838,
+      "step": 37010
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.698678970336914,
+      "learning_rate": 3.600596533439555e-06,
+      "loss": 0.459,
+      "step": 37011
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.5110323429107666,
+      "learning_rate": 3.6003434143618455e-06,
+      "loss": 0.5009,
+      "step": 37012
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.190552234649658,
+      "learning_rate": 3.6000902991760568e-06,
+      "loss": 0.5072,
+      "step": 37013
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.0028741359710693,
+      "learning_rate": 3.599837187882893e-06,
+      "loss": 0.4612,
+      "step": 37014
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.2123703956604004,
+      "learning_rate": 3.5995840804830575e-06,
+      "loss": 0.6511,
+      "step": 37015
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.409701108932495,
+      "learning_rate": 3.5993309769772534e-06,
+      "loss": 0.8304,
+      "step": 37016
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.9278337955474854,
+      "learning_rate": 3.5990778773661856e-06,
+      "loss": 0.5461,
+      "step": 37017
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.063199043273926,
+      "learning_rate": 3.5988247816505575e-06,
+      "loss": 0.6255,
+      "step": 37018
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.222663402557373,
+      "learning_rate": 3.5985716898310717e-06,
+      "loss": 0.7084,
+      "step": 37019
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.574615240097046,
+      "learning_rate": 3.598318601908434e-06,
+      "loss": 0.5585,
+      "step": 37020
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.514618992805481,
+      "learning_rate": 3.5980655178833466e-06,
+      "loss": 0.5667,
+      "step": 37021
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.7536768913269043,
+      "learning_rate": 3.597812437756514e-06,
+      "loss": 0.8119,
+      "step": 37022
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.1629397869110107,
+      "learning_rate": 3.5975593615286404e-06,
+      "loss": 0.6788,
+      "step": 37023
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.8280251026153564,
+      "learning_rate": 3.5973062892004284e-06,
+      "loss": 0.682,
+      "step": 37024
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.0501229763031006,
+      "learning_rate": 3.597053220772582e-06,
+      "loss": 0.5399,
+      "step": 37025
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 4.139041900634766,
+      "learning_rate": 3.5968001562458053e-06,
+      "loss": 0.8646,
+      "step": 37026
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.908874034881592,
+      "learning_rate": 3.596547095620802e-06,
+      "loss": 0.438,
+      "step": 37027
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.104703426361084,
+      "learning_rate": 3.5962940388982754e-06,
+      "loss": 0.4808,
+      "step": 37028
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.8550012111663818,
+      "learning_rate": 3.5960409860789293e-06,
+      "loss": 0.5401,
+      "step": 37029
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.3637025356292725,
+      "learning_rate": 3.5957879371634665e-06,
+      "loss": 0.485,
+      "step": 37030
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.6854288578033447,
+      "learning_rate": 3.5955348921525914e-06,
+      "loss": 0.4346,
+      "step": 37031
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.9754626750946045,
+      "learning_rate": 3.5952818510470066e-06,
+      "loss": 0.5298,
+      "step": 37032
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.183542251586914,
+      "learning_rate": 3.595028813847419e-06,
+      "loss": 0.6126,
+      "step": 37033
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.199582576751709,
+      "learning_rate": 3.5947757805545293e-06,
+      "loss": 0.7297,
+      "step": 37034
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 4.403595924377441,
+      "learning_rate": 3.5945227511690416e-06,
+      "loss": 0.8919,
+      "step": 37035
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.0042335987091064,
+      "learning_rate": 3.5942697256916594e-06,
+      "loss": 0.6943,
+      "step": 37036
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.641957998275757,
+      "learning_rate": 3.594016704123087e-06,
+      "loss": 0.3966,
+      "step": 37037
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.1368277072906494,
+      "learning_rate": 3.5937636864640275e-06,
+      "loss": 0.7739,
+      "step": 37038
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.8882687091827393,
+      "learning_rate": 3.5935106727151835e-06,
+      "loss": 0.6791,
+      "step": 37039
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.167543888092041,
+      "learning_rate": 3.593257662877261e-06,
+      "loss": 0.7584,
+      "step": 37040
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.3373806476593018,
+      "learning_rate": 3.5930046569509606e-06,
+      "loss": 0.5556,
+      "step": 37041
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.8572816848754883,
+      "learning_rate": 3.5927516549369877e-06,
+      "loss": 0.6892,
+      "step": 37042
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.5146772861480713,
+      "learning_rate": 3.5924986568360454e-06,
+      "loss": 0.62,
+      "step": 37043
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.454920768737793,
+      "learning_rate": 3.5922456626488365e-06,
+      "loss": 0.6727,
+      "step": 37044
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.9719318151474,
+      "learning_rate": 3.5919926723760654e-06,
+      "loss": 0.5701,
+      "step": 37045
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.8144900798797607,
+      "learning_rate": 3.591739686018436e-06,
+      "loss": 0.8833,
+      "step": 37046
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.1605570316314697,
+      "learning_rate": 3.59148670357665e-06,
+      "loss": 0.8896,
+      "step": 37047
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.5089879035949707,
+      "learning_rate": 3.5912337250514122e-06,
+      "loss": 0.5851,
+      "step": 37048
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.1366512775421143,
+      "learning_rate": 3.590980750443426e-06,
+      "loss": 0.571,
+      "step": 37049
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 4.998383522033691,
+      "learning_rate": 3.5907277797533933e-06,
+      "loss": 0.8569,
+      "step": 37050
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.9818127155303955,
+      "learning_rate": 3.5904748129820177e-06,
+      "loss": 0.5916,
+      "step": 37051
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.284757137298584,
+      "learning_rate": 3.590221850130006e-06,
+      "loss": 0.6301,
+      "step": 37052
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.922768473625183,
+      "learning_rate": 3.5899688911980586e-06,
+      "loss": 0.4499,
+      "step": 37053
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.7892415523529053,
+      "learning_rate": 3.5897159361868795e-06,
+      "loss": 0.4607,
+      "step": 37054
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.3476359844207764,
+      "learning_rate": 3.589462985097172e-06,
+      "loss": 0.602,
+      "step": 37055
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.206009864807129,
+      "learning_rate": 3.58921003792964e-06,
+      "loss": 0.6084,
+      "step": 37056
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.52240252494812,
+      "learning_rate": 3.588957094684986e-06,
+      "loss": 0.6288,
+      "step": 37057
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.4111785888671875,
+      "learning_rate": 3.5887041553639133e-06,
+      "loss": 0.583,
+      "step": 37058
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 4.427666187286377,
+      "learning_rate": 3.5884512199671268e-06,
+      "loss": 0.8737,
+      "step": 37059
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.9526640176773071,
+      "learning_rate": 3.5881982884953283e-06,
+      "loss": 0.5859,
+      "step": 37060
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.150972604751587,
+      "learning_rate": 3.587945360949221e-06,
+      "loss": 0.5904,
+      "step": 37061
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.9709891080856323,
+      "learning_rate": 3.5876924373295085e-06,
+      "loss": 0.6139,
+      "step": 37062
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.7181355953216553,
+      "learning_rate": 3.587439517636895e-06,
+      "loss": 0.7358,
+      "step": 37063
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.98971688747406,
+      "learning_rate": 3.5871866018720824e-06,
+      "loss": 0.3384,
+      "step": 37064
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.5750186443328857,
+      "learning_rate": 3.5869336900357752e-06,
+      "loss": 0.5975,
+      "step": 37065
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.4029359817504883,
+      "learning_rate": 3.586680782128676e-06,
+      "loss": 0.6812,
+      "step": 37066
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.951104760169983,
+      "learning_rate": 3.5864278781514882e-06,
+      "loss": 0.576,
+      "step": 37067
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.2524123191833496,
+      "learning_rate": 3.5861749781049145e-06,
+      "loss": 0.4537,
+      "step": 37068
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.354851722717285,
+      "learning_rate": 3.585922081989659e-06,
+      "loss": 0.5499,
+      "step": 37069
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 8.266767501831055,
+      "learning_rate": 3.5856691898064226e-06,
+      "loss": 0.8206,
+      "step": 37070
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.070830821990967,
+      "learning_rate": 3.585416301555911e-06,
+      "loss": 0.7104,
+      "step": 37071
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.9960501194000244,
+      "learning_rate": 3.585163417238828e-06,
+      "loss": 0.7407,
+      "step": 37072
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 4.258944034576416,
+      "learning_rate": 3.5849105368558746e-06,
+      "loss": 0.6437,
+      "step": 37073
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.676640033721924,
+      "learning_rate": 3.5846576604077556e-06,
+      "loss": 0.6791,
+      "step": 37074
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.8747828006744385,
+      "learning_rate": 3.5844047878951737e-06,
+      "loss": 0.5179,
+      "step": 37075
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.9294930696487427,
+      "learning_rate": 3.5841519193188313e-06,
+      "loss": 0.9774,
+      "step": 37076
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.1027040481567383,
+      "learning_rate": 3.583899054679432e-06,
+      "loss": 0.4746,
+      "step": 37077
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.4783971309661865,
+      "learning_rate": 3.583646193977679e-06,
+      "loss": 0.7758,
+      "step": 37078
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.375843048095703,
+      "learning_rate": 3.5833933372142758e-06,
+      "loss": 0.6498,
+      "step": 37079
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.453446388244629,
+      "learning_rate": 3.5831404843899248e-06,
+      "loss": 0.4438,
+      "step": 37080
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.506808280944824,
+      "learning_rate": 3.5828876355053286e-06,
+      "loss": 0.359,
+      "step": 37081
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.6349072456359863,
+      "learning_rate": 3.5826347905611915e-06,
+      "loss": 0.5641,
+      "step": 37082
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.322131633758545,
+      "learning_rate": 3.5823819495582157e-06,
+      "loss": 0.5517,
+      "step": 37083
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.085963487625122,
+      "learning_rate": 3.582129112497106e-06,
+      "loss": 0.5221,
+      "step": 37084
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.7724895477294922,
+      "learning_rate": 3.5818762793785627e-06,
+      "loss": 0.537,
+      "step": 37085
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.135967969894409,
+      "learning_rate": 3.5816234502032907e-06,
+      "loss": 0.343,
+      "step": 37086
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.3266303539276123,
+      "learning_rate": 3.581370624971993e-06,
+      "loss": 0.8335,
+      "step": 37087
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.268700122833252,
+      "learning_rate": 3.5811178036853723e-06,
+      "loss": 0.748,
+      "step": 37088
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.9684720039367676,
+      "learning_rate": 3.58086498634413e-06,
+      "loss": 0.4315,
+      "step": 37089
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.057417869567871,
+      "learning_rate": 3.5806121729489723e-06,
+      "loss": 0.8054,
+      "step": 37090
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.9317448139190674,
+      "learning_rate": 3.5803593635005996e-06,
+      "loss": 0.776,
+      "step": 37091
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.1626029014587402,
+      "learning_rate": 3.5801065579997147e-06,
+      "loss": 0.6731,
+      "step": 37092
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 5.391364574432373,
+      "learning_rate": 3.5798537564470214e-06,
+      "loss": 0.5886,
+      "step": 37093
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.356039524078369,
+      "learning_rate": 3.5796009588432246e-06,
+      "loss": 0.5085,
+      "step": 37094
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.3164383172988892,
+      "learning_rate": 3.579348165189025e-06,
+      "loss": 0.4227,
+      "step": 37095
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.0047805309295654,
+      "learning_rate": 3.5790953754851254e-06,
+      "loss": 0.4484,
+      "step": 37096
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.043247938156128,
+      "learning_rate": 3.57884258973223e-06,
+      "loss": 0.5645,
+      "step": 37097
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.9545241594314575,
+      "learning_rate": 3.578589807931041e-06,
+      "loss": 0.6323,
+      "step": 37098
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.2651023864746094,
+      "learning_rate": 3.5783370300822604e-06,
+      "loss": 0.7813,
+      "step": 37099
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 4.791755676269531,
+      "learning_rate": 3.5780842561865926e-06,
+      "loss": 0.9454,
+      "step": 37100
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.5044219493865967,
+      "learning_rate": 3.57783148624474e-06,
+      "loss": 0.5973,
+      "step": 37101
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.948539972305298,
+      "learning_rate": 3.5775787202574054e-06,
+      "loss": 0.4485,
+      "step": 37102
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 4.001760482788086,
+      "learning_rate": 3.5773259582252904e-06,
+      "loss": 0.6838,
+      "step": 37103
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.2536158561706543,
+      "learning_rate": 3.5770732001491005e-06,
+      "loss": 0.6443,
+      "step": 37104
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.9891445636749268,
+      "learning_rate": 3.5768204460295357e-06,
+      "loss": 0.4956,
+      "step": 37105
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.779483675956726,
+      "learning_rate": 3.576567695867301e-06,
+      "loss": 0.5357,
+      "step": 37106
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.6088104248046875,
+      "learning_rate": 3.576314949663098e-06,
+      "loss": 0.8028,
+      "step": 37107
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.0401480197906494,
+      "learning_rate": 3.5760622074176298e-06,
+      "loss": 0.6124,
+      "step": 37108
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.6862651109695435,
+      "learning_rate": 3.5758094691315993e-06,
+      "loss": 0.5299,
+      "step": 37109
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 4.873086929321289,
+      "learning_rate": 3.5755567348057084e-06,
+      "loss": 0.5772,
+      "step": 37110
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.708040237426758,
+      "learning_rate": 3.5753040044406617e-06,
+      "loss": 0.6315,
+      "step": 37111
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.1947598457336426,
+      "learning_rate": 3.5750512780371584e-06,
+      "loss": 0.6199,
+      "step": 37112
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.693861484527588,
+      "learning_rate": 3.5747985555959055e-06,
+      "loss": 0.5608,
+      "step": 37113
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 4.485633373260498,
+      "learning_rate": 3.574545837117604e-06,
+      "loss": 0.7412,
+      "step": 37114
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.349994659423828,
+      "learning_rate": 3.5742931226029567e-06,
+      "loss": 0.8441,
+      "step": 37115
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.9375429153442383,
+      "learning_rate": 3.574040412052666e-06,
+      "loss": 0.3999,
+      "step": 37116
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.4165027141571045,
+      "learning_rate": 3.573787705467435e-06,
+      "loss": 0.7362,
+      "step": 37117
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.0217738151550293,
+      "learning_rate": 3.5735350028479654e-06,
+      "loss": 0.6801,
+      "step": 37118
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 5.682142734527588,
+      "learning_rate": 3.5732823041949615e-06,
+      "loss": 0.6629,
+      "step": 37119
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.01041841506958,
+      "learning_rate": 3.5730296095091245e-06,
+      "loss": 0.7808,
+      "step": 37120
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.1823747158050537,
+      "learning_rate": 3.572776918791158e-06,
+      "loss": 0.5048,
+      "step": 37121
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.166072130203247,
+      "learning_rate": 3.5725242320417637e-06,
+      "loss": 0.5882,
+      "step": 37122
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.9145476818084717,
+      "learning_rate": 3.5722715492616448e-06,
+      "loss": 0.7185,
+      "step": 37123
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.9354751110076904,
+      "learning_rate": 3.5720188704515045e-06,
+      "loss": 0.5267,
+      "step": 37124
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.6389379501342773,
+      "learning_rate": 3.571766195612044e-06,
+      "loss": 0.4847,
+      "step": 37125
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.6046228408813477,
+      "learning_rate": 3.5715135247439676e-06,
+      "loss": 0.8016,
+      "step": 37126
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.9826217889785767,
+      "learning_rate": 3.5712608578479763e-06,
+      "loss": 0.6168,
+      "step": 37127
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.9103819131851196,
+      "learning_rate": 3.5710081949247733e-06,
+      "loss": 0.7045,
+      "step": 37128
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.6587467193603516,
+      "learning_rate": 3.570755535975062e-06,
+      "loss": 0.6566,
+      "step": 37129
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.9861350059509277,
+      "learning_rate": 3.5705028809995436e-06,
+      "loss": 0.5992,
+      "step": 37130
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.5871286392211914,
+      "learning_rate": 3.5702502299989205e-06,
+      "loss": 0.5112,
+      "step": 37131
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.1647160053253174,
+      "learning_rate": 3.569997582973895e-06,
+      "loss": 0.7613,
+      "step": 37132
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.067648410797119,
+      "learning_rate": 3.5697449399251717e-06,
+      "loss": 0.3372,
+      "step": 37133
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 4.383156776428223,
+      "learning_rate": 3.5694923008534525e-06,
+      "loss": 0.7225,
+      "step": 37134
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.919358491897583,
+      "learning_rate": 3.569239665759439e-06,
+      "loss": 0.253,
+      "step": 37135
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.4382379055023193,
+      "learning_rate": 3.5689870346438342e-06,
+      "loss": 0.651,
+      "step": 37136
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.519463539123535,
+      "learning_rate": 3.5687344075073403e-06,
+      "loss": 0.6345,
+      "step": 37137
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.7571661472320557,
+      "learning_rate": 3.5684817843506604e-06,
+      "loss": 1.0591,
+      "step": 37138
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.590398073196411,
+      "learning_rate": 3.568229165174496e-06,
+      "loss": 0.4867,
+      "step": 37139
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.0781350135803223,
+      "learning_rate": 3.5679765499795495e-06,
+      "loss": 0.6091,
+      "step": 37140
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.8855507373809814,
+      "learning_rate": 3.567723938766524e-06,
+      "loss": 0.7752,
+      "step": 37141
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.8135859966278076,
+      "learning_rate": 3.567471331536122e-06,
+      "loss": 0.4467,
+      "step": 37142
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.932502508163452,
+      "learning_rate": 3.5672187282890446e-06,
+      "loss": 0.7462,
+      "step": 37143
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.1331913471221924,
+      "learning_rate": 3.566966129025996e-06,
+      "loss": 0.4915,
+      "step": 37144
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.0055572986602783,
+      "learning_rate": 3.5667135337476776e-06,
+      "loss": 0.784,
+      "step": 37145
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.08870792388916,
+      "learning_rate": 3.566460942454792e-06,
+      "loss": 0.4458,
+      "step": 37146
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.218356132507324,
+      "learning_rate": 3.5662083551480412e-06,
+      "loss": 0.4663,
+      "step": 37147
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.717381238937378,
+      "learning_rate": 3.5659557718281284e-06,
+      "loss": 0.7738,
+      "step": 37148
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.7401492595672607,
+      "learning_rate": 3.5657031924957554e-06,
+      "loss": 0.5549,
+      "step": 37149
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 6.168588638305664,
+      "learning_rate": 3.5654506171516247e-06,
+      "loss": 0.4097,
+      "step": 37150
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.76908278465271,
+      "learning_rate": 3.5651980457964373e-06,
+      "loss": 0.5843,
+      "step": 37151
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.729943037033081,
+      "learning_rate": 3.564945478430898e-06,
+      "loss": 0.5873,
+      "step": 37152
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.569347858428955,
+      "learning_rate": 3.5646929150557066e-06,
+      "loss": 0.7074,
+      "step": 37153
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.46272873878479,
+      "learning_rate": 3.5644403556715668e-06,
+      "loss": 0.5093,
+      "step": 37154
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.353919506072998,
+      "learning_rate": 3.56418780027918e-06,
+      "loss": 0.4695,
+      "step": 37155
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.782112956047058,
+      "learning_rate": 3.5639352488792493e-06,
+      "loss": 0.9524,
+      "step": 37156
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.772401809692383,
+      "learning_rate": 3.563682701472477e-06,
+      "loss": 0.7957,
+      "step": 37157
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.6603620052337646,
+      "learning_rate": 3.563430158059566e-06,
+      "loss": 0.5733,
+      "step": 37158
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.0886623859405518,
+      "learning_rate": 3.563177618641217e-06,
+      "loss": 0.7667,
+      "step": 37159
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.09771466255188,
+      "learning_rate": 3.562925083218133e-06,
+      "loss": 0.7553,
+      "step": 37160
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 4.3108930587768555,
+      "learning_rate": 3.5626725517910155e-06,
+      "loss": 0.6416,
+      "step": 37161
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.0664896965026855,
+      "learning_rate": 3.5624200243605677e-06,
+      "loss": 0.7148,
+      "step": 37162
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.7090792655944824,
+      "learning_rate": 3.5621675009274916e-06,
+      "loss": 0.5947,
+      "step": 37163
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.4013898372650146,
+      "learning_rate": 3.561914981492488e-06,
+      "loss": 0.7822,
+      "step": 37164
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.041182279586792,
+      "learning_rate": 3.5616624660562605e-06,
+      "loss": 0.7017,
+      "step": 37165
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.159391164779663,
+      "learning_rate": 3.5614099546195112e-06,
+      "loss": 0.7139,
+      "step": 37166
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.940492630004883,
+      "learning_rate": 3.561157447182942e-06,
+      "loss": 0.7511,
+      "step": 37167
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.8583426475524902,
+      "learning_rate": 3.560904943747254e-06,
+      "loss": 0.6961,
+      "step": 37168
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.895597457885742,
+      "learning_rate": 3.5606524443131514e-06,
+      "loss": 0.6612,
+      "step": 37169
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.0375585556030273,
+      "learning_rate": 3.5603999488813345e-06,
+      "loss": 0.7635,
+      "step": 37170
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.8730387687683105,
+      "learning_rate": 3.5601474574525063e-06,
+      "loss": 0.847,
+      "step": 37171
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.667477607727051,
+      "learning_rate": 3.5598949700273687e-06,
+      "loss": 0.6251,
+      "step": 37172
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.3991291522979736,
+      "learning_rate": 3.5596424866066216e-06,
+      "loss": 0.7194,
+      "step": 37173
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 6.393105506896973,
+      "learning_rate": 3.559390007190971e-06,
+      "loss": 0.5026,
+      "step": 37174
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.7967082262039185,
+      "learning_rate": 3.559137531781117e-06,
+      "loss": 0.4256,
+      "step": 37175
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.5375938415527344,
+      "learning_rate": 3.5588850603777627e-06,
+      "loss": 0.4769,
+      "step": 37176
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.8759379386901855,
+      "learning_rate": 3.558632592981608e-06,
+      "loss": 0.4679,
+      "step": 37177
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.2050225734710693,
+      "learning_rate": 3.5583801295933567e-06,
+      "loss": 0.5444,
+      "step": 37178
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.8661389350891113,
+      "learning_rate": 3.5581276702137103e-06,
+      "loss": 0.4814,
+      "step": 37179
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.1496448516845703,
+      "learning_rate": 3.5578752148433703e-06,
+      "loss": 0.5919,
+      "step": 37180
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.0317680835723877,
+      "learning_rate": 3.5576227634830397e-06,
+      "loss": 0.8063,
+      "step": 37181
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.549353837966919,
+      "learning_rate": 3.5573703161334195e-06,
+      "loss": 0.4868,
+      "step": 37182
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.5151922702789307,
+      "learning_rate": 3.557117872795212e-06,
+      "loss": 0.7994,
+      "step": 37183
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.0038843154907227,
+      "learning_rate": 3.5568654334691195e-06,
+      "loss": 0.5882,
+      "step": 37184
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.762316942214966,
+      "learning_rate": 3.5566129981558434e-06,
+      "loss": 0.8246,
+      "step": 37185
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.965041399002075,
+      "learning_rate": 3.556360566856086e-06,
+      "loss": 0.6437,
+      "step": 37186
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.99259877204895,
+      "learning_rate": 3.556108139570549e-06,
+      "loss": 0.7075,
+      "step": 37187
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.7411422729492188,
+      "learning_rate": 3.555855716299934e-06,
+      "loss": 0.6415,
+      "step": 37188
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.97804594039917,
+      "learning_rate": 3.5556032970449435e-06,
+      "loss": 0.4803,
+      "step": 37189
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.1475000381469727,
+      "learning_rate": 3.5553508818062787e-06,
+      "loss": 0.5502,
+      "step": 37190
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 6.093835830688477,
+      "learning_rate": 3.5550984705846427e-06,
+      "loss": 0.5348,
+      "step": 37191
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.0522682666778564,
+      "learning_rate": 3.5548460633807364e-06,
+      "loss": 0.5765,
+      "step": 37192
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.8068927526474,
+      "learning_rate": 3.55459366019526e-06,
+      "loss": 0.5859,
+      "step": 37193
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.754573106765747,
+      "learning_rate": 3.5543412610289186e-06,
+      "loss": 0.7672,
+      "step": 37194
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.9006173610687256,
+      "learning_rate": 3.5540888658824124e-06,
+      "loss": 0.669,
+      "step": 37195
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.010298728942871,
+      "learning_rate": 3.553836474756444e-06,
+      "loss": 0.7541,
+      "step": 37196
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.0573959350585938,
+      "learning_rate": 3.5535840876517145e-06,
+      "loss": 0.4337,
+      "step": 37197
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.003920316696167,
+      "learning_rate": 3.5533317045689264e-06,
+      "loss": 0.6754,
+      "step": 37198
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.8972764015197754,
+      "learning_rate": 3.5530793255087803e-06,
+      "loss": 0.4902,
+      "step": 37199
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.259866237640381,
+      "learning_rate": 3.552826950471978e-06,
+      "loss": 0.9192,
+      "step": 37200
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.7259405851364136,
+      "learning_rate": 3.5525745794592226e-06,
+      "loss": 0.725,
+      "step": 37201
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.3917930126190186,
+      "learning_rate": 3.5523222124712154e-06,
+      "loss": 0.5099,
+      "step": 37202
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.0627150535583496,
+      "learning_rate": 3.552069849508657e-06,
+      "loss": 0.8322,
+      "step": 37203
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.5600242614746094,
+      "learning_rate": 3.55181749057225e-06,
+      "loss": 0.7895,
+      "step": 37204
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 4.104790687561035,
+      "learning_rate": 3.5515651356626956e-06,
+      "loss": 0.654,
+      "step": 37205
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.1094751358032227,
+      "learning_rate": 3.5513127847806963e-06,
+      "loss": 0.3962,
+      "step": 37206
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.6666643619537354,
+      "learning_rate": 3.551060437926954e-06,
+      "loss": 0.686,
+      "step": 37207
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.503504753112793,
+      "learning_rate": 3.5508080951021693e-06,
+      "loss": 0.6177,
+      "step": 37208
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.730466604232788,
+      "learning_rate": 3.5505557563070447e-06,
+      "loss": 0.6995,
+      "step": 37209
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.0760693550109863,
+      "learning_rate": 3.5503034215422817e-06,
+      "loss": 0.5296,
+      "step": 37210
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.005502939224243,
+      "learning_rate": 3.550051090808582e-06,
+      "loss": 0.8389,
+      "step": 37211
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.8953536748886108,
+      "learning_rate": 3.5497987641066466e-06,
+      "loss": 0.6949,
+      "step": 37212
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.2551794052124023,
+      "learning_rate": 3.549546441437178e-06,
+      "loss": 0.6829,
+      "step": 37213
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.596187114715576,
+      "learning_rate": 3.549294122800876e-06,
+      "loss": 0.6336,
+      "step": 37214
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.503537893295288,
+      "learning_rate": 3.5490418081984444e-06,
+      "loss": 0.4159,
+      "step": 37215
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.1220414638519287,
+      "learning_rate": 3.548789497630585e-06,
+      "loss": 0.6044,
+      "step": 37216
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.531731605529785,
+      "learning_rate": 3.5485371910979965e-06,
+      "loss": 0.5999,
+      "step": 37217
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.1458632946014404,
+      "learning_rate": 3.548284888601383e-06,
+      "loss": 0.5648,
+      "step": 37218
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.276257276535034,
+      "learning_rate": 3.5480325901414465e-06,
+      "loss": 0.5256,
+      "step": 37219
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.144970417022705,
+      "learning_rate": 3.5477802957188865e-06,
+      "loss": 0.7055,
+      "step": 37220
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.3029158115386963,
+      "learning_rate": 3.5475280053344062e-06,
+      "loss": 0.7408,
+      "step": 37221
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.409904956817627,
+      "learning_rate": 3.5472757189887065e-06,
+      "loss": 0.6097,
+      "step": 37222
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.359558343887329,
+      "learning_rate": 3.5470234366824883e-06,
+      "loss": 0.7325,
+      "step": 37223
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.341552257537842,
+      "learning_rate": 3.5467711584164533e-06,
+      "loss": 0.2884,
+      "step": 37224
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.8569767475128174,
+      "learning_rate": 3.546518884191304e-06,
+      "loss": 0.7574,
+      "step": 37225
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.6211860179901123,
+      "learning_rate": 3.5462666140077407e-06,
+      "loss": 0.6275,
+      "step": 37226
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.6587860584259033,
+      "learning_rate": 3.546014347866466e-06,
+      "loss": 0.373,
+      "step": 37227
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.7392020225524902,
+      "learning_rate": 3.54576208576818e-06,
+      "loss": 0.8795,
+      "step": 37228
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.724019765853882,
+      "learning_rate": 3.545509827713586e-06,
+      "loss": 0.4923,
+      "step": 37229
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.9776625633239746,
+      "learning_rate": 3.5452575737033833e-06,
+      "loss": 0.9036,
+      "step": 37230
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.636913299560547,
+      "learning_rate": 3.5450053237382742e-06,
+      "loss": 0.5719,
+      "step": 37231
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.8526577949523926,
+      "learning_rate": 3.544753077818961e-06,
+      "loss": 0.7249,
+      "step": 37232
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.2922415733337402,
+      "learning_rate": 3.5445008359461437e-06,
+      "loss": 0.7858,
+      "step": 37233
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.230883836746216,
+      "learning_rate": 3.544248598120522e-06,
+      "loss": 0.7482,
+      "step": 37234
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.1329362392425537,
+      "learning_rate": 3.5439963643428025e-06,
+      "loss": 0.9505,
+      "step": 37235
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.3788585662841797,
+      "learning_rate": 3.5437441346136837e-06,
+      "loss": 0.7676,
+      "step": 37236
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.7375659942626953,
+      "learning_rate": 3.543491908933866e-06,
+      "loss": 0.755,
+      "step": 37237
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.3601059913635254,
+      "learning_rate": 3.5432396873040525e-06,
+      "loss": 0.665,
+      "step": 37238
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.321230411529541,
+      "learning_rate": 3.5429874697249433e-06,
+      "loss": 0.7017,
+      "step": 37239
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.0300705432891846,
+      "learning_rate": 3.5427352561972405e-06,
+      "loss": 0.9757,
+      "step": 37240
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.5248537063598633,
+      "learning_rate": 3.5424830467216455e-06,
+      "loss": 0.8056,
+      "step": 37241
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.5935094356536865,
+      "learning_rate": 3.5422308412988583e-06,
+      "loss": 0.7828,
+      "step": 37242
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.106393814086914,
+      "learning_rate": 3.541978639929581e-06,
+      "loss": 0.7305,
+      "step": 37243
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.9992377758026123,
+      "learning_rate": 3.5417264426145155e-06,
+      "loss": 0.7248,
+      "step": 37244
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.012829303741455,
+      "learning_rate": 3.5414742493543617e-06,
+      "loss": 0.5751,
+      "step": 37245
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.082139730453491,
+      "learning_rate": 3.541222060149823e-06,
+      "loss": 0.4763,
+      "step": 37246
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.177649974822998,
+      "learning_rate": 3.5409698750015977e-06,
+      "loss": 0.4892,
+      "step": 37247
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.7285988330841064,
+      "learning_rate": 3.5407176939103893e-06,
+      "loss": 0.7292,
+      "step": 37248
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.2400267124176025,
+      "learning_rate": 3.5404655168768986e-06,
+      "loss": 0.5128,
+      "step": 37249
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.260831832885742,
+      "learning_rate": 3.540213343901826e-06,
+      "loss": 0.3733,
+      "step": 37250
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.3756613731384277,
+      "learning_rate": 3.539961174985873e-06,
+      "loss": 0.4018,
+      "step": 37251
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.6621509790420532,
+      "learning_rate": 3.539709010129742e-06,
+      "loss": 0.6878,
+      "step": 37252
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.074908494949341,
+      "learning_rate": 3.539456849334133e-06,
+      "loss": 0.5463,
+      "step": 37253
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.078319787979126,
+      "learning_rate": 3.5392046925997447e-06,
+      "loss": 0.4608,
+      "step": 37254
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.5929479598999023,
+      "learning_rate": 3.538952539927283e-06,
+      "loss": 0.5812,
+      "step": 37255
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.8976447582244873,
+      "learning_rate": 3.538700391317448e-06,
+      "loss": 0.7039,
+      "step": 37256
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.8623719215393066,
+      "learning_rate": 3.538448246770939e-06,
+      "loss": 0.5019,
+      "step": 37257
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.5997941493988037,
+      "learning_rate": 3.538196106288458e-06,
+      "loss": 0.7214,
+      "step": 37258
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.7624735832214355,
+      "learning_rate": 3.5379439698707064e-06,
+      "loss": 0.7773,
+      "step": 37259
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.707080841064453,
+      "learning_rate": 3.5376918375183845e-06,
+      "loss": 0.8417,
+      "step": 37260
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 4.875513076782227,
+      "learning_rate": 3.537439709232193e-06,
+      "loss": 0.7163,
+      "step": 37261
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.0285677909851074,
+      "learning_rate": 3.537187585012835e-06,
+      "loss": 0.7614,
+      "step": 37262
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.0413057804107666,
+      "learning_rate": 3.53693546486101e-06,
+      "loss": 0.4687,
+      "step": 37263
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.502194881439209,
+      "learning_rate": 3.53668334877742e-06,
+      "loss": 0.4446,
+      "step": 37264
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.1663360595703125,
+      "learning_rate": 3.536431236762764e-06,
+      "loss": 0.7882,
+      "step": 37265
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.4457013607025146,
+      "learning_rate": 3.5361791288177453e-06,
+      "loss": 0.4281,
+      "step": 37266
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.8367202281951904,
+      "learning_rate": 3.5359270249430633e-06,
+      "loss": 0.3159,
+      "step": 37267
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.699693202972412,
+      "learning_rate": 3.5356749251394206e-06,
+      "loss": 0.64,
+      "step": 37268
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.48472261428833,
+      "learning_rate": 3.5354228294075175e-06,
+      "loss": 0.7645,
+      "step": 37269
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.2094931602478027,
+      "learning_rate": 3.5351707377480547e-06,
+      "loss": 0.2925,
+      "step": 37270
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.5938236713409424,
+      "learning_rate": 3.5349186501617332e-06,
+      "loss": 0.7943,
+      "step": 37271
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.006558418273926,
+      "learning_rate": 3.534666566649254e-06,
+      "loss": 0.6403,
+      "step": 37272
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.0594663619995117,
+      "learning_rate": 3.5344144872113183e-06,
+      "loss": 0.7148,
+      "step": 37273
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.376152276992798,
+      "learning_rate": 3.534162411848627e-06,
+      "loss": 0.8112,
+      "step": 37274
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.8763291835784912,
+      "learning_rate": 3.53391034056188e-06,
+      "loss": 0.5399,
+      "step": 37275
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.4211363792419434,
+      "learning_rate": 3.5336582733517803e-06,
+      "loss": 0.6117,
+      "step": 37276
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.1156935691833496,
+      "learning_rate": 3.5334062102190277e-06,
+      "loss": 0.4221,
+      "step": 37277
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.6114490032196045,
+      "learning_rate": 3.5331541511643216e-06,
+      "loss": 0.7038,
+      "step": 37278
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.335731267929077,
+      "learning_rate": 3.5329020961883654e-06,
+      "loss": 0.7896,
+      "step": 37279
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.500560760498047,
+      "learning_rate": 3.5326500452918588e-06,
+      "loss": 0.8316,
+      "step": 37280
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.8799786567687988,
+      "learning_rate": 3.532397998475503e-06,
+      "loss": 0.5504,
+      "step": 37281
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 4.202545166015625,
+      "learning_rate": 3.5321459557399984e-06,
+      "loss": 0.7602,
+      "step": 37282
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.9769530296325684,
+      "learning_rate": 3.5318939170860457e-06,
+      "loss": 0.5616,
+      "step": 37283
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.4013257026672363,
+      "learning_rate": 3.5316418825143467e-06,
+      "loss": 0.5101,
+      "step": 37284
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.1096713542938232,
+      "learning_rate": 3.5313898520256014e-06,
+      "loss": 0.7284,
+      "step": 37285
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.706678867340088,
+      "learning_rate": 3.5311378256205108e-06,
+      "loss": 0.5828,
+      "step": 37286
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.158247947692871,
+      "learning_rate": 3.530885803299775e-06,
+      "loss": 0.6713,
+      "step": 37287
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.4563350677490234,
+      "learning_rate": 3.530633785064096e-06,
+      "loss": 0.666,
+      "step": 37288
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.9365530014038086,
+      "learning_rate": 3.530381770914174e-06,
+      "loss": 0.7553,
+      "step": 37289
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.6470582485198975,
+      "learning_rate": 3.53012976085071e-06,
+      "loss": 0.5844,
+      "step": 37290
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.992810010910034,
+      "learning_rate": 3.5298777548744047e-06,
+      "loss": 0.8099,
+      "step": 37291
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 5.031711101531982,
+      "learning_rate": 3.529625752985958e-06,
+      "loss": 0.5184,
+      "step": 37292
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.8802878856658936,
+      "learning_rate": 3.529373755186071e-06,
+      "loss": 0.8578,
+      "step": 37293
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 5.1000447273254395,
+      "learning_rate": 3.5291217614754454e-06,
+      "loss": 0.3129,
+      "step": 37294
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.4516348838806152,
+      "learning_rate": 3.528869771854778e-06,
+      "loss": 0.7434,
+      "step": 37295
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.6597678661346436,
+      "learning_rate": 3.528617786324776e-06,
+      "loss": 0.6076,
+      "step": 37296
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.165764808654785,
+      "learning_rate": 3.5283658048861362e-06,
+      "loss": 0.6516,
+      "step": 37297
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.1939749717712402,
+      "learning_rate": 3.52811382753956e-06,
+      "loss": 0.5805,
+      "step": 37298
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 5.094273090362549,
+      "learning_rate": 3.527861854285748e-06,
+      "loss": 0.6349,
+      "step": 37299
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.26937198638916,
+      "learning_rate": 3.5276098851254003e-06,
+      "loss": 0.5985,
+      "step": 37300
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.8102827072143555,
+      "learning_rate": 3.5273579200592183e-06,
+      "loss": 0.4852,
+      "step": 37301
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.209531307220459,
+      "learning_rate": 3.527105959087902e-06,
+      "loss": 0.6129,
+      "step": 37302
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.312303066253662,
+      "learning_rate": 3.5268540022121524e-06,
+      "loss": 0.8354,
+      "step": 37303
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.931273937225342,
+      "learning_rate": 3.5266020494326703e-06,
+      "loss": 0.6742,
+      "step": 37304
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 5.316593647003174,
+      "learning_rate": 3.5263501007501553e-06,
+      "loss": 0.6669,
+      "step": 37305
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.970053791999817,
+      "learning_rate": 3.5260981561653083e-06,
+      "loss": 0.5987,
+      "step": 37306
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.668506622314453,
+      "learning_rate": 3.5258462156788315e-06,
+      "loss": 0.5824,
+      "step": 37307
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.184582233428955,
+      "learning_rate": 3.5255942792914232e-06,
+      "loss": 0.5109,
+      "step": 37308
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.705465316772461,
+      "learning_rate": 3.525342347003785e-06,
+      "loss": 0.6569,
+      "step": 37309
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.351029634475708,
+      "learning_rate": 3.525090418816618e-06,
+      "loss": 0.5024,
+      "step": 37310
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.1580381393432617,
+      "learning_rate": 3.524838494730621e-06,
+      "loss": 0.582,
+      "step": 37311
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.1885123252868652,
+      "learning_rate": 3.524586574746496e-06,
+      "loss": 0.7153,
+      "step": 37312
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.5686302185058594,
+      "learning_rate": 3.524334658864943e-06,
+      "loss": 0.4178,
+      "step": 37313
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.226388692855835,
+      "learning_rate": 3.524082747086663e-06,
+      "loss": 0.3924,
+      "step": 37314
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.1792290210723877,
+      "learning_rate": 3.523830839412353e-06,
+      "loss": 0.5509,
+      "step": 37315
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.6884348392486572,
+      "learning_rate": 3.523578935842719e-06,
+      "loss": 0.757,
+      "step": 37316
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.101391077041626,
+      "learning_rate": 3.523327036378459e-06,
+      "loss": 0.7756,
+      "step": 37317
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.1646666526794434,
+      "learning_rate": 3.5230751410202734e-06,
+      "loss": 0.8686,
+      "step": 37318
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.9771710634231567,
+      "learning_rate": 3.5228232497688623e-06,
+      "loss": 0.6831,
+      "step": 37319
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.8591697216033936,
+      "learning_rate": 3.522571362624927e-06,
+      "loss": 0.612,
+      "step": 37320
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.290957450866699,
+      "learning_rate": 3.5223194795891662e-06,
+      "loss": 0.7769,
+      "step": 37321
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.4615628719329834,
+      "learning_rate": 3.5220676006622823e-06,
+      "loss": 0.559,
+      "step": 37322
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.2591352462768555,
+      "learning_rate": 3.5218157258449743e-06,
+      "loss": 0.67,
+      "step": 37323
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.228224277496338,
+      "learning_rate": 3.521563855137943e-06,
+      "loss": 0.4936,
+      "step": 37324
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.0554630756378174,
+      "learning_rate": 3.521311988541889e-06,
+      "loss": 0.6285,
+      "step": 37325
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.678426504135132,
+      "learning_rate": 3.5210601260575123e-06,
+      "loss": 0.6671,
+      "step": 37326
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.5161852836608887,
+      "learning_rate": 3.5208082676855133e-06,
+      "loss": 0.4954,
+      "step": 37327
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.185512065887451,
+      "learning_rate": 3.5205564134265925e-06,
+      "loss": 0.5812,
+      "step": 37328
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.6949615478515625,
+      "learning_rate": 3.5203045632814493e-06,
+      "loss": 0.4086,
+      "step": 37329
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.23553729057312,
+      "learning_rate": 3.520052717250785e-06,
+      "loss": 0.5904,
+      "step": 37330
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.0909907817840576,
+      "learning_rate": 3.5198008753353006e-06,
+      "loss": 0.7361,
+      "step": 37331
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.5221974849700928,
+      "learning_rate": 3.5195490375356957e-06,
+      "loss": 0.7144,
+      "step": 37332
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.9290966987609863,
+      "learning_rate": 3.5192972038526696e-06,
+      "loss": 0.7044,
+      "step": 37333
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.096797466278076,
+      "learning_rate": 3.5190453742869237e-06,
+      "loss": 0.7055,
+      "step": 37334
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.864717483520508,
+      "learning_rate": 3.518793548839157e-06,
+      "loss": 0.736,
+      "step": 37335
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.130566358566284,
+      "learning_rate": 3.5185417275100702e-06,
+      "loss": 0.6379,
+      "step": 37336
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.6084179878234863,
+      "learning_rate": 3.518289910300365e-06,
+      "loss": 0.7068,
+      "step": 37337
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.271200180053711,
+      "learning_rate": 3.51803809721074e-06,
+      "loss": 0.6463,
+      "step": 37338
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.0615692138671875,
+      "learning_rate": 3.5177862882418968e-06,
+      "loss": 0.7309,
+      "step": 37339
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.095855951309204,
+      "learning_rate": 3.517534483394533e-06,
+      "loss": 0.3481,
+      "step": 37340
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.1213009357452393,
+      "learning_rate": 3.517282682669352e-06,
+      "loss": 0.4403,
+      "step": 37341
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.4099860191345215,
+      "learning_rate": 3.5170308860670515e-06,
+      "loss": 0.9467,
+      "step": 37342
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.242072343826294,
+      "learning_rate": 3.516779093588333e-06,
+      "loss": 0.6763,
+      "step": 37343
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.090991258621216,
+      "learning_rate": 3.5165273052338967e-06,
+      "loss": 0.7815,
+      "step": 37344
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.070363998413086,
+      "learning_rate": 3.516275521004442e-06,
+      "loss": 0.7373,
+      "step": 37345
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.0001978874206543,
+      "learning_rate": 3.516023740900669e-06,
+      "loss": 0.776,
+      "step": 37346
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.6628258228302,
+      "learning_rate": 3.515771964923278e-06,
+      "loss": 0.7326,
+      "step": 37347
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.0777337551116943,
+      "learning_rate": 3.5155201930729693e-06,
+      "loss": 0.6678,
+      "step": 37348
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.667590379714966,
+      "learning_rate": 3.515268425350443e-06,
+      "loss": 0.6472,
+      "step": 37349
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.9571269750595093,
+      "learning_rate": 3.5150166617563987e-06,
+      "loss": 0.3506,
+      "step": 37350
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.2956252098083496,
+      "learning_rate": 3.514764902291537e-06,
+      "loss": 0.5178,
+      "step": 37351
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.1279733180999756,
+      "learning_rate": 3.5145131469565573e-06,
+      "loss": 0.7489,
+      "step": 37352
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.3450589179992676,
+      "learning_rate": 3.514261395752161e-06,
+      "loss": 0.5361,
+      "step": 37353
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.4984707832336426,
+      "learning_rate": 3.5140096486790464e-06,
+      "loss": 0.5722,
+      "step": 37354
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.318582773208618,
+      "learning_rate": 3.5137579057379146e-06,
+      "loss": 0.736,
+      "step": 37355
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.4774973392486572,
+      "learning_rate": 3.5135061669294633e-06,
+      "loss": 0.6643,
+      "step": 37356
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.915266990661621,
+      "learning_rate": 3.5132544322543964e-06,
+      "loss": 0.4447,
+      "step": 37357
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.893568515777588,
+      "learning_rate": 3.5130027017134123e-06,
+      "loss": 0.6655,
+      "step": 37358
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.228602647781372,
+      "learning_rate": 3.51275097530721e-06,
+      "loss": 0.5955,
+      "step": 37359
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 4.962152004241943,
+      "learning_rate": 3.512499253036491e-06,
+      "loss": 0.4531,
+      "step": 37360
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.716418981552124,
+      "learning_rate": 3.5122475349019536e-06,
+      "loss": 0.8203,
+      "step": 37361
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.2426443099975586,
+      "learning_rate": 3.511995820904299e-06,
+      "loss": 0.5136,
+      "step": 37362
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.8859138488769531,
+      "learning_rate": 3.5117441110442264e-06,
+      "loss": 0.4885,
+      "step": 37363
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.3228306770324707,
+      "learning_rate": 3.5114924053224354e-06,
+      "loss": 0.8105,
+      "step": 37364
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.194267749786377,
+      "learning_rate": 3.511240703739627e-06,
+      "loss": 0.661,
+      "step": 37365
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 6.975406646728516,
+      "learning_rate": 3.5109890062965e-06,
+      "loss": 0.6695,
+      "step": 37366
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.8898143768310547,
+      "learning_rate": 3.5107373129937555e-06,
+      "loss": 0.6436,
+      "step": 37367
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.641641616821289,
+      "learning_rate": 3.5104856238320923e-06,
+      "loss": 0.7903,
+      "step": 37368
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.757673740386963,
+      "learning_rate": 3.510233938812211e-06,
+      "loss": 0.5432,
+      "step": 37369
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.416447162628174,
+      "learning_rate": 3.50998225793481e-06,
+      "loss": 0.5699,
+      "step": 37370
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.8527355194091797,
+      "learning_rate": 3.509730581200591e-06,
+      "loss": 0.6097,
+      "step": 37371
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.886923313140869,
+      "learning_rate": 3.5094789086102527e-06,
+      "loss": 0.5333,
+      "step": 37372
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.6785902976989746,
+      "learning_rate": 3.509227240164495e-06,
+      "loss": 0.5075,
+      "step": 37373
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.3822224140167236,
+      "learning_rate": 3.5089755758640183e-06,
+      "loss": 0.5991,
+      "step": 37374
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.079374074935913,
+      "learning_rate": 3.5087239157095222e-06,
+      "loss": 0.603,
+      "step": 37375
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.0802013874053955,
+      "learning_rate": 3.508472259701703e-06,
+      "loss": 0.6758,
+      "step": 37376
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.907863140106201,
+      "learning_rate": 3.5082206078412672e-06,
+      "loss": 0.7396,
+      "step": 37377
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.950168490409851,
+      "learning_rate": 3.50796896012891e-06,
+      "loss": 0.7649,
+      "step": 37378
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.628650426864624,
+      "learning_rate": 3.5077173165653327e-06,
+      "loss": 0.6967,
+      "step": 37379
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.680914878845215,
+      "learning_rate": 3.5074656771512345e-06,
+      "loss": 0.5205,
+      "step": 37380
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.9881154298782349,
+      "learning_rate": 3.5072140418873145e-06,
+      "loss": 0.6037,
+      "step": 37381
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 5.566897869110107,
+      "learning_rate": 3.506962410774274e-06,
+      "loss": 0.9229,
+      "step": 37382
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.619218111038208,
+      "learning_rate": 3.506710783812811e-06,
+      "loss": 0.4603,
+      "step": 37383
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.4852583408355713,
+      "learning_rate": 3.5064591610036268e-06,
+      "loss": 0.7221,
+      "step": 37384
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.8731228113174438,
+      "learning_rate": 3.5062075423474195e-06,
+      "loss": 0.4558,
+      "step": 37385
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.905082941055298,
+      "learning_rate": 3.5059559278448894e-06,
+      "loss": 0.6574,
+      "step": 37386
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 11.471701622009277,
+      "learning_rate": 3.5057043174967366e-06,
+      "loss": 0.8431,
+      "step": 37387
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.8923258781433105,
+      "learning_rate": 3.5054527113036603e-06,
+      "loss": 0.8789,
+      "step": 37388
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.432025909423828,
+      "learning_rate": 3.5052011092663597e-06,
+      "loss": 0.569,
+      "step": 37389
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.8298494815826416,
+      "learning_rate": 3.5049495113855336e-06,
+      "loss": 0.777,
+      "step": 37390
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.9705138206481934,
+      "learning_rate": 3.5046979176618843e-06,
+      "loss": 0.5863,
+      "step": 37391
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.8499526977539062,
+      "learning_rate": 3.5044463280961105e-06,
+      "loss": 0.5253,
+      "step": 37392
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.301521062850952,
+      "learning_rate": 3.5041947426889106e-06,
+      "loss": 0.6678,
+      "step": 37393
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.913800001144409,
+      "learning_rate": 3.5039431614409846e-06,
+      "loss": 0.619,
+      "step": 37394
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.268580913543701,
+      "learning_rate": 3.5036915843530327e-06,
+      "loss": 0.6165,
+      "step": 37395
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.8717074394226074,
+      "learning_rate": 3.5034400114257537e-06,
+      "loss": 0.6869,
+      "step": 37396
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.427095651626587,
+      "learning_rate": 3.5031884426598463e-06,
+      "loss": 0.7938,
+      "step": 37397
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.894582748413086,
+      "learning_rate": 3.5029368780560123e-06,
+      "loss": 0.4745,
+      "step": 37398
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.464062213897705,
+      "learning_rate": 3.50268531761495e-06,
+      "loss": 0.5118,
+      "step": 37399
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.4114186763763428,
+      "learning_rate": 3.502433761337359e-06,
+      "loss": 0.6159,
+      "step": 37400
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.4522767066955566,
+      "learning_rate": 3.502182209223939e-06,
+      "loss": 0.4982,
+      "step": 37401
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.8648011684417725,
+      "learning_rate": 3.5019306612753874e-06,
+      "loss": 0.5851,
+      "step": 37402
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 4.142009258270264,
+      "learning_rate": 3.501679117492407e-06,
+      "loss": 0.721,
+      "step": 37403
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.299801826477051,
+      "learning_rate": 3.5014275778756957e-06,
+      "loss": 0.6443,
+      "step": 37404
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.7604950666427612,
+      "learning_rate": 3.5011760424259523e-06,
+      "loss": 0.4868,
+      "step": 37405
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 4.272071838378906,
+      "learning_rate": 3.5009245111438773e-06,
+      "loss": 0.6886,
+      "step": 37406
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.1659247875213623,
+      "learning_rate": 3.50067298403017e-06,
+      "loss": 0.4921,
+      "step": 37407
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.063706636428833,
+      "learning_rate": 3.5004214610855294e-06,
+      "loss": 0.7667,
+      "step": 37408
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.5867477655410767,
+      "learning_rate": 3.5001699423106544e-06,
+      "loss": 0.3078,
+      "step": 37409
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.307028293609619,
+      "learning_rate": 3.499918427706245e-06,
+      "loss": 0.6279,
+      "step": 37410
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 5.301944732666016,
+      "learning_rate": 3.499666917273001e-06,
+      "loss": 0.8441,
+      "step": 37411
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.838148832321167,
+      "learning_rate": 3.499415411011621e-06,
+      "loss": 0.7263,
+      "step": 37412
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.7250185012817383,
+      "learning_rate": 3.4991639089228045e-06,
+      "loss": 0.6208,
+      "step": 37413
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.2046709060668945,
+      "learning_rate": 3.4989124110072513e-06,
+      "loss": 0.5535,
+      "step": 37414
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.3870112895965576,
+      "learning_rate": 3.4986609172656605e-06,
+      "loss": 0.5689,
+      "step": 37415
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.4383046627044678,
+      "learning_rate": 3.49840942769873e-06,
+      "loss": 0.6305,
+      "step": 37416
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.623563528060913,
+      "learning_rate": 3.49815794230716e-06,
+      "loss": 0.5592,
+      "step": 37417
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.2253661155700684,
+      "learning_rate": 3.497906461091651e-06,
+      "loss": 0.5134,
+      "step": 37418
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.938682794570923,
+      "learning_rate": 3.4976549840529016e-06,
+      "loss": 0.7943,
+      "step": 37419
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.246056079864502,
+      "learning_rate": 3.4974035111916114e-06,
+      "loss": 0.5596,
+      "step": 37420
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.4358417987823486,
+      "learning_rate": 3.4971520425084794e-06,
+      "loss": 0.7027,
+      "step": 37421
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.5669726133346558,
+      "learning_rate": 3.4969005780042044e-06,
+      "loss": 0.327,
+      "step": 37422
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.27573823928833,
+      "learning_rate": 3.4966491176794846e-06,
+      "loss": 0.7651,
+      "step": 37423
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.951136827468872,
+      "learning_rate": 3.496397661535022e-06,
+      "loss": 0.4941,
+      "step": 37424
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.5108938217163086,
+      "learning_rate": 3.4961462095715136e-06,
+      "loss": 0.6785,
+      "step": 37425
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.0687673091888428,
+      "learning_rate": 3.4958947617896594e-06,
+      "loss": 0.6834,
+      "step": 37426
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.6024194955825806,
+      "learning_rate": 3.4956433181901583e-06,
+      "loss": 0.5771,
+      "step": 37427
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.9206949472427368,
+      "learning_rate": 3.4953918787737095e-06,
+      "loss": 0.6785,
+      "step": 37428
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.0087621212005615,
+      "learning_rate": 3.495140443541012e-06,
+      "loss": 0.8034,
+      "step": 37429
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.8868601322174072,
+      "learning_rate": 3.494889012492766e-06,
+      "loss": 0.6294,
+      "step": 37430
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.096414804458618,
+      "learning_rate": 3.494637585629669e-06,
+      "loss": 0.8119,
+      "step": 37431
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.479644536972046,
+      "learning_rate": 3.494386162952421e-06,
+      "loss": 0.506,
+      "step": 37432
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.7411696910858154,
+      "learning_rate": 3.494134744461722e-06,
+      "loss": 0.6972,
+      "step": 37433
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 17.83685302734375,
+      "learning_rate": 3.49388333015827e-06,
+      "loss": 0.7455,
+      "step": 37434
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.218803644180298,
+      "learning_rate": 3.4936319200427634e-06,
+      "loss": 0.6881,
+      "step": 37435
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.5873913764953613,
+      "learning_rate": 3.493380514115902e-06,
+      "loss": 0.5373,
+      "step": 37436
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.7433924674987793,
+      "learning_rate": 3.493129112378384e-06,
+      "loss": 0.5953,
+      "step": 37437
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.7082269191741943,
+      "learning_rate": 3.492877714830911e-06,
+      "loss": 0.537,
+      "step": 37438
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.9185116291046143,
+      "learning_rate": 3.49262632147418e-06,
+      "loss": 0.3506,
+      "step": 37439
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.0791049003601074,
+      "learning_rate": 3.492374932308891e-06,
+      "loss": 0.7508,
+      "step": 37440
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.476661443710327,
+      "learning_rate": 3.492123547335743e-06,
+      "loss": 0.5528,
+      "step": 37441
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.014011859893799,
+      "learning_rate": 3.4918721665554334e-06,
+      "loss": 0.4759,
+      "step": 37442
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.3050074577331543,
+      "learning_rate": 3.491620789968663e-06,
+      "loss": 0.5652,
+      "step": 37443
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.730175256729126,
+      "learning_rate": 3.4913694175761303e-06,
+      "loss": 0.6417,
+      "step": 37444
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.6124837398529053,
+      "learning_rate": 3.491118049378534e-06,
+      "loss": 0.6612,
+      "step": 37445
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 4.371466159820557,
+      "learning_rate": 3.490866685376573e-06,
+      "loss": 0.7371,
+      "step": 37446
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.4528632164001465,
+      "learning_rate": 3.490615325570947e-06,
+      "loss": 0.6985,
+      "step": 37447
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.716909885406494,
+      "learning_rate": 3.4903639699623537e-06,
+      "loss": 0.5506,
+      "step": 37448
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.1810405254364014,
+      "learning_rate": 3.4901126185514926e-06,
+      "loss": 0.5876,
+      "step": 37449
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.109224319458008,
+      "learning_rate": 3.4898612713390635e-06,
+      "loss": 0.9347,
+      "step": 37450
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.0195674896240234,
+      "learning_rate": 3.4896099283257634e-06,
+      "loss": 0.4882,
+      "step": 37451
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.118800401687622,
+      "learning_rate": 3.4893585895122915e-06,
+      "loss": 0.6031,
+      "step": 37452
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.0432965755462646,
+      "learning_rate": 3.489107254899349e-06,
+      "loss": 0.3826,
+      "step": 37453
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.9595491886138916,
+      "learning_rate": 3.488855924487633e-06,
+      "loss": 0.7508,
+      "step": 37454
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.3757412433624268,
+      "learning_rate": 3.488604598277843e-06,
+      "loss": 0.6126,
+      "step": 37455
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.7552762031555176,
+      "learning_rate": 3.4883532762706773e-06,
+      "loss": 0.6994,
+      "step": 37456
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.9243104457855225,
+      "learning_rate": 3.488101958466834e-06,
+      "loss": 0.4189,
+      "step": 37457
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.593928575515747,
+      "learning_rate": 3.4878506448670125e-06,
+      "loss": 0.6499,
+      "step": 37458
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.6717326641082764,
+      "learning_rate": 3.487599335471913e-06,
+      "loss": 0.7742,
+      "step": 37459
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.062572956085205,
+      "learning_rate": 3.4873480302822327e-06,
+      "loss": 0.4913,
+      "step": 37460
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.8333585262298584,
+      "learning_rate": 3.4870967292986714e-06,
+      "loss": 0.2999,
+      "step": 37461
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.1013877391815186,
+      "learning_rate": 3.486845432521927e-06,
+      "loss": 0.5365,
+      "step": 37462
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.3965091705322266,
+      "learning_rate": 3.486594139952699e-06,
+      "loss": 0.4996,
+      "step": 37463
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.1849565505981445,
+      "learning_rate": 3.4863428515916844e-06,
+      "loss": 0.71,
+      "step": 37464
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.1120362281799316,
+      "learning_rate": 3.4860915674395847e-06,
+      "loss": 0.5256,
+      "step": 37465
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.1914710998535156,
+      "learning_rate": 3.4858402874970964e-06,
+      "loss": 0.7461,
+      "step": 37466
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.6342735290527344,
+      "learning_rate": 3.4855890117649192e-06,
+      "loss": 0.6624,
+      "step": 37467
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.842708706855774,
+      "learning_rate": 3.485337740243753e-06,
+      "loss": 0.6338,
+      "step": 37468
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.650082588195801,
+      "learning_rate": 3.485086472934294e-06,
+      "loss": 0.6638,
+      "step": 37469
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.6978657245635986,
+      "learning_rate": 3.484835209837242e-06,
+      "loss": 0.7731,
+      "step": 37470
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.992923617362976,
+      "learning_rate": 3.4845839509532954e-06,
+      "loss": 0.6068,
+      "step": 37471
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.421950101852417,
+      "learning_rate": 3.4843326962831536e-06,
+      "loss": 0.7811,
+      "step": 37472
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.7022451162338257,
+      "learning_rate": 3.4840814458275154e-06,
+      "loss": 0.5238,
+      "step": 37473
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.128171682357788,
+      "learning_rate": 3.4838301995870777e-06,
+      "loss": 0.5212,
+      "step": 37474
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.7178176641464233,
+      "learning_rate": 3.4835789575625406e-06,
+      "loss": 0.5279,
+      "step": 37475
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 10.472213745117188,
+      "learning_rate": 3.483327719754602e-06,
+      "loss": 0.5539,
+      "step": 37476
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 4.844064235687256,
+      "learning_rate": 3.483076486163962e-06,
+      "loss": 0.6292,
+      "step": 37477
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.719676971435547,
+      "learning_rate": 3.482825256791315e-06,
+      "loss": 0.6424,
+      "step": 37478
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.7065980434417725,
+      "learning_rate": 3.482574031637365e-06,
+      "loss": 0.5073,
+      "step": 37479
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.7727150917053223,
+      "learning_rate": 3.4823228107028083e-06,
+      "loss": 0.4637,
+      "step": 37480
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.654954671859741,
+      "learning_rate": 3.4820715939883433e-06,
+      "loss": 0.5201,
+      "step": 37481
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.6689519882202148,
+      "learning_rate": 3.481820381494668e-06,
+      "loss": 0.4913,
+      "step": 37482
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.3955769538879395,
+      "learning_rate": 3.4815691732224822e-06,
+      "loss": 0.8491,
+      "step": 37483
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.9383186101913452,
+      "learning_rate": 3.4813179691724837e-06,
+      "loss": 0.5402,
+      "step": 37484
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.9730980396270752,
+      "learning_rate": 3.4810667693453705e-06,
+      "loss": 0.5868,
+      "step": 37485
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.750849723815918,
+      "learning_rate": 3.4808155737418427e-06,
+      "loss": 0.6247,
+      "step": 37486
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.718530535697937,
+      "learning_rate": 3.480564382362597e-06,
+      "loss": 0.6811,
+      "step": 37487
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.9248847961425781,
+      "learning_rate": 3.480313195208332e-06,
+      "loss": 0.5559,
+      "step": 37488
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.075627326965332,
+      "learning_rate": 3.480062012279747e-06,
+      "loss": 0.7328,
+      "step": 37489
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.347592353820801,
+      "learning_rate": 3.4798108335775404e-06,
+      "loss": 0.5864,
+      "step": 37490
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.601522445678711,
+      "learning_rate": 3.4795596591024105e-06,
+      "loss": 0.4922,
+      "step": 37491
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.1153926849365234,
+      "learning_rate": 3.4793084888550554e-06,
+      "loss": 0.4615,
+      "step": 37492
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.4639172554016113,
+      "learning_rate": 3.4790573228361734e-06,
+      "loss": 0.6949,
+      "step": 37493
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.1485965251922607,
+      "learning_rate": 3.4788061610464643e-06,
+      "loss": 0.7084,
+      "step": 37494
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.6257898807525635,
+      "learning_rate": 3.4785550034866246e-06,
+      "loss": 0.581,
+      "step": 37495
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.139256477355957,
+      "learning_rate": 3.478303850157353e-06,
+      "loss": 0.5299,
+      "step": 37496
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.660205602645874,
+      "learning_rate": 3.478052701059349e-06,
+      "loss": 0.8911,
+      "step": 37497
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.5652031898498535,
+      "learning_rate": 3.4778015561933087e-06,
+      "loss": 0.5795,
+      "step": 37498
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.085315227508545,
+      "learning_rate": 3.477550415559933e-06,
+      "loss": 0.8144,
+      "step": 37499
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.0319387912750244,
+      "learning_rate": 3.4772992791599204e-06,
+      "loss": 0.6456,
+      "step": 37500
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.8383307456970215,
+      "learning_rate": 3.4770481469939676e-06,
+      "loss": 0.6981,
+      "step": 37501
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.859347939491272,
+      "learning_rate": 3.476797019062773e-06,
+      "loss": 0.6604,
+      "step": 37502
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 4.131293296813965,
+      "learning_rate": 3.4765458953670357e-06,
+      "loss": 0.4403,
+      "step": 37503
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.5869693756103516,
+      "learning_rate": 3.4762947759074538e-06,
+      "loss": 0.6034,
+      "step": 37504
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.4808216094970703,
+      "learning_rate": 3.4760436606847247e-06,
+      "loss": 0.4826,
+      "step": 37505
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.937633514404297,
+      "learning_rate": 3.4757925496995478e-06,
+      "loss": 0.7771,
+      "step": 37506
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.4749335050582886,
+      "learning_rate": 3.47554144295262e-06,
+      "loss": 0.3518,
+      "step": 37507
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.8968796730041504,
+      "learning_rate": 3.4752903404446416e-06,
+      "loss": 0.4688,
+      "step": 37508
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.067774772644043,
+      "learning_rate": 3.475039242176309e-06,
+      "loss": 0.72,
+      "step": 37509
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.066106081008911,
+      "learning_rate": 3.4747881481483214e-06,
+      "loss": 0.5769,
+      "step": 37510
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 14.69554615020752,
+      "learning_rate": 3.474537058361376e-06,
+      "loss": 0.8107,
+      "step": 37511
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.182086229324341,
+      "learning_rate": 3.4742859728161716e-06,
+      "loss": 0.615,
+      "step": 37512
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.3878843784332275,
+      "learning_rate": 3.474034891513407e-06,
+      "loss": 0.8142,
+      "step": 37513
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.5479612350463867,
+      "learning_rate": 3.4737838144537785e-06,
+      "loss": 0.5151,
+      "step": 37514
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.9300179481506348,
+      "learning_rate": 3.473532741637986e-06,
+      "loss": 0.6159,
+      "step": 37515
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.5177886486053467,
+      "learning_rate": 3.4732816730667273e-06,
+      "loss": 0.5997,
+      "step": 37516
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.5872700214385986,
+      "learning_rate": 3.4730306087407008e-06,
+      "loss": 0.6758,
+      "step": 37517
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.7911629676818848,
+      "learning_rate": 3.4727795486606035e-06,
+      "loss": 0.7121,
+      "step": 37518
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.063810348510742,
+      "learning_rate": 3.4725284928271334e-06,
+      "loss": 0.6973,
+      "step": 37519
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 4.474408149719238,
+      "learning_rate": 3.4722774412409906e-06,
+      "loss": 0.7046,
+      "step": 37520
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.836474061012268,
+      "learning_rate": 3.472026393902872e-06,
+      "loss": 0.5849,
+      "step": 37521
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.670058012008667,
+      "learning_rate": 3.4717753508134753e-06,
+      "loss": 0.3471,
+      "step": 37522
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.6902363300323486,
+      "learning_rate": 3.4715243119734987e-06,
+      "loss": 0.6632,
+      "step": 37523
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.4753551483154297,
+      "learning_rate": 3.471273277383641e-06,
+      "loss": 0.6019,
+      "step": 37524
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.767339825630188,
+      "learning_rate": 3.471022247044598e-06,
+      "loss": 0.619,
+      "step": 37525
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.469543695449829,
+      "learning_rate": 3.470771220957071e-06,
+      "loss": 0.8708,
+      "step": 37526
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.0902938842773438,
+      "learning_rate": 3.4705201991217557e-06,
+      "loss": 0.6407,
+      "step": 37527
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.1718764305114746,
+      "learning_rate": 3.4702691815393507e-06,
+      "loss": 0.6572,
+      "step": 37528
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.076709032058716,
+      "learning_rate": 3.470018168210555e-06,
+      "loss": 0.6911,
+      "step": 37529
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 5.3692851066589355,
+      "learning_rate": 3.469767159136065e-06,
+      "loss": 0.7136,
+      "step": 37530
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.2441318035125732,
+      "learning_rate": 3.4695161543165793e-06,
+      "loss": 0.5069,
+      "step": 37531
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 4.443845748901367,
+      "learning_rate": 3.469265153752796e-06,
+      "loss": 0.4786,
+      "step": 37532
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.174258232116699,
+      "learning_rate": 3.469014157445413e-06,
+      "loss": 0.671,
+      "step": 37533
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.4858222007751465,
+      "learning_rate": 3.468763165395128e-06,
+      "loss": 0.5724,
+      "step": 37534
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.2728183269500732,
+      "learning_rate": 3.468512177602639e-06,
+      "loss": 0.5545,
+      "step": 37535
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.2052924633026123,
+      "learning_rate": 3.468261194068644e-06,
+      "loss": 0.6206,
+      "step": 37536
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.8944783210754395,
+      "learning_rate": 3.4680102147938406e-06,
+      "loss": 0.525,
+      "step": 37537
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.6817307472229004,
+      "learning_rate": 3.4677592397789265e-06,
+      "loss": 0.7347,
+      "step": 37538
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 4.332157611846924,
+      "learning_rate": 3.4675082690245988e-06,
+      "loss": 0.783,
+      "step": 37539
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.9878454208374023,
+      "learning_rate": 3.4672573025315582e-06,
+      "loss": 0.6039,
+      "step": 37540
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.7308803796768188,
+      "learning_rate": 3.4670063403005007e-06,
+      "loss": 0.6548,
+      "step": 37541
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.6209797859191895,
+      "learning_rate": 3.4667553823321243e-06,
+      "loss": 0.5817,
+      "step": 37542
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.7116289138793945,
+      "learning_rate": 3.4665044286271273e-06,
+      "loss": 0.6209,
+      "step": 37543
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.863097667694092,
+      "learning_rate": 3.4662534791862063e-06,
+      "loss": 0.4799,
+      "step": 37544
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.4936130046844482,
+      "learning_rate": 3.4660025340100602e-06,
+      "loss": 0.7452,
+      "step": 37545
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.6244401931762695,
+      "learning_rate": 3.465751593099387e-06,
+      "loss": 0.8526,
+      "step": 37546
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.0631680488586426,
+      "learning_rate": 3.4655006564548825e-06,
+      "loss": 0.7543,
+      "step": 37547
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.698371171951294,
+      "learning_rate": 3.4652497240772463e-06,
+      "loss": 0.5477,
+      "step": 37548
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.894904613494873,
+      "learning_rate": 3.4649987959671768e-06,
+      "loss": 0.6148,
+      "step": 37549
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.14025616645813,
+      "learning_rate": 3.464747872125369e-06,
+      "loss": 0.8425,
+      "step": 37550
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.678778648376465,
+      "learning_rate": 3.4644969525525234e-06,
+      "loss": 0.6011,
+      "step": 37551
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 4.769870758056641,
+      "learning_rate": 3.4642460372493363e-06,
+      "loss": 0.6342,
+      "step": 37552
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.0423576831817627,
+      "learning_rate": 3.4639951262165055e-06,
+      "loss": 0.672,
+      "step": 37553
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.641341209411621,
+      "learning_rate": 3.4637442194547288e-06,
+      "loss": 0.4235,
+      "step": 37554
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.0249691009521484,
+      "learning_rate": 3.4634933169647043e-06,
+      "loss": 0.5849,
+      "step": 37555
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.3923799991607666,
+      "learning_rate": 3.4632424187471293e-06,
+      "loss": 0.4592,
+      "step": 37556
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.8274458646774292,
+      "learning_rate": 3.4629915248027012e-06,
+      "loss": 0.5141,
+      "step": 37557
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.0903100967407227,
+      "learning_rate": 3.462740635132118e-06,
+      "loss": 0.6501,
+      "step": 37558
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.716102361679077,
+      "learning_rate": 3.462489749736075e-06,
+      "loss": 0.7762,
+      "step": 37559
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.7951081991195679,
+      "learning_rate": 3.462238868615274e-06,
+      "loss": 0.8121,
+      "step": 37560
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.185415744781494,
+      "learning_rate": 3.461987991770411e-06,
+      "loss": 0.5133,
+      "step": 37561
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.6115503311157227,
+      "learning_rate": 3.4617371192021833e-06,
+      "loss": 0.6094,
+      "step": 37562
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.1769795417785645,
+      "learning_rate": 3.4614862509112877e-06,
+      "loss": 0.8467,
+      "step": 37563
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.1005706787109375,
+      "learning_rate": 3.4612353868984227e-06,
+      "loss": 0.8368,
+      "step": 37564
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.3206608295440674,
+      "learning_rate": 3.460984527164286e-06,
+      "loss": 0.7382,
+      "step": 37565
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.009643077850342,
+      "learning_rate": 3.460733671709575e-06,
+      "loss": 0.5607,
+      "step": 37566
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.5018210411071777,
+      "learning_rate": 3.4604828205349874e-06,
+      "loss": 0.4935,
+      "step": 37567
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.576359987258911,
+      "learning_rate": 3.4602319736412194e-06,
+      "loss": 0.6205,
+      "step": 37568
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.065009593963623,
+      "learning_rate": 3.45998113102897e-06,
+      "loss": 0.5647,
+      "step": 37569
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.9258071184158325,
+      "learning_rate": 3.459730292698936e-06,
+      "loss": 0.6381,
+      "step": 37570
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.0755727291107178,
+      "learning_rate": 3.459479458651815e-06,
+      "loss": 0.5503,
+      "step": 37571
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.7719029188156128,
+      "learning_rate": 3.459228628888304e-06,
+      "loss": 0.6928,
+      "step": 37572
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.8916457891464233,
+      "learning_rate": 3.458977803409102e-06,
+      "loss": 0.8119,
+      "step": 37573
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.0363144874572754,
+      "learning_rate": 3.458726982214905e-06,
+      "loss": 0.4968,
+      "step": 37574
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.548767566680908,
+      "learning_rate": 3.4584761653064098e-06,
+      "loss": 0.389,
+      "step": 37575
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.196950674057007,
+      "learning_rate": 3.4582253526843156e-06,
+      "loss": 0.4686,
+      "step": 37576
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.8543184995651245,
+      "learning_rate": 3.45797454434932e-06,
+      "loss": 0.5456,
+      "step": 37577
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.5025522708892822,
+      "learning_rate": 3.457723740302119e-06,
+      "loss": 0.854,
+      "step": 37578
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.4920766353607178,
+      "learning_rate": 3.4574729405434106e-06,
+      "loss": 0.4519,
+      "step": 37579
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.275408983230591,
+      "learning_rate": 3.4572221450738912e-06,
+      "loss": 0.9399,
+      "step": 37580
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.612356662750244,
+      "learning_rate": 3.45697135389426e-06,
+      "loss": 0.6338,
+      "step": 37581
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.3774189949035645,
+      "learning_rate": 3.4567205670052135e-06,
+      "loss": 0.7462,
+      "step": 37582
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.1675379276275635,
+      "learning_rate": 3.456469784407449e-06,
+      "loss": 0.4101,
+      "step": 37583
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.6678919792175293,
+      "learning_rate": 3.4562190061016636e-06,
+      "loss": 0.8049,
+      "step": 37584
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.23293399810791,
+      "learning_rate": 3.4559682320885547e-06,
+      "loss": 0.4044,
+      "step": 37585
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.7228567600250244,
+      "learning_rate": 3.4557174623688204e-06,
+      "loss": 0.5247,
+      "step": 37586
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.340766429901123,
+      "learning_rate": 3.455466696943156e-06,
+      "loss": 0.7768,
+      "step": 37587
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 4.629790782928467,
+      "learning_rate": 3.4552159358122604e-06,
+      "loss": 0.7068,
+      "step": 37588
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.5876576900482178,
+      "learning_rate": 3.454965178976831e-06,
+      "loss": 0.615,
+      "step": 37589
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.129913330078125,
+      "learning_rate": 3.4547144264375655e-06,
+      "loss": 0.7982,
+      "step": 37590
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.213510751724243,
+      "learning_rate": 3.45446367819516e-06,
+      "loss": 0.5447,
+      "step": 37591
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.8497824668884277,
+      "learning_rate": 3.454212934250312e-06,
+      "loss": 0.6079,
+      "step": 37592
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.578681230545044,
+      "learning_rate": 3.4539621946037182e-06,
+      "loss": 0.5565,
+      "step": 37593
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.7408019304275513,
+      "learning_rate": 3.4537114592560765e-06,
+      "loss": 0.7142,
+      "step": 37594
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.7932541370391846,
+      "learning_rate": 3.4534607282080846e-06,
+      "loss": 0.5625,
+      "step": 37595
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.819165825843811,
+      "learning_rate": 3.4532100014604386e-06,
+      "loss": 0.6541,
+      "step": 37596
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.8739519119262695,
+      "learning_rate": 3.4529592790138354e-06,
+      "loss": 0.3813,
+      "step": 37597
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.9313561916351318,
+      "learning_rate": 3.452708560868974e-06,
+      "loss": 0.5463,
+      "step": 37598
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.916964292526245,
+      "learning_rate": 3.4524578470265503e-06,
+      "loss": 0.7179,
+      "step": 37599
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.883741855621338,
+      "learning_rate": 3.452207137487259e-06,
+      "loss": 0.5641,
+      "step": 37600
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.9320787191390991,
+      "learning_rate": 3.451956432251803e-06,
+      "loss": 0.6556,
+      "step": 37601
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.742703914642334,
+      "learning_rate": 3.451705731320875e-06,
+      "loss": 0.6144,
+      "step": 37602
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 4.285853385925293,
+      "learning_rate": 3.451455034695174e-06,
+      "loss": 0.6521,
+      "step": 37603
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.376516103744507,
+      "learning_rate": 3.451204342375396e-06,
+      "loss": 0.3969,
+      "step": 37604
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.6540565490722656,
+      "learning_rate": 3.450953654362239e-06,
+      "loss": 0.5356,
+      "step": 37605
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.869598865509033,
+      "learning_rate": 3.4507029706564e-06,
+      "loss": 0.5857,
+      "step": 37606
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.951702117919922,
+      "learning_rate": 3.4504522912585747e-06,
+      "loss": 0.4464,
+      "step": 37607
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.242995262145996,
+      "learning_rate": 3.450201616169462e-06,
+      "loss": 0.563,
+      "step": 37608
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 5.472290992736816,
+      "learning_rate": 3.4499509453897574e-06,
+      "loss": 0.397,
+      "step": 37609
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.8934524059295654,
+      "learning_rate": 3.4497002789201583e-06,
+      "loss": 0.54,
+      "step": 37610
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.8446900844573975,
+      "learning_rate": 3.4494496167613627e-06,
+      "loss": 0.7659,
+      "step": 37611
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.5152947902679443,
+      "learning_rate": 3.449198958914067e-06,
+      "loss": 0.4821,
+      "step": 37612
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.3606326580047607,
+      "learning_rate": 3.4489483053789675e-06,
+      "loss": 0.6176,
+      "step": 37613
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.2395474910736084,
+      "learning_rate": 3.448697656156762e-06,
+      "loss": 0.7634,
+      "step": 37614
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.046110153198242,
+      "learning_rate": 3.4484470112481466e-06,
+      "loss": 0.4712,
+      "step": 37615
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.930337905883789,
+      "learning_rate": 3.4481963706538193e-06,
+      "loss": 0.6103,
+      "step": 37616
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.3654983043670654,
+      "learning_rate": 3.4479457343744772e-06,
+      "loss": 0.7472,
+      "step": 37617
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.5267410278320312,
+      "learning_rate": 3.447695102410815e-06,
+      "loss": 0.2064,
+      "step": 37618
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.275768756866455,
+      "learning_rate": 3.447444474763533e-06,
+      "loss": 0.5084,
+      "step": 37619
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.9548403024673462,
+      "learning_rate": 3.447193851433323e-06,
+      "loss": 0.6323,
+      "step": 37620
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.9228787422180176,
+      "learning_rate": 3.4469432324208874e-06,
+      "loss": 0.4411,
+      "step": 37621
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.1679391860961914,
+      "learning_rate": 3.446692617726921e-06,
+      "loss": 0.5188,
+      "step": 37622
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.4101576805114746,
+      "learning_rate": 3.446442007352121e-06,
+      "loss": 0.3568,
+      "step": 37623
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.25962233543396,
+      "learning_rate": 3.446191401297184e-06,
+      "loss": 0.4439,
+      "step": 37624
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.8051989078521729,
+      "learning_rate": 3.4459407995628057e-06,
+      "loss": 0.5435,
+      "step": 37625
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.1847915649414062,
+      "learning_rate": 3.445690202149685e-06,
+      "loss": 0.6535,
+      "step": 37626
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.535954713821411,
+      "learning_rate": 3.445439609058516e-06,
+      "loss": 0.4685,
+      "step": 37627
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.9859718084335327,
+      "learning_rate": 3.4451890202899983e-06,
+      "loss": 0.4983,
+      "step": 37628
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.312854528427124,
+      "learning_rate": 3.4449384358448273e-06,
+      "loss": 0.668,
+      "step": 37629
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.484788179397583,
+      "learning_rate": 3.4446878557236998e-06,
+      "loss": 0.5671,
+      "step": 37630
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.8749176263809204,
+      "learning_rate": 3.444437279927313e-06,
+      "loss": 0.4008,
+      "step": 37631
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.556171178817749,
+      "learning_rate": 3.4441867084563628e-06,
+      "loss": 0.6375,
+      "step": 37632
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.362067222595215,
+      "learning_rate": 3.4439361413115475e-06,
+      "loss": 0.6044,
+      "step": 37633
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.2695205211639404,
+      "learning_rate": 3.443685578493562e-06,
+      "loss": 0.7121,
+      "step": 37634
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.9750683307647705,
+      "learning_rate": 3.4434350200031043e-06,
+      "loss": 0.582,
+      "step": 37635
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.1758432388305664,
+      "learning_rate": 3.443184465840871e-06,
+      "loss": 0.5872,
+      "step": 37636
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.8702892065048218,
+      "learning_rate": 3.442933916007557e-06,
+      "loss": 0.5318,
+      "step": 37637
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.050086498260498,
+      "learning_rate": 3.4426833705038615e-06,
+      "loss": 0.8315,
+      "step": 37638
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.6403326988220215,
+      "learning_rate": 3.4424328293304803e-06,
+      "loss": 0.5923,
+      "step": 37639
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.9276853799819946,
+      "learning_rate": 3.44218229248811e-06,
+      "loss": 0.5375,
+      "step": 37640
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.6977760791778564,
+      "learning_rate": 3.441931759977446e-06,
+      "loss": 0.5981,
+      "step": 37641
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.9589929580688477,
+      "learning_rate": 3.4416812317991872e-06,
+      "loss": 0.7246,
+      "step": 37642
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.478271484375,
+      "learning_rate": 3.4414307079540288e-06,
+      "loss": 0.6625,
+      "step": 37643
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.878880262374878,
+      "learning_rate": 3.4411801884426686e-06,
+      "loss": 0.3547,
+      "step": 37644
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.245481491088867,
+      "learning_rate": 3.4409296732658014e-06,
+      "loss": 0.3613,
+      "step": 37645
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.4038374423980713,
+      "learning_rate": 3.4406791624241254e-06,
+      "loss": 0.6366,
+      "step": 37646
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.873676061630249,
+      "learning_rate": 3.4404286559183366e-06,
+      "loss": 0.5008,
+      "step": 37647
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.019639253616333,
+      "learning_rate": 3.4401781537491307e-06,
+      "loss": 0.7109,
+      "step": 37648
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.537318468093872,
+      "learning_rate": 3.4399276559172045e-06,
+      "loss": 0.6581,
+      "step": 37649
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.9618233442306519,
+      "learning_rate": 3.4396771624232563e-06,
+      "loss": 0.6733,
+      "step": 37650
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 4.758909702301025,
+      "learning_rate": 3.439426673267981e-06,
+      "loss": 0.4202,
+      "step": 37651
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.590799331665039,
+      "learning_rate": 3.439176188452076e-06,
+      "loss": 0.7824,
+      "step": 37652
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.1021180152893066,
+      "learning_rate": 3.438925707976237e-06,
+      "loss": 0.6398,
+      "step": 37653
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.330950975418091,
+      "learning_rate": 3.4386752318411607e-06,
+      "loss": 0.7831,
+      "step": 37654
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.329601287841797,
+      "learning_rate": 3.438424760047544e-06,
+      "loss": 0.3849,
+      "step": 37655
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.3160126209259033,
+      "learning_rate": 3.438174292596083e-06,
+      "loss": 0.4703,
+      "step": 37656
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.35026478767395,
+      "learning_rate": 3.437923829487474e-06,
+      "loss": 0.4531,
+      "step": 37657
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.517606496810913,
+      "learning_rate": 3.4376733707224135e-06,
+      "loss": 0.5215,
+      "step": 37658
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.5360357761383057,
+      "learning_rate": 3.4374229163015984e-06,
+      "loss": 0.5919,
+      "step": 37659
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.6632298231124878,
+      "learning_rate": 3.4371724662257243e-06,
+      "loss": 0.5146,
+      "step": 37660
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.0336828231811523,
+      "learning_rate": 3.4369220204954867e-06,
+      "loss": 0.6298,
+      "step": 37661
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.188707113265991,
+      "learning_rate": 3.4366715791115856e-06,
+      "loss": 0.4915,
+      "step": 37662
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.7797855138778687,
+      "learning_rate": 3.436421142074715e-06,
+      "loss": 0.8479,
+      "step": 37663
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.070465326309204,
+      "learning_rate": 3.4361707093855713e-06,
+      "loss": 0.3684,
+      "step": 37664
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.1271839141845703,
+      "learning_rate": 3.435920281044851e-06,
+      "loss": 0.4673,
+      "step": 37665
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.141307830810547,
+      "learning_rate": 3.435669857053251e-06,
+      "loss": 0.6656,
+      "step": 37666
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.4493072032928467,
+      "learning_rate": 3.4354194374114664e-06,
+      "loss": 0.667,
+      "step": 37667
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.65706467628479,
+      "learning_rate": 3.4351690221201946e-06,
+      "loss": 0.626,
+      "step": 37668
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.2378957271575928,
+      "learning_rate": 3.4349186111801315e-06,
+      "loss": 0.56,
+      "step": 37669
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.7526922225952148,
+      "learning_rate": 3.4346682045919733e-06,
+      "loss": 0.564,
+      "step": 37670
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.3970658779144287,
+      "learning_rate": 3.4344178023564167e-06,
+      "loss": 0.6759,
+      "step": 37671
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.9684890508651733,
+      "learning_rate": 3.434167404474158e-06,
+      "loss": 0.4884,
+      "step": 37672
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.2217719554901123,
+      "learning_rate": 3.4339170109458926e-06,
+      "loss": 0.6874,
+      "step": 37673
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.8364733457565308,
+      "learning_rate": 3.433666621772317e-06,
+      "loss": 0.5722,
+      "step": 37674
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.9602344036102295,
+      "learning_rate": 3.4334162369541285e-06,
+      "loss": 0.4479,
+      "step": 37675
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.7781460285186768,
+      "learning_rate": 3.433165856492022e-06,
+      "loss": 0.6424,
+      "step": 37676
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.252182960510254,
+      "learning_rate": 3.4329154803866947e-06,
+      "loss": 0.5744,
+      "step": 37677
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.264859914779663,
+      "learning_rate": 3.432665108638843e-06,
+      "loss": 0.6857,
+      "step": 37678
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.1946935653686523,
+      "learning_rate": 3.4324147412491615e-06,
+      "loss": 0.623,
+      "step": 37679
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.9159013032913208,
+      "learning_rate": 3.432164378218348e-06,
+      "loss": 0.7253,
+      "step": 37680
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.8284225463867188,
+      "learning_rate": 3.431914019547096e-06,
+      "loss": 0.8983,
+      "step": 37681
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.1131486892700195,
+      "learning_rate": 3.4316636652361044e-06,
+      "loss": 0.5451,
+      "step": 37682
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.740288496017456,
+      "learning_rate": 3.4314133152860706e-06,
+      "loss": 0.7264,
+      "step": 37683
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.8418660163879395,
+      "learning_rate": 3.431162969697688e-06,
+      "loss": 0.5504,
+      "step": 37684
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 4.107572555541992,
+      "learning_rate": 3.430912628471653e-06,
+      "loss": 0.5117,
+      "step": 37685
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.1625101566314697,
+      "learning_rate": 3.4306622916086624e-06,
+      "loss": 0.6243,
+      "step": 37686
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.621659994125366,
+      "learning_rate": 3.430411959109412e-06,
+      "loss": 0.3999,
+      "step": 37687
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.124267578125,
+      "learning_rate": 3.430161630974599e-06,
+      "loss": 0.5882,
+      "step": 37688
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.7502901554107666,
+      "learning_rate": 3.429911307204917e-06,
+      "loss": 0.6643,
+      "step": 37689
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.108217239379883,
+      "learning_rate": 3.4296609878010645e-06,
+      "loss": 0.4631,
+      "step": 37690
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.8984878063201904,
+      "learning_rate": 3.4294106727637356e-06,
+      "loss": 0.6566,
+      "step": 37691
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.7855490446090698,
+      "learning_rate": 3.429160362093628e-06,
+      "loss": 0.3395,
+      "step": 37692
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.707547187805176,
+      "learning_rate": 3.4289100557914367e-06,
+      "loss": 0.7178,
+      "step": 37693
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.2985851764678955,
+      "learning_rate": 3.428659753857858e-06,
+      "loss": 0.77,
+      "step": 37694
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.395517349243164,
+      "learning_rate": 3.4284094562935876e-06,
+      "loss": 0.748,
+      "step": 37695
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 4.653256893157959,
+      "learning_rate": 3.428159163099323e-06,
+      "loss": 0.7648,
+      "step": 37696
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.4483706951141357,
+      "learning_rate": 3.4279088742757575e-06,
+      "loss": 0.4772,
+      "step": 37697
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.5595200061798096,
+      "learning_rate": 3.427658589823589e-06,
+      "loss": 0.7838,
+      "step": 37698
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.7702059745788574,
+      "learning_rate": 3.427408309743512e-06,
+      "loss": 0.4187,
+      "step": 37699
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.522467613220215,
+      "learning_rate": 3.4271580340362242e-06,
+      "loss": 0.6328,
+      "step": 37700
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.9917490482330322,
+      "learning_rate": 3.4269077627024217e-06,
+      "loss": 0.5349,
+      "step": 37701
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.8210550546646118,
+      "learning_rate": 3.4266574957427973e-06,
+      "loss": 0.2925,
+      "step": 37702
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.6854231357574463,
+      "learning_rate": 3.4264072331580502e-06,
+      "loss": 0.5679,
+      "step": 37703
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.8826377391815186,
+      "learning_rate": 3.4261569749488754e-06,
+      "loss": 0.5289,
+      "step": 37704
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 5.71334171295166,
+      "learning_rate": 3.425906721115968e-06,
+      "loss": 0.6058,
+      "step": 37705
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.3288004398345947,
+      "learning_rate": 3.425656471660025e-06,
+      "loss": 0.545,
+      "step": 37706
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.2564308643341064,
+      "learning_rate": 3.425406226581741e-06,
+      "loss": 0.7081,
+      "step": 37707
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.3209545612335205,
+      "learning_rate": 3.4251559858818127e-06,
+      "loss": 0.6602,
+      "step": 37708
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 6.4482293128967285,
+      "learning_rate": 3.4249057495609357e-06,
+      "loss": 0.506,
+      "step": 37709
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.1720688343048096,
+      "learning_rate": 3.424655517619805e-06,
+      "loss": 0.4574,
+      "step": 37710
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.218496322631836,
+      "learning_rate": 3.4244052900591175e-06,
+      "loss": 0.7734,
+      "step": 37711
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.1642117500305176,
+      "learning_rate": 3.4241550668795688e-06,
+      "loss": 0.5528,
+      "step": 37712
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 5.855498313903809,
+      "learning_rate": 3.423904848081855e-06,
+      "loss": 0.8283,
+      "step": 37713
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.8721072673797607,
+      "learning_rate": 3.423654633666671e-06,
+      "loss": 0.741,
+      "step": 37714
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.2331745624542236,
+      "learning_rate": 3.423404423634713e-06,
+      "loss": 0.4162,
+      "step": 37715
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.2564916610717773,
+      "learning_rate": 3.4231542179866773e-06,
+      "loss": 0.5842,
+      "step": 37716
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.9357051849365234,
+      "learning_rate": 3.4229040167232586e-06,
+      "loss": 0.7209,
+      "step": 37717
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.9852938652038574,
+      "learning_rate": 3.4226538198451527e-06,
+      "loss": 0.7257,
+      "step": 37718
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 6.383162021636963,
+      "learning_rate": 3.422403627353056e-06,
+      "loss": 0.591,
+      "step": 37719
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.0370163917541504,
+      "learning_rate": 3.4221534392476633e-06,
+      "loss": 0.7326,
+      "step": 37720
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.69468092918396,
+      "learning_rate": 3.4219032555296716e-06,
+      "loss": 0.6506,
+      "step": 37721
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 4.186865329742432,
+      "learning_rate": 3.4216530761997734e-06,
+      "loss": 0.6047,
+      "step": 37722
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.538445472717285,
+      "learning_rate": 3.4214029012586684e-06,
+      "loss": 0.7263,
+      "step": 37723
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.436922311782837,
+      "learning_rate": 3.4211527307070512e-06,
+      "loss": 0.571,
+      "step": 37724
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.395906686782837,
+      "learning_rate": 3.4209025645456164e-06,
+      "loss": 0.3282,
+      "step": 37725
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.3312859535217285,
+      "learning_rate": 3.4206524027750598e-06,
+      "loss": 0.5495,
+      "step": 37726
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.9029877185821533,
+      "learning_rate": 3.420402245396078e-06,
+      "loss": 0.5361,
+      "step": 37727
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.647533893585205,
+      "learning_rate": 3.4201520924093657e-06,
+      "loss": 0.6499,
+      "step": 37728
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.4090869426727295,
+      "learning_rate": 3.419901943815618e-06,
+      "loss": 0.4307,
+      "step": 37729
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.83652400970459,
+      "learning_rate": 3.419651799615531e-06,
+      "loss": 0.4053,
+      "step": 37730
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.1501481533050537,
+      "learning_rate": 3.419401659809801e-06,
+      "loss": 0.789,
+      "step": 37731
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.124302387237549,
+      "learning_rate": 3.4191515243991225e-06,
+      "loss": 0.6895,
+      "step": 37732
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.1991190910339355,
+      "learning_rate": 3.418901393384192e-06,
+      "loss": 0.553,
+      "step": 37733
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.730682849884033,
+      "learning_rate": 3.4186512667657034e-06,
+      "loss": 0.6263,
+      "step": 37734
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.1253271102905273,
+      "learning_rate": 3.4184011445443543e-06,
+      "loss": 0.7357,
+      "step": 37735
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.8450583219528198,
+      "learning_rate": 3.4181510267208383e-06,
+      "loss": 0.223,
+      "step": 37736
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 4.695634841918945,
+      "learning_rate": 3.417900913295852e-06,
+      "loss": 0.4067,
+      "step": 37737
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.3342788219451904,
+      "learning_rate": 3.4176508042700906e-06,
+      "loss": 0.6032,
+      "step": 37738
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.6951100826263428,
+      "learning_rate": 3.41740069964425e-06,
+      "loss": 0.6198,
+      "step": 37739
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.4046313762664795,
+      "learning_rate": 3.417150599419025e-06,
+      "loss": 0.5119,
+      "step": 37740
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.112388849258423,
+      "learning_rate": 3.416900503595111e-06,
+      "loss": 0.654,
+      "step": 37741
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.8888463973999023,
+      "learning_rate": 3.4166504121732012e-06,
+      "loss": 0.6263,
+      "step": 37742
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.9078574180603027,
+      "learning_rate": 3.4164003251539955e-06,
+      "loss": 0.841,
+      "step": 37743
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.027803659439087,
+      "learning_rate": 3.4161502425381875e-06,
+      "loss": 0.6607,
+      "step": 37744
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.081681966781616,
+      "learning_rate": 3.415900164326473e-06,
+      "loss": 0.6183,
+      "step": 37745
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.562781572341919,
+      "learning_rate": 3.4156500905195465e-06,
+      "loss": 0.6725,
+      "step": 37746
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.320345401763916,
+      "learning_rate": 3.4154000211181034e-06,
+      "loss": 0.6726,
+      "step": 37747
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.2571911811828613,
+      "learning_rate": 3.415149956122839e-06,
+      "loss": 0.4925,
+      "step": 37748
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.7815356254577637,
+      "learning_rate": 3.4148998955344493e-06,
+      "loss": 0.7536,
+      "step": 37749
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.198315143585205,
+      "learning_rate": 3.4146498393536286e-06,
+      "loss": 0.5316,
+      "step": 37750
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.925047516822815,
+      "learning_rate": 3.414399787581073e-06,
+      "loss": 0.6037,
+      "step": 37751
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.3567183017730713,
+      "learning_rate": 3.4141497402174783e-06,
+      "loss": 0.6812,
+      "step": 37752
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.245392084121704,
+      "learning_rate": 3.4138996972635387e-06,
+      "loss": 0.6387,
+      "step": 37753
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.5343728065490723,
+      "learning_rate": 3.41364965871995e-06,
+      "loss": 0.7133,
+      "step": 37754
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 5.42744779586792,
+      "learning_rate": 3.413399624587407e-06,
+      "loss": 0.5081,
+      "step": 37755
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.3380510807037354,
+      "learning_rate": 3.413149594866605e-06,
+      "loss": 0.6133,
+      "step": 37756
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.637223243713379,
+      "learning_rate": 3.41289956955824e-06,
+      "loss": 0.6376,
+      "step": 37757
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.0499765872955322,
+      "learning_rate": 3.412649548663007e-06,
+      "loss": 0.6075,
+      "step": 37758
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.6396615505218506,
+      "learning_rate": 3.4123995321816006e-06,
+      "loss": 0.7909,
+      "step": 37759
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.622246503829956,
+      "learning_rate": 3.4121495201147158e-06,
+      "loss": 0.7541,
+      "step": 37760
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.0443499088287354,
+      "learning_rate": 3.411899512463048e-06,
+      "loss": 0.692,
+      "step": 37761
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.4281539916992188,
+      "learning_rate": 3.4116495092272937e-06,
+      "loss": 0.8595,
+      "step": 37762
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.635345697402954,
+      "learning_rate": 3.4113995104081464e-06,
+      "loss": 0.7283,
+      "step": 37763
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.107451915740967,
+      "learning_rate": 3.4111495160063022e-06,
+      "loss": 0.8449,
+      "step": 37764
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 4.193329334259033,
+      "learning_rate": 3.410899526022457e-06,
+      "loss": 0.4247,
+      "step": 37765
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.0034868717193604,
+      "learning_rate": 3.4106495404573048e-06,
+      "loss": 0.5272,
+      "step": 37766
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 4.105732440948486,
+      "learning_rate": 3.41039955931154e-06,
+      "loss": 0.9031,
+      "step": 37767
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.7853429317474365,
+      "learning_rate": 3.4101495825858586e-06,
+      "loss": 0.4558,
+      "step": 37768
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.710280656814575,
+      "learning_rate": 3.409899610280957e-06,
+      "loss": 0.5786,
+      "step": 37769
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.9124951362609863,
+      "learning_rate": 3.409649642397528e-06,
+      "loss": 0.4475,
+      "step": 37770
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.5800423622131348,
+      "learning_rate": 3.4093996789362673e-06,
+      "loss": 0.5062,
+      "step": 37771
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.756556749343872,
+      "learning_rate": 3.4091497198978695e-06,
+      "loss": 0.6139,
+      "step": 37772
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.517608404159546,
+      "learning_rate": 3.4088997652830314e-06,
+      "loss": 1.0044,
+      "step": 37773
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.3139028549194336,
+      "learning_rate": 3.4086498150924467e-06,
+      "loss": 0.6416,
+      "step": 37774
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.3882896900177,
+      "learning_rate": 3.4083998693268113e-06,
+      "loss": 0.6072,
+      "step": 37775
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 3.0134756565093994,
+      "learning_rate": 3.408149927986819e-06,
+      "loss": 0.5917,
+      "step": 37776
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.2518198490142822,
+      "learning_rate": 3.407899991073166e-06,
+      "loss": 0.8498,
+      "step": 37777
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.908906936645508,
+      "learning_rate": 3.407650058586547e-06,
+      "loss": 0.6704,
+      "step": 37778
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.192593812942505,
+      "learning_rate": 3.4074001305276562e-06,
+      "loss": 0.7761,
+      "step": 37779
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.9553723335266113,
+      "learning_rate": 3.4071502068971885e-06,
+      "loss": 0.553,
+      "step": 37780
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.267612934112549,
+      "learning_rate": 3.40690028769584e-06,
+      "loss": 0.4222,
+      "step": 37781
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.9134234189987183,
+      "learning_rate": 3.406650372924305e-06,
+      "loss": 0.53,
+      "step": 37782
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 3.2073984146118164,
+      "learning_rate": 3.406400462583277e-06,
+      "loss": 0.5509,
+      "step": 37783
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.9760801792144775,
+      "learning_rate": 3.4061505566734525e-06,
+      "loss": 0.5795,
+      "step": 37784
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.6593676805496216,
+      "learning_rate": 3.4059006551955276e-06,
+      "loss": 0.3736,
+      "step": 37785
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 3.148566484451294,
+      "learning_rate": 3.4056507581501956e-06,
+      "loss": 0.6942,
+      "step": 37786
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.1558175086975098,
+      "learning_rate": 3.405400865538152e-06,
+      "loss": 0.5828,
+      "step": 37787
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 3.15368390083313,
+      "learning_rate": 3.4051509773600906e-06,
+      "loss": 0.8097,
+      "step": 37788
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.690682888031006,
+      "learning_rate": 3.404901093616707e-06,
+      "loss": 0.8039,
+      "step": 37789
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 3.0880932807922363,
+      "learning_rate": 3.404651214308696e-06,
+      "loss": 0.5552,
+      "step": 37790
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.871455192565918,
+      "learning_rate": 3.4044013394367525e-06,
+      "loss": 0.5591,
+      "step": 37791
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.200594663619995,
+      "learning_rate": 3.404151469001571e-06,
+      "loss": 0.7229,
+      "step": 37792
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.709946393966675,
+      "learning_rate": 3.4039016030038465e-06,
+      "loss": 0.503,
+      "step": 37793
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 3.017503499984741,
+      "learning_rate": 3.4036517414442737e-06,
+      "loss": 0.6273,
+      "step": 37794
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 6.637990474700928,
+      "learning_rate": 3.403401884323547e-06,
+      "loss": 0.5202,
+      "step": 37795
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.9962445497512817,
+      "learning_rate": 3.4031520316423627e-06,
+      "loss": 0.5304,
+      "step": 37796
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 3.0564072132110596,
+      "learning_rate": 3.4029021834014136e-06,
+      "loss": 0.885,
+      "step": 37797
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.7506109476089478,
+      "learning_rate": 3.4026523396013956e-06,
+      "loss": 0.5769,
+      "step": 37798
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.036930561065674,
+      "learning_rate": 3.4024025002430027e-06,
+      "loss": 0.7052,
+      "step": 37799
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 3.383258819580078,
+      "learning_rate": 3.40215266532693e-06,
+      "loss": 0.8536,
+      "step": 37800
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.920028805732727,
+      "learning_rate": 3.4019028348538718e-06,
+      "loss": 0.5319,
+      "step": 37801
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 3.3307723999023438,
+      "learning_rate": 3.4016530088245237e-06,
+      "loss": 0.5531,
+      "step": 37802
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.7799105644226074,
+      "learning_rate": 3.401403187239578e-06,
+      "loss": 0.8194,
+      "step": 37803
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.7969497442245483,
+      "learning_rate": 3.4011533700997324e-06,
+      "loss": 0.5478,
+      "step": 37804
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.8050612211227417,
+      "learning_rate": 3.4009035574056814e-06,
+      "loss": 0.4807,
+      "step": 37805
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.955859422683716,
+      "learning_rate": 3.4006537491581183e-06,
+      "loss": 0.4104,
+      "step": 37806
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 4.390568256378174,
+      "learning_rate": 3.4004039453577375e-06,
+      "loss": 0.6024,
+      "step": 37807
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 3.0192923545837402,
+      "learning_rate": 3.4001541460052338e-06,
+      "loss": 0.8838,
+      "step": 37808
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.2202517986297607,
+      "learning_rate": 3.399904351101303e-06,
+      "loss": 0.509,
+      "step": 37809
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 3.1774237155914307,
+      "learning_rate": 3.399654560646638e-06,
+      "loss": 0.4966,
+      "step": 37810
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.6146748065948486,
+      "learning_rate": 3.3994047746419347e-06,
+      "loss": 0.5623,
+      "step": 37811
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.428109884262085,
+      "learning_rate": 3.3991549930878874e-06,
+      "loss": 0.6605,
+      "step": 37812
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 3.6249608993530273,
+      "learning_rate": 3.3989052159851897e-06,
+      "loss": 0.6023,
+      "step": 37813
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 4.0128254890441895,
+      "learning_rate": 3.3986554433345374e-06,
+      "loss": 0.7002,
+      "step": 37814
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.9034398794174194,
+      "learning_rate": 3.398405675136624e-06,
+      "loss": 0.5732,
+      "step": 37815
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 3.038768768310547,
+      "learning_rate": 3.3981559113921447e-06,
+      "loss": 0.8472,
+      "step": 37816
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 3.168196439743042,
+      "learning_rate": 3.397906152101793e-06,
+      "loss": 0.7004,
+      "step": 37817
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 3.033825159072876,
+      "learning_rate": 3.397656397266265e-06,
+      "loss": 0.6569,
+      "step": 37818
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.309591293334961,
+      "learning_rate": 3.3974066468862544e-06,
+      "loss": 0.6105,
+      "step": 37819
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.8959656953811646,
+      "learning_rate": 3.397156900962455e-06,
+      "loss": 0.5398,
+      "step": 37820
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.3984925746917725,
+      "learning_rate": 3.396907159495562e-06,
+      "loss": 0.7063,
+      "step": 37821
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.6174681186676025,
+      "learning_rate": 3.3966574224862684e-06,
+      "loss": 0.4435,
+      "step": 37822
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.2239973545074463,
+      "learning_rate": 3.3964076899352714e-06,
+      "loss": 0.3769,
+      "step": 37823
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 5.048341274261475,
+      "learning_rate": 3.396157961843263e-06,
+      "loss": 0.8771,
+      "step": 37824
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.332021951675415,
+      "learning_rate": 3.3959082382109387e-06,
+      "loss": 0.6113,
+      "step": 37825
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.3474109172821045,
+      "learning_rate": 3.3956585190389935e-06,
+      "loss": 0.6953,
+      "step": 37826
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.9908967018127441,
+      "learning_rate": 3.395408804328121e-06,
+      "loss": 0.61,
+      "step": 37827
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.3341104984283447,
+      "learning_rate": 3.395159094079015e-06,
+      "loss": 0.6051,
+      "step": 37828
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.6879663467407227,
+      "learning_rate": 3.39490938829237e-06,
+      "loss": 0.5318,
+      "step": 37829
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.5920145511627197,
+      "learning_rate": 3.3946596869688816e-06,
+      "loss": 0.5495,
+      "step": 37830
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.9121003150939941,
+      "learning_rate": 3.3944099901092424e-06,
+      "loss": 0.6693,
+      "step": 37831
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.1254589557647705,
+      "learning_rate": 3.3941602977141486e-06,
+      "loss": 0.5611,
+      "step": 37832
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.847752094268799,
+      "learning_rate": 3.393910609784292e-06,
+      "loss": 0.5287,
+      "step": 37833
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.2349233627319336,
+      "learning_rate": 3.393660926320369e-06,
+      "loss": 0.6553,
+      "step": 37834
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 3.1113057136535645,
+      "learning_rate": 3.393411247323073e-06,
+      "loss": 0.5348,
+      "step": 37835
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.422821283340454,
+      "learning_rate": 3.393161572793099e-06,
+      "loss": 0.5882,
+      "step": 37836
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.1244070529937744,
+      "learning_rate": 3.392911902731141e-06,
+      "loss": 0.7153,
+      "step": 37837
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 3.032593011856079,
+      "learning_rate": 3.3926622371378926e-06,
+      "loss": 0.7232,
+      "step": 37838
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.4781134128570557,
+      "learning_rate": 3.3924125760140483e-06,
+      "loss": 0.3553,
+      "step": 37839
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.599396228790283,
+      "learning_rate": 3.392162919360302e-06,
+      "loss": 0.7976,
+      "step": 37840
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.3094334602355957,
+      "learning_rate": 3.3919132671773493e-06,
+      "loss": 0.6764,
+      "step": 37841
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.9379937648773193,
+      "learning_rate": 3.3916636194658824e-06,
+      "loss": 0.837,
+      "step": 37842
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.1759283542633057,
+      "learning_rate": 3.3914139762265975e-06,
+      "loss": 0.7779,
+      "step": 37843
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.319824695587158,
+      "learning_rate": 3.391164337460186e-06,
+      "loss": 0.6692,
+      "step": 37844
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 3.378070592880249,
+      "learning_rate": 3.390914703167344e-06,
+      "loss": 0.6944,
+      "step": 37845
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.9202419519424438,
+      "learning_rate": 3.390665073348767e-06,
+      "loss": 0.818,
+      "step": 37846
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 3.2370359897613525,
+      "learning_rate": 3.3904154480051465e-06,
+      "loss": 0.3178,
+      "step": 37847
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.2744407653808594,
+      "learning_rate": 3.3901658271371787e-06,
+      "loss": 0.4049,
+      "step": 37848
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.420405626296997,
+      "learning_rate": 3.389916210745555e-06,
+      "loss": 0.8046,
+      "step": 37849
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 3.3374595642089844,
+      "learning_rate": 3.389666598830973e-06,
+      "loss": 0.6601,
+      "step": 37850
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.0853309631347656,
+      "learning_rate": 3.3894169913941244e-06,
+      "loss": 0.477,
+      "step": 37851
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.6501736640930176,
+      "learning_rate": 3.3891673884357035e-06,
+      "loss": 0.6814,
+      "step": 37852
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 3.414310932159424,
+      "learning_rate": 3.388917789956405e-06,
+      "loss": 0.6805,
+      "step": 37853
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.2703709602355957,
+      "learning_rate": 3.388668195956922e-06,
+      "loss": 0.7356,
+      "step": 37854
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.9845662117004395,
+      "learning_rate": 3.38841860643795e-06,
+      "loss": 0.7016,
+      "step": 37855
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.314039707183838,
+      "learning_rate": 3.3881690214001818e-06,
+      "loss": 0.5011,
+      "step": 37856
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.5458664894104004,
+      "learning_rate": 3.387919440844312e-06,
+      "loss": 0.8339,
+      "step": 37857
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.1397721767425537,
+      "learning_rate": 3.387669864771034e-06,
+      "loss": 0.6322,
+      "step": 37858
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.7723946571350098,
+      "learning_rate": 3.3874202931810416e-06,
+      "loss": 0.4214,
+      "step": 37859
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.8858959674835205,
+      "learning_rate": 3.3871707260750296e-06,
+      "loss": 0.7976,
+      "step": 37860
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 4.381120204925537,
+      "learning_rate": 3.386921163453692e-06,
+      "loss": 0.6279,
+      "step": 37861
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.2427639961242676,
+      "learning_rate": 3.3866716053177224e-06,
+      "loss": 0.642,
+      "step": 37862
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.8679102659225464,
+      "learning_rate": 3.3864220516678143e-06,
+      "loss": 0.5255,
+      "step": 37863
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.9740257263183594,
+      "learning_rate": 3.38617250250466e-06,
+      "loss": 0.6743,
+      "step": 37864
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.6871180534362793,
+      "learning_rate": 3.3859229578289575e-06,
+      "loss": 0.4966,
+      "step": 37865
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.1722798347473145,
+      "learning_rate": 3.3856734176413993e-06,
+      "loss": 0.5444,
+      "step": 37866
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.321577310562134,
+      "learning_rate": 3.3854238819426777e-06,
+      "loss": 0.6285,
+      "step": 37867
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.2761378288269043,
+      "learning_rate": 3.385174350733488e-06,
+      "loss": 0.5191,
+      "step": 37868
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.6068434715270996,
+      "learning_rate": 3.3849248240145226e-06,
+      "loss": 0.564,
+      "step": 37869
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 4.314000129699707,
+      "learning_rate": 3.3846753017864773e-06,
+      "loss": 0.5725,
+      "step": 37870
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 3.981861114501953,
+      "learning_rate": 3.384425784050045e-06,
+      "loss": 0.7203,
+      "step": 37871
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.148474931716919,
+      "learning_rate": 3.3841762708059182e-06,
+      "loss": 0.7922,
+      "step": 37872
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.327098846435547,
+      "learning_rate": 3.3839267620547927e-06,
+      "loss": 0.5569,
+      "step": 37873
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.308929443359375,
+      "learning_rate": 3.3836772577973615e-06,
+      "loss": 0.6476,
+      "step": 37874
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 3.592982769012451,
+      "learning_rate": 3.3834277580343177e-06,
+      "loss": 0.6273,
+      "step": 37875
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.9301024675369263,
+      "learning_rate": 3.3831782627663566e-06,
+      "loss": 0.6428,
+      "step": 37876
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.1829071044921875,
+      "learning_rate": 3.3829287719941707e-06,
+      "loss": 0.5084,
+      "step": 37877
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.885972261428833,
+      "learning_rate": 3.382679285718454e-06,
+      "loss": 0.5455,
+      "step": 37878
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 3.820754289627075,
+      "learning_rate": 3.3824298039399006e-06,
+      "loss": 0.5507,
+      "step": 37879
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.785066843032837,
+      "learning_rate": 3.3821803266592042e-06,
+      "loss": 0.4572,
+      "step": 37880
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.6172075271606445,
+      "learning_rate": 3.381930853877058e-06,
+      "loss": 0.6818,
+      "step": 37881
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 3.3407795429229736,
+      "learning_rate": 3.3816813855941554e-06,
+      "loss": 0.6482,
+      "step": 37882
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.5364511013031006,
+      "learning_rate": 3.381431921811191e-06,
+      "loss": 0.4807,
+      "step": 37883
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 3.3752267360687256,
+      "learning_rate": 3.381182462528857e-06,
+      "loss": 0.6267,
+      "step": 37884
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.8399065732955933,
+      "learning_rate": 3.3809330077478484e-06,
+      "loss": 0.6671,
+      "step": 37885
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.521418333053589,
+      "learning_rate": 3.3806835574688595e-06,
+      "loss": 0.4993,
+      "step": 37886
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 3.4160919189453125,
+      "learning_rate": 3.3804341116925833e-06,
+      "loss": 0.9314,
+      "step": 37887
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.8735533952713013,
+      "learning_rate": 3.380184670419713e-06,
+      "loss": 0.6717,
+      "step": 37888
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 3.9427454471588135,
+      "learning_rate": 3.3799352336509417e-06,
+      "loss": 0.5146,
+      "step": 37889
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.73128080368042,
+      "learning_rate": 3.3796858013869645e-06,
+      "loss": 0.8587,
+      "step": 37890
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.6867809295654297,
+      "learning_rate": 3.3794363736284734e-06,
+      "loss": 0.487,
+      "step": 37891
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.15112042427063,
+      "learning_rate": 3.379186950376163e-06,
+      "loss": 0.6015,
+      "step": 37892
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.44653058052063,
+      "learning_rate": 3.3789375316307264e-06,
+      "loss": 0.6285,
+      "step": 37893
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.8418214321136475,
+      "learning_rate": 3.3786881173928576e-06,
+      "loss": 0.4669,
+      "step": 37894
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 5.473125457763672,
+      "learning_rate": 3.3784387076632497e-06,
+      "loss": 0.4728,
+      "step": 37895
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.4611361026763916,
+      "learning_rate": 3.378189302442595e-06,
+      "loss": 0.7523,
+      "step": 37896
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 4.064603805541992,
+      "learning_rate": 3.3779399017315897e-06,
+      "loss": 0.7998,
+      "step": 37897
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.9290252923965454,
+      "learning_rate": 3.377690505530925e-06,
+      "loss": 0.5779,
+      "step": 37898
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.7989171743392944,
+      "learning_rate": 3.3774411138412964e-06,
+      "loss": 0.6254,
+      "step": 37899
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.5257833003997803,
+      "learning_rate": 3.3771917266633957e-06,
+      "loss": 0.6225,
+      "step": 37900
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.9116058349609375,
+      "learning_rate": 3.3769423439979164e-06,
+      "loss": 0.7983,
+      "step": 37901
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.08497953414917,
+      "learning_rate": 3.376692965845554e-06,
+      "loss": 0.6031,
+      "step": 37902
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.5698415040969849,
+      "learning_rate": 3.376443592206999e-06,
+      "loss": 0.4377,
+      "step": 37903
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.7711448669433594,
+      "learning_rate": 3.3761942230829464e-06,
+      "loss": 0.4537,
+      "step": 37904
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.8109548091888428,
+      "learning_rate": 3.375944858474088e-06,
+      "loss": 0.5746,
+      "step": 37905
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.652692198753357,
+      "learning_rate": 3.375695498381121e-06,
+      "loss": 0.4453,
+      "step": 37906
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 3.254406690597534,
+      "learning_rate": 3.3754461428047336e-06,
+      "loss": 0.4869,
+      "step": 37907
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 10.146504402160645,
+      "learning_rate": 3.3751967917456243e-06,
+      "loss": 0.9187,
+      "step": 37908
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.188354253768921,
+      "learning_rate": 3.3749474452044828e-06,
+      "loss": 0.7018,
+      "step": 37909
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.9671969413757324,
+      "learning_rate": 3.3746981031820046e-06,
+      "loss": 0.6444,
+      "step": 37910
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.152235746383667,
+      "learning_rate": 3.3744487656788817e-06,
+      "loss": 0.5796,
+      "step": 37911
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.330061912536621,
+      "learning_rate": 3.3741994326958082e-06,
+      "loss": 0.5121,
+      "step": 37912
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.4144632816314697,
+      "learning_rate": 3.3739501042334767e-06,
+      "loss": 0.6955,
+      "step": 37913
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.342825174331665,
+      "learning_rate": 3.3737007802925815e-06,
+      "loss": 0.6329,
+      "step": 37914
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.3990085124969482,
+      "learning_rate": 3.3734514608738145e-06,
+      "loss": 0.5529,
+      "step": 37915
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 3.2765533924102783,
+      "learning_rate": 3.3732021459778697e-06,
+      "loss": 0.5333,
+      "step": 37916
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 3.0418808460235596,
+      "learning_rate": 3.3729528356054408e-06,
+      "loss": 0.9714,
+      "step": 37917
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 3.002546548843384,
+      "learning_rate": 3.37270352975722e-06,
+      "loss": 0.7426,
+      "step": 37918
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.110023260116577,
+      "learning_rate": 3.3724542284339014e-06,
+      "loss": 0.6107,
+      "step": 37919
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.6710011959075928,
+      "learning_rate": 3.3722049316361775e-06,
+      "loss": 0.7176,
+      "step": 37920
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 3.714172601699829,
+      "learning_rate": 3.3719556393647423e-06,
+      "loss": 0.7978,
+      "step": 37921
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.5061793327331543,
+      "learning_rate": 3.3717063516202887e-06,
+      "loss": 0.6329,
+      "step": 37922
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.7005066871643066,
+      "learning_rate": 3.37145706840351e-06,
+      "loss": 0.601,
+      "step": 37923
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 5.900761127471924,
+      "learning_rate": 3.3712077897150985e-06,
+      "loss": 0.6611,
+      "step": 37924
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.6329092979431152,
+      "learning_rate": 3.3709585155557455e-06,
+      "loss": 0.6822,
+      "step": 37925
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 3.1525495052337646,
+      "learning_rate": 3.3707092459261494e-06,
+      "loss": 0.7267,
+      "step": 37926
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 4.235641002655029,
+      "learning_rate": 3.3704599808270005e-06,
+      "loss": 0.7219,
+      "step": 37927
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.546384811401367,
+      "learning_rate": 3.3702107202589916e-06,
+      "loss": 0.8123,
+      "step": 37928
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.5689340829849243,
+      "learning_rate": 3.369961464222817e-06,
+      "loss": 0.2534,
+      "step": 37929
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.7638370990753174,
+      "learning_rate": 3.369712212719168e-06,
+      "loss": 0.7077,
+      "step": 37930
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 3.037301778793335,
+      "learning_rate": 3.369462965748739e-06,
+      "loss": 0.5782,
+      "step": 37931
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.315319299697876,
+      "learning_rate": 3.3692137233122225e-06,
+      "loss": 0.6455,
+      "step": 37932
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.869602680206299,
+      "learning_rate": 3.368964485410312e-06,
+      "loss": 0.5331,
+      "step": 37933
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 3.6323928833007812,
+      "learning_rate": 3.3687152520437004e-06,
+      "loss": 0.6069,
+      "step": 37934
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.8268420696258545,
+      "learning_rate": 3.3684660232130806e-06,
+      "loss": 0.6493,
+      "step": 37935
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.3940060138702393,
+      "learning_rate": 3.368216798919145e-06,
+      "loss": 0.4379,
+      "step": 37936
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.198779344558716,
+      "learning_rate": 3.3679675791625876e-06,
+      "loss": 0.5834,
+      "step": 37937
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.129289150238037,
+      "learning_rate": 3.3677183639441014e-06,
+      "loss": 0.4957,
+      "step": 37938
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 3.6306896209716797,
+      "learning_rate": 3.367469153264378e-06,
+      "loss": 0.5651,
+      "step": 37939
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.006922721862793,
+      "learning_rate": 3.3672199471241117e-06,
+      "loss": 0.5329,
+      "step": 37940
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.726858377456665,
+      "learning_rate": 3.366970745523995e-06,
+      "loss": 0.7279,
+      "step": 37941
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.9467856884002686,
+      "learning_rate": 3.3667215484647214e-06,
+      "loss": 0.4301,
+      "step": 37942
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.7900304794311523,
+      "learning_rate": 3.366472355946983e-06,
+      "loss": 0.3974,
+      "step": 37943
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 5.165938377380371,
+      "learning_rate": 3.366223167971473e-06,
+      "loss": 0.6651,
+      "step": 37944
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.8583409786224365,
+      "learning_rate": 3.365973984538885e-06,
+      "loss": 0.7547,
+      "step": 37945
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 3.534369945526123,
+      "learning_rate": 3.365724805649908e-06,
+      "loss": 0.5939,
+      "step": 37946
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.5848580598831177,
+      "learning_rate": 3.365475631305241e-06,
+      "loss": 0.3853,
+      "step": 37947
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.4794511795043945,
+      "learning_rate": 3.3652264615055742e-06,
+      "loss": 0.8153,
+      "step": 37948
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.624737024307251,
+      "learning_rate": 3.364977296251599e-06,
+      "loss": 0.7304,
+      "step": 37949
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 4.7287750244140625,
+      "learning_rate": 3.364728135544011e-06,
+      "loss": 0.4908,
+      "step": 37950
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.494613766670227,
+      "learning_rate": 3.3644789793835003e-06,
+      "loss": 0.4702,
+      "step": 37951
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.155841588973999,
+      "learning_rate": 3.364229827770761e-06,
+      "loss": 0.5997,
+      "step": 37952
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.376580238342285,
+      "learning_rate": 3.363980680706486e-06,
+      "loss": 0.5651,
+      "step": 37953
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 3.1096551418304443,
+      "learning_rate": 3.3637315381913684e-06,
+      "loss": 0.5441,
+      "step": 37954
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 3.5888617038726807,
+      "learning_rate": 3.3634824002260995e-06,
+      "loss": 0.6469,
+      "step": 37955
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.74922251701355,
+      "learning_rate": 3.3632332668113733e-06,
+      "loss": 0.4468,
+      "step": 37956
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.4472815990448,
+      "learning_rate": 3.3629841379478813e-06,
+      "loss": 0.8305,
+      "step": 37957
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.6608808040618896,
+      "learning_rate": 3.3627350136363185e-06,
+      "loss": 0.4879,
+      "step": 37958
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 3.106416702270508,
+      "learning_rate": 3.3624858938773757e-06,
+      "loss": 0.5564,
+      "step": 37959
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.420856475830078,
+      "learning_rate": 3.362236778671747e-06,
+      "loss": 0.393,
+      "step": 37960
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.3186707496643066,
+      "learning_rate": 3.3619876680201237e-06,
+      "loss": 0.5269,
+      "step": 37961
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.4298431873321533,
+      "learning_rate": 3.361738561923199e-06,
+      "loss": 0.6396,
+      "step": 37962
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.34318470954895,
+      "learning_rate": 3.361489460381666e-06,
+      "loss": 0.615,
+      "step": 37963
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.0932629108428955,
+      "learning_rate": 3.3612403633962167e-06,
+      "loss": 0.8786,
+      "step": 37964
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 3.3673038482666016,
+      "learning_rate": 3.360991270967544e-06,
+      "loss": 0.4527,
+      "step": 37965
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 4.390915870666504,
+      "learning_rate": 3.36074218309634e-06,
+      "loss": 0.9064,
+      "step": 37966
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 3.7510533332824707,
+      "learning_rate": 3.3604930997832987e-06,
+      "loss": 0.7037,
+      "step": 37967
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.6434950828552246,
+      "learning_rate": 3.360244021029112e-06,
+      "loss": 0.5682,
+      "step": 37968
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.2669949531555176,
+      "learning_rate": 3.359994946834471e-06,
+      "loss": 0.624,
+      "step": 37969
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 3.778944253921509,
+      "learning_rate": 3.3597458772000714e-06,
+      "loss": 0.5539,
+      "step": 37970
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 4.480681419372559,
+      "learning_rate": 3.3594968121266035e-06,
+      "loss": 0.6276,
+      "step": 37971
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.2176287174224854,
+      "learning_rate": 3.3592477516147605e-06,
+      "loss": 0.7746,
+      "step": 37972
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.504009485244751,
+      "learning_rate": 3.3589986956652347e-06,
+      "loss": 0.7978,
+      "step": 37973
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.2963695526123047,
+      "learning_rate": 3.3587496442787194e-06,
+      "loss": 0.5101,
+      "step": 37974
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 3.3061447143554688,
+      "learning_rate": 3.358500597455906e-06,
+      "loss": 0.5721,
+      "step": 37975
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 3.0502922534942627,
+      "learning_rate": 3.358251555197488e-06,
+      "loss": 0.9338,
+      "step": 37976
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 4.278554439544678,
+      "learning_rate": 3.358002517504157e-06,
+      "loss": 0.6549,
+      "step": 37977
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 3.5974457263946533,
+      "learning_rate": 3.3577534843766058e-06,
+      "loss": 0.859,
+      "step": 37978
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.2488794326782227,
+      "learning_rate": 3.3575044558155273e-06,
+      "loss": 0.7545,
+      "step": 37979
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 5.249450206756592,
+      "learning_rate": 3.3572554318216133e-06,
+      "loss": 0.5018,
+      "step": 37980
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.402869939804077,
+      "learning_rate": 3.357006412395557e-06,
+      "loss": 0.9028,
+      "step": 37981
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 3.3583405017852783,
+      "learning_rate": 3.3567573975380494e-06,
+      "loss": 0.5355,
+      "step": 37982
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.760540008544922,
+      "learning_rate": 3.356508387249785e-06,
+      "loss": 0.834,
+      "step": 37983
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.887011170387268,
+      "learning_rate": 3.3562593815314546e-06,
+      "loss": 0.5911,
+      "step": 37984
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 3.5275049209594727,
+      "learning_rate": 3.3560103803837518e-06,
+      "loss": 0.7725,
+      "step": 37985
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.116806745529175,
+      "learning_rate": 3.3557613838073655e-06,
+      "loss": 0.7006,
+      "step": 37986
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.5148811340332031,
+      "learning_rate": 3.3555123918029932e-06,
+      "loss": 0.4507,
+      "step": 37987
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.2639849185943604,
+      "learning_rate": 3.3552634043713245e-06,
+      "loss": 0.4662,
+      "step": 37988
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.8550848960876465,
+      "learning_rate": 3.355014421513052e-06,
+      "loss": 0.7583,
+      "step": 37989
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.6837050914764404,
+      "learning_rate": 3.354765443228869e-06,
+      "loss": 0.3253,
+      "step": 37990
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.0052571296691895,
+      "learning_rate": 3.3545164695194664e-06,
+      "loss": 0.4817,
+      "step": 37991
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.1869916915893555,
+      "learning_rate": 3.3542675003855373e-06,
+      "loss": 0.5963,
+      "step": 37992
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.3064124584198,
+      "learning_rate": 3.354018535827774e-06,
+      "loss": 0.7619,
+      "step": 37993
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.9442124366760254,
+      "learning_rate": 3.353769575846868e-06,
+      "loss": 0.5109,
+      "step": 37994
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.0213706493377686,
+      "learning_rate": 3.3535206204435124e-06,
+      "loss": 0.4362,
+      "step": 37995
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 8.753811836242676,
+      "learning_rate": 3.353271669618399e-06,
+      "loss": 0.6808,
+      "step": 37996
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.127575635910034,
+      "learning_rate": 3.35302272337222e-06,
+      "loss": 0.5858,
+      "step": 37997
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.8420093059539795,
+      "learning_rate": 3.3527737817056686e-06,
+      "loss": 0.6155,
+      "step": 37998
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.8548840284347534,
+      "learning_rate": 3.352524844619436e-06,
+      "loss": 0.6946,
+      "step": 37999
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.635105609893799,
+      "learning_rate": 3.3522759121142147e-06,
+      "loss": 0.6565,
+      "step": 38000
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.6742522716522217,
+      "learning_rate": 3.3520269841906962e-06,
+      "loss": 0.5297,
+      "step": 38001
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.6129109859466553,
+      "learning_rate": 3.3517780608495743e-06,
+      "loss": 0.696,
+      "step": 38002
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.2568445205688477,
+      "learning_rate": 3.3515291420915393e-06,
+      "loss": 0.5668,
+      "step": 38003
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.9822206497192383,
+      "learning_rate": 3.351280227917284e-06,
+      "loss": 0.5158,
+      "step": 38004
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 3.8559350967407227,
+      "learning_rate": 3.3510313183275016e-06,
+      "loss": 0.7467,
+      "step": 38005
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.9165098667144775,
+      "learning_rate": 3.3507824133228833e-06,
+      "loss": 0.4734,
+      "step": 38006
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.8520272970199585,
+      "learning_rate": 3.3505335129041187e-06,
+      "loss": 0.6614,
+      "step": 38007
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.954839825630188,
+      "learning_rate": 3.350284617071905e-06,
+      "loss": 0.5741,
+      "step": 38008
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 3.2605743408203125,
+      "learning_rate": 3.3500357258269316e-06,
+      "loss": 0.5655,
+      "step": 38009
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.9823881387710571,
+      "learning_rate": 3.3497868391698906e-06,
+      "loss": 0.4647,
+      "step": 38010
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.5560896396636963,
+      "learning_rate": 3.3495379571014748e-06,
+      "loss": 0.49,
+      "step": 38011
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.1775786876678467,
+      "learning_rate": 3.3492890796223754e-06,
+      "loss": 0.5073,
+      "step": 38012
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.048988103866577,
+      "learning_rate": 3.3490402067332845e-06,
+      "loss": 0.7645,
+      "step": 38013
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 3.5774059295654297,
+      "learning_rate": 3.348791338434895e-06,
+      "loss": 1.0413,
+      "step": 38014
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 3.077770948410034,
+      "learning_rate": 3.3485424747278973e-06,
+      "loss": 0.5371,
+      "step": 38015
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 3.800370931625366,
+      "learning_rate": 3.348293615612984e-06,
+      "loss": 0.4724,
+      "step": 38016
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 5.257506370544434,
+      "learning_rate": 3.3480447610908485e-06,
+      "loss": 0.5569,
+      "step": 38017
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.417736053466797,
+      "learning_rate": 3.3477959111621814e-06,
+      "loss": 0.4041,
+      "step": 38018
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.8958312273025513,
+      "learning_rate": 3.3475470658276737e-06,
+      "loss": 0.6238,
+      "step": 38019
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.241175413131714,
+      "learning_rate": 3.3472982250880194e-06,
+      "loss": 0.8042,
+      "step": 38020
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.645880103111267,
+      "learning_rate": 3.3470493889439105e-06,
+      "loss": 0.4055,
+      "step": 38021
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.752776622772217,
+      "learning_rate": 3.346800557396037e-06,
+      "loss": 0.4941,
+      "step": 38022
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.4380877017974854,
+      "learning_rate": 3.3465517304450924e-06,
+      "loss": 0.6651,
+      "step": 38023
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.2474594116210938,
+      "learning_rate": 3.346302908091768e-06,
+      "loss": 0.6673,
+      "step": 38024
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.070888042449951,
+      "learning_rate": 3.3460540903367555e-06,
+      "loss": 0.5885,
+      "step": 38025
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 3.452432155609131,
+      "learning_rate": 3.345805277180747e-06,
+      "loss": 0.6212,
+      "step": 38026
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.0200159549713135,
+      "learning_rate": 3.3455564686244334e-06,
+      "loss": 0.7459,
+      "step": 38027
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.8596925735473633,
+      "learning_rate": 3.345307664668509e-06,
+      "loss": 0.6249,
+      "step": 38028
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.086479425430298,
+      "learning_rate": 3.345058865313664e-06,
+      "loss": 0.722,
+      "step": 38029
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.5670995712280273,
+      "learning_rate": 3.34481007056059e-06,
+      "loss": 0.4287,
+      "step": 38030
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 4.41057014465332,
+      "learning_rate": 3.344561280409978e-06,
+      "loss": 0.9662,
+      "step": 38031
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.9190964698791504,
+      "learning_rate": 3.3443124948625223e-06,
+      "loss": 0.6392,
+      "step": 38032
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.9895507097244263,
+      "learning_rate": 3.344063713918913e-06,
+      "loss": 0.7893,
+      "step": 38033
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.278169870376587,
+      "learning_rate": 3.3438149375798423e-06,
+      "loss": 0.6887,
+      "step": 38034
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.786101222038269,
+      "learning_rate": 3.3435661658460017e-06,
+      "loss": 0.44,
+      "step": 38035
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.689387559890747,
+      "learning_rate": 3.3433173987180832e-06,
+      "loss": 0.851,
+      "step": 38036
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.5712568759918213,
+      "learning_rate": 3.343068636196779e-06,
+      "loss": 0.6033,
+      "step": 38037
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.3857297897338867,
+      "learning_rate": 3.3428198782827793e-06,
+      "loss": 0.63,
+      "step": 38038
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.125516891479492,
+      "learning_rate": 3.342571124976777e-06,
+      "loss": 0.8027,
+      "step": 38039
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 3.2009220123291016,
+      "learning_rate": 3.3423223762794637e-06,
+      "loss": 0.6153,
+      "step": 38040
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.6376607418060303,
+      "learning_rate": 3.342073632191531e-06,
+      "loss": 0.379,
+      "step": 38041
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.00886869430542,
+      "learning_rate": 3.3418248927136706e-06,
+      "loss": 0.4993,
+      "step": 38042
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.8911776542663574,
+      "learning_rate": 3.341576157846573e-06,
+      "loss": 0.6604,
+      "step": 38043
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.2723426818847656,
+      "learning_rate": 3.341327427590932e-06,
+      "loss": 0.6337,
+      "step": 38044
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 4.706402778625488,
+      "learning_rate": 3.341078701947438e-06,
+      "loss": 0.7037,
+      "step": 38045
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.7803374528884888,
+      "learning_rate": 3.340829980916782e-06,
+      "loss": 0.5917,
+      "step": 38046
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 3.0852763652801514,
+      "learning_rate": 3.340581264499654e-06,
+      "loss": 0.6663,
+      "step": 38047
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.6745429039001465,
+      "learning_rate": 3.3403325526967506e-06,
+      "loss": 0.6841,
+      "step": 38048
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.9928418397903442,
+      "learning_rate": 3.3400838455087603e-06,
+      "loss": 0.6314,
+      "step": 38049
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 3.207460880279541,
+      "learning_rate": 3.339835142936375e-06,
+      "loss": 0.7761,
+      "step": 38050
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.446578025817871,
+      "learning_rate": 3.3395864449802875e-06,
+      "loss": 0.7974,
+      "step": 38051
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.6927056312561035,
+      "learning_rate": 3.3393377516411863e-06,
+      "loss": 0.531,
+      "step": 38052
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.885878324508667,
+      "learning_rate": 3.339089062919766e-06,
+      "loss": 0.7802,
+      "step": 38053
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.9558682441711426,
+      "learning_rate": 3.3388403788167158e-06,
+      "loss": 0.568,
+      "step": 38054
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.8474810123443604,
+      "learning_rate": 3.3385916993327287e-06,
+      "loss": 0.7241,
+      "step": 38055
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 3.409952402114868,
+      "learning_rate": 3.3383430244684957e-06,
+      "loss": 0.6731,
+      "step": 38056
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.0446317195892334,
+      "learning_rate": 3.3380943542247084e-06,
+      "loss": 0.5695,
+      "step": 38057
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.3360111713409424,
+      "learning_rate": 3.337845688602058e-06,
+      "loss": 0.6712,
+      "step": 38058
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 3.619147539138794,
+      "learning_rate": 3.3375970276012358e-06,
+      "loss": 0.4891,
+      "step": 38059
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.826202392578125,
+      "learning_rate": 3.3373483712229344e-06,
+      "loss": 0.7643,
+      "step": 38060
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.6864768266677856,
+      "learning_rate": 3.3370997194678435e-06,
+      "loss": 0.5913,
+      "step": 38061
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.8763909339904785,
+      "learning_rate": 3.3368510723366555e-06,
+      "loss": 0.6362,
+      "step": 38062
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.8923380374908447,
+      "learning_rate": 3.336602429830062e-06,
+      "loss": 0.8509,
+      "step": 38063
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 3.563279628753662,
+      "learning_rate": 3.3363537919487537e-06,
+      "loss": 0.7635,
+      "step": 38064
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 3.4753775596618652,
+      "learning_rate": 3.3361051586934222e-06,
+      "loss": 0.8466,
+      "step": 38065
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.5458412170410156,
+      "learning_rate": 3.3358565300647592e-06,
+      "loss": 0.5832,
+      "step": 38066
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.0954432487487793,
+      "learning_rate": 3.3356079060634563e-06,
+      "loss": 0.63,
+      "step": 38067
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 4.912705421447754,
+      "learning_rate": 3.3353592866902008e-06,
+      "loss": 0.405,
+      "step": 38068
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.0597832202911377,
+      "learning_rate": 3.3351106719456906e-06,
+      "loss": 0.6023,
+      "step": 38069
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 3.086601734161377,
+      "learning_rate": 3.3348620618306136e-06,
+      "loss": 0.4634,
+      "step": 38070
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.9225023984909058,
+      "learning_rate": 3.3346134563456618e-06,
+      "loss": 0.4394,
+      "step": 38071
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.2139320373535156,
+      "learning_rate": 3.334364855491526e-06,
+      "loss": 0.707,
+      "step": 38072
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.5020785331726074,
+      "learning_rate": 3.3341162592688976e-06,
+      "loss": 0.6087,
+      "step": 38073
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.5040979385375977,
+      "learning_rate": 3.3338676676784676e-06,
+      "loss": 0.6114,
+      "step": 38074
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 3.749260187149048,
+      "learning_rate": 3.333619080720928e-06,
+      "loss": 0.5612,
+      "step": 38075
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.352184772491455,
+      "learning_rate": 3.3333704983969684e-06,
+      "loss": 0.4996,
+      "step": 38076
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.763318657875061,
+      "learning_rate": 3.3331219207072816e-06,
+      "loss": 0.4649,
+      "step": 38077
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.6837804317474365,
+      "learning_rate": 3.3328733476525587e-06,
+      "loss": 0.4805,
+      "step": 38078
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.2664942741394043,
+      "learning_rate": 3.3326247792334904e-06,
+      "loss": 0.5298,
+      "step": 38079
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 4.3150105476379395,
+      "learning_rate": 3.332376215450768e-06,
+      "loss": 0.7053,
+      "step": 38080
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.9521715641021729,
+      "learning_rate": 3.3321276563050816e-06,
+      "loss": 0.516,
+      "step": 38081
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 3.672743797302246,
+      "learning_rate": 3.331879101797124e-06,
+      "loss": 0.7594,
+      "step": 38082
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.4317468404769897,
+      "learning_rate": 3.331630551927586e-06,
+      "loss": 0.3043,
+      "step": 38083
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.218309164047241,
+      "learning_rate": 3.3313820066971587e-06,
+      "loss": 0.6571,
+      "step": 38084
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.3328857421875,
+      "learning_rate": 3.3311334661065333e-06,
+      "loss": 0.6878,
+      "step": 38085
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.0100138187408447,
+      "learning_rate": 3.3308849301563995e-06,
+      "loss": 0.6743,
+      "step": 38086
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.8607666492462158,
+      "learning_rate": 3.33063639884745e-06,
+      "loss": 0.5164,
+      "step": 38087
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.870944619178772,
+      "learning_rate": 3.330387872180374e-06,
+      "loss": 0.6894,
+      "step": 38088
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.7650249004364014,
+      "learning_rate": 3.3301393501558654e-06,
+      "loss": 0.4706,
+      "step": 38089
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.051149845123291,
+      "learning_rate": 3.329890832774613e-06,
+      "loss": 0.6177,
+      "step": 38090
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 3.1652886867523193,
+      "learning_rate": 3.3296423200373097e-06,
+      "loss": 0.6126,
+      "step": 38091
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.195737361907959,
+      "learning_rate": 3.3293938119446433e-06,
+      "loss": 0.6251,
+      "step": 38092
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.651289939880371,
+      "learning_rate": 3.329145308497308e-06,
+      "loss": 0.6698,
+      "step": 38093
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 5.6077094078063965,
+      "learning_rate": 3.328896809695994e-06,
+      "loss": 0.3208,
+      "step": 38094
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.7884786128997803,
+      "learning_rate": 3.328648315541392e-06,
+      "loss": 0.5544,
+      "step": 38095
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.540316104888916,
+      "learning_rate": 3.3283998260341932e-06,
+      "loss": 0.6355,
+      "step": 38096
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 3.3917722702026367,
+      "learning_rate": 3.328151341175088e-06,
+      "loss": 0.7175,
+      "step": 38097
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.2769217491149902,
+      "learning_rate": 3.3279028609647673e-06,
+      "loss": 0.505,
+      "step": 38098
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 4.123773097991943,
+      "learning_rate": 3.3276543854039224e-06,
+      "loss": 0.8019,
+      "step": 38099
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 3.8611960411071777,
+      "learning_rate": 3.327405914493245e-06,
+      "loss": 0.4382,
+      "step": 38100
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.5935781002044678,
+      "learning_rate": 3.3271574482334244e-06,
+      "loss": 0.8337,
+      "step": 38101
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.1117820739746094,
+      "learning_rate": 3.3269089866251523e-06,
+      "loss": 0.6581,
+      "step": 38102
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.0849435329437256,
+      "learning_rate": 3.32666052966912e-06,
+      "loss": 0.5664,
+      "step": 38103
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.596416711807251,
+      "learning_rate": 3.3264120773660175e-06,
+      "loss": 0.4,
+      "step": 38104
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 3.077411651611328,
+      "learning_rate": 3.326163629716536e-06,
+      "loss": 0.4277,
+      "step": 38105
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.6916890144348145,
+      "learning_rate": 3.3259151867213667e-06,
+      "loss": 0.5942,
+      "step": 38106
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.9676635265350342,
+      "learning_rate": 3.3256667483812e-06,
+      "loss": 0.6437,
+      "step": 38107
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.2851343154907227,
+      "learning_rate": 3.325418314696725e-06,
+      "loss": 0.4093,
+      "step": 38108
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.0051958560943604,
+      "learning_rate": 3.3251698856686367e-06,
+      "loss": 0.5117,
+      "step": 38109
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.235612630844116,
+      "learning_rate": 3.3249214612976228e-06,
+      "loss": 0.6009,
+      "step": 38110
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 3.1921286582946777,
+      "learning_rate": 3.3246730415843753e-06,
+      "loss": 0.7369,
+      "step": 38111
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.426952362060547,
+      "learning_rate": 3.324424626529584e-06,
+      "loss": 0.8529,
+      "step": 38112
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.2349774837493896,
+      "learning_rate": 3.324176216133941e-06,
+      "loss": 0.3837,
+      "step": 38113
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 3.977030038833618,
+      "learning_rate": 3.3239278103981355e-06,
+      "loss": 0.5392,
+      "step": 38114
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 6.1331400871276855,
+      "learning_rate": 3.3236794093228587e-06,
+      "loss": 0.409,
+      "step": 38115
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.311459541320801,
+      "learning_rate": 3.323431012908802e-06,
+      "loss": 0.8556,
+      "step": 38116
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.7248194217681885,
+      "learning_rate": 3.3231826211566555e-06,
+      "loss": 0.5591,
+      "step": 38117
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.088371515274048,
+      "learning_rate": 3.3229342340671094e-06,
+      "loss": 0.717,
+      "step": 38118
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.549528121948242,
+      "learning_rate": 3.3226858516408554e-06,
+      "loss": 0.8102,
+      "step": 38119
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.3745675086975098,
+      "learning_rate": 3.3224374738785837e-06,
+      "loss": 0.565,
+      "step": 38120
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 3.494250535964966,
+      "learning_rate": 3.3221891007809847e-06,
+      "loss": 0.4234,
+      "step": 38121
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.0610480308532715,
+      "learning_rate": 3.32194073234875e-06,
+      "loss": 0.6905,
+      "step": 38122
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.5538551807403564,
+      "learning_rate": 3.321692368582569e-06,
+      "loss": 0.5105,
+      "step": 38123
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.2095513343811035,
+      "learning_rate": 3.321444009483133e-06,
+      "loss": 0.6617,
+      "step": 38124
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.904226779937744,
+      "learning_rate": 3.3211956550511325e-06,
+      "loss": 0.3518,
+      "step": 38125
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.40568470954895,
+      "learning_rate": 3.3209473052872575e-06,
+      "loss": 0.66,
+      "step": 38126
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.8359615802764893,
+      "learning_rate": 3.3206989601921997e-06,
+      "loss": 0.7573,
+      "step": 38127
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 5.963071346282959,
+      "learning_rate": 3.3204506197666486e-06,
+      "loss": 0.5521,
+      "step": 38128
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 3.762284278869629,
+      "learning_rate": 3.320202284011293e-06,
+      "loss": 0.6013,
+      "step": 38129
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.7254966497421265,
+      "learning_rate": 3.3199539529268275e-06,
+      "loss": 0.4174,
+      "step": 38130
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.3127195835113525,
+      "learning_rate": 3.3197056265139415e-06,
+      "loss": 0.4898,
+      "step": 38131
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.5615506172180176,
+      "learning_rate": 3.319457304773324e-06,
+      "loss": 0.5958,
+      "step": 38132
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 4.076564788818359,
+      "learning_rate": 3.3192089877056665e-06,
+      "loss": 0.572,
+      "step": 38133
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 3.1878206729888916,
+      "learning_rate": 3.3189606753116587e-06,
+      "loss": 0.7052,
+      "step": 38134
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 3.7943015098571777,
+      "learning_rate": 3.318712367591992e-06,
+      "loss": 0.7449,
+      "step": 38135
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 3.342144250869751,
+      "learning_rate": 3.3184640645473567e-06,
+      "loss": 0.7346,
+      "step": 38136
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 3.320622682571411,
+      "learning_rate": 3.318215766178442e-06,
+      "loss": 0.4866,
+      "step": 38137
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.9940109252929688,
+      "learning_rate": 3.3179674724859403e-06,
+      "loss": 0.8407,
+      "step": 38138
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 4.173562526702881,
+      "learning_rate": 3.31771918347054e-06,
+      "loss": 0.6607,
+      "step": 38139
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.585315465927124,
+      "learning_rate": 3.317470899132933e-06,
+      "loss": 0.5965,
+      "step": 38140
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 3.13093638420105,
+      "learning_rate": 3.317222619473809e-06,
+      "loss": 0.7374,
+      "step": 38141
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.9848793745040894,
+      "learning_rate": 3.3169743444938573e-06,
+      "loss": 0.5327,
+      "step": 38142
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.9380589723587036,
+      "learning_rate": 3.3167260741937705e-06,
+      "loss": 0.719,
+      "step": 38143
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.2041144371032715,
+      "learning_rate": 3.3164778085742383e-06,
+      "loss": 0.7442,
+      "step": 38144
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.5195021629333496,
+      "learning_rate": 3.316229547635951e-06,
+      "loss": 0.4397,
+      "step": 38145
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.975986957550049,
+      "learning_rate": 3.3159812913795975e-06,
+      "loss": 0.6726,
+      "step": 38146
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.9777947664260864,
+      "learning_rate": 3.3157330398058697e-06,
+      "loss": 0.5634,
+      "step": 38147
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.900723934173584,
+      "learning_rate": 3.3154847929154577e-06,
+      "loss": 0.8883,
+      "step": 38148
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.425736665725708,
+      "learning_rate": 3.31523655070905e-06,
+      "loss": 0.7673,
+      "step": 38149
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 3.522141456604004,
+      "learning_rate": 3.3149883131873394e-06,
+      "loss": 0.7359,
+      "step": 38150
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 4.69472599029541,
+      "learning_rate": 3.314740080351015e-06,
+      "loss": 0.3663,
+      "step": 38151
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.4671247005462646,
+      "learning_rate": 3.3144918522007674e-06,
+      "loss": 0.6063,
+      "step": 38152
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.8277332782745361,
+      "learning_rate": 3.3142436287372857e-06,
+      "loss": 0.5185,
+      "step": 38153
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.0202059745788574,
+      "learning_rate": 3.3139954099612614e-06,
+      "loss": 0.6394,
+      "step": 38154
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 3.1889383792877197,
+      "learning_rate": 3.313747195873384e-06,
+      "loss": 0.5995,
+      "step": 38155
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.3273074626922607,
+      "learning_rate": 3.313498986474345e-06,
+      "loss": 0.5515,
+      "step": 38156
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.4658162593841553,
+      "learning_rate": 3.313250781764833e-06,
+      "loss": 0.7649,
+      "step": 38157
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.940488576889038,
+      "learning_rate": 3.3130025817455387e-06,
+      "loss": 0.5982,
+      "step": 38158
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.8251616954803467,
+      "learning_rate": 3.3127543864171516e-06,
+      "loss": 0.6965,
+      "step": 38159
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.386674404144287,
+      "learning_rate": 3.312506195780364e-06,
+      "loss": 0.492,
+      "step": 38160
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.281790018081665,
+      "learning_rate": 3.3122580098358636e-06,
+      "loss": 0.6571,
+      "step": 38161
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.7880609035491943,
+      "learning_rate": 3.3120098285843416e-06,
+      "loss": 0.827,
+      "step": 38162
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.3743340969085693,
+      "learning_rate": 3.311761652026488e-06,
+      "loss": 0.6764,
+      "step": 38163
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.6727731227874756,
+      "learning_rate": 3.311513480162992e-06,
+      "loss": 0.7989,
+      "step": 38164
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.0407838821411133,
+      "learning_rate": 3.311265312994546e-06,
+      "loss": 0.4174,
+      "step": 38165
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.5617640018463135,
+      "learning_rate": 3.3110171505218374e-06,
+      "loss": 0.6933,
+      "step": 38166
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 4.129068374633789,
+      "learning_rate": 3.310768992745558e-06,
+      "loss": 0.5967,
+      "step": 38167
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.577089786529541,
+      "learning_rate": 3.3105208396663966e-06,
+      "loss": 0.6456,
+      "step": 38168
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.357478141784668,
+      "learning_rate": 3.3102726912850423e-06,
+      "loss": 0.4917,
+      "step": 38169
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.172175884246826,
+      "learning_rate": 3.3100245476021886e-06,
+      "loss": 0.4974,
+      "step": 38170
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.7462947368621826,
+      "learning_rate": 3.3097764086185234e-06,
+      "loss": 0.8591,
+      "step": 38171
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.1076724529266357,
+      "learning_rate": 3.309528274334737e-06,
+      "loss": 0.6519,
+      "step": 38172
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.727778434753418,
+      "learning_rate": 3.309280144751519e-06,
+      "loss": 0.4108,
+      "step": 38173
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.5214526653289795,
+      "learning_rate": 3.3090320198695604e-06,
+      "loss": 0.6719,
+      "step": 38174
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.593202590942383,
+      "learning_rate": 3.3087838996895493e-06,
+      "loss": 0.5545,
+      "step": 38175
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.5136311054229736,
+      "learning_rate": 3.3085357842121766e-06,
+      "loss": 0.6331,
+      "step": 38176
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.8461081981658936,
+      "learning_rate": 3.308287673438133e-06,
+      "loss": 0.5302,
+      "step": 38177
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.276876211166382,
+      "learning_rate": 3.308039567368108e-06,
+      "loss": 0.5846,
+      "step": 38178
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.411743640899658,
+      "learning_rate": 3.3077914660027898e-06,
+      "loss": 0.5913,
+      "step": 38179
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.774305582046509,
+      "learning_rate": 3.3075433693428707e-06,
+      "loss": 0.7131,
+      "step": 38180
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.8201003074645996,
+      "learning_rate": 3.3072952773890386e-06,
+      "loss": 0.6915,
+      "step": 38181
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 5.663247585296631,
+      "learning_rate": 3.3070471901419845e-06,
+      "loss": 0.7307,
+      "step": 38182
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.9928287267684937,
+      "learning_rate": 3.306799107602398e-06,
+      "loss": 0.5517,
+      "step": 38183
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.8980129957199097,
+      "learning_rate": 3.306551029770969e-06,
+      "loss": 0.6142,
+      "step": 38184
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.250560760498047,
+      "learning_rate": 3.3063029566483873e-06,
+      "loss": 0.6471,
+      "step": 38185
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.2567996978759766,
+      "learning_rate": 3.306054888235343e-06,
+      "loss": 0.6102,
+      "step": 38186
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.9887274503707886,
+      "learning_rate": 3.305806824532525e-06,
+      "loss": 0.6364,
+      "step": 38187
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.3911967277526855,
+      "learning_rate": 3.305558765540623e-06,
+      "loss": 0.7967,
+      "step": 38188
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.314427375793457,
+      "learning_rate": 3.305310711260328e-06,
+      "loss": 0.6083,
+      "step": 38189
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 4.796970367431641,
+      "learning_rate": 3.3050626616923265e-06,
+      "loss": 0.7099,
+      "step": 38190
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.906484603881836,
+      "learning_rate": 3.3048146168373134e-06,
+      "loss": 0.4457,
+      "step": 38191
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.5320193767547607,
+      "learning_rate": 3.3045665766959756e-06,
+      "loss": 0.5443,
+      "step": 38192
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.592280864715576,
+      "learning_rate": 3.3043185412690038e-06,
+      "loss": 0.8999,
+      "step": 38193
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.8718122243881226,
+      "learning_rate": 3.3040705105570858e-06,
+      "loss": 0.6759,
+      "step": 38194
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.690004825592041,
+      "learning_rate": 3.303822484560913e-06,
+      "loss": 0.6463,
+      "step": 38195
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.9211502075195312,
+      "learning_rate": 3.303574463281174e-06,
+      "loss": 0.5341,
+      "step": 38196
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 3.1634180545806885,
+      "learning_rate": 3.3033264467185594e-06,
+      "loss": 0.4932,
+      "step": 38197
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 3.5212676525115967,
+      "learning_rate": 3.303078434873758e-06,
+      "loss": 0.7298,
+      "step": 38198
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 3.0284738540649414,
+      "learning_rate": 3.30283042774746e-06,
+      "loss": 0.5166,
+      "step": 38199
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 3.100641965866089,
+      "learning_rate": 3.302582425340355e-06,
+      "loss": 0.3439,
+      "step": 38200
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 5.073729515075684,
+      "learning_rate": 3.302334427653132e-06,
+      "loss": 0.7362,
+      "step": 38201
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.675708055496216,
+      "learning_rate": 3.302086434686481e-06,
+      "loss": 0.714,
+      "step": 38202
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.4825994968414307,
+      "learning_rate": 3.3018384464410918e-06,
+      "loss": 0.6671,
+      "step": 38203
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.3367373943328857,
+      "learning_rate": 3.3015904629176533e-06,
+      "loss": 0.3232,
+      "step": 38204
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.3546931743621826,
+      "learning_rate": 3.301342484116855e-06,
+      "loss": 0.631,
+      "step": 38205
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 3.5114455223083496,
+      "learning_rate": 3.301094510039388e-06,
+      "loss": 0.6397,
+      "step": 38206
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.0320191383361816,
+      "learning_rate": 3.300846540685941e-06,
+      "loss": 0.7125,
+      "step": 38207
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.451913833618164,
+      "learning_rate": 3.3005985760572024e-06,
+      "loss": 0.918,
+      "step": 38208
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 3.4827888011932373,
+      "learning_rate": 3.3003506161538633e-06,
+      "loss": 0.6206,
+      "step": 38209
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.9498995542526245,
+      "learning_rate": 3.3001026609766106e-06,
+      "loss": 0.498,
+      "step": 38210
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.8191115856170654,
+      "learning_rate": 3.299854710526137e-06,
+      "loss": 0.5734,
+      "step": 38211
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.5229580402374268,
+      "learning_rate": 3.2996067648031305e-06,
+      "loss": 0.5398,
+      "step": 38212
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.38718843460083,
+      "learning_rate": 3.2993588238082807e-06,
+      "loss": 0.6279,
+      "step": 38213
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.5179998874664307,
+      "learning_rate": 3.2991108875422774e-06,
+      "loss": 0.7302,
+      "step": 38214
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.080165386199951,
+      "learning_rate": 3.2988629560058096e-06,
+      "loss": 0.678,
+      "step": 38215
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.379300355911255,
+      "learning_rate": 3.2986150291995646e-06,
+      "loss": 0.6084,
+      "step": 38216
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.6071369647979736,
+      "learning_rate": 3.298367107124235e-06,
+      "loss": 0.3735,
+      "step": 38217
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.3568978309631348,
+      "learning_rate": 3.29811918978051e-06,
+      "loss": 0.7331,
+      "step": 38218
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.6656923294067383,
+      "learning_rate": 3.297871277169077e-06,
+      "loss": 0.5092,
+      "step": 38219
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.976576805114746,
+      "learning_rate": 3.297623369290627e-06,
+      "loss": 0.6711,
+      "step": 38220
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.4131040573120117,
+      "learning_rate": 3.297375466145849e-06,
+      "loss": 0.5207,
+      "step": 38221
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 3.636596441268921,
+      "learning_rate": 3.2971275677354307e-06,
+      "loss": 0.5684,
+      "step": 38222
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.2825591564178467,
+      "learning_rate": 3.296879674060064e-06,
+      "loss": 0.6114,
+      "step": 38223
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.218205451965332,
+      "learning_rate": 3.2966317851204366e-06,
+      "loss": 0.8598,
+      "step": 38224
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.1194024085998535,
+      "learning_rate": 3.2963839009172375e-06,
+      "loss": 0.6549,
+      "step": 38225
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.7573134899139404,
+      "learning_rate": 3.2961360214511575e-06,
+      "loss": 0.3255,
+      "step": 38226
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 5.299350738525391,
+      "learning_rate": 3.2958881467228842e-06,
+      "loss": 0.7989,
+      "step": 38227
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.4328994750976562,
+      "learning_rate": 3.295640276733108e-06,
+      "loss": 0.4942,
+      "step": 38228
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.25534987449646,
+      "learning_rate": 3.2953924114825173e-06,
+      "loss": 0.6442,
+      "step": 38229
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.522576093673706,
+      "learning_rate": 3.2951445509718005e-06,
+      "loss": 0.8423,
+      "step": 38230
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 4.069957256317139,
+      "learning_rate": 3.2948966952016493e-06,
+      "loss": 0.6974,
+      "step": 38231
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 4.83017110824585,
+      "learning_rate": 3.294648844172752e-06,
+      "loss": 0.762,
+      "step": 38232
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.086578130722046,
+      "learning_rate": 3.2944009978857973e-06,
+      "loss": 0.5462,
+      "step": 38233
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.0242185592651367,
+      "learning_rate": 3.2941531563414746e-06,
+      "loss": 0.3619,
+      "step": 38234
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.6391866207122803,
+      "learning_rate": 3.293905319540473e-06,
+      "loss": 0.4351,
+      "step": 38235
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 6.869139671325684,
+      "learning_rate": 3.2936574874834813e-06,
+      "loss": 0.7107,
+      "step": 38236
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.8537752628326416,
+      "learning_rate": 3.293409660171189e-06,
+      "loss": 0.7597,
+      "step": 38237
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.3276216983795166,
+      "learning_rate": 3.293161837604285e-06,
+      "loss": 0.5958,
+      "step": 38238
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.311807870864868,
+      "learning_rate": 3.2929140197834593e-06,
+      "loss": 0.7471,
+      "step": 38239
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 3.9414873123168945,
+      "learning_rate": 3.2926662067093997e-06,
+      "loss": 0.6024,
+      "step": 38240
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.8945196866989136,
+      "learning_rate": 3.292418398382796e-06,
+      "loss": 0.6906,
+      "step": 38241
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.7275042533874512,
+      "learning_rate": 3.2921705948043364e-06,
+      "loss": 0.4657,
+      "step": 38242
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.2599503993988037,
+      "learning_rate": 3.2919227959747113e-06,
+      "loss": 0.7817,
+      "step": 38243
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 3.1354265213012695,
+      "learning_rate": 3.2916750018946087e-06,
+      "loss": 0.5816,
+      "step": 38244
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.2290663719177246,
+      "learning_rate": 3.2914272125647177e-06,
+      "loss": 0.57,
+      "step": 38245
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.631654977798462,
+      "learning_rate": 3.2911794279857278e-06,
+      "loss": 0.8368,
+      "step": 38246
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.353036642074585,
+      "learning_rate": 3.290931648158329e-06,
+      "loss": 0.7546,
+      "step": 38247
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.6085991859436035,
+      "learning_rate": 3.290683873083207e-06,
+      "loss": 0.6192,
+      "step": 38248
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.7819690704345703,
+      "learning_rate": 3.290436102761054e-06,
+      "loss": 0.5442,
+      "step": 38249
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.291837453842163,
+      "learning_rate": 3.290188337192558e-06,
+      "loss": 0.4274,
+      "step": 38250
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 9.225150108337402,
+      "learning_rate": 3.2899405763784046e-06,
+      "loss": 0.6402,
+      "step": 38251
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.8869532346725464,
+      "learning_rate": 3.289692820319289e-06,
+      "loss": 0.5494,
+      "step": 38252
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.9351874589920044,
+      "learning_rate": 3.2894450690158964e-06,
+      "loss": 0.5586,
+      "step": 38253
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 3.0519254207611084,
+      "learning_rate": 3.2891973224689166e-06,
+      "loss": 0.4967,
+      "step": 38254
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.578486442565918,
+      "learning_rate": 3.2889495806790378e-06,
+      "loss": 0.6164,
+      "step": 38255
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.936967372894287,
+      "learning_rate": 3.2887018436469493e-06,
+      "loss": 0.6449,
+      "step": 38256
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.101452350616455,
+      "learning_rate": 3.2884541113733404e-06,
+      "loss": 0.6665,
+      "step": 38257
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 3.4352006912231445,
+      "learning_rate": 3.2882063838588983e-06,
+      "loss": 0.5628,
+      "step": 38258
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 4.622707366943359,
+      "learning_rate": 3.287958661104314e-06,
+      "loss": 0.6763,
+      "step": 38259
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.532654285430908,
+      "learning_rate": 3.287710943110275e-06,
+      "loss": 0.7627,
+      "step": 38260
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.031217098236084,
+      "learning_rate": 3.2874632298774713e-06,
+      "loss": 0.4751,
+      "step": 38261
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.8779888153076172,
+      "learning_rate": 3.2872155214065893e-06,
+      "loss": 0.6095,
+      "step": 38262
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 3.0546226501464844,
+      "learning_rate": 3.28696781769832e-06,
+      "loss": 0.6594,
+      "step": 38263
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.4934821128845215,
+      "learning_rate": 3.2867201187533517e-06,
+      "loss": 0.5253,
+      "step": 38264
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.123725414276123,
+      "learning_rate": 3.286472424572373e-06,
+      "loss": 0.6544,
+      "step": 38265
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.1144118309020996,
+      "learning_rate": 3.286224735156071e-06,
+      "loss": 0.442,
+      "step": 38266
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.1129331588745117,
+      "learning_rate": 3.2859770505051373e-06,
+      "loss": 0.5707,
+      "step": 38267
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 3.122297525405884,
+      "learning_rate": 3.28572937062026e-06,
+      "loss": 0.6885,
+      "step": 38268
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.6849937438964844,
+      "learning_rate": 3.285481695502126e-06,
+      "loss": 0.5317,
+      "step": 38269
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 5.26591157913208,
+      "learning_rate": 3.2852340251514258e-06,
+      "loss": 0.6359,
+      "step": 38270
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.9445128440856934,
+      "learning_rate": 3.2849863595688462e-06,
+      "loss": 0.7429,
+      "step": 38271
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.5434532165527344,
+      "learning_rate": 3.2847386987550783e-06,
+      "loss": 0.58,
+      "step": 38272
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.924072504043579,
+      "learning_rate": 3.284491042710809e-06,
+      "loss": 0.6317,
+      "step": 38273
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.5958869457244873,
+      "learning_rate": 3.284243391436728e-06,
+      "loss": 0.4012,
+      "step": 38274
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.5937628746032715,
+      "learning_rate": 3.2839957449335235e-06,
+      "loss": 0.7112,
+      "step": 38275
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 4.785977363586426,
+      "learning_rate": 3.2837481032018836e-06,
+      "loss": 0.7454,
+      "step": 38276
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.034693479537964,
+      "learning_rate": 3.2835004662424964e-06,
+      "loss": 0.5036,
+      "step": 38277
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 3.031010150909424,
+      "learning_rate": 3.283252834056052e-06,
+      "loss": 0.5974,
+      "step": 38278
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.5164170265197754,
+      "learning_rate": 3.2830052066432396e-06,
+      "loss": 0.6547,
+      "step": 38279
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.8661365509033203,
+      "learning_rate": 3.2827575840047455e-06,
+      "loss": 0.7705,
+      "step": 38280
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.128222703933716,
+      "learning_rate": 3.2825099661412595e-06,
+      "loss": 0.6358,
+      "step": 38281
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.230403184890747,
+      "learning_rate": 3.28226235305347e-06,
+      "loss": 0.7234,
+      "step": 38282
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 3.715045928955078,
+      "learning_rate": 3.2820147447420646e-06,
+      "loss": 0.9158,
+      "step": 38283
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.5940847396850586,
+      "learning_rate": 3.2817671412077333e-06,
+      "loss": 0.4883,
+      "step": 38284
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 3.14534854888916,
+      "learning_rate": 3.2815195424511647e-06,
+      "loss": 0.7336,
+      "step": 38285
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.0913941860198975,
+      "learning_rate": 3.2812719484730456e-06,
+      "loss": 0.8022,
+      "step": 38286
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 3.2373249530792236,
+      "learning_rate": 3.2810243592740655e-06,
+      "loss": 0.5299,
+      "step": 38287
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.428506851196289,
+      "learning_rate": 3.2807767748549123e-06,
+      "loss": 0.6728,
+      "step": 38288
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.109931707382202,
+      "learning_rate": 3.2805291952162754e-06,
+      "loss": 0.5074,
+      "step": 38289
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.5268962383270264,
+      "learning_rate": 3.280281620358843e-06,
+      "loss": 0.6912,
+      "step": 38290
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.837693452835083,
+      "learning_rate": 3.2800340502833008e-06,
+      "loss": 0.583,
+      "step": 38291
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 4.336659908294678,
+      "learning_rate": 3.279786484990341e-06,
+      "loss": 0.5612,
+      "step": 38292
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 3.394063949584961,
+      "learning_rate": 3.279538924480651e-06,
+      "loss": 0.5365,
+      "step": 38293
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.9902366399765015,
+      "learning_rate": 3.2792913687549187e-06,
+      "loss": 0.5411,
+      "step": 38294
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 4.011673927307129,
+      "learning_rate": 3.279043817813833e-06,
+      "loss": 0.664,
+      "step": 38295
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.1633660793304443,
+      "learning_rate": 3.278796271658081e-06,
+      "loss": 0.683,
+      "step": 38296
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.2432138919830322,
+      "learning_rate": 3.2785487302883524e-06,
+      "loss": 0.4176,
+      "step": 38297
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.9332892894744873,
+      "learning_rate": 3.278301193705335e-06,
+      "loss": 0.7375,
+      "step": 38298
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.6015595197677612,
+      "learning_rate": 3.278053661909717e-06,
+      "loss": 0.5806,
+      "step": 38299
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.8768304586410522,
+      "learning_rate": 3.277806134902186e-06,
+      "loss": 0.6276,
+      "step": 38300
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.9480314254760742,
+      "learning_rate": 3.2775586126834315e-06,
+      "loss": 0.3221,
+      "step": 38301
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.094647169113159,
+      "learning_rate": 3.277311095254141e-06,
+      "loss": 0.2194,
+      "step": 38302
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 3.739603042602539,
+      "learning_rate": 3.2770635826150033e-06,
+      "loss": 0.665,
+      "step": 38303
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.8649917840957642,
+      "learning_rate": 3.2768160747667065e-06,
+      "loss": 0.3828,
+      "step": 38304
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.806946039199829,
+      "learning_rate": 3.2765685717099383e-06,
+      "loss": 0.4473,
+      "step": 38305
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.552593469619751,
+      "learning_rate": 3.2763210734453877e-06,
+      "loss": 0.6425,
+      "step": 38306
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.8001158237457275,
+      "learning_rate": 3.276073579973742e-06,
+      "loss": 0.5046,
+      "step": 38307
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.9388999938964844,
+      "learning_rate": 3.2758260912956902e-06,
+      "loss": 0.3503,
+      "step": 38308
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.5414419174194336,
+      "learning_rate": 3.27557860741192e-06,
+      "loss": 0.5673,
+      "step": 38309
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 3.984999895095825,
+      "learning_rate": 3.27533112832312e-06,
+      "loss": 0.626,
+      "step": 38310
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 3.269493818283081,
+      "learning_rate": 3.2750836540299773e-06,
+      "loss": 0.7693,
+      "step": 38311
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 3.677612543106079,
+      "learning_rate": 3.2748361845331792e-06,
+      "loss": 0.5525,
+      "step": 38312
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.4561262130737305,
+      "learning_rate": 3.274588719833418e-06,
+      "loss": 0.5764,
+      "step": 38313
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.671987771987915,
+      "learning_rate": 3.2743412599313784e-06,
+      "loss": 0.5669,
+      "step": 38314
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.158475637435913,
+      "learning_rate": 3.2740938048277495e-06,
+      "loss": 0.633,
+      "step": 38315
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.979676365852356,
+      "learning_rate": 3.2738463545232196e-06,
+      "loss": 0.703,
+      "step": 38316
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.6504247188568115,
+      "learning_rate": 3.273598909018476e-06,
+      "loss": 0.4554,
+      "step": 38317
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.1004762649536133,
+      "learning_rate": 3.2733514683142077e-06,
+      "loss": 0.4986,
+      "step": 38318
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.9129374027252197,
+      "learning_rate": 3.273104032411101e-06,
+      "loss": 0.581,
+      "step": 38319
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.555962324142456,
+      "learning_rate": 3.272856601309846e-06,
+      "loss": 0.6618,
+      "step": 38320
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.247077226638794,
+      "learning_rate": 3.2726091750111293e-06,
+      "loss": 0.6272,
+      "step": 38321
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.6862258911132812,
+      "learning_rate": 3.272361753515639e-06,
+      "loss": 0.7429,
+      "step": 38322
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.234346628189087,
+      "learning_rate": 3.2721143368240647e-06,
+      "loss": 0.5899,
+      "step": 38323
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.0088350772857666,
+      "learning_rate": 3.2718669249370927e-06,
+      "loss": 0.488,
+      "step": 38324
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.9594508409500122,
+      "learning_rate": 3.271619517855411e-06,
+      "loss": 0.3915,
+      "step": 38325
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.6290063858032227,
+      "learning_rate": 3.2713721155797084e-06,
+      "loss": 0.5837,
+      "step": 38326
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.6970694065093994,
+      "learning_rate": 3.271124718110672e-06,
+      "loss": 0.4645,
+      "step": 38327
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 3.1875457763671875,
+      "learning_rate": 3.270877325448989e-06,
+      "loss": 0.7813,
+      "step": 38328
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.547149419784546,
+      "learning_rate": 3.27062993759535e-06,
+      "loss": 0.5506,
+      "step": 38329
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.584303855895996,
+      "learning_rate": 3.2703825545504406e-06,
+      "loss": 0.7017,
+      "step": 38330
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.379179835319519,
+      "learning_rate": 3.27013517631495e-06,
+      "loss": 0.2752,
+      "step": 38331
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.0801076889038086,
+      "learning_rate": 3.2698878028895644e-06,
+      "loss": 0.9019,
+      "step": 38332
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 3.184882879257202,
+      "learning_rate": 3.269640434274973e-06,
+      "loss": 0.7086,
+      "step": 38333
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.362771987915039,
+      "learning_rate": 3.2693930704718647e-06,
+      "loss": 0.515,
+      "step": 38334
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.252936601638794,
+      "learning_rate": 3.269145711480924e-06,
+      "loss": 0.5907,
+      "step": 38335
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.8815550804138184,
+      "learning_rate": 3.2688983573028427e-06,
+      "loss": 0.416,
+      "step": 38336
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 3.053365468978882,
+      "learning_rate": 3.2686510079383054e-06,
+      "loss": 0.6398,
+      "step": 38337
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.078989267349243,
+      "learning_rate": 3.268403663388001e-06,
+      "loss": 0.498,
+      "step": 38338
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.5065131187438965,
+      "learning_rate": 3.2681563236526172e-06,
+      "loss": 0.8373,
+      "step": 38339
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.391981840133667,
+      "learning_rate": 3.2679089887328424e-06,
+      "loss": 0.5751,
+      "step": 38340
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 3.006744384765625,
+      "learning_rate": 3.2676616586293642e-06,
+      "loss": 0.6969,
+      "step": 38341
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.144759178161621,
+      "learning_rate": 3.2674143333428697e-06,
+      "loss": 0.6052,
+      "step": 38342
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.599475860595703,
+      "learning_rate": 3.267167012874046e-06,
+      "loss": 0.7549,
+      "step": 38343
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.61191725730896,
+      "learning_rate": 3.266919697223583e-06,
+      "loss": 0.5911,
+      "step": 38344
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.725114583969116,
+      "learning_rate": 3.266672386392167e-06,
+      "loss": 0.5707,
+      "step": 38345
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.692716360092163,
+      "learning_rate": 3.266425080380485e-06,
+      "loss": 0.8662,
+      "step": 38346
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.2775609493255615,
+      "learning_rate": 3.2661777791892256e-06,
+      "loss": 0.7486,
+      "step": 38347
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.8626281023025513,
+      "learning_rate": 3.2659304828190772e-06,
+      "loss": 0.645,
+      "step": 38348
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 5.93283748626709,
+      "learning_rate": 3.2656831912707253e-06,
+      "loss": 0.5985,
+      "step": 38349
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.981210470199585,
+      "learning_rate": 3.2654359045448598e-06,
+      "loss": 0.4336,
+      "step": 38350
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.255218505859375,
+      "learning_rate": 3.265188622642167e-06,
+      "loss": 0.8428,
+      "step": 38351
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.5930991172790527,
+      "learning_rate": 3.264941345563332e-06,
+      "loss": 0.6354,
+      "step": 38352
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.1187081336975098,
+      "learning_rate": 3.2646940733090483e-06,
+      "loss": 0.5935,
+      "step": 38353
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.7516056299209595,
+      "learning_rate": 3.2644468058799994e-06,
+      "loss": 0.2807,
+      "step": 38354
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 3.130277633666992,
+      "learning_rate": 3.264199543276874e-06,
+      "loss": 0.5482,
+      "step": 38355
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.9438313245773315,
+      "learning_rate": 3.2639522855003598e-06,
+      "loss": 0.418,
+      "step": 38356
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.7938148975372314,
+      "learning_rate": 3.2637050325511434e-06,
+      "loss": 0.4881,
+      "step": 38357
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.096677303314209,
+      "learning_rate": 3.263457784429913e-06,
+      "loss": 0.8468,
+      "step": 38358
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.943732976913452,
+      "learning_rate": 3.2632105411373565e-06,
+      "loss": 0.6529,
+      "step": 38359
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.6239912509918213,
+      "learning_rate": 3.2629633026741603e-06,
+      "loss": 0.669,
+      "step": 38360
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.450406074523926,
+      "learning_rate": 3.262716069041013e-06,
+      "loss": 0.4661,
+      "step": 38361
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.54067325592041,
+      "learning_rate": 3.2624688402386013e-06,
+      "loss": 0.9264,
+      "step": 38362
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.2368955612182617,
+      "learning_rate": 3.2622216162676124e-06,
+      "loss": 0.6127,
+      "step": 38363
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.6816534996032715,
+      "learning_rate": 3.261974397128735e-06,
+      "loss": 0.7755,
+      "step": 38364
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 4.742109298706055,
+      "learning_rate": 3.2617271828226547e-06,
+      "loss": 0.6117,
+      "step": 38365
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.9094542264938354,
+      "learning_rate": 3.2614799733500606e-06,
+      "loss": 0.4348,
+      "step": 38366
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.434682846069336,
+      "learning_rate": 3.2612327687116395e-06,
+      "loss": 0.6552,
+      "step": 38367
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 4.809221267700195,
+      "learning_rate": 3.260985568908078e-06,
+      "loss": 0.464,
+      "step": 38368
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.5536587238311768,
+      "learning_rate": 3.2607383739400657e-06,
+      "loss": 0.5642,
+      "step": 38369
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.0182559490203857,
+      "learning_rate": 3.260491183808287e-06,
+      "loss": 0.4758,
+      "step": 38370
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.154268503189087,
+      "learning_rate": 3.2602439985134314e-06,
+      "loss": 0.5235,
+      "step": 38371
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.2416858673095703,
+      "learning_rate": 3.2599968180561848e-06,
+      "loss": 0.6965,
+      "step": 38372
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.1173253059387207,
+      "learning_rate": 3.259749642437234e-06,
+      "loss": 0.5658,
+      "step": 38373
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.8580503463745117,
+      "learning_rate": 3.259502471657269e-06,
+      "loss": 0.3772,
+      "step": 38374
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.193028211593628,
+      "learning_rate": 3.2592553057169764e-06,
+      "loss": 0.7848,
+      "step": 38375
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 4.003727436065674,
+      "learning_rate": 3.259008144617042e-06,
+      "loss": 0.6409,
+      "step": 38376
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.5912785530090332,
+      "learning_rate": 3.258760988358154e-06,
+      "loss": 0.4315,
+      "step": 38377
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.473172426223755,
+      "learning_rate": 3.2585138369409986e-06,
+      "loss": 0.6733,
+      "step": 38378
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.571563720703125,
+      "learning_rate": 3.2582666903662654e-06,
+      "loss": 0.9051,
+      "step": 38379
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.2786736488342285,
+      "learning_rate": 3.2580195486346387e-06,
+      "loss": 0.6904,
+      "step": 38380
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 4.737217903137207,
+      "learning_rate": 3.2577724117468077e-06,
+      "loss": 0.6358,
+      "step": 38381
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.178481101989746,
+      "learning_rate": 3.2575252797034584e-06,
+      "loss": 0.5974,
+      "step": 38382
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 3.055011034011841,
+      "learning_rate": 3.2572781525052794e-06,
+      "loss": 0.7003,
+      "step": 38383
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.460122585296631,
+      "learning_rate": 3.2570310301529564e-06,
+      "loss": 0.5385,
+      "step": 38384
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 3.139119863510132,
+      "learning_rate": 3.256783912647178e-06,
+      "loss": 0.6454,
+      "step": 38385
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.099993944168091,
+      "learning_rate": 3.256536799988629e-06,
+      "loss": 0.725,
+      "step": 38386
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.744925856590271,
+      "learning_rate": 3.2562896921779997e-06,
+      "loss": 0.6047,
+      "step": 38387
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.2394211292266846,
+      "learning_rate": 3.256042589215974e-06,
+      "loss": 0.8497,
+      "step": 38388
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.8786628246307373,
+      "learning_rate": 3.2557954911032407e-06,
+      "loss": 0.6618,
+      "step": 38389
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.2954089641571045,
+      "learning_rate": 3.2555483978404876e-06,
+      "loss": 0.62,
+      "step": 38390
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.5188169479370117,
+      "learning_rate": 3.255301309428401e-06,
+      "loss": 0.8058,
+      "step": 38391
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.159925699234009,
+      "learning_rate": 3.2550542258676678e-06,
+      "loss": 0.7772,
+      "step": 38392
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.9625115394592285,
+      "learning_rate": 3.2548071471589736e-06,
+      "loss": 0.5621,
+      "step": 38393
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.7617924213409424,
+      "learning_rate": 3.254560073303008e-06,
+      "loss": 0.3482,
+      "step": 38394
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.155303478240967,
+      "learning_rate": 3.254313004300458e-06,
+      "loss": 0.6213,
+      "step": 38395
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.3159546852111816,
+      "learning_rate": 3.254065940152008e-06,
+      "loss": 0.4295,
+      "step": 38396
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.3749475479125977,
+      "learning_rate": 3.2538188808583476e-06,
+      "loss": 0.5192,
+      "step": 38397
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.2083911895751953,
+      "learning_rate": 3.253571826420163e-06,
+      "loss": 0.5494,
+      "step": 38398
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 5.269611835479736,
+      "learning_rate": 3.2533247768381403e-06,
+      "loss": 0.6756,
+      "step": 38399
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.3399219512939453,
+      "learning_rate": 3.253077732112967e-06,
+      "loss": 0.5831,
+      "step": 38400
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.9938666820526123,
+      "learning_rate": 3.2528306922453295e-06,
+      "loss": 0.6802,
+      "step": 38401
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.4180028438568115,
+      "learning_rate": 3.2525836572359164e-06,
+      "loss": 0.6348,
+      "step": 38402
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.3079402446746826,
+      "learning_rate": 3.2523366270854133e-06,
+      "loss": 0.6504,
+      "step": 38403
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.6750857830047607,
+      "learning_rate": 3.2520896017945076e-06,
+      "loss": 0.5335,
+      "step": 38404
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.5781371593475342,
+      "learning_rate": 3.2518425813638855e-06,
+      "loss": 0.3552,
+      "step": 38405
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.2241413593292236,
+      "learning_rate": 3.251595565794235e-06,
+      "loss": 0.8288,
+      "step": 38406
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.270282745361328,
+      "learning_rate": 3.2513485550862417e-06,
+      "loss": 0.7693,
+      "step": 38407
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.8559850454330444,
+      "learning_rate": 3.251101549240593e-06,
+      "loss": 0.4803,
+      "step": 38408
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.6882359981536865,
+      "learning_rate": 3.250854548257976e-06,
+      "loss": 0.8563,
+      "step": 38409
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.8259117603302,
+      "learning_rate": 3.2506075521390774e-06,
+      "loss": 0.6598,
+      "step": 38410
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.987663984298706,
+      "learning_rate": 3.2503605608845835e-06,
+      "loss": 0.6027,
+      "step": 38411
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.7900463342666626,
+      "learning_rate": 3.2501135744951817e-06,
+      "loss": 0.3458,
+      "step": 38412
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.4527783393859863,
+      "learning_rate": 3.249866592971557e-06,
+      "loss": 0.6394,
+      "step": 38413
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.7575196027755737,
+      "learning_rate": 3.2496196163143985e-06,
+      "loss": 0.5804,
+      "step": 38414
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.993208646774292,
+      "learning_rate": 3.2493726445243922e-06,
+      "loss": 0.7283,
+      "step": 38415
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.4746932983398438,
+      "learning_rate": 3.249125677602226e-06,
+      "loss": 0.5206,
+      "step": 38416
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.3537302017211914,
+      "learning_rate": 3.2488787155485846e-06,
+      "loss": 0.609,
+      "step": 38417
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.5963244438171387,
+      "learning_rate": 3.248631758364156e-06,
+      "loss": 0.75,
+      "step": 38418
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.2822625637054443,
+      "learning_rate": 3.248384806049626e-06,
+      "loss": 0.6516,
+      "step": 38419
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.8720836639404297,
+      "learning_rate": 3.2481378586056813e-06,
+      "loss": 0.4594,
+      "step": 38420
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.199730634689331,
+      "learning_rate": 3.2478909160330097e-06,
+      "loss": 0.6119,
+      "step": 38421
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.1785261631011963,
+      "learning_rate": 3.2476439783322967e-06,
+      "loss": 0.5442,
+      "step": 38422
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.047471761703491,
+      "learning_rate": 3.247397045504229e-06,
+      "loss": 0.6354,
+      "step": 38423
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.1155526638031006,
+      "learning_rate": 3.2471501175494942e-06,
+      "loss": 0.424,
+      "step": 38424
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.7890005111694336,
+      "learning_rate": 3.246903194468778e-06,
+      "loss": 0.5748,
+      "step": 38425
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.2785258293151855,
+      "learning_rate": 3.2466562762627674e-06,
+      "loss": 0.6509,
+      "step": 38426
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 4.709211826324463,
+      "learning_rate": 3.246409362932149e-06,
+      "loss": 0.8348,
+      "step": 38427
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.0331945419311523,
+      "learning_rate": 3.2461624544776087e-06,
+      "loss": 0.5863,
+      "step": 38428
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.7033026218414307,
+      "learning_rate": 3.245915550899834e-06,
+      "loss": 0.5858,
+      "step": 38429
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.2624332904815674,
+      "learning_rate": 3.2456686521995113e-06,
+      "loss": 0.6442,
+      "step": 38430
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.6170525550842285,
+      "learning_rate": 3.2454217583773267e-06,
+      "loss": 0.7186,
+      "step": 38431
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.6210330724716187,
+      "learning_rate": 3.245174869433967e-06,
+      "loss": 0.6178,
+      "step": 38432
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.8907314538955688,
+      "learning_rate": 3.244927985370118e-06,
+      "loss": 0.5321,
+      "step": 38433
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.5174808502197266,
+      "learning_rate": 3.2446811061864657e-06,
+      "loss": 0.599,
+      "step": 38434
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.216188669204712,
+      "learning_rate": 3.2444342318836993e-06,
+      "loss": 0.687,
+      "step": 38435
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.748746871948242,
+      "learning_rate": 3.244187362462503e-06,
+      "loss": 0.8521,
+      "step": 38436
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.046255350112915,
+      "learning_rate": 3.243940497923565e-06,
+      "loss": 0.9652,
+      "step": 38437
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.6532795429229736,
+      "learning_rate": 3.24369363826757e-06,
+      "loss": 0.5135,
+      "step": 38438
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.8319165706634521,
+      "learning_rate": 3.2434467834952057e-06,
+      "loss": 0.5297,
+      "step": 38439
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.055016279220581,
+      "learning_rate": 3.2431999336071572e-06,
+      "loss": 0.7724,
+      "step": 38440
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.877416968345642,
+      "learning_rate": 3.2429530886041117e-06,
+      "loss": 0.7222,
+      "step": 38441
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.8999059200286865,
+      "learning_rate": 3.2427062484867556e-06,
+      "loss": 0.6846,
+      "step": 38442
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.111046314239502,
+      "learning_rate": 3.242459413255775e-06,
+      "loss": 0.7315,
+      "step": 38443
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.118833541870117,
+      "learning_rate": 3.2422125829118566e-06,
+      "loss": 0.4908,
+      "step": 38444
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.092580556869507,
+      "learning_rate": 3.241965757455686e-06,
+      "loss": 0.8763,
+      "step": 38445
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.6667280197143555,
+      "learning_rate": 3.241718936887951e-06,
+      "loss": 0.7779,
+      "step": 38446
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.2787160873413086,
+      "learning_rate": 3.241472121209336e-06,
+      "loss": 0.5169,
+      "step": 38447
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.645817995071411,
+      "learning_rate": 3.2412253104205287e-06,
+      "loss": 0.7436,
+      "step": 38448
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.807659387588501,
+      "learning_rate": 3.2409785045222152e-06,
+      "loss": 0.7744,
+      "step": 38449
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.26825213432312,
+      "learning_rate": 3.240731703515081e-06,
+      "loss": 0.8349,
+      "step": 38450
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.5601046085357666,
+      "learning_rate": 3.2404849073998124e-06,
+      "loss": 0.688,
+      "step": 38451
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.5818110704421997,
+      "learning_rate": 3.240238116177097e-06,
+      "loss": 0.5526,
+      "step": 38452
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.142972946166992,
+      "learning_rate": 3.23999132984762e-06,
+      "loss": 0.6767,
+      "step": 38453
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.141094923019409,
+      "learning_rate": 3.2397445484120666e-06,
+      "loss": 0.6672,
+      "step": 38454
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.9041881561279297,
+      "learning_rate": 3.2394977718711252e-06,
+      "loss": 0.6529,
+      "step": 38455
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.1258065700531006,
+      "learning_rate": 3.2392510002254816e-06,
+      "loss": 0.5917,
+      "step": 38456
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 7.905515193939209,
+      "learning_rate": 3.239004233475821e-06,
+      "loss": 0.5426,
+      "step": 38457
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.3101694583892822,
+      "learning_rate": 3.23875747162283e-06,
+      "loss": 0.4586,
+      "step": 38458
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.393587589263916,
+      "learning_rate": 3.2385107146671944e-06,
+      "loss": 0.6465,
+      "step": 38459
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.341986894607544,
+      "learning_rate": 3.2382639626096015e-06,
+      "loss": 0.8124,
+      "step": 38460
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 4.186516761779785,
+      "learning_rate": 3.238017215450736e-06,
+      "loss": 0.5963,
+      "step": 38461
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.632756233215332,
+      "learning_rate": 3.237770473191284e-06,
+      "loss": 0.5705,
+      "step": 38462
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.7966424226760864,
+      "learning_rate": 3.237523735831932e-06,
+      "loss": 0.5968,
+      "step": 38463
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.53555428981781,
+      "learning_rate": 3.237277003373367e-06,
+      "loss": 0.4548,
+      "step": 38464
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.619097948074341,
+      "learning_rate": 3.237030275816274e-06,
+      "loss": 0.3965,
+      "step": 38465
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.936361074447632,
+      "learning_rate": 3.23678355316134e-06,
+      "loss": 1.0027,
+      "step": 38466
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 4.127473831176758,
+      "learning_rate": 3.23653683540925e-06,
+      "loss": 0.7033,
+      "step": 38467
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 5.749996185302734,
+      "learning_rate": 3.2362901225606907e-06,
+      "loss": 0.5682,
+      "step": 38468
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.7009899616241455,
+      "learning_rate": 3.236043414616348e-06,
+      "loss": 0.5366,
+      "step": 38469
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.4528706073760986,
+      "learning_rate": 3.2357967115769074e-06,
+      "loss": 0.5706,
+      "step": 38470
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.2767982482910156,
+      "learning_rate": 3.2355500134430552e-06,
+      "loss": 0.5643,
+      "step": 38471
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.9748159646987915,
+      "learning_rate": 3.235303320215477e-06,
+      "loss": 0.61,
+      "step": 38472
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.8318359851837158,
+      "learning_rate": 3.2350566318948607e-06,
+      "loss": 0.6029,
+      "step": 38473
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.077160120010376,
+      "learning_rate": 3.2348099484818872e-06,
+      "loss": 0.6917,
+      "step": 38474
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.769523859024048,
+      "learning_rate": 3.234563269977249e-06,
+      "loss": 0.5345,
+      "step": 38475
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 4.523415565490723,
+      "learning_rate": 3.234316596381628e-06,
+      "loss": 0.6291,
+      "step": 38476
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.7728521823883057,
+      "learning_rate": 3.234069927695712e-06,
+      "loss": 0.7639,
+      "step": 38477
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.8783531188964844,
+      "learning_rate": 3.233823263920186e-06,
+      "loss": 0.4648,
+      "step": 38478
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.30671763420105,
+      "learning_rate": 3.2335766050557356e-06,
+      "loss": 0.6199,
+      "step": 38479
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.0551178455352783,
+      "learning_rate": 3.233329951103047e-06,
+      "loss": 0.534,
+      "step": 38480
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.7365713119506836,
+      "learning_rate": 3.2330833020628056e-06,
+      "loss": 0.6873,
+      "step": 38481
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.396862745285034,
+      "learning_rate": 3.232836657935698e-06,
+      "loss": 0.5849,
+      "step": 38482
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.447281837463379,
+      "learning_rate": 3.2325900187224096e-06,
+      "loss": 0.4418,
+      "step": 38483
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.073173999786377,
+      "learning_rate": 3.2323433844236264e-06,
+      "loss": 0.4829,
+      "step": 38484
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.720322370529175,
+      "learning_rate": 3.232096755040034e-06,
+      "loss": 0.5519,
+      "step": 38485
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.7497973442077637,
+      "learning_rate": 3.2318501305723182e-06,
+      "loss": 0.7136,
+      "step": 38486
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.322493076324463,
+      "learning_rate": 3.2316035110211646e-06,
+      "loss": 0.5109,
+      "step": 38487
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.625962972640991,
+      "learning_rate": 3.23135689638726e-06,
+      "loss": 0.4055,
+      "step": 38488
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.001520872116089,
+      "learning_rate": 3.2311102866712886e-06,
+      "loss": 0.6675,
+      "step": 38489
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.2385549545288086,
+      "learning_rate": 3.230863681873937e-06,
+      "loss": 0.6369,
+      "step": 38490
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.2714860439300537,
+      "learning_rate": 3.2306170819958906e-06,
+      "loss": 0.5665,
+      "step": 38491
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.3519673347473145,
+      "learning_rate": 3.230370487037835e-06,
+      "loss": 0.5585,
+      "step": 38492
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.9330214262008667,
+      "learning_rate": 3.2301238970004565e-06,
+      "loss": 0.7579,
+      "step": 38493
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.371351957321167,
+      "learning_rate": 3.2298773118844406e-06,
+      "loss": 0.3792,
+      "step": 38494
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.1320221424102783,
+      "learning_rate": 3.2296307316904705e-06,
+      "loss": 0.5685,
+      "step": 38495
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.2935616970062256,
+      "learning_rate": 3.2293841564192365e-06,
+      "loss": 0.5104,
+      "step": 38496
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.878533124923706,
+      "learning_rate": 3.2291375860714217e-06,
+      "loss": 0.8598,
+      "step": 38497
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.8908379077911377,
+      "learning_rate": 3.228891020647712e-06,
+      "loss": 0.6675,
+      "step": 38498
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.1603691577911377,
+      "learning_rate": 3.228644460148793e-06,
+      "loss": 0.7383,
+      "step": 38499
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.9614912271499634,
+      "learning_rate": 3.22839790457535e-06,
+      "loss": 0.658,
+      "step": 38500
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.149165153503418,
+      "learning_rate": 3.2281513539280685e-06,
+      "loss": 0.6408,
+      "step": 38501
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.722360372543335,
+      "learning_rate": 3.2279048082076346e-06,
+      "loss": 0.697,
+      "step": 38502
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.584482431411743,
+      "learning_rate": 3.227658267414734e-06,
+      "loss": 0.5824,
+      "step": 38503
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.3040435314178467,
+      "learning_rate": 3.227411731550051e-06,
+      "loss": 0.6177,
+      "step": 38504
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.4500505924224854,
+      "learning_rate": 3.2271652006142727e-06,
+      "loss": 0.4822,
+      "step": 38505
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.728736162185669,
+      "learning_rate": 3.2269186746080834e-06,
+      "loss": 0.6338,
+      "step": 38506
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.749702215194702,
+      "learning_rate": 3.226672153532169e-06,
+      "loss": 0.5186,
+      "step": 38507
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.1811814308166504,
+      "learning_rate": 3.2264256373872155e-06,
+      "loss": 0.4363,
+      "step": 38508
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.1689887046813965,
+      "learning_rate": 3.2261791261739073e-06,
+      "loss": 0.6961,
+      "step": 38509
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.9829148054122925,
+      "learning_rate": 3.225932619892931e-06,
+      "loss": 0.7662,
+      "step": 38510
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.9420764446258545,
+      "learning_rate": 3.225686118544971e-06,
+      "loss": 0.7163,
+      "step": 38511
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.5009145736694336,
+      "learning_rate": 3.2254396221307137e-06,
+      "loss": 0.7009,
+      "step": 38512
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.361112594604492,
+      "learning_rate": 3.225193130650843e-06,
+      "loss": 0.618,
+      "step": 38513
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.5334994792938232,
+      "learning_rate": 3.224946644106046e-06,
+      "loss": 0.3252,
+      "step": 38514
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.282958984375,
+      "learning_rate": 3.2247001624970065e-06,
+      "loss": 0.5122,
+      "step": 38515
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.92913818359375,
+      "learning_rate": 3.2244536858244123e-06,
+      "loss": 0.6702,
+      "step": 38516
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.783975601196289,
+      "learning_rate": 3.224207214088947e-06,
+      "loss": 0.5271,
+      "step": 38517
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.9353373050689697,
+      "learning_rate": 3.223960747291296e-06,
+      "loss": 0.6698,
+      "step": 38518
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.1333413124084473,
+      "learning_rate": 3.2237142854321445e-06,
+      "loss": 0.6309,
+      "step": 38519
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.546970844268799,
+      "learning_rate": 3.223467828512179e-06,
+      "loss": 0.8112,
+      "step": 38520
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.032493829727173,
+      "learning_rate": 3.2232213765320836e-06,
+      "loss": 0.8889,
+      "step": 38521
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.1516270637512207,
+      "learning_rate": 3.222974929492544e-06,
+      "loss": 0.6132,
+      "step": 38522
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.5824103355407715,
+      "learning_rate": 3.2227284873942453e-06,
+      "loss": 0.6198,
+      "step": 38523
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.659604549407959,
+      "learning_rate": 3.222482050237872e-06,
+      "loss": 0.5895,
+      "step": 38524
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.2021424770355225,
+      "learning_rate": 3.222235618024111e-06,
+      "loss": 0.5762,
+      "step": 38525
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.54337477684021,
+      "learning_rate": 3.2219891907536473e-06,
+      "loss": 0.3732,
+      "step": 38526
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.6249504089355469,
+      "learning_rate": 3.221742768427165e-06,
+      "loss": 0.4697,
+      "step": 38527
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.6340110301971436,
+      "learning_rate": 3.22149635104535e-06,
+      "loss": 0.6802,
+      "step": 38528
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 5.04023551940918,
+      "learning_rate": 3.2212499386088882e-06,
+      "loss": 0.4246,
+      "step": 38529
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.5941834449768066,
+      "learning_rate": 3.221003531118463e-06,
+      "loss": 0.583,
+      "step": 38530
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.800968885421753,
+      "learning_rate": 3.2207571285747606e-06,
+      "loss": 0.6129,
+      "step": 38531
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.1938202381134033,
+      "learning_rate": 3.2205107309784667e-06,
+      "loss": 0.7361,
+      "step": 38532
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.8111753463745117,
+      "learning_rate": 3.2202643383302657e-06,
+      "loss": 0.6914,
+      "step": 38533
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.255779981613159,
+      "learning_rate": 3.220017950630843e-06,
+      "loss": 0.6222,
+      "step": 38534
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.0051586627960205,
+      "learning_rate": 3.219771567880882e-06,
+      "loss": 0.6543,
+      "step": 38535
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.309364080429077,
+      "learning_rate": 3.21952519008107e-06,
+      "loss": 0.6171,
+      "step": 38536
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.910952568054199,
+      "learning_rate": 3.219278817232092e-06,
+      "loss": 0.6769,
+      "step": 38537
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.0227317810058594,
+      "learning_rate": 3.2190324493346325e-06,
+      "loss": 0.8769,
+      "step": 38538
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.9500659704208374,
+      "learning_rate": 3.2187860863893774e-06,
+      "loss": 0.4601,
+      "step": 38539
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.3229923248291016,
+      "learning_rate": 3.2185397283970097e-06,
+      "loss": 0.7075,
+      "step": 38540
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.2579400539398193,
+      "learning_rate": 3.218293375358217e-06,
+      "loss": 0.7599,
+      "step": 38541
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.101736068725586,
+      "learning_rate": 3.218047027273682e-06,
+      "loss": 0.6695,
+      "step": 38542
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.5942440032958984,
+      "learning_rate": 3.217800684144091e-06,
+      "loss": 0.6475,
+      "step": 38543
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.453892230987549,
+      "learning_rate": 3.2175543459701286e-06,
+      "loss": 0.5473,
+      "step": 38544
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.9844361543655396,
+      "learning_rate": 3.2173080127524803e-06,
+      "loss": 0.5362,
+      "step": 38545
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.177877902984619,
+      "learning_rate": 3.21706168449183e-06,
+      "loss": 0.6133,
+      "step": 38546
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.0514698028564453,
+      "learning_rate": 3.2168153611888638e-06,
+      "loss": 0.4923,
+      "step": 38547
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.364957571029663,
+      "learning_rate": 3.2165690428442652e-06,
+      "loss": 0.841,
+      "step": 38548
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.8699474334716797,
+      "learning_rate": 3.2163227294587206e-06,
+      "loss": 0.5233,
+      "step": 38549
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 4.773436069488525,
+      "learning_rate": 3.2160764210329143e-06,
+      "loss": 0.7737,
+      "step": 38550
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.092087984085083,
+      "learning_rate": 3.215830117567531e-06,
+      "loss": 0.4028,
+      "step": 38551
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.638795018196106,
+      "learning_rate": 3.215583819063256e-06,
+      "loss": 0.3377,
+      "step": 38552
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.262998342514038,
+      "learning_rate": 3.215337525520774e-06,
+      "loss": 0.6491,
+      "step": 38553
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.9173288345336914,
+      "learning_rate": 3.21509123694077e-06,
+      "loss": 0.4887,
+      "step": 38554
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.9468154907226562,
+      "learning_rate": 3.214844953323928e-06,
+      "loss": 0.7151,
+      "step": 38555
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.7355287075042725,
+      "learning_rate": 3.2145986746709316e-06,
+      "loss": 0.5675,
+      "step": 38556
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.6088310480117798,
+      "learning_rate": 3.21435240098247e-06,
+      "loss": 0.6766,
+      "step": 38557
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.3318188190460205,
+      "learning_rate": 3.2141061322592248e-06,
+      "loss": 0.4529,
+      "step": 38558
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.1805076599121094,
+      "learning_rate": 3.213859868501882e-06,
+      "loss": 0.4362,
+      "step": 38559
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 4.02670955657959,
+      "learning_rate": 3.2136136097111254e-06,
+      "loss": 0.4979,
+      "step": 38560
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.8296853303909302,
+      "learning_rate": 3.21336735588764e-06,
+      "loss": 0.3376,
+      "step": 38561
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.7698519229888916,
+      "learning_rate": 3.2131211070321112e-06,
+      "loss": 0.7816,
+      "step": 38562
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.2666423320770264,
+      "learning_rate": 3.2128748631452235e-06,
+      "loss": 0.5731,
+      "step": 38563
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.1669921875,
+      "learning_rate": 3.2126286242276605e-06,
+      "loss": 0.5299,
+      "step": 38564
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.2074034214019775,
+      "learning_rate": 3.2123823902801077e-06,
+      "loss": 0.9269,
+      "step": 38565
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.184887170791626,
+      "learning_rate": 3.2121361613032504e-06,
+      "loss": 0.5654,
+      "step": 38566
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.5169546604156494,
+      "learning_rate": 3.2118899372977722e-06,
+      "loss": 0.8216,
+      "step": 38567
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.121680736541748,
+      "learning_rate": 3.211643718264359e-06,
+      "loss": 0.3198,
+      "step": 38568
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.261704921722412,
+      "learning_rate": 3.2113975042036938e-06,
+      "loss": 0.6776,
+      "step": 38569
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.91127872467041,
+      "learning_rate": 3.2111512951164627e-06,
+      "loss": 0.5754,
+      "step": 38570
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.6949992179870605,
+      "learning_rate": 3.210905091003349e-06,
+      "loss": 0.6139,
+      "step": 38571
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.0102412700653076,
+      "learning_rate": 3.210658891865039e-06,
+      "loss": 0.6289,
+      "step": 38572
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 5.098623752593994,
+      "learning_rate": 3.2104126977022153e-06,
+      "loss": 0.3969,
+      "step": 38573
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.8466216325759888,
+      "learning_rate": 3.210166508515563e-06,
+      "loss": 0.5838,
+      "step": 38574
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.8500534296035767,
+      "learning_rate": 3.2099203243057687e-06,
+      "loss": 0.4647,
+      "step": 38575
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.235840082168579,
+      "learning_rate": 3.2096741450735136e-06,
+      "loss": 0.6784,
+      "step": 38576
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.8233864307403564,
+      "learning_rate": 3.2094279708194855e-06,
+      "loss": 0.3393,
+      "step": 38577
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.375870704650879,
+      "learning_rate": 3.209181801544368e-06,
+      "loss": 0.4446,
+      "step": 38578
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.8186638355255127,
+      "learning_rate": 3.2089356372488433e-06,
+      "loss": 0.5299,
+      "step": 38579
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 6.232175350189209,
+      "learning_rate": 3.208689477933599e-06,
+      "loss": 0.4979,
+      "step": 38580
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.0080950260162354,
+      "learning_rate": 3.208443323599318e-06,
+      "loss": 0.4325,
+      "step": 38581
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.239485263824463,
+      "learning_rate": 3.208197174246685e-06,
+      "loss": 0.5766,
+      "step": 38582
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.9157779216766357,
+      "learning_rate": 3.2079510298763836e-06,
+      "loss": 0.7453,
+      "step": 38583
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.0714681148529053,
+      "learning_rate": 3.2077048904891005e-06,
+      "loss": 0.6037,
+      "step": 38584
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.720820903778076,
+      "learning_rate": 3.2074587560855182e-06,
+      "loss": 0.5708,
+      "step": 38585
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.1927013397216797,
+      "learning_rate": 3.20721262666632e-06,
+      "loss": 0.5416,
+      "step": 38586
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 7.220386028289795,
+      "learning_rate": 3.206966502232193e-06,
+      "loss": 0.4216,
+      "step": 38587
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 5.643616676330566,
+      "learning_rate": 3.206720382783821e-06,
+      "loss": 0.5859,
+      "step": 38588
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.033411979675293,
+      "learning_rate": 3.2064742683218873e-06,
+      "loss": 0.758,
+      "step": 38589
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.5240936279296875,
+      "learning_rate": 3.206228158847078e-06,
+      "loss": 0.4885,
+      "step": 38590
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.6791422367095947,
+      "learning_rate": 3.205982054360075e-06,
+      "loss": 0.8479,
+      "step": 38591
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.287250518798828,
+      "learning_rate": 3.205735954861564e-06,
+      "loss": 0.6601,
+      "step": 38592
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.876827597618103,
+      "learning_rate": 3.205489860352229e-06,
+      "loss": 0.5255,
+      "step": 38593
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.664595127105713,
+      "learning_rate": 3.205243770832755e-06,
+      "loss": 0.7811,
+      "step": 38594
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 5.356510162353516,
+      "learning_rate": 3.2049976863038257e-06,
+      "loss": 0.5349,
+      "step": 38595
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.240802526473999,
+      "learning_rate": 3.2047516067661246e-06,
+      "loss": 0.4901,
+      "step": 38596
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.2397468090057373,
+      "learning_rate": 3.2045055322203367e-06,
+      "loss": 0.494,
+      "step": 38597
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.858781099319458,
+      "learning_rate": 3.2042594626671465e-06,
+      "loss": 0.9191,
+      "step": 38598
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.0391173362731934,
+      "learning_rate": 3.2040133981072384e-06,
+      "loss": 0.5261,
+      "step": 38599
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.390261173248291,
+      "learning_rate": 3.203767338541296e-06,
+      "loss": 0.6403,
+      "step": 38600
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 4.28156852722168,
+      "learning_rate": 3.203521283970005e-06,
+      "loss": 0.7148,
+      "step": 38601
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.1979596614837646,
+      "learning_rate": 3.2032752343940466e-06,
+      "loss": 0.7549,
+      "step": 38602
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.628390312194824,
+      "learning_rate": 3.2030291898141075e-06,
+      "loss": 0.4425,
+      "step": 38603
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.6647065877914429,
+      "learning_rate": 3.2027831502308717e-06,
+      "loss": 0.64,
+      "step": 38604
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.954887628555298,
+      "learning_rate": 3.202537115645022e-06,
+      "loss": 0.514,
+      "step": 38605
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.3430609703063965,
+      "learning_rate": 3.2022910860572432e-06,
+      "loss": 0.7088,
+      "step": 38606
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.9082396030426025,
+      "learning_rate": 3.20204506146822e-06,
+      "loss": 0.5782,
+      "step": 38607
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.7147587537765503,
+      "learning_rate": 3.2017990418786347e-06,
+      "loss": 0.679,
+      "step": 38608
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.9724104404449463,
+      "learning_rate": 3.2015530272891736e-06,
+      "loss": 0.608,
+      "step": 38609
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.9225728511810303,
+      "learning_rate": 3.20130701770052e-06,
+      "loss": 0.7992,
+      "step": 38610
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.247616767883301,
+      "learning_rate": 3.201061013113358e-06,
+      "loss": 0.8129,
+      "step": 38611
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.1876511573791504,
+      "learning_rate": 3.200815013528371e-06,
+      "loss": 0.6061,
+      "step": 38612
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.1113648414611816,
+      "learning_rate": 3.2005690189462433e-06,
+      "loss": 0.8626,
+      "step": 38613
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.161125421524048,
+      "learning_rate": 3.200323029367659e-06,
+      "loss": 0.5307,
+      "step": 38614
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.8369895219802856,
+      "learning_rate": 3.200077044793303e-06,
+      "loss": 0.5362,
+      "step": 38615
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.8635268211364746,
+      "learning_rate": 3.1998310652238574e-06,
+      "loss": 0.8298,
+      "step": 38616
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.7827633619308472,
+      "learning_rate": 3.1995850906600057e-06,
+      "loss": 0.432,
+      "step": 38617
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.2358100414276123,
+      "learning_rate": 3.199339121102435e-06,
+      "loss": 0.4439,
+      "step": 38618
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.091670513153076,
+      "learning_rate": 3.199093156551828e-06,
+      "loss": 0.5343,
+      "step": 38619
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.0983259677886963,
+      "learning_rate": 3.1988471970088686e-06,
+      "loss": 0.6635,
+      "step": 38620
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.418914556503296,
+      "learning_rate": 3.19860124247424e-06,
+      "loss": 0.7889,
+      "step": 38621
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.0914759635925293,
+      "learning_rate": 3.198355292948627e-06,
+      "loss": 0.597,
+      "step": 38622
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.642923593521118,
+      "learning_rate": 3.1981093484327127e-06,
+      "loss": 0.5339,
+      "step": 38623
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.5210022926330566,
+      "learning_rate": 3.197863408927181e-06,
+      "loss": 0.5088,
+      "step": 38624
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.151724100112915,
+      "learning_rate": 3.1976174744327167e-06,
+      "loss": 0.6089,
+      "step": 38625
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.5956201553344727,
+      "learning_rate": 3.1973715449500027e-06,
+      "loss": 0.6168,
+      "step": 38626
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.9874825477600098,
+      "learning_rate": 3.197125620479723e-06,
+      "loss": 0.4862,
+      "step": 38627
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.4131999015808105,
+      "learning_rate": 3.1968797010225618e-06,
+      "loss": 0.6464,
+      "step": 38628
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.621168851852417,
+      "learning_rate": 3.1966337865792024e-06,
+      "loss": 0.3328,
+      "step": 38629
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.543179512023926,
+      "learning_rate": 3.1963878771503293e-06,
+      "loss": 0.8422,
+      "step": 38630
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.369814872741699,
+      "learning_rate": 3.1961419727366256e-06,
+      "loss": 0.7316,
+      "step": 38631
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 4.130323886871338,
+      "learning_rate": 3.1958960733387757e-06,
+      "loss": 0.8725,
+      "step": 38632
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.5980095863342285,
+      "learning_rate": 3.1956501789574624e-06,
+      "loss": 0.7762,
+      "step": 38633
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.4373810291290283,
+      "learning_rate": 3.19540428959337e-06,
+      "loss": 0.8254,
+      "step": 38634
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.2252039909362793,
+      "learning_rate": 3.195158405247183e-06,
+      "loss": 0.6154,
+      "step": 38635
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.5064971446990967,
+      "learning_rate": 3.1949125259195823e-06,
+      "loss": 0.5137,
+      "step": 38636
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.302231788635254,
+      "learning_rate": 3.1946666516112545e-06,
+      "loss": 0.6998,
+      "step": 38637
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 8.729026794433594,
+      "learning_rate": 3.1944207823228835e-06,
+      "loss": 0.5323,
+      "step": 38638
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.764219045639038,
+      "learning_rate": 3.1941749180551517e-06,
+      "loss": 0.4135,
+      "step": 38639
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.912876844406128,
+      "learning_rate": 3.193929058808743e-06,
+      "loss": 0.6009,
+      "step": 38640
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.9288300275802612,
+      "learning_rate": 3.1936832045843406e-06,
+      "loss": 0.5528,
+      "step": 38641
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.045555591583252,
+      "learning_rate": 3.1934373553826292e-06,
+      "loss": 0.4719,
+      "step": 38642
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.857825756072998,
+      "learning_rate": 3.1931915112042912e-06,
+      "loss": 0.4371,
+      "step": 38643
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.8853633403778076,
+      "learning_rate": 3.1929456720500114e-06,
+      "loss": 0.5033,
+      "step": 38644
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 4.2211127281188965,
+      "learning_rate": 3.192699837920472e-06,
+      "loss": 0.8575,
+      "step": 38645
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.365297317504883,
+      "learning_rate": 3.192454008816358e-06,
+      "loss": 0.5747,
+      "step": 38646
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.9845527410507202,
+      "learning_rate": 3.192208184738352e-06,
+      "loss": 0.7264,
+      "step": 38647
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.484895944595337,
+      "learning_rate": 3.1919623656871367e-06,
+      "loss": 0.591,
+      "step": 38648
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.7816600799560547,
+      "learning_rate": 3.1917165516633973e-06,
+      "loss": 0.6258,
+      "step": 38649
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.834829092025757,
+      "learning_rate": 3.1914707426678175e-06,
+      "loss": 0.5335,
+      "step": 38650
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.3291549682617188,
+      "learning_rate": 3.1912249387010804e-06,
+      "loss": 0.707,
+      "step": 38651
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.3098466396331787,
+      "learning_rate": 3.1909791397638686e-06,
+      "loss": 0.6672,
+      "step": 38652
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.533792734146118,
+      "learning_rate": 3.1907333458568657e-06,
+      "loss": 0.6194,
+      "step": 38653
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.268164873123169,
+      "learning_rate": 3.1904875569807565e-06,
+      "loss": 0.6157,
+      "step": 38654
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.4864625930786133,
+      "learning_rate": 3.1902417731362234e-06,
+      "loss": 0.8202,
+      "step": 38655
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.9440228939056396,
+      "learning_rate": 3.1899959943239495e-06,
+      "loss": 0.5179,
+      "step": 38656
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.610313892364502,
+      "learning_rate": 3.1897502205446174e-06,
+      "loss": 0.763,
+      "step": 38657
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.34505295753479,
+      "learning_rate": 3.1895044517989137e-06,
+      "loss": 0.6831,
+      "step": 38658
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 4.547167778015137,
+      "learning_rate": 3.1892586880875186e-06,
+      "loss": 0.5375,
+      "step": 38659
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 4.535849094390869,
+      "learning_rate": 3.1890129294111173e-06,
+      "loss": 0.7777,
+      "step": 38660
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.6642369031906128,
+      "learning_rate": 3.188767175770393e-06,
+      "loss": 0.635,
+      "step": 38661
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.934607744216919,
+      "learning_rate": 3.188521427166029e-06,
+      "loss": 0.6696,
+      "step": 38662
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.8399906158447266,
+      "learning_rate": 3.188275683598708e-06,
+      "loss": 0.3915,
+      "step": 38663
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.881479024887085,
+      "learning_rate": 3.188029945069113e-06,
+      "loss": 0.4234,
+      "step": 38664
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.8017348051071167,
+      "learning_rate": 3.187784211577929e-06,
+      "loss": 0.4796,
+      "step": 38665
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.356919765472412,
+      "learning_rate": 3.187538483125838e-06,
+      "loss": 0.5197,
+      "step": 38666
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.8921476602554321,
+      "learning_rate": 3.1872927597135232e-06,
+      "loss": 0.506,
+      "step": 38667
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.059309720993042,
+      "learning_rate": 3.187047041341669e-06,
+      "loss": 0.6318,
+      "step": 38668
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.9388645887374878,
+      "learning_rate": 3.1868013280109567e-06,
+      "loss": 0.6809,
+      "step": 38669
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.5716731548309326,
+      "learning_rate": 3.1865556197220708e-06,
+      "loss": 0.5504,
+      "step": 38670
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.797445774078369,
+      "learning_rate": 3.186309916475695e-06,
+      "loss": 0.8541,
+      "step": 38671
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.718303918838501,
+      "learning_rate": 3.1860642182725116e-06,
+      "loss": 0.6868,
+      "step": 38672
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.919495105743408,
+      "learning_rate": 3.185818525113204e-06,
+      "loss": 0.6545,
+      "step": 38673
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.5630643367767334,
+      "learning_rate": 3.1855728369984555e-06,
+      "loss": 0.5228,
+      "step": 38674
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.776146411895752,
+      "learning_rate": 3.1853271539289494e-06,
+      "loss": 0.5684,
+      "step": 38675
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.306047201156616,
+      "learning_rate": 3.185081475905368e-06,
+      "loss": 0.6284,
+      "step": 38676
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.2020652294158936,
+      "learning_rate": 3.1848358029283964e-06,
+      "loss": 0.7436,
+      "step": 38677
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.1823720932006836,
+      "learning_rate": 3.1845901349987136e-06,
+      "loss": 0.617,
+      "step": 38678
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.432316541671753,
+      "learning_rate": 3.184344472117008e-06,
+      "loss": 0.7949,
+      "step": 38679
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.242511510848999,
+      "learning_rate": 3.1840988142839596e-06,
+      "loss": 0.4204,
+      "step": 38680
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.297182083129883,
+      "learning_rate": 3.1838531615002525e-06,
+      "loss": 0.6783,
+      "step": 38681
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.4598238468170166,
+      "learning_rate": 3.1836075137665703e-06,
+      "loss": 0.6199,
+      "step": 38682
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.01062273979187,
+      "learning_rate": 3.1833618710835934e-06,
+      "loss": 0.5607,
+      "step": 38683
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.386524200439453,
+      "learning_rate": 3.183116233452008e-06,
+      "loss": 0.6773,
+      "step": 38684
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.554854393005371,
+      "learning_rate": 3.1828706008724953e-06,
+      "loss": 0.7567,
+      "step": 38685
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.060849666595459,
+      "learning_rate": 3.1826249733457386e-06,
+      "loss": 0.6925,
+      "step": 38686
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.545809745788574,
+      "learning_rate": 3.182379350872421e-06,
+      "loss": 0.7477,
+      "step": 38687
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.480384588241577,
+      "learning_rate": 3.182133733453226e-06,
+      "loss": 0.5586,
+      "step": 38688
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.992431879043579,
+      "learning_rate": 3.181888121088836e-06,
+      "loss": 0.5737,
+      "step": 38689
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.118525743484497,
+      "learning_rate": 3.1816425137799332e-06,
+      "loss": 0.5246,
+      "step": 38690
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.4421193599700928,
+      "learning_rate": 3.1813969115272027e-06,
+      "loss": 0.6715,
+      "step": 38691
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.2250912189483643,
+      "learning_rate": 3.1811513143313253e-06,
+      "loss": 0.3429,
+      "step": 38692
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.519294023513794,
+      "learning_rate": 3.180905722192985e-06,
+      "loss": 0.6625,
+      "step": 38693
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.6736624240875244,
+      "learning_rate": 3.180660135112864e-06,
+      "loss": 0.3459,
+      "step": 38694
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.9726338386535645,
+      "learning_rate": 3.1804145530916462e-06,
+      "loss": 0.5795,
+      "step": 38695
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.5257949829101562,
+      "learning_rate": 3.180168976130014e-06,
+      "loss": 0.7439,
+      "step": 38696
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.79705011844635,
+      "learning_rate": 3.17992340422865e-06,
+      "loss": 0.4238,
+      "step": 38697
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.2514538764953613,
+      "learning_rate": 3.1796778373882344e-06,
+      "loss": 0.4087,
+      "step": 38698
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.5345799922943115,
+      "learning_rate": 3.1794322756094553e-06,
+      "loss": 0.6913,
+      "step": 38699
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 4.200467109680176,
+      "learning_rate": 3.1791867188929937e-06,
+      "loss": 0.7513,
+      "step": 38700
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.400010347366333,
+      "learning_rate": 3.178941167239531e-06,
+      "loss": 0.6561,
+      "step": 38701
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.862849235534668,
+      "learning_rate": 3.178695620649751e-06,
+      "loss": 0.7148,
+      "step": 38702
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.437992811203003,
+      "learning_rate": 3.178450079124336e-06,
+      "loss": 0.5088,
+      "step": 38703
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 4.264806747436523,
+      "learning_rate": 3.1782045426639697e-06,
+      "loss": 0.7911,
+      "step": 38704
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.408761978149414,
+      "learning_rate": 3.177959011269333e-06,
+      "loss": 0.6249,
+      "step": 38705
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.4026098251342773,
+      "learning_rate": 3.1777134849411106e-06,
+      "loss": 0.7289,
+      "step": 38706
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.135859966278076,
+      "learning_rate": 3.1774679636799837e-06,
+      "loss": 0.4019,
+      "step": 38707
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.253216505050659,
+      "learning_rate": 3.1772224474866366e-06,
+      "loss": 0.4174,
+      "step": 38708
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.137166738510132,
+      "learning_rate": 3.1769769363617487e-06,
+      "loss": 0.8613,
+      "step": 38709
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.6445531845092773,
+      "learning_rate": 3.176731430306007e-06,
+      "loss": 0.686,
+      "step": 38710
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.4888265132904053,
+      "learning_rate": 3.176485929320092e-06,
+      "loss": 0.7594,
+      "step": 38711
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.5824992656707764,
+      "learning_rate": 3.176240433404686e-06,
+      "loss": 0.707,
+      "step": 38712
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.7091221809387207,
+      "learning_rate": 3.175994942560473e-06,
+      "loss": 0.4192,
+      "step": 38713
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.3121020793914795,
+      "learning_rate": 3.1757494567881343e-06,
+      "loss": 0.8583,
+      "step": 38714
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 6.469398498535156,
+      "learning_rate": 3.175503976088353e-06,
+      "loss": 0.5768,
+      "step": 38715
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.3704700469970703,
+      "learning_rate": 3.175258500461812e-06,
+      "loss": 0.6068,
+      "step": 38716
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.4191462993621826,
+      "learning_rate": 3.175013029909193e-06,
+      "loss": 0.5446,
+      "step": 38717
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.426248550415039,
+      "learning_rate": 3.174767564431178e-06,
+      "loss": 0.7258,
+      "step": 38718
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.6063424348831177,
+      "learning_rate": 3.1745221040284522e-06,
+      "loss": 0.4304,
+      "step": 38719
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.7537386417388916,
+      "learning_rate": 3.1742766487016967e-06,
+      "loss": 0.6404,
+      "step": 38720
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.070862054824829,
+      "learning_rate": 3.1740311984515926e-06,
+      "loss": 0.6199,
+      "step": 38721
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 5.40814733505249,
+      "learning_rate": 3.1737857532788253e-06,
+      "loss": 0.466,
+      "step": 38722
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.9660571813583374,
+      "learning_rate": 3.173540313184075e-06,
+      "loss": 0.6823,
+      "step": 38723
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.099048376083374,
+      "learning_rate": 3.173294878168025e-06,
+      "loss": 0.7855,
+      "step": 38724
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.9501218795776367,
+      "learning_rate": 3.173049448231358e-06,
+      "loss": 0.5413,
+      "step": 38725
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.8710967302322388,
+      "learning_rate": 3.1728040233747556e-06,
+      "loss": 0.4829,
+      "step": 38726
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 4.363928318023682,
+      "learning_rate": 3.1725586035989008e-06,
+      "loss": 0.7213,
+      "step": 38727
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.9745705127716064,
+      "learning_rate": 3.1723131889044754e-06,
+      "loss": 0.4872,
+      "step": 38728
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.9979578256607056,
+      "learning_rate": 3.172067779292164e-06,
+      "loss": 0.4496,
+      "step": 38729
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.008317470550537,
+      "learning_rate": 3.171822374762646e-06,
+      "loss": 0.5629,
+      "step": 38730
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.1297147274017334,
+      "learning_rate": 3.1715769753166055e-06,
+      "loss": 0.5638,
+      "step": 38731
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.8544652462005615,
+      "learning_rate": 3.171331580954724e-06,
+      "loss": 0.88,
+      "step": 38732
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.967681646347046,
+      "learning_rate": 3.1710861916776852e-06,
+      "loss": 0.6444,
+      "step": 38733
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.2916207313537598,
+      "learning_rate": 3.1708408074861703e-06,
+      "loss": 0.5077,
+      "step": 38734
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.9108479022979736,
+      "learning_rate": 3.1705954283808617e-06,
+      "loss": 0.6711,
+      "step": 38735
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.9256296157836914,
+      "learning_rate": 3.170350054362442e-06,
+      "loss": 0.4839,
+      "step": 38736
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.076087474822998,
+      "learning_rate": 3.1701046854315932e-06,
+      "loss": 0.8294,
+      "step": 38737
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.860262393951416,
+      "learning_rate": 3.1698593215889984e-06,
+      "loss": 0.844,
+      "step": 38738
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.987799882888794,
+      "learning_rate": 3.169613962835336e-06,
+      "loss": 0.706,
+      "step": 38739
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.2969319820404053,
+      "learning_rate": 3.169368609171295e-06,
+      "loss": 0.6225,
+      "step": 38740
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 4.762645721435547,
+      "learning_rate": 3.169123260597553e-06,
+      "loss": 0.7132,
+      "step": 38741
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.7904380559921265,
+      "learning_rate": 3.1688779171147942e-06,
+      "loss": 0.7254,
+      "step": 38742
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.0277276039123535,
+      "learning_rate": 3.1686325787236993e-06,
+      "loss": 0.5836,
+      "step": 38743
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 4.828070640563965,
+      "learning_rate": 3.1683872454249516e-06,
+      "loss": 0.7463,
+      "step": 38744
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.571627378463745,
+      "learning_rate": 3.1681419172192326e-06,
+      "loss": 0.5417,
+      "step": 38745
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.9200849533081055,
+      "learning_rate": 3.167896594107225e-06,
+      "loss": 0.6953,
+      "step": 38746
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.840707778930664,
+      "learning_rate": 3.1676512760896098e-06,
+      "loss": 0.5123,
+      "step": 38747
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.8552396297454834,
+      "learning_rate": 3.1674059631670706e-06,
+      "loss": 0.6539,
+      "step": 38748
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.453460693359375,
+      "learning_rate": 3.1671606553402887e-06,
+      "loss": 0.5674,
+      "step": 38749
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.3602466583251953,
+      "learning_rate": 3.166915352609947e-06,
+      "loss": 0.7,
+      "step": 38750
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.813087821006775,
+      "learning_rate": 3.1666700549767265e-06,
+      "loss": 0.7425,
+      "step": 38751
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 5.035975456237793,
+      "learning_rate": 3.1664247624413102e-06,
+      "loss": 0.881,
+      "step": 38752
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.543584108352661,
+      "learning_rate": 3.1661794750043796e-06,
+      "loss": 0.5957,
+      "step": 38753
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.8540658950805664,
+      "learning_rate": 3.1659341926666166e-06,
+      "loss": 0.44,
+      "step": 38754
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.5025899410247803,
+      "learning_rate": 3.1656889154287034e-06,
+      "loss": 0.4886,
+      "step": 38755
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.8862383365631104,
+      "learning_rate": 3.165443643291323e-06,
+      "loss": 0.6703,
+      "step": 38756
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.665475845336914,
+      "learning_rate": 3.1651983762551554e-06,
+      "loss": 0.4397,
+      "step": 38757
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 4.033429145812988,
+      "learning_rate": 3.1649531143208846e-06,
+      "loss": 0.9475,
+      "step": 38758
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.523057699203491,
+      "learning_rate": 3.16470785748919e-06,
+      "loss": 0.5769,
+      "step": 38759
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.6685030460357666,
+      "learning_rate": 3.1644626057607564e-06,
+      "loss": 0.4702,
+      "step": 38760
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.649298667907715,
+      "learning_rate": 3.1642173591362657e-06,
+      "loss": 0.568,
+      "step": 38761
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.5114529132843018,
+      "learning_rate": 3.1639721176163982e-06,
+      "loss": 0.849,
+      "step": 38762
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.5909018516540527,
+      "learning_rate": 3.1637268812018364e-06,
+      "loss": 0.5184,
+      "step": 38763
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.536638021469116,
+      "learning_rate": 3.163481649893263e-06,
+      "loss": 0.6217,
+      "step": 38764
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.1430466175079346,
+      "learning_rate": 3.163236423691358e-06,
+      "loss": 0.6095,
+      "step": 38765
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.8987369537353516,
+      "learning_rate": 3.1629912025968055e-06,
+      "loss": 0.767,
+      "step": 38766
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.0531744956970215,
+      "learning_rate": 3.162745986610286e-06,
+      "loss": 0.6028,
+      "step": 38767
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.7464996576309204,
+      "learning_rate": 3.1625007757324812e-06,
+      "loss": 0.4294,
+      "step": 38768
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.6083924770355225,
+      "learning_rate": 3.1622555699640737e-06,
+      "loss": 0.513,
+      "step": 38769
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.801302433013916,
+      "learning_rate": 3.1620103693057452e-06,
+      "loss": 0.566,
+      "step": 38770
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.039964199066162,
+      "learning_rate": 3.161765173758177e-06,
+      "loss": 0.7595,
+      "step": 38771
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.918222188949585,
+      "learning_rate": 3.161519983322051e-06,
+      "loss": 0.7965,
+      "step": 38772
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.9240614175796509,
+      "learning_rate": 3.1612747979980497e-06,
+      "loss": 0.4501,
+      "step": 38773
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 7.167972564697266,
+      "learning_rate": 3.161029617786854e-06,
+      "loss": 0.555,
+      "step": 38774
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.9593662023544312,
+      "learning_rate": 3.1607844426891464e-06,
+      "loss": 0.6541,
+      "step": 38775
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.4714105129241943,
+      "learning_rate": 3.1605392727056083e-06,
+      "loss": 0.5742,
+      "step": 38776
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 5.167129993438721,
+      "learning_rate": 3.1602941078369214e-06,
+      "loss": 0.4705,
+      "step": 38777
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.9851973056793213,
+      "learning_rate": 3.1600489480837672e-06,
+      "loss": 0.5771,
+      "step": 38778
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.125967502593994,
+      "learning_rate": 3.159803793446827e-06,
+      "loss": 0.5013,
+      "step": 38779
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.846757173538208,
+      "learning_rate": 3.159558643926784e-06,
+      "loss": 0.8599,
+      "step": 38780
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.156442880630493,
+      "learning_rate": 3.1593134995243192e-06,
+      "loss": 0.609,
+      "step": 38781
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.1640076637268066,
+      "learning_rate": 3.159068360240113e-06,
+      "loss": 0.6291,
+      "step": 38782
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.5292372703552246,
+      "learning_rate": 3.158823226074848e-06,
+      "loss": 0.3918,
+      "step": 38783
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.838028907775879,
+      "learning_rate": 3.1585780970292067e-06,
+      "loss": 0.5879,
+      "step": 38784
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 4.873073101043701,
+      "learning_rate": 3.15833297310387e-06,
+      "loss": 0.4329,
+      "step": 38785
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 4.684968948364258,
+      "learning_rate": 3.1580878542995186e-06,
+      "loss": 0.6924,
+      "step": 38786
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.246431827545166,
+      "learning_rate": 3.1578427406168353e-06,
+      "loss": 0.6226,
+      "step": 38787
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.2740285396575928,
+      "learning_rate": 3.1575976320565017e-06,
+      "loss": 0.7338,
+      "step": 38788
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.6972229480743408,
+      "learning_rate": 3.1573525286191987e-06,
+      "loss": 0.3765,
+      "step": 38789
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.4518682956695557,
+      "learning_rate": 3.157107430305608e-06,
+      "loss": 0.6758,
+      "step": 38790
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.4214611053466797,
+      "learning_rate": 3.1568623371164114e-06,
+      "loss": 0.69,
+      "step": 38791
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.226142406463623,
+      "learning_rate": 3.15661724905229e-06,
+      "loss": 0.4535,
+      "step": 38792
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 5.377886772155762,
+      "learning_rate": 3.1563721661139255e-06,
+      "loss": 0.4838,
+      "step": 38793
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.1069304943084717,
+      "learning_rate": 3.156127088301999e-06,
+      "loss": 0.4996,
+      "step": 38794
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.086069107055664,
+      "learning_rate": 3.1558820156171933e-06,
+      "loss": 0.6889,
+      "step": 38795
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.4961252212524414,
+      "learning_rate": 3.1556369480601877e-06,
+      "loss": 0.5219,
+      "step": 38796
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.321981906890869,
+      "learning_rate": 3.1553918856316655e-06,
+      "loss": 0.7416,
+      "step": 38797
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.083866596221924,
+      "learning_rate": 3.1551468283323072e-06,
+      "loss": 0.5906,
+      "step": 38798
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.11942195892334,
+      "learning_rate": 3.154901776162795e-06,
+      "loss": 0.8223,
+      "step": 38799
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.1265077590942383,
+      "learning_rate": 3.154656729123807e-06,
+      "loss": 0.6855,
+      "step": 38800
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.8997974395751953,
+      "learning_rate": 3.1544116872160302e-06,
+      "loss": 0.7042,
+      "step": 38801
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.728489637374878,
+      "learning_rate": 3.1541666504401426e-06,
+      "loss": 0.5231,
+      "step": 38802
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 4.12200927734375,
+      "learning_rate": 3.1539216187968258e-06,
+      "loss": 0.9437,
+      "step": 38803
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.2951931953430176,
+      "learning_rate": 3.153676592286762e-06,
+      "loss": 0.8661,
+      "step": 38804
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.4207282066345215,
+      "learning_rate": 3.1534315709106317e-06,
+      "loss": 0.6955,
+      "step": 38805
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.937344551086426,
+      "learning_rate": 3.1531865546691165e-06,
+      "loss": 0.5952,
+      "step": 38806
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.699054002761841,
+      "learning_rate": 3.1529415435628974e-06,
+      "loss": 0.6284,
+      "step": 38807
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.684603214263916,
+      "learning_rate": 3.1526965375926566e-06,
+      "loss": 0.7013,
+      "step": 38808
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.9939898252487183,
+      "learning_rate": 3.152451536759074e-06,
+      "loss": 0.6187,
+      "step": 38809
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.2509963512420654,
+      "learning_rate": 3.152206541062832e-06,
+      "loss": 0.776,
+      "step": 38810
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.22340726852417,
+      "learning_rate": 3.1519615505046107e-06,
+      "loss": 0.5799,
+      "step": 38811
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.0196728706359863,
+      "learning_rate": 3.1517165650850934e-06,
+      "loss": 0.5271,
+      "step": 38812
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.94415283203125,
+      "learning_rate": 3.151471584804959e-06,
+      "loss": 0.6585,
+      "step": 38813
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.8714269399642944,
+      "learning_rate": 3.15122660966489e-06,
+      "loss": 0.6448,
+      "step": 38814
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.361743688583374,
+      "learning_rate": 3.1509816396655667e-06,
+      "loss": 0.4789,
+      "step": 38815
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.8094186782836914,
+      "learning_rate": 3.1507366748076713e-06,
+      "loss": 0.665,
+      "step": 38816
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.6519105434417725,
+      "learning_rate": 3.1504917150918847e-06,
+      "loss": 0.5187,
+      "step": 38817
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.325438976287842,
+      "learning_rate": 3.150246760518887e-06,
+      "loss": 0.3918,
+      "step": 38818
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.410127878189087,
+      "learning_rate": 3.1500018110893605e-06,
+      "loss": 0.3681,
+      "step": 38819
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.6778969764709473,
+      "learning_rate": 3.1497568668039837e-06,
+      "loss": 0.6118,
+      "step": 38820
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.0880675315856934,
+      "learning_rate": 3.149511927663442e-06,
+      "loss": 0.5308,
+      "step": 38821
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.693737506866455,
+      "learning_rate": 3.149266993668415e-06,
+      "loss": 0.7041,
+      "step": 38822
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.353571891784668,
+      "learning_rate": 3.1490220648195826e-06,
+      "loss": 0.6466,
+      "step": 38823
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.4458799362182617,
+      "learning_rate": 3.1487771411176273e-06,
+      "loss": 0.688,
+      "step": 38824
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.726633310317993,
+      "learning_rate": 3.148532222563228e-06,
+      "loss": 0.6761,
+      "step": 38825
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.99184250831604,
+      "learning_rate": 3.1482873091570675e-06,
+      "loss": 0.6903,
+      "step": 38826
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.1786680221557617,
+      "learning_rate": 3.1480424008998266e-06,
+      "loss": 0.7078,
+      "step": 38827
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.9274632930755615,
+      "learning_rate": 3.147797497792186e-06,
+      "loss": 0.5361,
+      "step": 38828
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.2845418453216553,
+      "learning_rate": 3.1475525998348266e-06,
+      "loss": 0.5023,
+      "step": 38829
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.8157488107681274,
+      "learning_rate": 3.14730770702843e-06,
+      "loss": 0.7633,
+      "step": 38830
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.1489219665527344,
+      "learning_rate": 3.1470628193736756e-06,
+      "loss": 0.7165,
+      "step": 38831
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.740290641784668,
+      "learning_rate": 3.1468179368712457e-06,
+      "loss": 0.6155,
+      "step": 38832
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.1126036643981934,
+      "learning_rate": 3.1465730595218204e-06,
+      "loss": 0.4757,
+      "step": 38833
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.492192268371582,
+      "learning_rate": 3.146328187326082e-06,
+      "loss": 0.5967,
+      "step": 38834
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.3115365505218506,
+      "learning_rate": 3.14608332028471e-06,
+      "loss": 0.431,
+      "step": 38835
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.7378731966018677,
+      "learning_rate": 3.145838458398386e-06,
+      "loss": 0.387,
+      "step": 38836
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.26969575881958,
+      "learning_rate": 3.1455936016677914e-06,
+      "loss": 0.6212,
+      "step": 38837
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.963334798812866,
+      "learning_rate": 3.1453487500936054e-06,
+      "loss": 0.5689,
+      "step": 38838
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.5088372230529785,
+      "learning_rate": 3.1451039036765096e-06,
+      "loss": 0.56,
+      "step": 38839
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.8160741329193115,
+      "learning_rate": 3.144859062417185e-06,
+      "loss": 0.679,
+      "step": 38840
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.1674768924713135,
+      "learning_rate": 3.1446142263163127e-06,
+      "loss": 0.7891,
+      "step": 38841
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.1590778827667236,
+      "learning_rate": 3.1443693953745736e-06,
+      "loss": 0.7738,
+      "step": 38842
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.0749809741973877,
+      "learning_rate": 3.1441245695926475e-06,
+      "loss": 0.4882,
+      "step": 38843
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.903027057647705,
+      "learning_rate": 3.1438797489712146e-06,
+      "loss": 0.4214,
+      "step": 38844
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.150585412979126,
+      "learning_rate": 3.1436349335109585e-06,
+      "loss": 0.5869,
+      "step": 38845
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.7127094268798828,
+      "learning_rate": 3.143390123212558e-06,
+      "loss": 0.7619,
+      "step": 38846
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.4014010429382324,
+      "learning_rate": 3.143145318076694e-06,
+      "loss": 0.5421,
+      "step": 38847
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.0072152614593506,
+      "learning_rate": 3.1429005181040474e-06,
+      "loss": 0.4492,
+      "step": 38848
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.9763731956481934,
+      "learning_rate": 3.1426557232952986e-06,
+      "loss": 0.6848,
+      "step": 38849
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.432408571243286,
+      "learning_rate": 3.142410933651129e-06,
+      "loss": 0.5507,
+      "step": 38850
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.204948902130127,
+      "learning_rate": 3.1421661491722176e-06,
+      "loss": 0.6339,
+      "step": 38851
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.8555290699005127,
+      "learning_rate": 3.141921369859247e-06,
+      "loss": 0.8479,
+      "step": 38852
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 4.232195854187012,
+      "learning_rate": 3.141676595712897e-06,
+      "loss": 0.8202,
+      "step": 38853
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.7559261322021484,
+      "learning_rate": 3.1414318267338483e-06,
+      "loss": 0.5071,
+      "step": 38854
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.945643901824951,
+      "learning_rate": 3.1411870629227814e-06,
+      "loss": 0.7994,
+      "step": 38855
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.0811767578125,
+      "learning_rate": 3.140942304280377e-06,
+      "loss": 0.5935,
+      "step": 38856
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.3175718784332275,
+      "learning_rate": 3.1406975508073152e-06,
+      "loss": 0.7633,
+      "step": 38857
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.887512683868408,
+      "learning_rate": 3.1404528025042768e-06,
+      "loss": 0.4702,
+      "step": 38858
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.9221551418304443,
+      "learning_rate": 3.140208059371943e-06,
+      "loss": 0.7098,
+      "step": 38859
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 4.60324239730835,
+      "learning_rate": 3.139963321410994e-06,
+      "loss": 0.6155,
+      "step": 38860
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.9017683267593384,
+      "learning_rate": 3.1397185886221083e-06,
+      "loss": 0.8015,
+      "step": 38861
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.596010684967041,
+      "learning_rate": 3.1394738610059704e-06,
+      "loss": 1.0087,
+      "step": 38862
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.065479040145874,
+      "learning_rate": 3.1392291385632583e-06,
+      "loss": 0.5447,
+      "step": 38863
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.8426669836044312,
+      "learning_rate": 3.138984421294653e-06,
+      "loss": 0.6125,
+      "step": 38864
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.735421061515808,
+      "learning_rate": 3.1387397092008354e-06,
+      "loss": 0.5818,
+      "step": 38865
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.1371262073516846,
+      "learning_rate": 3.1384950022824846e-06,
+      "loss": 0.6802,
+      "step": 38866
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.7705943584442139,
+      "learning_rate": 3.138250300540282e-06,
+      "loss": 0.5064,
+      "step": 38867
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.2753539085388184,
+      "learning_rate": 3.1380056039749086e-06,
+      "loss": 0.8257,
+      "step": 38868
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.8642003536224365,
+      "learning_rate": 3.1377609125870433e-06,
+      "loss": 0.8797,
+      "step": 38869
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 5.076835632324219,
+      "learning_rate": 3.1375162263773674e-06,
+      "loss": 0.6993,
+      "step": 38870
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.959674835205078,
+      "learning_rate": 3.137271545346562e-06,
+      "loss": 0.6736,
+      "step": 38871
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.365868091583252,
+      "learning_rate": 3.1370268694953056e-06,
+      "loss": 0.6565,
+      "step": 38872
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.7114825248718262,
+      "learning_rate": 3.1367821988242797e-06,
+      "loss": 0.523,
+      "step": 38873
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.236234188079834,
+      "learning_rate": 3.136537533334165e-06,
+      "loss": 0.6139,
+      "step": 38874
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.911304235458374,
+      "learning_rate": 3.1362928730256416e-06,
+      "loss": 0.7971,
+      "step": 38875
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.446429491043091,
+      "learning_rate": 3.136048217899389e-06,
+      "loss": 0.557,
+      "step": 38876
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.185241222381592,
+      "learning_rate": 3.1358035679560885e-06,
+      "loss": 0.6871,
+      "step": 38877
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.4951739311218262,
+      "learning_rate": 3.135558923196419e-06,
+      "loss": 0.2693,
+      "step": 38878
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.328599214553833,
+      "learning_rate": 3.1353142836210625e-06,
+      "loss": 0.7128,
+      "step": 38879
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.215047597885132,
+      "learning_rate": 3.1350696492306985e-06,
+      "loss": 0.4552,
+      "step": 38880
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 4.635457992553711,
+      "learning_rate": 3.134825020026005e-06,
+      "loss": 0.8314,
+      "step": 38881
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.054715394973755,
+      "learning_rate": 3.134580396007666e-06,
+      "loss": 0.6939,
+      "step": 38882
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 4.362473964691162,
+      "learning_rate": 3.1343357771763606e-06,
+      "loss": 0.5432,
+      "step": 38883
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.6468312740325928,
+      "learning_rate": 3.1340911635327685e-06,
+      "loss": 0.406,
+      "step": 38884
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.934523105621338,
+      "learning_rate": 3.13384655507757e-06,
+      "loss": 0.4652,
+      "step": 38885
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 4.494950771331787,
+      "learning_rate": 3.133601951811445e-06,
+      "loss": 0.6759,
+      "step": 38886
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.2721118927001953,
+      "learning_rate": 3.1333573537350735e-06,
+      "loss": 0.5662,
+      "step": 38887
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.2968366146087646,
+      "learning_rate": 3.1331127608491364e-06,
+      "loss": 0.5384,
+      "step": 38888
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 5.622738361358643,
+      "learning_rate": 3.132868173154313e-06,
+      "loss": 0.7819,
+      "step": 38889
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.4503564834594727,
+      "learning_rate": 3.1326235906512837e-06,
+      "loss": 0.6884,
+      "step": 38890
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.4330623149871826,
+      "learning_rate": 3.1323790133407294e-06,
+      "loss": 0.6535,
+      "step": 38891
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 5.396477222442627,
+      "learning_rate": 3.132134441223329e-06,
+      "loss": 0.6761,
+      "step": 38892
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.695486307144165,
+      "learning_rate": 3.1318898742997628e-06,
+      "loss": 0.7563,
+      "step": 38893
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.1153790950775146,
+      "learning_rate": 3.1316453125707103e-06,
+      "loss": 0.6164,
+      "step": 38894
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.9897119998931885,
+      "learning_rate": 3.131400756036853e-06,
+      "loss": 0.8339,
+      "step": 38895
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.446626663208008,
+      "learning_rate": 3.1311562046988697e-06,
+      "loss": 0.5519,
+      "step": 38896
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.0204098224639893,
+      "learning_rate": 3.130911658557442e-06,
+      "loss": 0.6192,
+      "step": 38897
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.3053367137908936,
+      "learning_rate": 3.1306671176132485e-06,
+      "loss": 0.5583,
+      "step": 38898
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.450448751449585,
+      "learning_rate": 3.1304225818669693e-06,
+      "loss": 0.7965,
+      "step": 38899
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.6915167570114136,
+      "learning_rate": 3.1301780513192846e-06,
+      "loss": 0.3445,
+      "step": 38900
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.890939950942993,
+      "learning_rate": 3.129933525970873e-06,
+      "loss": 0.7633,
+      "step": 38901
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.1625235080718994,
+      "learning_rate": 3.129689005822417e-06,
+      "loss": 0.6287,
+      "step": 38902
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.0068159103393555,
+      "learning_rate": 3.1294444908745953e-06,
+      "loss": 0.585,
+      "step": 38903
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.4545066356658936,
+      "learning_rate": 3.1291999811280878e-06,
+      "loss": 0.1832,
+      "step": 38904
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.5341672897338867,
+      "learning_rate": 3.128955476583574e-06,
+      "loss": 0.3765,
+      "step": 38905
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.6770172119140625,
+      "learning_rate": 3.1287109772417336e-06,
+      "loss": 0.7963,
+      "step": 38906
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.299438714981079,
+      "learning_rate": 3.1284664831032472e-06,
+      "loss": 0.5795,
+      "step": 38907
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.9323463439941406,
+      "learning_rate": 3.1282219941687954e-06,
+      "loss": 0.7081,
+      "step": 38908
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.4954254627227783,
+      "learning_rate": 3.1279775104390565e-06,
+      "loss": 0.8048,
+      "step": 38909
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.5021047592163086,
+      "learning_rate": 3.127733031914711e-06,
+      "loss": 0.4025,
+      "step": 38910
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.356720447540283,
+      "learning_rate": 3.127488558596438e-06,
+      "loss": 0.627,
+      "step": 38911
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.9486827850341797,
+      "learning_rate": 3.1272440904849187e-06,
+      "loss": 0.555,
+      "step": 38912
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.951572895050049,
+      "learning_rate": 3.1269996275808316e-06,
+      "loss": 0.7059,
+      "step": 38913
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 4.632079601287842,
+      "learning_rate": 3.126755169884857e-06,
+      "loss": 0.3452,
+      "step": 38914
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.285355806350708,
+      "learning_rate": 3.126510717397675e-06,
+      "loss": 0.5521,
+      "step": 38915
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.9139344692230225,
+      "learning_rate": 3.1262662701199643e-06,
+      "loss": 0.6759,
+      "step": 38916
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.6944546699523926,
+      "learning_rate": 3.1260218280524044e-06,
+      "loss": 0.673,
+      "step": 38917
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.2413928508758545,
+      "learning_rate": 3.1257773911956774e-06,
+      "loss": 0.3393,
+      "step": 38918
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.2557621002197266,
+      "learning_rate": 3.1255329595504602e-06,
+      "loss": 0.5698,
+      "step": 38919
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.7341904640197754,
+      "learning_rate": 3.125288533117434e-06,
+      "loss": 0.7098,
+      "step": 38920
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.178956985473633,
+      "learning_rate": 3.1250441118972787e-06,
+      "loss": 0.5914,
+      "step": 38921
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.093794345855713,
+      "learning_rate": 3.124799695890671e-06,
+      "loss": 0.3345,
+      "step": 38922
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.8324947357177734,
+      "learning_rate": 3.124555285098294e-06,
+      "loss": 0.319,
+      "step": 38923
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.313126564025879,
+      "learning_rate": 3.1243108795208275e-06,
+      "loss": 0.6699,
+      "step": 38924
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.6593793630599976,
+      "learning_rate": 3.124066479158949e-06,
+      "loss": 0.5919,
+      "step": 38925
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.3470377922058105,
+      "learning_rate": 3.123822084013339e-06,
+      "loss": 0.2655,
+      "step": 38926
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.330165147781372,
+      "learning_rate": 3.1235776940846773e-06,
+      "loss": 0.6563,
+      "step": 38927
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 4.772824764251709,
+      "learning_rate": 3.123333309373643e-06,
+      "loss": 0.6546,
+      "step": 38928
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.0427021980285645,
+      "learning_rate": 3.1230889298809153e-06,
+      "loss": 0.5539,
+      "step": 38929
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.2437853813171387,
+      "learning_rate": 3.1228445556071747e-06,
+      "loss": 0.4178,
+      "step": 38930
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.1257402896881104,
+      "learning_rate": 3.1226001865531e-06,
+      "loss": 0.8341,
+      "step": 38931
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.446744918823242,
+      "learning_rate": 3.1223558227193707e-06,
+      "loss": 0.7062,
+      "step": 38932
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.1700854301452637,
+      "learning_rate": 3.1221114641066662e-06,
+      "loss": 0.8249,
+      "step": 38933
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 4.790503025054932,
+      "learning_rate": 3.121867110715667e-06,
+      "loss": 0.8628,
+      "step": 38934
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.5777361392974854,
+      "learning_rate": 3.121622762547052e-06,
+      "loss": 0.6908,
+      "step": 38935
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.1552019119262695,
+      "learning_rate": 3.1213784196014997e-06,
+      "loss": 0.6702,
+      "step": 38936
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.4168319702148438,
+      "learning_rate": 3.1211340818796902e-06,
+      "loss": 0.5333,
+      "step": 38937
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.9973396062850952,
+      "learning_rate": 3.1208897493823033e-06,
+      "loss": 0.4323,
+      "step": 38938
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.3502235412597656,
+      "learning_rate": 3.1206454221100178e-06,
+      "loss": 0.5777,
+      "step": 38939
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 4.522439479827881,
+      "learning_rate": 3.120401100063514e-06,
+      "loss": 0.9443,
+      "step": 38940
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.2042336463928223,
+      "learning_rate": 3.1201567832434706e-06,
+      "loss": 0.5265,
+      "step": 38941
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.5727579593658447,
+      "learning_rate": 3.1199124716505645e-06,
+      "loss": 0.7852,
+      "step": 38942
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.8719305992126465,
+      "learning_rate": 3.1196681652854793e-06,
+      "loss": 0.483,
+      "step": 38943
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.0019147396087646,
+      "learning_rate": 3.1194238641488926e-06,
+      "loss": 0.5814,
+      "step": 38944
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 6.160887718200684,
+      "learning_rate": 3.1191795682414846e-06,
+      "loss": 0.4633,
+      "step": 38945
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.5102782249450684,
+      "learning_rate": 3.1189352775639324e-06,
+      "loss": 0.6643,
+      "step": 38946
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.7846221923828125,
+      "learning_rate": 3.1186909921169175e-06,
+      "loss": 0.3554,
+      "step": 38947
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.7918310165405273,
+      "learning_rate": 3.1184467119011175e-06,
+      "loss": 0.5582,
+      "step": 38948
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.3367817401885986,
+      "learning_rate": 3.118202436917213e-06,
+      "loss": 0.3847,
+      "step": 38949
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.682842493057251,
+      "learning_rate": 3.117958167165882e-06,
+      "loss": 0.5656,
+      "step": 38950
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.0882575511932373,
+      "learning_rate": 3.117713902647805e-06,
+      "loss": 0.5776,
+      "step": 38951
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.078418493270874,
+      "learning_rate": 3.11746964336366e-06,
+      "loss": 0.7073,
+      "step": 38952
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.830507278442383,
+      "learning_rate": 3.117225389314127e-06,
+      "loss": 0.6917,
+      "step": 38953
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.957106113433838,
+      "learning_rate": 3.116981140499884e-06,
+      "loss": 0.6315,
+      "step": 38954
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.191140651702881,
+      "learning_rate": 3.1167368969216123e-06,
+      "loss": 0.8295,
+      "step": 38955
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.9220092296600342,
+      "learning_rate": 3.116492658579988e-06,
+      "loss": 0.5985,
+      "step": 38956
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.5995450019836426,
+      "learning_rate": 3.116248425475693e-06,
+      "loss": 0.4181,
+      "step": 38957
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 11.48198413848877,
+      "learning_rate": 3.1160041976094057e-06,
+      "loss": 0.7122,
+      "step": 38958
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.6935198307037354,
+      "learning_rate": 3.115759974981805e-06,
+      "loss": 0.7494,
+      "step": 38959
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.0743603706359863,
+      "learning_rate": 3.11551575759357e-06,
+      "loss": 0.8027,
+      "step": 38960
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.183833360671997,
+      "learning_rate": 3.1152715454453797e-06,
+      "loss": 0.465,
+      "step": 38961
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.2715675830841064,
+      "learning_rate": 3.1150273385379126e-06,
+      "loss": 0.629,
+      "step": 38962
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 4.114660739898682,
+      "learning_rate": 3.1147831368718485e-06,
+      "loss": 0.826,
+      "step": 38963
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.869229555130005,
+      "learning_rate": 3.1145389404478663e-06,
+      "loss": 0.6704,
+      "step": 38964
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.5696604251861572,
+      "learning_rate": 3.1142947492666463e-06,
+      "loss": 0.7058,
+      "step": 38965
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.062331199645996,
+      "learning_rate": 3.1140505633288655e-06,
+      "loss": 0.3747,
+      "step": 38966
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.9762847423553467,
+      "learning_rate": 3.113806382635203e-06,
+      "loss": 0.5561,
+      "step": 38967
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.1054108142852783,
+      "learning_rate": 3.113562207186338e-06,
+      "loss": 0.5473,
+      "step": 38968
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 4.245697021484375,
+      "learning_rate": 3.11331803698295e-06,
+      "loss": 0.4125,
+      "step": 38969
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.2449135780334473,
+      "learning_rate": 3.1130738720257184e-06,
+      "loss": 0.6879,
+      "step": 38970
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 4.130159854888916,
+      "learning_rate": 3.112829712315322e-06,
+      "loss": 0.5015,
+      "step": 38971
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 5.165782451629639,
+      "learning_rate": 3.1125855578524384e-06,
+      "loss": 0.6698,
+      "step": 38972
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.1932998895645142,
+      "learning_rate": 3.1123414086377477e-06,
+      "loss": 0.3082,
+      "step": 38973
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.213066577911377,
+      "learning_rate": 3.1120972646719284e-06,
+      "loss": 0.6391,
+      "step": 38974
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.5355820655822754,
+      "learning_rate": 3.1118531259556593e-06,
+      "loss": 0.5259,
+      "step": 38975
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.0192480087280273,
+      "learning_rate": 3.1116089924896193e-06,
+      "loss": 0.4561,
+      "step": 38976
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.0041842460632324,
+      "learning_rate": 3.111364864274487e-06,
+      "loss": 0.5973,
+      "step": 38977
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.991026520729065,
+      "learning_rate": 3.1111207413109423e-06,
+      "loss": 0.8255,
+      "step": 38978
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.8871294260025024,
+      "learning_rate": 3.1108766235996622e-06,
+      "loss": 0.732,
+      "step": 38979
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.767524480819702,
+      "learning_rate": 3.1106325111413272e-06,
+      "loss": 0.4986,
+      "step": 38980
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.7670598030090332,
+      "learning_rate": 3.110388403936615e-06,
+      "loss": 0.5929,
+      "step": 38981
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.5930590629577637,
+      "learning_rate": 3.1101443019862053e-06,
+      "loss": 0.5457,
+      "step": 38982
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.0394623279571533,
+      "learning_rate": 3.1099002052907744e-06,
+      "loss": 0.6827,
+      "step": 38983
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.9179770946502686,
+      "learning_rate": 3.1096561138510047e-06,
+      "loss": 0.6288,
+      "step": 38984
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.4833736419677734,
+      "learning_rate": 3.1094120276675733e-06,
+      "loss": 0.6594,
+      "step": 38985
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.8595306873321533,
+      "learning_rate": 3.1091679467411583e-06,
+      "loss": 0.4944,
+      "step": 38986
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.9325050115585327,
+      "learning_rate": 3.1089238710724394e-06,
+      "loss": 0.493,
+      "step": 38987
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.8628990650177,
+      "learning_rate": 3.108679800662095e-06,
+      "loss": 0.5986,
+      "step": 38988
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.3916749954223633,
+      "learning_rate": 3.108435735510803e-06,
+      "loss": 0.7676,
+      "step": 38989
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.880063772201538,
+      "learning_rate": 3.108191675619243e-06,
+      "loss": 0.815,
+      "step": 38990
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 4.128800868988037,
+      "learning_rate": 3.107947620988093e-06,
+      "loss": 0.705,
+      "step": 38991
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.3332149982452393,
+      "learning_rate": 3.1077035716180328e-06,
+      "loss": 0.7072,
+      "step": 38992
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.0698695182800293,
+      "learning_rate": 3.107459527509739e-06,
+      "loss": 0.2782,
+      "step": 38993
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.42720103263855,
+      "learning_rate": 3.1072154886638917e-06,
+      "loss": 0.54,
+      "step": 38994
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.2355868816375732,
+      "learning_rate": 3.106971455081169e-06,
+      "loss": 0.588,
+      "step": 38995
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.2147440910339355,
+      "learning_rate": 3.10672742676225e-06,
+      "loss": 0.5263,
+      "step": 38996
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.923584461212158,
+      "learning_rate": 3.1064834037078124e-06,
+      "loss": 0.4921,
+      "step": 38997
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.6161364316940308,
+      "learning_rate": 3.106239385918535e-06,
+      "loss": 0.7707,
+      "step": 38998
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.4935381412506104,
+      "learning_rate": 3.1059953733950964e-06,
+      "loss": 0.577,
+      "step": 38999
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.3522958755493164,
+      "learning_rate": 3.1057513661381756e-06,
+      "loss": 0.4945,
+      "step": 39000
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.091693162918091,
+      "learning_rate": 3.10550736414845e-06,
+      "loss": 0.7102,
+      "step": 39001
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.1931653022766113,
+      "learning_rate": 3.105263367426599e-06,
+      "loss": 0.678,
+      "step": 39002
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.361546516418457,
+      "learning_rate": 3.1050193759732993e-06,
+      "loss": 0.6785,
+      "step": 39003
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 7.283430099487305,
+      "learning_rate": 3.1047753897892317e-06,
+      "loss": 0.5417,
+      "step": 39004
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.1972155570983887,
+      "learning_rate": 3.104531408875075e-06,
+      "loss": 0.5351,
+      "step": 39005
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.8359594345092773,
+      "learning_rate": 3.104287433231506e-06,
+      "loss": 0.5982,
+      "step": 39006
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.5075926780700684,
+      "learning_rate": 3.1040434628592033e-06,
+      "loss": 0.4302,
+      "step": 39007
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.062333822250366,
+      "learning_rate": 3.103799497758846e-06,
+      "loss": 0.6413,
+      "step": 39008
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.81556236743927,
+      "learning_rate": 3.103555537931111e-06,
+      "loss": 0.4767,
+      "step": 39009
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.392127752304077,
+      "learning_rate": 3.103311583376678e-06,
+      "loss": 0.6017,
+      "step": 39010
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.098012924194336,
+      "learning_rate": 3.103067634096225e-06,
+      "loss": 0.6266,
+      "step": 39011
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.759854316711426,
+      "learning_rate": 3.102823690090431e-06,
+      "loss": 0.6546,
+      "step": 39012
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.2161946296691895,
+      "learning_rate": 3.102579751359973e-06,
+      "loss": 0.4025,
+      "step": 39013
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.6115870475769043,
+      "learning_rate": 3.1023358179055296e-06,
+      "loss": 0.6003,
+      "step": 39014
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.911929726600647,
+      "learning_rate": 3.10209188972778e-06,
+      "loss": 0.2987,
+      "step": 39015
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 6.0732221603393555,
+      "learning_rate": 3.1018479668274015e-06,
+      "loss": 0.6757,
+      "step": 39016
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.540256977081299,
+      "learning_rate": 3.1016040492050737e-06,
+      "loss": 0.5792,
+      "step": 39017
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.354679584503174,
+      "learning_rate": 3.1013601368614714e-06,
+      "loss": 0.6349,
+      "step": 39018
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.1981594562530518,
+      "learning_rate": 3.1011162297972775e-06,
+      "loss": 0.4837,
+      "step": 39019
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.0833020210266113,
+      "learning_rate": 3.100872328013167e-06,
+      "loss": 0.7576,
+      "step": 39020
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.8800537586212158,
+      "learning_rate": 3.10062843150982e-06,
+      "loss": 0.6431,
+      "step": 39021
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 12.990979194641113,
+      "learning_rate": 3.1003845402879135e-06,
+      "loss": 0.5073,
+      "step": 39022
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.7067534923553467,
+      "learning_rate": 3.100140654348125e-06,
+      "loss": 0.6337,
+      "step": 39023
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.8254506587982178,
+      "learning_rate": 3.0998967736911346e-06,
+      "loss": 0.8662,
+      "step": 39024
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.6277215480804443,
+      "learning_rate": 3.0996528983176193e-06,
+      "loss": 0.8687,
+      "step": 39025
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.2964975833892822,
+      "learning_rate": 3.0994090282282572e-06,
+      "loss": 0.6204,
+      "step": 39026
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.5240583419799805,
+      "learning_rate": 3.0991651634237272e-06,
+      "loss": 0.484,
+      "step": 39027
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.2949955463409424,
+      "learning_rate": 3.0989213039047065e-06,
+      "loss": 0.5455,
+      "step": 39028
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.1638565063476562,
+      "learning_rate": 3.098677449671873e-06,
+      "loss": 0.6068,
+      "step": 39029
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.3320703506469727,
+      "learning_rate": 3.0984336007259053e-06,
+      "loss": 0.6193,
+      "step": 39030
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.8687329292297363,
+      "learning_rate": 3.0981897570674825e-06,
+      "loss": 0.4988,
+      "step": 39031
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.036620855331421,
+      "learning_rate": 3.0979459186972804e-06,
+      "loss": 0.7287,
+      "step": 39032
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.028444290161133,
+      "learning_rate": 3.097702085615979e-06,
+      "loss": 0.6722,
+      "step": 39033
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.902313709259033,
+      "learning_rate": 3.097458257824255e-06,
+      "loss": 0.6442,
+      "step": 39034
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.942796468734741,
+      "learning_rate": 3.0972144353227874e-06,
+      "loss": 0.8005,
+      "step": 39035
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.863281011581421,
+      "learning_rate": 3.096970618112254e-06,
+      "loss": 0.6079,
+      "step": 39036
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.7565516233444214,
+      "learning_rate": 3.0967268061933314e-06,
+      "loss": 0.5037,
+      "step": 39037
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 4.964381217956543,
+      "learning_rate": 3.096482999566699e-06,
+      "loss": 0.611,
+      "step": 39038
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.7159411907196045,
+      "learning_rate": 3.0962391982330342e-06,
+      "loss": 0.4673,
+      "step": 39039
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.19319224357605,
+      "learning_rate": 3.095995402193015e-06,
+      "loss": 0.6707,
+      "step": 39040
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 5.2381696701049805,
+      "learning_rate": 3.0957516114473198e-06,
+      "loss": 0.5108,
+      "step": 39041
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.1897284984588623,
+      "learning_rate": 3.0955078259966255e-06,
+      "loss": 0.5463,
+      "step": 39042
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 4.445154190063477,
+      "learning_rate": 3.0952640458416104e-06,
+      "loss": 0.6578,
+      "step": 39043
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 4.531074047088623,
+      "learning_rate": 3.0950202709829513e-06,
+      "loss": 0.7617,
+      "step": 39044
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.89665150642395,
+      "learning_rate": 3.0947765014213284e-06,
+      "loss": 0.709,
+      "step": 39045
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.857661247253418,
+      "learning_rate": 3.0945327371574184e-06,
+      "loss": 0.5584,
+      "step": 39046
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.0799782276153564,
+      "learning_rate": 3.0942889781918993e-06,
+      "loss": 0.7195,
+      "step": 39047
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.0643954277038574,
+      "learning_rate": 3.094045224525449e-06,
+      "loss": 0.9308,
+      "step": 39048
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.829763650894165,
+      "learning_rate": 3.0938014761587443e-06,
+      "loss": 0.6426,
+      "step": 39049
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.357861280441284,
+      "learning_rate": 3.0935577330924636e-06,
+      "loss": 0.4743,
+      "step": 39050
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.4029109477996826,
+      "learning_rate": 3.093313995327285e-06,
+      "loss": 0.684,
+      "step": 39051
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.783949136734009,
+      "learning_rate": 3.093070262863886e-06,
+      "loss": 0.689,
+      "step": 39052
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.2506518363952637,
+      "learning_rate": 3.0928265357029442e-06,
+      "loss": 0.4538,
+      "step": 39053
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.2943530082702637,
+      "learning_rate": 3.0925828138451373e-06,
+      "loss": 0.8075,
+      "step": 39054
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.99711537361145,
+      "learning_rate": 3.092339097291143e-06,
+      "loss": 0.5116,
+      "step": 39055
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 4.511089324951172,
+      "learning_rate": 3.0920953860416393e-06,
+      "loss": 0.5481,
+      "step": 39056
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.856606125831604,
+      "learning_rate": 3.091851680097303e-06,
+      "loss": 0.402,
+      "step": 39057
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.9997930526733398,
+      "learning_rate": 3.0916079794588128e-06,
+      "loss": 0.646,
+      "step": 39058
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.0979392528533936,
+      "learning_rate": 3.0913642841268455e-06,
+      "loss": 0.4633,
+      "step": 39059
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.573331594467163,
+      "learning_rate": 3.0911205941020796e-06,
+      "loss": 0.6483,
+      "step": 39060
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 4.779767990112305,
+      "learning_rate": 3.090876909385192e-06,
+      "loss": 0.5775,
+      "step": 39061
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.0085597038269043,
+      "learning_rate": 3.090633229976861e-06,
+      "loss": 0.4936,
+      "step": 39062
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.9201325178146362,
+      "learning_rate": 3.090389555877763e-06,
+      "loss": 0.5243,
+      "step": 39063
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.0650506019592285,
+      "learning_rate": 3.0901458870885746e-06,
+      "loss": 0.4198,
+      "step": 39064
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.09403395652771,
+      "learning_rate": 3.089902223609977e-06,
+      "loss": 0.6285,
+      "step": 39065
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.3543055057525635,
+      "learning_rate": 3.089658565442646e-06,
+      "loss": 0.6498,
+      "step": 39066
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.27337646484375,
+      "learning_rate": 3.0894149125872593e-06,
+      "loss": 0.5162,
+      "step": 39067
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.6634578704833984,
+      "learning_rate": 3.0891712650444932e-06,
+      "loss": 0.6555,
+      "step": 39068
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.165973663330078,
+      "learning_rate": 3.0889276228150268e-06,
+      "loss": 0.3875,
+      "step": 39069
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.664307713508606,
+      "learning_rate": 3.0886839858995354e-06,
+      "loss": 0.4665,
+      "step": 39070
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.229942798614502,
+      "learning_rate": 3.088440354298699e-06,
+      "loss": 0.6164,
+      "step": 39071
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.7686337232589722,
+      "learning_rate": 3.088196728013194e-06,
+      "loss": 0.4504,
+      "step": 39072
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.3983592987060547,
+      "learning_rate": 3.0879531070436973e-06,
+      "loss": 0.7865,
+      "step": 39073
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.980712890625,
+      "learning_rate": 3.0877094913908866e-06,
+      "loss": 0.3881,
+      "step": 39074
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 4.845913887023926,
+      "learning_rate": 3.087465881055439e-06,
+      "loss": 0.6286,
+      "step": 39075
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.737408399581909,
+      "learning_rate": 3.0872222760380332e-06,
+      "loss": 0.5896,
+      "step": 39076
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.8122950792312622,
+      "learning_rate": 3.086978676339346e-06,
+      "loss": 0.4046,
+      "step": 39077
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.4578869342803955,
+      "learning_rate": 3.086735081960054e-06,
+      "loss": 0.6461,
+      "step": 39078
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 4.368860721588135,
+      "learning_rate": 3.086491492900835e-06,
+      "loss": 0.6465,
+      "step": 39079
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 5.533369064331055,
+      "learning_rate": 3.086247909162365e-06,
+      "loss": 0.7401,
+      "step": 39080
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.0436527729034424,
+      "learning_rate": 3.0860043307453238e-06,
+      "loss": 0.4357,
+      "step": 39081
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.572746515274048,
+      "learning_rate": 3.085760757650388e-06,
+      "loss": 0.6674,
+      "step": 39082
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.3182952404022217,
+      "learning_rate": 3.0855171898782338e-06,
+      "loss": 0.671,
+      "step": 39083
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.695277214050293,
+      "learning_rate": 3.085273627429538e-06,
+      "loss": 0.7554,
+      "step": 39084
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.4947545528411865,
+      "learning_rate": 3.0850300703049807e-06,
+      "loss": 0.693,
+      "step": 39085
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.6136980056762695,
+      "learning_rate": 3.084786518505237e-06,
+      "loss": 0.5017,
+      "step": 39086
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.3346269130706787,
+      "learning_rate": 3.0845429720309848e-06,
+      "loss": 0.5942,
+      "step": 39087
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.1363332271575928,
+      "learning_rate": 3.084299430882901e-06,
+      "loss": 0.6863,
+      "step": 39088
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.7949320077896118,
+      "learning_rate": 3.0840558950616628e-06,
+      "loss": 0.7546,
+      "step": 39089
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.3810179233551025,
+      "learning_rate": 3.0838123645679474e-06,
+      "loss": 0.529,
+      "step": 39090
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.837561845779419,
+      "learning_rate": 3.083568839402431e-06,
+      "loss": 0.7645,
+      "step": 39091
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.1914899349212646,
+      "learning_rate": 3.0833253195657924e-06,
+      "loss": 0.6021,
+      "step": 39092
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.159691095352173,
+      "learning_rate": 3.0830818050587084e-06,
+      "loss": 0.6804,
+      "step": 39093
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.9339659214019775,
+      "learning_rate": 3.082838295881856e-06,
+      "loss": 0.5958,
+      "step": 39094
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 4.821043968200684,
+      "learning_rate": 3.0825947920359118e-06,
+      "loss": 0.5583,
+      "step": 39095
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.005894184112549,
+      "learning_rate": 3.082351293521553e-06,
+      "loss": 0.6352,
+      "step": 39096
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 4.388575077056885,
+      "learning_rate": 3.082107800339458e-06,
+      "loss": 0.8087,
+      "step": 39097
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.7514936923980713,
+      "learning_rate": 3.0818643124903014e-06,
+      "loss": 0.7129,
+      "step": 39098
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.360039472579956,
+      "learning_rate": 3.081620829974762e-06,
+      "loss": 0.5042,
+      "step": 39099
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.9593791961669922,
+      "learning_rate": 3.0813773527935164e-06,
+      "loss": 0.4307,
+      "step": 39100
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 4.065771102905273,
+      "learning_rate": 3.0811338809472426e-06,
+      "loss": 0.6778,
+      "step": 39101
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.4080605506896973,
+      "learning_rate": 3.080890414436616e-06,
+      "loss": 0.5216,
+      "step": 39102
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.6160712242126465,
+      "learning_rate": 3.080646953262314e-06,
+      "loss": 0.6931,
+      "step": 39103
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.993336796760559,
+      "learning_rate": 3.0804034974250137e-06,
+      "loss": 0.3833,
+      "step": 39104
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.1191515922546387,
+      "learning_rate": 3.08016004692539e-06,
+      "loss": 0.6747,
+      "step": 39105
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 9.160304069519043,
+      "learning_rate": 3.079916601764125e-06,
+      "loss": 0.4458,
+      "step": 39106
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.4136247634887695,
+      "learning_rate": 3.0796731619418916e-06,
+      "loss": 0.4854,
+      "step": 39107
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.9256420135498047,
+      "learning_rate": 3.079429727459369e-06,
+      "loss": 0.467,
+      "step": 39108
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.316141366958618,
+      "learning_rate": 3.0791862983172318e-06,
+      "loss": 0.4554,
+      "step": 39109
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.140855073928833,
+      "learning_rate": 3.0789428745161583e-06,
+      "loss": 0.4277,
+      "step": 39110
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.483555555343628,
+      "learning_rate": 3.078699456056825e-06,
+      "loss": 0.7026,
+      "step": 39111
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.2092154026031494,
+      "learning_rate": 3.0784560429399087e-06,
+      "loss": 0.5927,
+      "step": 39112
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.9450839757919312,
+      "learning_rate": 3.0782126351660864e-06,
+      "loss": 0.5877,
+      "step": 39113
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.3772122859954834,
+      "learning_rate": 3.0779692327360346e-06,
+      "loss": 0.6739,
+      "step": 39114
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.676121473312378,
+      "learning_rate": 3.0777258356504302e-06,
+      "loss": 0.6836,
+      "step": 39115
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.4412858486175537,
+      "learning_rate": 3.077482443909951e-06,
+      "loss": 0.6809,
+      "step": 39116
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.431929111480713,
+      "learning_rate": 3.0772390575152722e-06,
+      "loss": 0.8293,
+      "step": 39117
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.8121793270111084,
+      "learning_rate": 3.0769956764670717e-06,
+      "loss": 0.504,
+      "step": 39118
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.281151533126831,
+      "learning_rate": 3.076752300766026e-06,
+      "loss": 0.7722,
+      "step": 39119
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.9522204399108887,
+      "learning_rate": 3.0765089304128113e-06,
+      "loss": 0.7321,
+      "step": 39120
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 4.752688407897949,
+      "learning_rate": 3.0762655654081047e-06,
+      "loss": 0.6686,
+      "step": 39121
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 4.9623847007751465,
+      "learning_rate": 3.0760222057525835e-06,
+      "loss": 0.6109,
+      "step": 39122
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.410926580429077,
+      "learning_rate": 3.0757788514469233e-06,
+      "loss": 0.6645,
+      "step": 39123
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.460916042327881,
+      "learning_rate": 3.075535502491801e-06,
+      "loss": 0.4956,
+      "step": 39124
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.796027421951294,
+      "learning_rate": 3.0752921588878914e-06,
+      "loss": 0.5096,
+      "step": 39125
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.7223461866378784,
+      "learning_rate": 3.0750488206358764e-06,
+      "loss": 0.6462,
+      "step": 39126
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.350506067276001,
+      "learning_rate": 3.0748054877364285e-06,
+      "loss": 0.6674,
+      "step": 39127
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 4.956955432891846,
+      "learning_rate": 3.074562160190226e-06,
+      "loss": 0.7359,
+      "step": 39128
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.9557764530181885,
+      "learning_rate": 3.0743188379979446e-06,
+      "loss": 0.6612,
+      "step": 39129
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.9263787269592285,
+      "learning_rate": 3.074075521160261e-06,
+      "loss": 0.6435,
+      "step": 39130
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.2734060287475586,
+      "learning_rate": 3.0738322096778516e-06,
+      "loss": 0.7764,
+      "step": 39131
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.101588249206543,
+      "learning_rate": 3.0735889035513945e-06,
+      "loss": 0.7188,
+      "step": 39132
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 4.863804817199707,
+      "learning_rate": 3.073345602781564e-06,
+      "loss": 0.6763,
+      "step": 39133
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.5552704334259033,
+      "learning_rate": 3.0731023073690373e-06,
+      "loss": 0.7716,
+      "step": 39134
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.8404898643493652,
+      "learning_rate": 3.0728590173144925e-06,
+      "loss": 1.0236,
+      "step": 39135
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.54555606842041,
+      "learning_rate": 3.0726157326186033e-06,
+      "loss": 0.8261,
+      "step": 39136
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.5987236499786377,
+      "learning_rate": 3.0723724532820488e-06,
+      "loss": 0.676,
+      "step": 39137
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.497680902481079,
+      "learning_rate": 3.0721291793055043e-06,
+      "loss": 0.571,
+      "step": 39138
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.8100240230560303,
+      "learning_rate": 3.071885910689646e-06,
+      "loss": 0.6171,
+      "step": 39139
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.147892951965332,
+      "learning_rate": 3.0716426474351515e-06,
+      "loss": 0.6394,
+      "step": 39140
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.3603198528289795,
+      "learning_rate": 3.071399389542694e-06,
+      "loss": 0.7757,
+      "step": 39141
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.313877582550049,
+      "learning_rate": 3.071156137012955e-06,
+      "loss": 0.5782,
+      "step": 39142
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.681718111038208,
+      "learning_rate": 3.0709128898466066e-06,
+      "loss": 0.5828,
+      "step": 39143
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.708437442779541,
+      "learning_rate": 3.0706696480443283e-06,
+      "loss": 0.7843,
+      "step": 39144
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.0881359577178955,
+      "learning_rate": 3.070426411606793e-06,
+      "loss": 0.4865,
+      "step": 39145
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 7.544993877410889,
+      "learning_rate": 3.0701831805346805e-06,
+      "loss": 0.8606,
+      "step": 39146
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.7262290716171265,
+      "learning_rate": 3.069939954828665e-06,
+      "loss": 0.6007,
+      "step": 39147
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.3307790756225586,
+      "learning_rate": 3.069696734489424e-06,
+      "loss": 0.5639,
+      "step": 39148
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.900392532348633,
+      "learning_rate": 3.0694535195176333e-06,
+      "loss": 0.492,
+      "step": 39149
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.8799757957458496,
+      "learning_rate": 3.069210309913969e-06,
+      "loss": 0.6017,
+      "step": 39150
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.3437702655792236,
+      "learning_rate": 3.0689671056791076e-06,
+      "loss": 0.5924,
+      "step": 39151
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.3682827949523926,
+      "learning_rate": 3.0687239068137255e-06,
+      "loss": 0.5993,
+      "step": 39152
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.1366868019104004,
+      "learning_rate": 3.068480713318497e-06,
+      "loss": 0.612,
+      "step": 39153
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.0542054176330566,
+      "learning_rate": 3.068237525194102e-06,
+      "loss": 0.4569,
+      "step": 39154
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.5653069019317627,
+      "learning_rate": 3.067994342441214e-06,
+      "loss": 0.6171,
+      "step": 39155
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.319199800491333,
+      "learning_rate": 3.0677511650605106e-06,
+      "loss": 0.9304,
+      "step": 39156
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.076484441757202,
+      "learning_rate": 3.0675079930526668e-06,
+      "loss": 0.6904,
+      "step": 39157
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.5175050497055054,
+      "learning_rate": 3.06726482641836e-06,
+      "loss": 0.4626,
+      "step": 39158
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.6539440155029297,
+      "learning_rate": 3.0670216651582653e-06,
+      "loss": 0.6484,
+      "step": 39159
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.371553897857666,
+      "learning_rate": 3.0667785092730596e-06,
+      "loss": 0.8339,
+      "step": 39160
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.3361241817474365,
+      "learning_rate": 3.0665353587634184e-06,
+      "loss": 0.5019,
+      "step": 39161
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.8412666320800781,
+      "learning_rate": 3.066292213630018e-06,
+      "loss": 0.6491,
+      "step": 39162
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.4328534603118896,
+      "learning_rate": 3.0660490738735336e-06,
+      "loss": 0.5284,
+      "step": 39163
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.5795090198516846,
+      "learning_rate": 3.065805939494644e-06,
+      "loss": 0.5632,
+      "step": 39164
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.71537709236145,
+      "learning_rate": 3.0655628104940226e-06,
+      "loss": 0.7628,
+      "step": 39165
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.403263568878174,
+      "learning_rate": 3.0653196868723446e-06,
+      "loss": 0.9026,
+      "step": 39166
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 5.897456169128418,
+      "learning_rate": 3.0650765686302897e-06,
+      "loss": 0.8147,
+      "step": 39167
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.1098666191101074,
+      "learning_rate": 3.0648334557685323e-06,
+      "loss": 0.5618,
+      "step": 39168
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.796766996383667,
+      "learning_rate": 3.0645903482877477e-06,
+      "loss": 0.7049,
+      "step": 39169
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.234132766723633,
+      "learning_rate": 3.0643472461886127e-06,
+      "loss": 0.4765,
+      "step": 39170
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.750518798828125,
+      "learning_rate": 3.0641041494718026e-06,
+      "loss": 0.6634,
+      "step": 39171
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.7610208988189697,
+      "learning_rate": 3.0638610581379935e-06,
+      "loss": 0.6531,
+      "step": 39172
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.8381526470184326,
+      "learning_rate": 3.0636179721878623e-06,
+      "loss": 0.3758,
+      "step": 39173
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.712076187133789,
+      "learning_rate": 3.063374891622083e-06,
+      "loss": 0.7403,
+      "step": 39174
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.0453405380249023,
+      "learning_rate": 3.0631318164413336e-06,
+      "loss": 0.7437,
+      "step": 39175
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.5156869888305664,
+      "learning_rate": 3.0628887466462887e-06,
+      "loss": 0.7211,
+      "step": 39176
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.5372252464294434,
+      "learning_rate": 3.0626456822376245e-06,
+      "loss": 0.8025,
+      "step": 39177
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.239327907562256,
+      "learning_rate": 3.062402623216017e-06,
+      "loss": 0.6174,
+      "step": 39178
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.0356717109680176,
+      "learning_rate": 3.0621595695821415e-06,
+      "loss": 0.2444,
+      "step": 39179
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.212876319885254,
+      "learning_rate": 3.0619165213366752e-06,
+      "loss": 0.5426,
+      "step": 39180
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.468412160873413,
+      "learning_rate": 3.0616734784802927e-06,
+      "loss": 0.4873,
+      "step": 39181
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.802950382232666,
+      "learning_rate": 3.0614304410136697e-06,
+      "loss": 0.9725,
+      "step": 39182
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.025702476501465,
+      "learning_rate": 3.0611874089374826e-06,
+      "loss": 0.6024,
+      "step": 39183
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.137770891189575,
+      "learning_rate": 3.060944382252408e-06,
+      "loss": 0.5512,
+      "step": 39184
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.7201863527297974,
+      "learning_rate": 3.0607013609591196e-06,
+      "loss": 0.5184,
+      "step": 39185
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 4.553957462310791,
+      "learning_rate": 3.060458345058292e-06,
+      "loss": 0.5912,
+      "step": 39186
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.394641876220703,
+      "learning_rate": 3.060215334550606e-06,
+      "loss": 0.5233,
+      "step": 39187
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.1330504417419434,
+      "learning_rate": 3.059972329436734e-06,
+      "loss": 0.4397,
+      "step": 39188
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.829130172729492,
+      "learning_rate": 3.0597293297173523e-06,
+      "loss": 0.6497,
+      "step": 39189
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.6150784492492676,
+      "learning_rate": 3.0594863353931366e-06,
+      "loss": 0.6663,
+      "step": 39190
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.5475833415985107,
+      "learning_rate": 3.059243346464762e-06,
+      "loss": 0.4114,
+      "step": 39191
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.9485435485839844,
+      "learning_rate": 3.0590003629329057e-06,
+      "loss": 0.5587,
+      "step": 39192
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 4.560282230377197,
+      "learning_rate": 3.058757384798241e-06,
+      "loss": 0.5124,
+      "step": 39193
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.4350645542144775,
+      "learning_rate": 3.0585144120614457e-06,
+      "loss": 0.7884,
+      "step": 39194
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.8563268184661865,
+      "learning_rate": 3.0582714447231944e-06,
+      "loss": 0.5726,
+      "step": 39195
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.1657874584198,
+      "learning_rate": 3.0580284827841623e-06,
+      "loss": 0.5928,
+      "step": 39196
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 4.170591831207275,
+      "learning_rate": 3.0577855262450263e-06,
+      "loss": 0.8846,
+      "step": 39197
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 5.118432998657227,
+      "learning_rate": 3.05754257510646e-06,
+      "loss": 0.7197,
+      "step": 39198
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.599898099899292,
+      "learning_rate": 3.0572996293691414e-06,
+      "loss": 0.3806,
+      "step": 39199
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.7796140909194946,
+      "learning_rate": 3.0570566890337443e-06,
+      "loss": 0.5001,
+      "step": 39200
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.235046148300171,
+      "learning_rate": 3.0568137541009445e-06,
+      "loss": 0.5377,
+      "step": 39201
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.014414072036743,
+      "learning_rate": 3.0565708245714183e-06,
+      "loss": 0.4992,
+      "step": 39202
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.8267545700073242,
+      "learning_rate": 3.056327900445839e-06,
+      "loss": 0.4299,
+      "step": 39203
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 4.0340256690979,
+      "learning_rate": 3.056084981724885e-06,
+      "loss": 0.9024,
+      "step": 39204
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.588491439819336,
+      "learning_rate": 3.0558420684092306e-06,
+      "loss": 0.4762,
+      "step": 39205
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 4.362187385559082,
+      "learning_rate": 3.0555991604995495e-06,
+      "loss": 0.7882,
+      "step": 39206
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.506006956100464,
+      "learning_rate": 3.0553562579965203e-06,
+      "loss": 0.4187,
+      "step": 39207
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.74916410446167,
+      "learning_rate": 3.0551133609008165e-06,
+      "loss": 0.9866,
+      "step": 39208
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.8442416191101074,
+      "learning_rate": 3.0548704692131144e-06,
+      "loss": 0.6485,
+      "step": 39209
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 5.705957412719727,
+      "learning_rate": 3.0546275829340883e-06,
+      "loss": 0.554,
+      "step": 39210
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.997267484664917,
+      "learning_rate": 3.0543847020644144e-06,
+      "loss": 0.4226,
+      "step": 39211
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.492283821105957,
+      "learning_rate": 3.054141826604767e-06,
+      "loss": 0.6145,
+      "step": 39212
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.505507469177246,
+      "learning_rate": 3.0538989565558236e-06,
+      "loss": 0.5486,
+      "step": 39213
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.178866386413574,
+      "learning_rate": 3.053656091918258e-06,
+      "loss": 0.55,
+      "step": 39214
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.88726806640625,
+      "learning_rate": 3.0534132326927433e-06,
+      "loss": 0.4041,
+      "step": 39215
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.001582384109497,
+      "learning_rate": 3.0531703788799594e-06,
+      "loss": 0.6027,
+      "step": 39216
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.2302165031433105,
+      "learning_rate": 3.0529275304805795e-06,
+      "loss": 0.5108,
+      "step": 39217
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.0789499282836914,
+      "learning_rate": 3.052684687495279e-06,
+      "loss": 0.5567,
+      "step": 39218
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.5343544483184814,
+      "learning_rate": 3.0524418499247316e-06,
+      "loss": 0.5146,
+      "step": 39219
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.1215033531188965,
+      "learning_rate": 3.052199017769615e-06,
+      "loss": 0.33,
+      "step": 39220
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.4648287296295166,
+      "learning_rate": 3.0519561910306026e-06,
+      "loss": 0.6736,
+      "step": 39221
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.312346935272217,
+      "learning_rate": 3.05171336970837e-06,
+      "loss": 0.6637,
+      "step": 39222
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.8651337623596191,
+      "learning_rate": 3.051470553803594e-06,
+      "loss": 0.492,
+      "step": 39223
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.6939456462860107,
+      "learning_rate": 3.0512277433169467e-06,
+      "loss": 0.7185,
+      "step": 39224
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.5413711071014404,
+      "learning_rate": 3.0509849382491063e-06,
+      "loss": 0.7251,
+      "step": 39225
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.9756858348846436,
+      "learning_rate": 3.0507421386007463e-06,
+      "loss": 0.6316,
+      "step": 39226
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.716068744659424,
+      "learning_rate": 3.05049934437254e-06,
+      "loss": 0.769,
+      "step": 39227
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.3862478733062744,
+      "learning_rate": 3.050256555565167e-06,
+      "loss": 0.6861,
+      "step": 39228
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.3086163997650146,
+      "learning_rate": 3.0500137721793e-06,
+      "loss": 0.7538,
+      "step": 39229
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.7836122512817383,
+      "learning_rate": 3.0497709942156144e-06,
+      "loss": 0.6357,
+      "step": 39230
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.426722526550293,
+      "learning_rate": 3.0495282216747845e-06,
+      "loss": 0.6825,
+      "step": 39231
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.5576207637786865,
+      "learning_rate": 3.0492854545574867e-06,
+      "loss": 0.3439,
+      "step": 39232
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.399202585220337,
+      "learning_rate": 3.0490426928643952e-06,
+      "loss": 0.6003,
+      "step": 39233
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 4.6077399253845215,
+      "learning_rate": 3.048799936596185e-06,
+      "loss": 0.5143,
+      "step": 39234
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.6408376693725586,
+      "learning_rate": 3.048557185753531e-06,
+      "loss": 0.7218,
+      "step": 39235
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.3291471004486084,
+      "learning_rate": 3.048314440337109e-06,
+      "loss": 0.4001,
+      "step": 39236
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.446693181991577,
+      "learning_rate": 3.048071700347592e-06,
+      "loss": 0.7409,
+      "step": 39237
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.0700199604034424,
+      "learning_rate": 3.0478289657856584e-06,
+      "loss": 0.7823,
+      "step": 39238
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.8295609951019287,
+      "learning_rate": 3.0475862366519798e-06,
+      "loss": 0.8585,
+      "step": 39239
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.876402497291565,
+      "learning_rate": 3.047343512947233e-06,
+      "loss": 0.5089,
+      "step": 39240
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.860326051712036,
+      "learning_rate": 3.047100794672092e-06,
+      "loss": 0.6902,
+      "step": 39241
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.8139841556549072,
+      "learning_rate": 3.046858081827233e-06,
+      "loss": 0.6312,
+      "step": 39242
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.498094320297241,
+      "learning_rate": 3.046615374413329e-06,
+      "loss": 0.4693,
+      "step": 39243
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.3015568256378174,
+      "learning_rate": 3.046372672431056e-06,
+      "loss": 0.6099,
+      "step": 39244
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.6101057529449463,
+      "learning_rate": 3.046129975881089e-06,
+      "loss": 0.4125,
+      "step": 39245
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.5581483840942383,
+      "learning_rate": 3.045887284764103e-06,
+      "loss": 0.5464,
+      "step": 39246
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.2798478603363037,
+      "learning_rate": 3.0456445990807703e-06,
+      "loss": 0.5145,
+      "step": 39247
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.2278947830200195,
+      "learning_rate": 3.0454019188317695e-06,
+      "loss": 0.8354,
+      "step": 39248
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.8803430795669556,
+      "learning_rate": 3.0451592440177733e-06,
+      "loss": 0.5927,
+      "step": 39249
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.047837257385254,
+      "learning_rate": 3.0449165746394583e-06,
+      "loss": 0.4449,
+      "step": 39250
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.828634262084961,
+      "learning_rate": 3.0446739106974963e-06,
+      "loss": 0.6019,
+      "step": 39251
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.596423625946045,
+      "learning_rate": 3.044431252192565e-06,
+      "loss": 0.8202,
+      "step": 39252
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.503718137741089,
+      "learning_rate": 3.0441885991253374e-06,
+      "loss": 0.7534,
+      "step": 39253
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.458608865737915,
+      "learning_rate": 3.0439459514964885e-06,
+      "loss": 0.4673,
+      "step": 39254
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.004976749420166,
+      "learning_rate": 3.0437033093066935e-06,
+      "loss": 0.8121,
+      "step": 39255
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.540513753890991,
+      "learning_rate": 3.0434606725566262e-06,
+      "loss": 0.6486,
+      "step": 39256
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.978199005126953,
+      "learning_rate": 3.0432180412469627e-06,
+      "loss": 0.5558,
+      "step": 39257
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.546222448348999,
+      "learning_rate": 3.0429754153783758e-06,
+      "loss": 0.5898,
+      "step": 39258
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.414229393005371,
+      "learning_rate": 3.042732794951542e-06,
+      "loss": 0.7701,
+      "step": 39259
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 4.127131462097168,
+      "learning_rate": 3.0424901799671343e-06,
+      "loss": 0.6797,
+      "step": 39260
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.769726276397705,
+      "learning_rate": 3.0422475704258283e-06,
+      "loss": 0.8679,
+      "step": 39261
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.2480506896972656,
+      "learning_rate": 3.0420049663282984e-06,
+      "loss": 0.5557,
+      "step": 39262
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.999143362045288,
+      "learning_rate": 3.0417623676752193e-06,
+      "loss": 0.6605,
+      "step": 39263
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.222926378250122,
+      "learning_rate": 3.0415197744672657e-06,
+      "loss": 0.5162,
+      "step": 39264
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.7256851196289062,
+      "learning_rate": 3.0412771867051105e-06,
+      "loss": 0.7827,
+      "step": 39265
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.631450653076172,
+      "learning_rate": 3.0410346043894314e-06,
+      "loss": 0.5612,
+      "step": 39266
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.5880653858184814,
+      "learning_rate": 3.0407920275208996e-06,
+      "loss": 0.5463,
+      "step": 39267
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.512133836746216,
+      "learning_rate": 3.0405494561001923e-06,
+      "loss": 0.6651,
+      "step": 39268
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.5842002630233765,
+      "learning_rate": 3.040306890127983e-06,
+      "loss": 0.4393,
+      "step": 39269
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.9074649810791016,
+      "learning_rate": 3.0400643296049463e-06,
+      "loss": 0.6819,
+      "step": 39270
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.27207350730896,
+      "learning_rate": 3.0398217745317572e-06,
+      "loss": 0.4992,
+      "step": 39271
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.7581678628921509,
+      "learning_rate": 3.0395792249090884e-06,
+      "loss": 0.4852,
+      "step": 39272
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.0865256786346436,
+      "learning_rate": 3.0393366807376157e-06,
+      "loss": 0.4164,
+      "step": 39273
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.8064613342285156,
+      "learning_rate": 3.039094142018013e-06,
+      "loss": 0.646,
+      "step": 39274
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 4.625442981719971,
+      "learning_rate": 3.0388516087509545e-06,
+      "loss": 0.6179,
+      "step": 39275
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.070399284362793,
+      "learning_rate": 3.038609080937115e-06,
+      "loss": 0.5075,
+      "step": 39276
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.803879737854004,
+      "learning_rate": 3.0383665585771695e-06,
+      "loss": 0.5246,
+      "step": 39277
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.285593271255493,
+      "learning_rate": 3.0381240416717916e-06,
+      "loss": 0.4829,
+      "step": 39278
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.8367221355438232,
+      "learning_rate": 3.0378815302216557e-06,
+      "loss": 0.6079,
+      "step": 39279
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.4238970279693604,
+      "learning_rate": 3.037639024227437e-06,
+      "loss": 0.6398,
+      "step": 39280
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.276820421218872,
+      "learning_rate": 3.0373965236898085e-06,
+      "loss": 0.4876,
+      "step": 39281
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.6668187379837036,
+      "learning_rate": 3.037154028609445e-06,
+      "loss": 0.3726,
+      "step": 39282
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.7733350992202759,
+      "learning_rate": 3.0369115389870206e-06,
+      "loss": 0.7055,
+      "step": 39283
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.2335503101348877,
+      "learning_rate": 3.0366690548232104e-06,
+      "loss": 0.7386,
+      "step": 39284
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.9434666633605957,
+      "learning_rate": 3.0364265761186872e-06,
+      "loss": 0.8685,
+      "step": 39285
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.3683249950408936,
+      "learning_rate": 3.0361841028741274e-06,
+      "loss": 0.7593,
+      "step": 39286
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.8184940814971924,
+      "learning_rate": 3.035941635090203e-06,
+      "loss": 0.7177,
+      "step": 39287
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 4.05165958404541,
+      "learning_rate": 3.0356991727675876e-06,
+      "loss": 0.621,
+      "step": 39288
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.5834145545959473,
+      "learning_rate": 3.0354567159069586e-06,
+      "loss": 0.8115,
+      "step": 39289
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.8068339824676514,
+      "learning_rate": 3.0352142645089887e-06,
+      "loss": 0.7096,
+      "step": 39290
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.4393579959869385,
+      "learning_rate": 3.034971818574352e-06,
+      "loss": 0.8577,
+      "step": 39291
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.0225183963775635,
+      "learning_rate": 3.0347293781037226e-06,
+      "loss": 0.4761,
+      "step": 39292
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.9357377290725708,
+      "learning_rate": 3.034486943097774e-06,
+      "loss": 0.642,
+      "step": 39293
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.929640769958496,
+      "learning_rate": 3.0342445135571818e-06,
+      "loss": 0.8264,
+      "step": 39294
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 14.096561431884766,
+      "learning_rate": 3.034002089482619e-06,
+      "loss": 0.4095,
+      "step": 39295
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.2819368839263916,
+      "learning_rate": 3.03375967087476e-06,
+      "loss": 0.5856,
+      "step": 39296
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.8422739505767822,
+      "learning_rate": 3.0335172577342785e-06,
+      "loss": 0.4783,
+      "step": 39297
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.4933888912200928,
+      "learning_rate": 3.0332748500618485e-06,
+      "loss": 0.6963,
+      "step": 39298
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.018355131149292,
+      "learning_rate": 3.0330324478581453e-06,
+      "loss": 0.3588,
+      "step": 39299
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.7730369567871094,
+      "learning_rate": 3.0327900511238413e-06,
+      "loss": 0.701,
+      "step": 39300
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.008978843688965,
+      "learning_rate": 3.0325476598596113e-06,
+      "loss": 0.7915,
+      "step": 39301
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.8357410430908203,
+      "learning_rate": 3.032305274066129e-06,
+      "loss": 0.6263,
+      "step": 39302
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 9.168102264404297,
+      "learning_rate": 3.032062893744069e-06,
+      "loss": 0.4773,
+      "step": 39303
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.5121920108795166,
+      "learning_rate": 3.031820518894105e-06,
+      "loss": 0.6304,
+      "step": 39304
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 4.374455451965332,
+      "learning_rate": 3.0315781495169105e-06,
+      "loss": 0.754,
+      "step": 39305
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.9956927299499512,
+      "learning_rate": 3.0313357856131597e-06,
+      "loss": 0.4105,
+      "step": 39306
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.6675796508789062,
+      "learning_rate": 3.0310934271835275e-06,
+      "loss": 0.6048,
+      "step": 39307
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.4636430740356445,
+      "learning_rate": 3.030851074228684e-06,
+      "loss": 0.3819,
+      "step": 39308
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.705669403076172,
+      "learning_rate": 3.0306087267493086e-06,
+      "loss": 0.4513,
+      "step": 39309
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.905339241027832,
+      "learning_rate": 3.030366384746072e-06,
+      "loss": 0.3158,
+      "step": 39310
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 4.2412800788879395,
+      "learning_rate": 3.030124048219649e-06,
+      "loss": 0.7759,
+      "step": 39311
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.885033130645752,
+      "learning_rate": 3.0298817171707128e-06,
+      "loss": 0.3857,
+      "step": 39312
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.2359347343444824,
+      "learning_rate": 3.029639391599938e-06,
+      "loss": 0.7877,
+      "step": 39313
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.168998956680298,
+      "learning_rate": 3.029397071507998e-06,
+      "loss": 0.5137,
+      "step": 39314
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 4.633129119873047,
+      "learning_rate": 3.0291547568955657e-06,
+      "loss": 0.7353,
+      "step": 39315
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.0642361640930176,
+      "learning_rate": 3.028912447763316e-06,
+      "loss": 0.4391,
+      "step": 39316
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.122677803039551,
+      "learning_rate": 3.0286701441119227e-06,
+      "loss": 0.6208,
+      "step": 39317
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.9389498233795166,
+      "learning_rate": 3.0284278459420592e-06,
+      "loss": 0.73,
+      "step": 39318
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.9288971424102783,
+      "learning_rate": 3.0281855532543992e-06,
+      "loss": 0.6145,
+      "step": 39319
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.766654133796692,
+      "learning_rate": 3.0279432660496165e-06,
+      "loss": 0.7408,
+      "step": 39320
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.3710110187530518,
+      "learning_rate": 3.0277009843283854e-06,
+      "loss": 0.6604,
+      "step": 39321
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 5.409595489501953,
+      "learning_rate": 3.0274587080913787e-06,
+      "loss": 0.5078,
+      "step": 39322
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.5870299339294434,
+      "learning_rate": 3.02721643733927e-06,
+      "loss": 0.6729,
+      "step": 39323
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.3155031204223633,
+      "learning_rate": 3.026974172072733e-06,
+      "loss": 0.626,
+      "step": 39324
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.9872629642486572,
+      "learning_rate": 3.0267319122924427e-06,
+      "loss": 0.7566,
+      "step": 39325
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.4582080841064453,
+      "learning_rate": 3.0264896579990705e-06,
+      "loss": 0.4978,
+      "step": 39326
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.061098575592041,
+      "learning_rate": 3.026247409193292e-06,
+      "loss": 0.7172,
+      "step": 39327
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.133068799972534,
+      "learning_rate": 3.0260051658757795e-06,
+      "loss": 0.6266,
+      "step": 39328
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.6543071269989014,
+      "learning_rate": 3.0257629280472078e-06,
+      "loss": 0.6063,
+      "step": 39329
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.7342419624328613,
+      "learning_rate": 3.0255206957082494e-06,
+      "loss": 0.4865,
+      "step": 39330
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.3692739009857178,
+      "learning_rate": 3.0252784688595795e-06,
+      "loss": 0.7178,
+      "step": 39331
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.134922742843628,
+      "learning_rate": 3.0250362475018693e-06,
+      "loss": 0.7203,
+      "step": 39332
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.208930015563965,
+      "learning_rate": 3.024794031635794e-06,
+      "loss": 0.4625,
+      "step": 39333
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.8146973848342896,
+      "learning_rate": 3.024551821262026e-06,
+      "loss": 0.5202,
+      "step": 39334
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.438767910003662,
+      "learning_rate": 3.0243096163812404e-06,
+      "loss": 0.8208,
+      "step": 39335
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.2053723335266113,
+      "learning_rate": 3.0240674169941087e-06,
+      "loss": 0.6125,
+      "step": 39336
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.334639310836792,
+      "learning_rate": 3.0238252231013054e-06,
+      "loss": 0.6154,
+      "step": 39337
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.897704601287842,
+      "learning_rate": 3.0235830347035034e-06,
+      "loss": 0.6622,
+      "step": 39338
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.9467580318450928,
+      "learning_rate": 3.0233408518013774e-06,
+      "loss": 0.6134,
+      "step": 39339
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.4459986686706543,
+      "learning_rate": 3.0230986743956e-06,
+      "loss": 0.6605,
+      "step": 39340
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.20188045501709,
+      "learning_rate": 3.022856502486844e-06,
+      "loss": 0.6761,
+      "step": 39341
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.4374372959136963,
+      "learning_rate": 3.0226143360757845e-06,
+      "loss": 0.508,
+      "step": 39342
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.420032262802124,
+      "learning_rate": 3.022372175163093e-06,
+      "loss": 0.6516,
+      "step": 39343
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.6357582807540894,
+      "learning_rate": 3.022130019749444e-06,
+      "loss": 0.4882,
+      "step": 39344
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 4.232757568359375,
+      "learning_rate": 3.0218878698355105e-06,
+      "loss": 0.6983,
+      "step": 39345
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.0422956943511963,
+      "learning_rate": 3.0216457254219657e-06,
+      "loss": 0.4372,
+      "step": 39346
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.1609954833984375,
+      "learning_rate": 3.0214035865094828e-06,
+      "loss": 0.6653,
+      "step": 39347
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.8877835273742676,
+      "learning_rate": 3.0211614530987355e-06,
+      "loss": 0.5682,
+      "step": 39348
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.150144100189209,
+      "learning_rate": 3.0209193251903955e-06,
+      "loss": 0.5938,
+      "step": 39349
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.554175615310669,
+      "learning_rate": 3.0206772027851384e-06,
+      "loss": 0.4918,
+      "step": 39350
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.012645721435547,
+      "learning_rate": 3.0204350858836375e-06,
+      "loss": 0.5722,
+      "step": 39351
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.156430721282959,
+      "learning_rate": 3.020192974486564e-06,
+      "loss": 0.5238,
+      "step": 39352
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.442903995513916,
+      "learning_rate": 3.0199508685945934e-06,
+      "loss": 0.7182,
+      "step": 39353
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.213765859603882,
+      "learning_rate": 3.019708768208397e-06,
+      "loss": 0.6817,
+      "step": 39354
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.886171579360962,
+      "learning_rate": 3.0194666733286485e-06,
+      "loss": 0.656,
+      "step": 39355
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.2275960445404053,
+      "learning_rate": 3.019224583956022e-06,
+      "loss": 0.7222,
+      "step": 39356
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.360199213027954,
+      "learning_rate": 3.018982500091189e-06,
+      "loss": 0.6804,
+      "step": 39357
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.2857513427734375,
+      "learning_rate": 3.0187404217348244e-06,
+      "loss": 0.6522,
+      "step": 39358
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.5902204513549805,
+      "learning_rate": 3.0184983488876e-06,
+      "loss": 0.4458,
+      "step": 39359
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.616703748703003,
+      "learning_rate": 3.018256281550189e-06,
+      "loss": 0.7189,
+      "step": 39360
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.12221097946167,
+      "learning_rate": 3.018014219723266e-06,
+      "loss": 0.6091,
+      "step": 39361
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.231849431991577,
+      "learning_rate": 3.0177721634075017e-06,
+      "loss": 0.5438,
+      "step": 39362
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.615880250930786,
+      "learning_rate": 3.017530112603572e-06,
+      "loss": 0.54,
+      "step": 39363
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.0255796909332275,
+      "learning_rate": 3.0172880673121472e-06,
+      "loss": 0.8213,
+      "step": 39364
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.6312286853790283,
+      "learning_rate": 3.017046027533902e-06,
+      "loss": 0.8426,
+      "step": 39365
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.936476230621338,
+      "learning_rate": 3.0168039932695094e-06,
+      "loss": 0.5754,
+      "step": 39366
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 4.269798278808594,
+      "learning_rate": 3.016561964519641e-06,
+      "loss": 0.875,
+      "step": 39367
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.020432472229004,
+      "learning_rate": 3.0163199412849707e-06,
+      "loss": 0.6805,
+      "step": 39368
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 12.409317970275879,
+      "learning_rate": 3.0160779235661707e-06,
+      "loss": 0.7474,
+      "step": 39369
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.2715868949890137,
+      "learning_rate": 3.0158359113639158e-06,
+      "loss": 0.4574,
+      "step": 39370
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.280255079269409,
+      "learning_rate": 3.0155939046788786e-06,
+      "loss": 0.4004,
+      "step": 39371
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.4412336349487305,
+      "learning_rate": 3.015351903511731e-06,
+      "loss": 0.6439,
+      "step": 39372
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.847614049911499,
+      "learning_rate": 3.0151099078631463e-06,
+      "loss": 0.4494,
+      "step": 39373
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.3684332370758057,
+      "learning_rate": 3.0148679177337975e-06,
+      "loss": 0.5771,
+      "step": 39374
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.829040765762329,
+      "learning_rate": 3.014625933124358e-06,
+      "loss": 0.4545,
+      "step": 39375
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.6500768661499023,
+      "learning_rate": 3.0143839540354986e-06,
+      "loss": 0.7074,
+      "step": 39376
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.6822621822357178,
+      "learning_rate": 3.0141419804678945e-06,
+      "loss": 0.7498,
+      "step": 39377
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.0224339962005615,
+      "learning_rate": 3.0139000124222172e-06,
+      "loss": 0.7694,
+      "step": 39378
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.777007579803467,
+      "learning_rate": 3.013658049899141e-06,
+      "loss": 0.7781,
+      "step": 39379
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.05654239654541,
+      "learning_rate": 3.013416092899336e-06,
+      "loss": 0.6622,
+      "step": 39380
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.216125249862671,
+      "learning_rate": 3.0131741414234783e-06,
+      "loss": 0.8453,
+      "step": 39381
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 5.061877727508545,
+      "learning_rate": 3.012932195472238e-06,
+      "loss": 0.8015,
+      "step": 39382
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.0800230503082275,
+      "learning_rate": 3.012690255046289e-06,
+      "loss": 0.6691,
+      "step": 39383
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.0200366973876953,
+      "learning_rate": 3.012448320146304e-06,
+      "loss": 0.44,
+      "step": 39384
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.1707661151885986,
+      "learning_rate": 3.012206390772956e-06,
+      "loss": 0.3916,
+      "step": 39385
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.7393031120300293,
+      "learning_rate": 3.0119644669269177e-06,
+      "loss": 0.726,
+      "step": 39386
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.5670604705810547,
+      "learning_rate": 3.0117225486088608e-06,
+      "loss": 0.6621,
+      "step": 39387
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.594719409942627,
+      "learning_rate": 3.0114806358194574e-06,
+      "loss": 0.6448,
+      "step": 39388
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 4.747300624847412,
+      "learning_rate": 3.0112387285593818e-06,
+      "loss": 0.6327,
+      "step": 39389
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.058938503265381,
+      "learning_rate": 3.0109968268293077e-06,
+      "loss": 0.5026,
+      "step": 39390
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 4.267371654510498,
+      "learning_rate": 3.0107549306299056e-06,
+      "loss": 0.7558,
+      "step": 39391
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.6449346542358398,
+      "learning_rate": 3.0105130399618497e-06,
+      "loss": 0.4343,
+      "step": 39392
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.7499382495880127,
+      "learning_rate": 3.0102711548258114e-06,
+      "loss": 0.8406,
+      "step": 39393
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.392657995223999,
+      "learning_rate": 3.0100292752224634e-06,
+      "loss": 0.4276,
+      "step": 39394
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.843428134918213,
+      "learning_rate": 3.0097874011524785e-06,
+      "loss": 0.4523,
+      "step": 39395
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.7428715229034424,
+      "learning_rate": 3.0095455326165296e-06,
+      "loss": 0.4086,
+      "step": 39396
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.753488779067993,
+      "learning_rate": 3.0093036696152888e-06,
+      "loss": 0.5829,
+      "step": 39397
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.458122730255127,
+      "learning_rate": 3.0090618121494286e-06,
+      "loss": 0.6248,
+      "step": 39398
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.351752996444702,
+      "learning_rate": 3.0088199602196224e-06,
+      "loss": 0.7067,
+      "step": 39399
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 4.801937103271484,
+      "learning_rate": 3.0085781138265403e-06,
+      "loss": 0.7643,
+      "step": 39400
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.149477243423462,
+      "learning_rate": 3.0083362729708572e-06,
+      "loss": 0.7054,
+      "step": 39401
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.5765228271484375,
+      "learning_rate": 3.008094437653245e-06,
+      "loss": 0.6951,
+      "step": 39402
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.526358127593994,
+      "learning_rate": 3.007852607874376e-06,
+      "loss": 0.5557,
+      "step": 39403
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.503373146057129,
+      "learning_rate": 3.0076107836349232e-06,
+      "loss": 0.6234,
+      "step": 39404
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.3936309814453125,
+      "learning_rate": 3.0073689649355575e-06,
+      "loss": 0.765,
+      "step": 39405
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.6400647163391113,
+      "learning_rate": 3.0071271517769527e-06,
+      "loss": 0.68,
+      "step": 39406
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.4818880558013916,
+      "learning_rate": 3.006885344159781e-06,
+      "loss": 0.4228,
+      "step": 39407
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.0501577854156494,
+      "learning_rate": 3.006643542084715e-06,
+      "loss": 0.3036,
+      "step": 39408
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.7477033138275146,
+      "learning_rate": 3.006401745552426e-06,
+      "loss": 0.481,
+      "step": 39409
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.2194130420684814,
+      "learning_rate": 3.0061599545635856e-06,
+      "loss": 0.6115,
+      "step": 39410
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.495610237121582,
+      "learning_rate": 3.0059181691188676e-06,
+      "loss": 0.7729,
+      "step": 39411
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.659602165222168,
+      "learning_rate": 3.0056763892189454e-06,
+      "loss": 0.7106,
+      "step": 39412
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.7254865169525146,
+      "learning_rate": 3.00543461486449e-06,
+      "loss": 0.7735,
+      "step": 39413
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.419650077819824,
+      "learning_rate": 3.0051928460561737e-06,
+      "loss": 0.7639,
+      "step": 39414
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.8280898332595825,
+      "learning_rate": 3.0049510827946688e-06,
+      "loss": 0.5478,
+      "step": 39415
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.7756590843200684,
+      "learning_rate": 3.0047093250806473e-06,
+      "loss": 0.5963,
+      "step": 39416
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.0543103218078613,
+      "learning_rate": 3.004467572914782e-06,
+      "loss": 0.7269,
+      "step": 39417
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.863715648651123,
+      "learning_rate": 3.0042258262977455e-06,
+      "loss": 0.7701,
+      "step": 39418
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.934171199798584,
+      "learning_rate": 3.0039840852302084e-06,
+      "loss": 0.6478,
+      "step": 39419
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.0393590927124023,
+      "learning_rate": 3.0037423497128442e-06,
+      "loss": 0.5962,
+      "step": 39420
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.0574967861175537,
+      "learning_rate": 3.0035006197463245e-06,
+      "loss": 0.4475,
+      "step": 39421
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.675863027572632,
+      "learning_rate": 3.0032588953313218e-06,
+      "loss": 0.6934,
+      "step": 39422
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.995323896408081,
+      "learning_rate": 3.0030171764685082e-06,
+      "loss": 0.7379,
+      "step": 39423
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.1798293590545654,
+      "learning_rate": 3.002775463158556e-06,
+      "loss": 0.6144,
+      "step": 39424
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.0020663738250732,
+      "learning_rate": 3.0025337554021367e-06,
+      "loss": 0.2464,
+      "step": 39425
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.470946788787842,
+      "learning_rate": 3.002292053199923e-06,
+      "loss": 0.4661,
+      "step": 39426
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.071619749069214,
+      "learning_rate": 3.002050356552586e-06,
+      "loss": 0.2457,
+      "step": 39427
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.212552309036255,
+      "learning_rate": 3.0018086654607992e-06,
+      "loss": 0.4998,
+      "step": 39428
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.2171132564544678,
+      "learning_rate": 3.001566979925234e-06,
+      "loss": 0.519,
+      "step": 39429
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.0540928840637207,
+      "learning_rate": 3.00132529994656e-06,
+      "loss": 0.7085,
+      "step": 39430
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.0982825756073,
+      "learning_rate": 3.0010836255254537e-06,
+      "loss": 0.6599,
+      "step": 39431
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.6450226306915283,
+      "learning_rate": 3.000841956662586e-06,
+      "loss": 0.7974,
+      "step": 39432
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.0081381797790527,
+      "learning_rate": 3.0006002933586264e-06,
+      "loss": 0.7466,
+      "step": 39433
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.410865545272827,
+      "learning_rate": 3.000358635614249e-06,
+      "loss": 0.6352,
+      "step": 39434
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.7978154420852661,
+      "learning_rate": 3.0001169834301256e-06,
+      "loss": 0.6914,
+      "step": 39435
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.7756271362304688,
+      "learning_rate": 2.9998753368069274e-06,
+      "loss": 0.5996,
+      "step": 39436
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.173491954803467,
+      "learning_rate": 2.9996336957453263e-06,
+      "loss": 0.7676,
+      "step": 39437
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.552340507507324,
+      "learning_rate": 2.999392060245995e-06,
+      "loss": 0.608,
+      "step": 39438
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.842129707336426,
+      "learning_rate": 2.999150430309604e-06,
+      "loss": 0.548,
+      "step": 39439
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.5082356929779053,
+      "learning_rate": 2.9989088059368276e-06,
+      "loss": 0.5409,
+      "step": 39440
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.9756488800048828,
+      "learning_rate": 2.998667187128335e-06,
+      "loss": 0.8088,
+      "step": 39441
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.3231890201568604,
+      "learning_rate": 2.9984255738848e-06,
+      "loss": 0.7921,
+      "step": 39442
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.1064066886901855,
+      "learning_rate": 2.9981839662068933e-06,
+      "loss": 0.6059,
+      "step": 39443
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.536262035369873,
+      "learning_rate": 2.997942364095287e-06,
+      "loss": 0.628,
+      "step": 39444
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.301748514175415,
+      "learning_rate": 2.997700767550653e-06,
+      "loss": 0.7792,
+      "step": 39445
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.451554298400879,
+      "learning_rate": 2.997459176573663e-06,
+      "loss": 0.7441,
+      "step": 39446
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.0356082916259766,
+      "learning_rate": 2.997217591164989e-06,
+      "loss": 0.4977,
+      "step": 39447
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.7134761810302734,
+      "learning_rate": 2.996976011325303e-06,
+      "loss": 0.6709,
+      "step": 39448
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.4807674884796143,
+      "learning_rate": 2.9967344370552764e-06,
+      "loss": 0.5653,
+      "step": 39449
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.1073110103607178,
+      "learning_rate": 2.9964928683555785e-06,
+      "loss": 0.7701,
+      "step": 39450
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.074465274810791,
+      "learning_rate": 2.996251305226885e-06,
+      "loss": 0.8544,
+      "step": 39451
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.999763250350952,
+      "learning_rate": 2.9960097476698668e-06,
+      "loss": 0.6396,
+      "step": 39452
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.3482823371887207,
+      "learning_rate": 2.9957681956851943e-06,
+      "loss": 0.4708,
+      "step": 39453
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.5155739784240723,
+      "learning_rate": 2.9955266492735395e-06,
+      "loss": 0.7067,
+      "step": 39454
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.2537665367126465,
+      "learning_rate": 2.995285108435575e-06,
+      "loss": 0.7393,
+      "step": 39455
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.0387752056121826,
+      "learning_rate": 2.9950435731719707e-06,
+      "loss": 0.6911,
+      "step": 39456
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.8206721544265747,
+      "learning_rate": 2.9948020434833996e-06,
+      "loss": 0.4002,
+      "step": 39457
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.8678576946258545,
+      "learning_rate": 2.994560519370533e-06,
+      "loss": 0.5349,
+      "step": 39458
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.3499372005462646,
+      "learning_rate": 2.9943190008340418e-06,
+      "loss": 0.5856,
+      "step": 39459
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.47232985496521,
+      "learning_rate": 2.994077487874599e-06,
+      "loss": 0.582,
+      "step": 39460
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.1055350303649902,
+      "learning_rate": 2.993835980492873e-06,
+      "loss": 0.6886,
+      "step": 39461
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.934083938598633,
+      "learning_rate": 2.99359447868954e-06,
+      "loss": 0.6855,
+      "step": 39462
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.576328754425049,
+      "learning_rate": 2.9933529824652683e-06,
+      "loss": 0.6082,
+      "step": 39463
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.114428758621216,
+      "learning_rate": 2.9931114918207305e-06,
+      "loss": 0.3754,
+      "step": 39464
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.479585647583008,
+      "learning_rate": 2.9928700067565974e-06,
+      "loss": 0.5552,
+      "step": 39465
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.9805595874786377,
+      "learning_rate": 2.992628527273541e-06,
+      "loss": 0.7187,
+      "step": 39466
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.219968795776367,
+      "learning_rate": 2.992387053372233e-06,
+      "loss": 0.5675,
+      "step": 39467
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.9449694156646729,
+      "learning_rate": 2.9921455850533446e-06,
+      "loss": 0.8044,
+      "step": 39468
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.9510693550109863,
+      "learning_rate": 2.991904122317547e-06,
+      "loss": 0.4937,
+      "step": 39469
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.24869704246521,
+      "learning_rate": 2.9916626651655124e-06,
+      "loss": 0.7831,
+      "step": 39470
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.8880711793899536,
+      "learning_rate": 2.99142121359791e-06,
+      "loss": 0.7077,
+      "step": 39471
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.159503698348999,
+      "learning_rate": 2.9911797676154137e-06,
+      "loss": 0.6408,
+      "step": 39472
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.1769556999206543,
+      "learning_rate": 2.990938327218693e-06,
+      "loss": 0.6572,
+      "step": 39473
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.671262741088867,
+      "learning_rate": 2.9906968924084212e-06,
+      "loss": 0.3348,
+      "step": 39474
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.4931888580322266,
+      "learning_rate": 2.990455463185269e-06,
+      "loss": 0.6178,
+      "step": 39475
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.5113525390625,
+      "learning_rate": 2.990214039549907e-06,
+      "loss": 0.476,
+      "step": 39476
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.1010515689849854,
+      "learning_rate": 2.9899726215030066e-06,
+      "loss": 0.6097,
+      "step": 39477
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.5293521881103516,
+      "learning_rate": 2.98973120904524e-06,
+      "loss": 0.7919,
+      "step": 39478
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.3829867839813232,
+      "learning_rate": 2.9894898021772776e-06,
+      "loss": 0.7797,
+      "step": 39479
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.0279321670532227,
+      "learning_rate": 2.9892484008997907e-06,
+      "loss": 0.5455,
+      "step": 39480
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.980648994445801,
+      "learning_rate": 2.989007005213451e-06,
+      "loss": 0.8242,
+      "step": 39481
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.2831783294677734,
+      "learning_rate": 2.9887656151189293e-06,
+      "loss": 0.5794,
+      "step": 39482
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.239457845687866,
+      "learning_rate": 2.988524230616897e-06,
+      "loss": 0.5458,
+      "step": 39483
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.717690944671631,
+      "learning_rate": 2.9882828517080255e-06,
+      "loss": 0.656,
+      "step": 39484
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.9138503074645996,
+      "learning_rate": 2.9880414783929855e-06,
+      "loss": 0.7191,
+      "step": 39485
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.0917084217071533,
+      "learning_rate": 2.9878001106724486e-06,
+      "loss": 0.7303,
+      "step": 39486
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.1845359802246094,
+      "learning_rate": 2.9875587485470857e-06,
+      "loss": 0.7625,
+      "step": 39487
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.3409054279327393,
+      "learning_rate": 2.9873173920175685e-06,
+      "loss": 0.3503,
+      "step": 39488
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.3918302059173584,
+      "learning_rate": 2.9870760410845677e-06,
+      "loss": 0.4921,
+      "step": 39489
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.8353054523468018,
+      "learning_rate": 2.986834695748754e-06,
+      "loss": 0.5308,
+      "step": 39490
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.9439966678619385,
+      "learning_rate": 2.9865933560107962e-06,
+      "loss": 0.7352,
+      "step": 39491
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.2670326232910156,
+      "learning_rate": 2.9863520218713714e-06,
+      "loss": 0.7516,
+      "step": 39492
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.944812774658203,
+      "learning_rate": 2.986110693331146e-06,
+      "loss": 0.8219,
+      "step": 39493
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.020883321762085,
+      "learning_rate": 2.9858693703907936e-06,
+      "loss": 0.5687,
+      "step": 39494
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.9014160633087158,
+      "learning_rate": 2.9856280530509828e-06,
+      "loss": 0.6594,
+      "step": 39495
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.541583299636841,
+      "learning_rate": 2.9853867413123865e-06,
+      "loss": 0.5236,
+      "step": 39496
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 4.390120983123779,
+      "learning_rate": 2.985145435175675e-06,
+      "loss": 0.7435,
+      "step": 39497
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.045036554336548,
+      "learning_rate": 2.9849041346415186e-06,
+      "loss": 0.5615,
+      "step": 39498
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.8137457370758057,
+      "learning_rate": 2.98466283971059e-06,
+      "loss": 0.5003,
+      "step": 39499
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.8111237287521362,
+      "learning_rate": 2.9844215503835583e-06,
+      "loss": 0.3849,
+      "step": 39500
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.50942063331604,
+      "learning_rate": 2.984180266661095e-06,
+      "loss": 0.6693,
+      "step": 39501
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.9719693660736084,
+      "learning_rate": 2.9839389885438717e-06,
+      "loss": 0.583,
+      "step": 39502
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.3219547271728516,
+      "learning_rate": 2.9836977160325588e-06,
+      "loss": 0.3635,
+      "step": 39503
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 4.1758222579956055,
+      "learning_rate": 2.9834564491278272e-06,
+      "loss": 0.471,
+      "step": 39504
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 4.608645915985107,
+      "learning_rate": 2.983215187830348e-06,
+      "loss": 0.4762,
+      "step": 39505
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.6444878578186035,
+      "learning_rate": 2.9829739321407912e-06,
+      "loss": 0.7559,
+      "step": 39506
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.148270845413208,
+      "learning_rate": 2.982732682059829e-06,
+      "loss": 0.6781,
+      "step": 39507
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 4.51377534866333,
+      "learning_rate": 2.982491437588131e-06,
+      "loss": 0.5616,
+      "step": 39508
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.4554336071014404,
+      "learning_rate": 2.982250198726369e-06,
+      "loss": 0.7628,
+      "step": 39509
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.3689513206481934,
+      "learning_rate": 2.9820089654752132e-06,
+      "loss": 0.497,
+      "step": 39510
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.0636801719665527,
+      "learning_rate": 2.9817677378353326e-06,
+      "loss": 0.5656,
+      "step": 39511
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.221834659576416,
+      "learning_rate": 2.9815265158074014e-06,
+      "loss": 0.7621,
+      "step": 39512
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.1267998218536377,
+      "learning_rate": 2.9812852993920893e-06,
+      "loss": 0.4385,
+      "step": 39513
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.4582648277282715,
+      "learning_rate": 2.981044088590066e-06,
+      "loss": 0.5952,
+      "step": 39514
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.6450555324554443,
+      "learning_rate": 2.980802883402004e-06,
+      "loss": 0.3965,
+      "step": 39515
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.135122776031494,
+      "learning_rate": 2.980561683828571e-06,
+      "loss": 0.6522,
+      "step": 39516
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.454644203186035,
+      "learning_rate": 2.980320489870441e-06,
+      "loss": 0.508,
+      "step": 39517
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.218876600265503,
+      "learning_rate": 2.9800793015282824e-06,
+      "loss": 0.7689,
+      "step": 39518
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.5943500995635986,
+      "learning_rate": 2.979838118802767e-06,
+      "loss": 0.4966,
+      "step": 39519
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.225651741027832,
+      "learning_rate": 2.9795969416945645e-06,
+      "loss": 0.5378,
+      "step": 39520
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.8881545066833496,
+      "learning_rate": 2.9793557702043464e-06,
+      "loss": 0.5745,
+      "step": 39521
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.8421823978424072,
+      "learning_rate": 2.9791146043327823e-06,
+      "loss": 0.5018,
+      "step": 39522
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.23679780960083,
+      "learning_rate": 2.9788734440805433e-06,
+      "loss": 0.5521,
+      "step": 39523
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.3450684547424316,
+      "learning_rate": 2.978632289448301e-06,
+      "loss": 0.7306,
+      "step": 39524
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.7900230884552002,
+      "learning_rate": 2.978391140436725e-06,
+      "loss": 0.6654,
+      "step": 39525
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.9628331661224365,
+      "learning_rate": 2.9781499970464855e-06,
+      "loss": 0.5987,
+      "step": 39526
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.9963955879211426,
+      "learning_rate": 2.9779088592782537e-06,
+      "loss": 0.3979,
+      "step": 39527
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.8009848594665527,
+      "learning_rate": 2.9776677271327e-06,
+      "loss": 0.5733,
+      "step": 39528
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 5.189107418060303,
+      "learning_rate": 2.9774266006104948e-06,
+      "loss": 0.3835,
+      "step": 39529
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.876579999923706,
+      "learning_rate": 2.9771854797123077e-06,
+      "loss": 0.6807,
+      "step": 39530
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.5246039628982544,
+      "learning_rate": 2.976944364438811e-06,
+      "loss": 0.5093,
+      "step": 39531
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.152090549468994,
+      "learning_rate": 2.976703254790673e-06,
+      "loss": 0.457,
+      "step": 39532
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.606806993484497,
+      "learning_rate": 2.9764621507685663e-06,
+      "loss": 0.6973,
+      "step": 39533
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.4929473400115967,
+      "learning_rate": 2.9762210523731605e-06,
+      "loss": 0.8042,
+      "step": 39534
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.0473814010620117,
+      "learning_rate": 2.9759799596051244e-06,
+      "loss": 0.3572,
+      "step": 39535
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.433619499206543,
+      "learning_rate": 2.975738872465131e-06,
+      "loss": 0.6519,
+      "step": 39536
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.407011032104492,
+      "learning_rate": 2.9754977909538497e-06,
+      "loss": 0.6898,
+      "step": 39537
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.6807515621185303,
+      "learning_rate": 2.9752567150719503e-06,
+      "loss": 0.7127,
+      "step": 39538
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 4.251896381378174,
+      "learning_rate": 2.9750156448201035e-06,
+      "loss": 1.0141,
+      "step": 39539
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.335824728012085,
+      "learning_rate": 2.97477458019898e-06,
+      "loss": 0.6282,
+      "step": 39540
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.552507162094116,
+      "learning_rate": 2.9745335212092498e-06,
+      "loss": 0.625,
+      "step": 39541
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.8651154041290283,
+      "learning_rate": 2.9742924678515823e-06,
+      "loss": 0.7261,
+      "step": 39542
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.219656467437744,
+      "learning_rate": 2.974051420126649e-06,
+      "loss": 0.6379,
+      "step": 39543
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.2749135494232178,
+      "learning_rate": 2.97381037803512e-06,
+      "loss": 0.5642,
+      "step": 39544
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.4203755855560303,
+      "learning_rate": 2.973569341577665e-06,
+      "loss": 0.6504,
+      "step": 39545
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.9461820125579834,
+      "learning_rate": 2.973328310754955e-06,
+      "loss": 0.9158,
+      "step": 39546
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.2598307132720947,
+      "learning_rate": 2.9730872855676598e-06,
+      "loss": 0.6145,
+      "step": 39547
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 5.120326042175293,
+      "learning_rate": 2.9728462660164486e-06,
+      "loss": 0.5989,
+      "step": 39548
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.806899070739746,
+      "learning_rate": 2.972605252101993e-06,
+      "loss": 0.8124,
+      "step": 39549
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.2743444442749023,
+      "learning_rate": 2.972364243824963e-06,
+      "loss": 0.489,
+      "step": 39550
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 4.028199195861816,
+      "learning_rate": 2.9721232411860286e-06,
+      "loss": 0.904,
+      "step": 39551
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.6824225187301636,
+      "learning_rate": 2.9718822441858575e-06,
+      "loss": 0.4677,
+      "step": 39552
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.9623632431030273,
+      "learning_rate": 2.971641252825124e-06,
+      "loss": 0.5159,
+      "step": 39553
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.192227840423584,
+      "learning_rate": 2.971400267104496e-06,
+      "loss": 0.4766,
+      "step": 39554
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.0857484340667725,
+      "learning_rate": 2.971159287024645e-06,
+      "loss": 0.6934,
+      "step": 39555
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.2108612060546875,
+      "learning_rate": 2.9709183125862397e-06,
+      "loss": 0.7033,
+      "step": 39556
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.550816535949707,
+      "learning_rate": 2.97067734378995e-06,
+      "loss": 0.7694,
+      "step": 39557
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.1866891384124756,
+      "learning_rate": 2.970436380636447e-06,
+      "loss": 0.6148,
+      "step": 39558
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.2797317504882812,
+      "learning_rate": 2.9701954231263997e-06,
+      "loss": 0.4752,
+      "step": 39559
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.624518871307373,
+      "learning_rate": 2.9699544712604784e-06,
+      "loss": 0.7743,
+      "step": 39560
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 4.453732490539551,
+      "learning_rate": 2.969713525039353e-06,
+      "loss": 1.0593,
+      "step": 39561
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 4.211731433868408,
+      "learning_rate": 2.969472584463694e-06,
+      "loss": 0.3961,
+      "step": 39562
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.9024709463119507,
+      "learning_rate": 2.969231649534172e-06,
+      "loss": 0.5742,
+      "step": 39563
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.0455315113067627,
+      "learning_rate": 2.968990720251455e-06,
+      "loss": 0.7787,
+      "step": 39564
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.8462560176849365,
+      "learning_rate": 2.968749796616214e-06,
+      "loss": 0.7304,
+      "step": 39565
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.2362515926361084,
+      "learning_rate": 2.9685088786291196e-06,
+      "loss": 0.5533,
+      "step": 39566
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.37343430519104,
+      "learning_rate": 2.9682679662908403e-06,
+      "loss": 0.6036,
+      "step": 39567
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.1001319885253906,
+      "learning_rate": 2.9680270596020467e-06,
+      "loss": 0.6087,
+      "step": 39568
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.17801833152771,
+      "learning_rate": 2.9677861585634092e-06,
+      "loss": 0.6703,
+      "step": 39569
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.551621198654175,
+      "learning_rate": 2.9675452631755962e-06,
+      "loss": 0.5805,
+      "step": 39570
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.0786292552948,
+      "learning_rate": 2.9673043734392794e-06,
+      "loss": 0.5442,
+      "step": 39571
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.9485325813293457,
+      "learning_rate": 2.9670634893551253e-06,
+      "loss": 0.7558,
+      "step": 39572
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.1624956130981445,
+      "learning_rate": 2.9668226109238086e-06,
+      "loss": 0.8909,
+      "step": 39573
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.391901969909668,
+      "learning_rate": 2.966581738145996e-06,
+      "loss": 0.665,
+      "step": 39574
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 6.765557765960693,
+      "learning_rate": 2.9663408710223574e-06,
+      "loss": 0.6437,
+      "step": 39575
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.055382013320923,
+      "learning_rate": 2.966100009553564e-06,
+      "loss": 0.5683,
+      "step": 39576
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.775805950164795,
+      "learning_rate": 2.9658591537402837e-06,
+      "loss": 0.6976,
+      "step": 39577
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.2935984134674072,
+      "learning_rate": 2.9656183035831877e-06,
+      "loss": 0.6329,
+      "step": 39578
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.69338059425354,
+      "learning_rate": 2.965377459082945e-06,
+      "loss": 0.6034,
+      "step": 39579
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.797471523284912,
+      "learning_rate": 2.9651366202402254e-06,
+      "loss": 0.5174,
+      "step": 39580
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 4.547884941101074,
+      "learning_rate": 2.964895787055698e-06,
+      "loss": 0.5589,
+      "step": 39581
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 4.904017925262451,
+      "learning_rate": 2.9646549595300334e-06,
+      "loss": 0.5391,
+      "step": 39582
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.724527597427368,
+      "learning_rate": 2.9644141376639013e-06,
+      "loss": 0.7184,
+      "step": 39583
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.8010092973709106,
+      "learning_rate": 2.9641733214579704e-06,
+      "loss": 0.4859,
+      "step": 39584
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.5833474397659302,
+      "learning_rate": 2.96393251091291e-06,
+      "loss": 0.4213,
+      "step": 39585
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.4533557891845703,
+      "learning_rate": 2.963691706029392e-06,
+      "loss": 0.4776,
+      "step": 39586
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.8655264377593994,
+      "learning_rate": 2.963450906808084e-06,
+      "loss": 0.6169,
+      "step": 39587
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.0939152240753174,
+      "learning_rate": 2.963210113249656e-06,
+      "loss": 0.5477,
+      "step": 39588
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 4.1936163902282715,
+      "learning_rate": 2.9629693253547787e-06,
+      "loss": 0.6329,
+      "step": 39589
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.188737630844116,
+      "learning_rate": 2.9627285431241195e-06,
+      "loss": 0.8071,
+      "step": 39590
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.275336980819702,
+      "learning_rate": 2.9624877665583496e-06,
+      "loss": 0.6171,
+      "step": 39591
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.5972260236740112,
+      "learning_rate": 2.962246995658138e-06,
+      "loss": 0.7232,
+      "step": 39592
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.929098606109619,
+      "learning_rate": 2.9620062304241527e-06,
+      "loss": 0.6554,
+      "step": 39593
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.4093892574310303,
+      "learning_rate": 2.9617654708570657e-06,
+      "loss": 0.6636,
+      "step": 39594
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.874006509780884,
+      "learning_rate": 2.9615247169575456e-06,
+      "loss": 0.9664,
+      "step": 39595
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.427048683166504,
+      "learning_rate": 2.9612839687262607e-06,
+      "loss": 0.502,
+      "step": 39596
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.1921346187591553,
+      "learning_rate": 2.9610432261638822e-06,
+      "loss": 0.7195,
+      "step": 39597
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.5083839893341064,
+      "learning_rate": 2.9608024892710786e-06,
+      "loss": 0.7164,
+      "step": 39598
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 4.333594799041748,
+      "learning_rate": 2.96056175804852e-06,
+      "loss": 0.6675,
+      "step": 39599
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.5212807655334473,
+      "learning_rate": 2.9603210324968745e-06,
+      "loss": 0.5405,
+      "step": 39600
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.7622482776641846,
+      "learning_rate": 2.960080312616812e-06,
+      "loss": 0.6444,
+      "step": 39601
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.380323886871338,
+      "learning_rate": 2.9598395984090024e-06,
+      "loss": 0.3537,
+      "step": 39602
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 4.5265583992004395,
+      "learning_rate": 2.9595988898741144e-06,
+      "loss": 0.8578,
+      "step": 39603
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.3195040225982666,
+      "learning_rate": 2.9593581870128173e-06,
+      "loss": 0.7578,
+      "step": 39604
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.264875650405884,
+      "learning_rate": 2.9591174898257813e-06,
+      "loss": 0.5625,
+      "step": 39605
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.8014559745788574,
+      "learning_rate": 2.9588767983136746e-06,
+      "loss": 0.6414,
+      "step": 39606
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.1544189453125,
+      "learning_rate": 2.958636112477167e-06,
+      "loss": 0.4147,
+      "step": 39607
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.1414132118225098,
+      "learning_rate": 2.958395432316928e-06,
+      "loss": 0.7003,
+      "step": 39608
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.9131706953048706,
+      "learning_rate": 2.958154757833626e-06,
+      "loss": 0.6956,
+      "step": 39609
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.6021342277526855,
+      "learning_rate": 2.957914089027931e-06,
+      "loss": 0.8194,
+      "step": 39610
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.0030417442321777,
+      "learning_rate": 2.9576734259005113e-06,
+      "loss": 0.5567,
+      "step": 39611
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 4.619246006011963,
+      "learning_rate": 2.9574327684520375e-06,
+      "loss": 0.5494,
+      "step": 39612
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.2945597171783447,
+      "learning_rate": 2.957192116683176e-06,
+      "loss": 0.3635,
+      "step": 39613
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.97562575340271,
+      "learning_rate": 2.9569514705945995e-06,
+      "loss": 0.3689,
+      "step": 39614
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.8571034669876099,
+      "learning_rate": 2.9567108301869763e-06,
+      "loss": 0.577,
+      "step": 39615
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.3763816356658936,
+      "learning_rate": 2.9564701954609744e-06,
+      "loss": 0.6981,
+      "step": 39616
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 3.532323122024536,
+      "learning_rate": 2.956229566417263e-06,
+      "loss": 0.6337,
+      "step": 39617
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.746342897415161,
+      "learning_rate": 2.955988943056512e-06,
+      "loss": 0.738,
+      "step": 39618
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 3.119415283203125,
+      "learning_rate": 2.9557483253793896e-06,
+      "loss": 0.6181,
+      "step": 39619
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.7586276531219482,
+      "learning_rate": 2.955507713386566e-06,
+      "loss": 0.4756,
+      "step": 39620
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.208874225616455,
+      "learning_rate": 2.9552671070787094e-06,
+      "loss": 0.7515,
+      "step": 39621
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.8988759517669678,
+      "learning_rate": 2.955026506456489e-06,
+      "loss": 0.6733,
+      "step": 39622
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.0163424015045166,
+      "learning_rate": 2.9547859115205736e-06,
+      "loss": 0.5438,
+      "step": 39623
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 3.5958211421966553,
+      "learning_rate": 2.9545453222716326e-06,
+      "loss": 0.6719,
+      "step": 39624
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 3.343761920928955,
+      "learning_rate": 2.954304738710334e-06,
+      "loss": 0.494,
+      "step": 39625
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.8381106853485107,
+      "learning_rate": 2.9540641608373486e-06,
+      "loss": 0.4575,
+      "step": 39626
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.124643087387085,
+      "learning_rate": 2.953823588653344e-06,
+      "loss": 0.6712,
+      "step": 39627
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 3.7627434730529785,
+      "learning_rate": 2.9535830221589894e-06,
+      "loss": 0.4852,
+      "step": 39628
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.7585431337356567,
+      "learning_rate": 2.9533424613549546e-06,
+      "loss": 0.4243,
+      "step": 39629
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.083998441696167,
+      "learning_rate": 2.9531019062419063e-06,
+      "loss": 0.6325,
+      "step": 39630
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 12.372021675109863,
+      "learning_rate": 2.9528613568205155e-06,
+      "loss": 0.6377,
+      "step": 39631
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.9397414922714233,
+      "learning_rate": 2.9526208130914502e-06,
+      "loss": 0.6575,
+      "step": 39632
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.2520487308502197,
+      "learning_rate": 2.952380275055378e-06,
+      "loss": 0.5531,
+      "step": 39633
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.2840840816497803,
+      "learning_rate": 2.952139742712971e-06,
+      "loss": 0.5191,
+      "step": 39634
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.6686458587646484,
+      "learning_rate": 2.9518992160648965e-06,
+      "loss": 0.5837,
+      "step": 39635
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 3.3390045166015625,
+      "learning_rate": 2.951658695111822e-06,
+      "loss": 0.5003,
+      "step": 39636
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.9819605350494385,
+      "learning_rate": 2.9514181798544184e-06,
+      "loss": 0.7007,
+      "step": 39637
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.2726924419403076,
+      "learning_rate": 2.9511776702933524e-06,
+      "loss": 0.3352,
+      "step": 39638
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.9121170043945312,
+      "learning_rate": 2.9509371664292953e-06,
+      "loss": 0.5326,
+      "step": 39639
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 3.113564968109131,
+      "learning_rate": 2.9506966682629133e-06,
+      "loss": 0.4363,
+      "step": 39640
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.403528928756714,
+      "learning_rate": 2.9504561757948764e-06,
+      "loss": 0.8217,
+      "step": 39641
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 3.00949764251709,
+      "learning_rate": 2.9502156890258525e-06,
+      "loss": 0.7135,
+      "step": 39642
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.665223240852356,
+      "learning_rate": 2.949975207956512e-06,
+      "loss": 0.6201,
+      "step": 39643
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 10.308539390563965,
+      "learning_rate": 2.9497347325875212e-06,
+      "loss": 0.6583,
+      "step": 39644
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.976017951965332,
+      "learning_rate": 2.9494942629195513e-06,
+      "loss": 0.6038,
+      "step": 39645
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.1871066093444824,
+      "learning_rate": 2.9492537989532693e-06,
+      "loss": 0.4753,
+      "step": 39646
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.8987706899642944,
+      "learning_rate": 2.9490133406893427e-06,
+      "loss": 0.6089,
+      "step": 39647
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.905625343322754,
+      "learning_rate": 2.948772888128444e-06,
+      "loss": 0.5805,
+      "step": 39648
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 4.6704487800598145,
+      "learning_rate": 2.948532441271238e-06,
+      "loss": 0.7976,
+      "step": 39649
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.3176701068878174,
+      "learning_rate": 2.9482920001183957e-06,
+      "loss": 0.6839,
+      "step": 39650
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.9632740020751953,
+      "learning_rate": 2.9480515646705844e-06,
+      "loss": 0.5877,
+      "step": 39651
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.8247127532958984,
+      "learning_rate": 2.9478111349284734e-06,
+      "loss": 0.4702,
+      "step": 39652
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 3.2242114543914795,
+      "learning_rate": 2.9475707108927306e-06,
+      "loss": 0.7068,
+      "step": 39653
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 4.231012344360352,
+      "learning_rate": 2.9473302925640235e-06,
+      "loss": 0.7743,
+      "step": 39654
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.3014113903045654,
+      "learning_rate": 2.9470898799430237e-06,
+      "loss": 0.6889,
+      "step": 39655
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.082794427871704,
+      "learning_rate": 2.946849473030397e-06,
+      "loss": 0.6813,
+      "step": 39656
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 3.876113176345825,
+      "learning_rate": 2.9466090718268133e-06,
+      "loss": 0.7546,
+      "step": 39657
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.1907691955566406,
+      "learning_rate": 2.946368676332939e-06,
+      "loss": 0.5526,
+      "step": 39658
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.835172414779663,
+      "learning_rate": 2.9461282865494457e-06,
+      "loss": 0.7193,
+      "step": 39659
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.6229844093322754,
+      "learning_rate": 2.9458879024770002e-06,
+      "loss": 0.5405,
+      "step": 39660
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.195463180541992,
+      "learning_rate": 2.9456475241162708e-06,
+      "loss": 0.6127,
+      "step": 39661
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.514192581176758,
+      "learning_rate": 2.945407151467926e-06,
+      "loss": 0.7113,
+      "step": 39662
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.8664329051971436,
+      "learning_rate": 2.9451667845326346e-06,
+      "loss": 0.7961,
+      "step": 39663
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.3672099113464355,
+      "learning_rate": 2.944926423311064e-06,
+      "loss": 0.6857,
+      "step": 39664
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.4184625148773193,
+      "learning_rate": 2.9446860678038837e-06,
+      "loss": 0.7096,
+      "step": 39665
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.9886990785598755,
+      "learning_rate": 2.9444457180117612e-06,
+      "loss": 0.8041,
+      "step": 39666
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.8930273056030273,
+      "learning_rate": 2.9442053739353654e-06,
+      "loss": 0.4973,
+      "step": 39667
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.4129128456115723,
+      "learning_rate": 2.943965035575364e-06,
+      "loss": 0.7265,
+      "step": 39668
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.2959179878234863,
+      "learning_rate": 2.9437247029324256e-06,
+      "loss": 0.6272,
+      "step": 39669
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.831981658935547,
+      "learning_rate": 2.943484376007219e-06,
+      "loss": 0.6633,
+      "step": 39670
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.6215012073516846,
+      "learning_rate": 2.9432440548004114e-06,
+      "loss": 0.3597,
+      "step": 39671
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.3828262090682983,
+      "learning_rate": 2.943003739312672e-06,
+      "loss": 0.4922,
+      "step": 39672
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.7418439388275146,
+      "learning_rate": 2.9427634295446687e-06,
+      "loss": 0.4299,
+      "step": 39673
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 3.6342713832855225,
+      "learning_rate": 2.9425231254970674e-06,
+      "loss": 0.4671,
+      "step": 39674
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 3.099917411804199,
+      "learning_rate": 2.9422828271705406e-06,
+      "loss": 0.5867,
+      "step": 39675
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.425208806991577,
+      "learning_rate": 2.9420425345657544e-06,
+      "loss": 0.692,
+      "step": 39676
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.8336288928985596,
+      "learning_rate": 2.9418022476833774e-06,
+      "loss": 0.6173,
+      "step": 39677
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.9821676015853882,
+      "learning_rate": 2.941561966524077e-06,
+      "loss": 0.5738,
+      "step": 39678
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.106992483139038,
+      "learning_rate": 2.941321691088521e-06,
+      "loss": 0.4893,
+      "step": 39679
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.947711229324341,
+      "learning_rate": 2.9410814213773794e-06,
+      "loss": 0.7325,
+      "step": 39680
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.6157400608062744,
+      "learning_rate": 2.940841157391318e-06,
+      "loss": 0.617,
+      "step": 39681
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.6878671646118164,
+      "learning_rate": 2.9406008991310066e-06,
+      "loss": 0.7034,
+      "step": 39682
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.0262253284454346,
+      "learning_rate": 2.9403606465971125e-06,
+      "loss": 0.597,
+      "step": 39683
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.9025706052780151,
+      "learning_rate": 2.9401203997903038e-06,
+      "loss": 0.5904,
+      "step": 39684
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.2319602966308594,
+      "learning_rate": 2.9398801587112487e-06,
+      "loss": 0.5707,
+      "step": 39685
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.0788278579711914,
+      "learning_rate": 2.939639923360615e-06,
+      "loss": 0.5668,
+      "step": 39686
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.8042287826538086,
+      "learning_rate": 2.939399693739071e-06,
+      "loss": 0.4967,
+      "step": 39687
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 3.5014188289642334,
+      "learning_rate": 2.939159469847285e-06,
+      "loss": 0.5842,
+      "step": 39688
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.227186441421509,
+      "learning_rate": 2.9389192516859235e-06,
+      "loss": 0.6078,
+      "step": 39689
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 3.910309314727783,
+      "learning_rate": 2.938679039255656e-06,
+      "loss": 0.8388,
+      "step": 39690
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 4.9849019050598145,
+      "learning_rate": 2.93843883255715e-06,
+      "loss": 0.4376,
+      "step": 39691
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.9348301887512207,
+      "learning_rate": 2.9381986315910726e-06,
+      "loss": 0.5426,
+      "step": 39692
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.0300567150115967,
+      "learning_rate": 2.937958436358093e-06,
+      "loss": 0.4888,
+      "step": 39693
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 3.0822267532348633,
+      "learning_rate": 2.9377182468588768e-06,
+      "loss": 0.4369,
+      "step": 39694
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.318472146987915,
+      "learning_rate": 2.937478063094095e-06,
+      "loss": 0.5443,
+      "step": 39695
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.9593923091888428,
+      "learning_rate": 2.9372378850644144e-06,
+      "loss": 0.4349,
+      "step": 39696
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 4.7502899169921875,
+      "learning_rate": 2.936997712770503e-06,
+      "loss": 0.6218,
+      "step": 39697
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.3637757301330566,
+      "learning_rate": 2.936757546213028e-06,
+      "loss": 0.527,
+      "step": 39698
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.5960537195205688,
+      "learning_rate": 2.9365173853926566e-06,
+      "loss": 0.6013,
+      "step": 39699
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.846662998199463,
+      "learning_rate": 2.936277230310058e-06,
+      "loss": 0.6321,
+      "step": 39700
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.211076021194458,
+      "learning_rate": 2.936037080965899e-06,
+      "loss": 0.6012,
+      "step": 39701
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 3.156461477279663,
+      "learning_rate": 2.9357969373608476e-06,
+      "loss": 0.6682,
+      "step": 39702
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 3.7176687717437744,
+      "learning_rate": 2.935556799495572e-06,
+      "loss": 0.7863,
+      "step": 39703
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.2933151721954346,
+      "learning_rate": 2.9353166673707388e-06,
+      "loss": 0.5486,
+      "step": 39704
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.4963369369506836,
+      "learning_rate": 2.9350765409870175e-06,
+      "loss": 0.5492,
+      "step": 39705
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 3.6134870052337646,
+      "learning_rate": 2.9348364203450747e-06,
+      "loss": 0.6087,
+      "step": 39706
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 3.64113187789917,
+      "learning_rate": 2.9345963054455777e-06,
+      "loss": 0.5837,
+      "step": 39707
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.208707094192505,
+      "learning_rate": 2.934356196289194e-06,
+      "loss": 0.5896,
+      "step": 39708
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.211085557937622,
+      "learning_rate": 2.9341160928765926e-06,
+      "loss": 0.8425,
+      "step": 39709
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 3.0317769050598145,
+      "learning_rate": 2.9338759952084407e-06,
+      "loss": 0.5716,
+      "step": 39710
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.9733229875564575,
+      "learning_rate": 2.9336359032854057e-06,
+      "loss": 0.568,
+      "step": 39711
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.601289987564087,
+      "learning_rate": 2.9333958171081556e-06,
+      "loss": 0.6545,
+      "step": 39712
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.6824710369110107,
+      "learning_rate": 2.9331557366773567e-06,
+      "loss": 0.6851,
+      "step": 39713
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 3.9228947162628174,
+      "learning_rate": 2.932915661993678e-06,
+      "loss": 0.5158,
+      "step": 39714
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.9204643964767456,
+      "learning_rate": 2.9326755930577848e-06,
+      "loss": 0.4304,
+      "step": 39715
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 3.296414375305176,
+      "learning_rate": 2.9324355298703476e-06,
+      "loss": 0.7016,
+      "step": 39716
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.8710696697235107,
+      "learning_rate": 2.9321954724320335e-06,
+      "loss": 0.6537,
+      "step": 39717
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.422581434249878,
+      "learning_rate": 2.9319554207435087e-06,
+      "loss": 0.7359,
+      "step": 39718
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 3.7639567852020264,
+      "learning_rate": 2.9317153748054406e-06,
+      "loss": 0.5452,
+      "step": 39719
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.273385763168335,
+      "learning_rate": 2.9314753346184966e-06,
+      "loss": 0.7155,
+      "step": 39720
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 3.6805307865142822,
+      "learning_rate": 2.931235300183346e-06,
+      "loss": 0.5505,
+      "step": 39721
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 4.246488571166992,
+      "learning_rate": 2.930995271500655e-06,
+      "loss": 0.6852,
+      "step": 39722
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.9988415241241455,
+      "learning_rate": 2.930755248571091e-06,
+      "loss": 0.4606,
+      "step": 39723
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 3.6141276359558105,
+      "learning_rate": 2.9305152313953217e-06,
+      "loss": 0.9355,
+      "step": 39724
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.3892934322357178,
+      "learning_rate": 2.9302752199740135e-06,
+      "loss": 0.7901,
+      "step": 39725
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.206151008605957,
+      "learning_rate": 2.930035214307836e-06,
+      "loss": 0.8762,
+      "step": 39726
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.2432708740234375,
+      "learning_rate": 2.929795214397454e-06,
+      "loss": 0.7257,
+      "step": 39727
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.1989667415618896,
+      "learning_rate": 2.9295552202435364e-06,
+      "loss": 0.678,
+      "step": 39728
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.7844159603118896,
+      "learning_rate": 2.92931523184675e-06,
+      "loss": 0.7094,
+      "step": 39729
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.6948037147521973,
+      "learning_rate": 2.9290752492077623e-06,
+      "loss": 0.5298,
+      "step": 39730
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.5359723567962646,
+      "learning_rate": 2.9288352723272404e-06,
+      "loss": 0.5424,
+      "step": 39731
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.2641732692718506,
+      "learning_rate": 2.9285953012058523e-06,
+      "loss": 0.6474,
+      "step": 39732
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.2735085487365723,
+      "learning_rate": 2.928355335844264e-06,
+      "loss": 0.6951,
+      "step": 39733
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.481149673461914,
+      "learning_rate": 2.9281153762431436e-06,
+      "loss": 0.8304,
+      "step": 39734
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.371227502822876,
+      "learning_rate": 2.9278754224031563e-06,
+      "loss": 0.5708,
+      "step": 39735
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.035543441772461,
+      "learning_rate": 2.9276354743249734e-06,
+      "loss": 0.6315,
+      "step": 39736
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.4242429733276367,
+      "learning_rate": 2.9273955320092594e-06,
+      "loss": 0.6132,
+      "step": 39737
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 3.001673460006714,
+      "learning_rate": 2.9271555954566825e-06,
+      "loss": 0.4418,
+      "step": 39738
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.7238409519195557,
+      "learning_rate": 2.9269156646679096e-06,
+      "loss": 0.7356,
+      "step": 39739
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 3.339383363723755,
+      "learning_rate": 2.9266757396436075e-06,
+      "loss": 0.453,
+      "step": 39740
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.3558902740478516,
+      "learning_rate": 2.926435820384443e-06,
+      "loss": 0.6463,
+      "step": 39741
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 4.155591011047363,
+      "learning_rate": 2.926195906891084e-06,
+      "loss": 0.6811,
+      "step": 39742
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 10.423053741455078,
+      "learning_rate": 2.925955999164197e-06,
+      "loss": 0.5939,
+      "step": 39743
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.171536684036255,
+      "learning_rate": 2.9257160972044497e-06,
+      "loss": 0.6307,
+      "step": 39744
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.2325656414031982,
+      "learning_rate": 2.925476201012509e-06,
+      "loss": 0.5192,
+      "step": 39745
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.9565577507019043,
+      "learning_rate": 2.9252363105890417e-06,
+      "loss": 0.6162,
+      "step": 39746
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.5633437633514404,
+      "learning_rate": 2.9249964259347153e-06,
+      "loss": 0.5365,
+      "step": 39747
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.49029803276062,
+      "learning_rate": 2.924756547050196e-06,
+      "loss": 0.6454,
+      "step": 39748
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 3.175895929336548,
+      "learning_rate": 2.9245166739361518e-06,
+      "loss": 0.5813,
+      "step": 39749
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.8086175918579102,
+      "learning_rate": 2.924276806593249e-06,
+      "loss": 0.3483,
+      "step": 39750
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.7263948917388916,
+      "learning_rate": 2.924036945022155e-06,
+      "loss": 0.8636,
+      "step": 39751
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 4.457703590393066,
+      "learning_rate": 2.923797089223537e-06,
+      "loss": 0.5783,
+      "step": 39752
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 3.249814748764038,
+      "learning_rate": 2.9235572391980603e-06,
+      "loss": 0.6493,
+      "step": 39753
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.88759183883667,
+      "learning_rate": 2.923317394946394e-06,
+      "loss": 0.5728,
+      "step": 39754
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.906964898109436,
+      "learning_rate": 2.9230775564692015e-06,
+      "loss": 0.5388,
+      "step": 39755
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 3.2422492504119873,
+      "learning_rate": 2.9228377237671547e-06,
+      "loss": 0.8042,
+      "step": 39756
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.6037495136260986,
+      "learning_rate": 2.922597896840918e-06,
+      "loss": 0.7446,
+      "step": 39757
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.188183546066284,
+      "learning_rate": 2.9223580756911584e-06,
+      "loss": 0.5381,
+      "step": 39758
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 3.528945207595825,
+      "learning_rate": 2.922118260318543e-06,
+      "loss": 0.6606,
+      "step": 39759
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 3.26639461517334,
+      "learning_rate": 2.9218784507237375e-06,
+      "loss": 0.7651,
+      "step": 39760
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.7148685455322266,
+      "learning_rate": 2.9216386469074098e-06,
+      "loss": 0.6826,
+      "step": 39761
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.951446056365967,
+      "learning_rate": 2.921398848870227e-06,
+      "loss": 0.5919,
+      "step": 39762
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.6965482234954834,
+      "learning_rate": 2.9211590566128544e-06,
+      "loss": 0.466,
+      "step": 39763
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 4.308414459228516,
+      "learning_rate": 2.9209192701359607e-06,
+      "loss": 0.6434,
+      "step": 39764
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.935097098350525,
+      "learning_rate": 2.9206794894402114e-06,
+      "loss": 0.7025,
+      "step": 39765
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.691423773765564,
+      "learning_rate": 2.920439714526273e-06,
+      "loss": 0.3759,
+      "step": 39766
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.399205446243286,
+      "learning_rate": 2.920199945394813e-06,
+      "loss": 0.6305,
+      "step": 39767
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.776615619659424,
+      "learning_rate": 2.9199601820464985e-06,
+      "loss": 0.6742,
+      "step": 39768
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.2994225025177,
+      "learning_rate": 2.9197204244819945e-06,
+      "loss": 0.5968,
+      "step": 39769
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.797102689743042,
+      "learning_rate": 2.9194806727019688e-06,
+      "loss": 0.8764,
+      "step": 39770
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.1208503246307373,
+      "learning_rate": 2.9192409267070874e-06,
+      "loss": 0.5371,
+      "step": 39771
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.396237850189209,
+      "learning_rate": 2.91900118649802e-06,
+      "loss": 0.7121,
+      "step": 39772
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 3.1059155464172363,
+      "learning_rate": 2.918761452075428e-06,
+      "loss": 0.7357,
+      "step": 39773
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.2207915782928467,
+      "learning_rate": 2.918521723439983e-06,
+      "loss": 0.4999,
+      "step": 39774
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.571835994720459,
+      "learning_rate": 2.9182820005923475e-06,
+      "loss": 0.5035,
+      "step": 39775
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 5.119560241699219,
+      "learning_rate": 2.9180422835331895e-06,
+      "loss": 0.7603,
+      "step": 39776
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 5.015079021453857,
+      "learning_rate": 2.9178025722631764e-06,
+      "loss": 0.7116,
+      "step": 39777
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.8439176082611084,
+      "learning_rate": 2.9175628667829757e-06,
+      "loss": 0.5864,
+      "step": 39778
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.2862894535064697,
+      "learning_rate": 2.917323167093251e-06,
+      "loss": 0.5762,
+      "step": 39779
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.4246623516082764,
+      "learning_rate": 2.917083473194673e-06,
+      "loss": 0.8879,
+      "step": 39780
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.4791673421859741,
+      "learning_rate": 2.9168437850879026e-06,
+      "loss": 0.4606,
+      "step": 39781
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.9264119863510132,
+      "learning_rate": 2.9166041027736113e-06,
+      "loss": 0.6351,
+      "step": 39782
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 3.7388317584991455,
+      "learning_rate": 2.9163644262524625e-06,
+      "loss": 0.4174,
+      "step": 39783
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 3.4820706844329834,
+      "learning_rate": 2.916124755525125e-06,
+      "loss": 0.8402,
+      "step": 39784
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.4656081199645996,
+      "learning_rate": 2.915885090592262e-06,
+      "loss": 0.5602,
+      "step": 39785
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.984602689743042,
+      "learning_rate": 2.9156454314545435e-06,
+      "loss": 0.5814,
+      "step": 39786
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.9327847957611084,
+      "learning_rate": 2.9154057781126326e-06,
+      "loss": 0.584,
+      "step": 39787
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 5.917736530303955,
+      "learning_rate": 2.9151661305671997e-06,
+      "loss": 0.6801,
+      "step": 39788
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.6134202480316162,
+      "learning_rate": 2.914926488818906e-06,
+      "loss": 0.4004,
+      "step": 39789
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.2037367820739746,
+      "learning_rate": 2.914686852868423e-06,
+      "loss": 0.5577,
+      "step": 39790
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.3121771812438965,
+      "learning_rate": 2.914447222716413e-06,
+      "loss": 0.6992,
+      "step": 39791
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.004925489425659,
+      "learning_rate": 2.9142075983635455e-06,
+      "loss": 0.5015,
+      "step": 39792
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.5048134326934814,
+      "learning_rate": 2.9139679798104834e-06,
+      "loss": 0.775,
+      "step": 39793
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 3.311296224594116,
+      "learning_rate": 2.9137283670578975e-06,
+      "loss": 0.9612,
+      "step": 39794
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.4094502925872803,
+      "learning_rate": 2.9134887601064488e-06,
+      "loss": 0.5837,
+      "step": 39795
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.9955108165740967,
+      "learning_rate": 2.9132491589568064e-06,
+      "loss": 0.6407,
+      "step": 39796
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 3.617750406265259,
+      "learning_rate": 2.9130095636096367e-06,
+      "loss": 0.7146,
+      "step": 39797
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.9616779088974,
+      "learning_rate": 2.912769974065607e-06,
+      "loss": 0.3645,
+      "step": 39798
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.5606448650360107,
+      "learning_rate": 2.9125303903253805e-06,
+      "loss": 0.5924,
+      "step": 39799
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.4916398525238037,
+      "learning_rate": 2.9122908123896264e-06,
+      "loss": 0.4625,
+      "step": 39800
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.841046094894409,
+      "learning_rate": 2.9120512402590085e-06,
+      "loss": 0.6061,
+      "step": 39801
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.08278226852417,
+      "learning_rate": 2.911811673934195e-06,
+      "loss": 0.372,
+      "step": 39802
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 3.006477117538452,
+      "learning_rate": 2.9115721134158497e-06,
+      "loss": 0.5972,
+      "step": 39803
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 5.093843460083008,
+      "learning_rate": 2.9113325587046413e-06,
+      "loss": 0.7445,
+      "step": 39804
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.449101686477661,
+      "learning_rate": 2.9110930098012334e-06,
+      "loss": 0.7366,
+      "step": 39805
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 3.4333248138427734,
+      "learning_rate": 2.910853466706295e-06,
+      "loss": 0.5272,
+      "step": 39806
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 3.4149227142333984,
+      "learning_rate": 2.910613929420488e-06,
+      "loss": 0.5825,
+      "step": 39807
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.2871882915496826,
+      "learning_rate": 2.910374397944482e-06,
+      "loss": 0.7194,
+      "step": 39808
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.2651989459991455,
+      "learning_rate": 2.9101348722789434e-06,
+      "loss": 0.7264,
+      "step": 39809
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.89854097366333,
+      "learning_rate": 2.909895352424535e-06,
+      "loss": 0.4265,
+      "step": 39810
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.078219175338745,
+      "learning_rate": 2.9096558383819263e-06,
+      "loss": 0.4511,
+      "step": 39811
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.090775728225708,
+      "learning_rate": 2.9094163301517796e-06,
+      "loss": 0.5386,
+      "step": 39812
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 3.6160366535186768,
+      "learning_rate": 2.9091768277347654e-06,
+      "loss": 0.7215,
+      "step": 39813
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 4.448972702026367,
+      "learning_rate": 2.908937331131545e-06,
+      "loss": 0.7167,
+      "step": 39814
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.401731014251709,
+      "learning_rate": 2.908697840342789e-06,
+      "loss": 0.542,
+      "step": 39815
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 3.2828245162963867,
+      "learning_rate": 2.908458355369158e-06,
+      "loss": 0.7121,
+      "step": 39816
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.5675299167633057,
+      "learning_rate": 2.908218876211322e-06,
+      "loss": 0.7961,
+      "step": 39817
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.7822566032409668,
+      "learning_rate": 2.9079794028699443e-06,
+      "loss": 0.4227,
+      "step": 39818
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.0912415981292725,
+      "learning_rate": 2.9077399353456947e-06,
+      "loss": 0.8025,
+      "step": 39819
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.366201400756836,
+      "learning_rate": 2.9075004736392355e-06,
+      "loss": 0.4825,
+      "step": 39820
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.6268527507781982,
+      "learning_rate": 2.907261017751234e-06,
+      "loss": 0.7655,
+      "step": 39821
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.751073122024536,
+      "learning_rate": 2.9070215676823548e-06,
+      "loss": 0.6143,
+      "step": 39822
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.585134506225586,
+      "learning_rate": 2.906782123433266e-06,
+      "loss": 0.6373,
+      "step": 39823
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 6.842990875244141,
+      "learning_rate": 2.906542685004631e-06,
+      "loss": 0.6683,
+      "step": 39824
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.376234292984009,
+      "learning_rate": 2.906303252397117e-06,
+      "loss": 0.6737,
+      "step": 39825
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.213061809539795,
+      "learning_rate": 2.9060638256113884e-06,
+      "loss": 0.7007,
+      "step": 39826
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 4.602269172668457,
+      "learning_rate": 2.905824404648114e-06,
+      "loss": 0.536,
+      "step": 39827
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.3027114868164062,
+      "learning_rate": 2.9055849895079545e-06,
+      "loss": 0.5433,
+      "step": 39828
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 3.0907320976257324,
+      "learning_rate": 2.9053455801915816e-06,
+      "loss": 0.7978,
+      "step": 39829
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.525209426879883,
+      "learning_rate": 2.905106176699655e-06,
+      "loss": 0.8155,
+      "step": 39830
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.2720658779144287,
+      "learning_rate": 2.904866779032846e-06,
+      "loss": 0.6148,
+      "step": 39831
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.9526591300964355,
+      "learning_rate": 2.9046273871918153e-06,
+      "loss": 0.4221,
+      "step": 39832
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.8196065425872803,
+      "learning_rate": 2.904388001177233e-06,
+      "loss": 0.5701,
+      "step": 39833
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.9507970809936523,
+      "learning_rate": 2.9041486209897606e-06,
+      "loss": 0.7813,
+      "step": 39834
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.8774831295013428,
+      "learning_rate": 2.903909246630068e-06,
+      "loss": 0.9241,
+      "step": 39835
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.9498684406280518,
+      "learning_rate": 2.903669878098817e-06,
+      "loss": 0.6259,
+      "step": 39836
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.7542791366577148,
+      "learning_rate": 2.9034305153966734e-06,
+      "loss": 0.6486,
+      "step": 39837
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.107301950454712,
+      "learning_rate": 2.9031911585243057e-06,
+      "loss": 0.4583,
+      "step": 39838
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 3.4019858837127686,
+      "learning_rate": 2.902951807482379e-06,
+      "loss": 0.7574,
+      "step": 39839
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.3982198238372803,
+      "learning_rate": 2.9027124622715557e-06,
+      "loss": 0.6357,
+      "step": 39840
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.9075345993041992,
+      "learning_rate": 2.9024731228925058e-06,
+      "loss": 0.4655,
+      "step": 39841
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 3.8962628841400146,
+      "learning_rate": 2.90223378934589e-06,
+      "loss": 0.8576,
+      "step": 39842
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 4.530974864959717,
+      "learning_rate": 2.901994461632378e-06,
+      "loss": 0.6832,
+      "step": 39843
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.345245361328125,
+      "learning_rate": 2.9017551397526323e-06,
+      "loss": 0.5461,
+      "step": 39844
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.814795732498169,
+      "learning_rate": 2.9015158237073208e-06,
+      "loss": 0.4563,
+      "step": 39845
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.9225707054138184,
+      "learning_rate": 2.9012765134971066e-06,
+      "loss": 0.5521,
+      "step": 39846
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.036956787109375,
+      "learning_rate": 2.9010372091226568e-06,
+      "loss": 0.6113,
+      "step": 39847
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.582521677017212,
+      "learning_rate": 2.900797910584635e-06,
+      "loss": 0.7682,
+      "step": 39848
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.773068904876709,
+      "learning_rate": 2.9005586178837103e-06,
+      "loss": 0.6138,
+      "step": 39849
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.1079037189483643,
+      "learning_rate": 2.900319331020543e-06,
+      "loss": 0.4246,
+      "step": 39850
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.8337483406066895,
+      "learning_rate": 2.9000800499958036e-06,
+      "loss": 0.4703,
+      "step": 39851
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 4.534398555755615,
+      "learning_rate": 2.899840774810152e-06,
+      "loss": 0.6331,
+      "step": 39852
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.18241024017334,
+      "learning_rate": 2.899601505464259e-06,
+      "loss": 0.5541,
+      "step": 39853
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.9054162502288818,
+      "learning_rate": 2.8993622419587857e-06,
+      "loss": 0.7186,
+      "step": 39854
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.7491066455841064,
+      "learning_rate": 2.8991229842944005e-06,
+      "loss": 0.5726,
+      "step": 39855
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.3123416900634766,
+      "learning_rate": 2.8988837324717658e-06,
+      "loss": 0.4664,
+      "step": 39856
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.4063847064971924,
+      "learning_rate": 2.8986444864915488e-06,
+      "loss": 0.5442,
+      "step": 39857
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.549593925476074,
+      "learning_rate": 2.8984052463544132e-06,
+      "loss": 0.5858,
+      "step": 39858
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 3.3296194076538086,
+      "learning_rate": 2.898166012061028e-06,
+      "loss": 0.6246,
+      "step": 39859
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.0400843620300293,
+      "learning_rate": 2.897926783612053e-06,
+      "loss": 0.5457,
+      "step": 39860
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 3.3151328563690186,
+      "learning_rate": 2.897687561008159e-06,
+      "loss": 0.6072,
+      "step": 39861
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.676828384399414,
+      "learning_rate": 2.8974483442500056e-06,
+      "loss": 0.6211,
+      "step": 39862
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.528080463409424,
+      "learning_rate": 2.8972091333382634e-06,
+      "loss": 0.9063,
+      "step": 39863
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.1530096530914307,
+      "learning_rate": 2.896969928273593e-06,
+      "loss": 0.7059,
+      "step": 39864
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.0739777088165283,
+      "learning_rate": 2.896730729056663e-06,
+      "loss": 0.4874,
+      "step": 39865
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.6387755870819092,
+      "learning_rate": 2.8964915356881347e-06,
+      "loss": 0.4774,
+      "step": 39866
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.332756757736206,
+      "learning_rate": 2.8962523481686778e-06,
+      "loss": 0.6031,
+      "step": 39867
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 4.348408222198486,
+      "learning_rate": 2.8960131664989532e-06,
+      "loss": 0.6904,
+      "step": 39868
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.203904867172241,
+      "learning_rate": 2.8957739906796272e-06,
+      "loss": 0.5598,
+      "step": 39869
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 4.237043380737305,
+      "learning_rate": 2.8955348207113677e-06,
+      "loss": 0.6334,
+      "step": 39870
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.3861818313598633,
+      "learning_rate": 2.895295656594836e-06,
+      "loss": 0.5874,
+      "step": 39871
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.4628746509552,
+      "learning_rate": 2.8950564983307e-06,
+      "loss": 0.598,
+      "step": 39872
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.494877815246582,
+      "learning_rate": 2.8948173459196215e-06,
+      "loss": 0.4895,
+      "step": 39873
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 4.345871925354004,
+      "learning_rate": 2.89457819936227e-06,
+      "loss": 0.5462,
+      "step": 39874
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 3.789428234100342,
+      "learning_rate": 2.894339058659305e-06,
+      "loss": 0.4781,
+      "step": 39875
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.116363286972046,
+      "learning_rate": 2.8940999238113964e-06,
+      "loss": 0.5504,
+      "step": 39876
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.722256898880005,
+      "learning_rate": 2.8938607948192056e-06,
+      "loss": 0.6727,
+      "step": 39877
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 4.249483585357666,
+      "learning_rate": 2.8936216716833985e-06,
+      "loss": 0.7739,
+      "step": 39878
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 5.267790794372559,
+      "learning_rate": 2.893382554404642e-06,
+      "loss": 0.5951,
+      "step": 39879
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.4694037437438965,
+      "learning_rate": 2.893143442983598e-06,
+      "loss": 0.593,
+      "step": 39880
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.6923601627349854,
+      "learning_rate": 2.8929043374209333e-06,
+      "loss": 0.6123,
+      "step": 39881
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.0837955474853516,
+      "learning_rate": 2.8926652377173135e-06,
+      "loss": 0.6097,
+      "step": 39882
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.7169238328933716,
+      "learning_rate": 2.8924261438734008e-06,
+      "loss": 0.5559,
+      "step": 39883
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.030463695526123,
+      "learning_rate": 2.8921870558898634e-06,
+      "loss": 0.7403,
+      "step": 39884
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 4.360311031341553,
+      "learning_rate": 2.8919479737673615e-06,
+      "loss": 0.73,
+      "step": 39885
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.3509464263916016,
+      "learning_rate": 2.8917088975065653e-06,
+      "loss": 0.6685,
+      "step": 39886
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.4970345497131348,
+      "learning_rate": 2.8914698271081353e-06,
+      "loss": 0.4359,
+      "step": 39887
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.6291537284851074,
+      "learning_rate": 2.8912307625727386e-06,
+      "loss": 0.5527,
+      "step": 39888
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.1468303203582764,
+      "learning_rate": 2.890991703901038e-06,
+      "loss": 0.3753,
+      "step": 39889
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 4.434628009796143,
+      "learning_rate": 2.890752651093701e-06,
+      "loss": 1.0374,
+      "step": 39890
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.4069619178771973,
+      "learning_rate": 2.8905136041513893e-06,
+      "loss": 0.8851,
+      "step": 39891
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.9680733680725098,
+      "learning_rate": 2.890274563074771e-06,
+      "loss": 0.6985,
+      "step": 39892
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.8604629039764404,
+      "learning_rate": 2.8900355278645065e-06,
+      "loss": 0.61,
+      "step": 39893
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 6.289221286773682,
+      "learning_rate": 2.8897964985212646e-06,
+      "loss": 0.7348,
+      "step": 39894
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.1440274715423584,
+      "learning_rate": 2.8895574750457066e-06,
+      "loss": 0.6419,
+      "step": 39895
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.5518397092819214,
+      "learning_rate": 2.8893184574385004e-06,
+      "loss": 0.5084,
+      "step": 39896
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.6422817707061768,
+      "learning_rate": 2.889079445700307e-06,
+      "loss": 0.4769,
+      "step": 39897
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.3208346366882324,
+      "learning_rate": 2.8888404398317927e-06,
+      "loss": 0.3506,
+      "step": 39898
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.900862216949463,
+      "learning_rate": 2.8886014398336227e-06,
+      "loss": 0.5088,
+      "step": 39899
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.7586846351623535,
+      "learning_rate": 2.8883624457064625e-06,
+      "loss": 0.6455,
+      "step": 39900
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.0539791584014893,
+      "learning_rate": 2.8881234574509737e-06,
+      "loss": 0.4758,
+      "step": 39901
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 3.6844632625579834,
+      "learning_rate": 2.887884475067824e-06,
+      "loss": 0.5606,
+      "step": 39902
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.8145813941955566,
+      "learning_rate": 2.8876454985576752e-06,
+      "loss": 0.5096,
+      "step": 39903
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.3068196773529053,
+      "learning_rate": 2.8874065279211942e-06,
+      "loss": 0.7176,
+      "step": 39904
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 4.052393436431885,
+      "learning_rate": 2.8871675631590425e-06,
+      "loss": 0.5596,
+      "step": 39905
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.2020692825317383,
+      "learning_rate": 2.8869286042718885e-06,
+      "loss": 0.6436,
+      "step": 39906
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.5379194021224976,
+      "learning_rate": 2.8866896512603924e-06,
+      "loss": 0.6657,
+      "step": 39907
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.045797348022461,
+      "learning_rate": 2.8864507041252225e-06,
+      "loss": 0.5181,
+      "step": 39908
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 6.778326511383057,
+      "learning_rate": 2.8862117628670395e-06,
+      "loss": 0.509,
+      "step": 39909
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 3.092418909072876,
+      "learning_rate": 2.8859728274865118e-06,
+      "loss": 0.7102,
+      "step": 39910
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.188539743423462,
+      "learning_rate": 2.8857338979842995e-06,
+      "loss": 0.6828,
+      "step": 39911
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.5427908897399902,
+      "learning_rate": 2.8854949743610723e-06,
+      "loss": 0.6303,
+      "step": 39912
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.430023670196533,
+      "learning_rate": 2.8852560566174883e-06,
+      "loss": 0.4666,
+      "step": 39913
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.2861626148223877,
+      "learning_rate": 2.8850171447542175e-06,
+      "loss": 0.651,
+      "step": 39914
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.5336151123046875,
+      "learning_rate": 2.8847782387719193e-06,
+      "loss": 0.612,
+      "step": 39915
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.522043466567993,
+      "learning_rate": 2.884539338671263e-06,
+      "loss": 0.6776,
+      "step": 39916
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 4.064391136169434,
+      "learning_rate": 2.8843004444529077e-06,
+      "loss": 0.4835,
+      "step": 39917
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 4.6970062255859375,
+      "learning_rate": 2.884061556117521e-06,
+      "loss": 0.8662,
+      "step": 39918
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.1459174156188965,
+      "learning_rate": 2.883822673665766e-06,
+      "loss": 0.7362,
+      "step": 39919
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 3.1165595054626465,
+      "learning_rate": 2.8835837970983094e-06,
+      "loss": 0.6204,
+      "step": 39920
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.5142812728881836,
+      "learning_rate": 2.883344926415812e-06,
+      "loss": 0.3019,
+      "step": 39921
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.0757994651794434,
+      "learning_rate": 2.883106061618941e-06,
+      "loss": 0.5819,
+      "step": 39922
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.615989923477173,
+      "learning_rate": 2.882867202708357e-06,
+      "loss": 0.7807,
+      "step": 39923
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.890739679336548,
+      "learning_rate": 2.8826283496847287e-06,
+      "loss": 0.2749,
+      "step": 39924
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.081496477127075,
+      "learning_rate": 2.8823895025487147e-06,
+      "loss": 0.7672,
+      "step": 39925
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.9009125232696533,
+      "learning_rate": 2.8821506613009853e-06,
+      "loss": 0.6816,
+      "step": 39926
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 3.6244864463806152,
+      "learning_rate": 2.8819118259421987e-06,
+      "loss": 0.5306,
+      "step": 39927
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.2667953968048096,
+      "learning_rate": 2.8816729964730244e-06,
+      "loss": 0.7,
+      "step": 39928
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.5837392807006836,
+      "learning_rate": 2.881434172894122e-06,
+      "loss": 0.8276,
+      "step": 39929
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.2923085689544678,
+      "learning_rate": 2.8811953552061566e-06,
+      "loss": 0.4689,
+      "step": 39930
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.1372830867767334,
+      "learning_rate": 2.8809565434097963e-06,
+      "loss": 0.6695,
+      "step": 39931
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.3422391414642334,
+      "learning_rate": 2.8807177375056995e-06,
+      "loss": 0.4653,
+      "step": 39932
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 3.15738582611084,
+      "learning_rate": 2.880478937494535e-06,
+      "loss": 0.5938,
+      "step": 39933
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.9627206325531006,
+      "learning_rate": 2.8802401433769627e-06,
+      "loss": 0.5172,
+      "step": 39934
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.36557936668396,
+      "learning_rate": 2.8800013551536504e-06,
+      "loss": 0.8821,
+      "step": 39935
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.452092409133911,
+      "learning_rate": 2.8797625728252575e-06,
+      "loss": 0.5517,
+      "step": 39936
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.7808918952941895,
+      "learning_rate": 2.8795237963924537e-06,
+      "loss": 0.5973,
+      "step": 39937
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.9343228340148926,
+      "learning_rate": 2.8792850258558975e-06,
+      "loss": 0.5997,
+      "step": 39938
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.8788793087005615,
+      "learning_rate": 2.879046261216255e-06,
+      "loss": 0.6642,
+      "step": 39939
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.81160569190979,
+      "learning_rate": 2.8788075024741925e-06,
+      "loss": 0.579,
+      "step": 39940
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 3.299243927001953,
+      "learning_rate": 2.8785687496303694e-06,
+      "loss": 0.3537,
+      "step": 39941
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.8120970726013184,
+      "learning_rate": 2.8783300026854523e-06,
+      "loss": 0.8081,
+      "step": 39942
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.8173513412475586,
+      "learning_rate": 2.878091261640106e-06,
+      "loss": 0.7246,
+      "step": 39943
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 3.1511878967285156,
+      "learning_rate": 2.877852526494992e-06,
+      "loss": 0.7743,
+      "step": 39944
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.4149413108825684,
+      "learning_rate": 2.8776137972507766e-06,
+      "loss": 0.8068,
+      "step": 39945
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 3.4347240924835205,
+      "learning_rate": 2.8773750739081204e-06,
+      "loss": 0.5861,
+      "step": 39946
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 3.2363052368164062,
+      "learning_rate": 2.87713635646769e-06,
+      "loss": 0.8285,
+      "step": 39947
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 3.129652261734009,
+      "learning_rate": 2.876897644930148e-06,
+      "loss": 0.5855,
+      "step": 39948
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.1327013969421387,
+      "learning_rate": 2.8766589392961584e-06,
+      "loss": 0.4335,
+      "step": 39949
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 3.337480068206787,
+      "learning_rate": 2.876420239566384e-06,
+      "loss": 0.6771,
+      "step": 39950
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 3.2422614097595215,
+      "learning_rate": 2.876181545741491e-06,
+      "loss": 0.7862,
+      "step": 39951
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 3.2584521770477295,
+      "learning_rate": 2.8759428578221397e-06,
+      "loss": 0.728,
+      "step": 39952
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.470412254333496,
+      "learning_rate": 2.8757041758089976e-06,
+      "loss": 0.356,
+      "step": 39953
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.5033307075500488,
+      "learning_rate": 2.875465499702724e-06,
+      "loss": 0.5004,
+      "step": 39954
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.0296988487243652,
+      "learning_rate": 2.8752268295039874e-06,
+      "loss": 0.7875,
+      "step": 39955
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.5493171215057373,
+      "learning_rate": 2.874988165213447e-06,
+      "loss": 0.7646,
+      "step": 39956
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 5.270326137542725,
+      "learning_rate": 2.874749506831771e-06,
+      "loss": 0.9448,
+      "step": 39957
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 45.21434020996094,
+      "learning_rate": 2.874510854359618e-06,
+      "loss": 0.7561,
+      "step": 39958
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.474046230316162,
+      "learning_rate": 2.8742722077976537e-06,
+      "loss": 0.3218,
+      "step": 39959
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.2516751289367676,
+      "learning_rate": 2.8740335671465425e-06,
+      "loss": 0.7854,
+      "step": 39960
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 5.831579685211182,
+      "learning_rate": 2.873794932406949e-06,
+      "loss": 0.6609,
+      "step": 39961
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 7.421563625335693,
+      "learning_rate": 2.8735563035795343e-06,
+      "loss": 0.6506,
+      "step": 39962
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.080812931060791,
+      "learning_rate": 2.8733176806649643e-06,
+      "loss": 0.8029,
+      "step": 39963
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.768702507019043,
+      "learning_rate": 2.8730790636638994e-06,
+      "loss": 0.6464,
+      "step": 39964
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.0974841117858887,
+      "learning_rate": 2.872840452577006e-06,
+      "loss": 0.6767,
+      "step": 39965
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.5495452880859375,
+      "learning_rate": 2.8726018474049456e-06,
+      "loss": 0.5225,
+      "step": 39966
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.146743059158325,
+      "learning_rate": 2.8723632481483844e-06,
+      "loss": 0.5144,
+      "step": 39967
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.4744834899902344,
+      "learning_rate": 2.872124654807981e-06,
+      "loss": 0.5466,
+      "step": 39968
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.3798563480377197,
+      "learning_rate": 2.8718860673844047e-06,
+      "loss": 0.4176,
+      "step": 39969
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.0657618045806885,
+      "learning_rate": 2.871647485878314e-06,
+      "loss": 0.5316,
+      "step": 39970
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 3.8877029418945312,
+      "learning_rate": 2.8714089102903757e-06,
+      "loss": 0.554,
+      "step": 39971
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.6644670963287354,
+      "learning_rate": 2.8711703406212505e-06,
+      "loss": 0.6692,
+      "step": 39972
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.9883294105529785,
+      "learning_rate": 2.870931776871605e-06,
+      "loss": 0.3426,
+      "step": 39973
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 3.65134596824646,
+      "learning_rate": 2.8706932190420983e-06,
+      "loss": 0.9396,
+      "step": 39974
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 3.337681293487549,
+      "learning_rate": 2.8704546671333982e-06,
+      "loss": 0.68,
+      "step": 39975
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.6627771854400635,
+      "learning_rate": 2.8702161211461644e-06,
+      "loss": 0.564,
+      "step": 39976
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 3.5714030265808105,
+      "learning_rate": 2.869977581081063e-06,
+      "loss": 0.655,
+      "step": 39977
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.7364675998687744,
+      "learning_rate": 2.869739046938754e-06,
+      "loss": 0.7085,
+      "step": 39978
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.4568686485290527,
+      "learning_rate": 2.8695005187199032e-06,
+      "loss": 0.6638,
+      "step": 39979
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.2333505153656006,
+      "learning_rate": 2.8692619964251733e-06,
+      "loss": 0.4577,
+      "step": 39980
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 3.0476648807525635,
+      "learning_rate": 2.8690234800552295e-06,
+      "loss": 0.8031,
+      "step": 39981
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.5029447078704834,
+      "learning_rate": 2.868784969610732e-06,
+      "loss": 0.6421,
+      "step": 39982
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.746589422225952,
+      "learning_rate": 2.8685464650923457e-06,
+      "loss": 0.8816,
+      "step": 39983
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.2951700687408447,
+      "learning_rate": 2.868307966500732e-06,
+      "loss": 0.4675,
+      "step": 39984
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.2154431343078613,
+      "learning_rate": 2.8680694738365567e-06,
+      "loss": 0.4921,
+      "step": 39985
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.5454094409942627,
+      "learning_rate": 2.8678309871004807e-06,
+      "loss": 0.4799,
+      "step": 39986
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.40514874458313,
+      "learning_rate": 2.8675925062931687e-06,
+      "loss": 0.5935,
+      "step": 39987
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.8169450759887695,
+      "learning_rate": 2.8673540314152813e-06,
+      "loss": 0.497,
+      "step": 39988
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.695409059524536,
+      "learning_rate": 2.8671155624674863e-06,
+      "loss": 0.7779,
+      "step": 39989
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.521644353866577,
+      "learning_rate": 2.866877099450441e-06,
+      "loss": 0.5641,
+      "step": 39990
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 5.150658130645752,
+      "learning_rate": 2.866638642364814e-06,
+      "loss": 0.5972,
+      "step": 39991
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.283446788787842,
+      "learning_rate": 2.8664001912112637e-06,
+      "loss": 0.4436,
+      "step": 39992
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.9850918054580688,
+      "learning_rate": 2.8661617459904558e-06,
+      "loss": 0.7181,
+      "step": 39993
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.4777183532714844,
+      "learning_rate": 2.8659233067030534e-06,
+      "loss": 0.6003,
+      "step": 39994
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 4.607773303985596,
+      "learning_rate": 2.865684873349718e-06,
+      "loss": 0.7437,
+      "step": 39995
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.4381043910980225,
+      "learning_rate": 2.865446445931115e-06,
+      "loss": 0.7256,
+      "step": 39996
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.665151596069336,
+      "learning_rate": 2.865208024447903e-06,
+      "loss": 0.4552,
+      "step": 39997
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.898733377456665,
+      "learning_rate": 2.8649696089007505e-06,
+      "loss": 0.527,
+      "step": 39998
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.986029863357544,
+      "learning_rate": 2.864731199290316e-06,
+      "loss": 0.7912,
+      "step": 39999
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.6388614177703857,
+      "learning_rate": 2.8644927956172643e-06,
+      "loss": 0.8357,
+      "step": 40000
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.4595541954040527,
+      "learning_rate": 2.864254397882259e-06,
+      "loss": 0.6699,
+      "step": 40001
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.1666107177734375,
+      "learning_rate": 2.864016006085961e-06,
+      "loss": 0.4231,
+      "step": 40002
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.5339064598083496,
+      "learning_rate": 2.8637776202290345e-06,
+      "loss": 0.5143,
+      "step": 40003
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.7292323112487793,
+      "learning_rate": 2.8635392403121437e-06,
+      "loss": 0.6282,
+      "step": 40004
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.7557190656661987,
+      "learning_rate": 2.863300866335948e-06,
+      "loss": 0.6007,
+      "step": 40005
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.6703827381134033,
+      "learning_rate": 2.8630624983011146e-06,
+      "loss": 0.6654,
+      "step": 40006
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 3.610079526901245,
+      "learning_rate": 2.862824136208301e-06,
+      "loss": 0.9985,
+      "step": 40007
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.1826107501983643,
+      "learning_rate": 2.862585780058176e-06,
+      "loss": 0.6784,
+      "step": 40008
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.6161904335021973,
+      "learning_rate": 2.8623474298513965e-06,
+      "loss": 0.5652,
+      "step": 40009
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.595195770263672,
+      "learning_rate": 2.86210908558863e-06,
+      "loss": 0.7813,
+      "step": 40010
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.1927897930145264,
+      "learning_rate": 2.8618707472705355e-06,
+      "loss": 0.6844,
+      "step": 40011
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 4.445959091186523,
+      "learning_rate": 2.861632414897779e-06,
+      "loss": 0.6398,
+      "step": 40012
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.748743772506714,
+      "learning_rate": 2.8613940884710206e-06,
+      "loss": 1.0025,
+      "step": 40013
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.4903500080108643,
+      "learning_rate": 2.8611557679909256e-06,
+      "loss": 0.6428,
+      "step": 40014
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.538130283355713,
+      "learning_rate": 2.860917453458153e-06,
+      "loss": 0.4273,
+      "step": 40015
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 3.290402889251709,
+      "learning_rate": 2.8606791448733707e-06,
+      "loss": 0.4785,
+      "step": 40016
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.2753047943115234,
+      "learning_rate": 2.8604408422372356e-06,
+      "loss": 0.5533,
+      "step": 40017
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.6311378479003906,
+      "learning_rate": 2.8602025455504147e-06,
+      "loss": 0.7103,
+      "step": 40018
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.7725136280059814,
+      "learning_rate": 2.8599642548135665e-06,
+      "loss": 0.5983,
+      "step": 40019
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.1379754543304443,
+      "learning_rate": 2.8597259700273574e-06,
+      "loss": 0.5485,
+      "step": 40020
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.1522011756896973,
+      "learning_rate": 2.8594876911924474e-06,
+      "loss": 0.517,
+      "step": 40021
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.3330888748168945,
+      "learning_rate": 2.8592494183095024e-06,
+      "loss": 0.7622,
+      "step": 40022
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 4.0813889503479,
+      "learning_rate": 2.8590111513791815e-06,
+      "loss": 0.58,
+      "step": 40023
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.7181841135025024,
+      "learning_rate": 2.85877289040215e-06,
+      "loss": 0.5577,
+      "step": 40024
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.0498385429382324,
+      "learning_rate": 2.8585346353790674e-06,
+      "loss": 0.68,
+      "step": 40025
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.8892295360565186,
+      "learning_rate": 2.8582963863105994e-06,
+      "loss": 0.5023,
+      "step": 40026
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.1658549308776855,
+      "learning_rate": 2.8580581431974054e-06,
+      "loss": 0.6742,
+      "step": 40027
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.50217342376709,
+      "learning_rate": 2.8578199060401506e-06,
+      "loss": 0.5491,
+      "step": 40028
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 3.615199089050293,
+      "learning_rate": 2.8575816748394945e-06,
+      "loss": 0.7895,
+      "step": 40029
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 3.8373117446899414,
+      "learning_rate": 2.857343449596103e-06,
+      "loss": 0.7444,
+      "step": 40030
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 3.1443285942077637,
+      "learning_rate": 2.857105230310635e-06,
+      "loss": 0.66,
+      "step": 40031
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 3.0927345752716064,
+      "learning_rate": 2.8568670169837563e-06,
+      "loss": 0.5441,
+      "step": 40032
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.317624807357788,
+      "learning_rate": 2.856628809616126e-06,
+      "loss": 0.5123,
+      "step": 40033
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.79506254196167,
+      "learning_rate": 2.8563906082084096e-06,
+      "loss": 0.674,
+      "step": 40034
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.286414623260498,
+      "learning_rate": 2.8561524127612662e-06,
+      "loss": 0.4739,
+      "step": 40035
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.846692442893982,
+      "learning_rate": 2.8559142232753624e-06,
+      "loss": 0.6615,
+      "step": 40036
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.588106155395508,
+      "learning_rate": 2.855676039751355e-06,
+      "loss": 0.7615,
+      "step": 40037
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.580544352531433,
+      "learning_rate": 2.8554378621899113e-06,
+      "loss": 0.4304,
+      "step": 40038
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.4304449558258057,
+      "learning_rate": 2.85519969059169e-06,
+      "loss": 0.705,
+      "step": 40039
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.0456454753875732,
+      "learning_rate": 2.8549615249573547e-06,
+      "loss": 0.5541,
+      "step": 40040
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.406114339828491,
+      "learning_rate": 2.8547233652875676e-06,
+      "loss": 0.5593,
+      "step": 40041
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.464667320251465,
+      "learning_rate": 2.8544852115829935e-06,
+      "loss": 0.8006,
+      "step": 40042
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.28019118309021,
+      "learning_rate": 2.8542470638442905e-06,
+      "loss": 0.4739,
+      "step": 40043
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 4.013853073120117,
+      "learning_rate": 2.8540089220721245e-06,
+      "loss": 0.7239,
+      "step": 40044
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.9911223649978638,
+      "learning_rate": 2.8537707862671536e-06,
+      "loss": 0.5224,
+      "step": 40045
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.9610793590545654,
+      "learning_rate": 2.8535326564300435e-06,
+      "loss": 0.412,
+      "step": 40046
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 3.0300559997558594,
+      "learning_rate": 2.853294532561454e-06,
+      "loss": 0.5051,
+      "step": 40047
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.9318783283233643,
+      "learning_rate": 2.8530564146620498e-06,
+      "loss": 0.6666,
+      "step": 40048
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.1464061737060547,
+      "learning_rate": 2.8528183027324897e-06,
+      "loss": 0.3745,
+      "step": 40049
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.9734301567077637,
+      "learning_rate": 2.8525801967734395e-06,
+      "loss": 0.7396,
+      "step": 40050
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 3.0501623153686523,
+      "learning_rate": 2.8523420967855574e-06,
+      "loss": 0.5042,
+      "step": 40051
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.0874032974243164,
+      "learning_rate": 2.852104002769509e-06,
+      "loss": 0.3811,
+      "step": 40052
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.4798786640167236,
+      "learning_rate": 2.851865914725953e-06,
+      "loss": 0.6682,
+      "step": 40053
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 3.116185426712036,
+      "learning_rate": 2.8516278326555536e-06,
+      "loss": 0.4998,
+      "step": 40054
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 3.1343319416046143,
+      "learning_rate": 2.851389756558973e-06,
+      "loss": 0.5603,
+      "step": 40055
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.432847499847412,
+      "learning_rate": 2.8511516864368723e-06,
+      "loss": 0.692,
+      "step": 40056
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 5.093734264373779,
+      "learning_rate": 2.8509136222899146e-06,
+      "loss": 0.5314,
+      "step": 40057
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.1349191665649414,
+      "learning_rate": 2.85067556411876e-06,
+      "loss": 0.4306,
+      "step": 40058
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 3.4462831020355225,
+      "learning_rate": 2.8504375119240724e-06,
+      "loss": 0.752,
+      "step": 40059
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.454864740371704,
+      "learning_rate": 2.8501994657065118e-06,
+      "loss": 0.5393,
+      "step": 40060
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.3246424198150635,
+      "learning_rate": 2.8499614254667405e-06,
+      "loss": 0.6964,
+      "step": 40061
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.258759021759033,
+      "learning_rate": 2.8497233912054235e-06,
+      "loss": 0.7055,
+      "step": 40062
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.6593680381774902,
+      "learning_rate": 2.8494853629232177e-06,
+      "loss": 0.6105,
+      "step": 40063
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 3.12882661819458,
+      "learning_rate": 2.84924734062079e-06,
+      "loss": 0.7686,
+      "step": 40064
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.9975943565368652,
+      "learning_rate": 2.849009324298798e-06,
+      "loss": 0.5181,
+      "step": 40065
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.3589987754821777,
+      "learning_rate": 2.8487713139579046e-06,
+      "loss": 0.347,
+      "step": 40066
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.3629214763641357,
+      "learning_rate": 2.848533309598775e-06,
+      "loss": 0.6252,
+      "step": 40067
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.9586654901504517,
+      "learning_rate": 2.8482953112220657e-06,
+      "loss": 0.5043,
+      "step": 40068
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.9724520444869995,
+      "learning_rate": 2.8480573188284434e-06,
+      "loss": 0.6216,
+      "step": 40069
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 3.683028221130371,
+      "learning_rate": 2.8478193324185655e-06,
+      "loss": 0.6982,
+      "step": 40070
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.99442720413208,
+      "learning_rate": 2.8475813519930977e-06,
+      "loss": 0.611,
+      "step": 40071
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.6707046031951904,
+      "learning_rate": 2.8473433775526983e-06,
+      "loss": 0.7355,
+      "step": 40072
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 4.054272174835205,
+      "learning_rate": 2.847105409098032e-06,
+      "loss": 0.7366,
+      "step": 40073
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 3.5100677013397217,
+      "learning_rate": 2.8468674466297576e-06,
+      "loss": 0.9318,
+      "step": 40074
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.3169772624969482,
+      "learning_rate": 2.846629490148539e-06,
+      "loss": 0.6832,
+      "step": 40075
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.8105988502502441,
+      "learning_rate": 2.8463915396550367e-06,
+      "loss": 0.5809,
+      "step": 40076
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.009805679321289,
+      "learning_rate": 2.846153595149914e-06,
+      "loss": 0.5154,
+      "step": 40077
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 3.230058431625366,
+      "learning_rate": 2.845915656633829e-06,
+      "loss": 0.643,
+      "step": 40078
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.8287100791931152,
+      "learning_rate": 2.8456777241074485e-06,
+      "loss": 0.4898,
+      "step": 40079
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.8189570903778076,
+      "learning_rate": 2.845439797571429e-06,
+      "loss": 0.7545,
+      "step": 40080
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.9217184782028198,
+      "learning_rate": 2.845201877026433e-06,
+      "loss": 0.446,
+      "step": 40081
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.770899534225464,
+      "learning_rate": 2.8449639624731253e-06,
+      "loss": 0.5532,
+      "step": 40082
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.8615583181381226,
+      "learning_rate": 2.8447260539121664e-06,
+      "loss": 0.4666,
+      "step": 40083
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.1784465312957764,
+      "learning_rate": 2.844488151344215e-06,
+      "loss": 0.6753,
+      "step": 40084
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.319798469543457,
+      "learning_rate": 2.844250254769937e-06,
+      "loss": 0.7229,
+      "step": 40085
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.8628185987472534,
+      "learning_rate": 2.8440123641899892e-06,
+      "loss": 0.5244,
+      "step": 40086
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 3.147162675857544,
+      "learning_rate": 2.843774479605037e-06,
+      "loss": 0.7072,
+      "step": 40087
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.8367581367492676,
+      "learning_rate": 2.8435366010157396e-06,
+      "loss": 0.4227,
+      "step": 40088
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 3.707552909851074,
+      "learning_rate": 2.8432987284227596e-06,
+      "loss": 0.6594,
+      "step": 40089
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.049349308013916,
+      "learning_rate": 2.8430608618267564e-06,
+      "loss": 0.6767,
+      "step": 40090
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 3.9195218086242676,
+      "learning_rate": 2.8428230012283953e-06,
+      "loss": 0.7372,
+      "step": 40091
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 4.08953332901001,
+      "learning_rate": 2.8425851466283327e-06,
+      "loss": 0.6565,
+      "step": 40092
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.5165234804153442,
+      "learning_rate": 2.8423472980272348e-06,
+      "loss": 0.5986,
+      "step": 40093
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.926727294921875,
+      "learning_rate": 2.842109455425759e-06,
+      "loss": 0.8027,
+      "step": 40094
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 3.3621039390563965,
+      "learning_rate": 2.841871618824571e-06,
+      "loss": 0.7279,
+      "step": 40095
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.1087725162506104,
+      "learning_rate": 2.8416337882243267e-06,
+      "loss": 0.4703,
+      "step": 40096
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.272904634475708,
+      "learning_rate": 2.8413959636256925e-06,
+      "loss": 0.4673,
+      "step": 40097
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.9260997772216797,
+      "learning_rate": 2.8411581450293257e-06,
+      "loss": 0.5174,
+      "step": 40098
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.1899125576019287,
+      "learning_rate": 2.8409203324358915e-06,
+      "loss": 0.7976,
+      "step": 40099
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 3.078556537628174,
+      "learning_rate": 2.8406825258460464e-06,
+      "loss": 0.7635,
+      "step": 40100
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.3093128204345703,
+      "learning_rate": 2.8404447252604545e-06,
+      "loss": 0.7171,
+      "step": 40101
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.8840904235839844,
+      "learning_rate": 2.840206930679777e-06,
+      "loss": 0.8259,
+      "step": 40102
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.558544397354126,
+      "learning_rate": 2.839969142104677e-06,
+      "loss": 0.7892,
+      "step": 40103
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.0494484901428223,
+      "learning_rate": 2.839731359535812e-06,
+      "loss": 0.5467,
+      "step": 40104
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.1686837673187256,
+      "learning_rate": 2.8394935829738455e-06,
+      "loss": 0.4618,
+      "step": 40105
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.955077886581421,
+      "learning_rate": 2.8392558124194375e-06,
+      "loss": 0.4429,
+      "step": 40106
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.727900266647339,
+      "learning_rate": 2.839018047873251e-06,
+      "loss": 0.5216,
+      "step": 40107
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 3.0301895141601562,
+      "learning_rate": 2.838780289335944e-06,
+      "loss": 0.5585,
+      "step": 40108
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.4092886447906494,
+      "learning_rate": 2.838542536808181e-06,
+      "loss": 0.7133,
+      "step": 40109
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.2306530475616455,
+      "learning_rate": 2.83830479029062e-06,
+      "loss": 0.5277,
+      "step": 40110
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.702359914779663,
+      "learning_rate": 2.8380670497839254e-06,
+      "loss": 0.8647,
+      "step": 40111
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.008625030517578,
+      "learning_rate": 2.8378293152887543e-06,
+      "loss": 0.452,
+      "step": 40112
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.917937755584717,
+      "learning_rate": 2.837591586805772e-06,
+      "loss": 0.5739,
+      "step": 40113
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 3.929137706756592,
+      "learning_rate": 2.8373538643356356e-06,
+      "loss": 0.6691,
+      "step": 40114
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.2401838302612305,
+      "learning_rate": 2.8371161478790077e-06,
+      "loss": 0.6313,
+      "step": 40115
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.8330817222595215,
+      "learning_rate": 2.8368784374365522e-06,
+      "loss": 0.5992,
+      "step": 40116
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.376439094543457,
+      "learning_rate": 2.8366407330089243e-06,
+      "loss": 0.5029,
+      "step": 40117
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.131319284439087,
+      "learning_rate": 2.836403034596791e-06,
+      "loss": 0.5703,
+      "step": 40118
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 3.1994516849517822,
+      "learning_rate": 2.8361653422008085e-06,
+      "loss": 0.9153,
+      "step": 40119
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.162914514541626,
+      "learning_rate": 2.835927655821641e-06,
+      "loss": 0.5177,
+      "step": 40120
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 3.5689542293548584,
+      "learning_rate": 2.835689975459946e-06,
+      "loss": 0.6694,
+      "step": 40121
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 3.2569501399993896,
+      "learning_rate": 2.8354523011163864e-06,
+      "loss": 0.6212,
+      "step": 40122
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 3.259958267211914,
+      "learning_rate": 2.8352146327916252e-06,
+      "loss": 0.6524,
+      "step": 40123
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.938233494758606,
+      "learning_rate": 2.8349769704863194e-06,
+      "loss": 0.6924,
+      "step": 40124
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 5.231548309326172,
+      "learning_rate": 2.8347393142011337e-06,
+      "loss": 0.5741,
+      "step": 40125
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.1886818408966064,
+      "learning_rate": 2.8345016639367243e-06,
+      "loss": 0.6461,
+      "step": 40126
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 7.48549747467041,
+      "learning_rate": 2.834264019693755e-06,
+      "loss": 0.8896,
+      "step": 40127
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.203457832336426,
+      "learning_rate": 2.834026381472888e-06,
+      "loss": 0.7969,
+      "step": 40128
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.8869967460632324,
+      "learning_rate": 2.8337887492747805e-06,
+      "loss": 0.4973,
+      "step": 40129
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 4.070382118225098,
+      "learning_rate": 2.8335511231000965e-06,
+      "loss": 0.7484,
+      "step": 40130
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 5.236495494842529,
+      "learning_rate": 2.8333135029494935e-06,
+      "loss": 0.7413,
+      "step": 40131
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.7874338626861572,
+      "learning_rate": 2.8330758888236364e-06,
+      "loss": 0.7835,
+      "step": 40132
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.3263394832611084,
+      "learning_rate": 2.832838280723181e-06,
+      "loss": 0.5599,
+      "step": 40133
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.981163263320923,
+      "learning_rate": 2.8326006786487924e-06,
+      "loss": 0.6475,
+      "step": 40134
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 3.871894359588623,
+      "learning_rate": 2.8323630826011273e-06,
+      "loss": 0.7568,
+      "step": 40135
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.771184206008911,
+      "learning_rate": 2.8321254925808513e-06,
+      "loss": 0.542,
+      "step": 40136
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 4.313298225402832,
+      "learning_rate": 2.83188790858862e-06,
+      "loss": 0.7724,
+      "step": 40137
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.6327333450317383,
+      "learning_rate": 2.8316503306250976e-06,
+      "loss": 0.5736,
+      "step": 40138
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.796506404876709,
+      "learning_rate": 2.8314127586909413e-06,
+      "loss": 0.6361,
+      "step": 40139
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.764277935028076,
+      "learning_rate": 2.831175192786816e-06,
+      "loss": 0.5304,
+      "step": 40140
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.6440861225128174,
+      "learning_rate": 2.8309376329133785e-06,
+      "loss": 0.6515,
+      "step": 40141
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.247429132461548,
+      "learning_rate": 2.8307000790712903e-06,
+      "loss": 0.6797,
+      "step": 40142
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 3.944157838821411,
+      "learning_rate": 2.8304625312612127e-06,
+      "loss": 1.0107,
+      "step": 40143
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 4.99783992767334,
+      "learning_rate": 2.8302249894838086e-06,
+      "loss": 0.401,
+      "step": 40144
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 10.93120288848877,
+      "learning_rate": 2.829987453739733e-06,
+      "loss": 0.8309,
+      "step": 40145
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.433983325958252,
+      "learning_rate": 2.829749924029652e-06,
+      "loss": 0.6729,
+      "step": 40146
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 3.12941837310791,
+      "learning_rate": 2.829512400354222e-06,
+      "loss": 0.4939,
+      "step": 40147
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.439249038696289,
+      "learning_rate": 2.829274882714106e-06,
+      "loss": 0.6638,
+      "step": 40148
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.413008689880371,
+      "learning_rate": 2.8290373711099615e-06,
+      "loss": 0.5191,
+      "step": 40149
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 3.661975383758545,
+      "learning_rate": 2.828799865542453e-06,
+      "loss": 0.5773,
+      "step": 40150
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.4036343097686768,
+      "learning_rate": 2.828562366012236e-06,
+      "loss": 0.623,
+      "step": 40151
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.5210132598876953,
+      "learning_rate": 2.8283248725199767e-06,
+      "loss": 0.63,
+      "step": 40152
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.256692886352539,
+      "learning_rate": 2.8280873850663303e-06,
+      "loss": 0.5997,
+      "step": 40153
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 3.229494094848633,
+      "learning_rate": 2.8278499036519614e-06,
+      "loss": 0.7604,
+      "step": 40154
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.274367570877075,
+      "learning_rate": 2.8276124282775253e-06,
+      "loss": 0.636,
+      "step": 40155
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 4.808176040649414,
+      "learning_rate": 2.8273749589436877e-06,
+      "loss": 0.3421,
+      "step": 40156
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.89096999168396,
+      "learning_rate": 2.8271374956511042e-06,
+      "loss": 0.5345,
+      "step": 40157
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.5429600477218628,
+      "learning_rate": 2.8269000384004402e-06,
+      "loss": 0.3528,
+      "step": 40158
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.5210397243499756,
+      "learning_rate": 2.8266625871923505e-06,
+      "loss": 0.7047,
+      "step": 40159
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.9763821363449097,
+      "learning_rate": 2.8264251420275003e-06,
+      "loss": 0.311,
+      "step": 40160
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.130187749862671,
+      "learning_rate": 2.8261877029065453e-06,
+      "loss": 0.6551,
+      "step": 40161
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.7503724098205566,
+      "learning_rate": 2.825950269830148e-06,
+      "loss": 0.6531,
+      "step": 40162
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.0949695110321045,
+      "learning_rate": 2.825712842798968e-06,
+      "loss": 0.4616,
+      "step": 40163
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.954315423965454,
+      "learning_rate": 2.825475421813669e-06,
+      "loss": 0.4971,
+      "step": 40164
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.800520420074463,
+      "learning_rate": 2.8252380068749063e-06,
+      "loss": 0.7603,
+      "step": 40165
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.2134456634521484,
+      "learning_rate": 2.8250005979833434e-06,
+      "loss": 0.6226,
+      "step": 40166
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.2950589656829834,
+      "learning_rate": 2.8247631951396372e-06,
+      "loss": 0.5588,
+      "step": 40167
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.8482085466384888,
+      "learning_rate": 2.824525798344452e-06,
+      "loss": 0.3741,
+      "step": 40168
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 3.105167865753174,
+      "learning_rate": 2.8242884075984434e-06,
+      "loss": 0.7292,
+      "step": 40169
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 3.5830259323120117,
+      "learning_rate": 2.8240510229022754e-06,
+      "loss": 0.8101,
+      "step": 40170
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 10.662660598754883,
+      "learning_rate": 2.8238136442566043e-06,
+      "loss": 0.7795,
+      "step": 40171
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 3.6539862155914307,
+      "learning_rate": 2.823576271662095e-06,
+      "loss": 0.5284,
+      "step": 40172
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.7552008628845215,
+      "learning_rate": 2.8233389051194026e-06,
+      "loss": 0.7038,
+      "step": 40173
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.64640474319458,
+      "learning_rate": 2.823101544629191e-06,
+      "loss": 0.4233,
+      "step": 40174
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 3.170572280883789,
+      "learning_rate": 2.822864190192117e-06,
+      "loss": 0.7356,
+      "step": 40175
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 3.652961254119873,
+      "learning_rate": 2.8226268418088433e-06,
+      "loss": 0.4699,
+      "step": 40176
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.337583303451538,
+      "learning_rate": 2.822389499480027e-06,
+      "loss": 0.6524,
+      "step": 40177
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.2789719104766846,
+      "learning_rate": 2.8221521632063297e-06,
+      "loss": 0.5524,
+      "step": 40178
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.2334766387939453,
+      "learning_rate": 2.821914832988414e-06,
+      "loss": 0.3925,
+      "step": 40179
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 4.195079803466797,
+      "learning_rate": 2.8216775088269343e-06,
+      "loss": 0.8378,
+      "step": 40180
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.3746719360351562,
+      "learning_rate": 2.8214401907225565e-06,
+      "loss": 0.7028,
+      "step": 40181
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.804508924484253,
+      "learning_rate": 2.821202878675935e-06,
+      "loss": 0.736,
+      "step": 40182
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.1099588871002197,
+      "learning_rate": 2.8209655726877317e-06,
+      "loss": 0.6516,
+      "step": 40183
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 3.7503035068511963,
+      "learning_rate": 2.8207282727586086e-06,
+      "loss": 0.7582,
+      "step": 40184
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 3.1041531562805176,
+      "learning_rate": 2.820490978889223e-06,
+      "loss": 0.7756,
+      "step": 40185
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.344510555267334,
+      "learning_rate": 2.8202536910802362e-06,
+      "loss": 0.7037,
+      "step": 40186
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 4.113399982452393,
+      "learning_rate": 2.8200164093323057e-06,
+      "loss": 0.7895,
+      "step": 40187
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 3.606125831604004,
+      "learning_rate": 2.8197791336460935e-06,
+      "loss": 0.6084,
+      "step": 40188
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.8151462078094482,
+      "learning_rate": 2.8195418640222605e-06,
+      "loss": 0.3694,
+      "step": 40189
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 3.7286062240600586,
+      "learning_rate": 2.8193046004614627e-06,
+      "loss": 0.5975,
+      "step": 40190
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.9941461086273193,
+      "learning_rate": 2.819067342964364e-06,
+      "loss": 0.5013,
+      "step": 40191
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.6526174545288086,
+      "learning_rate": 2.8188300915316203e-06,
+      "loss": 0.4515,
+      "step": 40192
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.6538467407226562,
+      "learning_rate": 2.8185928461638946e-06,
+      "loss": 0.4408,
+      "step": 40193
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.9871622323989868,
+      "learning_rate": 2.818355606861843e-06,
+      "loss": 0.3629,
+      "step": 40194
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.8427703380584717,
+      "learning_rate": 2.818118373626129e-06,
+      "loss": 0.8204,
+      "step": 40195
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 4.367071151733398,
+      "learning_rate": 2.8178811464574085e-06,
+      "loss": 0.5041,
+      "step": 40196
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 4.108612537384033,
+      "learning_rate": 2.817643925356346e-06,
+      "loss": 0.7751,
+      "step": 40197
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.9206492900848389,
+      "learning_rate": 2.8174067103235953e-06,
+      "loss": 0.557,
+      "step": 40198
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 3.2329769134521484,
+      "learning_rate": 2.817169501359821e-06,
+      "loss": 0.4959,
+      "step": 40199
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.567699909210205,
+      "learning_rate": 2.816932298465679e-06,
+      "loss": 0.397,
+      "step": 40200
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.776050090789795,
+      "learning_rate": 2.816695101641832e-06,
+      "loss": 0.5089,
+      "step": 40201
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.773076057434082,
+      "learning_rate": 2.8164579108889366e-06,
+      "loss": 0.656,
+      "step": 40202
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.7102986574172974,
+      "learning_rate": 2.8162207262076533e-06,
+      "loss": 0.865,
+      "step": 40203
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 3.2367451190948486,
+      "learning_rate": 2.8159835475986426e-06,
+      "loss": 0.4983,
+      "step": 40204
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.783400535583496,
+      "learning_rate": 2.8157463750625646e-06,
+      "loss": 0.5662,
+      "step": 40205
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.8269413709640503,
+      "learning_rate": 2.815509208600077e-06,
+      "loss": 0.4332,
+      "step": 40206
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.1012041568756104,
+      "learning_rate": 2.815272048211841e-06,
+      "loss": 0.4641,
+      "step": 40207
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.310986280441284,
+      "learning_rate": 2.8150348938985126e-06,
+      "loss": 0.6941,
+      "step": 40208
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.9115883111953735,
+      "learning_rate": 2.8147977456607567e-06,
+      "loss": 0.5461,
+      "step": 40209
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 3.017484188079834,
+      "learning_rate": 2.8145606034992273e-06,
+      "loss": 0.586,
+      "step": 40210
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 4.489319801330566,
+      "learning_rate": 2.814323467414588e-06,
+      "loss": 0.7834,
+      "step": 40211
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 3.432910442352295,
+      "learning_rate": 2.8140863374074946e-06,
+      "loss": 0.6456,
+      "step": 40212
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.2603883743286133,
+      "learning_rate": 2.81384921347861e-06,
+      "loss": 0.472,
+      "step": 40213
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 3.0407423973083496,
+      "learning_rate": 2.8136120956285908e-06,
+      "loss": 0.5486,
+      "step": 40214
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.8815627098083496,
+      "learning_rate": 2.813374983858098e-06,
+      "loss": 0.6404,
+      "step": 40215
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.291598081588745,
+      "learning_rate": 2.8131378781677885e-06,
+      "loss": 0.85,
+      "step": 40216
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 3.0574636459350586,
+      "learning_rate": 2.812900778558326e-06,
+      "loss": 0.792,
+      "step": 40217
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.6965863704681396,
+      "learning_rate": 2.8126636850303645e-06,
+      "loss": 0.5973,
+      "step": 40218
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 3.205137014389038,
+      "learning_rate": 2.8124265975845677e-06,
+      "loss": 0.642,
+      "step": 40219
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.7676174640655518,
+      "learning_rate": 2.812189516221592e-06,
+      "loss": 0.7233,
+      "step": 40220
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.419515371322632,
+      "learning_rate": 2.811952440942098e-06,
+      "loss": 0.7231,
+      "step": 40221
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.9031574726104736,
+      "learning_rate": 2.8117153717467443e-06,
+      "loss": 0.8001,
+      "step": 40222
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 4.437264919281006,
+      "learning_rate": 2.8114783086361897e-06,
+      "loss": 0.6706,
+      "step": 40223
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.5840156078338623,
+      "learning_rate": 2.811241251611094e-06,
+      "loss": 0.3891,
+      "step": 40224
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.1953306198120117,
+      "learning_rate": 2.8110042006721196e-06,
+      "loss": 0.5622,
+      "step": 40225
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.7748284339904785,
+      "learning_rate": 2.810767155819919e-06,
+      "loss": 0.9028,
+      "step": 40226
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.1487486362457275,
+      "learning_rate": 2.8105301170551574e-06,
+      "loss": 0.5866,
+      "step": 40227
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.3044679164886475,
+      "learning_rate": 2.8102930843784894e-06,
+      "loss": 0.6776,
+      "step": 40228
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.124288320541382,
+      "learning_rate": 2.810056057790578e-06,
+      "loss": 0.67,
+      "step": 40229
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.7150523662567139,
+      "learning_rate": 2.8098190372920782e-06,
+      "loss": 0.4941,
+      "step": 40230
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.971877098083496,
+      "learning_rate": 2.8095820228836535e-06,
+      "loss": 0.6682,
+      "step": 40231
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.1757590770721436,
+      "learning_rate": 2.8093450145659586e-06,
+      "loss": 0.6263,
+      "step": 40232
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.2261595726013184,
+      "learning_rate": 2.809108012339656e-06,
+      "loss": 0.6791,
+      "step": 40233
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.8287596702575684,
+      "learning_rate": 2.808871016205402e-06,
+      "loss": 0.6659,
+      "step": 40234
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.682154893875122,
+      "learning_rate": 2.8086340261638586e-06,
+      "loss": 0.7197,
+      "step": 40235
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.136472702026367,
+      "learning_rate": 2.808397042215681e-06,
+      "loss": 0.4161,
+      "step": 40236
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.7845327854156494,
+      "learning_rate": 2.8081600643615323e-06,
+      "loss": 0.4203,
+      "step": 40237
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.337301015853882,
+      "learning_rate": 2.8079230926020674e-06,
+      "loss": 0.5075,
+      "step": 40238
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.7483553886413574,
+      "learning_rate": 2.807686126937947e-06,
+      "loss": 0.5983,
+      "step": 40239
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 4.5637311935424805,
+      "learning_rate": 2.807449167369832e-06,
+      "loss": 0.7146,
+      "step": 40240
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 4.054404258728027,
+      "learning_rate": 2.807212213898378e-06,
+      "loss": 0.5155,
+      "step": 40241
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.300405979156494,
+      "learning_rate": 2.806975266524248e-06,
+      "loss": 0.7226,
+      "step": 40242
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 4.132251262664795,
+      "learning_rate": 2.8067383252480953e-06,
+      "loss": 0.6454,
+      "step": 40243
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.156705856323242,
+      "learning_rate": 2.806501390070582e-06,
+      "loss": 0.7152,
+      "step": 40244
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.544529914855957,
+      "learning_rate": 2.8062644609923685e-06,
+      "loss": 0.6305,
+      "step": 40245
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.918315887451172,
+      "learning_rate": 2.80602753801411e-06,
+      "loss": 0.5476,
+      "step": 40246
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 4.796384811401367,
+      "learning_rate": 2.805790621136469e-06,
+      "loss": 0.631,
+      "step": 40247
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 4.314743995666504,
+      "learning_rate": 2.8055537103601e-06,
+      "loss": 0.7933,
+      "step": 40248
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.46331787109375,
+      "learning_rate": 2.8053168056856663e-06,
+      "loss": 0.7413,
+      "step": 40249
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.892155647277832,
+      "learning_rate": 2.8050799071138223e-06,
+      "loss": 0.5423,
+      "step": 40250
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.691023349761963,
+      "learning_rate": 2.804843014645229e-06,
+      "loss": 0.459,
+      "step": 40251
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.972259998321533,
+      "learning_rate": 2.804606128280547e-06,
+      "loss": 0.8105,
+      "step": 40252
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.194521427154541,
+      "learning_rate": 2.804369248020431e-06,
+      "loss": 0.5773,
+      "step": 40253
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.630314588546753,
+      "learning_rate": 2.8041323738655436e-06,
+      "loss": 0.5134,
+      "step": 40254
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.0259666442871094,
+      "learning_rate": 2.803895505816539e-06,
+      "loss": 0.601,
+      "step": 40255
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.430070400238037,
+      "learning_rate": 2.8036586438740808e-06,
+      "loss": 0.4386,
+      "step": 40256
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.808222770690918,
+      "learning_rate": 2.8034217880388225e-06,
+      "loss": 0.7588,
+      "step": 40257
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.4785077571868896,
+      "learning_rate": 2.8031849383114285e-06,
+      "loss": 0.5194,
+      "step": 40258
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.299394130706787,
+      "learning_rate": 2.8029480946925516e-06,
+      "loss": 0.5192,
+      "step": 40259
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.095203399658203,
+      "learning_rate": 2.8027112571828544e-06,
+      "loss": 0.3007,
+      "step": 40260
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.9851094484329224,
+      "learning_rate": 2.8024744257829927e-06,
+      "loss": 0.2792,
+      "step": 40261
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.113471746444702,
+      "learning_rate": 2.8022376004936284e-06,
+      "loss": 0.4704,
+      "step": 40262
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.7952535152435303,
+      "learning_rate": 2.802000781315416e-06,
+      "loss": 0.6044,
+      "step": 40263
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.877233624458313,
+      "learning_rate": 2.801763968249016e-06,
+      "loss": 0.3505,
+      "step": 40264
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.0392796993255615,
+      "learning_rate": 2.801527161295087e-06,
+      "loss": 0.6234,
+      "step": 40265
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.620450019836426,
+      "learning_rate": 2.8012903604542895e-06,
+      "loss": 0.7174,
+      "step": 40266
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.7077999114990234,
+      "learning_rate": 2.8010535657272775e-06,
+      "loss": 0.6805,
+      "step": 40267
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.7864532470703125,
+      "learning_rate": 2.8008167771147142e-06,
+      "loss": 0.647,
+      "step": 40268
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.2520480155944824,
+      "learning_rate": 2.800579994617254e-06,
+      "loss": 0.7581,
+      "step": 40269
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 4.089491367340088,
+      "learning_rate": 2.800343218235558e-06,
+      "loss": 0.4771,
+      "step": 40270
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.2440919876098633,
+      "learning_rate": 2.800106447970282e-06,
+      "loss": 0.4389,
+      "step": 40271
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.6459505558013916,
+      "learning_rate": 2.7998696838220873e-06,
+      "loss": 0.6254,
+      "step": 40272
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.15740704536438,
+      "learning_rate": 2.7996329257916295e-06,
+      "loss": 0.5557,
+      "step": 40273
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.1764767169952393,
+      "learning_rate": 2.79939617387957e-06,
+      "loss": 0.3887,
+      "step": 40274
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.3235321044921875,
+      "learning_rate": 2.799159428086563e-06,
+      "loss": 0.5687,
+      "step": 40275
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.9382431507110596,
+      "learning_rate": 2.798922688413272e-06,
+      "loss": 0.8235,
+      "step": 40276
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.810807704925537,
+      "learning_rate": 2.7986859548603496e-06,
+      "loss": 0.6851,
+      "step": 40277
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.090597152709961,
+      "learning_rate": 2.79844922742846e-06,
+      "loss": 0.3062,
+      "step": 40278
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.3359053134918213,
+      "learning_rate": 2.798212506118255e-06,
+      "loss": 0.6049,
+      "step": 40279
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.080223321914673,
+      "learning_rate": 2.7979757909303994e-06,
+      "loss": 0.5551,
+      "step": 40280
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.1087467670440674,
+      "learning_rate": 2.797739081865546e-06,
+      "loss": 0.5569,
+      "step": 40281
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.566599130630493,
+      "learning_rate": 2.7975023789243567e-06,
+      "loss": 0.7575,
+      "step": 40282
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.6949033737182617,
+      "learning_rate": 2.797265682107486e-06,
+      "loss": 0.4296,
+      "step": 40283
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.158137321472168,
+      "learning_rate": 2.7970289914155953e-06,
+      "loss": 0.4418,
+      "step": 40284
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.0410447120666504,
+      "learning_rate": 2.7967923068493406e-06,
+      "loss": 0.474,
+      "step": 40285
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.890509605407715,
+      "learning_rate": 2.796555628409384e-06,
+      "loss": 0.765,
+      "step": 40286
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.0141870975494385,
+      "learning_rate": 2.7963189560963788e-06,
+      "loss": 0.5713,
+      "step": 40287
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.036396026611328,
+      "learning_rate": 2.7960822899109863e-06,
+      "loss": 0.5161,
+      "step": 40288
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.4786746501922607,
+      "learning_rate": 2.7958456298538624e-06,
+      "loss": 0.5694,
+      "step": 40289
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.8707029819488525,
+      "learning_rate": 2.7956089759256673e-06,
+      "loss": 0.5971,
+      "step": 40290
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.966582775115967,
+      "learning_rate": 2.795372328127056e-06,
+      "loss": 0.7519,
+      "step": 40291
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.0318570137023926,
+      "learning_rate": 2.7951356864586906e-06,
+      "loss": 0.7357,
+      "step": 40292
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.0194661617279053,
+      "learning_rate": 2.794899050921225e-06,
+      "loss": 0.6575,
+      "step": 40293
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.7754604816436768,
+      "learning_rate": 2.7946624215153205e-06,
+      "loss": 0.5066,
+      "step": 40294
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.449767827987671,
+      "learning_rate": 2.7944257982416323e-06,
+      "loss": 0.4021,
+      "step": 40295
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.7057788372039795,
+      "learning_rate": 2.7941891811008214e-06,
+      "loss": 0.3938,
+      "step": 40296
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.9107576608657837,
+      "learning_rate": 2.793952570093542e-06,
+      "loss": 0.5205,
+      "step": 40297
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.21440052986145,
+      "learning_rate": 2.793715965220456e-06,
+      "loss": 0.6904,
+      "step": 40298
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.4763758182525635,
+      "learning_rate": 2.793479366482218e-06,
+      "loss": 0.5238,
+      "step": 40299
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.0368940830230713,
+      "learning_rate": 2.7932427738794864e-06,
+      "loss": 0.6664,
+      "step": 40300
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.7103044986724854,
+      "learning_rate": 2.793006187412923e-06,
+      "loss": 0.5572,
+      "step": 40301
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.9296456575393677,
+      "learning_rate": 2.79276960708318e-06,
+      "loss": 0.7251,
+      "step": 40302
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.813577890396118,
+      "learning_rate": 2.7925330328909193e-06,
+      "loss": 0.8424,
+      "step": 40303
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.5031750202178955,
+      "learning_rate": 2.7922964648367968e-06,
+      "loss": 0.775,
+      "step": 40304
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.9506475925445557,
+      "learning_rate": 2.79205990292147e-06,
+      "loss": 0.8118,
+      "step": 40305
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.9776451587677,
+      "learning_rate": 2.7918233471455992e-06,
+      "loss": 0.5741,
+      "step": 40306
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.8604191541671753,
+      "learning_rate": 2.791586797509839e-06,
+      "loss": 0.5211,
+      "step": 40307
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.8914591073989868,
+      "learning_rate": 2.79135025401485e-06,
+      "loss": 0.515,
+      "step": 40308
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.7437705993652344,
+      "learning_rate": 2.791113716661287e-06,
+      "loss": 0.604,
+      "step": 40309
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.0336740016937256,
+      "learning_rate": 2.7908771854498117e-06,
+      "loss": 0.5698,
+      "step": 40310
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.112079381942749,
+      "learning_rate": 2.7906406603810764e-06,
+      "loss": 0.5417,
+      "step": 40311
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.3285763263702393,
+      "learning_rate": 2.7904041414557424e-06,
+      "loss": 0.5447,
+      "step": 40312
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 4.909947872161865,
+      "learning_rate": 2.790167628674469e-06,
+      "loss": 0.8236,
+      "step": 40313
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.745042324066162,
+      "learning_rate": 2.7899311220379087e-06,
+      "loss": 0.5297,
+      "step": 40314
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.4298222064971924,
+      "learning_rate": 2.789694621546725e-06,
+      "loss": 0.5565,
+      "step": 40315
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.1000146865844727,
+      "learning_rate": 2.78945812720157e-06,
+      "loss": 0.5873,
+      "step": 40316
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.055905818939209,
+      "learning_rate": 2.7892216390031062e-06,
+      "loss": 0.472,
+      "step": 40317
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.935311198234558,
+      "learning_rate": 2.7889851569519865e-06,
+      "loss": 0.6543,
+      "step": 40318
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.3347573280334473,
+      "learning_rate": 2.7887486810488733e-06,
+      "loss": 0.4649,
+      "step": 40319
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.4843406677246094,
+      "learning_rate": 2.7885122112944198e-06,
+      "loss": 0.6892,
+      "step": 40320
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.269221782684326,
+      "learning_rate": 2.7882757476892865e-06,
+      "loss": 0.4968,
+      "step": 40321
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.267490863800049,
+      "learning_rate": 2.788039290234128e-06,
+      "loss": 0.548,
+      "step": 40322
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.1426219940185547,
+      "learning_rate": 2.7878028389296065e-06,
+      "loss": 0.32,
+      "step": 40323
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.5452466011047363,
+      "learning_rate": 2.7875663937763736e-06,
+      "loss": 0.6148,
+      "step": 40324
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 4.472209930419922,
+      "learning_rate": 2.7873299547750898e-06,
+      "loss": 0.6871,
+      "step": 40325
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 4.024651050567627,
+      "learning_rate": 2.7870935219264136e-06,
+      "loss": 0.5976,
+      "step": 40326
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.2874624729156494,
+      "learning_rate": 2.7868570952310027e-06,
+      "loss": 0.6204,
+      "step": 40327
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.5711028575897217,
+      "learning_rate": 2.7866206746895106e-06,
+      "loss": 0.586,
+      "step": 40328
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.8727481365203857,
+      "learning_rate": 2.7863842603025998e-06,
+      "loss": 0.5986,
+      "step": 40329
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.8341153860092163,
+      "learning_rate": 2.7861478520709227e-06,
+      "loss": 0.7114,
+      "step": 40330
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 4.9072465896606445,
+      "learning_rate": 2.7859114499951415e-06,
+      "loss": 0.8102,
+      "step": 40331
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.11651349067688,
+      "learning_rate": 2.785675054075909e-06,
+      "loss": 0.6259,
+      "step": 40332
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.8624651432037354,
+      "learning_rate": 2.785438664313887e-06,
+      "loss": 0.7066,
+      "step": 40333
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.1061267852783203,
+      "learning_rate": 2.785202280709728e-06,
+      "loss": 0.6172,
+      "step": 40334
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.626701831817627,
+      "learning_rate": 2.784965903264094e-06,
+      "loss": 0.432,
+      "step": 40335
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 4.123682498931885,
+      "learning_rate": 2.784729531977638e-06,
+      "loss": 0.4137,
+      "step": 40336
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.324495553970337,
+      "learning_rate": 2.784493166851022e-06,
+      "loss": 0.2896,
+      "step": 40337
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.437757730484009,
+      "learning_rate": 2.7842568078848973e-06,
+      "loss": 0.5011,
+      "step": 40338
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.2354066371917725,
+      "learning_rate": 2.784020455079927e-06,
+      "loss": 0.6874,
+      "step": 40339
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 4.079495906829834,
+      "learning_rate": 2.783784108436764e-06,
+      "loss": 0.8642,
+      "step": 40340
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 4.298977851867676,
+      "learning_rate": 2.7835477679560684e-06,
+      "loss": 0.7055,
+      "step": 40341
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.2947542667388916,
+      "learning_rate": 2.7833114336384947e-06,
+      "loss": 0.6244,
+      "step": 40342
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 5.0466179847717285,
+      "learning_rate": 2.783075105484704e-06,
+      "loss": 0.6352,
+      "step": 40343
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.9423881769180298,
+      "learning_rate": 2.7828387834953487e-06,
+      "loss": 0.5076,
+      "step": 40344
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 4.990593433380127,
+      "learning_rate": 2.7826024676710873e-06,
+      "loss": 0.6036,
+      "step": 40345
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.9274911880493164,
+      "learning_rate": 2.782366158012578e-06,
+      "loss": 0.434,
+      "step": 40346
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.233151435852051,
+      "learning_rate": 2.78212985452048e-06,
+      "loss": 0.559,
+      "step": 40347
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.7003250122070312,
+      "learning_rate": 2.7818935571954463e-06,
+      "loss": 0.4283,
+      "step": 40348
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.4160492420196533,
+      "learning_rate": 2.7816572660381374e-06,
+      "loss": 0.7572,
+      "step": 40349
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.3379623889923096,
+      "learning_rate": 2.7814209810492066e-06,
+      "loss": 0.801,
+      "step": 40350
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.632678508758545,
+      "learning_rate": 2.7811847022293155e-06,
+      "loss": 0.5955,
+      "step": 40351
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.8784477710723877,
+      "learning_rate": 2.7809484295791157e-06,
+      "loss": 0.5229,
+      "step": 40352
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.126101493835449,
+      "learning_rate": 2.7807121630992694e-06,
+      "loss": 0.6221,
+      "step": 40353
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.5580079555511475,
+      "learning_rate": 2.7804759027904295e-06,
+      "loss": 0.6811,
+      "step": 40354
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 4.54163932800293,
+      "learning_rate": 2.7802396486532567e-06,
+      "loss": 0.5634,
+      "step": 40355
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.83898663520813,
+      "learning_rate": 2.7800034006884037e-06,
+      "loss": 0.8064,
+      "step": 40356
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.6893218755722046,
+      "learning_rate": 2.7797671588965316e-06,
+      "loss": 0.5993,
+      "step": 40357
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.84248948097229,
+      "learning_rate": 2.779530923278293e-06,
+      "loss": 0.777,
+      "step": 40358
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.4712421894073486,
+      "learning_rate": 2.779294693834349e-06,
+      "loss": 0.8261,
+      "step": 40359
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.0395777225494385,
+      "learning_rate": 2.7790584705653535e-06,
+      "loss": 0.6203,
+      "step": 40360
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.457937479019165,
+      "learning_rate": 2.7788222534719656e-06,
+      "loss": 0.3714,
+      "step": 40361
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.9287488460540771,
+      "learning_rate": 2.7785860425548387e-06,
+      "loss": 0.7613,
+      "step": 40362
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 5.736876964569092,
+      "learning_rate": 2.7783498378146322e-06,
+      "loss": 0.424,
+      "step": 40363
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.450305700302124,
+      "learning_rate": 2.7781136392520047e-06,
+      "loss": 0.5852,
+      "step": 40364
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.580019474029541,
+      "learning_rate": 2.777877446867609e-06,
+      "loss": 0.8349,
+      "step": 40365
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.5191304683685303,
+      "learning_rate": 2.7776412606621033e-06,
+      "loss": 0.7584,
+      "step": 40366
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 4.629916667938232,
+      "learning_rate": 2.777405080636147e-06,
+      "loss": 0.68,
+      "step": 40367
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.624263286590576,
+      "learning_rate": 2.777168906790393e-06,
+      "loss": 0.903,
+      "step": 40368
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.1274478435516357,
+      "learning_rate": 2.776932739125501e-06,
+      "loss": 0.5735,
+      "step": 40369
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.641458034515381,
+      "learning_rate": 2.776696577642124e-06,
+      "loss": 0.6621,
+      "step": 40370
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.2939035892486572,
+      "learning_rate": 2.776460422340923e-06,
+      "loss": 0.7173,
+      "step": 40371
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.501418352127075,
+      "learning_rate": 2.7762242732225508e-06,
+      "loss": 0.699,
+      "step": 40372
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.601912260055542,
+      "learning_rate": 2.7759881302876678e-06,
+      "loss": 0.4864,
+      "step": 40373
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.2003087997436523,
+      "learning_rate": 2.775751993536927e-06,
+      "loss": 0.2819,
+      "step": 40374
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.7622032165527344,
+      "learning_rate": 2.7755158629709866e-06,
+      "loss": 0.7055,
+      "step": 40375
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.643674850463867,
+      "learning_rate": 2.775279738590505e-06,
+      "loss": 0.5298,
+      "step": 40376
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 6.348626613616943,
+      "learning_rate": 2.7750436203961357e-06,
+      "loss": 0.4887,
+      "step": 40377
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.8014729022979736,
+      "learning_rate": 2.774807508388538e-06,
+      "loss": 0.3835,
+      "step": 40378
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.3149731159210205,
+      "learning_rate": 2.774571402568365e-06,
+      "loss": 0.3617,
+      "step": 40379
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.0443649291992188,
+      "learning_rate": 2.7743353029362774e-06,
+      "loss": 0.386,
+      "step": 40380
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.4432971477508545,
+      "learning_rate": 2.7740992094929277e-06,
+      "loss": 0.3486,
+      "step": 40381
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.5561487674713135,
+      "learning_rate": 2.773863122238976e-06,
+      "loss": 0.744,
+      "step": 40382
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 4.839898109436035,
+      "learning_rate": 2.7736270411750755e-06,
+      "loss": 0.7337,
+      "step": 40383
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.3998911380767822,
+      "learning_rate": 2.7733909663018856e-06,
+      "loss": 0.6737,
+      "step": 40384
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.632437229156494,
+      "learning_rate": 2.7731548976200596e-06,
+      "loss": 0.6427,
+      "step": 40385
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.250420570373535,
+      "learning_rate": 2.772918835130256e-06,
+      "loss": 0.6968,
+      "step": 40386
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 4.968786716461182,
+      "learning_rate": 2.77268277883313e-06,
+      "loss": 0.6361,
+      "step": 40387
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.667039632797241,
+      "learning_rate": 2.772446728729341e-06,
+      "loss": 0.4532,
+      "step": 40388
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.5325355529785156,
+      "learning_rate": 2.7722106848195418e-06,
+      "loss": 0.551,
+      "step": 40389
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.4954590797424316,
+      "learning_rate": 2.7719746471043917e-06,
+      "loss": 0.654,
+      "step": 40390
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.659133195877075,
+      "learning_rate": 2.771738615584544e-06,
+      "loss": 0.9051,
+      "step": 40391
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.3210811614990234,
+      "learning_rate": 2.7715025902606574e-06,
+      "loss": 0.8051,
+      "step": 40392
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.1136558055877686,
+      "learning_rate": 2.7712665711333865e-06,
+      "loss": 0.3516,
+      "step": 40393
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.2852277755737305,
+      "learning_rate": 2.77103055820339e-06,
+      "loss": 0.5664,
+      "step": 40394
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.717855215072632,
+      "learning_rate": 2.7707945514713197e-06,
+      "loss": 0.5585,
+      "step": 40395
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.581212282180786,
+      "learning_rate": 2.7705585509378374e-06,
+      "loss": 0.3928,
+      "step": 40396
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.0615994930267334,
+      "learning_rate": 2.7703225566035945e-06,
+      "loss": 0.4967,
+      "step": 40397
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.0839779376983643,
+      "learning_rate": 2.770086568469251e-06,
+      "loss": 0.7343,
+      "step": 40398
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.3096299171447754,
+      "learning_rate": 2.7698505865354604e-06,
+      "loss": 0.8689,
+      "step": 40399
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.61125111579895,
+      "learning_rate": 2.769614610802881e-06,
+      "loss": 0.3945,
+      "step": 40400
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.4564709663391113,
+      "learning_rate": 2.7693786412721657e-06,
+      "loss": 0.56,
+      "step": 40401
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.538361072540283,
+      "learning_rate": 2.7691426779439754e-06,
+      "loss": 0.6202,
+      "step": 40402
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 6.471875190734863,
+      "learning_rate": 2.768906720818961e-06,
+      "loss": 0.7846,
+      "step": 40403
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.00942063331604,
+      "learning_rate": 2.7686707698977838e-06,
+      "loss": 0.7397,
+      "step": 40404
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.1466665267944336,
+      "learning_rate": 2.7684348251810957e-06,
+      "loss": 0.7613,
+      "step": 40405
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.6426515579223633,
+      "learning_rate": 2.768198886669553e-06,
+      "loss": 0.7819,
+      "step": 40406
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.9246290922164917,
+      "learning_rate": 2.7679629543638143e-06,
+      "loss": 0.4751,
+      "step": 40407
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 4.3136305809021,
+      "learning_rate": 2.7677270282645364e-06,
+      "loss": 0.7042,
+      "step": 40408
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.9973785877227783,
+      "learning_rate": 2.767491108372371e-06,
+      "loss": 0.5732,
+      "step": 40409
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 4.712673187255859,
+      "learning_rate": 2.7672551946879784e-06,
+      "loss": 0.6787,
+      "step": 40410
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.4549686908721924,
+      "learning_rate": 2.7670192872120104e-06,
+      "loss": 0.5292,
+      "step": 40411
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.18815541267395,
+      "learning_rate": 2.7667833859451284e-06,
+      "loss": 0.5545,
+      "step": 40412
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.8156113624572754,
+      "learning_rate": 2.7665474908879826e-06,
+      "loss": 0.3886,
+      "step": 40413
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.4417009353637695,
+      "learning_rate": 2.766311602041233e-06,
+      "loss": 0.5467,
+      "step": 40414
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.8410847187042236,
+      "learning_rate": 2.7660757194055333e-06,
+      "loss": 0.9405,
+      "step": 40415
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.2362959384918213,
+      "learning_rate": 2.7658398429815415e-06,
+      "loss": 0.6101,
+      "step": 40416
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.914377212524414,
+      "learning_rate": 2.76560397276991e-06,
+      "loss": 0.6523,
+      "step": 40417
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.2390594482421875,
+      "learning_rate": 2.765368108771299e-06,
+      "loss": 0.7564,
+      "step": 40418
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.69856858253479,
+      "learning_rate": 2.76513225098636e-06,
+      "loss": 0.5056,
+      "step": 40419
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.858345866203308,
+      "learning_rate": 2.7648963994157536e-06,
+      "loss": 0.5613,
+      "step": 40420
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 4.232519626617432,
+      "learning_rate": 2.7646605540601303e-06,
+      "loss": 0.6636,
+      "step": 40421
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.535752296447754,
+      "learning_rate": 2.7644247149201508e-06,
+      "loss": 0.6287,
+      "step": 40422
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.7992796897888184,
+      "learning_rate": 2.764188881996467e-06,
+      "loss": 0.6107,
+      "step": 40423
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.213146924972534,
+      "learning_rate": 2.7639530552897363e-06,
+      "loss": 0.3983,
+      "step": 40424
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.941915273666382,
+      "learning_rate": 2.7637172348006163e-06,
+      "loss": 0.5246,
+      "step": 40425
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.6761938333511353,
+      "learning_rate": 2.763481420529759e-06,
+      "loss": 0.5661,
+      "step": 40426
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.5517147779464722,
+      "learning_rate": 2.7632456124778222e-06,
+      "loss": 0.4382,
+      "step": 40427
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.5395078659057617,
+      "learning_rate": 2.7630098106454635e-06,
+      "loss": 0.5541,
+      "step": 40428
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.0732803344726562,
+      "learning_rate": 2.7627740150333345e-06,
+      "loss": 0.2386,
+      "step": 40429
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 4.3176493644714355,
+      "learning_rate": 2.7625382256420945e-06,
+      "loss": 0.6282,
+      "step": 40430
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.418741226196289,
+      "learning_rate": 2.7623024424723953e-06,
+      "loss": 0.7092,
+      "step": 40431
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.4820871353149414,
+      "learning_rate": 2.7620666655248973e-06,
+      "loss": 0.7111,
+      "step": 40432
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.661206007003784,
+      "learning_rate": 2.761830894800252e-06,
+      "loss": 0.6342,
+      "step": 40433
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.396831512451172,
+      "learning_rate": 2.7615951302991177e-06,
+      "loss": 0.6565,
+      "step": 40434
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.098764419555664,
+      "learning_rate": 2.7613593720221477e-06,
+      "loss": 0.8278,
+      "step": 40435
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.210815668106079,
+      "learning_rate": 2.7611236199699975e-06,
+      "loss": 0.6062,
+      "step": 40436
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.0986504554748535,
+      "learning_rate": 2.7608878741433264e-06,
+      "loss": 0.7278,
+      "step": 40437
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 4.111158847808838,
+      "learning_rate": 2.7606521345427854e-06,
+      "loss": 0.7198,
+      "step": 40438
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.461916208267212,
+      "learning_rate": 2.760416401169034e-06,
+      "loss": 0.7171,
+      "step": 40439
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.7845149040222168,
+      "learning_rate": 2.760180674022723e-06,
+      "loss": 0.7063,
+      "step": 40440
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.471848249435425,
+      "learning_rate": 2.7599449531045136e-06,
+      "loss": 0.6326,
+      "step": 40441
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.038414478302002,
+      "learning_rate": 2.7597092384150557e-06,
+      "loss": 0.5829,
+      "step": 40442
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.80027437210083,
+      "learning_rate": 2.759473529955009e-06,
+      "loss": 0.5341,
+      "step": 40443
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.2568578720092773,
+      "learning_rate": 2.7592378277250254e-06,
+      "loss": 0.6263,
+      "step": 40444
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.7386099100112915,
+      "learning_rate": 2.759002131725763e-06,
+      "loss": 0.4648,
+      "step": 40445
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 4.319820880889893,
+      "learning_rate": 2.758766441957875e-06,
+      "loss": 0.554,
+      "step": 40446
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.387965202331543,
+      "learning_rate": 2.758530758422018e-06,
+      "loss": 0.7322,
+      "step": 40447
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.5956032276153564,
+      "learning_rate": 2.758295081118847e-06,
+      "loss": 0.5303,
+      "step": 40448
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.322486162185669,
+      "learning_rate": 2.7580594100490197e-06,
+      "loss": 0.7297,
+      "step": 40449
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.2398133277893066,
+      "learning_rate": 2.7578237452131873e-06,
+      "loss": 0.6419,
+      "step": 40450
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.6901755332946777,
+      "learning_rate": 2.7575880866120095e-06,
+      "loss": 0.5972,
+      "step": 40451
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 4.098584175109863,
+      "learning_rate": 2.7573524342461364e-06,
+      "loss": 0.5488,
+      "step": 40452
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.2513208389282227,
+      "learning_rate": 2.7571167881162288e-06,
+      "loss": 0.6423,
+      "step": 40453
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.9580092430114746,
+      "learning_rate": 2.7568811482229374e-06,
+      "loss": 0.5373,
+      "step": 40454
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 4.145105361938477,
+      "learning_rate": 2.75664551456692e-06,
+      "loss": 0.5575,
+      "step": 40455
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.45111083984375,
+      "learning_rate": 2.7564098871488304e-06,
+      "loss": 0.6482,
+      "step": 40456
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.4093000888824463,
+      "learning_rate": 2.756174265969326e-06,
+      "loss": 0.7231,
+      "step": 40457
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.8259246349334717,
+      "learning_rate": 2.755938651029058e-06,
+      "loss": 0.6386,
+      "step": 40458
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.6042932271957397,
+      "learning_rate": 2.7557030423286864e-06,
+      "loss": 0.343,
+      "step": 40459
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.760324478149414,
+      "learning_rate": 2.7554674398688617e-06,
+      "loss": 0.5419,
+      "step": 40460
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.080591917037964,
+      "learning_rate": 2.7552318436502434e-06,
+      "loss": 0.5671,
+      "step": 40461
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.5723176002502441,
+      "learning_rate": 2.754996253673482e-06,
+      "loss": 0.4836,
+      "step": 40462
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.2575502395629883,
+      "learning_rate": 2.754760669939237e-06,
+      "loss": 0.7834,
+      "step": 40463
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.4679532051086426,
+      "learning_rate": 2.7545250924481605e-06,
+      "loss": 0.594,
+      "step": 40464
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.8161122798919678,
+      "learning_rate": 2.7542895212009095e-06,
+      "loss": 0.3451,
+      "step": 40465
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.8330872058868408,
+      "learning_rate": 2.7540539561981365e-06,
+      "loss": 0.5391,
+      "step": 40466
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.7646007537841797,
+      "learning_rate": 2.753818397440498e-06,
+      "loss": 0.4476,
+      "step": 40467
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.464750289916992,
+      "learning_rate": 2.7535828449286495e-06,
+      "loss": 0.6966,
+      "step": 40468
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.2204949855804443,
+      "learning_rate": 2.753347298663247e-06,
+      "loss": 0.6034,
+      "step": 40469
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.9949673414230347,
+      "learning_rate": 2.7531117586449424e-06,
+      "loss": 0.5065,
+      "step": 40470
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.4151527881622314,
+      "learning_rate": 2.752876224874394e-06,
+      "loss": 0.4862,
+      "step": 40471
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.112344980239868,
+      "learning_rate": 2.7526406973522533e-06,
+      "loss": 0.6995,
+      "step": 40472
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.3853538036346436,
+      "learning_rate": 2.7524051760791788e-06,
+      "loss": 0.7322,
+      "step": 40473
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.8611364364624023,
+      "learning_rate": 2.7521696610558212e-06,
+      "loss": 0.4736,
+      "step": 40474
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 4.620783805847168,
+      "learning_rate": 2.75193415228284e-06,
+      "loss": 0.711,
+      "step": 40475
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.903404474258423,
+      "learning_rate": 2.7516986497608867e-06,
+      "loss": 0.4131,
+      "step": 40476
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.5943652391433716,
+      "learning_rate": 2.751463153490618e-06,
+      "loss": 0.3618,
+      "step": 40477
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.3027400970458984,
+      "learning_rate": 2.7512276634726864e-06,
+      "loss": 0.5929,
+      "step": 40478
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.472304582595825,
+      "learning_rate": 2.75099217970775e-06,
+      "loss": 0.7477,
+      "step": 40479
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.7400166988372803,
+      "learning_rate": 2.75075670219646e-06,
+      "loss": 0.7413,
+      "step": 40480
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.4844110012054443,
+      "learning_rate": 2.750521230939475e-06,
+      "loss": 0.5149,
+      "step": 40481
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.1210618019104004,
+      "learning_rate": 2.750285765937446e-06,
+      "loss": 0.5303,
+      "step": 40482
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.942139744758606,
+      "learning_rate": 2.7500503071910313e-06,
+      "loss": 0.6176,
+      "step": 40483
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.737576723098755,
+      "learning_rate": 2.749814854700882e-06,
+      "loss": 0.8182,
+      "step": 40484
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.098024845123291,
+      "learning_rate": 2.7495794084676547e-06,
+      "loss": 0.4683,
+      "step": 40485
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 4.815184116363525,
+      "learning_rate": 2.7493439684920058e-06,
+      "loss": 0.4899,
+      "step": 40486
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.6429638862609863,
+      "learning_rate": 2.749108534774586e-06,
+      "loss": 0.5525,
+      "step": 40487
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.585188627243042,
+      "learning_rate": 2.748873107316053e-06,
+      "loss": 0.8156,
+      "step": 40488
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.07255220413208,
+      "learning_rate": 2.7486376861170616e-06,
+      "loss": 0.5567,
+      "step": 40489
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.6889753341674805,
+      "learning_rate": 2.748402271178264e-06,
+      "loss": 0.7508,
+      "step": 40490
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.740968942642212,
+      "learning_rate": 2.7481668625003187e-06,
+      "loss": 0.6097,
+      "step": 40491
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.246019124984741,
+      "learning_rate": 2.7479314600838747e-06,
+      "loss": 0.681,
+      "step": 40492
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.2835936546325684,
+      "learning_rate": 2.7476960639295926e-06,
+      "loss": 0.5416,
+      "step": 40493
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.631291627883911,
+      "learning_rate": 2.7474606740381214e-06,
+      "loss": 0.6969,
+      "step": 40494
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.5022757053375244,
+      "learning_rate": 2.7472252904101204e-06,
+      "loss": 0.6149,
+      "step": 40495
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.888326644897461,
+      "learning_rate": 2.74698991304624e-06,
+      "loss": 0.9072,
+      "step": 40496
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.163707733154297,
+      "learning_rate": 2.7467545419471364e-06,
+      "loss": 0.4725,
+      "step": 40497
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.2795255184173584,
+      "learning_rate": 2.746519177113467e-06,
+      "loss": 0.6711,
+      "step": 40498
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.468430519104004,
+      "learning_rate": 2.746283818545881e-06,
+      "loss": 0.565,
+      "step": 40499
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.5887277126312256,
+      "learning_rate": 2.7460484662450376e-06,
+      "loss": 0.7363,
+      "step": 40500
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 4.200196266174316,
+      "learning_rate": 2.745813120211587e-06,
+      "loss": 0.6764,
+      "step": 40501
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.2187869548797607,
+      "learning_rate": 2.745577780446188e-06,
+      "loss": 0.5767,
+      "step": 40502
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.375659704208374,
+      "learning_rate": 2.7453424469494903e-06,
+      "loss": 0.3986,
+      "step": 40503
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.6859285831451416,
+      "learning_rate": 2.745107119722152e-06,
+      "loss": 0.3908,
+      "step": 40504
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.117342710494995,
+      "learning_rate": 2.7448717987648256e-06,
+      "loss": 0.5318,
+      "step": 40505
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.7016804218292236,
+      "learning_rate": 2.744636484078167e-06,
+      "loss": 0.553,
+      "step": 40506
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.791813611984253,
+      "learning_rate": 2.744401175662828e-06,
+      "loss": 0.8685,
+      "step": 40507
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.933026909828186,
+      "learning_rate": 2.7441658735194642e-06,
+      "loss": 0.6393,
+      "step": 40508
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.282266616821289,
+      "learning_rate": 2.7439305776487298e-06,
+      "loss": 0.6768,
+      "step": 40509
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.261960983276367,
+      "learning_rate": 2.7436952880512814e-06,
+      "loss": 0.6573,
+      "step": 40510
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.2827460765838623,
+      "learning_rate": 2.7434600047277693e-06,
+      "loss": 0.8211,
+      "step": 40511
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.738720417022705,
+      "learning_rate": 2.7432247276788514e-06,
+      "loss": 0.4706,
+      "step": 40512
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.126640558242798,
+      "learning_rate": 2.742989456905178e-06,
+      "loss": 0.5139,
+      "step": 40513
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.191145181655884,
+      "learning_rate": 2.7427541924074077e-06,
+      "loss": 0.6339,
+      "step": 40514
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.4880874156951904,
+      "learning_rate": 2.74251893418619e-06,
+      "loss": 0.758,
+      "step": 40515
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.2915232181549072,
+      "learning_rate": 2.742283682242184e-06,
+      "loss": 0.5445,
+      "step": 40516
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.0360686779022217,
+      "learning_rate": 2.7420484365760393e-06,
+      "loss": 0.6963,
+      "step": 40517
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.7652463912963867,
+      "learning_rate": 2.7418131971884137e-06,
+      "loss": 0.8185,
+      "step": 40518
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.7553775310516357,
+      "learning_rate": 2.7415779640799576e-06,
+      "loss": 0.5837,
+      "step": 40519
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.640592098236084,
+      "learning_rate": 2.7413427372513297e-06,
+      "loss": 0.5474,
+      "step": 40520
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.9913289546966553,
+      "learning_rate": 2.7411075167031793e-06,
+      "loss": 0.869,
+      "step": 40521
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.544416666030884,
+      "learning_rate": 2.7408723024361644e-06,
+      "loss": 0.4947,
+      "step": 40522
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.908648133277893,
+      "learning_rate": 2.740637094450935e-06,
+      "loss": 0.2687,
+      "step": 40523
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.3861937522888184,
+      "learning_rate": 2.740401892748151e-06,
+      "loss": 0.8526,
+      "step": 40524
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.0850799083709717,
+      "learning_rate": 2.7401666973284597e-06,
+      "loss": 0.4613,
+      "step": 40525
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.1118710041046143,
+      "learning_rate": 2.7399315081925206e-06,
+      "loss": 0.6689,
+      "step": 40526
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.572204113006592,
+      "learning_rate": 2.739696325340983e-06,
+      "loss": 0.6528,
+      "step": 40527
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.6853481531143188,
+      "learning_rate": 2.739461148774504e-06,
+      "loss": 0.5492,
+      "step": 40528
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.654962420463562,
+      "learning_rate": 2.739225978493736e-06,
+      "loss": 0.5681,
+      "step": 40529
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.3518035411834717,
+      "learning_rate": 2.738990814499335e-06,
+      "loss": 0.5376,
+      "step": 40530
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.6305367946624756,
+      "learning_rate": 2.7387556567919525e-06,
+      "loss": 0.4732,
+      "step": 40531
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.5067765712738037,
+      "learning_rate": 2.7385205053722454e-06,
+      "loss": 0.5352,
+      "step": 40532
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.018644332885742,
+      "learning_rate": 2.738285360240863e-06,
+      "loss": 0.5268,
+      "step": 40533
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.514664649963379,
+      "learning_rate": 2.7380502213984645e-06,
+      "loss": 0.648,
+      "step": 40534
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.06662654876709,
+      "learning_rate": 2.737815088845699e-06,
+      "loss": 0.503,
+      "step": 40535
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.6552951335906982,
+      "learning_rate": 2.7375799625832235e-06,
+      "loss": 0.4917,
+      "step": 40536
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.877065420150757,
+      "learning_rate": 2.7373448426116894e-06,
+      "loss": 0.3105,
+      "step": 40537
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.9143357276916504,
+      "learning_rate": 2.7371097289317533e-06,
+      "loss": 0.4844,
+      "step": 40538
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.8488222360610962,
+      "learning_rate": 2.7368746215440655e-06,
+      "loss": 0.4412,
+      "step": 40539
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.861788749694824,
+      "learning_rate": 2.7366395204492836e-06,
+      "loss": 0.4772,
+      "step": 40540
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.592333197593689,
+      "learning_rate": 2.736404425648057e-06,
+      "loss": 0.5116,
+      "step": 40541
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.7311081886291504,
+      "learning_rate": 2.7361693371410443e-06,
+      "loss": 0.4716,
+      "step": 40542
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.8054730892181396,
+      "learning_rate": 2.7359342549288943e-06,
+      "loss": 0.5453,
+      "step": 40543
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.5755128860473633,
+      "learning_rate": 2.7356991790122646e-06,
+      "loss": 0.7035,
+      "step": 40544
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.716517686843872,
+      "learning_rate": 2.7354641093918057e-06,
+      "loss": 0.5037,
+      "step": 40545
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.9290709495544434,
+      "learning_rate": 2.7352290460681753e-06,
+      "loss": 0.6361,
+      "step": 40546
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.259720802307129,
+      "learning_rate": 2.734993989042022e-06,
+      "loss": 0.6194,
+      "step": 40547
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.6958054304122925,
+      "learning_rate": 2.7347589383140014e-06,
+      "loss": 0.693,
+      "step": 40548
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.9526716470718384,
+      "learning_rate": 2.7345238938847683e-06,
+      "loss": 0.5181,
+      "step": 40549
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.5421764850616455,
+      "learning_rate": 2.7342888557549775e-06,
+      "loss": 0.3531,
+      "step": 40550
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.588831663131714,
+      "learning_rate": 2.7340538239252786e-06,
+      "loss": 0.677,
+      "step": 40551
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.3179855346679688,
+      "learning_rate": 2.7338187983963286e-06,
+      "loss": 0.8525,
+      "step": 40552
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.7467739582061768,
+      "learning_rate": 2.733583779168778e-06,
+      "loss": 0.7621,
+      "step": 40553
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.6982691287994385,
+      "learning_rate": 2.7333487662432846e-06,
+      "loss": 0.6246,
+      "step": 40554
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.0379879474639893,
+      "learning_rate": 2.733113759620496e-06,
+      "loss": 0.6053,
+      "step": 40555
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.1937057971954346,
+      "learning_rate": 2.7328787593010718e-06,
+      "loss": 0.6257,
+      "step": 40556
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.31523060798645,
+      "learning_rate": 2.7326437652856595e-06,
+      "loss": 0.6619,
+      "step": 40557
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.3422858715057373,
+      "learning_rate": 2.732408777574919e-06,
+      "loss": 0.7043,
+      "step": 40558
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.511178970336914,
+      "learning_rate": 2.7321737961694973e-06,
+      "loss": 0.6253,
+      "step": 40559
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.1371500492095947,
+      "learning_rate": 2.73193882107005e-06,
+      "loss": 0.5422,
+      "step": 40560
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.1856276988983154,
+      "learning_rate": 2.7317038522772343e-06,
+      "loss": 0.6693,
+      "step": 40561
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.6309690475463867,
+      "learning_rate": 2.7314688897916985e-06,
+      "loss": 0.8084,
+      "step": 40562
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.8171863555908203,
+      "learning_rate": 2.7312339336140993e-06,
+      "loss": 0.5373,
+      "step": 40563
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.2195656299591064,
+      "learning_rate": 2.7309989837450866e-06,
+      "loss": 0.4836,
+      "step": 40564
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.0548267364501953,
+      "learning_rate": 2.730764040185318e-06,
+      "loss": 0.4732,
+      "step": 40565
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.278902053833008,
+      "learning_rate": 2.730529102935443e-06,
+      "loss": 0.6086,
+      "step": 40566
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.436748743057251,
+      "learning_rate": 2.730294171996117e-06,
+      "loss": 0.523,
+      "step": 40567
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.123922348022461,
+      "learning_rate": 2.730059247367992e-06,
+      "loss": 0.5354,
+      "step": 40568
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.9221079349517822,
+      "learning_rate": 2.7298243290517217e-06,
+      "loss": 0.6192,
+      "step": 40569
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.1413261890411377,
+      "learning_rate": 2.729589417047959e-06,
+      "loss": 0.6801,
+      "step": 40570
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.2663774490356445,
+      "learning_rate": 2.7293545113573606e-06,
+      "loss": 0.5942,
+      "step": 40571
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.5939531326293945,
+      "learning_rate": 2.729119611980574e-06,
+      "loss": 0.5405,
+      "step": 40572
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.5092382431030273,
+      "learning_rate": 2.7288847189182566e-06,
+      "loss": 0.64,
+      "step": 40573
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.458082437515259,
+      "learning_rate": 2.7286498321710587e-06,
+      "loss": 0.4593,
+      "step": 40574
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.919710397720337,
+      "learning_rate": 2.7284149517396365e-06,
+      "loss": 0.4671,
+      "step": 40575
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.973496675491333,
+      "learning_rate": 2.7281800776246394e-06,
+      "loss": 0.5268,
+      "step": 40576
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.5897252559661865,
+      "learning_rate": 2.7279452098267246e-06,
+      "loss": 0.6309,
+      "step": 40577
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.6481499671936035,
+      "learning_rate": 2.7277103483465407e-06,
+      "loss": 0.7217,
+      "step": 40578
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.031308174133301,
+      "learning_rate": 2.727475493184746e-06,
+      "loss": 0.6876,
+      "step": 40579
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.8959381580352783,
+      "learning_rate": 2.7272406443419874e-06,
+      "loss": 0.5905,
+      "step": 40580
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.2975845336914062,
+      "learning_rate": 2.727005801818924e-06,
+      "loss": 0.7978,
+      "step": 40581
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.2001404762268066,
+      "learning_rate": 2.7267709656162032e-06,
+      "loss": 0.7711,
+      "step": 40582
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.4875638484954834,
+      "learning_rate": 2.7265361357344834e-06,
+      "loss": 0.7932,
+      "step": 40583
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.9746155738830566,
+      "learning_rate": 2.726301312174413e-06,
+      "loss": 0.8595,
+      "step": 40584
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.557924747467041,
+      "learning_rate": 2.726066494936649e-06,
+      "loss": 0.7046,
+      "step": 40585
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.1428704261779785,
+      "learning_rate": 2.7258316840218394e-06,
+      "loss": 0.7843,
+      "step": 40586
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.1426148414611816,
+      "learning_rate": 2.725596879430643e-06,
+      "loss": 0.6225,
+      "step": 40587
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.673295497894287,
+      "learning_rate": 2.725362081163707e-06,
+      "loss": 0.6023,
+      "step": 40588
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.1664130687713623,
+      "learning_rate": 2.7251272892216873e-06,
+      "loss": 0.6601,
+      "step": 40589
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.0922205448150635,
+      "learning_rate": 2.7248925036052363e-06,
+      "loss": 0.5587,
+      "step": 40590
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.9094208478927612,
+      "learning_rate": 2.7246577243150085e-06,
+      "loss": 0.7201,
+      "step": 40591
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.002650260925293,
+      "learning_rate": 2.724422951351654e-06,
+      "loss": 0.5602,
+      "step": 40592
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.442915439605713,
+      "learning_rate": 2.7241881847158284e-06,
+      "loss": 0.767,
+      "step": 40593
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.765141725540161,
+      "learning_rate": 2.723953424408181e-06,
+      "loss": 0.6591,
+      "step": 40594
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.275245189666748,
+      "learning_rate": 2.7237186704293682e-06,
+      "loss": 0.6626,
+      "step": 40595
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.806046485900879,
+      "learning_rate": 2.7234839227800404e-06,
+      "loss": 0.7322,
+      "step": 40596
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.996568202972412,
+      "learning_rate": 2.723249181460851e-06,
+      "loss": 0.5655,
+      "step": 40597
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.554600238800049,
+      "learning_rate": 2.7230144464724522e-06,
+      "loss": 0.5456,
+      "step": 40598
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.4945950508117676,
+      "learning_rate": 2.722779717815499e-06,
+      "loss": 0.7211,
+      "step": 40599
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.1812498569488525,
+      "learning_rate": 2.72254499549064e-06,
+      "loss": 0.5752,
+      "step": 40600
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.147357225418091,
+      "learning_rate": 2.722310279498532e-06,
+      "loss": 0.6165,
+      "step": 40601
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.644392490386963,
+      "learning_rate": 2.7220755698398242e-06,
+      "loss": 0.4797,
+      "step": 40602
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.254970073699951,
+      "learning_rate": 2.7218408665151728e-06,
+      "loss": 0.9803,
+      "step": 40603
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.167391061782837,
+      "learning_rate": 2.721606169525227e-06,
+      "loss": 0.6154,
+      "step": 40604
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.7667877674102783,
+      "learning_rate": 2.7213714788706423e-06,
+      "loss": 0.6064,
+      "step": 40605
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 6.038074970245361,
+      "learning_rate": 2.7211367945520685e-06,
+      "loss": 0.7389,
+      "step": 40606
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.1493442058563232,
+      "learning_rate": 2.720902116570161e-06,
+      "loss": 0.8187,
+      "step": 40607
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.7229716777801514,
+      "learning_rate": 2.7206674449255695e-06,
+      "loss": 0.904,
+      "step": 40608
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.4761115312576294,
+      "learning_rate": 2.720432779618948e-06,
+      "loss": 0.5001,
+      "step": 40609
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.1976959705352783,
+      "learning_rate": 2.720198120650949e-06,
+      "loss": 0.5456,
+      "step": 40610
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.557891607284546,
+      "learning_rate": 2.719963468022226e-06,
+      "loss": 0.6339,
+      "step": 40611
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.1214821338653564,
+      "learning_rate": 2.7197288217334296e-06,
+      "loss": 0.6357,
+      "step": 40612
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.9149237871170044,
+      "learning_rate": 2.719494181785214e-06,
+      "loss": 0.4541,
+      "step": 40613
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.1286251544952393,
+      "learning_rate": 2.71925954817823e-06,
+      "loss": 0.436,
+      "step": 40614
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.8936258554458618,
+      "learning_rate": 2.719024920913132e-06,
+      "loss": 0.7227,
+      "step": 40615
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.9728095531463623,
+      "learning_rate": 2.7187902999905696e-06,
+      "loss": 0.5945,
+      "step": 40616
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.3577373027801514,
+      "learning_rate": 2.7185556854111983e-06,
+      "loss": 0.5182,
+      "step": 40617
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.17138934135437,
+      "learning_rate": 2.718321077175668e-06,
+      "loss": 0.6089,
+      "step": 40618
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.407881021499634,
+      "learning_rate": 2.7180864752846336e-06,
+      "loss": 0.6801,
+      "step": 40619
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 5.072948455810547,
+      "learning_rate": 2.717851879738743e-06,
+      "loss": 0.7305,
+      "step": 40620
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.2850589752197266,
+      "learning_rate": 2.7176172905386527e-06,
+      "loss": 0.4786,
+      "step": 40621
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.894999623298645,
+      "learning_rate": 2.717382707685015e-06,
+      "loss": 0.571,
+      "step": 40622
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.6675336360931396,
+      "learning_rate": 2.717148131178479e-06,
+      "loss": 0.6243,
+      "step": 40623
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 4.532458782196045,
+      "learning_rate": 2.7169135610197013e-06,
+      "loss": 0.6226,
+      "step": 40624
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.589481830596924,
+      "learning_rate": 2.7166789972093295e-06,
+      "loss": 0.9423,
+      "step": 40625
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.7854156494140625,
+      "learning_rate": 2.7164444397480193e-06,
+      "loss": 0.978,
+      "step": 40626
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.5725774765014648,
+      "learning_rate": 2.716209888636421e-06,
+      "loss": 0.4734,
+      "step": 40627
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.014800786972046,
+      "learning_rate": 2.7159753438751883e-06,
+      "loss": 0.4648,
+      "step": 40628
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.0985114574432373,
+      "learning_rate": 2.7157408054649713e-06,
+      "loss": 0.6231,
+      "step": 40629
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 5.115588188171387,
+      "learning_rate": 2.7155062734064232e-06,
+      "loss": 0.5642,
+      "step": 40630
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.1220664978027344,
+      "learning_rate": 2.7152717477001983e-06,
+      "loss": 0.3203,
+      "step": 40631
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.1838438510894775,
+      "learning_rate": 2.715037228346945e-06,
+      "loss": 0.6445,
+      "step": 40632
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.6518094539642334,
+      "learning_rate": 2.7148027153473166e-06,
+      "loss": 0.6316,
+      "step": 40633
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.768279790878296,
+      "learning_rate": 2.7145682087019685e-06,
+      "loss": 0.7479,
+      "step": 40634
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.4300074577331543,
+      "learning_rate": 2.7143337084115474e-06,
+      "loss": 0.4823,
+      "step": 40635
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 6.918795108795166,
+      "learning_rate": 2.714099214476711e-06,
+      "loss": 0.4793,
+      "step": 40636
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.488755226135254,
+      "learning_rate": 2.7138647268981053e-06,
+      "loss": 0.7237,
+      "step": 40637
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.4839839935302734,
+      "learning_rate": 2.713630245676388e-06,
+      "loss": 0.7236,
+      "step": 40638
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.280900478363037,
+      "learning_rate": 2.713395770812206e-06,
+      "loss": 0.6131,
+      "step": 40639
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.047302722930908,
+      "learning_rate": 2.7131613023062165e-06,
+      "loss": 0.6418,
+      "step": 40640
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.027440071105957,
+      "learning_rate": 2.7129268401590663e-06,
+      "loss": 0.6064,
+      "step": 40641
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.642348289489746,
+      "learning_rate": 2.712692384371412e-06,
+      "loss": 0.6375,
+      "step": 40642
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.948826789855957,
+      "learning_rate": 2.7124579349439013e-06,
+      "loss": 0.6158,
+      "step": 40643
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 4.098937511444092,
+      "learning_rate": 2.71222349187719e-06,
+      "loss": 0.714,
+      "step": 40644
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.6231598854064941,
+      "learning_rate": 2.7119890551719263e-06,
+      "loss": 0.263,
+      "step": 40645
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.747148275375366,
+      "learning_rate": 2.7117546248287656e-06,
+      "loss": 0.6488,
+      "step": 40646
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.20631742477417,
+      "learning_rate": 2.7115202008483567e-06,
+      "loss": 0.7158,
+      "step": 40647
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.8437705039978027,
+      "learning_rate": 2.7112857832313543e-06,
+      "loss": 0.6119,
+      "step": 40648
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 4.88289737701416,
+      "learning_rate": 2.711051371978407e-06,
+      "loss": 0.7086,
+      "step": 40649
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.1840455532073975,
+      "learning_rate": 2.7108169670901685e-06,
+      "loss": 0.7088,
+      "step": 40650
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.247788190841675,
+      "learning_rate": 2.71058256856729e-06,
+      "loss": 0.527,
+      "step": 40651
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.3536524772644043,
+      "learning_rate": 2.7103481764104255e-06,
+      "loss": 0.5302,
+      "step": 40652
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.6826999187469482,
+      "learning_rate": 2.710113790620224e-06,
+      "loss": 0.8982,
+      "step": 40653
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.9719358682632446,
+      "learning_rate": 2.7098794111973397e-06,
+      "loss": 0.4776,
+      "step": 40654
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.571355104446411,
+      "learning_rate": 2.709645038142421e-06,
+      "loss": 0.5469,
+      "step": 40655
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.72975754737854,
+      "learning_rate": 2.709410671456123e-06,
+      "loss": 0.6164,
+      "step": 40656
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.25563907623291,
+      "learning_rate": 2.7091763111390946e-06,
+      "loss": 0.7582,
+      "step": 40657
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.9250781536102295,
+      "learning_rate": 2.7089419571919907e-06,
+      "loss": 0.6983,
+      "step": 40658
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.267380714416504,
+      "learning_rate": 2.7087076096154583e-06,
+      "loss": 0.6912,
+      "step": 40659
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.8213932514190674,
+      "learning_rate": 2.708473268410154e-06,
+      "loss": 0.7524,
+      "step": 40660
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.8773695230484009,
+      "learning_rate": 2.708238933576725e-06,
+      "loss": 0.4818,
+      "step": 40661
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.016345262527466,
+      "learning_rate": 2.708004605115827e-06,
+      "loss": 0.4343,
+      "step": 40662
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.7930866479873657,
+      "learning_rate": 2.7077702830281084e-06,
+      "loss": 0.4863,
+      "step": 40663
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.2107350826263428,
+      "learning_rate": 2.707535967314223e-06,
+      "loss": 0.7662,
+      "step": 40664
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.0901639461517334,
+      "learning_rate": 2.707301657974819e-06,
+      "loss": 0.7371,
+      "step": 40665
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.2756946086883545,
+      "learning_rate": 2.707067355010553e-06,
+      "loss": 0.7383,
+      "step": 40666
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.1369786262512207,
+      "learning_rate": 2.7068330584220714e-06,
+      "loss": 0.6438,
+      "step": 40667
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.6103734970092773,
+      "learning_rate": 2.70659876821003e-06,
+      "loss": 0.4926,
+      "step": 40668
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.3413572311401367,
+      "learning_rate": 2.7063644843750767e-06,
+      "loss": 0.6146,
+      "step": 40669
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.333486557006836,
+      "learning_rate": 2.7061302069178647e-06,
+      "loss": 0.5928,
+      "step": 40670
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.10668683052063,
+      "learning_rate": 2.7058959358390446e-06,
+      "loss": 0.5998,
+      "step": 40671
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.831296682357788,
+      "learning_rate": 2.7056616711392715e-06,
+      "loss": 0.7095,
+      "step": 40672
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.9321908950805664,
+      "learning_rate": 2.7054274128191904e-06,
+      "loss": 0.5066,
+      "step": 40673
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.7169883251190186,
+      "learning_rate": 2.705193160879459e-06,
+      "loss": 0.5263,
+      "step": 40674
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.1555638313293457,
+      "learning_rate": 2.7049589153207236e-06,
+      "loss": 0.7947,
+      "step": 40675
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.4510713815689087,
+      "learning_rate": 2.7047246761436397e-06,
+      "loss": 0.4111,
+      "step": 40676
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.8825843334198,
+      "learning_rate": 2.704490443348855e-06,
+      "loss": 0.7073,
+      "step": 40677
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 4.34446907043457,
+      "learning_rate": 2.7042562169370235e-06,
+      "loss": 0.8658,
+      "step": 40678
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 4.618887901306152,
+      "learning_rate": 2.7040219969087944e-06,
+      "loss": 0.761,
+      "step": 40679
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.0535495281219482,
+      "learning_rate": 2.703787783264822e-06,
+      "loss": 0.5028,
+      "step": 40680
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.003648519515991,
+      "learning_rate": 2.703553576005753e-06,
+      "loss": 0.5051,
+      "step": 40681
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.8169429302215576,
+      "learning_rate": 2.703319375132242e-06,
+      "loss": 0.6865,
+      "step": 40682
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.366417407989502,
+      "learning_rate": 2.703085180644941e-06,
+      "loss": 0.5428,
+      "step": 40683
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.4397168159484863,
+      "learning_rate": 2.7028509925444974e-06,
+      "loss": 0.5923,
+      "step": 40684
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.9357733726501465,
+      "learning_rate": 2.7026168108315665e-06,
+      "loss": 0.754,
+      "step": 40685
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.396437168121338,
+      "learning_rate": 2.7023826355067963e-06,
+      "loss": 0.4951,
+      "step": 40686
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.2395050525665283,
+      "learning_rate": 2.7021484665708407e-06,
+      "loss": 0.4785,
+      "step": 40687
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 4.289028644561768,
+      "learning_rate": 2.7019143040243477e-06,
+      "loss": 0.4503,
+      "step": 40688
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.718654155731201,
+      "learning_rate": 2.701680147867972e-06,
+      "loss": 0.6745,
+      "step": 40689
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.211320400238037,
+      "learning_rate": 2.7014459981023617e-06,
+      "loss": 0.645,
+      "step": 40690
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.711256504058838,
+      "learning_rate": 2.7012118547281685e-06,
+      "loss": 0.4795,
+      "step": 40691
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.030259370803833,
+      "learning_rate": 2.7009777177460455e-06,
+      "loss": 0.5829,
+      "step": 40692
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.55155873298645,
+      "learning_rate": 2.7007435871566406e-06,
+      "loss": 0.5489,
+      "step": 40693
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.8680460453033447,
+      "learning_rate": 2.7005094629606066e-06,
+      "loss": 0.6459,
+      "step": 40694
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.3875598907470703,
+      "learning_rate": 2.700275345158596e-06,
+      "loss": 0.7265,
+      "step": 40695
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.9515533447265625,
+      "learning_rate": 2.7000412337512568e-06,
+      "loss": 0.6242,
+      "step": 40696
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.6452186107635498,
+      "learning_rate": 2.6998071287392425e-06,
+      "loss": 0.4814,
+      "step": 40697
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.210494041442871,
+      "learning_rate": 2.6995730301232015e-06,
+      "loss": 0.6025,
+      "step": 40698
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.141537666320801,
+      "learning_rate": 2.6993389379037882e-06,
+      "loss": 0.6729,
+      "step": 40699
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.6159329414367676,
+      "learning_rate": 2.699104852081649e-06,
+      "loss": 0.703,
+      "step": 40700
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.9986004829406738,
+      "learning_rate": 2.6988707726574394e-06,
+      "loss": 0.5263,
+      "step": 40701
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.453247308731079,
+      "learning_rate": 2.6986366996318062e-06,
+      "loss": 0.7881,
+      "step": 40702
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.011094093322754,
+      "learning_rate": 2.698402633005405e-06,
+      "loss": 0.5404,
+      "step": 40703
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.103715419769287,
+      "learning_rate": 2.698168572778881e-06,
+      "loss": 0.7304,
+      "step": 40704
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.7847940921783447,
+      "learning_rate": 2.6979345189528893e-06,
+      "loss": 0.5271,
+      "step": 40705
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.9416829347610474,
+      "learning_rate": 2.6977004715280786e-06,
+      "loss": 0.6077,
+      "step": 40706
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.2261404991149902,
+      "learning_rate": 2.697466430505102e-06,
+      "loss": 0.7363,
+      "step": 40707
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.9116265773773193,
+      "learning_rate": 2.6972323958846066e-06,
+      "loss": 0.8526,
+      "step": 40708
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.792858362197876,
+      "learning_rate": 2.696998367667247e-06,
+      "loss": 0.7387,
+      "step": 40709
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.818364143371582,
+      "learning_rate": 2.6967643458536706e-06,
+      "loss": 0.5053,
+      "step": 40710
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.1142680644989014,
+      "learning_rate": 2.69653033044453e-06,
+      "loss": 0.562,
+      "step": 40711
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.480639696121216,
+      "learning_rate": 2.6962963214404754e-06,
+      "loss": 0.9746,
+      "step": 40712
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.4085686206817627,
+      "learning_rate": 2.6960623188421598e-06,
+      "loss": 0.6226,
+      "step": 40713
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.5812501907348633,
+      "learning_rate": 2.695828322650229e-06,
+      "loss": 0.6454,
+      "step": 40714
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.691978693008423,
+      "learning_rate": 2.6955943328653393e-06,
+      "loss": 0.7199,
+      "step": 40715
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.140368938446045,
+      "learning_rate": 2.6953603494881363e-06,
+      "loss": 0.6893,
+      "step": 40716
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.95580792427063,
+      "learning_rate": 2.6951263725192754e-06,
+      "loss": 0.4994,
+      "step": 40717
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.1222422122955322,
+      "learning_rate": 2.694892401959402e-06,
+      "loss": 0.5015,
+      "step": 40718
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.8551253080368042,
+      "learning_rate": 2.694658437809171e-06,
+      "loss": 0.5517,
+      "step": 40719
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.390267848968506,
+      "learning_rate": 2.69442448006923e-06,
+      "loss": 0.7459,
+      "step": 40720
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.221062183380127,
+      "learning_rate": 2.6941905287402326e-06,
+      "loss": 0.833,
+      "step": 40721
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 5.942471504211426,
+      "learning_rate": 2.6939565838228253e-06,
+      "loss": 0.7175,
+      "step": 40722
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.1800105571746826,
+      "learning_rate": 2.6937226453176636e-06,
+      "loss": 0.5744,
+      "step": 40723
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.677215099334717,
+      "learning_rate": 2.6934887132253927e-06,
+      "loss": 0.6492,
+      "step": 40724
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.000051259994507,
+      "learning_rate": 2.6932547875466676e-06,
+      "loss": 0.7372,
+      "step": 40725
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.141115665435791,
+      "learning_rate": 2.6930208682821346e-06,
+      "loss": 0.688,
+      "step": 40726
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 4.128934860229492,
+      "learning_rate": 2.6927869554324494e-06,
+      "loss": 0.6773,
+      "step": 40727
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.2751917839050293,
+      "learning_rate": 2.6925530489982565e-06,
+      "loss": 0.4899,
+      "step": 40728
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.7457642555236816,
+      "learning_rate": 2.6923191489802116e-06,
+      "loss": 0.5574,
+      "step": 40729
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 4.864731788635254,
+      "learning_rate": 2.69208525537896e-06,
+      "loss": 0.592,
+      "step": 40730
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.3502659797668457,
+      "learning_rate": 2.6918513681951553e-06,
+      "loss": 0.66,
+      "step": 40731
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.03428316116333,
+      "learning_rate": 2.691617487429447e-06,
+      "loss": 0.6218,
+      "step": 40732
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.154695749282837,
+      "learning_rate": 2.691383613082488e-06,
+      "loss": 0.4573,
+      "step": 40733
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.150832176208496,
+      "learning_rate": 2.691149745154924e-06,
+      "loss": 0.7535,
+      "step": 40734
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.7753095626831055,
+      "learning_rate": 2.69091588364741e-06,
+      "loss": 0.5212,
+      "step": 40735
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.9397138357162476,
+      "learning_rate": 2.6906820285605918e-06,
+      "loss": 0.6872,
+      "step": 40736
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.0297961235046387,
+      "learning_rate": 2.6904481798951237e-06,
+      "loss": 0.5288,
+      "step": 40737
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.1449215412139893,
+      "learning_rate": 2.690214337651652e-06,
+      "loss": 0.6435,
+      "step": 40738
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.995143175125122,
+      "learning_rate": 2.6899805018308313e-06,
+      "loss": 0.5998,
+      "step": 40739
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.542816638946533,
+      "learning_rate": 2.689746672433307e-06,
+      "loss": 0.6761,
+      "step": 40740
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.4854042530059814,
+      "learning_rate": 2.6895128494597345e-06,
+      "loss": 0.6548,
+      "step": 40741
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.8032643795013428,
+      "learning_rate": 2.6892790329107586e-06,
+      "loss": 0.5053,
+      "step": 40742
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.214190721511841,
+      "learning_rate": 2.6890452227870346e-06,
+      "loss": 0.73,
+      "step": 40743
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.420354127883911,
+      "learning_rate": 2.6888114190892074e-06,
+      "loss": 0.8797,
+      "step": 40744
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.8882648944854736,
+      "learning_rate": 2.6885776218179306e-06,
+      "loss": 0.4482,
+      "step": 40745
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.4174649715423584,
+      "learning_rate": 2.6883438309738547e-06,
+      "loss": 0.4939,
+      "step": 40746
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.962496280670166,
+      "learning_rate": 2.688110046557627e-06,
+      "loss": 0.7147,
+      "step": 40747
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.848285675048828,
+      "learning_rate": 2.687876268569901e-06,
+      "loss": 0.5405,
+      "step": 40748
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 4.720349311828613,
+      "learning_rate": 2.6876424970113234e-06,
+      "loss": 0.6118,
+      "step": 40749
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.342947483062744,
+      "learning_rate": 2.687408731882548e-06,
+      "loss": 0.4061,
+      "step": 40750
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.870654344558716,
+      "learning_rate": 2.68717497318422e-06,
+      "loss": 0.5622,
+      "step": 40751
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.369868516921997,
+      "learning_rate": 2.6869412209169924e-06,
+      "loss": 0.3494,
+      "step": 40752
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.7939385175704956,
+      "learning_rate": 2.686707475081517e-06,
+      "loss": 0.4548,
+      "step": 40753
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.8447577953338623,
+      "learning_rate": 2.686473735678439e-06,
+      "loss": 0.6884,
+      "step": 40754
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.0638620853424072,
+      "learning_rate": 2.686240002708411e-06,
+      "loss": 0.5155,
+      "step": 40755
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.1754629611968994,
+      "learning_rate": 2.686006276172085e-06,
+      "loss": 0.6677,
+      "step": 40756
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.0644426345825195,
+      "learning_rate": 2.6857725560701065e-06,
+      "loss": 0.3977,
+      "step": 40757
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.35937762260437,
+      "learning_rate": 2.6855388424031293e-06,
+      "loss": 0.7176,
+      "step": 40758
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.258373737335205,
+      "learning_rate": 2.6853051351718003e-06,
+      "loss": 0.58,
+      "step": 40759
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.8647586107254028,
+      "learning_rate": 2.685071434376772e-06,
+      "loss": 0.5785,
+      "step": 40760
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.089459180831909,
+      "learning_rate": 2.684837740018691e-06,
+      "loss": 0.5886,
+      "step": 40761
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.0088107585906982,
+      "learning_rate": 2.684604052098211e-06,
+      "loss": 0.588,
+      "step": 40762
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.1357221603393555,
+      "learning_rate": 2.684370370615978e-06,
+      "loss": 0.5439,
+      "step": 40763
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.711454153060913,
+      "learning_rate": 2.6841366955726456e-06,
+      "loss": 0.5431,
+      "step": 40764
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 5.60703182220459,
+      "learning_rate": 2.683903026968859e-06,
+      "loss": 0.5475,
+      "step": 40765
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.329594612121582,
+      "learning_rate": 2.6836693648052726e-06,
+      "loss": 0.6105,
+      "step": 40766
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.1607730388641357,
+      "learning_rate": 2.6834357090825326e-06,
+      "loss": 0.6736,
+      "step": 40767
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 4.586357593536377,
+      "learning_rate": 2.683202059801291e-06,
+      "loss": 0.5749,
+      "step": 40768
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.023132801055908,
+      "learning_rate": 2.682968416962195e-06,
+      "loss": 0.636,
+      "step": 40769
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.2457265853881836,
+      "learning_rate": 2.682734780565898e-06,
+      "loss": 0.5316,
+      "step": 40770
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.4278323650360107,
+      "learning_rate": 2.682501150613046e-06,
+      "loss": 0.3348,
+      "step": 40771
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.4339168071746826,
+      "learning_rate": 2.68226752710429e-06,
+      "loss": 0.5449,
+      "step": 40772
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.092625617980957,
+      "learning_rate": 2.6820339100402793e-06,
+      "loss": 0.7405,
+      "step": 40773
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.59670352935791,
+      "learning_rate": 2.681800299421665e-06,
+      "loss": 0.6693,
+      "step": 40774
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.5005486011505127,
+      "learning_rate": 2.6815666952490953e-06,
+      "loss": 0.3712,
+      "step": 40775
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.0028586387634277,
+      "learning_rate": 2.6813330975232217e-06,
+      "loss": 0.6551,
+      "step": 40776
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.0622341632843018,
+      "learning_rate": 2.6810995062446898e-06,
+      "loss": 0.5355,
+      "step": 40777
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.5201022624969482,
+      "learning_rate": 2.680865921414154e-06,
+      "loss": 0.5103,
+      "step": 40778
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.7171428203582764,
+      "learning_rate": 2.6806323430322587e-06,
+      "loss": 0.4068,
+      "step": 40779
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 13.304229736328125,
+      "learning_rate": 2.680398771099658e-06,
+      "loss": 0.7077,
+      "step": 40780
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.6812095642089844,
+      "learning_rate": 2.680165205616998e-06,
+      "loss": 0.4489,
+      "step": 40781
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.4314212799072266,
+      "learning_rate": 2.679931646584931e-06,
+      "loss": 0.6221,
+      "step": 40782
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.886951208114624,
+      "learning_rate": 2.6796980940041035e-06,
+      "loss": 0.7197,
+      "step": 40783
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.7745118141174316,
+      "learning_rate": 2.679464547875168e-06,
+      "loss": 0.8533,
+      "step": 40784
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.2415223121643066,
+      "learning_rate": 2.6792310081987706e-06,
+      "loss": 0.69,
+      "step": 40785
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.157583475112915,
+      "learning_rate": 2.6789974749755643e-06,
+      "loss": 0.4869,
+      "step": 40786
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.411578416824341,
+      "learning_rate": 2.6787639482061943e-06,
+      "loss": 0.6234,
+      "step": 40787
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.782394528388977,
+      "learning_rate": 2.678530427891314e-06,
+      "loss": 0.4288,
+      "step": 40788
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.304253339767456,
+      "learning_rate": 2.6782969140315696e-06,
+      "loss": 0.6549,
+      "step": 40789
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.236053705215454,
+      "learning_rate": 2.6780634066276136e-06,
+      "loss": 0.5678,
+      "step": 40790
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.642582893371582,
+      "learning_rate": 2.677829905680091e-06,
+      "loss": 0.5424,
+      "step": 40791
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.300156831741333,
+      "learning_rate": 2.6775964111896534e-06,
+      "loss": 0.7248,
+      "step": 40792
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.2180097103118896,
+      "learning_rate": 2.677362923156951e-06,
+      "loss": 0.5709,
+      "step": 40793
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.0646655559539795,
+      "learning_rate": 2.677129441582634e-06,
+      "loss": 0.7781,
+      "step": 40794
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.7175562381744385,
+      "learning_rate": 2.6768959664673476e-06,
+      "loss": 0.6908,
+      "step": 40795
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 4.015214920043945,
+      "learning_rate": 2.676662497811745e-06,
+      "loss": 0.7025,
+      "step": 40796
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.8481647968292236,
+      "learning_rate": 2.6764290356164726e-06,
+      "loss": 0.5676,
+      "step": 40797
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.9447137117385864,
+      "learning_rate": 2.6761955798821815e-06,
+      "loss": 0.5851,
+      "step": 40798
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.4956417083740234,
+      "learning_rate": 2.6759621306095185e-06,
+      "loss": 0.4827,
+      "step": 40799
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 4.024681568145752,
+      "learning_rate": 2.6757286877991362e-06,
+      "loss": 0.7281,
+      "step": 40800
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.2957918643951416,
+      "learning_rate": 2.6754952514516798e-06,
+      "loss": 0.7225,
+      "step": 40801
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.597215175628662,
+      "learning_rate": 2.675261821567802e-06,
+      "loss": 0.7932,
+      "step": 40802
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.035309314727783,
+      "learning_rate": 2.6750283981481483e-06,
+      "loss": 0.5143,
+      "step": 40803
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.016692876815796,
+      "learning_rate": 2.6747949811933722e-06,
+      "loss": 0.7936,
+      "step": 40804
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 4.21224308013916,
+      "learning_rate": 2.6745615707041177e-06,
+      "loss": 0.6207,
+      "step": 40805
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 5.813479900360107,
+      "learning_rate": 2.6743281666810368e-06,
+      "loss": 0.9568,
+      "step": 40806
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.0393707752227783,
+      "learning_rate": 2.6740947691247797e-06,
+      "loss": 0.5946,
+      "step": 40807
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.860680341720581,
+      "learning_rate": 2.673861378035992e-06,
+      "loss": 0.4462,
+      "step": 40808
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 4.6744914054870605,
+      "learning_rate": 2.673627993415326e-06,
+      "loss": 0.5078,
+      "step": 40809
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.6404402256011963,
+      "learning_rate": 2.673394615263427e-06,
+      "loss": 0.5983,
+      "step": 40810
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.8253507614135742,
+      "learning_rate": 2.6731612435809483e-06,
+      "loss": 0.6867,
+      "step": 40811
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.8768826723098755,
+      "learning_rate": 2.6729278783685347e-06,
+      "loss": 0.6453,
+      "step": 40812
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.5062856674194336,
+      "learning_rate": 2.6726945196268365e-06,
+      "loss": 0.722,
+      "step": 40813
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 4.156707286834717,
+      "learning_rate": 2.6724611673565052e-06,
+      "loss": 0.8938,
+      "step": 40814
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.7458527088165283,
+      "learning_rate": 2.6722278215581854e-06,
+      "loss": 0.6387,
+      "step": 40815
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 5.319233417510986,
+      "learning_rate": 2.6719944822325305e-06,
+      "loss": 0.5139,
+      "step": 40816
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.165219306945801,
+      "learning_rate": 2.6717611493801844e-06,
+      "loss": 0.584,
+      "step": 40817
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.3061559200286865,
+      "learning_rate": 2.6715278230017982e-06,
+      "loss": 0.7455,
+      "step": 40818
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.8796160221099854,
+      "learning_rate": 2.671294503098023e-06,
+      "loss": 0.6775,
+      "step": 40819
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.066483497619629,
+      "learning_rate": 2.6710611896695036e-06,
+      "loss": 0.7408,
+      "step": 40820
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.116309881210327,
+      "learning_rate": 2.670827882716893e-06,
+      "loss": 0.5062,
+      "step": 40821
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.07328724861145,
+      "learning_rate": 2.670594582240835e-06,
+      "loss": 0.755,
+      "step": 40822
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.7589597702026367,
+      "learning_rate": 2.6703612882419825e-06,
+      "loss": 0.5186,
+      "step": 40823
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.642167568206787,
+      "learning_rate": 2.6701280007209817e-06,
+      "loss": 0.7034,
+      "step": 40824
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.3991503715515137,
+      "learning_rate": 2.669894719678483e-06,
+      "loss": 0.522,
+      "step": 40825
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.3627851009368896,
+      "learning_rate": 2.6696614451151328e-06,
+      "loss": 0.5532,
+      "step": 40826
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 4.463418483734131,
+      "learning_rate": 2.669428177031583e-06,
+      "loss": 0.6373,
+      "step": 40827
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.9143683910369873,
+      "learning_rate": 2.669194915428478e-06,
+      "loss": 0.5448,
+      "step": 40828
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.2069218158721924,
+      "learning_rate": 2.668961660306472e-06,
+      "loss": 0.683,
+      "step": 40829
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.7062554359436035,
+      "learning_rate": 2.6687284116662073e-06,
+      "loss": 0.6723,
+      "step": 40830
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.9021716117858887,
+      "learning_rate": 2.6684951695083376e-06,
+      "loss": 0.5504,
+      "step": 40831
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 4.331295967102051,
+      "learning_rate": 2.668261933833508e-06,
+      "loss": 0.6205,
+      "step": 40832
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.2501564025878906,
+      "learning_rate": 2.6680287046423682e-06,
+      "loss": 0.4026,
+      "step": 40833
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.107409715652466,
+      "learning_rate": 2.6677954819355666e-06,
+      "loss": 0.6902,
+      "step": 40834
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.4998817443847656,
+      "learning_rate": 2.6675622657137548e-06,
+      "loss": 0.5786,
+      "step": 40835
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 3.6096932888031006,
+      "learning_rate": 2.667329055977576e-06,
+      "loss": 0.6267,
+      "step": 40836
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.3574609756469727,
+      "learning_rate": 2.6670958527276835e-06,
+      "loss": 0.7449,
+      "step": 40837
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.9794833660125732,
+      "learning_rate": 2.6668626559647214e-06,
+      "loss": 0.4722,
+      "step": 40838
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.7159639596939087,
+      "learning_rate": 2.666629465689342e-06,
+      "loss": 0.5452,
+      "step": 40839
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.760582685470581,
+      "learning_rate": 2.6663962819021904e-06,
+      "loss": 0.6852,
+      "step": 40840
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.3067290782928467,
+      "learning_rate": 2.666163104603919e-06,
+      "loss": 0.4887,
+      "step": 40841
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.3628299236297607,
+      "learning_rate": 2.665929933795171e-06,
+      "loss": 0.4343,
+      "step": 40842
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.1884524822235107,
+      "learning_rate": 2.6656967694765994e-06,
+      "loss": 0.6608,
+      "step": 40843
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.2339603900909424,
+      "learning_rate": 2.6654636116488492e-06,
+      "loss": 0.5438,
+      "step": 40844
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.64125919342041,
+      "learning_rate": 2.6652304603125715e-06,
+      "loss": 0.8,
+      "step": 40845
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.8992117643356323,
+      "learning_rate": 2.6649973154684118e-06,
+      "loss": 0.4959,
+      "step": 40846
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.169872760772705,
+      "learning_rate": 2.6647641771170217e-06,
+      "loss": 0.5211,
+      "step": 40847
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.8739211559295654,
+      "learning_rate": 2.6645310452590455e-06,
+      "loss": 0.4445,
+      "step": 40848
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.179410457611084,
+      "learning_rate": 2.6642979198951358e-06,
+      "loss": 0.8213,
+      "step": 40849
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.0862882137298584,
+      "learning_rate": 2.6640648010259363e-06,
+      "loss": 0.634,
+      "step": 40850
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 4.085552215576172,
+      "learning_rate": 2.6638316886520997e-06,
+      "loss": 0.6261,
+      "step": 40851
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.537306308746338,
+      "learning_rate": 2.6635985827742706e-06,
+      "loss": 0.5792,
+      "step": 40852
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.267352819442749,
+      "learning_rate": 2.663365483393098e-06,
+      "loss": 0.7239,
+      "step": 40853
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.733753204345703,
+      "learning_rate": 2.6631323905092306e-06,
+      "loss": 0.6814,
+      "step": 40854
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.7512588500976562,
+      "learning_rate": 2.6628993041233186e-06,
+      "loss": 0.4503,
+      "step": 40855
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.8355878591537476,
+      "learning_rate": 2.6626662242360065e-06,
+      "loss": 0.3873,
+      "step": 40856
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.226766347885132,
+      "learning_rate": 2.6624331508479455e-06,
+      "loss": 0.746,
+      "step": 40857
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.6257824897766113,
+      "learning_rate": 2.6622000839597807e-06,
+      "loss": 0.6254,
+      "step": 40858
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.101679563522339,
+      "learning_rate": 2.6619670235721636e-06,
+      "loss": 0.6624,
+      "step": 40859
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.6680219173431396,
+      "learning_rate": 2.661733969685738e-06,
+      "loss": 0.5511,
+      "step": 40860
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.1968653202056885,
+      "learning_rate": 2.6615009223011566e-06,
+      "loss": 0.7002,
+      "step": 40861
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.266995906829834,
+      "learning_rate": 2.6612678814190633e-06,
+      "loss": 0.8179,
+      "step": 40862
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.835549831390381,
+      "learning_rate": 2.66103484704011e-06,
+      "loss": 0.5509,
+      "step": 40863
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.6086206436157227,
+      "learning_rate": 2.6608018191649397e-06,
+      "loss": 0.5056,
+      "step": 40864
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 5.38106632232666,
+      "learning_rate": 2.6605687977942062e-06,
+      "loss": 0.6027,
+      "step": 40865
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 4.278663158416748,
+      "learning_rate": 2.6603357829285524e-06,
+      "loss": 0.6613,
+      "step": 40866
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.7914693355560303,
+      "learning_rate": 2.660102774568628e-06,
+      "loss": 0.6559,
+      "step": 40867
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.587745428085327,
+      "learning_rate": 2.6598697727150836e-06,
+      "loss": 0.6173,
+      "step": 40868
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.4056081771850586,
+      "learning_rate": 2.659636777368562e-06,
+      "loss": 0.4685,
+      "step": 40869
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.3935537338256836,
+      "learning_rate": 2.6594037885297163e-06,
+      "loss": 0.7504,
+      "step": 40870
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.6872739791870117,
+      "learning_rate": 2.65917080619919e-06,
+      "loss": 0.7626,
+      "step": 40871
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.7680511474609375,
+      "learning_rate": 2.6589378303776347e-06,
+      "loss": 0.3918,
+      "step": 40872
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.32574462890625,
+      "learning_rate": 2.658704861065694e-06,
+      "loss": 0.4793,
+      "step": 40873
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.0713183879852295,
+      "learning_rate": 2.658471898264019e-06,
+      "loss": 0.4839,
+      "step": 40874
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.4755945205688477,
+      "learning_rate": 2.6582389419732573e-06,
+      "loss": 0.7312,
+      "step": 40875
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.7419946193695068,
+      "learning_rate": 2.6580059921940543e-06,
+      "loss": 0.4679,
+      "step": 40876
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.8683745861053467,
+      "learning_rate": 2.6577730489270613e-06,
+      "loss": 0.53,
+      "step": 40877
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.9480165243148804,
+      "learning_rate": 2.657540112172922e-06,
+      "loss": 0.4582,
+      "step": 40878
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.1254217624664307,
+      "learning_rate": 2.657307181932286e-06,
+      "loss": 0.6948,
+      "step": 40879
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.0684571266174316,
+      "learning_rate": 2.657074258205804e-06,
+      "loss": 0.7364,
+      "step": 40880
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.388441324234009,
+      "learning_rate": 2.656841340994118e-06,
+      "loss": 0.6353,
+      "step": 40881
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.8440699577331543,
+      "learning_rate": 2.65660843029788e-06,
+      "loss": 0.6151,
+      "step": 40882
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.371180772781372,
+      "learning_rate": 2.6563755261177347e-06,
+      "loss": 0.6951,
+      "step": 40883
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.7954821586608887,
+      "learning_rate": 2.6561426284543336e-06,
+      "loss": 0.7647,
+      "step": 40884
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 7.1784162521362305,
+      "learning_rate": 2.655909737308319e-06,
+      "loss": 0.8259,
+      "step": 40885
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.292724847793579,
+      "learning_rate": 2.6556768526803438e-06,
+      "loss": 0.4349,
+      "step": 40886
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.004399299621582,
+      "learning_rate": 2.6554439745710515e-06,
+      "loss": 0.3805,
+      "step": 40887
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.0727972984313965,
+      "learning_rate": 2.6552111029810923e-06,
+      "loss": 0.2969,
+      "step": 40888
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.2136173248291016,
+      "learning_rate": 2.654978237911111e-06,
+      "loss": 0.2066,
+      "step": 40889
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.7094147205352783,
+      "learning_rate": 2.654745379361758e-06,
+      "loss": 0.6457,
+      "step": 40890
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.566467523574829,
+      "learning_rate": 2.6545125273336793e-06,
+      "loss": 0.8142,
+      "step": 40891
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.143033742904663,
+      "learning_rate": 2.654279681827523e-06,
+      "loss": 0.6583,
+      "step": 40892
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.3732492923736572,
+      "learning_rate": 2.6540468428439347e-06,
+      "loss": 0.6206,
+      "step": 40893
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.205991506576538,
+      "learning_rate": 2.6538140103835637e-06,
+      "loss": 0.6248,
+      "step": 40894
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.0075156688690186,
+      "learning_rate": 2.6535811844470566e-06,
+      "loss": 0.6309,
+      "step": 40895
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.2724618911743164,
+      "learning_rate": 2.6533483650350634e-06,
+      "loss": 0.6642,
+      "step": 40896
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.1313047409057617,
+      "learning_rate": 2.6531155521482275e-06,
+      "loss": 0.6106,
+      "step": 40897
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.506967067718506,
+      "learning_rate": 2.6528827457872e-06,
+      "loss": 0.4098,
+      "step": 40898
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.580282688140869,
+      "learning_rate": 2.652649945952624e-06,
+      "loss": 0.6623,
+      "step": 40899
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.9394768476486206,
+      "learning_rate": 2.6524171526451513e-06,
+      "loss": 0.4889,
+      "step": 40900
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.6400296688079834,
+      "learning_rate": 2.652184365865425e-06,
+      "loss": 0.6968,
+      "step": 40901
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.5853652954101562,
+      "learning_rate": 2.6519515856140966e-06,
+      "loss": 0.6098,
+      "step": 40902
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.0227787494659424,
+      "learning_rate": 2.6517188118918093e-06,
+      "loss": 0.7765,
+      "step": 40903
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.091510772705078,
+      "learning_rate": 2.651486044699214e-06,
+      "loss": 0.5192,
+      "step": 40904
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.255492687225342,
+      "learning_rate": 2.6512532840369546e-06,
+      "loss": 0.6665,
+      "step": 40905
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.5592007637023926,
+      "learning_rate": 2.6510205299056823e-06,
+      "loss": 0.4552,
+      "step": 40906
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.34077525138855,
+      "learning_rate": 2.650787782306039e-06,
+      "loss": 0.4993,
+      "step": 40907
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.844991683959961,
+      "learning_rate": 2.6505550412386777e-06,
+      "loss": 0.5982,
+      "step": 40908
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.8704097270965576,
+      "learning_rate": 2.6503223067042406e-06,
+      "loss": 0.656,
+      "step": 40909
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.792940616607666,
+      "learning_rate": 2.6500895787033786e-06,
+      "loss": 0.6971,
+      "step": 40910
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.5275877714157104,
+      "learning_rate": 2.649856857236735e-06,
+      "loss": 0.3661,
+      "step": 40911
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.3644373416900635,
+      "learning_rate": 2.6496241423049616e-06,
+      "loss": 0.5859,
+      "step": 40912
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.759108781814575,
+      "learning_rate": 2.6493914339087005e-06,
+      "loss": 0.7291,
+      "step": 40913
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.7415032386779785,
+      "learning_rate": 2.6491587320486017e-06,
+      "loss": 0.3944,
+      "step": 40914
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.339266777038574,
+      "learning_rate": 2.6489260367253115e-06,
+      "loss": 0.745,
+      "step": 40915
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 5.647035598754883,
+      "learning_rate": 2.6486933479394792e-06,
+      "loss": 0.4135,
+      "step": 40916
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.6718451976776123,
+      "learning_rate": 2.648460665691749e-06,
+      "loss": 0.382,
+      "step": 40917
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.415358781814575,
+      "learning_rate": 2.6482279899827697e-06,
+      "loss": 0.5053,
+      "step": 40918
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.2894296646118164,
+      "learning_rate": 2.6479953208131854e-06,
+      "loss": 0.6125,
+      "step": 40919
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.4595746994018555,
+      "learning_rate": 2.647762658183647e-06,
+      "loss": 0.5106,
+      "step": 40920
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.9734957218170166,
+      "learning_rate": 2.647530002094798e-06,
+      "loss": 0.7512,
+      "step": 40921
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.819045066833496,
+      "learning_rate": 2.6472973525472877e-06,
+      "loss": 0.8606,
+      "step": 40922
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.3634727001190186,
+      "learning_rate": 2.6470647095417612e-06,
+      "loss": 0.689,
+      "step": 40923
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 4.481815338134766,
+      "learning_rate": 2.646832073078868e-06,
+      "loss": 0.9262,
+      "step": 40924
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.0004703998565674,
+      "learning_rate": 2.646599443159251e-06,
+      "loss": 0.5818,
+      "step": 40925
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.233656883239746,
+      "learning_rate": 2.6463668197835624e-06,
+      "loss": 0.6336,
+      "step": 40926
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.8686292171478271,
+      "learning_rate": 2.646134202952443e-06,
+      "loss": 0.5454,
+      "step": 40927
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.345381259918213,
+      "learning_rate": 2.6459015926665444e-06,
+      "loss": 0.586,
+      "step": 40928
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.9347264766693115,
+      "learning_rate": 2.64566898892651e-06,
+      "loss": 0.6426,
+      "step": 40929
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.526644706726074,
+      "learning_rate": 2.6454363917329885e-06,
+      "loss": 0.681,
+      "step": 40930
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.1246910095214844,
+      "learning_rate": 2.6452038010866284e-06,
+      "loss": 0.6222,
+      "step": 40931
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.7732973098754883,
+      "learning_rate": 2.6449712169880714e-06,
+      "loss": 0.7798,
+      "step": 40932
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.753751039505005,
+      "learning_rate": 2.64473863943797e-06,
+      "loss": 0.6185,
+      "step": 40933
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.7685073614120483,
+      "learning_rate": 2.644506068436966e-06,
+      "loss": 0.4719,
+      "step": 40934
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.6610546112060547,
+      "learning_rate": 2.6442735039857087e-06,
+      "loss": 0.581,
+      "step": 40935
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.8732492923736572,
+      "learning_rate": 2.644040946084846e-06,
+      "loss": 0.5402,
+      "step": 40936
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.5525929927825928,
+      "learning_rate": 2.6438083947350206e-06,
+      "loss": 0.55,
+      "step": 40937
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 7.153409004211426,
+      "learning_rate": 2.6435758499368836e-06,
+      "loss": 0.7797,
+      "step": 40938
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.5131945610046387,
+      "learning_rate": 2.643343311691078e-06,
+      "loss": 0.5561,
+      "step": 40939
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.637390375137329,
+      "learning_rate": 2.6431107799982524e-06,
+      "loss": 0.5306,
+      "step": 40940
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.096795082092285,
+      "learning_rate": 2.6428782548590513e-06,
+      "loss": 0.7322,
+      "step": 40941
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.538992166519165,
+      "learning_rate": 2.642645736274124e-06,
+      "loss": 0.7833,
+      "step": 40942
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.653534173965454,
+      "learning_rate": 2.642413224244117e-06,
+      "loss": 0.4075,
+      "step": 40943
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 11.077741622924805,
+      "learning_rate": 2.6421807187696735e-06,
+      "loss": 0.7299,
+      "step": 40944
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.2282304763793945,
+      "learning_rate": 2.641948219851444e-06,
+      "loss": 0.7569,
+      "step": 40945
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.880540132522583,
+      "learning_rate": 2.6417157274900713e-06,
+      "loss": 0.5562,
+      "step": 40946
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.9283949136734009,
+      "learning_rate": 2.6414832416862057e-06,
+      "loss": 0.603,
+      "step": 40947
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.647160053253174,
+      "learning_rate": 2.6412507624404897e-06,
+      "loss": 0.9304,
+      "step": 40948
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.7593146562576294,
+      "learning_rate": 2.6410182897535743e-06,
+      "loss": 0.7881,
+      "step": 40949
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.780128002166748,
+      "learning_rate": 2.6407858236261006e-06,
+      "loss": 0.6951,
+      "step": 40950
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.6487202644348145,
+      "learning_rate": 2.64055336405872e-06,
+      "loss": 0.6115,
+      "step": 40951
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.8174397945404053,
+      "learning_rate": 2.6403209110520744e-06,
+      "loss": 0.6554,
+      "step": 40952
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.1796486377716064,
+      "learning_rate": 2.640088464606815e-06,
+      "loss": 0.6152,
+      "step": 40953
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.4410970211029053,
+      "learning_rate": 2.6398560247235826e-06,
+      "loss": 0.6885,
+      "step": 40954
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.1966629028320312,
+      "learning_rate": 2.639623591403028e-06,
+      "loss": 0.5816,
+      "step": 40955
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.0779800415039062,
+      "learning_rate": 2.6393911646457947e-06,
+      "loss": 0.7555,
+      "step": 40956
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.761874198913574,
+      "learning_rate": 2.6391587444525324e-06,
+      "loss": 0.684,
+      "step": 40957
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.632110357284546,
+      "learning_rate": 2.6389263308238834e-06,
+      "loss": 0.6152,
+      "step": 40958
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.5147545337677,
+      "learning_rate": 2.638693923760498e-06,
+      "loss": 0.9119,
+      "step": 40959
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.4606404304504395,
+      "learning_rate": 2.638461523263018e-06,
+      "loss": 0.5615,
+      "step": 40960
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.3670222759246826,
+      "learning_rate": 2.6382291293320938e-06,
+      "loss": 0.5903,
+      "step": 40961
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.067542791366577,
+      "learning_rate": 2.637996741968368e-06,
+      "loss": 0.6361,
+      "step": 40962
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.832418918609619,
+      "learning_rate": 2.63776436117249e-06,
+      "loss": 0.5675,
+      "step": 40963
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.003523111343384,
+      "learning_rate": 2.6375319869451024e-06,
+      "loss": 0.4801,
+      "step": 40964
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.048980474472046,
+      "learning_rate": 2.6372996192868555e-06,
+      "loss": 0.6827,
+      "step": 40965
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.2657175064086914,
+      "learning_rate": 2.6370672581983914e-06,
+      "loss": 0.6042,
+      "step": 40966
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.4527530670166016,
+      "learning_rate": 2.6368349036803597e-06,
+      "loss": 0.6192,
+      "step": 40967
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.7668840885162354,
+      "learning_rate": 2.636602555733403e-06,
+      "loss": 0.8784,
+      "step": 40968
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.4129889011383057,
+      "learning_rate": 2.6363702143581717e-06,
+      "loss": 0.6429,
+      "step": 40969
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 23.661029815673828,
+      "learning_rate": 2.636137879555307e-06,
+      "loss": 0.6088,
+      "step": 40970
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.0051815509796143,
+      "learning_rate": 2.6359055513254592e-06,
+      "loss": 0.8953,
+      "step": 40971
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.3031296730041504,
+      "learning_rate": 2.635673229669271e-06,
+      "loss": 0.6824,
+      "step": 40972
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.4867985248565674,
+      "learning_rate": 2.6354409145873917e-06,
+      "loss": 0.7569,
+      "step": 40973
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 4.071038722991943,
+      "learning_rate": 2.6352086060804625e-06,
+      "loss": 0.6173,
+      "step": 40974
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.377547025680542,
+      "learning_rate": 2.634976304149134e-06,
+      "loss": 0.6453,
+      "step": 40975
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.567477822303772,
+      "learning_rate": 2.634744008794049e-06,
+      "loss": 0.6205,
+      "step": 40976
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.9388903379440308,
+      "learning_rate": 2.634511720015858e-06,
+      "loss": 0.5411,
+      "step": 40977
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.24239444732666,
+      "learning_rate": 2.6342794378152004e-06,
+      "loss": 0.6317,
+      "step": 40978
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.7145755290985107,
+      "learning_rate": 2.6340471621927282e-06,
+      "loss": 0.6566,
+      "step": 40979
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.9261815547943115,
+      "learning_rate": 2.6338148931490827e-06,
+      "loss": 0.7903,
+      "step": 40980
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.2249655723571777,
+      "learning_rate": 2.6335826306849133e-06,
+      "loss": 0.5645,
+      "step": 40981
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.7216222286224365,
+      "learning_rate": 2.6333503748008627e-06,
+      "loss": 0.8269,
+      "step": 40982
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.0985217094421387,
+      "learning_rate": 2.633118125497579e-06,
+      "loss": 0.8125,
+      "step": 40983
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.0701000690460205,
+      "learning_rate": 2.6328858827757065e-06,
+      "loss": 0.2913,
+      "step": 40984
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.188272714614868,
+      "learning_rate": 2.6326536466358926e-06,
+      "loss": 0.527,
+      "step": 40985
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.671060562133789,
+      "learning_rate": 2.6324214170787805e-06,
+      "loss": 0.449,
+      "step": 40986
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.523419141769409,
+      "learning_rate": 2.6321891941050193e-06,
+      "loss": 0.4232,
+      "step": 40987
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.1519112586975098,
+      "learning_rate": 2.6319569777152508e-06,
+      "loss": 0.4555,
+      "step": 40988
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.477297306060791,
+      "learning_rate": 2.6317247679101254e-06,
+      "loss": 0.475,
+      "step": 40989
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.470259189605713,
+      "learning_rate": 2.6314925646902843e-06,
+      "loss": 0.5883,
+      "step": 40990
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.177354574203491,
+      "learning_rate": 2.631260368056374e-06,
+      "loss": 0.5516,
+      "step": 40991
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.5154123306274414,
+      "learning_rate": 2.6310281780090444e-06,
+      "loss": 0.6778,
+      "step": 40992
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.0251059532165527,
+      "learning_rate": 2.6307959945489357e-06,
+      "loss": 0.4739,
+      "step": 40993
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 4.544609546661377,
+      "learning_rate": 2.630563817676698e-06,
+      "loss": 0.6562,
+      "step": 40994
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.397397994995117,
+      "learning_rate": 2.6303316473929722e-06,
+      "loss": 0.488,
+      "step": 40995
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 4.207374095916748,
+      "learning_rate": 2.6300994836984074e-06,
+      "loss": 0.7825,
+      "step": 40996
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.575350046157837,
+      "learning_rate": 2.6298673265936493e-06,
+      "loss": 0.6741,
+      "step": 40997
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.9244701862335205,
+      "learning_rate": 2.62963517607934e-06,
+      "loss": 0.3247,
+      "step": 40998
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.493131637573242,
+      "learning_rate": 2.62940303215613e-06,
+      "loss": 0.7686,
+      "step": 40999
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.396514892578125,
+      "learning_rate": 2.6291708948246597e-06,
+      "loss": 0.8552,
+      "step": 41000
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.0745153427124023,
+      "learning_rate": 2.628938764085579e-06,
+      "loss": 0.7097,
+      "step": 41001
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.2009780406951904,
+      "learning_rate": 2.62870663993953e-06,
+      "loss": 0.6212,
+      "step": 41002
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.685786008834839,
+      "learning_rate": 2.628474522387159e-06,
+      "loss": 0.3517,
+      "step": 41003
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 4.248183727264404,
+      "learning_rate": 2.628242411429114e-06,
+      "loss": 0.4846,
+      "step": 41004
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.7776732444763184,
+      "learning_rate": 2.6280103070660367e-06,
+      "loss": 0.6765,
+      "step": 41005
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.4096624851226807,
+      "learning_rate": 2.627778209298576e-06,
+      "loss": 0.5335,
+      "step": 41006
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.930119276046753,
+      "learning_rate": 2.627546118127373e-06,
+      "loss": 0.6191,
+      "step": 41007
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.8696463108062744,
+      "learning_rate": 2.627314033553078e-06,
+      "loss": 0.599,
+      "step": 41008
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.7975000143051147,
+      "learning_rate": 2.6270819555763315e-06,
+      "loss": 0.5855,
+      "step": 41009
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.442584753036499,
+      "learning_rate": 2.626849884197783e-06,
+      "loss": 0.827,
+      "step": 41010
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.622122287750244,
+      "learning_rate": 2.6266178194180746e-06,
+      "loss": 0.7305,
+      "step": 41011
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.144026279449463,
+      "learning_rate": 2.6263857612378548e-06,
+      "loss": 0.5127,
+      "step": 41012
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.1155130863189697,
+      "learning_rate": 2.6261537096577648e-06,
+      "loss": 0.6008,
+      "step": 41013
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.6725146770477295,
+      "learning_rate": 2.6259216646784543e-06,
+      "loss": 0.5878,
+      "step": 41014
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.3345160484313965,
+      "learning_rate": 2.6256896263005637e-06,
+      "loss": 0.6142,
+      "step": 41015
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.007209062576294,
+      "learning_rate": 2.625457594524742e-06,
+      "loss": 0.6078,
+      "step": 41016
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.643653631210327,
+      "learning_rate": 2.625225569351632e-06,
+      "loss": 0.4728,
+      "step": 41017
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.9353554248809814,
+      "learning_rate": 2.6249935507818825e-06,
+      "loss": 0.5564,
+      "step": 41018
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.582043170928955,
+      "learning_rate": 2.624761538816134e-06,
+      "loss": 0.7063,
+      "step": 41019
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.3672940731048584,
+      "learning_rate": 2.624529533455037e-06,
+      "loss": 0.5301,
+      "step": 41020
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.0746569633483887,
+      "learning_rate": 2.6242975346992305e-06,
+      "loss": 0.4676,
+      "step": 41021
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.74181866645813,
+      "learning_rate": 2.6240655425493645e-06,
+      "loss": 0.5681,
+      "step": 41022
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.1297693252563477,
+      "learning_rate": 2.6238335570060803e-06,
+      "loss": 0.7411,
+      "step": 41023
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.4835493564605713,
+      "learning_rate": 2.6236015780700274e-06,
+      "loss": 0.6237,
+      "step": 41024
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.5788331031799316,
+      "learning_rate": 2.6233696057418458e-06,
+      "loss": 0.5762,
+      "step": 41025
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.3645479679107666,
+      "learning_rate": 2.623137640022185e-06,
+      "loss": 0.5855,
+      "step": 41026
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.851001262664795,
+      "learning_rate": 2.622905680911686e-06,
+      "loss": 0.4486,
+      "step": 41027
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.9578208923339844,
+      "learning_rate": 2.622673728410998e-06,
+      "loss": 0.3421,
+      "step": 41028
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.1259450912475586,
+      "learning_rate": 2.6224417825207616e-06,
+      "loss": 0.8056,
+      "step": 41029
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.63059139251709,
+      "learning_rate": 2.6222098432416266e-06,
+      "loss": 0.7149,
+      "step": 41030
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.5289759635925293,
+      "learning_rate": 2.6219779105742326e-06,
+      "loss": 0.644,
+      "step": 41031
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.085550546646118,
+      "learning_rate": 2.6217459845192293e-06,
+      "loss": 0.715,
+      "step": 41032
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.799109697341919,
+      "learning_rate": 2.6215140650772573e-06,
+      "loss": 0.6568,
+      "step": 41033
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.8583054542541504,
+      "learning_rate": 2.621282152248966e-06,
+      "loss": 0.5674,
+      "step": 41034
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.8287501335144043,
+      "learning_rate": 2.6210502460349964e-06,
+      "loss": 0.5398,
+      "step": 41035
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.0431647300720215,
+      "learning_rate": 2.620818346435994e-06,
+      "loss": 0.6933,
+      "step": 41036
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.021127223968506,
+      "learning_rate": 2.6205864534526047e-06,
+      "loss": 0.5166,
+      "step": 41037
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.8434958457946777,
+      "learning_rate": 2.6203545670854758e-06,
+      "loss": 0.6887,
+      "step": 41038
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.0511739253997803,
+      "learning_rate": 2.620122687335247e-06,
+      "loss": 0.4846,
+      "step": 41039
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.0710811614990234,
+      "learning_rate": 2.619890814202567e-06,
+      "loss": 0.4168,
+      "step": 41040
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.5940678119659424,
+      "learning_rate": 2.619658947688078e-06,
+      "loss": 0.7539,
+      "step": 41041
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.786097526550293,
+      "learning_rate": 2.6194270877924272e-06,
+      "loss": 0.5758,
+      "step": 41042
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.9901442527770996,
+      "learning_rate": 2.6191952345162557e-06,
+      "loss": 0.8012,
+      "step": 41043
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 4.310606002807617,
+      "learning_rate": 2.618963387860213e-06,
+      "loss": 0.6885,
+      "step": 41044
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.2563211917877197,
+      "learning_rate": 2.6187315478249387e-06,
+      "loss": 0.7121,
+      "step": 41045
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.243407726287842,
+      "learning_rate": 2.618499714411082e-06,
+      "loss": 0.4527,
+      "step": 41046
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.019383430480957,
+      "learning_rate": 2.6182678876192836e-06,
+      "loss": 0.4962,
+      "step": 41047
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.418800115585327,
+      "learning_rate": 2.6180360674501915e-06,
+      "loss": 0.5559,
+      "step": 41048
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.63485050201416,
+      "learning_rate": 2.617804253904447e-06,
+      "loss": 0.7002,
+      "step": 41049
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.7916361093521118,
+      "learning_rate": 2.617572446982699e-06,
+      "loss": 0.5836,
+      "step": 41050
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.3532862663269043,
+      "learning_rate": 2.6173406466855874e-06,
+      "loss": 0.7333,
+      "step": 41051
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 5.190088748931885,
+      "learning_rate": 2.6171088530137585e-06,
+      "loss": 0.9779,
+      "step": 41052
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.6483733654022217,
+      "learning_rate": 2.6168770659678592e-06,
+      "loss": 0.6344,
+      "step": 41053
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.477294921875,
+      "learning_rate": 2.6166452855485304e-06,
+      "loss": 0.588,
+      "step": 41054
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.9570889472961426,
+      "learning_rate": 2.61641351175642e-06,
+      "loss": 0.4663,
+      "step": 41055
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.5614075660705566,
+      "learning_rate": 2.616181744592169e-06,
+      "loss": 0.6873,
+      "step": 41056
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.719642162322998,
+      "learning_rate": 2.615949984056423e-06,
+      "loss": 0.5744,
+      "step": 41057
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.458721876144409,
+      "learning_rate": 2.6157182301498296e-06,
+      "loss": 0.6604,
+      "step": 41058
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.037200689315796,
+      "learning_rate": 2.6154864828730285e-06,
+      "loss": 0.4125,
+      "step": 41059
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.0628552436828613,
+      "learning_rate": 2.6152547422266683e-06,
+      "loss": 0.4203,
+      "step": 41060
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.434232234954834,
+      "learning_rate": 2.6150230082113893e-06,
+      "loss": 0.6374,
+      "step": 41061
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.7859770059585571,
+      "learning_rate": 2.6147912808278392e-06,
+      "loss": 0.6031,
+      "step": 41062
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 5.210245132446289,
+      "learning_rate": 2.614559560076659e-06,
+      "loss": 0.7683,
+      "step": 41063
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.706421136856079,
+      "learning_rate": 2.6143278459584963e-06,
+      "loss": 0.7582,
+      "step": 41064
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.9580819606781006,
+      "learning_rate": 2.6140961384739953e-06,
+      "loss": 0.5359,
+      "step": 41065
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 4.347878932952881,
+      "learning_rate": 2.6138644376237976e-06,
+      "loss": 0.6906,
+      "step": 41066
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.7310099601745605,
+      "learning_rate": 2.613632743408551e-06,
+      "loss": 0.4447,
+      "step": 41067
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.6366193294525146,
+      "learning_rate": 2.6134010558288954e-06,
+      "loss": 0.6702,
+      "step": 41068
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.1701128482818604,
+      "learning_rate": 2.6131693748854792e-06,
+      "loss": 0.7482,
+      "step": 41069
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.3424694538116455,
+      "learning_rate": 2.6129377005789436e-06,
+      "loss": 0.6228,
+      "step": 41070
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.3647162914276123,
+      "learning_rate": 2.612706032909935e-06,
+      "loss": 0.6657,
+      "step": 41071
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.1004180908203125,
+      "learning_rate": 2.6124743718790955e-06,
+      "loss": 0.505,
+      "step": 41072
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.23132586479187,
+      "learning_rate": 2.6122427174870723e-06,
+      "loss": 0.6169,
+      "step": 41073
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.908555507659912,
+      "learning_rate": 2.612011069734506e-06,
+      "loss": 0.6597,
+      "step": 41074
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.3722164630889893,
+      "learning_rate": 2.6117794286220433e-06,
+      "loss": 0.6938,
+      "step": 41075
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.2974801063537598,
+      "learning_rate": 2.6115477941503266e-06,
+      "loss": 0.6622,
+      "step": 41076
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.8850042819976807,
+      "learning_rate": 2.611316166319999e-06,
+      "loss": 0.7543,
+      "step": 41077
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.342909812927246,
+      "learning_rate": 2.611084545131708e-06,
+      "loss": 0.7422,
+      "step": 41078
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.283273696899414,
+      "learning_rate": 2.6108529305860972e-06,
+      "loss": 0.4092,
+      "step": 41079
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.0040433406829834,
+      "learning_rate": 2.6106213226838077e-06,
+      "loss": 0.523,
+      "step": 41080
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.768343925476074,
+      "learning_rate": 2.610389721425487e-06,
+      "loss": 0.7034,
+      "step": 41081
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.569074869155884,
+      "learning_rate": 2.6101581268117756e-06,
+      "loss": 0.7454,
+      "step": 41082
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.474855899810791,
+      "learning_rate": 2.609926538843321e-06,
+      "loss": 0.9572,
+      "step": 41083
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.5595319271087646,
+      "learning_rate": 2.609694957520763e-06,
+      "loss": 0.6329,
+      "step": 41084
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.44659686088562,
+      "learning_rate": 2.6094633828447503e-06,
+      "loss": 0.6738,
+      "step": 41085
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.0007224082946777,
+      "learning_rate": 2.609231814815922e-06,
+      "loss": 0.6201,
+      "step": 41086
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.1522562503814697,
+      "learning_rate": 2.609000253434927e-06,
+      "loss": 0.7737,
+      "step": 41087
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.2488746643066406,
+      "learning_rate": 2.6087686987024047e-06,
+      "loss": 0.6709,
+      "step": 41088
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.067431688308716,
+      "learning_rate": 2.6085371506190028e-06,
+      "loss": 0.7359,
+      "step": 41089
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.182345151901245,
+      "learning_rate": 2.608305609185361e-06,
+      "loss": 0.8654,
+      "step": 41090
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.429884672164917,
+      "learning_rate": 2.608074074402127e-06,
+      "loss": 0.5781,
+      "step": 41091
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.101318359375,
+      "learning_rate": 2.6078425462699416e-06,
+      "loss": 0.5474,
+      "step": 41092
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.5978779792785645,
+      "learning_rate": 2.6076110247894516e-06,
+      "loss": 0.4297,
+      "step": 41093
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.009392738342285,
+      "learning_rate": 2.607379509961297e-06,
+      "loss": 0.6075,
+      "step": 41094
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.398655891418457,
+      "learning_rate": 2.607148001786126e-06,
+      "loss": 0.7357,
+      "step": 41095
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.8888800144195557,
+      "learning_rate": 2.606916500264578e-06,
+      "loss": 0.6839,
+      "step": 41096
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.9897236824035645,
+      "learning_rate": 2.606685005397298e-06,
+      "loss": 0.4646,
+      "step": 41097
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.4104809761047363,
+      "learning_rate": 2.6064535171849316e-06,
+      "loss": 0.7089,
+      "step": 41098
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.699817419052124,
+      "learning_rate": 2.6062220356281225e-06,
+      "loss": 0.4431,
+      "step": 41099
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.239403486251831,
+      "learning_rate": 2.6059905607275117e-06,
+      "loss": 0.556,
+      "step": 41100
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 9.367852210998535,
+      "learning_rate": 2.605759092483746e-06,
+      "loss": 0.7662,
+      "step": 41101
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.835887908935547,
+      "learning_rate": 2.6055276308974652e-06,
+      "loss": 0.7578,
+      "step": 41102
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.0435543060302734,
+      "learning_rate": 2.6052961759693175e-06,
+      "loss": 0.5313,
+      "step": 41103
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.2089617252349854,
+      "learning_rate": 2.6050647276999418e-06,
+      "loss": 0.5376,
+      "step": 41104
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.4388575553894043,
+      "learning_rate": 2.604833286089986e-06,
+      "loss": 0.476,
+      "step": 41105
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.0788745880126953,
+      "learning_rate": 2.60460185114009e-06,
+      "loss": 0.6651,
+      "step": 41106
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.049933910369873,
+      "learning_rate": 2.6043704228509006e-06,
+      "loss": 0.392,
+      "step": 41107
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.192440986633301,
+      "learning_rate": 2.6041390012230573e-06,
+      "loss": 0.7085,
+      "step": 41108
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.1163082122802734,
+      "learning_rate": 2.6039075862572083e-06,
+      "loss": 0.7763,
+      "step": 41109
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.460282564163208,
+      "learning_rate": 2.6036761779539925e-06,
+      "loss": 0.6384,
+      "step": 41110
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.639774799346924,
+      "learning_rate": 2.603444776314058e-06,
+      "loss": 0.6587,
+      "step": 41111
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.5550239086151123,
+      "learning_rate": 2.6032133813380433e-06,
+      "loss": 0.6279,
+      "step": 41112
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.9458484649658203,
+      "learning_rate": 2.6029819930265965e-06,
+      "loss": 0.662,
+      "step": 41113
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 4.571727752685547,
+      "learning_rate": 2.6027506113803567e-06,
+      "loss": 0.5519,
+      "step": 41114
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.9179575443267822,
+      "learning_rate": 2.6025192363999698e-06,
+      "loss": 1.0376,
+      "step": 41115
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.23982834815979,
+      "learning_rate": 2.6022878680860807e-06,
+      "loss": 0.6289,
+      "step": 41116
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.6702117919921875,
+      "learning_rate": 2.602056506439329e-06,
+      "loss": 0.7828,
+      "step": 41117
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.685896158218384,
+      "learning_rate": 2.6018251514603587e-06,
+      "loss": 0.4954,
+      "step": 41118
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.9470605850219727,
+      "learning_rate": 2.601593803149818e-06,
+      "loss": 0.6569,
+      "step": 41119
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.432115077972412,
+      "learning_rate": 2.6013624615083433e-06,
+      "loss": 0.6572,
+      "step": 41120
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.9846179485321045,
+      "learning_rate": 2.6011311265365835e-06,
+      "loss": 0.6187,
+      "step": 41121
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.664025068283081,
+      "learning_rate": 2.600899798235177e-06,
+      "loss": 0.6929,
+      "step": 41122
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.053295135498047,
+      "learning_rate": 2.6006684766047717e-06,
+      "loss": 0.6206,
+      "step": 41123
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.103102922439575,
+      "learning_rate": 2.6004371616460068e-06,
+      "loss": 0.391,
+      "step": 41124
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.033006429672241,
+      "learning_rate": 2.6002058533595287e-06,
+      "loss": 0.4609,
+      "step": 41125
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.9258489608764648,
+      "learning_rate": 2.5999745517459775e-06,
+      "loss": 0.4608,
+      "step": 41126
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.748640775680542,
+      "learning_rate": 2.5997432568059977e-06,
+      "loss": 0.5107,
+      "step": 41127
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.7952072620391846,
+      "learning_rate": 2.5995119685402347e-06,
+      "loss": 0.615,
+      "step": 41128
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.157916784286499,
+      "learning_rate": 2.599280686949328e-06,
+      "loss": 0.6019,
+      "step": 41129
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.321432113647461,
+      "learning_rate": 2.599049412033925e-06,
+      "loss": 0.4924,
+      "step": 41130
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.580568313598633,
+      "learning_rate": 2.598818143794663e-06,
+      "loss": 0.6116,
+      "step": 41131
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.2654902935028076,
+      "learning_rate": 2.598586882232191e-06,
+      "loss": 0.4846,
+      "step": 41132
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 4.614325523376465,
+      "learning_rate": 2.5983556273471468e-06,
+      "loss": 1.0468,
+      "step": 41133
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.6044061183929443,
+      "learning_rate": 2.5981243791401782e-06,
+      "loss": 0.6347,
+      "step": 41134
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.929884910583496,
+      "learning_rate": 2.5978931376119243e-06,
+      "loss": 0.9686,
+      "step": 41135
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.185633897781372,
+      "learning_rate": 2.597661902763031e-06,
+      "loss": 0.7439,
+      "step": 41136
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.249441385269165,
+      "learning_rate": 2.597430674594138e-06,
+      "loss": 0.6102,
+      "step": 41137
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.5942907333374023,
+      "learning_rate": 2.597199453105891e-06,
+      "loss": 0.6519,
+      "step": 41138
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.048224449157715,
+      "learning_rate": 2.5969682382989315e-06,
+      "loss": 0.6349,
+      "step": 41139
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.8644143342971802,
+      "learning_rate": 2.596737030173906e-06,
+      "loss": 0.4396,
+      "step": 41140
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.9817636013031006,
+      "learning_rate": 2.5965058287314515e-06,
+      "loss": 0.7563,
+      "step": 41141
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.2792060375213623,
+      "learning_rate": 2.5962746339722166e-06,
+      "loss": 0.6924,
+      "step": 41142
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.2792465686798096,
+      "learning_rate": 2.596043445896839e-06,
+      "loss": 0.6143,
+      "step": 41143
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.167958974838257,
+      "learning_rate": 2.595812264505967e-06,
+      "loss": 0.5034,
+      "step": 41144
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.6225078105926514,
+      "learning_rate": 2.5955810898002374e-06,
+      "loss": 0.6685,
+      "step": 41145
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.5251617431640625,
+      "learning_rate": 2.5953499217802987e-06,
+      "loss": 0.6717,
+      "step": 41146
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 4.759790420532227,
+      "learning_rate": 2.5951187604467885e-06,
+      "loss": 0.7448,
+      "step": 41147
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.4836440086364746,
+      "learning_rate": 2.594887605800355e-06,
+      "loss": 0.6227,
+      "step": 41148
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.3979332447052,
+      "learning_rate": 2.594656457841635e-06,
+      "loss": 0.6651,
+      "step": 41149
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.6790125370025635,
+      "learning_rate": 2.5944253165712776e-06,
+      "loss": 0.5474,
+      "step": 41150
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.513446807861328,
+      "learning_rate": 2.5941941819899186e-06,
+      "loss": 0.6923,
+      "step": 41151
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.018068313598633,
+      "learning_rate": 2.5939630540982076e-06,
+      "loss": 0.5839,
+      "step": 41152
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.865053653717041,
+      "learning_rate": 2.5937319328967814e-06,
+      "loss": 0.5719,
+      "step": 41153
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.8921236991882324,
+      "learning_rate": 2.5935008183862877e-06,
+      "loss": 0.734,
+      "step": 41154
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.3852570056915283,
+      "learning_rate": 2.593269710567365e-06,
+      "loss": 0.5594,
+      "step": 41155
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.753782033920288,
+      "learning_rate": 2.593038609440659e-06,
+      "loss": 0.8805,
+      "step": 41156
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.1927664279937744,
+      "learning_rate": 2.5928075150068095e-06,
+      "loss": 0.5342,
+      "step": 41157
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 4.7907867431640625,
+      "learning_rate": 2.59257642726646e-06,
+      "loss": 0.7582,
+      "step": 41158
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.2891929149627686,
+      "learning_rate": 2.592345346220254e-06,
+      "loss": 0.6424,
+      "step": 41159
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 4.528055191040039,
+      "learning_rate": 2.592114271868835e-06,
+      "loss": 0.3996,
+      "step": 41160
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.532374858856201,
+      "learning_rate": 2.5918832042128435e-06,
+      "loss": 0.6507,
+      "step": 41161
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.144641160964966,
+      "learning_rate": 2.591652143252924e-06,
+      "loss": 0.7383,
+      "step": 41162
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.508100986480713,
+      "learning_rate": 2.591421088989715e-06,
+      "loss": 0.5,
+      "step": 41163
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.3848140239715576,
+      "learning_rate": 2.5911900414238642e-06,
+      "loss": 0.6212,
+      "step": 41164
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 5.9356231689453125,
+      "learning_rate": 2.5909590005560104e-06,
+      "loss": 0.6648,
+      "step": 41165
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.0015523433685303,
+      "learning_rate": 2.590727966386798e-06,
+      "loss": 0.6179,
+      "step": 41166
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.3469574451446533,
+      "learning_rate": 2.5904969389168676e-06,
+      "loss": 0.4216,
+      "step": 41167
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.925265312194824,
+      "learning_rate": 2.5902659181468635e-06,
+      "loss": 0.6422,
+      "step": 41168
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.5879008769989014,
+      "learning_rate": 2.5900349040774258e-06,
+      "loss": 0.7154,
+      "step": 41169
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.243077516555786,
+      "learning_rate": 2.5898038967092e-06,
+      "loss": 0.5343,
+      "step": 41170
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.7244380712509155,
+      "learning_rate": 2.5895728960428252e-06,
+      "loss": 0.4371,
+      "step": 41171
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.3529834747314453,
+      "learning_rate": 2.589341902078947e-06,
+      "loss": 0.5404,
+      "step": 41172
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.231656551361084,
+      "learning_rate": 2.5891109148182037e-06,
+      "loss": 0.6259,
+      "step": 41173
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.6552000045776367,
+      "learning_rate": 2.5888799342612412e-06,
+      "loss": 0.5449,
+      "step": 41174
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.5942275524139404,
+      "learning_rate": 2.5886489604086996e-06,
+      "loss": 0.6732,
+      "step": 41175
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.0484914779663086,
+      "learning_rate": 2.588417993261221e-06,
+      "loss": 0.6726,
+      "step": 41176
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.922776222229004,
+      "learning_rate": 2.5881870328194515e-06,
+      "loss": 0.8089,
+      "step": 41177
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.2441093921661377,
+      "learning_rate": 2.5879560790840274e-06,
+      "loss": 0.4511,
+      "step": 41178
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.482722759246826,
+      "learning_rate": 2.587725132055594e-06,
+      "loss": 0.904,
+      "step": 41179
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.7428104877471924,
+      "learning_rate": 2.5874941917347964e-06,
+      "loss": 0.5357,
+      "step": 41180
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.8108588457107544,
+      "learning_rate": 2.58726325812227e-06,
+      "loss": 0.5109,
+      "step": 41181
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.3666062355041504,
+      "learning_rate": 2.587032331218664e-06,
+      "loss": 0.8553,
+      "step": 41182
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 4.4829020500183105,
+      "learning_rate": 2.586801411024615e-06,
+      "loss": 0.6411,
+      "step": 41183
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.763049602508545,
+      "learning_rate": 2.586570497540769e-06,
+      "loss": 0.3567,
+      "step": 41184
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.9864689111709595,
+      "learning_rate": 2.586339590767764e-06,
+      "loss": 0.5982,
+      "step": 41185
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.8771278858184814,
+      "learning_rate": 2.5861086907062473e-06,
+      "loss": 0.6854,
+      "step": 41186
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 5.413724422454834,
+      "learning_rate": 2.585877797356855e-06,
+      "loss": 0.7701,
+      "step": 41187
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.2017030715942383,
+      "learning_rate": 2.585646910720233e-06,
+      "loss": 0.6858,
+      "step": 41188
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.862591505050659,
+      "learning_rate": 2.5854160307970243e-06,
+      "loss": 0.8254,
+      "step": 41189
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.8781492710113525,
+      "learning_rate": 2.5851851575878672e-06,
+      "loss": 0.6216,
+      "step": 41190
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.6936841011047363,
+      "learning_rate": 2.584954291093407e-06,
+      "loss": 0.6245,
+      "step": 41191
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.10809326171875,
+      "learning_rate": 2.5847234313142827e-06,
+      "loss": 0.644,
+      "step": 41192
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 4.045553207397461,
+      "learning_rate": 2.5844925782511397e-06,
+      "loss": 0.554,
+      "step": 41193
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.4370319843292236,
+      "learning_rate": 2.584261731904616e-06,
+      "loss": 0.3125,
+      "step": 41194
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.632882595062256,
+      "learning_rate": 2.5840308922753577e-06,
+      "loss": 0.6486,
+      "step": 41195
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.3824472427368164,
+      "learning_rate": 2.5838000593640023e-06,
+      "loss": 0.6159,
+      "step": 41196
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.1916589736938477,
+      "learning_rate": 2.5835692331711953e-06,
+      "loss": 0.7708,
+      "step": 41197
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.2983102798461914,
+      "learning_rate": 2.5833384136975755e-06,
+      "loss": 0.6653,
+      "step": 41198
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.825007438659668,
+      "learning_rate": 2.583107600943786e-06,
+      "loss": 0.3721,
+      "step": 41199
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.765758991241455,
+      "learning_rate": 2.5828767949104695e-06,
+      "loss": 0.5041,
+      "step": 41200
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.720644474029541,
+      "learning_rate": 2.5826459955982684e-06,
+      "loss": 0.5163,
+      "step": 41201
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.29386043548584,
+      "learning_rate": 2.582415203007822e-06,
+      "loss": 0.7527,
+      "step": 41202
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.2280285358428955,
+      "learning_rate": 2.5821844171397746e-06,
+      "loss": 0.801,
+      "step": 41203
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.799253225326538,
+      "learning_rate": 2.5819536379947647e-06,
+      "loss": 0.5054,
+      "step": 41204
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 4.097208023071289,
+      "learning_rate": 2.5817228655734384e-06,
+      "loss": 0.6242,
+      "step": 41205
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 4.5002875328063965,
+      "learning_rate": 2.581492099876432e-06,
+      "loss": 0.6103,
+      "step": 41206
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.3490047454833984,
+      "learning_rate": 2.581261340904393e-06,
+      "loss": 0.4595,
+      "step": 41207
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.067239761352539,
+      "learning_rate": 2.5810305886579578e-06,
+      "loss": 0.5887,
+      "step": 41208
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.359968900680542,
+      "learning_rate": 2.5807998431377723e-06,
+      "loss": 0.5955,
+      "step": 41209
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.0219309329986572,
+      "learning_rate": 2.5805691043444735e-06,
+      "loss": 0.436,
+      "step": 41210
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 4.582586288452148,
+      "learning_rate": 2.580338372278709e-06,
+      "loss": 0.4593,
+      "step": 41211
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.229398488998413,
+      "learning_rate": 2.5801076469411147e-06,
+      "loss": 0.8139,
+      "step": 41212
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.2391278743743896,
+      "learning_rate": 2.5798769283323353e-06,
+      "loss": 0.5298,
+      "step": 41213
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.593294620513916,
+      "learning_rate": 2.579646216453011e-06,
+      "loss": 0.6158,
+      "step": 41214
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.4537837505340576,
+      "learning_rate": 2.5794155113037857e-06,
+      "loss": 0.5687,
+      "step": 41215
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.5517706871032715,
+      "learning_rate": 2.5791848128852966e-06,
+      "loss": 0.6422,
+      "step": 41216
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.285229444503784,
+      "learning_rate": 2.5789541211981893e-06,
+      "loss": 0.6606,
+      "step": 41217
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.374626636505127,
+      "learning_rate": 2.5787234362431026e-06,
+      "loss": 0.4042,
+      "step": 41218
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.516078472137451,
+      "learning_rate": 2.5784927580206788e-06,
+      "loss": 0.6178,
+      "step": 41219
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.6376826763153076,
+      "learning_rate": 2.5782620865315596e-06,
+      "loss": 0.5274,
+      "step": 41220
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.223449945449829,
+      "learning_rate": 2.578031421776388e-06,
+      "loss": 0.6507,
+      "step": 41221
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.0214924812316895,
+      "learning_rate": 2.5778007637558016e-06,
+      "loss": 0.6142,
+      "step": 41222
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.0322494506835938,
+      "learning_rate": 2.577570112470446e-06,
+      "loss": 0.7703,
+      "step": 41223
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.7245064973831177,
+      "learning_rate": 2.5773394679209586e-06,
+      "loss": 0.5951,
+      "step": 41224
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.691462278366089,
+      "learning_rate": 2.5771088301079843e-06,
+      "loss": 0.6815,
+      "step": 41225
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.5482003688812256,
+      "learning_rate": 2.576878199032161e-06,
+      "loss": 0.6499,
+      "step": 41226
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.2531752586364746,
+      "learning_rate": 2.5766475746941338e-06,
+      "loss": 0.9048,
+      "step": 41227
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.090759038925171,
+      "learning_rate": 2.5764169570945395e-06,
+      "loss": 0.5809,
+      "step": 41228
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.8817989826202393,
+      "learning_rate": 2.576186346234024e-06,
+      "loss": 0.642,
+      "step": 41229
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.5965631008148193,
+      "learning_rate": 2.5759557421132243e-06,
+      "loss": 0.5222,
+      "step": 41230
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 4.141682147979736,
+      "learning_rate": 2.575725144732786e-06,
+      "loss": 0.7387,
+      "step": 41231
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.895033121109009,
+      "learning_rate": 2.5754945540933452e-06,
+      "loss": 0.5907,
+      "step": 41232
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.581207513809204,
+      "learning_rate": 2.5752639701955482e-06,
+      "loss": 0.6505,
+      "step": 41233
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.54999041557312,
+      "learning_rate": 2.575033393040032e-06,
+      "loss": 0.5321,
+      "step": 41234
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.7996846437454224,
+      "learning_rate": 2.574802822627441e-06,
+      "loss": 0.673,
+      "step": 41235
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.708181858062744,
+      "learning_rate": 2.5745722589584134e-06,
+      "loss": 0.6087,
+      "step": 41236
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.456264019012451,
+      "learning_rate": 2.5743417020335924e-06,
+      "loss": 0.5723,
+      "step": 41237
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.1885452270507812,
+      "learning_rate": 2.5741111518536177e-06,
+      "loss": 0.6017,
+      "step": 41238
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.9369271993637085,
+      "learning_rate": 2.5738806084191314e-06,
+      "loss": 0.4826,
+      "step": 41239
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.441718101501465,
+      "learning_rate": 2.5736500717307733e-06,
+      "loss": 0.4796,
+      "step": 41240
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.8925492763519287,
+      "learning_rate": 2.5734195417891885e-06,
+      "loss": 0.5641,
+      "step": 41241
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 7.613796234130859,
+      "learning_rate": 2.5731890185950122e-06,
+      "loss": 0.7063,
+      "step": 41242
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.911808729171753,
+      "learning_rate": 2.57295850214889e-06,
+      "loss": 0.4901,
+      "step": 41243
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 4.279675483703613,
+      "learning_rate": 2.572727992451459e-06,
+      "loss": 0.749,
+      "step": 41244
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.2495357990264893,
+      "learning_rate": 2.572497489503365e-06,
+      "loss": 0.7191,
+      "step": 41245
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.3161826133728027,
+      "learning_rate": 2.5722669933052435e-06,
+      "loss": 0.9741,
+      "step": 41246
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.919642210006714,
+      "learning_rate": 2.57203650385774e-06,
+      "loss": 0.5675,
+      "step": 41247
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.5235345363616943,
+      "learning_rate": 2.5718060211614913e-06,
+      "loss": 0.4996,
+      "step": 41248
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.2173261642456055,
+      "learning_rate": 2.571575545217141e-06,
+      "loss": 0.4616,
+      "step": 41249
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.1373486518859863,
+      "learning_rate": 2.5713450760253312e-06,
+      "loss": 0.5752,
+      "step": 41250
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.9988291263580322,
+      "learning_rate": 2.5711146135866993e-06,
+      "loss": 0.52,
+      "step": 41251
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.8133554458618164,
+      "learning_rate": 2.5708841579018888e-06,
+      "loss": 0.5566,
+      "step": 41252
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.757298231124878,
+      "learning_rate": 2.570653708971538e-06,
+      "loss": 0.844,
+      "step": 41253
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.0479705333709717,
+      "learning_rate": 2.5704232667962913e-06,
+      "loss": 0.6278,
+      "step": 41254
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.2492828369140625,
+      "learning_rate": 2.5701928313767856e-06,
+      "loss": 0.7131,
+      "step": 41255
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.166860818862915,
+      "learning_rate": 2.569962402713665e-06,
+      "loss": 0.6415,
+      "step": 41256
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.1564838886260986,
+      "learning_rate": 2.5697319808075664e-06,
+      "loss": 0.3943,
+      "step": 41257
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.0258049964904785,
+      "learning_rate": 2.569501565659135e-06,
+      "loss": 0.3802,
+      "step": 41258
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.6022562980651855,
+      "learning_rate": 2.5692711572690075e-06,
+      "loss": 0.5348,
+      "step": 41259
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.94551682472229,
+      "learning_rate": 2.569040755637826e-06,
+      "loss": 0.3985,
+      "step": 41260
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.1710877418518066,
+      "learning_rate": 2.5688103607662317e-06,
+      "loss": 0.4421,
+      "step": 41261
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.1479241847991943,
+      "learning_rate": 2.568579972654867e-06,
+      "loss": 0.4714,
+      "step": 41262
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.764231562614441,
+      "learning_rate": 2.5683495913043686e-06,
+      "loss": 0.7977,
+      "step": 41263
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.017976999282837,
+      "learning_rate": 2.5681192167153814e-06,
+      "loss": 0.7326,
+      "step": 41264
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.583809733390808,
+      "learning_rate": 2.567888848888541e-06,
+      "loss": 0.4337,
+      "step": 41265
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.863717555999756,
+      "learning_rate": 2.5676584878244933e-06,
+      "loss": 0.5091,
+      "step": 41266
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.37545108795166,
+      "learning_rate": 2.567428133523874e-06,
+      "loss": 0.7593,
+      "step": 41267
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.8819501399993896,
+      "learning_rate": 2.567197785987327e-06,
+      "loss": 0.3747,
+      "step": 41268
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.6710257530212402,
+      "learning_rate": 2.5669674452154904e-06,
+      "loss": 0.6123,
+      "step": 41269
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.6412022113800049,
+      "learning_rate": 2.566737111209008e-06,
+      "loss": 0.6629,
+      "step": 41270
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.7013180255889893,
+      "learning_rate": 2.5665067839685153e-06,
+      "loss": 0.3627,
+      "step": 41271
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.349558115005493,
+      "learning_rate": 2.5662764634946585e-06,
+      "loss": 0.6105,
+      "step": 41272
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.7666971683502197,
+      "learning_rate": 2.5660461497880728e-06,
+      "loss": 0.6365,
+      "step": 41273
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 4.280027866363525,
+      "learning_rate": 2.565815842849403e-06,
+      "loss": 0.4653,
+      "step": 41274
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.16339111328125,
+      "learning_rate": 2.565585542679285e-06,
+      "loss": 0.7762,
+      "step": 41275
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.755110502243042,
+      "learning_rate": 2.565355249278365e-06,
+      "loss": 0.4488,
+      "step": 41276
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.078505277633667,
+      "learning_rate": 2.5651249626472767e-06,
+      "loss": 0.5287,
+      "step": 41277
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.2979516983032227,
+      "learning_rate": 2.564894682786666e-06,
+      "loss": 0.6552,
+      "step": 41278
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.212028980255127,
+      "learning_rate": 2.5646644096971697e-06,
+      "loss": 0.388,
+      "step": 41279
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.2780332565307617,
+      "learning_rate": 2.564434143379429e-06,
+      "loss": 0.3717,
+      "step": 41280
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.3328144550323486,
+      "learning_rate": 2.5642038838340845e-06,
+      "loss": 0.7986,
+      "step": 41281
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.5821332931518555,
+      "learning_rate": 2.563973631061778e-06,
+      "loss": 0.5723,
+      "step": 41282
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.6072750091552734,
+      "learning_rate": 2.563743385063147e-06,
+      "loss": 0.7746,
+      "step": 41283
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.512237310409546,
+      "learning_rate": 2.563513145838835e-06,
+      "loss": 0.424,
+      "step": 41284
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.0636649131774902,
+      "learning_rate": 2.563282913389478e-06,
+      "loss": 0.7439,
+      "step": 41285
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.957331895828247,
+      "learning_rate": 2.56305268771572e-06,
+      "loss": 0.7287,
+      "step": 41286
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 5.361445903778076,
+      "learning_rate": 2.562822468818198e-06,
+      "loss": 0.5599,
+      "step": 41287
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.6523263454437256,
+      "learning_rate": 2.562592256697556e-06,
+      "loss": 0.5547,
+      "step": 41288
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.419689893722534,
+      "learning_rate": 2.5623620513544294e-06,
+      "loss": 0.6343,
+      "step": 41289
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.8244062662124634,
+      "learning_rate": 2.562131852789463e-06,
+      "loss": 0.5434,
+      "step": 41290
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 4.445605754852295,
+      "learning_rate": 2.561901661003293e-06,
+      "loss": 0.5992,
+      "step": 41291
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.7552268505096436,
+      "learning_rate": 2.561671475996562e-06,
+      "loss": 0.7765,
+      "step": 41292
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.129793405532837,
+      "learning_rate": 2.5614412977699084e-06,
+      "loss": 0.4134,
+      "step": 41293
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.409003734588623,
+      "learning_rate": 2.561211126323975e-06,
+      "loss": 0.5258,
+      "step": 41294
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 4.586939334869385,
+      "learning_rate": 2.5609809616593974e-06,
+      "loss": 0.6511,
+      "step": 41295
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.1818313598632812,
+      "learning_rate": 2.560750803776819e-06,
+      "loss": 0.3636,
+      "step": 41296
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 4.191945552825928,
+      "learning_rate": 2.5605206526768787e-06,
+      "loss": 0.755,
+      "step": 41297
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.137019157409668,
+      "learning_rate": 2.5602905083602183e-06,
+      "loss": 0.644,
+      "step": 41298
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.07462477684021,
+      "learning_rate": 2.560060370827473e-06,
+      "loss": 0.7209,
+      "step": 41299
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.569009780883789,
+      "learning_rate": 2.559830240079286e-06,
+      "loss": 0.779,
+      "step": 41300
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.9399516582489014,
+      "learning_rate": 2.559600116116298e-06,
+      "loss": 0.6598,
+      "step": 41301
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.5859127044677734,
+      "learning_rate": 2.5593699989391497e-06,
+      "loss": 0.6258,
+      "step": 41302
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.562237501144409,
+      "learning_rate": 2.5591398885484763e-06,
+      "loss": 0.5942,
+      "step": 41303
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.404935836791992,
+      "learning_rate": 2.558909784944923e-06,
+      "loss": 0.7657,
+      "step": 41304
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 4.273112773895264,
+      "learning_rate": 2.5586796881291253e-06,
+      "loss": 0.6206,
+      "step": 41305
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.2134337425231934,
+      "learning_rate": 2.5584495981017266e-06,
+      "loss": 0.4987,
+      "step": 41306
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.772637367248535,
+      "learning_rate": 2.5582195148633627e-06,
+      "loss": 0.6362,
+      "step": 41307
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.9518353939056396,
+      "learning_rate": 2.557989438414678e-06,
+      "loss": 0.4127,
+      "step": 41308
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.8162643909454346,
+      "learning_rate": 2.557759368756308e-06,
+      "loss": 0.4761,
+      "step": 41309
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.427811861038208,
+      "learning_rate": 2.5575293058888963e-06,
+      "loss": 0.618,
+      "step": 41310
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.028616189956665,
+      "learning_rate": 2.5572992498130777e-06,
+      "loss": 0.6931,
+      "step": 41311
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.4001383781433105,
+      "learning_rate": 2.557069200529496e-06,
+      "loss": 0.4969,
+      "step": 41312
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.1638832092285156,
+      "learning_rate": 2.556839158038791e-06,
+      "loss": 0.4887,
+      "step": 41313
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.9856388568878174,
+      "learning_rate": 2.5566091223415996e-06,
+      "loss": 0.4742,
+      "step": 41314
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.3958804607391357,
+      "learning_rate": 2.5563790934385645e-06,
+      "loss": 0.6303,
+      "step": 41315
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.795931100845337,
+      "learning_rate": 2.5561490713303215e-06,
+      "loss": 0.5628,
+      "step": 41316
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.280102252960205,
+      "learning_rate": 2.555919056017515e-06,
+      "loss": 0.5285,
+      "step": 41317
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.28376841545105,
+      "learning_rate": 2.55568904750078e-06,
+      "loss": 0.7885,
+      "step": 41318
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.5710439682006836,
+      "learning_rate": 2.5554590457807606e-06,
+      "loss": 0.5327,
+      "step": 41319
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.3817391395568848,
+      "learning_rate": 2.555229050858091e-06,
+      "loss": 0.5554,
+      "step": 41320
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.9691579341888428,
+      "learning_rate": 2.5549990627334135e-06,
+      "loss": 0.5784,
+      "step": 41321
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.0156710147857666,
+      "learning_rate": 2.554769081407368e-06,
+      "loss": 0.4511,
+      "step": 41322
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.9537413120269775,
+      "learning_rate": 2.554539106880596e-06,
+      "loss": 0.5089,
+      "step": 41323
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.5791237354278564,
+      "learning_rate": 2.5543091391537323e-06,
+      "loss": 0.4507,
+      "step": 41324
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.701152801513672,
+      "learning_rate": 2.5540791782274206e-06,
+      "loss": 0.5969,
+      "step": 41325
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.9471619129180908,
+      "learning_rate": 2.553849224102296e-06,
+      "loss": 0.5859,
+      "step": 41326
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.5062036514282227,
+      "learning_rate": 2.5536192767790035e-06,
+      "loss": 0.2921,
+      "step": 41327
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.0193259716033936,
+      "learning_rate": 2.5533893362581763e-06,
+      "loss": 0.4808,
+      "step": 41328
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.693665027618408,
+      "learning_rate": 2.5531594025404594e-06,
+      "loss": 0.7599,
+      "step": 41329
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.851822018623352,
+      "learning_rate": 2.5529294756264876e-06,
+      "loss": 0.6223,
+      "step": 41330
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.318391799926758,
+      "learning_rate": 2.5526995555169043e-06,
+      "loss": 0.556,
+      "step": 41331
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.4899725914001465,
+      "learning_rate": 2.5524696422123447e-06,
+      "loss": 0.5212,
+      "step": 41332
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.884767532348633,
+      "learning_rate": 2.552239735713452e-06,
+      "loss": 0.6452,
+      "step": 41333
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.687514305114746,
+      "learning_rate": 2.5520098360208616e-06,
+      "loss": 0.6927,
+      "step": 41334
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 5.760036945343018,
+      "learning_rate": 2.551779943135217e-06,
+      "loss": 0.8051,
+      "step": 41335
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.0342371463775635,
+      "learning_rate": 2.551550057057153e-06,
+      "loss": 0.609,
+      "step": 41336
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.09006667137146,
+      "learning_rate": 2.551320177787313e-06,
+      "loss": 0.3774,
+      "step": 41337
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.741703510284424,
+      "learning_rate": 2.5510903053263325e-06,
+      "loss": 0.4126,
+      "step": 41338
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.3386154174804688,
+      "learning_rate": 2.550860439674855e-06,
+      "loss": 0.5266,
+      "step": 41339
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.2012624740600586,
+      "learning_rate": 2.550630580833514e-06,
+      "loss": 0.6172,
+      "step": 41340
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.1911072731018066,
+      "learning_rate": 2.550400728802953e-06,
+      "loss": 0.7394,
+      "step": 41341
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 4.20362663269043,
+      "learning_rate": 2.550170883583809e-06,
+      "loss": 0.632,
+      "step": 41342
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.7332091331481934,
+      "learning_rate": 2.549941045176725e-06,
+      "loss": 0.7293,
+      "step": 41343
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.5795421600341797,
+      "learning_rate": 2.549711213582334e-06,
+      "loss": 0.3905,
+      "step": 41344
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.202331304550171,
+      "learning_rate": 2.5494813888012815e-06,
+      "loss": 0.6899,
+      "step": 41345
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.598312497138977,
+      "learning_rate": 2.5492515708342e-06,
+      "loss": 0.5031,
+      "step": 41346
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.6524615287780762,
+      "learning_rate": 2.549021759681735e-06,
+      "loss": 0.5294,
+      "step": 41347
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.9383232593536377,
+      "learning_rate": 2.5487919553445194e-06,
+      "loss": 0.6458,
+      "step": 41348
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.957455635070801,
+      "learning_rate": 2.5485621578231973e-06,
+      "loss": 0.5214,
+      "step": 41349
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.8821898698806763,
+      "learning_rate": 2.5483323671184034e-06,
+      "loss": 0.6735,
+      "step": 41350
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.251958131790161,
+      "learning_rate": 2.5481025832307817e-06,
+      "loss": 0.7245,
+      "step": 41351
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.100811243057251,
+      "learning_rate": 2.5478728061609647e-06,
+      "loss": 0.7133,
+      "step": 41352
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.708726644515991,
+      "learning_rate": 2.547643035909598e-06,
+      "loss": 0.6983,
+      "step": 41353
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.424966335296631,
+      "learning_rate": 2.5474132724773147e-06,
+      "loss": 0.4792,
+      "step": 41354
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.8171074390411377,
+      "learning_rate": 2.547183515864759e-06,
+      "loss": 0.5543,
+      "step": 41355
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.9850540161132812,
+      "learning_rate": 2.546953766072565e-06,
+      "loss": 0.6746,
+      "step": 41356
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.0236001014709473,
+      "learning_rate": 2.546724023101375e-06,
+      "loss": 0.5184,
+      "step": 41357
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.9036673307418823,
+      "learning_rate": 2.546494286951825e-06,
+      "loss": 0.5549,
+      "step": 41358
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.0184733867645264,
+      "learning_rate": 2.546264557624557e-06,
+      "loss": 0.418,
+      "step": 41359
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.344656467437744,
+      "learning_rate": 2.5460348351202064e-06,
+      "loss": 0.6218,
+      "step": 41360
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.4646246433258057,
+      "learning_rate": 2.5458051194394135e-06,
+      "loss": 0.3546,
+      "step": 41361
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.6990292072296143,
+      "learning_rate": 2.545575410582817e-06,
+      "loss": 0.5852,
+      "step": 41362
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.0943050384521484,
+      "learning_rate": 2.545345708551058e-06,
+      "loss": 0.6567,
+      "step": 41363
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.7386209964752197,
+      "learning_rate": 2.545116013344771e-06,
+      "loss": 0.5317,
+      "step": 41364
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.9110097885131836,
+      "learning_rate": 2.544886324964599e-06,
+      "loss": 0.6245,
+      "step": 41365
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.2227022647857666,
+      "learning_rate": 2.5446566434111764e-06,
+      "loss": 0.5543,
+      "step": 41366
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.9504550695419312,
+      "learning_rate": 2.544426968685145e-06,
+      "loss": 0.7213,
+      "step": 41367
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 10.00892162322998,
+      "learning_rate": 2.544197300787141e-06,
+      "loss": 0.6483,
+      "step": 41368
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.6037843227386475,
+      "learning_rate": 2.5439676397178072e-06,
+      "loss": 0.5167,
+      "step": 41369
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 4.134668350219727,
+      "learning_rate": 2.543737985477776e-06,
+      "loss": 0.6404,
+      "step": 41370
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.472944498062134,
+      "learning_rate": 2.5435083380676917e-06,
+      "loss": 0.6231,
+      "step": 41371
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.867323637008667,
+      "learning_rate": 2.543278697488189e-06,
+      "loss": 0.6928,
+      "step": 41372
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.8677334785461426,
+      "learning_rate": 2.5430490637399072e-06,
+      "loss": 0.5051,
+      "step": 41373
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.544276237487793,
+      "learning_rate": 2.542819436823488e-06,
+      "loss": 0.6275,
+      "step": 41374
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.194160223007202,
+      "learning_rate": 2.5425898167395656e-06,
+      "loss": 0.6995,
+      "step": 41375
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.53067946434021,
+      "learning_rate": 2.542360203488782e-06,
+      "loss": 0.7978,
+      "step": 41376
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.7481234073638916,
+      "learning_rate": 2.5421305970717726e-06,
+      "loss": 0.6814,
+      "step": 41377
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.5111210346221924,
+      "learning_rate": 2.5419009974891783e-06,
+      "loss": 0.7253,
+      "step": 41378
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.18139386177063,
+      "learning_rate": 2.541671404741635e-06,
+      "loss": 0.6632,
+      "step": 41379
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.141284704208374,
+      "learning_rate": 2.541441818829785e-06,
+      "loss": 0.3352,
+      "step": 41380
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.8294405937194824,
+      "learning_rate": 2.5412122397542616e-06,
+      "loss": 0.7468,
+      "step": 41381
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.8391822576522827,
+      "learning_rate": 2.540982667515707e-06,
+      "loss": 0.5049,
+      "step": 41382
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.0398359298706055,
+      "learning_rate": 2.5407531021147592e-06,
+      "loss": 0.6501,
+      "step": 41383
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.6073522567749023,
+      "learning_rate": 2.5405235435520538e-06,
+      "loss": 0.67,
+      "step": 41384
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.3007984161376953,
+      "learning_rate": 2.540293991828232e-06,
+      "loss": 0.6845,
+      "step": 41385
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.6222941875457764,
+      "learning_rate": 2.5400644469439323e-06,
+      "loss": 0.6379,
+      "step": 41386
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.325162649154663,
+      "learning_rate": 2.53983490889979e-06,
+      "loss": 0.5693,
+      "step": 41387
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.968212842941284,
+      "learning_rate": 2.539605377696448e-06,
+      "loss": 0.7169,
+      "step": 41388
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.150209665298462,
+      "learning_rate": 2.5393758533345392e-06,
+      "loss": 0.6915,
+      "step": 41389
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.8009746074676514,
+      "learning_rate": 2.539146335814706e-06,
+      "loss": 0.5928,
+      "step": 41390
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.038546323776245,
+      "learning_rate": 2.538916825137584e-06,
+      "loss": 0.4891,
+      "step": 41391
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.3026676177978516,
+      "learning_rate": 2.5386873213038134e-06,
+      "loss": 0.7203,
+      "step": 41392
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.845593810081482,
+      "learning_rate": 2.5384578243140297e-06,
+      "loss": 0.5524,
+      "step": 41393
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.0483322143554688,
+      "learning_rate": 2.5382283341688744e-06,
+      "loss": 0.6887,
+      "step": 41394
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.9184151887893677,
+      "learning_rate": 2.5379988508689825e-06,
+      "loss": 0.5029,
+      "step": 41395
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.219391345977783,
+      "learning_rate": 2.537769374414995e-06,
+      "loss": 0.6043,
+      "step": 41396
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.454007625579834,
+      "learning_rate": 2.5375399048075466e-06,
+      "loss": 0.5978,
+      "step": 41397
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.91727614402771,
+      "learning_rate": 2.5373104420472795e-06,
+      "loss": 0.5801,
+      "step": 41398
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.8276960849761963,
+      "learning_rate": 2.5370809861348274e-06,
+      "loss": 0.8538,
+      "step": 41399
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.0555553436279297,
+      "learning_rate": 2.5368515370708324e-06,
+      "loss": 0.5703,
+      "step": 41400
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.837278127670288,
+      "learning_rate": 2.5366220948559284e-06,
+      "loss": 0.5724,
+      "step": 41401
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 4.7317962646484375,
+      "learning_rate": 2.5363926594907558e-06,
+      "loss": 0.5371,
+      "step": 41402
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.2848494052886963,
+      "learning_rate": 2.536163230975952e-06,
+      "loss": 0.6372,
+      "step": 41403
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.455214262008667,
+      "learning_rate": 2.5359338093121578e-06,
+      "loss": 0.5898,
+      "step": 41404
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.875307083129883,
+      "learning_rate": 2.535704394500006e-06,
+      "loss": 0.7969,
+      "step": 41405
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.791074514389038,
+      "learning_rate": 2.5354749865401396e-06,
+      "loss": 0.7233,
+      "step": 41406
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.1541483402252197,
+      "learning_rate": 2.535245585433192e-06,
+      "loss": 0.6742,
+      "step": 41407
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.192599296569824,
+      "learning_rate": 2.535016191179805e-06,
+      "loss": 0.5895,
+      "step": 41408
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.907514214515686,
+      "learning_rate": 2.5347868037806124e-06,
+      "loss": 0.465,
+      "step": 41409
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.9163718223571777,
+      "learning_rate": 2.534557423236257e-06,
+      "loss": 0.6991,
+      "step": 41410
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.5939342975616455,
+      "learning_rate": 2.534328049547371e-06,
+      "loss": 0.6498,
+      "step": 41411
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.0599935054779053,
+      "learning_rate": 2.534098682714597e-06,
+      "loss": 0.6424,
+      "step": 41412
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.1493475437164307,
+      "learning_rate": 2.533869322738569e-06,
+      "loss": 1.1079,
+      "step": 41413
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.3448662757873535,
+      "learning_rate": 2.533639969619929e-06,
+      "loss": 0.568,
+      "step": 41414
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.7787866592407227,
+      "learning_rate": 2.53341062335931e-06,
+      "loss": 0.6091,
+      "step": 41415
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.019749402999878,
+      "learning_rate": 2.533181283957353e-06,
+      "loss": 0.638,
+      "step": 41416
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.677910566329956,
+      "learning_rate": 2.5329519514146942e-06,
+      "loss": 0.8166,
+      "step": 41417
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.171670436859131,
+      "learning_rate": 2.532722625731974e-06,
+      "loss": 0.4074,
+      "step": 41418
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.9615321159362793,
+      "learning_rate": 2.5324933069098247e-06,
+      "loss": 0.9792,
+      "step": 41419
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.7657415866851807,
+      "learning_rate": 2.5322639949488903e-06,
+      "loss": 0.4466,
+      "step": 41420
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.081386089324951,
+      "learning_rate": 2.5320346898498028e-06,
+      "loss": 0.9277,
+      "step": 41421
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.3517632484436035,
+      "learning_rate": 2.5318053916132042e-06,
+      "loss": 0.8347,
+      "step": 41422
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 5.328925132751465,
+      "learning_rate": 2.5315761002397266e-06,
+      "loss": 0.6341,
+      "step": 41423
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.0799052715301514,
+      "learning_rate": 2.531346815730015e-06,
+      "loss": 0.5055,
+      "step": 41424
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 4.862311363220215,
+      "learning_rate": 2.5311175380847004e-06,
+      "loss": 0.5332,
+      "step": 41425
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.263871431350708,
+      "learning_rate": 2.5308882673044254e-06,
+      "loss": 0.5213,
+      "step": 41426
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.362088680267334,
+      "learning_rate": 2.530659003389824e-06,
+      "loss": 0.3673,
+      "step": 41427
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.546931743621826,
+      "learning_rate": 2.530429746341536e-06,
+      "loss": 0.5169,
+      "step": 41428
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.9262075424194336,
+      "learning_rate": 2.530200496160196e-06,
+      "loss": 0.5515,
+      "step": 41429
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.6675970554351807,
+      "learning_rate": 2.5299712528464453e-06,
+      "loss": 0.6734,
+      "step": 41430
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.5460665225982666,
+      "learning_rate": 2.5297420164009173e-06,
+      "loss": 0.4458,
+      "step": 41431
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 4.738644599914551,
+      "learning_rate": 2.529512786824253e-06,
+      "loss": 0.6836,
+      "step": 41432
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.111870527267456,
+      "learning_rate": 2.5292835641170864e-06,
+      "loss": 0.5014,
+      "step": 41433
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.8572592735290527,
+      "learning_rate": 2.5290543482800567e-06,
+      "loss": 0.4599,
+      "step": 41434
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.6326184272766113,
+      "learning_rate": 2.5288251393138038e-06,
+      "loss": 0.6123,
+      "step": 41435
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.978501796722412,
+      "learning_rate": 2.5285959372189595e-06,
+      "loss": 0.3975,
+      "step": 41436
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.8852438926696777,
+      "learning_rate": 2.5283667419961664e-06,
+      "loss": 0.6828,
+      "step": 41437
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.126675605773926,
+      "learning_rate": 2.528137553646058e-06,
+      "loss": 0.5158,
+      "step": 41438
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.405534029006958,
+      "learning_rate": 2.5279083721692743e-06,
+      "loss": 0.6394,
+      "step": 41439
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.665515661239624,
+      "learning_rate": 2.52767919756645e-06,
+      "loss": 0.4489,
+      "step": 41440
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.9053454399108887,
+      "learning_rate": 2.5274500298382253e-06,
+      "loss": 0.5783,
+      "step": 41441
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.517045021057129,
+      "learning_rate": 2.527220868985234e-06,
+      "loss": 0.4156,
+      "step": 41442
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.9530279636383057,
+      "learning_rate": 2.5269917150081152e-06,
+      "loss": 0.6404,
+      "step": 41443
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 8.179332733154297,
+      "learning_rate": 2.526762567907508e-06,
+      "loss": 0.446,
+      "step": 41444
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.793156623840332,
+      "learning_rate": 2.5265334276840457e-06,
+      "loss": 0.5117,
+      "step": 41445
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.1339361667633057,
+      "learning_rate": 2.526304294338367e-06,
+      "loss": 0.5931,
+      "step": 41446
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.915168046951294,
+      "learning_rate": 2.5260751678711116e-06,
+      "loss": 0.4636,
+      "step": 41447
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.6808388233184814,
+      "learning_rate": 2.5258460482829126e-06,
+      "loss": 0.8292,
+      "step": 41448
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.8997118473052979,
+      "learning_rate": 2.5256169355744104e-06,
+      "loss": 0.6686,
+      "step": 41449
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.8111331462860107,
+      "learning_rate": 2.5253878297462388e-06,
+      "loss": 0.4522,
+      "step": 41450
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.5761735439300537,
+      "learning_rate": 2.5251587307990388e-06,
+      "loss": 0.8574,
+      "step": 41451
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.2352797985076904,
+      "learning_rate": 2.524929638733443e-06,
+      "loss": 0.8069,
+      "step": 41452
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 4.404454708099365,
+      "learning_rate": 2.5247005535500928e-06,
+      "loss": 0.6303,
+      "step": 41453
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.4356868267059326,
+      "learning_rate": 2.524471475249621e-06,
+      "loss": 0.535,
+      "step": 41454
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.612640380859375,
+      "learning_rate": 2.524242403832669e-06,
+      "loss": 0.7644,
+      "step": 41455
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 4.031305313110352,
+      "learning_rate": 2.5240133392998697e-06,
+      "loss": 0.6371,
+      "step": 41456
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.3468286991119385,
+      "learning_rate": 2.5237842816518634e-06,
+      "loss": 0.5896,
+      "step": 41457
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.5911622047424316,
+      "learning_rate": 2.523555230889283e-06,
+      "loss": 0.5627,
+      "step": 41458
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.48930025100708,
+      "learning_rate": 2.52332618701277e-06,
+      "loss": 0.5528,
+      "step": 41459
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.9388999938964844,
+      "learning_rate": 2.523097150022957e-06,
+      "loss": 0.6651,
+      "step": 41460
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 1.5337580442428589,
+      "learning_rate": 2.522868119920485e-06,
+      "loss": 0.3139,
+      "step": 41461
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.584132194519043,
+      "learning_rate": 2.5226390967059877e-06,
+      "loss": 0.441,
+      "step": 41462
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.6417269706726074,
+      "learning_rate": 2.522410080380102e-06,
+      "loss": 0.6818,
+      "step": 41463
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.717938184738159,
+      "learning_rate": 2.5221810709434663e-06,
+      "loss": 0.6839,
+      "step": 41464
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.3040764331817627,
+      "learning_rate": 2.5219520683967184e-06,
+      "loss": 0.6818,
+      "step": 41465
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.620352268218994,
+      "learning_rate": 2.5217230727404916e-06,
+      "loss": 0.3594,
+      "step": 41466
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 4.124312400817871,
+      "learning_rate": 2.521494083975427e-06,
+      "loss": 0.8343,
+      "step": 41467
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.9921255111694336,
+      "learning_rate": 2.521265102102156e-06,
+      "loss": 0.5732,
+      "step": 41468
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.7943363189697266,
+      "learning_rate": 2.5210361271213213e-06,
+      "loss": 0.5257,
+      "step": 41469
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.470452308654785,
+      "learning_rate": 2.520807159033554e-06,
+      "loss": 0.6719,
+      "step": 41470
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.135434627532959,
+      "learning_rate": 2.520578197839495e-06,
+      "loss": 0.637,
+      "step": 41471
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.8214128017425537,
+      "learning_rate": 2.5203492435397775e-06,
+      "loss": 0.4621,
+      "step": 41472
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.492912530899048,
+      "learning_rate": 2.5201202961350414e-06,
+      "loss": 0.6899,
+      "step": 41473
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.2542014122009277,
+      "learning_rate": 2.5198913556259204e-06,
+      "loss": 0.8512,
+      "step": 41474
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.1456661224365234,
+      "learning_rate": 2.519662422013054e-06,
+      "loss": 0.6275,
+      "step": 41475
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.112947940826416,
+      "learning_rate": 2.5194334952970757e-06,
+      "loss": 0.7059,
+      "step": 41476
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.932105541229248,
+      "learning_rate": 2.5192045754786254e-06,
+      "loss": 0.5768,
+      "step": 41477
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.364474296569824,
+      "learning_rate": 2.5189756625583357e-06,
+      "loss": 0.6953,
+      "step": 41478
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.1218583583831787,
+      "learning_rate": 2.5187467565368473e-06,
+      "loss": 0.3364,
+      "step": 41479
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 1.8514070510864258,
+      "learning_rate": 2.518517857414793e-06,
+      "loss": 0.5301,
+      "step": 41480
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.1090331077575684,
+      "learning_rate": 2.518288965192812e-06,
+      "loss": 0.5458,
+      "step": 41481
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.1598622798919678,
+      "learning_rate": 2.5180600798715387e-06,
+      "loss": 0.5951,
+      "step": 41482
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.342466115951538,
+      "learning_rate": 2.517831201451613e-06,
+      "loss": 0.5621,
+      "step": 41483
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.6142966747283936,
+      "learning_rate": 2.517602329933664e-06,
+      "loss": 0.7313,
+      "step": 41484
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.2345073223114014,
+      "learning_rate": 2.517373465318337e-06,
+      "loss": 0.732,
+      "step": 41485
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 1.969719648361206,
+      "learning_rate": 2.5171446076062623e-06,
+      "loss": 0.6351,
+      "step": 41486
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 1.8676114082336426,
+      "learning_rate": 2.5169157567980808e-06,
+      "loss": 0.4478,
+      "step": 41487
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 4.914341449737549,
+      "learning_rate": 2.516686912894424e-06,
+      "loss": 0.5091,
+      "step": 41488
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.351327657699585,
+      "learning_rate": 2.516458075895933e-06,
+      "loss": 0.8081,
+      "step": 41489
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 4.8235764503479,
+      "learning_rate": 2.516229245803239e-06,
+      "loss": 0.6296,
+      "step": 41490
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.1208338737487793,
+      "learning_rate": 2.5160004226169836e-06,
+      "loss": 0.6016,
+      "step": 41491
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.648115873336792,
+      "learning_rate": 2.515771606337799e-06,
+      "loss": 0.7062,
+      "step": 41492
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.0009377002716064,
+      "learning_rate": 2.5155427969663237e-06,
+      "loss": 0.7473,
+      "step": 41493
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.188852310180664,
+      "learning_rate": 2.5153139945031922e-06,
+      "loss": 0.4694,
+      "step": 41494
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.446235179901123,
+      "learning_rate": 2.5150851989490434e-06,
+      "loss": 0.5741,
+      "step": 41495
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 4.1633429527282715,
+      "learning_rate": 2.51485641030451e-06,
+      "loss": 0.9547,
+      "step": 41496
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.623889207839966,
+      "learning_rate": 2.5146276285702297e-06,
+      "loss": 0.3755,
+      "step": 41497
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.4148447513580322,
+      "learning_rate": 2.5143988537468412e-06,
+      "loss": 0.639,
+      "step": 41498
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.338325262069702,
+      "learning_rate": 2.514170085834976e-06,
+      "loss": 0.5044,
+      "step": 41499
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.6010868549346924,
+      "learning_rate": 2.5139413248352746e-06,
+      "loss": 0.4374,
+      "step": 41500
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 1.781674861907959,
+      "learning_rate": 2.513712570748369e-06,
+      "loss": 0.6845,
+      "step": 41501
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.1436829566955566,
+      "learning_rate": 2.513483823574899e-06,
+      "loss": 0.6219,
+      "step": 41502
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.1828222274780273,
+      "learning_rate": 2.5132550833154977e-06,
+      "loss": 0.7247,
+      "step": 41503
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.0603911876678467,
+      "learning_rate": 2.5130263499708028e-06,
+      "loss": 0.6959,
+      "step": 41504
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 1.9792574644088745,
+      "learning_rate": 2.5127976235414513e-06,
+      "loss": 0.58,
+      "step": 41505
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.7375478744506836,
+      "learning_rate": 2.5125689040280763e-06,
+      "loss": 0.4829,
+      "step": 41506
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 1.9326971769332886,
+      "learning_rate": 2.512340191431317e-06,
+      "loss": 0.3405,
+      "step": 41507
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 4.047983169555664,
+      "learning_rate": 2.512111485751806e-06,
+      "loss": 0.6909,
+      "step": 41508
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.7107324600219727,
+      "learning_rate": 2.5118827869901807e-06,
+      "loss": 0.8234,
+      "step": 41509
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.1003177165985107,
+      "learning_rate": 2.5116540951470797e-06,
+      "loss": 0.819,
+      "step": 41510
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.6298232078552246,
+      "learning_rate": 2.5114254102231337e-06,
+      "loss": 0.4722,
+      "step": 41511
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.055230140686035,
+      "learning_rate": 2.5111967322189845e-06,
+      "loss": 0.7333,
+      "step": 41512
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.5260748863220215,
+      "learning_rate": 2.510968061135262e-06,
+      "loss": 0.6394,
+      "step": 41513
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.879490613937378,
+      "learning_rate": 2.5107393969726074e-06,
+      "loss": 0.524,
+      "step": 41514
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.5398991107940674,
+      "learning_rate": 2.510510739731652e-06,
+      "loss": 0.7483,
+      "step": 41515
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.517260789871216,
+      "learning_rate": 2.5102820894130354e-06,
+      "loss": 0.6859,
+      "step": 41516
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.185455322265625,
+      "learning_rate": 2.51005344601739e-06,
+      "loss": 0.4077,
+      "step": 41517
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.3975515365600586,
+      "learning_rate": 2.5098248095453547e-06,
+      "loss": 0.5656,
+      "step": 41518
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.541266441345215,
+      "learning_rate": 2.509596179997562e-06,
+      "loss": 0.6114,
+      "step": 41519
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.4530532360076904,
+      "learning_rate": 2.509367557374651e-06,
+      "loss": 0.5262,
+      "step": 41520
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 1.9495757818222046,
+      "learning_rate": 2.509138941677255e-06,
+      "loss": 0.8063,
+      "step": 41521
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.401496648788452,
+      "learning_rate": 2.5089103329060115e-06,
+      "loss": 0.678,
+      "step": 41522
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.5575947761535645,
+      "learning_rate": 2.508681731061553e-06,
+      "loss": 0.6243,
+      "step": 41523
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.6681506633758545,
+      "learning_rate": 2.508453136144518e-06,
+      "loss": 0.6746,
+      "step": 41524
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.031630039215088,
+      "learning_rate": 2.508224548155541e-06,
+      "loss": 0.7452,
+      "step": 41525
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.735233783721924,
+      "learning_rate": 2.50799596709526e-06,
+      "loss": 0.6445,
+      "step": 41526
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 5.196661472320557,
+      "learning_rate": 2.5077673929643074e-06,
+      "loss": 0.8766,
+      "step": 41527
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.16087007522583,
+      "learning_rate": 2.5075388257633214e-06,
+      "loss": 0.624,
+      "step": 41528
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.314392328262329,
+      "learning_rate": 2.507310265492935e-06,
+      "loss": 0.4708,
+      "step": 41529
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.680466413497925,
+      "learning_rate": 2.507081712153786e-06,
+      "loss": 0.5406,
+      "step": 41530
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 1.9806569814682007,
+      "learning_rate": 2.506853165746508e-06,
+      "loss": 0.6085,
+      "step": 41531
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.4663515090942383,
+      "learning_rate": 2.506624626271739e-06,
+      "loss": 0.7393,
+      "step": 41532
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.52354097366333,
+      "learning_rate": 2.50639609373011e-06,
+      "loss": 0.4854,
+      "step": 41533
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.927603006362915,
+      "learning_rate": 2.506167568122262e-06,
+      "loss": 0.538,
+      "step": 41534
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.3698229789733887,
+      "learning_rate": 2.5059390494488255e-06,
+      "loss": 0.7023,
+      "step": 41535
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.8567750453948975,
+      "learning_rate": 2.505710537710441e-06,
+      "loss": 0.6237,
+      "step": 41536
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.7133853435516357,
+      "learning_rate": 2.5054820329077386e-06,
+      "loss": 0.8067,
+      "step": 41537
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.4901771545410156,
+      "learning_rate": 2.5052535350413577e-06,
+      "loss": 0.654,
+      "step": 41538
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 1.637156367301941,
+      "learning_rate": 2.5050250441119307e-06,
+      "loss": 0.5931,
+      "step": 41539
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 5.034344673156738,
+      "learning_rate": 2.5047965601200965e-06,
+      "loss": 0.4566,
+      "step": 41540
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.1479363441467285,
+      "learning_rate": 2.504568083066486e-06,
+      "loss": 0.6682,
+      "step": 41541
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.2857344150543213,
+      "learning_rate": 2.5043396129517394e-06,
+      "loss": 0.7511,
+      "step": 41542
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.7778420448303223,
+      "learning_rate": 2.504111149776487e-06,
+      "loss": 0.5632,
+      "step": 41543
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.4883086681365967,
+      "learning_rate": 2.5038826935413685e-06,
+      "loss": 0.6615,
+      "step": 41544
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.9066410064697266,
+      "learning_rate": 2.5036542442470133e-06,
+      "loss": 0.6381,
+      "step": 41545
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.801379680633545,
+      "learning_rate": 2.5034258018940644e-06,
+      "loss": 0.6745,
+      "step": 41546
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 1.8820219039916992,
+      "learning_rate": 2.503197366483151e-06,
+      "loss": 0.5877,
+      "step": 41547
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.1656494140625,
+      "learning_rate": 2.5029689380149126e-06,
+      "loss": 0.6713,
+      "step": 41548
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.6454687118530273,
+      "learning_rate": 2.50274051648998e-06,
+      "loss": 0.5823,
+      "step": 41549
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.3386237621307373,
+      "learning_rate": 2.5025121019089925e-06,
+      "loss": 0.6458,
+      "step": 41550
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.0172016620635986,
+      "learning_rate": 2.5022836942725817e-06,
+      "loss": 0.4823,
+      "step": 41551
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.6121349334716797,
+      "learning_rate": 2.5020552935813857e-06,
+      "loss": 0.5655,
+      "step": 41552
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.7755508422851562,
+      "learning_rate": 2.501826899836037e-06,
+      "loss": 0.7622,
+      "step": 41553
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 1.933282732963562,
+      "learning_rate": 2.501598513037173e-06,
+      "loss": 0.5862,
+      "step": 41554
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.6781506538391113,
+      "learning_rate": 2.5013701331854255e-06,
+      "loss": 0.5508,
+      "step": 41555
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.4879937171936035,
+      "learning_rate": 2.5011417602814347e-06,
+      "loss": 0.8049,
+      "step": 41556
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.652827739715576,
+      "learning_rate": 2.50091339432583e-06,
+      "loss": 0.747,
+      "step": 41557
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 1.7832170724868774,
+      "learning_rate": 2.500685035319249e-06,
+      "loss": 0.5578,
+      "step": 41558
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.204944133758545,
+      "learning_rate": 2.5004566832623284e-06,
+      "loss": 0.7788,
+      "step": 41559
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.2939891815185547,
+      "learning_rate": 2.500228338155699e-06,
+      "loss": 0.6801,
+      "step": 41560
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.4932548999786377,
+      "learning_rate": 2.5000000000000015e-06,
+      "loss": 0.6926,
+      "step": 41561
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.872697114944458,
+      "learning_rate": 2.4997716687958644e-06,
+      "loss": 0.4575,
+      "step": 41562
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.166114568710327,
+      "learning_rate": 2.499543344543928e-06,
+      "loss": 0.4868,
+      "step": 41563
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.401109218597412,
+      "learning_rate": 2.499315027244823e-06,
+      "loss": 0.5145,
+      "step": 41564
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.224959373474121,
+      "learning_rate": 2.499086716899186e-06,
+      "loss": 0.5249,
+      "step": 41565
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 4.110263347625732,
+      "learning_rate": 2.4988584135076537e-06,
+      "loss": 0.7427,
+      "step": 41566
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 1.7180585861206055,
+      "learning_rate": 2.4986301170708577e-06,
+      "loss": 0.64,
+      "step": 41567
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.3278791904449463,
+      "learning_rate": 2.4984018275894363e-06,
+      "loss": 0.891,
+      "step": 41568
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.0690054893493652,
+      "learning_rate": 2.4981735450640197e-06,
+      "loss": 0.6294,
+      "step": 41569
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.448410749435425,
+      "learning_rate": 2.4979452694952462e-06,
+      "loss": 0.7145,
+      "step": 41570
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.3731915950775146,
+      "learning_rate": 2.497717000883751e-06,
+      "loss": 0.5677,
+      "step": 41571
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.3071799278259277,
+      "learning_rate": 2.4974887392301657e-06,
+      "loss": 0.5949,
+      "step": 41572
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 1.647376298904419,
+      "learning_rate": 2.4972604845351283e-06,
+      "loss": 0.5177,
+      "step": 41573
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 4.829641819000244,
+      "learning_rate": 2.49703223679927e-06,
+      "loss": 0.8502,
+      "step": 41574
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.0834860801696777,
+      "learning_rate": 2.49680399602323e-06,
+      "loss": 0.4272,
+      "step": 41575
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.140056848526001,
+      "learning_rate": 2.4965757622076373e-06,
+      "loss": 0.6452,
+      "step": 41576
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.9687469005584717,
+      "learning_rate": 2.4963475353531326e-06,
+      "loss": 0.7392,
+      "step": 41577
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.3085296154022217,
+      "learning_rate": 2.4961193154603455e-06,
+      "loss": 0.7048,
+      "step": 41578
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.1492650508880615,
+      "learning_rate": 2.495891102529914e-06,
+      "loss": 0.594,
+      "step": 41579
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 1.9731734991073608,
+      "learning_rate": 2.4956628965624694e-06,
+      "loss": 0.4586,
+      "step": 41580
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 1.920145034790039,
+      "learning_rate": 2.4954346975586507e-06,
+      "loss": 0.4685,
+      "step": 41581
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.456014394760132,
+      "learning_rate": 2.4952065055190877e-06,
+      "loss": 0.853,
+      "step": 41582
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.4179623126983643,
+      "learning_rate": 2.494978320444419e-06,
+      "loss": 0.665,
+      "step": 41583
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.0890274047851562,
+      "learning_rate": 2.4947501423352747e-06,
+      "loss": 0.5593,
+      "step": 41584
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.443347215652466,
+      "learning_rate": 2.4945219711922923e-06,
+      "loss": 0.6165,
+      "step": 41585
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.3187813758850098,
+      "learning_rate": 2.4942938070161056e-06,
+      "loss": 0.8478,
+      "step": 41586
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.586660861968994,
+      "learning_rate": 2.4940656498073513e-06,
+      "loss": 0.7325,
+      "step": 41587
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.161050796508789,
+      "learning_rate": 2.493837499566659e-06,
+      "loss": 0.5905,
+      "step": 41588
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 1.7457560300827026,
+      "learning_rate": 2.493609356294669e-06,
+      "loss": 0.3242,
+      "step": 41589
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.045283794403076,
+      "learning_rate": 2.4933812199920093e-06,
+      "loss": 0.6105,
+      "step": 41590
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.3527657985687256,
+      "learning_rate": 2.4931530906593203e-06,
+      "loss": 0.8247,
+      "step": 41591
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.6813483238220215,
+      "learning_rate": 2.4929249682972313e-06,
+      "loss": 0.4595,
+      "step": 41592
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.397392988204956,
+      "learning_rate": 2.4926968529063798e-06,
+      "loss": 0.4769,
+      "step": 41593
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.9642014503479004,
+      "learning_rate": 2.4924687444873984e-06,
+      "loss": 0.5191,
+      "step": 41594
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 6.824268341064453,
+      "learning_rate": 2.4922406430409233e-06,
+      "loss": 0.8314,
+      "step": 41595
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.9994194507598877,
+      "learning_rate": 2.492012548567586e-06,
+      "loss": 0.6563,
+      "step": 41596
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.07236647605896,
+      "learning_rate": 2.4917844610680242e-06,
+      "loss": 0.5857,
+      "step": 41597
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.467942953109741,
+      "learning_rate": 2.491556380542868e-06,
+      "loss": 0.714,
+      "step": 41598
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.8476197719573975,
+      "learning_rate": 2.491328306992756e-06,
+      "loss": 0.4667,
+      "step": 41599
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.423060417175293,
+      "learning_rate": 2.4911002404183184e-06,
+      "loss": 0.7232,
+      "step": 41600
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.3035051822662354,
+      "learning_rate": 2.4908721808201926e-06,
+      "loss": 0.7523,
+      "step": 41601
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.6596148014068604,
+      "learning_rate": 2.4906441281990095e-06,
+      "loss": 0.5499,
+      "step": 41602
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.159508466720581,
+      "learning_rate": 2.4904160825554067e-06,
+      "loss": 0.5057,
+      "step": 41603
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.0104873180389404,
+      "learning_rate": 2.4901880438900155e-06,
+      "loss": 0.6931,
+      "step": 41604
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.9569954872131348,
+      "learning_rate": 2.4899600122034726e-06,
+      "loss": 0.5791,
+      "step": 41605
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.0602498054504395,
+      "learning_rate": 2.489731987496407e-06,
+      "loss": 0.4552,
+      "step": 41606
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.985413074493408,
+      "learning_rate": 2.48950396976946e-06,
+      "loss": 0.7478,
+      "step": 41607
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.664036989212036,
+      "learning_rate": 2.4892759590232605e-06,
+      "loss": 0.579,
+      "step": 41608
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.472703218460083,
+      "learning_rate": 2.4890479552584456e-06,
+      "loss": 0.749,
+      "step": 41609
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 4.478497505187988,
+      "learning_rate": 2.4888199584756462e-06,
+      "loss": 0.6323,
+      "step": 41610
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.0488600730895996,
+      "learning_rate": 2.4885919686754993e-06,
+      "loss": 0.8207,
+      "step": 41611
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.296147346496582,
+      "learning_rate": 2.4883639858586355e-06,
+      "loss": 0.7778,
+      "step": 41612
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 8.498492240905762,
+      "learning_rate": 2.4881360100256925e-06,
+      "loss": 0.6882,
+      "step": 41613
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.3872649669647217,
+      "learning_rate": 2.4879080411773003e-06,
+      "loss": 0.6542,
+      "step": 41614
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 4.821911334991455,
+      "learning_rate": 2.4876800793140965e-06,
+      "loss": 0.8586,
+      "step": 41615
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.470616340637207,
+      "learning_rate": 2.487452124436712e-06,
+      "loss": 0.6158,
+      "step": 41616
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.8339171409606934,
+      "learning_rate": 2.4872241765457833e-06,
+      "loss": 0.496,
+      "step": 41617
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.1609444618225098,
+      "learning_rate": 2.486996235641941e-06,
+      "loss": 0.4604,
+      "step": 41618
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 4.004970550537109,
+      "learning_rate": 2.486768301725821e-06,
+      "loss": 0.9783,
+      "step": 41619
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.402353048324585,
+      "learning_rate": 2.4865403747980587e-06,
+      "loss": 0.5745,
+      "step": 41620
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.6532180309295654,
+      "learning_rate": 2.486312454859284e-06,
+      "loss": 0.6368,
+      "step": 41621
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.3313164710998535,
+      "learning_rate": 2.4860845419101347e-06,
+      "loss": 0.5605,
+      "step": 41622
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 5.481361389160156,
+      "learning_rate": 2.4858566359512406e-06,
+      "loss": 0.6405,
+      "step": 41623
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.2269351482391357,
+      "learning_rate": 2.4856287369832393e-06,
+      "loss": 0.7027,
+      "step": 41624
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.2231125831604004,
+      "learning_rate": 2.4854008450067605e-06,
+      "loss": 0.527,
+      "step": 41625
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.6029539108276367,
+      "learning_rate": 2.4851729600224406e-06,
+      "loss": 0.6124,
+      "step": 41626
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.5950422286987305,
+      "learning_rate": 2.484945082030914e-06,
+      "loss": 0.6972,
+      "step": 41627
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.070467948913574,
+      "learning_rate": 2.484717211032811e-06,
+      "loss": 0.8693,
+      "step": 41628
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.901916265487671,
+      "learning_rate": 2.4844893470287696e-06,
+      "loss": 0.7607,
+      "step": 41629
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.300804853439331,
+      "learning_rate": 2.4842614900194186e-06,
+      "loss": 0.5796,
+      "step": 41630
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.367842674255371,
+      "learning_rate": 2.484033640005394e-06,
+      "loss": 0.5767,
+      "step": 41631
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.909067392349243,
+      "learning_rate": 2.4838057969873314e-06,
+      "loss": 0.5686,
+      "step": 41632
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.9053099155426025,
+      "learning_rate": 2.4835779609658606e-06,
+      "loss": 0.3876,
+      "step": 41633
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.800783157348633,
+      "learning_rate": 2.483350131941618e-06,
+      "loss": 0.6012,
+      "step": 41634
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.289243221282959,
+      "learning_rate": 2.4831223099152347e-06,
+      "loss": 0.7007,
+      "step": 41635
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.1555051803588867,
+      "learning_rate": 2.482894494887347e-06,
+      "loss": 0.6342,
+      "step": 41636
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 4.039602756500244,
+      "learning_rate": 2.4826666868585852e-06,
+      "loss": 0.7201,
+      "step": 41637
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.4083309173583984,
+      "learning_rate": 2.4824388858295863e-06,
+      "loss": 0.4932,
+      "step": 41638
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.256871461868286,
+      "learning_rate": 2.4822110918009796e-06,
+      "loss": 0.7065,
+      "step": 41639
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.4158830642700195,
+      "learning_rate": 2.481983304773402e-06,
+      "loss": 0.7017,
+      "step": 41640
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.156020164489746,
+      "learning_rate": 2.4817555247474847e-06,
+      "loss": 0.7637,
+      "step": 41641
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.459763526916504,
+      "learning_rate": 2.481527751723863e-06,
+      "loss": 0.606,
+      "step": 41642
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.196084976196289,
+      "learning_rate": 2.481299985703168e-06,
+      "loss": 0.6051,
+      "step": 41643
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.3252580165863037,
+      "learning_rate": 2.4810722266860354e-06,
+      "loss": 0.7317,
+      "step": 41644
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.290754556655884,
+      "learning_rate": 2.480844474673096e-06,
+      "loss": 0.4498,
+      "step": 41645
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.447193145751953,
+      "learning_rate": 2.480616729664984e-06,
+      "loss": 0.5333,
+      "step": 41646
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.5578300952911377,
+      "learning_rate": 2.4803889916623336e-06,
+      "loss": 0.684,
+      "step": 41647
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.364968776702881,
+      "learning_rate": 2.480161260665779e-06,
+      "loss": 0.7124,
+      "step": 41648
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.572490930557251,
+      "learning_rate": 2.4799335366759504e-06,
+      "loss": 0.6492,
+      "step": 41649
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.5515592098236084,
+      "learning_rate": 2.4797058196934843e-06,
+      "loss": 0.6647,
+      "step": 41650
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 4.482801914215088,
+      "learning_rate": 2.47947810971901e-06,
+      "loss": 0.6386,
+      "step": 41651
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 1.8691368103027344,
+      "learning_rate": 2.479250406753165e-06,
+      "loss": 0.4122,
+      "step": 41652
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.2950851917266846,
+      "learning_rate": 2.4790227107965785e-06,
+      "loss": 0.6953,
+      "step": 41653
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.005837917327881,
+      "learning_rate": 2.4787950218498874e-06,
+      "loss": 0.5317,
+      "step": 41654
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.26182222366333,
+      "learning_rate": 2.478567339913721e-06,
+      "loss": 0.7652,
+      "step": 41655
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 1.7528162002563477,
+      "learning_rate": 2.478339664988716e-06,
+      "loss": 0.4305,
+      "step": 41656
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.797340154647827,
+      "learning_rate": 2.478111997075502e-06,
+      "loss": 0.7898,
+      "step": 41657
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.2966997623443604,
+      "learning_rate": 2.4778843361747157e-06,
+      "loss": 0.3979,
+      "step": 41658
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 4.079535484313965,
+      "learning_rate": 2.4776566822869864e-06,
+      "loss": 0.6772,
+      "step": 41659
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.690992593765259,
+      "learning_rate": 2.477429035412951e-06,
+      "loss": 0.5688,
+      "step": 41660
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.1299827098846436,
+      "learning_rate": 2.477201395553238e-06,
+      "loss": 0.3293,
+      "step": 41661
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.135934591293335,
+      "learning_rate": 2.476973762708485e-06,
+      "loss": 0.6397,
+      "step": 41662
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.436737537384033,
+      "learning_rate": 2.476746136879321e-06,
+      "loss": 0.5922,
+      "step": 41663
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.760255813598633,
+      "learning_rate": 2.476518518066383e-06,
+      "loss": 0.5708,
+      "step": 41664
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.735624074935913,
+      "learning_rate": 2.4762909062702994e-06,
+      "loss": 0.4695,
+      "step": 41665
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.3159725666046143,
+      "learning_rate": 2.4760633014917074e-06,
+      "loss": 0.6036,
+      "step": 41666
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.2589271068573,
+      "learning_rate": 2.4758357037312343e-06,
+      "loss": 0.6359,
+      "step": 41667
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.8902792930603027,
+      "learning_rate": 2.47560811298952e-06,
+      "loss": 0.4834,
+      "step": 41668
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.6887876987457275,
+      "learning_rate": 2.475380529267192e-06,
+      "loss": 0.5341,
+      "step": 41669
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.1255276203155518,
+      "learning_rate": 2.475152952564887e-06,
+      "loss": 0.744,
+      "step": 41670
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.2331771850585938,
+      "learning_rate": 2.4749253828832336e-06,
+      "loss": 0.5747,
+      "step": 41671
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.307591438293457,
+      "learning_rate": 2.4746978202228695e-06,
+      "loss": 0.5579,
+      "step": 41672
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 4.548707962036133,
+      "learning_rate": 2.474470264584422e-06,
+      "loss": 0.6331,
+      "step": 41673
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 1.8935627937316895,
+      "learning_rate": 2.474242715968529e-06,
+      "loss": 0.5185,
+      "step": 41674
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.355952739715576,
+      "learning_rate": 2.474015174375819e-06,
+      "loss": 0.4801,
+      "step": 41675
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.5909883975982666,
+      "learning_rate": 2.4737876398069277e-06,
+      "loss": 0.5734,
+      "step": 41676
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.2251524925231934,
+      "learning_rate": 2.4735601122624854e-06,
+      "loss": 0.4634,
+      "step": 41677
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.1302566528320312,
+      "learning_rate": 2.4733325917431277e-06,
+      "loss": 0.6879,
+      "step": 41678
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.9304590225219727,
+      "learning_rate": 2.4731050782494835e-06,
+      "loss": 0.6834,
+      "step": 41679
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.4082016944885254,
+      "learning_rate": 2.47287757178219e-06,
+      "loss": 0.6275,
+      "step": 41680
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.1121103763580322,
+      "learning_rate": 2.472650072341875e-06,
+      "loss": 0.6707,
+      "step": 41681
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 1.7415941953659058,
+      "learning_rate": 2.4724225799291736e-06,
+      "loss": 0.425,
+      "step": 41682
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.1836304664611816,
+      "learning_rate": 2.4721950945447197e-06,
+      "loss": 0.6586,
+      "step": 41683
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.4868123531341553,
+      "learning_rate": 2.4719676161891425e-06,
+      "loss": 0.7046,
+      "step": 41684
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.2761363983154297,
+      "learning_rate": 2.4717401448630786e-06,
+      "loss": 0.5937,
+      "step": 41685
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.39792799949646,
+      "learning_rate": 2.4715126805671557e-06,
+      "loss": 0.5783,
+      "step": 41686
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 1.6722168922424316,
+      "learning_rate": 2.471285223302009e-06,
+      "loss": 0.4782,
+      "step": 41687
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 1.9961826801300049,
+      "learning_rate": 2.471057773068273e-06,
+      "loss": 0.6423,
+      "step": 41688
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.050142288208008,
+      "learning_rate": 2.470830329866576e-06,
+      "loss": 0.7483,
+      "step": 41689
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.189514636993408,
+      "learning_rate": 2.470602893697554e-06,
+      "loss": 0.5034,
+      "step": 41690
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.2411701679229736,
+      "learning_rate": 2.470375464561836e-06,
+      "loss": 0.7242,
+      "step": 41691
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.0235657691955566,
+      "learning_rate": 2.4701480424600586e-06,
+      "loss": 0.4932,
+      "step": 41692
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.7947773933410645,
+      "learning_rate": 2.4699206273928495e-06,
+      "loss": 0.4601,
+      "step": 41693
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.5532407760620117,
+      "learning_rate": 2.4696932193608436e-06,
+      "loss": 0.5342,
+      "step": 41694
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.0165023803710938,
+      "learning_rate": 2.4694658183646745e-06,
+      "loss": 0.5022,
+      "step": 41695
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.745286464691162,
+      "learning_rate": 2.469238424404971e-06,
+      "loss": 0.6867,
+      "step": 41696
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.50982928276062,
+      "learning_rate": 2.46901103748237e-06,
+      "loss": 0.4041,
+      "step": 41697
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.0692026615142822,
+      "learning_rate": 2.468783657597499e-06,
+      "loss": 0.4609,
+      "step": 41698
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 4.028526306152344,
+      "learning_rate": 2.468556284750994e-06,
+      "loss": 0.5941,
+      "step": 41699
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 6.211596488952637,
+      "learning_rate": 2.468328918943484e-06,
+      "loss": 0.7176,
+      "step": 41700
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.665637731552124,
+      "learning_rate": 2.4681015601756046e-06,
+      "loss": 0.799,
+      "step": 41701
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 1.8597698211669922,
+      "learning_rate": 2.4678742084479844e-06,
+      "loss": 0.5117,
+      "step": 41702
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.0077829360961914,
+      "learning_rate": 2.46764686376126e-06,
+      "loss": 0.5822,
+      "step": 41703
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.132108211517334,
+      "learning_rate": 2.467419526116058e-06,
+      "loss": 0.7342,
+      "step": 41704
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.976372003555298,
+      "learning_rate": 2.467192195513017e-06,
+      "loss": 0.6565,
+      "step": 41705
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.628896474838257,
+      "learning_rate": 2.466964871952763e-06,
+      "loss": 0.6484,
+      "step": 41706
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.1430726051330566,
+      "learning_rate": 2.46673755543593e-06,
+      "loss": 0.6674,
+      "step": 41707
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.0630943775177,
+      "learning_rate": 2.4665102459631517e-06,
+      "loss": 0.5521,
+      "step": 41708
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.4661309719085693,
+      "learning_rate": 2.466282943535061e-06,
+      "loss": 0.633,
+      "step": 41709
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 1.9116557836532593,
+      "learning_rate": 2.466055648152287e-06,
+      "loss": 0.307,
+      "step": 41710
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.613295316696167,
+      "learning_rate": 2.4658283598154643e-06,
+      "loss": 0.5266,
+      "step": 41711
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.4598965644836426,
+      "learning_rate": 2.4656010785252215e-06,
+      "loss": 0.6074,
+      "step": 41712
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.4348676204681396,
+      "learning_rate": 2.465373804282195e-06,
+      "loss": 0.5657,
+      "step": 41713
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.2902936935424805,
+      "learning_rate": 2.4651465370870124e-06,
+      "loss": 0.6033,
+      "step": 41714
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 4.396202087402344,
+      "learning_rate": 2.4649192769403094e-06,
+      "loss": 0.6725,
+      "step": 41715
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 1.9481123685836792,
+      "learning_rate": 2.464692023842715e-06,
+      "loss": 0.5635,
+      "step": 41716
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.82681941986084,
+      "learning_rate": 2.4644647777948637e-06,
+      "loss": 0.5371,
+      "step": 41717
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.575639486312866,
+      "learning_rate": 2.464237538797384e-06,
+      "loss": 0.7513,
+      "step": 41718
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.0051372051239014,
+      "learning_rate": 2.4640103068509123e-06,
+      "loss": 0.3041,
+      "step": 41719
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.377781629562378,
+      "learning_rate": 2.463783081956076e-06,
+      "loss": 0.7022,
+      "step": 41720
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 1.645708441734314,
+      "learning_rate": 2.4635558641135105e-06,
+      "loss": 0.5739,
+      "step": 41721
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.1760969161987305,
+      "learning_rate": 2.463328653323844e-06,
+      "loss": 0.6169,
+      "step": 41722
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.7480413913726807,
+      "learning_rate": 2.463101449587712e-06,
+      "loss": 0.5523,
+      "step": 41723
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.24131178855896,
+      "learning_rate": 2.4628742529057424e-06,
+      "loss": 0.8053,
+      "step": 41724
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.312500476837158,
+      "learning_rate": 2.4626470632785714e-06,
+      "loss": 0.3392,
+      "step": 41725
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.0344135761260986,
+      "learning_rate": 2.4624198807068266e-06,
+      "loss": 0.508,
+      "step": 41726
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.7240989208221436,
+      "learning_rate": 2.462192705191143e-06,
+      "loss": 0.5436,
+      "step": 41727
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.3034474849700928,
+      "learning_rate": 2.4619655367321476e-06,
+      "loss": 0.5405,
+      "step": 41728
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.5388293266296387,
+      "learning_rate": 2.4617383753304793e-06,
+      "loss": 0.5159,
+      "step": 41729
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 1.8454515933990479,
+      "learning_rate": 2.461511220986763e-06,
+      "loss": 0.5465,
+      "step": 41730
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.234727144241333,
+      "learning_rate": 2.4612840737016353e-06,
+      "loss": 0.5409,
+      "step": 41731
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.3290040493011475,
+      "learning_rate": 2.4610569334757233e-06,
+      "loss": 0.7819,
+      "step": 41732
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.244370222091675,
+      "learning_rate": 2.4608298003096633e-06,
+      "loss": 0.5449,
+      "step": 41733
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.477036714553833,
+      "learning_rate": 2.4606026742040824e-06,
+      "loss": 0.7984,
+      "step": 41734
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.9834702014923096,
+      "learning_rate": 2.460375555159616e-06,
+      "loss": 0.582,
+      "step": 41735
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.1618995666503906,
+      "learning_rate": 2.460148443176892e-06,
+      "loss": 0.4671,
+      "step": 41736
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 1.7656726837158203,
+      "learning_rate": 2.459921338256545e-06,
+      "loss": 0.526,
+      "step": 41737
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.281421661376953,
+      "learning_rate": 2.4596942403992043e-06,
+      "loss": 0.7371,
+      "step": 41738
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.880958080291748,
+      "learning_rate": 2.4594671496055033e-06,
+      "loss": 0.6119,
+      "step": 41739
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.820204734802246,
+      "learning_rate": 2.4592400658760714e-06,
+      "loss": 0.5547,
+      "step": 41740
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.368302583694458,
+      "learning_rate": 2.4590129892115423e-06,
+      "loss": 0.5766,
+      "step": 41741
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.187429189682007,
+      "learning_rate": 2.4587859196125445e-06,
+      "loss": 0.574,
+      "step": 41742
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.8568501472473145,
+      "learning_rate": 2.4585588570797115e-06,
+      "loss": 0.6215,
+      "step": 41743
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.065993547439575,
+      "learning_rate": 2.4583318016136753e-06,
+      "loss": 0.5774,
+      "step": 41744
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.8420228958129883,
+      "learning_rate": 2.4581047532150646e-06,
+      "loss": 0.7579,
+      "step": 41745
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.9187586307525635,
+      "learning_rate": 2.457877711884515e-06,
+      "loss": 0.6789,
+      "step": 41746
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 1.8845715522766113,
+      "learning_rate": 2.4576506776226526e-06,
+      "loss": 0.5625,
+      "step": 41747
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.361501693725586,
+      "learning_rate": 2.457423650430112e-06,
+      "loss": 0.6154,
+      "step": 41748
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 1.4866611957550049,
+      "learning_rate": 2.4571966303075253e-06,
+      "loss": 0.3297,
+      "step": 41749
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.3884260654449463,
+      "learning_rate": 2.4569696172555195e-06,
+      "loss": 0.608,
+      "step": 41750
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.1620426177978516,
+      "learning_rate": 2.456742611274731e-06,
+      "loss": 0.518,
+      "step": 41751
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 1.8819524049758911,
+      "learning_rate": 2.4565156123657874e-06,
+      "loss": 0.3478,
+      "step": 41752
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.8772597312927246,
+      "learning_rate": 2.4562886205293222e-06,
+      "loss": 0.7881,
+      "step": 41753
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.076796054840088,
+      "learning_rate": 2.4560616357659638e-06,
+      "loss": 0.8186,
+      "step": 41754
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.7318673133850098,
+      "learning_rate": 2.4558346580763448e-06,
+      "loss": 0.447,
+      "step": 41755
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.5036747455596924,
+      "learning_rate": 2.4556076874610984e-06,
+      "loss": 0.7746,
+      "step": 41756
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.12050461769104,
+      "learning_rate": 2.455380723920852e-06,
+      "loss": 0.6366,
+      "step": 41757
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.401190996170044,
+      "learning_rate": 2.455153767456241e-06,
+      "loss": 0.6133,
+      "step": 41758
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.273491859436035,
+      "learning_rate": 2.4549268180678915e-06,
+      "loss": 0.5537,
+      "step": 41759
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 1.5088491439819336,
+      "learning_rate": 2.4546998757564394e-06,
+      "loss": 0.3481,
+      "step": 41760
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.475095748901367,
+      "learning_rate": 2.4544729405225114e-06,
+      "loss": 0.4798,
+      "step": 41761
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.471041440963745,
+      "learning_rate": 2.4542460123667426e-06,
+      "loss": 0.649,
+      "step": 41762
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.1637818813323975,
+      "learning_rate": 2.45401909128976e-06,
+      "loss": 0.5976,
+      "step": 41763
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.078843116760254,
+      "learning_rate": 2.4537921772921984e-06,
+      "loss": 0.7128,
+      "step": 41764
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.670539140701294,
+      "learning_rate": 2.4535652703746856e-06,
+      "loss": 0.5995,
+      "step": 41765
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.067051410675049,
+      "learning_rate": 2.453338370537855e-06,
+      "loss": 0.55,
+      "step": 41766
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.002509355545044,
+      "learning_rate": 2.4531114777823343e-06,
+      "loss": 0.5558,
+      "step": 41767
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.1886329650878906,
+      "learning_rate": 2.452884592108757e-06,
+      "loss": 0.5794,
+      "step": 41768
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.451752185821533,
+      "learning_rate": 2.4526577135177536e-06,
+      "loss": 0.7984,
+      "step": 41769
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 4.682397842407227,
+      "learning_rate": 2.4524308420099563e-06,
+      "loss": 0.5525,
+      "step": 41770
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.134843587875366,
+      "learning_rate": 2.452203977585993e-06,
+      "loss": 0.5581,
+      "step": 41771
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 4.141349792480469,
+      "learning_rate": 2.4519771202464976e-06,
+      "loss": 0.6668,
+      "step": 41772
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 1.9725786447525024,
+      "learning_rate": 2.4517502699920982e-06,
+      "loss": 0.4568,
+      "step": 41773
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.228154420852661,
+      "learning_rate": 2.4515234268234276e-06,
+      "loss": 0.3959,
+      "step": 41774
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.56013822555542,
+      "learning_rate": 2.451296590741114e-06,
+      "loss": 0.4623,
+      "step": 41775
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.157442331314087,
+      "learning_rate": 2.4510697617457918e-06,
+      "loss": 0.7971,
+      "step": 41776
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.2827625274658203,
+      "learning_rate": 2.4508429398380878e-06,
+      "loss": 0.5287,
+      "step": 41777
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.901954412460327,
+      "learning_rate": 2.4506161250186367e-06,
+      "loss": 0.5585,
+      "step": 41778
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.119670867919922,
+      "learning_rate": 2.4503893172880657e-06,
+      "loss": 0.639,
+      "step": 41779
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.355234146118164,
+      "learning_rate": 2.4501625166470076e-06,
+      "loss": 0.5318,
+      "step": 41780
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.697343349456787,
+      "learning_rate": 2.4499357230960914e-06,
+      "loss": 0.6899,
+      "step": 41781
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.514094591140747,
+      "learning_rate": 2.4497089366359505e-06,
+      "loss": 0.597,
+      "step": 41782
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.3401947021484375,
+      "learning_rate": 2.4494821572672118e-06,
+      "loss": 0.7535,
+      "step": 41783
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.6125993728637695,
+      "learning_rate": 2.4492553849905093e-06,
+      "loss": 0.6578,
+      "step": 41784
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.6004691123962402,
+      "learning_rate": 2.44902861980647e-06,
+      "loss": 0.6814,
+      "step": 41785
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.751157522201538,
+      "learning_rate": 2.448801861715729e-06,
+      "loss": 0.5703,
+      "step": 41786
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.6065361499786377,
+      "learning_rate": 2.448575110718912e-06,
+      "loss": 0.5627,
+      "step": 41787
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.175164222717285,
+      "learning_rate": 2.448348366816654e-06,
+      "loss": 0.7185,
+      "step": 41788
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.529711961746216,
+      "learning_rate": 2.4481216300095793e-06,
+      "loss": 0.8625,
+      "step": 41789
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.542752504348755,
+      "learning_rate": 2.447894900298326e-06,
+      "loss": 0.6969,
+      "step": 41790
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.0883705615997314,
+      "learning_rate": 2.44766817768352e-06,
+      "loss": 0.4714,
+      "step": 41791
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.0617945194244385,
+      "learning_rate": 2.447441462165793e-06,
+      "loss": 0.5166,
+      "step": 41792
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.4497458934783936,
+      "learning_rate": 2.4472147537457745e-06,
+      "loss": 0.5444,
+      "step": 41793
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.2519257068634033,
+      "learning_rate": 2.446988052424097e-06,
+      "loss": 0.8047,
+      "step": 41794
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.710832118988037,
+      "learning_rate": 2.446761358201387e-06,
+      "loss": 0.6714,
+      "step": 41795
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.074686050415039,
+      "learning_rate": 2.44653467107828e-06,
+      "loss": 0.6157,
+      "step": 41796
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.4993512630462646,
+      "learning_rate": 2.446307991055401e-06,
+      "loss": 0.685,
+      "step": 41797
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.895108461380005,
+      "learning_rate": 2.4460813181333848e-06,
+      "loss": 0.674,
+      "step": 41798
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.2245771884918213,
+      "learning_rate": 2.445854652312858e-06,
+      "loss": 0.5336,
+      "step": 41799
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.4588429927825928,
+      "learning_rate": 2.4456279935944542e-06,
+      "loss": 0.5974,
+      "step": 41800
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.586890697479248,
+      "learning_rate": 2.4454013419788004e-06,
+      "loss": 0.504,
+      "step": 41801
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.797530174255371,
+      "learning_rate": 2.4451746974665304e-06,
+      "loss": 0.7841,
+      "step": 41802
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.54353928565979,
+      "learning_rate": 2.4449480600582703e-06,
+      "loss": 0.793,
+      "step": 41803
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 1.8047847747802734,
+      "learning_rate": 2.4447214297546528e-06,
+      "loss": 0.4851,
+      "step": 41804
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.459188222885132,
+      "learning_rate": 2.44449480655631e-06,
+      "loss": 0.7711,
+      "step": 41805
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.209150552749634,
+      "learning_rate": 2.4442681904638675e-06,
+      "loss": 0.6676,
+      "step": 41806
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 1.6062376499176025,
+      "learning_rate": 2.4440415814779593e-06,
+      "loss": 0.495,
+      "step": 41807
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.232178211212158,
+      "learning_rate": 2.4438149795992127e-06,
+      "loss": 0.3225,
+      "step": 41808
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.2130184173583984,
+      "learning_rate": 2.4435883848282593e-06,
+      "loss": 0.6338,
+      "step": 41809
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.698676586151123,
+      "learning_rate": 2.4433617971657305e-06,
+      "loss": 0.6386,
+      "step": 41810
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.075469732284546,
+      "learning_rate": 2.4431352166122526e-06,
+      "loss": 0.7667,
+      "step": 41811
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.543519973754883,
+      "learning_rate": 2.44290864316846e-06,
+      "loss": 0.6516,
+      "step": 41812
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.4030022621154785,
+      "learning_rate": 2.4426820768349785e-06,
+      "loss": 0.7412,
+      "step": 41813
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.2359259128570557,
+      "learning_rate": 2.442455517612442e-06,
+      "loss": 0.5206,
+      "step": 41814
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.4634594917297363,
+      "learning_rate": 2.442228965501477e-06,
+      "loss": 0.5405,
+      "step": 41815
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.0904793739318848,
+      "learning_rate": 2.442002420502715e-06,
+      "loss": 0.6782,
+      "step": 41816
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 1.7412413358688354,
+      "learning_rate": 2.4417758826167883e-06,
+      "loss": 0.3236,
+      "step": 41817
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.396644353866577,
+      "learning_rate": 2.441549351844322e-06,
+      "loss": 0.5306,
+      "step": 41818
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.1290576457977295,
+      "learning_rate": 2.441322828185951e-06,
+      "loss": 0.7269,
+      "step": 41819
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.0198328495025635,
+      "learning_rate": 2.4410963116423005e-06,
+      "loss": 0.5996,
+      "step": 41820
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.1057612895965576,
+      "learning_rate": 2.4408698022140043e-06,
+      "loss": 0.8026,
+      "step": 41821
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.7969319820404053,
+      "learning_rate": 2.4406432999016886e-06,
+      "loss": 0.8268,
+      "step": 41822
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.7174532413482666,
+      "learning_rate": 2.4404168047059875e-06,
+      "loss": 0.5301,
+      "step": 41823
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.5228934288024902,
+      "learning_rate": 2.440190316627526e-06,
+      "loss": 0.7566,
+      "step": 41824
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.566453695297241,
+      "learning_rate": 2.4399638356669382e-06,
+      "loss": 0.619,
+      "step": 41825
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.210073232650757,
+      "learning_rate": 2.4397373618248504e-06,
+      "loss": 0.4465,
+      "step": 41826
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.3292787075042725,
+      "learning_rate": 2.439510895101896e-06,
+      "loss": 0.5092,
+      "step": 41827
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.6187384128570557,
+      "learning_rate": 2.4392844354987e-06,
+      "loss": 0.5851,
+      "step": 41828
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.254552125930786,
+      "learning_rate": 2.4390579830158952e-06,
+      "loss": 0.814,
+      "step": 41829
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.3072314262390137,
+      "learning_rate": 2.4388315376541104e-06,
+      "loss": 0.8778,
+      "step": 41830
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.2785542011260986,
+      "learning_rate": 2.4386050994139783e-06,
+      "loss": 0.5914,
+      "step": 41831
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.9525468349456787,
+      "learning_rate": 2.4383786682961235e-06,
+      "loss": 0.7155,
+      "step": 41832
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.2781312465667725,
+      "learning_rate": 2.43815224430118e-06,
+      "loss": 0.7098,
+      "step": 41833
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.4591753482818604,
+      "learning_rate": 2.4379258274297733e-06,
+      "loss": 0.9097,
+      "step": 41834
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.978055715560913,
+      "learning_rate": 2.4376994176825373e-06,
+      "loss": 0.595,
+      "step": 41835
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.5920217037200928,
+      "learning_rate": 2.437473015060098e-06,
+      "loss": 0.5913,
+      "step": 41836
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.7755966186523438,
+      "learning_rate": 2.437246619563087e-06,
+      "loss": 0.6386,
+      "step": 41837
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.003514528274536,
+      "learning_rate": 2.437020231192132e-06,
+      "loss": 0.7521,
+      "step": 41838
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.9365694522857666,
+      "learning_rate": 2.4367938499478656e-06,
+      "loss": 0.8522,
+      "step": 41839
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.7438371181488037,
+      "learning_rate": 2.4365674758309126e-06,
+      "loss": 0.7027,
+      "step": 41840
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.7257614135742188,
+      "learning_rate": 2.4363411088419075e-06,
+      "loss": 0.5144,
+      "step": 41841
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.1557669639587402,
+      "learning_rate": 2.4361147489814757e-06,
+      "loss": 0.881,
+      "step": 41842
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.0561907291412354,
+      "learning_rate": 2.4358883962502505e-06,
+      "loss": 0.4715,
+      "step": 41843
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 5.298894882202148,
+      "learning_rate": 2.4356620506488565e-06,
+      "loss": 0.6741,
+      "step": 41844
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 4.129962921142578,
+      "learning_rate": 2.435435712177928e-06,
+      "loss": 0.6656,
+      "step": 41845
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.125124454498291,
+      "learning_rate": 2.43520938083809e-06,
+      "loss": 0.6076,
+      "step": 41846
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 7.206005096435547,
+      "learning_rate": 2.4349830566299764e-06,
+      "loss": 0.4703,
+      "step": 41847
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 1.8223679065704346,
+      "learning_rate": 2.4347567395542114e-06,
+      "loss": 0.4919,
+      "step": 41848
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.361886978149414,
+      "learning_rate": 2.4345304296114294e-06,
+      "loss": 0.748,
+      "step": 41849
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.78861665725708,
+      "learning_rate": 2.4343041268022534e-06,
+      "loss": 0.5453,
+      "step": 41850
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.4336910247802734,
+      "learning_rate": 2.43407783112732e-06,
+      "loss": 0.6689,
+      "step": 41851
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.353821277618408,
+      "learning_rate": 2.433851542587253e-06,
+      "loss": 0.5701,
+      "step": 41852
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.0767199993133545,
+      "learning_rate": 2.4336252611826854e-06,
+      "loss": 0.699,
+      "step": 41853
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.4996824264526367,
+      "learning_rate": 2.4333989869142427e-06,
+      "loss": 0.7121,
+      "step": 41854
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.133908748626709,
+      "learning_rate": 2.433172719782558e-06,
+      "loss": 0.7036,
+      "step": 41855
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.5251269340515137,
+      "learning_rate": 2.432946459788257e-06,
+      "loss": 0.4291,
+      "step": 41856
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 4.412590503692627,
+      "learning_rate": 2.432720206931971e-06,
+      "loss": 0.5296,
+      "step": 41857
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 4.050530910491943,
+      "learning_rate": 2.4324939612143273e-06,
+      "loss": 0.5576,
+      "step": 41858
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 1.7180951833724976,
+      "learning_rate": 2.4322677226359583e-06,
+      "loss": 0.4489,
+      "step": 41859
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.793689727783203,
+      "learning_rate": 2.4320414911974883e-06,
+      "loss": 0.5798,
+      "step": 41860
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.7056093215942383,
+      "learning_rate": 2.431815266899551e-06,
+      "loss": 0.5803,
+      "step": 41861
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.698784828186035,
+      "learning_rate": 2.4315890497427714e-06,
+      "loss": 0.5889,
+      "step": 41862
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.4046034812927246,
+      "learning_rate": 2.4313628397277823e-06,
+      "loss": 0.4182,
+      "step": 41863
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.6168627738952637,
+      "learning_rate": 2.431136636855209e-06,
+      "loss": 0.4361,
+      "step": 41864
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.7940971851348877,
+      "learning_rate": 2.4309104411256852e-06,
+      "loss": 0.5593,
+      "step": 41865
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.8099400997161865,
+      "learning_rate": 2.430684252539834e-06,
+      "loss": 0.7952,
+      "step": 41866
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.0227718353271484,
+      "learning_rate": 2.430458071098288e-06,
+      "loss": 0.7724,
+      "step": 41867
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 1.7656161785125732,
+      "learning_rate": 2.4302318968016774e-06,
+      "loss": 0.5933,
+      "step": 41868
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.2891132831573486,
+      "learning_rate": 2.4300057296506267e-06,
+      "loss": 0.5039,
+      "step": 41869
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.0991499423980713,
+      "learning_rate": 2.429779569645768e-06,
+      "loss": 0.7488,
+      "step": 41870
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.6377716064453125,
+      "learning_rate": 2.4295534167877317e-06,
+      "loss": 0.7317,
+      "step": 41871
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 1.5772165060043335,
+      "learning_rate": 2.429327271077142e-06,
+      "loss": 0.5874,
+      "step": 41872
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.0298917293548584,
+      "learning_rate": 2.429101132514632e-06,
+      "loss": 0.5709,
+      "step": 41873
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.405508279800415,
+      "learning_rate": 2.428875001100827e-06,
+      "loss": 0.5611,
+      "step": 41874
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.9046225547790527,
+      "learning_rate": 2.4286488768363593e-06,
+      "loss": 0.6203,
+      "step": 41875
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.088167667388916,
+      "learning_rate": 2.428422759721854e-06,
+      "loss": 0.7403,
+      "step": 41876
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.440387010574341,
+      "learning_rate": 2.4281966497579436e-06,
+      "loss": 0.7393,
+      "step": 41877
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.3413572311401367,
+      "learning_rate": 2.4279705469452523e-06,
+      "loss": 0.5192,
+      "step": 41878
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.3818867206573486,
+      "learning_rate": 2.427744451284412e-06,
+      "loss": 0.4671,
+      "step": 41879
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 1.7447746992111206,
+      "learning_rate": 2.427518362776053e-06,
+      "loss": 0.3491,
+      "step": 41880
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.3847012519836426,
+      "learning_rate": 2.4272922814207995e-06,
+      "loss": 0.5544,
+      "step": 41881
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.280674695968628,
+      "learning_rate": 2.427066207219284e-06,
+      "loss": 0.7783,
+      "step": 41882
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.355217218399048,
+      "learning_rate": 2.426840140172132e-06,
+      "loss": 0.4983,
+      "step": 41883
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.219632387161255,
+      "learning_rate": 2.4266140802799758e-06,
+      "loss": 0.6404,
+      "step": 41884
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.175499439239502,
+      "learning_rate": 2.42638802754344e-06,
+      "loss": 0.6286,
+      "step": 41885
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.166294574737549,
+      "learning_rate": 2.426161981963156e-06,
+      "loss": 0.5045,
+      "step": 41886
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.2792935371398926,
+      "learning_rate": 2.42593594353975e-06,
+      "loss": 0.564,
+      "step": 41887
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.5576300621032715,
+      "learning_rate": 2.4257099122738537e-06,
+      "loss": 0.5516,
+      "step": 41888
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.5320262908935547,
+      "learning_rate": 2.425483888166092e-06,
+      "loss": 0.6113,
+      "step": 41889
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 9.462010383605957,
+      "learning_rate": 2.4252578712170955e-06,
+      "loss": 0.8603,
+      "step": 41890
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.6933352947235107,
+      "learning_rate": 2.4250318614274915e-06,
+      "loss": 0.7015,
+      "step": 41891
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 4.493791103363037,
+      "learning_rate": 2.4248058587979117e-06,
+      "loss": 0.6684,
+      "step": 41892
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.1140894889831543,
+      "learning_rate": 2.4245798633289803e-06,
+      "loss": 0.6267,
+      "step": 41893
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.773545980453491,
+      "learning_rate": 2.4243538750213292e-06,
+      "loss": 0.721,
+      "step": 41894
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.228109359741211,
+      "learning_rate": 2.4241278938755835e-06,
+      "loss": 0.9309,
+      "step": 41895
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 1.4967553615570068,
+      "learning_rate": 2.4239019198923747e-06,
+      "loss": 0.4317,
+      "step": 41896
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.4685776233673096,
+      "learning_rate": 2.4236759530723282e-06,
+      "loss": 0.5244,
+      "step": 41897
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.3248887062072754,
+      "learning_rate": 2.423449993416075e-06,
+      "loss": 0.5803,
+      "step": 41898
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.041869640350342,
+      "learning_rate": 2.4232240409242407e-06,
+      "loss": 0.5177,
+      "step": 41899
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 5.935493469238281,
+      "learning_rate": 2.4229980955974567e-06,
+      "loss": 0.6221,
+      "step": 41900
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.5390796661376953,
+      "learning_rate": 2.422772157436348e-06,
+      "loss": 0.765,
+      "step": 41901
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.417844295501709,
+      "learning_rate": 2.4225462264415467e-06,
+      "loss": 0.6193,
+      "step": 41902
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.273885488510132,
+      "learning_rate": 2.4223203026136765e-06,
+      "loss": 0.6133,
+      "step": 41903
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.3562159538269043,
+      "learning_rate": 2.42209438595337e-06,
+      "loss": 0.4599,
+      "step": 41904
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.7606117725372314,
+      "learning_rate": 2.421868476461251e-06,
+      "loss": 0.7145,
+      "step": 41905
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.3269355297088623,
+      "learning_rate": 2.421642574137952e-06,
+      "loss": 0.5261,
+      "step": 41906
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.3660404682159424,
+      "learning_rate": 2.4214166789840977e-06,
+      "loss": 0.6516,
+      "step": 41907
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 5.593632221221924,
+      "learning_rate": 2.421190791000319e-06,
+      "loss": 0.445,
+      "step": 41908
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.4844729900360107,
+      "learning_rate": 2.4209649101872406e-06,
+      "loss": 0.7765,
+      "step": 41909
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.2990076541900635,
+      "learning_rate": 2.4207390365454953e-06,
+      "loss": 0.7583,
+      "step": 41910
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.1153228282928467,
+      "learning_rate": 2.420513170075704e-06,
+      "loss": 0.5849,
+      "step": 41911
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.405845880508423,
+      "learning_rate": 2.4202873107785036e-06,
+      "loss": 0.6061,
+      "step": 41912
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.433353900909424,
+      "learning_rate": 2.4200614586545156e-06,
+      "loss": 0.5409,
+      "step": 41913
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.5817551612854004,
+      "learning_rate": 2.4198356137043727e-06,
+      "loss": 0.5662,
+      "step": 41914
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 5.944272518157959,
+      "learning_rate": 2.419609775928698e-06,
+      "loss": 0.6636,
+      "step": 41915
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.3347840309143066,
+      "learning_rate": 2.419383945328124e-06,
+      "loss": 0.3327,
+      "step": 41916
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 8.531962394714355,
+      "learning_rate": 2.419158121903274e-06,
+      "loss": 0.6978,
+      "step": 41917
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.160909414291382,
+      "learning_rate": 2.418932305654781e-06,
+      "loss": 0.6818,
+      "step": 41918
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 4.254855155944824,
+      "learning_rate": 2.418706496583268e-06,
+      "loss": 0.7667,
+      "step": 41919
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.587934732437134,
+      "learning_rate": 2.4184806946893674e-06,
+      "loss": 0.4993,
+      "step": 41920
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.2128875255584717,
+      "learning_rate": 2.418254899973703e-06,
+      "loss": 0.5162,
+      "step": 41921
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.756387710571289,
+      "learning_rate": 2.418029112436907e-06,
+      "loss": 0.4871,
+      "step": 41922
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.3951244354248047,
+      "learning_rate": 2.4178033320796022e-06,
+      "loss": 0.6931,
+      "step": 41923
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 4.739106178283691,
+      "learning_rate": 2.4175775589024213e-06,
+      "loss": 0.5845,
+      "step": 41924
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.2133820056915283,
+      "learning_rate": 2.417351792905988e-06,
+      "loss": 0.5454,
+      "step": 41925
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.2215516567230225,
+      "learning_rate": 2.4171260340909335e-06,
+      "loss": 0.73,
+      "step": 41926
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.3614914417266846,
+      "learning_rate": 2.416900282457882e-06,
+      "loss": 0.8537,
+      "step": 41927
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.770076036453247,
+      "learning_rate": 2.4166745380074635e-06,
+      "loss": 0.4939,
+      "step": 41928
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.1754419803619385,
+      "learning_rate": 2.416448800740307e-06,
+      "loss": 0.5797,
+      "step": 41929
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.50295090675354,
+      "learning_rate": 2.4162230706570366e-06,
+      "loss": 0.8308,
+      "step": 41930
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.005730390548706,
+      "learning_rate": 2.415997347758282e-06,
+      "loss": 0.7868,
+      "step": 41931
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.046299695968628,
+      "learning_rate": 2.4157716320446728e-06,
+      "loss": 0.6377,
+      "step": 41932
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.488577127456665,
+      "learning_rate": 2.4155459235168326e-06,
+      "loss": 0.6105,
+      "step": 41933
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.824463367462158,
+      "learning_rate": 2.415320222175393e-06,
+      "loss": 0.5986,
+      "step": 41934
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.2339835166931152,
+      "learning_rate": 2.4150945280209776e-06,
+      "loss": 0.5096,
+      "step": 41935
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.782604455947876,
+      "learning_rate": 2.414868841054218e-06,
+      "loss": 0.6236,
+      "step": 41936
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.100351333618164,
+      "learning_rate": 2.4146431612757375e-06,
+      "loss": 0.6069,
+      "step": 41937
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.12266206741333,
+      "learning_rate": 2.4144174886861683e-06,
+      "loss": 0.696,
+      "step": 41938
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.05989933013916,
+      "learning_rate": 2.4141918232861334e-06,
+      "loss": 0.563,
+      "step": 41939
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.4456207752227783,
+      "learning_rate": 2.4139661650762623e-06,
+      "loss": 0.4475,
+      "step": 41940
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.5043365955352783,
+      "learning_rate": 2.4137405140571845e-06,
+      "loss": 0.6107,
+      "step": 41941
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.1471025943756104,
+      "learning_rate": 2.4135148702295237e-06,
+      "loss": 0.4133,
+      "step": 41942
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.157226800918579,
+      "learning_rate": 2.413289233593911e-06,
+      "loss": 0.4093,
+      "step": 41943
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.3164405822753906,
+      "learning_rate": 2.4130636041509696e-06,
+      "loss": 0.4723,
+      "step": 41944
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 1.9590157270431519,
+      "learning_rate": 2.412837981901332e-06,
+      "loss": 0.5064,
+      "step": 41945
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.152862548828125,
+      "learning_rate": 2.4126123668456203e-06,
+      "loss": 0.5421,
+      "step": 41946
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.300886631011963,
+      "learning_rate": 2.412386758984466e-06,
+      "loss": 0.4681,
+      "step": 41947
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.25640606880188,
+      "learning_rate": 2.4121611583184936e-06,
+      "loss": 0.5912,
+      "step": 41948
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.626831531524658,
+      "learning_rate": 2.4119355648483327e-06,
+      "loss": 0.474,
+      "step": 41949
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.945671796798706,
+      "learning_rate": 2.4117099785746082e-06,
+      "loss": 0.7338,
+      "step": 41950
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.9246439933776855,
+      "learning_rate": 2.4114843994979486e-06,
+      "loss": 0.7415,
+      "step": 41951
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.528602361679077,
+      "learning_rate": 2.4112588276189807e-06,
+      "loss": 0.7124,
+      "step": 41952
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.47174072265625,
+      "learning_rate": 2.4110332629383347e-06,
+      "loss": 0.4451,
+      "step": 41953
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.520526170730591,
+      "learning_rate": 2.410807705456633e-06,
+      "loss": 0.4695,
+      "step": 41954
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.602325677871704,
+      "learning_rate": 2.410582155174508e-06,
+      "loss": 0.6087,
+      "step": 41955
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 4.11990213394165,
+      "learning_rate": 2.4103566120925807e-06,
+      "loss": 0.4147,
+      "step": 41956
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 4.301260471343994,
+      "learning_rate": 2.4101310762114843e-06,
+      "loss": 0.6149,
+      "step": 41957
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.5089778900146484,
+      "learning_rate": 2.4099055475318413e-06,
+      "loss": 0.8782,
+      "step": 41958
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 1.7274068593978882,
+      "learning_rate": 2.4096800260542823e-06,
+      "loss": 0.4402,
+      "step": 41959
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.9760901927948,
+      "learning_rate": 2.409454511779431e-06,
+      "loss": 0.5552,
+      "step": 41960
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.3319249153137207,
+      "learning_rate": 2.4092290047079183e-06,
+      "loss": 0.4991,
+      "step": 41961
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.0104894638061523,
+      "learning_rate": 2.409003504840367e-06,
+      "loss": 0.5646,
+      "step": 41962
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.495438814163208,
+      "learning_rate": 2.4087780121774085e-06,
+      "loss": 0.7889,
+      "step": 41963
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 1.8944518566131592,
+      "learning_rate": 2.408552526719666e-06,
+      "loss": 0.5001,
+      "step": 41964
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 1.8565363883972168,
+      "learning_rate": 2.4083270484677695e-06,
+      "loss": 0.764,
+      "step": 41965
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.072882652282715,
+      "learning_rate": 2.4081015774223433e-06,
+      "loss": 0.4888,
+      "step": 41966
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.383852958679199,
+      "learning_rate": 2.407876113584017e-06,
+      "loss": 0.4709,
+      "step": 41967
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.943068742752075,
+      "learning_rate": 2.4076506569534148e-06,
+      "loss": 0.5622,
+      "step": 41968
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.6045479774475098,
+      "learning_rate": 2.407425207531166e-06,
+      "loss": 0.8358,
+      "step": 41969
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.121014356613159,
+      "learning_rate": 2.4071997653178956e-06,
+      "loss": 0.6923,
+      "step": 41970
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.4208877086639404,
+      "learning_rate": 2.406974330314233e-06,
+      "loss": 0.5845,
+      "step": 41971
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.2205047607421875,
+      "learning_rate": 2.406748902520799e-06,
+      "loss": 0.6039,
+      "step": 41972
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.3999757766723633,
+      "learning_rate": 2.4065234819382293e-06,
+      "loss": 0.4731,
+      "step": 41973
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.533710479736328,
+      "learning_rate": 2.4062980685671444e-06,
+      "loss": 0.788,
+      "step": 41974
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.7707841396331787,
+      "learning_rate": 2.406072662408175e-06,
+      "loss": 0.6546,
+      "step": 41975
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.8954272270202637,
+      "learning_rate": 2.4058472634619433e-06,
+      "loss": 0.7695,
+      "step": 41976
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 1.7494486570358276,
+      "learning_rate": 2.4056218717290808e-06,
+      "loss": 0.3581,
+      "step": 41977
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 1.8286024332046509,
+      "learning_rate": 2.40539648721021e-06,
+      "loss": 0.5924,
+      "step": 41978
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 4.207449913024902,
+      "learning_rate": 2.4051711099059614e-06,
+      "loss": 0.6133,
+      "step": 41979
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 1.5548698902130127,
+      "learning_rate": 2.4049457398169582e-06,
+      "loss": 0.3587,
+      "step": 41980
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.796386241912842,
+      "learning_rate": 2.4047203769438304e-06,
+      "loss": 0.4533,
+      "step": 41981
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.254364013671875,
+      "learning_rate": 2.4044950212872014e-06,
+      "loss": 0.3932,
+      "step": 41982
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.915194034576416,
+      "learning_rate": 2.4042696728477018e-06,
+      "loss": 0.558,
+      "step": 41983
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.3025717735290527,
+      "learning_rate": 2.4040443316259533e-06,
+      "loss": 0.5366,
+      "step": 41984
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.244745969772339,
+      "learning_rate": 2.403818997622587e-06,
+      "loss": 0.5639,
+      "step": 41985
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.0982744693756104,
+      "learning_rate": 2.4035936708382258e-06,
+      "loss": 0.6182,
+      "step": 41986
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.3560400009155273,
+      "learning_rate": 2.4033683512734995e-06,
+      "loss": 0.5889,
+      "step": 41987
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.0944366455078125,
+      "learning_rate": 2.4031430389290316e-06,
+      "loss": 0.7553,
+      "step": 41988
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 1.9827340841293335,
+      "learning_rate": 2.4029177338054514e-06,
+      "loss": 0.5631,
+      "step": 41989
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 1.7152906656265259,
+      "learning_rate": 2.402692435903383e-06,
+      "loss": 0.5509,
+      "step": 41990
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.5673463344573975,
+      "learning_rate": 2.4024671452234537e-06,
+      "loss": 0.3763,
+      "step": 41991
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 1.8736385107040405,
+      "learning_rate": 2.40224186176629e-06,
+      "loss": 0.7062,
+      "step": 41992
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.409316062927246,
+      "learning_rate": 2.4020165855325207e-06,
+      "loss": 0.8443,
+      "step": 41993
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.298762321472168,
+      "learning_rate": 2.401791316522768e-06,
+      "loss": 0.4988,
+      "step": 41994
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.7792775630950928,
+      "learning_rate": 2.4015660547376623e-06,
+      "loss": 0.4726,
+      "step": 41995
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 4.436508655548096,
+      "learning_rate": 2.401340800177826e-06,
+      "loss": 0.638,
+      "step": 41996
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 1.9806333780288696,
+      "learning_rate": 2.4011155528438895e-06,
+      "loss": 0.4995,
+      "step": 41997
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 1.6400067806243896,
+      "learning_rate": 2.4008903127364754e-06,
+      "loss": 0.3503,
+      "step": 41998
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.300358295440674,
+      "learning_rate": 2.4006650798562133e-06,
+      "loss": 0.3374,
+      "step": 41999
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.3129336833953857,
+      "learning_rate": 2.4004398542037257e-06,
+      "loss": 0.7367,
+      "step": 42000
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 1.7623929977416992,
+      "learning_rate": 2.4002146357796415e-06,
+      "loss": 0.5974,
+      "step": 42001
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 1.9519343376159668,
+      "learning_rate": 2.399989424584588e-06,
+      "loss": 0.4624,
+      "step": 42002
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.4216268062591553,
+      "learning_rate": 2.3997642206191884e-06,
+      "loss": 0.7283,
+      "step": 42003
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.376444101333618,
+      "learning_rate": 2.3995390238840723e-06,
+      "loss": 0.5471,
+      "step": 42004
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.565027952194214,
+      "learning_rate": 2.3993138343798617e-06,
+      "loss": 0.4563,
+      "step": 42005
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.0496652126312256,
+      "learning_rate": 2.3990886521071873e-06,
+      "loss": 0.551,
+      "step": 42006
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.6382193565368652,
+      "learning_rate": 2.3988634770666712e-06,
+      "loss": 0.5013,
+      "step": 42007
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 5.512135028839111,
+      "learning_rate": 2.398638309258943e-06,
+      "loss": 0.6648,
+      "step": 42008
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 1.8322416543960571,
+      "learning_rate": 2.398413148684626e-06,
+      "loss": 0.6015,
+      "step": 42009
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.34586238861084,
+      "learning_rate": 2.398187995344348e-06,
+      "loss": 0.767,
+      "step": 42010
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.9252257347106934,
+      "learning_rate": 2.3979628492387337e-06,
+      "loss": 0.6524,
+      "step": 42011
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.063591241836548,
+      "learning_rate": 2.3977377103684095e-06,
+      "loss": 0.5553,
+      "step": 42012
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 1.8946890830993652,
+      "learning_rate": 2.3975125787340028e-06,
+      "loss": 0.6484,
+      "step": 42013
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.009573221206665,
+      "learning_rate": 2.3972874543361394e-06,
+      "loss": 0.8573,
+      "step": 42014
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.306870698928833,
+      "learning_rate": 2.397062337175443e-06,
+      "loss": 0.4725,
+      "step": 42015
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.227097272872925,
+      "learning_rate": 2.396837227252544e-06,
+      "loss": 0.6807,
+      "step": 42016
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 1.941971778869629,
+      "learning_rate": 2.3966121245680622e-06,
+      "loss": 0.5072,
+      "step": 42017
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 1.6322095394134521,
+      "learning_rate": 2.3963870291226295e-06,
+      "loss": 0.5269,
+      "step": 42018
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.265514612197876,
+      "learning_rate": 2.3961619409168675e-06,
+      "loss": 0.5899,
+      "step": 42019
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.5679421424865723,
+      "learning_rate": 2.395936859951405e-06,
+      "loss": 0.7511,
+      "step": 42020
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 1.8006798028945923,
+      "learning_rate": 2.3957117862268646e-06,
+      "loss": 0.4996,
+      "step": 42021
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.9362943172454834,
+      "learning_rate": 2.395486719743877e-06,
+      "loss": 0.5865,
+      "step": 42022
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 1.7854273319244385,
+      "learning_rate": 2.395261660503062e-06,
+      "loss": 0.5716,
+      "step": 42023
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.1151204109191895,
+      "learning_rate": 2.395036608505051e-06,
+      "loss": 0.496,
+      "step": 42024
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.632146120071411,
+      "learning_rate": 2.394811563750466e-06,
+      "loss": 0.7721,
+      "step": 42025
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.867018461227417,
+      "learning_rate": 2.3945865262399356e-06,
+      "loss": 0.763,
+      "step": 42026
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.2250587940216064,
+      "learning_rate": 2.394361495974082e-06,
+      "loss": 0.496,
+      "step": 42027
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 1.7362816333770752,
+      "learning_rate": 2.394136472953535e-06,
+      "loss": 0.4237,
+      "step": 42028
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 1.9037646055221558,
+      "learning_rate": 2.3939114571789162e-06,
+      "loss": 0.598,
+      "step": 42029
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.174816370010376,
+      "learning_rate": 2.3936864486508555e-06,
+      "loss": 0.5561,
+      "step": 42030
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.6108148097991943,
+      "learning_rate": 2.3934614473699747e-06,
+      "loss": 0.7135,
+      "step": 42031
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.817988872528076,
+      "learning_rate": 2.393236453336903e-06,
+      "loss": 0.7477,
+      "step": 42032
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.776695728302002,
+      "learning_rate": 2.39301146655226e-06,
+      "loss": 0.5526,
+      "step": 42033
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.857572317123413,
+      "learning_rate": 2.3927864870166793e-06,
+      "loss": 0.8528,
+      "step": 42034
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.0394506454467773,
+      "learning_rate": 2.3925615147307816e-06,
+      "loss": 0.6037,
+      "step": 42035
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.6389150619506836,
+      "learning_rate": 2.392336549695195e-06,
+      "loss": 0.6444,
+      "step": 42036
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.3386263847351074,
+      "learning_rate": 2.3921115919105415e-06,
+      "loss": 0.6684,
+      "step": 42037
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 1.7194322347640991,
+      "learning_rate": 2.391886641377451e-06,
+      "loss": 0.6442,
+      "step": 42038
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.506953239440918,
+      "learning_rate": 2.3916616980965445e-06,
+      "loss": 0.4946,
+      "step": 42039
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.737563133239746,
+      "learning_rate": 2.391436762068452e-06,
+      "loss": 0.5519,
+      "step": 42040
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 4.446508407592773,
+      "learning_rate": 2.3912118332937943e-06,
+      "loss": 0.6801,
+      "step": 42041
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 1.6693408489227295,
+      "learning_rate": 2.390986911773201e-06,
+      "loss": 0.6026,
+      "step": 42042
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 7.920770168304443,
+      "learning_rate": 2.390761997507294e-06,
+      "loss": 0.6028,
+      "step": 42043
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.5817127227783203,
+      "learning_rate": 2.390537090496703e-06,
+      "loss": 0.7317,
+      "step": 42044
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.9775185585021973,
+      "learning_rate": 2.3903121907420485e-06,
+      "loss": 0.7733,
+      "step": 42045
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.290400505065918,
+      "learning_rate": 2.39008729824396e-06,
+      "loss": 0.4961,
+      "step": 42046
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.731268882751465,
+      "learning_rate": 2.389862413003059e-06,
+      "loss": 0.7349,
+      "step": 42047
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.040782928466797,
+      "learning_rate": 2.3896375350199746e-06,
+      "loss": 0.4853,
+      "step": 42048
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.6781439781188965,
+      "learning_rate": 2.389412664295329e-06,
+      "loss": 0.7638,
+      "step": 42049
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.7039263248443604,
+      "learning_rate": 2.3891878008297502e-06,
+      "loss": 0.6637,
+      "step": 42050
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.505305051803589,
+      "learning_rate": 2.388962944623861e-06,
+      "loss": 0.6333,
+      "step": 42051
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.8918933868408203,
+      "learning_rate": 2.388738095678287e-06,
+      "loss": 0.6679,
+      "step": 42052
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.6802730560302734,
+      "learning_rate": 2.3885132539936545e-06,
+      "loss": 0.6316,
+      "step": 42053
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.0335421562194824,
+      "learning_rate": 2.38828841957059e-06,
+      "loss": 0.5289,
+      "step": 42054
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 4.23770809173584,
+      "learning_rate": 2.388063592409715e-06,
+      "loss": 0.6922,
+      "step": 42055
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 4.426294803619385,
+      "learning_rate": 2.387838772511659e-06,
+      "loss": 0.6392,
+      "step": 42056
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.2685353755950928,
+      "learning_rate": 2.3876139598770426e-06,
+      "loss": 0.6837,
+      "step": 42057
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.5695412158966064,
+      "learning_rate": 2.3873891545064953e-06,
+      "loss": 0.7356,
+      "step": 42058
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.841636896133423,
+      "learning_rate": 2.387164356400638e-06,
+      "loss": 0.6373,
+      "step": 42059
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.289555788040161,
+      "learning_rate": 2.3869395655601003e-06,
+      "loss": 0.6119,
+      "step": 42060
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.1001272201538086,
+      "learning_rate": 2.3867147819855018e-06,
+      "loss": 0.5578,
+      "step": 42061
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.2737698554992676,
+      "learning_rate": 2.386490005677473e-06,
+      "loss": 0.6963,
+      "step": 42062
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.4294774532318115,
+      "learning_rate": 2.386265236636635e-06,
+      "loss": 0.5194,
+      "step": 42063
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.4452972412109375,
+      "learning_rate": 2.386040474863614e-06,
+      "loss": 0.3732,
+      "step": 42064
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.025266408920288,
+      "learning_rate": 2.385815720359037e-06,
+      "loss": 0.7523,
+      "step": 42065
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.145991086959839,
+      "learning_rate": 2.3855909731235248e-06,
+      "loss": 0.508,
+      "step": 42066
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.1822187900543213,
+      "learning_rate": 2.3853662331577077e-06,
+      "loss": 0.5518,
+      "step": 42067
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 6.038313388824463,
+      "learning_rate": 2.3851415004622047e-06,
+      "loss": 0.5324,
+      "step": 42068
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 1.5424929857254028,
+      "learning_rate": 2.384916775037646e-06,
+      "loss": 0.479,
+      "step": 42069
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.201761484146118,
+      "learning_rate": 2.384692056884652e-06,
+      "loss": 0.6515,
+      "step": 42070
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.091331958770752,
+      "learning_rate": 2.384467346003852e-06,
+      "loss": 0.6856,
+      "step": 42071
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.746403694152832,
+      "learning_rate": 2.3842426423958666e-06,
+      "loss": 0.4229,
+      "step": 42072
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.2641687393188477,
+      "learning_rate": 2.3840179460613223e-06,
+      "loss": 0.517,
+      "step": 42073
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.4907195568084717,
+      "learning_rate": 2.3837932570008436e-06,
+      "loss": 0.65,
+      "step": 42074
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.3878965377807617,
+      "learning_rate": 2.3835685752150582e-06,
+      "loss": 0.8849,
+      "step": 42075
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.1683382987976074,
+      "learning_rate": 2.3833439007045867e-06,
+      "loss": 0.6712,
+      "step": 42076
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.4958364963531494,
+      "learning_rate": 2.3831192334700574e-06,
+      "loss": 0.421,
+      "step": 42077
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.4960949420928955,
+      "learning_rate": 2.3828945735120913e-06,
+      "loss": 0.6165,
+      "step": 42078
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.2884256839752197,
+      "learning_rate": 2.382669920831317e-06,
+      "loss": 0.6505,
+      "step": 42079
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.7034075260162354,
+      "learning_rate": 2.382445275428355e-06,
+      "loss": 0.5341,
+      "step": 42080
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.5064806938171387,
+      "learning_rate": 2.382220637303834e-06,
+      "loss": 0.5588,
+      "step": 42081
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.6040353775024414,
+      "learning_rate": 2.381996006458375e-06,
+      "loss": 0.3517,
+      "step": 42082
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.4407083988189697,
+      "learning_rate": 2.3817713828926067e-06,
+      "loss": 0.4711,
+      "step": 42083
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.5924222469329834,
+      "learning_rate": 2.381546766607149e-06,
+      "loss": 0.6232,
+      "step": 42084
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.9626264572143555,
+      "learning_rate": 2.3813221576026307e-06,
+      "loss": 0.5006,
+      "step": 42085
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.089564085006714,
+      "learning_rate": 2.381097555879673e-06,
+      "loss": 0.7367,
+      "step": 42086
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.9891722202301025,
+      "learning_rate": 2.3808729614389037e-06,
+      "loss": 0.7737,
+      "step": 42087
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.1746208667755127,
+      "learning_rate": 2.380648374280943e-06,
+      "loss": 0.6067,
+      "step": 42088
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 4.2366743087768555,
+      "learning_rate": 2.3804237944064206e-06,
+      "loss": 0.6588,
+      "step": 42089
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.266152858734131,
+      "learning_rate": 2.3801992218159563e-06,
+      "loss": 0.753,
+      "step": 42090
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.3700079917907715,
+      "learning_rate": 2.379974656510178e-06,
+      "loss": 0.6369,
+      "step": 42091
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.049056053161621,
+      "learning_rate": 2.3797500984897075e-06,
+      "loss": 0.6153,
+      "step": 42092
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.5438930988311768,
+      "learning_rate": 2.3795255477551717e-06,
+      "loss": 0.7063,
+      "step": 42093
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.9814016819000244,
+      "learning_rate": 2.37930100430719e-06,
+      "loss": 0.6648,
+      "step": 42094
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.2755322456359863,
+      "learning_rate": 2.3790764681463944e-06,
+      "loss": 0.765,
+      "step": 42095
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.2820093631744385,
+      "learning_rate": 2.3788519392734036e-06,
+      "loss": 0.7371,
+      "step": 42096
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.742913246154785,
+      "learning_rate": 2.378627417688845e-06,
+      "loss": 0.6905,
+      "step": 42097
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.166729688644409,
+      "learning_rate": 2.3784029033933405e-06,
+      "loss": 0.6121,
+      "step": 42098
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.6705455780029297,
+      "learning_rate": 2.378178396387517e-06,
+      "loss": 0.715,
+      "step": 42099
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.1233246326446533,
+      "learning_rate": 2.3779538966719953e-06,
+      "loss": 0.4625,
+      "step": 42100
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.0316085815429688,
+      "learning_rate": 2.377729404247403e-06,
+      "loss": 0.7127,
+      "step": 42101
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.901484966278076,
+      "learning_rate": 2.3775049191143616e-06,
+      "loss": 0.5088,
+      "step": 42102
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.1540215015411377,
+      "learning_rate": 2.3772804412734982e-06,
+      "loss": 0.6776,
+      "step": 42103
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.900743246078491,
+      "learning_rate": 2.3770559707254336e-06,
+      "loss": 0.5802,
+      "step": 42104
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.0802395343780518,
+      "learning_rate": 2.376831507470796e-06,
+      "loss": 0.7513,
+      "step": 42105
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.1110074520111084,
+      "learning_rate": 2.3766070515102045e-06,
+      "loss": 0.5055,
+      "step": 42106
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.5309455394744873,
+      "learning_rate": 2.3763826028442886e-06,
+      "loss": 0.7091,
+      "step": 42107
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.791186571121216,
+      "learning_rate": 2.3761581614736684e-06,
+      "loss": 0.5652,
+      "step": 42108
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.187998056411743,
+      "learning_rate": 2.3759337273989708e-06,
+      "loss": 0.4162,
+      "step": 42109
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.8560733795166016,
+      "learning_rate": 2.3757093006208165e-06,
+      "loss": 0.6695,
+      "step": 42110
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.059748411178589,
+      "learning_rate": 2.3754848811398333e-06,
+      "loss": 0.6906,
+      "step": 42111
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.3862850666046143,
+      "learning_rate": 2.375260468956642e-06,
+      "loss": 0.5332,
+      "step": 42112
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.869157075881958,
+      "learning_rate": 2.375036064071868e-06,
+      "loss": 0.5041,
+      "step": 42113
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.3364739418029785,
+      "learning_rate": 2.3748116664861354e-06,
+      "loss": 0.6329,
+      "step": 42114
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.1438779830932617,
+      "learning_rate": 2.3745872762000697e-06,
+      "loss": 0.6765,
+      "step": 42115
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 4.037053108215332,
+      "learning_rate": 2.374362893214292e-06,
+      "loss": 0.6689,
+      "step": 42116
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 4.330653667449951,
+      "learning_rate": 2.374138517529429e-06,
+      "loss": 0.4941,
+      "step": 42117
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.1143741607666016,
+      "learning_rate": 2.373914149146101e-06,
+      "loss": 0.4767,
+      "step": 42118
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.8934991359710693,
+      "learning_rate": 2.373689788064936e-06,
+      "loss": 0.829,
+      "step": 42119
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.035818099975586,
+      "learning_rate": 2.373465434286554e-06,
+      "loss": 0.5411,
+      "step": 42120
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 4.324370384216309,
+      "learning_rate": 2.373241087811583e-06,
+      "loss": 0.8297,
+      "step": 42121
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 5.417037010192871,
+      "learning_rate": 2.3730167486406425e-06,
+      "loss": 0.5546,
+      "step": 42122
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.971513271331787,
+      "learning_rate": 2.37279241677436e-06,
+      "loss": 0.8114,
+      "step": 42123
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.142443895339966,
+      "learning_rate": 2.372568092213356e-06,
+      "loss": 0.7027,
+      "step": 42124
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.0023882389068604,
+      "learning_rate": 2.3723437749582555e-06,
+      "loss": 0.4619,
+      "step": 42125
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.085303544998169,
+      "learning_rate": 2.3721194650096845e-06,
+      "loss": 0.6951,
+      "step": 42126
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.196549415588379,
+      "learning_rate": 2.3718951623682634e-06,
+      "loss": 0.5861,
+      "step": 42127
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 4.803333282470703,
+      "learning_rate": 2.371670867034619e-06,
+      "loss": 0.5354,
+      "step": 42128
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 4.119585990905762,
+      "learning_rate": 2.3714465790093715e-06,
+      "loss": 0.4162,
+      "step": 42129
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.3473565578460693,
+      "learning_rate": 2.3712222982931484e-06,
+      "loss": 0.6779,
+      "step": 42130
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.0503594875335693,
+      "learning_rate": 2.370998024886569e-06,
+      "loss": 0.4064,
+      "step": 42131
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.663972854614258,
+      "learning_rate": 2.370773758790261e-06,
+      "loss": 0.4312,
+      "step": 42132
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.902287006378174,
+      "learning_rate": 2.370549500004845e-06,
+      "loss": 0.4347,
+      "step": 42133
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.2022557258605957,
+      "learning_rate": 2.3703252485309458e-06,
+      "loss": 0.7539,
+      "step": 42134
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.3150112628936768,
+      "learning_rate": 2.3701010043691884e-06,
+      "loss": 0.5158,
+      "step": 42135
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.6794943809509277,
+      "learning_rate": 2.3698767675201935e-06,
+      "loss": 0.9011,
+      "step": 42136
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.742459774017334,
+      "learning_rate": 2.369652537984585e-06,
+      "loss": 0.5596,
+      "step": 42137
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.1223385334014893,
+      "learning_rate": 2.3694283157629904e-06,
+      "loss": 0.5987,
+      "step": 42138
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.907913088798523,
+      "learning_rate": 2.3692041008560286e-06,
+      "loss": 0.5418,
+      "step": 42139
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.056729316711426,
+      "learning_rate": 2.3689798932643254e-06,
+      "loss": 0.6996,
+      "step": 42140
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.0836374759674072,
+      "learning_rate": 2.3687556929885026e-06,
+      "loss": 0.5072,
+      "step": 42141
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.7959537506103516,
+      "learning_rate": 2.368531500029186e-06,
+      "loss": 0.4837,
+      "step": 42142
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.7009018659591675,
+      "learning_rate": 2.3683073143869957e-06,
+      "loss": 0.6895,
+      "step": 42143
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.215233564376831,
+      "learning_rate": 2.368083136062559e-06,
+      "loss": 0.7087,
+      "step": 42144
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.495424747467041,
+      "learning_rate": 2.367858965056495e-06,
+      "loss": 0.5302,
+      "step": 42145
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.379204034805298,
+      "learning_rate": 2.3676348013694312e-06,
+      "loss": 0.7341,
+      "step": 42146
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.03688907623291,
+      "learning_rate": 2.3674106450019867e-06,
+      "loss": 0.5727,
+      "step": 42147
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.5002241134643555,
+      "learning_rate": 2.3671864959547893e-06,
+      "loss": 0.8528,
+      "step": 42148
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 4.289061546325684,
+      "learning_rate": 2.3669623542284583e-06,
+      "loss": 0.6079,
+      "step": 42149
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.472383975982666,
+      "learning_rate": 2.36673821982362e-06,
+      "loss": 0.5316,
+      "step": 42150
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.5272607803344727,
+      "learning_rate": 2.3665140927408942e-06,
+      "loss": 0.6064,
+      "step": 42151
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.7569315433502197,
+      "learning_rate": 2.3662899729809084e-06,
+      "loss": 0.5393,
+      "step": 42152
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.233280897140503,
+      "learning_rate": 2.366065860544282e-06,
+      "loss": 0.4343,
+      "step": 42153
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 4.367917537689209,
+      "learning_rate": 2.365841755431641e-06,
+      "loss": 0.9668,
+      "step": 42154
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.983063220977783,
+      "learning_rate": 2.3656176576436043e-06,
+      "loss": 0.5284,
+      "step": 42155
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.410658597946167,
+      "learning_rate": 2.3653935671808008e-06,
+      "loss": 0.4911,
+      "step": 42156
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.652888059616089,
+      "learning_rate": 2.36516948404385e-06,
+      "loss": 0.7042,
+      "step": 42157
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.026315927505493,
+      "learning_rate": 2.3649454082333775e-06,
+      "loss": 0.5943,
+      "step": 42158
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.5626320838928223,
+      "learning_rate": 2.3647213397500024e-06,
+      "loss": 0.7468,
+      "step": 42159
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.3411240577697754,
+      "learning_rate": 2.364497278594352e-06,
+      "loss": 0.6952,
+      "step": 42160
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.172177314758301,
+      "learning_rate": 2.3642732247670457e-06,
+      "loss": 0.5767,
+      "step": 42161
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.2213168144226074,
+      "learning_rate": 2.36404917826871e-06,
+      "loss": 0.6103,
+      "step": 42162
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.104304075241089,
+      "learning_rate": 2.363825139099964e-06,
+      "loss": 0.7112,
+      "step": 42163
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.077948808670044,
+      "learning_rate": 2.363601107261435e-06,
+      "loss": 0.7389,
+      "step": 42164
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.4249353408813477,
+      "learning_rate": 2.3633770827537418e-06,
+      "loss": 0.4843,
+      "step": 42165
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.1819968223571777,
+      "learning_rate": 2.363153065577511e-06,
+      "loss": 0.4991,
+      "step": 42166
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.396310329437256,
+      "learning_rate": 2.362929055733362e-06,
+      "loss": 0.5073,
+      "step": 42167
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.1530325412750244,
+      "learning_rate": 2.362705053221921e-06,
+      "loss": 0.5628,
+      "step": 42168
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.993277072906494,
+      "learning_rate": 2.362481058043807e-06,
+      "loss": 0.4748,
+      "step": 42169
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.454169511795044,
+      "learning_rate": 2.362257070199648e-06,
+      "loss": 0.5272,
+      "step": 42170
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 4.977783679962158,
+      "learning_rate": 2.3620330896900617e-06,
+      "loss": 0.7428,
+      "step": 42171
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.2445616722106934,
+      "learning_rate": 2.361809116515675e-06,
+      "loss": 0.6738,
+      "step": 42172
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.8980207443237305,
+      "learning_rate": 2.3615851506771072e-06,
+      "loss": 0.7058,
+      "step": 42173
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 4.148143768310547,
+      "learning_rate": 2.361361192174984e-06,
+      "loss": 0.6742,
+      "step": 42174
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.1243488788604736,
+      "learning_rate": 2.361137241009924e-06,
+      "loss": 0.5002,
+      "step": 42175
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.1650805473327637,
+      "learning_rate": 2.3609132971825555e-06,
+      "loss": 0.5478,
+      "step": 42176
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.963902473449707,
+      "learning_rate": 2.3606893606934973e-06,
+      "loss": 0.8313,
+      "step": 42177
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.160465955734253,
+      "learning_rate": 2.360465431543375e-06,
+      "loss": 0.5847,
+      "step": 42178
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.0162436962127686,
+      "learning_rate": 2.360241509732808e-06,
+      "loss": 0.7055,
+      "step": 42179
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.3949992656707764,
+      "learning_rate": 2.3600175952624216e-06,
+      "loss": 0.5539,
+      "step": 42180
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.8029848337173462,
+      "learning_rate": 2.3597936881328358e-06,
+      "loss": 0.3073,
+      "step": 42181
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.1613476276397705,
+      "learning_rate": 2.359569788344677e-06,
+      "loss": 0.5632,
+      "step": 42182
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.2226405143737793,
+      "learning_rate": 2.359345895898563e-06,
+      "loss": 0.6129,
+      "step": 42183
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.970642566680908,
+      "learning_rate": 2.3591220107951213e-06,
+      "loss": 0.5068,
+      "step": 42184
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.9257426261901855,
+      "learning_rate": 2.3588981330349696e-06,
+      "loss": 0.657,
+      "step": 42185
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.7914750576019287,
+      "learning_rate": 2.358674262618733e-06,
+      "loss": 0.4611,
+      "step": 42186
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.6950736045837402,
+      "learning_rate": 2.358450399547036e-06,
+      "loss": 0.5878,
+      "step": 42187
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.580005168914795,
+      "learning_rate": 2.3582265438204966e-06,
+      "loss": 0.4775,
+      "step": 42188
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.0129125118255615,
+      "learning_rate": 2.3580026954397417e-06,
+      "loss": 0.5107,
+      "step": 42189
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.8023018836975098,
+      "learning_rate": 2.35777885440539e-06,
+      "loss": 0.5861,
+      "step": 42190
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.348742961883545,
+      "learning_rate": 2.3575550207180662e-06,
+      "loss": 0.4816,
+      "step": 42191
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.6326205730438232,
+      "learning_rate": 2.357331194378391e-06,
+      "loss": 0.7725,
+      "step": 42192
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.953936815261841,
+      "learning_rate": 2.35710737538699e-06,
+      "loss": 0.8097,
+      "step": 42193
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.614212989807129,
+      "learning_rate": 2.356883563744481e-06,
+      "loss": 0.628,
+      "step": 42194
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.4472811222076416,
+      "learning_rate": 2.3566597594514885e-06,
+      "loss": 0.7202,
+      "step": 42195
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 4.445185661315918,
+      "learning_rate": 2.3564359625086367e-06,
+      "loss": 0.4803,
+      "step": 42196
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.3250043392181396,
+      "learning_rate": 2.356212172916545e-06,
+      "loss": 0.4646,
+      "step": 42197
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.4583981037139893,
+      "learning_rate": 2.355988390675837e-06,
+      "loss": 0.6454,
+      "step": 42198
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.5118918418884277,
+      "learning_rate": 2.3557646157871363e-06,
+      "loss": 0.5599,
+      "step": 42199
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.0555622577667236,
+      "learning_rate": 2.355540848251061e-06,
+      "loss": 0.5086,
+      "step": 42200
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.189830780029297,
+      "learning_rate": 2.355317088068239e-06,
+      "loss": 0.4194,
+      "step": 42201
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.6873624324798584,
+      "learning_rate": 2.355093335239287e-06,
+      "loss": 0.6044,
+      "step": 42202
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.856623649597168,
+      "learning_rate": 2.3548695897648316e-06,
+      "loss": 0.547,
+      "step": 42203
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.7658920288085938,
+      "learning_rate": 2.354645851645491e-06,
+      "loss": 0.4623,
+      "step": 42204
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 8.006866455078125,
+      "learning_rate": 2.3544221208818914e-06,
+      "loss": 0.6059,
+      "step": 42205
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.4579575061798096,
+      "learning_rate": 2.354198397474651e-06,
+      "loss": 0.4978,
+      "step": 42206
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.1343836784362793,
+      "learning_rate": 2.353974681424395e-06,
+      "loss": 0.5974,
+      "step": 42207
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.369987726211548,
+      "learning_rate": 2.353750972731743e-06,
+      "loss": 0.6649,
+      "step": 42208
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.4236178398132324,
+      "learning_rate": 2.3535272713973202e-06,
+      "loss": 0.6109,
+      "step": 42209
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.7391703128814697,
+      "learning_rate": 2.3533035774217446e-06,
+      "loss": 0.5231,
+      "step": 42210
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.185436725616455,
+      "learning_rate": 2.353079890805642e-06,
+      "loss": 0.6788,
+      "step": 42211
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.5726146697998047,
+      "learning_rate": 2.352856211549631e-06,
+      "loss": 0.5078,
+      "step": 42212
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.230380058288574,
+      "learning_rate": 2.3526325396543365e-06,
+      "loss": 0.5548,
+      "step": 42213
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.5825512409210205,
+      "learning_rate": 2.3524088751203776e-06,
+      "loss": 0.7229,
+      "step": 42214
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.7953383922576904,
+      "learning_rate": 2.35218521794838e-06,
+      "loss": 0.563,
+      "step": 42215
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.421736717224121,
+      "learning_rate": 2.3519615681389596e-06,
+      "loss": 0.7735,
+      "step": 42216
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.0027499198913574,
+      "learning_rate": 2.351737925692746e-06,
+      "loss": 0.5664,
+      "step": 42217
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.6604881286621094,
+      "learning_rate": 2.351514290610355e-06,
+      "loss": 0.6134,
+      "step": 42218
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.010896682739258,
+      "learning_rate": 2.351290662892412e-06,
+      "loss": 0.4319,
+      "step": 42219
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 4.153412342071533,
+      "learning_rate": 2.3510670425395366e-06,
+      "loss": 0.8405,
+      "step": 42220
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.04838228225708,
+      "learning_rate": 2.3508434295523526e-06,
+      "loss": 0.4986,
+      "step": 42221
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.7292168140411377,
+      "learning_rate": 2.3506198239314788e-06,
+      "loss": 0.4195,
+      "step": 42222
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.385873794555664,
+      "learning_rate": 2.350396225677541e-06,
+      "loss": 0.6808,
+      "step": 42223
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.4600980281829834,
+      "learning_rate": 2.3501726347911563e-06,
+      "loss": 0.6956,
+      "step": 42224
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.747829794883728,
+      "learning_rate": 2.349949051272951e-06,
+      "loss": 0.3024,
+      "step": 42225
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.670311450958252,
+      "learning_rate": 2.349725475123543e-06,
+      "loss": 0.9756,
+      "step": 42226
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.625072717666626,
+      "learning_rate": 2.3495019063435575e-06,
+      "loss": 0.6473,
+      "step": 42227
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.007673978805542,
+      "learning_rate": 2.3492783449336127e-06,
+      "loss": 0.6017,
+      "step": 42228
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.6005468368530273,
+      "learning_rate": 2.3490547908943327e-06,
+      "loss": 0.7513,
+      "step": 42229
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.1771740913391113,
+      "learning_rate": 2.348831244226337e-06,
+      "loss": 0.3302,
+      "step": 42230
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.3026914596557617,
+      "learning_rate": 2.348607704930251e-06,
+      "loss": 0.6061,
+      "step": 42231
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.614776611328125,
+      "learning_rate": 2.348384173006691e-06,
+      "loss": 0.6354,
+      "step": 42232
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.8260124921798706,
+      "learning_rate": 2.3481606484562837e-06,
+      "loss": 0.453,
+      "step": 42233
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.0350492000579834,
+      "learning_rate": 2.3479371312796455e-06,
+      "loss": 0.4437,
+      "step": 42234
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.264807939529419,
+      "learning_rate": 2.347713621477403e-06,
+      "loss": 0.6187,
+      "step": 42235
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.766542911529541,
+      "learning_rate": 2.347490119050172e-06,
+      "loss": 0.7969,
+      "step": 42236
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.1073031425476074,
+      "learning_rate": 2.347266623998581e-06,
+      "loss": 0.7977,
+      "step": 42237
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.799323558807373,
+      "learning_rate": 2.3470431363232455e-06,
+      "loss": 0.2688,
+      "step": 42238
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.756115198135376,
+      "learning_rate": 2.3468196560247912e-06,
+      "loss": 0.5869,
+      "step": 42239
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.600745916366577,
+      "learning_rate": 2.346596183103835e-06,
+      "loss": 0.4836,
+      "step": 42240
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.7942006587982178,
+      "learning_rate": 2.346372717561004e-06,
+      "loss": 0.5844,
+      "step": 42241
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.9068729877471924,
+      "learning_rate": 2.3461492593969136e-06,
+      "loss": 0.5398,
+      "step": 42242
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.278012275695801,
+      "learning_rate": 2.3459258086121904e-06,
+      "loss": 0.6329,
+      "step": 42243
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.953247308731079,
+      "learning_rate": 2.3457023652074513e-06,
+      "loss": 0.5067,
+      "step": 42244
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.2778687477111816,
+      "learning_rate": 2.345478929183321e-06,
+      "loss": 0.7594,
+      "step": 42245
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.2456650733947754,
+      "learning_rate": 2.345255500540418e-06,
+      "loss": 0.3415,
+      "step": 42246
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.573624610900879,
+      "learning_rate": 2.3450320792793667e-06,
+      "loss": 0.5475,
+      "step": 42247
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.489863872528076,
+      "learning_rate": 2.344808665400785e-06,
+      "loss": 0.3879,
+      "step": 42248
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.9226127862930298,
+      "learning_rate": 2.3445852589052953e-06,
+      "loss": 0.5417,
+      "step": 42249
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.4190521240234375,
+      "learning_rate": 2.3443618597935214e-06,
+      "loss": 0.563,
+      "step": 42250
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.170745849609375,
+      "learning_rate": 2.3441384680660808e-06,
+      "loss": 0.5721,
+      "step": 42251
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.5489163398742676,
+      "learning_rate": 2.3439150837235973e-06,
+      "loss": 0.5208,
+      "step": 42252
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.440072536468506,
+      "learning_rate": 2.343691706766689e-06,
+      "loss": 0.8171,
+      "step": 42253
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.651834726333618,
+      "learning_rate": 2.3434683371959817e-06,
+      "loss": 0.6462,
+      "step": 42254
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.9636104106903076,
+      "learning_rate": 2.343244975012092e-06,
+      "loss": 0.6189,
+      "step": 42255
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.1223318576812744,
+      "learning_rate": 2.343021620215642e-06,
+      "loss": 0.5513,
+      "step": 42256
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.1439034938812256,
+      "learning_rate": 2.3427982728072558e-06,
+      "loss": 0.5607,
+      "step": 42257
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.253408670425415,
+      "learning_rate": 2.3425749327875504e-06,
+      "loss": 0.5332,
+      "step": 42258
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.402911424636841,
+      "learning_rate": 2.34235160015715e-06,
+      "loss": 0.6055,
+      "step": 42259
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.9631221294403076,
+      "learning_rate": 2.342128274916673e-06,
+      "loss": 0.9893,
+      "step": 42260
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.8910701274871826,
+      "learning_rate": 2.341904957066741e-06,
+      "loss": 0.4848,
+      "step": 42261
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.7225468158721924,
+      "learning_rate": 2.341681646607978e-06,
+      "loss": 0.5997,
+      "step": 42262
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.8628897666931152,
+      "learning_rate": 2.3414583435410005e-06,
+      "loss": 0.9114,
+      "step": 42263
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.228370428085327,
+      "learning_rate": 2.3412350478664333e-06,
+      "loss": 0.5085,
+      "step": 42264
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.913511276245117,
+      "learning_rate": 2.341011759584893e-06,
+      "loss": 0.815,
+      "step": 42265
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.0246522426605225,
+      "learning_rate": 2.3407884786970052e-06,
+      "loss": 0.6501,
+      "step": 42266
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.2199933528900146,
+      "learning_rate": 2.3405652052033866e-06,
+      "loss": 0.5716,
+      "step": 42267
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.2363407611846924,
+      "learning_rate": 2.3403419391046624e-06,
+      "loss": 0.6038,
+      "step": 42268
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.1472971439361572,
+      "learning_rate": 2.340118680401448e-06,
+      "loss": 0.557,
+      "step": 42269
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.0052952766418457,
+      "learning_rate": 2.3398954290943697e-06,
+      "loss": 0.8219,
+      "step": 42270
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.86681067943573,
+      "learning_rate": 2.3396721851840438e-06,
+      "loss": 0.5135,
+      "step": 42271
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.1283795833587646,
+      "learning_rate": 2.3394489486710946e-06,
+      "loss": 0.6298,
+      "step": 42272
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.831007480621338,
+      "learning_rate": 2.3392257195561396e-06,
+      "loss": 0.4333,
+      "step": 42273
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.9306492805480957,
+      "learning_rate": 2.3390024978398023e-06,
+      "loss": 0.6686,
+      "step": 42274
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.952502727508545,
+      "learning_rate": 2.3387792835227014e-06,
+      "loss": 0.7896,
+      "step": 42275
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.614104747772217,
+      "learning_rate": 2.3385560766054594e-06,
+      "loss": 0.5783,
+      "step": 42276
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.207327127456665,
+      "learning_rate": 2.3383328770886927e-06,
+      "loss": 0.5712,
+      "step": 42277
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.765490770339966,
+      "learning_rate": 2.3381096849730288e-06,
+      "loss": 0.8731,
+      "step": 42278
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.0610158443450928,
+      "learning_rate": 2.3378865002590826e-06,
+      "loss": 0.5925,
+      "step": 42279
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.5486505031585693,
+      "learning_rate": 2.3376633229474794e-06,
+      "loss": 0.6048,
+      "step": 42280
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.301229476928711,
+      "learning_rate": 2.3374401530388345e-06,
+      "loss": 0.5969,
+      "step": 42281
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.2112209796905518,
+      "learning_rate": 2.3372169905337733e-06,
+      "loss": 0.5974,
+      "step": 42282
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.938142776489258,
+      "learning_rate": 2.3369938354329118e-06,
+      "loss": 0.5136,
+      "step": 42283
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.105713129043579,
+      "learning_rate": 2.3367706877368758e-06,
+      "loss": 0.6007,
+      "step": 42284
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.7889931201934814,
+      "learning_rate": 2.3365475474462803e-06,
+      "loss": 0.6172,
+      "step": 42285
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 4.633311748504639,
+      "learning_rate": 2.33632441456175e-06,
+      "loss": 0.8377,
+      "step": 42286
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.8353583812713623,
+      "learning_rate": 2.336101289083902e-06,
+      "loss": 0.5455,
+      "step": 42287
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.492277145385742,
+      "learning_rate": 2.3358781710133603e-06,
+      "loss": 0.6298,
+      "step": 42288
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 4.32298469543457,
+      "learning_rate": 2.3356550603507415e-06,
+      "loss": 0.657,
+      "step": 42289
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.3864989280700684,
+      "learning_rate": 2.33543195709667e-06,
+      "loss": 0.5412,
+      "step": 42290
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.997661590576172,
+      "learning_rate": 2.335208861251762e-06,
+      "loss": 0.3488,
+      "step": 42291
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.94996976852417,
+      "learning_rate": 2.3349857728166416e-06,
+      "loss": 0.7668,
+      "step": 42292
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.226069688796997,
+      "learning_rate": 2.3347626917919258e-06,
+      "loss": 0.6389,
+      "step": 42293
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.0704588890075684,
+      "learning_rate": 2.334539618178238e-06,
+      "loss": 0.4935,
+      "step": 42294
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 5.796767711639404,
+      "learning_rate": 2.3343165519761956e-06,
+      "loss": 0.6159,
+      "step": 42295
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.0465619564056396,
+      "learning_rate": 2.3340934931864225e-06,
+      "loss": 0.558,
+      "step": 42296
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.506195306777954,
+      "learning_rate": 2.333870441809532e-06,
+      "loss": 0.7185,
+      "step": 42297
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.189607858657837,
+      "learning_rate": 2.333647397846153e-06,
+      "loss": 0.5404,
+      "step": 42298
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.859841823577881,
+      "learning_rate": 2.3334243612969004e-06,
+      "loss": 0.7209,
+      "step": 42299
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.237820625305176,
+      "learning_rate": 2.3332013321623975e-06,
+      "loss": 0.4725,
+      "step": 42300
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.99665904045105,
+      "learning_rate": 2.3329783104432606e-06,
+      "loss": 0.7109,
+      "step": 42301
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.1407628059387207,
+      "learning_rate": 2.332755296140114e-06,
+      "loss": 0.3945,
+      "step": 42302
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.8613479137420654,
+      "learning_rate": 2.3325322892535735e-06,
+      "loss": 0.7387,
+      "step": 42303
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 5.448537826538086,
+      "learning_rate": 2.332309289784264e-06,
+      "loss": 0.5797,
+      "step": 42304
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.8336126804351807,
+      "learning_rate": 2.3320862977328006e-06,
+      "loss": 0.4749,
+      "step": 42305
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.348559856414795,
+      "learning_rate": 2.3318633130998075e-06,
+      "loss": 0.6449,
+      "step": 42306
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.590116500854492,
+      "learning_rate": 2.3316403358859013e-06,
+      "loss": 0.4774,
+      "step": 42307
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.9518345594406128,
+      "learning_rate": 2.331417366091706e-06,
+      "loss": 0.5401,
+      "step": 42308
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.757333755493164,
+      "learning_rate": 2.3311944037178375e-06,
+      "loss": 0.5959,
+      "step": 42309
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.8411656618118286,
+      "learning_rate": 2.330971448764917e-06,
+      "loss": 0.6412,
+      "step": 42310
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.306185722351074,
+      "learning_rate": 2.3307485012335673e-06,
+      "loss": 0.6554,
+      "step": 42311
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.226551055908203,
+      "learning_rate": 2.3305255611244044e-06,
+      "loss": 0.811,
+      "step": 42312
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.3006560802459717,
+      "learning_rate": 2.3303026284380513e-06,
+      "loss": 0.5981,
+      "step": 42313
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.984966278076172,
+      "learning_rate": 2.330079703175125e-06,
+      "loss": 0.6327,
+      "step": 42314
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.6167354583740234,
+      "learning_rate": 2.3298567853362485e-06,
+      "loss": 0.5655,
+      "step": 42315
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.9525179862976074,
+      "learning_rate": 2.329633874922038e-06,
+      "loss": 0.6966,
+      "step": 42316
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.5522537231445312,
+      "learning_rate": 2.329410971933116e-06,
+      "loss": 0.7606,
+      "step": 42317
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.0213100910186768,
+      "learning_rate": 2.329188076370103e-06,
+      "loss": 0.4925,
+      "step": 42318
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.781489372253418,
+      "learning_rate": 2.3289651882336155e-06,
+      "loss": 0.5904,
+      "step": 42319
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.025614023208618,
+      "learning_rate": 2.3287423075242775e-06,
+      "loss": 0.4566,
+      "step": 42320
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.5561959743499756,
+      "learning_rate": 2.3285194342427044e-06,
+      "loss": 0.539,
+      "step": 42321
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.2939188480377197,
+      "learning_rate": 2.3282965683895183e-06,
+      "loss": 0.5168,
+      "step": 42322
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.1794207096099854,
+      "learning_rate": 2.32807370996534e-06,
+      "loss": 0.4512,
+      "step": 42323
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.4182252883911133,
+      "learning_rate": 2.327850858970786e-06,
+      "loss": 0.4797,
+      "step": 42324
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.164393901824951,
+      "learning_rate": 2.3276280154064797e-06,
+      "loss": 0.5674,
+      "step": 42325
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.749211072921753,
+      "learning_rate": 2.3274051792730366e-06,
+      "loss": 0.6891,
+      "step": 42326
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.789639711380005,
+      "learning_rate": 2.327182350571081e-06,
+      "loss": 0.7406,
+      "step": 42327
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.1226181983947754,
+      "learning_rate": 2.3269595293012277e-06,
+      "loss": 0.3987,
+      "step": 42328
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.506930112838745,
+      "learning_rate": 2.3267367154641008e-06,
+      "loss": 0.6343,
+      "step": 42329
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.0945351123809814,
+      "learning_rate": 2.3265139090603156e-06,
+      "loss": 0.3416,
+      "step": 42330
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.9472521543502808,
+      "learning_rate": 2.3262911100904955e-06,
+      "loss": 0.3687,
+      "step": 42331
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 8.235569953918457,
+      "learning_rate": 2.326068318555257e-06,
+      "loss": 0.7411,
+      "step": 42332
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.3075482845306396,
+      "learning_rate": 2.325845534455222e-06,
+      "loss": 0.6494,
+      "step": 42333
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.284273147583008,
+      "learning_rate": 2.325622757791007e-06,
+      "loss": 0.7607,
+      "step": 42334
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.0379889011383057,
+      "learning_rate": 2.325399988563235e-06,
+      "loss": 0.5471,
+      "step": 42335
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.522034168243408,
+      "learning_rate": 2.3251772267725224e-06,
+      "loss": 0.6028,
+      "step": 42336
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.084754228591919,
+      "learning_rate": 2.324954472419491e-06,
+      "loss": 0.6135,
+      "step": 42337
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 5.065917491912842,
+      "learning_rate": 2.3247317255047564e-06,
+      "loss": 0.6281,
+      "step": 42338
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.7377171516418457,
+      "learning_rate": 2.3245089860289437e-06,
+      "loss": 0.6831,
+      "step": 42339
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.555574893951416,
+      "learning_rate": 2.3242862539926675e-06,
+      "loss": 0.9454,
+      "step": 42340
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.588744878768921,
+      "learning_rate": 2.324063529396551e-06,
+      "loss": 0.8594,
+      "step": 42341
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.4088499546051025,
+      "learning_rate": 2.3238408122412095e-06,
+      "loss": 0.5863,
+      "step": 42342
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.0841097831726074,
+      "learning_rate": 2.323618102527266e-06,
+      "loss": 0.4667,
+      "step": 42343
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.349210023880005,
+      "learning_rate": 2.3233954002553354e-06,
+      "loss": 0.762,
+      "step": 42344
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.8084003925323486,
+      "learning_rate": 2.3231727054260424e-06,
+      "loss": 0.7502,
+      "step": 42345
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 5.0472331047058105,
+      "learning_rate": 2.322950018040001e-06,
+      "loss": 0.7974,
+      "step": 42346
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.7005796432495117,
+      "learning_rate": 2.3227273380978345e-06,
+      "loss": 0.8296,
+      "step": 42347
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.823551654815674,
+      "learning_rate": 2.322504665600159e-06,
+      "loss": 0.5197,
+      "step": 42348
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.4196534156799316,
+      "learning_rate": 2.322282000547596e-06,
+      "loss": 0.6408,
+      "step": 42349
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.0240297317504883,
+      "learning_rate": 2.322059342940763e-06,
+      "loss": 0.6818,
+      "step": 42350
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 6.249327182769775,
+      "learning_rate": 2.3218366927802805e-06,
+      "loss": 0.6901,
+      "step": 42351
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 9.977884292602539,
+      "learning_rate": 2.3216140500667655e-06,
+      "loss": 0.4126,
+      "step": 42352
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 4.52648401260376,
+      "learning_rate": 2.3213914148008403e-06,
+      "loss": 0.748,
+      "step": 42353
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.108651876449585,
+      "learning_rate": 2.3211687869831206e-06,
+      "loss": 0.6454,
+      "step": 42354
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.5101301670074463,
+      "learning_rate": 2.3209461666142285e-06,
+      "loss": 0.4963,
+      "step": 42355
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.545999526977539,
+      "learning_rate": 2.32072355369478e-06,
+      "loss": 0.4417,
+      "step": 42356
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.614488959312439,
+      "learning_rate": 2.3205009482253973e-06,
+      "loss": 0.6906,
+      "step": 42357
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.3462562561035156,
+      "learning_rate": 2.3202783502066935e-06,
+      "loss": 0.6551,
+      "step": 42358
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.2195351123809814,
+      "learning_rate": 2.3200557596392957e-06,
+      "loss": 0.4854,
+      "step": 42359
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.3966128826141357,
+      "learning_rate": 2.3198331765238173e-06,
+      "loss": 0.5109,
+      "step": 42360
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.769334316253662,
+      "learning_rate": 2.3196106008608808e-06,
+      "loss": 0.3992,
+      "step": 42361
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 5.676971912384033,
+      "learning_rate": 2.319388032651101e-06,
+      "loss": 0.5217,
+      "step": 42362
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.031519651412964,
+      "learning_rate": 2.3191654718951008e-06,
+      "loss": 0.5118,
+      "step": 42363
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.9973688125610352,
+      "learning_rate": 2.3189429185934952e-06,
+      "loss": 0.609,
+      "step": 42364
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.8549224138259888,
+      "learning_rate": 2.318720372746907e-06,
+      "loss": 0.6154,
+      "step": 42365
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.3255391120910645,
+      "learning_rate": 2.318497834355951e-06,
+      "loss": 0.7025,
+      "step": 42366
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.1635191440582275,
+      "learning_rate": 2.31827530342125e-06,
+      "loss": 0.603,
+      "step": 42367
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.2456040382385254,
+      "learning_rate": 2.3180527799434196e-06,
+      "loss": 0.5413,
+      "step": 42368
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.517408609390259,
+      "learning_rate": 2.3178302639230803e-06,
+      "loss": 0.8252,
+      "step": 42369
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.8162620067596436,
+      "learning_rate": 2.3176077553608495e-06,
+      "loss": 0.6019,
+      "step": 42370
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.4058542251586914,
+      "learning_rate": 2.317385254257346e-06,
+      "loss": 0.5449,
+      "step": 42371
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.2162704467773438,
+      "learning_rate": 2.3171627606131913e-06,
+      "loss": 0.6983,
+      "step": 42372
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.1708791255950928,
+      "learning_rate": 2.3169402744289997e-06,
+      "loss": 0.4907,
+      "step": 42373
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.861002206802368,
+      "learning_rate": 2.3167177957053943e-06,
+      "loss": 0.5627,
+      "step": 42374
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.2567386627197266,
+      "learning_rate": 2.31649532444299e-06,
+      "loss": 0.5037,
+      "step": 42375
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.8645339012145996,
+      "learning_rate": 2.3162728606424084e-06,
+      "loss": 0.7408,
+      "step": 42376
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.7248945236206055,
+      "learning_rate": 2.3160504043042646e-06,
+      "loss": 0.6578,
+      "step": 42377
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 6.486734390258789,
+      "learning_rate": 2.3158279554291795e-06,
+      "loss": 0.7726,
+      "step": 42378
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.965691089630127,
+      "learning_rate": 2.3156055140177735e-06,
+      "loss": 0.4793,
+      "step": 42379
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 4.530704498291016,
+      "learning_rate": 2.3153830800706607e-06,
+      "loss": 0.5844,
+      "step": 42380
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.2781277894973755,
+      "learning_rate": 2.3151606535884636e-06,
+      "loss": 0.4854,
+      "step": 42381
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.9345221519470215,
+      "learning_rate": 2.3149382345717973e-06,
+      "loss": 0.7841,
+      "step": 42382
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.142164468765259,
+      "learning_rate": 2.3147158230212817e-06,
+      "loss": 0.7851,
+      "step": 42383
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.845179319381714,
+      "learning_rate": 2.314493418937538e-06,
+      "loss": 0.7431,
+      "step": 42384
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.865678548812866,
+      "learning_rate": 2.3142710223211794e-06,
+      "loss": 0.6217,
+      "step": 42385
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.9599320888519287,
+      "learning_rate": 2.3140486331728298e-06,
+      "loss": 0.3056,
+      "step": 42386
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.5495872497558594,
+      "learning_rate": 2.313826251493102e-06,
+      "loss": 0.6399,
+      "step": 42387
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.1448705196380615,
+      "learning_rate": 2.313603877282619e-06,
+      "loss": 0.713,
+      "step": 42388
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.0459530353546143,
+      "learning_rate": 2.313381510541996e-06,
+      "loss": 0.544,
+      "step": 42389
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.199575185775757,
+      "learning_rate": 2.3131591512718543e-06,
+      "loss": 0.5305,
+      "step": 42390
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.226534128189087,
+      "learning_rate": 2.3129367994728084e-06,
+      "loss": 0.51,
+      "step": 42391
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.214874267578125,
+      "learning_rate": 2.3127144551454804e-06,
+      "loss": 0.7906,
+      "step": 42392
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.865345001220703,
+      "learning_rate": 2.312492118290485e-06,
+      "loss": 0.526,
+      "step": 42393
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.796647787094116,
+      "learning_rate": 2.3122697889084445e-06,
+      "loss": 0.465,
+      "step": 42394
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.870713949203491,
+      "learning_rate": 2.3120474669999733e-06,
+      "loss": 0.4909,
+      "step": 42395
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.020714521408081,
+      "learning_rate": 2.311825152565692e-06,
+      "loss": 0.4918,
+      "step": 42396
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 4.087702751159668,
+      "learning_rate": 2.311602845606217e-06,
+      "loss": 0.6549,
+      "step": 42397
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.8415520191192627,
+      "learning_rate": 2.311380546122169e-06,
+      "loss": 0.7937,
+      "step": 42398
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.9670720100402832,
+      "learning_rate": 2.31115825411416e-06,
+      "loss": 0.4993,
+      "step": 42399
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.259608745574951,
+      "learning_rate": 2.3109359695828172e-06,
+      "loss": 0.4574,
+      "step": 42400
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.6833322048187256,
+      "learning_rate": 2.310713692528752e-06,
+      "loss": 0.8836,
+      "step": 42401
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 4.311830520629883,
+      "learning_rate": 2.3104914229525865e-06,
+      "loss": 0.7028,
+      "step": 42402
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.157008409500122,
+      "learning_rate": 2.3102691608549354e-06,
+      "loss": 0.6631,
+      "step": 42403
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.129992961883545,
+      "learning_rate": 2.3100469062364196e-06,
+      "loss": 0.6474,
+      "step": 42404
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.11230206489563,
+      "learning_rate": 2.3098246590976537e-06,
+      "loss": 0.6053,
+      "step": 42405
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.906066417694092,
+      "learning_rate": 2.3096024194392597e-06,
+      "loss": 0.6464,
+      "step": 42406
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.0137109756469727,
+      "learning_rate": 2.3093801872618514e-06,
+      "loss": 0.6959,
+      "step": 42407
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.960400104522705,
+      "learning_rate": 2.309157962566051e-06,
+      "loss": 0.4288,
+      "step": 42408
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.5344603061676025,
+      "learning_rate": 2.3089357453524724e-06,
+      "loss": 0.675,
+      "step": 42409
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.9195311069488525,
+      "learning_rate": 2.308713535621737e-06,
+      "loss": 0.489,
+      "step": 42410
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.399094581604004,
+      "learning_rate": 2.3084913333744595e-06,
+      "loss": 0.566,
+      "step": 42411
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 4.125487804412842,
+      "learning_rate": 2.3082691386112614e-06,
+      "loss": 0.5973,
+      "step": 42412
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.3910248279571533,
+      "learning_rate": 2.3080469513327564e-06,
+      "loss": 0.525,
+      "step": 42413
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.9046945571899414,
+      "learning_rate": 2.3078247715395663e-06,
+      "loss": 0.6252,
+      "step": 42414
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.1399292945861816,
+      "learning_rate": 2.3076025992323055e-06,
+      "loss": 0.3877,
+      "step": 42415
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.3507144451141357,
+      "learning_rate": 2.3073804344115946e-06,
+      "loss": 0.4057,
+      "step": 42416
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.3300795555114746,
+      "learning_rate": 2.307158277078048e-06,
+      "loss": 0.8359,
+      "step": 42417
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.2145512104034424,
+      "learning_rate": 2.3069361272322875e-06,
+      "loss": 0.741,
+      "step": 42418
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 4.748915672302246,
+      "learning_rate": 2.3067139848749255e-06,
+      "loss": 0.5078,
+      "step": 42419
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.0600972175598145,
+      "learning_rate": 2.3064918500065863e-06,
+      "loss": 0.9496,
+      "step": 42420
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.3963701725006104,
+      "learning_rate": 2.3062697226278826e-06,
+      "loss": 0.7325,
+      "step": 42421
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 4.314637184143066,
+      "learning_rate": 2.3060476027394353e-06,
+      "loss": 0.6689,
+      "step": 42422
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.2816808223724365,
+      "learning_rate": 2.3058254903418586e-06,
+      "loss": 0.593,
+      "step": 42423
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.85096275806427,
+      "learning_rate": 2.305603385435774e-06,
+      "loss": 0.3572,
+      "step": 42424
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.7972447872161865,
+      "learning_rate": 2.3053812880217953e-06,
+      "loss": 0.8932,
+      "step": 42425
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.6471829414367676,
+      "learning_rate": 2.3051591981005432e-06,
+      "loss": 0.6367,
+      "step": 42426
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.942152261734009,
+      "learning_rate": 2.3049371156726325e-06,
+      "loss": 0.6846,
+      "step": 42427
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.5483479499816895,
+      "learning_rate": 2.3047150407386835e-06,
+      "loss": 0.5915,
+      "step": 42428
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.8431077003479004,
+      "learning_rate": 2.304492973299311e-06,
+      "loss": 0.6051,
+      "step": 42429
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.965409994125366,
+      "learning_rate": 2.3042709133551355e-06,
+      "loss": 0.6103,
+      "step": 42430
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.845533609390259,
+      "learning_rate": 2.304048860906771e-06,
+      "loss": 0.5065,
+      "step": 42431
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.7518997192382812,
+      "learning_rate": 2.3038268159548376e-06,
+      "loss": 0.3517,
+      "step": 42432
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.9459127187728882,
+      "learning_rate": 2.303604778499951e-06,
+      "loss": 0.7053,
+      "step": 42433
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.657754421234131,
+      "learning_rate": 2.303382748542729e-06,
+      "loss": 0.64,
+      "step": 42434
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.5632483959198,
+      "learning_rate": 2.3031607260837913e-06,
+      "loss": 0.6307,
+      "step": 42435
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.5744080543518066,
+      "learning_rate": 2.302938711123751e-06,
+      "loss": 0.7226,
+      "step": 42436
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.3160555362701416,
+      "learning_rate": 2.3027167036632303e-06,
+      "loss": 0.4355,
+      "step": 42437
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.1404643058776855,
+      "learning_rate": 2.302494703702842e-06,
+      "loss": 0.6051,
+      "step": 42438
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.9250656366348267,
+      "learning_rate": 2.3022727112432058e-06,
+      "loss": 0.4446,
+      "step": 42439
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.3856070041656494,
+      "learning_rate": 2.30205072628494e-06,
+      "loss": 0.7125,
+      "step": 42440
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.803131103515625,
+      "learning_rate": 2.301828748828659e-06,
+      "loss": 0.6804,
+      "step": 42441
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.3105273246765137,
+      "learning_rate": 2.301606778874983e-06,
+      "loss": 0.714,
+      "step": 42442
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.256067991256714,
+      "learning_rate": 2.301384816424526e-06,
+      "loss": 0.7208,
+      "step": 42443
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.523491144180298,
+      "learning_rate": 2.3011628614779082e-06,
+      "loss": 0.816,
+      "step": 42444
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.0739939212799072,
+      "learning_rate": 2.300940914035744e-06,
+      "loss": 0.3819,
+      "step": 42445
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.1803841590881348,
+      "learning_rate": 2.3007189740986525e-06,
+      "loss": 0.5733,
+      "step": 42446
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.603013277053833,
+      "learning_rate": 2.300497041667252e-06,
+      "loss": 0.5268,
+      "step": 42447
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.471067190170288,
+      "learning_rate": 2.3002751167421556e-06,
+      "loss": 0.6119,
+      "step": 42448
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.66062593460083,
+      "learning_rate": 2.300053199323985e-06,
+      "loss": 0.8618,
+      "step": 42449
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.5629141330718994,
+      "learning_rate": 2.299831289413353e-06,
+      "loss": 0.6562,
+      "step": 42450
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.8728795051574707,
+      "learning_rate": 2.2996093870108804e-06,
+      "loss": 0.5832,
+      "step": 42451
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.3296892642974854,
+      "learning_rate": 2.299387492117181e-06,
+      "loss": 0.4883,
+      "step": 42452
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.272925853729248,
+      "learning_rate": 2.2991656047328755e-06,
+      "loss": 0.7385,
+      "step": 42453
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.659639596939087,
+      "learning_rate": 2.2989437248585762e-06,
+      "loss": 0.5371,
+      "step": 42454
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.476396083831787,
+      "learning_rate": 2.2987218524949046e-06,
+      "loss": 0.7395,
+      "step": 42455
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.545088529586792,
+      "learning_rate": 2.2984999876424734e-06,
+      "loss": 0.6838,
+      "step": 42456
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.7377147674560547,
+      "learning_rate": 2.2982781303019043e-06,
+      "loss": 0.5117,
+      "step": 42457
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.9287210702896118,
+      "learning_rate": 2.2980562804738094e-06,
+      "loss": 0.6948,
+      "step": 42458
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.1213483810424805,
+      "learning_rate": 2.29783443815881e-06,
+      "loss": 0.5908,
+      "step": 42459
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.3666017055511475,
+      "learning_rate": 2.2976126033575164e-06,
+      "loss": 0.6824,
+      "step": 42460
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.270698308944702,
+      "learning_rate": 2.297390776070554e-06,
+      "loss": 0.7418,
+      "step": 42461
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.8857338428497314,
+      "learning_rate": 2.297168956298533e-06,
+      "loss": 0.3611,
+      "step": 42462
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.386141061782837,
+      "learning_rate": 2.2969471440420754e-06,
+      "loss": 0.4613,
+      "step": 42463
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 4.146292686462402,
+      "learning_rate": 2.296725339301793e-06,
+      "loss": 0.4992,
+      "step": 42464
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.747943878173828,
+      "learning_rate": 2.296503542078306e-06,
+      "loss": 0.4554,
+      "step": 42465
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 4.053061485290527,
+      "learning_rate": 2.2962817523722284e-06,
+      "loss": 0.6619,
+      "step": 42466
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.2460055351257324,
+      "learning_rate": 2.2960599701841807e-06,
+      "loss": 0.5255,
+      "step": 42467
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.60452401638031,
+      "learning_rate": 2.2958381955147752e-06,
+      "loss": 0.2753,
+      "step": 42468
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 5.069471836090088,
+      "learning_rate": 2.2956164283646324e-06,
+      "loss": 0.5578,
+      "step": 42469
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.7081422805786133,
+      "learning_rate": 2.295394668734366e-06,
+      "loss": 0.7039,
+      "step": 42470
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.1809136867523193,
+      "learning_rate": 2.295172916624595e-06,
+      "loss": 0.564,
+      "step": 42471
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.2595152854919434,
+      "learning_rate": 2.2949511720359328e-06,
+      "loss": 0.6253,
+      "step": 42472
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.8571972846984863,
+      "learning_rate": 2.2947294349690007e-06,
+      "loss": 0.8696,
+      "step": 42473
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.9386239051818848,
+      "learning_rate": 2.29450770542441e-06,
+      "loss": 0.8255,
+      "step": 42474
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.3651044368743896,
+      "learning_rate": 2.2942859834027827e-06,
+      "loss": 0.5388,
+      "step": 42475
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 5.562702655792236,
+      "learning_rate": 2.29406426890473e-06,
+      "loss": 0.6257,
+      "step": 42476
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.630807638168335,
+      "learning_rate": 2.2938425619308724e-06,
+      "loss": 0.7795,
+      "step": 42477
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.961563229560852,
+      "learning_rate": 2.2936208624818233e-06,
+      "loss": 0.4014,
+      "step": 42478
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.532435417175293,
+      "learning_rate": 2.2933991705582033e-06,
+      "loss": 0.7191,
+      "step": 42479
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.3716907501220703,
+      "learning_rate": 2.2931774861606214e-06,
+      "loss": 0.4693,
+      "step": 42480
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 9.877543449401855,
+      "learning_rate": 2.2929558092897035e-06,
+      "loss": 0.7535,
+      "step": 42481
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.559626340866089,
+      "learning_rate": 2.2927341399460595e-06,
+      "loss": 0.7668,
+      "step": 42482
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.1725199222564697,
+      "learning_rate": 2.2925124781303095e-06,
+      "loss": 0.5346,
+      "step": 42483
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.7525951862335205,
+      "learning_rate": 2.2922908238430658e-06,
+      "loss": 0.7937,
+      "step": 42484
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.867403507232666,
+      "learning_rate": 2.292069177084949e-06,
+      "loss": 0.6818,
+      "step": 42485
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 4.121917724609375,
+      "learning_rate": 2.291847537856572e-06,
+      "loss": 0.8368,
+      "step": 42486
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.386525869369507,
+      "learning_rate": 2.291625906158554e-06,
+      "loss": 0.6391,
+      "step": 42487
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.6845154762268066,
+      "learning_rate": 2.291404281991508e-06,
+      "loss": 0.5831,
+      "step": 42488
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.1273577213287354,
+      "learning_rate": 2.291182665356053e-06,
+      "loss": 0.4533,
+      "step": 42489
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.753697633743286,
+      "learning_rate": 2.2909610562528033e-06,
+      "loss": 0.5005,
+      "step": 42490
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.096518039703369,
+      "learning_rate": 2.290739454682378e-06,
+      "loss": 0.6839,
+      "step": 42491
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.050398349761963,
+      "learning_rate": 2.2905178606453895e-06,
+      "loss": 0.7638,
+      "step": 42492
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.0662784576416016,
+      "learning_rate": 2.290296274142457e-06,
+      "loss": 0.6139,
+      "step": 42493
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.4517998695373535,
+      "learning_rate": 2.290074695174195e-06,
+      "loss": 0.6451,
+      "step": 42494
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.6806212663650513,
+      "learning_rate": 2.2898531237412192e-06,
+      "loss": 0.6014,
+      "step": 42495
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.2672786712646484,
+      "learning_rate": 2.2896315598441483e-06,
+      "loss": 0.3602,
+      "step": 42496
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.443068027496338,
+      "learning_rate": 2.2894100034835953e-06,
+      "loss": 0.6561,
+      "step": 42497
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.193204402923584,
+      "learning_rate": 2.289188454660179e-06,
+      "loss": 0.605,
+      "step": 42498
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.7534699440002441,
+      "learning_rate": 2.288966913374513e-06,
+      "loss": 0.5028,
+      "step": 42499
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.200899362564087,
+      "learning_rate": 2.2887453796272137e-06,
+      "loss": 0.3188,
+      "step": 42500
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 4.454464435577393,
+      "learning_rate": 2.2885238534188995e-06,
+      "loss": 0.8051,
+      "step": 42501
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.673097610473633,
+      "learning_rate": 2.2883023347501836e-06,
+      "loss": 0.6321,
+      "step": 42502
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.6622036695480347,
+      "learning_rate": 2.288080823621685e-06,
+      "loss": 0.4112,
+      "step": 42503
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.793297529220581,
+      "learning_rate": 2.2878593200340155e-06,
+      "loss": 0.7809,
+      "step": 42504
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.532541036605835,
+      "learning_rate": 2.287637823987795e-06,
+      "loss": 0.6562,
+      "step": 42505
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 4.003903388977051,
+      "learning_rate": 2.287416335483636e-06,
+      "loss": 0.7999,
+      "step": 42506
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.428901195526123,
+      "learning_rate": 2.287194854522156e-06,
+      "loss": 0.4637,
+      "step": 42507
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.2479987144470215,
+      "learning_rate": 2.286973381103973e-06,
+      "loss": 0.5,
+      "step": 42508
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 4.5584845542907715,
+      "learning_rate": 2.2867519152296984e-06,
+      "loss": 0.6368,
+      "step": 42509
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.004955291748047,
+      "learning_rate": 2.2865304568999526e-06,
+      "loss": 0.8412,
+      "step": 42510
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.4109368324279785,
+      "learning_rate": 2.2863090061153463e-06,
+      "loss": 0.62,
+      "step": 42511
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.957158327102661,
+      "learning_rate": 2.2860875628765012e-06,
+      "loss": 0.6101,
+      "step": 42512
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.6462864875793457,
+      "learning_rate": 2.2858661271840273e-06,
+      "loss": 0.3899,
+      "step": 42513
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.1651880741119385,
+      "learning_rate": 2.285644699038545e-06,
+      "loss": 0.5246,
+      "step": 42514
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.2089099884033203,
+      "learning_rate": 2.285423278440666e-06,
+      "loss": 0.6659,
+      "step": 42515
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.8972620964050293,
+      "learning_rate": 2.2852018653910104e-06,
+      "loss": 0.6457,
+      "step": 42516
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.4095001220703125,
+      "learning_rate": 2.2849804598901885e-06,
+      "loss": 0.6094,
+      "step": 42517
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.247607469558716,
+      "learning_rate": 2.2847590619388214e-06,
+      "loss": 0.6629,
+      "step": 42518
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.196119546890259,
+      "learning_rate": 2.2845376715375206e-06,
+      "loss": 0.482,
+      "step": 42519
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.9661625623703003,
+      "learning_rate": 2.284316288686905e-06,
+      "loss": 0.4289,
+      "step": 42520
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.587310791015625,
+      "learning_rate": 2.2840949133875844e-06,
+      "loss": 0.7198,
+      "step": 42521
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.3489699363708496,
+      "learning_rate": 2.2838735456401823e-06,
+      "loss": 0.4014,
+      "step": 42522
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.5486805438995361,
+      "learning_rate": 2.2836521854453084e-06,
+      "loss": 0.3114,
+      "step": 42523
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.2257206439971924,
+      "learning_rate": 2.2834308328035828e-06,
+      "loss": 0.4874,
+      "step": 42524
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.863147735595703,
+      "learning_rate": 2.283209487715615e-06,
+      "loss": 0.7543,
+      "step": 42525
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.610809803009033,
+      "learning_rate": 2.2829881501820274e-06,
+      "loss": 0.7513,
+      "step": 42526
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.4667696952819824,
+      "learning_rate": 2.2827668202034293e-06,
+      "loss": 0.5218,
+      "step": 42527
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.6695756912231445,
+      "learning_rate": 2.282545497780441e-06,
+      "loss": 0.6895,
+      "step": 42528
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.8268113136291504,
+      "learning_rate": 2.2823241829136733e-06,
+      "loss": 0.5554,
+      "step": 42529
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.836182713508606,
+      "learning_rate": 2.282102875603746e-06,
+      "loss": 0.6086,
+      "step": 42530
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.735034942626953,
+      "learning_rate": 2.281881575851271e-06,
+      "loss": 0.8027,
+      "step": 42531
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.116905689239502,
+      "learning_rate": 2.281660283656866e-06,
+      "loss": 0.6678,
+      "step": 42532
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.6526505947113037,
+      "learning_rate": 2.281438999021145e-06,
+      "loss": 0.4892,
+      "step": 42533
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.293919801712036,
+      "learning_rate": 2.2812177219447247e-06,
+      "loss": 0.7558,
+      "step": 42534
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.2725002765655518,
+      "learning_rate": 2.2809964524282176e-06,
+      "loss": 0.7876,
+      "step": 42535
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.607473850250244,
+      "learning_rate": 2.280775190472243e-06,
+      "loss": 0.7332,
+      "step": 42536
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.2836544513702393,
+      "learning_rate": 2.2805539360774116e-06,
+      "loss": 0.8032,
+      "step": 42537
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.8411765098571777,
+      "learning_rate": 2.2803326892443435e-06,
+      "loss": 0.4761,
+      "step": 42538
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.543461322784424,
+      "learning_rate": 2.2801114499736486e-06,
+      "loss": 0.5563,
+      "step": 42539
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.08032488822937,
+      "learning_rate": 2.279890218265948e-06,
+      "loss": 0.4698,
+      "step": 42540
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.9049875736236572,
+      "learning_rate": 2.2796689941218486e-06,
+      "loss": 0.538,
+      "step": 42541
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.8854929208755493,
+      "learning_rate": 2.2794477775419754e-06,
+      "loss": 0.6687,
+      "step": 42542
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.853165864944458,
+      "learning_rate": 2.2792265685269366e-06,
+      "loss": 0.598,
+      "step": 42543
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.6202456951141357,
+      "learning_rate": 2.2790053670773517e-06,
+      "loss": 0.4487,
+      "step": 42544
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.9420759677886963,
+      "learning_rate": 2.278784173193831e-06,
+      "loss": 0.5394,
+      "step": 42545
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.5882108211517334,
+      "learning_rate": 2.2785629868769945e-06,
+      "loss": 0.6308,
+      "step": 42546
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.3038315773010254,
+      "learning_rate": 2.278341808127453e-06,
+      "loss": 0.4405,
+      "step": 42547
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.8277345895767212,
+      "learning_rate": 2.2781206369458247e-06,
+      "loss": 0.7268,
+      "step": 42548
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.217245101928711,
+      "learning_rate": 2.277899473332722e-06,
+      "loss": 0.7565,
+      "step": 42549
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.913831114768982,
+      "learning_rate": 2.2776783172887622e-06,
+      "loss": 0.599,
+      "step": 42550
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.3169074058532715,
+      "learning_rate": 2.277457168814558e-06,
+      "loss": 0.5025,
+      "step": 42551
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.495182514190674,
+      "learning_rate": 2.2772360279107276e-06,
+      "loss": 0.5047,
+      "step": 42552
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.017176628112793,
+      "learning_rate": 2.2770148945778814e-06,
+      "loss": 0.5199,
+      "step": 42553
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.0759875774383545,
+      "learning_rate": 2.276793768816639e-06,
+      "loss": 0.6835,
+      "step": 42554
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 4.517948150634766,
+      "learning_rate": 2.276572650627611e-06,
+      "loss": 0.7346,
+      "step": 42555
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.0254933834075928,
+      "learning_rate": 2.276351540011416e-06,
+      "loss": 0.4691,
+      "step": 42556
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.172330379486084,
+      "learning_rate": 2.2761304369686656e-06,
+      "loss": 0.6767,
+      "step": 42557
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.0662572383880615,
+      "learning_rate": 2.2759093414999756e-06,
+      "loss": 0.5861,
+      "step": 42558
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.431758403778076,
+      "learning_rate": 2.2756882536059633e-06,
+      "loss": 0.5467,
+      "step": 42559
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.1790995597839355,
+      "learning_rate": 2.2754671732872398e-06,
+      "loss": 0.4736,
+      "step": 42560
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.5968456268310547,
+      "learning_rate": 2.275246100544421e-06,
+      "loss": 0.9021,
+      "step": 42561
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.0433437824249268,
+      "learning_rate": 2.275025035378124e-06,
+      "loss": 0.5679,
+      "step": 42562
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 4.1594672203063965,
+      "learning_rate": 2.2748039777889597e-06,
+      "loss": 0.5798,
+      "step": 42563
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.156813383102417,
+      "learning_rate": 2.2745829277775472e-06,
+      "loss": 0.4076,
+      "step": 42564
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.7841527462005615,
+      "learning_rate": 2.274361885344496e-06,
+      "loss": 0.8449,
+      "step": 42565
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.0491132736206055,
+      "learning_rate": 2.2741408504904254e-06,
+      "loss": 0.6425,
+      "step": 42566
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.905564069747925,
+      "learning_rate": 2.273919823215946e-06,
+      "loss": 0.7443,
+      "step": 42567
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.2045187950134277,
+      "learning_rate": 2.2736988035216744e-06,
+      "loss": 0.4562,
+      "step": 42568
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.1123995780944824,
+      "learning_rate": 2.2734777914082275e-06,
+      "loss": 0.587,
+      "step": 42569
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.8966385126113892,
+      "learning_rate": 2.273256786876215e-06,
+      "loss": 0.5041,
+      "step": 42570
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.690359115600586,
+      "learning_rate": 2.273035789926256e-06,
+      "loss": 0.51,
+      "step": 42571
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.62764310836792,
+      "learning_rate": 2.2728148005589613e-06,
+      "loss": 0.7175,
+      "step": 42572
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.018935203552246,
+      "learning_rate": 2.2725938187749486e-06,
+      "loss": 0.732,
+      "step": 42573
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.9923489093780518,
+      "learning_rate": 2.272372844574829e-06,
+      "loss": 0.6379,
+      "step": 42574
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.1462087631225586,
+      "learning_rate": 2.27215187795922e-06,
+      "loss": 0.5263,
+      "step": 42575
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.8347524404525757,
+      "learning_rate": 2.2719309189287334e-06,
+      "loss": 0.7073,
+      "step": 42576
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.2077949047088623,
+      "learning_rate": 2.271709967483987e-06,
+      "loss": 0.7397,
+      "step": 42577
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.7458336353302,
+      "learning_rate": 2.271489023625591e-06,
+      "loss": 0.7619,
+      "step": 42578
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.0357019901275635,
+      "learning_rate": 2.271268087354164e-06,
+      "loss": 0.7062,
+      "step": 42579
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.4838709831237793,
+      "learning_rate": 2.2710471586703166e-06,
+      "loss": 0.8475,
+      "step": 42580
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.795513153076172,
+      "learning_rate": 2.270826237574666e-06,
+      "loss": 0.6483,
+      "step": 42581
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.401123046875,
+      "learning_rate": 2.2706053240678217e-06,
+      "loss": 0.5924,
+      "step": 42582
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.6849420070648193,
+      "learning_rate": 2.2703844181504054e-06,
+      "loss": 0.3588,
+      "step": 42583
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 6.517140865325928,
+      "learning_rate": 2.270163519823026e-06,
+      "loss": 0.5039,
+      "step": 42584
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.9255030155181885,
+      "learning_rate": 2.2699426290863004e-06,
+      "loss": 0.5636,
+      "step": 42585
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.890636444091797,
+      "learning_rate": 2.2697217459408398e-06,
+      "loss": 0.5862,
+      "step": 42586
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.573467969894409,
+      "learning_rate": 2.269500870387263e-06,
+      "loss": 0.8974,
+      "step": 42587
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.7942681312561035,
+      "learning_rate": 2.2692800024261784e-06,
+      "loss": 0.5336,
+      "step": 42588
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.64651620388031,
+      "learning_rate": 2.269059142058206e-06,
+      "loss": 0.3336,
+      "step": 42589
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.136288642883301,
+      "learning_rate": 2.2688382892839543e-06,
+      "loss": 0.3218,
+      "step": 42590
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.3460440635681152,
+      "learning_rate": 2.268617444104043e-06,
+      "loss": 0.5789,
+      "step": 42591
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.3882715702056885,
+      "learning_rate": 2.268396606519081e-06,
+      "loss": 0.6434,
+      "step": 42592
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.5686683654785156,
+      "learning_rate": 2.268175776529687e-06,
+      "loss": 0.4074,
+      "step": 42593
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.6382062435150146,
+      "learning_rate": 2.2679549541364706e-06,
+      "loss": 0.3748,
+      "step": 42594
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.7969372272491455,
+      "learning_rate": 2.26773413934005e-06,
+      "loss": 0.6386,
+      "step": 42595
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.9809019565582275,
+      "learning_rate": 2.2675133321410348e-06,
+      "loss": 0.5553,
+      "step": 42596
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.8284740447998047,
+      "learning_rate": 2.267292532540044e-06,
+      "loss": 0.3355,
+      "step": 42597
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.821728229522705,
+      "learning_rate": 2.267071740537687e-06,
+      "loss": 0.762,
+      "step": 42598
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.3727293014526367,
+      "learning_rate": 2.2668509561345815e-06,
+      "loss": 0.5551,
+      "step": 42599
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.2919552326202393,
+      "learning_rate": 2.2666301793313376e-06,
+      "loss": 0.8557,
+      "step": 42600
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.2562825679779053,
+      "learning_rate": 2.2664094101285733e-06,
+      "loss": 0.6506,
+      "step": 42601
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.5208663940429688,
+      "learning_rate": 2.2661886485268963e-06,
+      "loss": 0.6254,
+      "step": 42602
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.8552017211914062,
+      "learning_rate": 2.265967894526929e-06,
+      "loss": 0.4746,
+      "step": 42603
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.8316349983215332,
+      "learning_rate": 2.265747148129278e-06,
+      "loss": 0.5473,
+      "step": 42604
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.7650809288024902,
+      "learning_rate": 2.2655264093345625e-06,
+      "loss": 0.5819,
+      "step": 42605
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.059394598007202,
+      "learning_rate": 2.265305678143392e-06,
+      "loss": 0.7583,
+      "step": 42606
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.298295497894287,
+      "learning_rate": 2.265084954556383e-06,
+      "loss": 0.5989,
+      "step": 42607
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.9306596517562866,
+      "learning_rate": 2.264864238574147e-06,
+      "loss": 0.5717,
+      "step": 42608
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.2986884117126465,
+      "learning_rate": 2.2646435301973003e-06,
+      "loss": 0.5237,
+      "step": 42609
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.095914363861084,
+      "learning_rate": 2.2644228294264535e-06,
+      "loss": 0.5637,
+      "step": 42610
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.999254584312439,
+      "learning_rate": 2.2642021362622244e-06,
+      "loss": 0.6158,
+      "step": 42611
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.5208137035369873,
+      "learning_rate": 2.263981450705222e-06,
+      "loss": 0.6429,
+      "step": 42612
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.0864734649658203,
+      "learning_rate": 2.2637607727560644e-06,
+      "loss": 0.3942,
+      "step": 42613
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.8494272232055664,
+      "learning_rate": 2.2635401024153612e-06,
+      "loss": 0.5918,
+      "step": 42614
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.09143328666687,
+      "learning_rate": 2.2633194396837295e-06,
+      "loss": 0.415,
+      "step": 42615
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.8328098058700562,
+      "learning_rate": 2.2630987845617796e-06,
+      "loss": 0.5152,
+      "step": 42616
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.1427578926086426,
+      "learning_rate": 2.2628781370501284e-06,
+      "loss": 0.383,
+      "step": 42617
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.4906046390533447,
+      "learning_rate": 2.262657497149386e-06,
+      "loss": 0.4105,
+      "step": 42618
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.284952402114868,
+      "learning_rate": 2.2624368648601675e-06,
+      "loss": 0.4078,
+      "step": 42619
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.0153093338012695,
+      "learning_rate": 2.262216240183089e-06,
+      "loss": 0.5786,
+      "step": 42620
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.543041467666626,
+      "learning_rate": 2.261995623118759e-06,
+      "loss": 0.8049,
+      "step": 42621
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.8986138105392456,
+      "learning_rate": 2.2617750136677936e-06,
+      "loss": 0.5944,
+      "step": 42622
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.4232594966888428,
+      "learning_rate": 2.261554411830808e-06,
+      "loss": 0.7093,
+      "step": 42623
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.7381436824798584,
+      "learning_rate": 2.261333817608412e-06,
+      "loss": 0.5278,
+      "step": 42624
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.0123915672302246,
+      "learning_rate": 2.2611132310012228e-06,
+      "loss": 0.4983,
+      "step": 42625
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.9452428817749023,
+      "learning_rate": 2.2608926520098496e-06,
+      "loss": 0.541,
+      "step": 42626
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.4657697677612305,
+      "learning_rate": 2.26067208063491e-06,
+      "loss": 0.7505,
+      "step": 42627
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.899425745010376,
+      "learning_rate": 2.2604515168770125e-06,
+      "loss": 0.8244,
+      "step": 42628
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 4.41438627243042,
+      "learning_rate": 2.2602309607367756e-06,
+      "loss": 0.5828,
+      "step": 42629
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.2520620822906494,
+      "learning_rate": 2.260010412214808e-06,
+      "loss": 0.6888,
+      "step": 42630
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.7302298545837402,
+      "learning_rate": 2.259789871311725e-06,
+      "loss": 0.7373,
+      "step": 42631
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.505073070526123,
+      "learning_rate": 2.259569338028142e-06,
+      "loss": 0.7064,
+      "step": 42632
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.8475472927093506,
+      "learning_rate": 2.2593488123646677e-06,
+      "loss": 0.4893,
+      "step": 42633
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.2287673950195312,
+      "learning_rate": 2.25912829432192e-06,
+      "loss": 0.5117,
+      "step": 42634
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.4943220615386963,
+      "learning_rate": 2.258907783900508e-06,
+      "loss": 0.7797,
+      "step": 42635
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.9842727184295654,
+      "learning_rate": 2.258687281101048e-06,
+      "loss": 0.5741,
+      "step": 42636
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.92899751663208,
+      "learning_rate": 2.25846678592415e-06,
+      "loss": 0.7453,
+      "step": 42637
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.7261245250701904,
+      "learning_rate": 2.258246298370431e-06,
+      "loss": 0.6566,
+      "step": 42638
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.239079475402832,
+      "learning_rate": 2.2580258184405e-06,
+      "loss": 0.4756,
+      "step": 42639
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 4.478928565979004,
+      "learning_rate": 2.2578053461349738e-06,
+      "loss": 0.6316,
+      "step": 42640
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.9152727127075195,
+      "learning_rate": 2.2575848814544614e-06,
+      "loss": 0.5614,
+      "step": 42641
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.936683177947998,
+      "learning_rate": 2.257364424399581e-06,
+      "loss": 0.5364,
+      "step": 42642
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 4.094759941101074,
+      "learning_rate": 2.2571439749709382e-06,
+      "loss": 0.6769,
+      "step": 42643
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 4.311003684997559,
+      "learning_rate": 2.2569235331691547e-06,
+      "loss": 0.9061,
+      "step": 42644
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.2819714546203613,
+      "learning_rate": 2.2567030989948367e-06,
+      "loss": 0.7789,
+      "step": 42645
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.2388622760772705,
+      "learning_rate": 2.2564826724486024e-06,
+      "loss": 0.5656,
+      "step": 42646
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 4.089638710021973,
+      "learning_rate": 2.2562622535310597e-06,
+      "loss": 0.6406,
+      "step": 42647
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.345109462738037,
+      "learning_rate": 2.256041842242826e-06,
+      "loss": 0.6114,
+      "step": 42648
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 4.523830890655518,
+      "learning_rate": 2.25582143858451e-06,
+      "loss": 0.501,
+      "step": 42649
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.3383820056915283,
+      "learning_rate": 2.2556010425567283e-06,
+      "loss": 0.4103,
+      "step": 42650
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.926642656326294,
+      "learning_rate": 2.2553806541600905e-06,
+      "loss": 0.6583,
+      "step": 42651
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.2818174362182617,
+      "learning_rate": 2.2551602733952127e-06,
+      "loss": 0.4054,
+      "step": 42652
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.246181011199951,
+      "learning_rate": 2.2549399002627036e-06,
+      "loss": 0.4506,
+      "step": 42653
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.6626336574554443,
+      "learning_rate": 2.2547195347631806e-06,
+      "loss": 0.5481,
+      "step": 42654
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.1126105785369873,
+      "learning_rate": 2.2544991768972524e-06,
+      "loss": 0.5935,
+      "step": 42655
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.87842059135437,
+      "learning_rate": 2.254278826665535e-06,
+      "loss": 0.8797,
+      "step": 42656
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 4.0555267333984375,
+      "learning_rate": 2.254058484068639e-06,
+      "loss": 0.539,
+      "step": 42657
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.8944931030273438,
+      "learning_rate": 2.253838149107178e-06,
+      "loss": 0.5028,
+      "step": 42658
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.8743195533752441,
+      "learning_rate": 2.2536178217817628e-06,
+      "loss": 0.3433,
+      "step": 42659
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.502427816390991,
+      "learning_rate": 2.25339750209301e-06,
+      "loss": 0.8167,
+      "step": 42660
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 4.97465705871582,
+      "learning_rate": 2.253177190041528e-06,
+      "loss": 0.7621,
+      "step": 42661
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.408615827560425,
+      "learning_rate": 2.2529568856279327e-06,
+      "loss": 0.5078,
+      "step": 42662
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.1037487983703613,
+      "learning_rate": 2.252736588852831e-06,
+      "loss": 0.5332,
+      "step": 42663
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.9869874715805054,
+      "learning_rate": 2.252516299716845e-06,
+      "loss": 0.5082,
+      "step": 42664
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.945896625518799,
+      "learning_rate": 2.252296018220579e-06,
+      "loss": 0.6509,
+      "step": 42665
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.0603578090667725,
+      "learning_rate": 2.2520757443646503e-06,
+      "loss": 0.5894,
+      "step": 42666
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.073436975479126,
+      "learning_rate": 2.251855478149668e-06,
+      "loss": 0.7051,
+      "step": 42667
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.26027774810791,
+      "learning_rate": 2.2516352195762477e-06,
+      "loss": 0.7437,
+      "step": 42668
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.0951340198516846,
+      "learning_rate": 2.251414968644999e-06,
+      "loss": 0.4552,
+      "step": 42669
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.2643115520477295,
+      "learning_rate": 2.251194725356537e-06,
+      "loss": 0.6655,
+      "step": 42670
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.5789785385131836,
+      "learning_rate": 2.2509744897114706e-06,
+      "loss": 0.623,
+      "step": 42671
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.0198049545288086,
+      "learning_rate": 2.250754261710417e-06,
+      "loss": 0.6352,
+      "step": 42672
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.008293867111206,
+      "learning_rate": 2.2505340413539836e-06,
+      "loss": 0.5183,
+      "step": 42673
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.0654332637786865,
+      "learning_rate": 2.2503138286427867e-06,
+      "loss": 0.5011,
+      "step": 42674
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 4.259353160858154,
+      "learning_rate": 2.250093623577435e-06,
+      "loss": 0.5632,
+      "step": 42675
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.7334187030792236,
+      "learning_rate": 2.2498734261585453e-06,
+      "loss": 0.8015,
+      "step": 42676
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.1682233810424805,
+      "learning_rate": 2.249653236386725e-06,
+      "loss": 0.5974,
+      "step": 42677
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.7341465950012207,
+      "learning_rate": 2.2494330542625915e-06,
+      "loss": 0.4829,
+      "step": 42678
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.6888980865478516,
+      "learning_rate": 2.249212879786751e-06,
+      "loss": 0.3453,
+      "step": 42679
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.924778461456299,
+      "learning_rate": 2.2489927129598203e-06,
+      "loss": 0.5702,
+      "step": 42680
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.270051956176758,
+      "learning_rate": 2.2487725537824118e-06,
+      "loss": 0.6968,
+      "step": 42681
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.565706968307495,
+      "learning_rate": 2.2485524022551342e-06,
+      "loss": 0.5243,
+      "step": 42682
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.9362554550170898,
+      "learning_rate": 2.2483322583786018e-06,
+      "loss": 0.4894,
+      "step": 42683
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.4097354412078857,
+      "learning_rate": 2.248112122153428e-06,
+      "loss": 0.668,
+      "step": 42684
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.5821635723114014,
+      "learning_rate": 2.247891993580222e-06,
+      "loss": 0.7017,
+      "step": 42685
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.0926742553710938,
+      "learning_rate": 2.247671872659599e-06,
+      "loss": 0.5132,
+      "step": 42686
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 1.7597769498825073,
+      "learning_rate": 2.247451759392168e-06,
+      "loss": 0.5255,
+      "step": 42687
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.0704798698425293,
+      "learning_rate": 2.247231653778544e-06,
+      "loss": 0.538,
+      "step": 42688
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.6481194496154785,
+      "learning_rate": 2.247011555819336e-06,
+      "loss": 0.4065,
+      "step": 42689
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.2614762783050537,
+      "learning_rate": 2.2467914655151596e-06,
+      "loss": 0.5903,
+      "step": 42690
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.1759567260742188,
+      "learning_rate": 2.2465713828666227e-06,
+      "loss": 0.7748,
+      "step": 42691
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.1566853523254395,
+      "learning_rate": 2.2463513078743395e-06,
+      "loss": 0.7475,
+      "step": 42692
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.161104202270508,
+      "learning_rate": 2.246131240538923e-06,
+      "loss": 0.5547,
+      "step": 42693
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.567235231399536,
+      "learning_rate": 2.245911180860982e-06,
+      "loss": 0.6209,
+      "step": 42694
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.4841701984405518,
+      "learning_rate": 2.2456911288411333e-06,
+      "loss": 0.5176,
+      "step": 42695
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.0751495361328125,
+      "learning_rate": 2.245471084479983e-06,
+      "loss": 0.6264,
+      "step": 42696
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.125648260116577,
+      "learning_rate": 2.245251047778148e-06,
+      "loss": 0.3836,
+      "step": 42697
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 1.9356162548065186,
+      "learning_rate": 2.2450310187362357e-06,
+      "loss": 0.554,
+      "step": 42698
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.422724485397339,
+      "learning_rate": 2.244810997354862e-06,
+      "loss": 0.4615,
+      "step": 42699
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.310245990753174,
+      "learning_rate": 2.2445909836346353e-06,
+      "loss": 0.797,
+      "step": 42700
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.7070305347442627,
+      "learning_rate": 2.24437097757617e-06,
+      "loss": 0.6856,
+      "step": 42701
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.317241907119751,
+      "learning_rate": 2.244150979180076e-06,
+      "loss": 0.3845,
+      "step": 42702
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.3959784507751465,
+      "learning_rate": 2.2439309884469666e-06,
+      "loss": 0.4853,
+      "step": 42703
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.3291015625,
+      "learning_rate": 2.2437110053774495e-06,
+      "loss": 0.6254,
+      "step": 42704
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.7240490913391113,
+      "learning_rate": 2.243491029972143e-06,
+      "loss": 0.609,
+      "step": 42705
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.5551366806030273,
+      "learning_rate": 2.243271062231654e-06,
+      "loss": 0.6454,
+      "step": 42706
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 1.9753795862197876,
+      "learning_rate": 2.2430511021565965e-06,
+      "loss": 0.5789,
+      "step": 42707
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.0886151790618896,
+      "learning_rate": 2.2428311497475796e-06,
+      "loss": 0.7733,
+      "step": 42708
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.514183759689331,
+      "learning_rate": 2.2426112050052184e-06,
+      "loss": 0.6344,
+      "step": 42709
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.3428125381469727,
+      "learning_rate": 2.2423912679301206e-06,
+      "loss": 0.5693,
+      "step": 42710
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 1.829693078994751,
+      "learning_rate": 2.2421713385229017e-06,
+      "loss": 0.5512,
+      "step": 42711
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.1152007579803467,
+      "learning_rate": 2.2419514167841694e-06,
+      "loss": 0.5435,
+      "step": 42712
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 1.7821775674819946,
+      "learning_rate": 2.2417315027145386e-06,
+      "loss": 0.5308,
+      "step": 42713
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.308335542678833,
+      "learning_rate": 2.241511596314617e-06,
+      "loss": 0.6334,
+      "step": 42714
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 1.9468212127685547,
+      "learning_rate": 2.2412916975850206e-06,
+      "loss": 0.4158,
+      "step": 42715
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 1.9597182273864746,
+      "learning_rate": 2.241071806526357e-06,
+      "loss": 0.7383,
+      "step": 42716
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 4.299075603485107,
+      "learning_rate": 2.2408519231392406e-06,
+      "loss": 0.8214,
+      "step": 42717
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.789308786392212,
+      "learning_rate": 2.2406320474242793e-06,
+      "loss": 0.5214,
+      "step": 42718
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.041940689086914,
+      "learning_rate": 2.2404121793820894e-06,
+      "loss": 0.5365,
+      "step": 42719
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 4.999785900115967,
+      "learning_rate": 2.2401923190132768e-06,
+      "loss": 0.5401,
+      "step": 42720
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.889116048812866,
+      "learning_rate": 2.2399724663184573e-06,
+      "loss": 0.6242,
+      "step": 42721
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.122983694076538,
+      "learning_rate": 2.239752621298239e-06,
+      "loss": 0.7156,
+      "step": 42722
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.7838454246520996,
+      "learning_rate": 2.2395327839532364e-06,
+      "loss": 0.8545,
+      "step": 42723
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.8789913654327393,
+      "learning_rate": 2.239312954284055e-06,
+      "loss": 0.4811,
+      "step": 42724
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.1698527336120605,
+      "learning_rate": 2.2390931322913145e-06,
+      "loss": 0.4268,
+      "step": 42725
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.345515489578247,
+      "learning_rate": 2.2388733179756193e-06,
+      "loss": 0.8751,
+      "step": 42726
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.1690831184387207,
+      "learning_rate": 2.2386535113375853e-06,
+      "loss": 0.8022,
+      "step": 42727
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.393510580062866,
+      "learning_rate": 2.2384337123778194e-06,
+      "loss": 0.4995,
+      "step": 42728
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.3905816078186035,
+      "learning_rate": 2.2382139210969366e-06,
+      "loss": 0.816,
+      "step": 42729
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 1.6356925964355469,
+      "learning_rate": 2.2379941374955446e-06,
+      "loss": 0.6907,
+      "step": 42730
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.898543357849121,
+      "learning_rate": 2.237774361574258e-06,
+      "loss": 0.7372,
+      "step": 42731
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.3174691200256348,
+      "learning_rate": 2.237554593333684e-06,
+      "loss": 0.8025,
+      "step": 42732
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.7740445137023926,
+      "learning_rate": 2.2373348327744383e-06,
+      "loss": 0.6476,
+      "step": 42733
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.0246047973632812,
+      "learning_rate": 2.237115079897127e-06,
+      "loss": 0.5952,
+      "step": 42734
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.4587793350219727,
+      "learning_rate": 2.236895334702366e-06,
+      "loss": 0.5997,
+      "step": 42735
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.479668378829956,
+      "learning_rate": 2.2366755971907624e-06,
+      "loss": 0.6898,
+      "step": 42736
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.425119400024414,
+      "learning_rate": 2.23645586736293e-06,
+      "loss": 0.5099,
+      "step": 42737
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.9387576580047607,
+      "learning_rate": 2.236236145219477e-06,
+      "loss": 0.7291,
+      "step": 42738
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 1.7675471305847168,
+      "learning_rate": 2.2360164307610175e-06,
+      "loss": 0.4524,
+      "step": 42739
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.266265630722046,
+      "learning_rate": 2.2357967239881596e-06,
+      "loss": 0.7786,
+      "step": 42740
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.072885513305664,
+      "learning_rate": 2.2355770249015174e-06,
+      "loss": 0.4814,
+      "step": 42741
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.4426512718200684,
+      "learning_rate": 2.2353573335016975e-06,
+      "loss": 0.6082,
+      "step": 42742
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.0893499851226807,
+      "learning_rate": 2.2351376497893137e-06,
+      "loss": 0.5762,
+      "step": 42743
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.4016950130462646,
+      "learning_rate": 2.234917973764976e-06,
+      "loss": 0.423,
+      "step": 42744
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.5037455558776855,
+      "learning_rate": 2.2346983054292975e-06,
+      "loss": 0.4308,
+      "step": 42745
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.066166400909424,
+      "learning_rate": 2.2344786447828854e-06,
+      "loss": 0.7407,
+      "step": 42746
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.7186100482940674,
+      "learning_rate": 2.2342589918263543e-06,
+      "loss": 0.5355,
+      "step": 42747
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.417811393737793,
+      "learning_rate": 2.2340393465603105e-06,
+      "loss": 0.6783,
+      "step": 42748
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 1.933435320854187,
+      "learning_rate": 2.233819708985369e-06,
+      "loss": 0.5649,
+      "step": 42749
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.6703429222106934,
+      "learning_rate": 2.2336000791021374e-06,
+      "loss": 0.6252,
+      "step": 42750
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.248466968536377,
+      "learning_rate": 2.2333804569112286e-06,
+      "loss": 0.3597,
+      "step": 42751
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 1.8625270128250122,
+      "learning_rate": 2.233160842413251e-06,
+      "loss": 0.6283,
+      "step": 42752
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.098538875579834,
+      "learning_rate": 2.2329412356088164e-06,
+      "loss": 0.6667,
+      "step": 42753
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 1.8024300336837769,
+      "learning_rate": 2.232721636498538e-06,
+      "loss": 0.4878,
+      "step": 42754
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.2093722820281982,
+      "learning_rate": 2.232502045083022e-06,
+      "loss": 0.4627,
+      "step": 42755
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.3188650608062744,
+      "learning_rate": 2.232282461362883e-06,
+      "loss": 0.7772,
+      "step": 42756
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 1.8693677186965942,
+      "learning_rate": 2.2320628853387278e-06,
+      "loss": 0.5631,
+      "step": 42757
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.4857006072998047,
+      "learning_rate": 2.23184331701117e-06,
+      "loss": 0.8476,
+      "step": 42758
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 1.94464910030365,
+      "learning_rate": 2.231623756380818e-06,
+      "loss": 0.4106,
+      "step": 42759
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.2038793563842773,
+      "learning_rate": 2.2314042034482846e-06,
+      "loss": 0.7809,
+      "step": 42760
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.274439573287964,
+      "learning_rate": 2.2311846582141777e-06,
+      "loss": 0.5218,
+      "step": 42761
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 1.8051002025604248,
+      "learning_rate": 2.2309651206791104e-06,
+      "loss": 0.6017,
+      "step": 42762
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 4.110780239105225,
+      "learning_rate": 2.2307455908436902e-06,
+      "loss": 0.5648,
+      "step": 42763
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.2670581340789795,
+      "learning_rate": 2.2305260687085305e-06,
+      "loss": 0.5389,
+      "step": 42764
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.6887552738189697,
+      "learning_rate": 2.2303065542742374e-06,
+      "loss": 0.6356,
+      "step": 42765
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 1.9685848951339722,
+      "learning_rate": 2.230087047541427e-06,
+      "loss": 0.6151,
+      "step": 42766
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.6487298011779785,
+      "learning_rate": 2.229867548510706e-06,
+      "loss": 0.5159,
+      "step": 42767
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 7.506643772125244,
+      "learning_rate": 2.2296480571826867e-06,
+      "loss": 0.7009,
+      "step": 42768
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.9423515796661377,
+      "learning_rate": 2.229428573557977e-06,
+      "loss": 0.7418,
+      "step": 42769
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.443411350250244,
+      "learning_rate": 2.2292090976371903e-06,
+      "loss": 0.4736,
+      "step": 42770
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.2378530502319336,
+      "learning_rate": 2.2289896294209336e-06,
+      "loss": 0.7252,
+      "step": 42771
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.1659653186798096,
+      "learning_rate": 2.2287701689098196e-06,
+      "loss": 0.5441,
+      "step": 42772
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.640864372253418,
+      "learning_rate": 2.2285507161044566e-06,
+      "loss": 0.4193,
+      "step": 42773
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.36210298538208,
+      "learning_rate": 2.2283312710054574e-06,
+      "loss": 0.4789,
+      "step": 42774
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.4114537239074707,
+      "learning_rate": 2.228111833613429e-06,
+      "loss": 0.5559,
+      "step": 42775
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 4.466555595397949,
+      "learning_rate": 2.2278924039289846e-06,
+      "loss": 0.7262,
+      "step": 42776
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 4.181646347045898,
+      "learning_rate": 2.2276729819527317e-06,
+      "loss": 0.4265,
+      "step": 42777
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.9209136962890625,
+      "learning_rate": 2.227453567685283e-06,
+      "loss": 0.8053,
+      "step": 42778
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.5738589763641357,
+      "learning_rate": 2.2272341611272457e-06,
+      "loss": 0.6229,
+      "step": 42779
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.2279348373413086,
+      "learning_rate": 2.2270147622792337e-06,
+      "loss": 0.4443,
+      "step": 42780
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.0525951385498047,
+      "learning_rate": 2.226795371141853e-06,
+      "loss": 0.457,
+      "step": 42781
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.2137699127197266,
+      "learning_rate": 2.226575987715717e-06,
+      "loss": 0.55,
+      "step": 42782
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.2513508796691895,
+      "learning_rate": 2.2263566120014325e-06,
+      "loss": 0.7174,
+      "step": 42783
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.0707552433013916,
+      "learning_rate": 2.2261372439996127e-06,
+      "loss": 0.7058,
+      "step": 42784
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.443859815597534,
+      "learning_rate": 2.2259178837108626e-06,
+      "loss": 0.4857,
+      "step": 42785
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 4.221761703491211,
+      "learning_rate": 2.2256985311357997e-06,
+      "loss": 0.8687,
+      "step": 42786
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.5292041301727295,
+      "learning_rate": 2.225479186275028e-06,
+      "loss": 0.7041,
+      "step": 42787
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 1.9246296882629395,
+      "learning_rate": 2.225259849129161e-06,
+      "loss": 0.6171,
+      "step": 42788
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.516573667526245,
+      "learning_rate": 2.225040519698805e-06,
+      "loss": 0.5061,
+      "step": 42789
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.594820499420166,
+      "learning_rate": 2.2248211979845737e-06,
+      "loss": 0.622,
+      "step": 42790
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.7732017040252686,
+      "learning_rate": 2.224601883987073e-06,
+      "loss": 0.6808,
+      "step": 42791
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.392915964126587,
+      "learning_rate": 2.2243825777069172e-06,
+      "loss": 0.7776,
+      "step": 42792
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.357720136642456,
+      "learning_rate": 2.2241632791447115e-06,
+      "loss": 0.8397,
+      "step": 42793
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.17098069190979,
+      "learning_rate": 2.2239439883010693e-06,
+      "loss": 0.6202,
+      "step": 42794
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.887439250946045,
+      "learning_rate": 2.2237247051765976e-06,
+      "loss": 0.4976,
+      "step": 42795
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.7371838092803955,
+      "learning_rate": 2.2235054297719092e-06,
+      "loss": 0.7088,
+      "step": 42796
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.851611375808716,
+      "learning_rate": 2.2232861620876107e-06,
+      "loss": 0.7097,
+      "step": 42797
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.5433716773986816,
+      "learning_rate": 2.2230669021243147e-06,
+      "loss": 0.7247,
+      "step": 42798
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.6906564235687256,
+      "learning_rate": 2.2228476498826275e-06,
+      "loss": 0.7174,
+      "step": 42799
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.735887289047241,
+      "learning_rate": 2.222628405363162e-06,
+      "loss": 0.6056,
+      "step": 42800
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.0488340854644775,
+      "learning_rate": 2.222409168566525e-06,
+      "loss": 0.5798,
+      "step": 42801
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 1.9160064458847046,
+      "learning_rate": 2.2221899394933295e-06,
+      "loss": 0.5671,
+      "step": 42802
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.5682427883148193,
+      "learning_rate": 2.2219707181441815e-06,
+      "loss": 0.6718,
+      "step": 42803
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 4.067669868469238,
+      "learning_rate": 2.221751504519692e-06,
+      "loss": 0.8537,
+      "step": 42804
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.839707374572754,
+      "learning_rate": 2.2215322986204714e-06,
+      "loss": 0.5792,
+      "step": 42805
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 1.947675347328186,
+      "learning_rate": 2.2213131004471295e-06,
+      "loss": 0.6986,
+      "step": 42806
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.9168920516967773,
+      "learning_rate": 2.221093910000274e-06,
+      "loss": 0.6729,
+      "step": 42807
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.3066444396972656,
+      "learning_rate": 2.2208747272805162e-06,
+      "loss": 0.6984,
+      "step": 42808
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.413820505142212,
+      "learning_rate": 2.2206555522884637e-06,
+      "loss": 0.572,
+      "step": 42809
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.6888577938079834,
+      "learning_rate": 2.2204363850247284e-06,
+      "loss": 0.6614,
+      "step": 42810
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.3739876747131348,
+      "learning_rate": 2.2202172254899167e-06,
+      "loss": 0.5124,
+      "step": 42811
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.095571517944336,
+      "learning_rate": 2.2199980736846415e-06,
+      "loss": 0.6359,
+      "step": 42812
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.640641450881958,
+      "learning_rate": 2.219778929609508e-06,
+      "loss": 0.6271,
+      "step": 42813
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 1.7981775999069214,
+      "learning_rate": 2.21955979326513e-06,
+      "loss": 0.3973,
+      "step": 42814
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.179558753967285,
+      "learning_rate": 2.2193406646521127e-06,
+      "loss": 0.5627,
+      "step": 42815
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.1693859100341797,
+      "learning_rate": 2.219121543771067e-06,
+      "loss": 0.5719,
+      "step": 42816
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.451533555984497,
+      "learning_rate": 2.218902430622605e-06,
+      "loss": 0.6021,
+      "step": 42817
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.4256272315979004,
+      "learning_rate": 2.218683325207331e-06,
+      "loss": 0.6131,
+      "step": 42818
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 4.220524311065674,
+      "learning_rate": 2.2184642275258588e-06,
+      "loss": 0.5127,
+      "step": 42819
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.5036392211914062,
+      "learning_rate": 2.218245137578794e-06,
+      "loss": 0.656,
+      "step": 42820
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.750354290008545,
+      "learning_rate": 2.218026055366749e-06,
+      "loss": 0.5844,
+      "step": 42821
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.3502392768859863,
+      "learning_rate": 2.2178069808903295e-06,
+      "loss": 0.6884,
+      "step": 42822
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.4159011840820312,
+      "learning_rate": 2.217587914150149e-06,
+      "loss": 0.5589,
+      "step": 42823
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.1382405757904053,
+      "learning_rate": 2.217368855146812e-06,
+      "loss": 0.3944,
+      "step": 42824
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.8777878284454346,
+      "learning_rate": 2.2171498038809315e-06,
+      "loss": 0.5927,
+      "step": 42825
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.127124071121216,
+      "learning_rate": 2.2169307603531133e-06,
+      "loss": 0.5704,
+      "step": 42826
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.242316484451294,
+      "learning_rate": 2.216711724563968e-06,
+      "loss": 0.5565,
+      "step": 42827
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 5.491335868835449,
+      "learning_rate": 2.216492696514104e-06,
+      "loss": 0.6331,
+      "step": 42828
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.172175884246826,
+      "learning_rate": 2.216273676204134e-06,
+      "loss": 0.5907,
+      "step": 42829
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.9664549827575684,
+      "learning_rate": 2.216054663634662e-06,
+      "loss": 0.7181,
+      "step": 42830
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.8563685417175293,
+      "learning_rate": 2.215835658806301e-06,
+      "loss": 0.6796,
+      "step": 42831
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 1.9228429794311523,
+      "learning_rate": 2.2156166617196555e-06,
+      "loss": 0.3637,
+      "step": 42832
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 4.44430685043335,
+      "learning_rate": 2.2153976723753394e-06,
+      "loss": 0.6372,
+      "step": 42833
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.704909324645996,
+      "learning_rate": 2.2151786907739577e-06,
+      "loss": 0.4544,
+      "step": 42834
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 1.6290239095687866,
+      "learning_rate": 2.2149597169161224e-06,
+      "loss": 0.509,
+      "step": 42835
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.4505176544189453,
+      "learning_rate": 2.2147407508024387e-06,
+      "loss": 0.666,
+      "step": 42836
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 5.25993013381958,
+      "learning_rate": 2.21452179243352e-06,
+      "loss": 0.7341,
+      "step": 42837
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 1.9619543552398682,
+      "learning_rate": 2.2143028418099703e-06,
+      "loss": 0.5807,
+      "step": 42838
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.206315755844116,
+      "learning_rate": 2.2140838989324032e-06,
+      "loss": 0.5572,
+      "step": 42839
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.2209062576293945,
+      "learning_rate": 2.213864963801423e-06,
+      "loss": 0.6503,
+      "step": 42840
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.4357681274414062,
+      "learning_rate": 2.2136460364176427e-06,
+      "loss": 0.6112,
+      "step": 42841
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 4.115105152130127,
+      "learning_rate": 2.213427116781667e-06,
+      "loss": 0.6452,
+      "step": 42842
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.6230902671813965,
+      "learning_rate": 2.2132082048941084e-06,
+      "loss": 0.4202,
+      "step": 42843
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.854757785797119,
+      "learning_rate": 2.212989300755572e-06,
+      "loss": 0.6941,
+      "step": 42844
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.3069026470184326,
+      "learning_rate": 2.2127704043666704e-06,
+      "loss": 0.4243,
+      "step": 42845
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 1.9598637819290161,
+      "learning_rate": 2.212551515728006e-06,
+      "loss": 0.4809,
+      "step": 42846
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 1.954159140586853,
+      "learning_rate": 2.2123326348401957e-06,
+      "loss": 0.6809,
+      "step": 42847
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.955406427383423,
+      "learning_rate": 2.212113761703842e-06,
+      "loss": 0.4957,
+      "step": 42848
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.4284422397613525,
+      "learning_rate": 2.2118948963195576e-06,
+      "loss": 0.6007,
+      "step": 42849
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.108102798461914,
+      "learning_rate": 2.211676038687947e-06,
+      "loss": 0.4833,
+      "step": 42850
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 1.8440303802490234,
+      "learning_rate": 2.2114571888096225e-06,
+      "loss": 0.3221,
+      "step": 42851
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.490013360977173,
+      "learning_rate": 2.2112383466851896e-06,
+      "loss": 0.6115,
+      "step": 42852
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 5.291816711425781,
+      "learning_rate": 2.21101951231526e-06,
+      "loss": 0.6028,
+      "step": 42853
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.6751348972320557,
+      "learning_rate": 2.210800685700438e-06,
+      "loss": 0.5519,
+      "step": 42854
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 1.9744901657104492,
+      "learning_rate": 2.210581866841337e-06,
+      "loss": 0.702,
+      "step": 42855
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.2800252437591553,
+      "learning_rate": 2.21036305573856e-06,
+      "loss": 0.8042,
+      "step": 42856
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 4.874251842498779,
+      "learning_rate": 2.210144252392721e-06,
+      "loss": 0.648,
+      "step": 42857
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.7460360527038574,
+      "learning_rate": 2.2099254568044236e-06,
+      "loss": 0.6575,
+      "step": 42858
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.7764015197753906,
+      "learning_rate": 2.209706668974281e-06,
+      "loss": 0.6722,
+      "step": 42859
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 4.846865653991699,
+      "learning_rate": 2.209487888902896e-06,
+      "loss": 0.6322,
+      "step": 42860
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.450871229171753,
+      "learning_rate": 2.209269116590882e-06,
+      "loss": 0.816,
+      "step": 42861
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.544003963470459,
+      "learning_rate": 2.209050352038843e-06,
+      "loss": 0.8646,
+      "step": 42862
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 1.7620218992233276,
+      "learning_rate": 2.2088315952473915e-06,
+      "loss": 0.4586,
+      "step": 42863
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.1992852687835693,
+      "learning_rate": 2.208612846217132e-06,
+      "loss": 0.6188,
+      "step": 42864
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.6246156692504883,
+      "learning_rate": 2.2083941049486745e-06,
+      "loss": 0.5267,
+      "step": 42865
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.3307509422302246,
+      "learning_rate": 2.2081753714426272e-06,
+      "loss": 0.6315,
+      "step": 42866
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.6386969089508057,
+      "learning_rate": 2.2079566456996005e-06,
+      "loss": 0.7273,
+      "step": 42867
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.6313602924346924,
+      "learning_rate": 2.2077379277201983e-06,
+      "loss": 0.6179,
+      "step": 42868
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.0706398487091064,
+      "learning_rate": 2.2075192175050326e-06,
+      "loss": 0.3345,
+      "step": 42869
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.5843043327331543,
+      "learning_rate": 2.2073005150547076e-06,
+      "loss": 0.5229,
+      "step": 42870
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.9421744346618652,
+      "learning_rate": 2.2070818203698363e-06,
+      "loss": 0.6011,
+      "step": 42871
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 6.054944038391113,
+      "learning_rate": 2.2068631334510226e-06,
+      "loss": 0.5003,
+      "step": 42872
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.331066846847534,
+      "learning_rate": 2.2066444542988776e-06,
+      "loss": 0.7558,
+      "step": 42873
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.261086940765381,
+      "learning_rate": 2.2064257829140056e-06,
+      "loss": 0.5628,
+      "step": 42874
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.853713035583496,
+      "learning_rate": 2.2062071192970196e-06,
+      "loss": 0.5689,
+      "step": 42875
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.512032985687256,
+      "learning_rate": 2.205988463448523e-06,
+      "loss": 0.3559,
+      "step": 42876
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.6568877696990967,
+      "learning_rate": 2.205769815369126e-06,
+      "loss": 0.7852,
+      "step": 42877
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 4.1046295166015625,
+      "learning_rate": 2.2055511750594377e-06,
+      "loss": 0.5305,
+      "step": 42878
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.4965827465057373,
+      "learning_rate": 2.2053325425200634e-06,
+      "loss": 0.3831,
+      "step": 42879
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.306284189224243,
+      "learning_rate": 2.205113917751614e-06,
+      "loss": 0.4241,
+      "step": 42880
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.7308287620544434,
+      "learning_rate": 2.2048953007546942e-06,
+      "loss": 0.605,
+      "step": 42881
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 4.492716312408447,
+      "learning_rate": 2.204676691529915e-06,
+      "loss": 0.5489,
+      "step": 42882
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.647841215133667,
+      "learning_rate": 2.2044580900778813e-06,
+      "loss": 0.653,
+      "step": 42883
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.5952889919281006,
+      "learning_rate": 2.2042394963992042e-06,
+      "loss": 0.8405,
+      "step": 42884
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.7238657474517822,
+      "learning_rate": 2.2040209104944875e-06,
+      "loss": 0.5364,
+      "step": 42885
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.0001919269561768,
+      "learning_rate": 2.203802332364343e-06,
+      "loss": 0.6907,
+      "step": 42886
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.3054897785186768,
+      "learning_rate": 2.203583762009376e-06,
+      "loss": 0.5731,
+      "step": 42887
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.4233505725860596,
+      "learning_rate": 2.203365199430194e-06,
+      "loss": 0.4866,
+      "step": 42888
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 1.715331792831421,
+      "learning_rate": 2.2031466446274057e-06,
+      "loss": 0.5515,
+      "step": 42889
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.2449676990509033,
+      "learning_rate": 2.202928097601621e-06,
+      "loss": 0.5685,
+      "step": 42890
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.8954319953918457,
+      "learning_rate": 2.2027095583534434e-06,
+      "loss": 0.6483,
+      "step": 42891
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.78533935546875,
+      "learning_rate": 2.202491026883485e-06,
+      "loss": 0.4602,
+      "step": 42892
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.1040828227996826,
+      "learning_rate": 2.2022725031923484e-06,
+      "loss": 0.6925,
+      "step": 42893
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.2002811431884766,
+      "learning_rate": 2.2020539872806467e-06,
+      "loss": 0.6552,
+      "step": 42894
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.9767112731933594,
+      "learning_rate": 2.201835479148983e-06,
+      "loss": 0.3493,
+      "step": 42895
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.4523427486419678,
+      "learning_rate": 2.2016169787979675e-06,
+      "loss": 0.4582,
+      "step": 42896
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.928392171859741,
+      "learning_rate": 2.201398486228205e-06,
+      "loss": 0.7167,
+      "step": 42897
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.0402920246124268,
+      "learning_rate": 2.201180001440307e-06,
+      "loss": 0.4534,
+      "step": 42898
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.7737834453582764,
+      "learning_rate": 2.200961524434878e-06,
+      "loss": 0.4772,
+      "step": 42899
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.748260498046875,
+      "learning_rate": 2.2007430552125273e-06,
+      "loss": 0.8076,
+      "step": 42900
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.0377931594848633,
+      "learning_rate": 2.20052459377386e-06,
+      "loss": 0.7001,
+      "step": 42901
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.41085147857666,
+      "learning_rate": 2.200306140119487e-06,
+      "loss": 0.6569,
+      "step": 42902
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.3213515281677246,
+      "learning_rate": 2.2000876942500123e-06,
+      "loss": 0.5586,
+      "step": 42903
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.78812575340271,
+      "learning_rate": 2.199869256166046e-06,
+      "loss": 0.7226,
+      "step": 42904
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.7419958114624023,
+      "learning_rate": 2.1996508258681932e-06,
+      "loss": 0.575,
+      "step": 42905
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.1717917919158936,
+      "learning_rate": 2.1994324033570633e-06,
+      "loss": 0.6305,
+      "step": 42906
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 1.844173550605774,
+      "learning_rate": 2.1992139886332604e-06,
+      "loss": 0.7,
+      "step": 42907
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.8400049209594727,
+      "learning_rate": 2.198995581697397e-06,
+      "loss": 0.7013,
+      "step": 42908
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 1.9429484605789185,
+      "learning_rate": 2.198777182550076e-06,
+      "loss": 0.5302,
+      "step": 42909
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.814396619796753,
+      "learning_rate": 2.198558791191908e-06,
+      "loss": 0.605,
+      "step": 42910
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.538863182067871,
+      "learning_rate": 2.1983404076234966e-06,
+      "loss": 0.5337,
+      "step": 42911
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.307652473449707,
+      "learning_rate": 2.1981220318454534e-06,
+      "loss": 0.69,
+      "step": 42912
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.027827501296997,
+      "learning_rate": 2.1979036638583807e-06,
+      "loss": 0.5444,
+      "step": 42913
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.701787233352661,
+      "learning_rate": 2.1976853036628904e-06,
+      "loss": 0.6765,
+      "step": 42914
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.325420379638672,
+      "learning_rate": 2.1974669512595857e-06,
+      "loss": 0.7983,
+      "step": 42915
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.1331586837768555,
+      "learning_rate": 2.1972486066490774e-06,
+      "loss": 0.4427,
+      "step": 42916
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.9020211696624756,
+      "learning_rate": 2.1970302698319688e-06,
+      "loss": 0.7154,
+      "step": 42917
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.5478875637054443,
+      "learning_rate": 2.196811940808871e-06,
+      "loss": 0.6308,
+      "step": 42918
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.4834506511688232,
+      "learning_rate": 2.196593619580387e-06,
+      "loss": 0.6856,
+      "step": 42919
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.06063175201416,
+      "learning_rate": 2.196375306147128e-06,
+      "loss": 0.3912,
+      "step": 42920
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 4.112982273101807,
+      "learning_rate": 2.196157000509697e-06,
+      "loss": 0.7927,
+      "step": 42921
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 1.9158680438995361,
+      "learning_rate": 2.195938702668706e-06,
+      "loss": 0.6122,
+      "step": 42922
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 4.514766216278076,
+      "learning_rate": 2.1957204126247553e-06,
+      "loss": 0.7391,
+      "step": 42923
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.698803424835205,
+      "learning_rate": 2.195502130378459e-06,
+      "loss": 0.4626,
+      "step": 42924
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.3895583152770996,
+      "learning_rate": 2.195283855930418e-06,
+      "loss": 0.5546,
+      "step": 42925
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.093839406967163,
+      "learning_rate": 2.195065589281244e-06,
+      "loss": 0.6466,
+      "step": 42926
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.253862142562866,
+      "learning_rate": 2.1948473304315384e-06,
+      "loss": 0.5553,
+      "step": 42927
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 1.5198791027069092,
+      "learning_rate": 2.1946290793819147e-06,
+      "loss": 0.5746,
+      "step": 42928
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.335515260696411,
+      "learning_rate": 2.1944108361329753e-06,
+      "loss": 0.6214,
+      "step": 42929
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.09997820854187,
+      "learning_rate": 2.1941926006853296e-06,
+      "loss": 0.5521,
+      "step": 42930
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 1.9833991527557373,
+      "learning_rate": 2.1939743730395816e-06,
+      "loss": 0.4941,
+      "step": 42931
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.7282073497772217,
+      "learning_rate": 2.193756153196342e-06,
+      "loss": 0.6857,
+      "step": 42932
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.910198450088501,
+      "learning_rate": 2.1935379411562133e-06,
+      "loss": 0.5525,
+      "step": 42933
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 1.896315336227417,
+      "learning_rate": 2.193319736919806e-06,
+      "loss": 0.4394,
+      "step": 42934
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.6554157733917236,
+      "learning_rate": 2.193101540487723e-06,
+      "loss": 0.5891,
+      "step": 42935
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.978313684463501,
+      "learning_rate": 2.1928833518605754e-06,
+      "loss": 0.8117,
+      "step": 42936
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.1537818908691406,
+      "learning_rate": 2.192665171038966e-06,
+      "loss": 0.4541,
+      "step": 42937
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.9893665313720703,
+      "learning_rate": 2.1924469980235018e-06,
+      "loss": 0.422,
+      "step": 42938
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.4644932746887207,
+      "learning_rate": 2.192228832814794e-06,
+      "loss": 0.8074,
+      "step": 42939
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.2704036235809326,
+      "learning_rate": 2.1920106754134434e-06,
+      "loss": 0.6863,
+      "step": 42940
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.149772882461548,
+      "learning_rate": 2.191792525820061e-06,
+      "loss": 0.4474,
+      "step": 42941
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.809662342071533,
+      "learning_rate": 2.1915743840352503e-06,
+      "loss": 0.7889,
+      "step": 42942
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.3754043579101562,
+      "learning_rate": 2.1913562500596207e-06,
+      "loss": 0.593,
+      "step": 42943
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.1501259803771973,
+      "learning_rate": 2.1911381238937756e-06,
+      "loss": 0.7538,
+      "step": 42944
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.4784114360809326,
+      "learning_rate": 2.190920005538325e-06,
+      "loss": 0.7847,
+      "step": 42945
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.8705179691314697,
+      "learning_rate": 2.190701894993871e-06,
+      "loss": 0.7144,
+      "step": 42946
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.5689713954925537,
+      "learning_rate": 2.1904837922610256e-06,
+      "loss": 0.8597,
+      "step": 42947
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 1.8963927030563354,
+      "learning_rate": 2.1902656973403903e-06,
+      "loss": 0.5725,
+      "step": 42948
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.1637001037597656,
+      "learning_rate": 2.190047610232573e-06,
+      "loss": 0.7302,
+      "step": 42949
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.5554163455963135,
+      "learning_rate": 2.1898295309381804e-06,
+      "loss": 0.6644,
+      "step": 42950
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.1472628116607666,
+      "learning_rate": 2.1896114594578217e-06,
+      "loss": 0.6031,
+      "step": 42951
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.420682907104492,
+      "learning_rate": 2.1893933957920987e-06,
+      "loss": 0.4632,
+      "step": 42952
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.7017855644226074,
+      "learning_rate": 2.1891753399416217e-06,
+      "loss": 0.6128,
+      "step": 42953
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.423476457595825,
+      "learning_rate": 2.188957291906993e-06,
+      "loss": 0.9147,
+      "step": 42954
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.4566402435302734,
+      "learning_rate": 2.188739251688823e-06,
+      "loss": 0.6051,
+      "step": 42955
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 1.7047427892684937,
+      "learning_rate": 2.188521219287714e-06,
+      "loss": 0.3096,
+      "step": 42956
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.4416074752807617,
+      "learning_rate": 2.1883031947042764e-06,
+      "loss": 0.7748,
+      "step": 42957
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.3208420276641846,
+      "learning_rate": 2.188085177939113e-06,
+      "loss": 0.5734,
+      "step": 42958
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.8193931579589844,
+      "learning_rate": 2.187867168992832e-06,
+      "loss": 0.7742,
+      "step": 42959
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.940364122390747,
+      "learning_rate": 2.1876491678660376e-06,
+      "loss": 0.5291,
+      "step": 42960
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.7897300720214844,
+      "learning_rate": 2.1874311745593397e-06,
+      "loss": 0.6278,
+      "step": 42961
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.0311355590820312,
+      "learning_rate": 2.1872131890733394e-06,
+      "loss": 0.4938,
+      "step": 42962
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.498605489730835,
+      "learning_rate": 2.1869952114086477e-06,
+      "loss": 0.4035,
+      "step": 42963
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.401928663253784,
+      "learning_rate": 2.186777241565867e-06,
+      "loss": 0.8082,
+      "step": 42964
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 4.649927139282227,
+      "learning_rate": 2.186559279545607e-06,
+      "loss": 0.7726,
+      "step": 42965
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.07501220703125,
+      "learning_rate": 2.1863413253484694e-06,
+      "loss": 0.8224,
+      "step": 42966
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.523768663406372,
+      "learning_rate": 2.186123378975064e-06,
+      "loss": 0.6917,
+      "step": 42967
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.8279500007629395,
+      "learning_rate": 2.1859054404259923e-06,
+      "loss": 0.5714,
+      "step": 42968
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.7835259437561035,
+      "learning_rate": 2.185687509701867e-06,
+      "loss": 0.44,
+      "step": 42969
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.0391886234283447,
+      "learning_rate": 2.1854695868032884e-06,
+      "loss": 0.6283,
+      "step": 42970
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.6144418716430664,
+      "learning_rate": 2.1852516717308664e-06,
+      "loss": 0.501,
+      "step": 42971
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.4213874340057373,
+      "learning_rate": 2.1850337644852037e-06,
+      "loss": 0.68,
+      "step": 42972
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.6649246215820312,
+      "learning_rate": 2.184815865066909e-06,
+      "loss": 0.616,
+      "step": 42973
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.1172540187835693,
+      "learning_rate": 2.1845979734765847e-06,
+      "loss": 0.56,
+      "step": 42974
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.4984254837036133,
+      "learning_rate": 2.184380089714841e-06,
+      "loss": 0.6427,
+      "step": 42975
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.0016121864318848,
+      "learning_rate": 2.18416221378228e-06,
+      "loss": 0.6458,
+      "step": 42976
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.2357146739959717,
+      "learning_rate": 2.18394434567951e-06,
+      "loss": 0.665,
+      "step": 42977
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.1022651195526123,
+      "learning_rate": 2.183726485407135e-06,
+      "loss": 0.4021,
+      "step": 42978
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.600229263305664,
+      "learning_rate": 2.1835086329657627e-06,
+      "loss": 0.7104,
+      "step": 42979
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.5055792331695557,
+      "learning_rate": 2.183290788355996e-06,
+      "loss": 0.5147,
+      "step": 42980
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.2628326416015625,
+      "learning_rate": 2.183072951578445e-06,
+      "loss": 0.6308,
+      "step": 42981
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.4001266956329346,
+      "learning_rate": 2.18285512263371e-06,
+      "loss": 0.4986,
+      "step": 42982
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 4.017605781555176,
+      "learning_rate": 2.1826373015224022e-06,
+      "loss": 0.5565,
+      "step": 42983
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.5056684017181396,
+      "learning_rate": 2.182419488245123e-06,
+      "loss": 0.5549,
+      "step": 42984
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.217109441757202,
+      "learning_rate": 2.182201682802481e-06,
+      "loss": 0.4108,
+      "step": 42985
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 4.4981689453125,
+      "learning_rate": 2.181983885195079e-06,
+      "loss": 0.4342,
+      "step": 42986
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.3090121746063232,
+      "learning_rate": 2.1817660954235257e-06,
+      "loss": 0.6372,
+      "step": 42987
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.139423131942749,
+      "learning_rate": 2.181548313488422e-06,
+      "loss": 0.3941,
+      "step": 42988
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 1.6034427881240845,
+      "learning_rate": 2.1813305393903794e-06,
+      "loss": 0.2525,
+      "step": 42989
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.550123691558838,
+      "learning_rate": 2.18111277313e-06,
+      "loss": 0.6033,
+      "step": 42990
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 4.422347068786621,
+      "learning_rate": 2.180895014707891e-06,
+      "loss": 0.8866,
+      "step": 42991
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.6864476203918457,
+      "learning_rate": 2.1806772641246556e-06,
+      "loss": 0.5878,
+      "step": 42992
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 4.1628851890563965,
+      "learning_rate": 2.1804595213809015e-06,
+      "loss": 0.4767,
+      "step": 42993
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.7930667400360107,
+      "learning_rate": 2.1802417864772324e-06,
+      "loss": 0.6052,
+      "step": 42994
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.414167881011963,
+      "learning_rate": 2.180024059414256e-06,
+      "loss": 0.5901,
+      "step": 42995
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.799929618835449,
+      "learning_rate": 2.1798063401925746e-06,
+      "loss": 0.6106,
+      "step": 42996
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.1229686737060547,
+      "learning_rate": 2.1795886288127972e-06,
+      "loss": 0.6079,
+      "step": 42997
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.1165060997009277,
+      "learning_rate": 2.179370925275525e-06,
+      "loss": 0.5052,
+      "step": 42998
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.1301026344299316,
+      "learning_rate": 2.1791532295813677e-06,
+      "loss": 0.7541,
+      "step": 42999
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.933253765106201,
+      "learning_rate": 2.1789355417309264e-06,
+      "loss": 0.7074,
+      "step": 43000
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.732417345046997,
+      "learning_rate": 2.1787178617248094e-06,
+      "loss": 0.4711,
+      "step": 43001
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.9751291275024414,
+      "learning_rate": 2.1785001895636226e-06,
+      "loss": 0.5888,
+      "step": 43002
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.4839537143707275,
+      "learning_rate": 2.1782825252479676e-06,
+      "loss": 0.6443,
+      "step": 43003
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.5426523685455322,
+      "learning_rate": 2.1780648687784538e-06,
+      "loss": 0.7471,
+      "step": 43004
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 1.9194446802139282,
+      "learning_rate": 2.1778472201556833e-06,
+      "loss": 0.3796,
+      "step": 43005
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.944814682006836,
+      "learning_rate": 2.177629579380264e-06,
+      "loss": 0.341,
+      "step": 43006
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.1040842533111572,
+      "learning_rate": 2.177411946452797e-06,
+      "loss": 0.4883,
+      "step": 43007
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 5.919045925140381,
+      "learning_rate": 2.1771943213738928e-06,
+      "loss": 0.6402,
+      "step": 43008
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.6374988555908203,
+      "learning_rate": 2.1769767041441513e-06,
+      "loss": 0.6121,
+      "step": 43009
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.8215272426605225,
+      "learning_rate": 2.1767590947641805e-06,
+      "loss": 0.4963,
+      "step": 43010
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.5760858058929443,
+      "learning_rate": 2.176541493234586e-06,
+      "loss": 0.7543,
+      "step": 43011
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.876892566680908,
+      "learning_rate": 2.176323899555971e-06,
+      "loss": 0.6398,
+      "step": 43012
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.0537328720092773,
+      "learning_rate": 2.176106313728941e-06,
+      "loss": 0.7313,
+      "step": 43013
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.5900723934173584,
+      "learning_rate": 2.1758887357541042e-06,
+      "loss": 0.7139,
+      "step": 43014
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.06156063079834,
+      "learning_rate": 2.17567116563206e-06,
+      "loss": 0.7161,
+      "step": 43015
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.5311999320983887,
+      "learning_rate": 2.1754536033634188e-06,
+      "loss": 0.5688,
+      "step": 43016
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.486130475997925,
+      "learning_rate": 2.1752360489487805e-06,
+      "loss": 0.6846,
+      "step": 43017
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.0042831897735596,
+      "learning_rate": 2.175018502388755e-06,
+      "loss": 0.4335,
+      "step": 43018
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 4.1760687828063965,
+      "learning_rate": 2.174800963683943e-06,
+      "loss": 0.6257,
+      "step": 43019
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.8643548488616943,
+      "learning_rate": 2.1745834328349525e-06,
+      "loss": 0.8066,
+      "step": 43020
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 1.994994878768921,
+      "learning_rate": 2.1743659098423855e-06,
+      "loss": 0.5672,
+      "step": 43021
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.2618534564971924,
+      "learning_rate": 2.1741483947068503e-06,
+      "loss": 0.3528,
+      "step": 43022
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.4189984798431396,
+      "learning_rate": 2.1739308874289476e-06,
+      "loss": 0.3448,
+      "step": 43023
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.366098165512085,
+      "learning_rate": 2.1737133880092863e-06,
+      "loss": 0.5626,
+      "step": 43024
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 1.9526002407073975,
+      "learning_rate": 2.1734958964484683e-06,
+      "loss": 0.6796,
+      "step": 43025
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.8366470336914062,
+      "learning_rate": 2.173278412747101e-06,
+      "loss": 0.5022,
+      "step": 43026
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.105013132095337,
+      "learning_rate": 2.1730609369057846e-06,
+      "loss": 0.7843,
+      "step": 43027
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.1602864265441895,
+      "learning_rate": 2.1728434689251293e-06,
+      "loss": 0.5998,
+      "step": 43028
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.1430370807647705,
+      "learning_rate": 2.172626008805734e-06,
+      "loss": 0.4779,
+      "step": 43029
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.5387449264526367,
+      "learning_rate": 2.1724085565482093e-06,
+      "loss": 0.6554,
+      "step": 43030
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.6247718334198,
+      "learning_rate": 2.1721911121531563e-06,
+      "loss": 0.6939,
+      "step": 43031
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 4.140024185180664,
+      "learning_rate": 2.171973675621182e-06,
+      "loss": 0.4483,
+      "step": 43032
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.555011034011841,
+      "learning_rate": 2.1717562469528875e-06,
+      "loss": 0.6276,
+      "step": 43033
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.6245741844177246,
+      "learning_rate": 2.171538826148881e-06,
+      "loss": 0.6793,
+      "step": 43034
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.8701324462890625,
+      "learning_rate": 2.1713214132097642e-06,
+      "loss": 0.5652,
+      "step": 43035
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.360265016555786,
+      "learning_rate": 2.1711040081361444e-06,
+      "loss": 0.3199,
+      "step": 43036
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.1747310161590576,
+      "learning_rate": 2.170886610928623e-06,
+      "loss": 0.6221,
+      "step": 43037
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.4010088443756104,
+      "learning_rate": 2.1706692215878077e-06,
+      "loss": 0.6326,
+      "step": 43038
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 4.8950724601745605,
+      "learning_rate": 2.1704518401143e-06,
+      "loss": 0.6123,
+      "step": 43039
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.4436960220336914,
+      "learning_rate": 2.170234466508707e-06,
+      "loss": 0.6725,
+      "step": 43040
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.3105194568634033,
+      "learning_rate": 2.1700171007716302e-06,
+      "loss": 0.8103,
+      "step": 43041
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.8150413036346436,
+      "learning_rate": 2.1697997429036775e-06,
+      "loss": 0.6172,
+      "step": 43042
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.952212333679199,
+      "learning_rate": 2.16958239290545e-06,
+      "loss": 0.4619,
+      "step": 43043
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.774165153503418,
+      "learning_rate": 2.1693650507775555e-06,
+      "loss": 0.5418,
+      "step": 43044
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 1.8201169967651367,
+      "learning_rate": 2.1691477165205945e-06,
+      "loss": 0.5888,
+      "step": 43045
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.216986656188965,
+      "learning_rate": 2.168930390135174e-06,
+      "loss": 0.4252,
+      "step": 43046
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.132843017578125,
+      "learning_rate": 2.1687130716218973e-06,
+      "loss": 0.6324,
+      "step": 43047
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.0557382106781006,
+      "learning_rate": 2.16849576098137e-06,
+      "loss": 0.4759,
+      "step": 43048
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.3183484077453613,
+      "learning_rate": 2.168278458214192e-06,
+      "loss": 0.4784,
+      "step": 43049
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.3819541931152344,
+      "learning_rate": 2.1680611633209737e-06,
+      "loss": 0.5608,
+      "step": 43050
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.5130462646484375,
+      "learning_rate": 2.167843876302315e-06,
+      "loss": 0.5093,
+      "step": 43051
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.7051339149475098,
+      "learning_rate": 2.1676265971588234e-06,
+      "loss": 0.5426,
+      "step": 43052
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.4603750705718994,
+      "learning_rate": 2.1674093258910995e-06,
+      "loss": 0.8424,
+      "step": 43053
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.591155767440796,
+      "learning_rate": 2.167192062499751e-06,
+      "loss": 0.58,
+      "step": 43054
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.5070137977600098,
+      "learning_rate": 2.1669748069853785e-06,
+      "loss": 0.4242,
+      "step": 43055
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 4.783707618713379,
+      "learning_rate": 2.166757559348589e-06,
+      "loss": 0.6592,
+      "step": 43056
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.9703967571258545,
+      "learning_rate": 2.166540319589984e-06,
+      "loss": 0.625,
+      "step": 43057
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.7809088230133057,
+      "learning_rate": 2.1663230877101705e-06,
+      "loss": 0.7355,
+      "step": 43058
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.254438877105713,
+      "learning_rate": 2.1661058637097493e-06,
+      "loss": 0.8645,
+      "step": 43059
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.910398006439209,
+      "learning_rate": 2.165888647589328e-06,
+      "loss": 0.438,
+      "step": 43060
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.79079532623291,
+      "learning_rate": 2.165671439349507e-06,
+      "loss": 0.6205,
+      "step": 43061
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.9409303665161133,
+      "learning_rate": 2.165454238990892e-06,
+      "loss": 0.4766,
+      "step": 43062
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 1.9311214685440063,
+      "learning_rate": 2.165237046514088e-06,
+      "loss": 0.5446,
+      "step": 43063
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.592040777206421,
+      "learning_rate": 2.1650198619196965e-06,
+      "loss": 0.5743,
+      "step": 43064
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.7629055976867676,
+      "learning_rate": 2.1648026852083247e-06,
+      "loss": 0.8953,
+      "step": 43065
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 1.9126088619232178,
+      "learning_rate": 2.1645855163805723e-06,
+      "loss": 0.2687,
+      "step": 43066
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.5854380130767822,
+      "learning_rate": 2.1643683554370475e-06,
+      "loss": 0.6072,
+      "step": 43067
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.5483908653259277,
+      "learning_rate": 2.1641512023783508e-06,
+      "loss": 0.6643,
+      "step": 43068
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.406378984451294,
+      "learning_rate": 2.163934057205088e-06,
+      "loss": 0.7046,
+      "step": 43069
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.9299893379211426,
+      "learning_rate": 2.1637169199178616e-06,
+      "loss": 0.5592,
+      "step": 43070
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.9069087505340576,
+      "learning_rate": 2.163499790517275e-06,
+      "loss": 0.6093,
+      "step": 43071
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.8217756748199463,
+      "learning_rate": 2.1632826690039347e-06,
+      "loss": 0.6849,
+      "step": 43072
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.01419997215271,
+      "learning_rate": 2.1630655553784414e-06,
+      "loss": 0.6925,
+      "step": 43073
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.480583667755127,
+      "learning_rate": 2.1628484496413998e-06,
+      "loss": 0.5743,
+      "step": 43074
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.1597437858581543,
+      "learning_rate": 2.1626313517934155e-06,
+      "loss": 0.6709,
+      "step": 43075
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.8996825218200684,
+      "learning_rate": 2.162414261835089e-06,
+      "loss": 0.6593,
+      "step": 43076
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.3392670154571533,
+      "learning_rate": 2.162197179767027e-06,
+      "loss": 0.7991,
+      "step": 43077
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.126776933670044,
+      "learning_rate": 2.1619801055898296e-06,
+      "loss": 0.3785,
+      "step": 43078
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.0419301986694336,
+      "learning_rate": 2.1617630393041044e-06,
+      "loss": 0.5837,
+      "step": 43079
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.7809338569641113,
+      "learning_rate": 2.1615459809104515e-06,
+      "loss": 0.4727,
+      "step": 43080
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 4.150852680206299,
+      "learning_rate": 2.1613289304094774e-06,
+      "loss": 0.5058,
+      "step": 43081
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.765913724899292,
+      "learning_rate": 2.1611118878017823e-06,
+      "loss": 0.5843,
+      "step": 43082
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.0205230712890625,
+      "learning_rate": 2.1608948530879735e-06,
+      "loss": 0.5762,
+      "step": 43083
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 1.7087733745574951,
+      "learning_rate": 2.1606778262686506e-06,
+      "loss": 0.4161,
+      "step": 43084
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.547745704650879,
+      "learning_rate": 2.160460807344421e-06,
+      "loss": 0.855,
+      "step": 43085
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.4916632175445557,
+      "learning_rate": 2.160243796315884e-06,
+      "loss": 0.5792,
+      "step": 43086
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 4.343216896057129,
+      "learning_rate": 2.160026793183648e-06,
+      "loss": 0.7046,
+      "step": 43087
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.7310903072357178,
+      "learning_rate": 2.159809797948311e-06,
+      "loss": 0.5771,
+      "step": 43088
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.6185781955718994,
+      "learning_rate": 2.159592810610481e-06,
+      "loss": 0.5604,
+      "step": 43089
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.1207966804504395,
+      "learning_rate": 2.159375831170755e-06,
+      "loss": 0.509,
+      "step": 43090
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.059596300125122,
+      "learning_rate": 2.1591588596297452e-06,
+      "loss": 0.8479,
+      "step": 43091
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.484966278076172,
+      "learning_rate": 2.158941895988048e-06,
+      "loss": 0.5733,
+      "step": 43092
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.945577621459961,
+      "learning_rate": 2.158724940246272e-06,
+      "loss": 0.3778,
+      "step": 43093
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.0917766094207764,
+      "learning_rate": 2.1585079924050147e-06,
+      "loss": 0.5766,
+      "step": 43094
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.2859132289886475,
+      "learning_rate": 2.158291052464884e-06,
+      "loss": 0.5963,
+      "step": 43095
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.4726288318634033,
+      "learning_rate": 2.1580741204264805e-06,
+      "loss": 0.6161,
+      "step": 43096
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 4.183126926422119,
+      "learning_rate": 2.1578571962904093e-06,
+      "loss": 0.4148,
+      "step": 43097
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.1728203296661377,
+      "learning_rate": 2.157640280057271e-06,
+      "loss": 0.4512,
+      "step": 43098
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.214794397354126,
+      "learning_rate": 2.157423371727672e-06,
+      "loss": 0.6334,
+      "step": 43099
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.7676680088043213,
+      "learning_rate": 2.1572064713022118e-06,
+      "loss": 0.3834,
+      "step": 43100
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.6232948303222656,
+      "learning_rate": 2.1569895787814975e-06,
+      "loss": 0.782,
+      "step": 43101
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.0845744609832764,
+      "learning_rate": 2.1567726941661282e-06,
+      "loss": 0.4774,
+      "step": 43102
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.153050184249878,
+      "learning_rate": 2.156555817456711e-06,
+      "loss": 0.5402,
+      "step": 43103
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.7317216396331787,
+      "learning_rate": 2.156338948653845e-06,
+      "loss": 0.6288,
+      "step": 43104
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.288712978363037,
+      "learning_rate": 2.156122087758137e-06,
+      "loss": 0.6202,
+      "step": 43105
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.910273790359497,
+      "learning_rate": 2.155905234770186e-06,
+      "loss": 0.8284,
+      "step": 43106
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.6924145221710205,
+      "learning_rate": 2.1556883896905988e-06,
+      "loss": 0.4831,
+      "step": 43107
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.7913808822631836,
+      "learning_rate": 2.1554715525199754e-06,
+      "loss": 0.7852,
+      "step": 43108
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.794689893722534,
+      "learning_rate": 2.1552547232589216e-06,
+      "loss": 0.6335,
+      "step": 43109
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.6667425632476807,
+      "learning_rate": 2.1550379019080353e-06,
+      "loss": 0.6164,
+      "step": 43110
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.605973482131958,
+      "learning_rate": 2.1548210884679268e-06,
+      "loss": 0.5225,
+      "step": 43111
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.0386860370635986,
+      "learning_rate": 2.1546042829391924e-06,
+      "loss": 0.5652,
+      "step": 43112
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.353900194168091,
+      "learning_rate": 2.1543874853224395e-06,
+      "loss": 0.5354,
+      "step": 43113
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.8127291202545166,
+      "learning_rate": 2.1541706956182677e-06,
+      "loss": 0.5026,
+      "step": 43114
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.3310298919677734,
+      "learning_rate": 2.1539539138272823e-06,
+      "loss": 0.7711,
+      "step": 43115
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.1952614784240723,
+      "learning_rate": 2.1537371399500838e-06,
+      "loss": 0.6558,
+      "step": 43116
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.641800880432129,
+      "learning_rate": 2.1535203739872778e-06,
+      "loss": 0.4394,
+      "step": 43117
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.1215853691101074,
+      "learning_rate": 2.153303615939463e-06,
+      "loss": 0.531,
+      "step": 43118
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.6956090927124023,
+      "learning_rate": 2.153086865807247e-06,
+      "loss": 0.6235,
+      "step": 43119
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.9742836952209473,
+      "learning_rate": 2.1528701235912276e-06,
+      "loss": 0.4716,
+      "step": 43120
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 1.792365312576294,
+      "learning_rate": 2.1526533892920116e-06,
+      "loss": 0.3527,
+      "step": 43121
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.3199570178985596,
+      "learning_rate": 2.1524366629101976e-06,
+      "loss": 0.6102,
+      "step": 43122
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.1833417415618896,
+      "learning_rate": 2.152219944446393e-06,
+      "loss": 0.6016,
+      "step": 43123
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.9176599979400635,
+      "learning_rate": 2.1520032339011966e-06,
+      "loss": 0.5794,
+      "step": 43124
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.1442081928253174,
+      "learning_rate": 2.151786531275212e-06,
+      "loss": 0.6294,
+      "step": 43125
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.685238838195801,
+      "learning_rate": 2.1515698365690436e-06,
+      "loss": 0.7207,
+      "step": 43126
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.0656392574310303,
+      "learning_rate": 2.15135314978329e-06,
+      "loss": 0.4598,
+      "step": 43127
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 1.6685038805007935,
+      "learning_rate": 2.151136470918559e-06,
+      "loss": 0.2564,
+      "step": 43128
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.3214120864868164,
+      "learning_rate": 2.150919799975448e-06,
+      "loss": 0.6212,
+      "step": 43129
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 1.5097019672393799,
+      "learning_rate": 2.1507031369545637e-06,
+      "loss": 0.2651,
+      "step": 43130
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 1.7907665967941284,
+      "learning_rate": 2.150486481856505e-06,
+      "loss": 0.667,
+      "step": 43131
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.947307586669922,
+      "learning_rate": 2.1502698346818755e-06,
+      "loss": 0.442,
+      "step": 43132
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.319896697998047,
+      "learning_rate": 2.1500531954312805e-06,
+      "loss": 0.6895,
+      "step": 43133
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.4185564517974854,
+      "learning_rate": 2.1498365641053178e-06,
+      "loss": 0.7288,
+      "step": 43134
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 4.190271854400635,
+      "learning_rate": 2.1496199407045913e-06,
+      "loss": 0.5569,
+      "step": 43135
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.8527073860168457,
+      "learning_rate": 2.1494033252297062e-06,
+      "loss": 0.6793,
+      "step": 43136
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 1.9855852127075195,
+      "learning_rate": 2.1491867176812615e-06,
+      "loss": 0.2914,
+      "step": 43137
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.454990863800049,
+      "learning_rate": 2.1489701180598614e-06,
+      "loss": 0.6346,
+      "step": 43138
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.525116443634033,
+      "learning_rate": 2.1487535263661065e-06,
+      "loss": 0.6806,
+      "step": 43139
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.4614131450653076,
+      "learning_rate": 2.148536942600601e-06,
+      "loss": 0.5765,
+      "step": 43140
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.711294412612915,
+      "learning_rate": 2.1483203667639447e-06,
+      "loss": 0.6493,
+      "step": 43141
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.9661834239959717,
+      "learning_rate": 2.1481037988567427e-06,
+      "loss": 0.528,
+      "step": 43142
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.9678761959075928,
+      "learning_rate": 2.147887238879594e-06,
+      "loss": 0.5504,
+      "step": 43143
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.7214269638061523,
+      "learning_rate": 2.1476706868331043e-06,
+      "loss": 0.4893,
+      "step": 43144
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.1610443592071533,
+      "learning_rate": 2.147454142717872e-06,
+      "loss": 0.4778,
+      "step": 43145
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.0030603408813477,
+      "learning_rate": 2.147237606534503e-06,
+      "loss": 0.4211,
+      "step": 43146
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.0649666786193848,
+      "learning_rate": 2.1470210782835955e-06,
+      "loss": 0.522,
+      "step": 43147
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.3992040157318115,
+      "learning_rate": 2.146804557965756e-06,
+      "loss": 0.5026,
+      "step": 43148
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.375652551651001,
+      "learning_rate": 2.1465880455815818e-06,
+      "loss": 0.5436,
+      "step": 43149
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.027869462966919,
+      "learning_rate": 2.1463715411316792e-06,
+      "loss": 0.4659,
+      "step": 43150
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.595081329345703,
+      "learning_rate": 2.1461550446166453e-06,
+      "loss": 0.7316,
+      "step": 43151
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.2157578468322754,
+      "learning_rate": 2.1459385560370884e-06,
+      "loss": 0.6437,
+      "step": 43152
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.2557547092437744,
+      "learning_rate": 2.1457220753936054e-06,
+      "loss": 0.6583,
+      "step": 43153
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.6757359504699707,
+      "learning_rate": 2.145505602686802e-06,
+      "loss": 0.673,
+      "step": 43154
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.633843421936035,
+      "learning_rate": 2.1452891379172764e-06,
+      "loss": 0.9331,
+      "step": 43155
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.3586223125457764,
+      "learning_rate": 2.1450726810856337e-06,
+      "loss": 0.6678,
+      "step": 43156
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.618706226348877,
+      "learning_rate": 2.144856232192473e-06,
+      "loss": 0.7291,
+      "step": 43157
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.0879299640655518,
+      "learning_rate": 2.1446397912383993e-06,
+      "loss": 0.6957,
+      "step": 43158
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.5137016773223877,
+      "learning_rate": 2.144423358224011e-06,
+      "loss": 0.7268,
+      "step": 43159
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.3003430366516113,
+      "learning_rate": 2.144206933149913e-06,
+      "loss": 0.5526,
+      "step": 43160
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.262134552001953,
+      "learning_rate": 2.143990516016705e-06,
+      "loss": 0.7894,
+      "step": 43161
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 1.9789276123046875,
+      "learning_rate": 2.14377410682499e-06,
+      "loss": 0.832,
+      "step": 43162
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.6381731033325195,
+      "learning_rate": 2.1435577055753678e-06,
+      "loss": 0.6891,
+      "step": 43163
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.12052059173584,
+      "learning_rate": 2.1433413122684436e-06,
+      "loss": 0.5706,
+      "step": 43164
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.5796022415161133,
+      "learning_rate": 2.143124926904815e-06,
+      "loss": 0.7136,
+      "step": 43165
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.932316303253174,
+      "learning_rate": 2.1429085494850876e-06,
+      "loss": 0.6053,
+      "step": 43166
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.797632932662964,
+      "learning_rate": 2.1426921800098597e-06,
+      "loss": 0.8989,
+      "step": 43167
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.7045772075653076,
+      "learning_rate": 2.142475818479736e-06,
+      "loss": 0.7729,
+      "step": 43168
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.4687724113464355,
+      "learning_rate": 2.1422594648953152e-06,
+      "loss": 0.7632,
+      "step": 43169
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.0005412101745605,
+      "learning_rate": 2.1420431192572018e-06,
+      "loss": 0.9093,
+      "step": 43170
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.5526275634765625,
+      "learning_rate": 2.1418267815659925e-06,
+      "loss": 0.789,
+      "step": 43171
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 1.9860817193984985,
+      "learning_rate": 2.1416104518222956e-06,
+      "loss": 0.3359,
+      "step": 43172
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.7518041133880615,
+      "learning_rate": 2.141394130026708e-06,
+      "loss": 0.5784,
+      "step": 43173
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.5605270862579346,
+      "learning_rate": 2.141177816179833e-06,
+      "loss": 0.5261,
+      "step": 43174
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 5.407590389251709,
+      "learning_rate": 2.1409615102822706e-06,
+      "loss": 0.4711,
+      "step": 43175
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.022726058959961,
+      "learning_rate": 2.1407452123346252e-06,
+      "loss": 0.4478,
+      "step": 43176
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 1.9466348886489868,
+      "learning_rate": 2.140528922337494e-06,
+      "loss": 0.4819,
+      "step": 43177
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.0026183128356934,
+      "learning_rate": 2.1403126402914824e-06,
+      "loss": 0.5478,
+      "step": 43178
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.026402711868286,
+      "learning_rate": 2.140096366197189e-06,
+      "loss": 0.4817,
+      "step": 43179
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.6309995651245117,
+      "learning_rate": 2.139880100055218e-06,
+      "loss": 0.4452,
+      "step": 43180
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 4.749413013458252,
+      "learning_rate": 2.139663841866166e-06,
+      "loss": 0.9457,
+      "step": 43181
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 1.9875105619430542,
+      "learning_rate": 2.1394475916306408e-06,
+      "loss": 0.4757,
+      "step": 43182
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.6567962169647217,
+      "learning_rate": 2.139231349349237e-06,
+      "loss": 0.5125,
+      "step": 43183
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 4.259876728057861,
+      "learning_rate": 2.139015115022562e-06,
+      "loss": 0.4917,
+      "step": 43184
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 6.908333778381348,
+      "learning_rate": 2.1387988886512116e-06,
+      "loss": 0.5278,
+      "step": 43185
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.7366654872894287,
+      "learning_rate": 2.1385826702357904e-06,
+      "loss": 0.5803,
+      "step": 43186
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.454498529434204,
+      "learning_rate": 2.1383664597769004e-06,
+      "loss": 0.7428,
+      "step": 43187
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.4501776695251465,
+      "learning_rate": 2.1381502572751396e-06,
+      "loss": 0.7487,
+      "step": 43188
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.2331221103668213,
+      "learning_rate": 2.1379340627311123e-06,
+      "loss": 0.6651,
+      "step": 43189
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.576162815093994,
+      "learning_rate": 2.1377178761454164e-06,
+      "loss": 0.5484,
+      "step": 43190
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.1695828437805176,
+      "learning_rate": 2.1375016975186573e-06,
+      "loss": 0.542,
+      "step": 43191
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.1595520973205566,
+      "learning_rate": 2.1372855268514312e-06,
+      "loss": 0.806,
+      "step": 43192
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.6210217475891113,
+      "learning_rate": 2.1370693641443417e-06,
+      "loss": 0.6159,
+      "step": 43193
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 4.539292335510254,
+      "learning_rate": 2.1368532093979922e-06,
+      "loss": 0.7616,
+      "step": 43194
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.1739373207092285,
+      "learning_rate": 2.136637062612979e-06,
+      "loss": 0.5442,
+      "step": 43195
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 4.721682071685791,
+      "learning_rate": 2.1364209237899075e-06,
+      "loss": 0.9724,
+      "step": 43196
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.577411651611328,
+      "learning_rate": 2.136204792929375e-06,
+      "loss": 0.4684,
+      "step": 43197
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.317002534866333,
+      "learning_rate": 2.1359886700319837e-06,
+      "loss": 0.8585,
+      "step": 43198
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.29325795173645,
+      "learning_rate": 2.135772555098337e-06,
+      "loss": 0.6214,
+      "step": 43199
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.0687711238861084,
+      "learning_rate": 2.1355564481290327e-06,
+      "loss": 0.6634,
+      "step": 43200
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.4933109283447266,
+      "learning_rate": 2.135340349124674e-06,
+      "loss": 0.6435,
+      "step": 43201
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.406662940979004,
+      "learning_rate": 2.135124258085859e-06,
+      "loss": 0.4654,
+      "step": 43202
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.0779829025268555,
+      "learning_rate": 2.1349081750131923e-06,
+      "loss": 0.6291,
+      "step": 43203
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.0958220958709717,
+      "learning_rate": 2.134692099907271e-06,
+      "loss": 0.4443,
+      "step": 43204
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.3809633255004883,
+      "learning_rate": 2.1344760327686984e-06,
+      "loss": 0.5599,
+      "step": 43205
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.1859171390533447,
+      "learning_rate": 2.1342599735980735e-06,
+      "loss": 0.449,
+      "step": 43206
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.817387104034424,
+      "learning_rate": 2.134043922396e-06,
+      "loss": 0.4649,
+      "step": 43207
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.485891103744507,
+      "learning_rate": 2.1338278791630744e-06,
+      "loss": 0.6659,
+      "step": 43208
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.340503692626953,
+      "learning_rate": 2.133611843899902e-06,
+      "loss": 0.7103,
+      "step": 43209
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.813155174255371,
+      "learning_rate": 2.1333958166070794e-06,
+      "loss": 0.6113,
+      "step": 43210
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.6931142807006836,
+      "learning_rate": 2.133179797285211e-06,
+      "loss": 0.8105,
+      "step": 43211
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.956937074661255,
+      "learning_rate": 2.132963785934892e-06,
+      "loss": 0.6165,
+      "step": 43212
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.7667713165283203,
+      "learning_rate": 2.13274778255673e-06,
+      "loss": 0.6581,
+      "step": 43213
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 4.2010931968688965,
+      "learning_rate": 2.132531787151321e-06,
+      "loss": 0.5851,
+      "step": 43214
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.5309436321258545,
+      "learning_rate": 2.1323157997192683e-06,
+      "loss": 0.6455,
+      "step": 43215
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 4.0516862869262695,
+      "learning_rate": 2.13209982026117e-06,
+      "loss": 0.5294,
+      "step": 43216
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.558934211730957,
+      "learning_rate": 2.131883848777629e-06,
+      "loss": 0.5344,
+      "step": 43217
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.8886685371398926,
+      "learning_rate": 2.1316678852692423e-06,
+      "loss": 0.534,
+      "step": 43218
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.651949167251587,
+      "learning_rate": 2.1314519297366153e-06,
+      "loss": 0.6778,
+      "step": 43219
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 1.8839614391326904,
+      "learning_rate": 2.1312359821803442e-06,
+      "loss": 0.6481,
+      "step": 43220
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.4755373001098633,
+      "learning_rate": 2.1310200426010326e-06,
+      "loss": 0.5871,
+      "step": 43221
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.390040159225464,
+      "learning_rate": 2.1308041109992777e-06,
+      "loss": 0.8568,
+      "step": 43222
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 4.217093467712402,
+      "learning_rate": 2.1305881873756833e-06,
+      "loss": 0.5071,
+      "step": 43223
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 1.7136682271957397,
+      "learning_rate": 2.1303722717308466e-06,
+      "loss": 0.4243,
+      "step": 43224
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.1580398082733154,
+      "learning_rate": 2.1301563640653724e-06,
+      "loss": 0.5222,
+      "step": 43225
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.3744025230407715,
+      "learning_rate": 2.129940464379855e-06,
+      "loss": 0.475,
+      "step": 43226
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 1.6786142587661743,
+      "learning_rate": 2.1297245726749005e-06,
+      "loss": 0.4913,
+      "step": 43227
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.0378952026367188,
+      "learning_rate": 2.129508688951105e-06,
+      "loss": 0.4991,
+      "step": 43228
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.378122329711914,
+      "learning_rate": 2.1292928132090725e-06,
+      "loss": 0.4088,
+      "step": 43229
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.889812707901001,
+      "learning_rate": 2.1290769454493995e-06,
+      "loss": 0.4093,
+      "step": 43230
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.106419324874878,
+      "learning_rate": 2.1288610856726893e-06,
+      "loss": 0.3498,
+      "step": 43231
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.168020725250244,
+      "learning_rate": 2.1286452338795373e-06,
+      "loss": 0.6303,
+      "step": 43232
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 5.764580726623535,
+      "learning_rate": 2.128429390070551e-06,
+      "loss": 0.9073,
+      "step": 43233
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.397857666015625,
+      "learning_rate": 2.1282135542463257e-06,
+      "loss": 0.7081,
+      "step": 43234
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.484037160873413,
+      "learning_rate": 2.1279977264074637e-06,
+      "loss": 0.4737,
+      "step": 43235
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.231430768966675,
+      "learning_rate": 2.127781906554562e-06,
+      "loss": 0.6993,
+      "step": 43236
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.6186976432800293,
+      "learning_rate": 2.127566094688226e-06,
+      "loss": 0.614,
+      "step": 43237
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.1950600147247314,
+      "learning_rate": 2.12735029080905e-06,
+      "loss": 0.603,
+      "step": 43238
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.3352861404418945,
+      "learning_rate": 2.1271344949176386e-06,
+      "loss": 0.5141,
+      "step": 43239
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 6.971691608428955,
+      "learning_rate": 2.126918707014588e-06,
+      "loss": 0.5158,
+      "step": 43240
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.46730375289917,
+      "learning_rate": 2.1267029271005026e-06,
+      "loss": 0.5561,
+      "step": 43241
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 1.7920899391174316,
+      "learning_rate": 2.1264871551759774e-06,
+      "loss": 0.4503,
+      "step": 43242
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.5302186012268066,
+      "learning_rate": 2.126271391241618e-06,
+      "loss": 0.5512,
+      "step": 43243
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.236754894256592,
+      "learning_rate": 2.1260556352980183e-06,
+      "loss": 0.6457,
+      "step": 43244
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.2544631958007812,
+      "learning_rate": 2.1258398873457834e-06,
+      "loss": 0.5947,
+      "step": 43245
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 5.718999862670898,
+      "learning_rate": 2.1256241473855093e-06,
+      "loss": 0.501,
+      "step": 43246
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.2123634815216064,
+      "learning_rate": 2.125408415417798e-06,
+      "loss": 0.451,
+      "step": 43247
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.7550265789031982,
+      "learning_rate": 2.125192691443251e-06,
+      "loss": 0.798,
+      "step": 43248
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.131619691848755,
+      "learning_rate": 2.124976975462464e-06,
+      "loss": 0.4175,
+      "step": 43249
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.490609645843506,
+      "learning_rate": 2.1247612674760403e-06,
+      "loss": 0.5964,
+      "step": 43250
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.051470994949341,
+      "learning_rate": 2.1245455674845773e-06,
+      "loss": 0.5366,
+      "step": 43251
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.2378015518188477,
+      "learning_rate": 2.1243298754886776e-06,
+      "loss": 0.4274,
+      "step": 43252
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.420426845550537,
+      "learning_rate": 2.1241141914889375e-06,
+      "loss": 0.8574,
+      "step": 43253
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.821434259414673,
+      "learning_rate": 2.123898515485958e-06,
+      "loss": 0.4889,
+      "step": 43254
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 1.8688538074493408,
+      "learning_rate": 2.1236828474803416e-06,
+      "loss": 0.5477,
+      "step": 43255
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.3261876106262207,
+      "learning_rate": 2.1234671874726844e-06,
+      "loss": 0.7176,
+      "step": 43256
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.7321298122406006,
+      "learning_rate": 2.1232515354635885e-06,
+      "loss": 0.5091,
+      "step": 43257
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 4.2326579093933105,
+      "learning_rate": 2.123035891453651e-06,
+      "loss": 0.7544,
+      "step": 43258
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.181788444519043,
+      "learning_rate": 2.1228202554434723e-06,
+      "loss": 0.7108,
+      "step": 43259
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.668640613555908,
+      "learning_rate": 2.1226046274336548e-06,
+      "loss": 0.6461,
+      "step": 43260
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.704880714416504,
+      "learning_rate": 2.1223890074247945e-06,
+      "loss": 0.5437,
+      "step": 43261
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.808676242828369,
+      "learning_rate": 2.122173395417494e-06,
+      "loss": 0.6373,
+      "step": 43262
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.4569239616394043,
+      "learning_rate": 2.121957791412349e-06,
+      "loss": 0.6298,
+      "step": 43263
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.1641902923583984,
+      "learning_rate": 2.1217421954099637e-06,
+      "loss": 0.5022,
+      "step": 43264
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.2434778213500977,
+      "learning_rate": 2.121526607410933e-06,
+      "loss": 0.5006,
+      "step": 43265
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 5.339548110961914,
+      "learning_rate": 2.1213110274158604e-06,
+      "loss": 0.657,
+      "step": 43266
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.0924456119537354,
+      "learning_rate": 2.1210954554253414e-06,
+      "loss": 0.4323,
+      "step": 43267
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.4623396396636963,
+      "learning_rate": 2.1208798914399797e-06,
+      "loss": 0.5305,
+      "step": 43268
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.6995906829833984,
+      "learning_rate": 2.1206643354603702e-06,
+      "loss": 0.9163,
+      "step": 43269
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.092491626739502,
+      "learning_rate": 2.1204487874871165e-06,
+      "loss": 0.575,
+      "step": 43270
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.5735597610473633,
+      "learning_rate": 2.1202332475208144e-06,
+      "loss": 0.4955,
+      "step": 43271
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.621197462081909,
+      "learning_rate": 2.1200177155620665e-06,
+      "loss": 0.737,
+      "step": 43272
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.8117668628692627,
+      "learning_rate": 2.1198021916114668e-06,
+      "loss": 0.5537,
+      "step": 43273
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 4.392402172088623,
+      "learning_rate": 2.119586675669622e-06,
+      "loss": 0.8254,
+      "step": 43274
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.648946762084961,
+      "learning_rate": 2.119371167737126e-06,
+      "loss": 0.7022,
+      "step": 43275
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.3213865756988525,
+      "learning_rate": 2.119155667814581e-06,
+      "loss": 0.5206,
+      "step": 43276
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.650193691253662,
+      "learning_rate": 2.1189401759025834e-06,
+      "loss": 0.7112,
+      "step": 43277
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.6319966316223145,
+      "learning_rate": 2.1187246920017355e-06,
+      "loss": 0.5848,
+      "step": 43278
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.108241319656372,
+      "learning_rate": 2.1185092161126335e-06,
+      "loss": 0.5209,
+      "step": 43279
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.6353085041046143,
+      "learning_rate": 2.1182937482358795e-06,
+      "loss": 0.5175,
+      "step": 43280
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 1.9642479419708252,
+      "learning_rate": 2.1180782883720695e-06,
+      "loss": 0.66,
+      "step": 43281
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 1.8740471601486206,
+      "learning_rate": 2.1178628365218057e-06,
+      "loss": 0.2904,
+      "step": 43282
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.8397254943847656,
+      "learning_rate": 2.117647392685684e-06,
+      "loss": 0.5321,
+      "step": 43283
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.238391876220703,
+      "learning_rate": 2.117431956864307e-06,
+      "loss": 0.7074,
+      "step": 43284
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.217136859893799,
+      "learning_rate": 2.11721652905827e-06,
+      "loss": 0.6953,
+      "step": 43285
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.647493362426758,
+      "learning_rate": 2.117001109268175e-06,
+      "loss": 0.6972,
+      "step": 43286
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.387341260910034,
+      "learning_rate": 2.1167856974946193e-06,
+      "loss": 0.4022,
+      "step": 43287
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.8281404972076416,
+      "learning_rate": 2.116570293738204e-06,
+      "loss": 0.9352,
+      "step": 43288
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.5458426475524902,
+      "learning_rate": 2.1163548979995247e-06,
+      "loss": 0.692,
+      "step": 43289
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.4999775886535645,
+      "learning_rate": 2.1161395102791837e-06,
+      "loss": 0.5299,
+      "step": 43290
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.4629645347595215,
+      "learning_rate": 2.115924130577777e-06,
+      "loss": 0.501,
+      "step": 43291
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 4.943614482879639,
+      "learning_rate": 2.1157087588959065e-06,
+      "loss": 0.5694,
+      "step": 43292
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.2657227516174316,
+      "learning_rate": 2.1154933952341654e-06,
+      "loss": 0.7498,
+      "step": 43293
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.119198799133301,
+      "learning_rate": 2.1152780395931606e-06,
+      "loss": 0.3747,
+      "step": 43294
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.5944066047668457,
+      "learning_rate": 2.115062691973485e-06,
+      "loss": 0.5692,
+      "step": 43295
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.404571533203125,
+      "learning_rate": 2.1148473523757416e-06,
+      "loss": 0.5341,
+      "step": 43296
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 4.107334136962891,
+      "learning_rate": 2.114632020800525e-06,
+      "loss": 0.7524,
+      "step": 43297
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.8809564113616943,
+      "learning_rate": 2.114416697248437e-06,
+      "loss": 0.6616,
+      "step": 43298
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 5.823273181915283,
+      "learning_rate": 2.114201381720074e-06,
+      "loss": 0.6548,
+      "step": 43299
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.3527278900146484,
+      "learning_rate": 2.113986074216038e-06,
+      "loss": 0.6122,
+      "step": 43300
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.228349447250366,
+      "learning_rate": 2.113770774736923e-06,
+      "loss": 0.6143,
+      "step": 43301
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.0513293743133545,
+      "learning_rate": 2.113555483283333e-06,
+      "loss": 0.6117,
+      "step": 43302
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.2967731952667236,
+      "learning_rate": 2.113340199855861e-06,
+      "loss": 0.4563,
+      "step": 43303
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.0292773246765137,
+      "learning_rate": 2.113124924455111e-06,
+      "loss": 0.741,
+      "step": 43304
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.964509963989258,
+      "learning_rate": 2.112909657081677e-06,
+      "loss": 0.4924,
+      "step": 43305
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.6354665756225586,
+      "learning_rate": 2.112694397736162e-06,
+      "loss": 0.7544,
+      "step": 43306
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.708836793899536,
+      "learning_rate": 2.1124791464191602e-06,
+      "loss": 0.5699,
+      "step": 43307
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.208040475845337,
+      "learning_rate": 2.112263903131274e-06,
+      "loss": 0.6,
+      "step": 43308
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.7967169284820557,
+      "learning_rate": 2.112048667873098e-06,
+      "loss": 0.6816,
+      "step": 43309
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.706303834915161,
+      "learning_rate": 2.1118334406452333e-06,
+      "loss": 0.488,
+      "step": 43310
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.2797558307647705,
+      "learning_rate": 2.1116182214482792e-06,
+      "loss": 0.566,
+      "step": 43311
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.68172550201416,
+      "learning_rate": 2.111403010282832e-06,
+      "loss": 0.5002,
+      "step": 43312
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.70345139503479,
+      "learning_rate": 2.1111878071494917e-06,
+      "loss": 0.4873,
+      "step": 43313
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.12373423576355,
+      "learning_rate": 2.110972612048855e-06,
+      "loss": 0.488,
+      "step": 43314
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.2928614616394043,
+      "learning_rate": 2.1107574249815205e-06,
+      "loss": 0.4964,
+      "step": 43315
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.9144253730773926,
+      "learning_rate": 2.110542245948089e-06,
+      "loss": 0.5625,
+      "step": 43316
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.9902191162109375,
+      "learning_rate": 2.1103270749491556e-06,
+      "loss": 0.4645,
+      "step": 43317
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.49371600151062,
+      "learning_rate": 2.110111911985322e-06,
+      "loss": 0.8329,
+      "step": 43318
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.186931610107422,
+      "learning_rate": 2.1098967570571827e-06,
+      "loss": 0.5754,
+      "step": 43319
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.360218048095703,
+      "learning_rate": 2.1096816101653378e-06,
+      "loss": 0.5902,
+      "step": 43320
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.9190802574157715,
+      "learning_rate": 2.1094664713103875e-06,
+      "loss": 0.564,
+      "step": 43321
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.361421823501587,
+      "learning_rate": 2.109251340492926e-06,
+      "loss": 0.8397,
+      "step": 43322
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.572232723236084,
+      "learning_rate": 2.109036217713556e-06,
+      "loss": 0.6391,
+      "step": 43323
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.021792411804199,
+      "learning_rate": 2.1088211029728713e-06,
+      "loss": 0.6439,
+      "step": 43324
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.0003786087036133,
+      "learning_rate": 2.1086059962714734e-06,
+      "loss": 0.4955,
+      "step": 43325
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.5228850841522217,
+      "learning_rate": 2.1083908976099577e-06,
+      "loss": 0.7061,
+      "step": 43326
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.6771345138549805,
+      "learning_rate": 2.108175806988926e-06,
+      "loss": 0.8028,
+      "step": 43327
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.9218764305114746,
+      "learning_rate": 2.107960724408971e-06,
+      "loss": 0.6443,
+      "step": 43328
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.603621006011963,
+      "learning_rate": 2.1077456498706966e-06,
+      "loss": 0.6759,
+      "step": 43329
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.971961498260498,
+      "learning_rate": 2.1075305833746963e-06,
+      "loss": 0.5009,
+      "step": 43330
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.2837438583374023,
+      "learning_rate": 2.107315524921571e-06,
+      "loss": 0.4109,
+      "step": 43331
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.142481565475464,
+      "learning_rate": 2.107100474511916e-06,
+      "loss": 0.6304,
+      "step": 43332
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.4637227058410645,
+      "learning_rate": 2.106885432146333e-06,
+      "loss": 0.7495,
+      "step": 43333
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 4.233763694763184,
+      "learning_rate": 2.1066703978254143e-06,
+      "loss": 0.3844,
+      "step": 43334
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.53735089302063,
+      "learning_rate": 2.1064553715497647e-06,
+      "loss": 0.4576,
+      "step": 43335
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.4702723026275635,
+      "learning_rate": 2.1062403533199773e-06,
+      "loss": 0.6282,
+      "step": 43336
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.7453560829162598,
+      "learning_rate": 2.106025343136653e-06,
+      "loss": 0.73,
+      "step": 43337
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 5.393277645111084,
+      "learning_rate": 2.105810341000386e-06,
+      "loss": 0.6346,
+      "step": 43338
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.9829052686691284,
+      "learning_rate": 2.1055953469117785e-06,
+      "loss": 0.4717,
+      "step": 43339
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.8307173252105713,
+      "learning_rate": 2.1053803608714242e-06,
+      "loss": 0.308,
+      "step": 43340
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.728332281112671,
+      "learning_rate": 2.1051653828799247e-06,
+      "loss": 0.7626,
+      "step": 43341
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.515876531600952,
+      "learning_rate": 2.104950412937874e-06,
+      "loss": 0.4851,
+      "step": 43342
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.3848018646240234,
+      "learning_rate": 2.1047354510458727e-06,
+      "loss": 0.5033,
+      "step": 43343
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 4.373477935791016,
+      "learning_rate": 2.104520497204517e-06,
+      "loss": 0.4552,
+      "step": 43344
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.2762272357940674,
+      "learning_rate": 2.104305551414406e-06,
+      "loss": 0.4468,
+      "step": 43345
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.9770383834838867,
+      "learning_rate": 2.104090613676135e-06,
+      "loss": 0.4611,
+      "step": 43346
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.4375691413879395,
+      "learning_rate": 2.103875683990305e-06,
+      "loss": 0.5489,
+      "step": 43347
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.314628839492798,
+      "learning_rate": 2.10366076235751e-06,
+      "loss": 0.7923,
+      "step": 43348
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.4929347038269043,
+      "learning_rate": 2.103445848778351e-06,
+      "loss": 0.7527,
+      "step": 43349
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.650029182434082,
+      "learning_rate": 2.1032309432534227e-06,
+      "loss": 0.5975,
+      "step": 43350
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.078429937362671,
+      "learning_rate": 2.1030160457833256e-06,
+      "loss": 0.7376,
+      "step": 43351
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.172576427459717,
+      "learning_rate": 2.1028011563686533e-06,
+      "loss": 0.6425,
+      "step": 43352
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 4.5026750564575195,
+      "learning_rate": 2.1025862750100084e-06,
+      "loss": 0.5732,
+      "step": 43353
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.9027085304260254,
+      "learning_rate": 2.102371401707982e-06,
+      "loss": 0.7918,
+      "step": 43354
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.5141565799713135,
+      "learning_rate": 2.102156536463178e-06,
+      "loss": 0.7918,
+      "step": 43355
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.0242934226989746,
+      "learning_rate": 2.1019416792761897e-06,
+      "loss": 0.4093,
+      "step": 43356
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.4017860889434814,
+      "learning_rate": 2.1017268301476178e-06,
+      "loss": 0.7008,
+      "step": 43357
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.555079460144043,
+      "learning_rate": 2.101511989078056e-06,
+      "loss": 0.635,
+      "step": 43358
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.536543607711792,
+      "learning_rate": 2.1012971560681056e-06,
+      "loss": 0.7697,
+      "step": 43359
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.6964550018310547,
+      "learning_rate": 2.1010823311183598e-06,
+      "loss": 0.7907,
+      "step": 43360
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.4556524753570557,
+      "learning_rate": 2.10086751422942e-06,
+      "loss": 0.5765,
+      "step": 43361
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.260948896408081,
+      "learning_rate": 2.1006527054018798e-06,
+      "loss": 0.5891,
+      "step": 43362
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.5522527694702148,
+      "learning_rate": 2.1004379046363404e-06,
+      "loss": 0.4091,
+      "step": 43363
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.470761299133301,
+      "learning_rate": 2.100223111933395e-06,
+      "loss": 0.5209,
+      "step": 43364
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.2611758708953857,
+      "learning_rate": 2.1000083272936443e-06,
+      "loss": 0.6622,
+      "step": 43365
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.850825071334839,
+      "learning_rate": 2.0997935507176824e-06,
+      "loss": 0.48,
+      "step": 43366
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.7763867378234863,
+      "learning_rate": 2.0995787822061103e-06,
+      "loss": 0.6926,
+      "step": 43367
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 4.648265361785889,
+      "learning_rate": 2.0993640217595213e-06,
+      "loss": 0.6831,
+      "step": 43368
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.480689525604248,
+      "learning_rate": 2.099149269378516e-06,
+      "loss": 0.483,
+      "step": 43369
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.5070812702178955,
+      "learning_rate": 2.0989345250636885e-06,
+      "loss": 0.5703,
+      "step": 43370
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.8874287605285645,
+      "learning_rate": 2.098719788815637e-06,
+      "loss": 0.6098,
+      "step": 43371
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.110046148300171,
+      "learning_rate": 2.0985050606349605e-06,
+      "loss": 0.5042,
+      "step": 43372
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.341588020324707,
+      "learning_rate": 2.0982903405222526e-06,
+      "loss": 0.7649,
+      "step": 43373
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.3621702194213867,
+      "learning_rate": 2.0980756284781145e-06,
+      "loss": 0.5849,
+      "step": 43374
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.5028774738311768,
+      "learning_rate": 2.097860924503139e-06,
+      "loss": 0.791,
+      "step": 43375
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.8746616840362549,
+      "learning_rate": 2.097646228597925e-06,
+      "loss": 0.4216,
+      "step": 43376
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.809682846069336,
+      "learning_rate": 2.0974315407630713e-06,
+      "loss": 0.6794,
+      "step": 43377
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.463428020477295,
+      "learning_rate": 2.0972168609991717e-06,
+      "loss": 0.6532,
+      "step": 43378
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.777003049850464,
+      "learning_rate": 2.0970021893068265e-06,
+      "loss": 0.6894,
+      "step": 43379
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.3729782104492188,
+      "learning_rate": 2.096787525686628e-06,
+      "loss": 0.5238,
+      "step": 43380
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.719365358352661,
+      "learning_rate": 2.0965728701391785e-06,
+      "loss": 0.6546,
+      "step": 43381
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.8970369100570679,
+      "learning_rate": 2.09635822266507e-06,
+      "loss": 0.4677,
+      "step": 43382
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.7458443641662598,
+      "learning_rate": 2.0961435832649023e-06,
+      "loss": 0.8132,
+      "step": 43383
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.6772806644439697,
+      "learning_rate": 2.095928951939273e-06,
+      "loss": 0.6849,
+      "step": 43384
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.329827308654785,
+      "learning_rate": 2.095714328688775e-06,
+      "loss": 0.6348,
+      "step": 43385
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 4.942448616027832,
+      "learning_rate": 2.09549971351401e-06,
+      "loss": 0.7063,
+      "step": 43386
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.0290791988372803,
+      "learning_rate": 2.09528510641557e-06,
+      "loss": 0.7746,
+      "step": 43387
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.5193254947662354,
+      "learning_rate": 2.0950705073940553e-06,
+      "loss": 0.6905,
+      "step": 43388
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.83674693107605,
+      "learning_rate": 2.09485591645006e-06,
+      "loss": 0.5833,
+      "step": 43389
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.348839521408081,
+      "learning_rate": 2.0946413335841844e-06,
+      "loss": 0.4416,
+      "step": 43390
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.9388660192489624,
+      "learning_rate": 2.094426758797021e-06,
+      "loss": 0.4253,
+      "step": 43391
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.9671454429626465,
+      "learning_rate": 2.0942121920891697e-06,
+      "loss": 0.559,
+      "step": 43392
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.1850221157073975,
+      "learning_rate": 2.093997633461224e-06,
+      "loss": 0.8294,
+      "step": 43393
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.560638666152954,
+      "learning_rate": 2.093783082913784e-06,
+      "loss": 0.6555,
+      "step": 43394
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.076080083847046,
+      "learning_rate": 2.093568540447441e-06,
+      "loss": 0.6217,
+      "step": 43395
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.541964292526245,
+      "learning_rate": 2.093354006062798e-06,
+      "loss": 0.7816,
+      "step": 43396
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.3572165966033936,
+      "learning_rate": 2.093139479760448e-06,
+      "loss": 0.6721,
+      "step": 43397
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.7770808935165405,
+      "learning_rate": 2.0929249615409895e-06,
+      "loss": 0.4705,
+      "step": 43398
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.1647298336029053,
+      "learning_rate": 2.0927104514050162e-06,
+      "loss": 0.5404,
+      "step": 43399
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.171767473220825,
+      "learning_rate": 2.092495949353128e-06,
+      "loss": 0.6787,
+      "step": 43400
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.7283432483673096,
+      "learning_rate": 2.0922814553859167e-06,
+      "loss": 0.5472,
+      "step": 43401
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.3234899044036865,
+      "learning_rate": 2.0920669695039835e-06,
+      "loss": 0.5719,
+      "step": 43402
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.919945001602173,
+      "learning_rate": 2.0918524917079213e-06,
+      "loss": 0.4195,
+      "step": 43403
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.8513355255126953,
+      "learning_rate": 2.0916380219983297e-06,
+      "loss": 0.575,
+      "step": 43404
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.1268649101257324,
+      "learning_rate": 2.0914235603758013e-06,
+      "loss": 0.6291,
+      "step": 43405
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 7.758397579193115,
+      "learning_rate": 2.091209106840936e-06,
+      "loss": 0.4615,
+      "step": 43406
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.529973030090332,
+      "learning_rate": 2.0909946613943273e-06,
+      "loss": 0.5822,
+      "step": 43407
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.8780862092971802,
+      "learning_rate": 2.090780224036574e-06,
+      "loss": 0.5141,
+      "step": 43408
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.5013097524642944,
+      "learning_rate": 2.0905657947682696e-06,
+      "loss": 0.4316,
+      "step": 43409
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.8461177349090576,
+      "learning_rate": 2.090351373590013e-06,
+      "loss": 0.509,
+      "step": 43410
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 4.031273365020752,
+      "learning_rate": 2.090136960502398e-06,
+      "loss": 1.0043,
+      "step": 43411
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.4963982105255127,
+      "learning_rate": 2.0899225555060237e-06,
+      "loss": 0.6359,
+      "step": 43412
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.0933830738067627,
+      "learning_rate": 2.089708158601483e-06,
+      "loss": 0.4738,
+      "step": 43413
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.3617494106292725,
+      "learning_rate": 2.0894937697893752e-06,
+      "loss": 0.5489,
+      "step": 43414
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.036885976791382,
+      "learning_rate": 2.0892793890702916e-06,
+      "loss": 0.8545,
+      "step": 43415
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.5114924907684326,
+      "learning_rate": 2.0890650164448352e-06,
+      "loss": 0.8582,
+      "step": 43416
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.2529337406158447,
+      "learning_rate": 2.0888506519135964e-06,
+      "loss": 0.5554,
+      "step": 43417
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.2972633838653564,
+      "learning_rate": 2.0886362954771754e-06,
+      "loss": 0.733,
+      "step": 43418
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.7588616609573364,
+      "learning_rate": 2.0884219471361642e-06,
+      "loss": 0.346,
+      "step": 43419
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.998542547225952,
+      "learning_rate": 2.0882076068911634e-06,
+      "loss": 0.5771,
+      "step": 43420
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.149451494216919,
+      "learning_rate": 2.0879932747427637e-06,
+      "loss": 0.2608,
+      "step": 43421
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.0303590297698975,
+      "learning_rate": 2.087778950691567e-06,
+      "loss": 0.4762,
+      "step": 43422
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.4683077335357666,
+      "learning_rate": 2.087564634738163e-06,
+      "loss": 0.5745,
+      "step": 43423
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.002924919128418,
+      "learning_rate": 2.087350326883153e-06,
+      "loss": 0.5713,
+      "step": 43424
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.432497501373291,
+      "learning_rate": 2.087136027127129e-06,
+      "loss": 0.4416,
+      "step": 43425
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.052036762237549,
+      "learning_rate": 2.08692173547069e-06,
+      "loss": 0.6443,
+      "step": 43426
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.0837440490722656,
+      "learning_rate": 2.0867074519144287e-06,
+      "loss": 0.9161,
+      "step": 43427
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.8620977401733398,
+      "learning_rate": 2.0864931764589446e-06,
+      "loss": 0.7132,
+      "step": 43428
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.0106101036071777,
+      "learning_rate": 2.0862789091048295e-06,
+      "loss": 0.6001,
+      "step": 43429
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.4104561805725098,
+      "learning_rate": 2.0860646498526833e-06,
+      "loss": 0.4821,
+      "step": 43430
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.568140983581543,
+      "learning_rate": 2.0858503987030976e-06,
+      "loss": 0.7001,
+      "step": 43431
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.09651780128479,
+      "learning_rate": 2.0856361556566707e-06,
+      "loss": 0.8242,
+      "step": 43432
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.586782693862915,
+      "learning_rate": 2.085421920713999e-06,
+      "loss": 0.6758,
+      "step": 43433
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 4.621092796325684,
+      "learning_rate": 2.085207693875676e-06,
+      "loss": 1.0543,
+      "step": 43434
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.8695298433303833,
+      "learning_rate": 2.0849934751423e-06,
+      "loss": 0.6665,
+      "step": 43435
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.218132257461548,
+      "learning_rate": 2.0847792645144633e-06,
+      "loss": 0.4702,
+      "step": 43436
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.2243337631225586,
+      "learning_rate": 2.084565061992764e-06,
+      "loss": 0.6162,
+      "step": 43437
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.6967532634735107,
+      "learning_rate": 2.084350867577798e-06,
+      "loss": 0.5994,
+      "step": 43438
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.315284013748169,
+      "learning_rate": 2.0841366812701587e-06,
+      "loss": 0.5375,
+      "step": 43439
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.9303786754608154,
+      "learning_rate": 2.083922503070445e-06,
+      "loss": 0.5951,
+      "step": 43440
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.3160336017608643,
+      "learning_rate": 2.083708332979248e-06,
+      "loss": 0.3957,
+      "step": 43441
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.3193182945251465,
+      "learning_rate": 2.0834941709971677e-06,
+      "loss": 0.6514,
+      "step": 43442
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 4.074737548828125,
+      "learning_rate": 2.083280017124796e-06,
+      "loss": 0.6525,
+      "step": 43443
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.642132043838501,
+      "learning_rate": 2.08306587136273e-06,
+      "loss": 0.7336,
+      "step": 43444
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.0284342765808105,
+      "learning_rate": 2.0828517337115667e-06,
+      "loss": 0.5545,
+      "step": 43445
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.9262748956680298,
+      "learning_rate": 2.0826376041718984e-06,
+      "loss": 0.5324,
+      "step": 43446
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.342801332473755,
+      "learning_rate": 2.0824234827443234e-06,
+      "loss": 0.3727,
+      "step": 43447
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 4.552038192749023,
+      "learning_rate": 2.0822093694294344e-06,
+      "loss": 0.5452,
+      "step": 43448
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.3606553077697754,
+      "learning_rate": 2.0819952642278297e-06,
+      "loss": 0.5432,
+      "step": 43449
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.4622890949249268,
+      "learning_rate": 2.081781167140101e-06,
+      "loss": 0.7424,
+      "step": 43450
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.2876758575439453,
+      "learning_rate": 2.081567078166848e-06,
+      "loss": 0.591,
+      "step": 43451
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.1322522163391113,
+      "learning_rate": 2.0813529973086615e-06,
+      "loss": 0.8042,
+      "step": 43452
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.472648859024048,
+      "learning_rate": 2.0811389245661408e-06,
+      "loss": 0.5472,
+      "step": 43453
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.9040112495422363,
+      "learning_rate": 2.0809248599398774e-06,
+      "loss": 0.426,
+      "step": 43454
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.938697099685669,
+      "learning_rate": 2.0807108034304708e-06,
+      "loss": 0.5184,
+      "step": 43455
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.509852409362793,
+      "learning_rate": 2.0804967550385095e-06,
+      "loss": 0.4384,
+      "step": 43456
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.635976791381836,
+      "learning_rate": 2.080282714764597e-06,
+      "loss": 0.6221,
+      "step": 43457
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.5740177631378174,
+      "learning_rate": 2.080068682609323e-06,
+      "loss": 0.5845,
+      "step": 43458
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.1858127117156982,
+      "learning_rate": 2.079854658573286e-06,
+      "loss": 0.933,
+      "step": 43459
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.57759690284729,
+      "learning_rate": 2.079640642657077e-06,
+      "loss": 0.7259,
+      "step": 43460
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.4893574714660645,
+      "learning_rate": 2.0794266348612955e-06,
+      "loss": 0.6207,
+      "step": 43461
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.931922674179077,
+      "learning_rate": 2.079212635186533e-06,
+      "loss": 0.6316,
+      "step": 43462
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.2439517974853516,
+      "learning_rate": 2.078998643633387e-06,
+      "loss": 0.5,
+      "step": 43463
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.014636278152466,
+      "learning_rate": 2.0787846602024504e-06,
+      "loss": 0.6578,
+      "step": 43464
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.507538318634033,
+      "learning_rate": 2.0785706848943205e-06,
+      "loss": 0.6533,
+      "step": 43465
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.87609601020813,
+      "learning_rate": 2.07835671770959e-06,
+      "loss": 0.6986,
+      "step": 43466
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.6083507537841797,
+      "learning_rate": 2.078142758648856e-06,
+      "loss": 0.7729,
+      "step": 43467
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.885054588317871,
+      "learning_rate": 2.0779288077127113e-06,
+      "loss": 0.828,
+      "step": 43468
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.297945022583008,
+      "learning_rate": 2.0777148649017536e-06,
+      "loss": 0.6256,
+      "step": 43469
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.567176103591919,
+      "learning_rate": 2.0775009302165743e-06,
+      "loss": 0.3153,
+      "step": 43470
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.9999175071716309,
+      "learning_rate": 2.077287003657772e-06,
+      "loss": 0.4168,
+      "step": 43471
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.0730295181274414,
+      "learning_rate": 2.077073085225938e-06,
+      "loss": 0.6193,
+      "step": 43472
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.038501739501953,
+      "learning_rate": 2.0768591749216704e-06,
+      "loss": 0.4509,
+      "step": 43473
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.054619789123535,
+      "learning_rate": 2.0766452727455606e-06,
+      "loss": 0.8245,
+      "step": 43474
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.022035598754883,
+      "learning_rate": 2.076431378698207e-06,
+      "loss": 0.3876,
+      "step": 43475
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.799131155014038,
+      "learning_rate": 2.076217492780199e-06,
+      "loss": 0.5601,
+      "step": 43476
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.3302881717681885,
+      "learning_rate": 2.0760036149921385e-06,
+      "loss": 0.5403,
+      "step": 43477
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 4.1519012451171875,
+      "learning_rate": 2.075789745334615e-06,
+      "loss": 0.62,
+      "step": 43478
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.760094165802002,
+      "learning_rate": 2.0755758838082257e-06,
+      "loss": 0.474,
+      "step": 43479
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.016875982284546,
+      "learning_rate": 2.0753620304135636e-06,
+      "loss": 0.7109,
+      "step": 43480
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.865549325942993,
+      "learning_rate": 2.075148185151225e-06,
+      "loss": 0.8662,
+      "step": 43481
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.3947396278381348,
+      "learning_rate": 2.0749343480218025e-06,
+      "loss": 0.3284,
+      "step": 43482
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.944486141204834,
+      "learning_rate": 2.0747205190258936e-06,
+      "loss": 0.447,
+      "step": 43483
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 4.0038228034973145,
+      "learning_rate": 2.0745066981640887e-06,
+      "loss": 0.4069,
+      "step": 43484
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.648252010345459,
+      "learning_rate": 2.074292885436987e-06,
+      "loss": 0.6796,
+      "step": 43485
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 4.115322589874268,
+      "learning_rate": 2.0740790808451793e-06,
+      "loss": 0.9387,
+      "step": 43486
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.423954486846924,
+      "learning_rate": 2.0738652843892633e-06,
+      "loss": 0.6212,
+      "step": 43487
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.32084584236145,
+      "learning_rate": 2.0736514960698294e-06,
+      "loss": 0.8824,
+      "step": 43488
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.9175441265106201,
+      "learning_rate": 2.0734377158874763e-06,
+      "loss": 0.6728,
+      "step": 43489
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 6.145782470703125,
+      "learning_rate": 2.0732239438427953e-06,
+      "loss": 0.5371,
+      "step": 43490
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 4.472445964813232,
+      "learning_rate": 2.073010179936384e-06,
+      "loss": 0.673,
+      "step": 43491
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.8548945188522339,
+      "learning_rate": 2.0727964241688323e-06,
+      "loss": 0.5673,
+      "step": 43492
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.4492111206054688,
+      "learning_rate": 2.0725826765407396e-06,
+      "loss": 0.3308,
+      "step": 43493
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.079460859298706,
+      "learning_rate": 2.072368937052696e-06,
+      "loss": 0.75,
+      "step": 43494
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.0512020587921143,
+      "learning_rate": 2.072155205705298e-06,
+      "loss": 0.6058,
+      "step": 43495
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.713609218597412,
+      "learning_rate": 2.0719414824991408e-06,
+      "loss": 0.7376,
+      "step": 43496
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.877023696899414,
+      "learning_rate": 2.071727767434816e-06,
+      "loss": 0.6805,
+      "step": 43497
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.5745534896850586,
+      "learning_rate": 2.0715140605129193e-06,
+      "loss": 0.5064,
+      "step": 43498
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.3305039405822754,
+      "learning_rate": 2.0713003617340467e-06,
+      "loss": 0.6818,
+      "step": 43499
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.184805393218994,
+      "learning_rate": 2.0710866710987886e-06,
+      "loss": 0.5616,
+      "step": 43500
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.82955002784729,
+      "learning_rate": 2.0708729886077434e-06,
+      "loss": 0.6039,
+      "step": 43501
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.276372194290161,
+      "learning_rate": 2.070659314261501e-06,
+      "loss": 0.6444,
+      "step": 43502
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.2624428272247314,
+      "learning_rate": 2.07044564806066e-06,
+      "loss": 0.7179,
+      "step": 43503
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.014488935470581,
+      "learning_rate": 2.0702319900058097e-06,
+      "loss": 0.6775,
+      "step": 43504
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.7536892890930176,
+      "learning_rate": 2.070018340097547e-06,
+      "loss": 0.3986,
+      "step": 43505
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.5933115482330322,
+      "learning_rate": 2.0698046983364677e-06,
+      "loss": 0.5895,
+      "step": 43506
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.331106662750244,
+      "learning_rate": 2.0695910647231627e-06,
+      "loss": 0.5095,
+      "step": 43507
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.973111629486084,
+      "learning_rate": 2.0693774392582278e-06,
+      "loss": 0.6285,
+      "step": 43508
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.653581380844116,
+      "learning_rate": 2.0691638219422555e-06,
+      "loss": 0.5304,
+      "step": 43509
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.799574851989746,
+      "learning_rate": 2.068950212775842e-06,
+      "loss": 0.6602,
+      "step": 43510
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.734360694885254,
+      "learning_rate": 2.068736611759578e-06,
+      "loss": 0.7027,
+      "step": 43511
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.4677475690841675,
+      "learning_rate": 2.068523018894062e-06,
+      "loss": 0.2911,
+      "step": 43512
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.2084765434265137,
+      "learning_rate": 2.0683094341798826e-06,
+      "loss": 0.545,
+      "step": 43513
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.8441526889801025,
+      "learning_rate": 2.0680958576176385e-06,
+      "loss": 0.6398,
+      "step": 43514
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.1461987495422363,
+      "learning_rate": 2.0678822892079205e-06,
+      "loss": 0.5526,
+      "step": 43515
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.7726328372955322,
+      "learning_rate": 2.067668728951324e-06,
+      "loss": 0.4807,
+      "step": 43516
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.7341179847717285,
+      "learning_rate": 2.0674551768484396e-06,
+      "loss": 0.492,
+      "step": 43517
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.681006908416748,
+      "learning_rate": 2.0672416328998668e-06,
+      "loss": 0.647,
+      "step": 43518
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.845297932624817,
+      "learning_rate": 2.067028097106195e-06,
+      "loss": 0.5701,
+      "step": 43519
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 4.100619316101074,
+      "learning_rate": 2.066814569468021e-06,
+      "loss": 0.4962,
+      "step": 43520
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.2416412830352783,
+      "learning_rate": 2.066601049985935e-06,
+      "loss": 0.5923,
+      "step": 43521
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.269827365875244,
+      "learning_rate": 2.0663875386605346e-06,
+      "loss": 0.6421,
+      "step": 43522
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.1320102214813232,
+      "learning_rate": 2.0661740354924096e-06,
+      "loss": 0.5129,
+      "step": 43523
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.9934778213500977,
+      "learning_rate": 2.0659605404821576e-06,
+      "loss": 0.6185,
+      "step": 43524
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.8499789237976074,
+      "learning_rate": 2.065747053630368e-06,
+      "loss": 0.8878,
+      "step": 43525
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 5.430211544036865,
+      "learning_rate": 2.0655335749376388e-06,
+      "loss": 0.6587,
+      "step": 43526
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.7301299571990967,
+      "learning_rate": 2.0653201044045595e-06,
+      "loss": 0.527,
+      "step": 43527
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.269676923751831,
+      "learning_rate": 2.0651066420317274e-06,
+      "loss": 0.7949,
+      "step": 43528
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.454836130142212,
+      "learning_rate": 2.0648931878197325e-06,
+      "loss": 0.9198,
+      "step": 43529
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.2125461101531982,
+      "learning_rate": 2.0646797417691716e-06,
+      "loss": 0.7251,
+      "step": 43530
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.6451090574264526,
+      "learning_rate": 2.0644663038806357e-06,
+      "loss": 0.3604,
+      "step": 43531
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.423354387283325,
+      "learning_rate": 2.064252874154721e-06,
+      "loss": 0.4089,
+      "step": 43532
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.876682996749878,
+      "learning_rate": 2.0640394525920163e-06,
+      "loss": 0.6348,
+      "step": 43533
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.217982530593872,
+      "learning_rate": 2.0638260391931216e-06,
+      "loss": 0.5743,
+      "step": 43534
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.8442387580871582,
+      "learning_rate": 2.0636126339586236e-06,
+      "loss": 0.4831,
+      "step": 43535
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.6299192905426025,
+      "learning_rate": 2.0633992368891214e-06,
+      "loss": 0.6702,
+      "step": 43536
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.1757874488830566,
+      "learning_rate": 2.063185847985202e-06,
+      "loss": 0.4212,
+      "step": 43537
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.1312618255615234,
+      "learning_rate": 2.0629724672474668e-06,
+      "loss": 0.6924,
+      "step": 43538
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 4.0743088722229,
+      "learning_rate": 2.062759094676502e-06,
+      "loss": 0.8253,
+      "step": 43539
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.85859751701355,
+      "learning_rate": 2.062545730272906e-06,
+      "loss": 0.8578,
+      "step": 43540
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.1466097831726074,
+      "learning_rate": 2.0623323740372687e-06,
+      "loss": 0.6116,
+      "step": 43541
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.081955671310425,
+      "learning_rate": 2.062119025970186e-06,
+      "loss": 0.4538,
+      "step": 43542
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.120628833770752,
+      "learning_rate": 2.061905686072248e-06,
+      "loss": 0.4751,
+      "step": 43543
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.887425661087036,
+      "learning_rate": 2.0616923543440513e-06,
+      "loss": 0.5799,
+      "step": 43544
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.1766269207000732,
+      "learning_rate": 2.061479030786186e-06,
+      "loss": 0.622,
+      "step": 43545
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.8898990154266357,
+      "learning_rate": 2.0612657153992485e-06,
+      "loss": 0.4582,
+      "step": 43546
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.473888635635376,
+      "learning_rate": 2.061052408183828e-06,
+      "loss": 0.6066,
+      "step": 43547
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.592820167541504,
+      "learning_rate": 2.0608391091405216e-06,
+      "loss": 0.5719,
+      "step": 43548
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.7424256801605225,
+      "learning_rate": 2.060625818269919e-06,
+      "loss": 0.5462,
+      "step": 43549
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.180253505706787,
+      "learning_rate": 2.0604125355726168e-06,
+      "loss": 0.5436,
+      "step": 43550
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.6457409858703613,
+      "learning_rate": 2.0601992610492037e-06,
+      "loss": 0.6156,
+      "step": 43551
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.104851245880127,
+      "learning_rate": 2.059985994700278e-06,
+      "loss": 0.4315,
+      "step": 43552
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.317117214202881,
+      "learning_rate": 2.0597727365264275e-06,
+      "loss": 0.5441,
+      "step": 43553
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.841566562652588,
+      "learning_rate": 2.059559486528249e-06,
+      "loss": 0.5347,
+      "step": 43554
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.5320558547973633,
+      "learning_rate": 2.0593462447063335e-06,
+      "loss": 0.4183,
+      "step": 43555
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.489358425140381,
+      "learning_rate": 2.0591330110612734e-06,
+      "loss": 0.4306,
+      "step": 43556
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.603658676147461,
+      "learning_rate": 2.058919785593665e-06,
+      "loss": 0.6444,
+      "step": 43557
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.8744038343429565,
+      "learning_rate": 2.0587065683040967e-06,
+      "loss": 0.362,
+      "step": 43558
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.1179394721984863,
+      "learning_rate": 2.0584933591931638e-06,
+      "loss": 0.6708,
+      "step": 43559
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.7541072368621826,
+      "learning_rate": 2.058280158261461e-06,
+      "loss": 0.577,
+      "step": 43560
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.375614643096924,
+      "learning_rate": 2.058066965509577e-06,
+      "loss": 0.4578,
+      "step": 43561
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.1999764442443848,
+      "learning_rate": 2.0578537809381088e-06,
+      "loss": 0.622,
+      "step": 43562
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.902922511100769,
+      "learning_rate": 2.057640604547645e-06,
+      "loss": 0.586,
+      "step": 43563
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.798417091369629,
+      "learning_rate": 2.0574274363387827e-06,
+      "loss": 0.7088,
+      "step": 43564
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.1010799407958984,
+      "learning_rate": 2.05721427631211e-06,
+      "loss": 0.5308,
+      "step": 43565
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.519136667251587,
+      "learning_rate": 2.0570011244682235e-06,
+      "loss": 0.815,
+      "step": 43566
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.951786994934082,
+      "learning_rate": 2.0567879808077136e-06,
+      "loss": 0.5836,
+      "step": 43567
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.9296563863754272,
+      "learning_rate": 2.056574845331173e-06,
+      "loss": 0.6346,
+      "step": 43568
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.8896775245666504,
+      "learning_rate": 2.0563617180391975e-06,
+      "loss": 0.5176,
+      "step": 43569
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.2072339057922363,
+      "learning_rate": 2.056148598932375e-06,
+      "loss": 0.6062,
+      "step": 43570
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.4324445724487305,
+      "learning_rate": 2.0559354880113025e-06,
+      "loss": 0.7211,
+      "step": 43571
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 4.254815578460693,
+      "learning_rate": 2.055722385276569e-06,
+      "loss": 0.6615,
+      "step": 43572
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.8070716857910156,
+      "learning_rate": 2.05550929072877e-06,
+      "loss": 0.648,
+      "step": 43573
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.341904640197754,
+      "learning_rate": 2.0552962043684947e-06,
+      "loss": 0.5473,
+      "step": 43574
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 6.367609977722168,
+      "learning_rate": 2.05508312619634e-06,
+      "loss": 0.7153,
+      "step": 43575
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.6031529903411865,
+      "learning_rate": 2.0548700562128935e-06,
+      "loss": 1.1592,
+      "step": 43576
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.584199905395508,
+      "learning_rate": 2.0546569944187524e-06,
+      "loss": 0.5984,
+      "step": 43577
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.2858874797821045,
+      "learning_rate": 2.0544439408145047e-06,
+      "loss": 0.6205,
+      "step": 43578
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.584399461746216,
+      "learning_rate": 2.054230895400745e-06,
+      "loss": 0.6329,
+      "step": 43579
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.0162546634674072,
+      "learning_rate": 2.0540178581780657e-06,
+      "loss": 0.4176,
+      "step": 43580
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.400689721107483,
+      "learning_rate": 2.053804829147061e-06,
+      "loss": 0.3207,
+      "step": 43581
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.442214012145996,
+      "learning_rate": 2.0535918083083196e-06,
+      "loss": 0.4979,
+      "step": 43582
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.777024745941162,
+      "learning_rate": 2.0533787956624373e-06,
+      "loss": 0.651,
+      "step": 43583
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.1693167686462402,
+      "learning_rate": 2.0531657912100034e-06,
+      "loss": 0.5843,
+      "step": 43584
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.2496562004089355,
+      "learning_rate": 2.0529527949516133e-06,
+      "loss": 0.9065,
+      "step": 43585
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.248727798461914,
+      "learning_rate": 2.0527398068878553e-06,
+      "loss": 0.6862,
+      "step": 43586
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.3197102546691895,
+      "learning_rate": 2.0525268270193253e-06,
+      "loss": 0.7724,
+      "step": 43587
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.5470528602600098,
+      "learning_rate": 2.0523138553466124e-06,
+      "loss": 0.6327,
+      "step": 43588
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.770073175430298,
+      "learning_rate": 2.052100891870313e-06,
+      "loss": 0.602,
+      "step": 43589
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.32218074798584,
+      "learning_rate": 2.0518879365910143e-06,
+      "loss": 0.8636,
+      "step": 43590
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.5220377445220947,
+      "learning_rate": 2.051674989509313e-06,
+      "loss": 0.6181,
+      "step": 43591
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.6220569610595703,
+      "learning_rate": 2.051462050625797e-06,
+      "loss": 0.6476,
+      "step": 43592
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.0514469146728516,
+      "learning_rate": 2.0512491199410624e-06,
+      "loss": 0.7501,
+      "step": 43593
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.6058857440948486,
+      "learning_rate": 2.0510361974556976e-06,
+      "loss": 0.5527,
+      "step": 43594
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 4.777805805206299,
+      "learning_rate": 2.050823283170298e-06,
+      "loss": 0.7899,
+      "step": 43595
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.121168613433838,
+      "learning_rate": 2.0506103770854522e-06,
+      "loss": 0.4843,
+      "step": 43596
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.81040096282959,
+      "learning_rate": 2.050397479201756e-06,
+      "loss": 0.4155,
+      "step": 43597
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.114739418029785,
+      "learning_rate": 2.050184589519796e-06,
+      "loss": 0.59,
+      "step": 43598
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.5907275676727295,
+      "learning_rate": 2.049971708040171e-06,
+      "loss": 0.3835,
+      "step": 43599
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.8115689754486084,
+      "learning_rate": 2.0497588347634683e-06,
+      "loss": 0.5752,
+      "step": 43600
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 4.0231852531433105,
+      "learning_rate": 2.0495459696902825e-06,
+      "loss": 0.3935,
+      "step": 43601
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.119666337966919,
+      "learning_rate": 2.0493331128212025e-06,
+      "loss": 0.5066,
+      "step": 43602
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.7053964138031006,
+      "learning_rate": 2.049120264156823e-06,
+      "loss": 0.6441,
+      "step": 43603
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.224886417388916,
+      "learning_rate": 2.0489074236977336e-06,
+      "loss": 0.4948,
+      "step": 43604
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.9646570682525635,
+      "learning_rate": 2.0486945914445283e-06,
+      "loss": 0.6978,
+      "step": 43605
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.561577081680298,
+      "learning_rate": 2.0484817673977963e-06,
+      "loss": 0.7373,
+      "step": 43606
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.6412659883499146,
+      "learning_rate": 2.048268951558133e-06,
+      "loss": 0.6034,
+      "step": 43607
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.546492338180542,
+      "learning_rate": 2.0480561439261253e-06,
+      "loss": 0.5944,
+      "step": 43608
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.919500470161438,
+      "learning_rate": 2.04784334450237e-06,
+      "loss": 0.2986,
+      "step": 43609
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.473276138305664,
+      "learning_rate": 2.0476305532874545e-06,
+      "loss": 0.5781,
+      "step": 43610
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.697129964828491,
+      "learning_rate": 2.0474177702819743e-06,
+      "loss": 0.6774,
+      "step": 43611
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.6926307678222656,
+      "learning_rate": 2.0472049954865176e-06,
+      "loss": 0.6338,
+      "step": 43612
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.119443416595459,
+      "learning_rate": 2.046992228901679e-06,
+      "loss": 0.8178,
+      "step": 43613
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.155017375946045,
+      "learning_rate": 2.0467794705280474e-06,
+      "loss": 0.4734,
+      "step": 43614
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.8574696779251099,
+      "learning_rate": 2.0465667203662177e-06,
+      "loss": 0.6435,
+      "step": 43615
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.077160120010376,
+      "learning_rate": 2.0463539784167776e-06,
+      "loss": 0.4767,
+      "step": 43616
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.5521135330200195,
+      "learning_rate": 2.0461412446803202e-06,
+      "loss": 0.4856,
+      "step": 43617
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.7643675804138184,
+      "learning_rate": 2.0459285191574397e-06,
+      "loss": 0.4859,
+      "step": 43618
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.5122625827789307,
+      "learning_rate": 2.0457158018487234e-06,
+      "loss": 0.5912,
+      "step": 43619
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 4.134620189666748,
+      "learning_rate": 2.0455030927547647e-06,
+      "loss": 0.8368,
+      "step": 43620
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.6738569736480713,
+      "learning_rate": 2.0452903918761574e-06,
+      "loss": 0.6368,
+      "step": 43621
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.2927932739257812,
+      "learning_rate": 2.0450776992134885e-06,
+      "loss": 0.5555,
+      "step": 43622
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.5884716510772705,
+      "learning_rate": 2.044865014767353e-06,
+      "loss": 0.6499,
+      "step": 43623
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.832080364227295,
+      "learning_rate": 2.04465233853834e-06,
+      "loss": 0.4212,
+      "step": 43624
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 4.187076568603516,
+      "learning_rate": 2.044439670527043e-06,
+      "loss": 0.4243,
+      "step": 43625
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.201775550842285,
+      "learning_rate": 2.04422701073405e-06,
+      "loss": 0.5605,
+      "step": 43626
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.9287936687469482,
+      "learning_rate": 2.044014359159957e-06,
+      "loss": 0.4655,
+      "step": 43627
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 6.817383766174316,
+      "learning_rate": 2.0438017158053502e-06,
+      "loss": 0.6089,
+      "step": 43628
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.0832595825195312,
+      "learning_rate": 2.043589080670824e-06,
+      "loss": 0.4442,
+      "step": 43629
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.859567642211914,
+      "learning_rate": 2.0433764537569707e-06,
+      "loss": 0.35,
+      "step": 43630
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 4.712203502655029,
+      "learning_rate": 2.043163835064378e-06,
+      "loss": 0.5685,
+      "step": 43631
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.540008783340454,
+      "learning_rate": 2.0429512245936408e-06,
+      "loss": 0.6578,
+      "step": 43632
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.9990123510360718,
+      "learning_rate": 2.0427386223453464e-06,
+      "loss": 0.3769,
+      "step": 43633
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.65250301361084,
+      "learning_rate": 2.0425260283200903e-06,
+      "loss": 0.5815,
+      "step": 43634
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.4495105743408203,
+      "learning_rate": 2.0423134425184593e-06,
+      "loss": 0.74,
+      "step": 43635
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.0806798934936523,
+      "learning_rate": 2.0421008649410486e-06,
+      "loss": 0.5715,
+      "step": 43636
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.123260974884033,
+      "learning_rate": 2.0418882955884455e-06,
+      "loss": 0.4743,
+      "step": 43637
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.2630770206451416,
+      "learning_rate": 2.0416757344612447e-06,
+      "loss": 0.3803,
+      "step": 43638
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.6609320640563965,
+      "learning_rate": 2.041463181560034e-06,
+      "loss": 0.6198,
+      "step": 43639
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.3865792751312256,
+      "learning_rate": 2.0412506368854056e-06,
+      "loss": 0.5954,
+      "step": 43640
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 5.528050422668457,
+      "learning_rate": 2.041038100437951e-06,
+      "loss": 0.9688,
+      "step": 43641
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.4771311283111572,
+      "learning_rate": 2.040825572218262e-06,
+      "loss": 0.8023,
+      "step": 43642
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 4.221319675445557,
+      "learning_rate": 2.040613052226928e-06,
+      "loss": 0.482,
+      "step": 43643
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.9599382877349854,
+      "learning_rate": 2.0404005404645415e-06,
+      "loss": 0.663,
+      "step": 43644
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 4.038708209991455,
+      "learning_rate": 2.040188036931691e-06,
+      "loss": 0.3115,
+      "step": 43645
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.7726895809173584,
+      "learning_rate": 2.0399755416289703e-06,
+      "loss": 0.822,
+      "step": 43646
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.3594744205474854,
+      "learning_rate": 2.0397630545569673e-06,
+      "loss": 0.6124,
+      "step": 43647
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.76960825920105,
+      "learning_rate": 2.039550575716276e-06,
+      "loss": 0.5464,
+      "step": 43648
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.8139655590057373,
+      "learning_rate": 2.0393381051074836e-06,
+      "loss": 0.7582,
+      "step": 43649
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.4529144763946533,
+      "learning_rate": 2.0391256427311844e-06,
+      "loss": 0.5525,
+      "step": 43650
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.384249210357666,
+      "learning_rate": 2.038913188587966e-06,
+      "loss": 0.6686,
+      "step": 43651
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.3597819805145264,
+      "learning_rate": 2.0387007426784226e-06,
+      "loss": 0.5264,
+      "step": 43652
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.019455671310425,
+      "learning_rate": 2.038488305003142e-06,
+      "loss": 0.4976,
+      "step": 43653
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.2042288780212402,
+      "learning_rate": 2.038275875562717e-06,
+      "loss": 0.6873,
+      "step": 43654
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 5.355259895324707,
+      "learning_rate": 2.0380634543577353e-06,
+      "loss": 0.7354,
+      "step": 43655
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.394172191619873,
+      "learning_rate": 2.037851041388792e-06,
+      "loss": 0.7046,
+      "step": 43656
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.8992645740509033,
+      "learning_rate": 2.037638636656473e-06,
+      "loss": 0.6561,
+      "step": 43657
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.604465961456299,
+      "learning_rate": 2.037426240161374e-06,
+      "loss": 0.614,
+      "step": 43658
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.594780445098877,
+      "learning_rate": 2.0372138519040778e-06,
+      "loss": 0.5845,
+      "step": 43659
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.6024045944213867,
+      "learning_rate": 2.0370014718851842e-06,
+      "loss": 0.7698,
+      "step": 43660
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.6569730043411255,
+      "learning_rate": 2.036789100105278e-06,
+      "loss": 0.5237,
+      "step": 43661
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.2245521545410156,
+      "learning_rate": 2.036576736564953e-06,
+      "loss": 0.8378,
+      "step": 43662
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.81648588180542,
+      "learning_rate": 2.036364381264796e-06,
+      "loss": 0.6016,
+      "step": 43663
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.2244253158569336,
+      "learning_rate": 2.0361520342054016e-06,
+      "loss": 0.7681,
+      "step": 43664
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.3753600120544434,
+      "learning_rate": 2.035939695387356e-06,
+      "loss": 0.5656,
+      "step": 43665
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.546661138534546,
+      "learning_rate": 2.0357273648112534e-06,
+      "loss": 0.6316,
+      "step": 43666
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 4.044595241546631,
+      "learning_rate": 2.035515042477681e-06,
+      "loss": 0.8206,
+      "step": 43667
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.802558422088623,
+      "learning_rate": 2.0353027283872325e-06,
+      "loss": 0.8724,
+      "step": 43668
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.1753547191619873,
+      "learning_rate": 2.035090422540495e-06,
+      "loss": 0.7038,
+      "step": 43669
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.378455638885498,
+      "learning_rate": 2.0348781249380615e-06,
+      "loss": 0.6954,
+      "step": 43670
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.6443119049072266,
+      "learning_rate": 2.03466583558052e-06,
+      "loss": 0.5398,
+      "step": 43671
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.4178354740142822,
+      "learning_rate": 2.0344535544684635e-06,
+      "loss": 0.6411,
+      "step": 43672
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.887899160385132,
+      "learning_rate": 2.0342412816024785e-06,
+      "loss": 0.6409,
+      "step": 43673
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.1337411403656006,
+      "learning_rate": 2.03402901698316e-06,
+      "loss": 0.4314,
+      "step": 43674
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.001072406768799,
+      "learning_rate": 2.0338167606110938e-06,
+      "loss": 0.5315,
+      "step": 43675
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.6735100746154785,
+      "learning_rate": 2.0336045124868734e-06,
+      "loss": 0.4418,
+      "step": 43676
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.9681808948516846,
+      "learning_rate": 2.033392272611086e-06,
+      "loss": 0.5372,
+      "step": 43677
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.618190288543701,
+      "learning_rate": 2.0331800409843254e-06,
+      "loss": 0.4253,
+      "step": 43678
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 4.922057628631592,
+      "learning_rate": 2.032967817607178e-06,
+      "loss": 0.484,
+      "step": 43679
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.027881145477295,
+      "learning_rate": 2.0327556024802354e-06,
+      "loss": 0.7222,
+      "step": 43680
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.5558300018310547,
+      "learning_rate": 2.032543395604088e-06,
+      "loss": 0.5881,
+      "step": 43681
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.290377140045166,
+      "learning_rate": 2.0323311969793273e-06,
+      "loss": 0.754,
+      "step": 43682
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.465590476989746,
+      "learning_rate": 2.0321190066065405e-06,
+      "loss": 0.6814,
+      "step": 43683
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.532888889312744,
+      "learning_rate": 2.0319068244863203e-06,
+      "loss": 0.5232,
+      "step": 43684
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.9894596338272095,
+      "learning_rate": 2.0316946506192538e-06,
+      "loss": 0.4789,
+      "step": 43685
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.730137586593628,
+      "learning_rate": 2.0314824850059346e-06,
+      "loss": 0.5937,
+      "step": 43686
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.3445677757263184,
+      "learning_rate": 2.031270327646948e-06,
+      "loss": 0.6114,
+      "step": 43687
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.24524188041687,
+      "learning_rate": 2.031058178542889e-06,
+      "loss": 0.5675,
+      "step": 43688
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.714644193649292,
+      "learning_rate": 2.0308460376943427e-06,
+      "loss": 0.495,
+      "step": 43689
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.1881630420684814,
+      "learning_rate": 2.030633905101902e-06,
+      "loss": 0.5209,
+      "step": 43690
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.337876081466675,
+      "learning_rate": 2.030421780766157e-06,
+      "loss": 0.6212,
+      "step": 43691
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.9106180667877197,
+      "learning_rate": 2.030209664687696e-06,
+      "loss": 0.6751,
+      "step": 43692
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.9572657346725464,
+      "learning_rate": 2.02999755686711e-06,
+      "loss": 0.5614,
+      "step": 43693
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.1962273120880127,
+      "learning_rate": 2.029785457304987e-06,
+      "loss": 0.4298,
+      "step": 43694
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.2593188285827637,
+      "learning_rate": 2.0295733660019196e-06,
+      "loss": 0.439,
+      "step": 43695
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.943368434906006,
+      "learning_rate": 2.029361282958494e-06,
+      "loss": 0.5382,
+      "step": 43696
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.914353132247925,
+      "learning_rate": 2.029149208175303e-06,
+      "loss": 0.5066,
+      "step": 43697
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.3383126258850098,
+      "learning_rate": 2.0289371416529346e-06,
+      "loss": 0.306,
+      "step": 43698
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.572437286376953,
+      "learning_rate": 2.0287250833919804e-06,
+      "loss": 0.8843,
+      "step": 43699
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.1479010581970215,
+      "learning_rate": 2.028513033393027e-06,
+      "loss": 0.5222,
+      "step": 43700
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 4.3442277908325195,
+      "learning_rate": 2.0283009916566653e-06,
+      "loss": 0.9417,
+      "step": 43701
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.968998432159424,
+      "learning_rate": 2.0280889581834856e-06,
+      "loss": 0.5206,
+      "step": 43702
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.017275094985962,
+      "learning_rate": 2.027876932974079e-06,
+      "loss": 0.4291,
+      "step": 43703
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.497898817062378,
+      "learning_rate": 2.027664916029032e-06,
+      "loss": 0.6665,
+      "step": 43704
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 5.227789878845215,
+      "learning_rate": 2.027452907348936e-06,
+      "loss": 0.6521,
+      "step": 43705
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.5485281944274902,
+      "learning_rate": 2.027240906934379e-06,
+      "loss": 0.5777,
+      "step": 43706
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.5637359619140625,
+      "learning_rate": 2.027028914785953e-06,
+      "loss": 0.6963,
+      "step": 43707
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.968231678009033,
+      "learning_rate": 2.026816930904244e-06,
+      "loss": 0.5676,
+      "step": 43708
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.6531124114990234,
+      "learning_rate": 2.0266049552898454e-06,
+      "loss": 0.6491,
+      "step": 43709
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.3284480571746826,
+      "learning_rate": 2.0263929879433426e-06,
+      "loss": 0.4953,
+      "step": 43710
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.8919625282287598,
+      "learning_rate": 2.0261810288653284e-06,
+      "loss": 0.4078,
+      "step": 43711
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.5001420974731445,
+      "learning_rate": 2.0259690780563895e-06,
+      "loss": 0.7782,
+      "step": 43712
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.4880176782608032,
+      "learning_rate": 2.025757135517118e-06,
+      "loss": 0.1982,
+      "step": 43713
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.9319676160812378,
+      "learning_rate": 2.0255452012480996e-06,
+      "loss": 0.5288,
+      "step": 43714
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.3605740070343018,
+      "learning_rate": 2.025333275249928e-06,
+      "loss": 0.7064,
+      "step": 43715
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.501715898513794,
+      "learning_rate": 2.0251213575231887e-06,
+      "loss": 0.6254,
+      "step": 43716
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.0075035095214844,
+      "learning_rate": 2.024909448068474e-06,
+      "loss": 0.6586,
+      "step": 43717
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.805342435836792,
+      "learning_rate": 2.0246975468863696e-06,
+      "loss": 0.4002,
+      "step": 43718
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.4345550537109375,
+      "learning_rate": 2.0244856539774686e-06,
+      "loss": 0.5291,
+      "step": 43719
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 4.913769721984863,
+      "learning_rate": 2.024273769342355e-06,
+      "loss": 0.6253,
+      "step": 43720
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.4049644470214844,
+      "learning_rate": 2.024061892981625e-06,
+      "loss": 0.4375,
+      "step": 43721
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.54537296295166,
+      "learning_rate": 2.023850024895862e-06,
+      "loss": 0.7598,
+      "step": 43722
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 4.745248794555664,
+      "learning_rate": 2.023638165085659e-06,
+      "loss": 0.9051,
+      "step": 43723
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.220665454864502,
+      "learning_rate": 2.0234263135516014e-06,
+      "loss": 0.3368,
+      "step": 43724
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.8225195407867432,
+      "learning_rate": 2.023214470294282e-06,
+      "loss": 0.4367,
+      "step": 43725
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.2155139446258545,
+      "learning_rate": 2.0230026353142864e-06,
+      "loss": 0.7861,
+      "step": 43726
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.470954656600952,
+      "learning_rate": 2.022790808612207e-06,
+      "loss": 0.7544,
+      "step": 43727
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.7252037525177,
+      "learning_rate": 2.0225789901886293e-06,
+      "loss": 0.4807,
+      "step": 43728
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.3693995475769043,
+      "learning_rate": 2.022367180044146e-06,
+      "loss": 0.5977,
+      "step": 43729
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.812122344970703,
+      "learning_rate": 2.022155378179342e-06,
+      "loss": 0.8315,
+      "step": 43730
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.1696176528930664,
+      "learning_rate": 2.0219435845948105e-06,
+      "loss": 0.6581,
+      "step": 43731
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.067903757095337,
+      "learning_rate": 2.021731799291136e-06,
+      "loss": 0.5386,
+      "step": 43732
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.188056230545044,
+      "learning_rate": 2.021520022268912e-06,
+      "loss": 0.7808,
+      "step": 43733
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 6.066166877746582,
+      "learning_rate": 2.0213082535287224e-06,
+      "loss": 0.6072,
+      "step": 43734
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.3220436573028564,
+      "learning_rate": 2.0210964930711612e-06,
+      "loss": 0.6921,
+      "step": 43735
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.12863826751709,
+      "learning_rate": 2.0208847408968128e-06,
+      "loss": 0.6258,
+      "step": 43736
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.5243377685546875,
+      "learning_rate": 2.0206729970062703e-06,
+      "loss": 0.6587,
+      "step": 43737
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.7478079795837402,
+      "learning_rate": 2.020461261400118e-06,
+      "loss": 0.4107,
+      "step": 43738
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.0481204986572266,
+      "learning_rate": 2.0202495340789478e-06,
+      "loss": 0.6879,
+      "step": 43739
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.909482955932617,
+      "learning_rate": 2.0200378150433463e-06,
+      "loss": 0.6967,
+      "step": 43740
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.187070608139038,
+      "learning_rate": 2.0198261042939034e-06,
+      "loss": 0.5723,
+      "step": 43741
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.773911714553833,
+      "learning_rate": 2.0196144018312076e-06,
+      "loss": 0.7487,
+      "step": 43742
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.6678507328033447,
+      "learning_rate": 2.0194027076558494e-06,
+      "loss": 0.745,
+      "step": 43743
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 4.250921726226807,
+      "learning_rate": 2.019191021768414e-06,
+      "loss": 0.2654,
+      "step": 43744
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.866208553314209,
+      "learning_rate": 2.018979344169493e-06,
+      "loss": 0.6832,
+      "step": 43745
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.541828155517578,
+      "learning_rate": 2.0187676748596723e-06,
+      "loss": 0.6558,
+      "step": 43746
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.5112671852111816,
+      "learning_rate": 2.0185560138395444e-06,
+      "loss": 0.692,
+      "step": 43747
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.6514482498168945,
+      "learning_rate": 2.0183443611096926e-06,
+      "loss": 0.4333,
+      "step": 43748
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.86857533454895,
+      "learning_rate": 2.0181327166707102e-06,
+      "loss": 0.6017,
+      "step": 43749
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.486156463623047,
+      "learning_rate": 2.0179210805231818e-06,
+      "loss": 0.6826,
+      "step": 43750
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.7950117588043213,
+      "learning_rate": 2.0177094526676997e-06,
+      "loss": 0.7536,
+      "step": 43751
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.4458396434783936,
+      "learning_rate": 2.017497833104849e-06,
+      "loss": 0.5821,
+      "step": 43752
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.4580488204956055,
+      "learning_rate": 2.017286221835219e-06,
+      "loss": 0.5948,
+      "step": 43753
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.64145565032959,
+      "learning_rate": 2.017074618859401e-06,
+      "loss": 0.6622,
+      "step": 43754
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.3483078479766846,
+      "learning_rate": 2.0168630241779792e-06,
+      "loss": 0.5731,
+      "step": 43755
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.9200496673583984,
+      "learning_rate": 2.0166514377915447e-06,
+      "loss": 0.6619,
+      "step": 43756
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.7615249156951904,
+      "learning_rate": 2.016439859700684e-06,
+      "loss": 0.5289,
+      "step": 43757
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.8841595649719238,
+      "learning_rate": 2.016228289905988e-06,
+      "loss": 0.4833,
+      "step": 43758
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.173177480697632,
+      "learning_rate": 2.0160167284080416e-06,
+      "loss": 0.5849,
+      "step": 43759
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.5169425010681152,
+      "learning_rate": 2.0158051752074363e-06,
+      "loss": 0.5737,
+      "step": 43760
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.8177802562713623,
+      "learning_rate": 2.015593630304758e-06,
+      "loss": 0.6222,
+      "step": 43761
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.8951621055603027,
+      "learning_rate": 2.0153820937005953e-06,
+      "loss": 0.3931,
+      "step": 43762
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.258521795272827,
+      "learning_rate": 2.015170565395538e-06,
+      "loss": 0.6059,
+      "step": 43763
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.7000725269317627,
+      "learning_rate": 2.0149590453901723e-06,
+      "loss": 0.7,
+      "step": 43764
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.3662216663360596,
+      "learning_rate": 2.0147475336850874e-06,
+      "loss": 0.6777,
+      "step": 43765
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.9062745571136475,
+      "learning_rate": 2.0145360302808727e-06,
+      "loss": 0.5201,
+      "step": 43766
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.6922855377197266,
+      "learning_rate": 2.0143245351781133e-06,
+      "loss": 0.332,
+      "step": 43767
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.418760299682617,
+      "learning_rate": 2.014113048377401e-06,
+      "loss": 0.5659,
+      "step": 43768
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.425326347351074,
+      "learning_rate": 2.013901569879319e-06,
+      "loss": 0.4467,
+      "step": 43769
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.316826820373535,
+      "learning_rate": 2.0136900996844603e-06,
+      "loss": 0.682,
+      "step": 43770
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.394253730773926,
+      "learning_rate": 2.0134786377934096e-06,
+      "loss": 0.718,
+      "step": 43771
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.7665369510650635,
+      "learning_rate": 2.0132671842067576e-06,
+      "loss": 0.8552,
+      "step": 43772
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.9687927961349487,
+      "learning_rate": 2.0130557389250883e-06,
+      "loss": 0.3634,
+      "step": 43773
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.7911227941513062,
+      "learning_rate": 2.0128443019489946e-06,
+      "loss": 0.4796,
+      "step": 43774
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.703068971633911,
+      "learning_rate": 2.0126328732790594e-06,
+      "loss": 0.4845,
+      "step": 43775
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.9196641445159912,
+      "learning_rate": 2.0124214529158754e-06,
+      "loss": 0.5035,
+      "step": 43776
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.4147963523864746,
+      "learning_rate": 2.012210040860026e-06,
+      "loss": 0.6376,
+      "step": 43777
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.478778839111328,
+      "learning_rate": 2.011998637112103e-06,
+      "loss": 0.7208,
+      "step": 43778
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.4331903457641602,
+      "learning_rate": 2.0117872416726913e-06,
+      "loss": 0.452,
+      "step": 43779
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.4522523880004883,
+      "learning_rate": 2.011575854542381e-06,
+      "loss": 0.8401,
+      "step": 43780
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.649799346923828,
+      "learning_rate": 2.0113644757217554e-06,
+      "loss": 0.6631,
+      "step": 43781
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.290306568145752,
+      "learning_rate": 2.011153105211409e-06,
+      "loss": 0.6932,
+      "step": 43782
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.7885477542877197,
+      "learning_rate": 2.010941743011925e-06,
+      "loss": 0.6515,
+      "step": 43783
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.363175868988037,
+      "learning_rate": 2.0107303891238935e-06,
+      "loss": 0.4294,
+      "step": 43784
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.123812675476074,
+      "learning_rate": 2.0105190435478992e-06,
+      "loss": 0.4054,
+      "step": 43785
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.9448440074920654,
+      "learning_rate": 2.010307706284534e-06,
+      "loss": 0.7052,
+      "step": 43786
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 4.534937858581543,
+      "learning_rate": 2.0100963773343806e-06,
+      "loss": 0.8362,
+      "step": 43787
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.6602632999420166,
+      "learning_rate": 2.009885056698031e-06,
+      "loss": 0.6664,
+      "step": 43788
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 7.245453357696533,
+      "learning_rate": 2.0096737443760696e-06,
+      "loss": 0.5352,
+      "step": 43789
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.6255414485931396,
+      "learning_rate": 2.0094624403690867e-06,
+      "loss": 0.5219,
+      "step": 43790
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.2294554710388184,
+      "learning_rate": 2.0092511446776664e-06,
+      "loss": 0.4052,
+      "step": 43791
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.1024250984191895,
+      "learning_rate": 2.0090398573024007e-06,
+      "loss": 0.5737,
+      "step": 43792
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.473362445831299,
+      "learning_rate": 2.0088285782438725e-06,
+      "loss": 0.3569,
+      "step": 43793
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.1967902183532715,
+      "learning_rate": 2.0086173075026737e-06,
+      "loss": 0.442,
+      "step": 43794
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.74483585357666,
+      "learning_rate": 2.0084060450793872e-06,
+      "loss": 0.7108,
+      "step": 43795
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.159954786300659,
+      "learning_rate": 2.008194790974605e-06,
+      "loss": 0.6759,
+      "step": 43796
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.019899368286133,
+      "learning_rate": 2.007983545188911e-06,
+      "loss": 0.2927,
+      "step": 43797
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.25642728805542,
+      "learning_rate": 2.007772307722895e-06,
+      "loss": 0.6934,
+      "step": 43798
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.348574638366699,
+      "learning_rate": 2.0075610785771417e-06,
+      "loss": 0.5703,
+      "step": 43799
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.599066138267517,
+      "learning_rate": 2.0073498577522416e-06,
+      "loss": 0.3101,
+      "step": 43800
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.3533754348754883,
+      "learning_rate": 2.0071386452487788e-06,
+      "loss": 0.6001,
+      "step": 43801
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.694854259490967,
+      "learning_rate": 2.0069274410673426e-06,
+      "loss": 0.6813,
+      "step": 43802
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.2045626640319824,
+      "learning_rate": 2.0067162452085198e-06,
+      "loss": 0.5737,
+      "step": 43803
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.046992301940918,
+      "learning_rate": 2.0065050576728997e-06,
+      "loss": 0.4636,
+      "step": 43804
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 4.063770771026611,
+      "learning_rate": 2.0062938784610656e-06,
+      "loss": 0.3381,
+      "step": 43805
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.121269941329956,
+      "learning_rate": 2.006082707573609e-06,
+      "loss": 0.5289,
+      "step": 43806
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 5.480382442474365,
+      "learning_rate": 2.005871545011113e-06,
+      "loss": 0.5018,
+      "step": 43807
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.900762677192688,
+      "learning_rate": 2.005660390774168e-06,
+      "loss": 0.3473,
+      "step": 43808
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.883727788925171,
+      "learning_rate": 2.0054492448633576e-06,
+      "loss": 0.5225,
+      "step": 43809
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.6192768812179565,
+      "learning_rate": 2.0052381072792737e-06,
+      "loss": 0.3874,
+      "step": 43810
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.1847846508026123,
+      "learning_rate": 2.0050269780224986e-06,
+      "loss": 0.487,
+      "step": 43811
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.505537748336792,
+      "learning_rate": 2.004815857093623e-06,
+      "loss": 0.524,
+      "step": 43812
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.027256488800049,
+      "learning_rate": 2.0046047444932316e-06,
+      "loss": 0.3715,
+      "step": 43813
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.3348448276519775,
+      "learning_rate": 2.0043936402219117e-06,
+      "loss": 0.5941,
+      "step": 43814
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.6366970539093018,
+      "learning_rate": 2.0041825442802527e-06,
+      "loss": 0.7568,
+      "step": 43815
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.094589948654175,
+      "learning_rate": 2.003971456668838e-06,
+      "loss": 0.6391,
+      "step": 43816
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.9719792604446411,
+      "learning_rate": 2.003760377388258e-06,
+      "loss": 0.5552,
+      "step": 43817
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.8117401599884033,
+      "learning_rate": 2.0035493064390967e-06,
+      "loss": 0.7117,
+      "step": 43818
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.551696300506592,
+      "learning_rate": 2.0033382438219433e-06,
+      "loss": 0.6907,
+      "step": 43819
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.626009702682495,
+      "learning_rate": 2.003127189537382e-06,
+      "loss": 0.5504,
+      "step": 43820
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.921584129333496,
+      "learning_rate": 2.0029161435860038e-06,
+      "loss": 0.5631,
+      "step": 43821
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.984027862548828,
+      "learning_rate": 2.0027051059683913e-06,
+      "loss": 0.3626,
+      "step": 43822
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.6086812019348145,
+      "learning_rate": 2.002494076685132e-06,
+      "loss": 0.5877,
+      "step": 43823
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.6853981018066406,
+      "learning_rate": 2.0022830557368165e-06,
+      "loss": 0.7053,
+      "step": 43824
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.0829644203186035,
+      "learning_rate": 2.002072043124027e-06,
+      "loss": 0.5766,
+      "step": 43825
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.7679457664489746,
+      "learning_rate": 2.001861038847352e-06,
+      "loss": 0.6789,
+      "step": 43826
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.1236393451690674,
+      "learning_rate": 2.00165004290738e-06,
+      "loss": 0.6134,
+      "step": 43827
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.9965240955352783,
+      "learning_rate": 2.001439055304694e-06,
+      "loss": 0.8075,
+      "step": 43828
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.7216756343841553,
+      "learning_rate": 2.001228076039885e-06,
+      "loss": 0.5344,
+      "step": 43829
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.496338367462158,
+      "learning_rate": 2.001017105113535e-06,
+      "loss": 0.5646,
+      "step": 43830
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.964496612548828,
+      "learning_rate": 2.0008061425262353e-06,
+      "loss": 0.691,
+      "step": 43831
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.8853694200515747,
+      "learning_rate": 2.000595188278568e-06,
+      "loss": 0.5142,
+      "step": 43832
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.312338352203369,
+      "learning_rate": 2.000384242371124e-06,
+      "loss": 0.5319,
+      "step": 43833
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.753112316131592,
+      "learning_rate": 2.0001733048044857e-06,
+      "loss": 0.7707,
+      "step": 43834
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.8907129764556885,
+      "learning_rate": 1.9999623755792434e-06,
+      "loss": 0.5928,
+      "step": 43835
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.2686495780944824,
+      "learning_rate": 1.9997514546959805e-06,
+      "loss": 0.7115,
+      "step": 43836
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.0581533908843994,
+      "learning_rate": 1.9995405421552865e-06,
+      "loss": 0.7054,
+      "step": 43837
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.836247205734253,
+      "learning_rate": 1.9993296379577443e-06,
+      "loss": 0.6483,
+      "step": 43838
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.8248698711395264,
+      "learning_rate": 1.999118742103944e-06,
+      "loss": 0.7892,
+      "step": 43839
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.713738441467285,
+      "learning_rate": 1.998907854594469e-06,
+      "loss": 0.6342,
+      "step": 43840
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.726465940475464,
+      "learning_rate": 1.9986969754299084e-06,
+      "loss": 0.6227,
+      "step": 43841
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.132756233215332,
+      "learning_rate": 1.9984861046108437e-06,
+      "loss": 0.2678,
+      "step": 43842
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.2880594730377197,
+      "learning_rate": 1.9982752421378688e-06,
+      "loss": 0.6879,
+      "step": 43843
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.23711895942688,
+      "learning_rate": 1.9980643880115637e-06,
+      "loss": 0.4623,
+      "step": 43844
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.063847541809082,
+      "learning_rate": 1.9978535422325192e-06,
+      "loss": 0.7691,
+      "step": 43845
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.5908267498016357,
+      "learning_rate": 1.997642704801317e-06,
+      "loss": 0.3788,
+      "step": 43846
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.8365485668182373,
+      "learning_rate": 1.997431875718548e-06,
+      "loss": 0.5296,
+      "step": 43847
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.9393540620803833,
+      "learning_rate": 1.9972210549847936e-06,
+      "loss": 0.5062,
+      "step": 43848
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.096174478530884,
+      "learning_rate": 1.9970102426006457e-06,
+      "loss": 0.4156,
+      "step": 43849
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.4611852169036865,
+      "learning_rate": 1.9967994385666845e-06,
+      "loss": 0.8011,
+      "step": 43850
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.0610697269439697,
+      "learning_rate": 1.996588642883501e-06,
+      "loss": 0.6605,
+      "step": 43851
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.3049232959747314,
+      "learning_rate": 1.996377855551678e-06,
+      "loss": 0.3898,
+      "step": 43852
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.3581714630126953,
+      "learning_rate": 1.9961670765718045e-06,
+      "loss": 0.6073,
+      "step": 43853
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.19897723197937,
+      "learning_rate": 1.995956305944463e-06,
+      "loss": 0.5555,
+      "step": 43854
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.339000701904297,
+      "learning_rate": 1.995745543670244e-06,
+      "loss": 0.6605,
+      "step": 43855
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.5525286197662354,
+      "learning_rate": 1.995534789749729e-06,
+      "loss": 0.6102,
+      "step": 43856
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.4938085079193115,
+      "learning_rate": 1.995324044183508e-06,
+      "loss": 0.5485,
+      "step": 43857
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.879350423812866,
+      "learning_rate": 1.995113306972164e-06,
+      "loss": 0.5663,
+      "step": 43858
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.800755262374878,
+      "learning_rate": 1.9949025781162857e-06,
+      "loss": 0.5677,
+      "step": 43859
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.6602330207824707,
+      "learning_rate": 1.994691857616456e-06,
+      "loss": 0.4191,
+      "step": 43860
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.0589756965637207,
+      "learning_rate": 1.994481145473264e-06,
+      "loss": 0.6573,
+      "step": 43861
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.365234613418579,
+      "learning_rate": 1.994270441687292e-06,
+      "loss": 0.6719,
+      "step": 43862
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.786130666732788,
+      "learning_rate": 1.994059746259128e-06,
+      "loss": 0.6001,
+      "step": 43863
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.684525966644287,
+      "learning_rate": 1.9938490591893585e-06,
+      "loss": 0.6798,
+      "step": 43864
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.8976852893829346,
+      "learning_rate": 1.9936383804785704e-06,
+      "loss": 0.5672,
+      "step": 43865
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.9862393140792847,
+      "learning_rate": 1.9934277101273452e-06,
+      "loss": 0.5016,
+      "step": 43866
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.130186080932617,
+      "learning_rate": 1.9932170481362735e-06,
+      "loss": 0.5281,
+      "step": 43867
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.3499884605407715,
+      "learning_rate": 1.993006394505937e-06,
+      "loss": 0.7376,
+      "step": 43868
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.567749500274658,
+      "learning_rate": 1.992795749236925e-06,
+      "loss": 0.8475,
+      "step": 43869
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.4895846843719482,
+      "learning_rate": 1.9925851123298194e-06,
+      "loss": 0.6937,
+      "step": 43870
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.9803935289382935,
+      "learning_rate": 1.9923744837852098e-06,
+      "loss": 0.4322,
+      "step": 43871
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.4137818813323975,
+      "learning_rate": 1.9921638636036787e-06,
+      "loss": 0.7028,
+      "step": 43872
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.7271308898925781,
+      "learning_rate": 1.9919532517858146e-06,
+      "loss": 0.3326,
+      "step": 43873
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.517348527908325,
+      "learning_rate": 1.9917426483322004e-06,
+      "loss": 0.6,
+      "step": 43874
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.6443371772766113,
+      "learning_rate": 1.9915320532434246e-06,
+      "loss": 0.6673,
+      "step": 43875
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.220245599746704,
+      "learning_rate": 1.9913214665200694e-06,
+      "loss": 0.5695,
+      "step": 43876
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.848456859588623,
+      "learning_rate": 1.991110888162722e-06,
+      "loss": 0.602,
+      "step": 43877
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.0886576175689697,
+      "learning_rate": 1.9909003181719692e-06,
+      "loss": 0.579,
+      "step": 43878
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.8259358406066895,
+      "learning_rate": 1.990689756548394e-06,
+      "loss": 0.5683,
+      "step": 43879
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.3049070835113525,
+      "learning_rate": 1.990479203292586e-06,
+      "loss": 0.5888,
+      "step": 43880
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.320880651473999,
+      "learning_rate": 1.990268658405125e-06,
+      "loss": 0.4553,
+      "step": 43881
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.39255428314209,
+      "learning_rate": 1.9900581218866015e-06,
+      "loss": 0.7105,
+      "step": 43882
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.298344373703003,
+      "learning_rate": 1.9898475937375973e-06,
+      "loss": 0.6111,
+      "step": 43883
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.7473578453063965,
+      "learning_rate": 1.989637073958699e-06,
+      "loss": 0.7016,
+      "step": 43884
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.5672335624694824,
+      "learning_rate": 1.989426562550494e-06,
+      "loss": 0.6225,
+      "step": 43885
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.8623881340026855,
+      "learning_rate": 1.9892160595135644e-06,
+      "loss": 0.5262,
+      "step": 43886
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.5745420455932617,
+      "learning_rate": 1.989005564848497e-06,
+      "loss": 0.6719,
+      "step": 43887
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.925106406211853,
+      "learning_rate": 1.988795078555879e-06,
+      "loss": 0.3779,
+      "step": 43888
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 6.500831127166748,
+      "learning_rate": 1.988584600636292e-06,
+      "loss": 0.6037,
+      "step": 43889
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.8960713148117065,
+      "learning_rate": 1.988374131090325e-06,
+      "loss": 0.6151,
+      "step": 43890
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.94566011428833,
+      "learning_rate": 1.9881636699185595e-06,
+      "loss": 0.5564,
+      "step": 43891
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.2360715866088867,
+      "learning_rate": 1.9879532171215843e-06,
+      "loss": 0.4337,
+      "step": 43892
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.12790584564209,
+      "learning_rate": 1.9877427726999814e-06,
+      "loss": 0.5285,
+      "step": 43893
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 4.669090747833252,
+      "learning_rate": 1.9875323366543385e-06,
+      "loss": 0.5028,
+      "step": 43894
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.4400227069854736,
+      "learning_rate": 1.9873219089852385e-06,
+      "loss": 0.5567,
+      "step": 43895
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.8265511989593506,
+      "learning_rate": 1.98711148969327e-06,
+      "loss": 0.4371,
+      "step": 43896
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.89251708984375,
+      "learning_rate": 1.9869010787790127e-06,
+      "loss": 0.5253,
+      "step": 43897
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.157301902770996,
+      "learning_rate": 1.986690676243058e-06,
+      "loss": 0.5268,
+      "step": 43898
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.612882137298584,
+      "learning_rate": 1.9864802820859846e-06,
+      "loss": 0.2869,
+      "step": 43899
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.2774155139923096,
+      "learning_rate": 1.986269896308383e-06,
+      "loss": 0.4146,
+      "step": 43900
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.947032928466797,
+      "learning_rate": 1.986059518910834e-06,
+      "loss": 0.5226,
+      "step": 43901
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.8947871923446655,
+      "learning_rate": 1.9858491498939264e-06,
+      "loss": 0.6435,
+      "step": 43902
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.9576165676116943,
+      "learning_rate": 1.9856387892582397e-06,
+      "loss": 0.6991,
+      "step": 43903
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.34026837348938,
+      "learning_rate": 1.9854284370043654e-06,
+      "loss": 0.6305,
+      "step": 43904
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 4.859524726867676,
+      "learning_rate": 1.985218093132884e-06,
+      "loss": 0.7839,
+      "step": 43905
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.6409401893615723,
+      "learning_rate": 1.9850077576443834e-06,
+      "loss": 0.7922,
+      "step": 43906
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.2924294471740723,
+      "learning_rate": 1.9847974305394446e-06,
+      "loss": 0.6186,
+      "step": 43907
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.0245869159698486,
+      "learning_rate": 1.9845871118186565e-06,
+      "loss": 0.6348,
+      "step": 43908
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.808877944946289,
+      "learning_rate": 1.984376801482601e-06,
+      "loss": 0.5771,
+      "step": 43909
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.3384110927581787,
+      "learning_rate": 1.984166499531865e-06,
+      "loss": 0.654,
+      "step": 43910
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.5250842571258545,
+      "learning_rate": 1.9839562059670307e-06,
+      "loss": 0.6728,
+      "step": 43911
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.392928123474121,
+      "learning_rate": 1.9837459207886862e-06,
+      "loss": 0.6192,
+      "step": 43912
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.127748966217041,
+      "learning_rate": 1.9835356439974126e-06,
+      "loss": 0.6303,
+      "step": 43913
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.9163116216659546,
+      "learning_rate": 1.9833253755937976e-06,
+      "loss": 0.7401,
+      "step": 43914
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.2498583793640137,
+      "learning_rate": 1.9831151155784236e-06,
+      "loss": 0.8269,
+      "step": 43915
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.2323668003082275,
+      "learning_rate": 1.982904863951878e-06,
+      "loss": 0.5402,
+      "step": 43916
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.9938833713531494,
+      "learning_rate": 1.9826946207147413e-06,
+      "loss": 0.3802,
+      "step": 43917
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 1.6931581497192383,
+      "learning_rate": 1.982484385867603e-06,
+      "loss": 0.1801,
+      "step": 43918
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.169710397720337,
+      "learning_rate": 1.9822741594110435e-06,
+      "loss": 0.5939,
+      "step": 43919
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.5220208168029785,
+      "learning_rate": 1.9820639413456504e-06,
+      "loss": 0.5825,
+      "step": 43920
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.9684159755706787,
+      "learning_rate": 1.9818537316720055e-06,
+      "loss": 0.8,
+      "step": 43921
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.870692491531372,
+      "learning_rate": 1.981643530390696e-06,
+      "loss": 0.8982,
+      "step": 43922
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.558464765548706,
+      "learning_rate": 1.9814333375023036e-06,
+      "loss": 0.6117,
+      "step": 43923
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.234966993331909,
+      "learning_rate": 1.981223153007415e-06,
+      "loss": 0.6324,
+      "step": 43924
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.285738706588745,
+      "learning_rate": 1.9810129769066135e-06,
+      "loss": 0.8437,
+      "step": 43925
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.1314921379089355,
+      "learning_rate": 1.9808028092004856e-06,
+      "loss": 0.6033,
+      "step": 43926
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.9036505222320557,
+      "learning_rate": 1.980592649889612e-06,
+      "loss": 0.7526,
+      "step": 43927
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.005342483520508,
+      "learning_rate": 1.9803824989745817e-06,
+      "loss": 0.5788,
+      "step": 43928
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.9062671661376953,
+      "learning_rate": 1.9801723564559746e-06,
+      "loss": 0.7368,
+      "step": 43929
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.239006996154785,
+      "learning_rate": 1.9799622223343778e-06,
+      "loss": 0.5366,
+      "step": 43930
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 1.6629550457000732,
+      "learning_rate": 1.979752096610374e-06,
+      "loss": 0.3482,
+      "step": 43931
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.6652607917785645,
+      "learning_rate": 1.9795419792845495e-06,
+      "loss": 0.7701,
+      "step": 43932
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.7446272373199463,
+      "learning_rate": 1.979331870357485e-06,
+      "loss": 0.7662,
+      "step": 43933
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.359945058822632,
+      "learning_rate": 1.97912176982977e-06,
+      "loss": 0.6701,
+      "step": 43934
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.160202980041504,
+      "learning_rate": 1.978911677701983e-06,
+      "loss": 0.728,
+      "step": 43935
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.3163278102874756,
+      "learning_rate": 1.978701593974713e-06,
+      "loss": 0.7115,
+      "step": 43936
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.209279775619507,
+      "learning_rate": 1.97849151864854e-06,
+      "loss": 0.7417,
+      "step": 43937
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.6416449546813965,
+      "learning_rate": 1.9782814517240505e-06,
+      "loss": 0.6299,
+      "step": 43938
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.305102825164795,
+      "learning_rate": 1.9780713932018302e-06,
+      "loss": 0.6727,
+      "step": 43939
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 4.019337177276611,
+      "learning_rate": 1.9778613430824593e-06,
+      "loss": 0.8933,
+      "step": 43940
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.25032901763916,
+      "learning_rate": 1.9776513013665255e-06,
+      "loss": 0.7617,
+      "step": 43941
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.4555559158325195,
+      "learning_rate": 1.9774412680546096e-06,
+      "loss": 0.5424,
+      "step": 43942
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.1246798038482666,
+      "learning_rate": 1.9772312431472997e-06,
+      "loss": 0.5987,
+      "step": 43943
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.2422261238098145,
+      "learning_rate": 1.9770212266451744e-06,
+      "loss": 0.651,
+      "step": 43944
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.5897765159606934,
+      "learning_rate": 1.9768112185488216e-06,
+      "loss": 0.805,
+      "step": 43945
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.155219316482544,
+      "learning_rate": 1.9766012188588256e-06,
+      "loss": 0.3972,
+      "step": 43946
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 4.290799140930176,
+      "learning_rate": 1.9763912275757676e-06,
+      "loss": 0.7555,
+      "step": 43947
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.3100764751434326,
+      "learning_rate": 1.9761812447002345e-06,
+      "loss": 0.7352,
+      "step": 43948
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 4.0039567947387695,
+      "learning_rate": 1.9759712702328066e-06,
+      "loss": 0.6151,
+      "step": 43949
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 1.5059821605682373,
+      "learning_rate": 1.97576130417407e-06,
+      "loss": 0.2816,
+      "step": 43950
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.811694383621216,
+      "learning_rate": 1.9755513465246095e-06,
+      "loss": 0.7429,
+      "step": 43951
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.268646001815796,
+      "learning_rate": 1.9753413972850068e-06,
+      "loss": 0.539,
+      "step": 43952
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.904146671295166,
+      "learning_rate": 1.9751314564558483e-06,
+      "loss": 0.6915,
+      "step": 43953
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.477562189102173,
+      "learning_rate": 1.974921524037714e-06,
+      "loss": 0.4985,
+      "step": 43954
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.888510227203369,
+      "learning_rate": 1.974711600031191e-06,
+      "loss": 0.3427,
+      "step": 43955
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.281104564666748,
+      "learning_rate": 1.9745016844368607e-06,
+      "loss": 0.6253,
+      "step": 43956
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.0240747928619385,
+      "learning_rate": 1.9742917772553093e-06,
+      "loss": 0.6076,
+      "step": 43957
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.5404675006866455,
+      "learning_rate": 1.974081878487117e-06,
+      "loss": 0.7064,
+      "step": 43958
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.45223331451416,
+      "learning_rate": 1.9738719881328722e-06,
+      "loss": 0.411,
+      "step": 43959
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.4109878540039062,
+      "learning_rate": 1.9736621061931534e-06,
+      "loss": 0.7203,
+      "step": 43960
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.057332754135132,
+      "learning_rate": 1.9734522326685485e-06,
+      "loss": 0.8177,
+      "step": 43961
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.317762613296509,
+      "learning_rate": 1.9732423675596373e-06,
+      "loss": 0.7095,
+      "step": 43962
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.2823641300201416,
+      "learning_rate": 1.973032510867007e-06,
+      "loss": 0.5932,
+      "step": 43963
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.5343971252441406,
+      "learning_rate": 1.9728226625912356e-06,
+      "loss": 0.6202,
+      "step": 43964
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.5543200969696045,
+      "learning_rate": 1.9726128227329144e-06,
+      "loss": 0.6127,
+      "step": 43965
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 6.815539360046387,
+      "learning_rate": 1.9724029912926206e-06,
+      "loss": 0.7954,
+      "step": 43966
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.2916388511657715,
+      "learning_rate": 1.972193168270942e-06,
+      "loss": 0.6171,
+      "step": 43967
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 1.9956625699996948,
+      "learning_rate": 1.971983353668458e-06,
+      "loss": 0.3704,
+      "step": 43968
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.0892250537872314,
+      "learning_rate": 1.971773547485756e-06,
+      "loss": 0.6035,
+      "step": 43969
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.2775957584381104,
+      "learning_rate": 1.9715637497234153e-06,
+      "loss": 0.7207,
+      "step": 43970
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 1.9388136863708496,
+      "learning_rate": 1.971353960382023e-06,
+      "loss": 0.468,
+      "step": 43971
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.886359930038452,
+      "learning_rate": 1.9711441794621593e-06,
+      "loss": 0.7317,
+      "step": 43972
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.5939536094665527,
+      "learning_rate": 1.970934406964411e-06,
+      "loss": 0.7943,
+      "step": 43973
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.052931070327759,
+      "learning_rate": 1.9707246428893573e-06,
+      "loss": 0.5106,
+      "step": 43974
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.6423394680023193,
+      "learning_rate": 1.970514887237585e-06,
+      "loss": 0.4175,
+      "step": 43975
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.5455188751220703,
+      "learning_rate": 1.9703051400096744e-06,
+      "loss": 0.656,
+      "step": 43976
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.3726260662078857,
+      "learning_rate": 1.9700954012062124e-06,
+      "loss": 0.6533,
+      "step": 43977
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.9446303844451904,
+      "learning_rate": 1.969885670827778e-06,
+      "loss": 0.9858,
+      "step": 43978
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 1.9908353090286255,
+      "learning_rate": 1.9696759488749583e-06,
+      "loss": 0.5916,
+      "step": 43979
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.8049373626708984,
+      "learning_rate": 1.969466235348333e-06,
+      "loss": 0.3314,
+      "step": 43980
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.5332040786743164,
+      "learning_rate": 1.9692565302484886e-06,
+      "loss": 0.6052,
+      "step": 43981
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.431806802749634,
+      "learning_rate": 1.9690468335760037e-06,
+      "loss": 0.4257,
+      "step": 43982
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.6509323120117188,
+      "learning_rate": 1.968837145331467e-06,
+      "loss": 0.5363,
+      "step": 43983
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.7643771171569824,
+      "learning_rate": 1.9686274655154565e-06,
+      "loss": 0.9986,
+      "step": 43984
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.344010353088379,
+      "learning_rate": 1.9684177941285576e-06,
+      "loss": 0.6493,
+      "step": 43985
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.3083345890045166,
+      "learning_rate": 1.968208131171353e-06,
+      "loss": 0.6282,
+      "step": 43986
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.2594239711761475,
+      "learning_rate": 1.9679984766444278e-06,
+      "loss": 0.7587,
+      "step": 43987
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.9765470027923584,
+      "learning_rate": 1.967788830548361e-06,
+      "loss": 0.6407,
+      "step": 43988
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.6341207027435303,
+      "learning_rate": 1.9675791928837395e-06,
+      "loss": 0.9332,
+      "step": 43989
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 4.574825763702393,
+      "learning_rate": 1.9673695636511424e-06,
+      "loss": 0.8869,
+      "step": 43990
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.70493221282959,
+      "learning_rate": 1.967159942851157e-06,
+      "loss": 0.3954,
+      "step": 43991
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.650665521621704,
+      "learning_rate": 1.9669503304843614e-06,
+      "loss": 0.6524,
+      "step": 43992
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.553516149520874,
+      "learning_rate": 1.966740726551342e-06,
+      "loss": 0.8477,
+      "step": 43993
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 1.757808804512024,
+      "learning_rate": 1.966531131052679e-06,
+      "loss": 0.5804,
+      "step": 43994
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.6745338439941406,
+      "learning_rate": 1.9663215439889584e-06,
+      "loss": 0.645,
+      "step": 43995
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.5027847290039062,
+      "learning_rate": 1.966111965360759e-06,
+      "loss": 0.6366,
+      "step": 43996
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.694272756576538,
+      "learning_rate": 1.965902395168668e-06,
+      "loss": 0.6893,
+      "step": 43997
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 1.7536834478378296,
+      "learning_rate": 1.9656928334132642e-06,
+      "loss": 0.4439,
+      "step": 43998
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.7018685340881348,
+      "learning_rate": 1.965483280095131e-06,
+      "loss": 0.6982,
+      "step": 43999
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 1.8936845064163208,
+      "learning_rate": 1.9652737352148547e-06,
+      "loss": 0.4823,
+      "step": 44000
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 5.822364330291748,
+      "learning_rate": 1.965064198773013e-06,
+      "loss": 0.559,
+      "step": 44001
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.131640672683716,
+      "learning_rate": 1.964854670770193e-06,
+      "loss": 0.8354,
+      "step": 44002
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.0038487911224365,
+      "learning_rate": 1.964645151206973e-06,
+      "loss": 0.649,
+      "step": 44003
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 1.6379101276397705,
+      "learning_rate": 1.96443564008394e-06,
+      "loss": 0.3846,
+      "step": 44004
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.795602321624756,
+      "learning_rate": 1.9642261374016724e-06,
+      "loss": 0.6264,
+      "step": 44005
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 4.103457450866699,
+      "learning_rate": 1.964016643160755e-06,
+      "loss": 0.4942,
+      "step": 44006
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.351439952850342,
+      "learning_rate": 1.9638071573617713e-06,
+      "loss": 0.5306,
+      "step": 44007
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.7356624603271484,
+      "learning_rate": 1.963597680005301e-06,
+      "loss": 0.4188,
+      "step": 44008
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.9719109535217285,
+      "learning_rate": 1.9633882110919296e-06,
+      "loss": 0.6384,
+      "step": 44009
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.9562439918518066,
+      "learning_rate": 1.9631787506222362e-06,
+      "loss": 0.6179,
+      "step": 44010
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.6646013259887695,
+      "learning_rate": 1.9629692985968053e-06,
+      "loss": 0.5615,
+      "step": 44011
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.7844228744506836,
+      "learning_rate": 1.962759855016221e-06,
+      "loss": 0.6297,
+      "step": 44012
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.26016902923584,
+      "learning_rate": 1.962550419881061e-06,
+      "loss": 0.3981,
+      "step": 44013
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.2292184829711914,
+      "learning_rate": 1.9623409931919137e-06,
+      "loss": 0.5585,
+      "step": 44014
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.3093786239624023,
+      "learning_rate": 1.962131574949356e-06,
+      "loss": 0.577,
+      "step": 44015
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 6.727781295776367,
+      "learning_rate": 1.9619221651539733e-06,
+      "loss": 0.3259,
+      "step": 44016
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.802884578704834,
+      "learning_rate": 1.961712763806346e-06,
+      "loss": 0.6562,
+      "step": 44017
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.2707319259643555,
+      "learning_rate": 1.9615033709070587e-06,
+      "loss": 0.5801,
+      "step": 44018
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.281355857849121,
+      "learning_rate": 1.961293986456691e-06,
+      "loss": 0.6915,
+      "step": 44019
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.077387809753418,
+      "learning_rate": 1.9610846104558274e-06,
+      "loss": 0.6569,
+      "step": 44020
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.481736183166504,
+      "learning_rate": 1.9608752429050483e-06,
+      "loss": 0.4548,
+      "step": 44021
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.479677677154541,
+      "learning_rate": 1.9606658838049377e-06,
+      "loss": 0.7701,
+      "step": 44022
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 6.881393909454346,
+      "learning_rate": 1.960456533156075e-06,
+      "loss": 0.4605,
+      "step": 44023
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.9818739891052246,
+      "learning_rate": 1.960247190959046e-06,
+      "loss": 0.6075,
+      "step": 44024
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.3348820209503174,
+      "learning_rate": 1.9600378572144275e-06,
+      "loss": 0.7113,
+      "step": 44025
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.4929025173187256,
+      "learning_rate": 1.9598285319228078e-06,
+      "loss": 0.7396,
+      "step": 44026
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 1.731176733970642,
+      "learning_rate": 1.959619215084765e-06,
+      "loss": 0.6992,
+      "step": 44027
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.590301275253296,
+      "learning_rate": 1.9594099067008837e-06,
+      "loss": 0.5991,
+      "step": 44028
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.643418073654175,
+      "learning_rate": 1.959200606771743e-06,
+      "loss": 0.7281,
+      "step": 44029
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.622083902359009,
+      "learning_rate": 1.9589913152979275e-06,
+      "loss": 0.7573,
+      "step": 44030
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.8455545902252197,
+      "learning_rate": 1.958782032280016e-06,
+      "loss": 0.622,
+      "step": 44031
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.1335318088531494,
+      "learning_rate": 1.9585727577185947e-06,
+      "loss": 0.664,
+      "step": 44032
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.5265989303588867,
+      "learning_rate": 1.958363491614241e-06,
+      "loss": 0.6296,
+      "step": 44033
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.2390964031219482,
+      "learning_rate": 1.958154233967542e-06,
+      "loss": 0.6314,
+      "step": 44034
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.66161847114563,
+      "learning_rate": 1.957944984779074e-06,
+      "loss": 0.4103,
+      "step": 44035
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.2273738384246826,
+      "learning_rate": 1.9577357440494227e-06,
+      "loss": 0.4624,
+      "step": 44036
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 1.8451961278915405,
+      "learning_rate": 1.9575265117791674e-06,
+      "loss": 0.3903,
+      "step": 44037
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.7097556591033936,
+      "learning_rate": 1.957317287968893e-06,
+      "loss": 0.9307,
+      "step": 44038
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.3797764778137207,
+      "learning_rate": 1.957108072619177e-06,
+      "loss": 0.6289,
+      "step": 44039
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.7524776458740234,
+      "learning_rate": 1.956898865730606e-06,
+      "loss": 0.6737,
+      "step": 44040
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.1742093563079834,
+      "learning_rate": 1.956689667303757e-06,
+      "loss": 0.931,
+      "step": 44041
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.3362576961517334,
+      "learning_rate": 1.956480477339216e-06,
+      "loss": 0.4796,
+      "step": 44042
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.409745931625366,
+      "learning_rate": 1.9562712958375614e-06,
+      "loss": 0.5,
+      "step": 44043
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.4641780853271484,
+      "learning_rate": 1.956062122799377e-06,
+      "loss": 0.5755,
+      "step": 44044
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.7442193031311035,
+      "learning_rate": 1.955852958225242e-06,
+      "loss": 0.4035,
+      "step": 44045
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.128420352935791,
+      "learning_rate": 1.9556438021157397e-06,
+      "loss": 0.4037,
+      "step": 44046
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.8365166187286377,
+      "learning_rate": 1.955434654471452e-06,
+      "loss": 0.4854,
+      "step": 44047
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 1.868829607963562,
+      "learning_rate": 1.955225515292961e-06,
+      "loss": 0.4111,
+      "step": 44048
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.6405680179595947,
+      "learning_rate": 1.955016384580846e-06,
+      "loss": 0.6576,
+      "step": 44049
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.4491825103759766,
+      "learning_rate": 1.9548072623356916e-06,
+      "loss": 0.6563,
+      "step": 44050
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.5307042598724365,
+      "learning_rate": 1.9545981485580756e-06,
+      "loss": 0.4841,
+      "step": 44051
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 1.9767305850982666,
+      "learning_rate": 1.954389043248583e-06,
+      "loss": 0.3468,
+      "step": 44052
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.2237019538879395,
+      "learning_rate": 1.9541799464077915e-06,
+      "loss": 0.828,
+      "step": 44053
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.3177170753479004,
+      "learning_rate": 1.9539708580362867e-06,
+      "loss": 0.4951,
+      "step": 44054
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 5.330734729766846,
+      "learning_rate": 1.953761778134646e-06,
+      "loss": 0.5686,
+      "step": 44055
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.1625046730041504,
+      "learning_rate": 1.9535527067034545e-06,
+      "loss": 0.6047,
+      "step": 44056
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.0654590129852295,
+      "learning_rate": 1.9533436437432896e-06,
+      "loss": 0.5462,
+      "step": 44057
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.5721092224121094,
+      "learning_rate": 1.9531345892547367e-06,
+      "loss": 0.8222,
+      "step": 44058
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.985065221786499,
+      "learning_rate": 1.9529255432383733e-06,
+      "loss": 0.4871,
+      "step": 44059
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.139035701751709,
+      "learning_rate": 1.9527165056947843e-06,
+      "loss": 0.5534,
+      "step": 44060
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 1.7429170608520508,
+      "learning_rate": 1.952507476624547e-06,
+      "loss": 0.4027,
+      "step": 44061
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.5101706981658936,
+      "learning_rate": 1.9522984560282454e-06,
+      "loss": 0.4505,
+      "step": 44062
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.533625602722168,
+      "learning_rate": 1.952089443906462e-06,
+      "loss": 0.807,
+      "step": 44063
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.1478641033172607,
+      "learning_rate": 1.951880440259774e-06,
+      "loss": 0.7869,
+      "step": 44064
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.088902473449707,
+      "learning_rate": 1.9516714450887665e-06,
+      "loss": 0.3018,
+      "step": 44065
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.8631198406219482,
+      "learning_rate": 1.9514624583940167e-06,
+      "loss": 0.5901,
+      "step": 44066
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.391171932220459,
+      "learning_rate": 1.951253480176108e-06,
+      "loss": 0.5037,
+      "step": 44067
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.456270217895508,
+      "learning_rate": 1.951044510435623e-06,
+      "loss": 0.543,
+      "step": 44068
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.3307111263275146,
+      "learning_rate": 1.9508355491731396e-06,
+      "loss": 0.5428,
+      "step": 44069
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.4922397136688232,
+      "learning_rate": 1.9506265963892416e-06,
+      "loss": 0.6263,
+      "step": 44070
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.265049934387207,
+      "learning_rate": 1.9504176520845077e-06,
+      "loss": 0.9126,
+      "step": 44071
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.917389154434204,
+      "learning_rate": 1.950208716259519e-06,
+      "loss": 0.484,
+      "step": 44072
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.344189167022705,
+      "learning_rate": 1.9499997889148596e-06,
+      "loss": 0.6783,
+      "step": 44073
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.5299103260040283,
+      "learning_rate": 1.9497908700511067e-06,
+      "loss": 0.6789,
+      "step": 44074
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.5423121452331543,
+      "learning_rate": 1.9495819596688437e-06,
+      "loss": 0.4556,
+      "step": 44075
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.9382026195526123,
+      "learning_rate": 1.9493730577686493e-06,
+      "loss": 0.6572,
+      "step": 44076
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 1.9658005237579346,
+      "learning_rate": 1.949164164351108e-06,
+      "loss": 0.4462,
+      "step": 44077
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.322002410888672,
+      "learning_rate": 1.9489552794167952e-06,
+      "loss": 0.6907,
+      "step": 44078
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 1.8099994659423828,
+      "learning_rate": 1.9487464029662978e-06,
+      "loss": 0.5241,
+      "step": 44079
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.365560293197632,
+      "learning_rate": 1.948537535000191e-06,
+      "loss": 0.5405,
+      "step": 44080
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.4701690673828125,
+      "learning_rate": 1.9483286755190605e-06,
+      "loss": 0.5062,
+      "step": 44081
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.803483009338379,
+      "learning_rate": 1.948119824523483e-06,
+      "loss": 0.6742,
+      "step": 44082
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.7531092166900635,
+      "learning_rate": 1.9479109820140424e-06,
+      "loss": 0.5321,
+      "step": 44083
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.0090060234069824,
+      "learning_rate": 1.947702147991317e-06,
+      "loss": 0.758,
+      "step": 44084
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 4.0643439292907715,
+      "learning_rate": 1.9474933224558893e-06,
+      "loss": 0.5684,
+      "step": 44085
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.4325547218322754,
+      "learning_rate": 1.947284505408336e-06,
+      "loss": 0.5539,
+      "step": 44086
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.91729736328125,
+      "learning_rate": 1.947075696849245e-06,
+      "loss": 0.6816,
+      "step": 44087
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.9695868492126465,
+      "learning_rate": 1.9468668967791904e-06,
+      "loss": 0.5933,
+      "step": 44088
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.118513822555542,
+      "learning_rate": 1.946658105198757e-06,
+      "loss": 0.5232,
+      "step": 44089
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 5.1204023361206055,
+      "learning_rate": 1.9464493221085224e-06,
+      "loss": 0.6311,
+      "step": 44090
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.6699159145355225,
+      "learning_rate": 1.9462405475090697e-06,
+      "loss": 0.6261,
+      "step": 44091
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.1457607746124268,
+      "learning_rate": 1.9460317814009766e-06,
+      "loss": 0.6012,
+      "step": 44092
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.1796770095825195,
+      "learning_rate": 1.9458230237848263e-06,
+      "loss": 0.6179,
+      "step": 44093
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 4.901657581329346,
+      "learning_rate": 1.945614274661197e-06,
+      "loss": 0.6682,
+      "step": 44094
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 1.5535964965820312,
+      "learning_rate": 1.9454055340306715e-06,
+      "loss": 0.5213,
+      "step": 44095
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.105745315551758,
+      "learning_rate": 1.945196801893827e-06,
+      "loss": 0.5008,
+      "step": 44096
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.228332996368408,
+      "learning_rate": 1.944988078251248e-06,
+      "loss": 0.6286,
+      "step": 44097
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.1366639137268066,
+      "learning_rate": 1.9447793631035106e-06,
+      "loss": 0.656,
+      "step": 44098
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 8.163475036621094,
+      "learning_rate": 1.9445706564511995e-06,
+      "loss": 0.6577,
+      "step": 44099
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.090022563934326,
+      "learning_rate": 1.944361958294891e-06,
+      "loss": 0.4409,
+      "step": 44100
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.982530355453491,
+      "learning_rate": 1.9441532686351682e-06,
+      "loss": 0.5725,
+      "step": 44101
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.454939365386963,
+      "learning_rate": 1.9439445874726094e-06,
+      "loss": 0.5014,
+      "step": 44102
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.788416624069214,
+      "learning_rate": 1.9437359148077977e-06,
+      "loss": 0.536,
+      "step": 44103
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 4.043964862823486,
+      "learning_rate": 1.9435272506413095e-06,
+      "loss": 0.7447,
+      "step": 44104
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.0021109580993652,
+      "learning_rate": 1.9433185949737287e-06,
+      "loss": 0.3738,
+      "step": 44105
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.4051425457000732,
+      "learning_rate": 1.943109947805632e-06,
+      "loss": 0.7423,
+      "step": 44106
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.245375394821167,
+      "learning_rate": 1.942901309137602e-06,
+      "loss": 0.6541,
+      "step": 44107
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.6935250759124756,
+      "learning_rate": 1.9426926789702178e-06,
+      "loss": 0.5076,
+      "step": 44108
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.5123860836029053,
+      "learning_rate": 1.9424840573040617e-06,
+      "loss": 0.5463,
+      "step": 44109
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 1.9463391304016113,
+      "learning_rate": 1.9422754441397103e-06,
+      "loss": 0.3721,
+      "step": 44110
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.4032809734344482,
+      "learning_rate": 1.942066839477747e-06,
+      "loss": 0.6383,
+      "step": 44111
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.081162452697754,
+      "learning_rate": 1.941858243318749e-06,
+      "loss": 0.748,
+      "step": 44112
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.7935168743133545,
+      "learning_rate": 1.941649655663299e-06,
+      "loss": 0.8908,
+      "step": 44113
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.152747392654419,
+      "learning_rate": 1.941441076511974e-06,
+      "loss": 0.5507,
+      "step": 44114
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.171609401702881,
+      "learning_rate": 1.9412325058653575e-06,
+      "loss": 0.4809,
+      "step": 44115
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.1202187538146973,
+      "learning_rate": 1.941023943724026e-06,
+      "loss": 0.4658,
+      "step": 44116
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.3970119953155518,
+      "learning_rate": 1.940815390088562e-06,
+      "loss": 0.7297,
+      "step": 44117
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.4436163902282715,
+      "learning_rate": 1.9406068449595434e-06,
+      "loss": 0.8031,
+      "step": 44118
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.0288994312286377,
+      "learning_rate": 1.9403983083375527e-06,
+      "loss": 0.6602,
+      "step": 44119
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 13.889643669128418,
+      "learning_rate": 1.9401897802231663e-06,
+      "loss": 0.658,
+      "step": 44120
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 4.709784030914307,
+      "learning_rate": 1.939981260616967e-06,
+      "loss": 0.5669,
+      "step": 44121
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.421086549758911,
+      "learning_rate": 1.939772749519533e-06,
+      "loss": 0.4535,
+      "step": 44122
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.703423500061035,
+      "learning_rate": 1.9395642469314433e-06,
+      "loss": 0.6101,
+      "step": 44123
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.4121294021606445,
+      "learning_rate": 1.939355752853282e-06,
+      "loss": 0.6862,
+      "step": 44124
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.9542806148529053,
+      "learning_rate": 1.939147267285623e-06,
+      "loss": 0.5598,
+      "step": 44125
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 1.7558029890060425,
+      "learning_rate": 1.9389387902290513e-06,
+      "loss": 0.2734,
+      "step": 44126
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.37461519241333,
+      "learning_rate": 1.9387303216841417e-06,
+      "loss": 0.7948,
+      "step": 44127
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.4638078212738037,
+      "learning_rate": 1.938521861651477e-06,
+      "loss": 0.4415,
+      "step": 44128
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.176330089569092,
+      "learning_rate": 1.9383134101316377e-06,
+      "loss": 0.481,
+      "step": 44129
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.0894269943237305,
+      "learning_rate": 1.9381049671251996e-06,
+      "loss": 0.3706,
+      "step": 44130
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.2779700756073,
+      "learning_rate": 1.937896532632747e-06,
+      "loss": 0.6748,
+      "step": 44131
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.1259303092956543,
+      "learning_rate": 1.9376881066548547e-06,
+      "loss": 0.8064,
+      "step": 44132
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.633479595184326,
+      "learning_rate": 1.9374796891921066e-06,
+      "loss": 0.3495,
+      "step": 44133
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 1.8389724493026733,
+      "learning_rate": 1.937271280245079e-06,
+      "loss": 0.2886,
+      "step": 44134
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.530278205871582,
+      "learning_rate": 1.937062879814352e-06,
+      "loss": 0.5816,
+      "step": 44135
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.1555938720703125,
+      "learning_rate": 1.9368544879005076e-06,
+      "loss": 0.4777,
+      "step": 44136
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.0870652198791504,
+      "learning_rate": 1.936646104504122e-06,
+      "loss": 0.8157,
+      "step": 44137
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.1449923515319824,
+      "learning_rate": 1.9364377296257775e-06,
+      "loss": 0.6241,
+      "step": 44138
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 5.512479782104492,
+      "learning_rate": 1.93622936326605e-06,
+      "loss": 0.762,
+      "step": 44139
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 7.542909145355225,
+      "learning_rate": 1.936021005425523e-06,
+      "loss": 0.5628,
+      "step": 44140
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.192105293273926,
+      "learning_rate": 1.935812656104772e-06,
+      "loss": 0.6672,
+      "step": 44141
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.0463314056396484,
+      "learning_rate": 1.93560431530438e-06,
+      "loss": 0.6202,
+      "step": 44142
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.260669231414795,
+      "learning_rate": 1.9353959830249223e-06,
+      "loss": 0.8082,
+      "step": 44143
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.901526689529419,
+      "learning_rate": 1.9351876592669826e-06,
+      "loss": 0.6569,
+      "step": 44144
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.1148111820220947,
+      "learning_rate": 1.934979344031136e-06,
+      "loss": 0.728,
+      "step": 44145
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.4462428092956543,
+      "learning_rate": 1.934771037317965e-06,
+      "loss": 0.3991,
+      "step": 44146
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.6263535022735596,
+      "learning_rate": 1.934562739128045e-06,
+      "loss": 0.5559,
+      "step": 44147
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 1.726152777671814,
+      "learning_rate": 1.93435444946196e-06,
+      "loss": 0.5732,
+      "step": 44148
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.1469573974609375,
+      "learning_rate": 1.9341461683202856e-06,
+      "loss": 0.8539,
+      "step": 44149
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.2057740688323975,
+      "learning_rate": 1.933937895703604e-06,
+      "loss": 0.5174,
+      "step": 44150
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.0488693714141846,
+      "learning_rate": 1.933729631612491e-06,
+      "loss": 0.4851,
+      "step": 44151
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.5952811241149902,
+      "learning_rate": 1.9335213760475285e-06,
+      "loss": 0.66,
+      "step": 44152
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.829702615737915,
+      "learning_rate": 1.9333131290092927e-06,
+      "loss": 0.4264,
+      "step": 44153
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.6464157104492188,
+      "learning_rate": 1.933104890498366e-06,
+      "loss": 0.456,
+      "step": 44154
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.2247891426086426,
+      "learning_rate": 1.932896660515324e-06,
+      "loss": 0.6854,
+      "step": 44155
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 4.2982306480407715,
+      "learning_rate": 1.93268843906075e-06,
+      "loss": 0.6757,
+      "step": 44156
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 10.422322273254395,
+      "learning_rate": 1.9324802261352176e-06,
+      "loss": 0.5613,
+      "step": 44157
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.685236692428589,
+      "learning_rate": 1.932272021739311e-06,
+      "loss": 0.5055,
+      "step": 44158
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 8.18406867980957,
+      "learning_rate": 1.9320638258736044e-06,
+      "loss": 0.7227,
+      "step": 44159
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.5215225219726562,
+      "learning_rate": 1.931855638538681e-06,
+      "loss": 0.7216,
+      "step": 44160
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.057371139526367,
+      "learning_rate": 1.931647459735116e-06,
+      "loss": 0.5237,
+      "step": 44161
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.1298344135284424,
+      "learning_rate": 1.931439289463492e-06,
+      "loss": 0.5405,
+      "step": 44162
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.508040189743042,
+      "learning_rate": 1.931231127724384e-06,
+      "loss": 0.4925,
+      "step": 44163
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.7767207622528076,
+      "learning_rate": 1.931022974518374e-06,
+      "loss": 0.6015,
+      "step": 44164
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.323225736618042,
+      "learning_rate": 1.930814829846038e-06,
+      "loss": 0.7597,
+      "step": 44165
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.0970492362976074,
+      "learning_rate": 1.9306066937079576e-06,
+      "loss": 0.6928,
+      "step": 44166
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.7679998874664307,
+      "learning_rate": 1.930398566104709e-06,
+      "loss": 0.572,
+      "step": 44167
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.986797571182251,
+      "learning_rate": 1.9301904470368717e-06,
+      "loss": 0.3775,
+      "step": 44168
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.344473361968994,
+      "learning_rate": 1.929982336505025e-06,
+      "loss": 0.6364,
+      "step": 44169
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.434084415435791,
+      "learning_rate": 1.9297742345097487e-06,
+      "loss": 0.6624,
+      "step": 44170
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.9761557579040527,
+      "learning_rate": 1.9295661410516186e-06,
+      "loss": 0.5264,
+      "step": 44171
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.7492337226867676,
+      "learning_rate": 1.9293580561312164e-06,
+      "loss": 0.4545,
+      "step": 44172
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.054774761199951,
+      "learning_rate": 1.9291499797491176e-06,
+      "loss": 0.4302,
+      "step": 44173
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.51967716217041,
+      "learning_rate": 1.9289419119059042e-06,
+      "loss": 0.8856,
+      "step": 44174
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.1283774375915527,
+      "learning_rate": 1.928733852602151e-06,
+      "loss": 0.7296,
+      "step": 44175
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.0828945636749268,
+      "learning_rate": 1.928525801838441e-06,
+      "loss": 0.3336,
+      "step": 44176
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.9109079837799072,
+      "learning_rate": 1.9283177596153474e-06,
+      "loss": 0.5824,
+      "step": 44177
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.501671075820923,
+      "learning_rate": 1.928109725933453e-06,
+      "loss": 0.7211,
+      "step": 44178
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.5704197883605957,
+      "learning_rate": 1.9279017007933336e-06,
+      "loss": 0.6086,
+      "step": 44179
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.8084335327148438,
+      "learning_rate": 1.9276936841955703e-06,
+      "loss": 0.6012,
+      "step": 44180
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.07163667678833,
+      "learning_rate": 1.927485676140738e-06,
+      "loss": 0.5693,
+      "step": 44181
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.4031121730804443,
+      "learning_rate": 1.9272776766294192e-06,
+      "loss": 0.4472,
+      "step": 44182
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.0416722297668457,
+      "learning_rate": 1.9270696856621877e-06,
+      "loss": 0.4937,
+      "step": 44183
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 4.144235134124756,
+      "learning_rate": 1.9268617032396254e-06,
+      "loss": 0.6627,
+      "step": 44184
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.164731025695801,
+      "learning_rate": 1.92665372936231e-06,
+      "loss": 0.5117,
+      "step": 44185
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 1.5930955410003662,
+      "learning_rate": 1.9264457640308183e-06,
+      "loss": 0.3609,
+      "step": 44186
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.8350491523742676,
+      "learning_rate": 1.926237807245731e-06,
+      "loss": 0.5399,
+      "step": 44187
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.723578453063965,
+      "learning_rate": 1.926029859007623e-06,
+      "loss": 0.6102,
+      "step": 44188
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 1.6915655136108398,
+      "learning_rate": 1.9258219193170748e-06,
+      "loss": 0.3656,
+      "step": 44189
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.439262866973877,
+      "learning_rate": 1.925613988174666e-06,
+      "loss": 0.8392,
+      "step": 44190
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.9711532592773438,
+      "learning_rate": 1.9254060655809707e-06,
+      "loss": 0.4838,
+      "step": 44191
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.4780304431915283,
+      "learning_rate": 1.925198151536571e-06,
+      "loss": 0.7274,
+      "step": 44192
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.273170232772827,
+      "learning_rate": 1.9249902460420427e-06,
+      "loss": 0.805,
+      "step": 44193
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.5753045082092285,
+      "learning_rate": 1.924782349097966e-06,
+      "loss": 0.5304,
+      "step": 44194
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.7776286602020264,
+      "learning_rate": 1.9245744607049155e-06,
+      "loss": 0.5322,
+      "step": 44195
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.621575355529785,
+      "learning_rate": 1.9243665808634714e-06,
+      "loss": 0.4899,
+      "step": 44196
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.6322832107543945,
+      "learning_rate": 1.9241587095742136e-06,
+      "loss": 0.5568,
+      "step": 44197
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.0218443870544434,
+      "learning_rate": 1.923950846837716e-06,
+      "loss": 0.652,
+      "step": 44198
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 4.022764682769775,
+      "learning_rate": 1.923742992654561e-06,
+      "loss": 0.3001,
+      "step": 44199
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 6.905288219451904,
+      "learning_rate": 1.923535147025323e-06,
+      "loss": 0.4715,
+      "step": 44200
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.3914554119110107,
+      "learning_rate": 1.9233273099505824e-06,
+      "loss": 0.5406,
+      "step": 44201
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 1.9918146133422852,
+      "learning_rate": 1.923119481430914e-06,
+      "loss": 0.4032,
+      "step": 44202
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 1.9000767469406128,
+      "learning_rate": 1.9229116614668995e-06,
+      "loss": 0.5083,
+      "step": 44203
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.7138302326202393,
+      "learning_rate": 1.922703850059114e-06,
+      "loss": 0.6512,
+      "step": 44204
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 1.8960223197937012,
+      "learning_rate": 1.922496047208137e-06,
+      "loss": 0.6803,
+      "step": 44205
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.3817083835601807,
+      "learning_rate": 1.922288252914545e-06,
+      "loss": 0.5451,
+      "step": 44206
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.033625841140747,
+      "learning_rate": 1.9220804671789173e-06,
+      "loss": 0.4637,
+      "step": 44207
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.1060800552368164,
+      "learning_rate": 1.9218726900018275e-06,
+      "loss": 0.5903,
+      "step": 44208
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.489621639251709,
+      "learning_rate": 1.921664921383861e-06,
+      "loss": 0.7884,
+      "step": 44209
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 1.883476734161377,
+      "learning_rate": 1.9214571613255885e-06,
+      "loss": 0.3876,
+      "step": 44210
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.728362560272217,
+      "learning_rate": 1.9212494098275917e-06,
+      "loss": 0.4885,
+      "step": 44211
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.980403184890747,
+      "learning_rate": 1.9210416668904464e-06,
+      "loss": 0.5021,
+      "step": 44212
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 1.9305545091629028,
+      "learning_rate": 1.920833932514732e-06,
+      "loss": 0.5277,
+      "step": 44213
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.317110538482666,
+      "learning_rate": 1.920626206701023e-06,
+      "loss": 0.6139,
+      "step": 44214
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.491948366165161,
+      "learning_rate": 1.920418489449901e-06,
+      "loss": 0.45,
+      "step": 44215
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.9857823848724365,
+      "learning_rate": 1.9202107807619396e-06,
+      "loss": 0.6303,
+      "step": 44216
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.0161690711975098,
+      "learning_rate": 1.92000308063772e-06,
+      "loss": 0.5588,
+      "step": 44217
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.3301262855529785,
+      "learning_rate": 1.919795389077817e-06,
+      "loss": 0.5946,
+      "step": 44218
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 1.8101136684417725,
+      "learning_rate": 1.9195877060828105e-06,
+      "loss": 0.5585,
+      "step": 44219
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.219438076019287,
+      "learning_rate": 1.9193800316532745e-06,
+      "loss": 0.6676,
+      "step": 44220
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.3760764598846436,
+      "learning_rate": 1.919172365789791e-06,
+      "loss": 0.4311,
+      "step": 44221
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 1.8973915576934814,
+      "learning_rate": 1.9189647084929326e-06,
+      "loss": 0.4804,
+      "step": 44222
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.4565377235412598,
+      "learning_rate": 1.918757059763281e-06,
+      "loss": 0.7032,
+      "step": 44223
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 4.379382133483887,
+      "learning_rate": 1.9185494196014105e-06,
+      "loss": 0.548,
+      "step": 44224
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.393908977508545,
+      "learning_rate": 1.9183417880079014e-06,
+      "loss": 0.2896,
+      "step": 44225
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.0854132175445557,
+      "learning_rate": 1.9181341649833275e-06,
+      "loss": 0.5574,
+      "step": 44226
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.519660234451294,
+      "learning_rate": 1.9179265505282697e-06,
+      "loss": 0.7958,
+      "step": 44227
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.165825605392456,
+      "learning_rate": 1.9177189446433016e-06,
+      "loss": 0.6577,
+      "step": 44228
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 5.02900505065918,
+      "learning_rate": 1.9175113473290025e-06,
+      "loss": 0.659,
+      "step": 44229
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.4412150382995605,
+      "learning_rate": 1.91730375858595e-06,
+      "loss": 0.545,
+      "step": 44230
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.760355234146118,
+      "learning_rate": 1.9170961784147223e-06,
+      "loss": 0.8836,
+      "step": 44231
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.8016245365142822,
+      "learning_rate": 1.9168886068158937e-06,
+      "loss": 0.6694,
+      "step": 44232
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.535933494567871,
+      "learning_rate": 1.916681043790045e-06,
+      "loss": 0.7414,
+      "step": 44233
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 7.008730411529541,
+      "learning_rate": 1.916473489337749e-06,
+      "loss": 0.8662,
+      "step": 44234
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.3852193355560303,
+      "learning_rate": 1.916265943459587e-06,
+      "loss": 0.5368,
+      "step": 44235
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.4243969917297363,
+      "learning_rate": 1.9160584061561326e-06,
+      "loss": 0.8042,
+      "step": 44236
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.468106746673584,
+      "learning_rate": 1.9158508774279665e-06,
+      "loss": 0.9229,
+      "step": 44237
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.3807756900787354,
+      "learning_rate": 1.9156433572756618e-06,
+      "loss": 0.8304,
+      "step": 44238
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.0627737045288086,
+      "learning_rate": 1.915435845699799e-06,
+      "loss": 0.3989,
+      "step": 44239
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.6791272163391113,
+      "learning_rate": 1.915228342700952e-06,
+      "loss": 0.4676,
+      "step": 44240
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.7598910331726074,
+      "learning_rate": 1.915020848279701e-06,
+      "loss": 0.4767,
+      "step": 44241
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.2606072425842285,
+      "learning_rate": 1.91481336243662e-06,
+      "loss": 0.5993,
+      "step": 44242
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.441833734512329,
+      "learning_rate": 1.9146058851722885e-06,
+      "loss": 0.5853,
+      "step": 44243
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.9782567024230957,
+      "learning_rate": 1.9143984164872807e-06,
+      "loss": 0.6729,
+      "step": 44244
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.640138626098633,
+      "learning_rate": 1.914190956382177e-06,
+      "loss": 0.6724,
+      "step": 44245
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.328817367553711,
+      "learning_rate": 1.9139835048575507e-06,
+      "loss": 0.8101,
+      "step": 44246
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 6.929866313934326,
+      "learning_rate": 1.9137760619139796e-06,
+      "loss": 0.6298,
+      "step": 44247
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.6035265922546387,
+      "learning_rate": 1.913568627552043e-06,
+      "loss": 0.5432,
+      "step": 44248
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.9387197494506836,
+      "learning_rate": 1.913361201772315e-06,
+      "loss": 0.797,
+      "step": 44249
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.9657974243164062,
+      "learning_rate": 1.9131537845753723e-06,
+      "loss": 0.6595,
+      "step": 44250
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.1750733852386475,
+      "learning_rate": 1.9129463759617946e-06,
+      "loss": 0.4339,
+      "step": 44251
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.4100100994110107,
+      "learning_rate": 1.9127389759321546e-06,
+      "loss": 0.6555,
+      "step": 44252
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 18.785249710083008,
+      "learning_rate": 1.912531584487033e-06,
+      "loss": 0.5492,
+      "step": 44253
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 6.1456122398376465,
+      "learning_rate": 1.912324201627002e-06,
+      "loss": 0.6457,
+      "step": 44254
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 4.023388385772705,
+      "learning_rate": 1.9121168273526436e-06,
+      "loss": 0.7613,
+      "step": 44255
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.301820993423462,
+      "learning_rate": 1.911909461664529e-06,
+      "loss": 0.5394,
+      "step": 44256
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.237161636352539,
+      "learning_rate": 1.911702104563238e-06,
+      "loss": 0.6072,
+      "step": 44257
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 6.305270671844482,
+      "learning_rate": 1.911494756049348e-06,
+      "loss": 0.4879,
+      "step": 44258
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 4.743282318115234,
+      "learning_rate": 1.9112874161234325e-06,
+      "loss": 0.5467,
+      "step": 44259
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.9324402809143066,
+      "learning_rate": 1.911080084786071e-06,
+      "loss": 0.6736,
+      "step": 44260
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.2138423919677734,
+      "learning_rate": 1.910872762037837e-06,
+      "loss": 0.7603,
+      "step": 44261
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.8030261993408203,
+      "learning_rate": 1.9106654478793107e-06,
+      "loss": 0.4841,
+      "step": 44262
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.8867404460906982,
+      "learning_rate": 1.910458142311064e-06,
+      "loss": 0.5328,
+      "step": 44263
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.4714066982269287,
+      "learning_rate": 1.910250845333678e-06,
+      "loss": 0.4453,
+      "step": 44264
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.049257278442383,
+      "learning_rate": 1.910043556947726e-06,
+      "loss": 0.6199,
+      "step": 44265
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.035794973373413,
+      "learning_rate": 1.909836277153786e-06,
+      "loss": 0.3686,
+      "step": 44266
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.2024941444396973,
+      "learning_rate": 1.9096290059524326e-06,
+      "loss": 0.6603,
+      "step": 44267
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.1665472984313965,
+      "learning_rate": 1.909421743344245e-06,
+      "loss": 0.7366,
+      "step": 44268
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.750532388687134,
+      "learning_rate": 1.9092144893297938e-06,
+      "loss": 0.5789,
+      "step": 44269
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.1551144123077393,
+      "learning_rate": 1.909007243909663e-06,
+      "loss": 0.7628,
+      "step": 44270
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 4.375731468200684,
+      "learning_rate": 1.9088000070844237e-06,
+      "loss": 0.45,
+      "step": 44271
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.8546130657196045,
+      "learning_rate": 1.908592778854655e-06,
+      "loss": 0.7722,
+      "step": 44272
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.5581793785095215,
+      "learning_rate": 1.90838555922093e-06,
+      "loss": 0.4809,
+      "step": 44273
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.0201656818389893,
+      "learning_rate": 1.9081783481838286e-06,
+      "loss": 0.3869,
+      "step": 44274
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.448866367340088,
+      "learning_rate": 1.907971145743923e-06,
+      "loss": 0.739,
+      "step": 44275
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.7465932369232178,
+      "learning_rate": 1.9077639519017927e-06,
+      "loss": 0.7207,
+      "step": 44276
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.343313217163086,
+      "learning_rate": 1.9075567666580114e-06,
+      "loss": 0.534,
+      "step": 44277
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.117673873901367,
+      "learning_rate": 1.907349590013158e-06,
+      "loss": 0.6496,
+      "step": 44278
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.2366855144500732,
+      "learning_rate": 1.9071424219678048e-06,
+      "loss": 0.3891,
+      "step": 44279
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.8014001846313477,
+      "learning_rate": 1.9069352625225318e-06,
+      "loss": 0.6889,
+      "step": 44280
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.195197105407715,
+      "learning_rate": 1.9067281116779113e-06,
+      "loss": 0.5892,
+      "step": 44281
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 4.3851799964904785,
+      "learning_rate": 1.9065209694345226e-06,
+      "loss": 0.7105,
+      "step": 44282
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.89345383644104,
+      "learning_rate": 1.9063138357929388e-06,
+      "loss": 0.5124,
+      "step": 44283
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.3735697269439697,
+      "learning_rate": 1.906106710753739e-06,
+      "loss": 0.8383,
+      "step": 44284
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 1.5201116800308228,
+      "learning_rate": 1.9058995943174964e-06,
+      "loss": 0.5091,
+      "step": 44285
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.4530487060546875,
+      "learning_rate": 1.9056924864847886e-06,
+      "loss": 0.8341,
+      "step": 44286
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.1590428352355957,
+      "learning_rate": 1.9054853872561896e-06,
+      "loss": 0.5001,
+      "step": 44287
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.4316463470458984,
+      "learning_rate": 1.9052782966322786e-06,
+      "loss": 0.8061,
+      "step": 44288
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.723438262939453,
+      "learning_rate": 1.9050712146136268e-06,
+      "loss": 0.7137,
+      "step": 44289
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.273526668548584,
+      "learning_rate": 1.9048641412008135e-06,
+      "loss": 0.6492,
+      "step": 44290
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.1919729709625244,
+      "learning_rate": 1.9046570763944127e-06,
+      "loss": 0.461,
+      "step": 44291
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.4637577533721924,
+      "learning_rate": 1.9044500201950034e-06,
+      "loss": 0.6832,
+      "step": 44292
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.5514440536499023,
+      "learning_rate": 1.9042429726031569e-06,
+      "loss": 0.5572,
+      "step": 44293
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.692840337753296,
+      "learning_rate": 1.9040359336194525e-06,
+      "loss": 0.4985,
+      "step": 44294
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.940708875656128,
+      "learning_rate": 1.9038289032444628e-06,
+      "loss": 0.7844,
+      "step": 44295
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.8216943740844727,
+      "learning_rate": 1.903621881478767e-06,
+      "loss": 0.7167,
+      "step": 44296
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.7279584407806396,
+      "learning_rate": 1.903414868322937e-06,
+      "loss": 0.4533,
+      "step": 44297
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 1.9351369142532349,
+      "learning_rate": 1.9032078637775524e-06,
+      "loss": 0.5779,
+      "step": 44298
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.689138650894165,
+      "learning_rate": 1.9030008678431843e-06,
+      "loss": 0.5009,
+      "step": 44299
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 4.976554870605469,
+      "learning_rate": 1.9027938805204126e-06,
+      "loss": 0.518,
+      "step": 44300
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.271589994430542,
+      "learning_rate": 1.9025869018098093e-06,
+      "loss": 0.604,
+      "step": 44301
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.6416103839874268,
+      "learning_rate": 1.902379931711953e-06,
+      "loss": 0.49,
+      "step": 44302
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.436964511871338,
+      "learning_rate": 1.902172970227416e-06,
+      "loss": 0.5925,
+      "step": 44303
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.128178119659424,
+      "learning_rate": 1.9019660173567773e-06,
+      "loss": 0.6128,
+      "step": 44304
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.8455100059509277,
+      "learning_rate": 1.9017590731006085e-06,
+      "loss": 0.6638,
+      "step": 44305
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 1.8451611995697021,
+      "learning_rate": 1.9015521374594892e-06,
+      "loss": 0.641,
+      "step": 44306
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.451035499572754,
+      "learning_rate": 1.9013452104339913e-06,
+      "loss": 0.7109,
+      "step": 44307
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.145550012588501,
+      "learning_rate": 1.9011382920246913e-06,
+      "loss": 0.4881,
+      "step": 44308
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 5.943131923675537,
+      "learning_rate": 1.9009313822321662e-06,
+      "loss": 0.6372,
+      "step": 44309
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.8721723556518555,
+      "learning_rate": 1.9007244810569886e-06,
+      "loss": 0.5103,
+      "step": 44310
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 10.118925094604492,
+      "learning_rate": 1.9005175884997352e-06,
+      "loss": 0.7651,
+      "step": 44311
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.3897151947021484,
+      "learning_rate": 1.9003107045609826e-06,
+      "loss": 0.6644,
+      "step": 44312
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.724527597427368,
+      "learning_rate": 1.9001038292413039e-06,
+      "loss": 0.7835,
+      "step": 44313
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.1035749912261963,
+      "learning_rate": 1.899896962541276e-06,
+      "loss": 0.7102,
+      "step": 44314
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.3448166847229004,
+      "learning_rate": 1.899690104461472e-06,
+      "loss": 0.4771,
+      "step": 44315
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 4.38736629486084,
+      "learning_rate": 1.89948325500247e-06,
+      "loss": 0.6891,
+      "step": 44316
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.64866304397583,
+      "learning_rate": 1.8992764141648423e-06,
+      "loss": 0.8748,
+      "step": 44317
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 1.9898232221603394,
+      "learning_rate": 1.8990695819491667e-06,
+      "loss": 0.4518,
+      "step": 44318
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.3274059295654297,
+      "learning_rate": 1.8988627583560155e-06,
+      "loss": 0.6748,
+      "step": 44319
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.743640661239624,
+      "learning_rate": 1.8986559433859646e-06,
+      "loss": 0.5687,
+      "step": 44320
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.648577928543091,
+      "learning_rate": 1.8984491370395924e-06,
+      "loss": 0.4444,
+      "step": 44321
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 1.9229884147644043,
+      "learning_rate": 1.8982423393174686e-06,
+      "loss": 0.4164,
+      "step": 44322
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.5206382274627686,
+      "learning_rate": 1.8980355502201736e-06,
+      "loss": 0.5622,
+      "step": 44323
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 4.103919506072998,
+      "learning_rate": 1.8978287697482773e-06,
+      "loss": 0.6511,
+      "step": 44324
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.8575172424316406,
+      "learning_rate": 1.8976219979023591e-06,
+      "loss": 0.6511,
+      "step": 44325
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.878737211227417,
+      "learning_rate": 1.89741523468299e-06,
+      "loss": 0.7281,
+      "step": 44326
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 1.9748014211654663,
+      "learning_rate": 1.8972084800907486e-06,
+      "loss": 0.4926,
+      "step": 44327
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.8864526748657227,
+      "learning_rate": 1.8970017341262064e-06,
+      "loss": 0.6978,
+      "step": 44328
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.821488857269287,
+      "learning_rate": 1.896794996789942e-06,
+      "loss": 0.5962,
+      "step": 44329
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.2634522914886475,
+      "learning_rate": 1.8965882680825264e-06,
+      "loss": 0.4372,
+      "step": 44330
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 1.7715380191802979,
+      "learning_rate": 1.896381548004536e-06,
+      "loss": 0.5669,
+      "step": 44331
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.882988929748535,
+      "learning_rate": 1.8961748365565463e-06,
+      "loss": 0.5503,
+      "step": 44332
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 4.315252304077148,
+      "learning_rate": 1.8959681337391329e-06,
+      "loss": 0.6723,
+      "step": 44333
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 1.996585488319397,
+      "learning_rate": 1.8957614395528678e-06,
+      "loss": 0.5758,
+      "step": 44334
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 4.534004211425781,
+      "learning_rate": 1.8955547539983293e-06,
+      "loss": 0.5467,
+      "step": 44335
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.61276912689209,
+      "learning_rate": 1.895348077076088e-06,
+      "loss": 0.4213,
+      "step": 44336
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.2465121746063232,
+      "learning_rate": 1.895141408786722e-06,
+      "loss": 0.5388,
+      "step": 44337
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.2862062454223633,
+      "learning_rate": 1.8949347491308035e-06,
+      "loss": 0.7695,
+      "step": 44338
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.983097791671753,
+      "learning_rate": 1.89472809810891e-06,
+      "loss": 0.4761,
+      "step": 44339
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.996922492980957,
+      "learning_rate": 1.8945214557216124e-06,
+      "loss": 0.684,
+      "step": 44340
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.3215432167053223,
+      "learning_rate": 1.894314821969489e-06,
+      "loss": 0.5555,
+      "step": 44341
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.2419629096984863,
+      "learning_rate": 1.8941081968531111e-06,
+      "loss": 0.779,
+      "step": 44342
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.1361501216888428,
+      "learning_rate": 1.8939015803730564e-06,
+      "loss": 0.3661,
+      "step": 44343
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.317160129547119,
+      "learning_rate": 1.8936949725298963e-06,
+      "loss": 0.4799,
+      "step": 44344
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.524625539779663,
+      "learning_rate": 1.8934883733242087e-06,
+      "loss": 0.6662,
+      "step": 44345
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 6.485041618347168,
+      "learning_rate": 1.8932817827565642e-06,
+      "loss": 0.6648,
+      "step": 44346
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.625958204269409,
+      "learning_rate": 1.893075200827541e-06,
+      "loss": 0.6199,
+      "step": 44347
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 4.892845630645752,
+      "learning_rate": 1.89286862753771e-06,
+      "loss": 0.5965,
+      "step": 44348
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.2303884029388428,
+      "learning_rate": 1.8926620628876496e-06,
+      "loss": 0.739,
+      "step": 44349
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.7433836460113525,
+      "learning_rate": 1.8924555068779298e-06,
+      "loss": 0.696,
+      "step": 44350
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 4.571873188018799,
+      "learning_rate": 1.892248959509127e-06,
+      "loss": 0.4638,
+      "step": 44351
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.6141536235809326,
+      "learning_rate": 1.892042420781816e-06,
+      "loss": 0.8989,
+      "step": 44352
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.734781265258789,
+      "learning_rate": 1.8918358906965722e-06,
+      "loss": 0.6244,
+      "step": 44353
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.1296088695526123,
+      "learning_rate": 1.8916293692539673e-06,
+      "loss": 0.5774,
+      "step": 44354
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 1.6661417484283447,
+      "learning_rate": 1.8914228564545777e-06,
+      "loss": 0.3665,
+      "step": 44355
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 1.7991615533828735,
+      "learning_rate": 1.8912163522989751e-06,
+      "loss": 0.4117,
+      "step": 44356
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.5766851902008057,
+      "learning_rate": 1.8910098567877372e-06,
+      "loss": 0.4881,
+      "step": 44357
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.4412083625793457,
+      "learning_rate": 1.8908033699214346e-06,
+      "loss": 0.6692,
+      "step": 44358
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 4.17774772644043,
+      "learning_rate": 1.8905968917006445e-06,
+      "loss": 0.4753,
+      "step": 44359
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.1561689376831055,
+      "learning_rate": 1.8903904221259378e-06,
+      "loss": 0.6238,
+      "step": 44360
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.548469305038452,
+      "learning_rate": 1.8901839611978922e-06,
+      "loss": 0.7231,
+      "step": 44361
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.0542094707489014,
+      "learning_rate": 1.8899775089170786e-06,
+      "loss": 0.749,
+      "step": 44362
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.4290363788604736,
+      "learning_rate": 1.8897710652840746e-06,
+      "loss": 0.861,
+      "step": 44363
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 1.8243091106414795,
+      "learning_rate": 1.8895646302994498e-06,
+      "loss": 0.4428,
+      "step": 44364
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 1.8777905702590942,
+      "learning_rate": 1.889358203963782e-06,
+      "loss": 0.4677,
+      "step": 44365
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.199140787124634,
+      "learning_rate": 1.8891517862776426e-06,
+      "loss": 0.5073,
+      "step": 44366
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.5948336124420166,
+      "learning_rate": 1.8889453772416083e-06,
+      "loss": 0.6791,
+      "step": 44367
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.1217072010040283,
+      "learning_rate": 1.8887389768562497e-06,
+      "loss": 0.6519,
+      "step": 44368
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.7000484466552734,
+      "learning_rate": 1.8885325851221425e-06,
+      "loss": 0.6447,
+      "step": 44369
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.541492223739624,
+      "learning_rate": 1.888326202039863e-06,
+      "loss": 0.7164,
+      "step": 44370
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.9623117446899414,
+      "learning_rate": 1.88811982760998e-06,
+      "loss": 0.6343,
+      "step": 44371
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.507298469543457,
+      "learning_rate": 1.8879134618330702e-06,
+      "loss": 0.6871,
+      "step": 44372
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.3516976833343506,
+      "learning_rate": 1.8877071047097095e-06,
+      "loss": 0.5125,
+      "step": 44373
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.3658740520477295,
+      "learning_rate": 1.8875007562404673e-06,
+      "loss": 0.5032,
+      "step": 44374
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.7708616256713867,
+      "learning_rate": 1.8872944164259216e-06,
+      "loss": 0.7624,
+      "step": 44375
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 5.505009174346924,
+      "learning_rate": 1.8870880852666417e-06,
+      "loss": 0.5937,
+      "step": 44376
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.2775659561157227,
+      "learning_rate": 1.8868817627632057e-06,
+      "loss": 0.9515,
+      "step": 44377
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.773305892944336,
+      "learning_rate": 1.8866754489161836e-06,
+      "loss": 0.6267,
+      "step": 44378
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.5668883323669434,
+      "learning_rate": 1.886469143726153e-06,
+      "loss": 0.7528,
+      "step": 44379
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.1602210998535156,
+      "learning_rate": 1.8862628471936833e-06,
+      "loss": 0.6512,
+      "step": 44380
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.9465434551239014,
+      "learning_rate": 1.8860565593193504e-06,
+      "loss": 0.5436,
+      "step": 44381
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 1.8032824993133545,
+      "learning_rate": 1.8858502801037288e-06,
+      "loss": 0.326,
+      "step": 44382
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.7719876766204834,
+      "learning_rate": 1.8856440095473898e-06,
+      "loss": 0.5747,
+      "step": 44383
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.6876635551452637,
+      "learning_rate": 1.8854377476509095e-06,
+      "loss": 0.5181,
+      "step": 44384
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.207148551940918,
+      "learning_rate": 1.8852314944148586e-06,
+      "loss": 0.4034,
+      "step": 44385
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.0164451599121094,
+      "learning_rate": 1.885025249839814e-06,
+      "loss": 0.4902,
+      "step": 44386
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 4.046955585479736,
+      "learning_rate": 1.884819013926345e-06,
+      "loss": 0.4856,
+      "step": 44387
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.7674007415771484,
+      "learning_rate": 1.8846127866750297e-06,
+      "loss": 0.7889,
+      "step": 44388
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.33524751663208,
+      "learning_rate": 1.884406568086437e-06,
+      "loss": 0.7072,
+      "step": 44389
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.5216758251190186,
+      "learning_rate": 1.8842003581611445e-06,
+      "loss": 0.6517,
+      "step": 44390
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.483297348022461,
+      "learning_rate": 1.8839941568997223e-06,
+      "loss": 0.5302,
+      "step": 44391
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.3336358070373535,
+      "learning_rate": 1.883787964302744e-06,
+      "loss": 0.7903,
+      "step": 44392
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.5691072940826416,
+      "learning_rate": 1.8835817803707845e-06,
+      "loss": 0.5395,
+      "step": 44393
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.9660251140594482,
+      "learning_rate": 1.8833756051044188e-06,
+      "loss": 0.5924,
+      "step": 44394
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.0328445434570312,
+      "learning_rate": 1.883169438504216e-06,
+      "loss": 0.4669,
+      "step": 44395
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.288501501083374,
+      "learning_rate": 1.8829632805707526e-06,
+      "loss": 0.7868,
+      "step": 44396
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.041530132293701,
+      "learning_rate": 1.8827571313045994e-06,
+      "loss": 0.6389,
+      "step": 44397
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 1.8848326206207275,
+      "learning_rate": 1.8825509907063328e-06,
+      "loss": 0.5716,
+      "step": 44398
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.618027925491333,
+      "learning_rate": 1.882344858776522e-06,
+      "loss": 0.741,
+      "step": 44399
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.918735980987549,
+      "learning_rate": 1.8821387355157443e-06,
+      "loss": 0.712,
+      "step": 44400
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.4978485107421875,
+      "learning_rate": 1.8819326209245685e-06,
+      "loss": 0.5871,
+      "step": 44401
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.145627498626709,
+      "learning_rate": 1.8817265150035724e-06,
+      "loss": 0.6138,
+      "step": 44402
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.1921873092651367,
+      "learning_rate": 1.8815204177533248e-06,
+      "loss": 0.511,
+      "step": 44403
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.0901567935943604,
+      "learning_rate": 1.8813143291744025e-06,
+      "loss": 0.5216,
+      "step": 44404
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 4.307076930999756,
+      "learning_rate": 1.8811082492673749e-06,
+      "loss": 0.6359,
+      "step": 44405
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.587275266647339,
+      "learning_rate": 1.8809021780328184e-06,
+      "loss": 0.8103,
+      "step": 44406
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.619713068008423,
+      "learning_rate": 1.8806961154713034e-06,
+      "loss": 0.6189,
+      "step": 44407
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 1.8287876844406128,
+      "learning_rate": 1.8804900615834048e-06,
+      "loss": 0.4561,
+      "step": 44408
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.4530293941497803,
+      "learning_rate": 1.880284016369694e-06,
+      "loss": 0.6553,
+      "step": 44409
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 1.6981728076934814,
+      "learning_rate": 1.8800779798307456e-06,
+      "loss": 0.4161,
+      "step": 44410
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.174588203430176,
+      "learning_rate": 1.8798719519671304e-06,
+      "loss": 0.5686,
+      "step": 44411
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.4069368839263916,
+      "learning_rate": 1.879665932779422e-06,
+      "loss": 0.8978,
+      "step": 44412
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 16.065074920654297,
+      "learning_rate": 1.8794599222681942e-06,
+      "loss": 0.4106,
+      "step": 44413
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.5491743087768555,
+      "learning_rate": 1.879253920434021e-06,
+      "loss": 0.6986,
+      "step": 44414
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.8173766136169434,
+      "learning_rate": 1.8790479272774715e-06,
+      "loss": 0.5758,
+      "step": 44415
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.1861464977264404,
+      "learning_rate": 1.8788419427991229e-06,
+      "loss": 0.4498,
+      "step": 44416
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.074998617172241,
+      "learning_rate": 1.8786359669995435e-06,
+      "loss": 0.575,
+      "step": 44417
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.1035261154174805,
+      "learning_rate": 1.87842999987931e-06,
+      "loss": 0.4954,
+      "step": 44418
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.367098331451416,
+      "learning_rate": 1.8782240414389913e-06,
+      "loss": 0.5235,
+      "step": 44419
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.5522494316101074,
+      "learning_rate": 1.878018091679164e-06,
+      "loss": 0.38,
+      "step": 44420
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.104081869125366,
+      "learning_rate": 1.8778121506003978e-06,
+      "loss": 0.6537,
+      "step": 44421
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 1.575339913368225,
+      "learning_rate": 1.8776062182032673e-06,
+      "loss": 0.2555,
+      "step": 44422
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.2811477184295654,
+      "learning_rate": 1.8774002944883429e-06,
+      "loss": 0.654,
+      "step": 44423
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.517430067062378,
+      "learning_rate": 1.8771943794561997e-06,
+      "loss": 0.5773,
+      "step": 44424
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.435544729232788,
+      "learning_rate": 1.8769884731074078e-06,
+      "loss": 0.5108,
+      "step": 44425
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.054006338119507,
+      "learning_rate": 1.876782575442543e-06,
+      "loss": 0.5768,
+      "step": 44426
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.3505654335021973,
+      "learning_rate": 1.8765766864621743e-06,
+      "loss": 0.5022,
+      "step": 44427
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.4738824367523193,
+      "learning_rate": 1.8763708061668766e-06,
+      "loss": 0.583,
+      "step": 44428
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.5559277534484863,
+      "learning_rate": 1.8761649345572208e-06,
+      "loss": 0.6108,
+      "step": 44429
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.7846155166625977,
+      "learning_rate": 1.8759590716337812e-06,
+      "loss": 0.6018,
+      "step": 44430
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.442448854446411,
+      "learning_rate": 1.8757532173971277e-06,
+      "loss": 0.5839,
+      "step": 44431
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 4.080482006072998,
+      "learning_rate": 1.875547371847834e-06,
+      "loss": 0.6638,
+      "step": 44432
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.5142173767089844,
+      "learning_rate": 1.8753415349864723e-06,
+      "loss": 0.516,
+      "step": 44433
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.424570083618164,
+      "learning_rate": 1.8751357068136177e-06,
+      "loss": 0.8005,
+      "step": 44434
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.4940760135650635,
+      "learning_rate": 1.8749298873298377e-06,
+      "loss": 0.6082,
+      "step": 44435
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.7019007205963135,
+      "learning_rate": 1.8747240765357095e-06,
+      "loss": 0.5824,
+      "step": 44436
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 1.9288232326507568,
+      "learning_rate": 1.8745182744318003e-06,
+      "loss": 0.5535,
+      "step": 44437
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 6.915557384490967,
+      "learning_rate": 1.8743124810186875e-06,
+      "loss": 0.4843,
+      "step": 44438
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.1949779987335205,
+      "learning_rate": 1.8741066962969383e-06,
+      "loss": 0.7771,
+      "step": 44439
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.307931900024414,
+      "learning_rate": 1.8739009202671293e-06,
+      "loss": 0.6492,
+      "step": 44440
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.27496075630188,
+      "learning_rate": 1.873695152929829e-06,
+      "loss": 0.7342,
+      "step": 44441
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.222526788711548,
+      "learning_rate": 1.8734893942856136e-06,
+      "loss": 0.5574,
+      "step": 44442
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.6671998500823975,
+      "learning_rate": 1.873283644335051e-06,
+      "loss": 0.6535,
+      "step": 44443
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.588895320892334,
+      "learning_rate": 1.8730779030787149e-06,
+      "loss": 0.4973,
+      "step": 44444
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.4486217498779297,
+      "learning_rate": 1.8728721705171798e-06,
+      "loss": 0.5581,
+      "step": 44445
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.331935167312622,
+      "learning_rate": 1.8726664466510142e-06,
+      "loss": 0.4335,
+      "step": 44446
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.3679754734039307,
+      "learning_rate": 1.8724607314807935e-06,
+      "loss": 0.6382,
+      "step": 44447
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.0328166484832764,
+      "learning_rate": 1.872255025007086e-06,
+      "loss": 0.5374,
+      "step": 44448
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 1.8731673955917358,
+      "learning_rate": 1.8720493272304668e-06,
+      "loss": 0.3611,
+      "step": 44449
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.0324416160583496,
+      "learning_rate": 1.8718436381515053e-06,
+      "loss": 0.5697,
+      "step": 44450
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.859611749649048,
+      "learning_rate": 1.8716379577707766e-06,
+      "loss": 0.794,
+      "step": 44451
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.120805501937866,
+      "learning_rate": 1.8714322860888495e-06,
+      "loss": 0.5255,
+      "step": 44452
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.352741241455078,
+      "learning_rate": 1.8712266231062964e-06,
+      "loss": 0.5583,
+      "step": 44453
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.459658622741699,
+      "learning_rate": 1.8710209688236908e-06,
+      "loss": 0.547,
+      "step": 44454
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.0882840156555176,
+      "learning_rate": 1.8708153232416049e-06,
+      "loss": 0.7388,
+      "step": 44455
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.4122331142425537,
+      "learning_rate": 1.8706096863606077e-06,
+      "loss": 0.6225,
+      "step": 44456
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.3354268074035645,
+      "learning_rate": 1.8704040581812748e-06,
+      "loss": 0.6921,
+      "step": 44457
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.247692108154297,
+      "learning_rate": 1.8701984387041738e-06,
+      "loss": 0.7123,
+      "step": 44458
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.3177621364593506,
+      "learning_rate": 1.8699928279298802e-06,
+      "loss": 0.5999,
+      "step": 44459
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.5230259895324707,
+      "learning_rate": 1.8697872258589621e-06,
+      "loss": 0.5772,
+      "step": 44460
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.5533089637756348,
+      "learning_rate": 1.8695816324919952e-06,
+      "loss": 0.8759,
+      "step": 44461
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.839672088623047,
+      "learning_rate": 1.8693760478295475e-06,
+      "loss": 0.496,
+      "step": 44462
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.0775277614593506,
+      "learning_rate": 1.8691704718721936e-06,
+      "loss": 0.631,
+      "step": 44463
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.5667967796325684,
+      "learning_rate": 1.8689649046205022e-06,
+      "loss": 0.523,
+      "step": 44464
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.051253318786621,
+      "learning_rate": 1.8687593460750485e-06,
+      "loss": 0.6536,
+      "step": 44465
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.565215587615967,
+      "learning_rate": 1.8685537962363998e-06,
+      "loss": 0.722,
+      "step": 44466
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.763529062271118,
+      "learning_rate": 1.8683482551051318e-06,
+      "loss": 0.7496,
+      "step": 44467
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 4.041291236877441,
+      "learning_rate": 1.8681427226818127e-06,
+      "loss": 0.5317,
+      "step": 44468
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.3242440223693848,
+      "learning_rate": 1.867937198967017e-06,
+      "loss": 0.6522,
+      "step": 44469
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.688416004180908,
+      "learning_rate": 1.867731683961313e-06,
+      "loss": 0.6634,
+      "step": 44470
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 4.496126651763916,
+      "learning_rate": 1.8675261776652759e-06,
+      "loss": 0.5952,
+      "step": 44471
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.0553693771362305,
+      "learning_rate": 1.8673206800794724e-06,
+      "loss": 0.572,
+      "step": 44472
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.337601900100708,
+      "learning_rate": 1.8671151912044772e-06,
+      "loss": 0.4894,
+      "step": 44473
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.395385980606079,
+      "learning_rate": 1.8669097110408613e-06,
+      "loss": 0.5513,
+      "step": 44474
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.587878942489624,
+      "learning_rate": 1.866704239589197e-06,
+      "loss": 0.5906,
+      "step": 44475
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.5810580253601074,
+      "learning_rate": 1.866498776850053e-06,
+      "loss": 0.5586,
+      "step": 44476
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.2732365131378174,
+      "learning_rate": 1.8662933228240038e-06,
+      "loss": 0.577,
+      "step": 44477
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.5558736324310303,
+      "learning_rate": 1.866087877511617e-06,
+      "loss": 0.5946,
+      "step": 44478
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.0805253982543945,
+      "learning_rate": 1.8658824409134675e-06,
+      "loss": 0.7782,
+      "step": 44479
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.326148271560669,
+      "learning_rate": 1.8656770130301232e-06,
+      "loss": 0.8168,
+      "step": 44480
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 1.9815973043441772,
+      "learning_rate": 1.8654715938621588e-06,
+      "loss": 0.5262,
+      "step": 44481
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.515082836151123,
+      "learning_rate": 1.865266183410142e-06,
+      "loss": 0.6478,
+      "step": 44482
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.0289993286132812,
+      "learning_rate": 1.8650607816746474e-06,
+      "loss": 0.4875,
+      "step": 44483
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.3102433681488037,
+      "learning_rate": 1.8648553886562427e-06,
+      "loss": 0.5297,
+      "step": 44484
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.280719757080078,
+      "learning_rate": 1.864650004355502e-06,
+      "loss": 0.7812,
+      "step": 44485
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.868061065673828,
+      "learning_rate": 1.864444628772994e-06,
+      "loss": 0.2728,
+      "step": 44486
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.375303268432617,
+      "learning_rate": 1.8642392619092925e-06,
+      "loss": 0.7164,
+      "step": 44487
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.344029664993286,
+      "learning_rate": 1.8640339037649647e-06,
+      "loss": 0.497,
+      "step": 44488
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.360654830932617,
+      "learning_rate": 1.8638285543405859e-06,
+      "loss": 0.4955,
+      "step": 44489
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.0714921951293945,
+      "learning_rate": 1.8636232136367238e-06,
+      "loss": 0.6882,
+      "step": 44490
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.319819927215576,
+      "learning_rate": 1.8634178816539516e-06,
+      "loss": 0.5009,
+      "step": 44491
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.68827223777771,
+      "learning_rate": 1.8632125583928378e-06,
+      "loss": 0.6736,
+      "step": 44492
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.09905743598938,
+      "learning_rate": 1.8630072438539548e-06,
+      "loss": 0.5593,
+      "step": 44493
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 1.9486078023910522,
+      "learning_rate": 1.8628019380378736e-06,
+      "loss": 0.7228,
+      "step": 44494
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 1.9915846586227417,
+      "learning_rate": 1.8625966409451669e-06,
+      "loss": 0.4503,
+      "step": 44495
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 1.7726439237594604,
+      "learning_rate": 1.8623913525764014e-06,
+      "loss": 0.4267,
+      "step": 44496
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.0490972995758057,
+      "learning_rate": 1.8621860729321522e-06,
+      "loss": 0.4757,
+      "step": 44497
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.3721089363098145,
+      "learning_rate": 1.8619808020129864e-06,
+      "loss": 0.6795,
+      "step": 44498
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.8134219646453857,
+      "learning_rate": 1.8617755398194781e-06,
+      "loss": 0.6346,
+      "step": 44499
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 4.230406284332275,
+      "learning_rate": 1.8615702863521951e-06,
+      "loss": 0.6059,
+      "step": 44500
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.2065508365631104,
+      "learning_rate": 1.8613650416117107e-06,
+      "loss": 0.5981,
+      "step": 44501
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.330068588256836,
+      "learning_rate": 1.861159805598593e-06,
+      "loss": 0.4805,
+      "step": 44502
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.79152512550354,
+      "learning_rate": 1.8609545783134154e-06,
+      "loss": 0.7044,
+      "step": 44503
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.4676320552825928,
+      "learning_rate": 1.8607493597567456e-06,
+      "loss": 0.6386,
+      "step": 44504
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 5.403720855712891,
+      "learning_rate": 1.8605441499291555e-06,
+      "loss": 0.8173,
+      "step": 44505
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.3044369220733643,
+      "learning_rate": 1.8603389488312185e-06,
+      "loss": 0.5185,
+      "step": 44506
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 1.908448576927185,
+      "learning_rate": 1.8601337564635003e-06,
+      "loss": 0.6117,
+      "step": 44507
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.4971203804016113,
+      "learning_rate": 1.859928572826576e-06,
+      "loss": 0.6853,
+      "step": 44508
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.072861433029175,
+      "learning_rate": 1.8597233979210122e-06,
+      "loss": 0.5346,
+      "step": 44509
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 4.353245258331299,
+      "learning_rate": 1.859518231747383e-06,
+      "loss": 0.7327,
+      "step": 44510
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.40938663482666,
+      "learning_rate": 1.8593130743062555e-06,
+      "loss": 0.5964,
+      "step": 44511
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.5344173908233643,
+      "learning_rate": 1.8591079255982036e-06,
+      "loss": 0.7202,
+      "step": 44512
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.83514404296875,
+      "learning_rate": 1.858902785623794e-06,
+      "loss": 1.0115,
+      "step": 44513
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.0439531803131104,
+      "learning_rate": 1.8586976543835995e-06,
+      "loss": 0.715,
+      "step": 44514
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.380735158920288,
+      "learning_rate": 1.8584925318781915e-06,
+      "loss": 0.3437,
+      "step": 44515
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.1848108768463135,
+      "learning_rate": 1.8582874181081373e-06,
+      "loss": 0.5032,
+      "step": 44516
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 8.0432767868042,
+      "learning_rate": 1.8580823130740084e-06,
+      "loss": 0.7835,
+      "step": 44517
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.061636209487915,
+      "learning_rate": 1.857877216776378e-06,
+      "loss": 0.6411,
+      "step": 44518
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 1.6354758739471436,
+      "learning_rate": 1.8576721292158117e-06,
+      "loss": 0.4992,
+      "step": 44519
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 1.7974679470062256,
+      "learning_rate": 1.8574670503928843e-06,
+      "loss": 0.4404,
+      "step": 44520
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.3200042247772217,
+      "learning_rate": 1.8572619803081615e-06,
+      "loss": 0.3834,
+      "step": 44521
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.78861403465271,
+      "learning_rate": 1.8570569189622184e-06,
+      "loss": 0.5849,
+      "step": 44522
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 1.9752918481826782,
+      "learning_rate": 1.85685186635562e-06,
+      "loss": 0.7242,
+      "step": 44523
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 1.5259041786193848,
+      "learning_rate": 1.8566468224889417e-06,
+      "loss": 0.4443,
+      "step": 44524
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 4.849477767944336,
+      "learning_rate": 1.8564417873627488e-06,
+      "loss": 0.7173,
+      "step": 44525
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.1409499645233154,
+      "learning_rate": 1.8562367609776154e-06,
+      "loss": 0.4725,
+      "step": 44526
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.09975266456604,
+      "learning_rate": 1.8560317433341085e-06,
+      "loss": 0.5359,
+      "step": 44527
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.265399694442749,
+      "learning_rate": 1.8558267344328007e-06,
+      "loss": 0.5039,
+      "step": 44528
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.3605399131774902,
+      "learning_rate": 1.8556217342742595e-06,
+      "loss": 0.6871,
+      "step": 44529
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 1.996935486793518,
+      "learning_rate": 1.855416742859058e-06,
+      "loss": 0.6293,
+      "step": 44530
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.718047857284546,
+      "learning_rate": 1.8552117601877634e-06,
+      "loss": 0.5846,
+      "step": 44531
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 4.044548034667969,
+      "learning_rate": 1.8550067862609478e-06,
+      "loss": 0.5644,
+      "step": 44532
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.78702974319458,
+      "learning_rate": 1.854801821079179e-06,
+      "loss": 0.7903,
+      "step": 44533
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 3.172999382019043,
+      "learning_rate": 1.8545968646430279e-06,
+      "loss": 0.712,
+      "step": 44534
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 1.417446494102478,
+      "learning_rate": 1.8543919169530644e-06,
+      "loss": 0.4999,
+      "step": 44535
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.4955310821533203,
+      "learning_rate": 1.8541869780098603e-06,
+      "loss": 0.7665,
+      "step": 44536
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.2071239948272705,
+      "learning_rate": 1.8539820478139825e-06,
+      "loss": 0.4756,
+      "step": 44537
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 1.438037633895874,
+      "learning_rate": 1.8537771263660038e-06,
+      "loss": 0.3615,
+      "step": 44538
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.383078098297119,
+      "learning_rate": 1.8535722136664902e-06,
+      "loss": 0.6312,
+      "step": 44539
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 3.6523489952087402,
+      "learning_rate": 1.8533673097160154e-06,
+      "loss": 0.4555,
+      "step": 44540
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 3.0404343605041504,
+      "learning_rate": 1.853162414515146e-06,
+      "loss": 0.7955,
+      "step": 44541
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.4727795124053955,
+      "learning_rate": 1.8529575280644545e-06,
+      "loss": 0.6387,
+      "step": 44542
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 3.0983011722564697,
+      "learning_rate": 1.8527526503645078e-06,
+      "loss": 0.394,
+      "step": 44543
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.334202766418457,
+      "learning_rate": 1.8525477814158782e-06,
+      "loss": 0.669,
+      "step": 44544
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.4928855895996094,
+      "learning_rate": 1.8523429212191325e-06,
+      "loss": 0.7021,
+      "step": 44545
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.746788501739502,
+      "learning_rate": 1.8521380697748443e-06,
+      "loss": 0.5745,
+      "step": 44546
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 3.368377685546875,
+      "learning_rate": 1.8519332270835788e-06,
+      "loss": 0.6748,
+      "step": 44547
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 3.4523470401763916,
+      "learning_rate": 1.851728393145909e-06,
+      "loss": 0.7519,
+      "step": 44548
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 3.8307442665100098,
+      "learning_rate": 1.8515235679624022e-06,
+      "loss": 0.7733,
+      "step": 44549
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.257375955581665,
+      "learning_rate": 1.85131875153363e-06,
+      "loss": 0.6936,
+      "step": 44550
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.6873791217803955,
+      "learning_rate": 1.8511139438601594e-06,
+      "loss": 0.7328,
+      "step": 44551
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.8911752700805664,
+      "learning_rate": 1.850909144942563e-06,
+      "loss": 0.6596,
+      "step": 44552
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 4.094413757324219,
+      "learning_rate": 1.8507043547814068e-06,
+      "loss": 0.7243,
+      "step": 44553
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 3.273123264312744,
+      "learning_rate": 1.8504995733772625e-06,
+      "loss": 0.8143,
+      "step": 44554
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.530428647994995,
+      "learning_rate": 1.8502948007306988e-06,
+      "loss": 0.4098,
+      "step": 44555
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.260857105255127,
+      "learning_rate": 1.850090036842287e-06,
+      "loss": 0.5697,
+      "step": 44556
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.8717916011810303,
+      "learning_rate": 1.8498852817125929e-06,
+      "loss": 0.7681,
+      "step": 44557
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.425327777862549,
+      "learning_rate": 1.8496805353421897e-06,
+      "loss": 0.5522,
+      "step": 44558
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.4761953353881836,
+      "learning_rate": 1.849475797731643e-06,
+      "loss": 0.7452,
+      "step": 44559
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 3.329319953918457,
+      "learning_rate": 1.8492710688815257e-06,
+      "loss": 0.586,
+      "step": 44560
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 3.3341732025146484,
+      "learning_rate": 1.8490663487924037e-06,
+      "loss": 0.7136,
+      "step": 44561
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.546320676803589,
+      "learning_rate": 1.8488616374648494e-06,
+      "loss": 0.4498,
+      "step": 44562
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.915856122970581,
+      "learning_rate": 1.8486569348994287e-06,
+      "loss": 0.7601,
+      "step": 44563
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.1448025703430176,
+      "learning_rate": 1.8484522410967143e-06,
+      "loss": 0.6502,
+      "step": 44564
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.965508460998535,
+      "learning_rate": 1.8482475560572716e-06,
+      "loss": 0.7526,
+      "step": 44565
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 3.1892762184143066,
+      "learning_rate": 1.8480428797816718e-06,
+      "loss": 0.4175,
+      "step": 44566
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 1.7243804931640625,
+      "learning_rate": 1.8478382122704862e-06,
+      "loss": 0.4846,
+      "step": 44567
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.9344940185546875,
+      "learning_rate": 1.8476335535242795e-06,
+      "loss": 0.6119,
+      "step": 44568
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.6816492080688477,
+      "learning_rate": 1.847428903543625e-06,
+      "loss": 0.4928,
+      "step": 44569
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.7450411319732666,
+      "learning_rate": 1.8472242623290882e-06,
+      "loss": 0.7786,
+      "step": 44570
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.1710753440856934,
+      "learning_rate": 1.847019629881241e-06,
+      "loss": 0.2806,
+      "step": 44571
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.3248560428619385,
+      "learning_rate": 1.8468150062006497e-06,
+      "loss": 0.4748,
+      "step": 44572
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.5016353130340576,
+      "learning_rate": 1.8466103912878857e-06,
+      "loss": 0.7019,
+      "step": 44573
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 1.6152658462524414,
+      "learning_rate": 1.8464057851435157e-06,
+      "loss": 0.5835,
+      "step": 44574
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.2254221439361572,
+      "learning_rate": 1.8462011877681102e-06,
+      "loss": 0.7964,
+      "step": 44575
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 3.070107936859131,
+      "learning_rate": 1.8459965991622392e-06,
+      "loss": 0.71,
+      "step": 44576
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.436424732208252,
+      "learning_rate": 1.8457920193264684e-06,
+      "loss": 0.5314,
+      "step": 44577
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.4183030128479004,
+      "learning_rate": 1.8455874482613678e-06,
+      "loss": 0.81,
+      "step": 44578
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 1.7873913049697876,
+      "learning_rate": 1.8453828859675094e-06,
+      "loss": 0.6025,
+      "step": 44579
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 6.090214729309082,
+      "learning_rate": 1.8451783324454575e-06,
+      "loss": 0.6877,
+      "step": 44580
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 1.8380613327026367,
+      "learning_rate": 1.844973787695784e-06,
+      "loss": 0.5347,
+      "step": 44581
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.549718141555786,
+      "learning_rate": 1.8447692517190552e-06,
+      "loss": 0.5562,
+      "step": 44582
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.3754985332489014,
+      "learning_rate": 1.8445647245158433e-06,
+      "loss": 0.6344,
+      "step": 44583
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.4438531398773193,
+      "learning_rate": 1.8443602060867128e-06,
+      "loss": 0.7444,
+      "step": 44584
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.6268038749694824,
+      "learning_rate": 1.8441556964322361e-06,
+      "loss": 0.8515,
+      "step": 44585
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.9136290550231934,
+      "learning_rate": 1.8439511955529782e-06,
+      "loss": 0.5014,
+      "step": 44586
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.4742395877838135,
+      "learning_rate": 1.8437467034495116e-06,
+      "loss": 0.4595,
+      "step": 44587
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 3.6975831985473633,
+      "learning_rate": 1.8435422201224012e-06,
+      "loss": 0.5344,
+      "step": 44588
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.1442952156066895,
+      "learning_rate": 1.8433377455722195e-06,
+      "loss": 0.5904,
+      "step": 44589
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.753901958465576,
+      "learning_rate": 1.8431332797995317e-06,
+      "loss": 0.615,
+      "step": 44590
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.232656717300415,
+      "learning_rate": 1.8429288228049086e-06,
+      "loss": 0.6179,
+      "step": 44591
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.4849634170532227,
+      "learning_rate": 1.8427243745889162e-06,
+      "loss": 0.5225,
+      "step": 44592
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.096346139907837,
+      "learning_rate": 1.8425199351521266e-06,
+      "loss": 0.4317,
+      "step": 44593
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.7359459400177,
+      "learning_rate": 1.8423155044951041e-06,
+      "loss": 0.4878,
+      "step": 44594
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.003835439682007,
+      "learning_rate": 1.8421110826184197e-06,
+      "loss": 0.6053,
+      "step": 44595
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.3106160163879395,
+      "learning_rate": 1.8419066695226412e-06,
+      "loss": 0.5105,
+      "step": 44596
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.124436378479004,
+      "learning_rate": 1.8417022652083393e-06,
+      "loss": 0.5822,
+      "step": 44597
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.8861374855041504,
+      "learning_rate": 1.8414978696760783e-06,
+      "loss": 0.6124,
+      "step": 44598
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.904527425765991,
+      "learning_rate": 1.8412934829264301e-06,
+      "loss": 0.6771,
+      "step": 44599
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.1105093955993652,
+      "learning_rate": 1.84108910495996e-06,
+      "loss": 0.5748,
+      "step": 44600
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 3.8133254051208496,
+      "learning_rate": 1.8408847357772392e-06,
+      "loss": 0.7052,
+      "step": 44601
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 3.4716405868530273,
+      "learning_rate": 1.8406803753788328e-06,
+      "loss": 0.6335,
+      "step": 44602
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.9990460872650146,
+      "learning_rate": 1.8404760237653124e-06,
+      "loss": 0.5985,
+      "step": 44603
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.900789976119995,
+      "learning_rate": 1.8402716809372428e-06,
+      "loss": 0.6858,
+      "step": 44604
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 3.913534641265869,
+      "learning_rate": 1.8400673468951962e-06,
+      "loss": 0.7438,
+      "step": 44605
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.218158006668091,
+      "learning_rate": 1.8398630216397368e-06,
+      "loss": 0.7241,
+      "step": 44606
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.6885030269622803,
+      "learning_rate": 1.839658705171436e-06,
+      "loss": 0.5607,
+      "step": 44607
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 3.745588541030884,
+      "learning_rate": 1.8394543974908585e-06,
+      "loss": 0.7372,
+      "step": 44608
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.7533175945281982,
+      "learning_rate": 1.839250098598576e-06,
+      "loss": 0.5057,
+      "step": 44609
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 1.8108711242675781,
+      "learning_rate": 1.8390458084951541e-06,
+      "loss": 0.5718,
+      "step": 44610
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.6691787242889404,
+      "learning_rate": 1.8388415271811626e-06,
+      "loss": 0.6004,
+      "step": 44611
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 3.5549542903900146,
+      "learning_rate": 1.8386372546571674e-06,
+      "loss": 0.6363,
+      "step": 44612
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 1.5761325359344482,
+      "learning_rate": 1.8384329909237393e-06,
+      "loss": 0.4039,
+      "step": 44613
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.6205554008483887,
+      "learning_rate": 1.8382287359814427e-06,
+      "loss": 0.6697,
+      "step": 44614
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.672633647918701,
+      "learning_rate": 1.8380244898308474e-06,
+      "loss": 0.7224,
+      "step": 44615
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 3.5289340019226074,
+      "learning_rate": 1.837820252472522e-06,
+      "loss": 0.6652,
+      "step": 44616
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.8388803005218506,
+      "learning_rate": 1.8376160239070357e-06,
+      "loss": 0.643,
+      "step": 44617
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 3.844679355621338,
+      "learning_rate": 1.8374118041349525e-06,
+      "loss": 0.4598,
+      "step": 44618
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.4046804904937744,
+      "learning_rate": 1.837207593156844e-06,
+      "loss": 0.6138,
+      "step": 44619
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.883284330368042,
+      "learning_rate": 1.837003390973275e-06,
+      "loss": 0.754,
+      "step": 44620
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.3316452503204346,
+      "learning_rate": 1.836799197584816e-06,
+      "loss": 0.4858,
+      "step": 44621
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.0615720748901367,
+      "learning_rate": 1.8365950129920323e-06,
+      "loss": 0.4337,
+      "step": 44622
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 4.277628421783447,
+      "learning_rate": 1.8363908371954942e-06,
+      "loss": 0.7443,
+      "step": 44623
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.396324396133423,
+      "learning_rate": 1.8361866701957659e-06,
+      "loss": 0.5978,
+      "step": 44624
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 1.9837454557418823,
+      "learning_rate": 1.83598251199342e-06,
+      "loss": 0.5166,
+      "step": 44625
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.207603693008423,
+      "learning_rate": 1.835778362589019e-06,
+      "loss": 0.4116,
+      "step": 44626
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.463528871536255,
+      "learning_rate": 1.835574221983135e-06,
+      "loss": 0.7324,
+      "step": 44627
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.1173758506774902,
+      "learning_rate": 1.8353700901763317e-06,
+      "loss": 0.6654,
+      "step": 44628
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 3.3110857009887695,
+      "learning_rate": 1.835165967169179e-06,
+      "loss": 0.8279,
+      "step": 44629
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.1619784832000732,
+      "learning_rate": 1.8349618529622458e-06,
+      "loss": 0.5244,
+      "step": 44630
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.32344126701355,
+      "learning_rate": 1.8347577475560963e-06,
+      "loss": 0.648,
+      "step": 44631
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.097379684448242,
+      "learning_rate": 1.8345536509513012e-06,
+      "loss": 0.4838,
+      "step": 44632
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 4.532800674438477,
+      "learning_rate": 1.8343495631484254e-06,
+      "loss": 0.5933,
+      "step": 44633
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.7931387424468994,
+      "learning_rate": 1.8341454841480393e-06,
+      "loss": 0.4732,
+      "step": 44634
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.223282814025879,
+      "learning_rate": 1.8339414139507067e-06,
+      "loss": 0.6448,
+      "step": 44635
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.7440171241760254,
+      "learning_rate": 1.8337373525569969e-06,
+      "loss": 0.5749,
+      "step": 44636
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.2131314277648926,
+      "learning_rate": 1.8335332999674792e-06,
+      "loss": 0.6911,
+      "step": 44637
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.6719374656677246,
+      "learning_rate": 1.8333292561827176e-06,
+      "loss": 0.4043,
+      "step": 44638
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 3.2423453330993652,
+      "learning_rate": 1.8331252212032807e-06,
+      "loss": 0.778,
+      "step": 44639
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.6327009201049805,
+      "learning_rate": 1.8329211950297382e-06,
+      "loss": 0.7245,
+      "step": 44640
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.6154024600982666,
+      "learning_rate": 1.832717177662654e-06,
+      "loss": 0.6569,
+      "step": 44641
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 3.3016469478607178,
+      "learning_rate": 1.8325131691025982e-06,
+      "loss": 0.4724,
+      "step": 44642
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.5829734802246094,
+      "learning_rate": 1.8323091693501354e-06,
+      "loss": 0.4649,
+      "step": 44643
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.9797396659851074,
+      "learning_rate": 1.8321051784058352e-06,
+      "loss": 0.425,
+      "step": 44644
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.585991621017456,
+      "learning_rate": 1.8319011962702627e-06,
+      "loss": 0.7643,
+      "step": 44645
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.369743585586548,
+      "learning_rate": 1.8316972229439878e-06,
+      "loss": 0.3828,
+      "step": 44646
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.494669198989868,
+      "learning_rate": 1.831493258427574e-06,
+      "loss": 0.5841,
+      "step": 44647
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 1.8373147249221802,
+      "learning_rate": 1.8312893027215921e-06,
+      "loss": 0.6296,
+      "step": 44648
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.8063559532165527,
+      "learning_rate": 1.8310853558266068e-06,
+      "loss": 0.5006,
+      "step": 44649
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 3.7835283279418945,
+      "learning_rate": 1.8308814177431867e-06,
+      "loss": 0.6162,
+      "step": 44650
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.0398197174072266,
+      "learning_rate": 1.830677488471897e-06,
+      "loss": 0.3463,
+      "step": 44651
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 5.831177711486816,
+      "learning_rate": 1.8304735680133072e-06,
+      "loss": 0.6621,
+      "step": 44652
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.6981728076934814,
+      "learning_rate": 1.8302696563679824e-06,
+      "loss": 0.465,
+      "step": 44653
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.1700799465179443,
+      "learning_rate": 1.8300657535364912e-06,
+      "loss": 0.6479,
+      "step": 44654
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.2231640815734863,
+      "learning_rate": 1.829861859519398e-06,
+      "loss": 0.6219,
+      "step": 44655
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.6671783924102783,
+      "learning_rate": 1.8296579743172715e-06,
+      "loss": 0.654,
+      "step": 44656
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.372087240219116,
+      "learning_rate": 1.8294540979306786e-06,
+      "loss": 0.498,
+      "step": 44657
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 3.061483860015869,
+      "learning_rate": 1.8292502303601877e-06,
+      "loss": 0.8596,
+      "step": 44658
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 3.970811367034912,
+      "learning_rate": 1.8290463716063622e-06,
+      "loss": 0.6359,
+      "step": 44659
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.5221643447875977,
+      "learning_rate": 1.828842521669773e-06,
+      "loss": 0.577,
+      "step": 44660
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 4.3941426277160645,
+      "learning_rate": 1.8286386805509832e-06,
+      "loss": 0.6649,
+      "step": 44661
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 1.7376329898834229,
+      "learning_rate": 1.8284348482505626e-06,
+      "loss": 0.3431,
+      "step": 44662
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 3.6102547645568848,
+      "learning_rate": 1.8282310247690748e-06,
+      "loss": 0.7901,
+      "step": 44663
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.2263057231903076,
+      "learning_rate": 1.82802721010709e-06,
+      "loss": 0.6627,
+      "step": 44664
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.0749363899230957,
+      "learning_rate": 1.8278234042651722e-06,
+      "loss": 0.6388,
+      "step": 44665
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 3.024144172668457,
+      "learning_rate": 1.8276196072438902e-06,
+      "loss": 0.6183,
+      "step": 44666
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.672266721725464,
+      "learning_rate": 1.8274158190438085e-06,
+      "loss": 0.4188,
+      "step": 44667
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.2034335136413574,
+      "learning_rate": 1.8272120396654964e-06,
+      "loss": 0.312,
+      "step": 44668
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 1.7895358800888062,
+      "learning_rate": 1.827008269109517e-06,
+      "loss": 0.3693,
+      "step": 44669
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 1.9186599254608154,
+      "learning_rate": 1.8268045073764413e-06,
+      "loss": 0.5066,
+      "step": 44670
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 3.010962724685669,
+      "learning_rate": 1.8266007544668312e-06,
+      "loss": 0.7853,
+      "step": 44671
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.20888614654541,
+      "learning_rate": 1.8263970103812577e-06,
+      "loss": 0.5093,
+      "step": 44672
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 1.0862096548080444,
+      "learning_rate": 1.8261932751202838e-06,
+      "loss": 0.2117,
+      "step": 44673
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 4.797669410705566,
+      "learning_rate": 1.8259895486844792e-06,
+      "loss": 0.6433,
+      "step": 44674
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.667151927947998,
+      "learning_rate": 1.8257858310744065e-06,
+      "loss": 0.6116,
+      "step": 44675
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.260221242904663,
+      "learning_rate": 1.8255821222906345e-06,
+      "loss": 0.6116,
+      "step": 44676
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 5.455267429351807,
+      "learning_rate": 1.8253784223337295e-06,
+      "loss": 0.6624,
+      "step": 44677
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.0779504776000977,
+      "learning_rate": 1.8251747312042595e-06,
+      "loss": 0.5254,
+      "step": 44678
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.6566436290740967,
+      "learning_rate": 1.824971048902787e-06,
+      "loss": 0.4108,
+      "step": 44679
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.4845492839813232,
+      "learning_rate": 1.8247673754298834e-06,
+      "loss": 0.653,
+      "step": 44680
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 4.541635036468506,
+      "learning_rate": 1.8245637107861098e-06,
+      "loss": 0.4834,
+      "step": 44681
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 3.638140916824341,
+      "learning_rate": 1.824360054972037e-06,
+      "loss": 0.8497,
+      "step": 44682
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.682365655899048,
+      "learning_rate": 1.8241564079882274e-06,
+      "loss": 0.842,
+      "step": 44683
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.7777910232543945,
+      "learning_rate": 1.823952769835251e-06,
+      "loss": 0.6765,
+      "step": 44684
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.303804397583008,
+      "learning_rate": 1.8237491405136704e-06,
+      "loss": 0.4832,
+      "step": 44685
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 5.472179889678955,
+      "learning_rate": 1.8235455200240555e-06,
+      "loss": 0.5919,
+      "step": 44686
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.5681474208831787,
+      "learning_rate": 1.8233419083669685e-06,
+      "loss": 0.7549,
+      "step": 44687
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.5823071002960205,
+      "learning_rate": 1.8231383055429796e-06,
+      "loss": 0.3804,
+      "step": 44688
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 1.9666039943695068,
+      "learning_rate": 1.8229347115526513e-06,
+      "loss": 0.553,
+      "step": 44689
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.848517656326294,
+      "learning_rate": 1.8227311263965514e-06,
+      "loss": 0.544,
+      "step": 44690
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 4.207837104797363,
+      "learning_rate": 1.8225275500752471e-06,
+      "loss": 0.7167,
+      "step": 44691
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.579434394836426,
+      "learning_rate": 1.8223239825893025e-06,
+      "loss": 0.7378,
+      "step": 44692
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.1675350666046143,
+      "learning_rate": 1.8221204239392858e-06,
+      "loss": 0.5447,
+      "step": 44693
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.4965693950653076,
+      "learning_rate": 1.8219168741257603e-06,
+      "loss": 0.6206,
+      "step": 44694
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 1.9368194341659546,
+      "learning_rate": 1.8217133331492947e-06,
+      "loss": 0.6471,
+      "step": 44695
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.185102939605713,
+      "learning_rate": 1.8215098010104525e-06,
+      "loss": 0.5231,
+      "step": 44696
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.757047414779663,
+      "learning_rate": 1.8213062777098005e-06,
+      "loss": 0.6413,
+      "step": 44697
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.7991607189178467,
+      "learning_rate": 1.8211027632479066e-06,
+      "loss": 0.657,
+      "step": 44698
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 1.8440470695495605,
+      "learning_rate": 1.8208992576253337e-06,
+      "loss": 0.411,
+      "step": 44699
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.2375614643096924,
+      "learning_rate": 1.8206957608426507e-06,
+      "loss": 0.3614,
+      "step": 44700
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.6411726474761963,
+      "learning_rate": 1.8204922729004205e-06,
+      "loss": 0.4801,
+      "step": 44701
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.7841687202453613,
+      "learning_rate": 1.8202887937992091e-06,
+      "loss": 0.5523,
+      "step": 44702
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 4.85037899017334,
+      "learning_rate": 1.820085323539586e-06,
+      "loss": 0.5076,
+      "step": 44703
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.4574365615844727,
+      "learning_rate": 1.8198818621221125e-06,
+      "loss": 0.6199,
+      "step": 44704
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.7346668243408203,
+      "learning_rate": 1.819678409547358e-06,
+      "loss": 0.7241,
+      "step": 44705
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 1.6601227521896362,
+      "learning_rate": 1.8194749658158844e-06,
+      "loss": 0.5114,
+      "step": 44706
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.720223903656006,
+      "learning_rate": 1.8192715309282617e-06,
+      "loss": 0.4864,
+      "step": 44707
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.4531326293945312,
+      "learning_rate": 1.8190681048850516e-06,
+      "loss": 0.6758,
+      "step": 44708
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.865325450897217,
+      "learning_rate": 1.8188646876868226e-06,
+      "loss": 0.5053,
+      "step": 44709
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.364449977874756,
+      "learning_rate": 1.8186612793341379e-06,
+      "loss": 0.7601,
+      "step": 44710
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.520989179611206,
+      "learning_rate": 1.8184578798275665e-06,
+      "loss": 0.5507,
+      "step": 44711
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.576446056365967,
+      "learning_rate": 1.8182544891676695e-06,
+      "loss": 0.6461,
+      "step": 44712
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.248643159866333,
+      "learning_rate": 1.8180511073550171e-06,
+      "loss": 0.6891,
+      "step": 44713
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 1.5817967653274536,
+      "learning_rate": 1.8178477343901706e-06,
+      "loss": 0.4612,
+      "step": 44714
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.5476553440093994,
+      "learning_rate": 1.8176443702736996e-06,
+      "loss": 0.6846,
+      "step": 44715
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 3.1936376094818115,
+      "learning_rate": 1.817441015006165e-06,
+      "loss": 0.4195,
+      "step": 44716
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 3.8824172019958496,
+      "learning_rate": 1.8172376685881354e-06,
+      "loss": 0.6585,
+      "step": 44717
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.261549472808838,
+      "learning_rate": 1.8170343310201755e-06,
+      "loss": 0.4407,
+      "step": 44718
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.846686840057373,
+      "learning_rate": 1.816831002302852e-06,
+      "loss": 0.5415,
+      "step": 44719
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.225634813308716,
+      "learning_rate": 1.8166276824367274e-06,
+      "loss": 0.5535,
+      "step": 44720
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 3.6209187507629395,
+      "learning_rate": 1.8164243714223705e-06,
+      "loss": 0.4684,
+      "step": 44721
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 1.668834924697876,
+      "learning_rate": 1.8162210692603438e-06,
+      "loss": 0.4065,
+      "step": 44722
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.5686042308807373,
+      "learning_rate": 1.8160177759512142e-06,
+      "loss": 0.6543,
+      "step": 44723
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.387458086013794,
+      "learning_rate": 1.8158144914955456e-06,
+      "loss": 0.7631,
+      "step": 44724
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.820763111114502,
+      "learning_rate": 1.8156112158939054e-06,
+      "loss": 0.4633,
+      "step": 44725
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.3877875804901123,
+      "learning_rate": 1.8154079491468558e-06,
+      "loss": 0.4584,
+      "step": 44726
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.5143635272979736,
+      "learning_rate": 1.815204691254966e-06,
+      "loss": 0.4416,
+      "step": 44727
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.6069514751434326,
+      "learning_rate": 1.8150014422187962e-06,
+      "loss": 0.7451,
+      "step": 44728
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.4422426223754883,
+      "learning_rate": 1.8147982020389165e-06,
+      "loss": 0.5666,
+      "step": 44729
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.9278347492218018,
+      "learning_rate": 1.8145949707158882e-06,
+      "loss": 0.8216,
+      "step": 44730
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.4285740852355957,
+      "learning_rate": 1.8143917482502793e-06,
+      "loss": 0.5328,
+      "step": 44731
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.74910306930542,
+      "learning_rate": 1.8141885346426524e-06,
+      "loss": 0.778,
+      "step": 44732
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 3.766540288925171,
+      "learning_rate": 1.813985329893575e-06,
+      "loss": 0.7004,
+      "step": 44733
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 3.3203909397125244,
+      "learning_rate": 1.8137821340036093e-06,
+      "loss": 0.6205,
+      "step": 44734
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.360914945602417,
+      "learning_rate": 1.813578946973324e-06,
+      "loss": 0.6161,
+      "step": 44735
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 3.2107253074645996,
+      "learning_rate": 1.81337576880328e-06,
+      "loss": 0.9219,
+      "step": 44736
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.752323865890503,
+      "learning_rate": 1.8131725994940441e-06,
+      "loss": 0.7154,
+      "step": 44737
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.442584991455078,
+      "learning_rate": 1.8129694390461816e-06,
+      "loss": 0.3908,
+      "step": 44738
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.4379053115844727,
+      "learning_rate": 1.8127662874602587e-06,
+      "loss": 0.3724,
+      "step": 44739
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.6581153869628906,
+      "learning_rate": 1.8125631447368375e-06,
+      "loss": 0.7631,
+      "step": 44740
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 4.201959133148193,
+      "learning_rate": 1.8123600108764855e-06,
+      "loss": 0.6666,
+      "step": 44741
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.102142095565796,
+      "learning_rate": 1.8121568858797644e-06,
+      "loss": 0.6839,
+      "step": 44742
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 3.4791324138641357,
+      "learning_rate": 1.8119537697472428e-06,
+      "loss": 0.4996,
+      "step": 44743
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.8715810775756836,
+      "learning_rate": 1.8117506624794818e-06,
+      "loss": 0.6057,
+      "step": 44744
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 3.6992805004119873,
+      "learning_rate": 1.8115475640770492e-06,
+      "loss": 0.8259,
+      "step": 44745
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 1.9164820909500122,
+      "learning_rate": 1.8113444745405069e-06,
+      "loss": 0.4167,
+      "step": 44746
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.6647791862487793,
+      "learning_rate": 1.8111413938704232e-06,
+      "loss": 0.6917,
+      "step": 44747
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 8.642288208007812,
+      "learning_rate": 1.8109383220673582e-06,
+      "loss": 0.324,
+      "step": 44748
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.304288864135742,
+      "learning_rate": 1.810735259131881e-06,
+      "loss": 0.2955,
+      "step": 44749
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.212106227874756,
+      "learning_rate": 1.810532205064553e-06,
+      "loss": 0.5323,
+      "step": 44750
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.7559540271759033,
+      "learning_rate": 1.81032915986594e-06,
+      "loss": 0.4491,
+      "step": 44751
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.143887758255005,
+      "learning_rate": 1.8101261235366085e-06,
+      "loss": 0.5975,
+      "step": 44752
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 1.9212923049926758,
+      "learning_rate": 1.8099230960771198e-06,
+      "loss": 0.6387,
+      "step": 44753
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 3.9438576698303223,
+      "learning_rate": 1.8097200774880414e-06,
+      "loss": 0.9075,
+      "step": 44754
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.634235382080078,
+      "learning_rate": 1.8095170677699343e-06,
+      "loss": 0.5502,
+      "step": 44755
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.5813791751861572,
+      "learning_rate": 1.809314066923367e-06,
+      "loss": 0.5724,
+      "step": 44756
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 3.5319671630859375,
+      "learning_rate": 1.8091110749489004e-06,
+      "loss": 0.9367,
+      "step": 44757
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 5.620607852935791,
+      "learning_rate": 1.808908091847101e-06,
+      "loss": 0.472,
+      "step": 44758
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 1.8724545240402222,
+      "learning_rate": 1.808705117618534e-06,
+      "loss": 0.4788,
+      "step": 44759
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.265697479248047,
+      "learning_rate": 1.8085021522637609e-06,
+      "loss": 0.6718,
+      "step": 44760
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 1.882214903831482,
+      "learning_rate": 1.808299195783349e-06,
+      "loss": 0.4519,
+      "step": 44761
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 1.6054201126098633,
+      "learning_rate": 1.8080962481778602e-06,
+      "loss": 0.2664,
+      "step": 44762
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 3.16349458694458,
+      "learning_rate": 1.80789330944786e-06,
+      "loss": 0.4913,
+      "step": 44763
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 4.347250461578369,
+      "learning_rate": 1.8076903795939143e-06,
+      "loss": 0.5452,
+      "step": 44764
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.4615020751953125,
+      "learning_rate": 1.8074874586165842e-06,
+      "loss": 0.481,
+      "step": 44765
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.606743812561035,
+      "learning_rate": 1.8072845465164374e-06,
+      "loss": 0.733,
+      "step": 44766
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 1.5751806497573853,
+      "learning_rate": 1.8070816432940342e-06,
+      "loss": 0.3499,
+      "step": 44767
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 3.635679006576538,
+      "learning_rate": 1.8068787489499428e-06,
+      "loss": 0.7236,
+      "step": 44768
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.4270777702331543,
+      "learning_rate": 1.8066758634847237e-06,
+      "loss": 0.5981,
+      "step": 44769
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.2341198921203613,
+      "learning_rate": 1.8064729868989445e-06,
+      "loss": 0.6299,
+      "step": 44770
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 5.141119003295898,
+      "learning_rate": 1.806270119193166e-06,
+      "loss": 0.5191,
+      "step": 44771
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 1.9240097999572754,
+      "learning_rate": 1.8060672603679551e-06,
+      "loss": 0.5023,
+      "step": 44772
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 3.487790822982788,
+      "learning_rate": 1.8058644104238737e-06,
+      "loss": 0.5551,
+      "step": 44773
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 1.6598384380340576,
+      "learning_rate": 1.8056615693614882e-06,
+      "loss": 0.6182,
+      "step": 44774
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 3.3471603393554688,
+      "learning_rate": 1.8054587371813591e-06,
+      "loss": 0.6989,
+      "step": 44775
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.442155361175537,
+      "learning_rate": 1.8052559138840548e-06,
+      "loss": 0.7496,
+      "step": 44776
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 1.8640072345733643,
+      "learning_rate": 1.8050530994701354e-06,
+      "loss": 0.5176,
+      "step": 44777
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 3.0058069229125977,
+      "learning_rate": 1.8048502939401658e-06,
+      "loss": 0.4732,
+      "step": 44778
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.5470330715179443,
+      "learning_rate": 1.8046474972947114e-06,
+      "loss": 0.6737,
+      "step": 44779
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.395461082458496,
+      "learning_rate": 1.8044447095343365e-06,
+      "loss": 0.6933,
+      "step": 44780
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 3.6674070358276367,
+      "learning_rate": 1.804241930659602e-06,
+      "loss": 0.5827,
+      "step": 44781
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 3.3112380504608154,
+      "learning_rate": 1.8040391606710756e-06,
+      "loss": 0.5246,
+      "step": 44782
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 7.370336055755615,
+      "learning_rate": 1.8038363995693175e-06,
+      "loss": 0.5757,
+      "step": 44783
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 3.298192262649536,
+      "learning_rate": 1.8036336473548943e-06,
+      "loss": 0.4507,
+      "step": 44784
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.8051235675811768,
+      "learning_rate": 1.803430904028367e-06,
+      "loss": 0.6253,
+      "step": 44785
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.4584944248199463,
+      "learning_rate": 1.8032281695903025e-06,
+      "loss": 0.6514,
+      "step": 44786
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 3.916378974914551,
+      "learning_rate": 1.8030254440412614e-06,
+      "loss": 0.6148,
+      "step": 44787
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 3.0313310623168945,
+      "learning_rate": 1.8028227273818106e-06,
+      "loss": 0.5525,
+      "step": 44788
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.6425232887268066,
+      "learning_rate": 1.8026200196125105e-06,
+      "loss": 0.5061,
+      "step": 44789
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 3.7286179065704346,
+      "learning_rate": 1.802417320733928e-06,
+      "loss": 0.818,
+      "step": 44790
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 3.859247922897339,
+      "learning_rate": 1.8022146307466231e-06,
+      "loss": 0.6036,
+      "step": 44791
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.153010368347168,
+      "learning_rate": 1.8020119496511635e-06,
+      "loss": 0.6137,
+      "step": 44792
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.569385528564453,
+      "learning_rate": 1.801809277448109e-06,
+      "loss": 0.4115,
+      "step": 44793
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.8292508125305176,
+      "learning_rate": 1.8016066141380262e-06,
+      "loss": 0.573,
+      "step": 44794
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.5532522201538086,
+      "learning_rate": 1.8014039597214754e-06,
+      "loss": 0.7296,
+      "step": 44795
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.142245292663574,
+      "learning_rate": 1.8012013141990243e-06,
+      "loss": 0.7026,
+      "step": 44796
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 1.6702114343643188,
+      "learning_rate": 1.8009986775712318e-06,
+      "loss": 0.4346,
+      "step": 44797
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.7929530143737793,
+      "learning_rate": 1.8007960498386634e-06,
+      "loss": 0.5543,
+      "step": 44798
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.2741384506225586,
+      "learning_rate": 1.8005934310018835e-06,
+      "loss": 0.4193,
+      "step": 44799
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.0131733417510986,
+      "learning_rate": 1.8003908210614563e-06,
+      "loss": 0.4183,
+      "step": 44800
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 3.8232781887054443,
+      "learning_rate": 1.8001882200179416e-06,
+      "loss": 0.8081,
+      "step": 44801
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.621215343475342,
+      "learning_rate": 1.7999856278719063e-06,
+      "loss": 0.6555,
+      "step": 44802
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.2531652450561523,
+      "learning_rate": 1.7997830446239107e-06,
+      "loss": 0.663,
+      "step": 44803
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.6148698329925537,
+      "learning_rate": 1.7995804702745217e-06,
+      "loss": 0.5957,
+      "step": 44804
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.348134994506836,
+      "learning_rate": 1.7993779048242981e-06,
+      "loss": 0.6836,
+      "step": 44805
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.2713799476623535,
+      "learning_rate": 1.7991753482738079e-06,
+      "loss": 0.5795,
+      "step": 44806
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.4466357231140137,
+      "learning_rate": 1.79897280062361e-06,
+      "loss": 0.7784,
+      "step": 44807
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.3625566959381104,
+      "learning_rate": 1.7987702618742715e-06,
+      "loss": 0.5127,
+      "step": 44808
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 5.116898059844971,
+      "learning_rate": 1.7985677320263523e-06,
+      "loss": 0.5506,
+      "step": 44809
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.9104700088500977,
+      "learning_rate": 1.7983652110804179e-06,
+      "loss": 0.74,
+      "step": 44810
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.1652634143829346,
+      "learning_rate": 1.7981626990370294e-06,
+      "loss": 0.5018,
+      "step": 44811
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.3369014263153076,
+      "learning_rate": 1.7979601958967524e-06,
+      "loss": 0.7485,
+      "step": 44812
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.8329570293426514,
+      "learning_rate": 1.7977577016601467e-06,
+      "loss": 0.6898,
+      "step": 44813
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.714663505554199,
+      "learning_rate": 1.7975552163277775e-06,
+      "loss": 0.572,
+      "step": 44814
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 1.7534453868865967,
+      "learning_rate": 1.7973527399002094e-06,
+      "loss": 0.4015,
+      "step": 44815
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 1.8628034591674805,
+      "learning_rate": 1.7971502723780015e-06,
+      "loss": 0.3867,
+      "step": 44816
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.3265559673309326,
+      "learning_rate": 1.7969478137617209e-06,
+      "loss": 0.8202,
+      "step": 44817
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 3.6982343196868896,
+      "learning_rate": 1.7967453640519266e-06,
+      "loss": 0.8328,
+      "step": 44818
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.8644306659698486,
+      "learning_rate": 1.7965429232491838e-06,
+      "loss": 0.506,
+      "step": 44819
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 3.3690333366394043,
+      "learning_rate": 1.7963404913540567e-06,
+      "loss": 0.502,
+      "step": 44820
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.4844472408294678,
+      "learning_rate": 1.7961380683671048e-06,
+      "loss": 0.6391,
+      "step": 44821
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.390125036239624,
+      "learning_rate": 1.7959356542888945e-06,
+      "loss": 0.8141,
+      "step": 44822
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.425786256790161,
+      "learning_rate": 1.7957332491199848e-06,
+      "loss": 0.561,
+      "step": 44823
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.804232597351074,
+      "learning_rate": 1.7955308528609411e-06,
+      "loss": 0.4471,
+      "step": 44824
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.5577616691589355,
+      "learning_rate": 1.795328465512327e-06,
+      "loss": 0.7456,
+      "step": 44825
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.4331014156341553,
+      "learning_rate": 1.7951260870747029e-06,
+      "loss": 0.572,
+      "step": 44826
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.386491060256958,
+      "learning_rate": 1.7949237175486333e-06,
+      "loss": 0.6283,
+      "step": 44827
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 1.6616692543029785,
+      "learning_rate": 1.794721356934679e-06,
+      "loss": 0.4168,
+      "step": 44828
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.8951327800750732,
+      "learning_rate": 1.7945190052334056e-06,
+      "loss": 0.3794,
+      "step": 44829
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.4423139095306396,
+      "learning_rate": 1.7943166624453723e-06,
+      "loss": 0.4107,
+      "step": 44830
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 9.26303482055664,
+      "learning_rate": 1.7941143285711454e-06,
+      "loss": 0.6396,
+      "step": 44831
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.6249656677246094,
+      "learning_rate": 1.7939120036112835e-06,
+      "loss": 0.6301,
+      "step": 44832
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.685603380203247,
+      "learning_rate": 1.7937096875663535e-06,
+      "loss": 0.6661,
+      "step": 44833
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.3159615993499756,
+      "learning_rate": 1.7935073804369135e-06,
+      "loss": 0.5711,
+      "step": 44834
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.418952226638794,
+      "learning_rate": 1.7933050822235304e-06,
+      "loss": 0.525,
+      "step": 44835
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.8175241947174072,
+      "learning_rate": 1.7931027929267625e-06,
+      "loss": 0.6605,
+      "step": 44836
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.588589668273926,
+      "learning_rate": 1.7929005125471765e-06,
+      "loss": 0.8047,
+      "step": 44837
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.461038112640381,
+      "learning_rate": 1.792698241085331e-06,
+      "loss": 0.6558,
+      "step": 44838
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.5157673358917236,
+      "learning_rate": 1.79249597854179e-06,
+      "loss": 0.6646,
+      "step": 44839
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.8248274326324463,
+      "learning_rate": 1.7922937249171163e-06,
+      "loss": 0.6364,
+      "step": 44840
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.082592487335205,
+      "learning_rate": 1.7920914802118738e-06,
+      "loss": 0.4439,
+      "step": 44841
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 3.388096809387207,
+      "learning_rate": 1.7918892444266212e-06,
+      "loss": 0.6846,
+      "step": 44842
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.7389109134674072,
+      "learning_rate": 1.791687017561925e-06,
+      "loss": 0.5027,
+      "step": 44843
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.5909945964813232,
+      "learning_rate": 1.7914847996183433e-06,
+      "loss": 0.5681,
+      "step": 44844
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.5690324306488037,
+      "learning_rate": 1.7912825905964425e-06,
+      "loss": 0.6297,
+      "step": 44845
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.8434669971466064,
+      "learning_rate": 1.7910803904967805e-06,
+      "loss": 0.4329,
+      "step": 44846
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 3.173771858215332,
+      "learning_rate": 1.7908781993199242e-06,
+      "loss": 0.9684,
+      "step": 44847
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.0822112560272217,
+      "learning_rate": 1.790676017066431e-06,
+      "loss": 0.6478,
+      "step": 44848
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.7359871864318848,
+      "learning_rate": 1.790473843736868e-06,
+      "loss": 0.4656,
+      "step": 44849
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.405763864517212,
+      "learning_rate": 1.790271679331793e-06,
+      "loss": 0.5643,
+      "step": 44850
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.477292060852051,
+      "learning_rate": 1.7900695238517718e-06,
+      "loss": 0.5083,
+      "step": 44851
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.1941912174224854,
+      "learning_rate": 1.789867377297363e-06,
+      "loss": 0.4843,
+      "step": 44852
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.2235586643218994,
+      "learning_rate": 1.789665239669132e-06,
+      "loss": 0.7054,
+      "step": 44853
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.348693370819092,
+      "learning_rate": 1.789463110967638e-06,
+      "loss": 0.4865,
+      "step": 44854
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 3.2208876609802246,
+      "learning_rate": 1.789260991193446e-06,
+      "loss": 0.7474,
+      "step": 44855
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 3.119001865386963,
+      "learning_rate": 1.7890588803471142e-06,
+      "loss": 0.6665,
+      "step": 44856
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 3.0472331047058105,
+      "learning_rate": 1.7888567784292094e-06,
+      "loss": 0.7041,
+      "step": 44857
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.3492355346679688,
+      "learning_rate": 1.7886546854402885e-06,
+      "loss": 0.4559,
+      "step": 44858
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.2277112007141113,
+      "learning_rate": 1.7884526013809167e-06,
+      "loss": 0.5731,
+      "step": 44859
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 1.8896570205688477,
+      "learning_rate": 1.7882505262516541e-06,
+      "loss": 0.5793,
+      "step": 44860
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.1804027557373047,
+      "learning_rate": 1.7880484600530663e-06,
+      "loss": 0.6559,
+      "step": 44861
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.720353126525879,
+      "learning_rate": 1.7878464027857106e-06,
+      "loss": 0.5878,
+      "step": 44862
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.5525145530700684,
+      "learning_rate": 1.787644354450152e-06,
+      "loss": 0.4098,
+      "step": 44863
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.6933705806732178,
+      "learning_rate": 1.78744231504695e-06,
+      "loss": 0.8296,
+      "step": 44864
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 12.572893142700195,
+      "learning_rate": 1.7872402845766683e-06,
+      "loss": 0.3402,
+      "step": 44865
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 3.5315425395965576,
+      "learning_rate": 1.7870382630398669e-06,
+      "loss": 0.5499,
+      "step": 44866
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.4625346660614014,
+      "learning_rate": 1.78683625043711e-06,
+      "loss": 0.6257,
+      "step": 44867
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.66507625579834,
+      "learning_rate": 1.7866342467689563e-06,
+      "loss": 0.4596,
+      "step": 44868
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.482574939727783,
+      "learning_rate": 1.78643225203597e-06,
+      "loss": 0.5083,
+      "step": 44869
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 3.067939519882202,
+      "learning_rate": 1.786230266238711e-06,
+      "loss": 0.8491,
+      "step": 44870
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.5207698345184326,
+      "learning_rate": 1.7860282893777425e-06,
+      "loss": 0.8034,
+      "step": 44871
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 3.812969446182251,
+      "learning_rate": 1.7858263214536243e-06,
+      "loss": 0.6336,
+      "step": 44872
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.8399224281311035,
+      "learning_rate": 1.7856243624669205e-06,
+      "loss": 0.8416,
+      "step": 44873
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.9483838081359863,
+      "learning_rate": 1.7854224124181896e-06,
+      "loss": 0.4013,
+      "step": 44874
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.5232462882995605,
+      "learning_rate": 1.785220471307995e-06,
+      "loss": 0.3928,
+      "step": 44875
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 3.1109604835510254,
+      "learning_rate": 1.785018539136899e-06,
+      "loss": 0.6716,
+      "step": 44876
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.408389091491699,
+      "learning_rate": 1.7848166159054614e-06,
+      "loss": 0.5113,
+      "step": 44877
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.709376573562622,
+      "learning_rate": 1.784614701614245e-06,
+      "loss": 0.6008,
+      "step": 44878
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.932522773742676,
+      "learning_rate": 1.7844127962638097e-06,
+      "loss": 0.4616,
+      "step": 44879
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.0621132850646973,
+      "learning_rate": 1.7842108998547175e-06,
+      "loss": 0.5499,
+      "step": 44880
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 4.133431911468506,
+      "learning_rate": 1.7840090123875314e-06,
+      "loss": 1.0676,
+      "step": 44881
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.100193977355957,
+      "learning_rate": 1.7838071338628105e-06,
+      "loss": 0.5594,
+      "step": 44882
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.2589194774627686,
+      "learning_rate": 1.7836052642811186e-06,
+      "loss": 0.6974,
+      "step": 44883
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 5.294414043426514,
+      "learning_rate": 1.7834034036430137e-06,
+      "loss": 0.8236,
+      "step": 44884
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.498544931411743,
+      "learning_rate": 1.7832015519490608e-06,
+      "loss": 0.6771,
+      "step": 44885
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.54721736907959,
+      "learning_rate": 1.782999709199818e-06,
+      "loss": 0.491,
+      "step": 44886
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.4481942653656006,
+      "learning_rate": 1.7827978753958474e-06,
+      "loss": 0.6484,
+      "step": 44887
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.741112232208252,
+      "learning_rate": 1.7825960505377128e-06,
+      "loss": 0.518,
+      "step": 44888
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.5952348709106445,
+      "learning_rate": 1.782394234625971e-06,
+      "loss": 0.7016,
+      "step": 44889
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.9121580123901367,
+      "learning_rate": 1.7821924276611874e-06,
+      "loss": 0.8374,
+      "step": 44890
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 3.048309564590454,
+      "learning_rate": 1.78199062964392e-06,
+      "loss": 0.4749,
+      "step": 44891
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.818934917449951,
+      "learning_rate": 1.7817888405747324e-06,
+      "loss": 0.6285,
+      "step": 44892
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.3500003814697266,
+      "learning_rate": 1.781587060454183e-06,
+      "loss": 0.6046,
+      "step": 44893
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.3845784664154053,
+      "learning_rate": 1.7813852892828358e-06,
+      "loss": 0.5582,
+      "step": 44894
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.8792243003845215,
+      "learning_rate": 1.7811835270612492e-06,
+      "loss": 0.8552,
+      "step": 44895
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 1.871087670326233,
+      "learning_rate": 1.7809817737899866e-06,
+      "loss": 0.2515,
+      "step": 44896
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.0974414348602295,
+      "learning_rate": 1.7807800294696065e-06,
+      "loss": 0.5491,
+      "step": 44897
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.222788095474243,
+      "learning_rate": 1.7805782941006728e-06,
+      "loss": 0.5788,
+      "step": 44898
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.0618789196014404,
+      "learning_rate": 1.7803765676837432e-06,
+      "loss": 0.5497,
+      "step": 44899
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.6783254146575928,
+      "learning_rate": 1.78017485021938e-06,
+      "loss": 0.4947,
+      "step": 44900
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 3.558370590209961,
+      "learning_rate": 1.7799731417081439e-06,
+      "loss": 0.4647,
+      "step": 44901
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.676427125930786,
+      "learning_rate": 1.7797714421505991e-06,
+      "loss": 0.6304,
+      "step": 44902
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 1.966463327407837,
+      "learning_rate": 1.779569751547301e-06,
+      "loss": 0.5911,
+      "step": 44903
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 3.0474746227264404,
+      "learning_rate": 1.7793680698988152e-06,
+      "loss": 0.6081,
+      "step": 44904
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 5.181955337524414,
+      "learning_rate": 1.7791663972056984e-06,
+      "loss": 0.5329,
+      "step": 44905
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.1205031871795654,
+      "learning_rate": 1.778964733468515e-06,
+      "loss": 0.6514,
+      "step": 44906
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 1.8814035654067993,
+      "learning_rate": 1.7787630786878224e-06,
+      "loss": 0.615,
+      "step": 44907
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.5475454330444336,
+      "learning_rate": 1.7785614328641848e-06,
+      "loss": 0.4295,
+      "step": 44908
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.8569586277008057,
+      "learning_rate": 1.778359795998159e-06,
+      "loss": 0.6702,
+      "step": 44909
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 3.0378222465515137,
+      "learning_rate": 1.7781581680903099e-06,
+      "loss": 0.4493,
+      "step": 44910
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.2819507122039795,
+      "learning_rate": 1.7779565491411938e-06,
+      "loss": 0.5616,
+      "step": 44911
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.1622252464294434,
+      "learning_rate": 1.7777549391513754e-06,
+      "loss": 0.5409,
+      "step": 44912
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 3.5300004482269287,
+      "learning_rate": 1.777553338121412e-06,
+      "loss": 0.7547,
+      "step": 44913
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.194200277328491,
+      "learning_rate": 1.7773517460518674e-06,
+      "loss": 0.6118,
+      "step": 44914
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.696873903274536,
+      "learning_rate": 1.7771501629432986e-06,
+      "loss": 0.5763,
+      "step": 44915
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 3.91776967048645,
+      "learning_rate": 1.776948588796269e-06,
+      "loss": 0.5933,
+      "step": 44916
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 1.951844573020935,
+      "learning_rate": 1.776747023611337e-06,
+      "loss": 0.5849,
+      "step": 44917
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 4.091453552246094,
+      "learning_rate": 1.7765454673890652e-06,
+      "loss": 0.8529,
+      "step": 44918
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.4076924324035645,
+      "learning_rate": 1.776343920130012e-06,
+      "loss": 0.6766,
+      "step": 44919
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.236569404602051,
+      "learning_rate": 1.7761423818347378e-06,
+      "loss": 0.6998,
+      "step": 44920
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.988630771636963,
+      "learning_rate": 1.7759408525038046e-06,
+      "loss": 0.4874,
+      "step": 44921
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.384361982345581,
+      "learning_rate": 1.7757393321377737e-06,
+      "loss": 0.475,
+      "step": 44922
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 3.574354887008667,
+      "learning_rate": 1.775537820737202e-06,
+      "loss": 0.5314,
+      "step": 44923
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 3.423156261444092,
+      "learning_rate": 1.7753363183026533e-06,
+      "loss": 0.5775,
+      "step": 44924
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 4.255834102630615,
+      "learning_rate": 1.775134824834685e-06,
+      "loss": 0.4936,
+      "step": 44925
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.8856637477874756,
+      "learning_rate": 1.7749333403338603e-06,
+      "loss": 0.5435,
+      "step": 44926
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.438889503479004,
+      "learning_rate": 1.7747318648007361e-06,
+      "loss": 0.5034,
+      "step": 44927
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 3.555485486984253,
+      "learning_rate": 1.7745303982358763e-06,
+      "loss": 0.6155,
+      "step": 44928
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.7116189002990723,
+      "learning_rate": 1.7743289406398373e-06,
+      "loss": 0.6509,
+      "step": 44929
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.0337817668914795,
+      "learning_rate": 1.7741274920131825e-06,
+      "loss": 0.5066,
+      "step": 44930
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.9361066818237305,
+      "learning_rate": 1.7739260523564695e-06,
+      "loss": 0.7375,
+      "step": 44931
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 1.815078854560852,
+      "learning_rate": 1.7737246216702609e-06,
+      "loss": 0.4874,
+      "step": 44932
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 3.693488597869873,
+      "learning_rate": 1.773523199955114e-06,
+      "loss": 0.4377,
+      "step": 44933
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.782639265060425,
+      "learning_rate": 1.7733217872115922e-06,
+      "loss": 0.5986,
+      "step": 44934
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 3.050609827041626,
+      "learning_rate": 1.773120383440252e-06,
+      "loss": 0.7235,
+      "step": 44935
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 3.3500614166259766,
+      "learning_rate": 1.7729189886416553e-06,
+      "loss": 0.6744,
+      "step": 44936
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.1135196685791016,
+      "learning_rate": 1.7727176028163635e-06,
+      "loss": 0.3665,
+      "step": 44937
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.6516237258911133,
+      "learning_rate": 1.7725162259649336e-06,
+      "loss": 0.5164,
+      "step": 44938
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.288414478302002,
+      "learning_rate": 1.7723148580879284e-06,
+      "loss": 0.6298,
+      "step": 44939
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.478142023086548,
+      "learning_rate": 1.7721134991859052e-06,
+      "loss": 0.6073,
+      "step": 44940
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 3.751483917236328,
+      "learning_rate": 1.7719121492594249e-06,
+      "loss": 0.7803,
+      "step": 44941
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.27337384223938,
+      "learning_rate": 1.7717108083090494e-06,
+      "loss": 0.624,
+      "step": 44942
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.6087536811828613,
+      "learning_rate": 1.7715094763353352e-06,
+      "loss": 0.7313,
+      "step": 44943
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 5.685116291046143,
+      "learning_rate": 1.7713081533388455e-06,
+      "loss": 0.4975,
+      "step": 44944
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.8872711658477783,
+      "learning_rate": 1.7711068393201363e-06,
+      "loss": 0.6841,
+      "step": 44945
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.1991450786590576,
+      "learning_rate": 1.7709055342797716e-06,
+      "loss": 0.6353,
+      "step": 44946
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 1.897519588470459,
+      "learning_rate": 1.7707042382183064e-06,
+      "loss": 0.5619,
+      "step": 44947
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.929555892944336,
+      "learning_rate": 1.7705029511363037e-06,
+      "loss": 0.3729,
+      "step": 44948
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 1.9728211164474487,
+      "learning_rate": 1.7703016730343242e-06,
+      "loss": 0.3658,
+      "step": 44949
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 3.8266518115997314,
+      "learning_rate": 1.7701004039129237e-06,
+      "loss": 0.7097,
+      "step": 44950
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 4.995919704437256,
+      "learning_rate": 1.7698991437726664e-06,
+      "loss": 0.5838,
+      "step": 44951
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.2263481616973877,
+      "learning_rate": 1.7696978926141073e-06,
+      "loss": 0.4953,
+      "step": 44952
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.8775973320007324,
+      "learning_rate": 1.7694966504378102e-06,
+      "loss": 0.5398,
+      "step": 44953
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 3.304670810699463,
+      "learning_rate": 1.769295417244331e-06,
+      "loss": 0.4688,
+      "step": 44954
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.9276914596557617,
+      "learning_rate": 1.7690941930342326e-06,
+      "loss": 0.6131,
+      "step": 44955
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.061753273010254,
+      "learning_rate": 1.7688929778080715e-06,
+      "loss": 0.66,
+      "step": 44956
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.227288007736206,
+      "learning_rate": 1.7686917715664099e-06,
+      "loss": 0.3967,
+      "step": 44957
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 1.8589789867401123,
+      "learning_rate": 1.7684905743098046e-06,
+      "loss": 0.5472,
+      "step": 44958
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.70725154876709,
+      "learning_rate": 1.7682893860388184e-06,
+      "loss": 0.6951,
+      "step": 44959
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.4600119590759277,
+      "learning_rate": 1.7680882067540068e-06,
+      "loss": 0.5679,
+      "step": 44960
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.5304758548736572,
+      "learning_rate": 1.7678870364559308e-06,
+      "loss": 0.6202,
+      "step": 44961
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.441878318786621,
+      "learning_rate": 1.7676858751451509e-06,
+      "loss": 0.6511,
+      "step": 44962
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.378871202468872,
+      "learning_rate": 1.767484722822227e-06,
+      "loss": 0.4978,
+      "step": 44963
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.031874895095825,
+      "learning_rate": 1.7672835794877157e-06,
+      "loss": 0.6263,
+      "step": 44964
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 3.710142135620117,
+      "learning_rate": 1.7670824451421793e-06,
+      "loss": 0.5119,
+      "step": 44965
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.0936806201934814,
+      "learning_rate": 1.766881319786174e-06,
+      "loss": 0.5681,
+      "step": 44966
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.010413885116577,
+      "learning_rate": 1.7666802034202618e-06,
+      "loss": 0.4581,
+      "step": 44967
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.5058367252349854,
+      "learning_rate": 1.7664790960449996e-06,
+      "loss": 0.7532,
+      "step": 44968
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.332641839981079,
+      "learning_rate": 1.766277997660949e-06,
+      "loss": 0.5914,
+      "step": 44969
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.421447515487671,
+      "learning_rate": 1.7660769082686664e-06,
+      "loss": 0.4591,
+      "step": 44970
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.541806936264038,
+      "learning_rate": 1.7658758278687144e-06,
+      "loss": 0.4779,
+      "step": 44971
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.842559576034546,
+      "learning_rate": 1.7656747564616484e-06,
+      "loss": 0.6269,
+      "step": 44972
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.553013324737549,
+      "learning_rate": 1.7654736940480305e-06,
+      "loss": 0.6761,
+      "step": 44973
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.400437116622925,
+      "learning_rate": 1.765272640628417e-06,
+      "loss": 0.6957,
+      "step": 44974
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.5574233531951904,
+      "learning_rate": 1.765071596203371e-06,
+      "loss": 0.4494,
+      "step": 44975
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 3.0642707347869873,
+      "learning_rate": 1.7648705607734467e-06,
+      "loss": 0.6987,
+      "step": 44976
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.421720027923584,
+      "learning_rate": 1.7646695343392073e-06,
+      "loss": 0.6593,
+      "step": 44977
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 5.261477470397949,
+      "learning_rate": 1.7644685169012083e-06,
+      "loss": 0.7523,
+      "step": 44978
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.098926544189453,
+      "learning_rate": 1.764267508460012e-06,
+      "loss": 0.5953,
+      "step": 44979
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.5478105545043945,
+      "learning_rate": 1.7640665090161734e-06,
+      "loss": 0.6138,
+      "step": 44980
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.5977742671966553,
+      "learning_rate": 1.7638655185702542e-06,
+      "loss": 0.7403,
+      "step": 44981
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.7907705307006836,
+      "learning_rate": 1.7636645371228127e-06,
+      "loss": 0.7662,
+      "step": 44982
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.815272092819214,
+      "learning_rate": 1.7634635646744092e-06,
+      "loss": 0.5049,
+      "step": 44983
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.6809165477752686,
+      "learning_rate": 1.7632626012256e-06,
+      "loss": 0.5753,
+      "step": 44984
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.5902798175811768,
+      "learning_rate": 1.7630616467769457e-06,
+      "loss": 0.5606,
+      "step": 44985
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 3.517813205718994,
+      "learning_rate": 1.7628607013290034e-06,
+      "loss": 0.5222,
+      "step": 44986
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 1.405430793762207,
+      "learning_rate": 1.7626597648823346e-06,
+      "loss": 0.2245,
+      "step": 44987
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.277026891708374,
+      "learning_rate": 1.762458837437494e-06,
+      "loss": 0.6306,
+      "step": 44988
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 1.9459091424942017,
+      "learning_rate": 1.7622579189950444e-06,
+      "loss": 0.5818,
+      "step": 44989
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.6324071884155273,
+      "learning_rate": 1.7620570095555416e-06,
+      "loss": 0.4909,
+      "step": 44990
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.193136692047119,
+      "learning_rate": 1.7618561091195464e-06,
+      "loss": 0.3649,
+      "step": 44991
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.390122890472412,
+      "learning_rate": 1.7616552176876144e-06,
+      "loss": 0.5558,
+      "step": 44992
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.872999429702759,
+      "learning_rate": 1.7614543352603087e-06,
+      "loss": 0.5081,
+      "step": 44993
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 4.199204921722412,
+      "learning_rate": 1.7612534618381828e-06,
+      "loss": 0.6002,
+      "step": 44994
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.08860445022583,
+      "learning_rate": 1.7610525974217996e-06,
+      "loss": 0.2732,
+      "step": 44995
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.1079885959625244,
+      "learning_rate": 1.7608517420117145e-06,
+      "loss": 0.5919,
+      "step": 44996
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 1.9847381114959717,
+      "learning_rate": 1.7606508956084883e-06,
+      "loss": 0.4816,
+      "step": 44997
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.2579336166381836,
+      "learning_rate": 1.7604500582126777e-06,
+      "loss": 0.5614,
+      "step": 44998
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.746194839477539,
+      "learning_rate": 1.7602492298248414e-06,
+      "loss": 0.5864,
+      "step": 44999
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 3.2796223163604736,
+      "learning_rate": 1.7600484104455401e-06,
+      "loss": 0.752,
+      "step": 45000
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.5839335918426514,
+      "learning_rate": 1.7598476000753285e-06,
+      "loss": 0.5477,
+      "step": 45001
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 3.123793601989746,
+      "learning_rate": 1.759646798714767e-06,
+      "loss": 0.5329,
+      "step": 45002
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.6472678184509277,
+      "learning_rate": 1.7594460063644159e-06,
+      "loss": 0.7719,
+      "step": 45003
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 6.157271385192871,
+      "learning_rate": 1.7592452230248298e-06,
+      "loss": 0.7867,
+      "step": 45004
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.1790294647216797,
+      "learning_rate": 1.7590444486965701e-06,
+      "loss": 0.6073,
+      "step": 45005
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.873936414718628,
+      "learning_rate": 1.758843683380192e-06,
+      "loss": 0.5077,
+      "step": 45006
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.7545247077941895,
+      "learning_rate": 1.7586429270762573e-06,
+      "loss": 0.5514,
+      "step": 45007
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 3.8965353965759277,
+      "learning_rate": 1.7584421797853206e-06,
+      "loss": 0.5926,
+      "step": 45008
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.8069963455200195,
+      "learning_rate": 1.758241441507944e-06,
+      "loss": 0.6615,
+      "step": 45009
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 1.8398629426956177,
+      "learning_rate": 1.7580407122446818e-06,
+      "loss": 0.5523,
+      "step": 45010
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 3.1787078380584717,
+      "learning_rate": 1.7578399919960936e-06,
+      "loss": 0.8722,
+      "step": 45011
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.887291431427002,
+      "learning_rate": 1.7576392807627395e-06,
+      "loss": 0.7497,
+      "step": 45012
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.051269292831421,
+      "learning_rate": 1.7574385785451742e-06,
+      "loss": 0.6153,
+      "step": 45013
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.8284034729003906,
+      "learning_rate": 1.7572378853439597e-06,
+      "loss": 0.6987,
+      "step": 45014
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.521902322769165,
+      "learning_rate": 1.7570372011596499e-06,
+      "loss": 0.5294,
+      "step": 45015
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.6594808101654053,
+      "learning_rate": 1.7568365259928061e-06,
+      "loss": 0.7489,
+      "step": 45016
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.202974319458008,
+      "learning_rate": 1.7566358598439836e-06,
+      "loss": 0.5516,
+      "step": 45017
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 1.9460481405258179,
+      "learning_rate": 1.7564352027137439e-06,
+      "loss": 0.3392,
+      "step": 45018
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.4361042976379395,
+      "learning_rate": 1.7562345546026404e-06,
+      "loss": 0.6617,
+      "step": 45019
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 3.151888370513916,
+      "learning_rate": 1.7560339155112355e-06,
+      "loss": 0.8395,
+      "step": 45020
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.013227939605713,
+      "learning_rate": 1.7558332854400833e-06,
+      "loss": 0.5946,
+      "step": 45021
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 3.3963778018951416,
+      "learning_rate": 1.7556326643897437e-06,
+      "loss": 0.6386,
+      "step": 45022
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.907017707824707,
+      "learning_rate": 1.7554320523607744e-06,
+      "loss": 0.4657,
+      "step": 45023
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.5467467308044434,
+      "learning_rate": 1.7552314493537343e-06,
+      "loss": 0.6923,
+      "step": 45024
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.9080710411071777,
+      "learning_rate": 1.7550308553691786e-06,
+      "loss": 0.6048,
+      "step": 45025
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.3248395919799805,
+      "learning_rate": 1.754830270407668e-06,
+      "loss": 0.7517,
+      "step": 45026
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.0496158599853516,
+      "learning_rate": 1.7546296944697577e-06,
+      "loss": 0.606,
+      "step": 45027
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 3.526970386505127,
+      "learning_rate": 1.7544291275560077e-06,
+      "loss": 0.4275,
+      "step": 45028
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 3.4118330478668213,
+      "learning_rate": 1.7542285696669724e-06,
+      "loss": 0.6932,
+      "step": 45029
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.473288059234619,
+      "learning_rate": 1.754028020803214e-06,
+      "loss": 0.5212,
+      "step": 45030
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.197695255279541,
+      "learning_rate": 1.7538274809652855e-06,
+      "loss": 0.5853,
+      "step": 45031
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.7456672191619873,
+      "learning_rate": 1.7536269501537483e-06,
+      "loss": 0.6006,
+      "step": 45032
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.7577075958251953,
+      "learning_rate": 1.753426428369157e-06,
+      "loss": 0.5888,
+      "step": 45033
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 4.307689666748047,
+      "learning_rate": 1.7532259156120722e-06,
+      "loss": 0.7628,
+      "step": 45034
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.803534984588623,
+      "learning_rate": 1.7530254118830487e-06,
+      "loss": 0.613,
+      "step": 45035
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.4414174556732178,
+      "learning_rate": 1.7528249171826462e-06,
+      "loss": 0.4298,
+      "step": 45036
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.1019129753112793,
+      "learning_rate": 1.7526244315114198e-06,
+      "loss": 0.5318,
+      "step": 45037
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.649329900741577,
+      "learning_rate": 1.7524239548699302e-06,
+      "loss": 0.819,
+      "step": 45038
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 3.2714247703552246,
+      "learning_rate": 1.7522234872587308e-06,
+      "loss": 0.6629,
+      "step": 45039
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.805896759033203,
+      "learning_rate": 1.7520230286783834e-06,
+      "loss": 0.633,
+      "step": 45040
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 3.1823065280914307,
+      "learning_rate": 1.7518225791294414e-06,
+      "loss": 0.6948,
+      "step": 45041
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.370939254760742,
+      "learning_rate": 1.7516221386124638e-06,
+      "loss": 0.5282,
+      "step": 45042
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.1535563468933105,
+      "learning_rate": 1.7514217071280086e-06,
+      "loss": 0.5481,
+      "step": 45043
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 1.9725303649902344,
+      "learning_rate": 1.751221284676634e-06,
+      "loss": 0.5571,
+      "step": 45044
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 3.605093479156494,
+      "learning_rate": 1.7510208712588944e-06,
+      "loss": 0.4826,
+      "step": 45045
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 3.0837135314941406,
+      "learning_rate": 1.7508204668753503e-06,
+      "loss": 0.7179,
+      "step": 45046
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.4784140586853027,
+      "learning_rate": 1.7506200715265558e-06,
+      "loss": 0.6062,
+      "step": 45047
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.972600221633911,
+      "learning_rate": 1.750419685213071e-06,
+      "loss": 0.4965,
+      "step": 45048
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 1.809492588043213,
+      "learning_rate": 1.7502193079354507e-06,
+      "loss": 0.4674,
+      "step": 45049
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 3.1644723415374756,
+      "learning_rate": 1.750018939694254e-06,
+      "loss": 0.7587,
+      "step": 45050
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.9457590579986572,
+      "learning_rate": 1.749818580490036e-06,
+      "loss": 0.6731,
+      "step": 45051
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 4.241537094116211,
+      "learning_rate": 1.7496182303233567e-06,
+      "loss": 0.4982,
+      "step": 45052
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.3233492374420166,
+      "learning_rate": 1.7494178891947694e-06,
+      "loss": 0.4886,
+      "step": 45053
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.5126121044158936,
+      "learning_rate": 1.7492175571048353e-06,
+      "loss": 0.7689,
+      "step": 45054
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.2231266498565674,
+      "learning_rate": 1.7490172340541074e-06,
+      "loss": 0.4148,
+      "step": 45055
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.212050437927246,
+      "learning_rate": 1.7488169200431464e-06,
+      "loss": 0.4085,
+      "step": 45056
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 3.61607027053833,
+      "learning_rate": 1.7486166150725065e-06,
+      "loss": 0.6789,
+      "step": 45057
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.404385805130005,
+      "learning_rate": 1.7484163191427472e-06,
+      "loss": 0.3728,
+      "step": 45058
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 3.327493190765381,
+      "learning_rate": 1.7482160322544222e-06,
+      "loss": 0.5643,
+      "step": 45059
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.6861119270324707,
+      "learning_rate": 1.7480157544080901e-06,
+      "loss": 0.7964,
+      "step": 45060
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.7267565727233887,
+      "learning_rate": 1.7478154856043101e-06,
+      "loss": 0.6117,
+      "step": 45061
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 4.109808444976807,
+      "learning_rate": 1.7476152258436347e-06,
+      "loss": 0.6692,
+      "step": 45062
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 6.250727653503418,
+      "learning_rate": 1.7474149751266234e-06,
+      "loss": 0.6031,
+      "step": 45063
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 4.06169319152832,
+      "learning_rate": 1.7472147334538343e-06,
+      "loss": 0.5378,
+      "step": 45064
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.667856454849243,
+      "learning_rate": 1.7470145008258205e-06,
+      "loss": 0.4907,
+      "step": 45065
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.093832492828369,
+      "learning_rate": 1.7468142772431429e-06,
+      "loss": 0.4169,
+      "step": 45066
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 3.0233657360076904,
+      "learning_rate": 1.7466140627063544e-06,
+      "loss": 0.6541,
+      "step": 45067
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 1.916818618774414,
+      "learning_rate": 1.7464138572160149e-06,
+      "loss": 0.6709,
+      "step": 45068
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.1703855991363525,
+      "learning_rate": 1.7462136607726782e-06,
+      "loss": 0.3299,
+      "step": 45069
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.054979085922241,
+      "learning_rate": 1.7460134733769036e-06,
+      "loss": 0.6253,
+      "step": 45070
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.4299213886260986,
+      "learning_rate": 1.745813295029245e-06,
+      "loss": 0.5966,
+      "step": 45071
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 4.079265117645264,
+      "learning_rate": 1.7456131257302611e-06,
+      "loss": 0.9853,
+      "step": 45072
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.3862483501434326,
+      "learning_rate": 1.745412965480509e-06,
+      "loss": 0.53,
+      "step": 45073
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 1.9834924936294556,
+      "learning_rate": 1.745212814280543e-06,
+      "loss": 0.5389,
+      "step": 45074
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.1181588172912598,
+      "learning_rate": 1.7450126721309223e-06,
+      "loss": 0.6235,
+      "step": 45075
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.575420618057251,
+      "learning_rate": 1.7448125390322007e-06,
+      "loss": 0.7417,
+      "step": 45076
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.473958969116211,
+      "learning_rate": 1.7446124149849375e-06,
+      "loss": 0.6909,
+      "step": 45077
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 4.093597888946533,
+      "learning_rate": 1.7444122999896862e-06,
+      "loss": 0.683,
+      "step": 45078
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.440068244934082,
+      "learning_rate": 1.7442121940470058e-06,
+      "loss": 0.5388,
+      "step": 45079
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.548552989959717,
+      "learning_rate": 1.7440120971574508e-06,
+      "loss": 0.5108,
+      "step": 45080
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.3202836513519287,
+      "learning_rate": 1.7438120093215794e-06,
+      "loss": 0.5446,
+      "step": 45081
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.7733781337738037,
+      "learning_rate": 1.7436119305399458e-06,
+      "loss": 0.467,
+      "step": 45082
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 1.9534130096435547,
+      "learning_rate": 1.7434118608131073e-06,
+      "loss": 0.5949,
+      "step": 45083
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.003896713256836,
+      "learning_rate": 1.7432118001416204e-06,
+      "loss": 0.5602,
+      "step": 45084
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.676361083984375,
+      "learning_rate": 1.7430117485260434e-06,
+      "loss": 0.7404,
+      "step": 45085
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 3.567049503326416,
+      "learning_rate": 1.7428117059669286e-06,
+      "loss": 0.7149,
+      "step": 45086
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 1.872851014137268,
+      "learning_rate": 1.7426116724648363e-06,
+      "loss": 0.4804,
+      "step": 45087
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.3278379440307617,
+      "learning_rate": 1.742411648020319e-06,
+      "loss": 0.1668,
+      "step": 45088
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 3.2317938804626465,
+      "learning_rate": 1.742211632633936e-06,
+      "loss": 0.8433,
+      "step": 45089
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.325812578201294,
+      "learning_rate": 1.7420116263062404e-06,
+      "loss": 0.6696,
+      "step": 45090
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 1.8231120109558105,
+      "learning_rate": 1.7418116290377918e-06,
+      "loss": 0.4628,
+      "step": 45091
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 3.3436315059661865,
+      "learning_rate": 1.7416116408291428e-06,
+      "loss": 0.6329,
+      "step": 45092
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.4043891429901123,
+      "learning_rate": 1.741411661680853e-06,
+      "loss": 0.5603,
+      "step": 45093
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 1.953282117843628,
+      "learning_rate": 1.7412116915934752e-06,
+      "loss": 0.6302,
+      "step": 45094
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.8105742931365967,
+      "learning_rate": 1.7410117305675683e-06,
+      "loss": 0.4716,
+      "step": 45095
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.7231228351593018,
+      "learning_rate": 1.740811778603685e-06,
+      "loss": 0.4877,
+      "step": 45096
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.3697314262390137,
+      "learning_rate": 1.7406118357023855e-06,
+      "loss": 0.4461,
+      "step": 45097
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 1.8789491653442383,
+      "learning_rate": 1.7404119018642213e-06,
+      "loss": 0.4544,
+      "step": 45098
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.2229902744293213,
+      "learning_rate": 1.7402119770897525e-06,
+      "loss": 0.4982,
+      "step": 45099
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 4.7166900634765625,
+      "learning_rate": 1.7400120613795308e-06,
+      "loss": 0.7851,
+      "step": 45100
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 3.302358627319336,
+      "learning_rate": 1.7398121547341168e-06,
+      "loss": 0.6982,
+      "step": 45101
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.5086145401000977,
+      "learning_rate": 1.7396122571540619e-06,
+      "loss": 0.7783,
+      "step": 45102
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.5039870738983154,
+      "learning_rate": 1.7394123686399239e-06,
+      "loss": 0.7644,
+      "step": 45103
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 4.349203109741211,
+      "learning_rate": 1.7392124891922586e-06,
+      "loss": 0.6727,
+      "step": 45104
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.5949130058288574,
+      "learning_rate": 1.7390126188116236e-06,
+      "loss": 0.485,
+      "step": 45105
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.3981173038482666,
+      "learning_rate": 1.7388127574985714e-06,
+      "loss": 0.6882,
+      "step": 45106
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.304288864135742,
+      "learning_rate": 1.7386129052536604e-06,
+      "loss": 0.5952,
+      "step": 45107
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.077627182006836,
+      "learning_rate": 1.7384130620774441e-06,
+      "loss": 0.4374,
+      "step": 45108
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 5.1699323654174805,
+      "learning_rate": 1.7382132279704805e-06,
+      "loss": 0.7846,
+      "step": 45109
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.4478538036346436,
+      "learning_rate": 1.7380134029333228e-06,
+      "loss": 0.5559,
+      "step": 45110
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.7031028270721436,
+      "learning_rate": 1.737813586966529e-06,
+      "loss": 0.8229,
+      "step": 45111
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.3170969486236572,
+      "learning_rate": 1.7376137800706522e-06,
+      "loss": 0.6795,
+      "step": 45112
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.630561113357544,
+      "learning_rate": 1.737413982246251e-06,
+      "loss": 0.6517,
+      "step": 45113
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.2552404403686523,
+      "learning_rate": 1.737214193493878e-06,
+      "loss": 0.4637,
+      "step": 45114
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.683828115463257,
+      "learning_rate": 1.7370144138140916e-06,
+      "loss": 0.964,
+      "step": 45115
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 1.7655484676361084,
+      "learning_rate": 1.7368146432074434e-06,
+      "loss": 0.4434,
+      "step": 45116
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.0412354469299316,
+      "learning_rate": 1.7366148816744937e-06,
+      "loss": 0.4331,
+      "step": 45117
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.811645030975342,
+      "learning_rate": 1.7364151292157939e-06,
+      "loss": 0.4191,
+      "step": 45118
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 3.003791332244873,
+      "learning_rate": 1.736215385831902e-06,
+      "loss": 0.2978,
+      "step": 45119
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 1.6034284830093384,
+      "learning_rate": 1.7360156515233712e-06,
+      "loss": 0.3518,
+      "step": 45120
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 1.79056978225708,
+      "learning_rate": 1.7358159262907582e-06,
+      "loss": 0.5471,
+      "step": 45121
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.36303448677063,
+      "learning_rate": 1.7356162101346203e-06,
+      "loss": 0.7334,
+      "step": 45122
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.577540636062622,
+      "learning_rate": 1.7354165030555086e-06,
+      "loss": 0.6595,
+      "step": 45123
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 3.821737766265869,
+      "learning_rate": 1.7352168050539808e-06,
+      "loss": 0.7927,
+      "step": 45124
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.392991542816162,
+      "learning_rate": 1.735017116130594e-06,
+      "loss": 0.5111,
+      "step": 45125
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.5363359451293945,
+      "learning_rate": 1.7348174362858993e-06,
+      "loss": 0.861,
+      "step": 45126
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.833916425704956,
+      "learning_rate": 1.7346177655204565e-06,
+      "loss": 0.7144,
+      "step": 45127
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 1.5308901071548462,
+      "learning_rate": 1.7344181038348162e-06,
+      "loss": 0.372,
+      "step": 45128
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.2322728633880615,
+      "learning_rate": 1.7342184512295378e-06,
+      "loss": 0.5908,
+      "step": 45129
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 3.001786708831787,
+      "learning_rate": 1.734018807705173e-06,
+      "loss": 0.6664,
+      "step": 45130
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 3.1656064987182617,
+      "learning_rate": 1.73381917326228e-06,
+      "loss": 0.8069,
+      "step": 45131
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.7328338623046875,
+      "learning_rate": 1.7336195479014106e-06,
+      "loss": 0.6346,
+      "step": 45132
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 3.201303005218506,
+      "learning_rate": 1.7334199316231215e-06,
+      "loss": 0.5188,
+      "step": 45133
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 3.1098546981811523,
+      "learning_rate": 1.7332203244279693e-06,
+      "loss": 0.7829,
+      "step": 45134
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.5243513584136963,
+      "learning_rate": 1.7330207263165065e-06,
+      "loss": 0.5163,
+      "step": 45135
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.29604172706604,
+      "learning_rate": 1.7328211372892906e-06,
+      "loss": 0.6474,
+      "step": 45136
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 3.5409836769104004,
+      "learning_rate": 1.7326215573468734e-06,
+      "loss": 0.6272,
+      "step": 45137
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 6.445659637451172,
+      "learning_rate": 1.732421986489814e-06,
+      "loss": 0.4461,
+      "step": 45138
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.1326942443847656,
+      "learning_rate": 1.7322224247186625e-06,
+      "loss": 0.5598,
+      "step": 45139
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 4.401742935180664,
+      "learning_rate": 1.7320228720339788e-06,
+      "loss": 0.8176,
+      "step": 45140
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 3.4161484241485596,
+      "learning_rate": 1.7318233284363128e-06,
+      "loss": 0.8057,
+      "step": 45141
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.93176531791687,
+      "learning_rate": 1.7316237939262236e-06,
+      "loss": 0.734,
+      "step": 45142
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.3119747638702393,
+      "learning_rate": 1.731424268504263e-06,
+      "loss": 0.6423,
+      "step": 45143
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.4730288982391357,
+      "learning_rate": 1.7312247521709868e-06,
+      "loss": 0.6623,
+      "step": 45144
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 4.600944995880127,
+      "learning_rate": 1.7310252449269504e-06,
+      "loss": 0.6414,
+      "step": 45145
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 13.326473236083984,
+      "learning_rate": 1.7308257467727096e-06,
+      "loss": 0.6365,
+      "step": 45146
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 4.386501789093018,
+      "learning_rate": 1.7306262577088163e-06,
+      "loss": 0.6052,
+      "step": 45147
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.1761679649353027,
+      "learning_rate": 1.730426777735828e-06,
+      "loss": 0.5336,
+      "step": 45148
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.0817928314208984,
+      "learning_rate": 1.7302273068542964e-06,
+      "loss": 0.5602,
+      "step": 45149
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 1.8256484270095825,
+      "learning_rate": 1.7300278450647794e-06,
+      "loss": 0.5016,
+      "step": 45150
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.038846969604492,
+      "learning_rate": 1.729828392367829e-06,
+      "loss": 0.5709,
+      "step": 45151
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.5960731506347656,
+      "learning_rate": 1.7296289487640016e-06,
+      "loss": 0.4125,
+      "step": 45152
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.895582437515259,
+      "learning_rate": 1.7294295142538497e-06,
+      "loss": 0.6411,
+      "step": 45153
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.61151385307312,
+      "learning_rate": 1.729230088837931e-06,
+      "loss": 0.6186,
+      "step": 45154
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.074040174484253,
+      "learning_rate": 1.729030672516796e-06,
+      "loss": 0.7316,
+      "step": 45155
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.5330560207366943,
+      "learning_rate": 1.7288312652910033e-06,
+      "loss": 0.5361,
+      "step": 45156
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.3372621536254883,
+      "learning_rate": 1.7286318671611035e-06,
+      "loss": 0.4454,
+      "step": 45157
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.0211355686187744,
+      "learning_rate": 1.7284324781276552e-06,
+      "loss": 0.4928,
+      "step": 45158
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.3761866092681885,
+      "learning_rate": 1.7282330981912088e-06,
+      "loss": 0.5681,
+      "step": 45159
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.0252163410186768,
+      "learning_rate": 1.7280337273523217e-06,
+      "loss": 0.4258,
+      "step": 45160
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.01462984085083,
+      "learning_rate": 1.727834365611546e-06,
+      "loss": 0.5836,
+      "step": 45161
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.0950114727020264,
+      "learning_rate": 1.7276350129694387e-06,
+      "loss": 0.6191,
+      "step": 45162
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 5.432683944702148,
+      "learning_rate": 1.7274356694265508e-06,
+      "loss": 0.7839,
+      "step": 45163
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.8783905506134033,
+      "learning_rate": 1.7272363349834388e-06,
+      "loss": 0.5388,
+      "step": 45164
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.113635301589966,
+      "learning_rate": 1.7270370096406557e-06,
+      "loss": 0.8157,
+      "step": 45165
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.9048285484313965,
+      "learning_rate": 1.7268376933987585e-06,
+      "loss": 0.6484,
+      "step": 45166
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 1.9818899631500244,
+      "learning_rate": 1.726638386258298e-06,
+      "loss": 0.7119,
+      "step": 45167
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.5436110496520996,
+      "learning_rate": 1.7264390882198318e-06,
+      "loss": 0.5872,
+      "step": 45168
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.2186379432678223,
+      "learning_rate": 1.7262397992839103e-06,
+      "loss": 0.5403,
+      "step": 45169
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.649625062942505,
+      "learning_rate": 1.7260405194510916e-06,
+      "loss": 0.5751,
+      "step": 45170
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.43369722366333,
+      "learning_rate": 1.7258412487219255e-06,
+      "loss": 0.5447,
+      "step": 45171
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.169809103012085,
+      "learning_rate": 1.7256419870969704e-06,
+      "loss": 0.5038,
+      "step": 45172
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.5696167945861816,
+      "learning_rate": 1.7254427345767765e-06,
+      "loss": 0.5608,
+      "step": 45173
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.1435883045196533,
+      "learning_rate": 1.7252434911619016e-06,
+      "loss": 0.6542,
+      "step": 45174
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 4.010092258453369,
+      "learning_rate": 1.7250442568528957e-06,
+      "loss": 0.569,
+      "step": 45175
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 1.7168946266174316,
+      "learning_rate": 1.7248450316503167e-06,
+      "loss": 0.3772,
+      "step": 45176
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 6.049144268035889,
+      "learning_rate": 1.7246458155547151e-06,
+      "loss": 0.6,
+      "step": 45177
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.56227707862854,
+      "learning_rate": 1.724446608566649e-06,
+      "loss": 0.6962,
+      "step": 45178
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.0721840858459473,
+      "learning_rate": 1.724247410686667e-06,
+      "loss": 0.6052,
+      "step": 45179
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.6252806186676025,
+      "learning_rate": 1.7240482219153282e-06,
+      "loss": 0.8041,
+      "step": 45180
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.0394341945648193,
+      "learning_rate": 1.7238490422531817e-06,
+      "loss": 0.765,
+      "step": 45181
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 4.9430317878723145,
+      "learning_rate": 1.723649871700786e-06,
+      "loss": 0.6725,
+      "step": 45182
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.625483274459839,
+      "learning_rate": 1.723450710258691e-06,
+      "loss": 0.7009,
+      "step": 45183
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.5434682369232178,
+      "learning_rate": 1.7232515579274522e-06,
+      "loss": 0.7255,
+      "step": 45184
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.8389742374420166,
+      "learning_rate": 1.7230524147076228e-06,
+      "loss": 0.6937,
+      "step": 45185
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.9721474647521973,
+      "learning_rate": 1.7228532805997595e-06,
+      "loss": 0.5216,
+      "step": 45186
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.4291112422943115,
+      "learning_rate": 1.7226541556044112e-06,
+      "loss": 0.5269,
+      "step": 45187
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.0400242805480957,
+      "learning_rate": 1.7224550397221357e-06,
+      "loss": 0.4476,
+      "step": 45188
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.2856838703155518,
+      "learning_rate": 1.7222559329534838e-06,
+      "loss": 0.6022,
+      "step": 45189
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.4715635776519775,
+      "learning_rate": 1.7220568352990113e-06,
+      "loss": 0.8677,
+      "step": 45190
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.7786753177642822,
+      "learning_rate": 1.7218577467592696e-06,
+      "loss": 0.4594,
+      "step": 45191
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 4.341715335845947,
+      "learning_rate": 1.721658667334815e-06,
+      "loss": 0.5187,
+      "step": 45192
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.0296380519866943,
+      "learning_rate": 1.7214595970261977e-06,
+      "loss": 0.4342,
+      "step": 45193
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.611736297607422,
+      "learning_rate": 1.7212605358339751e-06,
+      "loss": 0.3857,
+      "step": 45194
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.166935920715332,
+      "learning_rate": 1.721061483758697e-06,
+      "loss": 0.4657,
+      "step": 45195
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.277005910873413,
+      "learning_rate": 1.7208624408009184e-06,
+      "loss": 0.6932,
+      "step": 45196
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.980576276779175,
+      "learning_rate": 1.720663406961195e-06,
+      "loss": 0.454,
+      "step": 45197
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.927513837814331,
+      "learning_rate": 1.7204643822400763e-06,
+      "loss": 0.8108,
+      "step": 45198
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.038609504699707,
+      "learning_rate": 1.720265366638119e-06,
+      "loss": 0.3685,
+      "step": 45199
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 4.005927085876465,
+      "learning_rate": 1.7200663601558738e-06,
+      "loss": 0.4436,
+      "step": 45200
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.532658100128174,
+      "learning_rate": 1.7198673627938967e-06,
+      "loss": 0.7051,
+      "step": 45201
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 1.9168281555175781,
+      "learning_rate": 1.7196683745527382e-06,
+      "loss": 0.3662,
+      "step": 45202
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.1747283935546875,
+      "learning_rate": 1.719469395432955e-06,
+      "loss": 0.5247,
+      "step": 45203
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.14650821685791,
+      "learning_rate": 1.7192704254350962e-06,
+      "loss": 0.6442,
+      "step": 45204
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.3366663455963135,
+      "learning_rate": 1.719071464559718e-06,
+      "loss": 0.69,
+      "step": 45205
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 1.8424582481384277,
+      "learning_rate": 1.718872512807373e-06,
+      "loss": 0.4918,
+      "step": 45206
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.1241958141326904,
+      "learning_rate": 1.7186735701786155e-06,
+      "loss": 0.6102,
+      "step": 45207
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 4.31905460357666,
+      "learning_rate": 1.7184746366739962e-06,
+      "loss": 0.4982,
+      "step": 45208
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.395763874053955,
+      "learning_rate": 1.7182757122940713e-06,
+      "loss": 0.5193,
+      "step": 45209
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.15071702003479,
+      "learning_rate": 1.7180767970393908e-06,
+      "loss": 0.49,
+      "step": 45210
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.417097806930542,
+      "learning_rate": 1.717877890910511e-06,
+      "loss": 0.7115,
+      "step": 45211
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.395843029022217,
+      "learning_rate": 1.7176789939079812e-06,
+      "loss": 0.9443,
+      "step": 45212
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.2815730571746826,
+      "learning_rate": 1.7174801060323587e-06,
+      "loss": 0.7382,
+      "step": 45213
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.127718448638916,
+      "learning_rate": 1.7172812272841927e-06,
+      "loss": 0.5076,
+      "step": 45214
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.6242973804473877,
+      "learning_rate": 1.7170823576640388e-06,
+      "loss": 0.5734,
+      "step": 45215
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.535334825515747,
+      "learning_rate": 1.716883497172448e-06,
+      "loss": 0.3968,
+      "step": 45216
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.4170620441436768,
+      "learning_rate": 1.7166846458099762e-06,
+      "loss": 0.5665,
+      "step": 45217
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.0235817432403564,
+      "learning_rate": 1.716485803577172e-06,
+      "loss": 0.5815,
+      "step": 45218
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.003990411758423,
+      "learning_rate": 1.7162869704745933e-06,
+      "loss": 0.3613,
+      "step": 45219
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.0086660385131836,
+      "learning_rate": 1.716088146502788e-06,
+      "loss": 0.4346,
+      "step": 45220
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.1429641246795654,
+      "learning_rate": 1.7158893316623132e-06,
+      "loss": 0.5524,
+      "step": 45221
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.1651673316955566,
+      "learning_rate": 1.7156905259537188e-06,
+      "loss": 0.606,
+      "step": 45222
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.9210567474365234,
+      "learning_rate": 1.7154917293775597e-06,
+      "loss": 0.5989,
+      "step": 45223
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 1.9434940814971924,
+      "learning_rate": 1.7152929419343866e-06,
+      "loss": 0.3732,
+      "step": 45224
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.05155611038208,
+      "learning_rate": 1.715094163624753e-06,
+      "loss": 0.5513,
+      "step": 45225
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.4934942722320557,
+      "learning_rate": 1.714895394449212e-06,
+      "loss": 0.7384,
+      "step": 45226
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 4.358802318572998,
+      "learning_rate": 1.7146966344083177e-06,
+      "loss": 0.5535,
+      "step": 45227
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.3158276081085205,
+      "learning_rate": 1.71449788350262e-06,
+      "loss": 0.5607,
+      "step": 45228
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.362246513366699,
+      "learning_rate": 1.7142991417326748e-06,
+      "loss": 0.5955,
+      "step": 45229
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.1965534687042236,
+      "learning_rate": 1.714100409099031e-06,
+      "loss": 0.5954,
+      "step": 45230
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.2724428176879883,
+      "learning_rate": 1.7139016856022445e-06,
+      "loss": 0.7317,
+      "step": 45231
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.675419807434082,
+      "learning_rate": 1.713702971242865e-06,
+      "loss": 0.7823,
+      "step": 45232
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.429163932800293,
+      "learning_rate": 1.713504266021448e-06,
+      "loss": 0.4688,
+      "step": 45233
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.323716878890991,
+      "learning_rate": 1.7133055699385425e-06,
+      "loss": 0.5239,
+      "step": 45234
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.7455739974975586,
+      "learning_rate": 1.713106882994705e-06,
+      "loss": 0.6502,
+      "step": 45235
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.813744068145752,
+      "learning_rate": 1.7129082051904843e-06,
+      "loss": 0.3047,
+      "step": 45236
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.320452928543091,
+      "learning_rate": 1.7127095365264363e-06,
+      "loss": 0.5977,
+      "step": 45237
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.573091745376587,
+      "learning_rate": 1.7125108770031095e-06,
+      "loss": 0.5838,
+      "step": 45238
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.274036407470703,
+      "learning_rate": 1.7123122266210606e-06,
+      "loss": 0.7212,
+      "step": 45239
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.688741445541382,
+      "learning_rate": 1.712113585380838e-06,
+      "loss": 0.9191,
+      "step": 45240
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.5345966815948486,
+      "learning_rate": 1.7119149532829975e-06,
+      "loss": 0.4315,
+      "step": 45241
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 1.7299737930297852,
+      "learning_rate": 1.711716330328088e-06,
+      "loss": 0.3375,
+      "step": 45242
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.3657805919647217,
+      "learning_rate": 1.7115177165166653e-06,
+      "loss": 0.6263,
+      "step": 45243
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.312676429748535,
+      "learning_rate": 1.7113191118492784e-06,
+      "loss": 0.7138,
+      "step": 45244
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.762185573577881,
+      "learning_rate": 1.711120516326481e-06,
+      "loss": 0.6701,
+      "step": 45245
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.4794485569000244,
+      "learning_rate": 1.710921929948825e-06,
+      "loss": 0.5889,
+      "step": 45246
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.5272703170776367,
+      "learning_rate": 1.7107233527168654e-06,
+      "loss": 0.7297,
+      "step": 45247
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.2757568359375,
+      "learning_rate": 1.7105247846311495e-06,
+      "loss": 0.66,
+      "step": 45248
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.3008737564086914,
+      "learning_rate": 1.7103262256922342e-06,
+      "loss": 0.5574,
+      "step": 45249
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.90775465965271,
+      "learning_rate": 1.7101276759006673e-06,
+      "loss": 0.4435,
+      "step": 45250
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.642561912536621,
+      "learning_rate": 1.7099291352570047e-06,
+      "loss": 0.6913,
+      "step": 45251
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.0493695735931396,
+      "learning_rate": 1.709730603761795e-06,
+      "loss": 0.4953,
+      "step": 45252
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.208249568939209,
+      "learning_rate": 1.7095320814155937e-06,
+      "loss": 0.5702,
+      "step": 45253
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.432565212249756,
+      "learning_rate": 1.7093335682189494e-06,
+      "loss": 0.6897,
+      "step": 45254
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.727891206741333,
+      "learning_rate": 1.7091350641724176e-06,
+      "loss": 0.6129,
+      "step": 45255
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.930662155151367,
+      "learning_rate": 1.7089365692765464e-06,
+      "loss": 0.7019,
+      "step": 45256
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.6746792793273926,
+      "learning_rate": 1.7087380835318895e-06,
+      "loss": 0.6245,
+      "step": 45257
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.0216586589813232,
+      "learning_rate": 1.708539606939001e-06,
+      "loss": 0.5791,
+      "step": 45258
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.9861090183258057,
+      "learning_rate": 1.7083411394984294e-06,
+      "loss": 0.6357,
+      "step": 45259
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 1.6212708950042725,
+      "learning_rate": 1.7081426812107294e-06,
+      "loss": 0.3203,
+      "step": 45260
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.274721384048462,
+      "learning_rate": 1.70794423207645e-06,
+      "loss": 0.6359,
+      "step": 45261
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 1.944104790687561,
+      "learning_rate": 1.707745792096146e-06,
+      "loss": 0.6743,
+      "step": 45262
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.523749589920044,
+      "learning_rate": 1.7075473612703658e-06,
+      "loss": 0.5897,
+      "step": 45263
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.332792282104492,
+      "learning_rate": 1.7073489395996645e-06,
+      "loss": 0.753,
+      "step": 45264
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.237745761871338,
+      "learning_rate": 1.7071505270845912e-06,
+      "loss": 0.5107,
+      "step": 45265
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.3600690364837646,
+      "learning_rate": 1.7069521237256986e-06,
+      "loss": 0.5257,
+      "step": 45266
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.399526357650757,
+      "learning_rate": 1.7067537295235404e-06,
+      "loss": 0.6916,
+      "step": 45267
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.211458683013916,
+      "learning_rate": 1.7065553444786641e-06,
+      "loss": 0.4431,
+      "step": 45268
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.6148681640625,
+      "learning_rate": 1.7063569685916242e-06,
+      "loss": 0.6991,
+      "step": 45269
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.6690783500671387,
+      "learning_rate": 1.7061586018629734e-06,
+      "loss": 0.6115,
+      "step": 45270
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.1783382892608643,
+      "learning_rate": 1.7059602442932594e-06,
+      "loss": 0.5195,
+      "step": 45271
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.304753303527832,
+      "learning_rate": 1.705761895883038e-06,
+      "loss": 0.5529,
+      "step": 45272
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.0047965049743652,
+      "learning_rate": 1.705563556632857e-06,
+      "loss": 0.4511,
+      "step": 45273
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 1.7751288414001465,
+      "learning_rate": 1.7053652265432718e-06,
+      "loss": 0.6678,
+      "step": 45274
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.4039461612701416,
+      "learning_rate": 1.7051669056148296e-06,
+      "loss": 0.6923,
+      "step": 45275
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.11926531791687,
+      "learning_rate": 1.7049685938480853e-06,
+      "loss": 0.4811,
+      "step": 45276
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.3217647075653076,
+      "learning_rate": 1.7047702912435881e-06,
+      "loss": 0.61,
+      "step": 45277
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.8269970417022705,
+      "learning_rate": 1.7045719978018915e-06,
+      "loss": 0.5243,
+      "step": 45278
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.2607874870300293,
+      "learning_rate": 1.7043737135235438e-06,
+      "loss": 0.3848,
+      "step": 45279
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.1090915203094482,
+      "learning_rate": 1.7041754384091004e-06,
+      "loss": 0.7,
+      "step": 45280
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.5638983249664307,
+      "learning_rate": 1.7039771724591087e-06,
+      "loss": 0.6935,
+      "step": 45281
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.1015758514404297,
+      "learning_rate": 1.703778915674123e-06,
+      "loss": 0.552,
+      "step": 45282
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.1914632320404053,
+      "learning_rate": 1.7035806680546917e-06,
+      "loss": 0.5463,
+      "step": 45283
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.734795570373535,
+      "learning_rate": 1.70338242960137e-06,
+      "loss": 0.5203,
+      "step": 45284
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 1.7346266508102417,
+      "learning_rate": 1.7031842003147048e-06,
+      "loss": 0.2982,
+      "step": 45285
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.1241140365600586,
+      "learning_rate": 1.7029859801952491e-06,
+      "loss": 0.6139,
+      "step": 45286
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.5477309226989746,
+      "learning_rate": 1.7027877692435546e-06,
+      "loss": 0.6079,
+      "step": 45287
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 4.296140670776367,
+      "learning_rate": 1.702589567460174e-06,
+      "loss": 0.9092,
+      "step": 45288
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.334459066390991,
+      "learning_rate": 1.7023913748456544e-06,
+      "loss": 0.7271,
+      "step": 45289
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 1.9500843286514282,
+      "learning_rate": 1.7021931914005512e-06,
+      "loss": 0.5566,
+      "step": 45290
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.8405234813690186,
+      "learning_rate": 1.7019950171254113e-06,
+      "loss": 0.6987,
+      "step": 45291
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.0115761756896973,
+      "learning_rate": 1.7017968520207895e-06,
+      "loss": 0.5198,
+      "step": 45292
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.246699094772339,
+      "learning_rate": 1.7015986960872339e-06,
+      "loss": 0.5805,
+      "step": 45293
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.026991605758667,
+      "learning_rate": 1.7014005493252978e-06,
+      "loss": 0.4812,
+      "step": 45294
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 4.43538236618042,
+      "learning_rate": 1.7012024117355296e-06,
+      "loss": 0.6832,
+      "step": 45295
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.9313852787017822,
+      "learning_rate": 1.7010042833184836e-06,
+      "loss": 0.2847,
+      "step": 45296
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 1.917319893836975,
+      "learning_rate": 1.700806164074707e-06,
+      "loss": 0.512,
+      "step": 45297
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.3070483207702637,
+      "learning_rate": 1.7006080540047543e-06,
+      "loss": 0.8597,
+      "step": 45298
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.7282650470733643,
+      "learning_rate": 1.7004099531091729e-06,
+      "loss": 0.7934,
+      "step": 45299
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.1945042610168457,
+      "learning_rate": 1.700211861388517e-06,
+      "loss": 0.6769,
+      "step": 45300
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.0494794845581055,
+      "learning_rate": 1.700013778843334e-06,
+      "loss": 0.4665,
+      "step": 45301
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.801687717437744,
+      "learning_rate": 1.6998157054741787e-06,
+      "loss": 0.4881,
+      "step": 45302
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.7566516399383545,
+      "learning_rate": 1.6996176412815973e-06,
+      "loss": 0.7338,
+      "step": 45303
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 4.532413005828857,
+      "learning_rate": 1.699419586266145e-06,
+      "loss": 0.5627,
+      "step": 45304
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.3998236656188965,
+      "learning_rate": 1.6992215404283685e-06,
+      "loss": 0.6167,
+      "step": 45305
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.0015759468078613,
+      "learning_rate": 1.6990235037688207e-06,
+      "loss": 0.6635,
+      "step": 45306
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.7541823387145996,
+      "learning_rate": 1.6988254762880518e-06,
+      "loss": 0.7347,
+      "step": 45307
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 4.510593891143799,
+      "learning_rate": 1.6986274579866141e-06,
+      "loss": 0.4498,
+      "step": 45308
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.5692684650421143,
+      "learning_rate": 1.6984294488650555e-06,
+      "loss": 0.8993,
+      "step": 45309
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 1.6264036893844604,
+      "learning_rate": 1.6982314489239288e-06,
+      "loss": 0.3294,
+      "step": 45310
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.2596166133880615,
+      "learning_rate": 1.6980334581637824e-06,
+      "loss": 0.6277,
+      "step": 45311
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.250805139541626,
+      "learning_rate": 1.6978354765851696e-06,
+      "loss": 0.5116,
+      "step": 45312
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.2489185333251953,
+      "learning_rate": 1.6976375041886374e-06,
+      "loss": 0.5924,
+      "step": 45313
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 1.8704012632369995,
+      "learning_rate": 1.6974395409747402e-06,
+      "loss": 0.4395,
+      "step": 45314
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.583211898803711,
+      "learning_rate": 1.6972415869440245e-06,
+      "loss": 0.7284,
+      "step": 45315
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.0145552158355713,
+      "learning_rate": 1.697043642097045e-06,
+      "loss": 0.581,
+      "step": 45316
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.7181050777435303,
+      "learning_rate": 1.6968457064343475e-06,
+      "loss": 0.6755,
+      "step": 45317
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.600558042526245,
+      "learning_rate": 1.6966477799564846e-06,
+      "loss": 0.7036,
+      "step": 45318
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.1462149620056152,
+      "learning_rate": 1.696449862664009e-06,
+      "loss": 0.5932,
+      "step": 45319
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.899792194366455,
+      "learning_rate": 1.6962519545574669e-06,
+      "loss": 0.7588,
+      "step": 45320
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.9600820541381836,
+      "learning_rate": 1.6960540556374116e-06,
+      "loss": 0.6982,
+      "step": 45321
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.174116849899292,
+      "learning_rate": 1.6958561659043915e-06,
+      "loss": 0.6842,
+      "step": 45322
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.745129108428955,
+      "learning_rate": 1.6956582853589587e-06,
+      "loss": 0.6763,
+      "step": 45323
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.3812167644500732,
+      "learning_rate": 1.6954604140016611e-06,
+      "loss": 0.6368,
+      "step": 45324
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.6603918075561523,
+      "learning_rate": 1.6952625518330517e-06,
+      "loss": 0.5604,
+      "step": 45325
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.6875932216644287,
+      "learning_rate": 1.6950646988536772e-06,
+      "loss": 0.4602,
+      "step": 45326
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.219372510910034,
+      "learning_rate": 1.6948668550640901e-06,
+      "loss": 0.4763,
+      "step": 45327
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.8070662021636963,
+      "learning_rate": 1.6946690204648418e-06,
+      "loss": 0.6429,
+      "step": 45328
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.059526205062866,
+      "learning_rate": 1.6944711950564792e-06,
+      "loss": 0.4463,
+      "step": 45329
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.663210391998291,
+      "learning_rate": 1.6942733788395539e-06,
+      "loss": 0.5556,
+      "step": 45330
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.6793227195739746,
+      "learning_rate": 1.6940755718146174e-06,
+      "loss": 0.6043,
+      "step": 45331
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.4012019634246826,
+      "learning_rate": 1.6938777739822166e-06,
+      "loss": 0.4096,
+      "step": 45332
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.673550605773926,
+      "learning_rate": 1.6936799853429049e-06,
+      "loss": 0.6316,
+      "step": 45333
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.234506130218506,
+      "learning_rate": 1.6934822058972289e-06,
+      "loss": 0.8164,
+      "step": 45334
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.381969690322876,
+      "learning_rate": 1.6932844356457423e-06,
+      "loss": 0.7942,
+      "step": 45335
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 1.6089614629745483,
+      "learning_rate": 1.6930866745889906e-06,
+      "loss": 0.72,
+      "step": 45336
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.7003512382507324,
+      "learning_rate": 1.692888922727528e-06,
+      "loss": 0.6354,
+      "step": 45337
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.5650079250335693,
+      "learning_rate": 1.692691180061901e-06,
+      "loss": 0.593,
+      "step": 45338
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.433866500854492,
+      "learning_rate": 1.6924934465926624e-06,
+      "loss": 0.5542,
+      "step": 45339
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.5177009105682373,
+      "learning_rate": 1.692295722320359e-06,
+      "loss": 0.6026,
+      "step": 45340
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 1.7387454509735107,
+      "learning_rate": 1.692098007245544e-06,
+      "loss": 0.3356,
+      "step": 45341
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.6826155185699463,
+      "learning_rate": 1.691900301368763e-06,
+      "loss": 0.5332,
+      "step": 45342
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.2822022438049316,
+      "learning_rate": 1.6917026046905698e-06,
+      "loss": 0.5867,
+      "step": 45343
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.6064453125,
+      "learning_rate": 1.691504917211511e-06,
+      "loss": 0.5637,
+      "step": 45344
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.9207699298858643,
+      "learning_rate": 1.6913072389321388e-06,
+      "loss": 0.5048,
+      "step": 45345
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.3111252784729004,
+      "learning_rate": 1.6911095698530006e-06,
+      "loss": 0.7346,
+      "step": 45346
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.11892032623291,
+      "learning_rate": 1.6909119099746468e-06,
+      "loss": 0.4241,
+      "step": 45347
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 4.600301265716553,
+      "learning_rate": 1.6907142592976278e-06,
+      "loss": 0.4318,
+      "step": 45348
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.1754536628723145,
+      "learning_rate": 1.6905166178224935e-06,
+      "loss": 0.8027,
+      "step": 45349
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 1.7733347415924072,
+      "learning_rate": 1.690318985549792e-06,
+      "loss": 0.4224,
+      "step": 45350
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 4.394627571105957,
+      "learning_rate": 1.6901213624800749e-06,
+      "loss": 0.9927,
+      "step": 45351
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.0794906616210938,
+      "learning_rate": 1.6899237486138887e-06,
+      "loss": 0.3209,
+      "step": 45352
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.6826529502868652,
+      "learning_rate": 1.689726143951786e-06,
+      "loss": 0.4919,
+      "step": 45353
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.190296173095703,
+      "learning_rate": 1.6895285484943135e-06,
+      "loss": 0.4761,
+      "step": 45354
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.2091004848480225,
+      "learning_rate": 1.6893309622420235e-06,
+      "loss": 0.5986,
+      "step": 45355
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.4548702239990234,
+      "learning_rate": 1.6891333851954623e-06,
+      "loss": 0.4844,
+      "step": 45356
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.64972186088562,
+      "learning_rate": 1.6889358173551824e-06,
+      "loss": 0.8081,
+      "step": 45357
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.2910168170928955,
+      "learning_rate": 1.6887382587217304e-06,
+      "loss": 0.5409,
+      "step": 45358
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.25162935256958,
+      "learning_rate": 1.6885407092956579e-06,
+      "loss": 0.6325,
+      "step": 45359
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.1269326210021973,
+      "learning_rate": 1.688343169077512e-06,
+      "loss": 0.6147,
+      "step": 45360
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.2265021800994873,
+      "learning_rate": 1.6881456380678445e-06,
+      "loss": 0.5323,
+      "step": 45361
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.1980748176574707,
+      "learning_rate": 1.6879481162672019e-06,
+      "loss": 0.581,
+      "step": 45362
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.670912265777588,
+      "learning_rate": 1.6877506036761365e-06,
+      "loss": 0.7192,
+      "step": 45363
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.644684314727783,
+      "learning_rate": 1.6875531002951944e-06,
+      "loss": 0.6469,
+      "step": 45364
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.5388174057006836,
+      "learning_rate": 1.6873556061249274e-06,
+      "loss": 0.4982,
+      "step": 45365
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.6216063499450684,
+      "learning_rate": 1.6871581211658822e-06,
+      "loss": 0.5329,
+      "step": 45366
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.669691562652588,
+      "learning_rate": 1.6869606454186094e-06,
+      "loss": 0.7795,
+      "step": 45367
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.3832340240478516,
+      "learning_rate": 1.6867631788836575e-06,
+      "loss": 0.6902,
+      "step": 45368
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.674956798553467,
+      "learning_rate": 1.6865657215615783e-06,
+      "loss": 0.6226,
+      "step": 45369
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.961637258529663,
+      "learning_rate": 1.6863682734529164e-06,
+      "loss": 0.8105,
+      "step": 45370
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.586531162261963,
+      "learning_rate": 1.6861708345582251e-06,
+      "loss": 0.6071,
+      "step": 45371
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.064600706100464,
+      "learning_rate": 1.685973404878049e-06,
+      "loss": 0.4808,
+      "step": 45372
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.2705373764038086,
+      "learning_rate": 1.6857759844129418e-06,
+      "loss": 0.6103,
+      "step": 45373
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.392789363861084,
+      "learning_rate": 1.6855785731634478e-06,
+      "loss": 0.6834,
+      "step": 45374
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.267892837524414,
+      "learning_rate": 1.6853811711301205e-06,
+      "loss": 0.5316,
+      "step": 45375
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.8877148628234863,
+      "learning_rate": 1.6851837783135045e-06,
+      "loss": 0.5199,
+      "step": 45376
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.4857518672943115,
+      "learning_rate": 1.684986394714152e-06,
+      "loss": 0.5696,
+      "step": 45377
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.3788235187530518,
+      "learning_rate": 1.684789020332609e-06,
+      "loss": 0.6894,
+      "step": 45378
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.1667497158050537,
+      "learning_rate": 1.6845916551694274e-06,
+      "loss": 0.5447,
+      "step": 45379
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.327989101409912,
+      "learning_rate": 1.6843942992251533e-06,
+      "loss": 0.7521,
+      "step": 45380
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.4931273460388184,
+      "learning_rate": 1.684196952500336e-06,
+      "loss": 0.5975,
+      "step": 45381
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.1939926147460938,
+      "learning_rate": 1.6839996149955267e-06,
+      "loss": 0.6007,
+      "step": 45382
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.9535067081451416,
+      "learning_rate": 1.6838022867112708e-06,
+      "loss": 0.7109,
+      "step": 45383
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.115082263946533,
+      "learning_rate": 1.6836049676481197e-06,
+      "loss": 0.8808,
+      "step": 45384
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.1061058044433594,
+      "learning_rate": 1.6834076578066195e-06,
+      "loss": 0.6317,
+      "step": 45385
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.2960116863250732,
+      "learning_rate": 1.683210357187322e-06,
+      "loss": 0.5394,
+      "step": 45386
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 1.9641268253326416,
+      "learning_rate": 1.6830130657907723e-06,
+      "loss": 0.5819,
+      "step": 45387
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.7598159313201904,
+      "learning_rate": 1.6828157836175202e-06,
+      "loss": 0.5776,
+      "step": 45388
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.2234842777252197,
+      "learning_rate": 1.682618510668117e-06,
+      "loss": 0.5654,
+      "step": 45389
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.397301197052002,
+      "learning_rate": 1.6824212469431068e-06,
+      "loss": 0.6613,
+      "step": 45390
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 1.757134199142456,
+      "learning_rate": 1.6822239924430405e-06,
+      "loss": 0.385,
+      "step": 45391
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.5955231189727783,
+      "learning_rate": 1.6820267471684682e-06,
+      "loss": 0.7958,
+      "step": 45392
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.391540288925171,
+      "learning_rate": 1.6818295111199351e-06,
+      "loss": 0.6912,
+      "step": 45393
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.258782386779785,
+      "learning_rate": 1.6816322842979927e-06,
+      "loss": 0.7487,
+      "step": 45394
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.941216468811035,
+      "learning_rate": 1.6814350667031858e-06,
+      "loss": 0.4872,
+      "step": 45395
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.0176358222961426,
+      "learning_rate": 1.6812378583360667e-06,
+      "loss": 0.7215,
+      "step": 45396
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.433283567428589,
+      "learning_rate": 1.68104065919718e-06,
+      "loss": 0.5461,
+      "step": 45397
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.5620131492614746,
+      "learning_rate": 1.6808434692870778e-06,
+      "loss": 0.5918,
+      "step": 45398
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.771756410598755,
+      "learning_rate": 1.6806462886063052e-06,
+      "loss": 0.6059,
+      "step": 45399
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 1.9206019639968872,
+      "learning_rate": 1.6804491171554132e-06,
+      "loss": 0.5391,
+      "step": 45400
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.4879348278045654,
+      "learning_rate": 1.6802519549349467e-06,
+      "loss": 0.5001,
+      "step": 45401
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.7258992195129395,
+      "learning_rate": 1.6800548019454576e-06,
+      "loss": 0.4444,
+      "step": 45402
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.2992305755615234,
+      "learning_rate": 1.6798576581874915e-06,
+      "loss": 0.532,
+      "step": 45403
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.1904380321502686,
+      "learning_rate": 1.6796605236615986e-06,
+      "loss": 0.5713,
+      "step": 45404
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.22765851020813,
+      "learning_rate": 1.679463398368324e-06,
+      "loss": 0.4814,
+      "step": 45405
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.378601551055908,
+      "learning_rate": 1.67926628230822e-06,
+      "loss": 0.6714,
+      "step": 45406
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.722628355026245,
+      "learning_rate": 1.6790691754818305e-06,
+      "loss": 0.6865,
+      "step": 45407
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 1.8423298597335815,
+      "learning_rate": 1.6788720778897056e-06,
+      "loss": 0.309,
+      "step": 45408
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.358079433441162,
+      "learning_rate": 1.6786749895323934e-06,
+      "loss": 0.7364,
+      "step": 45409
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.8955516815185547,
+      "learning_rate": 1.6784779104104438e-06,
+      "loss": 0.7289,
+      "step": 45410
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.6769721508026123,
+      "learning_rate": 1.6782808405244011e-06,
+      "loss": 0.6054,
+      "step": 45411
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.474451780319214,
+      "learning_rate": 1.6780837798748163e-06,
+      "loss": 0.4735,
+      "step": 45412
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.894721746444702,
+      "learning_rate": 1.6778867284622348e-06,
+      "loss": 0.908,
+      "step": 45413
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.5211122035980225,
+      "learning_rate": 1.6776896862872072e-06,
+      "loss": 0.5032,
+      "step": 45414
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.2746942043304443,
+      "learning_rate": 1.6774926533502784e-06,
+      "loss": 0.72,
+      "step": 45415
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.6729207038879395,
+      "learning_rate": 1.677295629652e-06,
+      "loss": 0.7201,
+      "step": 45416
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.738978147506714,
+      "learning_rate": 1.6770986151929153e-06,
+      "loss": 0.5596,
+      "step": 45417
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 4.33591365814209,
+      "learning_rate": 1.676901609973577e-06,
+      "loss": 0.4688,
+      "step": 45418
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.5104217529296875,
+      "learning_rate": 1.6767046139945281e-06,
+      "loss": 0.7139,
+      "step": 45419
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.5552420616149902,
+      "learning_rate": 1.6765076272563208e-06,
+      "loss": 0.6436,
+      "step": 45420
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.587890386581421,
+      "learning_rate": 1.6763106497594988e-06,
+      "loss": 0.5844,
+      "step": 45421
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.00545334815979,
+      "learning_rate": 1.6761136815046143e-06,
+      "loss": 0.547,
+      "step": 45422
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.4960389137268066,
+      "learning_rate": 1.6759167224922102e-06,
+      "loss": 0.693,
+      "step": 45423
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.8035218715667725,
+      "learning_rate": 1.675719772722838e-06,
+      "loss": 0.4794,
+      "step": 45424
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.167818546295166,
+      "learning_rate": 1.6755228321970424e-06,
+      "loss": 0.6886,
+      "step": 45425
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.162583827972412,
+      "learning_rate": 1.675325900915374e-06,
+      "loss": 0.492,
+      "step": 45426
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 1.8370344638824463,
+      "learning_rate": 1.675128978878377e-06,
+      "loss": 0.2892,
+      "step": 45427
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.200470447540283,
+      "learning_rate": 1.6749320660866008e-06,
+      "loss": 0.3534,
+      "step": 45428
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.2870991230010986,
+      "learning_rate": 1.6747351625405933e-06,
+      "loss": 0.5986,
+      "step": 45429
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.793191432952881,
+      "learning_rate": 1.6745382682409033e-06,
+      "loss": 0.5401,
+      "step": 45430
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.347956657409668,
+      "learning_rate": 1.6743413831880744e-06,
+      "loss": 0.6429,
+      "step": 45431
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.4914724826812744,
+      "learning_rate": 1.6741445073826583e-06,
+      "loss": 0.724,
+      "step": 45432
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.1643013954162598,
+      "learning_rate": 1.6739476408251992e-06,
+      "loss": 0.61,
+      "step": 45433
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.493715286254883,
+      "learning_rate": 1.673750783516247e-06,
+      "loss": 0.5182,
+      "step": 45434
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.425391912460327,
+      "learning_rate": 1.6735539354563456e-06,
+      "loss": 0.7763,
+      "step": 45435
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 4.174607276916504,
+      "learning_rate": 1.6733570966460472e-06,
+      "loss": 0.5889,
+      "step": 45436
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.452211380004883,
+      "learning_rate": 1.6731602670858943e-06,
+      "loss": 0.5718,
+      "step": 45437
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.282322645187378,
+      "learning_rate": 1.6729634467764383e-06,
+      "loss": 0.6685,
+      "step": 45438
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.5663459300994873,
+      "learning_rate": 1.6727666357182233e-06,
+      "loss": 0.6358,
+      "step": 45439
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.391244411468506,
+      "learning_rate": 1.672569833911799e-06,
+      "loss": 0.5069,
+      "step": 45440
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.0913403034210205,
+      "learning_rate": 1.6723730413577099e-06,
+      "loss": 0.3299,
+      "step": 45441
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.176987886428833,
+      "learning_rate": 1.6721762580565048e-06,
+      "loss": 0.5316,
+      "step": 45442
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.9038915634155273,
+      "learning_rate": 1.6719794840087329e-06,
+      "loss": 0.7324,
+      "step": 45443
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.8643298149108887,
+      "learning_rate": 1.6717827192149373e-06,
+      "loss": 0.8033,
+      "step": 45444
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 4.293125629425049,
+      "learning_rate": 1.6715859636756693e-06,
+      "loss": 0.5972,
+      "step": 45445
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 1.6851260662078857,
+      "learning_rate": 1.6713892173914715e-06,
+      "loss": 0.3143,
+      "step": 45446
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.315110921859741,
+      "learning_rate": 1.6711924803628954e-06,
+      "loss": 0.5923,
+      "step": 45447
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 1.5151851177215576,
+      "learning_rate": 1.6709957525904847e-06,
+      "loss": 0.3569,
+      "step": 45448
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.8387975692749023,
+      "learning_rate": 1.6707990340747871e-06,
+      "loss": 0.5047,
+      "step": 45449
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.6295180320739746,
+      "learning_rate": 1.6706023248163523e-06,
+      "loss": 0.8807,
+      "step": 45450
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.6939384937286377,
+      "learning_rate": 1.6704056248157236e-06,
+      "loss": 0.6002,
+      "step": 45451
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.1147379875183105,
+      "learning_rate": 1.6702089340734512e-06,
+      "loss": 0.7607,
+      "step": 45452
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.488511800765991,
+      "learning_rate": 1.6700122525900786e-06,
+      "loss": 0.7684,
+      "step": 45453
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 1.8510452508926392,
+      "learning_rate": 1.6698155803661548e-06,
+      "loss": 0.3448,
+      "step": 45454
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.1907999515533447,
+      "learning_rate": 1.6696189174022276e-06,
+      "loss": 0.444,
+      "step": 45455
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.295866012573242,
+      "learning_rate": 1.6694222636988412e-06,
+      "loss": 0.4898,
+      "step": 45456
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.0259528160095215,
+      "learning_rate": 1.6692256192565454e-06,
+      "loss": 0.5706,
+      "step": 45457
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.5746803283691406,
+      "learning_rate": 1.6690289840758833e-06,
+      "loss": 0.5798,
+      "step": 45458
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.909346580505371,
+      "learning_rate": 1.6688323581574061e-06,
+      "loss": 0.7339,
+      "step": 45459
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.9067556858062744,
+      "learning_rate": 1.6686357415016563e-06,
+      "loss": 0.6367,
+      "step": 45460
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.174513578414917,
+      "learning_rate": 1.6684391341091843e-06,
+      "loss": 0.502,
+      "step": 45461
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.3315770626068115,
+      "learning_rate": 1.6682425359805332e-06,
+      "loss": 0.7564,
+      "step": 45462
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 1.8841235637664795,
+      "learning_rate": 1.668045947116253e-06,
+      "loss": 0.3942,
+      "step": 45463
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.7519493103027344,
+      "learning_rate": 1.6678493675168879e-06,
+      "loss": 0.561,
+      "step": 45464
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.2602880001068115,
+      "learning_rate": 1.6676527971829866e-06,
+      "loss": 0.5238,
+      "step": 45465
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.427009105682373,
+      "learning_rate": 1.6674562361150925e-06,
+      "loss": 0.5146,
+      "step": 45466
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 4.000092029571533,
+      "learning_rate": 1.6672596843137567e-06,
+      "loss": 0.5056,
+      "step": 45467
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.123152494430542,
+      "learning_rate": 1.6670631417795214e-06,
+      "loss": 0.6029,
+      "step": 45468
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.922182559967041,
+      "learning_rate": 1.6668666085129349e-06,
+      "loss": 0.6433,
+      "step": 45469
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 5.151956558227539,
+      "learning_rate": 1.6666700845145435e-06,
+      "loss": 0.7103,
+      "step": 45470
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.7478690147399902,
+      "learning_rate": 1.6664735697848955e-06,
+      "loss": 0.6405,
+      "step": 45471
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 6.155113697052002,
+      "learning_rate": 1.6662770643245342e-06,
+      "loss": 0.6777,
+      "step": 45472
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.275759220123291,
+      "learning_rate": 1.6660805681340091e-06,
+      "loss": 0.4174,
+      "step": 45473
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.9727721214294434,
+      "learning_rate": 1.6658840812138637e-06,
+      "loss": 0.8058,
+      "step": 45474
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.7455008029937744,
+      "learning_rate": 1.6656876035646468e-06,
+      "loss": 0.5369,
+      "step": 45475
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.0796964168548584,
+      "learning_rate": 1.6654911351869024e-06,
+      "loss": 0.6482,
+      "step": 45476
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.5208089351654053,
+      "learning_rate": 1.66529467608118e-06,
+      "loss": 0.6951,
+      "step": 45477
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 1.914951205253601,
+      "learning_rate": 1.6650982262480215e-06,
+      "loss": 0.4309,
+      "step": 45478
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.531682014465332,
+      "learning_rate": 1.664901785687978e-06,
+      "loss": 0.6872,
+      "step": 45479
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.59464693069458,
+      "learning_rate": 1.6647053544015907e-06,
+      "loss": 0.3725,
+      "step": 45480
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.4796035289764404,
+      "learning_rate": 1.6645089323894104e-06,
+      "loss": 0.6679,
+      "step": 45481
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 4.660586833953857,
+      "learning_rate": 1.66431251965198e-06,
+      "loss": 0.6041,
+      "step": 45482
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.5653181076049805,
+      "learning_rate": 1.6641161161898477e-06,
+      "loss": 0.59,
+      "step": 45483
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.0192582607269287,
+      "learning_rate": 1.663919722003558e-06,
+      "loss": 0.4394,
+      "step": 45484
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.3556771278381348,
+      "learning_rate": 1.6637233370936595e-06,
+      "loss": 0.484,
+      "step": 45485
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.664912700653076,
+      "learning_rate": 1.6635269614606948e-06,
+      "loss": 0.5302,
+      "step": 45486
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.046710252761841,
+      "learning_rate": 1.663330595105213e-06,
+      "loss": 0.5472,
+      "step": 45487
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.3845748901367188,
+      "learning_rate": 1.6631342380277576e-06,
+      "loss": 0.6214,
+      "step": 45488
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 9.878766059875488,
+      "learning_rate": 1.6629378902288757e-06,
+      "loss": 0.6604,
+      "step": 45489
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.0229411125183105,
+      "learning_rate": 1.6627415517091134e-06,
+      "loss": 0.4521,
+      "step": 45490
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.0433692932128906,
+      "learning_rate": 1.6625452224690181e-06,
+      "loss": 0.5084,
+      "step": 45491
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.335292339324951,
+      "learning_rate": 1.6623489025091332e-06,
+      "loss": 0.8311,
+      "step": 45492
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.2710413932800293,
+      "learning_rate": 1.662152591830007e-06,
+      "loss": 0.7341,
+      "step": 45493
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.138157367706299,
+      "learning_rate": 1.6619562904321822e-06,
+      "loss": 0.5614,
+      "step": 45494
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.2091660499572754,
+      "learning_rate": 1.6617599983162087e-06,
+      "loss": 0.6918,
+      "step": 45495
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.0932788848876953,
+      "learning_rate": 1.661563715482628e-06,
+      "loss": 0.4727,
+      "step": 45496
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.7220330238342285,
+      "learning_rate": 1.6613674419319904e-06,
+      "loss": 0.7245,
+      "step": 45497
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.3953118324279785,
+      "learning_rate": 1.661171177664837e-06,
+      "loss": 0.5009,
+      "step": 45498
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.1775925159454346,
+      "learning_rate": 1.6609749226817173e-06,
+      "loss": 0.6129,
+      "step": 45499
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.577500343322754,
+      "learning_rate": 1.660778676983174e-06,
+      "loss": 0.553,
+      "step": 45500
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.6654021739959717,
+      "learning_rate": 1.6605824405697564e-06,
+      "loss": 0.7313,
+      "step": 45501
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.2316083908081055,
+      "learning_rate": 1.6603862134420056e-06,
+      "loss": 0.5568,
+      "step": 45502
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.266951084136963,
+      "learning_rate": 1.6601899956004702e-06,
+      "loss": 0.5072,
+      "step": 45503
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.6957578659057617,
+      "learning_rate": 1.6599937870456972e-06,
+      "loss": 0.7092,
+      "step": 45504
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.7099032402038574,
+      "learning_rate": 1.659797587778228e-06,
+      "loss": 0.729,
+      "step": 45505
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 1.6526150703430176,
+      "learning_rate": 1.6596013977986126e-06,
+      "loss": 0.7041,
+      "step": 45506
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.500455141067505,
+      "learning_rate": 1.6594052171073927e-06,
+      "loss": 0.7287,
+      "step": 45507
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.1367459297180176,
+      "learning_rate": 1.6592090457051168e-06,
+      "loss": 0.6188,
+      "step": 45508
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.0806519985198975,
+      "learning_rate": 1.6590128835923276e-06,
+      "loss": 0.49,
+      "step": 45509
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.1528494358062744,
+      "learning_rate": 1.6588167307695718e-06,
+      "loss": 0.6254,
+      "step": 45510
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.4266042709350586,
+      "learning_rate": 1.6586205872373968e-06,
+      "loss": 0.5426,
+      "step": 45511
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.9611189365386963,
+      "learning_rate": 1.6584244529963444e-06,
+      "loss": 0.5883,
+      "step": 45512
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.518259286880493,
+      "learning_rate": 1.658228328046964e-06,
+      "loss": 0.6358,
+      "step": 45513
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.9019274711608887,
+      "learning_rate": 1.6580322123897967e-06,
+      "loss": 0.69,
+      "step": 45514
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.5253050327301025,
+      "learning_rate": 1.65783610602539e-06,
+      "loss": 0.754,
+      "step": 45515
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 1.7830922603607178,
+      "learning_rate": 1.657640008954291e-06,
+      "loss": 0.5353,
+      "step": 45516
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.00346302986145,
+      "learning_rate": 1.657443921177041e-06,
+      "loss": 0.5445,
+      "step": 45517
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.678952217102051,
+      "learning_rate": 1.6572478426941891e-06,
+      "loss": 0.6564,
+      "step": 45518
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.124228000640869,
+      "learning_rate": 1.657051773506277e-06,
+      "loss": 0.6277,
+      "step": 45519
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.267272710800171,
+      "learning_rate": 1.6568557136138536e-06,
+      "loss": 0.883,
+      "step": 45520
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.331716537475586,
+      "learning_rate": 1.6566596630174604e-06,
+      "loss": 0.474,
+      "step": 45521
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.804227590560913,
+      "learning_rate": 1.6564636217176455e-06,
+      "loss": 0.5867,
+      "step": 45522
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.86019229888916,
+      "learning_rate": 1.656267589714951e-06,
+      "loss": 0.4664,
+      "step": 45523
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.53108549118042,
+      "learning_rate": 1.656071567009926e-06,
+      "loss": 0.7154,
+      "step": 45524
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.5208873748779297,
+      "learning_rate": 1.6558755536031112e-06,
+      "loss": 0.5204,
+      "step": 45525
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.142416477203369,
+      "learning_rate": 1.6556795494950555e-06,
+      "loss": 0.5191,
+      "step": 45526
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 5.419830322265625,
+      "learning_rate": 1.6554835546863001e-06,
+      "loss": 0.7088,
+      "step": 45527
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.0526652336120605,
+      "learning_rate": 1.6552875691773945e-06,
+      "loss": 0.5268,
+      "step": 45528
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.5301589965820312,
+      "learning_rate": 1.6550915929688793e-06,
+      "loss": 0.6186,
+      "step": 45529
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.3159732818603516,
+      "learning_rate": 1.6548956260613009e-06,
+      "loss": 0.7302,
+      "step": 45530
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.6988747119903564,
+      "learning_rate": 1.6546996684552053e-06,
+      "loss": 0.6734,
+      "step": 45531
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.4438140392303467,
+      "learning_rate": 1.6545037201511381e-06,
+      "loss": 0.5948,
+      "step": 45532
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.7810189723968506,
+      "learning_rate": 1.654307781149641e-06,
+      "loss": 0.4628,
+      "step": 45533
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.898833751678467,
+      "learning_rate": 1.6541118514512628e-06,
+      "loss": 0.7284,
+      "step": 45534
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.8672399520874023,
+      "learning_rate": 1.653915931056544e-06,
+      "loss": 0.33,
+      "step": 45535
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.560129165649414,
+      "learning_rate": 1.653720019966033e-06,
+      "loss": 0.6162,
+      "step": 45536
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.1165857315063477,
+      "learning_rate": 1.6535241181802719e-06,
+      "loss": 0.4943,
+      "step": 45537
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 4.0103983879089355,
+      "learning_rate": 1.653328225699808e-06,
+      "loss": 0.8041,
+      "step": 45538
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.8934600353240967,
+      "learning_rate": 1.6531323425251832e-06,
+      "loss": 0.7342,
+      "step": 45539
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.4346730709075928,
+      "learning_rate": 1.6529364686569448e-06,
+      "loss": 0.4166,
+      "step": 45540
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.3228704929351807,
+      "learning_rate": 1.652740604095635e-06,
+      "loss": 0.799,
+      "step": 45541
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.0197060108184814,
+      "learning_rate": 1.6525447488418012e-06,
+      "loss": 0.5438,
+      "step": 45542
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.2098655700683594,
+      "learning_rate": 1.6523489028959849e-06,
+      "loss": 0.4682,
+      "step": 45543
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.2710418701171875,
+      "learning_rate": 1.6521530662587337e-06,
+      "loss": 0.4298,
+      "step": 45544
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 4.198516845703125,
+      "learning_rate": 1.6519572389305893e-06,
+      "loss": 0.4785,
+      "step": 45545
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 10.63905143737793,
+      "learning_rate": 1.651761420912099e-06,
+      "loss": 0.6777,
+      "step": 45546
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.9135582447052,
+      "learning_rate": 1.6515656122038044e-06,
+      "loss": 0.6441,
+      "step": 45547
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 4.111457347869873,
+      "learning_rate": 1.6513698128062533e-06,
+      "loss": 0.4987,
+      "step": 45548
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.703817367553711,
+      "learning_rate": 1.6511740227199862e-06,
+      "loss": 0.7403,
+      "step": 45549
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.9138941764831543,
+      "learning_rate": 1.65097824194555e-06,
+      "loss": 0.6709,
+      "step": 45550
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.612624168395996,
+      "learning_rate": 1.6507824704834884e-06,
+      "loss": 0.5801,
+      "step": 45551
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.1118791103363037,
+      "learning_rate": 1.6505867083343484e-06,
+      "loss": 0.4369,
+      "step": 45552
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 1.7907894849777222,
+      "learning_rate": 1.6503909554986697e-06,
+      "loss": 0.4939,
+      "step": 45553
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.600501298904419,
+      "learning_rate": 1.650195211977001e-06,
+      "loss": 0.5358,
+      "step": 45554
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.6898837089538574,
+      "learning_rate": 1.649999477769883e-06,
+      "loss": 0.6867,
+      "step": 45555
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.4109363555908203,
+      "learning_rate": 1.6498037528778632e-06,
+      "loss": 0.628,
+      "step": 45556
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.8860013484954834,
+      "learning_rate": 1.6496080373014822e-06,
+      "loss": 0.7442,
+      "step": 45557
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.0339064598083496,
+      "learning_rate": 1.649412331041288e-06,
+      "loss": 0.6604,
+      "step": 45558
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.131514310836792,
+      "learning_rate": 1.649216634097821e-06,
+      "loss": 0.767,
+      "step": 45559
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.070204734802246,
+      "learning_rate": 1.649020946471629e-06,
+      "loss": 0.8869,
+      "step": 45560
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.4028351306915283,
+      "learning_rate": 1.6488252681632533e-06,
+      "loss": 0.7032,
+      "step": 45561
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.134620428085327,
+      "learning_rate": 1.6486295991732404e-06,
+      "loss": 0.6871,
+      "step": 45562
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 4.051201343536377,
+      "learning_rate": 1.6484339395021314e-06,
+      "loss": 0.9744,
+      "step": 45563
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.508324146270752,
+      "learning_rate": 1.648238289150474e-06,
+      "loss": 0.8224,
+      "step": 45564
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.108142375946045,
+      "learning_rate": 1.6480426481188083e-06,
+      "loss": 0.71,
+      "step": 45565
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.3530993461608887,
+      "learning_rate": 1.6478470164076803e-06,
+      "loss": 0.6004,
+      "step": 45566
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.6637728214263916,
+      "learning_rate": 1.6476513940176365e-06,
+      "loss": 0.4437,
+      "step": 45567
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.13826322555542,
+      "learning_rate": 1.6474557809492158e-06,
+      "loss": 0.6134,
+      "step": 45568
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.6376070976257324,
+      "learning_rate": 1.6472601772029666e-06,
+      "loss": 0.3339,
+      "step": 45569
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.0560922622680664,
+      "learning_rate": 1.6470645827794297e-06,
+      "loss": 0.6947,
+      "step": 45570
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.3205440044403076,
+      "learning_rate": 1.6468689976791497e-06,
+      "loss": 0.5099,
+      "step": 45571
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 4.469331741333008,
+      "learning_rate": 1.6466734219026725e-06,
+      "loss": 0.7168,
+      "step": 45572
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.8293497562408447,
+      "learning_rate": 1.6464778554505384e-06,
+      "loss": 0.713,
+      "step": 45573
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.0655717849731445,
+      "learning_rate": 1.646282298323295e-06,
+      "loss": 0.7474,
+      "step": 45574
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 4.3210978507995605,
+      "learning_rate": 1.6460867505214828e-06,
+      "loss": 0.521,
+      "step": 45575
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 1.769047498703003,
+      "learning_rate": 1.6458912120456488e-06,
+      "loss": 0.3784,
+      "step": 45576
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.4435391426086426,
+      "learning_rate": 1.6456956828963328e-06,
+      "loss": 0.5915,
+      "step": 45577
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.2179977893829346,
+      "learning_rate": 1.6455001630740809e-06,
+      "loss": 0.4249,
+      "step": 45578
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 1.8721987009048462,
+      "learning_rate": 1.6453046525794375e-06,
+      "loss": 0.5377,
+      "step": 45579
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.2043285369873047,
+      "learning_rate": 1.6451091514129442e-06,
+      "loss": 0.6437,
+      "step": 45580
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.0923125743865967,
+      "learning_rate": 1.6449136595751464e-06,
+      "loss": 0.7481,
+      "step": 45581
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.556469678878784,
+      "learning_rate": 1.644718177066586e-06,
+      "loss": 0.6776,
+      "step": 45582
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.4037024974823,
+      "learning_rate": 1.6445227038878087e-06,
+      "loss": 0.5991,
+      "step": 45583
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 1.9702903032302856,
+      "learning_rate": 1.644327240039355e-06,
+      "loss": 0.5645,
+      "step": 45584
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.1002326011657715,
+      "learning_rate": 1.644131785521772e-06,
+      "loss": 0.6253,
+      "step": 45585
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.5657429695129395,
+      "learning_rate": 1.6439363403356001e-06,
+      "loss": 0.6178,
+      "step": 45586
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.1386849880218506,
+      "learning_rate": 1.643740904481385e-06,
+      "loss": 0.4414,
+      "step": 45587
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.5243964195251465,
+      "learning_rate": 1.6435454779596677e-06,
+      "loss": 0.4564,
+      "step": 45588
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 1.9811406135559082,
+      "learning_rate": 1.6433500607709951e-06,
+      "loss": 0.2993,
+      "step": 45589
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.5976905822753906,
+      "learning_rate": 1.6431546529159065e-06,
+      "loss": 0.5521,
+      "step": 45590
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.0968217849731445,
+      "learning_rate": 1.6429592543949473e-06,
+      "loss": 0.6438,
+      "step": 45591
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.1659977436065674,
+      "learning_rate": 1.6427638652086613e-06,
+      "loss": 0.4114,
+      "step": 45592
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.7769317626953125,
+      "learning_rate": 1.6425684853575924e-06,
+      "loss": 0.582,
+      "step": 45593
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.6200599670410156,
+      "learning_rate": 1.6423731148422818e-06,
+      "loss": 0.7688,
+      "step": 45594
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.1409730911254883,
+      "learning_rate": 1.6421777536632749e-06,
+      "loss": 0.7786,
+      "step": 45595
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.409467935562134,
+      "learning_rate": 1.6419824018211122e-06,
+      "loss": 0.6674,
+      "step": 45596
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.6521449089050293,
+      "learning_rate": 1.6417870593163404e-06,
+      "loss": 0.5637,
+      "step": 45597
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.7951815128326416,
+      "learning_rate": 1.641591726149499e-06,
+      "loss": 0.611,
+      "step": 45598
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.4155354499816895,
+      "learning_rate": 1.641396402321135e-06,
+      "loss": 0.7948,
+      "step": 45599
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.7613627910614014,
+      "learning_rate": 1.6412010878317869e-06,
+      "loss": 0.6231,
+      "step": 45600
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.173856735229492,
+      "learning_rate": 1.6410057826820026e-06,
+      "loss": 0.5002,
+      "step": 45601
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.1894421577453613,
+      "learning_rate": 1.6408104868723213e-06,
+      "loss": 0.6258,
+      "step": 45602
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.493231773376465,
+      "learning_rate": 1.6406152004032888e-06,
+      "loss": 0.7347,
+      "step": 45603
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.2811052799224854,
+      "learning_rate": 1.640419923275446e-06,
+      "loss": 0.6493,
+      "step": 45604
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.854837417602539,
+      "learning_rate": 1.640224655489338e-06,
+      "loss": 0.7754,
+      "step": 45605
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.113243341445923,
+      "learning_rate": 1.640029397045505e-06,
+      "loss": 0.5816,
+      "step": 45606
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.724078416824341,
+      "learning_rate": 1.6398341479444935e-06,
+      "loss": 0.5216,
+      "step": 45607
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.2402915954589844,
+      "learning_rate": 1.6396389081868424e-06,
+      "loss": 0.5839,
+      "step": 45608
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.302961587905884,
+      "learning_rate": 1.6394436777730987e-06,
+      "loss": 0.5713,
+      "step": 45609
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.83951735496521,
+      "learning_rate": 1.6392484567038013e-06,
+      "loss": 0.673,
+      "step": 45610
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.9182937145233154,
+      "learning_rate": 1.6390532449794955e-06,
+      "loss": 0.7061,
+      "step": 45611
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.7443814277648926,
+      "learning_rate": 1.6388580426007233e-06,
+      "loss": 0.6769,
+      "step": 45612
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.675149440765381,
+      "learning_rate": 1.6386628495680295e-06,
+      "loss": 0.7118,
+      "step": 45613
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.1690125465393066,
+      "learning_rate": 1.6384676658819531e-06,
+      "loss": 0.617,
+      "step": 45614
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.016720771789551,
+      "learning_rate": 1.638272491543041e-06,
+      "loss": 0.5149,
+      "step": 45615
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.8515028953552246,
+      "learning_rate": 1.6380773265518318e-06,
+      "loss": 0.5308,
+      "step": 45616
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.4532222747802734,
+      "learning_rate": 1.6378821709088722e-06,
+      "loss": 0.7316,
+      "step": 45617
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.6001689434051514,
+      "learning_rate": 1.6376870246147009e-06,
+      "loss": 0.3919,
+      "step": 45618
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.8709394931793213,
+      "learning_rate": 1.6374918876698636e-06,
+      "loss": 0.6159,
+      "step": 45619
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.1564412117004395,
+      "learning_rate": 1.6372967600749007e-06,
+      "loss": 0.6331,
+      "step": 45620
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 4.707334995269775,
+      "learning_rate": 1.637101641830357e-06,
+      "loss": 0.7309,
+      "step": 45621
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.3625457286834717,
+      "learning_rate": 1.6369065329367728e-06,
+      "loss": 0.7007,
+      "step": 45622
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.3802638053894043,
+      "learning_rate": 1.6367114333946932e-06,
+      "loss": 0.4594,
+      "step": 45623
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 4.221340656280518,
+      "learning_rate": 1.6365163432046572e-06,
+      "loss": 0.6236,
+      "step": 45624
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 1.948193907737732,
+      "learning_rate": 1.636321262367211e-06,
+      "loss": 0.6104,
+      "step": 45625
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.684410333633423,
+      "learning_rate": 1.636126190882894e-06,
+      "loss": 0.5486,
+      "step": 45626
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.004779100418091,
+      "learning_rate": 1.6359311287522495e-06,
+      "loss": 0.4754,
+      "step": 45627
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 5.362265110015869,
+      "learning_rate": 1.6357360759758224e-06,
+      "loss": 0.8183,
+      "step": 45628
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.831134080886841,
+      "learning_rate": 1.635541032554151e-06,
+      "loss": 0.5119,
+      "step": 45629
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.0042619705200195,
+      "learning_rate": 1.635345998487781e-06,
+      "loss": 0.4907,
+      "step": 45630
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.3717358112335205,
+      "learning_rate": 1.6351509737772524e-06,
+      "loss": 0.763,
+      "step": 45631
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.16408371925354,
+      "learning_rate": 1.634955958423108e-06,
+      "loss": 0.2943,
+      "step": 45632
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.79335355758667,
+      "learning_rate": 1.6347609524258922e-06,
+      "loss": 0.4075,
+      "step": 45633
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.9007785320281982,
+      "learning_rate": 1.6345659557861442e-06,
+      "loss": 0.6016,
+      "step": 45634
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.365952968597412,
+      "learning_rate": 1.634370968504409e-06,
+      "loss": 0.7687,
+      "step": 45635
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 1.7088967561721802,
+      "learning_rate": 1.6341759905812254e-06,
+      "loss": 0.3401,
+      "step": 45636
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.618222951889038,
+      "learning_rate": 1.6339810220171392e-06,
+      "loss": 0.5971,
+      "step": 45637
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.2328600883483887,
+      "learning_rate": 1.6337860628126895e-06,
+      "loss": 0.5857,
+      "step": 45638
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.2603652477264404,
+      "learning_rate": 1.6335911129684196e-06,
+      "loss": 0.6831,
+      "step": 45639
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.4910330772399902,
+      "learning_rate": 1.6333961724848734e-06,
+      "loss": 0.5319,
+      "step": 45640
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.1852986812591553,
+      "learning_rate": 1.6332012413625896e-06,
+      "loss": 0.5532,
+      "step": 45641
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.1363775730133057,
+      "learning_rate": 1.633006319602114e-06,
+      "loss": 0.4194,
+      "step": 45642
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.5174572467803955,
+      "learning_rate": 1.6328114072039846e-06,
+      "loss": 0.7476,
+      "step": 45643
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.420560598373413,
+      "learning_rate": 1.632616504168747e-06,
+      "loss": 0.383,
+      "step": 45644
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.0268051624298096,
+      "learning_rate": 1.6324216104969393e-06,
+      "loss": 0.4956,
+      "step": 45645
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.020386219024658,
+      "learning_rate": 1.632226726189108e-06,
+      "loss": 0.5347,
+      "step": 45646
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.975843667984009,
+      "learning_rate": 1.63203185124579e-06,
+      "loss": 0.646,
+      "step": 45647
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.8337957859039307,
+      "learning_rate": 1.631836985667532e-06,
+      "loss": 0.5458,
+      "step": 45648
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.878939151763916,
+      "learning_rate": 1.6316421294548717e-06,
+      "loss": 0.6355,
+      "step": 45649
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.0157032012939453,
+      "learning_rate": 1.6314472826083549e-06,
+      "loss": 0.5401,
+      "step": 45650
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.2354557514190674,
+      "learning_rate": 1.6312524451285194e-06,
+      "loss": 0.5523,
+      "step": 45651
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.6434216499328613,
+      "learning_rate": 1.6310576170159088e-06,
+      "loss": 0.6561,
+      "step": 45652
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.5686631202697754,
+      "learning_rate": 1.630862798271065e-06,
+      "loss": 0.6034,
+      "step": 45653
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.9109959602355957,
+      "learning_rate": 1.6306679888945314e-06,
+      "loss": 0.5503,
+      "step": 45654
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 1.9502545595169067,
+      "learning_rate": 1.6304731888868463e-06,
+      "loss": 0.5123,
+      "step": 45655
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.518143653869629,
+      "learning_rate": 1.6302783982485543e-06,
+      "loss": 0.4906,
+      "step": 45656
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 6.673252582550049,
+      "learning_rate": 1.6300836169801943e-06,
+      "loss": 1.0662,
+      "step": 45657
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.0725595951080322,
+      "learning_rate": 1.6298888450823114e-06,
+      "loss": 0.6198,
+      "step": 45658
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.996145248413086,
+      "learning_rate": 1.629694082555443e-06,
+      "loss": 0.531,
+      "step": 45659
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 1.8165838718414307,
+      "learning_rate": 1.6294993294001348e-06,
+      "loss": 0.5675,
+      "step": 45660
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.322411298751831,
+      "learning_rate": 1.6293045856169242e-06,
+      "loss": 0.7077,
+      "step": 45661
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.1171340942382812,
+      "learning_rate": 1.629109851206357e-06,
+      "loss": 0.5622,
+      "step": 45662
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.877923011779785,
+      "learning_rate": 1.6289151261689706e-06,
+      "loss": 0.6749,
+      "step": 45663
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.4467856884002686,
+      "learning_rate": 1.6287204105053105e-06,
+      "loss": 0.6268,
+      "step": 45664
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.4169890880584717,
+      "learning_rate": 1.6285257042159142e-06,
+      "loss": 0.6747,
+      "step": 45665
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.5607264041900635,
+      "learning_rate": 1.6283310073013263e-06,
+      "loss": 0.5879,
+      "step": 45666
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.9289231300354004,
+      "learning_rate": 1.6281363197620853e-06,
+      "loss": 0.4822,
+      "step": 45667
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.352288246154785,
+      "learning_rate": 1.6279416415987353e-06,
+      "loss": 0.5638,
+      "step": 45668
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.408795118331909,
+      "learning_rate": 1.6277469728118157e-06,
+      "loss": 0.4335,
+      "step": 45669
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 5.203841209411621,
+      "learning_rate": 1.6275523134018695e-06,
+      "loss": 0.6747,
+      "step": 45670
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.4551756381988525,
+      "learning_rate": 1.6273576633694355e-06,
+      "loss": 0.5185,
+      "step": 45671
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.0675885677337646,
+      "learning_rate": 1.6271630227150559e-06,
+      "loss": 0.8052,
+      "step": 45672
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 5.393537998199463,
+      "learning_rate": 1.626968391439273e-06,
+      "loss": 0.4171,
+      "step": 45673
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.4537572860717773,
+      "learning_rate": 1.6267737695426289e-06,
+      "loss": 0.4128,
+      "step": 45674
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.336538314819336,
+      "learning_rate": 1.6265791570256616e-06,
+      "loss": 0.5166,
+      "step": 45675
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 1.9452868700027466,
+      "learning_rate": 1.6263845538889162e-06,
+      "loss": 0.5441,
+      "step": 45676
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.7113983631134033,
+      "learning_rate": 1.6261899601329295e-06,
+      "loss": 0.7247,
+      "step": 45677
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.548217296600342,
+      "learning_rate": 1.625995375758246e-06,
+      "loss": 0.7021,
+      "step": 45678
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.5935943126678467,
+      "learning_rate": 1.625800800765404e-06,
+      "loss": 0.6978,
+      "step": 45679
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.2750515937805176,
+      "learning_rate": 1.625606235154948e-06,
+      "loss": 0.636,
+      "step": 45680
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.7907462120056152,
+      "learning_rate": 1.6254116789274148e-06,
+      "loss": 0.3591,
+      "step": 45681
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 4.190042972564697,
+      "learning_rate": 1.6252171320833493e-06,
+      "loss": 0.4951,
+      "step": 45682
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.3896639347076416,
+      "learning_rate": 1.6250225946232895e-06,
+      "loss": 0.5905,
+      "step": 45683
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.692704200744629,
+      "learning_rate": 1.624828066547779e-06,
+      "loss": 0.6722,
+      "step": 45684
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 1.933980941772461,
+      "learning_rate": 1.624633547857355e-06,
+      "loss": 0.547,
+      "step": 45685
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.027712345123291,
+      "learning_rate": 1.6244390385525632e-06,
+      "loss": 0.4063,
+      "step": 45686
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.757265567779541,
+      "learning_rate": 1.6242445386339394e-06,
+      "loss": 0.8388,
+      "step": 45687
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.2036874294281006,
+      "learning_rate": 1.6240500481020278e-06,
+      "loss": 0.4736,
+      "step": 45688
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.387103319168091,
+      "learning_rate": 1.6238555669573698e-06,
+      "loss": 0.74,
+      "step": 45689
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.273736000061035,
+      "learning_rate": 1.6236610952005027e-06,
+      "loss": 0.546,
+      "step": 45690
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.1030170917510986,
+      "learning_rate": 1.6234666328319715e-06,
+      "loss": 0.2654,
+      "step": 45691
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.925766944885254,
+      "learning_rate": 1.6232721798523127e-06,
+      "loss": 0.6296,
+      "step": 45692
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 1.664108395576477,
+      "learning_rate": 1.6230777362620691e-06,
+      "loss": 0.4196,
+      "step": 45693
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.084494113922119,
+      "learning_rate": 1.622883302061783e-06,
+      "loss": 0.6514,
+      "step": 45694
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.2351949214935303,
+      "learning_rate": 1.6226888772519916e-06,
+      "loss": 0.4114,
+      "step": 45695
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.666767120361328,
+      "learning_rate": 1.6224944618332389e-06,
+      "loss": 0.8045,
+      "step": 45696
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.2461588382720947,
+      "learning_rate": 1.622300055806062e-06,
+      "loss": 0.4567,
+      "step": 45697
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.8992221355438232,
+      "learning_rate": 1.6221056591710056e-06,
+      "loss": 0.7403,
+      "step": 45698
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.1061086654663086,
+      "learning_rate": 1.6219112719286056e-06,
+      "loss": 0.6027,
+      "step": 45699
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.675851345062256,
+      "learning_rate": 1.6217168940794053e-06,
+      "loss": 0.4567,
+      "step": 45700
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.7226572036743164,
+      "learning_rate": 1.6215225256239464e-06,
+      "loss": 0.6046,
+      "step": 45701
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 1.8988012075424194,
+      "learning_rate": 1.621328166562766e-06,
+      "loss": 0.5314,
+      "step": 45702
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.796133279800415,
+      "learning_rate": 1.621133816896408e-06,
+      "loss": 0.7005,
+      "step": 45703
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.4332449436187744,
+      "learning_rate": 1.6209394766254094e-06,
+      "loss": 0.5503,
+      "step": 45704
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.4043240547180176,
+      "learning_rate": 1.6207451457503142e-06,
+      "loss": 0.5737,
+      "step": 45705
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.0796079635620117,
+      "learning_rate": 1.6205508242716588e-06,
+      "loss": 0.6989,
+      "step": 45706
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.3059446811676025,
+      "learning_rate": 1.6203565121899878e-06,
+      "loss": 0.4332,
+      "step": 45707
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.73634934425354,
+      "learning_rate": 1.6201622095058373e-06,
+      "loss": 0.5338,
+      "step": 45708
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.000072717666626,
+      "learning_rate": 1.6199679162197513e-06,
+      "loss": 0.7579,
+      "step": 45709
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.571709156036377,
+      "learning_rate": 1.6197736323322667e-06,
+      "loss": 0.6972,
+      "step": 45710
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.0003862380981445,
+      "learning_rate": 1.6195793578439268e-06,
+      "loss": 0.7235,
+      "step": 45711
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.8025553226470947,
+      "learning_rate": 1.6193850927552684e-06,
+      "loss": 0.4674,
+      "step": 45712
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.8508152961730957,
+      "learning_rate": 1.619190837066834e-06,
+      "loss": 0.679,
+      "step": 45713
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.2246854305267334,
+      "learning_rate": 1.6189965907791638e-06,
+      "loss": 0.5127,
+      "step": 45714
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.5460128784179688,
+      "learning_rate": 1.6188023538927988e-06,
+      "loss": 0.59,
+      "step": 45715
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.659006118774414,
+      "learning_rate": 1.6186081264082765e-06,
+      "loss": 0.4272,
+      "step": 45716
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.845515251159668,
+      "learning_rate": 1.6184139083261398e-06,
+      "loss": 0.4209,
+      "step": 45717
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.910095453262329,
+      "learning_rate": 1.6182196996469252e-06,
+      "loss": 0.5515,
+      "step": 45718
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 1.9097048044204712,
+      "learning_rate": 1.618025500371177e-06,
+      "loss": 0.5544,
+      "step": 45719
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.16939640045166,
+      "learning_rate": 1.6178313104994308e-06,
+      "loss": 0.6367,
+      "step": 45720
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.064058780670166,
+      "learning_rate": 1.6176371300322309e-06,
+      "loss": 0.6733,
+      "step": 45721
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.0606865882873535,
+      "learning_rate": 1.617442958970113e-06,
+      "loss": 0.8294,
+      "step": 45722
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.7658863067626953,
+      "learning_rate": 1.6172487973136208e-06,
+      "loss": 0.5222,
+      "step": 45723
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.662471294403076,
+      "learning_rate": 1.6170546450632912e-06,
+      "loss": 0.7597,
+      "step": 45724
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.158406972885132,
+      "learning_rate": 1.6168605022196666e-06,
+      "loss": 0.5589,
+      "step": 45725
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.5035619735717773,
+      "learning_rate": 1.616666368783284e-06,
+      "loss": 0.6454,
+      "step": 45726
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.523057460784912,
+      "learning_rate": 1.6164722447546865e-06,
+      "loss": 0.6788,
+      "step": 45727
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.4266128540039062,
+      "learning_rate": 1.6162781301344105e-06,
+      "loss": 0.7335,
+      "step": 45728
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 1.8354344367980957,
+      "learning_rate": 1.6160840249229992e-06,
+      "loss": 0.6374,
+      "step": 45729
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.093902111053467,
+      "learning_rate": 1.615889929120989e-06,
+      "loss": 0.3739,
+      "step": 45730
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.803656578063965,
+      "learning_rate": 1.6156958427289227e-06,
+      "loss": 0.6623,
+      "step": 45731
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.180771827697754,
+      "learning_rate": 1.6155017657473371e-06,
+      "loss": 0.5783,
+      "step": 45732
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.910834312438965,
+      "learning_rate": 1.615307698176773e-06,
+      "loss": 0.3909,
+      "step": 45733
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.3728272914886475,
+      "learning_rate": 1.6151136400177703e-06,
+      "loss": 0.637,
+      "step": 45734
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.2626941204071045,
+      "learning_rate": 1.61491959127087e-06,
+      "loss": 0.5673,
+      "step": 45735
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 1.9162938594818115,
+      "learning_rate": 1.6147255519366084e-06,
+      "loss": 0.6226,
+      "step": 45736
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.4260125160217285,
+      "learning_rate": 1.6145315220155288e-06,
+      "loss": 0.7599,
+      "step": 45737
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.6199660301208496,
+      "learning_rate": 1.6143375015081674e-06,
+      "loss": 0.5896,
+      "step": 45738
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.1680526733398438,
+      "learning_rate": 1.6141434904150666e-06,
+      "loss": 0.5835,
+      "step": 45739
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.532957077026367,
+      "learning_rate": 1.6139494887367629e-06,
+      "loss": 0.5847,
+      "step": 45740
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.2956459522247314,
+      "learning_rate": 1.613755496473799e-06,
+      "loss": 0.6425,
+      "step": 45741
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.848483085632324,
+      "learning_rate": 1.6135615136267103e-06,
+      "loss": 0.8663,
+      "step": 45742
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.203137159347534,
+      "learning_rate": 1.6133675401960403e-06,
+      "loss": 0.6772,
+      "step": 45743
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.2539658546447754,
+      "learning_rate": 1.6131735761823253e-06,
+      "loss": 0.438,
+      "step": 45744
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 4.177491664886475,
+      "learning_rate": 1.6129796215861071e-06,
+      "loss": 0.7297,
+      "step": 45745
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.551687002182007,
+      "learning_rate": 1.612785676407922e-06,
+      "loss": 0.7296,
+      "step": 45746
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.6041808128356934,
+      "learning_rate": 1.6125917406483128e-06,
+      "loss": 0.4844,
+      "step": 45747
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.2699663639068604,
+      "learning_rate": 1.6123978143078157e-06,
+      "loss": 0.4672,
+      "step": 45748
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.4015846252441406,
+      "learning_rate": 1.6122038973869726e-06,
+      "loss": 0.4456,
+      "step": 45749
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.937046527862549,
+      "learning_rate": 1.6120099898863195e-06,
+      "loss": 0.6237,
+      "step": 45750
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.576040744781494,
+      "learning_rate": 1.611816091806398e-06,
+      "loss": 0.7606,
+      "step": 45751
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.2951133251190186,
+      "learning_rate": 1.6116222031477479e-06,
+      "loss": 0.3736,
+      "step": 45752
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.33088755607605,
+      "learning_rate": 1.6114283239109057e-06,
+      "loss": 0.5464,
+      "step": 45753
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.3829185962677,
+      "learning_rate": 1.611234454096412e-06,
+      "loss": 0.6808,
+      "step": 45754
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.2332100868225098,
+      "learning_rate": 1.6110405937048073e-06,
+      "loss": 0.7829,
+      "step": 45755
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.3952274322509766,
+      "learning_rate": 1.6108467427366275e-06,
+      "loss": 0.5162,
+      "step": 45756
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.140254020690918,
+      "learning_rate": 1.6106529011924154e-06,
+      "loss": 0.6013,
+      "step": 45757
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.6156580448150635,
+      "learning_rate": 1.6104590690727056e-06,
+      "loss": 0.6848,
+      "step": 45758
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.1516029834747314,
+      "learning_rate": 1.6102652463780416e-06,
+      "loss": 0.6912,
+      "step": 45759
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.676719903945923,
+      "learning_rate": 1.610071433108958e-06,
+      "loss": 0.6203,
+      "step": 45760
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.8573367595672607,
+      "learning_rate": 1.6098776292659979e-06,
+      "loss": 0.5773,
+      "step": 45761
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.0535593032836914,
+      "learning_rate": 1.6096838348496963e-06,
+      "loss": 0.5694,
+      "step": 45762
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.029405355453491,
+      "learning_rate": 1.6094900498605937e-06,
+      "loss": 0.543,
+      "step": 45763
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.8410253524780273,
+      "learning_rate": 1.6092962742992314e-06,
+      "loss": 0.8153,
+      "step": 45764
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.6740846633911133,
+      "learning_rate": 1.6091025081661438e-06,
+      "loss": 0.8042,
+      "step": 45765
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.0188584327697754,
+      "learning_rate": 1.6089087514618733e-06,
+      "loss": 1.0307,
+      "step": 45766
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.809481620788574,
+      "learning_rate": 1.6087150041869564e-06,
+      "loss": 0.3727,
+      "step": 45767
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 1.9875648021697998,
+      "learning_rate": 1.6085212663419336e-06,
+      "loss": 0.5396,
+      "step": 45768
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.354353427886963,
+      "learning_rate": 1.6083275379273416e-06,
+      "loss": 0.5405,
+      "step": 45769
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.6607747077941895,
+      "learning_rate": 1.6081338189437213e-06,
+      "loss": 0.5542,
+      "step": 45770
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.1285452842712402,
+      "learning_rate": 1.6079401093916085e-06,
+      "loss": 0.7717,
+      "step": 45771
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.283229351043701,
+      "learning_rate": 1.6077464092715456e-06,
+      "loss": 0.7266,
+      "step": 45772
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.0632474422454834,
+      "learning_rate": 1.6075527185840678e-06,
+      "loss": 0.6609,
+      "step": 45773
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.868767261505127,
+      "learning_rate": 1.6073590373297144e-06,
+      "loss": 0.6633,
+      "step": 45774
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.9493794441223145,
+      "learning_rate": 1.6071653655090252e-06,
+      "loss": 0.7351,
+      "step": 45775
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.577772855758667,
+      "learning_rate": 1.606971703122539e-06,
+      "loss": 0.5925,
+      "step": 45776
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.7194108963012695,
+      "learning_rate": 1.6067780501707926e-06,
+      "loss": 0.7729,
+      "step": 45777
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.041964292526245,
+      "learning_rate": 1.6065844066543261e-06,
+      "loss": 0.6945,
+      "step": 45778
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.329477310180664,
+      "learning_rate": 1.6063907725736765e-06,
+      "loss": 0.5377,
+      "step": 45779
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.48442006111145,
+      "learning_rate": 1.6061971479293841e-06,
+      "loss": 0.5362,
+      "step": 45780
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.048569917678833,
+      "learning_rate": 1.6060035327219843e-06,
+      "loss": 0.5273,
+      "step": 45781
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.4390761852264404,
+      "learning_rate": 1.6058099269520188e-06,
+      "loss": 0.8561,
+      "step": 45782
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 1.8887931108474731,
+      "learning_rate": 1.605616330620023e-06,
+      "loss": 0.5153,
+      "step": 45783
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.090930938720703,
+      "learning_rate": 1.6054227437265384e-06,
+      "loss": 0.4712,
+      "step": 45784
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.138152837753296,
+      "learning_rate": 1.6052291662720998e-06,
+      "loss": 0.7064,
+      "step": 45785
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.5401244163513184,
+      "learning_rate": 1.6050355982572486e-06,
+      "loss": 0.7385,
+      "step": 45786
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.4282758235931396,
+      "learning_rate": 1.6048420396825199e-06,
+      "loss": 0.3161,
+      "step": 45787
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.5109972953796387,
+      "learning_rate": 1.6046484905484555e-06,
+      "loss": 0.794,
+      "step": 45788
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.6884374618530273,
+      "learning_rate": 1.6044549508555902e-06,
+      "loss": 0.6755,
+      "step": 45789
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 1.515094518661499,
+      "learning_rate": 1.6042614206044648e-06,
+      "loss": 0.4751,
+      "step": 45790
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.1577999591827393,
+      "learning_rate": 1.6040678997956154e-06,
+      "loss": 0.5841,
+      "step": 45791
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.5461647510528564,
+      "learning_rate": 1.6038743884295821e-06,
+      "loss": 0.5888,
+      "step": 45792
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.4718496799468994,
+      "learning_rate": 1.6036808865069004e-06,
+      "loss": 0.8189,
+      "step": 45793
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 1.981400489807129,
+      "learning_rate": 1.6034873940281098e-06,
+      "loss": 0.7632,
+      "step": 45794
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.3157260417938232,
+      "learning_rate": 1.603293910993749e-06,
+      "loss": 0.5882,
+      "step": 45795
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.857106924057007,
+      "learning_rate": 1.6031004374043563e-06,
+      "loss": 0.5427,
+      "step": 45796
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.791037082672119,
+      "learning_rate": 1.6029069732604674e-06,
+      "loss": 0.6551,
+      "step": 45797
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.3445534706115723,
+      "learning_rate": 1.6027135185626237e-06,
+      "loss": 0.4318,
+      "step": 45798
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 1.9320169687271118,
+      "learning_rate": 1.602520073311359e-06,
+      "loss": 0.3273,
+      "step": 45799
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.289376974105835,
+      "learning_rate": 1.6023266375072144e-06,
+      "loss": 0.6889,
+      "step": 45800
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.5692648887634277,
+      "learning_rate": 1.6021332111507255e-06,
+      "loss": 0.4991,
+      "step": 45801
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 4.170596122741699,
+      "learning_rate": 1.6019397942424331e-06,
+      "loss": 0.5689,
+      "step": 45802
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.891449213027954,
+      "learning_rate": 1.6017463867828708e-06,
+      "loss": 0.4875,
+      "step": 45803
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.1891822814941406,
+      "learning_rate": 1.601552988772581e-06,
+      "loss": 0.5455,
+      "step": 45804
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.287539005279541,
+      "learning_rate": 1.601359600212098e-06,
+      "loss": 0.6819,
+      "step": 45805
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.2725703716278076,
+      "learning_rate": 1.6011662211019618e-06,
+      "loss": 0.7302,
+      "step": 45806
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.812899112701416,
+      "learning_rate": 1.6009728514427076e-06,
+      "loss": 0.4112,
+      "step": 45807
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.795436382293701,
+      "learning_rate": 1.6007794912348761e-06,
+      "loss": 0.4595,
+      "step": 45808
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.213879108428955,
+      "learning_rate": 1.6005861404790023e-06,
+      "loss": 0.5347,
+      "step": 45809
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.677426338195801,
+      "learning_rate": 1.6003927991756263e-06,
+      "loss": 0.5483,
+      "step": 45810
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.0555596351623535,
+      "learning_rate": 1.6001994673252825e-06,
+      "loss": 0.4898,
+      "step": 45811
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.5411646366119385,
+      "learning_rate": 1.600006144928511e-06,
+      "loss": 0.565,
+      "step": 45812
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.024212121963501,
+      "learning_rate": 1.5998128319858498e-06,
+      "loss": 0.556,
+      "step": 45813
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.1218998432159424,
+      "learning_rate": 1.5996195284978345e-06,
+      "loss": 0.7341,
+      "step": 45814
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.58953857421875,
+      "learning_rate": 1.5994262344650024e-06,
+      "loss": 0.2987,
+      "step": 45815
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 1.8860735893249512,
+      "learning_rate": 1.5992329498878945e-06,
+      "loss": 0.416,
+      "step": 45816
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.339163303375244,
+      "learning_rate": 1.599039674767044e-06,
+      "loss": 0.5901,
+      "step": 45817
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.8994221687316895,
+      "learning_rate": 1.5988464091029915e-06,
+      "loss": 0.4845,
+      "step": 45818
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.342359781265259,
+      "learning_rate": 1.5986531528962718e-06,
+      "loss": 0.9163,
+      "step": 45819
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.6543052196502686,
+      "learning_rate": 1.5984599061474249e-06,
+      "loss": 0.8828,
+      "step": 45820
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.535463571548462,
+      "learning_rate": 1.5982666688569854e-06,
+      "loss": 0.5947,
+      "step": 45821
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.4193553924560547,
+      "learning_rate": 1.5980734410254933e-06,
+      "loss": 0.7304,
+      "step": 45822
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 6.0067458152771,
+      "learning_rate": 1.5978802226534828e-06,
+      "loss": 0.3506,
+      "step": 45823
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.14857816696167,
+      "learning_rate": 1.5976870137414935e-06,
+      "loss": 0.5309,
+      "step": 45824
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.2512717247009277,
+      "learning_rate": 1.5974938142900632e-06,
+      "loss": 0.639,
+      "step": 45825
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.075267791748047,
+      "learning_rate": 1.5973006242997263e-06,
+      "loss": 0.823,
+      "step": 45826
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.183945655822754,
+      "learning_rate": 1.5971074437710237e-06,
+      "loss": 0.5195,
+      "step": 45827
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.4021079540252686,
+      "learning_rate": 1.596914272704489e-06,
+      "loss": 0.3877,
+      "step": 45828
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.379178047180176,
+      "learning_rate": 1.596721111100662e-06,
+      "loss": 0.645,
+      "step": 45829
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.590315580368042,
+      "learning_rate": 1.5965279589600774e-06,
+      "loss": 0.8041,
+      "step": 45830
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.6951260566711426,
+      "learning_rate": 1.5963348162832753e-06,
+      "loss": 0.6108,
+      "step": 45831
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.6825554370880127,
+      "learning_rate": 1.5961416830707893e-06,
+      "loss": 0.781,
+      "step": 45832
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.113631010055542,
+      "learning_rate": 1.59594855932316e-06,
+      "loss": 0.8518,
+      "step": 45833
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.5132758617401123,
+      "learning_rate": 1.5957554450409203e-06,
+      "loss": 0.5139,
+      "step": 45834
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 4.121192932128906,
+      "learning_rate": 1.59556234022461e-06,
+      "loss": 0.6279,
+      "step": 45835
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.621551990509033,
+      "learning_rate": 1.5953692448747655e-06,
+      "loss": 0.6004,
+      "step": 45836
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.8307127952575684,
+      "learning_rate": 1.595176158991925e-06,
+      "loss": 0.5525,
+      "step": 45837
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.125417947769165,
+      "learning_rate": 1.5949830825766228e-06,
+      "loss": 0.7441,
+      "step": 45838
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.634032726287842,
+      "learning_rate": 1.5947900156293983e-06,
+      "loss": 0.6135,
+      "step": 45839
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.915884017944336,
+      "learning_rate": 1.5945969581507858e-06,
+      "loss": 0.5187,
+      "step": 45840
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.26354718208313,
+      "learning_rate": 1.5944039101413255e-06,
+      "loss": 0.4499,
+      "step": 45841
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.0759572982788086,
+      "learning_rate": 1.5942108716015497e-06,
+      "loss": 0.5638,
+      "step": 45842
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.7965269088745117,
+      "learning_rate": 1.5940178425320003e-06,
+      "loss": 0.6941,
+      "step": 45843
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.6693029403686523,
+      "learning_rate": 1.5938248229332087e-06,
+      "loss": 0.483,
+      "step": 45844
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.592113494873047,
+      "learning_rate": 1.5936318128057164e-06,
+      "loss": 0.5643,
+      "step": 45845
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.7718892097473145,
+      "learning_rate": 1.5934388121500566e-06,
+      "loss": 0.5298,
+      "step": 45846
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 1.9297653436660767,
+      "learning_rate": 1.5932458209667683e-06,
+      "loss": 0.4253,
+      "step": 45847
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.07051682472229,
+      "learning_rate": 1.593052839256386e-06,
+      "loss": 0.7898,
+      "step": 45848
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.7919113636016846,
+      "learning_rate": 1.5928598670194496e-06,
+      "loss": 0.7752,
+      "step": 45849
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.180305242538452,
+      "learning_rate": 1.5926669042564913e-06,
+      "loss": 0.6157,
+      "step": 45850
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 4.334164142608643,
+      "learning_rate": 1.5924739509680515e-06,
+      "loss": 0.539,
+      "step": 45851
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.3677263259887695,
+      "learning_rate": 1.5922810071546636e-06,
+      "loss": 0.4796,
+      "step": 45852
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.796476125717163,
+      "learning_rate": 1.5920880728168675e-06,
+      "loss": 0.5259,
+      "step": 45853
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.456027030944824,
+      "learning_rate": 1.591895147955196e-06,
+      "loss": 0.6045,
+      "step": 45854
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 1.9527407884597778,
+      "learning_rate": 1.591702232570187e-06,
+      "loss": 0.4412,
+      "step": 45855
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.132981061935425,
+      "learning_rate": 1.5915093266623782e-06,
+      "loss": 0.5129,
+      "step": 45856
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.6725826263427734,
+      "learning_rate": 1.5913164302323058e-06,
+      "loss": 0.71,
+      "step": 45857
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.0515429973602295,
+      "learning_rate": 1.5911235432805043e-06,
+      "loss": 0.5145,
+      "step": 45858
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.337090015411377,
+      "learning_rate": 1.590930665807513e-06,
+      "loss": 0.6855,
+      "step": 45859
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.2663779258728027,
+      "learning_rate": 1.590737797813865e-06,
+      "loss": 0.6392,
+      "step": 45860
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.536323070526123,
+      "learning_rate": 1.5905449393000992e-06,
+      "loss": 0.532,
+      "step": 45861
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.1989457607269287,
+      "learning_rate": 1.5903520902667496e-06,
+      "loss": 0.6233,
+      "step": 45862
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.346776008605957,
+      "learning_rate": 1.5901592507143548e-06,
+      "loss": 0.5615,
+      "step": 45863
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.5716025829315186,
+      "learning_rate": 1.5899664206434483e-06,
+      "loss": 0.4528,
+      "step": 45864
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 30.374765396118164,
+      "learning_rate": 1.5897736000545694e-06,
+      "loss": 0.3987,
+      "step": 45865
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.900995969772339,
+      "learning_rate": 1.5895807889482506e-06,
+      "loss": 0.5973,
+      "step": 45866
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.711226463317871,
+      "learning_rate": 1.5893879873250317e-06,
+      "loss": 0.6191,
+      "step": 45867
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.481642723083496,
+      "learning_rate": 1.5891951951854462e-06,
+      "loss": 0.7094,
+      "step": 45868
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.633594274520874,
+      "learning_rate": 1.5890024125300324e-06,
+      "loss": 0.5995,
+      "step": 45869
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.1399099826812744,
+      "learning_rate": 1.5888096393593233e-06,
+      "loss": 0.4285,
+      "step": 45870
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.3353171348571777,
+      "learning_rate": 1.5886168756738584e-06,
+      "loss": 0.5339,
+      "step": 45871
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.0512311458587646,
+      "learning_rate": 1.5884241214741708e-06,
+      "loss": 0.4539,
+      "step": 45872
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.5590672492980957,
+      "learning_rate": 1.5882313767607971e-06,
+      "loss": 0.6319,
+      "step": 45873
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.2467193603515625,
+      "learning_rate": 1.5880386415342757e-06,
+      "loss": 0.8154,
+      "step": 45874
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 4.411537170410156,
+      "learning_rate": 1.5878459157951398e-06,
+      "loss": 0.5204,
+      "step": 45875
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.0548951625823975,
+      "learning_rate": 1.5876531995439254e-06,
+      "loss": 0.4525,
+      "step": 45876
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.0353662967681885,
+      "learning_rate": 1.587460492781171e-06,
+      "loss": 0.7225,
+      "step": 45877
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.5529959201812744,
+      "learning_rate": 1.587267795507409e-06,
+      "loss": 0.6946,
+      "step": 45878
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 1.7949830293655396,
+      "learning_rate": 1.5870751077231783e-06,
+      "loss": 0.464,
+      "step": 45879
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.9892737865448,
+      "learning_rate": 1.5868824294290113e-06,
+      "loss": 0.511,
+      "step": 45880
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.0539193153381348,
+      "learning_rate": 1.5866897606254478e-06,
+      "loss": 0.4239,
+      "step": 45881
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 4.290554523468018,
+      "learning_rate": 1.58649710131302e-06,
+      "loss": 0.569,
+      "step": 45882
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.2775416374206543,
+      "learning_rate": 1.586304451492266e-06,
+      "loss": 0.4659,
+      "step": 45883
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.8002283573150635,
+      "learning_rate": 1.586111811163719e-06,
+      "loss": 0.5026,
+      "step": 45884
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.608154535293579,
+      "learning_rate": 1.5859191803279168e-06,
+      "loss": 0.8081,
+      "step": 45885
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.3745529651641846,
+      "learning_rate": 1.5857265589853955e-06,
+      "loss": 0.6713,
+      "step": 45886
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.5502562522888184,
+      "learning_rate": 1.5855339471366881e-06,
+      "loss": 0.4642,
+      "step": 45887
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.726377487182617,
+      "learning_rate": 1.5853413447823335e-06,
+      "loss": 0.5574,
+      "step": 45888
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.685765504837036,
+      "learning_rate": 1.5851487519228636e-06,
+      "loss": 0.4991,
+      "step": 45889
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.217092752456665,
+      "learning_rate": 1.5849561685588178e-06,
+      "loss": 0.6739,
+      "step": 45890
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.8408284187316895,
+      "learning_rate": 1.5847635946907275e-06,
+      "loss": 0.6879,
+      "step": 45891
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.840676784515381,
+      "learning_rate": 1.584571030319132e-06,
+      "loss": 0.6881,
+      "step": 45892
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.4350311756134033,
+      "learning_rate": 1.5843784754445635e-06,
+      "loss": 0.7786,
+      "step": 45893
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.266740322113037,
+      "learning_rate": 1.5841859300675604e-06,
+      "loss": 0.642,
+      "step": 45894
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.9716453552246094,
+      "learning_rate": 1.5839933941886554e-06,
+      "loss": 0.7753,
+      "step": 45895
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 1.9434822797775269,
+      "learning_rate": 1.583800867808385e-06,
+      "loss": 0.4379,
+      "step": 45896
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.017021417617798,
+      "learning_rate": 1.5836083509272842e-06,
+      "loss": 0.4948,
+      "step": 45897
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.008690357208252,
+      "learning_rate": 1.5834158435458906e-06,
+      "loss": 0.5368,
+      "step": 45898
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.0078468322753906,
+      "learning_rate": 1.583223345664736e-06,
+      "loss": 0.4172,
+      "step": 45899
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.4602224826812744,
+      "learning_rate": 1.5830308572843595e-06,
+      "loss": 0.5336,
+      "step": 45900
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.860828161239624,
+      "learning_rate": 1.582838378405292e-06,
+      "loss": 0.702,
+      "step": 45901
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.5099101066589355,
+      "learning_rate": 1.5826459090280727e-06,
+      "loss": 0.5938,
+      "step": 45902
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.1195194721221924,
+      "learning_rate": 1.5824534491532328e-06,
+      "loss": 0.8759,
+      "step": 45903
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.3315908908843994,
+      "learning_rate": 1.582260998781312e-06,
+      "loss": 0.6294,
+      "step": 45904
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 4.011035919189453,
+      "learning_rate": 1.5820685579128409e-06,
+      "loss": 0.4402,
+      "step": 45905
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.6476235389709473,
+      "learning_rate": 1.581876126548359e-06,
+      "loss": 0.7424,
+      "step": 45906
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.153503894805908,
+      "learning_rate": 1.581683704688397e-06,
+      "loss": 0.8052,
+      "step": 45907
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.57922625541687,
+      "learning_rate": 1.5814912923334936e-06,
+      "loss": 0.5652,
+      "step": 45908
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.7063260078430176,
+      "learning_rate": 1.5812988894841807e-06,
+      "loss": 0.8016,
+      "step": 45909
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.4328324794769287,
+      "learning_rate": 1.5811064961409972e-06,
+      "loss": 0.5888,
+      "step": 45910
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.1513864994049072,
+      "learning_rate": 1.5809141123044735e-06,
+      "loss": 0.5483,
+      "step": 45911
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 4.8802947998046875,
+      "learning_rate": 1.5807217379751487e-06,
+      "loss": 0.7094,
+      "step": 45912
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.205348253250122,
+      "learning_rate": 1.5805293731535542e-06,
+      "loss": 0.7924,
+      "step": 45913
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.3044369220733643,
+      "learning_rate": 1.5803370178402277e-06,
+      "loss": 0.4269,
+      "step": 45914
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.284891366958618,
+      "learning_rate": 1.5801446720357022e-06,
+      "loss": 0.5833,
+      "step": 45915
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.0548999309539795,
+      "learning_rate": 1.5799523357405127e-06,
+      "loss": 0.7186,
+      "step": 45916
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 4.033734321594238,
+      "learning_rate": 1.5797600089551945e-06,
+      "loss": 0.6545,
+      "step": 45917
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.2704973220825195,
+      "learning_rate": 1.5795676916802844e-06,
+      "loss": 0.5936,
+      "step": 45918
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.8613483905792236,
+      "learning_rate": 1.5793753839163135e-06,
+      "loss": 0.5192,
+      "step": 45919
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.60353422164917,
+      "learning_rate": 1.5791830856638195e-06,
+      "loss": 0.5447,
+      "step": 45920
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.496958017349243,
+      "learning_rate": 1.5789907969233348e-06,
+      "loss": 0.5219,
+      "step": 45921
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.587681770324707,
+      "learning_rate": 1.5787985176953967e-06,
+      "loss": 0.6878,
+      "step": 45922
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.391392946243286,
+      "learning_rate": 1.5786062479805365e-06,
+      "loss": 0.5966,
+      "step": 45923
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.4786791801452637,
+      "learning_rate": 1.5784139877792926e-06,
+      "loss": 0.6541,
+      "step": 45924
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.034416437149048,
+      "learning_rate": 1.5782217370921955e-06,
+      "loss": 0.5046,
+      "step": 45925
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.079301118850708,
+      "learning_rate": 1.5780294959197839e-06,
+      "loss": 0.5251,
+      "step": 45926
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.319481134414673,
+      "learning_rate": 1.577837264262589e-06,
+      "loss": 0.6594,
+      "step": 45927
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.965174436569214,
+      "learning_rate": 1.5776450421211476e-06,
+      "loss": 0.5116,
+      "step": 45928
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.578752040863037,
+      "learning_rate": 1.5774528294959918e-06,
+      "loss": 0.2837,
+      "step": 45929
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.7016007900238037,
+      "learning_rate": 1.5772606263876595e-06,
+      "loss": 0.6165,
+      "step": 45930
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 1.5312554836273193,
+      "learning_rate": 1.5770684327966817e-06,
+      "loss": 0.5348,
+      "step": 45931
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.3009748458862305,
+      "learning_rate": 1.5768762487235956e-06,
+      "loss": 0.48,
+      "step": 45932
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.172337293624878,
+      "learning_rate": 1.5766840741689332e-06,
+      "loss": 0.6051,
+      "step": 45933
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.161691904067993,
+      "learning_rate": 1.576491909133231e-06,
+      "loss": 0.7192,
+      "step": 45934
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.379199981689453,
+      "learning_rate": 1.576299753617021e-06,
+      "loss": 0.6426,
+      "step": 45935
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.1387128829956055,
+      "learning_rate": 1.5761076076208387e-06,
+      "loss": 0.5201,
+      "step": 45936
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.424539089202881,
+      "learning_rate": 1.575915471145219e-06,
+      "loss": 0.544,
+      "step": 45937
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 1.855141043663025,
+      "learning_rate": 1.5757233441906965e-06,
+      "loss": 0.7075,
+      "step": 45938
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 1.7879753112792969,
+      "learning_rate": 1.5755312267578037e-06,
+      "loss": 0.4958,
+      "step": 45939
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.075411081314087,
+      "learning_rate": 1.5753391188470774e-06,
+      "loss": 0.4985,
+      "step": 45940
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.582991123199463,
+      "learning_rate": 1.575147020459048e-06,
+      "loss": 0.6698,
+      "step": 45941
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.177341938018799,
+      "learning_rate": 1.5749549315942536e-06,
+      "loss": 0.7014,
+      "step": 45942
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.532028913497925,
+      "learning_rate": 1.5747628522532244e-06,
+      "loss": 0.6714,
+      "step": 45943
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 1.419345736503601,
+      "learning_rate": 1.5745707824364986e-06,
+      "loss": 0.4527,
+      "step": 45944
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.3264122009277344,
+      "learning_rate": 1.5743787221446065e-06,
+      "loss": 0.4798,
+      "step": 45945
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.0766186714172363,
+      "learning_rate": 1.5741866713780862e-06,
+      "loss": 0.7623,
+      "step": 45946
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.2437970638275146,
+      "learning_rate": 1.573994630137467e-06,
+      "loss": 0.5276,
+      "step": 45947
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.3433492183685303,
+      "learning_rate": 1.5738025984232858e-06,
+      "loss": 0.6522,
+      "step": 45948
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 1.9332624673843384,
+      "learning_rate": 1.5736105762360775e-06,
+      "loss": 0.4729,
+      "step": 45949
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.457559108734131,
+      "learning_rate": 1.5734185635763732e-06,
+      "loss": 0.6919,
+      "step": 45950
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.314423084259033,
+      "learning_rate": 1.5732265604447095e-06,
+      "loss": 0.748,
+      "step": 45951
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.671096086502075,
+      "learning_rate": 1.5730345668416175e-06,
+      "loss": 0.5755,
+      "step": 45952
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.190343141555786,
+      "learning_rate": 1.5728425827676347e-06,
+      "loss": 0.4094,
+      "step": 45953
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.4095237255096436,
+      "learning_rate": 1.5726506082232906e-06,
+      "loss": 0.5121,
+      "step": 45954
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 4.787145614624023,
+      "learning_rate": 1.5724586432091233e-06,
+      "loss": 0.6328,
+      "step": 45955
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 1.7098349332809448,
+      "learning_rate": 1.5722666877256626e-06,
+      "loss": 0.3466,
+      "step": 45956
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.4115118980407715,
+      "learning_rate": 1.5720747417734445e-06,
+      "loss": 0.5511,
+      "step": 45957
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.386000156402588,
+      "learning_rate": 1.5718828053530022e-06,
+      "loss": 0.7802,
+      "step": 45958
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.598862648010254,
+      "learning_rate": 1.571690878464871e-06,
+      "loss": 0.6173,
+      "step": 45959
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.503891706466675,
+      "learning_rate": 1.5714989611095816e-06,
+      "loss": 0.4019,
+      "step": 45960
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.054044246673584,
+      "learning_rate": 1.5713070532876712e-06,
+      "loss": 0.5665,
+      "step": 45961
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.3216910362243652,
+      "learning_rate": 1.57111515499967e-06,
+      "loss": 0.4516,
+      "step": 45962
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.674104928970337,
+      "learning_rate": 1.5709232662461143e-06,
+      "loss": 0.6441,
+      "step": 45963
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.368694305419922,
+      "learning_rate": 1.5707313870275348e-06,
+      "loss": 0.6957,
+      "step": 45964
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 4.039913177490234,
+      "learning_rate": 1.5705395173444687e-06,
+      "loss": 0.4629,
+      "step": 45965
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.6331002712249756,
+      "learning_rate": 1.5703476571974453e-06,
+      "loss": 0.6887,
+      "step": 45966
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.982278347015381,
+      "learning_rate": 1.5701558065870025e-06,
+      "loss": 0.5649,
+      "step": 45967
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.465958595275879,
+      "learning_rate": 1.5699639655136694e-06,
+      "loss": 0.7145,
+      "step": 45968
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 5.539287567138672,
+      "learning_rate": 1.5697721339779836e-06,
+      "loss": 0.601,
+      "step": 45969
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.44739031791687,
+      "learning_rate": 1.5695803119804747e-06,
+      "loss": 0.6142,
+      "step": 45970
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.19549560546875,
+      "learning_rate": 1.5693884995216796e-06,
+      "loss": 0.4356,
+      "step": 45971
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.5274038314819336,
+      "learning_rate": 1.5691966966021282e-06,
+      "loss": 0.5783,
+      "step": 45972
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.6533446311950684,
+      "learning_rate": 1.5690049032223576e-06,
+      "loss": 0.6738,
+      "step": 45973
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.484470844268799,
+      "learning_rate": 1.568813119382897e-06,
+      "loss": 0.4241,
+      "step": 45974
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.5883278846740723,
+      "learning_rate": 1.5686213450842835e-06,
+      "loss": 0.7156,
+      "step": 45975
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.2050795555114746,
+      "learning_rate": 1.568429580327047e-06,
+      "loss": 0.6165,
+      "step": 45976
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.8098111152648926,
+      "learning_rate": 1.5682378251117226e-06,
+      "loss": 0.8832,
+      "step": 45977
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.7436928749084473,
+      "learning_rate": 1.5680460794388425e-06,
+      "loss": 0.5251,
+      "step": 45978
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.4426419734954834,
+      "learning_rate": 1.5678543433089427e-06,
+      "loss": 0.5164,
+      "step": 45979
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 1.9237253665924072,
+      "learning_rate": 1.5676626167225527e-06,
+      "loss": 0.6488,
+      "step": 45980
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 1.991904377937317,
+      "learning_rate": 1.567470899680209e-06,
+      "loss": 0.2879,
+      "step": 45981
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.412055253982544,
+      "learning_rate": 1.5672791921824409e-06,
+      "loss": 0.6854,
+      "step": 45982
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 4.31644868850708,
+      "learning_rate": 1.5670874942297854e-06,
+      "loss": 0.6527,
+      "step": 45983
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.2699549198150635,
+      "learning_rate": 1.5668958058227717e-06,
+      "loss": 0.7147,
+      "step": 45984
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.0504813194274902,
+      "learning_rate": 1.5667041269619365e-06,
+      "loss": 0.6207,
+      "step": 45985
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.5622825622558594,
+      "learning_rate": 1.566512457647809e-06,
+      "loss": 0.487,
+      "step": 45986
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.2445595264434814,
+      "learning_rate": 1.5663207978809258e-06,
+      "loss": 0.69,
+      "step": 45987
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.040351390838623,
+      "learning_rate": 1.5661291476618168e-06,
+      "loss": 0.5143,
+      "step": 45988
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.7206742763519287,
+      "learning_rate": 1.5659375069910182e-06,
+      "loss": 0.5084,
+      "step": 45989
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.2049918174743652,
+      "learning_rate": 1.5657458758690585e-06,
+      "loss": 0.4945,
+      "step": 45990
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.7504427433013916,
+      "learning_rate": 1.5655542542964753e-06,
+      "loss": 0.8731,
+      "step": 45991
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 1.761308193206787,
+      "learning_rate": 1.5653626422737978e-06,
+      "loss": 0.4502,
+      "step": 45992
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.0306577682495117,
+      "learning_rate": 1.5651710398015612e-06,
+      "loss": 0.3424,
+      "step": 45993
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.391707420349121,
+      "learning_rate": 1.5649794468802953e-06,
+      "loss": 0.7783,
+      "step": 45994
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.6947903633117676,
+      "learning_rate": 1.5647878635105368e-06,
+      "loss": 0.4705,
+      "step": 45995
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.9956607818603516,
+      "learning_rate": 1.5645962896928151e-06,
+      "loss": 0.6883,
+      "step": 45996
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.048546314239502,
+      "learning_rate": 1.564404725427664e-06,
+      "loss": 0.3663,
+      "step": 45997
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.639608144760132,
+      "learning_rate": 1.5642131707156161e-06,
+      "loss": 0.4956,
+      "step": 45998
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.4799911975860596,
+      "learning_rate": 1.564021625557206e-06,
+      "loss": 0.7697,
+      "step": 45999
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 1.8775171041488647,
+      "learning_rate": 1.5638300899529628e-06,
+      "loss": 0.4444,
+      "step": 46000
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.691443681716919,
+      "learning_rate": 1.5636385639034229e-06,
+      "loss": 0.6972,
+      "step": 46001
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.6952297687530518,
+      "learning_rate": 1.563447047409115e-06,
+      "loss": 0.6935,
+      "step": 46002
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.3286378383636475,
+      "learning_rate": 1.563255540470575e-06,
+      "loss": 0.7349,
+      "step": 46003
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.3682851791381836,
+      "learning_rate": 1.563064043088333e-06,
+      "loss": 0.5119,
+      "step": 46004
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.4758596420288086,
+      "learning_rate": 1.5628725552629232e-06,
+      "loss": 0.7592,
+      "step": 46005
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.6538960933685303,
+      "learning_rate": 1.5626810769948765e-06,
+      "loss": 0.764,
+      "step": 46006
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.5945885181427,
+      "learning_rate": 1.5624896082847268e-06,
+      "loss": 0.5483,
+      "step": 46007
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 1.685456395149231,
+      "learning_rate": 1.5622981491330047e-06,
+      "loss": 0.6316,
+      "step": 46008
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.7676761150360107,
+      "learning_rate": 1.5621066995402434e-06,
+      "loss": 0.7669,
+      "step": 46009
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 1.9070066213607788,
+      "learning_rate": 1.561915259506977e-06,
+      "loss": 0.5662,
+      "step": 46010
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.790781259536743,
+      "learning_rate": 1.561723829033735e-06,
+      "loss": 0.6943,
+      "step": 46011
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.782411575317383,
+      "learning_rate": 1.5615324081210525e-06,
+      "loss": 0.4132,
+      "step": 46012
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 5.750025749206543,
+      "learning_rate": 1.5613409967694587e-06,
+      "loss": 0.6899,
+      "step": 46013
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.6603596210479736,
+      "learning_rate": 1.561149594979489e-06,
+      "loss": 0.6268,
+      "step": 46014
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.109668016433716,
+      "learning_rate": 1.5609582027516722e-06,
+      "loss": 0.6786,
+      "step": 46015
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.23551344871521,
+      "learning_rate": 1.5607668200865444e-06,
+      "loss": 0.5888,
+      "step": 46016
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.663350820541382,
+      "learning_rate": 1.560575446984634e-06,
+      "loss": 0.642,
+      "step": 46017
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.3339719772338867,
+      "learning_rate": 1.560384083446475e-06,
+      "loss": 0.7013,
+      "step": 46018
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.6724982261657715,
+      "learning_rate": 1.5601927294726004e-06,
+      "loss": 0.6435,
+      "step": 46019
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.913250684738159,
+      "learning_rate": 1.5600013850635403e-06,
+      "loss": 0.6254,
+      "step": 46020
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 1.9297527074813843,
+      "learning_rate": 1.5598100502198272e-06,
+      "loss": 0.4433,
+      "step": 46021
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.5230085849761963,
+      "learning_rate": 1.5596187249419953e-06,
+      "loss": 0.5029,
+      "step": 46022
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.305192708969116,
+      "learning_rate": 1.5594274092305739e-06,
+      "loss": 0.5253,
+      "step": 46023
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.617952585220337,
+      "learning_rate": 1.559236103086097e-06,
+      "loss": 0.6257,
+      "step": 46024
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.554992914199829,
+      "learning_rate": 1.559044806509094e-06,
+      "loss": 0.5003,
+      "step": 46025
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.7502408027648926,
+      "learning_rate": 1.5588535195001004e-06,
+      "loss": 0.5866,
+      "step": 46026
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.4799914360046387,
+      "learning_rate": 1.5586622420596437e-06,
+      "loss": 0.6275,
+      "step": 46027
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.5896224975585938,
+      "learning_rate": 1.5584709741882604e-06,
+      "loss": 0.9297,
+      "step": 46028
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.8861773014068604,
+      "learning_rate": 1.5582797158864782e-06,
+      "loss": 0.6767,
+      "step": 46029
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.791412591934204,
+      "learning_rate": 1.5580884671548325e-06,
+      "loss": 0.9406,
+      "step": 46030
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.586581230163574,
+      "learning_rate": 1.557897227993852e-06,
+      "loss": 0.7459,
+      "step": 46031
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.4334561824798584,
+      "learning_rate": 1.557705998404071e-06,
+      "loss": 0.3716,
+      "step": 46032
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.1924567222595215,
+      "learning_rate": 1.557514778386019e-06,
+      "loss": 0.5724,
+      "step": 46033
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.861318349838257,
+      "learning_rate": 1.5573235679402304e-06,
+      "loss": 0.6631,
+      "step": 46034
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.150303363800049,
+      "learning_rate": 1.5571323670672333e-06,
+      "loss": 0.5652,
+      "step": 46035
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.325716733932495,
+      "learning_rate": 1.5569411757675635e-06,
+      "loss": 0.5067,
+      "step": 46036
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.304380416870117,
+      "learning_rate": 1.5567499940417485e-06,
+      "loss": 0.7273,
+      "step": 46037
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.296842098236084,
+      "learning_rate": 1.5565588218903215e-06,
+      "loss": 0.7346,
+      "step": 46038
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 4.051694393157959,
+      "learning_rate": 1.556367659313815e-06,
+      "loss": 0.6943,
+      "step": 46039
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 1.9794301986694336,
+      "learning_rate": 1.5561765063127615e-06,
+      "loss": 0.7055,
+      "step": 46040
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.051443576812744,
+      "learning_rate": 1.5559853628876897e-06,
+      "loss": 0.7706,
+      "step": 46041
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.1390490531921387,
+      "learning_rate": 1.5557942290391337e-06,
+      "loss": 0.7418,
+      "step": 46042
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.4698116779327393,
+      "learning_rate": 1.5556031047676224e-06,
+      "loss": 0.5227,
+      "step": 46043
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.2492311000823975,
+      "learning_rate": 1.5554119900736902e-06,
+      "loss": 0.7082,
+      "step": 46044
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.4310572147369385,
+      "learning_rate": 1.5552208849578648e-06,
+      "loss": 0.5763,
+      "step": 46045
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.764479875564575,
+      "learning_rate": 1.5550297894206818e-06,
+      "loss": 0.6425,
+      "step": 46046
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.534705400466919,
+      "learning_rate": 1.554838703462669e-06,
+      "loss": 0.8317,
+      "step": 46047
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 8.838528633117676,
+      "learning_rate": 1.5546476270843602e-06,
+      "loss": 0.562,
+      "step": 46048
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.8965439796447754,
+      "learning_rate": 1.5544565602862843e-06,
+      "loss": 0.8039,
+      "step": 46049
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.088264226913452,
+      "learning_rate": 1.5542655030689758e-06,
+      "loss": 0.4706,
+      "step": 46050
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.4176793098449707,
+      "learning_rate": 1.5540744554329622e-06,
+      "loss": 0.7097,
+      "step": 46051
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.847759246826172,
+      "learning_rate": 1.553883417378778e-06,
+      "loss": 0.5418,
+      "step": 46052
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.282132625579834,
+      "learning_rate": 1.5536923889069522e-06,
+      "loss": 0.4684,
+      "step": 46053
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.322357654571533,
+      "learning_rate": 1.553501370018018e-06,
+      "loss": 0.491,
+      "step": 46054
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.532792568206787,
+      "learning_rate": 1.553310360712504e-06,
+      "loss": 0.3823,
+      "step": 46055
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.2313222885131836,
+      "learning_rate": 1.553119360990944e-06,
+      "loss": 0.9395,
+      "step": 46056
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.1144824028015137,
+      "learning_rate": 1.5529283708538668e-06,
+      "loss": 0.5902,
+      "step": 46057
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.9824318885803223,
+      "learning_rate": 1.5527373903018055e-06,
+      "loss": 0.5459,
+      "step": 46058
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 7.1100873947143555,
+      "learning_rate": 1.5525464193352873e-06,
+      "loss": 0.4675,
+      "step": 46059
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.760800838470459,
+      "learning_rate": 1.5523554579548495e-06,
+      "loss": 0.5242,
+      "step": 46060
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.695415735244751,
+      "learning_rate": 1.5521645061610175e-06,
+      "loss": 0.5758,
+      "step": 46061
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 5.381547451019287,
+      "learning_rate": 1.551973563954326e-06,
+      "loss": 0.6524,
+      "step": 46062
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.087660789489746,
+      "learning_rate": 1.551782631335303e-06,
+      "loss": 0.6848,
+      "step": 46063
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 1.7598066329956055,
+      "learning_rate": 1.5515917083044824e-06,
+      "loss": 0.5229,
+      "step": 46064
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 1.8521009683609009,
+      "learning_rate": 1.5514007948623922e-06,
+      "loss": 0.3419,
+      "step": 46065
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 8.996923446655273,
+      "learning_rate": 1.5512098910095652e-06,
+      "loss": 0.6455,
+      "step": 46066
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 4.9951863288879395,
+      "learning_rate": 1.5510189967465305e-06,
+      "loss": 0.616,
+      "step": 46067
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.2340846061706543,
+      "learning_rate": 1.5508281120738216e-06,
+      "loss": 0.5119,
+      "step": 46068
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.443901777267456,
+      "learning_rate": 1.5506372369919664e-06,
+      "loss": 0.5645,
+      "step": 46069
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.9952592849731445,
+      "learning_rate": 1.5504463715014962e-06,
+      "loss": 0.5435,
+      "step": 46070
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.602546453475952,
+      "learning_rate": 1.5502555156029442e-06,
+      "loss": 0.7685,
+      "step": 46071
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.2438416481018066,
+      "learning_rate": 1.5500646692968379e-06,
+      "loss": 0.6239,
+      "step": 46072
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.8141021728515625,
+      "learning_rate": 1.5498738325837109e-06,
+      "loss": 0.7439,
+      "step": 46073
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.945977210998535,
+      "learning_rate": 1.5496830054640904e-06,
+      "loss": 0.802,
+      "step": 46074
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.1416373252868652,
+      "learning_rate": 1.5494921879385111e-06,
+      "loss": 0.7991,
+      "step": 46075
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 1.8031119108200073,
+      "learning_rate": 1.5493013800074997e-06,
+      "loss": 0.4209,
+      "step": 46076
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.164677143096924,
+      "learning_rate": 1.5491105816715896e-06,
+      "loss": 0.6108,
+      "step": 46077
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.275359630584717,
+      "learning_rate": 1.5489197929313092e-06,
+      "loss": 0.6633,
+      "step": 46078
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.0062973499298096,
+      "learning_rate": 1.5487290137871896e-06,
+      "loss": 0.4541,
+      "step": 46079
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.6397247314453125,
+      "learning_rate": 1.548538244239764e-06,
+      "loss": 0.4836,
+      "step": 46080
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.2938380241394043,
+      "learning_rate": 1.5483474842895586e-06,
+      "loss": 0.5746,
+      "step": 46081
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.3635125160217285,
+      "learning_rate": 1.5481567339371057e-06,
+      "loss": 0.3074,
+      "step": 46082
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.4239814281463623,
+      "learning_rate": 1.5479659931829371e-06,
+      "loss": 0.5895,
+      "step": 46083
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.8746542930603027,
+      "learning_rate": 1.547775262027581e-06,
+      "loss": 0.7055,
+      "step": 46084
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.126500368118286,
+      "learning_rate": 1.54758454047157e-06,
+      "loss": 0.5994,
+      "step": 46085
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.730827569961548,
+      "learning_rate": 1.5473938285154317e-06,
+      "loss": 0.705,
+      "step": 46086
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.1300039291381836,
+      "learning_rate": 1.5472031261596993e-06,
+      "loss": 0.5664,
+      "step": 46087
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.672280788421631,
+      "learning_rate": 1.5470124334048996e-06,
+      "loss": 0.4482,
+      "step": 46088
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.1420600414276123,
+      "learning_rate": 1.5468217502515665e-06,
+      "loss": 0.5363,
+      "step": 46089
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.969058036804199,
+      "learning_rate": 1.5466310767002273e-06,
+      "loss": 0.6576,
+      "step": 46090
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.1240265369415283,
+      "learning_rate": 1.546440412751415e-06,
+      "loss": 0.4792,
+      "step": 46091
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.713179588317871,
+      "learning_rate": 1.5462497584056563e-06,
+      "loss": 0.6875,
+      "step": 46092
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.7402257919311523,
+      "learning_rate": 1.5460591136634844e-06,
+      "loss": 0.7511,
+      "step": 46093
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.4699442386627197,
+      "learning_rate": 1.5458684785254273e-06,
+      "loss": 0.4974,
+      "step": 46094
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.1322569847106934,
+      "learning_rate": 1.5456778529920175e-06,
+      "loss": 0.4129,
+      "step": 46095
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 4.003979206085205,
+      "learning_rate": 1.5454872370637818e-06,
+      "loss": 0.5632,
+      "step": 46096
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.360095500946045,
+      "learning_rate": 1.5452966307412537e-06,
+      "loss": 0.7924,
+      "step": 46097
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.980199098587036,
+      "learning_rate": 1.5451060340249603e-06,
+      "loss": 0.7155,
+      "step": 46098
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 4.379270553588867,
+      "learning_rate": 1.544915446915432e-06,
+      "loss": 0.6404,
+      "step": 46099
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 1.8657000064849854,
+      "learning_rate": 1.5447248694132e-06,
+      "loss": 0.4849,
+      "step": 46100
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.5082337856292725,
+      "learning_rate": 1.5445343015187952e-06,
+      "loss": 0.6556,
+      "step": 46101
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 6.966772556304932,
+      "learning_rate": 1.5443437432327447e-06,
+      "loss": 0.8233,
+      "step": 46102
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.2658708095550537,
+      "learning_rate": 1.544153194555581e-06,
+      "loss": 0.4652,
+      "step": 46103
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.718411684036255,
+      "learning_rate": 1.5439626554878318e-06,
+      "loss": 0.5425,
+      "step": 46104
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.8070013523101807,
+      "learning_rate": 1.5437721260300286e-06,
+      "loss": 0.5471,
+      "step": 46105
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.975325584411621,
+      "learning_rate": 1.5435816061826992e-06,
+      "loss": 0.5773,
+      "step": 46106
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.4347074031829834,
+      "learning_rate": 1.543391095946376e-06,
+      "loss": 0.6381,
+      "step": 46107
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.8360419273376465,
+      "learning_rate": 1.5432005953215856e-06,
+      "loss": 0.7594,
+      "step": 46108
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.082298517227173,
+      "learning_rate": 1.5430101043088607e-06,
+      "loss": 0.8007,
+      "step": 46109
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.154709577560425,
+      "learning_rate": 1.5428196229087288e-06,
+      "loss": 0.6189,
+      "step": 46110
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.4213197231292725,
+      "learning_rate": 1.5426291511217216e-06,
+      "loss": 0.6807,
+      "step": 46111
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.0506036281585693,
+      "learning_rate": 1.542438688948366e-06,
+      "loss": 0.5784,
+      "step": 46112
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.872502326965332,
+      "learning_rate": 1.542248236389195e-06,
+      "loss": 0.828,
+      "step": 46113
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.4876694679260254,
+      "learning_rate": 1.5420577934447346e-06,
+      "loss": 0.413,
+      "step": 46114
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.898085355758667,
+      "learning_rate": 1.5418673601155177e-06,
+      "loss": 0.6391,
+      "step": 46115
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.381737232208252,
+      "learning_rate": 1.5416769364020706e-06,
+      "loss": 0.6006,
+      "step": 46116
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 7.309082508087158,
+      "learning_rate": 1.5414865223049257e-06,
+      "loss": 0.4132,
+      "step": 46117
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.445800304412842,
+      "learning_rate": 1.5412961178246105e-06,
+      "loss": 0.6655,
+      "step": 46118
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.6123485565185547,
+      "learning_rate": 1.5411057229616562e-06,
+      "loss": 0.5102,
+      "step": 46119
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 4.478095531463623,
+      "learning_rate": 1.5409153377165875e-06,
+      "loss": 0.4742,
+      "step": 46120
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.058307647705078,
+      "learning_rate": 1.5407249620899412e-06,
+      "loss": 0.808,
+      "step": 46121
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 4.04439640045166,
+      "learning_rate": 1.540534596082241e-06,
+      "loss": 0.524,
+      "step": 46122
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.4249820709228516,
+      "learning_rate": 1.54034423969402e-06,
+      "loss": 0.7432,
+      "step": 46123
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.584264039993286,
+      "learning_rate": 1.5401538929258042e-06,
+      "loss": 0.6585,
+      "step": 46124
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.1278316974639893,
+      "learning_rate": 1.5399635557781257e-06,
+      "loss": 0.6432,
+      "step": 46125
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.282214641571045,
+      "learning_rate": 1.539773228251511e-06,
+      "loss": 0.5969,
+      "step": 46126
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.3008198738098145,
+      "learning_rate": 1.539582910346492e-06,
+      "loss": 0.5543,
+      "step": 46127
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 1.908426284790039,
+      "learning_rate": 1.5393926020635951e-06,
+      "loss": 0.4565,
+      "step": 46128
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 5.172653675079346,
+      "learning_rate": 1.5392023034033527e-06,
+      "loss": 0.6537,
+      "step": 46129
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.471926212310791,
+      "learning_rate": 1.5390120143662907e-06,
+      "loss": 0.6674,
+      "step": 46130
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.1251261234283447,
+      "learning_rate": 1.5388217349529411e-06,
+      "loss": 0.6058,
+      "step": 46131
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.356802225112915,
+      "learning_rate": 1.53863146516383e-06,
+      "loss": 0.5869,
+      "step": 46132
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.0626583099365234,
+      "learning_rate": 1.5384412049994884e-06,
+      "loss": 0.6229,
+      "step": 46133
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.153533935546875,
+      "learning_rate": 1.5382509544604468e-06,
+      "loss": 0.6186,
+      "step": 46134
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.2019267082214355,
+      "learning_rate": 1.5380607135472308e-06,
+      "loss": 0.699,
+      "step": 46135
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.919981002807617,
+      "learning_rate": 1.5378704822603724e-06,
+      "loss": 0.5519,
+      "step": 46136
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.589540958404541,
+      "learning_rate": 1.5376802606003982e-06,
+      "loss": 0.7963,
+      "step": 46137
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.2313148975372314,
+      "learning_rate": 1.5374900485678397e-06,
+      "loss": 0.591,
+      "step": 46138
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.279789686203003,
+      "learning_rate": 1.5372998461632221e-06,
+      "loss": 0.7084,
+      "step": 46139
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.559220552444458,
+      "learning_rate": 1.5371096533870765e-06,
+      "loss": 0.4902,
+      "step": 46140
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 4.585363864898682,
+      "learning_rate": 1.5369194702399337e-06,
+      "loss": 0.7442,
+      "step": 46141
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.8315112590789795,
+      "learning_rate": 1.536729296722319e-06,
+      "loss": 0.7307,
+      "step": 46142
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.1396591663360596,
+      "learning_rate": 1.536539132834763e-06,
+      "loss": 0.5683,
+      "step": 46143
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.8788368701934814,
+      "learning_rate": 1.5363489785777953e-06,
+      "loss": 0.8352,
+      "step": 46144
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.684837579727173,
+      "learning_rate": 1.5361588339519423e-06,
+      "loss": 0.4598,
+      "step": 46145
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.406125783920288,
+      "learning_rate": 1.5359686989577354e-06,
+      "loss": 0.2893,
+      "step": 46146
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.1451125144958496,
+      "learning_rate": 1.5357785735957003e-06,
+      "loss": 0.626,
+      "step": 46147
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.629601001739502,
+      "learning_rate": 1.5355884578663687e-06,
+      "loss": 0.4705,
+      "step": 46148
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.6987860202789307,
+      "learning_rate": 1.5353983517702663e-06,
+      "loss": 0.5549,
+      "step": 46149
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.5530545711517334,
+      "learning_rate": 1.535208255307925e-06,
+      "loss": 0.5043,
+      "step": 46150
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.645094633102417,
+      "learning_rate": 1.5350181684798697e-06,
+      "loss": 0.5871,
+      "step": 46151
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.4106953144073486,
+      "learning_rate": 1.534828091286633e-06,
+      "loss": 0.5543,
+      "step": 46152
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.600308656692505,
+      "learning_rate": 1.5346380237287396e-06,
+      "loss": 0.5155,
+      "step": 46153
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 6.587682247161865,
+      "learning_rate": 1.5344479658067206e-06,
+      "loss": 0.5535,
+      "step": 46154
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.741007089614868,
+      "learning_rate": 1.5342579175211025e-06,
+      "loss": 0.7165,
+      "step": 46155
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.0883545875549316,
+      "learning_rate": 1.5340678788724166e-06,
+      "loss": 0.7991,
+      "step": 46156
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.5721564292907715,
+      "learning_rate": 1.5338778498611878e-06,
+      "loss": 0.7757,
+      "step": 46157
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.1160640716552734,
+      "learning_rate": 1.5336878304879477e-06,
+      "loss": 0.5702,
+      "step": 46158
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.8022937774658203,
+      "learning_rate": 1.5334978207532219e-06,
+      "loss": 0.5946,
+      "step": 46159
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.2571749687194824,
+      "learning_rate": 1.5333078206575402e-06,
+      "loss": 0.5852,
+      "step": 46160
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.071457624435425,
+      "learning_rate": 1.5331178302014304e-06,
+      "loss": 0.6894,
+      "step": 46161
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.9390971660614014,
+      "learning_rate": 1.5329278493854227e-06,
+      "loss": 0.6671,
+      "step": 46162
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.081915855407715,
+      "learning_rate": 1.5327378782100427e-06,
+      "loss": 0.6139,
+      "step": 46163
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.2284533977508545,
+      "learning_rate": 1.5325479166758212e-06,
+      "loss": 0.755,
+      "step": 46164
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 5.117013454437256,
+      "learning_rate": 1.532357964783283e-06,
+      "loss": 0.2679,
+      "step": 46165
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.8949756622314453,
+      "learning_rate": 1.5321680225329605e-06,
+      "loss": 0.4968,
+      "step": 46166
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.888292074203491,
+      "learning_rate": 1.5319780899253777e-06,
+      "loss": 0.6506,
+      "step": 46167
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.3937833309173584,
+      "learning_rate": 1.5317881669610668e-06,
+      "loss": 0.4864,
+      "step": 46168
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 4.903634071350098,
+      "learning_rate": 1.5315982536405515e-06,
+      "loss": 0.9265,
+      "step": 46169
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.1257872581481934,
+      "learning_rate": 1.531408349964364e-06,
+      "loss": 0.552,
+      "step": 46170
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 4.930338382720947,
+      "learning_rate": 1.531218455933029e-06,
+      "loss": 0.4017,
+      "step": 46171
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.5657103061676025,
+      "learning_rate": 1.5310285715470774e-06,
+      "loss": 0.6204,
+      "step": 46172
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.6740217208862305,
+      "learning_rate": 1.530838696807035e-06,
+      "loss": 0.4663,
+      "step": 46173
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.236438512802124,
+      "learning_rate": 1.5306488317134315e-06,
+      "loss": 0.5551,
+      "step": 46174
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 4.803705215454102,
+      "learning_rate": 1.5304589762667926e-06,
+      "loss": 0.5495,
+      "step": 46175
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.275989055633545,
+      "learning_rate": 1.5302691304676486e-06,
+      "loss": 0.4727,
+      "step": 46176
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.1693389415740967,
+      "learning_rate": 1.5300792943165254e-06,
+      "loss": 0.6412,
+      "step": 46177
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.8591346740722656,
+      "learning_rate": 1.529889467813953e-06,
+      "loss": 0.5522,
+      "step": 46178
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.169302225112915,
+      "learning_rate": 1.5296996509604572e-06,
+      "loss": 0.6179,
+      "step": 46179
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 4.233739852905273,
+      "learning_rate": 1.5295098437565674e-06,
+      "loss": 0.7176,
+      "step": 46180
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.4862520694732666,
+      "learning_rate": 1.5293200462028078e-06,
+      "loss": 0.6043,
+      "step": 46181
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 5.202991962432861,
+      "learning_rate": 1.5291302582997124e-06,
+      "loss": 0.5275,
+      "step": 46182
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.7178823947906494,
+      "learning_rate": 1.5289404800478035e-06,
+      "loss": 0.5618,
+      "step": 46183
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.5059101581573486,
+      "learning_rate": 1.528750711447613e-06,
+      "loss": 0.5349,
+      "step": 46184
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.4263193607330322,
+      "learning_rate": 1.5285609524996642e-06,
+      "loss": 0.5462,
+      "step": 46185
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.8038899898529053,
+      "learning_rate": 1.5283712032044889e-06,
+      "loss": 0.5251,
+      "step": 46186
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 8.034693717956543,
+      "learning_rate": 1.5281814635626114e-06,
+      "loss": 0.5756,
+      "step": 46187
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.166408061981201,
+      "learning_rate": 1.527991733574562e-06,
+      "loss": 0.4367,
+      "step": 46188
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.569882869720459,
+      "learning_rate": 1.5278020132408655e-06,
+      "loss": 0.7767,
+      "step": 46189
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.154224395751953,
+      "learning_rate": 1.527612302562052e-06,
+      "loss": 0.6816,
+      "step": 46190
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.2258286476135254,
+      "learning_rate": 1.5274226015386468e-06,
+      "loss": 0.5319,
+      "step": 46191
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 5.7144775390625,
+      "learning_rate": 1.5272329101711802e-06,
+      "loss": 0.6249,
+      "step": 46192
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.14662504196167,
+      "learning_rate": 1.5270432284601761e-06,
+      "loss": 0.6426,
+      "step": 46193
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.3170101642608643,
+      "learning_rate": 1.5268535564061638e-06,
+      "loss": 0.4882,
+      "step": 46194
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 1.8646001815795898,
+      "learning_rate": 1.5266638940096717e-06,
+      "loss": 0.5091,
+      "step": 46195
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.8553731441497803,
+      "learning_rate": 1.5264742412712252e-06,
+      "loss": 0.7779,
+      "step": 46196
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 4.348776340484619,
+      "learning_rate": 1.5262845981913544e-06,
+      "loss": 0.5043,
+      "step": 46197
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 1.8829238414764404,
+      "learning_rate": 1.526094964770583e-06,
+      "loss": 0.3415,
+      "step": 46198
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.6569225788116455,
+      "learning_rate": 1.5259053410094415e-06,
+      "loss": 0.5388,
+      "step": 46199
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.3664746284484863,
+      "learning_rate": 1.5257157269084548e-06,
+      "loss": 0.4508,
+      "step": 46200
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.5820682048797607,
+      "learning_rate": 1.5255261224681506e-06,
+      "loss": 0.7054,
+      "step": 46201
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.290959596633911,
+      "learning_rate": 1.5253365276890586e-06,
+      "loss": 0.4899,
+      "step": 46202
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.562342405319214,
+      "learning_rate": 1.5251469425717024e-06,
+      "loss": 0.4685,
+      "step": 46203
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.6415481567382812,
+      "learning_rate": 1.5249573671166123e-06,
+      "loss": 0.7306,
+      "step": 46204
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.4337093830108643,
+      "learning_rate": 1.5247678013243128e-06,
+      "loss": 0.6452,
+      "step": 46205
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.4483139514923096,
+      "learning_rate": 1.5245782451953317e-06,
+      "loss": 0.5525,
+      "step": 46206
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.7245519161224365,
+      "learning_rate": 1.5243886987301988e-06,
+      "loss": 0.6318,
+      "step": 46207
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 1.9230287075042725,
+      "learning_rate": 1.5241991619294365e-06,
+      "loss": 0.4543,
+      "step": 46208
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.2526440620422363,
+      "learning_rate": 1.5240096347935769e-06,
+      "loss": 0.522,
+      "step": 46209
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.5476362705230713,
+      "learning_rate": 1.5238201173231415e-06,
+      "loss": 0.8749,
+      "step": 46210
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.8026039600372314,
+      "learning_rate": 1.5236306095186625e-06,
+      "loss": 0.5705,
+      "step": 46211
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.0586252212524414,
+      "learning_rate": 1.523441111380663e-06,
+      "loss": 0.7111,
+      "step": 46212
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.072742462158203,
+      "learning_rate": 1.5232516229096727e-06,
+      "loss": 0.7841,
+      "step": 46213
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.482372999191284,
+      "learning_rate": 1.523062144106216e-06,
+      "loss": 0.4454,
+      "step": 46214
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.984616279602051,
+      "learning_rate": 1.5228726749708222e-06,
+      "loss": 0.685,
+      "step": 46215
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 4.268986701965332,
+      "learning_rate": 1.5226832155040156e-06,
+      "loss": 0.6493,
+      "step": 46216
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.44150710105896,
+      "learning_rate": 1.5224937657063255e-06,
+      "loss": 0.6411,
+      "step": 46217
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.0643696784973145,
+      "learning_rate": 1.5223043255782765e-06,
+      "loss": 0.5364,
+      "step": 46218
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.2007453441619873,
+      "learning_rate": 1.5221148951203978e-06,
+      "loss": 0.5287,
+      "step": 46219
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.5330207347869873,
+      "learning_rate": 1.5219254743332129e-06,
+      "loss": 0.4767,
+      "step": 46220
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.6907172203063965,
+      "learning_rate": 1.5217360632172506e-06,
+      "loss": 0.4571,
+      "step": 46221
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.6238162517547607,
+      "learning_rate": 1.5215466617730373e-06,
+      "loss": 0.9478,
+      "step": 46222
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.2436225414276123,
+      "learning_rate": 1.5213572700011009e-06,
+      "loss": 0.549,
+      "step": 46223
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.8930442333221436,
+      "learning_rate": 1.5211678879019658e-06,
+      "loss": 0.7476,
+      "step": 46224
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.4178435802459717,
+      "learning_rate": 1.5209785154761609e-06,
+      "loss": 0.5475,
+      "step": 46225
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.3465139865875244,
+      "learning_rate": 1.5207891527242097e-06,
+      "loss": 0.521,
+      "step": 46226
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.3857271671295166,
+      "learning_rate": 1.5205997996466426e-06,
+      "loss": 0.7214,
+      "step": 46227
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.3347327709198,
+      "learning_rate": 1.520410456243982e-06,
+      "loss": 0.6688,
+      "step": 46228
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.296574592590332,
+      "learning_rate": 1.520221122516758e-06,
+      "loss": 0.6356,
+      "step": 46229
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.843456506729126,
+      "learning_rate": 1.5200317984654944e-06,
+      "loss": 0.7092,
+      "step": 46230
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.973210334777832,
+      "learning_rate": 1.51984248409072e-06,
+      "loss": 0.4825,
+      "step": 46231
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.3257126808166504,
+      "learning_rate": 1.519653179392958e-06,
+      "loss": 0.395,
+      "step": 46232
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.7355401515960693,
+      "learning_rate": 1.5194638843727388e-06,
+      "loss": 0.4937,
+      "step": 46233
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.151707172393799,
+      "learning_rate": 1.5192745990305852e-06,
+      "loss": 0.354,
+      "step": 46234
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.7609310150146484,
+      "learning_rate": 1.5190853233670266e-06,
+      "loss": 0.6905,
+      "step": 46235
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.758199691772461,
+      "learning_rate": 1.5188960573825861e-06,
+      "loss": 0.4428,
+      "step": 46236
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.4405877590179443,
+      "learning_rate": 1.5187068010777927e-06,
+      "loss": 0.5099,
+      "step": 46237
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 4.166130542755127,
+      "learning_rate": 1.5185175544531706e-06,
+      "loss": 0.6628,
+      "step": 46238
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 4.741642475128174,
+      "learning_rate": 1.5183283175092483e-06,
+      "loss": 0.8101,
+      "step": 46239
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.9708476066589355,
+      "learning_rate": 1.5181390902465493e-06,
+      "loss": 0.4721,
+      "step": 46240
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.742645740509033,
+      "learning_rate": 1.5179498726656027e-06,
+      "loss": 0.4216,
+      "step": 46241
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.536454677581787,
+      "learning_rate": 1.5177606647669296e-06,
+      "loss": 0.5742,
+      "step": 46242
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.539644718170166,
+      "learning_rate": 1.5175714665510633e-06,
+      "loss": 0.5686,
+      "step": 46243
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.489553689956665,
+      "learning_rate": 1.5173822780185243e-06,
+      "loss": 0.6922,
+      "step": 46244
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.3916399478912354,
+      "learning_rate": 1.5171930991698419e-06,
+      "loss": 0.5381,
+      "step": 46245
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.748629093170166,
+      "learning_rate": 1.5170039300055395e-06,
+      "loss": 0.5755,
+      "step": 46246
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 4.502225399017334,
+      "learning_rate": 1.516814770526146e-06,
+      "loss": 0.6121,
+      "step": 46247
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.245819091796875,
+      "learning_rate": 1.5166256207321844e-06,
+      "loss": 0.71,
+      "step": 46248
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.4721193313598633,
+      "learning_rate": 1.5164364806241832e-06,
+      "loss": 0.6408,
+      "step": 46249
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.5989456176757812,
+      "learning_rate": 1.516247350202666e-06,
+      "loss": 0.7226,
+      "step": 46250
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.4665987491607666,
+      "learning_rate": 1.5160582294681614e-06,
+      "loss": 0.6688,
+      "step": 46251
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.7207441329956055,
+      "learning_rate": 1.5158691184211915e-06,
+      "loss": 0.4932,
+      "step": 46252
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.6148440837860107,
+      "learning_rate": 1.515680017062287e-06,
+      "loss": 0.4947,
+      "step": 46253
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.749173879623413,
+      "learning_rate": 1.515490925391969e-06,
+      "loss": 0.8495,
+      "step": 46254
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 1.8935655355453491,
+      "learning_rate": 1.5153018434107652e-06,
+      "loss": 0.3943,
+      "step": 46255
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.399402141571045,
+      "learning_rate": 1.5151127711192038e-06,
+      "loss": 0.7064,
+      "step": 46256
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.3712265491485596,
+      "learning_rate": 1.514923708517806e-06,
+      "loss": 0.5928,
+      "step": 46257
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.6717374324798584,
+      "learning_rate": 1.5147346556071019e-06,
+      "loss": 0.4255,
+      "step": 46258
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.0595381259918213,
+      "learning_rate": 1.5145456123876134e-06,
+      "loss": 0.7563,
+      "step": 46259
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.766552448272705,
+      "learning_rate": 1.5143565788598696e-06,
+      "loss": 0.6782,
+      "step": 46260
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.2490007877349854,
+      "learning_rate": 1.5141675550243928e-06,
+      "loss": 0.5758,
+      "step": 46261
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.0001254081726074,
+      "learning_rate": 1.5139785408817097e-06,
+      "loss": 0.5661,
+      "step": 46262
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.2316668033599854,
+      "learning_rate": 1.5137895364323484e-06,
+      "loss": 0.5547,
+      "step": 46263
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.6187267303466797,
+      "learning_rate": 1.5136005416768307e-06,
+      "loss": 0.593,
+      "step": 46264
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.8049817085266113,
+      "learning_rate": 1.513411556615686e-06,
+      "loss": 0.7321,
+      "step": 46265
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.5045504570007324,
+      "learning_rate": 1.5132225812494356e-06,
+      "loss": 0.6252,
+      "step": 46266
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.6699986457824707,
+      "learning_rate": 1.5130336155786068e-06,
+      "loss": 0.9067,
+      "step": 46267
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.8401472568511963,
+      "learning_rate": 1.5128446596037272e-06,
+      "loss": 0.6822,
+      "step": 46268
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.139632225036621,
+      "learning_rate": 1.5126557133253184e-06,
+      "loss": 0.6401,
+      "step": 46269
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 4.464308261871338,
+      "learning_rate": 1.5124667767439093e-06,
+      "loss": 0.556,
+      "step": 46270
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.932501792907715,
+      "learning_rate": 1.5122778498600221e-06,
+      "loss": 0.6564,
+      "step": 46271
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.6691691875457764,
+      "learning_rate": 1.5120889326741855e-06,
+      "loss": 0.6026,
+      "step": 46272
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.263869285583496,
+      "learning_rate": 1.5119000251869214e-06,
+      "loss": 0.6818,
+      "step": 46273
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.635510206222534,
+      "learning_rate": 1.5117111273987578e-06,
+      "loss": 0.568,
+      "step": 46274
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.3772952556610107,
+      "learning_rate": 1.511522239310218e-06,
+      "loss": 0.6465,
+      "step": 46275
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.1799745559692383,
+      "learning_rate": 1.5113333609218288e-06,
+      "loss": 0.7184,
+      "step": 46276
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 1.9088294506072998,
+      "learning_rate": 1.5111444922341134e-06,
+      "loss": 0.5174,
+      "step": 46277
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 1.883633017539978,
+      "learning_rate": 1.5109556332476e-06,
+      "loss": 0.3807,
+      "step": 46278
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 1.8568363189697266,
+      "learning_rate": 1.5107667839628097e-06,
+      "loss": 0.5308,
+      "step": 46279
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.2453176975250244,
+      "learning_rate": 1.510577944380272e-06,
+      "loss": 0.4287,
+      "step": 46280
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.459021806716919,
+      "learning_rate": 1.5103891145005085e-06,
+      "loss": 0.7907,
+      "step": 46281
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.3991522789001465,
+      "learning_rate": 1.5102002943240451e-06,
+      "loss": 0.4726,
+      "step": 46282
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 1.6636576652526855,
+      "learning_rate": 1.5100114838514079e-06,
+      "loss": 0.4461,
+      "step": 46283
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.2601706981658936,
+      "learning_rate": 1.5098226830831225e-06,
+      "loss": 0.6673,
+      "step": 46284
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.107617139816284,
+      "learning_rate": 1.5096338920197111e-06,
+      "loss": 0.9312,
+      "step": 46285
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.3615000247955322,
+      "learning_rate": 1.5094451106617026e-06,
+      "loss": 0.8946,
+      "step": 46286
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.0901315212249756,
+      "learning_rate": 1.5092563390096176e-06,
+      "loss": 0.5044,
+      "step": 46287
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.4894325733184814,
+      "learning_rate": 1.5090675770639845e-06,
+      "loss": 0.5271,
+      "step": 46288
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.705152750015259,
+      "learning_rate": 1.5088788248253255e-06,
+      "loss": 0.3965,
+      "step": 46289
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.3294448852539062,
+      "learning_rate": 1.508690082294168e-06,
+      "loss": 0.5729,
+      "step": 46290
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.4348952770233154,
+      "learning_rate": 1.5085013494710343e-06,
+      "loss": 0.6252,
+      "step": 46291
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.51008939743042,
+      "learning_rate": 1.5083126263564512e-06,
+      "loss": 0.5412,
+      "step": 46292
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.6196773052215576,
+      "learning_rate": 1.5081239129509418e-06,
+      "loss": 0.62,
+      "step": 46293
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.633652448654175,
+      "learning_rate": 1.507935209255033e-06,
+      "loss": 0.5264,
+      "step": 46294
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.3016209602355957,
+      "learning_rate": 1.5077465152692473e-06,
+      "loss": 0.5957,
+      "step": 46295
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 4.255184173583984,
+      "learning_rate": 1.5075578309941108e-06,
+      "loss": 0.8557,
+      "step": 46296
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.390338659286499,
+      "learning_rate": 1.5073691564301467e-06,
+      "loss": 0.5567,
+      "step": 46297
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.6404218673706055,
+      "learning_rate": 1.507180491577882e-06,
+      "loss": 0.7743,
+      "step": 46298
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.3978729248046875,
+      "learning_rate": 1.506991836437839e-06,
+      "loss": 0.4741,
+      "step": 46299
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.9640347957611084,
+      "learning_rate": 1.506803191010544e-06,
+      "loss": 0.5557,
+      "step": 46300
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.615206003189087,
+      "learning_rate": 1.5066145552965194e-06,
+      "loss": 0.6075,
+      "step": 46301
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.1403250694274902,
+      "learning_rate": 1.5064259292962924e-06,
+      "loss": 0.5217,
+      "step": 46302
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.7840771675109863,
+      "learning_rate": 1.5062373130103836e-06,
+      "loss": 0.7577,
+      "step": 46303
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.8910470008850098,
+      "learning_rate": 1.5060487064393225e-06,
+      "loss": 0.7165,
+      "step": 46304
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.2099313735961914,
+      "learning_rate": 1.5058601095836301e-06,
+      "loss": 0.5309,
+      "step": 46305
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.313047170639038,
+      "learning_rate": 1.5056715224438328e-06,
+      "loss": 0.6324,
+      "step": 46306
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.199317455291748,
+      "learning_rate": 1.5054829450204528e-06,
+      "loss": 0.5917,
+      "step": 46307
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.473027467727661,
+      "learning_rate": 1.5052943773140172e-06,
+      "loss": 0.4762,
+      "step": 46308
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 5.7269463539123535,
+      "learning_rate": 1.5051058193250472e-06,
+      "loss": 0.6953,
+      "step": 46309
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.3058242797851562,
+      "learning_rate": 1.5049172710540704e-06,
+      "loss": 0.7244,
+      "step": 46310
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.2647106647491455,
+      "learning_rate": 1.5047287325016074e-06,
+      "loss": 0.6934,
+      "step": 46311
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.842794179916382,
+      "learning_rate": 1.5045402036681862e-06,
+      "loss": 0.9484,
+      "step": 46312
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.914076566696167,
+      "learning_rate": 1.5043516845543276e-06,
+      "loss": 0.5769,
+      "step": 46313
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.2126052379608154,
+      "learning_rate": 1.5041631751605596e-06,
+      "loss": 0.8723,
+      "step": 46314
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.012617826461792,
+      "learning_rate": 1.5039746754874023e-06,
+      "loss": 0.5308,
+      "step": 46315
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.015275716781616,
+      "learning_rate": 1.503786185535383e-06,
+      "loss": 0.5582,
+      "step": 46316
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.354130268096924,
+      "learning_rate": 1.503597705305024e-06,
+      "loss": 0.4344,
+      "step": 46317
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.3850486278533936,
+      "learning_rate": 1.5034092347968487e-06,
+      "loss": 0.6428,
+      "step": 46318
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.008416175842285,
+      "learning_rate": 1.5032207740113846e-06,
+      "loss": 0.6846,
+      "step": 46319
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.7462363243103027,
+      "learning_rate": 1.5030323229491523e-06,
+      "loss": 0.7157,
+      "step": 46320
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.677128791809082,
+      "learning_rate": 1.5028438816106789e-06,
+      "loss": 0.7658,
+      "step": 46321
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.6132194995880127,
+      "learning_rate": 1.5026554499964846e-06,
+      "loss": 0.5362,
+      "step": 46322
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.943629741668701,
+      "learning_rate": 1.5024670281070953e-06,
+      "loss": 0.5227,
+      "step": 46323
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.3988406658172607,
+      "learning_rate": 1.5022786159430363e-06,
+      "loss": 0.4142,
+      "step": 46324
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.2070536613464355,
+      "learning_rate": 1.5020902135048293e-06,
+      "loss": 0.6717,
+      "step": 46325
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.4916462898254395,
+      "learning_rate": 1.5019018207930003e-06,
+      "loss": 0.5242,
+      "step": 46326
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.1829986572265625,
+      "learning_rate": 1.5017134378080706e-06,
+      "loss": 0.4457,
+      "step": 46327
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.3937385082244873,
+      "learning_rate": 1.5015250645505663e-06,
+      "loss": 0.6705,
+      "step": 46328
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.057396650314331,
+      "learning_rate": 1.501336701021009e-06,
+      "loss": 0.4131,
+      "step": 46329
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.2885801792144775,
+      "learning_rate": 1.501148347219924e-06,
+      "loss": 0.373,
+      "step": 46330
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.4717626571655273,
+      "learning_rate": 1.500960003147836e-06,
+      "loss": 0.4937,
+      "step": 46331
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.437243938446045,
+      "learning_rate": 1.500771668805266e-06,
+      "loss": 0.6366,
+      "step": 46332
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.0577147006988525,
+      "learning_rate": 1.5005833441927409e-06,
+      "loss": 0.5951,
+      "step": 46333
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 1.9816832542419434,
+      "learning_rate": 1.5003950293107806e-06,
+      "loss": 0.4581,
+      "step": 46334
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.0866522789001465,
+      "learning_rate": 1.5002067241599123e-06,
+      "loss": 0.4196,
+      "step": 46335
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.9554615020751953,
+      "learning_rate": 1.5000184287406566e-06,
+      "loss": 0.5585,
+      "step": 46336
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.085231065750122,
+      "learning_rate": 1.4998301430535405e-06,
+      "loss": 0.4491,
+      "step": 46337
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.183617353439331,
+      "learning_rate": 1.4996418670990837e-06,
+      "loss": 0.6555,
+      "step": 46338
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.711346387863159,
+      "learning_rate": 1.4994536008778133e-06,
+      "loss": 0.6828,
+      "step": 46339
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.0605764389038086,
+      "learning_rate": 1.4992653443902493e-06,
+      "loss": 0.6046,
+      "step": 46340
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.5627503395080566,
+      "learning_rate": 1.4990770976369184e-06,
+      "loss": 0.5602,
+      "step": 46341
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.186408519744873,
+      "learning_rate": 1.4988888606183417e-06,
+      "loss": 0.5412,
+      "step": 46342
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.2889821529388428,
+      "learning_rate": 1.4987006333350428e-06,
+      "loss": 0.3059,
+      "step": 46343
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.7795515060424805,
+      "learning_rate": 1.4985124157875464e-06,
+      "loss": 0.5035,
+      "step": 46344
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 1.9246859550476074,
+      "learning_rate": 1.4983242079763765e-06,
+      "loss": 0.6265,
+      "step": 46345
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.2283825874328613,
+      "learning_rate": 1.4981360099020536e-06,
+      "loss": 0.577,
+      "step": 46346
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.7057371139526367,
+      "learning_rate": 1.4979478215651044e-06,
+      "loss": 0.5752,
+      "step": 46347
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.0077874660491943,
+      "learning_rate": 1.4977596429660491e-06,
+      "loss": 0.5773,
+      "step": 46348
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.633068561553955,
+      "learning_rate": 1.4975714741054132e-06,
+      "loss": 0.4421,
+      "step": 46349
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.236147403717041,
+      "learning_rate": 1.4973833149837181e-06,
+      "loss": 0.858,
+      "step": 46350
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.307918071746826,
+      "learning_rate": 1.4971951656014889e-06,
+      "loss": 0.459,
+      "step": 46351
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.2569241523742676,
+      "learning_rate": 1.4970070259592468e-06,
+      "loss": 0.6027,
+      "step": 46352
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 1.6143436431884766,
+      "learning_rate": 1.4968188960575169e-06,
+      "loss": 0.4531,
+      "step": 46353
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.968097686767578,
+      "learning_rate": 1.4966307758968196e-06,
+      "loss": 0.4488,
+      "step": 46354
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.8732709884643555,
+      "learning_rate": 1.496442665477682e-06,
+      "loss": 0.5717,
+      "step": 46355
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.1604716777801514,
+      "learning_rate": 1.4962545648006226e-06,
+      "loss": 0.5909,
+      "step": 46356
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 1.9881813526153564,
+      "learning_rate": 1.496066473866169e-06,
+      "loss": 0.1617,
+      "step": 46357
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.111717939376831,
+      "learning_rate": 1.49587839267484e-06,
+      "loss": 0.4805,
+      "step": 46358
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 7.492537021636963,
+      "learning_rate": 1.4956903212271618e-06,
+      "loss": 0.2536,
+      "step": 46359
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.725761651992798,
+      "learning_rate": 1.4955022595236552e-06,
+      "loss": 0.6024,
+      "step": 46360
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.311736822128296,
+      "learning_rate": 1.4953142075648446e-06,
+      "loss": 0.5643,
+      "step": 46361
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 1.920175313949585,
+      "learning_rate": 1.4951261653512517e-06,
+      "loss": 0.394,
+      "step": 46362
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.4846320152282715,
+      "learning_rate": 1.4949381328834013e-06,
+      "loss": 0.7786,
+      "step": 46363
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.446798801422119,
+      "learning_rate": 1.4947501101618112e-06,
+      "loss": 0.7076,
+      "step": 46364
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 1.7820639610290527,
+      "learning_rate": 1.4945620971870116e-06,
+      "loss": 0.4567,
+      "step": 46365
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.163370370864868,
+      "learning_rate": 1.4943740939595202e-06,
+      "loss": 0.6292,
+      "step": 46366
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.8044180870056152,
+      "learning_rate": 1.4941861004798624e-06,
+      "loss": 0.7402,
+      "step": 46367
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.3434977531433105,
+      "learning_rate": 1.4939981167485583e-06,
+      "loss": 0.5997,
+      "step": 46368
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.0189883708953857,
+      "learning_rate": 1.4938101427661338e-06,
+      "loss": 0.3871,
+      "step": 46369
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.0361533164978027,
+      "learning_rate": 1.4936221785331085e-06,
+      "loss": 0.7866,
+      "step": 46370
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.3594753742218018,
+      "learning_rate": 1.4934342240500072e-06,
+      "loss": 0.5232,
+      "step": 46371
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.2380502223968506,
+      "learning_rate": 1.4932462793173509e-06,
+      "loss": 0.5539,
+      "step": 46372
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.625539541244507,
+      "learning_rate": 1.493058344335664e-06,
+      "loss": 0.5198,
+      "step": 46373
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 4.1900200843811035,
+      "learning_rate": 1.4928704191054671e-06,
+      "loss": 0.8355,
+      "step": 46374
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.387859344482422,
+      "learning_rate": 1.4926825036272845e-06,
+      "loss": 0.5575,
+      "step": 46375
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 1.8257746696472168,
+      "learning_rate": 1.4924945979016374e-06,
+      "loss": 0.3788,
+      "step": 46376
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.594229221343994,
+      "learning_rate": 1.4923067019290498e-06,
+      "loss": 0.6076,
+      "step": 46377
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.3704044818878174,
+      "learning_rate": 1.492118815710042e-06,
+      "loss": 0.6374,
+      "step": 46378
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.4115188121795654,
+      "learning_rate": 1.4919309392451376e-06,
+      "loss": 0.5084,
+      "step": 46379
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.401989459991455,
+      "learning_rate": 1.4917430725348609e-06,
+      "loss": 0.6823,
+      "step": 46380
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.4519543647766113,
+      "learning_rate": 1.4915552155797308e-06,
+      "loss": 0.6409,
+      "step": 46381
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.1442811489105225,
+      "learning_rate": 1.4913673683802727e-06,
+      "loss": 0.3631,
+      "step": 46382
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.3651182651519775,
+      "learning_rate": 1.491179530937006e-06,
+      "loss": 0.5155,
+      "step": 46383
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.2282440662384033,
+      "learning_rate": 1.490991703250455e-06,
+      "loss": 0.832,
+      "step": 46384
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.689864158630371,
+      "learning_rate": 1.4908038853211426e-06,
+      "loss": 0.2918,
+      "step": 46385
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.37408447265625,
+      "learning_rate": 1.4906160771495887e-06,
+      "loss": 0.5428,
+      "step": 46386
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.197331666946411,
+      "learning_rate": 1.4904282787363184e-06,
+      "loss": 0.5347,
+      "step": 46387
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.179079532623291,
+      "learning_rate": 1.4902404900818512e-06,
+      "loss": 0.5146,
+      "step": 46388
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.8813769817352295,
+      "learning_rate": 1.4900527111867113e-06,
+      "loss": 0.7296,
+      "step": 46389
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.375119686126709,
+      "learning_rate": 1.489864942051419e-06,
+      "loss": 0.7704,
+      "step": 46390
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 1.5771145820617676,
+      "learning_rate": 1.4896771826764966e-06,
+      "loss": 0.2687,
+      "step": 46391
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.416048526763916,
+      "learning_rate": 1.4894894330624693e-06,
+      "loss": 0.4481,
+      "step": 46392
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.6156005859375,
+      "learning_rate": 1.4893016932098547e-06,
+      "loss": 0.7905,
+      "step": 46393
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.451296091079712,
+      "learning_rate": 1.4891139631191786e-06,
+      "loss": 0.455,
+      "step": 46394
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.4166970252990723,
+      "learning_rate": 1.4889262427909602e-06,
+      "loss": 0.4017,
+      "step": 46395
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 1.8484867811203003,
+      "learning_rate": 1.4887385322257235e-06,
+      "loss": 0.4429,
+      "step": 46396
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.4495885372161865,
+      "learning_rate": 1.4885508314239888e-06,
+      "loss": 0.6498,
+      "step": 46397
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.377565860748291,
+      "learning_rate": 1.4883631403862802e-06,
+      "loss": 0.4978,
+      "step": 46398
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.718048572540283,
+      "learning_rate": 1.4881754591131164e-06,
+      "loss": 0.6577,
+      "step": 46399
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.411724090576172,
+      "learning_rate": 1.487987787605023e-06,
+      "loss": 0.5779,
+      "step": 46400
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.6330738067626953,
+      "learning_rate": 1.487800125862518e-06,
+      "loss": 0.5229,
+      "step": 46401
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.198137044906616,
+      "learning_rate": 1.487612473886127e-06,
+      "loss": 0.5958,
+      "step": 46402
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.6841681003570557,
+      "learning_rate": 1.4874248316763678e-06,
+      "loss": 0.585,
+      "step": 46403
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.062062978744507,
+      "learning_rate": 1.4872371992337647e-06,
+      "loss": 0.5885,
+      "step": 46404
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.1046533584594727,
+      "learning_rate": 1.4870495765588395e-06,
+      "loss": 0.4064,
+      "step": 46405
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.2742393016815186,
+      "learning_rate": 1.4868619636521142e-06,
+      "loss": 0.7724,
+      "step": 46406
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.1830978393554688,
+      "learning_rate": 1.4866743605141088e-06,
+      "loss": 0.5913,
+      "step": 46407
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.3623313903808594,
+      "learning_rate": 1.486486767145347e-06,
+      "loss": 0.7297,
+      "step": 46408
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.924917221069336,
+      "learning_rate": 1.486299183546348e-06,
+      "loss": 0.6726,
+      "step": 46409
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.777252674102783,
+      "learning_rate": 1.4861116097176359e-06,
+      "loss": 0.5981,
+      "step": 46410
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.1400723457336426,
+      "learning_rate": 1.4859240456597302e-06,
+      "loss": 0.6086,
+      "step": 46411
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 6.843619346618652,
+      "learning_rate": 1.4857364913731542e-06,
+      "loss": 0.4466,
+      "step": 46412
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.516223907470703,
+      "learning_rate": 1.4855489468584271e-06,
+      "loss": 0.6388,
+      "step": 46413
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 1.924195647239685,
+      "learning_rate": 1.4853614121160736e-06,
+      "loss": 0.6885,
+      "step": 46414
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.954787015914917,
+      "learning_rate": 1.4851738871466121e-06,
+      "loss": 0.8119,
+      "step": 46415
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.163536310195923,
+      "learning_rate": 1.4849863719505664e-06,
+      "loss": 0.57,
+      "step": 46416
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.4779415130615234,
+      "learning_rate": 1.4847988665284557e-06,
+      "loss": 0.6975,
+      "step": 46417
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.3863632678985596,
+      "learning_rate": 1.484611370880804e-06,
+      "loss": 0.4467,
+      "step": 46418
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.432999610900879,
+      "learning_rate": 1.4844238850081295e-06,
+      "loss": 0.5962,
+      "step": 46419
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.5937464237213135,
+      "learning_rate": 1.4842364089109567e-06,
+      "loss": 0.4211,
+      "step": 46420
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.6128578186035156,
+      "learning_rate": 1.484048942589804e-06,
+      "loss": 0.4449,
+      "step": 46421
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.4984240531921387,
+      "learning_rate": 1.4838614860451955e-06,
+      "loss": 0.3221,
+      "step": 46422
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.5137224197387695,
+      "learning_rate": 1.4836740392776494e-06,
+      "loss": 0.6457,
+      "step": 46423
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.1548585891723633,
+      "learning_rate": 1.4834866022876905e-06,
+      "loss": 0.5397,
+      "step": 46424
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.2960009574890137,
+      "learning_rate": 1.4832991750758346e-06,
+      "loss": 0.4033,
+      "step": 46425
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.257668972015381,
+      "learning_rate": 1.4831117576426096e-06,
+      "loss": 0.5404,
+      "step": 46426
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.0410678386688232,
+      "learning_rate": 1.4829243499885314e-06,
+      "loss": 0.4219,
+      "step": 46427
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.7529430389404297,
+      "learning_rate": 1.482736952114125e-06,
+      "loss": 0.7711,
+      "step": 46428
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.282273769378662,
+      "learning_rate": 1.4825495640199078e-06,
+      "loss": 0.7028,
+      "step": 46429
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.540860652923584,
+      "learning_rate": 1.4823621857064047e-06,
+      "loss": 0.7187,
+      "step": 46430
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 6.086682319641113,
+      "learning_rate": 1.4821748171741324e-06,
+      "loss": 0.7669,
+      "step": 46431
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.992673635482788,
+      "learning_rate": 1.481987458423616e-06,
+      "loss": 0.4532,
+      "step": 46432
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.904531240463257,
+      "learning_rate": 1.481800109455373e-06,
+      "loss": 0.611,
+      "step": 46433
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.4105677604675293,
+      "learning_rate": 1.4816127702699273e-06,
+      "loss": 0.63,
+      "step": 46434
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.149970293045044,
+      "learning_rate": 1.481425440867797e-06,
+      "loss": 0.5784,
+      "step": 46435
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.4120657444000244,
+      "learning_rate": 1.4812381212495058e-06,
+      "loss": 0.7203,
+      "step": 46436
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.1897594928741455,
+      "learning_rate": 1.481050811415572e-06,
+      "loss": 0.4809,
+      "step": 46437
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.21057391166687,
+      "learning_rate": 1.4808635113665194e-06,
+      "loss": 0.621,
+      "step": 46438
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.20000958442688,
+      "learning_rate": 1.4806762211028653e-06,
+      "loss": 0.716,
+      "step": 46439
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.4490227699279785,
+      "learning_rate": 1.480488940625132e-06,
+      "loss": 0.6436,
+      "step": 46440
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.97988224029541,
+      "learning_rate": 1.4803016699338424e-06,
+      "loss": 0.5849,
+      "step": 46441
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.469883680343628,
+      "learning_rate": 1.480114409029514e-06,
+      "loss": 0.5381,
+      "step": 46442
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.5390400886535645,
+      "learning_rate": 1.4799271579126701e-06,
+      "loss": 0.5871,
+      "step": 46443
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.8374338150024414,
+      "learning_rate": 1.4797399165838284e-06,
+      "loss": 0.4246,
+      "step": 46444
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.161818742752075,
+      "learning_rate": 1.4795526850435116e-06,
+      "loss": 0.524,
+      "step": 46445
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.2617294788360596,
+      "learning_rate": 1.4793654632922411e-06,
+      "loss": 0.6944,
+      "step": 46446
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.508044958114624,
+      "learning_rate": 1.4791782513305353e-06,
+      "loss": 0.6228,
+      "step": 46447
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 4.056554794311523,
+      "learning_rate": 1.4789910491589171e-06,
+      "loss": 0.6549,
+      "step": 46448
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.9515128135681152,
+      "learning_rate": 1.4788038567779044e-06,
+      "loss": 0.4947,
+      "step": 46449
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.6022891998291016,
+      "learning_rate": 1.4786166741880208e-06,
+      "loss": 0.3785,
+      "step": 46450
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.4853179454803467,
+      "learning_rate": 1.478429501389783e-06,
+      "loss": 0.4565,
+      "step": 46451
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.1085879802703857,
+      "learning_rate": 1.4782423383837141e-06,
+      "loss": 0.5824,
+      "step": 46452
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.2122037410736084,
+      "learning_rate": 1.4780551851703356e-06,
+      "loss": 0.5584,
+      "step": 46453
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.3841042518615723,
+      "learning_rate": 1.4778680417501644e-06,
+      "loss": 0.4375,
+      "step": 46454
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.4014229774475098,
+      "learning_rate": 1.4776809081237242e-06,
+      "loss": 0.4069,
+      "step": 46455
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.6245336532592773,
+      "learning_rate": 1.4774937842915326e-06,
+      "loss": 0.4879,
+      "step": 46456
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.1422410011291504,
+      "learning_rate": 1.4773066702541127e-06,
+      "loss": 0.5321,
+      "step": 46457
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.751859188079834,
+      "learning_rate": 1.477119566011982e-06,
+      "loss": 0.8522,
+      "step": 46458
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 1.6260359287261963,
+      "learning_rate": 1.476932471565663e-06,
+      "loss": 0.4226,
+      "step": 46459
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.9465019702911377,
+      "learning_rate": 1.4767453869156738e-06,
+      "loss": 0.5133,
+      "step": 46460
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.7431883811950684,
+      "learning_rate": 1.4765583120625377e-06,
+      "loss": 0.7868,
+      "step": 46461
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.7006332874298096,
+      "learning_rate": 1.4763712470067716e-06,
+      "loss": 0.5558,
+      "step": 46462
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.3282630443573,
+      "learning_rate": 1.476184191748898e-06,
+      "loss": 0.5083,
+      "step": 46463
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.6305196285247803,
+      "learning_rate": 1.4759971462894351e-06,
+      "loss": 0.4761,
+      "step": 46464
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.6335761547088623,
+      "learning_rate": 1.4758101106289035e-06,
+      "loss": 0.5723,
+      "step": 46465
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.2773184776306152,
+      "learning_rate": 1.4756230847678243e-06,
+      "loss": 0.6533,
+      "step": 46466
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.3834571838378906,
+      "learning_rate": 1.4754360687067182e-06,
+      "loss": 0.519,
+      "step": 46467
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.187401533126831,
+      "learning_rate": 1.4752490624461024e-06,
+      "loss": 0.6351,
+      "step": 46468
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 4.408424377441406,
+      "learning_rate": 1.4750620659865005e-06,
+      "loss": 0.6387,
+      "step": 46469
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.21685528755188,
+      "learning_rate": 1.474875079328429e-06,
+      "loss": 0.6852,
+      "step": 46470
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.28544282913208,
+      "learning_rate": 1.4746881024724107e-06,
+      "loss": 0.6184,
+      "step": 46471
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.073770046234131,
+      "learning_rate": 1.4745011354189626e-06,
+      "loss": 0.5479,
+      "step": 46472
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 1.7419471740722656,
+      "learning_rate": 1.4743141781686076e-06,
+      "loss": 0.2975,
+      "step": 46473
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.2289419174194336,
+      "learning_rate": 1.474127230721863e-06,
+      "loss": 0.8042,
+      "step": 46474
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.7942750453948975,
+      "learning_rate": 1.4739402930792507e-06,
+      "loss": 0.6658,
+      "step": 46475
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.816174030303955,
+      "learning_rate": 1.4737533652412882e-06,
+      "loss": 0.6148,
+      "step": 46476
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 1.8892632722854614,
+      "learning_rate": 1.4735664472084988e-06,
+      "loss": 0.4223,
+      "step": 46477
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.20742130279541,
+      "learning_rate": 1.4733795389813976e-06,
+      "loss": 0.648,
+      "step": 46478
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 1.9341622591018677,
+      "learning_rate": 1.473192640560509e-06,
+      "loss": 0.6466,
+      "step": 46479
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.3351762294769287,
+      "learning_rate": 1.4730057519463487e-06,
+      "loss": 0.5775,
+      "step": 46480
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.3768577575683594,
+      "learning_rate": 1.47281887313944e-06,
+      "loss": 0.7693,
+      "step": 46481
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.046562910079956,
+      "learning_rate": 1.4726320041402987e-06,
+      "loss": 0.6243,
+      "step": 46482
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.6899325847625732,
+      "learning_rate": 1.472445144949448e-06,
+      "loss": 0.535,
+      "step": 46483
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.26629376411438,
+      "learning_rate": 1.4722582955674048e-06,
+      "loss": 0.5937,
+      "step": 46484
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.0499513149261475,
+      "learning_rate": 1.4720714559946909e-06,
+      "loss": 0.6273,
+      "step": 46485
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.285945177078247,
+      "learning_rate": 1.471884626231822e-06,
+      "loss": 0.5855,
+      "step": 46486
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.221113681793213,
+      "learning_rate": 1.471697806279323e-06,
+      "loss": 0.7495,
+      "step": 46487
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.642237424850464,
+      "learning_rate": 1.4715109961377094e-06,
+      "loss": 0.5624,
+      "step": 46488
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.4370222091674805,
+      "learning_rate": 1.471324195807503e-06,
+      "loss": 0.6773,
+      "step": 46489
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.6352734565734863,
+      "learning_rate": 1.4711374052892207e-06,
+      "loss": 0.59,
+      "step": 46490
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.391533851623535,
+      "learning_rate": 1.4709506245833855e-06,
+      "loss": 0.7444,
+      "step": 46491
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.456993818283081,
+      "learning_rate": 1.4707638536905123e-06,
+      "loss": 0.4732,
+      "step": 46492
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.14386248588562,
+      "learning_rate": 1.4705770926111246e-06,
+      "loss": 0.5859,
+      "step": 46493
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.4028384685516357,
+      "learning_rate": 1.4703903413457382e-06,
+      "loss": 0.6247,
+      "step": 46494
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.636138677597046,
+      "learning_rate": 1.470203599894876e-06,
+      "loss": 0.4635,
+      "step": 46495
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.0494866371154785,
+      "learning_rate": 1.4700168682590533e-06,
+      "loss": 0.6054,
+      "step": 46496
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.3745553493499756,
+      "learning_rate": 1.4698301464387931e-06,
+      "loss": 0.7052,
+      "step": 46497
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.5759081840515137,
+      "learning_rate": 1.469643434434611e-06,
+      "loss": 0.4351,
+      "step": 46498
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.467017412185669,
+      "learning_rate": 1.46945673224703e-06,
+      "loss": 0.7732,
+      "step": 46499
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.1649692058563232,
+      "learning_rate": 1.4692700398765653e-06,
+      "loss": 0.504,
+      "step": 46500
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 1.8099853992462158,
+      "learning_rate": 1.4690833573237401e-06,
+      "loss": 0.3907,
+      "step": 46501
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.2254371643066406,
+      "learning_rate": 1.468896684589069e-06,
+      "loss": 0.5513,
+      "step": 46502
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.2154555320739746,
+      "learning_rate": 1.4687100216730742e-06,
+      "loss": 0.7082,
+      "step": 46503
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.0768706798553467,
+      "learning_rate": 1.4685233685762751e-06,
+      "loss": 0.5648,
+      "step": 46504
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.7607810497283936,
+      "learning_rate": 1.468336725299188e-06,
+      "loss": 0.6773,
+      "step": 46505
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.108118772506714,
+      "learning_rate": 1.4681500918423335e-06,
+      "loss": 0.766,
+      "step": 46506
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.6263844966888428,
+      "learning_rate": 1.4679634682062322e-06,
+      "loss": 0.8666,
+      "step": 46507
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.062143325805664,
+      "learning_rate": 1.4677768543914e-06,
+      "loss": 0.6825,
+      "step": 46508
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.3351900577545166,
+      "learning_rate": 1.4675902503983585e-06,
+      "loss": 0.7898,
+      "step": 46509
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.9555141925811768,
+      "learning_rate": 1.4674036562276239e-06,
+      "loss": 0.6687,
+      "step": 46510
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.694251537322998,
+      "learning_rate": 1.4672170718797175e-06,
+      "loss": 0.4609,
+      "step": 46511
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.980803966522217,
+      "learning_rate": 1.4670304973551557e-06,
+      "loss": 0.7124,
+      "step": 46512
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.1641387939453125,
+      "learning_rate": 1.4668439326544604e-06,
+      "loss": 0.425,
+      "step": 46513
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 1.625478982925415,
+      "learning_rate": 1.4666573777781467e-06,
+      "loss": 0.467,
+      "step": 46514
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.9870173931121826,
+      "learning_rate": 1.4664708327267351e-06,
+      "loss": 0.587,
+      "step": 46515
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.5879883766174316,
+      "learning_rate": 1.4662842975007463e-06,
+      "loss": 0.5412,
+      "step": 46516
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.975005626678467,
+      "learning_rate": 1.4660977721006953e-06,
+      "loss": 0.6243,
+      "step": 46517
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.8983640670776367,
+      "learning_rate": 1.4659112565271043e-06,
+      "loss": 0.7099,
+      "step": 46518
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.162728786468506,
+      "learning_rate": 1.4657247507804883e-06,
+      "loss": 0.4415,
+      "step": 46519
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.5747265815734863,
+      "learning_rate": 1.46553825486137e-06,
+      "loss": 0.5416,
+      "step": 46520
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 4.811347007751465,
+      "learning_rate": 1.465351768770264e-06,
+      "loss": 0.9307,
+      "step": 46521
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 4.445894241333008,
+      "learning_rate": 1.465165292507692e-06,
+      "loss": 0.4918,
+      "step": 46522
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.2856903076171875,
+      "learning_rate": 1.4649788260741694e-06,
+      "loss": 0.733,
+      "step": 46523
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.953402280807495,
+      "learning_rate": 1.4647923694702182e-06,
+      "loss": 0.7633,
+      "step": 46524
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.562363386154175,
+      "learning_rate": 1.4646059226963538e-06,
+      "loss": 0.6942,
+      "step": 46525
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.1773171424865723,
+      "learning_rate": 1.4644194857530957e-06,
+      "loss": 0.4911,
+      "step": 46526
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 4.145198822021484,
+      "learning_rate": 1.4642330586409625e-06,
+      "loss": 0.6107,
+      "step": 46527
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.7953784465789795,
+      "learning_rate": 1.4640466413604743e-06,
+      "loss": 0.5964,
+      "step": 46528
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.232548236846924,
+      "learning_rate": 1.463860233912146e-06,
+      "loss": 0.5389,
+      "step": 46529
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.6981279850006104,
+      "learning_rate": 1.4636738362964997e-06,
+      "loss": 0.7133,
+      "step": 46530
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.3108279705047607,
+      "learning_rate": 1.46348744851405e-06,
+      "loss": 0.6865,
+      "step": 46531
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.781751871109009,
+      "learning_rate": 1.4633010705653184e-06,
+      "loss": 0.4071,
+      "step": 46532
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 4.386685848236084,
+      "learning_rate": 1.4631147024508197e-06,
+      "loss": 0.6883,
+      "step": 46533
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.3438191413879395,
+      "learning_rate": 1.4629283441710762e-06,
+      "loss": 0.5619,
+      "step": 46534
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.8934056758880615,
+      "learning_rate": 1.462741995726602e-06,
+      "loss": 0.728,
+      "step": 46535
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.12869930267334,
+      "learning_rate": 1.462555657117919e-06,
+      "loss": 0.6574,
+      "step": 46536
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.643679141998291,
+      "learning_rate": 1.462369328345542e-06,
+      "loss": 0.393,
+      "step": 46537
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.196373462677002,
+      "learning_rate": 1.4621830094099926e-06,
+      "loss": 0.551,
+      "step": 46538
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.2080888748168945,
+      "learning_rate": 1.4619967003117846e-06,
+      "loss": 0.5858,
+      "step": 46539
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.138533592224121,
+      "learning_rate": 1.4618104010514406e-06,
+      "loss": 0.8396,
+      "step": 46540
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.648556709289551,
+      "learning_rate": 1.4616241116294745e-06,
+      "loss": 0.7975,
+      "step": 46541
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 1.9747272729873657,
+      "learning_rate": 1.4614378320464078e-06,
+      "loss": 0.548,
+      "step": 46542
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.5200047492980957,
+      "learning_rate": 1.4612515623027558e-06,
+      "loss": 0.4785,
+      "step": 46543
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 5.798317909240723,
+      "learning_rate": 1.4610653023990385e-06,
+      "loss": 0.4794,
+      "step": 46544
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.2874155044555664,
+      "learning_rate": 1.4608790523357718e-06,
+      "loss": 0.7339,
+      "step": 46545
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.432758331298828,
+      "learning_rate": 1.4606928121134762e-06,
+      "loss": 0.63,
+      "step": 46546
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.0975401401519775,
+      "learning_rate": 1.4605065817326647e-06,
+      "loss": 0.5959,
+      "step": 46547
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.7559728622436523,
+      "learning_rate": 1.4603203611938616e-06,
+      "loss": 0.6235,
+      "step": 46548
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.5826244354248047,
+      "learning_rate": 1.4601341504975802e-06,
+      "loss": 0.71,
+      "step": 46549
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.229219913482666,
+      "learning_rate": 1.4599479496443409e-06,
+      "loss": 0.4159,
+      "step": 46550
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.7378485202789307,
+      "learning_rate": 1.459761758634659e-06,
+      "loss": 0.6111,
+      "step": 46551
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.3863751888275146,
+      "learning_rate": 1.459575577469055e-06,
+      "loss": 0.5426,
+      "step": 46552
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.5368189811706543,
+      "learning_rate": 1.4593894061480434e-06,
+      "loss": 0.5081,
+      "step": 46553
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.633819818496704,
+      "learning_rate": 1.4592032446721448e-06,
+      "loss": 0.4225,
+      "step": 46554
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.458850383758545,
+      "learning_rate": 1.4590170930418745e-06,
+      "loss": 0.5151,
+      "step": 46555
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.89963960647583,
+      "learning_rate": 1.4588309512577526e-06,
+      "loss": 0.5329,
+      "step": 46556
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.51265811920166,
+      "learning_rate": 1.4586448193202934e-06,
+      "loss": 0.6692,
+      "step": 46557
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.2511770725250244,
+      "learning_rate": 1.458458697230018e-06,
+      "loss": 0.5629,
+      "step": 46558
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.203362226486206,
+      "learning_rate": 1.4582725849874407e-06,
+      "loss": 0.7413,
+      "step": 46559
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.184609889984131,
+      "learning_rate": 1.4580864825930825e-06,
+      "loss": 0.6566,
+      "step": 46560
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.641141414642334,
+      "learning_rate": 1.457900390047457e-06,
+      "loss": 0.7774,
+      "step": 46561
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.4794180393218994,
+      "learning_rate": 1.4577143073510858e-06,
+      "loss": 0.6566,
+      "step": 46562
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 4.334826946258545,
+      "learning_rate": 1.4575282345044818e-06,
+      "loss": 0.6571,
+      "step": 46563
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.5374131202697754,
+      "learning_rate": 1.457342171508165e-06,
+      "loss": 0.6996,
+      "step": 46564
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.7755343914031982,
+      "learning_rate": 1.4571561183626543e-06,
+      "loss": 0.8633,
+      "step": 46565
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.554980993270874,
+      "learning_rate": 1.4569700750684635e-06,
+      "loss": 0.4841,
+      "step": 46566
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 10.056532859802246,
+      "learning_rate": 1.4567840416261115e-06,
+      "loss": 0.7064,
+      "step": 46567
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.946571111679077,
+      "learning_rate": 1.4565980180361179e-06,
+      "loss": 0.4148,
+      "step": 46568
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 1.871227741241455,
+      "learning_rate": 1.4564120042989955e-06,
+      "loss": 0.4545,
+      "step": 46569
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.599783182144165,
+      "learning_rate": 1.4562260004152655e-06,
+      "loss": 0.7346,
+      "step": 46570
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.6505515575408936,
+      "learning_rate": 1.4560400063854423e-06,
+      "loss": 0.6891,
+      "step": 46571
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.1487514972686768,
+      "learning_rate": 1.455854022210046e-06,
+      "loss": 0.6738,
+      "step": 46572
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.6846301555633545,
+      "learning_rate": 1.4556680478895901e-06,
+      "loss": 0.5199,
+      "step": 46573
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.420469045639038,
+      "learning_rate": 1.4554820834245954e-06,
+      "loss": 0.6255,
+      "step": 46574
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.8975818157196045,
+      "learning_rate": 1.4552961288155754e-06,
+      "loss": 0.6328,
+      "step": 46575
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.0437815189361572,
+      "learning_rate": 1.4551101840630488e-06,
+      "loss": 0.4757,
+      "step": 46576
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.424330234527588,
+      "learning_rate": 1.4549242491675347e-06,
+      "loss": 0.6006,
+      "step": 46577
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.608790159225464,
+      "learning_rate": 1.4547383241295466e-06,
+      "loss": 0.7476,
+      "step": 46578
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.4535064697265625,
+      "learning_rate": 1.454552408949605e-06,
+      "loss": 0.5435,
+      "step": 46579
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.553514242172241,
+      "learning_rate": 1.4543665036282229e-06,
+      "loss": 0.7795,
+      "step": 46580
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 1.6032658815383911,
+      "learning_rate": 1.4541806081659216e-06,
+      "loss": 0.5573,
+      "step": 46581
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.1511433124542236,
+      "learning_rate": 1.4539947225632135e-06,
+      "loss": 0.528,
+      "step": 46582
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.6232213973999023,
+      "learning_rate": 1.4538088468206196e-06,
+      "loss": 0.7494,
+      "step": 46583
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.8577377796173096,
+      "learning_rate": 1.4536229809386532e-06,
+      "loss": 0.8253,
+      "step": 46584
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.691986083984375,
+      "learning_rate": 1.4534371249178342e-06,
+      "loss": 0.4132,
+      "step": 46585
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.543672561645508,
+      "learning_rate": 1.4532512787586766e-06,
+      "loss": 0.5999,
+      "step": 46586
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.2737133502960205,
+      "learning_rate": 1.4530654424616985e-06,
+      "loss": 0.3901,
+      "step": 46587
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.0318050384521484,
+      "learning_rate": 1.4528796160274166e-06,
+      "loss": 0.7826,
+      "step": 46588
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.3187873363494873,
+      "learning_rate": 1.4526937994563495e-06,
+      "loss": 0.5515,
+      "step": 46589
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.9769299030303955,
+      "learning_rate": 1.4525079927490105e-06,
+      "loss": 0.6542,
+      "step": 46590
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.767979860305786,
+      "learning_rate": 1.452322195905919e-06,
+      "loss": 0.6279,
+      "step": 46591
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.571916341781616,
+      "learning_rate": 1.4521364089275896e-06,
+      "loss": 0.6287,
+      "step": 46592
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.142266035079956,
+      "learning_rate": 1.451950631814541e-06,
+      "loss": 0.5421,
+      "step": 46593
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.0788447856903076,
+      "learning_rate": 1.4517648645672872e-06,
+      "loss": 0.5779,
+      "step": 46594
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.5746254920959473,
+      "learning_rate": 1.4515791071863472e-06,
+      "loss": 0.626,
+      "step": 46595
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.618213653564453,
+      "learning_rate": 1.4513933596722351e-06,
+      "loss": 0.6198,
+      "step": 46596
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 1.9645438194274902,
+      "learning_rate": 1.4512076220254706e-06,
+      "loss": 0.348,
+      "step": 46597
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.7651939392089844,
+      "learning_rate": 1.4510218942465664e-06,
+      "loss": 0.6496,
+      "step": 46598
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 4.509129047393799,
+      "learning_rate": 1.4508361763360424e-06,
+      "loss": 0.5815,
+      "step": 46599
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.428388833999634,
+      "learning_rate": 1.4506504682944122e-06,
+      "loss": 0.5208,
+      "step": 46600
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 1.833486795425415,
+      "learning_rate": 1.4504647701221947e-06,
+      "loss": 0.2559,
+      "step": 46601
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 4.264679908752441,
+      "learning_rate": 1.4502790818199037e-06,
+      "loss": 0.5305,
+      "step": 46602
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.0820984840393066,
+      "learning_rate": 1.4500934033880582e-06,
+      "loss": 0.8241,
+      "step": 46603
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.5190610885620117,
+      "learning_rate": 1.4499077348271713e-06,
+      "loss": 0.7389,
+      "step": 46604
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.9495866298675537,
+      "learning_rate": 1.4497220761377628e-06,
+      "loss": 0.5158,
+      "step": 46605
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.117457151412964,
+      "learning_rate": 1.4495364273203455e-06,
+      "loss": 0.5288,
+      "step": 46606
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.4269912242889404,
+      "learning_rate": 1.4493507883754394e-06,
+      "loss": 0.8372,
+      "step": 46607
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.550318956375122,
+      "learning_rate": 1.4491651593035556e-06,
+      "loss": 0.6951,
+      "step": 46608
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.8930301666259766,
+      "learning_rate": 1.4489795401052158e-06,
+      "loss": 0.6744,
+      "step": 46609
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 4.470608711242676,
+      "learning_rate": 1.4487939307809324e-06,
+      "loss": 0.663,
+      "step": 46610
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.023237466812134,
+      "learning_rate": 1.4486083313312244e-06,
+      "loss": 0.3972,
+      "step": 46611
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.119218349456787,
+      "learning_rate": 1.448422741756605e-06,
+      "loss": 0.4566,
+      "step": 46612
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 4.093811988830566,
+      "learning_rate": 1.4482371620575924e-06,
+      "loss": 0.9207,
+      "step": 46613
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.256910800933838,
+      "learning_rate": 1.4480515922347004e-06,
+      "loss": 0.4107,
+      "step": 46614
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.3082492351531982,
+      "learning_rate": 1.447866032288448e-06,
+      "loss": 0.4068,
+      "step": 46615
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.6479403972625732,
+      "learning_rate": 1.447680482219348e-06,
+      "loss": 0.6704,
+      "step": 46616
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.4779107570648193,
+      "learning_rate": 1.4474949420279195e-06,
+      "loss": 0.5939,
+      "step": 46617
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.882641077041626,
+      "learning_rate": 1.4473094117146752e-06,
+      "loss": 0.7219,
+      "step": 46618
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.026007652282715,
+      "learning_rate": 1.447123891280134e-06,
+      "loss": 0.5031,
+      "step": 46619
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 1.9094537496566772,
+      "learning_rate": 1.4469383807248084e-06,
+      "loss": 0.4343,
+      "step": 46620
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 4.114984035491943,
+      "learning_rate": 1.4467528800492181e-06,
+      "loss": 0.4553,
+      "step": 46621
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.1197686195373535,
+      "learning_rate": 1.4465673892538757e-06,
+      "loss": 0.7399,
+      "step": 46622
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.0456771850585938,
+      "learning_rate": 1.4463819083392994e-06,
+      "loss": 0.8003,
+      "step": 46623
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.5308377742767334,
+      "learning_rate": 1.4461964373060022e-06,
+      "loss": 0.6759,
+      "step": 46624
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.2327382564544678,
+      "learning_rate": 1.4460109761545026e-06,
+      "loss": 0.6783,
+      "step": 46625
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 4.57419490814209,
+      "learning_rate": 1.4458255248853142e-06,
+      "loss": 0.5576,
+      "step": 46626
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.5454790592193604,
+      "learning_rate": 1.4456400834989537e-06,
+      "loss": 0.3341,
+      "step": 46627
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 1.741686224937439,
+      "learning_rate": 1.4454546519959362e-06,
+      "loss": 0.6375,
+      "step": 46628
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 4.012472152709961,
+      "learning_rate": 1.4452692303767797e-06,
+      "loss": 0.5335,
+      "step": 46629
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 4.960309982299805,
+      "learning_rate": 1.4450838186419958e-06,
+      "loss": 0.6344,
+      "step": 46630
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.4181010723114014,
+      "learning_rate": 1.4448984167921038e-06,
+      "loss": 0.6898,
+      "step": 46631
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.5928564071655273,
+      "learning_rate": 1.444713024827616e-06,
+      "loss": 0.6012,
+      "step": 46632
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.3094658851623535,
+      "learning_rate": 1.4445276427490511e-06,
+      "loss": 0.6454,
+      "step": 46633
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.662287473678589,
+      "learning_rate": 1.4443422705569215e-06,
+      "loss": 0.439,
+      "step": 46634
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.312976598739624,
+      "learning_rate": 1.4441569082517454e-06,
+      "loss": 0.4391,
+      "step": 46635
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.040447473526001,
+      "learning_rate": 1.4439715558340356e-06,
+      "loss": 0.7887,
+      "step": 46636
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.3304765224456787,
+      "learning_rate": 1.4437862133043086e-06,
+      "loss": 0.7292,
+      "step": 46637
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 4.46419095993042,
+      "learning_rate": 1.4436008806630814e-06,
+      "loss": 0.5688,
+      "step": 46638
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.4030375480651855,
+      "learning_rate": 1.443415557910866e-06,
+      "loss": 0.7896,
+      "step": 46639
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 4.235069751739502,
+      "learning_rate": 1.4432302450481817e-06,
+      "loss": 0.6517,
+      "step": 46640
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 5.014414310455322,
+      "learning_rate": 1.4430449420755398e-06,
+      "loss": 0.6125,
+      "step": 46641
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.363795518875122,
+      "learning_rate": 1.4428596489934593e-06,
+      "loss": 0.4524,
+      "step": 46642
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.9878134727478027,
+      "learning_rate": 1.4426743658024522e-06,
+      "loss": 0.6131,
+      "step": 46643
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.3955094814300537,
+      "learning_rate": 1.4424890925030365e-06,
+      "loss": 0.4477,
+      "step": 46644
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.594163179397583,
+      "learning_rate": 1.4423038290957242e-06,
+      "loss": 0.7023,
+      "step": 46645
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.8379948139190674,
+      "learning_rate": 1.4421185755810342e-06,
+      "loss": 0.7009,
+      "step": 46646
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.9159042835235596,
+      "learning_rate": 1.4419333319594774e-06,
+      "loss": 0.6597,
+      "step": 46647
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.5870232582092285,
+      "learning_rate": 1.4417480982315713e-06,
+      "loss": 0.3236,
+      "step": 46648
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.4445724487304688,
+      "learning_rate": 1.4415628743978311e-06,
+      "loss": 0.703,
+      "step": 46649
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.3091840744018555,
+      "learning_rate": 1.4413776604587727e-06,
+      "loss": 0.4354,
+      "step": 46650
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.4365601539611816,
+      "learning_rate": 1.4411924564149087e-06,
+      "loss": 0.7512,
+      "step": 46651
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.6861207485198975,
+      "learning_rate": 1.441007262266757e-06,
+      "loss": 0.6525,
+      "step": 46652
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.161353349685669,
+      "learning_rate": 1.440822078014829e-06,
+      "loss": 0.615,
+      "step": 46653
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.826070785522461,
+      "learning_rate": 1.440636903659643e-06,
+      "loss": 0.6068,
+      "step": 46654
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.3852906227111816,
+      "learning_rate": 1.440451739201711e-06,
+      "loss": 0.7148,
+      "step": 46655
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.0265491008758545,
+      "learning_rate": 1.440266584641551e-06,
+      "loss": 0.5025,
+      "step": 46656
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.8340790271759033,
+      "learning_rate": 1.4400814399796742e-06,
+      "loss": 0.794,
+      "step": 46657
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.326796293258667,
+      "learning_rate": 1.4398963052165993e-06,
+      "loss": 0.5205,
+      "step": 46658
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.6899619102478027,
+      "learning_rate": 1.439711180352837e-06,
+      "loss": 0.6102,
+      "step": 46659
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.408561944961548,
+      "learning_rate": 1.4395260653889064e-06,
+      "loss": 0.8012,
+      "step": 46660
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.2654287815093994,
+      "learning_rate": 1.4393409603253177e-06,
+      "loss": 0.7438,
+      "step": 46661
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.286940336227417,
+      "learning_rate": 1.4391558651625903e-06,
+      "loss": 0.6092,
+      "step": 46662
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.875469446182251,
+      "learning_rate": 1.438970779901235e-06,
+      "loss": 0.538,
+      "step": 46663
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 1.7592549324035645,
+      "learning_rate": 1.4387857045417685e-06,
+      "loss": 0.5421,
+      "step": 46664
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.1472418308258057,
+      "learning_rate": 1.4386006390847041e-06,
+      "loss": 0.6381,
+      "step": 46665
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.3035435676574707,
+      "learning_rate": 1.4384155835305585e-06,
+      "loss": 0.5648,
+      "step": 46666
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 1.7005817890167236,
+      "learning_rate": 1.4382305378798438e-06,
+      "loss": 0.5529,
+      "step": 46667
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.287681818008423,
+      "learning_rate": 1.4380455021330768e-06,
+      "loss": 0.4898,
+      "step": 46668
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.0755162239074707,
+      "learning_rate": 1.4378604762907682e-06,
+      "loss": 0.5719,
+      "step": 46669
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 1.9870710372924805,
+      "learning_rate": 1.4376754603534382e-06,
+      "loss": 0.4472,
+      "step": 46670
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.7211458683013916,
+      "learning_rate": 1.4374904543215967e-06,
+      "loss": 0.5446,
+      "step": 46671
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.68399977684021,
+      "learning_rate": 1.4373054581957613e-06,
+      "loss": 0.7508,
+      "step": 46672
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.134981393814087,
+      "learning_rate": 1.437120471976443e-06,
+      "loss": 0.5442,
+      "step": 46673
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.304288148880005,
+      "learning_rate": 1.4369354956641597e-06,
+      "loss": 0.7765,
+      "step": 46674
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.3293585777282715,
+      "learning_rate": 1.4367505292594224e-06,
+      "loss": 0.4492,
+      "step": 46675
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.08996844291687,
+      "learning_rate": 1.4365655727627487e-06,
+      "loss": 0.6198,
+      "step": 46676
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.310284376144409,
+      "learning_rate": 1.4363806261746499e-06,
+      "loss": 0.4346,
+      "step": 46677
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.3329269886016846,
+      "learning_rate": 1.436195689495643e-06,
+      "loss": 0.6725,
+      "step": 46678
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.443297863006592,
+      "learning_rate": 1.4360107627262388e-06,
+      "loss": 0.36,
+      "step": 46679
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.5604617595672607,
+      "learning_rate": 1.4358258458669555e-06,
+      "loss": 0.4744,
+      "step": 46680
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.624454975128174,
+      "learning_rate": 1.4356409389183041e-06,
+      "loss": 0.5431,
+      "step": 46681
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.7919352054595947,
+      "learning_rate": 1.4354560418808011e-06,
+      "loss": 0.5004,
+      "step": 46682
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.6524434089660645,
+      "learning_rate": 1.4352711547549586e-06,
+      "loss": 0.6076,
+      "step": 46683
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.924072027206421,
+      "learning_rate": 1.4350862775412927e-06,
+      "loss": 0.6001,
+      "step": 46684
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.042372465133667,
+      "learning_rate": 1.4349014102403147e-06,
+      "loss": 0.4278,
+      "step": 46685
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.488504648208618,
+      "learning_rate": 1.4347165528525426e-06,
+      "loss": 0.7213,
+      "step": 46686
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.9262139797210693,
+      "learning_rate": 1.4345317053784862e-06,
+      "loss": 0.5522,
+      "step": 46687
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.477794647216797,
+      "learning_rate": 1.4343468678186612e-06,
+      "loss": 0.5213,
+      "step": 46688
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.3859047889709473,
+      "learning_rate": 1.4341620401735824e-06,
+      "loss": 0.8381,
+      "step": 46689
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.492414712905884,
+      "learning_rate": 1.4339772224437643e-06,
+      "loss": 0.7724,
+      "step": 46690
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.1568727493286133,
+      "learning_rate": 1.4337924146297183e-06,
+      "loss": 0.4916,
+      "step": 46691
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.3092775344848633,
+      "learning_rate": 1.4336076167319612e-06,
+      "loss": 0.58,
+      "step": 46692
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.215766668319702,
+      "learning_rate": 1.4334228287510039e-06,
+      "loss": 0.7052,
+      "step": 46693
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.102545738220215,
+      "learning_rate": 1.4332380506873629e-06,
+      "loss": 0.5416,
+      "step": 46694
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.426950693130493,
+      "learning_rate": 1.4330532825415495e-06,
+      "loss": 0.4688,
+      "step": 46695
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.927539825439453,
+      "learning_rate": 1.43286852431408e-06,
+      "loss": 0.6977,
+      "step": 46696
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.232959032058716,
+      "learning_rate": 1.4326837760054651e-06,
+      "loss": 0.43,
+      "step": 46697
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.9982056617736816,
+      "learning_rate": 1.4324990376162222e-06,
+      "loss": 0.6616,
+      "step": 46698
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.623466968536377,
+      "learning_rate": 1.4323143091468617e-06,
+      "loss": 0.6488,
+      "step": 46699
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.1400108337402344,
+      "learning_rate": 1.4321295905978983e-06,
+      "loss": 0.3777,
+      "step": 46700
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.850508213043213,
+      "learning_rate": 1.4319448819698473e-06,
+      "loss": 0.7389,
+      "step": 46701
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.5676074028015137,
+      "learning_rate": 1.43176018326322e-06,
+      "loss": 0.6865,
+      "step": 46702
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.0096731185913086,
+      "learning_rate": 1.4315754944785321e-06,
+      "loss": 0.4391,
+      "step": 46703
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.9595606327056885,
+      "learning_rate": 1.4313908156162942e-06,
+      "loss": 0.6597,
+      "step": 46704
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.1280810832977295,
+      "learning_rate": 1.431206146677024e-06,
+      "loss": 0.543,
+      "step": 46705
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.4578826427459717,
+      "learning_rate": 1.43102148766123e-06,
+      "loss": 0.4934,
+      "step": 46706
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.615142822265625,
+      "learning_rate": 1.4308368385694305e-06,
+      "loss": 0.3928,
+      "step": 46707
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 1.7345772981643677,
+      "learning_rate": 1.4306521994021349e-06,
+      "loss": 0.4221,
+      "step": 46708
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.343022584915161,
+      "learning_rate": 1.4304675701598586e-06,
+      "loss": 0.4662,
+      "step": 46709
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.0216658115386963,
+      "learning_rate": 1.4302829508431149e-06,
+      "loss": 0.4506,
+      "step": 46710
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.478487968444824,
+      "learning_rate": 1.4300983414524184e-06,
+      "loss": 0.4449,
+      "step": 46711
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 4.515380382537842,
+      "learning_rate": 1.4299137419882798e-06,
+      "loss": 0.4108,
+      "step": 46712
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 1.9539520740509033,
+      "learning_rate": 1.4297291524512146e-06,
+      "loss": 0.4641,
+      "step": 46713
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.0985066890716553,
+      "learning_rate": 1.429544572841734e-06,
+      "loss": 0.5684,
+      "step": 46714
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.542759656906128,
+      "learning_rate": 1.4293600031603544e-06,
+      "loss": 0.5005,
+      "step": 46715
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.5660219192504883,
+      "learning_rate": 1.429175443407585e-06,
+      "loss": 0.7033,
+      "step": 46716
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.050602436065674,
+      "learning_rate": 1.4289908935839425e-06,
+      "loss": 0.4012,
+      "step": 46717
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.8576009273529053,
+      "learning_rate": 1.4288063536899376e-06,
+      "loss": 0.5847,
+      "step": 46718
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.466158628463745,
+      "learning_rate": 1.4286218237260858e-06,
+      "loss": 0.4831,
+      "step": 46719
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.4812076091766357,
+      "learning_rate": 1.4284373036928967e-06,
+      "loss": 0.3793,
+      "step": 46720
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.7508513927459717,
+      "learning_rate": 1.4282527935908873e-06,
+      "loss": 0.6702,
+      "step": 46721
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.669419050216675,
+      "learning_rate": 1.4280682934205682e-06,
+      "loss": 0.4685,
+      "step": 46722
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.8782310485839844,
+      "learning_rate": 1.4278838031824538e-06,
+      "loss": 0.5728,
+      "step": 46723
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.65167236328125,
+      "learning_rate": 1.4276993228770547e-06,
+      "loss": 0.4339,
+      "step": 46724
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.7309396266937256,
+      "learning_rate": 1.4275148525048877e-06,
+      "loss": 0.6822,
+      "step": 46725
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.3062920570373535,
+      "learning_rate": 1.4273303920664616e-06,
+      "loss": 0.4424,
+      "step": 46726
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.0323894023895264,
+      "learning_rate": 1.4271459415622928e-06,
+      "loss": 0.612,
+      "step": 46727
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.4033961296081543,
+      "learning_rate": 1.4269615009928918e-06,
+      "loss": 0.599,
+      "step": 46728
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.9761955738067627,
+      "learning_rate": 1.4267770703587736e-06,
+      "loss": 0.6442,
+      "step": 46729
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.7025864124298096,
+      "learning_rate": 1.4265926496604466e-06,
+      "loss": 0.5396,
+      "step": 46730
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.4027557373046875,
+      "learning_rate": 1.42640823889843e-06,
+      "loss": 0.7183,
+      "step": 46731
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.48837947845459,
+      "learning_rate": 1.4262238380732318e-06,
+      "loss": 0.5706,
+      "step": 46732
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.1769068241119385,
+      "learning_rate": 1.426039447185368e-06,
+      "loss": 0.599,
+      "step": 46733
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 4.664279460906982,
+      "learning_rate": 1.4258550662353477e-06,
+      "loss": 0.7115,
+      "step": 46734
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.631319046020508,
+      "learning_rate": 1.4256706952236877e-06,
+      "loss": 0.5594,
+      "step": 46735
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.658356189727783,
+      "learning_rate": 1.4254863341508967e-06,
+      "loss": 0.4334,
+      "step": 46736
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 4.040036201477051,
+      "learning_rate": 1.4253019830174908e-06,
+      "loss": 0.7626,
+      "step": 46737
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.376293420791626,
+      "learning_rate": 1.4251176418239793e-06,
+      "loss": 0.5725,
+      "step": 46738
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.35774827003479,
+      "learning_rate": 1.424933310570878e-06,
+      "loss": 0.5407,
+      "step": 46739
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.3197271823883057,
+      "learning_rate": 1.4247489892586964e-06,
+      "loss": 0.6659,
+      "step": 46740
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.268648386001587,
+      "learning_rate": 1.4245646778879501e-06,
+      "loss": 0.3879,
+      "step": 46741
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.2573506832122803,
+      "learning_rate": 1.4243803764591484e-06,
+      "loss": 0.5842,
+      "step": 46742
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.8808352947235107,
+      "learning_rate": 1.424196084972807e-06,
+      "loss": 0.6262,
+      "step": 46743
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.55226731300354,
+      "learning_rate": 1.4240118034294353e-06,
+      "loss": 0.7808,
+      "step": 46744
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.042847156524658,
+      "learning_rate": 1.4238275318295485e-06,
+      "loss": 0.4707,
+      "step": 46745
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.0724992752075195,
+      "learning_rate": 1.4236432701736564e-06,
+      "loss": 0.6608,
+      "step": 46746
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.7877182960510254,
+      "learning_rate": 1.4234590184622744e-06,
+      "loss": 0.4858,
+      "step": 46747
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.534283399581909,
+      "learning_rate": 1.423274776695911e-06,
+      "loss": 0.61,
+      "step": 46748
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.200352907180786,
+      "learning_rate": 1.423090544875081e-06,
+      "loss": 0.4806,
+      "step": 46749
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.0055251121520996,
+      "learning_rate": 1.422906323000296e-06,
+      "loss": 0.6908,
+      "step": 46750
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.643441915512085,
+      "learning_rate": 1.4227221110720702e-06,
+      "loss": 0.7857,
+      "step": 46751
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.885974645614624,
+      "learning_rate": 1.422537909090912e-06,
+      "loss": 0.4952,
+      "step": 46752
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.7775514125823975,
+      "learning_rate": 1.422353717057338e-06,
+      "loss": 0.4258,
+      "step": 46753
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.2281453609466553,
+      "learning_rate": 1.422169534971856e-06,
+      "loss": 0.75,
+      "step": 46754
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.4839093685150146,
+      "learning_rate": 1.4219853628349822e-06,
+      "loss": 0.7209,
+      "step": 46755
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.3707218170166016,
+      "learning_rate": 1.4218012006472253e-06,
+      "loss": 0.6083,
+      "step": 46756
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.7143120765686035,
+      "learning_rate": 1.4216170484091003e-06,
+      "loss": 0.4479,
+      "step": 46757
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.6541285514831543,
+      "learning_rate": 1.4214329061211158e-06,
+      "loss": 0.7024,
+      "step": 46758
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.2864654064178467,
+      "learning_rate": 1.4212487737837876e-06,
+      "loss": 0.5579,
+      "step": 46759
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.5397050380706787,
+      "learning_rate": 1.4210646513976244e-06,
+      "loss": 0.5862,
+      "step": 46760
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.3069567680358887,
+      "learning_rate": 1.4208805389631397e-06,
+      "loss": 0.4934,
+      "step": 46761
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.284119129180908,
+      "learning_rate": 1.4206964364808473e-06,
+      "loss": 0.3968,
+      "step": 46762
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.0168933868408203,
+      "learning_rate": 1.4205123439512559e-06,
+      "loss": 0.5043,
+      "step": 46763
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.9530162811279297,
+      "learning_rate": 1.4203282613748797e-06,
+      "loss": 0.6783,
+      "step": 46764
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.3871591091156006,
+      "learning_rate": 1.4201441887522283e-06,
+      "loss": 0.592,
+      "step": 46765
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 4.109877109527588,
+      "learning_rate": 1.4199601260838164e-06,
+      "loss": 0.8315,
+      "step": 46766
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.1155004501342773,
+      "learning_rate": 1.4197760733701526e-06,
+      "loss": 0.8089,
+      "step": 46767
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 1.8519643545150757,
+      "learning_rate": 1.4195920306117522e-06,
+      "loss": 0.5122,
+      "step": 46768
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.531193971633911,
+      "learning_rate": 1.4194079978091236e-06,
+      "loss": 0.6146,
+      "step": 46769
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.326014995574951,
+      "learning_rate": 1.4192239749627796e-06,
+      "loss": 0.5467,
+      "step": 46770
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.8312010765075684,
+      "learning_rate": 1.4190399620732343e-06,
+      "loss": 0.707,
+      "step": 46771
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.507441520690918,
+      "learning_rate": 1.4188559591409955e-06,
+      "loss": 0.4633,
+      "step": 46772
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.560157299041748,
+      "learning_rate": 1.418671966166577e-06,
+      "loss": 0.4944,
+      "step": 46773
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.136263132095337,
+      "learning_rate": 1.4184879831504917e-06,
+      "loss": 0.5429,
+      "step": 46774
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 1.6667848825454712,
+      "learning_rate": 1.4183040100932477e-06,
+      "loss": 0.6131,
+      "step": 46775
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 9.74250602722168,
+      "learning_rate": 1.4181200469953604e-06,
+      "loss": 0.4669,
+      "step": 46776
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.235163927078247,
+      "learning_rate": 1.4179360938573383e-06,
+      "loss": 0.5448,
+      "step": 46777
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 4.294399261474609,
+      "learning_rate": 1.4177521506796949e-06,
+      "loss": 0.6544,
+      "step": 46778
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.5815188884735107,
+      "learning_rate": 1.4175682174629395e-06,
+      "loss": 0.6132,
+      "step": 46779
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.289215326309204,
+      "learning_rate": 1.4173842942075866e-06,
+      "loss": 0.7114,
+      "step": 46780
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.1599628925323486,
+      "learning_rate": 1.4172003809141443e-06,
+      "loss": 0.6583,
+      "step": 46781
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 4.501870632171631,
+      "learning_rate": 1.4170164775831274e-06,
+      "loss": 0.4872,
+      "step": 46782
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.94315767288208,
+      "learning_rate": 1.4168325842150432e-06,
+      "loss": 0.5825,
+      "step": 46783
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.22061824798584,
+      "learning_rate": 1.4166487008104074e-06,
+      "loss": 0.898,
+      "step": 46784
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.7072558403015137,
+      "learning_rate": 1.4164648273697279e-06,
+      "loss": 0.7463,
+      "step": 46785
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 1.9789235591888428,
+      "learning_rate": 1.4162809638935183e-06,
+      "loss": 0.5247,
+      "step": 46786
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.604905366897583,
+      "learning_rate": 1.4160971103822873e-06,
+      "loss": 0.5965,
+      "step": 46787
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.549229145050049,
+      "learning_rate": 1.4159132668365493e-06,
+      "loss": 0.4789,
+      "step": 46788
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.4633305072784424,
+      "learning_rate": 1.4157294332568128e-06,
+      "loss": 0.4873,
+      "step": 46789
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 17.002544403076172,
+      "learning_rate": 1.415545609643591e-06,
+      "loss": 0.4296,
+      "step": 46790
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.013721227645874,
+      "learning_rate": 1.4153617959973915e-06,
+      "loss": 0.6707,
+      "step": 46791
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.838503122329712,
+      "learning_rate": 1.4151779923187303e-06,
+      "loss": 0.7887,
+      "step": 46792
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.711577892303467,
+      "learning_rate": 1.4149941986081156e-06,
+      "loss": 0.4805,
+      "step": 46793
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.592637777328491,
+      "learning_rate": 1.4148104148660601e-06,
+      "loss": 0.8361,
+      "step": 46794
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.3217849731445312,
+      "learning_rate": 1.414626641093072e-06,
+      "loss": 0.5472,
+      "step": 46795
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.9744248390197754,
+      "learning_rate": 1.4144428772896662e-06,
+      "loss": 0.4686,
+      "step": 46796
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.801664113998413,
+      "learning_rate": 1.4142591234563496e-06,
+      "loss": 0.6497,
+      "step": 46797
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.6756279468536377,
+      "learning_rate": 1.4140753795936368e-06,
+      "loss": 0.7152,
+      "step": 46798
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.478402614593506,
+      "learning_rate": 1.4138916457020357e-06,
+      "loss": 0.4581,
+      "step": 46799
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.6059725284576416,
+      "learning_rate": 1.4137079217820604e-06,
+      "loss": 0.8266,
+      "step": 46800
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.28824520111084,
+      "learning_rate": 1.4135242078342177e-06,
+      "loss": 0.6931,
+      "step": 46801
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.5988337993621826,
+      "learning_rate": 1.4133405038590226e-06,
+      "loss": 0.6405,
+      "step": 46802
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.722485303878784,
+      "learning_rate": 1.413156809856982e-06,
+      "loss": 0.4616,
+      "step": 46803
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 7.414172649383545,
+      "learning_rate": 1.4129731258286105e-06,
+      "loss": 0.6603,
+      "step": 46804
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.120701313018799,
+      "learning_rate": 1.4127894517744155e-06,
+      "loss": 0.8276,
+      "step": 46805
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.430741786956787,
+      "learning_rate": 1.4126057876949107e-06,
+      "loss": 0.6121,
+      "step": 46806
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.377676248550415,
+      "learning_rate": 1.4124221335906036e-06,
+      "loss": 0.4348,
+      "step": 46807
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.179274320602417,
+      "learning_rate": 1.4122384894620084e-06,
+      "loss": 0.6656,
+      "step": 46808
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.4191112518310547,
+      "learning_rate": 1.4120548553096325e-06,
+      "loss": 0.7508,
+      "step": 46809
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.125885486602783,
+      "learning_rate": 1.4118712311339893e-06,
+      "loss": 0.7116,
+      "step": 46810
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 10.089384078979492,
+      "learning_rate": 1.4116876169355853e-06,
+      "loss": 0.2631,
+      "step": 46811
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.357147216796875,
+      "learning_rate": 1.4115040127149366e-06,
+      "loss": 0.8877,
+      "step": 46812
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.3340671062469482,
+      "learning_rate": 1.4113204184725494e-06,
+      "loss": 0.5062,
+      "step": 46813
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.839843273162842,
+      "learning_rate": 1.4111368342089371e-06,
+      "loss": 0.6439,
+      "step": 46814
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.3165063858032227,
+      "learning_rate": 1.410953259924608e-06,
+      "loss": 0.6171,
+      "step": 46815
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.3524551391601562,
+      "learning_rate": 1.4107696956200745e-06,
+      "loss": 0.7342,
+      "step": 46816
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.6953396797180176,
+      "learning_rate": 1.4105861412958444e-06,
+      "loss": 0.6601,
+      "step": 46817
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.214449644088745,
+      "learning_rate": 1.410402596952431e-06,
+      "loss": 0.6231,
+      "step": 46818
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.108156442642212,
+      "learning_rate": 1.4102190625903417e-06,
+      "loss": 0.6421,
+      "step": 46819
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.4653618335723877,
+      "learning_rate": 1.4100355382100904e-06,
+      "loss": 0.5497,
+      "step": 46820
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.838777542114258,
+      "learning_rate": 1.4098520238121839e-06,
+      "loss": 0.6767,
+      "step": 46821
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.105832099914551,
+      "learning_rate": 1.4096685193971332e-06,
+      "loss": 0.4393,
+      "step": 46822
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.567868232727051,
+      "learning_rate": 1.4094850249654519e-06,
+      "loss": 0.5274,
+      "step": 46823
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 4.360894203186035,
+      "learning_rate": 1.4093015405176458e-06,
+      "loss": 0.73,
+      "step": 46824
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 4.181014060974121,
+      "learning_rate": 1.409118066054228e-06,
+      "loss": 0.5586,
+      "step": 46825
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 4.570506572723389,
+      "learning_rate": 1.4089346015757071e-06,
+      "loss": 0.8513,
+      "step": 46826
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.688525915145874,
+      "learning_rate": 1.408751147082595e-06,
+      "loss": 0.7079,
+      "step": 46827
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.729733943939209,
+      "learning_rate": 1.4085677025753991e-06,
+      "loss": 0.8505,
+      "step": 46828
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.201887369155884,
+      "learning_rate": 1.4083842680546322e-06,
+      "loss": 0.6226,
+      "step": 46829
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.175732135772705,
+      "learning_rate": 1.4082008435208021e-06,
+      "loss": 0.6419,
+      "step": 46830
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.377363920211792,
+      "learning_rate": 1.4080174289744197e-06,
+      "loss": 0.7296,
+      "step": 46831
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.5693962574005127,
+      "learning_rate": 1.4078340244159972e-06,
+      "loss": 0.5179,
+      "step": 46832
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 1.9792108535766602,
+      "learning_rate": 1.4076506298460407e-06,
+      "loss": 0.6123,
+      "step": 46833
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 1.7057604789733887,
+      "learning_rate": 1.4074672452650618e-06,
+      "loss": 0.4312,
+      "step": 46834
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.6846201419830322,
+      "learning_rate": 1.4072838706735725e-06,
+      "loss": 0.4577,
+      "step": 46835
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.51987624168396,
+      "learning_rate": 1.40710050607208e-06,
+      "loss": 0.4908,
+      "step": 46836
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 1.9966557025909424,
+      "learning_rate": 1.4069171514610958e-06,
+      "loss": 0.6251,
+      "step": 46837
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.3286020755767822,
+      "learning_rate": 1.4067338068411275e-06,
+      "loss": 0.6601,
+      "step": 46838
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.853025197982788,
+      "learning_rate": 1.4065504722126882e-06,
+      "loss": 0.5854,
+      "step": 46839
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.1946160793304443,
+      "learning_rate": 1.406367147576284e-06,
+      "loss": 0.6103,
+      "step": 46840
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.395425319671631,
+      "learning_rate": 1.4061838329324289e-06,
+      "loss": 0.4135,
+      "step": 46841
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.2390706539154053,
+      "learning_rate": 1.406000528281628e-06,
+      "loss": 0.5172,
+      "step": 46842
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.3714535236358643,
+      "learning_rate": 1.4058172336243953e-06,
+      "loss": 0.429,
+      "step": 46843
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.98354172706604,
+      "learning_rate": 1.405633948961237e-06,
+      "loss": 0.6409,
+      "step": 46844
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 1.9185270071029663,
+      "learning_rate": 1.4054506742926649e-06,
+      "loss": 0.4051,
+      "step": 46845
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.3323371410369873,
+      "learning_rate": 1.405267409619187e-06,
+      "loss": 0.3992,
+      "step": 46846
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.0143790245056152,
+      "learning_rate": 1.4050841549413151e-06,
+      "loss": 0.5395,
+      "step": 46847
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.2419064044952393,
+      "learning_rate": 1.4049009102595567e-06,
+      "loss": 0.7019,
+      "step": 46848
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.5486459732055664,
+      "learning_rate": 1.404717675574423e-06,
+      "loss": 0.5058,
+      "step": 46849
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.5740115642547607,
+      "learning_rate": 1.4045344508864211e-06,
+      "loss": 0.5968,
+      "step": 46850
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.8139467239379883,
+      "learning_rate": 1.4043512361960638e-06,
+      "loss": 0.6146,
+      "step": 46851
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 6.037817478179932,
+      "learning_rate": 1.4041680315038552e-06,
+      "loss": 0.7122,
+      "step": 46852
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 4.664681434631348,
+      "learning_rate": 1.4039848368103116e-06,
+      "loss": 0.516,
+      "step": 46853
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.2690539360046387,
+      "learning_rate": 1.4038016521159375e-06,
+      "loss": 0.4602,
+      "step": 46854
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.1642255783081055,
+      "learning_rate": 1.4036184774212458e-06,
+      "loss": 0.7083,
+      "step": 46855
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.261831760406494,
+      "learning_rate": 1.403435312726742e-06,
+      "loss": 0.6419,
+      "step": 46856
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.7334251403808594,
+      "learning_rate": 1.4032521580329388e-06,
+      "loss": 0.5673,
+      "step": 46857
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.031355142593384,
+      "learning_rate": 1.4030690133403424e-06,
+      "loss": 0.7001,
+      "step": 46858
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.825557231903076,
+      "learning_rate": 1.4028858786494654e-06,
+      "loss": 0.7988,
+      "step": 46859
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 4.959197521209717,
+      "learning_rate": 1.4027027539608135e-06,
+      "loss": 0.6365,
+      "step": 46860
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.94449520111084,
+      "learning_rate": 1.4025196392748991e-06,
+      "loss": 0.6777,
+      "step": 46861
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.7369377613067627,
+      "learning_rate": 1.4023365345922286e-06,
+      "loss": 0.8034,
+      "step": 46862
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.839956521987915,
+      "learning_rate": 1.4021534399133142e-06,
+      "loss": 0.6515,
+      "step": 46863
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.918104887008667,
+      "learning_rate": 1.401970355238661e-06,
+      "loss": 0.6128,
+      "step": 46864
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.328584671020508,
+      "learning_rate": 1.4017872805687827e-06,
+      "loss": 0.3778,
+      "step": 46865
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.6813642978668213,
+      "learning_rate": 1.4016042159041844e-06,
+      "loss": 0.5077,
+      "step": 46866
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.227424144744873,
+      "learning_rate": 1.4014211612453783e-06,
+      "loss": 0.4864,
+      "step": 46867
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.206272602081299,
+      "learning_rate": 1.4012381165928701e-06,
+      "loss": 0.5026,
+      "step": 46868
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.5719189643859863,
+      "learning_rate": 1.401055081947172e-06,
+      "loss": 0.5387,
+      "step": 46869
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.4669957160949707,
+      "learning_rate": 1.4008720573087903e-06,
+      "loss": 0.3539,
+      "step": 46870
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.93078351020813,
+      "learning_rate": 1.4006890426782365e-06,
+      "loss": 0.5808,
+      "step": 46871
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.3131589889526367,
+      "learning_rate": 1.4005060380560154e-06,
+      "loss": 0.3828,
+      "step": 46872
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.066713333129883,
+      "learning_rate": 1.4003230434426412e-06,
+      "loss": 0.6763,
+      "step": 46873
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.2296805381774902,
+      "learning_rate": 1.4001400588386194e-06,
+      "loss": 0.4648,
+      "step": 46874
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 1.912635326385498,
+      "learning_rate": 1.3999570842444604e-06,
+      "loss": 0.4702,
+      "step": 46875
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.218669891357422,
+      "learning_rate": 1.3997741196606706e-06,
+      "loss": 0.366,
+      "step": 46876
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.5506253242492676,
+      "learning_rate": 1.3995911650877619e-06,
+      "loss": 0.613,
+      "step": 46877
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.189552068710327,
+      "learning_rate": 1.3994082205262405e-06,
+      "loss": 0.606,
+      "step": 46878
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.11104416847229,
+      "learning_rate": 1.399225285976617e-06,
+      "loss": 0.5202,
+      "step": 46879
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.8927204608917236,
+      "learning_rate": 1.3990423614393972e-06,
+      "loss": 0.5966,
+      "step": 46880
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.0818333625793457,
+      "learning_rate": 1.398859446915094e-06,
+      "loss": 0.6571,
+      "step": 46881
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.9084956645965576,
+      "learning_rate": 1.3986765424042114e-06,
+      "loss": 0.5458,
+      "step": 46882
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 1.8100426197052002,
+      "learning_rate": 1.3984936479072625e-06,
+      "loss": 0.4737,
+      "step": 46883
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.8374738693237305,
+      "learning_rate": 1.3983107634247517e-06,
+      "loss": 0.5984,
+      "step": 46884
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.9610955715179443,
+      "learning_rate": 1.3981278889571891e-06,
+      "loss": 0.6351,
+      "step": 46885
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.3123154640197754,
+      "learning_rate": 1.3979450245050858e-06,
+      "loss": 0.5694,
+      "step": 46886
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 5.634582996368408,
+      "learning_rate": 1.397762170068946e-06,
+      "loss": 0.6163,
+      "step": 46887
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.4827823638916016,
+      "learning_rate": 1.3975793256492815e-06,
+      "loss": 0.8083,
+      "step": 46888
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.3247084617614746,
+      "learning_rate": 1.3973964912465982e-06,
+      "loss": 0.6866,
+      "step": 46889
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.5816144943237305,
+      "learning_rate": 1.3972136668614073e-06,
+      "loss": 0.6307,
+      "step": 46890
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.480745792388916,
+      "learning_rate": 1.397030852494214e-06,
+      "loss": 0.697,
+      "step": 46891
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.432243824005127,
+      "learning_rate": 1.3968480481455282e-06,
+      "loss": 0.6552,
+      "step": 46892
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.7518868446350098,
+      "learning_rate": 1.3966652538158598e-06,
+      "loss": 0.6852,
+      "step": 46893
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.6583898067474365,
+      "learning_rate": 1.396482469505714e-06,
+      "loss": 0.3791,
+      "step": 46894
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.392282247543335,
+      "learning_rate": 1.3962996952156016e-06,
+      "loss": 0.7604,
+      "step": 46895
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.0194029808044434,
+      "learning_rate": 1.396116930946028e-06,
+      "loss": 0.5556,
+      "step": 46896
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.0989878177642822,
+      "learning_rate": 1.3959341766975043e-06,
+      "loss": 0.5029,
+      "step": 46897
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 1.8366013765335083,
+      "learning_rate": 1.3957514324705385e-06,
+      "loss": 0.3639,
+      "step": 46898
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.752192258834839,
+      "learning_rate": 1.3955686982656359e-06,
+      "loss": 0.7736,
+      "step": 46899
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.777104139328003,
+      "learning_rate": 1.3953859740833086e-06,
+      "loss": 0.5632,
+      "step": 46900
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.439727544784546,
+      "learning_rate": 1.3952032599240605e-06,
+      "loss": 0.5798,
+      "step": 46901
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.2616167068481445,
+      "learning_rate": 1.395020555788404e-06,
+      "loss": 0.6077,
+      "step": 46902
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.097346782684326,
+      "learning_rate": 1.3948378616768426e-06,
+      "loss": 0.5875,
+      "step": 46903
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.4576869010925293,
+      "learning_rate": 1.3946551775898887e-06,
+      "loss": 0.6302,
+      "step": 46904
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 1.8506871461868286,
+      "learning_rate": 1.3944725035280465e-06,
+      "loss": 0.5131,
+      "step": 46905
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.484020948410034,
+      "learning_rate": 1.394289839491827e-06,
+      "loss": 0.772,
+      "step": 46906
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.5310170650482178,
+      "learning_rate": 1.394107185481735e-06,
+      "loss": 0.8649,
+      "step": 46907
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.248180627822876,
+      "learning_rate": 1.3939245414982823e-06,
+      "loss": 0.4305,
+      "step": 46908
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.0599653720855713,
+      "learning_rate": 1.3937419075419723e-06,
+      "loss": 0.4287,
+      "step": 46909
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.2456014156341553,
+      "learning_rate": 1.3935592836133172e-06,
+      "loss": 0.5057,
+      "step": 46910
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.4500694274902344,
+      "learning_rate": 1.3933766697128208e-06,
+      "loss": 0.6725,
+      "step": 46911
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 1.8681738376617432,
+      "learning_rate": 1.3931940658409943e-06,
+      "loss": 0.5103,
+      "step": 46912
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.261375665664673,
+      "learning_rate": 1.3930114719983412e-06,
+      "loss": 0.376,
+      "step": 46913
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.6223268508911133,
+      "learning_rate": 1.3928288881853753e-06,
+      "loss": 0.7354,
+      "step": 46914
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.4841856956481934,
+      "learning_rate": 1.3926463144025987e-06,
+      "loss": 0.7886,
+      "step": 46915
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.2471160888671875,
+      "learning_rate": 1.3924637506505234e-06,
+      "loss": 0.4765,
+      "step": 46916
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.5214593410491943,
+      "learning_rate": 1.3922811969296536e-06,
+      "loss": 0.5795,
+      "step": 46917
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.110973596572876,
+      "learning_rate": 1.3920986532404996e-06,
+      "loss": 0.6389,
+      "step": 46918
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.7444615364074707,
+      "learning_rate": 1.3919161195835656e-06,
+      "loss": 0.8996,
+      "step": 46919
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.811333656311035,
+      "learning_rate": 1.3917335959593637e-06,
+      "loss": 0.613,
+      "step": 46920
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.1558029651641846,
+      "learning_rate": 1.391551082368397e-06,
+      "loss": 0.5359,
+      "step": 46921
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 4.154633522033691,
+      "learning_rate": 1.3913685788111769e-06,
+      "loss": 0.7254,
+      "step": 46922
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.5478546619415283,
+      "learning_rate": 1.391186085288207e-06,
+      "loss": 0.8237,
+      "step": 46923
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.8966329097747803,
+      "learning_rate": 1.3910036017999984e-06,
+      "loss": 0.4012,
+      "step": 46924
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.612729549407959,
+      "learning_rate": 1.390821128347055e-06,
+      "loss": 0.6324,
+      "step": 46925
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 4.7107133865356445,
+      "learning_rate": 1.3906386649298882e-06,
+      "loss": 0.7038,
+      "step": 46926
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.3922348022460938,
+      "learning_rate": 1.3904562115490012e-06,
+      "loss": 0.6795,
+      "step": 46927
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.202094316482544,
+      "learning_rate": 1.390273768204905e-06,
+      "loss": 0.7218,
+      "step": 46928
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.8344311714172363,
+      "learning_rate": 1.3900913348981033e-06,
+      "loss": 0.5625,
+      "step": 46929
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.498753309249878,
+      "learning_rate": 1.389908911629107e-06,
+      "loss": 0.4343,
+      "step": 46930
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.2243759632110596,
+      "learning_rate": 1.3897264983984204e-06,
+      "loss": 0.8157,
+      "step": 46931
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.945258855819702,
+      "learning_rate": 1.3895440952065532e-06,
+      "loss": 0.7135,
+      "step": 46932
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 5.5160980224609375,
+      "learning_rate": 1.3893617020540085e-06,
+      "loss": 0.6347,
+      "step": 46933
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.5654916763305664,
+      "learning_rate": 1.3891793189412994e-06,
+      "loss": 0.6255,
+      "step": 46934
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.818699359893799,
+      "learning_rate": 1.3889969458689283e-06,
+      "loss": 0.5686,
+      "step": 46935
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.0571107864379883,
+      "learning_rate": 1.3888145828374055e-06,
+      "loss": 0.55,
+      "step": 46936
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 4.2861857414245605,
+      "learning_rate": 1.3886322298472348e-06,
+      "loss": 0.5069,
+      "step": 46937
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 4.533139705657959,
+      "learning_rate": 1.388449886898927e-06,
+      "loss": 0.7752,
+      "step": 46938
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.3379743099212646,
+      "learning_rate": 1.3882675539929857e-06,
+      "loss": 0.8332,
+      "step": 46939
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.9038381576538086,
+      "learning_rate": 1.3880852311299209e-06,
+      "loss": 0.6804,
+      "step": 46940
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.9601054191589355,
+      "learning_rate": 1.387902918310236e-06,
+      "loss": 0.464,
+      "step": 46941
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 4.960270404815674,
+      "learning_rate": 1.387720615534442e-06,
+      "loss": 0.3773,
+      "step": 46942
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.8027145862579346,
+      "learning_rate": 1.3875383228030414e-06,
+      "loss": 0.7283,
+      "step": 46943
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.2626590728759766,
+      "learning_rate": 1.3873560401165458e-06,
+      "loss": 0.6893,
+      "step": 46944
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.1850225925445557,
+      "learning_rate": 1.387173767475458e-06,
+      "loss": 0.6729,
+      "step": 46945
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.4544899463653564,
+      "learning_rate": 1.3869915048802863e-06,
+      "loss": 0.6464,
+      "step": 46946
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.204441785812378,
+      "learning_rate": 1.3868092523315396e-06,
+      "loss": 0.6819,
+      "step": 46947
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.3236207962036133,
+      "learning_rate": 1.3866270098297213e-06,
+      "loss": 0.8298,
+      "step": 46948
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.6941521167755127,
+      "learning_rate": 1.386444777375341e-06,
+      "loss": 0.7593,
+      "step": 46949
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 1.9949398040771484,
+      "learning_rate": 1.3862625549689029e-06,
+      "loss": 0.5871,
+      "step": 46950
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.381385087966919,
+      "learning_rate": 1.386080342610916e-06,
+      "loss": 0.6575,
+      "step": 46951
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.124382495880127,
+      "learning_rate": 1.3858981403018846e-06,
+      "loss": 0.7031,
+      "step": 46952
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.3526899814605713,
+      "learning_rate": 1.3857159480423167e-06,
+      "loss": 0.5139,
+      "step": 46953
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.4702720642089844,
+      "learning_rate": 1.38553376583272e-06,
+      "loss": 0.5617,
+      "step": 46954
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 4.412678241729736,
+      "learning_rate": 1.3853515936735989e-06,
+      "loss": 0.9707,
+      "step": 46955
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.4164445400238037,
+      "learning_rate": 1.3851694315654619e-06,
+      "loss": 0.4803,
+      "step": 46956
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.518538475036621,
+      "learning_rate": 1.3849872795088131e-06,
+      "loss": 0.475,
+      "step": 46957
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.9604194164276123,
+      "learning_rate": 1.384805137504161e-06,
+      "loss": 0.6001,
+      "step": 46958
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.649973154067993,
+      "learning_rate": 1.3846230055520126e-06,
+      "loss": 0.818,
+      "step": 46959
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.813009738922119,
+      "learning_rate": 1.3844408836528716e-06,
+      "loss": 0.6457,
+      "step": 46960
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.910101890563965,
+      "learning_rate": 1.384258771807248e-06,
+      "loss": 0.5858,
+      "step": 46961
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.316710948944092,
+      "learning_rate": 1.3840766700156444e-06,
+      "loss": 0.4271,
+      "step": 46962
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.0833561420440674,
+      "learning_rate": 1.3838945782785712e-06,
+      "loss": 0.6381,
+      "step": 46963
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.1297850608825684,
+      "learning_rate": 1.3837124965965304e-06,
+      "loss": 0.9547,
+      "step": 46964
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.2347564697265625,
+      "learning_rate": 1.383530424970032e-06,
+      "loss": 0.5672,
+      "step": 46965
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.5193703174591064,
+      "learning_rate": 1.3833483633995793e-06,
+      "loss": 0.7809,
+      "step": 46966
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.26564359664917,
+      "learning_rate": 1.3831663118856814e-06,
+      "loss": 0.4947,
+      "step": 46967
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.677522659301758,
+      "learning_rate": 1.382984270428842e-06,
+      "loss": 0.5829,
+      "step": 46968
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.5219953060150146,
+      "learning_rate": 1.3828022390295698e-06,
+      "loss": 0.6181,
+      "step": 46969
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.1795389652252197,
+      "learning_rate": 1.3826202176883674e-06,
+      "loss": 0.5096,
+      "step": 46970
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 1.8872395753860474,
+      "learning_rate": 1.382438206405745e-06,
+      "loss": 0.4251,
+      "step": 46971
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.6161251068115234,
+      "learning_rate": 1.3822562051822052e-06,
+      "loss": 0.5657,
+      "step": 46972
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 4.0085601806640625,
+      "learning_rate": 1.3820742140182575e-06,
+      "loss": 0.7254,
+      "step": 46973
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 1.9195276498794556,
+      "learning_rate": 1.381892232914403e-06,
+      "loss": 0.5168,
+      "step": 46974
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.5224156379699707,
+      "learning_rate": 1.3817102618711536e-06,
+      "loss": 0.4002,
+      "step": 46975
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.972034454345703,
+      "learning_rate": 1.3815283008890106e-06,
+      "loss": 0.5196,
+      "step": 46976
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.660768508911133,
+      "learning_rate": 1.3813463499684842e-06,
+      "loss": 0.6333,
+      "step": 46977
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.3382203578948975,
+      "learning_rate": 1.3811644091100761e-06,
+      "loss": 0.6264,
+      "step": 46978
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 4.42432165145874,
+      "learning_rate": 1.3809824783142962e-06,
+      "loss": 0.7731,
+      "step": 46979
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.5684125423431396,
+      "learning_rate": 1.3808005575816459e-06,
+      "loss": 0.6603,
+      "step": 46980
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.410889148712158,
+      "learning_rate": 1.3806186469126353e-06,
+      "loss": 0.4329,
+      "step": 46981
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.094660758972168,
+      "learning_rate": 1.3804367463077674e-06,
+      "loss": 0.6253,
+      "step": 46982
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.1269822120666504,
+      "learning_rate": 1.3802548557675499e-06,
+      "loss": 0.5165,
+      "step": 46983
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.9031147956848145,
+      "learning_rate": 1.3800729752924862e-06,
+      "loss": 0.6641,
+      "step": 46984
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.3563859462738037,
+      "learning_rate": 1.3798911048830855e-06,
+      "loss": 0.5052,
+      "step": 46985
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 5.37381649017334,
+      "learning_rate": 1.3797092445398496e-06,
+      "loss": 0.6086,
+      "step": 46986
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.907336950302124,
+      "learning_rate": 1.3795273942632874e-06,
+      "loss": 0.6534,
+      "step": 46987
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.7646000385284424,
+      "learning_rate": 1.379345554053902e-06,
+      "loss": 0.6937,
+      "step": 46988
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.7166638374328613,
+      "learning_rate": 1.3791637239122018e-06,
+      "loss": 0.6824,
+      "step": 46989
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.236438274383545,
+      "learning_rate": 1.3789819038386892e-06,
+      "loss": 0.7122,
+      "step": 46990
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.9955806732177734,
+      "learning_rate": 1.3788000938338735e-06,
+      "loss": 0.5936,
+      "step": 46991
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.9099390506744385,
+      "learning_rate": 1.3786182938982562e-06,
+      "loss": 0.6586,
+      "step": 46992
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.792147159576416,
+      "learning_rate": 1.3784365040323466e-06,
+      "loss": 0.6621,
+      "step": 46993
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 1.9896947145462036,
+      "learning_rate": 1.378254724236645e-06,
+      "loss": 0.5735,
+      "step": 46994
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.6979076862335205,
+      "learning_rate": 1.378072954511664e-06,
+      "loss": 0.6906,
+      "step": 46995
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 4.036843299865723,
+      "learning_rate": 1.3778911948579031e-06,
+      "loss": 0.5492,
+      "step": 46996
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.5495338439941406,
+      "learning_rate": 1.3777094452758717e-06,
+      "loss": 0.6201,
+      "step": 46997
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.0281941890716553,
+      "learning_rate": 1.3775277057660718e-06,
+      "loss": 0.4986,
+      "step": 46998
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.3274714946746826,
+      "learning_rate": 1.377345976329012e-06,
+      "loss": 0.428,
+      "step": 46999
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.3477044105529785,
+      "learning_rate": 1.3771642569651943e-06,
+      "loss": 0.2782,
+      "step": 47000
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 1.9413703680038452,
+      "learning_rate": 1.3769825476751274e-06,
+      "loss": 0.4781,
+      "step": 47001
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 5.649561882019043,
+      "learning_rate": 1.3768008484593131e-06,
+      "loss": 0.6505,
+      "step": 47002
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 4.688913345336914,
+      "learning_rate": 1.37661915931826e-06,
+      "loss": 0.6888,
+      "step": 47003
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.040032386779785,
+      "learning_rate": 1.37643748025247e-06,
+      "loss": 0.6183,
+      "step": 47004
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.9627115726470947,
+      "learning_rate": 1.3762558112624518e-06,
+      "loss": 0.4172,
+      "step": 47005
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.2909488677978516,
+      "learning_rate": 1.376074152348707e-06,
+      "loss": 0.6179,
+      "step": 47006
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.789436101913452,
+      "learning_rate": 1.375892503511742e-06,
+      "loss": 0.7058,
+      "step": 47007
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.1029915809631348,
+      "learning_rate": 1.3757108647520645e-06,
+      "loss": 0.6091,
+      "step": 47008
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.081732749938965,
+      "learning_rate": 1.3755292360701756e-06,
+      "loss": 0.5535,
+      "step": 47009
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.3054661750793457,
+      "learning_rate": 1.3753476174665837e-06,
+      "loss": 0.47,
+      "step": 47010
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.0154857635498047,
+      "learning_rate": 1.3751660089417907e-06,
+      "loss": 0.6685,
+      "step": 47011
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.21897029876709,
+      "learning_rate": 1.3749844104963045e-06,
+      "loss": 0.5396,
+      "step": 47012
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.5865414142608643,
+      "learning_rate": 1.374802822130627e-06,
+      "loss": 0.3145,
+      "step": 47013
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.179518699645996,
+      "learning_rate": 1.3746212438452654e-06,
+      "loss": 0.6576,
+      "step": 47014
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.5956344604492188,
+      "learning_rate": 1.3744396756407253e-06,
+      "loss": 0.5208,
+      "step": 47015
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.382566452026367,
+      "learning_rate": 1.3742581175175085e-06,
+      "loss": 0.4859,
+      "step": 47016
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.57949161529541,
+      "learning_rate": 1.3740765694761233e-06,
+      "loss": 0.6341,
+      "step": 47017
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.209195375442505,
+      "learning_rate": 1.3738950315170719e-06,
+      "loss": 0.6198,
+      "step": 47018
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 1.9774366617202759,
+      "learning_rate": 1.3737135036408594e-06,
+      "loss": 0.3015,
+      "step": 47019
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.413090705871582,
+      "learning_rate": 1.3735319858479928e-06,
+      "loss": 0.5396,
+      "step": 47020
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.5243046283721924,
+      "learning_rate": 1.373350478138974e-06,
+      "loss": 0.5846,
+      "step": 47021
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.4031529426574707,
+      "learning_rate": 1.3731689805143101e-06,
+      "loss": 0.747,
+      "step": 47022
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 1.559027075767517,
+      "learning_rate": 1.3729874929745035e-06,
+      "loss": 0.3513,
+      "step": 47023
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.5380711555480957,
+      "learning_rate": 1.3728060155200612e-06,
+      "loss": 0.5484,
+      "step": 47024
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.5388612747192383,
+      "learning_rate": 1.372624548151485e-06,
+      "loss": 0.7035,
+      "step": 47025
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.4506430625915527,
+      "learning_rate": 1.3724430908692832e-06,
+      "loss": 0.6266,
+      "step": 47026
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.3776392936706543,
+      "learning_rate": 1.372261643673956e-06,
+      "loss": 0.3772,
+      "step": 47027
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.263502836227417,
+      "learning_rate": 1.3720802065660121e-06,
+      "loss": 0.425,
+      "step": 47028
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.7644922733306885,
+      "learning_rate": 1.3718987795459526e-06,
+      "loss": 0.6371,
+      "step": 47029
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 1.8639533519744873,
+      "learning_rate": 1.3717173626142848e-06,
+      "loss": 0.4662,
+      "step": 47030
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.8126139640808105,
+      "learning_rate": 1.3715359557715102e-06,
+      "loss": 0.4189,
+      "step": 47031
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.598712682723999,
+      "learning_rate": 1.3713545590181366e-06,
+      "loss": 0.505,
+      "step": 47032
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.853579044342041,
+      "learning_rate": 1.3711731723546652e-06,
+      "loss": 0.5668,
+      "step": 47033
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 5.348483085632324,
+      "learning_rate": 1.3709917957816033e-06,
+      "loss": 0.74,
+      "step": 47034
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.4328548908233643,
+      "learning_rate": 1.3708104292994506e-06,
+      "loss": 0.5334,
+      "step": 47035
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.970595121383667,
+      "learning_rate": 1.3706290729087173e-06,
+      "loss": 0.4323,
+      "step": 47036
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 4.367378234863281,
+      "learning_rate": 1.3704477266099041e-06,
+      "loss": 0.6758,
+      "step": 47037
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.3684608936309814,
+      "learning_rate": 1.3702663904035175e-06,
+      "loss": 0.582,
+      "step": 47038
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 1.9109351634979248,
+      "learning_rate": 1.3700850642900586e-06,
+      "loss": 0.5247,
+      "step": 47039
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.0961484909057617,
+      "learning_rate": 1.369903748270035e-06,
+      "loss": 0.4014,
+      "step": 47040
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.4047651290893555,
+      "learning_rate": 1.3697224423439476e-06,
+      "loss": 0.4967,
+      "step": 47041
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.6776602268218994,
+      "learning_rate": 1.3695411465123037e-06,
+      "loss": 0.7703,
+      "step": 47042
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.218411684036255,
+      "learning_rate": 1.3693598607756042e-06,
+      "loss": 0.7495,
+      "step": 47043
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.945683002471924,
+      "learning_rate": 1.3691785851343565e-06,
+      "loss": 0.6379,
+      "step": 47044
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 5.207914352416992,
+      "learning_rate": 1.3689973195890615e-06,
+      "loss": 0.523,
+      "step": 47045
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.0439271926879883,
+      "learning_rate": 1.3688160641402258e-06,
+      "loss": 0.5361,
+      "step": 47046
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.043088436126709,
+      "learning_rate": 1.368634818788352e-06,
+      "loss": 0.7254,
+      "step": 47047
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.883842706680298,
+      "learning_rate": 1.3684535835339452e-06,
+      "loss": 0.4903,
+      "step": 47048
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.3217806816101074,
+      "learning_rate": 1.3682723583775071e-06,
+      "loss": 0.5314,
+      "step": 47049
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.212690591812134,
+      "learning_rate": 1.3680911433195443e-06,
+      "loss": 0.6185,
+      "step": 47050
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.182854413986206,
+      "learning_rate": 1.3679099383605581e-06,
+      "loss": 0.5319,
+      "step": 47051
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.3862006664276123,
+      "learning_rate": 1.3677287435010556e-06,
+      "loss": 0.7273,
+      "step": 47052
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.276930093765259,
+      "learning_rate": 1.367547558741537e-06,
+      "loss": 0.4783,
+      "step": 47053
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.3485333919525146,
+      "learning_rate": 1.3673663840825098e-06,
+      "loss": 0.5867,
+      "step": 47054
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.724087953567505,
+      "learning_rate": 1.3671852195244723e-06,
+      "loss": 0.4313,
+      "step": 47055
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.1577749252319336,
+      "learning_rate": 1.3670040650679357e-06,
+      "loss": 0.645,
+      "step": 47056
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.9671788215637207,
+      "learning_rate": 1.3668229207133975e-06,
+      "loss": 0.4995,
+      "step": 47057
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.6517505645751953,
+      "learning_rate": 1.3666417864613652e-06,
+      "loss": 0.7206,
+      "step": 47058
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.034925937652588,
+      "learning_rate": 1.36646066231234e-06,
+      "loss": 0.5855,
+      "step": 47059
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.9227867126464844,
+      "learning_rate": 1.3662795482668277e-06,
+      "loss": 0.5738,
+      "step": 47060
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.1386711597442627,
+      "learning_rate": 1.3660984443253295e-06,
+      "loss": 0.6126,
+      "step": 47061
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.080415964126587,
+      "learning_rate": 1.3659173504883515e-06,
+      "loss": 0.5078,
+      "step": 47062
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.2068872451782227,
+      "learning_rate": 1.365736266756395e-06,
+      "loss": 0.6647,
+      "step": 47063
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.198239803314209,
+      "learning_rate": 1.3655551931299654e-06,
+      "loss": 0.6637,
+      "step": 47064
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.0115561485290527,
+      "learning_rate": 1.365374129609564e-06,
+      "loss": 0.3265,
+      "step": 47065
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 4.497077941894531,
+      "learning_rate": 1.3651930761956972e-06,
+      "loss": 0.6648,
+      "step": 47066
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.693000316619873,
+      "learning_rate": 1.3650120328888654e-06,
+      "loss": 0.5374,
+      "step": 47067
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 5.8944010734558105,
+      "learning_rate": 1.3648309996895748e-06,
+      "loss": 0.517,
+      "step": 47068
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 4.382537364959717,
+      "learning_rate": 1.3646499765983257e-06,
+      "loss": 0.4776,
+      "step": 47069
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.019512414932251,
+      "learning_rate": 1.3644689636156233e-06,
+      "loss": 0.5668,
+      "step": 47070
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.2293546199798584,
+      "learning_rate": 1.3642879607419729e-06,
+      "loss": 0.5847,
+      "step": 47071
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.4066474437713623,
+      "learning_rate": 1.3641069679778735e-06,
+      "loss": 0.7515,
+      "step": 47072
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.8775458335876465,
+      "learning_rate": 1.3639259853238325e-06,
+      "loss": 0.5598,
+      "step": 47073
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.409801483154297,
+      "learning_rate": 1.3637450127803492e-06,
+      "loss": 0.4424,
+      "step": 47074
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.8214879035949707,
+      "learning_rate": 1.3635640503479291e-06,
+      "loss": 0.388,
+      "step": 47075
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.646587610244751,
+      "learning_rate": 1.3633830980270767e-06,
+      "loss": 0.6166,
+      "step": 47076
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.2888007164001465,
+      "learning_rate": 1.3632021558182923e-06,
+      "loss": 0.5536,
+      "step": 47077
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.2400760650634766,
+      "learning_rate": 1.363021223722082e-06,
+      "loss": 0.5722,
+      "step": 47078
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.680715560913086,
+      "learning_rate": 1.3628403017389453e-06,
+      "loss": 0.545,
+      "step": 47079
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.758549451828003,
+      "learning_rate": 1.3626593898693885e-06,
+      "loss": 0.6182,
+      "step": 47080
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.937424659729004,
+      "learning_rate": 1.3624784881139126e-06,
+      "loss": 0.8796,
+      "step": 47081
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 5.18743371963501,
+      "learning_rate": 1.3622975964730207e-06,
+      "loss": 0.6367,
+      "step": 47082
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.4459924697875977,
+      "learning_rate": 1.362116714947218e-06,
+      "loss": 0.6603,
+      "step": 47083
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.0829453468322754,
+      "learning_rate": 1.3619358435370044e-06,
+      "loss": 0.5572,
+      "step": 47084
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.859575033187866,
+      "learning_rate": 1.3617549822428867e-06,
+      "loss": 0.6556,
+      "step": 47085
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 4.7596635818481445,
+      "learning_rate": 1.3615741310653629e-06,
+      "loss": 0.7211,
+      "step": 47086
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.5858747959136963,
+      "learning_rate": 1.361393290004941e-06,
+      "loss": 0.5306,
+      "step": 47087
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.9448049068450928,
+      "learning_rate": 1.361212459062119e-06,
+      "loss": 0.4638,
+      "step": 47088
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 4.3384504318237305,
+      "learning_rate": 1.3610316382374038e-06,
+      "loss": 0.5039,
+      "step": 47089
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.5584380626678467,
+      "learning_rate": 1.3608508275312948e-06,
+      "loss": 0.3973,
+      "step": 47090
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 1.7895628213882446,
+      "learning_rate": 1.360670026944298e-06,
+      "loss": 0.1887,
+      "step": 47091
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.0546185970306396,
+      "learning_rate": 1.3604892364769129e-06,
+      "loss": 0.3349,
+      "step": 47092
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.2369847297668457,
+      "learning_rate": 1.3603084561296453e-06,
+      "loss": 0.5993,
+      "step": 47093
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.1065163612365723,
+      "learning_rate": 1.3601276859029954e-06,
+      "loss": 0.5933,
+      "step": 47094
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.433844566345215,
+      "learning_rate": 1.359946925797468e-06,
+      "loss": 0.4814,
+      "step": 47095
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.8871755599975586,
+      "learning_rate": 1.3597661758135617e-06,
+      "loss": 0.5991,
+      "step": 47096
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.587761402130127,
+      "learning_rate": 1.3595854359517852e-06,
+      "loss": 0.7887,
+      "step": 47097
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.1870527267456055,
+      "learning_rate": 1.359404706212636e-06,
+      "loss": 0.3745,
+      "step": 47098
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.5845744609832764,
+      "learning_rate": 1.35922398659662e-06,
+      "loss": 0.7221,
+      "step": 47099
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.7330873012542725,
+      "learning_rate": 1.3590432771042367e-06,
+      "loss": 0.6379,
+      "step": 47100
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 4.543331146240234,
+      "learning_rate": 1.3588625777359921e-06,
+      "loss": 0.6806,
+      "step": 47101
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 1.954142689704895,
+      "learning_rate": 1.3586818884923847e-06,
+      "loss": 0.5525,
+      "step": 47102
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.041677951812744,
+      "learning_rate": 1.3585012093739208e-06,
+      "loss": 0.6465,
+      "step": 47103
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.4881606101989746,
+      "learning_rate": 1.358320540381099e-06,
+      "loss": 0.552,
+      "step": 47104
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.274374008178711,
+      "learning_rate": 1.3581398815144252e-06,
+      "loss": 0.5892,
+      "step": 47105
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 4.80048131942749,
+      "learning_rate": 1.357959232774399e-06,
+      "loss": 0.4871,
+      "step": 47106
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.3865091800689697,
+      "learning_rate": 1.3577785941615245e-06,
+      "loss": 0.628,
+      "step": 47107
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.733447790145874,
+      "learning_rate": 1.3575979656763027e-06,
+      "loss": 0.6045,
+      "step": 47108
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.456601619720459,
+      "learning_rate": 1.3574173473192377e-06,
+      "loss": 0.6621,
+      "step": 47109
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.1922194957733154,
+      "learning_rate": 1.3572367390908287e-06,
+      "loss": 0.5963,
+      "step": 47110
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.2739596366882324,
+      "learning_rate": 1.3570561409915817e-06,
+      "loss": 0.4997,
+      "step": 47111
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.2329792976379395,
+      "learning_rate": 1.3568755530219945e-06,
+      "loss": 0.6552,
+      "step": 47112
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.0991456508636475,
+      "learning_rate": 1.3566949751825743e-06,
+      "loss": 0.6701,
+      "step": 47113
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.910244941711426,
+      "learning_rate": 1.3565144074738185e-06,
+      "loss": 0.5276,
+      "step": 47114
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 4.224465370178223,
+      "learning_rate": 1.3563338498962325e-06,
+      "loss": 0.4112,
+      "step": 47115
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.8526999950408936,
+      "learning_rate": 1.3561533024503143e-06,
+      "loss": 0.4714,
+      "step": 47116
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.5198450088500977,
+      "learning_rate": 1.3559727651365722e-06,
+      "loss": 0.567,
+      "step": 47117
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.3637547492980957,
+      "learning_rate": 1.3557922379555028e-06,
+      "loss": 0.6654,
+      "step": 47118
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.0338284969329834,
+      "learning_rate": 1.3556117209076114e-06,
+      "loss": 0.5499,
+      "step": 47119
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.669527053833008,
+      "learning_rate": 1.3554312139933973e-06,
+      "loss": 0.6057,
+      "step": 47120
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.897010326385498,
+      "learning_rate": 1.3552507172133655e-06,
+      "loss": 0.7238,
+      "step": 47121
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.3110647201538086,
+      "learning_rate": 1.3550702305680141e-06,
+      "loss": 0.6382,
+      "step": 47122
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.244365930557251,
+      "learning_rate": 1.3548897540578488e-06,
+      "loss": 0.6446,
+      "step": 47123
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.639329195022583,
+      "learning_rate": 1.3547092876833678e-06,
+      "loss": 0.606,
+      "step": 47124
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.7688913345336914,
+      "learning_rate": 1.3545288314450766e-06,
+      "loss": 0.5548,
+      "step": 47125
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 4.742210388183594,
+      "learning_rate": 1.3543483853434735e-06,
+      "loss": 0.489,
+      "step": 47126
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.037222146987915,
+      "learning_rate": 1.3541679493790633e-06,
+      "loss": 0.6604,
+      "step": 47127
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.7793564796447754,
+      "learning_rate": 1.3539875235523447e-06,
+      "loss": 0.4955,
+      "step": 47128
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.0449862480163574,
+      "learning_rate": 1.3538071078638225e-06,
+      "loss": 0.4716,
+      "step": 47129
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.4791994094848633,
+      "learning_rate": 1.3536267023139955e-06,
+      "loss": 0.5836,
+      "step": 47130
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.017813205718994,
+      "learning_rate": 1.3534463069033664e-06,
+      "loss": 0.4415,
+      "step": 47131
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.0609729290008545,
+      "learning_rate": 1.3532659216324384e-06,
+      "loss": 0.59,
+      "step": 47132
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.9984893798828125,
+      "learning_rate": 1.3530855465017106e-06,
+      "loss": 0.7409,
+      "step": 47133
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 4.504227161407471,
+      "learning_rate": 1.3529051815116874e-06,
+      "loss": 0.8865,
+      "step": 47134
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.64583158493042,
+      "learning_rate": 1.3527248266628669e-06,
+      "loss": 0.5407,
+      "step": 47135
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.8781981468200684,
+      "learning_rate": 1.3525444819557527e-06,
+      "loss": 0.4801,
+      "step": 47136
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 4.4170989990234375,
+      "learning_rate": 1.3523641473908467e-06,
+      "loss": 0.5563,
+      "step": 47137
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.613959550857544,
+      "learning_rate": 1.3521838229686484e-06,
+      "loss": 0.5349,
+      "step": 47138
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 4.2218523025512695,
+      "learning_rate": 1.3520035086896622e-06,
+      "loss": 0.5851,
+      "step": 47139
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.3865716457366943,
+      "learning_rate": 1.351823204554386e-06,
+      "loss": 0.7958,
+      "step": 47140
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.082141637802124,
+      "learning_rate": 1.3516429105633245e-06,
+      "loss": 0.5437,
+      "step": 47141
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.3610599040985107,
+      "learning_rate": 1.3514626267169755e-06,
+      "loss": 0.5899,
+      "step": 47142
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 1.8954700231552124,
+      "learning_rate": 1.3512823530158426e-06,
+      "loss": 0.5267,
+      "step": 47143
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.2013494968414307,
+      "learning_rate": 1.3511020894604276e-06,
+      "loss": 0.6913,
+      "step": 47144
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.5527520179748535,
+      "learning_rate": 1.3509218360512295e-06,
+      "loss": 0.3973,
+      "step": 47145
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.2626240253448486,
+      "learning_rate": 1.3507415927887524e-06,
+      "loss": 0.3039,
+      "step": 47146
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 4.096871852874756,
+      "learning_rate": 1.3505613596734945e-06,
+      "loss": 0.8496,
+      "step": 47147
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.412093162536621,
+      "learning_rate": 1.3503811367059594e-06,
+      "loss": 0.5378,
+      "step": 47148
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 16.890174865722656,
+      "learning_rate": 1.350200923886646e-06,
+      "loss": 0.8393,
+      "step": 47149
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 5.980049133300781,
+      "learning_rate": 1.3500207212160583e-06,
+      "loss": 0.681,
+      "step": 47150
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.847060441970825,
+      "learning_rate": 1.3498405286946937e-06,
+      "loss": 0.3402,
+      "step": 47151
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.960378408432007,
+      "learning_rate": 1.3496603463230568e-06,
+      "loss": 0.7635,
+      "step": 47152
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.43617582321167,
+      "learning_rate": 1.3494801741016455e-06,
+      "loss": 0.6348,
+      "step": 47153
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.1192986965179443,
+      "learning_rate": 1.3493000120309641e-06,
+      "loss": 0.3933,
+      "step": 47154
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 1.9200495481491089,
+      "learning_rate": 1.3491198601115102e-06,
+      "loss": 0.4595,
+      "step": 47155
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.8175690174102783,
+      "learning_rate": 1.348939718343788e-06,
+      "loss": 0.4575,
+      "step": 47156
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.0129473209381104,
+      "learning_rate": 1.3487595867282932e-06,
+      "loss": 0.7622,
+      "step": 47157
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.623685598373413,
+      "learning_rate": 1.3485794652655336e-06,
+      "loss": 0.4769,
+      "step": 47158
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.7526471614837646,
+      "learning_rate": 1.3483993539560054e-06,
+      "loss": 0.6626,
+      "step": 47159
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.1724660396575928,
+      "learning_rate": 1.3482192528002113e-06,
+      "loss": 0.7013,
+      "step": 47160
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.0340065956115723,
+      "learning_rate": 1.3480391617986506e-06,
+      "loss": 0.5835,
+      "step": 47161
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.1518054008483887,
+      "learning_rate": 1.3478590809518266e-06,
+      "loss": 0.5287,
+      "step": 47162
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.6448826789855957,
+      "learning_rate": 1.3476790102602366e-06,
+      "loss": 0.6396,
+      "step": 47163
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.949791431427002,
+      "learning_rate": 1.3474989497243846e-06,
+      "loss": 0.7102,
+      "step": 47164
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.025623083114624,
+      "learning_rate": 1.3473188993447679e-06,
+      "loss": 0.6338,
+      "step": 47165
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.9306700229644775,
+      "learning_rate": 1.3471388591218908e-06,
+      "loss": 0.7627,
+      "step": 47166
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.7296652793884277,
+      "learning_rate": 1.3469588290562508e-06,
+      "loss": 0.6857,
+      "step": 47167
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.395106792449951,
+      "learning_rate": 1.3467788091483513e-06,
+      "loss": 0.7147,
+      "step": 47168
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 1.9471237659454346,
+      "learning_rate": 1.34659879939869e-06,
+      "loss": 0.459,
+      "step": 47169
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.9995744228363037,
+      "learning_rate": 1.3464187998077698e-06,
+      "loss": 0.7613,
+      "step": 47170
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.742587089538574,
+      "learning_rate": 1.3462388103760893e-06,
+      "loss": 0.8433,
+      "step": 47171
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.0220787525177,
+      "learning_rate": 1.3460588311041511e-06,
+      "loss": 0.4919,
+      "step": 47172
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 1.8156216144561768,
+      "learning_rate": 1.3458788619924534e-06,
+      "loss": 0.6063,
+      "step": 47173
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.979257583618164,
+      "learning_rate": 1.3456989030414991e-06,
+      "loss": 0.5715,
+      "step": 47174
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.856017827987671,
+      "learning_rate": 1.3455189542517854e-06,
+      "loss": 0.6202,
+      "step": 47175
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.778592109680176,
+      "learning_rate": 1.345339015623816e-06,
+      "loss": 0.7308,
+      "step": 47176
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.3848683834075928,
+      "learning_rate": 1.3451590871580872e-06,
+      "loss": 0.8284,
+      "step": 47177
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.7845683097839355,
+      "learning_rate": 1.3449791688551044e-06,
+      "loss": 0.578,
+      "step": 47178
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.4186997413635254,
+      "learning_rate": 1.3447992607153643e-06,
+      "loss": 0.4707,
+      "step": 47179
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.4947426319122314,
+      "learning_rate": 1.3446193627393695e-06,
+      "loss": 0.6164,
+      "step": 47180
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 6.202176094055176,
+      "learning_rate": 1.3444394749276167e-06,
+      "loss": 0.6663,
+      "step": 47181
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.421048402786255,
+      "learning_rate": 1.3442595972806104e-06,
+      "loss": 0.5258,
+      "step": 47182
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.481485366821289,
+      "learning_rate": 1.3440797297988472e-06,
+      "loss": 0.584,
+      "step": 47183
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.133894443511963,
+      "learning_rate": 1.3438998724828305e-06,
+      "loss": 0.6181,
+      "step": 47184
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.3079793453216553,
+      "learning_rate": 1.3437200253330568e-06,
+      "loss": 0.6349,
+      "step": 47185
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.004960775375366,
+      "learning_rate": 1.3435401883500298e-06,
+      "loss": 1.0004,
+      "step": 47186
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.101865768432617,
+      "learning_rate": 1.3433603615342461e-06,
+      "loss": 0.5749,
+      "step": 47187
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.179779529571533,
+      "learning_rate": 1.343180544886209e-06,
+      "loss": 0.6884,
+      "step": 47188
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.229417324066162,
+      "learning_rate": 1.3430007384064153e-06,
+      "loss": 0.5705,
+      "step": 47189
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.625931739807129,
+      "learning_rate": 1.3428209420953681e-06,
+      "loss": 0.5485,
+      "step": 47190
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.2834677696228027,
+      "learning_rate": 1.3426411559535647e-06,
+      "loss": 0.6477,
+      "step": 47191
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.3686530590057373,
+      "learning_rate": 1.3424613799815073e-06,
+      "loss": 0.629,
+      "step": 47192
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.7157328128814697,
+      "learning_rate": 1.342281614179693e-06,
+      "loss": 0.7506,
+      "step": 47193
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.5921385288238525,
+      "learning_rate": 1.3421018585486234e-06,
+      "loss": 0.4764,
+      "step": 47194
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.826371192932129,
+      "learning_rate": 1.3419221130887998e-06,
+      "loss": 0.7731,
+      "step": 47195
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.1179122924804688,
+      "learning_rate": 1.3417423778007194e-06,
+      "loss": 0.6296,
+      "step": 47196
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.57975435256958,
+      "learning_rate": 1.341562652684883e-06,
+      "loss": 0.6443,
+      "step": 47197
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.8793275356292725,
+      "learning_rate": 1.3413829377417915e-06,
+      "loss": 0.5107,
+      "step": 47198
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.5069186687469482,
+      "learning_rate": 1.3412032329719422e-06,
+      "loss": 0.6583,
+      "step": 47199
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.5972588062286377,
+      "learning_rate": 1.3410235383758374e-06,
+      "loss": 0.6353,
+      "step": 47200
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.1497504711151123,
+      "learning_rate": 1.3408438539539737e-06,
+      "loss": 0.5256,
+      "step": 47201
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.5355169773101807,
+      "learning_rate": 1.3406641797068542e-06,
+      "loss": 0.3926,
+      "step": 47202
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.1607306003570557,
+      "learning_rate": 1.3404845156349755e-06,
+      "loss": 0.7269,
+      "step": 47203
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.988308906555176,
+      "learning_rate": 1.3403048617388387e-06,
+      "loss": 0.603,
+      "step": 47204
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.4207515716552734,
+      "learning_rate": 1.340125218018944e-06,
+      "loss": 0.5397,
+      "step": 47205
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.924725294113159,
+      "learning_rate": 1.339945584475789e-06,
+      "loss": 0.637,
+      "step": 47206
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.4435009956359863,
+      "learning_rate": 1.3397659611098756e-06,
+      "loss": 0.7885,
+      "step": 47207
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.2895708084106445,
+      "learning_rate": 1.3395863479217004e-06,
+      "loss": 0.6343,
+      "step": 47208
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.0209569931030273,
+      "learning_rate": 1.339406744911766e-06,
+      "loss": 0.5444,
+      "step": 47209
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.3691911697387695,
+      "learning_rate": 1.3392271520805683e-06,
+      "loss": 0.6033,
+      "step": 47210
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.909827470779419,
+      "learning_rate": 1.3390475694286104e-06,
+      "loss": 0.584,
+      "step": 47211
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.0752952098846436,
+      "learning_rate": 1.3388679969563885e-06,
+      "loss": 0.5362,
+      "step": 47212
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.7189078330993652,
+      "learning_rate": 1.3386884346644041e-06,
+      "loss": 0.7013,
+      "step": 47213
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.1805548667907715,
+      "learning_rate": 1.3385088825531544e-06,
+      "loss": 0.5712,
+      "step": 47214
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.544769525527954,
+      "learning_rate": 1.3383293406231413e-06,
+      "loss": 0.6397,
+      "step": 47215
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.2564711570739746,
+      "learning_rate": 1.3381498088748611e-06,
+      "loss": 0.7274,
+      "step": 47216
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.417447566986084,
+      "learning_rate": 1.3379702873088158e-06,
+      "loss": 0.7238,
+      "step": 47217
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.235250949859619,
+      "learning_rate": 1.3377907759255004e-06,
+      "loss": 0.5691,
+      "step": 47218
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.451220989227295,
+      "learning_rate": 1.3376112747254205e-06,
+      "loss": 0.5416,
+      "step": 47219
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.3279433250427246,
+      "learning_rate": 1.3374317837090694e-06,
+      "loss": 0.475,
+      "step": 47220
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.9030041694641113,
+      "learning_rate": 1.3372523028769507e-06,
+      "loss": 0.6881,
+      "step": 47221
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 4.365026473999023,
+      "learning_rate": 1.3370728322295595e-06,
+      "loss": 0.5345,
+      "step": 47222
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.6970667839050293,
+      "learning_rate": 1.3368933717673977e-06,
+      "loss": 0.5595,
+      "step": 47223
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.6907565593719482,
+      "learning_rate": 1.3367139214909624e-06,
+      "loss": 0.7529,
+      "step": 47224
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.2849013805389404,
+      "learning_rate": 1.3365344814007542e-06,
+      "loss": 0.5592,
+      "step": 47225
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.3125264644622803,
+      "learning_rate": 1.33635505149727e-06,
+      "loss": 0.7434,
+      "step": 47226
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 4.005770683288574,
+      "learning_rate": 1.3361756317810114e-06,
+      "loss": 0.7717,
+      "step": 47227
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.940077304840088,
+      "learning_rate": 1.3359962222524747e-06,
+      "loss": 0.8053,
+      "step": 47228
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.920103073120117,
+      "learning_rate": 1.3358168229121615e-06,
+      "loss": 0.4717,
+      "step": 47229
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.286968946456909,
+      "learning_rate": 1.335637433760567e-06,
+      "loss": 0.6443,
+      "step": 47230
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.2898268699645996,
+      "learning_rate": 1.3354580547981937e-06,
+      "loss": 0.6576,
+      "step": 47231
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.0358054637908936,
+      "learning_rate": 1.335278686025538e-06,
+      "loss": 0.5178,
+      "step": 47232
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.873746395111084,
+      "learning_rate": 1.3350993274431e-06,
+      "loss": 0.4651,
+      "step": 47233
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.8734562397003174,
+      "learning_rate": 1.3349199790513766e-06,
+      "loss": 0.7554,
+      "step": 47234
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.367218255996704,
+      "learning_rate": 1.3347406408508695e-06,
+      "loss": 0.4826,
+      "step": 47235
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.185396671295166,
+      "learning_rate": 1.3345613128420742e-06,
+      "loss": 0.5023,
+      "step": 47236
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.743544340133667,
+      "learning_rate": 1.3343819950254922e-06,
+      "loss": 0.6976,
+      "step": 47237
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.7149765491485596,
+      "learning_rate": 1.334202687401618e-06,
+      "loss": 1.0133,
+      "step": 47238
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.924966335296631,
+      "learning_rate": 1.3340233899709554e-06,
+      "loss": 0.4279,
+      "step": 47239
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.9469728469848633,
+      "learning_rate": 1.3338441027339993e-06,
+      "loss": 0.5815,
+      "step": 47240
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.510814905166626,
+      "learning_rate": 1.3336648256912505e-06,
+      "loss": 0.4976,
+      "step": 47241
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.0958757400512695,
+      "learning_rate": 1.333485558843205e-06,
+      "loss": 0.5284,
+      "step": 47242
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.482450246810913,
+      "learning_rate": 1.333306302190364e-06,
+      "loss": 0.4173,
+      "step": 47243
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 4.369383335113525,
+      "learning_rate": 1.3331270557332237e-06,
+      "loss": 0.5757,
+      "step": 47244
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.6610305309295654,
+      "learning_rate": 1.3329478194722845e-06,
+      "loss": 0.3955,
+      "step": 47245
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.6123580932617188,
+      "learning_rate": 1.3327685934080419e-06,
+      "loss": 0.5847,
+      "step": 47246
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.4591572284698486,
+      "learning_rate": 1.332589377540998e-06,
+      "loss": 0.5321,
+      "step": 47247
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.537881851196289,
+      "learning_rate": 1.3324101718716476e-06,
+      "loss": 0.6318,
+      "step": 47248
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.185668706893921,
+      "learning_rate": 1.332230976400492e-06,
+      "loss": 0.4602,
+      "step": 47249
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.540454149246216,
+      "learning_rate": 1.3320517911280268e-06,
+      "loss": 0.8006,
+      "step": 47250
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 4.302577018737793,
+      "learning_rate": 1.331872616054753e-06,
+      "loss": 0.5036,
+      "step": 47251
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.698957681655884,
+      "learning_rate": 1.3316934511811657e-06,
+      "loss": 0.4463,
+      "step": 47252
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.243650197982788,
+      "learning_rate": 1.3315142965077665e-06,
+      "loss": 0.7527,
+      "step": 47253
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.2784366607666016,
+      "learning_rate": 1.33133515203505e-06,
+      "loss": 0.2888,
+      "step": 47254
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.493558406829834,
+      "learning_rate": 1.3311560177635164e-06,
+      "loss": 0.4425,
+      "step": 47255
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.482938289642334,
+      "learning_rate": 1.3309768936936645e-06,
+      "loss": 0.4632,
+      "step": 47256
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 1.977530598640442,
+      "learning_rate": 1.3307977798259902e-06,
+      "loss": 0.3027,
+      "step": 47257
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 4.491275787353516,
+      "learning_rate": 1.330618676160993e-06,
+      "loss": 0.6722,
+      "step": 47258
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.7986438274383545,
+      "learning_rate": 1.3304395826991717e-06,
+      "loss": 0.5924,
+      "step": 47259
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.6961302757263184,
+      "learning_rate": 1.3302604994410223e-06,
+      "loss": 0.7927,
+      "step": 47260
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.24950909614563,
+      "learning_rate": 1.3300814263870449e-06,
+      "loss": 0.6356,
+      "step": 47261
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.4161736965179443,
+      "learning_rate": 1.329902363537735e-06,
+      "loss": 0.6282,
+      "step": 47262
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.649881601333618,
+      "learning_rate": 1.329723310893593e-06,
+      "loss": 0.7525,
+      "step": 47263
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.177716016769409,
+      "learning_rate": 1.329544268455114e-06,
+      "loss": 0.5304,
+      "step": 47264
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 1.984608769416809,
+      "learning_rate": 1.3293652362227993e-06,
+      "loss": 0.3788,
+      "step": 47265
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.0478930473327637,
+      "learning_rate": 1.329186214197143e-06,
+      "loss": 0.546,
+      "step": 47266
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.7979040145874023,
+      "learning_rate": 1.3290072023786449e-06,
+      "loss": 0.6802,
+      "step": 47267
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.747544288635254,
+      "learning_rate": 1.3288282007678043e-06,
+      "loss": 0.709,
+      "step": 47268
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.8250482082366943,
+      "learning_rate": 1.3286492093651154e-06,
+      "loss": 0.5035,
+      "step": 47269
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.091862440109253,
+      "learning_rate": 1.3284702281710793e-06,
+      "loss": 0.6578,
+      "step": 47270
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.366553544998169,
+      "learning_rate": 1.3282912571861905e-06,
+      "loss": 0.6698,
+      "step": 47271
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.844175338745117,
+      "learning_rate": 1.32811229641095e-06,
+      "loss": 0.5472,
+      "step": 47272
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.482726812362671,
+      "learning_rate": 1.3279333458458521e-06,
+      "loss": 0.7815,
+      "step": 47273
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.4375202655792236,
+      "learning_rate": 1.3277544054913972e-06,
+      "loss": 0.6823,
+      "step": 47274
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 4.869172096252441,
+      "learning_rate": 1.3275754753480802e-06,
+      "loss": 0.4217,
+      "step": 47275
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.902290105819702,
+      "learning_rate": 1.327396555416402e-06,
+      "loss": 0.7925,
+      "step": 47276
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.3463072776794434,
+      "learning_rate": 1.327217645696856e-06,
+      "loss": 0.6483,
+      "step": 47277
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.50057315826416,
+      "learning_rate": 1.3270387461899443e-06,
+      "loss": 0.5846,
+      "step": 47278
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.423980236053467,
+      "learning_rate": 1.3268598568961583e-06,
+      "loss": 0.5497,
+      "step": 47279
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.406050205230713,
+      "learning_rate": 1.3266809778160022e-06,
+      "loss": 0.6369,
+      "step": 47280
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 4.041303634643555,
+      "learning_rate": 1.3265021089499686e-06,
+      "loss": 0.409,
+      "step": 47281
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.5060856342315674,
+      "learning_rate": 1.326323250298558e-06,
+      "loss": 0.671,
+      "step": 47282
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.3046841621398926,
+      "learning_rate": 1.3261444018622648e-06,
+      "loss": 0.6656,
+      "step": 47283
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 1.8185206651687622,
+      "learning_rate": 1.3259655636415898e-06,
+      "loss": 0.4254,
+      "step": 47284
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.1676712036132812,
+      "learning_rate": 1.3257867356370257e-06,
+      "loss": 0.9335,
+      "step": 47285
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.171912670135498,
+      "learning_rate": 1.325607917849075e-06,
+      "loss": 0.4383,
+      "step": 47286
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.188462734222412,
+      "learning_rate": 1.3254291102782295e-06,
+      "loss": 0.6172,
+      "step": 47287
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.209618330001831,
+      "learning_rate": 1.3252503129249915e-06,
+      "loss": 0.6444,
+      "step": 47288
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.5876541137695312,
+      "learning_rate": 1.325071525789854e-06,
+      "loss": 0.6724,
+      "step": 47289
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.308290719985962,
+      "learning_rate": 1.3248927488733176e-06,
+      "loss": 0.5513,
+      "step": 47290
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.1156790256500244,
+      "learning_rate": 1.324713982175876e-06,
+      "loss": 0.6151,
+      "step": 47291
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.9603261947631836,
+      "learning_rate": 1.3245352256980292e-06,
+      "loss": 0.7029,
+      "step": 47292
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.9313125610351562,
+      "learning_rate": 1.3243564794402724e-06,
+      "loss": 0.6959,
+      "step": 47293
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.8695857524871826,
+      "learning_rate": 1.3241777434031045e-06,
+      "loss": 0.4986,
+      "step": 47294
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.56807279586792,
+      "learning_rate": 1.3239990175870193e-06,
+      "loss": 0.6441,
+      "step": 47295
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 1.7654919624328613,
+      "learning_rate": 1.3238203019925178e-06,
+      "loss": 0.3742,
+      "step": 47296
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.5154173374176025,
+      "learning_rate": 1.3236415966200934e-06,
+      "loss": 0.4528,
+      "step": 47297
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.3208506107330322,
+      "learning_rate": 1.3234629014702455e-06,
+      "loss": 0.606,
+      "step": 47298
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 4.425257205963135,
+      "learning_rate": 1.3232842165434678e-06,
+      "loss": 0.5763,
+      "step": 47299
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.0201940536499023,
+      "learning_rate": 1.3231055418402617e-06,
+      "loss": 0.5715,
+      "step": 47300
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.7721009254455566,
+      "learning_rate": 1.3229268773611203e-06,
+      "loss": 0.9097,
+      "step": 47301
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.2545900344848633,
+      "learning_rate": 1.3227482231065436e-06,
+      "loss": 0.6083,
+      "step": 47302
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.3591148853302,
+      "learning_rate": 1.3225695790770249e-06,
+      "loss": 0.6228,
+      "step": 47303
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.2756571769714355,
+      "learning_rate": 1.3223909452730643e-06,
+      "loss": 0.5294,
+      "step": 47304
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.5910773277282715,
+      "learning_rate": 1.322212321695155e-06,
+      "loss": 0.5278,
+      "step": 47305
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.539198875427246,
+      "learning_rate": 1.3220337083437972e-06,
+      "loss": 0.2568,
+      "step": 47306
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.0635294914245605,
+      "learning_rate": 1.321855105219484e-06,
+      "loss": 0.5199,
+      "step": 47307
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.82669997215271,
+      "learning_rate": 1.321676512322716e-06,
+      "loss": 0.5675,
+      "step": 47308
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.2956509590148926,
+      "learning_rate": 1.3214979296539859e-06,
+      "loss": 0.7933,
+      "step": 47309
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.0521838665008545,
+      "learning_rate": 1.3213193572137934e-06,
+      "loss": 0.5684,
+      "step": 47310
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.339344024658203,
+      "learning_rate": 1.3211407950026322e-06,
+      "loss": 0.6521,
+      "step": 47311
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.8051767349243164,
+      "learning_rate": 1.3209622430210016e-06,
+      "loss": 0.4983,
+      "step": 47312
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.181446075439453,
+      "learning_rate": 1.3207837012693958e-06,
+      "loss": 0.5133,
+      "step": 47313
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.6252880096435547,
+      "learning_rate": 1.3206051697483135e-06,
+      "loss": 0.7556,
+      "step": 47314
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 1.7791892290115356,
+      "learning_rate": 1.320426648458249e-06,
+      "loss": 0.6158,
+      "step": 47315
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.3852500915527344,
+      "learning_rate": 1.3202481373996989e-06,
+      "loss": 0.9698,
+      "step": 47316
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.7868828773498535,
+      "learning_rate": 1.3200696365731618e-06,
+      "loss": 0.5639,
+      "step": 47317
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.396435260772705,
+      "learning_rate": 1.3198911459791314e-06,
+      "loss": 0.7695,
+      "step": 47318
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.5066325664520264,
+      "learning_rate": 1.3197126656181046e-06,
+      "loss": 0.708,
+      "step": 47319
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.2162506580352783,
+      "learning_rate": 1.31953419549058e-06,
+      "loss": 0.4201,
+      "step": 47320
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.802779197692871,
+      "learning_rate": 1.3193557355970505e-06,
+      "loss": 0.6011,
+      "step": 47321
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.908163547515869,
+      "learning_rate": 1.3191772859380158e-06,
+      "loss": 0.4546,
+      "step": 47322
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.0635690689086914,
+      "learning_rate": 1.318998846513968e-06,
+      "loss": 0.7688,
+      "step": 47323
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.0874762535095215,
+      "learning_rate": 1.3188204173254076e-06,
+      "loss": 0.613,
+      "step": 47324
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.9061241149902344,
+      "learning_rate": 1.3186419983728265e-06,
+      "loss": 0.6042,
+      "step": 47325
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.5597217082977295,
+      "learning_rate": 1.3184635896567249e-06,
+      "loss": 0.4543,
+      "step": 47326
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.6667542457580566,
+      "learning_rate": 1.3182851911775956e-06,
+      "loss": 0.6481,
+      "step": 47327
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.9563794136047363,
+      "learning_rate": 1.3181068029359357e-06,
+      "loss": 0.5317,
+      "step": 47328
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.473592519760132,
+      "learning_rate": 1.3179284249322427e-06,
+      "loss": 0.5677,
+      "step": 47329
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.879298448562622,
+      "learning_rate": 1.3177500571670104e-06,
+      "loss": 0.8629,
+      "step": 47330
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.879770517349243,
+      "learning_rate": 1.3175716996407366e-06,
+      "loss": 0.5386,
+      "step": 47331
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.5665721893310547,
+      "learning_rate": 1.3173933523539157e-06,
+      "loss": 0.5012,
+      "step": 47332
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.1043310165405273,
+      "learning_rate": 1.3172150153070456e-06,
+      "loss": 0.6784,
+      "step": 47333
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.1789491176605225,
+      "learning_rate": 1.3170366885006197e-06,
+      "loss": 0.3447,
+      "step": 47334
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.2621500492095947,
+      "learning_rate": 1.3168583719351364e-06,
+      "loss": 0.6182,
+      "step": 47335
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.328127384185791,
+      "learning_rate": 1.3166800656110884e-06,
+      "loss": 0.4756,
+      "step": 47336
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.231271982192993,
+      "learning_rate": 1.316501769528975e-06,
+      "loss": 0.5305,
+      "step": 47337
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.775465488433838,
+      "learning_rate": 1.3163234836892891e-06,
+      "loss": 0.6486,
+      "step": 47338
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.5333027839660645,
+      "learning_rate": 1.3161452080925285e-06,
+      "loss": 0.7553,
+      "step": 47339
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.189945936203003,
+      "learning_rate": 1.3159669427391859e-06,
+      "loss": 0.6308,
+      "step": 47340
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 4.80195951461792,
+      "learning_rate": 1.3157886876297615e-06,
+      "loss": 0.4676,
+      "step": 47341
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 4.082613468170166,
+      "learning_rate": 1.3156104427647476e-06,
+      "loss": 0.4057,
+      "step": 47342
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.357891798019409,
+      "learning_rate": 1.3154322081446424e-06,
+      "loss": 0.42,
+      "step": 47343
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.166494607925415,
+      "learning_rate": 1.3152539837699379e-06,
+      "loss": 0.433,
+      "step": 47344
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.9712653160095215,
+      "learning_rate": 1.3150757696411336e-06,
+      "loss": 0.5814,
+      "step": 47345
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 4.2416863441467285,
+      "learning_rate": 1.3148975657587216e-06,
+      "loss": 0.7697,
+      "step": 47346
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 4.288116455078125,
+      "learning_rate": 1.3147193721232004e-06,
+      "loss": 0.7092,
+      "step": 47347
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.594245672225952,
+      "learning_rate": 1.3145411887350628e-06,
+      "loss": 0.5464,
+      "step": 47348
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.683336019515991,
+      "learning_rate": 1.3143630155948073e-06,
+      "loss": 0.5861,
+      "step": 47349
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.7447571754455566,
+      "learning_rate": 1.3141848527029255e-06,
+      "loss": 0.4839,
+      "step": 47350
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.312197685241699,
+      "learning_rate": 1.3140067000599166e-06,
+      "loss": 0.6504,
+      "step": 47351
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.6884968280792236,
+      "learning_rate": 1.313828557666273e-06,
+      "loss": 0.5287,
+      "step": 47352
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.9345226287841797,
+      "learning_rate": 1.3136504255224923e-06,
+      "loss": 0.4389,
+      "step": 47353
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.3457624912261963,
+      "learning_rate": 1.313472303629068e-06,
+      "loss": 0.7413,
+      "step": 47354
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.558799982070923,
+      "learning_rate": 1.3132941919864971e-06,
+      "loss": 0.7296,
+      "step": 47355
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.947742462158203,
+      "learning_rate": 1.3131160905952728e-06,
+      "loss": 0.6102,
+      "step": 47356
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.1761093139648438,
+      "learning_rate": 1.3129379994558927e-06,
+      "loss": 0.6852,
+      "step": 47357
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 1.7987453937530518,
+      "learning_rate": 1.31275991856885e-06,
+      "loss": 0.576,
+      "step": 47358
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.7587995529174805,
+      "learning_rate": 1.3125818479346414e-06,
+      "loss": 0.7405,
+      "step": 47359
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.9184799194335938,
+      "learning_rate": 1.3124037875537588e-06,
+      "loss": 0.5529,
+      "step": 47360
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.912921667098999,
+      "learning_rate": 1.3122257374267028e-06,
+      "loss": 0.6491,
+      "step": 47361
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.5904369354248047,
+      "learning_rate": 1.3120476975539642e-06,
+      "loss": 0.4431,
+      "step": 47362
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.9996962547302246,
+      "learning_rate": 1.3118696679360405e-06,
+      "loss": 0.5126,
+      "step": 47363
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.2489616870880127,
+      "learning_rate": 1.3116916485734244e-06,
+      "loss": 0.7313,
+      "step": 47364
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 4.5716400146484375,
+      "learning_rate": 1.3115136394666138e-06,
+      "loss": 0.5252,
+      "step": 47365
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.0922811031341553,
+      "learning_rate": 1.3113356406161004e-06,
+      "loss": 0.6085,
+      "step": 47366
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.606098175048828,
+      "learning_rate": 1.3111576520223824e-06,
+      "loss": 0.4242,
+      "step": 47367
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.8969078063964844,
+      "learning_rate": 1.3109796736859515e-06,
+      "loss": 0.6228,
+      "step": 47368
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.245032787322998,
+      "learning_rate": 1.3108017056073052e-06,
+      "loss": 0.5883,
+      "step": 47369
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 1.6079810857772827,
+      "learning_rate": 1.3106237477869367e-06,
+      "loss": 0.3167,
+      "step": 47370
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.2690329551696777,
+      "learning_rate": 1.3104458002253422e-06,
+      "loss": 0.5093,
+      "step": 47371
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.635197162628174,
+      "learning_rate": 1.3102678629230148e-06,
+      "loss": 0.6007,
+      "step": 47372
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.575690269470215,
+      "learning_rate": 1.3100899358804513e-06,
+      "loss": 0.572,
+      "step": 47373
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.227084159851074,
+      "learning_rate": 1.3099120190981447e-06,
+      "loss": 0.4712,
+      "step": 47374
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.2085318565368652,
+      "learning_rate": 1.3097341125765912e-06,
+      "loss": 0.6819,
+      "step": 47375
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 1.6966007947921753,
+      "learning_rate": 1.309556216316284e-06,
+      "loss": 0.4673,
+      "step": 47376
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.1156928539276123,
+      "learning_rate": 1.3093783303177194e-06,
+      "loss": 0.6681,
+      "step": 47377
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.571434259414673,
+      "learning_rate": 1.3092004545813897e-06,
+      "loss": 0.5597,
+      "step": 47378
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.5162622928619385,
+      "learning_rate": 1.3090225891077906e-06,
+      "loss": 0.7735,
+      "step": 47379
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.412093162536621,
+      "learning_rate": 1.3088447338974175e-06,
+      "loss": 0.6689,
+      "step": 47380
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.225346565246582,
+      "learning_rate": 1.308666888950766e-06,
+      "loss": 0.6076,
+      "step": 47381
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.2193639278411865,
+      "learning_rate": 1.308489054268327e-06,
+      "loss": 0.437,
+      "step": 47382
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.2168071269989014,
+      "learning_rate": 1.308311229850599e-06,
+      "loss": 0.7814,
+      "step": 47383
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.326564311981201,
+      "learning_rate": 1.3081334156980729e-06,
+      "loss": 0.5286,
+      "step": 47384
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.0580713748931885,
+      "learning_rate": 1.307955611811247e-06,
+      "loss": 0.6196,
+      "step": 47385
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.553835153579712,
+      "learning_rate": 1.3077778181906108e-06,
+      "loss": 0.5706,
+      "step": 47386
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.4436142444610596,
+      "learning_rate": 1.3076000348366635e-06,
+      "loss": 0.8497,
+      "step": 47387
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.161560297012329,
+      "learning_rate": 1.3074222617498956e-06,
+      "loss": 0.8105,
+      "step": 47388
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 4.098446369171143,
+      "learning_rate": 1.3072444989308037e-06,
+      "loss": 0.5594,
+      "step": 47389
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.897578477859497,
+      "learning_rate": 1.307066746379882e-06,
+      "loss": 0.5799,
+      "step": 47390
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.6960184574127197,
+      "learning_rate": 1.3068890040976234e-06,
+      "loss": 0.6382,
+      "step": 47391
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.8060967922210693,
+      "learning_rate": 1.3067112720845243e-06,
+      "loss": 0.482,
+      "step": 47392
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.402815341949463,
+      "learning_rate": 1.3065335503410765e-06,
+      "loss": 0.5102,
+      "step": 47393
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.16867733001709,
+      "learning_rate": 1.3063558388677766e-06,
+      "loss": 0.7757,
+      "step": 47394
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.290276527404785,
+      "learning_rate": 1.3061781376651156e-06,
+      "loss": 0.811,
+      "step": 47395
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.0536601543426514,
+      "learning_rate": 1.3060004467335913e-06,
+      "loss": 0.4126,
+      "step": 47396
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.3486812114715576,
+      "learning_rate": 1.3058227660736945e-06,
+      "loss": 0.4829,
+      "step": 47397
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.9615514278411865,
+      "learning_rate": 1.3056450956859218e-06,
+      "loss": 0.6752,
+      "step": 47398
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.477651596069336,
+      "learning_rate": 1.3054674355707646e-06,
+      "loss": 0.7819,
+      "step": 47399
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 6.387734889984131,
+      "learning_rate": 1.3052897857287206e-06,
+      "loss": 0.5731,
+      "step": 47400
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.000056743621826,
+      "learning_rate": 1.3051121461602783e-06,
+      "loss": 0.516,
+      "step": 47401
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.290929079055786,
+      "learning_rate": 1.3049345168659383e-06,
+      "loss": 0.7074,
+      "step": 47402
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.5442209243774414,
+      "learning_rate": 1.3047568978461893e-06,
+      "loss": 0.5392,
+      "step": 47403
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.1152961254119873,
+      "learning_rate": 1.3045792891015285e-06,
+      "loss": 0.5413,
+      "step": 47404
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 1.8949018716812134,
+      "learning_rate": 1.3044016906324468e-06,
+      "loss": 0.5428,
+      "step": 47405
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.7949202060699463,
+      "learning_rate": 1.3042241024394419e-06,
+      "loss": 0.6123,
+      "step": 47406
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.47094464302063,
+      "learning_rate": 1.3040465245230034e-06,
+      "loss": 0.6631,
+      "step": 47407
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.479024887084961,
+      "learning_rate": 1.3038689568836283e-06,
+      "loss": 0.5136,
+      "step": 47408
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 4.062494277954102,
+      "learning_rate": 1.3036913995218076e-06,
+      "loss": 0.6545,
+      "step": 47409
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.6915950775146484,
+      "learning_rate": 1.3035138524380375e-06,
+      "loss": 0.3245,
+      "step": 47410
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.099797487258911,
+      "learning_rate": 1.3033363156328098e-06,
+      "loss": 0.4465,
+      "step": 47411
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.013660430908203,
+      "learning_rate": 1.30315878910662e-06,
+      "loss": 0.4531,
+      "step": 47412
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.1960842609405518,
+      "learning_rate": 1.3029812728599594e-06,
+      "loss": 0.5942,
+      "step": 47413
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.087514638900757,
+      "learning_rate": 1.3028037668933247e-06,
+      "loss": 0.4092,
+      "step": 47414
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.0239384174346924,
+      "learning_rate": 1.3026262712072062e-06,
+      "loss": 0.5113,
+      "step": 47415
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.282999038696289,
+      "learning_rate": 1.3024487858021007e-06,
+      "loss": 0.4507,
+      "step": 47416
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.6836965084075928,
+      "learning_rate": 1.3022713106784979e-06,
+      "loss": 0.6031,
+      "step": 47417
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.3596885204315186,
+      "learning_rate": 1.3020938458368948e-06,
+      "loss": 0.7208,
+      "step": 47418
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.14662504196167,
+      "learning_rate": 1.3019163912777826e-06,
+      "loss": 0.4953,
+      "step": 47419
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.0696983337402344,
+      "learning_rate": 1.3017389470016567e-06,
+      "loss": 0.4511,
+      "step": 47420
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.7345564365386963,
+      "learning_rate": 1.3015615130090065e-06,
+      "loss": 0.4466,
+      "step": 47421
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.656787872314453,
+      "learning_rate": 1.301384089300331e-06,
+      "loss": 0.7583,
+      "step": 47422
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.097996234893799,
+      "learning_rate": 1.30120667587612e-06,
+      "loss": 0.4374,
+      "step": 47423
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.3312437534332275,
+      "learning_rate": 1.3010292727368685e-06,
+      "loss": 0.8032,
+      "step": 47424
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.042344808578491,
+      "learning_rate": 1.3008518798830672e-06,
+      "loss": 0.5608,
+      "step": 47425
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.685633897781372,
+      "learning_rate": 1.300674497315213e-06,
+      "loss": 0.4397,
+      "step": 47426
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 1.893415927886963,
+      "learning_rate": 1.3004971250337961e-06,
+      "loss": 0.4743,
+      "step": 47427
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.3966751098632812,
+      "learning_rate": 1.3003197630393117e-06,
+      "loss": 0.5312,
+      "step": 47428
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.1672492027282715,
+      "learning_rate": 1.3001424113322508e-06,
+      "loss": 0.6067,
+      "step": 47429
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.8007545471191406,
+      "learning_rate": 1.2999650699131094e-06,
+      "loss": 0.753,
+      "step": 47430
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.3030858039855957,
+      "learning_rate": 1.2997877387823777e-06,
+      "loss": 0.4458,
+      "step": 47431
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 6.902207851409912,
+      "learning_rate": 1.2996104179405517e-06,
+      "loss": 0.5948,
+      "step": 47432
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.2981979846954346,
+      "learning_rate": 1.2994331073881212e-06,
+      "loss": 0.7102,
+      "step": 47433
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.4305367469787598,
+      "learning_rate": 1.2992558071255823e-06,
+      "loss": 0.5552,
+      "step": 47434
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.2325377464294434,
+      "learning_rate": 1.2990785171534253e-06,
+      "loss": 0.4418,
+      "step": 47435
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.7206172943115234,
+      "learning_rate": 1.298901237472146e-06,
+      "loss": 0.5448,
+      "step": 47436
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.4150378704071045,
+      "learning_rate": 1.298723968082235e-06,
+      "loss": 0.7847,
+      "step": 47437
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.784432888031006,
+      "learning_rate": 1.2985467089841868e-06,
+      "loss": 0.7588,
+      "step": 47438
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.2960662841796875,
+      "learning_rate": 1.2983694601784924e-06,
+      "loss": 0.4694,
+      "step": 47439
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.4127633571624756,
+      "learning_rate": 1.298192221665646e-06,
+      "loss": 0.6206,
+      "step": 47440
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.4113433361053467,
+      "learning_rate": 1.2980149934461405e-06,
+      "loss": 0.5828,
+      "step": 47441
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.6284255981445312,
+      "learning_rate": 1.2978377755204696e-06,
+      "loss": 0.6881,
+      "step": 47442
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.6503000259399414,
+      "learning_rate": 1.2976605678891236e-06,
+      "loss": 0.5611,
+      "step": 47443
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.3912692070007324,
+      "learning_rate": 1.297483370552598e-06,
+      "loss": 0.5535,
+      "step": 47444
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.2185747623443604,
+      "learning_rate": 1.2973061835113832e-06,
+      "loss": 0.4387,
+      "step": 47445
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.882899284362793,
+      "learning_rate": 1.2971290067659736e-06,
+      "loss": 0.5311,
+      "step": 47446
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.6418955326080322,
+      "learning_rate": 1.2969518403168602e-06,
+      "loss": 0.4075,
+      "step": 47447
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.899047613143921,
+      "learning_rate": 1.296774684164538e-06,
+      "loss": 0.6936,
+      "step": 47448
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.5627224445343018,
+      "learning_rate": 1.2965975383094965e-06,
+      "loss": 0.7381,
+      "step": 47449
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.3295488357543945,
+      "learning_rate": 1.296420402752231e-06,
+      "loss": 0.7057,
+      "step": 47450
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.5337908267974854,
+      "learning_rate": 1.296243277493232e-06,
+      "loss": 0.5793,
+      "step": 47451
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.4654483795166016,
+      "learning_rate": 1.296066162532993e-06,
+      "loss": 0.743,
+      "step": 47452
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.5404229164123535,
+      "learning_rate": 1.2958890578720075e-06,
+      "loss": 0.6105,
+      "step": 47453
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.396235227584839,
+      "learning_rate": 1.2957119635107658e-06,
+      "loss": 0.6237,
+      "step": 47454
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.0662386417388916,
+      "learning_rate": 1.295534879449763e-06,
+      "loss": 0.6327,
+      "step": 47455
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 5.040616989135742,
+      "learning_rate": 1.2953578056894883e-06,
+      "loss": 0.601,
+      "step": 47456
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.187983274459839,
+      "learning_rate": 1.2951807422304374e-06,
+      "loss": 0.8343,
+      "step": 47457
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.4272005558013916,
+      "learning_rate": 1.2950036890730994e-06,
+      "loss": 0.5249,
+      "step": 47458
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.525878667831421,
+      "learning_rate": 1.2948266462179693e-06,
+      "loss": 0.8733,
+      "step": 47459
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 5.660074710845947,
+      "learning_rate": 1.2946496136655372e-06,
+      "loss": 0.7663,
+      "step": 47460
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 5.002546310424805,
+      "learning_rate": 1.2944725914162981e-06,
+      "loss": 0.6306,
+      "step": 47461
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.3255374431610107,
+      "learning_rate": 1.2942955794707413e-06,
+      "loss": 0.682,
+      "step": 47462
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.6524288654327393,
+      "learning_rate": 1.2941185778293596e-06,
+      "loss": 0.4774,
+      "step": 47463
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.029590606689453,
+      "learning_rate": 1.293941586492647e-06,
+      "loss": 0.431,
+      "step": 47464
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.5426955223083496,
+      "learning_rate": 1.2937646054610947e-06,
+      "loss": 0.7267,
+      "step": 47465
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.3192737102508545,
+      "learning_rate": 1.293587634735194e-06,
+      "loss": 0.5822,
+      "step": 47466
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.5826327800750732,
+      "learning_rate": 1.293410674315439e-06,
+      "loss": 0.8245,
+      "step": 47467
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.0813519954681396,
+      "learning_rate": 1.2932337242023186e-06,
+      "loss": 0.5033,
+      "step": 47468
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.7110722064971924,
+      "learning_rate": 1.2930567843963287e-06,
+      "loss": 0.4882,
+      "step": 47469
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.2385692596435547,
+      "learning_rate": 1.2928798548979571e-06,
+      "loss": 0.7885,
+      "step": 47470
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.9505133628845215,
+      "learning_rate": 1.2927029357076997e-06,
+      "loss": 0.5712,
+      "step": 47471
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.3282899856567383,
+      "learning_rate": 1.2925260268260452e-06,
+      "loss": 0.5892,
+      "step": 47472
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.5317599773406982,
+      "learning_rate": 1.2923491282534884e-06,
+      "loss": 0.6495,
+      "step": 47473
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.4084174633026123,
+      "learning_rate": 1.2921722399905185e-06,
+      "loss": 0.4594,
+      "step": 47474
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.748305320739746,
+      "learning_rate": 1.2919953620376296e-06,
+      "loss": 0.5349,
+      "step": 47475
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.4644932746887207,
+      "learning_rate": 1.291818494395311e-06,
+      "loss": 0.4799,
+      "step": 47476
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.5740277767181396,
+      "learning_rate": 1.2916416370640583e-06,
+      "loss": 0.4329,
+      "step": 47477
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.714738368988037,
+      "learning_rate": 1.291464790044359e-06,
+      "loss": 0.6282,
+      "step": 47478
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.2056286334991455,
+      "learning_rate": 1.2912879533367084e-06,
+      "loss": 0.4955,
+      "step": 47479
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 4.567300319671631,
+      "learning_rate": 1.291111126941595e-06,
+      "loss": 0.754,
+      "step": 47480
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.801032781600952,
+      "learning_rate": 1.2909343108595141e-06,
+      "loss": 0.5797,
+      "step": 47481
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 4.513594627380371,
+      "learning_rate": 1.2907575050909522e-06,
+      "loss": 0.3785,
+      "step": 47482
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.4035985469818115,
+      "learning_rate": 1.290580709636407e-06,
+      "loss": 0.3469,
+      "step": 47483
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.840860605239868,
+      "learning_rate": 1.2904039244963662e-06,
+      "loss": 0.7367,
+      "step": 47484
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.248263120651245,
+      "learning_rate": 1.2902271496713237e-06,
+      "loss": 0.5278,
+      "step": 47485
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.4532155990600586,
+      "learning_rate": 1.2900503851617685e-06,
+      "loss": 0.6495,
+      "step": 47486
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.78009295463562,
+      "learning_rate": 1.2898736309681946e-06,
+      "loss": 0.5052,
+      "step": 47487
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.832443952560425,
+      "learning_rate": 1.289696887091091e-06,
+      "loss": 0.3374,
+      "step": 47488
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.1914522647857666,
+      "learning_rate": 1.2895201535309514e-06,
+      "loss": 0.6217,
+      "step": 47489
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 7.690083026885986,
+      "learning_rate": 1.2893434302882657e-06,
+      "loss": 0.5608,
+      "step": 47490
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.665403127670288,
+      "learning_rate": 1.2891667173635264e-06,
+      "loss": 0.4667,
+      "step": 47491
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.2471706867218018,
+      "learning_rate": 1.2889900147572238e-06,
+      "loss": 0.8425,
+      "step": 47492
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.0703165531158447,
+      "learning_rate": 1.2888133224698508e-06,
+      "loss": 0.4491,
+      "step": 47493
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.9539637565612793,
+      "learning_rate": 1.2886366405018958e-06,
+      "loss": 0.4923,
+      "step": 47494
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 4.275958061218262,
+      "learning_rate": 1.288459968853854e-06,
+      "loss": 1.115,
+      "step": 47495
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.5731990337371826,
+      "learning_rate": 1.288283307526213e-06,
+      "loss": 0.5175,
+      "step": 47496
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.5774343013763428,
+      "learning_rate": 1.288106656519467e-06,
+      "loss": 0.6283,
+      "step": 47497
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.319072723388672,
+      "learning_rate": 1.2879300158341052e-06,
+      "loss": 0.6447,
+      "step": 47498
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 1.9469207525253296,
+      "learning_rate": 1.2877533854706204e-06,
+      "loss": 0.3791,
+      "step": 47499
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.62040114402771,
+      "learning_rate": 1.2875767654295012e-06,
+      "loss": 0.6555,
+      "step": 47500
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.6876587867736816,
+      "learning_rate": 1.2874001557112404e-06,
+      "loss": 0.6636,
+      "step": 47501
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.9342806339263916,
+      "learning_rate": 1.2872235563163287e-06,
+      "loss": 0.7767,
+      "step": 47502
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 4.553009510040283,
+      "learning_rate": 1.2870469672452596e-06,
+      "loss": 0.6088,
+      "step": 47503
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.5214223861694336,
+      "learning_rate": 1.28687038849852e-06,
+      "loss": 0.7119,
+      "step": 47504
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.306697368621826,
+      "learning_rate": 1.2866938200766043e-06,
+      "loss": 0.7129,
+      "step": 47505
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.271317720413208,
+      "learning_rate": 1.286517261980001e-06,
+      "loss": 0.3601,
+      "step": 47506
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.2743396759033203,
+      "learning_rate": 1.2863407142092033e-06,
+      "loss": 0.6433,
+      "step": 47507
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.714653730392456,
+      "learning_rate": 1.2861641767646994e-06,
+      "loss": 0.5447,
+      "step": 47508
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.233992099761963,
+      "learning_rate": 1.285987649646983e-06,
+      "loss": 0.4998,
+      "step": 47509
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.3239800930023193,
+      "learning_rate": 1.2858111328565426e-06,
+      "loss": 0.6803,
+      "step": 47510
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.9105045795440674,
+      "learning_rate": 1.2856346263938712e-06,
+      "loss": 0.6963,
+      "step": 47511
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.942135810852051,
+      "learning_rate": 1.2854581302594576e-06,
+      "loss": 0.7984,
+      "step": 47512
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.366816997528076,
+      "learning_rate": 1.285281644453793e-06,
+      "loss": 0.6794,
+      "step": 47513
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.7595880031585693,
+      "learning_rate": 1.2851051689773696e-06,
+      "loss": 0.482,
+      "step": 47514
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.3815743923187256,
+      "learning_rate": 1.2849287038306763e-06,
+      "loss": 0.5405,
+      "step": 47515
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 4.952095031738281,
+      "learning_rate": 1.2847522490142057e-06,
+      "loss": 0.7859,
+      "step": 47516
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.6119282245635986,
+      "learning_rate": 1.2845758045284461e-06,
+      "loss": 0.4119,
+      "step": 47517
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.644381284713745,
+      "learning_rate": 1.2843993703738905e-06,
+      "loss": 0.5965,
+      "step": 47518
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.5726239681243896,
+      "learning_rate": 1.2842229465510275e-06,
+      "loss": 0.4568,
+      "step": 47519
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.950331449508667,
+      "learning_rate": 1.284046533060349e-06,
+      "loss": 0.9626,
+      "step": 47520
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.1713428497314453,
+      "learning_rate": 1.2838701299023443e-06,
+      "loss": 0.5569,
+      "step": 47521
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.575690984725952,
+      "learning_rate": 1.283693737077506e-06,
+      "loss": 0.6307,
+      "step": 47522
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.6107418537139893,
+      "learning_rate": 1.2835173545863223e-06,
+      "loss": 0.5599,
+      "step": 47523
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 5.667503833770752,
+      "learning_rate": 1.2833409824292836e-06,
+      "loss": 0.5015,
+      "step": 47524
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.7137277126312256,
+      "learning_rate": 1.283164620606882e-06,
+      "loss": 0.748,
+      "step": 47525
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.0313949584960938,
+      "learning_rate": 1.2829882691196088e-06,
+      "loss": 0.6149,
+      "step": 47526
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.5569984912872314,
+      "learning_rate": 1.2828119279679507e-06,
+      "loss": 0.463,
+      "step": 47527
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.2610363960266113,
+      "learning_rate": 1.2826355971524025e-06,
+      "loss": 0.4271,
+      "step": 47528
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.6261849403381348,
+      "learning_rate": 1.2824592766734501e-06,
+      "loss": 0.7305,
+      "step": 47529
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.188836097717285,
+      "learning_rate": 1.2822829665315873e-06,
+      "loss": 0.6468,
+      "step": 47530
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.8919031620025635,
+      "learning_rate": 1.2821066667273013e-06,
+      "loss": 0.6755,
+      "step": 47531
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.409040689468384,
+      "learning_rate": 1.2819303772610858e-06,
+      "loss": 0.7364,
+      "step": 47532
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 1.9206326007843018,
+      "learning_rate": 1.281754098133427e-06,
+      "loss": 0.5959,
+      "step": 47533
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.24552321434021,
+      "learning_rate": 1.281577829344819e-06,
+      "loss": 0.8406,
+      "step": 47534
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.9464480876922607,
+      "learning_rate": 1.2814015708957489e-06,
+      "loss": 0.8378,
+      "step": 47535
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 1.6775054931640625,
+      "learning_rate": 1.2812253227867094e-06,
+      "loss": 0.512,
+      "step": 47536
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 1.7497471570968628,
+      "learning_rate": 1.2810490850181872e-06,
+      "loss": 0.3549,
+      "step": 47537
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.7796008586883545,
+      "learning_rate": 1.2808728575906765e-06,
+      "loss": 0.6289,
+      "step": 47538
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 1.8558176755905151,
+      "learning_rate": 1.2806966405046635e-06,
+      "loss": 0.7267,
+      "step": 47539
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.1065540313720703,
+      "learning_rate": 1.2805204337606408e-06,
+      "loss": 0.7568,
+      "step": 47540
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.0671205520629883,
+      "learning_rate": 1.2803442373590963e-06,
+      "loss": 0.535,
+      "step": 47541
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.397766351699829,
+      "learning_rate": 1.2801680513005227e-06,
+      "loss": 0.6986,
+      "step": 47542
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 1.9651825428009033,
+      "learning_rate": 1.279991875585405e-06,
+      "loss": 0.4028,
+      "step": 47543
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.8505475521087646,
+      "learning_rate": 1.2798157102142394e-06,
+      "loss": 0.6988,
+      "step": 47544
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.410858392715454,
+      "learning_rate": 1.2796395551875117e-06,
+      "loss": 0.6143,
+      "step": 47545
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.532052755355835,
+      "learning_rate": 1.2794634105057136e-06,
+      "loss": 0.6743,
+      "step": 47546
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.0317928791046143,
+      "learning_rate": 1.2792872761693327e-06,
+      "loss": 0.6132,
+      "step": 47547
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.0841548442840576,
+      "learning_rate": 1.2791111521788618e-06,
+      "loss": 0.6444,
+      "step": 47548
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.7156896591186523,
+      "learning_rate": 1.2789350385347872e-06,
+      "loss": 0.4566,
+      "step": 47549
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.3481476306915283,
+      "learning_rate": 1.2787589352376018e-06,
+      "loss": 0.6887,
+      "step": 47550
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.0911548137664795,
+      "learning_rate": 1.278582842287792e-06,
+      "loss": 0.4362,
+      "step": 47551
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.1789724826812744,
+      "learning_rate": 1.2784067596858512e-06,
+      "loss": 0.4088,
+      "step": 47552
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.3377797603607178,
+      "learning_rate": 1.2782306874322658e-06,
+      "loss": 0.5943,
+      "step": 47553
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.57330322265625,
+      "learning_rate": 1.2780546255275278e-06,
+      "loss": 0.6082,
+      "step": 47554
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 1.8371104001998901,
+      "learning_rate": 1.2778785739721238e-06,
+      "loss": 0.305,
+      "step": 47555
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.9392969608306885,
+      "learning_rate": 1.2777025327665472e-06,
+      "loss": 0.5547,
+      "step": 47556
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 4.2006306648254395,
+      "learning_rate": 1.2775265019112837e-06,
+      "loss": 0.6823,
+      "step": 47557
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.1392879486083984,
+      "learning_rate": 1.277350481406826e-06,
+      "loss": 0.623,
+      "step": 47558
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 1.771342158317566,
+      "learning_rate": 1.2771744712536604e-06,
+      "loss": 0.3426,
+      "step": 47559
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.649716854095459,
+      "learning_rate": 1.2769984714522799e-06,
+      "loss": 0.4627,
+      "step": 47560
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.621577739715576,
+      "learning_rate": 1.2768224820031698e-06,
+      "loss": 0.6024,
+      "step": 47561
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.1500768661499023,
+      "learning_rate": 1.2766465029068236e-06,
+      "loss": 0.3886,
+      "step": 47562
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 1.9210162162780762,
+      "learning_rate": 1.2764705341637262e-06,
+      "loss": 0.4049,
+      "step": 47563
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.9806456565856934,
+      "learning_rate": 1.2762945757743716e-06,
+      "loss": 0.4907,
+      "step": 47564
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.5114309787750244,
+      "learning_rate": 1.2761186277392457e-06,
+      "loss": 0.472,
+      "step": 47565
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.4193851947784424,
+      "learning_rate": 1.27594269005884e-06,
+      "loss": 0.5406,
+      "step": 47566
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.006152868270874,
+      "learning_rate": 1.2757667627336418e-06,
+      "loss": 0.6853,
+      "step": 47567
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.5273964405059814,
+      "learning_rate": 1.275590845764142e-06,
+      "loss": 0.553,
+      "step": 47568
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.1051149368286133,
+      "learning_rate": 1.275414939150828e-06,
+      "loss": 0.7209,
+      "step": 47569
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.7697341442108154,
+      "learning_rate": 1.2752390428941907e-06,
+      "loss": 0.6189,
+      "step": 47570
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.634211778640747,
+      "learning_rate": 1.275063156994717e-06,
+      "loss": 0.6384,
+      "step": 47571
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.7251768112182617,
+      "learning_rate": 1.274887281452899e-06,
+      "loss": 0.3787,
+      "step": 47572
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.921398878097534,
+      "learning_rate": 1.274711416269222e-06,
+      "loss": 0.7694,
+      "step": 47573
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.6011033058166504,
+      "learning_rate": 1.2745355614441773e-06,
+      "loss": 0.4486,
+      "step": 47574
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.246295690536499,
+      "learning_rate": 1.274359716978255e-06,
+      "loss": 0.7494,
+      "step": 47575
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.324800968170166,
+      "learning_rate": 1.2741838828719416e-06,
+      "loss": 0.6547,
+      "step": 47576
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.469407081604004,
+      "learning_rate": 1.2740080591257282e-06,
+      "loss": 0.3758,
+      "step": 47577
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.6301631927490234,
+      "learning_rate": 1.273832245740101e-06,
+      "loss": 0.7549,
+      "step": 47578
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.540111541748047,
+      "learning_rate": 1.2736564427155524e-06,
+      "loss": 0.54,
+      "step": 47579
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.429703712463379,
+      "learning_rate": 1.2734806500525676e-06,
+      "loss": 0.7258,
+      "step": 47580
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.205723524093628,
+      "learning_rate": 1.2733048677516385e-06,
+      "loss": 0.5396,
+      "step": 47581
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.405303955078125,
+      "learning_rate": 1.2731290958132509e-06,
+      "loss": 0.6501,
+      "step": 47582
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.1271698474884033,
+      "learning_rate": 1.2729533342378969e-06,
+      "loss": 0.6543,
+      "step": 47583
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.6765284538269043,
+      "learning_rate": 1.272777583026062e-06,
+      "loss": 0.5254,
+      "step": 47584
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.947098731994629,
+      "learning_rate": 1.272601842178236e-06,
+      "loss": 0.715,
+      "step": 47585
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.390254497528076,
+      "learning_rate": 1.2724261116949088e-06,
+      "loss": 0.4587,
+      "step": 47586
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.1487679481506348,
+      "learning_rate": 1.2722503915765688e-06,
+      "loss": 0.4253,
+      "step": 47587
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.4684152603149414,
+      "learning_rate": 1.272074681823703e-06,
+      "loss": 0.5644,
+      "step": 47588
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.705202341079712,
+      "learning_rate": 1.2718989824368028e-06,
+      "loss": 0.7738,
+      "step": 47589
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.4364476203918457,
+      "learning_rate": 1.2717232934163526e-06,
+      "loss": 0.6622,
+      "step": 47590
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.863199472427368,
+      "learning_rate": 1.2715476147628452e-06,
+      "loss": 0.6489,
+      "step": 47591
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.987837553024292,
+      "learning_rate": 1.271371946476766e-06,
+      "loss": 0.5203,
+      "step": 47592
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.17969012260437,
+      "learning_rate": 1.271196288558606e-06,
+      "loss": 0.5601,
+      "step": 47593
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 4.6016154289245605,
+      "learning_rate": 1.2710206410088504e-06,
+      "loss": 0.681,
+      "step": 47594
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.400015115737915,
+      "learning_rate": 1.2708450038279913e-06,
+      "loss": 0.5834,
+      "step": 47595
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.3715710639953613,
+      "learning_rate": 1.2706693770165135e-06,
+      "loss": 0.5891,
+      "step": 47596
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.6023027896881104,
+      "learning_rate": 1.270493760574909e-06,
+      "loss": 0.4342,
+      "step": 47597
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 4.883970737457275,
+      "learning_rate": 1.2703181545036624e-06,
+      "loss": 0.5689,
+      "step": 47598
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.752375602722168,
+      "learning_rate": 1.2701425588032656e-06,
+      "loss": 0.7283,
+      "step": 47599
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.675891637802124,
+      "learning_rate": 1.2699669734742036e-06,
+      "loss": 0.7149,
+      "step": 47600
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.4448561668395996,
+      "learning_rate": 1.2697913985169673e-06,
+      "loss": 0.4361,
+      "step": 47601
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.577131986618042,
+      "learning_rate": 1.2696158339320431e-06,
+      "loss": 0.462,
+      "step": 47602
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.3254857063293457,
+      "learning_rate": 1.2694402797199206e-06,
+      "loss": 0.5477,
+      "step": 47603
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.375607490539551,
+      "learning_rate": 1.2692647358810845e-06,
+      "loss": 0.9108,
+      "step": 47604
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.397803544998169,
+      "learning_rate": 1.2690892024160283e-06,
+      "loss": 0.4692,
+      "step": 47605
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 4.200033664703369,
+      "learning_rate": 1.268913679325236e-06,
+      "loss": 0.9935,
+      "step": 47606
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.184967279434204,
+      "learning_rate": 1.2687381666091986e-06,
+      "loss": 0.5993,
+      "step": 47607
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.6443002223968506,
+      "learning_rate": 1.2685626642684013e-06,
+      "loss": 0.5795,
+      "step": 47608
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.7487030029296875,
+      "learning_rate": 1.2683871723033342e-06,
+      "loss": 0.6489,
+      "step": 47609
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.355513572692871,
+      "learning_rate": 1.2682116907144838e-06,
+      "loss": 0.6037,
+      "step": 47610
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.9013547897338867,
+      "learning_rate": 1.2680362195023398e-06,
+      "loss": 0.6194,
+      "step": 47611
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 5.252048015594482,
+      "learning_rate": 1.2678607586673876e-06,
+      "loss": 0.6677,
+      "step": 47612
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.5603785514831543,
+      "learning_rate": 1.2676853082101182e-06,
+      "loss": 0.7687,
+      "step": 47613
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.1637415885925293,
+      "learning_rate": 1.267509868131016e-06,
+      "loss": 0.6765,
+      "step": 47614
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.7358994483947754,
+      "learning_rate": 1.2673344384305724e-06,
+      "loss": 0.5092,
+      "step": 47615
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.2170636653900146,
+      "learning_rate": 1.267159019109272e-06,
+      "loss": 0.6227,
+      "step": 47616
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.2604265213012695,
+      "learning_rate": 1.2669836101676053e-06,
+      "loss": 0.496,
+      "step": 47617
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.7594940662384033,
+      "learning_rate": 1.266808211606057e-06,
+      "loss": 0.7608,
+      "step": 47618
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.9950168132781982,
+      "learning_rate": 1.2666328234251185e-06,
+      "loss": 0.535,
+      "step": 47619
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.3587453365325928,
+      "learning_rate": 1.2664574456252742e-06,
+      "loss": 0.7574,
+      "step": 47620
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.272653341293335,
+      "learning_rate": 1.2662820782070145e-06,
+      "loss": 0.5907,
+      "step": 47621
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.3948466777801514,
+      "learning_rate": 1.2661067211708239e-06,
+      "loss": 0.6251,
+      "step": 47622
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.5572316646575928,
+      "learning_rate": 1.2659313745171936e-06,
+      "loss": 0.5958,
+      "step": 47623
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.690401792526245,
+      "learning_rate": 1.265756038246606e-06,
+      "loss": 0.65,
+      "step": 47624
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.5131995677948,
+      "learning_rate": 1.2655807123595555e-06,
+      "loss": 1.0099,
+      "step": 47625
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.6759653091430664,
+      "learning_rate": 1.2654053968565238e-06,
+      "loss": 0.5559,
+      "step": 47626
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.500459909439087,
+      "learning_rate": 1.2652300917380028e-06,
+      "loss": 0.6934,
+      "step": 47627
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.043933868408203,
+      "learning_rate": 1.2650547970044764e-06,
+      "loss": 0.7211,
+      "step": 47628
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.6468887329101562,
+      "learning_rate": 1.2648795126564344e-06,
+      "loss": 0.7049,
+      "step": 47629
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.636803388595581,
+      "learning_rate": 1.2647042386943619e-06,
+      "loss": 0.6628,
+      "step": 47630
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.917518138885498,
+      "learning_rate": 1.2645289751187495e-06,
+      "loss": 0.549,
+      "step": 47631
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.4909934997558594,
+      "learning_rate": 1.2643537219300812e-06,
+      "loss": 0.5621,
+      "step": 47632
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.9019691944122314,
+      "learning_rate": 1.264178479128847e-06,
+      "loss": 0.4375,
+      "step": 47633
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.6315860748291016,
+      "learning_rate": 1.2640032467155317e-06,
+      "loss": 0.7766,
+      "step": 47634
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.652832508087158,
+      "learning_rate": 1.2638280246906248e-06,
+      "loss": 0.6492,
+      "step": 47635
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.683032989501953,
+      "learning_rate": 1.2636528130546117e-06,
+      "loss": 0.6927,
+      "step": 47636
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.409288167953491,
+      "learning_rate": 1.26347761180798e-06,
+      "loss": 0.6476,
+      "step": 47637
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.329714059829712,
+      "learning_rate": 1.263302420951219e-06,
+      "loss": 0.5176,
+      "step": 47638
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.984208106994629,
+      "learning_rate": 1.2631272404848128e-06,
+      "loss": 0.5807,
+      "step": 47639
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.406752824783325,
+      "learning_rate": 1.2629520704092512e-06,
+      "loss": 0.5035,
+      "step": 47640
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.0963358879089355,
+      "learning_rate": 1.2627769107250188e-06,
+      "loss": 0.4536,
+      "step": 47641
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.4303226470947266,
+      "learning_rate": 1.2626017614326053e-06,
+      "loss": 0.8195,
+      "step": 47642
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.52717661857605,
+      "learning_rate": 1.2624266225324944e-06,
+      "loss": 0.4209,
+      "step": 47643
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.5637285709381104,
+      "learning_rate": 1.2622514940251761e-06,
+      "loss": 0.483,
+      "step": 47644
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.402604579925537,
+      "learning_rate": 1.2620763759111355e-06,
+      "loss": 0.7235,
+      "step": 47645
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.0794639587402344,
+      "learning_rate": 1.26190126819086e-06,
+      "loss": 0.5768,
+      "step": 47646
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.5219995975494385,
+      "learning_rate": 1.2617261708648382e-06,
+      "loss": 0.6877,
+      "step": 47647
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.4343020915985107,
+      "learning_rate": 1.2615510839335538e-06,
+      "loss": 0.4898,
+      "step": 47648
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.0818560123443604,
+      "learning_rate": 1.2613760073974956e-06,
+      "loss": 0.4817,
+      "step": 47649
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.0743966102600098,
+      "learning_rate": 1.2612009412571518e-06,
+      "loss": 0.8236,
+      "step": 47650
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.454274892807007,
+      "learning_rate": 1.2610258855130058e-06,
+      "loss": 0.6618,
+      "step": 47651
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.9321510791778564,
+      "learning_rate": 1.2608508401655478e-06,
+      "loss": 0.6901,
+      "step": 47652
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.8806915283203125,
+      "learning_rate": 1.2606758052152618e-06,
+      "loss": 0.4905,
+      "step": 47653
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.564868450164795,
+      "learning_rate": 1.2605007806626363e-06,
+      "loss": 0.5311,
+      "step": 47654
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.3321070671081543,
+      "learning_rate": 1.2603257665081564e-06,
+      "loss": 0.5234,
+      "step": 47655
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.802316904067993,
+      "learning_rate": 1.2601507627523107e-06,
+      "loss": 0.7807,
+      "step": 47656
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 4.083763599395752,
+      "learning_rate": 1.259975769395584e-06,
+      "loss": 0.6932,
+      "step": 47657
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.7684292793273926,
+      "learning_rate": 1.2598007864384644e-06,
+      "loss": 0.6471,
+      "step": 47658
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.9252641201019287,
+      "learning_rate": 1.259625813881436e-06,
+      "loss": 0.5447,
+      "step": 47659
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 1.8186157941818237,
+      "learning_rate": 1.2594508517249894e-06,
+      "loss": 0.4726,
+      "step": 47660
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.8346784114837646,
+      "learning_rate": 1.2592758999696063e-06,
+      "loss": 0.4457,
+      "step": 47661
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 4.449652194976807,
+      "learning_rate": 1.2591009586157782e-06,
+      "loss": 0.4074,
+      "step": 47662
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.717270612716675,
+      "learning_rate": 1.2589260276639864e-06,
+      "loss": 0.7177,
+      "step": 47663
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.754772901535034,
+      "learning_rate": 1.2587511071147218e-06,
+      "loss": 0.6611,
+      "step": 47664
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.019869565963745,
+      "learning_rate": 1.2585761969684663e-06,
+      "loss": 0.5399,
+      "step": 47665
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.001697540283203,
+      "learning_rate": 1.2584012972257114e-06,
+      "loss": 0.5363,
+      "step": 47666
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.9860198497772217,
+      "learning_rate": 1.2582264078869395e-06,
+      "loss": 0.832,
+      "step": 47667
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.5307910442352295,
+      "learning_rate": 1.2580515289526395e-06,
+      "loss": 0.5959,
+      "step": 47668
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.59897518157959,
+      "learning_rate": 1.257876660423295e-06,
+      "loss": 0.6316,
+      "step": 47669
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.2776756286621094,
+      "learning_rate": 1.2577018022993952e-06,
+      "loss": 0.6029,
+      "step": 47670
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.3307013511657715,
+      "learning_rate": 1.2575269545814233e-06,
+      "loss": 0.5437,
+      "step": 47671
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.1721878051757812,
+      "learning_rate": 1.2573521172698688e-06,
+      "loss": 0.5028,
+      "step": 47672
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.7173349857330322,
+      "learning_rate": 1.2571772903652141e-06,
+      "loss": 0.6728,
+      "step": 47673
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.962153434753418,
+      "learning_rate": 1.2570024738679493e-06,
+      "loss": 0.3519,
+      "step": 47674
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.5920681953430176,
+      "learning_rate": 1.2568276677785563e-06,
+      "loss": 0.4497,
+      "step": 47675
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.577531576156616,
+      "learning_rate": 1.2566528720975252e-06,
+      "loss": 0.6868,
+      "step": 47676
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 8.59455394744873,
+      "learning_rate": 1.2564780868253384e-06,
+      "loss": 0.7609,
+      "step": 47677
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.7983620166778564,
+      "learning_rate": 1.2563033119624856e-06,
+      "loss": 0.6803,
+      "step": 47678
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.0638701915740967,
+      "learning_rate": 1.2561285475094493e-06,
+      "loss": 0.5376,
+      "step": 47679
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 8.178328514099121,
+      "learning_rate": 1.2559537934667182e-06,
+      "loss": 0.8636,
+      "step": 47680
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.0904977321624756,
+      "learning_rate": 1.2557790498347756e-06,
+      "loss": 0.5727,
+      "step": 47681
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 4.570457458496094,
+      "learning_rate": 1.2556043166141108e-06,
+      "loss": 0.6834,
+      "step": 47682
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.554882526397705,
+      "learning_rate": 1.2554295938052058e-06,
+      "loss": 0.4718,
+      "step": 47683
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.5109479427337646,
+      "learning_rate": 1.2552548814085496e-06,
+      "loss": 0.6761,
+      "step": 47684
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.9732699394226074,
+      "learning_rate": 1.2550801794246242e-06,
+      "loss": 0.6552,
+      "step": 47685
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.612299919128418,
+      "learning_rate": 1.2549054878539207e-06,
+      "loss": 0.6777,
+      "step": 47686
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.8039801120758057,
+      "learning_rate": 1.254730806696921e-06,
+      "loss": 0.75,
+      "step": 47687
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.378679037094116,
+      "learning_rate": 1.254556135954113e-06,
+      "loss": 0.5665,
+      "step": 47688
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.941004514694214,
+      "learning_rate": 1.2543814756259803e-06,
+      "loss": 0.6478,
+      "step": 47689
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 1.9990342855453491,
+      "learning_rate": 1.2542068257130108e-06,
+      "loss": 0.7368,
+      "step": 47690
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.225036859512329,
+      "learning_rate": 1.2540321862156868e-06,
+      "loss": 0.6624,
+      "step": 47691
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.1829590797424316,
+      "learning_rate": 1.2538575571344985e-06,
+      "loss": 0.7328,
+      "step": 47692
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.085477352142334,
+      "learning_rate": 1.2536829384699268e-06,
+      "loss": 0.5134,
+      "step": 47693
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.0069468021392822,
+      "learning_rate": 1.2535083302224616e-06,
+      "loss": 0.3319,
+      "step": 47694
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.5935800075531006,
+      "learning_rate": 1.2533337323925842e-06,
+      "loss": 0.6413,
+      "step": 47695
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.4926598072052,
+      "learning_rate": 1.2531591449807834e-06,
+      "loss": 0.5831,
+      "step": 47696
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.767479419708252,
+      "learning_rate": 1.2529845679875425e-06,
+      "loss": 0.5057,
+      "step": 47697
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.008488178253174,
+      "learning_rate": 1.2528100014133477e-06,
+      "loss": 0.769,
+      "step": 47698
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.5029327869415283,
+      "learning_rate": 1.2526354452586864e-06,
+      "loss": 0.5804,
+      "step": 47699
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 1.8368918895721436,
+      "learning_rate": 1.25246089952404e-06,
+      "loss": 0.4151,
+      "step": 47700
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.206364870071411,
+      "learning_rate": 1.2522863642098977e-06,
+      "loss": 0.681,
+      "step": 47701
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.3377742767333984,
+      "learning_rate": 1.252111839316742e-06,
+      "loss": 0.6843,
+      "step": 47702
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.0120248794555664,
+      "learning_rate": 1.2519373248450601e-06,
+      "loss": 0.6744,
+      "step": 47703
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.820845127105713,
+      "learning_rate": 1.2517628207953352e-06,
+      "loss": 0.4701,
+      "step": 47704
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.398256301879883,
+      "learning_rate": 1.2515883271680557e-06,
+      "loss": 0.6233,
+      "step": 47705
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 4.4568095207214355,
+      "learning_rate": 1.251413843963703e-06,
+      "loss": 0.7537,
+      "step": 47706
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.8565518856048584,
+      "learning_rate": 1.2512393711827643e-06,
+      "loss": 0.599,
+      "step": 47707
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.2727277278900146,
+      "learning_rate": 1.251064908825726e-06,
+      "loss": 0.5684,
+      "step": 47708
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.6319339275360107,
+      "learning_rate": 1.25089045689307e-06,
+      "loss": 0.5545,
+      "step": 47709
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 4.971432685852051,
+      "learning_rate": 1.2507160153852839e-06,
+      "loss": 0.7054,
+      "step": 47710
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.417853832244873,
+      "learning_rate": 1.2505415843028528e-06,
+      "loss": 0.7434,
+      "step": 47711
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.068896532058716,
+      "learning_rate": 1.25036716364626e-06,
+      "loss": 0.9389,
+      "step": 47712
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.670931816101074,
+      "learning_rate": 1.2501927534159925e-06,
+      "loss": 0.6311,
+      "step": 47713
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.6052281856536865,
+      "learning_rate": 1.2500183536125332e-06,
+      "loss": 0.7401,
+      "step": 47714
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.1093015670776367,
+      "learning_rate": 1.249843964236369e-06,
+      "loss": 0.6376,
+      "step": 47715
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.145893096923828,
+      "learning_rate": 1.2496695852879831e-06,
+      "loss": 0.7257,
+      "step": 47716
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.5480287075042725,
+      "learning_rate": 1.249495216767862e-06,
+      "loss": 0.6032,
+      "step": 47717
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.3325159549713135,
+      "learning_rate": 1.2493208586764887e-06,
+      "loss": 0.4927,
+      "step": 47718
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 7.049435138702393,
+      "learning_rate": 1.2491465110143507e-06,
+      "loss": 0.501,
+      "step": 47719
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.349458694458008,
+      "learning_rate": 1.2489721737819293e-06,
+      "loss": 0.6932,
+      "step": 47720
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.4097206592559814,
+      "learning_rate": 1.2487978469797129e-06,
+      "loss": 0.6778,
+      "step": 47721
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.228187084197998,
+      "learning_rate": 1.2486235306081828e-06,
+      "loss": 0.732,
+      "step": 47722
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.9383928775787354,
+      "learning_rate": 1.2484492246678264e-06,
+      "loss": 0.5958,
+      "step": 47723
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.421973705291748,
+      "learning_rate": 1.2482749291591268e-06,
+      "loss": 0.4041,
+      "step": 47724
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.2448740005493164,
+      "learning_rate": 1.24810064408257e-06,
+      "loss": 0.499,
+      "step": 47725
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.0031867027282715,
+      "learning_rate": 1.2479263694386368e-06,
+      "loss": 0.4803,
+      "step": 47726
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 1.9222649335861206,
+      "learning_rate": 1.247752105227818e-06,
+      "loss": 0.2322,
+      "step": 47727
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 4.387913703918457,
+      "learning_rate": 1.2475778514505937e-06,
+      "loss": 0.8816,
+      "step": 47728
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 1.6104214191436768,
+      "learning_rate": 1.2474036081074509e-06,
+      "loss": 0.3079,
+      "step": 47729
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.8115944862365723,
+      "learning_rate": 1.2472293751988713e-06,
+      "loss": 0.54,
+      "step": 47730
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 1.8868000507354736,
+      "learning_rate": 1.2470551527253427e-06,
+      "loss": 0.3474,
+      "step": 47731
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.4048807621002197,
+      "learning_rate": 1.2468809406873467e-06,
+      "loss": 0.4722,
+      "step": 47732
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 1.8447630405426025,
+      "learning_rate": 1.24670673908537e-06,
+      "loss": 0.5872,
+      "step": 47733
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.751099109649658,
+      "learning_rate": 1.2465325479198941e-06,
+      "loss": 0.7356,
+      "step": 47734
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 1.8910081386566162,
+      "learning_rate": 1.2463583671914065e-06,
+      "loss": 0.6312,
+      "step": 47735
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.319477081298828,
+      "learning_rate": 1.2461841969003891e-06,
+      "loss": 0.5178,
+      "step": 47736
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.4160351753234863,
+      "learning_rate": 1.2460100370473287e-06,
+      "loss": 0.6613,
+      "step": 47737
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.113708972930908,
+      "learning_rate": 1.2458358876327066e-06,
+      "loss": 0.6622,
+      "step": 47738
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.374138832092285,
+      "learning_rate": 1.24566174865701e-06,
+      "loss": 0.5978,
+      "step": 47739
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.613563060760498,
+      "learning_rate": 1.24548762012072e-06,
+      "loss": 0.6938,
+      "step": 47740
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.70624041557312,
+      "learning_rate": 1.245313502024324e-06,
+      "loss": 0.6994,
+      "step": 47741
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.9226276874542236,
+      "learning_rate": 1.2451393943683032e-06,
+      "loss": 0.8311,
+      "step": 47742
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.6704964637756348,
+      "learning_rate": 1.2449652971531446e-06,
+      "loss": 0.793,
+      "step": 47743
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.4007327556610107,
+      "learning_rate": 1.2447912103793291e-06,
+      "loss": 0.7052,
+      "step": 47744
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.5426061153411865,
+      "learning_rate": 1.2446171340473445e-06,
+      "loss": 0.4067,
+      "step": 47745
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.2983877658843994,
+      "learning_rate": 1.2444430681576697e-06,
+      "loss": 0.5003,
+      "step": 47746
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 4.325018405914307,
+      "learning_rate": 1.2442690127107948e-06,
+      "loss": 0.6104,
+      "step": 47747
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.7012884616851807,
+      "learning_rate": 1.2440949677071996e-06,
+      "loss": 0.6255,
+      "step": 47748
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 1.5768814086914062,
+      "learning_rate": 1.24392093314737e-06,
+      "loss": 0.5081,
+      "step": 47749
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.39255952835083,
+      "learning_rate": 1.2437469090317884e-06,
+      "loss": 0.4456,
+      "step": 47750
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.9126715660095215,
+      "learning_rate": 1.2435728953609412e-06,
+      "loss": 0.6058,
+      "step": 47751
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.7223961353302,
+      "learning_rate": 1.2433988921353085e-06,
+      "loss": 0.5576,
+      "step": 47752
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.2627739906311035,
+      "learning_rate": 1.243224899355378e-06,
+      "loss": 0.5444,
+      "step": 47753
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.6234023571014404,
+      "learning_rate": 1.2430509170216304e-06,
+      "loss": 0.7292,
+      "step": 47754
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.9560580253601074,
+      "learning_rate": 1.2428769451345517e-06,
+      "loss": 0.7026,
+      "step": 47755
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.824875831604004,
+      "learning_rate": 1.242702983694624e-06,
+      "loss": 0.4851,
+      "step": 47756
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.103571653366089,
+      "learning_rate": 1.242529032702333e-06,
+      "loss": 0.4926,
+      "step": 47757
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.0201056003570557,
+      "learning_rate": 1.2423550921581596e-06,
+      "loss": 0.5729,
+      "step": 47758
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.7233452796936035,
+      "learning_rate": 1.242181162062589e-06,
+      "loss": 0.6375,
+      "step": 47759
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.5724740028381348,
+      "learning_rate": 1.2420072424161068e-06,
+      "loss": 0.7125,
+      "step": 47760
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.645385265350342,
+      "learning_rate": 1.2418333332191929e-06,
+      "loss": 0.6001,
+      "step": 47761
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.7305564880371094,
+      "learning_rate": 1.2416594344723343e-06,
+      "loss": 0.4873,
+      "step": 47762
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.177420139312744,
+      "learning_rate": 1.241485546176011e-06,
+      "loss": 0.5282,
+      "step": 47763
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.3798840045928955,
+      "learning_rate": 1.2413116683307103e-06,
+      "loss": 0.6394,
+      "step": 47764
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.6753926277160645,
+      "learning_rate": 1.2411378009369118e-06,
+      "loss": 0.5531,
+      "step": 47765
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.161524534225464,
+      "learning_rate": 1.2409639439951032e-06,
+      "loss": 0.405,
+      "step": 47766
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.4839563369750977,
+      "learning_rate": 1.2407900975057636e-06,
+      "loss": 0.6412,
+      "step": 47767
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.0078811645507812,
+      "learning_rate": 1.240616261469379e-06,
+      "loss": 0.4444,
+      "step": 47768
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.2038276195526123,
+      "learning_rate": 1.2404424358864336e-06,
+      "loss": 0.8079,
+      "step": 47769
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.4562079906463623,
+      "learning_rate": 1.2402686207574076e-06,
+      "loss": 0.6633,
+      "step": 47770
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.5316810607910156,
+      "learning_rate": 1.2400948160827863e-06,
+      "loss": 0.8001,
+      "step": 47771
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.229417324066162,
+      "learning_rate": 1.2399210218630547e-06,
+      "loss": 0.4087,
+      "step": 47772
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.6214215755462646,
+      "learning_rate": 1.2397472380986925e-06,
+      "loss": 0.481,
+      "step": 47773
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.283162832260132,
+      "learning_rate": 1.2395734647901859e-06,
+      "loss": 0.5673,
+      "step": 47774
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.786999225616455,
+      "learning_rate": 1.2393997019380155e-06,
+      "loss": 0.6461,
+      "step": 47775
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 4.244977951049805,
+      "learning_rate": 1.2392259495426673e-06,
+      "loss": 0.924,
+      "step": 47776
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.012812376022339,
+      "learning_rate": 1.2390522076046213e-06,
+      "loss": 0.5804,
+      "step": 47777
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.026752233505249,
+      "learning_rate": 1.2388784761243639e-06,
+      "loss": 0.4279,
+      "step": 47778
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.531830310821533,
+      "learning_rate": 1.2387047551023746e-06,
+      "loss": 0.5095,
+      "step": 47779
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 1.8862581253051758,
+      "learning_rate": 1.2385310445391401e-06,
+      "loss": 0.4652,
+      "step": 47780
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.816284656524658,
+      "learning_rate": 1.2383573444351404e-06,
+      "loss": 0.7734,
+      "step": 47781
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.3059611320495605,
+      "learning_rate": 1.238183654790861e-06,
+      "loss": 0.5854,
+      "step": 47782
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.394056558609009,
+      "learning_rate": 1.2380099756067821e-06,
+      "loss": 0.5113,
+      "step": 47783
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.772901773452759,
+      "learning_rate": 1.23783630688339e-06,
+      "loss": 0.7079,
+      "step": 47784
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.4634783267974854,
+      "learning_rate": 1.237662648621164e-06,
+      "loss": 0.6243,
+      "step": 47785
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.358458995819092,
+      "learning_rate": 1.2374890008205904e-06,
+      "loss": 0.6016,
+      "step": 47786
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.9967730045318604,
+      "learning_rate": 1.2373153634821472e-06,
+      "loss": 0.602,
+      "step": 47787
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 1.758400797843933,
+      "learning_rate": 1.2371417366063237e-06,
+      "loss": 0.5032,
+      "step": 47788
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.7829692363739014,
+      "learning_rate": 1.236968120193598e-06,
+      "loss": 0.6193,
+      "step": 47789
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 1.7957921028137207,
+      "learning_rate": 1.2367945142444554e-06,
+      "loss": 0.4167,
+      "step": 47790
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 4.504117012023926,
+      "learning_rate": 1.2366209187593765e-06,
+      "loss": 0.7109,
+      "step": 47791
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.7774031162261963,
+      "learning_rate": 1.2364473337388467e-06,
+      "loss": 0.5384,
+      "step": 47792
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.3556835651397705,
+      "learning_rate": 1.2362737591833446e-06,
+      "loss": 0.5611,
+      "step": 47793
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.035360097885132,
+      "learning_rate": 1.2361001950933577e-06,
+      "loss": 0.7803,
+      "step": 47794
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.6883738040924072,
+      "learning_rate": 1.235926641469364e-06,
+      "loss": 0.6359,
+      "step": 47795
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.8150229454040527,
+      "learning_rate": 1.2357530983118498e-06,
+      "loss": 0.5856,
+      "step": 47796
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.1985907554626465,
+      "learning_rate": 1.2355795656212948e-06,
+      "loss": 0.6356,
+      "step": 47797
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.6916189193725586,
+      "learning_rate": 1.2354060433981845e-06,
+      "loss": 0.6807,
+      "step": 47798
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.298140525817871,
+      "learning_rate": 1.235232531642998e-06,
+      "loss": 0.8052,
+      "step": 47799
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.7010960578918457,
+      "learning_rate": 1.2350590303562215e-06,
+      "loss": 0.6692,
+      "step": 47800
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.3856496810913086,
+      "learning_rate": 1.2348855395383335e-06,
+      "loss": 0.6751,
+      "step": 47801
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.1421682834625244,
+      "learning_rate": 1.23471205918982e-06,
+      "loss": 0.5492,
+      "step": 47802
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.981426477432251,
+      "learning_rate": 1.23453858931116e-06,
+      "loss": 0.5759,
+      "step": 47803
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.517554998397827,
+      "learning_rate": 1.234365129902839e-06,
+      "loss": 0.6521,
+      "step": 47804
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 1.989172101020813,
+      "learning_rate": 1.234191680965337e-06,
+      "loss": 0.4087,
+      "step": 47805
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.5434343814849854,
+      "learning_rate": 1.2340182424991388e-06,
+      "loss": 0.4819,
+      "step": 47806
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.6371219158172607,
+      "learning_rate": 1.2338448145047216e-06,
+      "loss": 0.6325,
+      "step": 47807
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.1646976470947266,
+      "learning_rate": 1.2336713969825742e-06,
+      "loss": 0.4808,
+      "step": 47808
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 4.223740100860596,
+      "learning_rate": 1.2334979899331744e-06,
+      "loss": 0.5328,
+      "step": 47809
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.4239819049835205,
+      "learning_rate": 1.233324593357007e-06,
+      "loss": 0.4322,
+      "step": 47810
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.6077663898468018,
+      "learning_rate": 1.233151207254552e-06,
+      "loss": 0.6767,
+      "step": 47811
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.9599061012268066,
+      "learning_rate": 1.2329778316262937e-06,
+      "loss": 0.4789,
+      "step": 47812
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.3076224327087402,
+      "learning_rate": 1.2328044664727106e-06,
+      "loss": 0.5166,
+      "step": 47813
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 7.370284080505371,
+      "learning_rate": 1.232631111794289e-06,
+      "loss": 0.7854,
+      "step": 47814
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.7337865829467773,
+      "learning_rate": 1.2324577675915077e-06,
+      "loss": 0.7361,
+      "step": 47815
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.939467191696167,
+      "learning_rate": 1.2322844338648516e-06,
+      "loss": 0.2451,
+      "step": 47816
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.5190820693969727,
+      "learning_rate": 1.2321111106147993e-06,
+      "loss": 0.5881,
+      "step": 47817
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.5765106678009033,
+      "learning_rate": 1.2319377978418362e-06,
+      "loss": 0.7118,
+      "step": 47818
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.6181881427764893,
+      "learning_rate": 1.231764495546441e-06,
+      "loss": 0.5552,
+      "step": 47819
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.8648054599761963,
+      "learning_rate": 1.2315912037290984e-06,
+      "loss": 0.5299,
+      "step": 47820
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.8311727046966553,
+      "learning_rate": 1.2314179223902878e-06,
+      "loss": 0.8585,
+      "step": 47821
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.3547542095184326,
+      "learning_rate": 1.2312446515304921e-06,
+      "loss": 0.4794,
+      "step": 47822
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.509427547454834,
+      "learning_rate": 1.2310713911501942e-06,
+      "loss": 0.6557,
+      "step": 47823
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.9525773525238037,
+      "learning_rate": 1.2308981412498744e-06,
+      "loss": 0.4029,
+      "step": 47824
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.9358513355255127,
+      "learning_rate": 1.2307249018300156e-06,
+      "loss": 0.971,
+      "step": 47825
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 5.08320951461792,
+      "learning_rate": 1.2305516728910977e-06,
+      "loss": 0.4653,
+      "step": 47826
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.4737253189086914,
+      "learning_rate": 1.230378454433605e-06,
+      "loss": 0.6153,
+      "step": 47827
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.260706901550293,
+      "learning_rate": 1.2302052464580162e-06,
+      "loss": 0.5214,
+      "step": 47828
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.8881869316101074,
+      "learning_rate": 1.2300320489648142e-06,
+      "loss": 0.4855,
+      "step": 47829
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.380859375,
+      "learning_rate": 1.2298588619544827e-06,
+      "loss": 0.6726,
+      "step": 47830
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 6.539068222045898,
+      "learning_rate": 1.2296856854274991e-06,
+      "loss": 0.5504,
+      "step": 47831
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 4.744349002838135,
+      "learning_rate": 1.2295125193843493e-06,
+      "loss": 0.4709,
+      "step": 47832
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.0302131175994873,
+      "learning_rate": 1.2293393638255113e-06,
+      "loss": 0.703,
+      "step": 47833
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.3927409648895264,
+      "learning_rate": 1.229166218751468e-06,
+      "loss": 0.3805,
+      "step": 47834
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.290393590927124,
+      "learning_rate": 1.2289930841627024e-06,
+      "loss": 0.4946,
+      "step": 47835
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.204237699508667,
+      "learning_rate": 1.228819960059693e-06,
+      "loss": 0.7725,
+      "step": 47836
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.4634058475494385,
+      "learning_rate": 1.2286468464429236e-06,
+      "loss": 0.617,
+      "step": 47837
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 5.420424938201904,
+      "learning_rate": 1.228473743312873e-06,
+      "loss": 0.6965,
+      "step": 47838
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.001283645629883,
+      "learning_rate": 1.2283006506700257e-06,
+      "loss": 0.506,
+      "step": 47839
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.8920342922210693,
+      "learning_rate": 1.2281275685148603e-06,
+      "loss": 0.6193,
+      "step": 47840
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.6681103706359863,
+      "learning_rate": 1.2279544968478601e-06,
+      "loss": 0.5933,
+      "step": 47841
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.9885571002960205,
+      "learning_rate": 1.2277814356695044e-06,
+      "loss": 0.6754,
+      "step": 47842
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.2333478927612305,
+      "learning_rate": 1.2276083849802768e-06,
+      "loss": 0.5966,
+      "step": 47843
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.5229299068450928,
+      "learning_rate": 1.2274353447806552e-06,
+      "loss": 0.734,
+      "step": 47844
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.7979490756988525,
+      "learning_rate": 1.2272623150711244e-06,
+      "loss": 0.7587,
+      "step": 47845
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 8.899096488952637,
+      "learning_rate": 1.2270892958521625e-06,
+      "loss": 0.5928,
+      "step": 47846
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.0290749073028564,
+      "learning_rate": 1.226916287124253e-06,
+      "loss": 0.6021,
+      "step": 47847
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.8852310180664062,
+      "learning_rate": 1.2267432888878732e-06,
+      "loss": 0.5989,
+      "step": 47848
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.670056104660034,
+      "learning_rate": 1.2265703011435098e-06,
+      "loss": 0.6169,
+      "step": 47849
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.313336133956909,
+      "learning_rate": 1.2263973238916393e-06,
+      "loss": 0.514,
+      "step": 47850
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.9720165729522705,
+      "learning_rate": 1.226224357132746e-06,
+      "loss": 0.8484,
+      "step": 47851
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 6.004665374755859,
+      "learning_rate": 1.226051400867307e-06,
+      "loss": 0.5008,
+      "step": 47852
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.2009201049804688,
+      "learning_rate": 1.2258784550958075e-06,
+      "loss": 0.6442,
+      "step": 47853
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.1492631435394287,
+      "learning_rate": 1.225705519818724e-06,
+      "loss": 0.3738,
+      "step": 47854
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.3501040935516357,
+      "learning_rate": 1.2255325950365416e-06,
+      "loss": 0.6422,
+      "step": 47855
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 6.96660041809082,
+      "learning_rate": 1.225359680749737e-06,
+      "loss": 0.5629,
+      "step": 47856
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.87727689743042,
+      "learning_rate": 1.225186776958795e-06,
+      "loss": 0.7054,
+      "step": 47857
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.7579102516174316,
+      "learning_rate": 1.2250138836641928e-06,
+      "loss": 0.7661,
+      "step": 47858
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.2480623722076416,
+      "learning_rate": 1.2248410008664147e-06,
+      "loss": 0.629,
+      "step": 47859
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.19260835647583,
+      "learning_rate": 1.2246681285659378e-06,
+      "loss": 0.6274,
+      "step": 47860
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.4285435676574707,
+      "learning_rate": 1.2244952667632464e-06,
+      "loss": 0.4363,
+      "step": 47861
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.273528814315796,
+      "learning_rate": 1.2243224154588174e-06,
+      "loss": 0.626,
+      "step": 47862
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.3841309547424316,
+      "learning_rate": 1.2241495746531351e-06,
+      "loss": 0.6451,
+      "step": 47863
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.165706157684326,
+      "learning_rate": 1.2239767443466766e-06,
+      "loss": 0.6527,
+      "step": 47864
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.169552803039551,
+      "learning_rate": 1.2238039245399264e-06,
+      "loss": 0.4622,
+      "step": 47865
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.613316535949707,
+      "learning_rate": 1.2236311152333607e-06,
+      "loss": 0.5945,
+      "step": 47866
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.933102607727051,
+      "learning_rate": 1.2234583164274639e-06,
+      "loss": 0.6975,
+      "step": 47867
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.6624743938446045,
+      "learning_rate": 1.2232855281227118e-06,
+      "loss": 0.5896,
+      "step": 47868
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.2576992511749268,
+      "learning_rate": 1.2231127503195911e-06,
+      "loss": 0.455,
+      "step": 47869
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.8462908267974854,
+      "learning_rate": 1.2229399830185774e-06,
+      "loss": 0.603,
+      "step": 47870
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.858027458190918,
+      "learning_rate": 1.222767226220154e-06,
+      "loss": 0.4622,
+      "step": 47871
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.292189359664917,
+      "learning_rate": 1.2225944799247986e-06,
+      "loss": 0.577,
+      "step": 47872
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 4.115898609161377,
+      "learning_rate": 1.2224217441329944e-06,
+      "loss": 0.5726,
+      "step": 47873
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.4696807861328125,
+      "learning_rate": 1.2222490188452185e-06,
+      "loss": 0.5381,
+      "step": 47874
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.1284399032592773,
+      "learning_rate": 1.2220763040619542e-06,
+      "loss": 0.6115,
+      "step": 47875
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.38967227935791,
+      "learning_rate": 1.2219035997836797e-06,
+      "loss": 0.5043,
+      "step": 47876
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.6050498485565186,
+      "learning_rate": 1.2217309060108767e-06,
+      "loss": 0.5339,
+      "step": 47877
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 1.9186822175979614,
+      "learning_rate": 1.2215582227440237e-06,
+      "loss": 0.6195,
+      "step": 47878
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.2772867679595947,
+      "learning_rate": 1.2213855499836035e-06,
+      "loss": 0.4254,
+      "step": 47879
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.190833330154419,
+      "learning_rate": 1.2212128877300927e-06,
+      "loss": 0.6455,
+      "step": 47880
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.7197043895721436,
+      "learning_rate": 1.2210402359839747e-06,
+      "loss": 0.5178,
+      "step": 47881
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.3098394870758057,
+      "learning_rate": 1.2208675947457272e-06,
+      "loss": 0.68,
+      "step": 47882
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.4635958671569824,
+      "learning_rate": 1.2206949640158306e-06,
+      "loss": 0.5475,
+      "step": 47883
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.7366089820861816,
+      "learning_rate": 1.2205223437947673e-06,
+      "loss": 0.6473,
+      "step": 47884
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.592191219329834,
+      "learning_rate": 1.2203497340830145e-06,
+      "loss": 0.657,
+      "step": 47885
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.8204715251922607,
+      "learning_rate": 1.2201771348810543e-06,
+      "loss": 0.6863,
+      "step": 47886
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.0370032787323,
+      "learning_rate": 1.220004546189364e-06,
+      "loss": 0.5955,
+      "step": 47887
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.8556501865386963,
+      "learning_rate": 1.2198319680084264e-06,
+      "loss": 0.39,
+      "step": 47888
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.7058792114257812,
+      "learning_rate": 1.2196594003387186e-06,
+      "loss": 0.7564,
+      "step": 47889
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.4799439907073975,
+      "learning_rate": 1.2194868431807216e-06,
+      "loss": 0.6592,
+      "step": 47890
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.1754820346832275,
+      "learning_rate": 1.2193142965349175e-06,
+      "loss": 0.5723,
+      "step": 47891
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.2541964054107666,
+      "learning_rate": 1.2191417604017824e-06,
+      "loss": 0.6085,
+      "step": 47892
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.618677854537964,
+      "learning_rate": 1.2189692347817994e-06,
+      "loss": 0.4318,
+      "step": 47893
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 1.930888056755066,
+      "learning_rate": 1.2187967196754448e-06,
+      "loss": 0.5237,
+      "step": 47894
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.8290693759918213,
+      "learning_rate": 1.2186242150832002e-06,
+      "loss": 0.7939,
+      "step": 47895
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.452383041381836,
+      "learning_rate": 1.2184517210055463e-06,
+      "loss": 0.7088,
+      "step": 47896
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.8699333667755127,
+      "learning_rate": 1.2182792374429603e-06,
+      "loss": 0.4562,
+      "step": 47897
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.755885601043701,
+      "learning_rate": 1.2181067643959244e-06,
+      "loss": 0.8381,
+      "step": 47898
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.9407732486724854,
+      "learning_rate": 1.2179343018649153e-06,
+      "loss": 0.5893,
+      "step": 47899
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 5.045597553253174,
+      "learning_rate": 1.2177618498504162e-06,
+      "loss": 0.5891,
+      "step": 47900
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.212608814239502,
+      "learning_rate": 1.2175894083529028e-06,
+      "loss": 0.7428,
+      "step": 47901
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.668245792388916,
+      "learning_rate": 1.2174169773728572e-06,
+      "loss": 0.6093,
+      "step": 47902
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.238662004470825,
+      "learning_rate": 1.2172445569107572e-06,
+      "loss": 0.546,
+      "step": 47903
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 4.844436168670654,
+      "learning_rate": 1.2170721469670843e-06,
+      "loss": 0.5414,
+      "step": 47904
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.4612369537353516,
+      "learning_rate": 1.216899747542315e-06,
+      "loss": 0.5353,
+      "step": 47905
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.3068199157714844,
+      "learning_rate": 1.2167273586369322e-06,
+      "loss": 0.6448,
+      "step": 47906
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.6668014526367188,
+      "learning_rate": 1.2165549802514115e-06,
+      "loss": 0.457,
+      "step": 47907
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.0794517993927,
+      "learning_rate": 1.216382612386236e-06,
+      "loss": 0.5127,
+      "step": 47908
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.4360804557800293,
+      "learning_rate": 1.2162102550418798e-06,
+      "loss": 0.5414,
+      "step": 47909
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 1.9430726766586304,
+      "learning_rate": 1.2160379082188284e-06,
+      "loss": 0.6014,
+      "step": 47910
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 4.711847305297852,
+      "learning_rate": 1.2158655719175565e-06,
+      "loss": 0.595,
+      "step": 47911
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.274538040161133,
+      "learning_rate": 1.2156932461385468e-06,
+      "loss": 0.634,
+      "step": 47912
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.604363441467285,
+      "learning_rate": 1.2155209308822746e-06,
+      "loss": 0.5619,
+      "step": 47913
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.3581364154815674,
+      "learning_rate": 1.2153486261492225e-06,
+      "loss": 0.6445,
+      "step": 47914
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.399538278579712,
+      "learning_rate": 1.2151763319398667e-06,
+      "loss": 0.5382,
+      "step": 47915
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 5.464311599731445,
+      "learning_rate": 1.2150040482546893e-06,
+      "loss": 0.3615,
+      "step": 47916
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.124809980392456,
+      "learning_rate": 1.2148317750941663e-06,
+      "loss": 0.4403,
+      "step": 47917
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.3170340061187744,
+      "learning_rate": 1.2146595124587796e-06,
+      "loss": 0.417,
+      "step": 47918
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.325840711593628,
+      "learning_rate": 1.2144872603490055e-06,
+      "loss": 0.7054,
+      "step": 47919
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.660341501235962,
+      "learning_rate": 1.2143150187653252e-06,
+      "loss": 0.4653,
+      "step": 47920
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.093798875808716,
+      "learning_rate": 1.2141427877082157e-06,
+      "loss": 0.6238,
+      "step": 47921
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.5735087394714355,
+      "learning_rate": 1.2139705671781581e-06,
+      "loss": 0.555,
+      "step": 47922
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.9499847888946533,
+      "learning_rate": 1.2137983571756285e-06,
+      "loss": 0.7228,
+      "step": 47923
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 7.759801864624023,
+      "learning_rate": 1.213626157701109e-06,
+      "loss": 0.4774,
+      "step": 47924
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.3613016605377197,
+      "learning_rate": 1.213453968755075e-06,
+      "loss": 0.4798,
+      "step": 47925
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.7977373600006104,
+      "learning_rate": 1.213281790338009e-06,
+      "loss": 0.4456,
+      "step": 47926
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.58305287361145,
+      "learning_rate": 1.2131096224503857e-06,
+      "loss": 0.6579,
+      "step": 47927
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 8.568351745605469,
+      "learning_rate": 1.2129374650926884e-06,
+      "loss": 0.4981,
+      "step": 47928
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.082160472869873,
+      "learning_rate": 1.2127653182653899e-06,
+      "loss": 0.5797,
+      "step": 47929
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.297855854034424,
+      "learning_rate": 1.2125931819689745e-06,
+      "loss": 0.6514,
+      "step": 47930
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 1.9104961156845093,
+      "learning_rate": 1.2124210562039179e-06,
+      "loss": 0.4688,
+      "step": 47931
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.350581407546997,
+      "learning_rate": 1.2122489409707011e-06,
+      "loss": 0.5062,
+      "step": 47932
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.0754168033599854,
+      "learning_rate": 1.2120768362697993e-06,
+      "loss": 0.6585,
+      "step": 47933
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.9631102085113525,
+      "learning_rate": 1.2119047421016944e-06,
+      "loss": 0.6789,
+      "step": 47934
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.17153263092041,
+      "learning_rate": 1.211732658466862e-06,
+      "loss": 0.7257,
+      "step": 47935
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.337373733520508,
+      "learning_rate": 1.211560585365783e-06,
+      "loss": 0.6205,
+      "step": 47936
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.181823253631592,
+      "learning_rate": 1.2113885227989342e-06,
+      "loss": 0.5595,
+      "step": 47937
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.1424498558044434,
+      "learning_rate": 1.2112164707667956e-06,
+      "loss": 0.4078,
+      "step": 47938
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.238525390625,
+      "learning_rate": 1.2110444292698431e-06,
+      "loss": 0.5613,
+      "step": 47939
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.6593079566955566,
+      "learning_rate": 1.210872398308558e-06,
+      "loss": 0.7553,
+      "step": 47940
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.3221023082733154,
+      "learning_rate": 1.2107003778834165e-06,
+      "loss": 0.6966,
+      "step": 47941
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.341600179672241,
+      "learning_rate": 1.2105283679948986e-06,
+      "loss": 0.6113,
+      "step": 47942
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.8245327472686768,
+      "learning_rate": 1.2103563686434806e-06,
+      "loss": 0.6347,
+      "step": 47943
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.1908445358276367,
+      "learning_rate": 1.210184379829643e-06,
+      "loss": 0.3656,
+      "step": 47944
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.4532649517059326,
+      "learning_rate": 1.2100124015538622e-06,
+      "loss": 0.6245,
+      "step": 47945
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.266418695449829,
+      "learning_rate": 1.2098404338166165e-06,
+      "loss": 0.5056,
+      "step": 47946
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.5041797161102295,
+      "learning_rate": 1.2096684766183864e-06,
+      "loss": 0.6354,
+      "step": 47947
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.979870557785034,
+      "learning_rate": 1.2094965299596466e-06,
+      "loss": 0.4645,
+      "step": 47948
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.560331106185913,
+      "learning_rate": 1.2093245938408792e-06,
+      "loss": 0.6277,
+      "step": 47949
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.1405410766601562,
+      "learning_rate": 1.2091526682625582e-06,
+      "loss": 0.6419,
+      "step": 47950
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.3011016845703125,
+      "learning_rate": 1.2089807532251635e-06,
+      "loss": 0.6017,
+      "step": 47951
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.2259299755096436,
+      "learning_rate": 1.2088088487291743e-06,
+      "loss": 0.7017,
+      "step": 47952
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.2784197330474854,
+      "learning_rate": 1.2086369547750665e-06,
+      "loss": 0.5495,
+      "step": 47953
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.7267489433288574,
+      "learning_rate": 1.2084650713633206e-06,
+      "loss": 0.5527,
+      "step": 47954
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.5582470893859863,
+      "learning_rate": 1.2082931984944113e-06,
+      "loss": 0.5951,
+      "step": 47955
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.6919643878936768,
+      "learning_rate": 1.2081213361688181e-06,
+      "loss": 0.786,
+      "step": 47956
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.1786186695098877,
+      "learning_rate": 1.207949484387021e-06,
+      "loss": 0.5683,
+      "step": 47957
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.5993096828460693,
+      "learning_rate": 1.2077776431494937e-06,
+      "loss": 0.5038,
+      "step": 47958
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.4937820434570312,
+      "learning_rate": 1.207605812456718e-06,
+      "loss": 0.5659,
+      "step": 47959
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.9670042991638184,
+      "learning_rate": 1.2074339923091682e-06,
+      "loss": 0.5252,
+      "step": 47960
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.4006872177124023,
+      "learning_rate": 1.2072621827073256e-06,
+      "loss": 0.6904,
+      "step": 47961
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.2247321605682373,
+      "learning_rate": 1.207090383651664e-06,
+      "loss": 0.5398,
+      "step": 47962
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.186713218688965,
+      "learning_rate": 1.2069185951426654e-06,
+      "loss": 0.5837,
+      "step": 47963
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.7039995193481445,
+      "learning_rate": 1.2067468171808033e-06,
+      "loss": 0.6057,
+      "step": 47964
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.703965425491333,
+      "learning_rate": 1.2065750497665591e-06,
+      "loss": 0.6843,
+      "step": 47965
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.7785677909851074,
+      "learning_rate": 1.2064032929004065e-06,
+      "loss": 0.6952,
+      "step": 47966
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.2770447731018066,
+      "learning_rate": 1.2062315465828273e-06,
+      "loss": 0.6335,
+      "step": 47967
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.3402087688446045,
+      "learning_rate": 1.2060598108142952e-06,
+      "loss": 0.4636,
+      "step": 47968
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.166292190551758,
+      "learning_rate": 1.2058880855952914e-06,
+      "loss": 0.7145,
+      "step": 47969
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.7311129570007324,
+      "learning_rate": 1.2057163709262882e-06,
+      "loss": 0.555,
+      "step": 47970
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.439249277114868,
+      "learning_rate": 1.2055446668077698e-06,
+      "loss": 0.7058,
+      "step": 47971
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.918078660964966,
+      "learning_rate": 1.205372973240208e-06,
+      "loss": 0.5606,
+      "step": 47972
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.1006104946136475,
+      "learning_rate": 1.2052012902240844e-06,
+      "loss": 0.5252,
+      "step": 47973
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.1887717247009277,
+      "learning_rate": 1.2050296177598731e-06,
+      "loss": 0.4444,
+      "step": 47974
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.600501298904419,
+      "learning_rate": 1.2048579558480544e-06,
+      "loss": 0.637,
+      "step": 47975
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.5577123165130615,
+      "learning_rate": 1.2046863044891022e-06,
+      "loss": 0.5845,
+      "step": 47976
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.038968801498413,
+      "learning_rate": 1.204514663683497e-06,
+      "loss": 0.3463,
+      "step": 47977
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.679590940475464,
+      "learning_rate": 1.2043430334317136e-06,
+      "loss": 0.6038,
+      "step": 47978
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.8202009201049805,
+      "learning_rate": 1.2041714137342319e-06,
+      "loss": 0.4938,
+      "step": 47979
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.6052606105804443,
+      "learning_rate": 1.2039998045915258e-06,
+      "loss": 0.875,
+      "step": 47980
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.6021289825439453,
+      "learning_rate": 1.2038282060040758e-06,
+      "loss": 0.6335,
+      "step": 47981
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.0675952434539795,
+      "learning_rate": 1.2036566179723558e-06,
+      "loss": 0.5918,
+      "step": 47982
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 5.557544708251953,
+      "learning_rate": 1.2034850404968462e-06,
+      "loss": 0.6669,
+      "step": 47983
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.294536590576172,
+      "learning_rate": 1.203313473578021e-06,
+      "loss": 0.4758,
+      "step": 47984
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.83487868309021,
+      "learning_rate": 1.2031419172163605e-06,
+      "loss": 0.5996,
+      "step": 47985
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 1.9731470346450806,
+      "learning_rate": 1.2029703714123381e-06,
+      "loss": 0.4192,
+      "step": 47986
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.6668646335601807,
+      "learning_rate": 1.2027988361664345e-06,
+      "loss": 0.588,
+      "step": 47987
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.7372982501983643,
+      "learning_rate": 1.2026273114791231e-06,
+      "loss": 0.4515,
+      "step": 47988
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.367757558822632,
+      "learning_rate": 1.2024557973508844e-06,
+      "loss": 0.5887,
+      "step": 47989
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 1.9357397556304932,
+      "learning_rate": 1.2022842937821905e-06,
+      "loss": 0.5684,
+      "step": 47990
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.1041271686553955,
+      "learning_rate": 1.2021128007735238e-06,
+      "loss": 0.4905,
+      "step": 47991
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.729520559310913,
+      "learning_rate": 1.201941318325358e-06,
+      "loss": 0.5867,
+      "step": 47992
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 4.414118766784668,
+      "learning_rate": 1.2017698464381716e-06,
+      "loss": 0.5109,
+      "step": 47993
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 4.320833206176758,
+      "learning_rate": 1.201598385112439e-06,
+      "loss": 0.3743,
+      "step": 47994
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.5430071353912354,
+      "learning_rate": 1.2014269343486402e-06,
+      "loss": 0.6905,
+      "step": 47995
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.251938819885254,
+      "learning_rate": 1.2012554941472482e-06,
+      "loss": 0.4745,
+      "step": 47996
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.0784382820129395,
+      "learning_rate": 1.2010840645087428e-06,
+      "loss": 0.7278,
+      "step": 47997
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.9204394817352295,
+      "learning_rate": 1.200912645433599e-06,
+      "loss": 0.4744,
+      "step": 47998
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.60300874710083,
+      "learning_rate": 1.2007412369222949e-06,
+      "loss": 0.425,
+      "step": 47999
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.7236239910125732,
+      "learning_rate": 1.2005698389753045e-06,
+      "loss": 0.6479,
+      "step": 48000
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.5357422828674316,
+      "learning_rate": 1.2003984515931077e-06,
+      "loss": 0.3805,
+      "step": 48001
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.60917067527771,
+      "learning_rate": 1.200227074776178e-06,
+      "loss": 0.4926,
+      "step": 48002
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.640561819076538,
+      "learning_rate": 1.2000557085249948e-06,
+      "loss": 0.6192,
+      "step": 48003
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.2610244750976562,
+      "learning_rate": 1.1998843528400321e-06,
+      "loss": 0.4194,
+      "step": 48004
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.1162092685699463,
+      "learning_rate": 1.1997130077217684e-06,
+      "loss": 0.8432,
+      "step": 48005
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 1.999771237373352,
+      "learning_rate": 1.1995416731706782e-06,
+      "loss": 0.5345,
+      "step": 48006
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.1278703212738037,
+      "learning_rate": 1.1993703491872388e-06,
+      "loss": 0.4899,
+      "step": 48007
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.925697088241577,
+      "learning_rate": 1.199199035771928e-06,
+      "loss": 0.7826,
+      "step": 48008
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.707091808319092,
+      "learning_rate": 1.1990277329252193e-06,
+      "loss": 0.395,
+      "step": 48009
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.6368377208709717,
+      "learning_rate": 1.1988564406475923e-06,
+      "loss": 0.5939,
+      "step": 48010
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.7329084873199463,
+      "learning_rate": 1.1986851589395198e-06,
+      "loss": 0.6627,
+      "step": 48011
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.5414419174194336,
+      "learning_rate": 1.19851388780148e-06,
+      "loss": 0.4911,
+      "step": 48012
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.0288095474243164,
+      "learning_rate": 1.198342627233951e-06,
+      "loss": 0.6075,
+      "step": 48013
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 1.761100172996521,
+      "learning_rate": 1.1981713772374048e-06,
+      "loss": 0.4193,
+      "step": 48014
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.4552319049835205,
+      "learning_rate": 1.1980001378123217e-06,
+      "loss": 0.4883,
+      "step": 48015
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.120049238204956,
+      "learning_rate": 1.1978289089591749e-06,
+      "loss": 0.475,
+      "step": 48016
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.3340885639190674,
+      "learning_rate": 1.197657690678442e-06,
+      "loss": 0.4411,
+      "step": 48017
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.9057118892669678,
+      "learning_rate": 1.197486482970598e-06,
+      "loss": 0.6375,
+      "step": 48018
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.6961171627044678,
+      "learning_rate": 1.1973152858361198e-06,
+      "loss": 0.3256,
+      "step": 48019
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.060392379760742,
+      "learning_rate": 1.197144099275484e-06,
+      "loss": 0.7179,
+      "step": 48020
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 1.7053507566452026,
+      "learning_rate": 1.1969729232891646e-06,
+      "loss": 0.474,
+      "step": 48021
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.7797794342041016,
+      "learning_rate": 1.196801757877641e-06,
+      "loss": 0.5252,
+      "step": 48022
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.501737356185913,
+      "learning_rate": 1.1966306030413848e-06,
+      "loss": 0.6947,
+      "step": 48023
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.63615345954895,
+      "learning_rate": 1.1964594587808759e-06,
+      "loss": 0.4485,
+      "step": 48024
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.5402920246124268,
+      "learning_rate": 1.1962883250965867e-06,
+      "loss": 0.6667,
+      "step": 48025
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.9634106159210205,
+      "learning_rate": 1.196117201988996e-06,
+      "loss": 0.6584,
+      "step": 48026
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.671823501586914,
+      "learning_rate": 1.1959460894585778e-06,
+      "loss": 0.4831,
+      "step": 48027
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.748095989227295,
+      "learning_rate": 1.1957749875058094e-06,
+      "loss": 0.8233,
+      "step": 48028
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.8276572227478027,
+      "learning_rate": 1.1956038961311639e-06,
+      "loss": 0.7996,
+      "step": 48029
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.0524511337280273,
+      "learning_rate": 1.1954328153351207e-06,
+      "loss": 0.5436,
+      "step": 48030
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.056971549987793,
+      "learning_rate": 1.1952617451181508e-06,
+      "loss": 0.4098,
+      "step": 48031
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.500899314880371,
+      "learning_rate": 1.1950906854807354e-06,
+      "loss": 0.6958,
+      "step": 48032
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.0969042778015137,
+      "learning_rate": 1.1949196364233463e-06,
+      "loss": 0.5825,
+      "step": 48033
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.5999059677124023,
+      "learning_rate": 1.1947485979464612e-06,
+      "loss": 0.6457,
+      "step": 48034
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.397637367248535,
+      "learning_rate": 1.1945775700505536e-06,
+      "loss": 0.5926,
+      "step": 48035
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 1.9435957670211792,
+      "learning_rate": 1.1944065527361015e-06,
+      "loss": 0.4636,
+      "step": 48036
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.819216012954712,
+      "learning_rate": 1.1942355460035782e-06,
+      "loss": 0.5333,
+      "step": 48037
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 4.536293029785156,
+      "learning_rate": 1.1940645498534608e-06,
+      "loss": 0.8542,
+      "step": 48038
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.0955140590667725,
+      "learning_rate": 1.1938935642862237e-06,
+      "loss": 0.3196,
+      "step": 48039
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.3808233737945557,
+      "learning_rate": 1.1937225893023435e-06,
+      "loss": 0.572,
+      "step": 48040
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.448014259338379,
+      "learning_rate": 1.1935516249022938e-06,
+      "loss": 0.4635,
+      "step": 48041
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.5838510990142822,
+      "learning_rate": 1.1933806710865525e-06,
+      "loss": 0.7383,
+      "step": 48042
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.2500150203704834,
+      "learning_rate": 1.193209727855592e-06,
+      "loss": 0.4696,
+      "step": 48043
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.854274272918701,
+      "learning_rate": 1.193038795209891e-06,
+      "loss": 0.3766,
+      "step": 48044
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.586609125137329,
+      "learning_rate": 1.192867873149921e-06,
+      "loss": 0.6716,
+      "step": 48045
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.0011770725250244,
+      "learning_rate": 1.1926969616761607e-06,
+      "loss": 0.5952,
+      "step": 48046
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.4568257331848145,
+      "learning_rate": 1.192526060789083e-06,
+      "loss": 0.564,
+      "step": 48047
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.4334492683410645,
+      "learning_rate": 1.1923551704891651e-06,
+      "loss": 0.451,
+      "step": 48048
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.8693864345550537,
+      "learning_rate": 1.19218429077688e-06,
+      "loss": 0.6254,
+      "step": 48049
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.792712926864624,
+      "learning_rate": 1.1920134216527052e-06,
+      "loss": 0.8776,
+      "step": 48050
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.155998706817627,
+      "learning_rate": 1.1918425631171115e-06,
+      "loss": 0.6228,
+      "step": 48051
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.8133597373962402,
+      "learning_rate": 1.1916717151705797e-06,
+      "loss": 0.55,
+      "step": 48052
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 4.312992572784424,
+      "learning_rate": 1.1915008778135811e-06,
+      "loss": 0.6625,
+      "step": 48053
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.5916409492492676,
+      "learning_rate": 1.1913300510465936e-06,
+      "loss": 0.4368,
+      "step": 48054
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.496168613433838,
+      "learning_rate": 1.191159234870089e-06,
+      "loss": 0.4444,
+      "step": 48055
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.6857433319091797,
+      "learning_rate": 1.1909884292845448e-06,
+      "loss": 0.3301,
+      "step": 48056
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.3526711463928223,
+      "learning_rate": 1.1908176342904337e-06,
+      "loss": 0.4269,
+      "step": 48057
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.3256208896636963,
+      "learning_rate": 1.190646849888233e-06,
+      "loss": 0.4029,
+      "step": 48058
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.8060643672943115,
+      "learning_rate": 1.190476076078415e-06,
+      "loss": 0.6464,
+      "step": 48059
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.5561153888702393,
+      "learning_rate": 1.1903053128614578e-06,
+      "loss": 0.5319,
+      "step": 48060
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.1221086978912354,
+      "learning_rate": 1.1901345602378323e-06,
+      "loss": 0.6014,
+      "step": 48061
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.307483434677124,
+      "learning_rate": 1.1899638182080176e-06,
+      "loss": 0.5824,
+      "step": 48062
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.652101993560791,
+      "learning_rate": 1.1897930867724838e-06,
+      "loss": 0.7101,
+      "step": 48063
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.6401712894439697,
+      "learning_rate": 1.1896223659317101e-06,
+      "loss": 0.6094,
+      "step": 48064
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.146883249282837,
+      "learning_rate": 1.189451655686168e-06,
+      "loss": 0.6321,
+      "step": 48065
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.1442439556121826,
+      "learning_rate": 1.1892809560363345e-06,
+      "loss": 0.7151,
+      "step": 48066
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.9981353282928467,
+      "learning_rate": 1.189110266982682e-06,
+      "loss": 0.6754,
+      "step": 48067
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.793874740600586,
+      "learning_rate": 1.188939588525686e-06,
+      "loss": 0.6589,
+      "step": 48068
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 4.03119421005249,
+      "learning_rate": 1.1887689206658226e-06,
+      "loss": 0.4814,
+      "step": 48069
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.064800262451172,
+      "learning_rate": 1.1885982634035643e-06,
+      "loss": 0.4521,
+      "step": 48070
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.3985788822174072,
+      "learning_rate": 1.188427616739387e-06,
+      "loss": 0.6734,
+      "step": 48071
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.4378418922424316,
+      "learning_rate": 1.1882569806737638e-06,
+      "loss": 0.4496,
+      "step": 48072
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.384749412536621,
+      "learning_rate": 1.1880863552071697e-06,
+      "loss": 0.5319,
+      "step": 48073
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.854123592376709,
+      "learning_rate": 1.1879157403400814e-06,
+      "loss": 0.6163,
+      "step": 48074
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.837568759918213,
+      "learning_rate": 1.1877451360729697e-06,
+      "loss": 0.6703,
+      "step": 48075
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.6951851844787598,
+      "learning_rate": 1.1875745424063122e-06,
+      "loss": 0.7675,
+      "step": 48076
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.196831226348877,
+      "learning_rate": 1.18740395934058e-06,
+      "loss": 0.6077,
+      "step": 48077
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 4.047252178192139,
+      "learning_rate": 1.187233386876251e-06,
+      "loss": 0.6358,
+      "step": 48078
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.2999966144561768,
+      "learning_rate": 1.1870628250137956e-06,
+      "loss": 0.6051,
+      "step": 48079
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.8452274799346924,
+      "learning_rate": 1.1868922737536908e-06,
+      "loss": 0.6768,
+      "step": 48080
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.4062623977661133,
+      "learning_rate": 1.1867217330964116e-06,
+      "loss": 0.5443,
+      "step": 48081
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 1.5409601926803589,
+      "learning_rate": 1.1865512030424286e-06,
+      "loss": 0.3883,
+      "step": 48082
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.637016773223877,
+      "learning_rate": 1.1863806835922203e-06,
+      "loss": 0.579,
+      "step": 48083
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.47859787940979,
+      "learning_rate": 1.1862101747462568e-06,
+      "loss": 0.5937,
+      "step": 48084
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.475356101989746,
+      "learning_rate": 1.1860396765050157e-06,
+      "loss": 0.4275,
+      "step": 48085
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.9788055419921875,
+      "learning_rate": 1.1858691888689683e-06,
+      "loss": 0.6393,
+      "step": 48086
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 12.41601276397705,
+      "learning_rate": 1.185698711838591e-06,
+      "loss": 0.4589,
+      "step": 48087
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.3449361324310303,
+      "learning_rate": 1.1855282454143558e-06,
+      "loss": 0.5967,
+      "step": 48088
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.903503656387329,
+      "learning_rate": 1.1853577895967383e-06,
+      "loss": 0.6103,
+      "step": 48089
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.073273181915283,
+      "learning_rate": 1.1851873443862112e-06,
+      "loss": 0.8017,
+      "step": 48090
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.181269884109497,
+      "learning_rate": 1.1850169097832498e-06,
+      "loss": 0.7229,
+      "step": 48091
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.432692766189575,
+      "learning_rate": 1.1848464857883247e-06,
+      "loss": 0.5308,
+      "step": 48092
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.246926784515381,
+      "learning_rate": 1.184676072401915e-06,
+      "loss": 0.5282,
+      "step": 48093
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.379622459411621,
+      "learning_rate": 1.1845056696244905e-06,
+      "loss": 0.3641,
+      "step": 48094
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.575822353363037,
+      "learning_rate": 1.184335277456528e-06,
+      "loss": 0.6924,
+      "step": 48095
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.5892133712768555,
+      "learning_rate": 1.184164895898498e-06,
+      "loss": 0.6036,
+      "step": 48096
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.4796106815338135,
+      "learning_rate": 1.1839945249508778e-06,
+      "loss": 0.5448,
+      "step": 48097
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.7288854122161865,
+      "learning_rate": 1.1838241646141374e-06,
+      "loss": 0.4751,
+      "step": 48098
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.9022130966186523,
+      "learning_rate": 1.1836538148887544e-06,
+      "loss": 0.5284,
+      "step": 48099
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.2457454204559326,
+      "learning_rate": 1.183483475775199e-06,
+      "loss": 0.523,
+      "step": 48100
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.8539352416992188,
+      "learning_rate": 1.1833131472739473e-06,
+      "loss": 0.5848,
+      "step": 48101
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.020115852355957,
+      "learning_rate": 1.1831428293854707e-06,
+      "loss": 0.384,
+      "step": 48102
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.1697707176208496,
+      "learning_rate": 1.1829725221102456e-06,
+      "loss": 0.582,
+      "step": 48103
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.75537109375,
+      "learning_rate": 1.1828022254487427e-06,
+      "loss": 0.6367,
+      "step": 48104
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.5813663005828857,
+      "learning_rate": 1.1826319394014386e-06,
+      "loss": 0.7901,
+      "step": 48105
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.4586262702941895,
+      "learning_rate": 1.182461663968803e-06,
+      "loss": 0.652,
+      "step": 48106
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.1491997241973877,
+      "learning_rate": 1.182291399151313e-06,
+      "loss": 0.4462,
+      "step": 48107
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.806309461593628,
+      "learning_rate": 1.1821211449494391e-06,
+      "loss": 0.4699,
+      "step": 48108
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.201099157333374,
+      "learning_rate": 1.1819509013636577e-06,
+      "loss": 0.7007,
+      "step": 48109
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.1478378772735596,
+      "learning_rate": 1.181780668394439e-06,
+      "loss": 0.4728,
+      "step": 48110
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 4.337803840637207,
+      "learning_rate": 1.1816104460422596e-06,
+      "loss": 0.737,
+      "step": 48111
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.116039276123047,
+      "learning_rate": 1.1814402343075876e-06,
+      "loss": 0.5148,
+      "step": 48112
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 4.541484355926514,
+      "learning_rate": 1.1812700331909033e-06,
+      "loss": 0.8528,
+      "step": 48113
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 1.9625189304351807,
+      "learning_rate": 1.181099842692675e-06,
+      "loss": 0.5016,
+      "step": 48114
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 4.1426615715026855,
+      "learning_rate": 1.1809296628133787e-06,
+      "loss": 0.6895,
+      "step": 48115
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.93664288520813,
+      "learning_rate": 1.180759493553485e-06,
+      "loss": 0.5399,
+      "step": 48116
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.5074574947357178,
+      "learning_rate": 1.18058933491347e-06,
+      "loss": 0.6237,
+      "step": 48117
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.542391061782837,
+      "learning_rate": 1.1804191868938037e-06,
+      "loss": 0.5944,
+      "step": 48118
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.6874277591705322,
+      "learning_rate": 1.1802490494949619e-06,
+      "loss": 0.5978,
+      "step": 48119
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.8964734077453613,
+      "learning_rate": 1.1800789227174159e-06,
+      "loss": 0.6705,
+      "step": 48120
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.2289886474609375,
+      "learning_rate": 1.1799088065616403e-06,
+      "loss": 0.6417,
+      "step": 48121
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.2807915210723877,
+      "learning_rate": 1.1797387010281065e-06,
+      "loss": 0.7714,
+      "step": 48122
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.7107303142547607,
+      "learning_rate": 1.179568606117289e-06,
+      "loss": 0.5115,
+      "step": 48123
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.650667428970337,
+      "learning_rate": 1.179398521829659e-06,
+      "loss": 0.8088,
+      "step": 48124
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.338107109069824,
+      "learning_rate": 1.1792284481656912e-06,
+      "loss": 0.6614,
+      "step": 48125
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.940690040588379,
+      "learning_rate": 1.1790583851258568e-06,
+      "loss": 0.5063,
+      "step": 48126
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 4.811199188232422,
+      "learning_rate": 1.1788883327106315e-06,
+      "loss": 0.5147,
+      "step": 48127
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.622149705886841,
+      "learning_rate": 1.1787182909204848e-06,
+      "loss": 0.4885,
+      "step": 48128
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.53680419921875,
+      "learning_rate": 1.1785482597558923e-06,
+      "loss": 0.656,
+      "step": 48129
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.7382326126098633,
+      "learning_rate": 1.1783782392173238e-06,
+      "loss": 0.7085,
+      "step": 48130
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.311321973800659,
+      "learning_rate": 1.1782082293052543e-06,
+      "loss": 0.6313,
+      "step": 48131
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.443741798400879,
+      "learning_rate": 1.178038230020157e-06,
+      "loss": 0.7843,
+      "step": 48132
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 5.48897647857666,
+      "learning_rate": 1.1778682413625026e-06,
+      "loss": 0.6085,
+      "step": 48133
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 6.511795997619629,
+      "learning_rate": 1.1776982633327644e-06,
+      "loss": 0.7144,
+      "step": 48134
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.1828858852386475,
+      "learning_rate": 1.1775282959314171e-06,
+      "loss": 0.5849,
+      "step": 48135
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.2591073513031006,
+      "learning_rate": 1.1773583391589304e-06,
+      "loss": 0.7786,
+      "step": 48136
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.703263759613037,
+      "learning_rate": 1.1771883930157797e-06,
+      "loss": 0.5932,
+      "step": 48137
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.999833822250366,
+      "learning_rate": 1.1770184575024346e-06,
+      "loss": 0.5383,
+      "step": 48138
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.469355821609497,
+      "learning_rate": 1.1768485326193702e-06,
+      "loss": 0.5502,
+      "step": 48139
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 4.429220199584961,
+      "learning_rate": 1.1766786183670565e-06,
+      "loss": 0.5861,
+      "step": 48140
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 4.173505783081055,
+      "learning_rate": 1.1765087147459674e-06,
+      "loss": 0.4459,
+      "step": 48141
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.1055729389190674,
+      "learning_rate": 1.1763388217565764e-06,
+      "loss": 0.7403,
+      "step": 48142
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.1474978923797607,
+      "learning_rate": 1.1761689393993535e-06,
+      "loss": 0.4963,
+      "step": 48143
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.5586750507354736,
+      "learning_rate": 1.175999067674774e-06,
+      "loss": 0.5251,
+      "step": 48144
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 1.9265689849853516,
+      "learning_rate": 1.1758292065833072e-06,
+      "loss": 0.5963,
+      "step": 48145
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.8373570442199707,
+      "learning_rate": 1.1756593561254276e-06,
+      "loss": 0.4044,
+      "step": 48146
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.148803949356079,
+      "learning_rate": 1.1754895163016057e-06,
+      "loss": 0.7941,
+      "step": 48147
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 4.4070210456848145,
+      "learning_rate": 1.1753196871123163e-06,
+      "loss": 0.6539,
+      "step": 48148
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.427628517150879,
+      "learning_rate": 1.1751498685580287e-06,
+      "loss": 0.5023,
+      "step": 48149
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.4555020332336426,
+      "learning_rate": 1.1749800606392175e-06,
+      "loss": 0.6573,
+      "step": 48150
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 6.413444519042969,
+      "learning_rate": 1.1748102633563525e-06,
+      "loss": 0.7405,
+      "step": 48151
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.761988401412964,
+      "learning_rate": 1.174640476709909e-06,
+      "loss": 0.4333,
+      "step": 48152
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 1.950305700302124,
+      "learning_rate": 1.1744707007003541e-06,
+      "loss": 0.5275,
+      "step": 48153
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.0765557289123535,
+      "learning_rate": 1.1743009353281664e-06,
+      "loss": 0.6132,
+      "step": 48154
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.932182788848877,
+      "learning_rate": 1.1741311805938128e-06,
+      "loss": 0.7871,
+      "step": 48155
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.6479179859161377,
+      "learning_rate": 1.1739614364977693e-06,
+      "loss": 0.5595,
+      "step": 48156
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.1321916580200195,
+      "learning_rate": 1.1737917030405032e-06,
+      "loss": 0.5197,
+      "step": 48157
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.172346353530884,
+      "learning_rate": 1.1736219802224914e-06,
+      "loss": 0.7672,
+      "step": 48158
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.9462568759918213,
+      "learning_rate": 1.1734522680442017e-06,
+      "loss": 0.3786,
+      "step": 48159
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.068218946456909,
+      "learning_rate": 1.1732825665061093e-06,
+      "loss": 0.7032,
+      "step": 48160
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.394587755203247,
+      "learning_rate": 1.173112875608683e-06,
+      "loss": 0.6718,
+      "step": 48161
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.349005937576294,
+      "learning_rate": 1.1729431953523978e-06,
+      "loss": 0.6891,
+      "step": 48162
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.734757661819458,
+      "learning_rate": 1.1727735257377227e-06,
+      "loss": 0.5748,
+      "step": 48163
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.159226417541504,
+      "learning_rate": 1.172603866765132e-06,
+      "loss": 0.6097,
+      "step": 48164
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.363363027572632,
+      "learning_rate": 1.1724342184350946e-06,
+      "loss": 0.6128,
+      "step": 48165
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.2480859756469727,
+      "learning_rate": 1.1722645807480847e-06,
+      "loss": 0.5538,
+      "step": 48166
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.4184796810150146,
+      "learning_rate": 1.1720949537045724e-06,
+      "loss": 0.4189,
+      "step": 48167
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.0231165885925293,
+      "learning_rate": 1.1719253373050315e-06,
+      "loss": 0.7935,
+      "step": 48168
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.241075277328491,
+      "learning_rate": 1.1717557315499306e-06,
+      "loss": 0.6842,
+      "step": 48169
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.377668619155884,
+      "learning_rate": 1.171586136439744e-06,
+      "loss": 0.534,
+      "step": 48170
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.3621439933776855,
+      "learning_rate": 1.1714165519749416e-06,
+      "loss": 0.5898,
+      "step": 48171
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.9488577842712402,
+      "learning_rate": 1.1712469781559964e-06,
+      "loss": 0.5319,
+      "step": 48172
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 4.718915939331055,
+      "learning_rate": 1.171077414983376e-06,
+      "loss": 0.485,
+      "step": 48173
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.8364572525024414,
+      "learning_rate": 1.1709078624575581e-06,
+      "loss": 0.6844,
+      "step": 48174
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 1.7628175020217896,
+      "learning_rate": 1.1707383205790097e-06,
+      "loss": 0.5158,
+      "step": 48175
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.1324193477630615,
+      "learning_rate": 1.170568789348205e-06,
+      "loss": 0.454,
+      "step": 48176
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 1.992601990699768,
+      "learning_rate": 1.1703992687656125e-06,
+      "loss": 0.4843,
+      "step": 48177
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.36007022857666,
+      "learning_rate": 1.1702297588317063e-06,
+      "loss": 0.5438,
+      "step": 48178
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.9640707969665527,
+      "learning_rate": 1.1700602595469552e-06,
+      "loss": 0.5452,
+      "step": 48179
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.5351903438568115,
+      "learning_rate": 1.1698907709118329e-06,
+      "loss": 0.4849,
+      "step": 48180
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.33071231842041,
+      "learning_rate": 1.1697212929268088e-06,
+      "loss": 0.5545,
+      "step": 48181
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 5.644040584564209,
+      "learning_rate": 1.169551825592356e-06,
+      "loss": 0.9259,
+      "step": 48182
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 1.874239206314087,
+      "learning_rate": 1.1693823689089428e-06,
+      "loss": 0.3176,
+      "step": 48183
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.1118576526641846,
+      "learning_rate": 1.1692129228770444e-06,
+      "loss": 0.76,
+      "step": 48184
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.2522013187408447,
+      "learning_rate": 1.169043487497128e-06,
+      "loss": 0.6118,
+      "step": 48185
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.2992196083068848,
+      "learning_rate": 1.1688740627696682e-06,
+      "loss": 0.6505,
+      "step": 48186
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.1225242614746094,
+      "learning_rate": 1.1687046486951325e-06,
+      "loss": 0.6225,
+      "step": 48187
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.6844310760498047,
+      "learning_rate": 1.1685352452739956e-06,
+      "loss": 0.5137,
+      "step": 48188
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.1052262783050537,
+      "learning_rate": 1.168365852506726e-06,
+      "loss": 0.5754,
+      "step": 48189
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.360076904296875,
+      "learning_rate": 1.1681964703937964e-06,
+      "loss": 0.666,
+      "step": 48190
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.7207188606262207,
+      "learning_rate": 1.1680270989356757e-06,
+      "loss": 0.5916,
+      "step": 48191
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.313387393951416,
+      "learning_rate": 1.1678577381328364e-06,
+      "loss": 0.4548,
+      "step": 48192
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.686671018600464,
+      "learning_rate": 1.1676883879857504e-06,
+      "loss": 0.7732,
+      "step": 48193
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 6.15772819519043,
+      "learning_rate": 1.167519048494886e-06,
+      "loss": 0.895,
+      "step": 48194
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.067227602005005,
+      "learning_rate": 1.1673497196607152e-06,
+      "loss": 0.8813,
+      "step": 48195
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.1769423484802246,
+      "learning_rate": 1.1671804014837106e-06,
+      "loss": 0.632,
+      "step": 48196
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.576162576675415,
+      "learning_rate": 1.1670110939643404e-06,
+      "loss": 0.6328,
+      "step": 48197
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.639608860015869,
+      "learning_rate": 1.1668417971030777e-06,
+      "loss": 0.7575,
+      "step": 48198
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.359300374984741,
+      "learning_rate": 1.1666725109003907e-06,
+      "loss": 0.8487,
+      "step": 48199
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.176814317703247,
+      "learning_rate": 1.1665032353567529e-06,
+      "loss": 0.2694,
+      "step": 48200
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.3981568813323975,
+      "learning_rate": 1.1663339704726318e-06,
+      "loss": 0.5287,
+      "step": 48201
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.4880764484405518,
+      "learning_rate": 1.1661647162485017e-06,
+      "loss": 0.6207,
+      "step": 48202
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.8805429935455322,
+      "learning_rate": 1.1659954726848293e-06,
+      "loss": 0.7166,
+      "step": 48203
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.4541778564453125,
+      "learning_rate": 1.1658262397820879e-06,
+      "loss": 0.5717,
+      "step": 48204
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.4528913497924805,
+      "learning_rate": 1.1656570175407483e-06,
+      "loss": 0.7968,
+      "step": 48205
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.342224359512329,
+      "learning_rate": 1.165487805961279e-06,
+      "loss": 0.5447,
+      "step": 48206
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 4.671661376953125,
+      "learning_rate": 1.1653186050441534e-06,
+      "loss": 0.7929,
+      "step": 48207
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.8151535987854004,
+      "learning_rate": 1.1651494147898385e-06,
+      "loss": 0.6226,
+      "step": 48208
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 4.506983757019043,
+      "learning_rate": 1.1649802351988076e-06,
+      "loss": 0.6284,
+      "step": 48209
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.9411628246307373,
+      "learning_rate": 1.1648110662715289e-06,
+      "loss": 0.5755,
+      "step": 48210
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.408552885055542,
+      "learning_rate": 1.1646419080084752e-06,
+      "loss": 0.4978,
+      "step": 48211
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 4.066072940826416,
+      "learning_rate": 1.1644727604101146e-06,
+      "loss": 0.4559,
+      "step": 48212
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.019136667251587,
+      "learning_rate": 1.1643036234769195e-06,
+      "loss": 0.4831,
+      "step": 48213
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.2156667709350586,
+      "learning_rate": 1.1641344972093583e-06,
+      "loss": 0.5501,
+      "step": 48214
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 4.593224048614502,
+      "learning_rate": 1.163965381607901e-06,
+      "loss": 0.458,
+      "step": 48215
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.6019484996795654,
+      "learning_rate": 1.1637962766730198e-06,
+      "loss": 0.4897,
+      "step": 48216
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.161318063735962,
+      "learning_rate": 1.1636271824051853e-06,
+      "loss": 0.4232,
+      "step": 48217
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.2376747131347656,
+      "learning_rate": 1.1634580988048649e-06,
+      "loss": 0.4415,
+      "step": 48218
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.4830756187438965,
+      "learning_rate": 1.163289025872532e-06,
+      "loss": 0.5762,
+      "step": 48219
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 4.38596248626709,
+      "learning_rate": 1.1631199636086538e-06,
+      "loss": 0.5708,
+      "step": 48220
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 1.8488965034484863,
+      "learning_rate": 1.1629509120137028e-06,
+      "loss": 0.3525,
+      "step": 48221
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.604675769805908,
+      "learning_rate": 1.1627818710881462e-06,
+      "loss": 0.5234,
+      "step": 48222
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.5933644771575928,
+      "learning_rate": 1.1626128408324577e-06,
+      "loss": 0.6076,
+      "step": 48223
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.704491138458252,
+      "learning_rate": 1.162443821247104e-06,
+      "loss": 0.4959,
+      "step": 48224
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.5711348056793213,
+      "learning_rate": 1.1622748123325572e-06,
+      "loss": 0.3116,
+      "step": 48225
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 4.306114673614502,
+      "learning_rate": 1.1621058140892855e-06,
+      "loss": 0.5153,
+      "step": 48226
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.4822049140930176,
+      "learning_rate": 1.161936826517761e-06,
+      "loss": 0.5436,
+      "step": 48227
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.498488426208496,
+      "learning_rate": 1.1617678496184516e-06,
+      "loss": 0.5959,
+      "step": 48228
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.4892160892486572,
+      "learning_rate": 1.161598883391829e-06,
+      "loss": 0.7109,
+      "step": 48229
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 4.023843765258789,
+      "learning_rate": 1.1614299278383606e-06,
+      "loss": 0.5421,
+      "step": 48230
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.475623846054077,
+      "learning_rate": 1.1612609829585191e-06,
+      "loss": 0.4886,
+      "step": 48231
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.1557693481445312,
+      "learning_rate": 1.161092048752771e-06,
+      "loss": 0.6866,
+      "step": 48232
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 4.101363182067871,
+      "learning_rate": 1.1609231252215896e-06,
+      "loss": 0.6604,
+      "step": 48233
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.201840877532959,
+      "learning_rate": 1.1607542123654402e-06,
+      "loss": 0.4988,
+      "step": 48234
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.6536333560943604,
+      "learning_rate": 1.1605853101847974e-06,
+      "loss": 0.6467,
+      "step": 48235
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.345064640045166,
+      "learning_rate": 1.1604164186801276e-06,
+      "loss": 0.6357,
+      "step": 48236
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 8.475223541259766,
+      "learning_rate": 1.1602475378519028e-06,
+      "loss": 0.2681,
+      "step": 48237
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.435046911239624,
+      "learning_rate": 1.1600786677005898e-06,
+      "loss": 0.5584,
+      "step": 48238
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 5.855592250823975,
+      "learning_rate": 1.1599098082266608e-06,
+      "loss": 0.5097,
+      "step": 48239
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.9927992820739746,
+      "learning_rate": 1.1597409594305825e-06,
+      "loss": 0.7951,
+      "step": 48240
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.463038444519043,
+      "learning_rate": 1.1595721213128275e-06,
+      "loss": 0.5385,
+      "step": 48241
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.2141518592834473,
+      "learning_rate": 1.1594032938738626e-06,
+      "loss": 0.6496,
+      "step": 48242
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.702927827835083,
+      "learning_rate": 1.15923447711416e-06,
+      "loss": 0.5056,
+      "step": 48243
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.071782350540161,
+      "learning_rate": 1.1590656710341858e-06,
+      "loss": 0.4909,
+      "step": 48244
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.476167917251587,
+      "learning_rate": 1.158896875634412e-06,
+      "loss": 0.6035,
+      "step": 48245
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.4461922645568848,
+      "learning_rate": 1.1587280909153064e-06,
+      "loss": 0.6573,
+      "step": 48246
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.893704414367676,
+      "learning_rate": 1.1585593168773407e-06,
+      "loss": 0.7061,
+      "step": 48247
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.82116436958313,
+      "learning_rate": 1.1583905535209805e-06,
+      "loss": 0.4267,
+      "step": 48248
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.4974524974823,
+      "learning_rate": 1.1582218008466982e-06,
+      "loss": 0.6756,
+      "step": 48249
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.6851632595062256,
+      "learning_rate": 1.158053058854961e-06,
+      "loss": 0.618,
+      "step": 48250
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.6606650352478027,
+      "learning_rate": 1.1578843275462403e-06,
+      "loss": 0.6216,
+      "step": 48251
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.6112751960754395,
+      "learning_rate": 1.1577156069210027e-06,
+      "loss": 0.6308,
+      "step": 48252
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.796945810317993,
+      "learning_rate": 1.1575468969797188e-06,
+      "loss": 0.7254,
+      "step": 48253
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 4.033026218414307,
+      "learning_rate": 1.1573781977228587e-06,
+      "loss": 0.5158,
+      "step": 48254
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.2119219303131104,
+      "learning_rate": 1.157209509150889e-06,
+      "loss": 0.5083,
+      "step": 48255
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.6901512145996094,
+      "learning_rate": 1.15704083126428e-06,
+      "loss": 0.3608,
+      "step": 48256
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.203770875930786,
+      "learning_rate": 1.1568721640635028e-06,
+      "loss": 0.3325,
+      "step": 48257
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.9042155742645264,
+      "learning_rate": 1.1567035075490224e-06,
+      "loss": 0.7771,
+      "step": 48258
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.327300548553467,
+      "learning_rate": 1.1565348617213112e-06,
+      "loss": 0.6996,
+      "step": 48259
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.6005990505218506,
+      "learning_rate": 1.1563662265808356e-06,
+      "loss": 0.7093,
+      "step": 48260
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.9593775272369385,
+      "learning_rate": 1.156197602128067e-06,
+      "loss": 0.6985,
+      "step": 48261
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.93607497215271,
+      "learning_rate": 1.1560289883634717e-06,
+      "loss": 0.6609,
+      "step": 48262
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.416146993637085,
+      "learning_rate": 1.155860385287521e-06,
+      "loss": 0.5138,
+      "step": 48263
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.844113349914551,
+      "learning_rate": 1.1556917929006811e-06,
+      "loss": 0.6849,
+      "step": 48264
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.1739161014556885,
+      "learning_rate": 1.1555232112034226e-06,
+      "loss": 0.5634,
+      "step": 48265
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.6747002601623535,
+      "learning_rate": 1.1553546401962152e-06,
+      "loss": 0.7945,
+      "step": 48266
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.272613525390625,
+      "learning_rate": 1.1551860798795245e-06,
+      "loss": 0.3778,
+      "step": 48267
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.1691527366638184,
+      "learning_rate": 1.1550175302538225e-06,
+      "loss": 0.5245,
+      "step": 48268
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.1232616901397705,
+      "learning_rate": 1.1548489913195748e-06,
+      "loss": 0.4534,
+      "step": 48269
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.624488115310669,
+      "learning_rate": 1.1546804630772535e-06,
+      "loss": 0.6492,
+      "step": 48270
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.050123691558838,
+      "learning_rate": 1.1545119455273241e-06,
+      "loss": 0.523,
+      "step": 48271
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.246485471725464,
+      "learning_rate": 1.1543434386702573e-06,
+      "loss": 0.6273,
+      "step": 48272
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.62337064743042,
+      "learning_rate": 1.1541749425065196e-06,
+      "loss": 0.7632,
+      "step": 48273
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.6122076511383057,
+      "learning_rate": 1.1540064570365822e-06,
+      "loss": 0.4807,
+      "step": 48274
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.992605686187744,
+      "learning_rate": 1.1538379822609103e-06,
+      "loss": 0.5101,
+      "step": 48275
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 1.9509387016296387,
+      "learning_rate": 1.1536695181799746e-06,
+      "loss": 0.3987,
+      "step": 48276
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 4.6465911865234375,
+      "learning_rate": 1.153501064794243e-06,
+      "loss": 0.611,
+      "step": 48277
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.4805264472961426,
+      "learning_rate": 1.1533326221041853e-06,
+      "loss": 0.6569,
+      "step": 48278
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 4.749670505523682,
+      "learning_rate": 1.153164190110267e-06,
+      "loss": 0.6033,
+      "step": 48279
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.3890602588653564,
+      "learning_rate": 1.1529957688129601e-06,
+      "loss": 0.5777,
+      "step": 48280
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.263758659362793,
+      "learning_rate": 1.1528273582127287e-06,
+      "loss": 0.6029,
+      "step": 48281
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 4.975203514099121,
+      "learning_rate": 1.152658958310045e-06,
+      "loss": 0.7501,
+      "step": 48282
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.591465711593628,
+      "learning_rate": 1.152490569105374e-06,
+      "loss": 0.7676,
+      "step": 48283
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 4.036629676818848,
+      "learning_rate": 1.152322190599187e-06,
+      "loss": 0.7685,
+      "step": 48284
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.103107213973999,
+      "learning_rate": 1.1521538227919489e-06,
+      "loss": 0.5002,
+      "step": 48285
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.6881470680236816,
+      "learning_rate": 1.1519854656841312e-06,
+      "loss": 0.4583,
+      "step": 48286
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.4035468101501465,
+      "learning_rate": 1.1518171192761989e-06,
+      "loss": 0.5569,
+      "step": 48287
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.7919187545776367,
+      "learning_rate": 1.151648783568623e-06,
+      "loss": 0.5933,
+      "step": 48288
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.205251455307007,
+      "learning_rate": 1.1514804585618689e-06,
+      "loss": 0.684,
+      "step": 48289
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.2998461723327637,
+      "learning_rate": 1.1513121442564074e-06,
+      "loss": 0.5875,
+      "step": 48290
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 4.076081275939941,
+      "learning_rate": 1.1511438406527038e-06,
+      "loss": 0.605,
+      "step": 48291
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.168905735015869,
+      "learning_rate": 1.1509755477512286e-06,
+      "loss": 0.3639,
+      "step": 48292
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 1.9819896221160889,
+      "learning_rate": 1.1508072655524472e-06,
+      "loss": 0.4458,
+      "step": 48293
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.9227421283721924,
+      "learning_rate": 1.1506389940568302e-06,
+      "loss": 0.4821,
+      "step": 48294
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.574394702911377,
+      "learning_rate": 1.1504707332648412e-06,
+      "loss": 0.7493,
+      "step": 48295
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.557567834854126,
+      "learning_rate": 1.150302483176954e-06,
+      "loss": 0.6061,
+      "step": 48296
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.047316312789917,
+      "learning_rate": 1.150134243793632e-06,
+      "loss": 0.6448,
+      "step": 48297
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.37260365486145,
+      "learning_rate": 1.1499660151153458e-06,
+      "loss": 0.7392,
+      "step": 48298
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.1397202014923096,
+      "learning_rate": 1.1497977971425607e-06,
+      "loss": 0.6824,
+      "step": 48299
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 1.825634479522705,
+      "learning_rate": 1.149629589875747e-06,
+      "loss": 0.4222,
+      "step": 48300
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.8469882011413574,
+      "learning_rate": 1.1494613933153692e-06,
+      "loss": 0.8921,
+      "step": 48301
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.9132208824157715,
+      "learning_rate": 1.1492932074618985e-06,
+      "loss": 0.54,
+      "step": 48302
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 1.6992019414901733,
+      "learning_rate": 1.1491250323157998e-06,
+      "loss": 0.3543,
+      "step": 48303
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.961632251739502,
+      "learning_rate": 1.1489568678775425e-06,
+      "loss": 0.3851,
+      "step": 48304
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.351943254470825,
+      "learning_rate": 1.1487887141475928e-06,
+      "loss": 0.5367,
+      "step": 48305
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.6453607082366943,
+      "learning_rate": 1.1486205711264204e-06,
+      "loss": 0.739,
+      "step": 48306
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.908267021179199,
+      "learning_rate": 1.1484524388144897e-06,
+      "loss": 0.5249,
+      "step": 48307
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.0295116901397705,
+      "learning_rate": 1.148284317212271e-06,
+      "loss": 0.5792,
+      "step": 48308
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.948150634765625,
+      "learning_rate": 1.1481162063202301e-06,
+      "loss": 0.6916,
+      "step": 48309
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.701314687728882,
+      "learning_rate": 1.147948106138836e-06,
+      "loss": 0.5627,
+      "step": 48310
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.1540987491607666,
+      "learning_rate": 1.1477800166685538e-06,
+      "loss": 0.6247,
+      "step": 48311
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 1.940808892250061,
+      "learning_rate": 1.1476119379098538e-06,
+      "loss": 0.4506,
+      "step": 48312
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.6371870040893555,
+      "learning_rate": 1.1474438698632007e-06,
+      "loss": 0.5852,
+      "step": 48313
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 1.8886239528656006,
+      "learning_rate": 1.147275812529064e-06,
+      "loss": 0.4814,
+      "step": 48314
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.9408998489379883,
+      "learning_rate": 1.1471077659079089e-06,
+      "loss": 0.6164,
+      "step": 48315
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.712637424468994,
+      "learning_rate": 1.1469397300002034e-06,
+      "loss": 0.5898,
+      "step": 48316
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.688673973083496,
+      "learning_rate": 1.1467717048064152e-06,
+      "loss": 0.5632,
+      "step": 48317
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.3780367374420166,
+      "learning_rate": 1.1466036903270134e-06,
+      "loss": 0.388,
+      "step": 48318
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.171339988708496,
+      "learning_rate": 1.1464356865624614e-06,
+      "loss": 0.9827,
+      "step": 48319
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.187640428543091,
+      "learning_rate": 1.1462676935132295e-06,
+      "loss": 0.666,
+      "step": 48320
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.5792529582977295,
+      "learning_rate": 1.1460997111797821e-06,
+      "loss": 0.7488,
+      "step": 48321
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.1812644004821777,
+      "learning_rate": 1.145931739562589e-06,
+      "loss": 0.7186,
+      "step": 48322
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 1.605380654335022,
+      "learning_rate": 1.145763778662115e-06,
+      "loss": 0.4035,
+      "step": 48323
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.562380313873291,
+      "learning_rate": 1.1455958284788294e-06,
+      "loss": 0.5794,
+      "step": 48324
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.9689598083496094,
+      "learning_rate": 1.1454278890131965e-06,
+      "loss": 0.6686,
+      "step": 48325
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.9448556900024414,
+      "learning_rate": 1.1452599602656845e-06,
+      "loss": 0.757,
+      "step": 48326
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.9938278198242188,
+      "learning_rate": 1.1450920422367618e-06,
+      "loss": 0.6213,
+      "step": 48327
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.8038077354431152,
+      "learning_rate": 1.144924134926893e-06,
+      "loss": 0.6836,
+      "step": 48328
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.0281004905700684,
+      "learning_rate": 1.1447562383365473e-06,
+      "loss": 0.5556,
+      "step": 48329
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 6.1489129066467285,
+      "learning_rate": 1.1445883524661888e-06,
+      "loss": 0.6399,
+      "step": 48330
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.4127650260925293,
+      "learning_rate": 1.144420477316287e-06,
+      "loss": 0.5507,
+      "step": 48331
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 4.110171794891357,
+      "learning_rate": 1.1442526128873067e-06,
+      "loss": 0.5083,
+      "step": 48332
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 4.232868194580078,
+      "learning_rate": 1.1440847591797165e-06,
+      "loss": 0.7238,
+      "step": 48333
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.884706497192383,
+      "learning_rate": 1.143916916193981e-06,
+      "loss": 0.3427,
+      "step": 48334
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.462583303451538,
+      "learning_rate": 1.1437490839305693e-06,
+      "loss": 0.4502,
+      "step": 48335
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.7971649169921875,
+      "learning_rate": 1.1435812623899451e-06,
+      "loss": 0.5604,
+      "step": 48336
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.4643425941467285,
+      "learning_rate": 1.143413451572577e-06,
+      "loss": 0.6381,
+      "step": 48337
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.5645439624786377,
+      "learning_rate": 1.1432456514789315e-06,
+      "loss": 0.7166,
+      "step": 48338
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.9596917629241943,
+      "learning_rate": 1.1430778621094763e-06,
+      "loss": 0.5924,
+      "step": 48339
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.5016021728515625,
+      "learning_rate": 1.1429100834646756e-06,
+      "loss": 0.5571,
+      "step": 48340
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.6522464752197266,
+      "learning_rate": 1.1427423155449985e-06,
+      "loss": 0.6641,
+      "step": 48341
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 4.305973052978516,
+      "learning_rate": 1.1425745583509085e-06,
+      "loss": 0.7663,
+      "step": 48342
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.4394071102142334,
+      "learning_rate": 1.1424068118828746e-06,
+      "loss": 0.5335,
+      "step": 48343
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.2361202239990234,
+      "learning_rate": 1.1422390761413615e-06,
+      "loss": 0.5509,
+      "step": 48344
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.114654541015625,
+      "learning_rate": 1.1420713511268372e-06,
+      "loss": 0.6877,
+      "step": 48345
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.485980272293091,
+      "learning_rate": 1.1419036368397663e-06,
+      "loss": 0.5886,
+      "step": 48346
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.7750163078308105,
+      "learning_rate": 1.1417359332806178e-06,
+      "loss": 0.6467,
+      "step": 48347
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.954586982727051,
+      "learning_rate": 1.1415682404498541e-06,
+      "loss": 0.6661,
+      "step": 48348
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.7742316722869873,
+      "learning_rate": 1.1414005583479454e-06,
+      "loss": 0.6,
+      "step": 48349
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.4123423099517822,
+      "learning_rate": 1.1412328869753553e-06,
+      "loss": 0.482,
+      "step": 48350
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 29.42215919494629,
+      "learning_rate": 1.1410652263325522e-06,
+      "loss": 0.5647,
+      "step": 48351
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.0336947441101074,
+      "learning_rate": 1.1408975764199992e-06,
+      "loss": 0.569,
+      "step": 48352
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.0871713161468506,
+      "learning_rate": 1.1407299372381664e-06,
+      "loss": 0.5266,
+      "step": 48353
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.0380043983459473,
+      "learning_rate": 1.1405623087875168e-06,
+      "loss": 0.6115,
+      "step": 48354
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.7288424968719482,
+      "learning_rate": 1.1403946910685182e-06,
+      "loss": 0.619,
+      "step": 48355
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.0941784381866455,
+      "learning_rate": 1.1402270840816338e-06,
+      "loss": 0.467,
+      "step": 48356
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.709392547607422,
+      "learning_rate": 1.1400594878273348e-06,
+      "loss": 0.5857,
+      "step": 48357
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.547166585922241,
+      "learning_rate": 1.139891902306083e-06,
+      "loss": 0.5478,
+      "step": 48358
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.715541124343872,
+      "learning_rate": 1.1397243275183468e-06,
+      "loss": 0.638,
+      "step": 48359
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 6.232597351074219,
+      "learning_rate": 1.1395567634645904e-06,
+      "loss": 0.4695,
+      "step": 48360
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 4.010549545288086,
+      "learning_rate": 1.1393892101452813e-06,
+      "loss": 0.8809,
+      "step": 48361
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.5057947635650635,
+      "learning_rate": 1.1392216675608836e-06,
+      "loss": 0.6084,
+      "step": 48362
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.5638861656188965,
+      "learning_rate": 1.1390541357118651e-06,
+      "loss": 0.7065,
+      "step": 48363
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.198206901550293,
+      "learning_rate": 1.1388866145986898e-06,
+      "loss": 0.6258,
+      "step": 48364
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.7620248794555664,
+      "learning_rate": 1.1387191042218255e-06,
+      "loss": 0.6327,
+      "step": 48365
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.617293119430542,
+      "learning_rate": 1.1385516045817356e-06,
+      "loss": 0.79,
+      "step": 48366
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.3840432167053223,
+      "learning_rate": 1.1383841156788883e-06,
+      "loss": 0.7014,
+      "step": 48367
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.381493091583252,
+      "learning_rate": 1.1382166375137471e-06,
+      "loss": 0.5575,
+      "step": 48368
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.19490647315979,
+      "learning_rate": 1.1380491700867802e-06,
+      "loss": 0.531,
+      "step": 48369
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.8446059226989746,
+      "learning_rate": 1.1378817133984504e-06,
+      "loss": 0.5567,
+      "step": 48370
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.5345213413238525,
+      "learning_rate": 1.1377142674492264e-06,
+      "loss": 0.5535,
+      "step": 48371
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.7783656120300293,
+      "learning_rate": 1.1375468322395705e-06,
+      "loss": 0.5964,
+      "step": 48372
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.9047749042510986,
+      "learning_rate": 1.1373794077699513e-06,
+      "loss": 0.5377,
+      "step": 48373
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.108670949935913,
+      "learning_rate": 1.1372119940408316e-06,
+      "loss": 0.4482,
+      "step": 48374
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 4.250259876251221,
+      "learning_rate": 1.1370445910526795e-06,
+      "loss": 0.7122,
+      "step": 48375
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 1.6441951990127563,
+      "learning_rate": 1.1368771988059586e-06,
+      "loss": 0.2904,
+      "step": 48376
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.3372983932495117,
+      "learning_rate": 1.1367098173011342e-06,
+      "loss": 0.674,
+      "step": 48377
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.2434799671173096,
+      "learning_rate": 1.1365424465386732e-06,
+      "loss": 0.5006,
+      "step": 48378
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 4.084616184234619,
+      "learning_rate": 1.1363750865190415e-06,
+      "loss": 0.747,
+      "step": 48379
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.1236448287963867,
+      "learning_rate": 1.136207737242702e-06,
+      "loss": 0.7275,
+      "step": 48380
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 1.7845333814620972,
+      "learning_rate": 1.1360403987101226e-06,
+      "loss": 0.5631,
+      "step": 48381
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.2725446224212646,
+      "learning_rate": 1.1358730709217664e-06,
+      "loss": 0.4739,
+      "step": 48382
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.042149305343628,
+      "learning_rate": 1.1357057538781008e-06,
+      "loss": 0.4669,
+      "step": 48383
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.516510009765625,
+      "learning_rate": 1.1355384475795882e-06,
+      "loss": 0.5436,
+      "step": 48384
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.7268764972686768,
+      "learning_rate": 1.1353711520266976e-06,
+      "loss": 0.5551,
+      "step": 48385
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.7202751636505127,
+      "learning_rate": 1.13520386721989e-06,
+      "loss": 0.7029,
+      "step": 48386
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.0875821113586426,
+      "learning_rate": 1.1350365931596346e-06,
+      "loss": 0.3994,
+      "step": 48387
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.435026168823242,
+      "learning_rate": 1.1348693298463932e-06,
+      "loss": 0.3218,
+      "step": 48388
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.2495648860931396,
+      "learning_rate": 1.1347020772806317e-06,
+      "loss": 0.5724,
+      "step": 48389
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.2931604385375977,
+      "learning_rate": 1.1345348354628177e-06,
+      "loss": 0.7399,
+      "step": 48390
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.1854147911071777,
+      "learning_rate": 1.1343676043934126e-06,
+      "loss": 0.7272,
+      "step": 48391
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.5342648029327393,
+      "learning_rate": 1.1342003840728844e-06,
+      "loss": 0.3548,
+      "step": 48392
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.569321870803833,
+      "learning_rate": 1.1340331745016958e-06,
+      "loss": 0.4756,
+      "step": 48393
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.3805623054504395,
+      "learning_rate": 1.1338659756803138e-06,
+      "loss": 0.6099,
+      "step": 48394
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.279029607772827,
+      "learning_rate": 1.133698787609201e-06,
+      "loss": 0.3646,
+      "step": 48395
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.1315245628356934,
+      "learning_rate": 1.1335316102888244e-06,
+      "loss": 0.7475,
+      "step": 48396
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.1864914894104004,
+      "learning_rate": 1.1333644437196473e-06,
+      "loss": 0.7135,
+      "step": 48397
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.3714492321014404,
+      "learning_rate": 1.1331972879021348e-06,
+      "loss": 0.8351,
+      "step": 48398
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.041451930999756,
+      "learning_rate": 1.1330301428367534e-06,
+      "loss": 0.7644,
+      "step": 48399
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.3748011589050293,
+      "learning_rate": 1.1328630085239655e-06,
+      "loss": 0.7875,
+      "step": 48400
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.311298131942749,
+      "learning_rate": 1.1326958849642372e-06,
+      "loss": 0.7076,
+      "step": 48401
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.089384078979492,
+      "learning_rate": 1.1325287721580336e-06,
+      "loss": 0.7473,
+      "step": 48402
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.0714008808135986,
+      "learning_rate": 1.1323616701058177e-06,
+      "loss": 0.5362,
+      "step": 48403
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.0622591972351074,
+      "learning_rate": 1.1321945788080563e-06,
+      "loss": 0.4923,
+      "step": 48404
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.311194896697998,
+      "learning_rate": 1.1320274982652119e-06,
+      "loss": 0.5595,
+      "step": 48405
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.546602964401245,
+      "learning_rate": 1.1318604284777513e-06,
+      "loss": 0.8208,
+      "step": 48406
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.3934319019317627,
+      "learning_rate": 1.1316933694461364e-06,
+      "loss": 0.609,
+      "step": 48407
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 1.578608512878418,
+      "learning_rate": 1.1315263211708338e-06,
+      "loss": 0.3302,
+      "step": 48408
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.6025993824005127,
+      "learning_rate": 1.131359283652307e-06,
+      "loss": 0.5417,
+      "step": 48409
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.492046356201172,
+      "learning_rate": 1.1311922568910216e-06,
+      "loss": 0.2679,
+      "step": 48410
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.376126766204834,
+      "learning_rate": 1.13102524088744e-06,
+      "loss": 0.5663,
+      "step": 48411
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.86867094039917,
+      "learning_rate": 1.1308582356420294e-06,
+      "loss": 0.5662,
+      "step": 48412
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.885382890701294,
+      "learning_rate": 1.130691241155251e-06,
+      "loss": 0.7375,
+      "step": 48413
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.342169761657715,
+      "learning_rate": 1.130524257427572e-06,
+      "loss": 0.3899,
+      "step": 48414
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.9724349975585938,
+      "learning_rate": 1.1303572844594547e-06,
+      "loss": 0.6385,
+      "step": 48415
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.5563173294067383,
+      "learning_rate": 1.1301903222513655e-06,
+      "loss": 0.8197,
+      "step": 48416
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.581209659576416,
+      "learning_rate": 1.1300233708037638e-06,
+      "loss": 0.4843,
+      "step": 48417
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 6.069243907928467,
+      "learning_rate": 1.129856430117121e-06,
+      "loss": 0.7272,
+      "step": 48418
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.21292781829834,
+      "learning_rate": 1.1296895001918961e-06,
+      "loss": 0.7341,
+      "step": 48419
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.6284053325653076,
+      "learning_rate": 1.129522581028556e-06,
+      "loss": 0.9068,
+      "step": 48420
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.8995940685272217,
+      "learning_rate": 1.1293556726275623e-06,
+      "loss": 0.7384,
+      "step": 48421
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 4.1773858070373535,
+      "learning_rate": 1.129188774989382e-06,
+      "loss": 0.4843,
+      "step": 48422
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 6.871764183044434,
+      "learning_rate": 1.1290218881144766e-06,
+      "loss": 0.3416,
+      "step": 48423
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 5.115994930267334,
+      "learning_rate": 1.128855012003312e-06,
+      "loss": 0.7757,
+      "step": 48424
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.175678253173828,
+      "learning_rate": 1.1286881466563505e-06,
+      "loss": 0.5979,
+      "step": 48425
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.502065420150757,
+      "learning_rate": 1.1285212920740584e-06,
+      "loss": 0.5646,
+      "step": 48426
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 4.909942626953125,
+      "learning_rate": 1.1283544482568964e-06,
+      "loss": 0.7722,
+      "step": 48427
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 4.191215515136719,
+      "learning_rate": 1.1281876152053324e-06,
+      "loss": 0.6748,
+      "step": 48428
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.7574758529663086,
+      "learning_rate": 1.1280207929198266e-06,
+      "loss": 0.5982,
+      "step": 48429
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.577568292617798,
+      "learning_rate": 1.127853981400846e-06,
+      "loss": 0.8714,
+      "step": 48430
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.551100969314575,
+      "learning_rate": 1.127687180648851e-06,
+      "loss": 0.7892,
+      "step": 48431
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.171283006668091,
+      "learning_rate": 1.1275203906643096e-06,
+      "loss": 0.8388,
+      "step": 48432
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.1736466884613037,
+      "learning_rate": 1.1273536114476812e-06,
+      "loss": 0.4202,
+      "step": 48433
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.0529751777648926,
+      "learning_rate": 1.1271868429994331e-06,
+      "loss": 0.6906,
+      "step": 48434
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.551006555557251,
+      "learning_rate": 1.127020085320027e-06,
+      "loss": 0.7113,
+      "step": 48435
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.4694604873657227,
+      "learning_rate": 1.1268533384099278e-06,
+      "loss": 0.7074,
+      "step": 48436
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.557981014251709,
+      "learning_rate": 1.126686602269597e-06,
+      "loss": 0.5541,
+      "step": 48437
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.612584114074707,
+      "learning_rate": 1.1265198768995005e-06,
+      "loss": 0.7584,
+      "step": 48438
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.4921600818634033,
+      "learning_rate": 1.126353162300101e-06,
+      "loss": 0.4224,
+      "step": 48439
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.1356139183044434,
+      "learning_rate": 1.1261864584718635e-06,
+      "loss": 0.3996,
+      "step": 48440
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.004694700241089,
+      "learning_rate": 1.126019765415249e-06,
+      "loss": 0.4505,
+      "step": 48441
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 4.9623823165893555,
+      "learning_rate": 1.1258530831307235e-06,
+      "loss": 0.755,
+      "step": 48442
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.819287061691284,
+      "learning_rate": 1.1256864116187483e-06,
+      "loss": 0.4341,
+      "step": 48443
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.7705488204956055,
+      "learning_rate": 1.125519750879789e-06,
+      "loss": 0.6925,
+      "step": 48444
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.490074872970581,
+      "learning_rate": 1.125353100914306e-06,
+      "loss": 0.4634,
+      "step": 48445
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.3697054386138916,
+      "learning_rate": 1.1251864617227664e-06,
+      "loss": 0.4418,
+      "step": 48446
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 4.030575752258301,
+      "learning_rate": 1.1250198333056301e-06,
+      "loss": 0.6285,
+      "step": 48447
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 4.320137977600098,
+      "learning_rate": 1.124853215663364e-06,
+      "loss": 0.4892,
+      "step": 48448
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.965216636657715,
+      "learning_rate": 1.1246866087964275e-06,
+      "loss": 0.5964,
+      "step": 48449
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.4317986965179443,
+      "learning_rate": 1.1245200127052857e-06,
+      "loss": 0.614,
+      "step": 48450
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 4.051488876342773,
+      "learning_rate": 1.124353427390404e-06,
+      "loss": 0.6031,
+      "step": 48451
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.6656508445739746,
+      "learning_rate": 1.1241868528522415e-06,
+      "loss": 0.7197,
+      "step": 48452
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.0696377754211426,
+      "learning_rate": 1.1240202890912654e-06,
+      "loss": 0.511,
+      "step": 48453
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.7251296043395996,
+      "learning_rate": 1.123853736107935e-06,
+      "loss": 0.5865,
+      "step": 48454
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 1.6020244359970093,
+      "learning_rate": 1.1236871939027173e-06,
+      "loss": 0.4884,
+      "step": 48455
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.022402048110962,
+      "learning_rate": 1.1235206624760714e-06,
+      "loss": 0.4011,
+      "step": 48456
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.253444194793701,
+      "learning_rate": 1.1233541418284644e-06,
+      "loss": 0.5916,
+      "step": 48457
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.7821996212005615,
+      "learning_rate": 1.1231876319603552e-06,
+      "loss": 0.4656,
+      "step": 48458
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.1050922870635986,
+      "learning_rate": 1.1230211328722096e-06,
+      "loss": 0.558,
+      "step": 48459
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.8859260082244873,
+      "learning_rate": 1.1228546445644906e-06,
+      "loss": 0.5824,
+      "step": 48460
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.9873733520507812,
+      "learning_rate": 1.1226881670376593e-06,
+      "loss": 0.6686,
+      "step": 48461
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.170438051223755,
+      "learning_rate": 1.1225217002921802e-06,
+      "loss": 0.5082,
+      "step": 48462
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.517087697982788,
+      "learning_rate": 1.1223552443285163e-06,
+      "loss": 0.6552,
+      "step": 48463
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.0032198429107666,
+      "learning_rate": 1.1221887991471291e-06,
+      "loss": 0.3834,
+      "step": 48464
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.832261323928833,
+      "learning_rate": 1.1220223647484835e-06,
+      "loss": 0.7202,
+      "step": 48465
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.8016397953033447,
+      "learning_rate": 1.1218559411330394e-06,
+      "loss": 0.453,
+      "step": 48466
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.747450351715088,
+      "learning_rate": 1.1216895283012624e-06,
+      "loss": 0.4865,
+      "step": 48467
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.889227867126465,
+      "learning_rate": 1.1215231262536131e-06,
+      "loss": 0.5845,
+      "step": 48468
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.6234617233276367,
+      "learning_rate": 1.1213567349905558e-06,
+      "loss": 0.8309,
+      "step": 48469
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.8362042903900146,
+      "learning_rate": 1.1211903545125514e-06,
+      "loss": 0.4949,
+      "step": 48470
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.821803092956543,
+      "learning_rate": 1.1210239848200648e-06,
+      "loss": 0.6288,
+      "step": 48471
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 8.144173622131348,
+      "learning_rate": 1.1208576259135566e-06,
+      "loss": 0.4829,
+      "step": 48472
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.193225622177124,
+      "learning_rate": 1.1206912777934909e-06,
+      "loss": 0.5223,
+      "step": 48473
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.3415591716766357,
+      "learning_rate": 1.1205249404603285e-06,
+      "loss": 0.4639,
+      "step": 48474
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.1985208988189697,
+      "learning_rate": 1.1203586139145345e-06,
+      "loss": 0.5568,
+      "step": 48475
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.169344186782837,
+      "learning_rate": 1.1201922981565682e-06,
+      "loss": 0.6532,
+      "step": 48476
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.0176753997802734,
+      "learning_rate": 1.120025993186895e-06,
+      "loss": 0.5436,
+      "step": 48477
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.098851203918457,
+      "learning_rate": 1.1198596990059734e-06,
+      "loss": 0.5599,
+      "step": 48478
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.5379576683044434,
+      "learning_rate": 1.1196934156142715e-06,
+      "loss": 0.5981,
+      "step": 48479
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.3949832916259766,
+      "learning_rate": 1.1195271430122468e-06,
+      "loss": 0.5246,
+      "step": 48480
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.5935683250427246,
+      "learning_rate": 1.1193608812003647e-06,
+      "loss": 0.5313,
+      "step": 48481
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.69926381111145,
+      "learning_rate": 1.1191946301790852e-06,
+      "loss": 0.6346,
+      "step": 48482
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.8031229972839355,
+      "learning_rate": 1.119028389948873e-06,
+      "loss": 0.6933,
+      "step": 48483
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.1728177070617676,
+      "learning_rate": 1.1188621605101874e-06,
+      "loss": 0.5607,
+      "step": 48484
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.379026412963867,
+      "learning_rate": 1.1186959418634942e-06,
+      "loss": 0.6456,
+      "step": 48485
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 4.1815714836120605,
+      "learning_rate": 1.1185297340092515e-06,
+      "loss": 0.7222,
+      "step": 48486
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.2728865146636963,
+      "learning_rate": 1.1183635369479252e-06,
+      "loss": 0.732,
+      "step": 48487
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.214475631713867,
+      "learning_rate": 1.1181973506799748e-06,
+      "loss": 0.8465,
+      "step": 48488
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.1853127479553223,
+      "learning_rate": 1.1180311752058642e-06,
+      "loss": 0.4336,
+      "step": 48489
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.1634953022003174,
+      "learning_rate": 1.1178650105260535e-06,
+      "loss": 0.644,
+      "step": 48490
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.203313112258911,
+      "learning_rate": 1.117698856641008e-06,
+      "loss": 0.5198,
+      "step": 48491
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.785046339035034,
+      "learning_rate": 1.1175327135511854e-06,
+      "loss": 0.7931,
+      "step": 48492
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.724606990814209,
+      "learning_rate": 1.1173665812570517e-06,
+      "loss": 0.2695,
+      "step": 48493
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 1.8103622198104858,
+      "learning_rate": 1.1172004597590651e-06,
+      "loss": 0.5617,
+      "step": 48494
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.303971767425537,
+      "learning_rate": 1.1170343490576919e-06,
+      "loss": 0.5363,
+      "step": 48495
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.2512567043304443,
+      "learning_rate": 1.1168682491533894e-06,
+      "loss": 0.5294,
+      "step": 48496
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.0467305183410645,
+      "learning_rate": 1.1167021600466233e-06,
+      "loss": 0.5645,
+      "step": 48497
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.027212381362915,
+      "learning_rate": 1.1165360817378522e-06,
+      "loss": 0.642,
+      "step": 48498
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.6142616271972656,
+      "learning_rate": 1.1163700142275396e-06,
+      "loss": 0.5823,
+      "step": 48499
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 5.006716251373291,
+      "learning_rate": 1.1162039575161476e-06,
+      "loss": 0.5612,
+      "step": 48500
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 1.9888031482696533,
+      "learning_rate": 1.1160379116041382e-06,
+      "loss": 0.5945,
+      "step": 48501
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.2537851333618164,
+      "learning_rate": 1.1158718764919712e-06,
+      "loss": 0.5408,
+      "step": 48502
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.5383963584899902,
+      "learning_rate": 1.1157058521801112e-06,
+      "loss": 0.5485,
+      "step": 48503
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.139080762863159,
+      "learning_rate": 1.1155398386690163e-06,
+      "loss": 0.3579,
+      "step": 48504
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.8732821941375732,
+      "learning_rate": 1.1153738359591515e-06,
+      "loss": 0.6086,
+      "step": 48505
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.4717657566070557,
+      "learning_rate": 1.1152078440509755e-06,
+      "loss": 0.6973,
+      "step": 48506
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.8747429847717285,
+      "learning_rate": 1.1150418629449529e-06,
+      "loss": 0.7076,
+      "step": 48507
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.3995044231414795,
+      "learning_rate": 1.1148758926415416e-06,
+      "loss": 0.5147,
+      "step": 48508
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.574920415878296,
+      "learning_rate": 1.114709933141207e-06,
+      "loss": 0.7231,
+      "step": 48509
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.5083909034729004,
+      "learning_rate": 1.1145439844444066e-06,
+      "loss": 0.5111,
+      "step": 48510
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 11.470301628112793,
+      "learning_rate": 1.1143780465516057e-06,
+      "loss": 0.6338,
+      "step": 48511
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.8303110599517822,
+      "learning_rate": 1.114212119463262e-06,
+      "loss": 0.6973,
+      "step": 48512
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.0473387241363525,
+      "learning_rate": 1.1140462031798393e-06,
+      "loss": 0.5377,
+      "step": 48513
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.6487948894500732,
+      "learning_rate": 1.1138802977017999e-06,
+      "loss": 0.6762,
+      "step": 48514
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.5032613277435303,
+      "learning_rate": 1.1137144030296015e-06,
+      "loss": 0.4728,
+      "step": 48515
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.3837316036224365,
+      "learning_rate": 1.1135485191637096e-06,
+      "loss": 0.6257,
+      "step": 48516
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.289853572845459,
+      "learning_rate": 1.1133826461045816e-06,
+      "loss": 0.8088,
+      "step": 48517
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.024374008178711,
+      "learning_rate": 1.113216783852682e-06,
+      "loss": 0.3935,
+      "step": 48518
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 1.746585488319397,
+      "learning_rate": 1.1130509324084688e-06,
+      "loss": 0.2865,
+      "step": 48519
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.267871618270874,
+      "learning_rate": 1.1128850917724055e-06,
+      "loss": 0.3622,
+      "step": 48520
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.168614387512207,
+      "learning_rate": 1.1127192619449534e-06,
+      "loss": 0.6133,
+      "step": 48521
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.7607436180114746,
+      "learning_rate": 1.1125534429265717e-06,
+      "loss": 0.5063,
+      "step": 48522
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.1054494380950928,
+      "learning_rate": 1.1123876347177225e-06,
+      "loss": 0.4158,
+      "step": 48523
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.4569814205169678,
+      "learning_rate": 1.1122218373188688e-06,
+      "loss": 0.4468,
+      "step": 48524
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.3173763751983643,
+      "learning_rate": 1.1120560507304678e-06,
+      "loss": 0.7363,
+      "step": 48525
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 5.929540634155273,
+      "learning_rate": 1.1118902749529842e-06,
+      "loss": 0.6417,
+      "step": 48526
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.159557342529297,
+      "learning_rate": 1.1117245099868757e-06,
+      "loss": 0.5602,
+      "step": 48527
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.4622890949249268,
+      "learning_rate": 1.1115587558326057e-06,
+      "loss": 0.5852,
+      "step": 48528
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.657769203186035,
+      "learning_rate": 1.1113930124906336e-06,
+      "loss": 0.7409,
+      "step": 48529
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.3682103157043457,
+      "learning_rate": 1.1112272799614216e-06,
+      "loss": 0.6354,
+      "step": 48530
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.585020065307617,
+      "learning_rate": 1.1110615582454288e-06,
+      "loss": 0.4924,
+      "step": 48531
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.455821990966797,
+      "learning_rate": 1.1108958473431181e-06,
+      "loss": 0.6827,
+      "step": 48532
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.882552146911621,
+      "learning_rate": 1.110730147254948e-06,
+      "loss": 0.7295,
+      "step": 48533
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 1.9180716276168823,
+      "learning_rate": 1.1105644579813818e-06,
+      "loss": 0.4902,
+      "step": 48534
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.552006959915161,
+      "learning_rate": 1.1103987795228772e-06,
+      "loss": 0.6265,
+      "step": 48535
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.8764231204986572,
+      "learning_rate": 1.1102331118798982e-06,
+      "loss": 0.7445,
+      "step": 48536
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.9436826705932617,
+      "learning_rate": 1.1100674550529016e-06,
+      "loss": 0.7412,
+      "step": 48537
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.1785483360290527,
+      "learning_rate": 1.1099018090423526e-06,
+      "loss": 0.7068,
+      "step": 48538
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.5735843181610107,
+      "learning_rate": 1.1097361738487062e-06,
+      "loss": 0.6024,
+      "step": 48539
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.5259952545166016,
+      "learning_rate": 1.1095705494724285e-06,
+      "loss": 0.7513,
+      "step": 48540
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.4700381755828857,
+      "learning_rate": 1.1094049359139768e-06,
+      "loss": 0.6606,
+      "step": 48541
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.9065513610839844,
+      "learning_rate": 1.1092393331738139e-06,
+      "loss": 0.4561,
+      "step": 48542
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.155758857727051,
+      "learning_rate": 1.1090737412523979e-06,
+      "loss": 0.484,
+      "step": 48543
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 6.013664722442627,
+      "learning_rate": 1.108908160150191e-06,
+      "loss": 0.6563,
+      "step": 48544
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.0648229122161865,
+      "learning_rate": 1.1087425898676517e-06,
+      "loss": 0.5413,
+      "step": 48545
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 4.5317230224609375,
+      "learning_rate": 1.108577030405243e-06,
+      "loss": 0.4952,
+      "step": 48546
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.764819383621216,
+      "learning_rate": 1.1084114817634222e-06,
+      "loss": 0.4805,
+      "step": 48547
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.4327151775360107,
+      "learning_rate": 1.108245943942653e-06,
+      "loss": 0.4369,
+      "step": 48548
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.5445375442504883,
+      "learning_rate": 1.1080804169433924e-06,
+      "loss": 0.7157,
+      "step": 48549
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.5875978469848633,
+      "learning_rate": 1.1079149007661039e-06,
+      "loss": 0.5564,
+      "step": 48550
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.47916316986084,
+      "learning_rate": 1.107749395411244e-06,
+      "loss": 0.7354,
+      "step": 48551
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.492588758468628,
+      "learning_rate": 1.1075839008792766e-06,
+      "loss": 0.5919,
+      "step": 48552
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.392204523086548,
+      "learning_rate": 1.1074184171706593e-06,
+      "loss": 0.6687,
+      "step": 48553
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.5722415447235107,
+      "learning_rate": 1.1072529442858543e-06,
+      "loss": 0.8808,
+      "step": 48554
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.810434103012085,
+      "learning_rate": 1.107087482225319e-06,
+      "loss": 0.4828,
+      "step": 48555
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.0297374725341797,
+      "learning_rate": 1.1069220309895168e-06,
+      "loss": 0.3667,
+      "step": 48556
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 1.9637539386749268,
+      "learning_rate": 1.106756590578904e-06,
+      "loss": 0.5665,
+      "step": 48557
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 5.177717208862305,
+      "learning_rate": 1.1065911609939451e-06,
+      "loss": 0.6627,
+      "step": 48558
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.3113629817962646,
+      "learning_rate": 1.1064257422350954e-06,
+      "loss": 0.6943,
+      "step": 48559
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.909693956375122,
+      "learning_rate": 1.1062603343028177e-06,
+      "loss": 0.647,
+      "step": 48560
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 4.228485584259033,
+      "learning_rate": 1.106094937197571e-06,
+      "loss": 0.6782,
+      "step": 48561
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.5104563236236572,
+      "learning_rate": 1.105929550919817e-06,
+      "loss": 0.5463,
+      "step": 48562
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.211246967315674,
+      "learning_rate": 1.1057641754700133e-06,
+      "loss": 0.9186,
+      "step": 48563
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.177522659301758,
+      "learning_rate": 1.1055988108486215e-06,
+      "loss": 0.6909,
+      "step": 48564
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.1243741512298584,
+      "learning_rate": 1.1054334570560994e-06,
+      "loss": 0.7473,
+      "step": 48565
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.598795175552368,
+      "learning_rate": 1.105268114092909e-06,
+      "loss": 0.6534,
+      "step": 48566
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 4.243252754211426,
+      "learning_rate": 1.1051027819595077e-06,
+      "loss": 0.9213,
+      "step": 48567
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.6773834228515625,
+      "learning_rate": 1.104937460656358e-06,
+      "loss": 0.6717,
+      "step": 48568
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.0418412685394287,
+      "learning_rate": 1.104772150183917e-06,
+      "loss": 0.6952,
+      "step": 48569
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.438183546066284,
+      "learning_rate": 1.104606850542646e-06,
+      "loss": 0.4208,
+      "step": 48570
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.417309284210205,
+      "learning_rate": 1.1044415617330035e-06,
+      "loss": 0.5071,
+      "step": 48571
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.959512948989868,
+      "learning_rate": 1.1042762837554511e-06,
+      "loss": 0.7156,
+      "step": 48572
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.2327044010162354,
+      "learning_rate": 1.104111016610445e-06,
+      "loss": 0.3148,
+      "step": 48573
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.5425217151641846,
+      "learning_rate": 1.103945760298447e-06,
+      "loss": 0.5785,
+      "step": 48574
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.148313522338867,
+      "learning_rate": 1.1037805148199177e-06,
+      "loss": 0.7992,
+      "step": 48575
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.5879085063934326,
+      "learning_rate": 1.103615280175314e-06,
+      "loss": 0.4412,
+      "step": 48576
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.0720338821411133,
+      "learning_rate": 1.1034500563650979e-06,
+      "loss": 0.6859,
+      "step": 48577
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 4.390686988830566,
+      "learning_rate": 1.1032848433897265e-06,
+      "loss": 0.5964,
+      "step": 48578
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.526806354522705,
+      "learning_rate": 1.103119641249661e-06,
+      "loss": 0.3614,
+      "step": 48579
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.318495273590088,
+      "learning_rate": 1.1029544499453588e-06,
+      "loss": 0.5056,
+      "step": 48580
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.7945969104766846,
+      "learning_rate": 1.1027892694772806e-06,
+      "loss": 0.7768,
+      "step": 48581
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.42679762840271,
+      "learning_rate": 1.1026240998458864e-06,
+      "loss": 0.6121,
+      "step": 48582
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.294589042663574,
+      "learning_rate": 1.102458941051634e-06,
+      "loss": 0.7142,
+      "step": 48583
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.2770943641662598,
+      "learning_rate": 1.102293793094984e-06,
+      "loss": 0.4701,
+      "step": 48584
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.2626497745513916,
+      "learning_rate": 1.1021286559763934e-06,
+      "loss": 0.3851,
+      "step": 48585
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.781754970550537,
+      "learning_rate": 1.1019635296963233e-06,
+      "loss": 0.523,
+      "step": 48586
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.24704909324646,
+      "learning_rate": 1.1017984142552334e-06,
+      "loss": 0.6225,
+      "step": 48587
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.8295509815216064,
+      "learning_rate": 1.1016333096535803e-06,
+      "loss": 0.5053,
+      "step": 48588
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.5603389739990234,
+      "learning_rate": 1.1014682158918266e-06,
+      "loss": 0.5579,
+      "step": 48589
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.3264248371124268,
+      "learning_rate": 1.101303132970427e-06,
+      "loss": 0.4357,
+      "step": 48590
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.9542672634124756,
+      "learning_rate": 1.1011380608898453e-06,
+      "loss": 0.4898,
+      "step": 48591
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.743133544921875,
+      "learning_rate": 1.1009729996505365e-06,
+      "loss": 0.5062,
+      "step": 48592
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 6.008751392364502,
+      "learning_rate": 1.1008079492529622e-06,
+      "loss": 0.7195,
+      "step": 48593
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.721757650375366,
+      "learning_rate": 1.100642909697579e-06,
+      "loss": 0.78,
+      "step": 48594
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.303856134414673,
+      "learning_rate": 1.1004778809848487e-06,
+      "loss": 0.5161,
+      "step": 48595
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.347231388092041,
+      "learning_rate": 1.100312863115227e-06,
+      "loss": 0.4772,
+      "step": 48596
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.4840006828308105,
+      "learning_rate": 1.1001478560891759e-06,
+      "loss": 0.6005,
+      "step": 48597
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.835336446762085,
+      "learning_rate": 1.0999828599071516e-06,
+      "loss": 0.7379,
+      "step": 48598
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.08953595161438,
+      "learning_rate": 1.0998178745696147e-06,
+      "loss": 0.5383,
+      "step": 48599
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.612851142883301,
+      "learning_rate": 1.0996529000770212e-06,
+      "loss": 0.7102,
+      "step": 48600
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.5967631340026855,
+      "learning_rate": 1.0994879364298344e-06,
+      "loss": 0.7481,
+      "step": 48601
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.5419468879699707,
+      "learning_rate": 1.099322983628509e-06,
+      "loss": 0.4567,
+      "step": 48602
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.9410128593444824,
+      "learning_rate": 1.0991580416735065e-06,
+      "loss": 0.5684,
+      "step": 48603
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 4.3665690422058105,
+      "learning_rate": 1.0989931105652829e-06,
+      "loss": 0.6291,
+      "step": 48604
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.343961477279663,
+      "learning_rate": 1.0988281903042997e-06,
+      "loss": 0.6382,
+      "step": 48605
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.752030372619629,
+      "learning_rate": 1.098663280891012e-06,
+      "loss": 0.6223,
+      "step": 48606
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.972018003463745,
+      "learning_rate": 1.0984983823258822e-06,
+      "loss": 0.7725,
+      "step": 48607
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.3572661876678467,
+      "learning_rate": 1.098333494609365e-06,
+      "loss": 0.444,
+      "step": 48608
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.7304272651672363,
+      "learning_rate": 1.098168617741922e-06,
+      "loss": 0.6482,
+      "step": 48609
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.4374582767486572,
+      "learning_rate": 1.0980037517240093e-06,
+      "loss": 0.5735,
+      "step": 48610
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.9443767070770264,
+      "learning_rate": 1.0978388965560883e-06,
+      "loss": 0.6555,
+      "step": 48611
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.877141237258911,
+      "learning_rate": 1.0976740522386136e-06,
+      "loss": 0.7308,
+      "step": 48612
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.190729856491089,
+      "learning_rate": 1.097509218772047e-06,
+      "loss": 0.4633,
+      "step": 48613
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.1744980812072754,
+      "learning_rate": 1.0973443961568442e-06,
+      "loss": 0.5087,
+      "step": 48614
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.2923574447631836,
+      "learning_rate": 1.0971795843934658e-06,
+      "loss": 0.5875,
+      "step": 48615
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.840686798095703,
+      "learning_rate": 1.0970147834823675e-06,
+      "loss": 0.9471,
+      "step": 48616
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 4.790170192718506,
+      "learning_rate": 1.0968499934240107e-06,
+      "loss": 0.5906,
+      "step": 48617
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.3937673568725586,
+      "learning_rate": 1.0966852142188506e-06,
+      "loss": 0.7772,
+      "step": 48618
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.517726421356201,
+      "learning_rate": 1.096520445867348e-06,
+      "loss": 0.5977,
+      "step": 48619
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.70175838470459,
+      "learning_rate": 1.0963556883699577e-06,
+      "loss": 0.4733,
+      "step": 48620
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.892000913619995,
+      "learning_rate": 1.0961909417271406e-06,
+      "loss": 0.591,
+      "step": 48621
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.1932668685913086,
+      "learning_rate": 1.0960262059393539e-06,
+      "loss": 0.5919,
+      "step": 48622
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.6080121994018555,
+      "learning_rate": 1.0958614810070572e-06,
+      "loss": 0.4974,
+      "step": 48623
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.2440197467803955,
+      "learning_rate": 1.0956967669307057e-06,
+      "loss": 0.4728,
+      "step": 48624
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 4.9738054275512695,
+      "learning_rate": 1.0955320637107603e-06,
+      "loss": 0.9262,
+      "step": 48625
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.6493468284606934,
+      "learning_rate": 1.0953673713476764e-06,
+      "loss": 0.6191,
+      "step": 48626
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.8325417041778564,
+      "learning_rate": 1.0952026898419144e-06,
+      "loss": 0.7309,
+      "step": 48627
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.094252824783325,
+      "learning_rate": 1.0950380191939297e-06,
+      "loss": 0.6447,
+      "step": 48628
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 5.078554630279541,
+      "learning_rate": 1.0948733594041826e-06,
+      "loss": 0.6211,
+      "step": 48629
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.0028493404388428,
+      "learning_rate": 1.094708710473129e-06,
+      "loss": 0.56,
+      "step": 48630
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.3571863174438477,
+      "learning_rate": 1.0945440724012286e-06,
+      "loss": 0.4812,
+      "step": 48631
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.6580166816711426,
+      "learning_rate": 1.0943794451889361e-06,
+      "loss": 0.5476,
+      "step": 48632
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.429788112640381,
+      "learning_rate": 1.0942148288367132e-06,
+      "loss": 0.6593,
+      "step": 48633
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.612773895263672,
+      "learning_rate": 1.0940502233450145e-06,
+      "loss": 0.7785,
+      "step": 48634
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.5860090255737305,
+      "learning_rate": 1.0938856287142984e-06,
+      "loss": 0.5973,
+      "step": 48635
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.4774234294891357,
+      "learning_rate": 1.0937210449450252e-06,
+      "loss": 0.6878,
+      "step": 48636
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.5109598636627197,
+      "learning_rate": 1.0935564720376486e-06,
+      "loss": 0.6035,
+      "step": 48637
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.533395290374756,
+      "learning_rate": 1.0933919099926294e-06,
+      "loss": 0.5873,
+      "step": 48638
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.311757802963257,
+      "learning_rate": 1.0932273588104226e-06,
+      "loss": 0.485,
+      "step": 48639
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.8019425868988037,
+      "learning_rate": 1.0930628184914883e-06,
+      "loss": 0.6139,
+      "step": 48640
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.3209116458892822,
+      "learning_rate": 1.0928982890362816e-06,
+      "loss": 0.5068,
+      "step": 48641
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.7048094272613525,
+      "learning_rate": 1.0927337704452607e-06,
+      "loss": 0.5703,
+      "step": 48642
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.268988609313965,
+      "learning_rate": 1.0925692627188856e-06,
+      "loss": 0.794,
+      "step": 48643
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.136317014694214,
+      "learning_rate": 1.0924047658576091e-06,
+      "loss": 0.5321,
+      "step": 48644
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.923509359359741,
+      "learning_rate": 1.0922402798618937e-06,
+      "loss": 0.6582,
+      "step": 48645
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.3737781047821045,
+      "learning_rate": 1.0920758047321922e-06,
+      "loss": 0.9048,
+      "step": 48646
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.849513053894043,
+      "learning_rate": 1.0919113404689642e-06,
+      "loss": 0.8027,
+      "step": 48647
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.9512476921081543,
+      "learning_rate": 1.0917468870726678e-06,
+      "loss": 0.6523,
+      "step": 48648
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.216026782989502,
+      "learning_rate": 1.0915824445437578e-06,
+      "loss": 0.5006,
+      "step": 48649
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.191910982131958,
+      "learning_rate": 1.0914180128826945e-06,
+      "loss": 0.7073,
+      "step": 48650
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.7785580158233643,
+      "learning_rate": 1.0912535920899325e-06,
+      "loss": 0.7128,
+      "step": 48651
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.864541530609131,
+      "learning_rate": 1.0910891821659307e-06,
+      "loss": 0.587,
+      "step": 48652
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.5669753551483154,
+      "learning_rate": 1.0909247831111447e-06,
+      "loss": 0.6779,
+      "step": 48653
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.725370407104492,
+      "learning_rate": 1.0907603949260337e-06,
+      "loss": 0.773,
+      "step": 48654
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 1.6646168231964111,
+      "learning_rate": 1.090596017611052e-06,
+      "loss": 0.226,
+      "step": 48655
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.520298957824707,
+      "learning_rate": 1.0904316511666602e-06,
+      "loss": 0.494,
+      "step": 48656
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.7737278938293457,
+      "learning_rate": 1.0902672955933119e-06,
+      "loss": 0.5624,
+      "step": 48657
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.5921876430511475,
+      "learning_rate": 1.0901029508914669e-06,
+      "loss": 0.6057,
+      "step": 48658
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.3222715854644775,
+      "learning_rate": 1.0899386170615788e-06,
+      "loss": 0.6279,
+      "step": 48659
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.634291648864746,
+      "learning_rate": 1.0897742941041089e-06,
+      "loss": 0.8445,
+      "step": 48660
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 1.951351284980774,
+      "learning_rate": 1.0896099820195094e-06,
+      "loss": 0.5441,
+      "step": 48661
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.309997081756592,
+      "learning_rate": 1.0894456808082416e-06,
+      "loss": 0.5509,
+      "step": 48662
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 4.056211471557617,
+      "learning_rate": 1.0892813904707595e-06,
+      "loss": 0.727,
+      "step": 48663
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.0581915378570557,
+      "learning_rate": 1.089117111007522e-06,
+      "loss": 0.4191,
+      "step": 48664
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.559527635574341,
+      "learning_rate": 1.088952842418984e-06,
+      "loss": 0.4087,
+      "step": 48665
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 4.214475631713867,
+      "learning_rate": 1.088788584705604e-06,
+      "loss": 0.7363,
+      "step": 48666
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.6865010261535645,
+      "learning_rate": 1.0886243378678362e-06,
+      "loss": 0.5173,
+      "step": 48667
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 4.113267421722412,
+      "learning_rate": 1.0884601019061402e-06,
+      "loss": 0.6631,
+      "step": 48668
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.6393980979919434,
+      "learning_rate": 1.0882958768209707e-06,
+      "loss": 0.5199,
+      "step": 48669
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.9538321495056152,
+      "learning_rate": 1.0881316626127859e-06,
+      "loss": 0.5658,
+      "step": 48670
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.9440786838531494,
+      "learning_rate": 1.08796745928204e-06,
+      "loss": 0.6458,
+      "step": 48671
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.163050413131714,
+      "learning_rate": 1.0878032668291927e-06,
+      "loss": 0.5517,
+      "step": 48672
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.104461669921875,
+      "learning_rate": 1.0876390852546975e-06,
+      "loss": 0.4455,
+      "step": 48673
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.665569543838501,
+      "learning_rate": 1.0874749145590136e-06,
+      "loss": 0.5347,
+      "step": 48674
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.6319241523742676,
+      "learning_rate": 1.0873107547425955e-06,
+      "loss": 0.4844,
+      "step": 48675
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 4.022953510284424,
+      "learning_rate": 1.087146605805901e-06,
+      "loss": 0.7469,
+      "step": 48676
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.324777364730835,
+      "learning_rate": 1.086982467749385e-06,
+      "loss": 0.4228,
+      "step": 48677
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.6235125064849854,
+      "learning_rate": 1.0868183405735067e-06,
+      "loss": 0.6406,
+      "step": 48678
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.6175620555877686,
+      "learning_rate": 1.0866542242787192e-06,
+      "loss": 0.6478,
+      "step": 48679
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.1906440258026123,
+      "learning_rate": 1.0864901188654814e-06,
+      "loss": 0.8482,
+      "step": 48680
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.7042856216430664,
+      "learning_rate": 1.086326024334247e-06,
+      "loss": 0.6196,
+      "step": 48681
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 1.759860634803772,
+      "learning_rate": 1.086161940685474e-06,
+      "loss": 0.3933,
+      "step": 48682
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.8096539974212646,
+      "learning_rate": 1.0859978679196186e-06,
+      "loss": 0.5295,
+      "step": 48683
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.504598379135132,
+      "learning_rate": 1.0858338060371387e-06,
+      "loss": 0.6675,
+      "step": 48684
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.9740712642669678,
+      "learning_rate": 1.0856697550384865e-06,
+      "loss": 0.6356,
+      "step": 48685
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.6463887691497803,
+      "learning_rate": 1.085505714924122e-06,
+      "loss": 0.339,
+      "step": 48686
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.8102996349334717,
+      "learning_rate": 1.0853416856944988e-06,
+      "loss": 0.5531,
+      "step": 48687
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.871105194091797,
+      "learning_rate": 1.0851776673500753e-06,
+      "loss": 0.4099,
+      "step": 48688
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.7763831615448,
+      "learning_rate": 1.085013659891304e-06,
+      "loss": 0.5865,
+      "step": 48689
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.785719871520996,
+      "learning_rate": 1.0848496633186451e-06,
+      "loss": 0.306,
+      "step": 48690
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.1251139640808105,
+      "learning_rate": 1.0846856776325509e-06,
+      "loss": 0.2481,
+      "step": 48691
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 1.9451972246170044,
+      "learning_rate": 1.084521702833481e-06,
+      "loss": 0.4928,
+      "step": 48692
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 5.429105758666992,
+      "learning_rate": 1.084357738921888e-06,
+      "loss": 0.5483,
+      "step": 48693
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 4.903088092803955,
+      "learning_rate": 1.0841937858982305e-06,
+      "loss": 0.6196,
+      "step": 48694
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.0813989639282227,
+      "learning_rate": 1.0840298437629614e-06,
+      "loss": 0.5213,
+      "step": 48695
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 4.055647850036621,
+      "learning_rate": 1.08386591251654e-06,
+      "loss": 0.5293,
+      "step": 48696
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.1960976123809814,
+      "learning_rate": 1.0837019921594194e-06,
+      "loss": 0.6875,
+      "step": 48697
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.406790256500244,
+      "learning_rate": 1.0835380826920566e-06,
+      "loss": 0.4865,
+      "step": 48698
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.473726987838745,
+      "learning_rate": 1.083374184114908e-06,
+      "loss": 0.6578,
+      "step": 48699
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.7507801055908203,
+      "learning_rate": 1.0832102964284275e-06,
+      "loss": 0.5406,
+      "step": 48700
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.29012131690979,
+      "learning_rate": 1.0830464196330737e-06,
+      "loss": 0.5259,
+      "step": 48701
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.7080025672912598,
+      "learning_rate": 1.0828825537292987e-06,
+      "loss": 0.5685,
+      "step": 48702
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.6179587841033936,
+      "learning_rate": 1.0827186987175598e-06,
+      "loss": 0.5119,
+      "step": 48703
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.1204378604888916,
+      "learning_rate": 1.0825548545983144e-06,
+      "loss": 0.8196,
+      "step": 48704
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.0875134468078613,
+      "learning_rate": 1.0823910213720151e-06,
+      "loss": 0.3696,
+      "step": 48705
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.961042881011963,
+      "learning_rate": 1.0822271990391199e-06,
+      "loss": 0.5982,
+      "step": 48706
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.637305498123169,
+      "learning_rate": 1.0820633876000818e-06,
+      "loss": 0.8326,
+      "step": 48707
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 1.7100707292556763,
+      "learning_rate": 1.081899587055358e-06,
+      "loss": 0.5525,
+      "step": 48708
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.9603874683380127,
+      "learning_rate": 1.081735797405405e-06,
+      "loss": 0.6745,
+      "step": 48709
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 1.5699996948242188,
+      "learning_rate": 1.0815720186506755e-06,
+      "loss": 0.2743,
+      "step": 48710
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 4.260879039764404,
+      "learning_rate": 1.0814082507916279e-06,
+      "loss": 0.6712,
+      "step": 48711
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.3262031078338623,
+      "learning_rate": 1.0812444938287142e-06,
+      "loss": 0.5695,
+      "step": 48712
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.9724085330963135,
+      "learning_rate": 1.0810807477623937e-06,
+      "loss": 0.6521,
+      "step": 48713
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.4475560188293457,
+      "learning_rate": 1.0809170125931172e-06,
+      "loss": 0.5336,
+      "step": 48714
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.4077885150909424,
+      "learning_rate": 1.0807532883213444e-06,
+      "loss": 0.4412,
+      "step": 48715
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.166766881942749,
+      "learning_rate": 1.080589574947527e-06,
+      "loss": 0.6577,
+      "step": 48716
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.8976633548736572,
+      "learning_rate": 1.0804258724721234e-06,
+      "loss": 0.9087,
+      "step": 48717
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.0447328090667725,
+      "learning_rate": 1.0802621808955854e-06,
+      "loss": 0.7051,
+      "step": 48718
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.7261157035827637,
+      "learning_rate": 1.0800985002183718e-06,
+      "loss": 0.7056,
+      "step": 48719
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.8645505905151367,
+      "learning_rate": 1.079934830440934e-06,
+      "loss": 0.3703,
+      "step": 48720
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.5879476070404053,
+      "learning_rate": 1.0797711715637298e-06,
+      "loss": 0.4735,
+      "step": 48721
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.9526801109313965,
+      "learning_rate": 1.0796075235872116e-06,
+      "loss": 0.7235,
+      "step": 48722
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 6.54984188079834,
+      "learning_rate": 1.0794438865118389e-06,
+      "loss": 0.608,
+      "step": 48723
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.3726537227630615,
+      "learning_rate": 1.0792802603380624e-06,
+      "loss": 0.5401,
+      "step": 48724
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.833921194076538,
+      "learning_rate": 1.07911664506634e-06,
+      "loss": 0.4257,
+      "step": 48725
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.256112813949585,
+      "learning_rate": 1.0789530406971243e-06,
+      "loss": 0.5409,
+      "step": 48726
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 1.9918570518493652,
+      "learning_rate": 1.078789447230873e-06,
+      "loss": 0.375,
+      "step": 48727
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.7728261947631836,
+      "learning_rate": 1.0786258646680375e-06,
+      "loss": 0.5314,
+      "step": 48728
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.8736953735351562,
+      "learning_rate": 1.078462293009076e-06,
+      "loss": 0.6042,
+      "step": 48729
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 7.469000816345215,
+      "learning_rate": 1.0782987322544402e-06,
+      "loss": 0.7831,
+      "step": 48730
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.402907133102417,
+      "learning_rate": 1.0781351824045883e-06,
+      "loss": 0.7009,
+      "step": 48731
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.7980594635009766,
+      "learning_rate": 1.0779716434599718e-06,
+      "loss": 0.5778,
+      "step": 48732
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.671496868133545,
+      "learning_rate": 1.077808115421048e-06,
+      "loss": 0.5785,
+      "step": 48733
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.3748245239257812,
+      "learning_rate": 1.0776445982882693e-06,
+      "loss": 0.4099,
+      "step": 48734
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.0875580310821533,
+      "learning_rate": 1.077481092062093e-06,
+      "loss": 0.4931,
+      "step": 48735
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.419363498687744,
+      "learning_rate": 1.0773175967429717e-06,
+      "loss": 0.7693,
+      "step": 48736
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 8.524060249328613,
+      "learning_rate": 1.0771541123313612e-06,
+      "loss": 0.556,
+      "step": 48737
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.5789682865142822,
+      "learning_rate": 1.0769906388277146e-06,
+      "loss": 0.5847,
+      "step": 48738
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.6779794692993164,
+      "learning_rate": 1.0768271762324889e-06,
+      "loss": 0.584,
+      "step": 48739
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 1.9852296113967896,
+      "learning_rate": 1.0766637245461354e-06,
+      "loss": 0.4489,
+      "step": 48740
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.6294305324554443,
+      "learning_rate": 1.076500283769112e-06,
+      "loss": 0.6843,
+      "step": 48741
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.9023492336273193,
+      "learning_rate": 1.0763368539018698e-06,
+      "loss": 0.6658,
+      "step": 48742
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.5923702716827393,
+      "learning_rate": 1.0761734349448654e-06,
+      "loss": 0.6832,
+      "step": 48743
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.796938896179199,
+      "learning_rate": 1.076010026898553e-06,
+      "loss": 0.5068,
+      "step": 48744
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.237105369567871,
+      "learning_rate": 1.0758466297633874e-06,
+      "loss": 0.5789,
+      "step": 48745
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.592900514602661,
+      "learning_rate": 1.0756832435398213e-06,
+      "loss": 0.7096,
+      "step": 48746
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.7161989212036133,
+      "learning_rate": 1.0755198682283114e-06,
+      "loss": 0.4328,
+      "step": 48747
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.571936845779419,
+      "learning_rate": 1.0753565038293096e-06,
+      "loss": 0.5267,
+      "step": 48748
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.008481025695801,
+      "learning_rate": 1.0751931503432722e-06,
+      "loss": 0.6425,
+      "step": 48749
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.1732144355773926,
+      "learning_rate": 1.0750298077706506e-06,
+      "loss": 0.6855,
+      "step": 48750
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.848813533782959,
+      "learning_rate": 1.0748664761119026e-06,
+      "loss": 0.7066,
+      "step": 48751
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 4.118247032165527,
+      "learning_rate": 1.074703155367479e-06,
+      "loss": 0.6512,
+      "step": 48752
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.4528753757476807,
+      "learning_rate": 1.074539845537837e-06,
+      "loss": 0.4634,
+      "step": 48753
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.0076115131378174,
+      "learning_rate": 1.074376546623428e-06,
+      "loss": 0.2963,
+      "step": 48754
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.7114100456237793,
+      "learning_rate": 1.0742132586247084e-06,
+      "loss": 0.5348,
+      "step": 48755
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.08078932762146,
+      "learning_rate": 1.0740499815421296e-06,
+      "loss": 0.4761,
+      "step": 48756
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.075927495956421,
+      "learning_rate": 1.0738867153761485e-06,
+      "loss": 0.6627,
+      "step": 48757
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.601882219314575,
+      "learning_rate": 1.0737234601272163e-06,
+      "loss": 0.5109,
+      "step": 48758
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.7109317779541016,
+      "learning_rate": 1.0735602157957887e-06,
+      "loss": 0.5363,
+      "step": 48759
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.1638729572296143,
+      "learning_rate": 1.0733969823823204e-06,
+      "loss": 0.6865,
+      "step": 48760
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 4.1392951011657715,
+      "learning_rate": 1.0732337598872627e-06,
+      "loss": 0.6344,
+      "step": 48761
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.041933059692383,
+      "learning_rate": 1.073070548311072e-06,
+      "loss": 0.6423,
+      "step": 48762
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 4.412180423736572,
+      "learning_rate": 1.0729073476542002e-06,
+      "loss": 0.7797,
+      "step": 48763
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 4.006550312042236,
+      "learning_rate": 1.0727441579171015e-06,
+      "loss": 0.3913,
+      "step": 48764
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.7511470317840576,
+      "learning_rate": 1.0725809791002318e-06,
+      "loss": 0.4272,
+      "step": 48765
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.2875077724456787,
+      "learning_rate": 1.0724178112040418e-06,
+      "loss": 0.6539,
+      "step": 48766
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.2777857780456543,
+      "learning_rate": 1.072254654228988e-06,
+      "loss": 0.4783,
+      "step": 48767
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 4.105441570281982,
+      "learning_rate": 1.0720915081755213e-06,
+      "loss": 0.5963,
+      "step": 48768
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.2037107944488525,
+      "learning_rate": 1.0719283730440978e-06,
+      "loss": 0.6372,
+      "step": 48769
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.5070950984954834,
+      "learning_rate": 1.0717652488351681e-06,
+      "loss": 0.4406,
+      "step": 48770
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.3346750736236572,
+      "learning_rate": 1.0716021355491885e-06,
+      "loss": 0.6637,
+      "step": 48771
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.6990997791290283,
+      "learning_rate": 1.071439033186613e-06,
+      "loss": 0.5661,
+      "step": 48772
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.0069918632507324,
+      "learning_rate": 1.0712759417478925e-06,
+      "loss": 0.6179,
+      "step": 48773
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.100238561630249,
+      "learning_rate": 1.0711128612334831e-06,
+      "loss": 0.6362,
+      "step": 48774
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.813931941986084,
+      "learning_rate": 1.0709497916438355e-06,
+      "loss": 0.5376,
+      "step": 48775
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.7329585552215576,
+      "learning_rate": 1.0707867329794063e-06,
+      "loss": 0.5795,
+      "step": 48776
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.7098135948181152,
+      "learning_rate": 1.0706236852406459e-06,
+      "loss": 0.5574,
+      "step": 48777
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.6259171962738037,
+      "learning_rate": 1.0704606484280106e-06,
+      "loss": 0.6845,
+      "step": 48778
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.0566630363464355,
+      "learning_rate": 1.070297622541951e-06,
+      "loss": 0.5021,
+      "step": 48779
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.8821563720703125,
+      "learning_rate": 1.0701346075829221e-06,
+      "loss": 0.5139,
+      "step": 48780
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.1229782104492188,
+      "learning_rate": 1.0699716035513758e-06,
+      "loss": 0.5983,
+      "step": 48781
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.7711710929870605,
+      "learning_rate": 1.0698086104477679e-06,
+      "loss": 0.7998,
+      "step": 48782
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.1426589488983154,
+      "learning_rate": 1.069645628272547e-06,
+      "loss": 0.6433,
+      "step": 48783
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.418632745742798,
+      "learning_rate": 1.0694826570261718e-06,
+      "loss": 0.5035,
+      "step": 48784
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.9743905067443848,
+      "learning_rate": 1.0693196967090919e-06,
+      "loss": 0.6489,
+      "step": 48785
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.6983048915863037,
+      "learning_rate": 1.0691567473217628e-06,
+      "loss": 0.5532,
+      "step": 48786
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.407632350921631,
+      "learning_rate": 1.0689938088646345e-06,
+      "loss": 0.6574,
+      "step": 48787
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.812772035598755,
+      "learning_rate": 1.0688308813381637e-06,
+      "loss": 0.7838,
+      "step": 48788
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.349736213684082,
+      "learning_rate": 1.0686679647428e-06,
+      "loss": 0.424,
+      "step": 48789
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.4785361289978027,
+      "learning_rate": 1.0685050590789992e-06,
+      "loss": 0.5501,
+      "step": 48790
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.1292030811309814,
+      "learning_rate": 1.068342164347212e-06,
+      "loss": 0.7628,
+      "step": 48791
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.128779649734497,
+      "learning_rate": 1.0681792805478935e-06,
+      "loss": 0.4112,
+      "step": 48792
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.9398512840270996,
+      "learning_rate": 1.0680164076814942e-06,
+      "loss": 0.7345,
+      "step": 48793
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.032966375350952,
+      "learning_rate": 1.0678535457484696e-06,
+      "loss": 0.4896,
+      "step": 48794
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 4.197638034820557,
+      "learning_rate": 1.06769069474927e-06,
+      "loss": 0.3895,
+      "step": 48795
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.8585314750671387,
+      "learning_rate": 1.067527854684351e-06,
+      "loss": 0.3194,
+      "step": 48796
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.4458088874816895,
+      "learning_rate": 1.0673650255541624e-06,
+      "loss": 0.4597,
+      "step": 48797
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.5160751342773438,
+      "learning_rate": 1.0672022073591597e-06,
+      "loss": 0.6187,
+      "step": 48798
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.3922817707061768,
+      "learning_rate": 1.0670394000997935e-06,
+      "loss": 0.6973,
+      "step": 48799
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 5.353832244873047,
+      "learning_rate": 1.0668766037765183e-06,
+      "loss": 0.7015,
+      "step": 48800
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.408381462097168,
+      "learning_rate": 1.0667138183897847e-06,
+      "loss": 0.5276,
+      "step": 48801
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.0566346645355225,
+      "learning_rate": 1.0665510439400478e-06,
+      "loss": 0.8413,
+      "step": 48802
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.0770580768585205,
+      "learning_rate": 1.0663882804277575e-06,
+      "loss": 0.4265,
+      "step": 48803
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.478766679763794,
+      "learning_rate": 1.0662255278533683e-06,
+      "loss": 0.8335,
+      "step": 48804
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.1409082412719727,
+      "learning_rate": 1.0660627862173317e-06,
+      "loss": 0.5071,
+      "step": 48805
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.633575201034546,
+      "learning_rate": 1.0659000555201027e-06,
+      "loss": 0.4888,
+      "step": 48806
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.764326810836792,
+      "learning_rate": 1.0657373357621304e-06,
+      "loss": 0.6063,
+      "step": 48807
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.3502533435821533,
+      "learning_rate": 1.0655746269438699e-06,
+      "loss": 0.5633,
+      "step": 48808
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.6675913333892822,
+      "learning_rate": 1.065411929065771e-06,
+      "loss": 0.5524,
+      "step": 48809
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.1131296157836914,
+      "learning_rate": 1.0652492421282896e-06,
+      "loss": 0.5482,
+      "step": 48810
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.347722291946411,
+      "learning_rate": 1.0650865661318739e-06,
+      "loss": 0.5831,
+      "step": 48811
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.399885416030884,
+      "learning_rate": 1.0649239010769802e-06,
+      "loss": 0.4372,
+      "step": 48812
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 5.407250881195068,
+      "learning_rate": 1.0647612469640577e-06,
+      "loss": 0.4947,
+      "step": 48813
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.612870931625366,
+      "learning_rate": 1.0645986037935613e-06,
+      "loss": 0.5021,
+      "step": 48814
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 1.8754428625106812,
+      "learning_rate": 1.0644359715659403e-06,
+      "loss": 0.464,
+      "step": 48815
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.394038200378418,
+      "learning_rate": 1.0642733502816505e-06,
+      "loss": 0.4887,
+      "step": 48816
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.235018014907837,
+      "learning_rate": 1.0641107399411404e-06,
+      "loss": 0.6763,
+      "step": 48817
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.0112738609313965,
+      "learning_rate": 1.0639481405448654e-06,
+      "loss": 0.5995,
+      "step": 48818
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.699460744857788,
+      "learning_rate": 1.0637855520932744e-06,
+      "loss": 0.613,
+      "step": 48819
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.88942551612854,
+      "learning_rate": 1.0636229745868214e-06,
+      "loss": 0.5164,
+      "step": 48820
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.3136374950408936,
+      "learning_rate": 1.0634604080259603e-06,
+      "loss": 0.6549,
+      "step": 48821
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.982578992843628,
+      "learning_rate": 1.0632978524111393e-06,
+      "loss": 0.7339,
+      "step": 48822
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 4.305220127105713,
+      "learning_rate": 1.0631353077428137e-06,
+      "loss": 0.7414,
+      "step": 48823
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 4.005337238311768,
+      "learning_rate": 1.0629727740214324e-06,
+      "loss": 0.5252,
+      "step": 48824
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 4.345390796661377,
+      "learning_rate": 1.0628102512474493e-06,
+      "loss": 0.4866,
+      "step": 48825
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.4209022521972656,
+      "learning_rate": 1.062647739421317e-06,
+      "loss": 0.6481,
+      "step": 48826
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.702866792678833,
+      "learning_rate": 1.0624852385434854e-06,
+      "loss": 0.7787,
+      "step": 48827
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 4.0545454025268555,
+      "learning_rate": 1.0623227486144084e-06,
+      "loss": 0.5641,
+      "step": 48828
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.845869779586792,
+      "learning_rate": 1.0621602696345356e-06,
+      "loss": 0.5915,
+      "step": 48829
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.3639471530914307,
+      "learning_rate": 1.0619978016043215e-06,
+      "loss": 0.6944,
+      "step": 48830
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.1957478523254395,
+      "learning_rate": 1.0618353445242148e-06,
+      "loss": 0.4047,
+      "step": 48831
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.8126790523529053,
+      "learning_rate": 1.0616728983946683e-06,
+      "loss": 0.5372,
+      "step": 48832
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.9276132583618164,
+      "learning_rate": 1.0615104632161356e-06,
+      "loss": 0.4063,
+      "step": 48833
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 1.797755479812622,
+      "learning_rate": 1.0613480389890656e-06,
+      "loss": 0.5196,
+      "step": 48834
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.2328925132751465,
+      "learning_rate": 1.061185625713913e-06,
+      "loss": 0.5558,
+      "step": 48835
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.495569944381714,
+      "learning_rate": 1.061023223391126e-06,
+      "loss": 0.7048,
+      "step": 48836
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.409180164337158,
+      "learning_rate": 1.0608608320211594e-06,
+      "loss": 0.5809,
+      "step": 48837
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 4.158745288848877,
+      "learning_rate": 1.0606984516044616e-06,
+      "loss": 0.6491,
+      "step": 48838
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.2517282962799072,
+      "learning_rate": 1.0605360821414872e-06,
+      "loss": 0.611,
+      "step": 48839
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.364039897918701,
+      "learning_rate": 1.0603737236326844e-06,
+      "loss": 0.5801,
+      "step": 48840
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.902498722076416,
+      "learning_rate": 1.0602113760785083e-06,
+      "loss": 0.804,
+      "step": 48841
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.5800089836120605,
+      "learning_rate": 1.0600490394794072e-06,
+      "loss": 0.5922,
+      "step": 48842
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.7314212322235107,
+      "learning_rate": 1.059886713835835e-06,
+      "loss": 0.6255,
+      "step": 48843
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.16379714012146,
+      "learning_rate": 1.0597243991482391e-06,
+      "loss": 0.6087,
+      "step": 48844
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.596407890319824,
+      "learning_rate": 1.0595620954170765e-06,
+      "loss": 0.5857,
+      "step": 48845
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.017425775527954,
+      "learning_rate": 1.0593998026427937e-06,
+      "loss": 0.4554,
+      "step": 48846
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.27201247215271,
+      "learning_rate": 1.059237520825846e-06,
+      "loss": 0.4925,
+      "step": 48847
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.7046916484832764,
+      "learning_rate": 1.0590752499666806e-06,
+      "loss": 0.5138,
+      "step": 48848
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 4.186049461364746,
+      "learning_rate": 1.0589129900657518e-06,
+      "loss": 0.544,
+      "step": 48849
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.587982416152954,
+      "learning_rate": 1.0587507411235087e-06,
+      "loss": 0.5537,
+      "step": 48850
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.5567386150360107,
+      "learning_rate": 1.0585885031404046e-06,
+      "loss": 0.5082,
+      "step": 48851
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.612999200820923,
+      "learning_rate": 1.058426276116888e-06,
+      "loss": 0.6439,
+      "step": 48852
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.1545894145965576,
+      "learning_rate": 1.0582640600534128e-06,
+      "loss": 0.5773,
+      "step": 48853
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.938532590866089,
+      "learning_rate": 1.058101854950427e-06,
+      "loss": 0.6261,
+      "step": 48854
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 4.084501266479492,
+      "learning_rate": 1.0579396608083847e-06,
+      "loss": 0.8029,
+      "step": 48855
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.5049872398376465,
+      "learning_rate": 1.0577774776277344e-06,
+      "loss": 0.5456,
+      "step": 48856
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.7034337520599365,
+      "learning_rate": 1.0576153054089295e-06,
+      "loss": 0.7947,
+      "step": 48857
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.155827283859253,
+      "learning_rate": 1.0574531441524182e-06,
+      "loss": 0.5509,
+      "step": 48858
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 4.710312366485596,
+      "learning_rate": 1.0572909938586539e-06,
+      "loss": 1.009,
+      "step": 48859
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.303617238998413,
+      "learning_rate": 1.057128854528085e-06,
+      "loss": 0.4961,
+      "step": 48860
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.436255931854248,
+      "learning_rate": 1.0569667261611654e-06,
+      "loss": 0.6528,
+      "step": 48861
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.748417377471924,
+      "learning_rate": 1.0568046087583429e-06,
+      "loss": 0.668,
+      "step": 48862
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.760781764984131,
+      "learning_rate": 1.0566425023200704e-06,
+      "loss": 0.6191,
+      "step": 48863
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.089712381362915,
+      "learning_rate": 1.056480406846797e-06,
+      "loss": 0.7467,
+      "step": 48864
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.3975040912628174,
+      "learning_rate": 1.0563183223389745e-06,
+      "loss": 0.885,
+      "step": 48865
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.532419443130493,
+      "learning_rate": 1.0561562487970534e-06,
+      "loss": 0.5329,
+      "step": 48866
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.808579683303833,
+      "learning_rate": 1.0559941862214852e-06,
+      "loss": 0.5073,
+      "step": 48867
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.4752249717712402,
+      "learning_rate": 1.0558321346127187e-06,
+      "loss": 0.4973,
+      "step": 48868
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 5.913089275360107,
+      "learning_rate": 1.0556700939712066e-06,
+      "loss": 0.5863,
+      "step": 48869
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.6725282669067383,
+      "learning_rate": 1.0555080642973974e-06,
+      "loss": 0.9001,
+      "step": 48870
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.1983978748321533,
+      "learning_rate": 1.0553460455917435e-06,
+      "loss": 0.2994,
+      "step": 48871
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.127624034881592,
+      "learning_rate": 1.055184037854694e-06,
+      "loss": 0.2574,
+      "step": 48872
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.7106661796569824,
+      "learning_rate": 1.0550220410867007e-06,
+      "loss": 0.6274,
+      "step": 48873
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.644080638885498,
+      "learning_rate": 1.0548600552882122e-06,
+      "loss": 0.5873,
+      "step": 48874
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.735989809036255,
+      "learning_rate": 1.0546980804596812e-06,
+      "loss": 0.5867,
+      "step": 48875
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.873354434967041,
+      "learning_rate": 1.0545361166015556e-06,
+      "loss": 0.8865,
+      "step": 48876
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.788510799407959,
+      "learning_rate": 1.0543741637142878e-06,
+      "loss": 0.572,
+      "step": 48877
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.2161529064178467,
+      "learning_rate": 1.054212221798327e-06,
+      "loss": 0.5811,
+      "step": 48878
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.53363037109375,
+      "learning_rate": 1.0540502908541244e-06,
+      "loss": 0.5095,
+      "step": 48879
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.4506213665008545,
+      "learning_rate": 1.053888370882129e-06,
+      "loss": 0.5097,
+      "step": 48880
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 4.1419806480407715,
+      "learning_rate": 1.0537264618827931e-06,
+      "loss": 0.6281,
+      "step": 48881
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 1.9081346988677979,
+      "learning_rate": 1.0535645638565638e-06,
+      "loss": 0.4784,
+      "step": 48882
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.1577725410461426,
+      "learning_rate": 1.0534026768038935e-06,
+      "loss": 0.6058,
+      "step": 48883
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.2832510471343994,
+      "learning_rate": 1.0532408007252331e-06,
+      "loss": 0.4624,
+      "step": 48884
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.677858829498291,
+      "learning_rate": 1.0530789356210302e-06,
+      "loss": 0.5115,
+      "step": 48885
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.5369789600372314,
+      "learning_rate": 1.0529170814917366e-06,
+      "loss": 0.581,
+      "step": 48886
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.046124219894409,
+      "learning_rate": 1.052755238337803e-06,
+      "loss": 0.5342,
+      "step": 48887
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 1.833775520324707,
+      "learning_rate": 1.052593406159677e-06,
+      "loss": 0.2904,
+      "step": 48888
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.719578981399536,
+      "learning_rate": 1.0524315849578116e-06,
+      "loss": 0.7269,
+      "step": 48889
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 1.887885332107544,
+      "learning_rate": 1.0522697747326538e-06,
+      "loss": 0.459,
+      "step": 48890
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.680607557296753,
+      "learning_rate": 1.0521079754846558e-06,
+      "loss": 0.5988,
+      "step": 48891
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.1415324211120605,
+      "learning_rate": 1.0519461872142656e-06,
+      "loss": 0.5983,
+      "step": 48892
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.481736660003662,
+      "learning_rate": 1.0517844099219343e-06,
+      "loss": 0.6739,
+      "step": 48893
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.9956722259521484,
+      "learning_rate": 1.0516226436081124e-06,
+      "loss": 0.6123,
+      "step": 48894
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.8957784175872803,
+      "learning_rate": 1.051460888273248e-06,
+      "loss": 0.6128,
+      "step": 48895
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.4793944358825684,
+      "learning_rate": 1.0512991439177928e-06,
+      "loss": 0.6916,
+      "step": 48896
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.4448647499084473,
+      "learning_rate": 1.051137410542194e-06,
+      "loss": 0.657,
+      "step": 48897
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.8220205307006836,
+      "learning_rate": 1.0509756881469042e-06,
+      "loss": 0.6158,
+      "step": 48898
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.868896722793579,
+      "learning_rate": 1.05081397673237e-06,
+      "loss": 0.7537,
+      "step": 48899
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.2382733821868896,
+      "learning_rate": 1.0506522762990446e-06,
+      "loss": 0.6496,
+      "step": 48900
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.998289108276367,
+      "learning_rate": 1.0504905868473742e-06,
+      "loss": 0.6842,
+      "step": 48901
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.0427193641662598,
+      "learning_rate": 1.0503289083778112e-06,
+      "loss": 0.6135,
+      "step": 48902
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 4.075410842895508,
+      "learning_rate": 1.0501672408908026e-06,
+      "loss": 0.4186,
+      "step": 48903
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.3502793312072754,
+      "learning_rate": 1.0500055843868012e-06,
+      "loss": 0.5397,
+      "step": 48904
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.922490358352661,
+      "learning_rate": 1.0498439388662513e-06,
+      "loss": 0.4941,
+      "step": 48905
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 1.733574628829956,
+      "learning_rate": 1.0496823043296083e-06,
+      "loss": 0.4359,
+      "step": 48906
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.1275715827941895,
+      "learning_rate": 1.0495206807773178e-06,
+      "loss": 0.6846,
+      "step": 48907
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.779858112335205,
+      "learning_rate": 1.0493590682098315e-06,
+      "loss": 0.5258,
+      "step": 48908
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.8551039695739746,
+      "learning_rate": 1.0491974666275967e-06,
+      "loss": 0.6405,
+      "step": 48909
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 8.30483627319336,
+      "learning_rate": 1.0490358760310649e-06,
+      "loss": 0.8468,
+      "step": 48910
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 6.029536247253418,
+      "learning_rate": 1.0488742964206833e-06,
+      "loss": 0.6748,
+      "step": 48911
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 1.960012435913086,
+      "learning_rate": 1.048712727796903e-06,
+      "loss": 0.5281,
+      "step": 48912
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.3812415599823,
+      "learning_rate": 1.0485511701601713e-06,
+      "loss": 0.6414,
+      "step": 48913
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.2183189392089844,
+      "learning_rate": 1.0483896235109397e-06,
+      "loss": 0.4761,
+      "step": 48914
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.5991322994232178,
+      "learning_rate": 1.0482280878496548e-06,
+      "loss": 0.7327,
+      "step": 48915
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.759380340576172,
+      "learning_rate": 1.048066563176769e-06,
+      "loss": 0.6171,
+      "step": 48916
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.456961154937744,
+      "learning_rate": 1.0479050494927283e-06,
+      "loss": 0.8467,
+      "step": 48917
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.3698689937591553,
+      "learning_rate": 1.0477435467979842e-06,
+      "loss": 0.6112,
+      "step": 48918
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.927501678466797,
+      "learning_rate": 1.0475820550929838e-06,
+      "loss": 0.4601,
+      "step": 48919
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.3349127769470215,
+      "learning_rate": 1.047420574378178e-06,
+      "loss": 0.587,
+      "step": 48920
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 1.8982040882110596,
+      "learning_rate": 1.0472591046540136e-06,
+      "loss": 0.3916,
+      "step": 48921
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.0289194583892822,
+      "learning_rate": 1.0470976459209426e-06,
+      "loss": 0.7149,
+      "step": 48922
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 1.7798287868499756,
+      "learning_rate": 1.0469361981794108e-06,
+      "loss": 0.4263,
+      "step": 48923
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.622459888458252,
+      "learning_rate": 1.0467747614298695e-06,
+      "loss": 0.5634,
+      "step": 48924
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 5.0827813148498535,
+      "learning_rate": 1.046613335672766e-06,
+      "loss": 0.7669,
+      "step": 48925
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.735257625579834,
+      "learning_rate": 1.0464519209085495e-06,
+      "loss": 0.6314,
+      "step": 48926
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.2239575386047363,
+      "learning_rate": 1.0462905171376692e-06,
+      "loss": 0.5964,
+      "step": 48927
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.003755569458008,
+      "learning_rate": 1.0461291243605747e-06,
+      "loss": 0.648,
+      "step": 48928
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.5535199642181396,
+      "learning_rate": 1.0459677425777132e-06,
+      "loss": 0.5769,
+      "step": 48929
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 1.9708168506622314,
+      "learning_rate": 1.0458063717895356e-06,
+      "loss": 0.3125,
+      "step": 48930
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.412006378173828,
+      "learning_rate": 1.0456450119964878e-06,
+      "loss": 0.5881,
+      "step": 48931
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.3093981742858887,
+      "learning_rate": 1.045483663199021e-06,
+      "loss": 0.7179,
+      "step": 48932
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 4.274967193603516,
+      "learning_rate": 1.0453223253975813e-06,
+      "loss": 0.725,
+      "step": 48933
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.547022819519043,
+      "learning_rate": 1.0451609985926204e-06,
+      "loss": 0.7134,
+      "step": 48934
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.7977516651153564,
+      "learning_rate": 1.0449996827845842e-06,
+      "loss": 0.7296,
+      "step": 48935
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.9656732082366943,
+      "learning_rate": 1.044838377973923e-06,
+      "loss": 0.6744,
+      "step": 48936
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.0099945068359375,
+      "learning_rate": 1.0446770841610837e-06,
+      "loss": 0.7881,
+      "step": 48937
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.5302181243896484,
+      "learning_rate": 1.0445158013465167e-06,
+      "loss": 0.5344,
+      "step": 48938
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.14713978767395,
+      "learning_rate": 1.0443545295306685e-06,
+      "loss": 0.5346,
+      "step": 48939
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.8999669551849365,
+      "learning_rate": 1.04419326871399e-06,
+      "loss": 0.6558,
+      "step": 48940
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.3712244033813477,
+      "learning_rate": 1.0440320188969265e-06,
+      "loss": 0.5698,
+      "step": 48941
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.4531145095825195,
+      "learning_rate": 1.0438707800799292e-06,
+      "loss": 0.383,
+      "step": 48942
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.353050947189331,
+      "learning_rate": 1.0437095522634444e-06,
+      "loss": 0.557,
+      "step": 48943
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.1701250076293945,
+      "learning_rate": 1.043548335447921e-06,
+      "loss": 0.5855,
+      "step": 48944
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.5617523193359375,
+      "learning_rate": 1.0433871296338088e-06,
+      "loss": 0.5105,
+      "step": 48945
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.7765581607818604,
+      "learning_rate": 1.0432259348215534e-06,
+      "loss": 0.5484,
+      "step": 48946
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.969188690185547,
+      "learning_rate": 1.043064751011605e-06,
+      "loss": 0.4365,
+      "step": 48947
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 4.293878555297852,
+      "learning_rate": 1.0429035782044122e-06,
+      "loss": 0.7558,
+      "step": 48948
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.1315648555755615,
+      "learning_rate": 1.0427424164004208e-06,
+      "loss": 0.6119,
+      "step": 48949
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.959510326385498,
+      "learning_rate": 1.0425812656000816e-06,
+      "loss": 0.8037,
+      "step": 48950
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.2561659812927246,
+      "learning_rate": 1.0424201258038402e-06,
+      "loss": 0.7388,
+      "step": 48951
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.3866536617279053,
+      "learning_rate": 1.042258997012147e-06,
+      "loss": 0.7428,
+      "step": 48952
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.9793238639831543,
+      "learning_rate": 1.0420978792254482e-06,
+      "loss": 0.564,
+      "step": 48953
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 1.8866878747940063,
+      "learning_rate": 1.041936772444193e-06,
+      "loss": 0.4596,
+      "step": 48954
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 5.42389440536499,
+      "learning_rate": 1.0417756766688286e-06,
+      "loss": 0.6299,
+      "step": 48955
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.365342617034912,
+      "learning_rate": 1.0416145918998023e-06,
+      "loss": 0.7383,
+      "step": 48956
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.1473941802978516,
+      "learning_rate": 1.0414535181375646e-06,
+      "loss": 0.5756,
+      "step": 48957
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.4823451042175293,
+      "learning_rate": 1.0412924553825609e-06,
+      "loss": 0.6941,
+      "step": 48958
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 4.327762126922607,
+      "learning_rate": 1.041131403635241e-06,
+      "loss": 0.6727,
+      "step": 48959
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 4.263260364532471,
+      "learning_rate": 1.0409703628960498e-06,
+      "loss": 0.5993,
+      "step": 48960
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.2086193561553955,
+      "learning_rate": 1.040809333165439e-06,
+      "loss": 0.4907,
+      "step": 48961
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.2100141048431396,
+      "learning_rate": 1.0406483144438522e-06,
+      "loss": 0.4963,
+      "step": 48962
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.204873561859131,
+      "learning_rate": 1.0404873067317411e-06,
+      "loss": 0.4239,
+      "step": 48963
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.2889482975006104,
+      "learning_rate": 1.0403263100295502e-06,
+      "loss": 0.5505,
+      "step": 48964
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.3621320724487305,
+      "learning_rate": 1.0401653243377296e-06,
+      "loss": 0.784,
+      "step": 48965
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.083463191986084,
+      "learning_rate": 1.0400043496567248e-06,
+      "loss": 0.5792,
+      "step": 48966
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.0187935829162598,
+      "learning_rate": 1.0398433859869845e-06,
+      "loss": 0.6702,
+      "step": 48967
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.455667495727539,
+      "learning_rate": 1.0396824333289557e-06,
+      "loss": 0.4326,
+      "step": 48968
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.4011528491973877,
+      "learning_rate": 1.0395214916830881e-06,
+      "loss": 0.5296,
+      "step": 48969
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.5284695625305176,
+      "learning_rate": 1.0393605610498264e-06,
+      "loss": 0.4091,
+      "step": 48970
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.455054521560669,
+      "learning_rate": 1.0391996414296206e-06,
+      "loss": 0.778,
+      "step": 48971
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.2121543884277344,
+      "learning_rate": 1.0390387328229152e-06,
+      "loss": 0.8225,
+      "step": 48972
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.768847703933716,
+      "learning_rate": 1.038877835230161e-06,
+      "loss": 0.5741,
+      "step": 48973
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.705646514892578,
+      "learning_rate": 1.0387169486518022e-06,
+      "loss": 0.6499,
+      "step": 48974
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.6596906185150146,
+      "learning_rate": 1.0385560730882887e-06,
+      "loss": 0.5623,
+      "step": 48975
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.933852434158325,
+      "learning_rate": 1.0383952085400662e-06,
+      "loss": 0.6658,
+      "step": 48976
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.807314157485962,
+      "learning_rate": 1.038234355007583e-06,
+      "loss": 0.7645,
+      "step": 48977
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.3122236728668213,
+      "learning_rate": 1.038073512491285e-06,
+      "loss": 0.8691,
+      "step": 48978
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.261471748352051,
+      "learning_rate": 1.0379126809916218e-06,
+      "loss": 0.5331,
+      "step": 48979
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.1755800247192383,
+      "learning_rate": 1.0377518605090381e-06,
+      "loss": 0.6397,
+      "step": 48980
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.961073160171509,
+      "learning_rate": 1.0375910510439829e-06,
+      "loss": 0.5187,
+      "step": 48981
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.257228374481201,
+      "learning_rate": 1.0374302525969016e-06,
+      "loss": 0.5805,
+      "step": 48982
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.480745792388916,
+      "learning_rate": 1.0372694651682441e-06,
+      "loss": 0.5063,
+      "step": 48983
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.204185962677002,
+      "learning_rate": 1.0371086887584536e-06,
+      "loss": 0.4138,
+      "step": 48984
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.5602481365203857,
+      "learning_rate": 1.0369479233679812e-06,
+      "loss": 0.5683,
+      "step": 48985
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.0528855323791504,
+      "learning_rate": 1.0367871689972708e-06,
+      "loss": 0.4709,
+      "step": 48986
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.077752113342285,
+      "learning_rate": 1.03662642564677e-06,
+      "loss": 0.5601,
+      "step": 48987
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.005582332611084,
+      "learning_rate": 1.036465693316927e-06,
+      "loss": 0.5031,
+      "step": 48988
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 1.7863221168518066,
+      "learning_rate": 1.036304972008189e-06,
+      "loss": 0.6144,
+      "step": 48989
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.7456910610198975,
+      "learning_rate": 1.036144261721001e-06,
+      "loss": 0.7714,
+      "step": 48990
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.8409528732299805,
+      "learning_rate": 1.0359835624558124e-06,
+      "loss": 0.5331,
+      "step": 48991
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.2864696979522705,
+      "learning_rate": 1.0358228742130672e-06,
+      "loss": 0.5568,
+      "step": 48992
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.319498300552368,
+      "learning_rate": 1.0356621969932145e-06,
+      "loss": 0.6041,
+      "step": 48993
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.3607091903686523,
+      "learning_rate": 1.0355015307966992e-06,
+      "loss": 0.4494,
+      "step": 48994
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 1.811964988708496,
+      "learning_rate": 1.0353408756239702e-06,
+      "loss": 0.5538,
+      "step": 48995
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.809298515319824,
+      "learning_rate": 1.0351802314754716e-06,
+      "loss": 0.5378,
+      "step": 48996
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 1.880034327507019,
+      "learning_rate": 1.0350195983516531e-06,
+      "loss": 0.3637,
+      "step": 48997
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.7579028606414795,
+      "learning_rate": 1.0348589762529582e-06,
+      "loss": 0.3941,
+      "step": 48998
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.3526203632354736,
+      "learning_rate": 1.0346983651798364e-06,
+      "loss": 0.6684,
+      "step": 48999
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.6683363914489746,
+      "learning_rate": 1.034537765132732e-06,
+      "loss": 0.6136,
+      "step": 49000
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.309908390045166,
+      "learning_rate": 1.0343771761120936e-06,
+      "loss": 0.6033,
+      "step": 49001
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.263489246368408,
+      "learning_rate": 1.0342165981183655e-06,
+      "loss": 0.5763,
+      "step": 49002
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.8471527099609375,
+      "learning_rate": 1.0340560311519971e-06,
+      "loss": 0.6794,
+      "step": 49003
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.2846903800964355,
+      "learning_rate": 1.033895475213431e-06,
+      "loss": 0.586,
+      "step": 49004
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.517291784286499,
+      "learning_rate": 1.033734930303117e-06,
+      "loss": 0.68,
+      "step": 49005
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.4716904163360596,
+      "learning_rate": 1.0335743964215012e-06,
+      "loss": 0.716,
+      "step": 49006
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.1471943855285645,
+      "learning_rate": 1.0334138735690274e-06,
+      "loss": 0.5649,
+      "step": 49007
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.066340684890747,
+      "learning_rate": 1.0332533617461437e-06,
+      "loss": 0.4412,
+      "step": 49008
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.431736707687378,
+      "learning_rate": 1.0330928609532981e-06,
+      "loss": 0.4678,
+      "step": 49009
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.1077678203582764,
+      "learning_rate": 1.0329323711909338e-06,
+      "loss": 0.5718,
+      "step": 49010
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.933990716934204,
+      "learning_rate": 1.0327718924595003e-06,
+      "loss": 0.6934,
+      "step": 49011
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.787630558013916,
+      "learning_rate": 1.0326114247594398e-06,
+      "loss": 0.8997,
+      "step": 49012
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.155573844909668,
+      "learning_rate": 1.0324509680912027e-06,
+      "loss": 0.5561,
+      "step": 49013
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 4.037215232849121,
+      "learning_rate": 1.0322905224552315e-06,
+      "loss": 0.554,
+      "step": 49014
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.6075031757354736,
+      "learning_rate": 1.0321300878519758e-06,
+      "loss": 0.7223,
+      "step": 49015
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.4639008045196533,
+      "learning_rate": 1.0319696642818783e-06,
+      "loss": 0.8061,
+      "step": 49016
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.0400919914245605,
+      "learning_rate": 1.0318092517453865e-06,
+      "loss": 0.4484,
+      "step": 49017
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.471935749053955,
+      "learning_rate": 1.0316488502429479e-06,
+      "loss": 0.4323,
+      "step": 49018
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.5299084186553955,
+      "learning_rate": 1.0314884597750063e-06,
+      "loss": 0.4509,
+      "step": 49019
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.924595355987549,
+      "learning_rate": 1.03132808034201e-06,
+      "loss": 0.3784,
+      "step": 49020
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 1.6473451852798462,
+      "learning_rate": 1.031167711944402e-06,
+      "loss": 0.3109,
+      "step": 49021
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.4983339309692383,
+      "learning_rate": 1.0310073545826316e-06,
+      "loss": 0.5657,
+      "step": 49022
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.534031629562378,
+      "learning_rate": 1.030847008257141e-06,
+      "loss": 0.5624,
+      "step": 49023
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.11775803565979,
+      "learning_rate": 1.0306866729683796e-06,
+      "loss": 0.5045,
+      "step": 49024
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.6912686824798584,
+      "learning_rate": 1.0305263487167904e-06,
+      "loss": 0.8443,
+      "step": 49025
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 4.444975852966309,
+      "learning_rate": 1.0303660355028217e-06,
+      "loss": 0.3952,
+      "step": 49026
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.7588627338409424,
+      "learning_rate": 1.0302057333269166e-06,
+      "loss": 0.5178,
+      "step": 49027
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 4.581035614013672,
+      "learning_rate": 1.0300454421895223e-06,
+      "loss": 0.5785,
+      "step": 49028
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.260390281677246,
+      "learning_rate": 1.0298851620910843e-06,
+      "loss": 0.5823,
+      "step": 49029
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.415473222732544,
+      "learning_rate": 1.0297248930320503e-06,
+      "loss": 0.6633,
+      "step": 49030
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.144566059112549,
+      "learning_rate": 1.0295646350128624e-06,
+      "loss": 0.4702,
+      "step": 49031
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.4675652980804443,
+      "learning_rate": 1.0294043880339689e-06,
+      "loss": 0.7485,
+      "step": 49032
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.802285671234131,
+      "learning_rate": 1.0292441520958135e-06,
+      "loss": 0.6698,
+      "step": 49033
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.192460775375366,
+      "learning_rate": 1.0290839271988434e-06,
+      "loss": 0.4068,
+      "step": 49034
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.6807963848114014,
+      "learning_rate": 1.0289237133435025e-06,
+      "loss": 0.6293,
+      "step": 49035
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.5964536666870117,
+      "learning_rate": 1.0287635105302386e-06,
+      "loss": 0.8178,
+      "step": 49036
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.86263108253479,
+      "learning_rate": 1.028603318759494e-06,
+      "loss": 0.3312,
+      "step": 49037
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.7176363468170166,
+      "learning_rate": 1.0284431380317173e-06,
+      "loss": 0.5587,
+      "step": 49038
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.7584424018859863,
+      "learning_rate": 1.028282968347351e-06,
+      "loss": 0.6958,
+      "step": 49039
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.6228322982788086,
+      "learning_rate": 1.0281228097068435e-06,
+      "loss": 0.5858,
+      "step": 49040
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 4.938702583312988,
+      "learning_rate": 1.0279626621106376e-06,
+      "loss": 0.6386,
+      "step": 49041
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 4.771124839782715,
+      "learning_rate": 1.0278025255591805e-06,
+      "loss": 0.4664,
+      "step": 49042
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.3018980026245117,
+      "learning_rate": 1.027642400052915e-06,
+      "loss": 0.4071,
+      "step": 49043
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.25150203704834,
+      "learning_rate": 1.0274822855922894e-06,
+      "loss": 0.4093,
+      "step": 49044
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.4674715995788574,
+      "learning_rate": 1.0273221821777464e-06,
+      "loss": 0.4752,
+      "step": 49045
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.461221218109131,
+      "learning_rate": 1.027162089809733e-06,
+      "loss": 0.5585,
+      "step": 49046
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.9178338050842285,
+      "learning_rate": 1.0270020084886929e-06,
+      "loss": 0.4608,
+      "step": 49047
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.027871608734131,
+      "learning_rate": 1.0268419382150713e-06,
+      "loss": 0.6582,
+      "step": 49048
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.789158821105957,
+      "learning_rate": 1.0266818789893141e-06,
+      "loss": 0.559,
+      "step": 49049
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.6579339504241943,
+      "learning_rate": 1.0265218308118673e-06,
+      "loss": 0.6034,
+      "step": 49050
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.6563494205474854,
+      "learning_rate": 1.0263617936831738e-06,
+      "loss": 0.5263,
+      "step": 49051
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 1.7467646598815918,
+      "learning_rate": 1.0262017676036806e-06,
+      "loss": 0.5043,
+      "step": 49052
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.421283006668091,
+      "learning_rate": 1.0260417525738308e-06,
+      "loss": 0.5562,
+      "step": 49053
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.845670700073242,
+      "learning_rate": 1.0258817485940709e-06,
+      "loss": 0.4922,
+      "step": 49054
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.14719557762146,
+      "learning_rate": 1.025721755664844e-06,
+      "loss": 0.5495,
+      "step": 49055
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.1702895164489746,
+      "learning_rate": 1.025561773786598e-06,
+      "loss": 0.5496,
+      "step": 49056
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.8650102615356445,
+      "learning_rate": 1.025401802959774e-06,
+      "loss": 0.6499,
+      "step": 49057
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.431985378265381,
+      "learning_rate": 1.0252418431848198e-06,
+      "loss": 0.5959,
+      "step": 49058
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 4.000428199768066,
+      "learning_rate": 1.0250818944621781e-06,
+      "loss": 0.4095,
+      "step": 49059
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 4.106083869934082,
+      "learning_rate": 1.0249219567922957e-06,
+      "loss": 0.4392,
+      "step": 49060
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 1.8905348777770996,
+      "learning_rate": 1.024762030175615e-06,
+      "loss": 0.428,
+      "step": 49061
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.4694125652313232,
+      "learning_rate": 1.0246021146125835e-06,
+      "loss": 0.5696,
+      "step": 49062
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 4.5688862800598145,
+      "learning_rate": 1.0244422101036428e-06,
+      "loss": 0.5479,
+      "step": 49063
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 4.2277116775512695,
+      "learning_rate": 1.0242823166492405e-06,
+      "loss": 0.7768,
+      "step": 49064
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.6215760707855225,
+      "learning_rate": 1.0241224342498185e-06,
+      "loss": 0.5615,
+      "step": 49065
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.121708869934082,
+      "learning_rate": 1.0239625629058236e-06,
+      "loss": 0.3214,
+      "step": 49066
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.0619802474975586,
+      "learning_rate": 1.0238027026176978e-06,
+      "loss": 0.4014,
+      "step": 49067
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.3046038150787354,
+      "learning_rate": 1.0236428533858878e-06,
+      "loss": 0.4317,
+      "step": 49068
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.189256429672241,
+      "learning_rate": 1.0234830152108372e-06,
+      "loss": 0.5082,
+      "step": 49069
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.437074661254883,
+      "learning_rate": 1.0233231880929918e-06,
+      "loss": 0.4809,
+      "step": 49070
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.512171506881714,
+      "learning_rate": 1.0231633720327938e-06,
+      "loss": 0.5698,
+      "step": 49071
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.3614554405212402,
+      "learning_rate": 1.0230035670306898e-06,
+      "loss": 0.6661,
+      "step": 49072
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.281555414199829,
+      "learning_rate": 1.022843773087122e-06,
+      "loss": 0.677,
+      "step": 49073
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 6.5908894538879395,
+      "learning_rate": 1.0226839902025371e-06,
+      "loss": 0.5671,
+      "step": 49074
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.1175484657287598,
+      "learning_rate": 1.0225242183773764e-06,
+      "loss": 0.6912,
+      "step": 49075
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.147197961807251,
+      "learning_rate": 1.022364457612088e-06,
+      "loss": 0.7405,
+      "step": 49076
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.554358959197998,
+      "learning_rate": 1.0222047079071119e-06,
+      "loss": 0.4991,
+      "step": 49077
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.6288535594940186,
+      "learning_rate": 1.0220449692628958e-06,
+      "loss": 0.4729,
+      "step": 49078
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.2743005752563477,
+      "learning_rate": 1.0218852416798814e-06,
+      "loss": 0.4885,
+      "step": 49079
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.8171699047088623,
+      "learning_rate": 1.021725525158514e-06,
+      "loss": 0.5817,
+      "step": 49080
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 1.9402213096618652,
+      "learning_rate": 1.021565819699239e-06,
+      "loss": 0.473,
+      "step": 49081
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.7465648651123047,
+      "learning_rate": 1.021406125302497e-06,
+      "loss": 0.744,
+      "step": 49082
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.6935062408447266,
+      "learning_rate": 1.0212464419687362e-06,
+      "loss": 0.5719,
+      "step": 49083
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.935147523880005,
+      "learning_rate": 1.0210867696983973e-06,
+      "loss": 0.5816,
+      "step": 49084
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.5493946075439453,
+      "learning_rate": 1.0209271084919264e-06,
+      "loss": 0.6158,
+      "step": 49085
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.6202783584594727,
+      "learning_rate": 1.0207674583497657e-06,
+      "loss": 0.5222,
+      "step": 49086
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.205782651901245,
+      "learning_rate": 1.0206078192723612e-06,
+      "loss": 0.5063,
+      "step": 49087
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.8895249366760254,
+      "learning_rate": 1.0204481912601543e-06,
+      "loss": 0.6192,
+      "step": 49088
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.8346457481384277,
+      "learning_rate": 1.0202885743135899e-06,
+      "loss": 0.613,
+      "step": 49089
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.858372926712036,
+      "learning_rate": 1.0201289684331128e-06,
+      "loss": 0.5151,
+      "step": 49090
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 4.849925994873047,
+      "learning_rate": 1.0199693736191668e-06,
+      "loss": 0.61,
+      "step": 49091
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.423307418823242,
+      "learning_rate": 1.0198097898721943e-06,
+      "loss": 0.6315,
+      "step": 49092
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.4471821784973145,
+      "learning_rate": 1.0196502171926408e-06,
+      "loss": 0.491,
+      "step": 49093
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.850796937942505,
+      "learning_rate": 1.0194906555809475e-06,
+      "loss": 0.5312,
+      "step": 49094
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.5767316818237305,
+      "learning_rate": 1.019331105037561e-06,
+      "loss": 0.6954,
+      "step": 49095
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.5582523345947266,
+      "learning_rate": 1.0191715655629224e-06,
+      "loss": 0.641,
+      "step": 49096
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 16.59507942199707,
+      "learning_rate": 1.0190120371574774e-06,
+      "loss": 0.5124,
+      "step": 49097
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.098273992538452,
+      "learning_rate": 1.0188525198216676e-06,
+      "loss": 0.6425,
+      "step": 49098
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 4.0291314125061035,
+      "learning_rate": 1.0186930135559387e-06,
+      "loss": 0.5728,
+      "step": 49099
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.1989998817443848,
+      "learning_rate": 1.0185335183607315e-06,
+      "loss": 0.6057,
+      "step": 49100
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 5.838716506958008,
+      "learning_rate": 1.0183740342364928e-06,
+      "loss": 0.7045,
+      "step": 49101
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.163698434829712,
+      "learning_rate": 1.018214561183663e-06,
+      "loss": 0.3843,
+      "step": 49102
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.241689920425415,
+      "learning_rate": 1.0180550992026883e-06,
+      "loss": 0.4474,
+      "step": 49103
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.8192920684814453,
+      "learning_rate": 1.0178956482940095e-06,
+      "loss": 0.4817,
+      "step": 49104
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.3770639896392822,
+      "learning_rate": 1.017736208458072e-06,
+      "loss": 0.4509,
+      "step": 49105
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.3247976303100586,
+      "learning_rate": 1.0175767796953172e-06,
+      "loss": 0.6218,
+      "step": 49106
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.45950984954834,
+      "learning_rate": 1.017417362006191e-06,
+      "loss": 0.7056,
+      "step": 49107
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.5777511596679688,
+      "learning_rate": 1.0172579553911338e-06,
+      "loss": 0.4566,
+      "step": 49108
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.580915927886963,
+      "learning_rate": 1.01709855985059e-06,
+      "loss": 0.4987,
+      "step": 49109
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.6697001457214355,
+      "learning_rate": 1.0169391753850032e-06,
+      "loss": 0.5427,
+      "step": 49110
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.2516305446624756,
+      "learning_rate": 1.0167798019948182e-06,
+      "loss": 0.602,
+      "step": 49111
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.499980926513672,
+      "learning_rate": 1.0166204396804746e-06,
+      "loss": 0.5269,
+      "step": 49112
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.93778133392334,
+      "learning_rate": 1.0164610884424193e-06,
+      "loss": 0.518,
+      "step": 49113
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.763338088989258,
+      "learning_rate": 1.0163017482810916e-06,
+      "loss": 0.8928,
+      "step": 49114
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.9159090518951416,
+      "learning_rate": 1.0161424191969376e-06,
+      "loss": 0.4177,
+      "step": 49115
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.550936222076416,
+      "learning_rate": 1.015983101190398e-06,
+      "loss": 0.4455,
+      "step": 49116
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.208674669265747,
+      "learning_rate": 1.0158237942619187e-06,
+      "loss": 0.4825,
+      "step": 49117
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.949669599533081,
+      "learning_rate": 1.015664498411939e-06,
+      "loss": 0.5774,
+      "step": 49118
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.3002371788024902,
+      "learning_rate": 1.0155052136409054e-06,
+      "loss": 0.4228,
+      "step": 49119
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.6128036975860596,
+      "learning_rate": 1.0153459399492577e-06,
+      "loss": 0.7831,
+      "step": 49120
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.7919929027557373,
+      "learning_rate": 1.0151866773374413e-06,
+      "loss": 0.5508,
+      "step": 49121
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.821025848388672,
+      "learning_rate": 1.0150274258058973e-06,
+      "loss": 0.5734,
+      "step": 49122
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.00976824760437,
+      "learning_rate": 1.0148681853550702e-06,
+      "loss": 0.5197,
+      "step": 49123
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.513880491256714,
+      "learning_rate": 1.0147089559854006e-06,
+      "loss": 0.5926,
+      "step": 49124
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.660022258758545,
+      "learning_rate": 1.014549737697334e-06,
+      "loss": 0.6713,
+      "step": 49125
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.900872230529785,
+      "learning_rate": 1.01439053049131e-06,
+      "loss": 0.6953,
+      "step": 49126
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.7663254737854004,
+      "learning_rate": 1.014231334367774e-06,
+      "loss": 0.7518,
+      "step": 49127
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.2046620845794678,
+      "learning_rate": 1.0140721493271665e-06,
+      "loss": 0.624,
+      "step": 49128
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.8689773082733154,
+      "learning_rate": 1.0139129753699312e-06,
+      "loss": 0.4985,
+      "step": 49129
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.7215700149536133,
+      "learning_rate": 1.0137538124965107e-06,
+      "loss": 0.5725,
+      "step": 49130
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.3302764892578125,
+      "learning_rate": 1.0135946607073488e-06,
+      "loss": 0.5903,
+      "step": 49131
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.1197001934051514,
+      "learning_rate": 1.0134355200028856e-06,
+      "loss": 0.5104,
+      "step": 49132
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 4.446012020111084,
+      "learning_rate": 1.0132763903835658e-06,
+      "loss": 0.6976,
+      "step": 49133
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.7494142055511475,
+      "learning_rate": 1.01311727184983e-06,
+      "loss": 0.6859,
+      "step": 49134
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.2732255458831787,
+      "learning_rate": 1.0129581644021225e-06,
+      "loss": 0.2945,
+      "step": 49135
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.340944766998291,
+      "learning_rate": 1.0127990680408838e-06,
+      "loss": 0.5822,
+      "step": 49136
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.188347578048706,
+      "learning_rate": 1.0126399827665585e-06,
+      "loss": 0.5636,
+      "step": 49137
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.5584216117858887,
+      "learning_rate": 1.0124809085795861e-06,
+      "loss": 0.6703,
+      "step": 49138
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.6921122074127197,
+      "learning_rate": 1.0123218454804123e-06,
+      "loss": 0.7198,
+      "step": 49139
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.072728157043457,
+      "learning_rate": 1.0121627934694761e-06,
+      "loss": 0.3746,
+      "step": 49140
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.0781538486480713,
+      "learning_rate": 1.012003752547221e-06,
+      "loss": 0.3886,
+      "step": 49141
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.4418728351593018,
+      "learning_rate": 1.0118447227140916e-06,
+      "loss": 0.362,
+      "step": 49142
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.584444284439087,
+      "learning_rate": 1.0116857039705258e-06,
+      "loss": 0.4827,
+      "step": 49143
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.1969847679138184,
+      "learning_rate": 1.0115266963169701e-06,
+      "loss": 0.506,
+      "step": 49144
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.0980098247528076,
+      "learning_rate": 1.0113676997538625e-06,
+      "loss": 0.5168,
+      "step": 49145
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.941631317138672,
+      "learning_rate": 1.011208714281649e-06,
+      "loss": 0.793,
+      "step": 49146
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.891087293624878,
+      "learning_rate": 1.0110497399007686e-06,
+      "loss": 0.6518,
+      "step": 49147
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.002599000930786,
+      "learning_rate": 1.0108907766116655e-06,
+      "loss": 0.4638,
+      "step": 49148
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.9613397121429443,
+      "learning_rate": 1.0107318244147795e-06,
+      "loss": 0.6364,
+      "step": 49149
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.1615498065948486,
+      "learning_rate": 1.0105728833105537e-06,
+      "loss": 0.5968,
+      "step": 49150
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.3770158290863037,
+      "learning_rate": 1.0104139532994323e-06,
+      "loss": 0.4507,
+      "step": 49151
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.0473618507385254,
+      "learning_rate": 1.0102550343818535e-06,
+      "loss": 0.582,
+      "step": 49152
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.472202777862549,
+      "learning_rate": 1.0100961265582604e-06,
+      "loss": 0.7144,
+      "step": 49153
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.4812264442443848,
+      "learning_rate": 1.0099372298290971e-06,
+      "loss": 0.7904,
+      "step": 49154
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 1.6335692405700684,
+      "learning_rate": 1.0097783441948022e-06,
+      "loss": 0.5369,
+      "step": 49155
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.635641574859619,
+      "learning_rate": 1.0096194696558204e-06,
+      "loss": 0.5066,
+      "step": 49156
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.5860774517059326,
+      "learning_rate": 1.0094606062125905e-06,
+      "loss": 0.5309,
+      "step": 49157
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.7887487411499023,
+      "learning_rate": 1.009301753865557e-06,
+      "loss": 0.6948,
+      "step": 49158
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.807847499847412,
+      "learning_rate": 1.0091429126151598e-06,
+      "loss": 0.6098,
+      "step": 49159
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.3204197883605957,
+      "learning_rate": 1.0089840824618418e-06,
+      "loss": 0.7068,
+      "step": 49160
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.3702142238616943,
+      "learning_rate": 1.0088252634060424e-06,
+      "loss": 0.4108,
+      "step": 49161
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 1.9845467805862427,
+      "learning_rate": 1.0086664554482068e-06,
+      "loss": 0.3979,
+      "step": 49162
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 4.653339862823486,
+      "learning_rate": 1.0085076585887732e-06,
+      "loss": 0.8518,
+      "step": 49163
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.246830463409424,
+      "learning_rate": 1.0083488728281854e-06,
+      "loss": 0.788,
+      "step": 49164
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.2917733192443848,
+      "learning_rate": 1.0081900981668829e-06,
+      "loss": 0.6676,
+      "step": 49165
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.2796287536621094,
+      "learning_rate": 1.00803133460531e-06,
+      "loss": 0.5178,
+      "step": 49166
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.554230213165283,
+      "learning_rate": 1.007872582143905e-06,
+      "loss": 0.4529,
+      "step": 49167
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 4.689078330993652,
+      "learning_rate": 1.0077138407831122e-06,
+      "loss": 0.5367,
+      "step": 49168
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.3532917499542236,
+      "learning_rate": 1.0075551105233705e-06,
+      "loss": 0.4702,
+      "step": 49169
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.2778804302215576,
+      "learning_rate": 1.0073963913651225e-06,
+      "loss": 0.5887,
+      "step": 49170
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.6113786697387695,
+      "learning_rate": 1.0072376833088088e-06,
+      "loss": 0.4634,
+      "step": 49171
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.446974277496338,
+      "learning_rate": 1.0070789863548736e-06,
+      "loss": 0.5525,
+      "step": 49172
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.246504306793213,
+      "learning_rate": 1.0069203005037542e-06,
+      "loss": 0.5522,
+      "step": 49173
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 4.405975341796875,
+      "learning_rate": 1.0067616257558948e-06,
+      "loss": 0.8598,
+      "step": 49174
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.2522008419036865,
+      "learning_rate": 1.006602962111734e-06,
+      "loss": 0.4968,
+      "step": 49175
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.3507497310638428,
+      "learning_rate": 1.0064443095717164e-06,
+      "loss": 0.5728,
+      "step": 49176
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.080514430999756,
+      "learning_rate": 1.0062856681362792e-06,
+      "loss": 0.6234,
+      "step": 49177
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 1.9675465822219849,
+      "learning_rate": 1.0061270378058669e-06,
+      "loss": 0.4049,
+      "step": 49178
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.4804623126983643,
+      "learning_rate": 1.0059684185809183e-06,
+      "loss": 0.4211,
+      "step": 49179
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.2024261951446533,
+      "learning_rate": 1.0058098104618758e-06,
+      "loss": 0.6051,
+      "step": 49180
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.6986515522003174,
+      "learning_rate": 1.0056512134491792e-06,
+      "loss": 0.6903,
+      "step": 49181
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.7301957607269287,
+      "learning_rate": 1.0054926275432715e-06,
+      "loss": 0.5133,
+      "step": 49182
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.5846569538116455,
+      "learning_rate": 1.0053340527445909e-06,
+      "loss": 0.7247,
+      "step": 49183
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.999175786972046,
+      "learning_rate": 1.0051754890535819e-06,
+      "loss": 0.6609,
+      "step": 49184
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.6524336338043213,
+      "learning_rate": 1.0050169364706813e-06,
+      "loss": 0.8234,
+      "step": 49185
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.0296425819396973,
+      "learning_rate": 1.0048583949963337e-06,
+      "loss": 0.4982,
+      "step": 49186
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.904881477355957,
+      "learning_rate": 1.0046998646309769e-06,
+      "loss": 0.7348,
+      "step": 49187
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.33076810836792,
+      "learning_rate": 1.0045413453750541e-06,
+      "loss": 0.7152,
+      "step": 49188
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.1445486545562744,
+      "learning_rate": 1.0043828372290042e-06,
+      "loss": 0.7525,
+      "step": 49189
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.740783214569092,
+      "learning_rate": 1.0042243401932688e-06,
+      "loss": 0.5396,
+      "step": 49190
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.559861421585083,
+      "learning_rate": 1.0040658542682884e-06,
+      "loss": 0.5632,
+      "step": 49191
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.509305477142334,
+      "learning_rate": 1.0039073794545056e-06,
+      "loss": 0.5697,
+      "step": 49192
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.7116153240203857,
+      "learning_rate": 1.0037489157523579e-06,
+      "loss": 0.6124,
+      "step": 49193
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.9790070056915283,
+      "learning_rate": 1.003590463162289e-06,
+      "loss": 0.7453,
+      "step": 49194
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 5.665945053100586,
+      "learning_rate": 1.003432021684736e-06,
+      "loss": 0.6577,
+      "step": 49195
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.205477714538574,
+      "learning_rate": 1.003273591320143e-06,
+      "loss": 0.7206,
+      "step": 49196
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.387336492538452,
+      "learning_rate": 1.003115172068948e-06,
+      "loss": 0.3426,
+      "step": 49197
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.473092794418335,
+      "learning_rate": 1.0029567639315935e-06,
+      "loss": 0.6052,
+      "step": 49198
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.1633460521698,
+      "learning_rate": 1.0027983669085173e-06,
+      "loss": 0.7466,
+      "step": 49199
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.4765474796295166,
+      "learning_rate": 1.002639981000163e-06,
+      "loss": 0.4456,
+      "step": 49200
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.1103711128234863,
+      "learning_rate": 1.0024816062069686e-06,
+      "loss": 0.6053,
+      "step": 49201
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.2104570865631104,
+      "learning_rate": 1.002323242529375e-06,
+      "loss": 0.8035,
+      "step": 49202
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.2415378093719482,
+      "learning_rate": 1.0021648899678238e-06,
+      "loss": 0.457,
+      "step": 49203
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.210224151611328,
+      "learning_rate": 1.0020065485227537e-06,
+      "loss": 0.465,
+      "step": 49204
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 1.9557814598083496,
+      "learning_rate": 1.0018482181946065e-06,
+      "loss": 0.6096,
+      "step": 49205
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.5247104167938232,
+      "learning_rate": 1.0016898989838208e-06,
+      "loss": 0.665,
+      "step": 49206
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.3047871589660645,
+      "learning_rate": 1.001531590890839e-06,
+      "loss": 0.3691,
+      "step": 49207
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.7407519817352295,
+      "learning_rate": 1.0013732939160986e-06,
+      "loss": 0.5912,
+      "step": 49208
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.038957357406616,
+      "learning_rate": 1.001215008060042e-06,
+      "loss": 0.4323,
+      "step": 49209
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.4777863025665283,
+      "learning_rate": 1.0010567333231074e-06,
+      "loss": 0.5726,
+      "step": 49210
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.523545980453491,
+      "learning_rate": 1.0008984697057367e-06,
+      "loss": 0.2949,
+      "step": 49211
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.805311441421509,
+      "learning_rate": 1.0007402172083696e-06,
+      "loss": 0.7966,
+      "step": 49212
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 4.59672737121582,
+      "learning_rate": 1.000581975831445e-06,
+      "loss": 0.6604,
+      "step": 49213
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.2326695919036865,
+      "learning_rate": 1.0004237455754035e-06,
+      "loss": 0.4278,
+      "step": 49214
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 1.9910894632339478,
+      "learning_rate": 1.000265526440687e-06,
+      "loss": 0.3628,
+      "step": 49215
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.827362537384033,
+      "learning_rate": 1.0001073184277322e-06,
+      "loss": 0.7518,
+      "step": 49216
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.539144277572632,
+      "learning_rate": 9.999491215369816e-07,
+      "loss": 0.4962,
+      "step": 49217
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.712437629699707,
+      "learning_rate": 9.99790935768873e-07,
+      "loss": 0.9151,
+      "step": 49218
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.6483049392700195,
+      "learning_rate": 9.996327611238481e-07,
+      "loss": 0.455,
+      "step": 49219
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.7230257987976074,
+      "learning_rate": 9.99474597602345e-07,
+      "loss": 0.4623,
+      "step": 49220
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.53340482711792,
+      "learning_rate": 9.99316445204806e-07,
+      "loss": 0.6503,
+      "step": 49221
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.9510979652404785,
+      "learning_rate": 9.991583039316678e-07,
+      "loss": 0.5423,
+      "step": 49222
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.1163604259490967,
+      "learning_rate": 9.990001737833727e-07,
+      "loss": 0.7116,
+      "step": 49223
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.4291672706604004,
+      "learning_rate": 9.988420547603583e-07,
+      "loss": 0.4744,
+      "step": 49224
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.0292444229125977,
+      "learning_rate": 9.98683946863066e-07,
+      "loss": 0.492,
+      "step": 49225
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 4.5227203369140625,
+      "learning_rate": 9.985258500919342e-07,
+      "loss": 0.7049,
+      "step": 49226
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.5609657764434814,
+      "learning_rate": 9.983677644474039e-07,
+      "loss": 0.5917,
+      "step": 49227
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.209101438522339,
+      "learning_rate": 9.982096899299127e-07,
+      "loss": 0.5226,
+      "step": 49228
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.6542694568634033,
+      "learning_rate": 9.98051626539902e-07,
+      "loss": 0.5007,
+      "step": 49229
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.7446539402008057,
+      "learning_rate": 9.9789357427781e-07,
+      "loss": 0.6363,
+      "step": 49230
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 9.893770217895508,
+      "learning_rate": 9.977355331440763e-07,
+      "loss": 0.5834,
+      "step": 49231
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.7942051887512207,
+      "learning_rate": 9.97577503139141e-07,
+      "loss": 0.6614,
+      "step": 49232
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.1789231300354004,
+      "learning_rate": 9.974194842634443e-07,
+      "loss": 0.8289,
+      "step": 49233
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 5.350790977478027,
+      "learning_rate": 9.972614765174238e-07,
+      "loss": 0.5767,
+      "step": 49234
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.3719191551208496,
+      "learning_rate": 9.971034799015206e-07,
+      "loss": 0.593,
+      "step": 49235
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.7578322887420654,
+      "learning_rate": 9.969454944161716e-07,
+      "loss": 0.5985,
+      "step": 49236
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.3720057010650635,
+      "learning_rate": 9.967875200618194e-07,
+      "loss": 0.4679,
+      "step": 49237
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.5008385181427,
+      "learning_rate": 9.966295568389001e-07,
+      "loss": 0.5688,
+      "step": 49238
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.5118837356567383,
+      "learning_rate": 9.964716047478556e-07,
+      "loss": 0.7509,
+      "step": 49239
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.6499295234680176,
+      "learning_rate": 9.963136637891224e-07,
+      "loss": 0.6112,
+      "step": 49240
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.325979471206665,
+      "learning_rate": 9.961557339631422e-07,
+      "loss": 0.3814,
+      "step": 49241
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.303253650665283,
+      "learning_rate": 9.95997815270352e-07,
+      "loss": 0.7421,
+      "step": 49242
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.7057437896728516,
+      "learning_rate": 9.95839907711193e-07,
+      "loss": 0.7401,
+      "step": 49243
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.84419322013855,
+      "learning_rate": 9.95682011286102e-07,
+      "loss": 0.7307,
+      "step": 49244
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.6175596714019775,
+      "learning_rate": 9.955241259955212e-07,
+      "loss": 0.5793,
+      "step": 49245
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.186501979827881,
+      "learning_rate": 9.953662518398854e-07,
+      "loss": 0.5701,
+      "step": 49246
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 4.814248085021973,
+      "learning_rate": 9.95208388819638e-07,
+      "loss": 0.5809,
+      "step": 49247
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.4136605262756348,
+      "learning_rate": 9.950505369352143e-07,
+      "loss": 0.722,
+      "step": 49248
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.8571062088012695,
+      "learning_rate": 9.948926961870558e-07,
+      "loss": 0.5607,
+      "step": 49249
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.562512159347534,
+      "learning_rate": 9.94734866575599e-07,
+      "loss": 0.5186,
+      "step": 49250
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.9954299926757812,
+      "learning_rate": 9.945770481012845e-07,
+      "loss": 0.5804,
+      "step": 49251
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.2921459674835205,
+      "learning_rate": 9.944192407645502e-07,
+      "loss": 0.6479,
+      "step": 49252
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.3692195415496826,
+      "learning_rate": 9.942614445658378e-07,
+      "loss": 0.5062,
+      "step": 49253
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.8435518741607666,
+      "learning_rate": 9.941036595055814e-07,
+      "loss": 0.553,
+      "step": 49254
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.746737003326416,
+      "learning_rate": 9.939458855842239e-07,
+      "loss": 0.6199,
+      "step": 49255
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.319756031036377,
+      "learning_rate": 9.937881228022006e-07,
+      "loss": 0.5237,
+      "step": 49256
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.622527837753296,
+      "learning_rate": 9.936303711599531e-07,
+      "loss": 0.4891,
+      "step": 49257
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.1165874004364014,
+      "learning_rate": 9.93472630657918e-07,
+      "loss": 0.4775,
+      "step": 49258
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.6173524856567383,
+      "learning_rate": 9.933149012965355e-07,
+      "loss": 0.7378,
+      "step": 49259
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.4528648853302,
+      "learning_rate": 9.931571830762422e-07,
+      "loss": 0.4669,
+      "step": 49260
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.1451773643493652,
+      "learning_rate": 9.929994759974787e-07,
+      "loss": 0.6786,
+      "step": 49261
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.3672146797180176,
+      "learning_rate": 9.928417800606816e-07,
+      "loss": 0.4889,
+      "step": 49262
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.101524591445923,
+      "learning_rate": 9.926840952662913e-07,
+      "loss": 0.6868,
+      "step": 49263
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.499382495880127,
+      "learning_rate": 9.925264216147446e-07,
+      "loss": 0.3816,
+      "step": 49264
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.397594690322876,
+      "learning_rate": 9.923687591064802e-07,
+      "loss": 0.6961,
+      "step": 49265
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.142782688140869,
+      "learning_rate": 9.922111077419388e-07,
+      "loss": 0.5248,
+      "step": 49266
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 1.9494199752807617,
+      "learning_rate": 9.920534675215553e-07,
+      "loss": 0.3805,
+      "step": 49267
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.1305603981018066,
+      "learning_rate": 9.918958384457711e-07,
+      "loss": 0.5002,
+      "step": 49268
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.0090034008026123,
+      "learning_rate": 9.917382205150216e-07,
+      "loss": 0.478,
+      "step": 49269
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.048764705657959,
+      "learning_rate": 9.915806137297484e-07,
+      "loss": 0.4681,
+      "step": 49270
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.3435261249542236,
+      "learning_rate": 9.91423018090386e-07,
+      "loss": 0.5486,
+      "step": 49271
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.5416178703308105,
+      "learning_rate": 9.912654335973749e-07,
+      "loss": 0.3624,
+      "step": 49272
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.654681444168091,
+      "learning_rate": 9.911078602511542e-07,
+      "loss": 0.6867,
+      "step": 49273
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.2126760482788086,
+      "learning_rate": 9.909502980521597e-07,
+      "loss": 0.4617,
+      "step": 49274
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.782564401626587,
+      "learning_rate": 9.9079274700083e-07,
+      "loss": 0.9737,
+      "step": 49275
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.358572006225586,
+      "learning_rate": 9.906352070976055e-07,
+      "loss": 0.7517,
+      "step": 49276
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.5710835456848145,
+      "learning_rate": 9.904776783429209e-07,
+      "loss": 0.5793,
+      "step": 49277
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 4.486443519592285,
+      "learning_rate": 9.903201607372176e-07,
+      "loss": 0.4792,
+      "step": 49278
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.1350300312042236,
+      "learning_rate": 9.901626542809307e-07,
+      "loss": 0.7853,
+      "step": 49279
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 4.059205532073975,
+      "learning_rate": 9.900051589745003e-07,
+      "loss": 0.6689,
+      "step": 49280
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.626811981201172,
+      "learning_rate": 9.898476748183617e-07,
+      "loss": 0.6448,
+      "step": 49281
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.144482374191284,
+      "learning_rate": 9.896902018129561e-07,
+      "loss": 0.5467,
+      "step": 49282
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.206563949584961,
+      "learning_rate": 9.89532739958719e-07,
+      "loss": 0.502,
+      "step": 49283
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.1111207008361816,
+      "learning_rate": 9.893752892560899e-07,
+      "loss": 0.8358,
+      "step": 49284
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 1.926931381225586,
+      "learning_rate": 9.892178497055037e-07,
+      "loss": 0.6097,
+      "step": 49285
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.483917713165283,
+      "learning_rate": 9.890604213074023e-07,
+      "loss": 0.5718,
+      "step": 49286
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 1.9721673727035522,
+      "learning_rate": 9.8890300406222e-07,
+      "loss": 0.5372,
+      "step": 49287
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.7382357120513916,
+      "learning_rate": 9.88745597970397e-07,
+      "loss": 0.659,
+      "step": 49288
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.18595027923584,
+      "learning_rate": 9.885882030323685e-07,
+      "loss": 0.6324,
+      "step": 49289
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.537140369415283,
+      "learning_rate": 9.884308192485754e-07,
+      "loss": 0.394,
+      "step": 49290
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.224529981613159,
+      "learning_rate": 9.88273446619451e-07,
+      "loss": 0.489,
+      "step": 49291
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.335660219192505,
+      "learning_rate": 9.881160851454363e-07,
+      "loss": 0.3549,
+      "step": 49292
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.3720204830169678,
+      "learning_rate": 9.879587348269675e-07,
+      "loss": 0.5102,
+      "step": 49293
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.5138254165649414,
+      "learning_rate": 9.878013956644838e-07,
+      "loss": 0.6802,
+      "step": 49294
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.8171167373657227,
+      "learning_rate": 9.876440676584198e-07,
+      "loss": 0.5226,
+      "step": 49295
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.6688480377197266,
+      "learning_rate": 9.874867508092162e-07,
+      "loss": 0.6228,
+      "step": 49296
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.926278591156006,
+      "learning_rate": 9.873294451173077e-07,
+      "loss": 0.5858,
+      "step": 49297
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.7095541954040527,
+      "learning_rate": 9.87172150583134e-07,
+      "loss": 0.6051,
+      "step": 49298
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.165951728820801,
+      "learning_rate": 9.870148672071301e-07,
+      "loss": 0.4002,
+      "step": 49299
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.6991195678710938,
+      "learning_rate": 9.868575949897357e-07,
+      "loss": 0.6074,
+      "step": 49300
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 4.643580436706543,
+      "learning_rate": 9.867003339313858e-07,
+      "loss": 0.6309,
+      "step": 49301
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.7109367847442627,
+      "learning_rate": 9.865430840325197e-07,
+      "loss": 0.7065,
+      "step": 49302
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.0287704467773438,
+      "learning_rate": 9.863858452935731e-07,
+      "loss": 0.5101,
+      "step": 49303
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.861356496810913,
+      "learning_rate": 9.862286177149849e-07,
+      "loss": 0.7267,
+      "step": 49304
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.8278920650482178,
+      "learning_rate": 9.8607140129719e-07,
+      "loss": 0.6263,
+      "step": 49305
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.4739882946014404,
+      "learning_rate": 9.859141960406281e-07,
+      "loss": 0.7833,
+      "step": 49306
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.7941854000091553,
+      "learning_rate": 9.85757001945734e-07,
+      "loss": 0.511,
+      "step": 49307
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 8.52450180053711,
+      "learning_rate": 9.855998190129473e-07,
+      "loss": 0.7555,
+      "step": 49308
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.302203893661499,
+      "learning_rate": 9.854426472427026e-07,
+      "loss": 0.6307,
+      "step": 49309
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.843705177307129,
+      "learning_rate": 9.852854866354388e-07,
+      "loss": 0.7408,
+      "step": 49310
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.546844005584717,
+      "learning_rate": 9.85128337191591e-07,
+      "loss": 0.5461,
+      "step": 49311
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.5595381259918213,
+      "learning_rate": 9.849711989115974e-07,
+      "loss": 0.425,
+      "step": 49312
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.6255414485931396,
+      "learning_rate": 9.848140717958948e-07,
+      "loss": 0.4327,
+      "step": 49313
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 1.9901231527328491,
+      "learning_rate": 9.846569558449214e-07,
+      "loss": 0.4537,
+      "step": 49314
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 1.9016430377960205,
+      "learning_rate": 9.844998510591113e-07,
+      "loss": 0.6255,
+      "step": 49315
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.414959192276001,
+      "learning_rate": 9.843427574389041e-07,
+      "loss": 0.6065,
+      "step": 49316
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.4383509159088135,
+      "learning_rate": 9.841856749847344e-07,
+      "loss": 0.282,
+      "step": 49317
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.1850225925445557,
+      "learning_rate": 9.840286036970415e-07,
+      "loss": 0.4655,
+      "step": 49318
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.9444735050201416,
+      "learning_rate": 9.838715435762592e-07,
+      "loss": 0.8042,
+      "step": 49319
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.442002058029175,
+      "learning_rate": 9.837144946228266e-07,
+      "loss": 0.6741,
+      "step": 49320
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.3577768802642822,
+      "learning_rate": 9.835574568371786e-07,
+      "loss": 0.7753,
+      "step": 49321
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.3575878143310547,
+      "learning_rate": 9.834004302197536e-07,
+      "loss": 0.7241,
+      "step": 49322
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.5562903881073,
+      "learning_rate": 9.832434147709862e-07,
+      "loss": 0.6544,
+      "step": 49323
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.707493305206299,
+      "learning_rate": 9.830864104913157e-07,
+      "loss": 0.529,
+      "step": 49324
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.621000051498413,
+      "learning_rate": 9.829294173811754e-07,
+      "loss": 0.8293,
+      "step": 49325
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.7092347145080566,
+      "learning_rate": 9.827724354410039e-07,
+      "loss": 0.6556,
+      "step": 49326
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.590717077255249,
+      "learning_rate": 9.826154646712388e-07,
+      "loss": 0.6369,
+      "step": 49327
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.3251264095306396,
+      "learning_rate": 9.824585050723134e-07,
+      "loss": 0.5134,
+      "step": 49328
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 1.9957313537597656,
+      "learning_rate": 9.823015566446675e-07,
+      "loss": 0.3249,
+      "step": 49329
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.720507860183716,
+      "learning_rate": 9.821446193887346e-07,
+      "loss": 0.7946,
+      "step": 49330
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.8099327087402344,
+      "learning_rate": 9.81987693304954e-07,
+      "loss": 0.5609,
+      "step": 49331
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 1.9910411834716797,
+      "learning_rate": 9.81830778393758e-07,
+      "loss": 0.5127,
+      "step": 49332
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.696829080581665,
+      "learning_rate": 9.816738746555865e-07,
+      "loss": 0.6534,
+      "step": 49333
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.058990716934204,
+      "learning_rate": 9.815169820908754e-07,
+      "loss": 0.6552,
+      "step": 49334
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.8946220874786377,
+      "learning_rate": 9.81360100700059e-07,
+      "loss": 0.8325,
+      "step": 49335
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.7065889835357666,
+      "learning_rate": 9.81203230483576e-07,
+      "loss": 0.7966,
+      "step": 49336
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.5098800659179688,
+      "learning_rate": 9.810463714418606e-07,
+      "loss": 0.644,
+      "step": 49337
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 1.9352465867996216,
+      "learning_rate": 9.80889523575349e-07,
+      "loss": 0.5686,
+      "step": 49338
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.8821024894714355,
+      "learning_rate": 9.807326868844796e-07,
+      "loss": 0.8316,
+      "step": 49339
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.912076950073242,
+      "learning_rate": 9.805758613696858e-07,
+      "loss": 0.7931,
+      "step": 49340
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.5040760040283203,
+      "learning_rate": 9.804190470314056e-07,
+      "loss": 0.5266,
+      "step": 49341
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.6178641319274902,
+      "learning_rate": 9.802622438700738e-07,
+      "loss": 0.4575,
+      "step": 49342
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.54168438911438,
+      "learning_rate": 9.801054518861275e-07,
+      "loss": 0.6845,
+      "step": 49343
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.40692138671875,
+      "learning_rate": 9.79948671080001e-07,
+      "loss": 0.7579,
+      "step": 49344
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.9798741340637207,
+      "learning_rate": 9.79791901452133e-07,
+      "loss": 0.5215,
+      "step": 49345
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.357576608657837,
+      "learning_rate": 9.796351430029555e-07,
+      "loss": 0.8612,
+      "step": 49346
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.0841057300567627,
+      "learning_rate": 9.794783957329086e-07,
+      "loss": 0.5794,
+      "step": 49347
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 1.9263345003128052,
+      "learning_rate": 9.793216596424243e-07,
+      "loss": 0.4614,
+      "step": 49348
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.198875665664673,
+      "learning_rate": 9.79164934731942e-07,
+      "loss": 0.6465,
+      "step": 49349
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.4316914081573486,
+      "learning_rate": 9.790082210018947e-07,
+      "loss": 0.8387,
+      "step": 49350
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.3161051273345947,
+      "learning_rate": 9.788515184527202e-07,
+      "loss": 0.6277,
+      "step": 49351
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.3266513347625732,
+      "learning_rate": 9.786948270848518e-07,
+      "loss": 0.638,
+      "step": 49352
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.663372755050659,
+      "learning_rate": 9.78538146898727e-07,
+      "loss": 0.4841,
+      "step": 49353
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.3939239978790283,
+      "learning_rate": 9.78381477894781e-07,
+      "loss": 0.6314,
+      "step": 49354
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.4591684341430664,
+      "learning_rate": 9.782248200734507e-07,
+      "loss": 0.6232,
+      "step": 49355
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.505858898162842,
+      "learning_rate": 9.780681734351698e-07,
+      "loss": 0.6016,
+      "step": 49356
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.378095865249634,
+      "learning_rate": 9.779115379803754e-07,
+      "loss": 0.6923,
+      "step": 49357
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.4882285594940186,
+      "learning_rate": 9.777549137095017e-07,
+      "loss": 0.3967,
+      "step": 49358
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.0882651805877686,
+      "learning_rate": 9.775983006229855e-07,
+      "loss": 0.5615,
+      "step": 49359
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.799288511276245,
+      "learning_rate": 9.774416987212604e-07,
+      "loss": 0.5935,
+      "step": 49360
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.3564488887786865,
+      "learning_rate": 9.77285108004764e-07,
+      "loss": 0.5113,
+      "step": 49361
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.360133409500122,
+      "learning_rate": 9.7712852847393e-07,
+      "loss": 0.5137,
+      "step": 49362
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.6857783794403076,
+      "learning_rate": 9.76971960129196e-07,
+      "loss": 0.5565,
+      "step": 49363
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.359790802001953,
+      "learning_rate": 9.768154029709942e-07,
+      "loss": 0.4611,
+      "step": 49364
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.406763792037964,
+      "learning_rate": 9.76658856999763e-07,
+      "loss": 0.7057,
+      "step": 49365
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.2130634784698486,
+      "learning_rate": 9.765023222159354e-07,
+      "loss": 0.4668,
+      "step": 49366
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.7884323596954346,
+      "learning_rate": 9.763457986199487e-07,
+      "loss": 0.4707,
+      "step": 49367
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.1131203174591064,
+      "learning_rate": 9.761892862122357e-07,
+      "loss": 0.8048,
+      "step": 49368
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.556173086166382,
+      "learning_rate": 9.760327849932344e-07,
+      "loss": 0.5889,
+      "step": 49369
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.9984707832336426,
+      "learning_rate": 9.758762949633766e-07,
+      "loss": 0.6207,
+      "step": 49370
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.2141788005828857,
+      "learning_rate": 9.757198161231013e-07,
+      "loss": 0.7129,
+      "step": 49371
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.5810956954956055,
+      "learning_rate": 9.755633484728399e-07,
+      "loss": 0.6284,
+      "step": 49372
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.5019285678863525,
+      "learning_rate": 9.7540689201303e-07,
+      "loss": 0.5449,
+      "step": 49373
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.374114751815796,
+      "learning_rate": 9.752504467441053e-07,
+      "loss": 0.3469,
+      "step": 49374
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.773749351501465,
+      "learning_rate": 9.750940126665026e-07,
+      "loss": 0.5684,
+      "step": 49375
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.732232093811035,
+      "learning_rate": 9.749375897806546e-07,
+      "loss": 0.8778,
+      "step": 49376
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.8168976306915283,
+      "learning_rate": 9.747811780869986e-07,
+      "loss": 0.6587,
+      "step": 49377
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.7425591945648193,
+      "learning_rate": 9.74624777585967e-07,
+      "loss": 0.683,
+      "step": 49378
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.630716562271118,
+      "learning_rate": 9.74468388277997e-07,
+      "loss": 0.6399,
+      "step": 49379
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.3599958419799805,
+      "learning_rate": 9.743120101635217e-07,
+      "loss": 0.609,
+      "step": 49380
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 1.7944105863571167,
+      "learning_rate": 9.741556432429778e-07,
+      "loss": 0.3741,
+      "step": 49381
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.789598226547241,
+      "learning_rate": 9.73999287516797e-07,
+      "loss": 0.6225,
+      "step": 49382
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.0072553157806396,
+      "learning_rate": 9.738429429854178e-07,
+      "loss": 0.5435,
+      "step": 49383
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.9818506240844727,
+      "learning_rate": 9.73686609649272e-07,
+      "loss": 0.6361,
+      "step": 49384
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.6858255863189697,
+      "learning_rate": 9.735302875087965e-07,
+      "loss": 0.7163,
+      "step": 49385
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.750850200653076,
+      "learning_rate": 9.733739765644235e-07,
+      "loss": 0.7598,
+      "step": 49386
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.9402897357940674,
+      "learning_rate": 9.732176768165891e-07,
+      "loss": 0.8389,
+      "step": 49387
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.733210563659668,
+      "learning_rate": 9.730613882657292e-07,
+      "loss": 0.6362,
+      "step": 49388
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.5165300369262695,
+      "learning_rate": 9.729051109122762e-07,
+      "loss": 0.5834,
+      "step": 49389
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.936178684234619,
+      "learning_rate": 9.727488447566664e-07,
+      "loss": 0.6732,
+      "step": 49390
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.5183908939361572,
+      "learning_rate": 9.725925897993322e-07,
+      "loss": 0.6058,
+      "step": 49391
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.436821222305298,
+      "learning_rate": 9.724363460407104e-07,
+      "loss": 0.5061,
+      "step": 49392
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 4.0434088706970215,
+      "learning_rate": 9.722801134812333e-07,
+      "loss": 0.9348,
+      "step": 49393
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.8343403339385986,
+      "learning_rate": 9.72123892121336e-07,
+      "loss": 0.7098,
+      "step": 49394
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.589830160140991,
+      "learning_rate": 9.719676819614548e-07,
+      "loss": 0.5058,
+      "step": 49395
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.601820945739746,
+      "learning_rate": 9.718114830020214e-07,
+      "loss": 0.4988,
+      "step": 49396
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.850193738937378,
+      "learning_rate": 9.716552952434727e-07,
+      "loss": 0.4163,
+      "step": 49397
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 4.354026794433594,
+      "learning_rate": 9.7149911868624e-07,
+      "loss": 0.6559,
+      "step": 49398
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.92488169670105,
+      "learning_rate": 9.713429533307594e-07,
+      "loss": 0.6218,
+      "step": 49399
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.3925862312316895,
+      "learning_rate": 9.711867991774659e-07,
+      "loss": 0.5021,
+      "step": 49400
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.1777188777923584,
+      "learning_rate": 9.710306562267913e-07,
+      "loss": 0.5461,
+      "step": 49401
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.6761107444763184,
+      "learning_rate": 9.70874524479173e-07,
+      "loss": 0.6663,
+      "step": 49402
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.778353214263916,
+      "learning_rate": 9.707184039350414e-07,
+      "loss": 0.7854,
+      "step": 49403
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.5613231658935547,
+      "learning_rate": 9.70562294594834e-07,
+      "loss": 0.6353,
+      "step": 49404
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.2057745456695557,
+      "learning_rate": 9.704061964589823e-07,
+      "loss": 0.619,
+      "step": 49405
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.7714078426361084,
+      "learning_rate": 9.702501095279226e-07,
+      "loss": 0.716,
+      "step": 49406
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.57053542137146,
+      "learning_rate": 9.700940338020865e-07,
+      "loss": 0.523,
+      "step": 49407
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.128685474395752,
+      "learning_rate": 9.699379692819106e-07,
+      "loss": 0.4125,
+      "step": 49408
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.5111262798309326,
+      "learning_rate": 9.697819159678261e-07,
+      "loss": 0.8589,
+      "step": 49409
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.629215955734253,
+      "learning_rate": 9.696258738602693e-07,
+      "loss": 0.7004,
+      "step": 49410
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.221506118774414,
+      "learning_rate": 9.69469842959672e-07,
+      "loss": 0.602,
+      "step": 49411
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.5341227054595947,
+      "learning_rate": 9.69313823266471e-07,
+      "loss": 0.526,
+      "step": 49412
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.7789390087127686,
+      "learning_rate": 9.69157814781097e-07,
+      "loss": 0.5366,
+      "step": 49413
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.81636381149292,
+      "learning_rate": 9.690018175039845e-07,
+      "loss": 0.7725,
+      "step": 49414
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.738342761993408,
+      "learning_rate": 9.688458314355685e-07,
+      "loss": 0.569,
+      "step": 49415
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.3360116481781006,
+      "learning_rate": 9.68689856576283e-07,
+      "loss": 0.5407,
+      "step": 49416
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.5247514247894287,
+      "learning_rate": 9.685338929265599e-07,
+      "loss": 0.8538,
+      "step": 49417
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.578826904296875,
+      "learning_rate": 9.683779404868349e-07,
+      "loss": 0.6556,
+      "step": 49418
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.7311270236968994,
+      "learning_rate": 9.682219992575393e-07,
+      "loss": 0.7602,
+      "step": 49419
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 4.011478424072266,
+      "learning_rate": 9.68066069239109e-07,
+      "loss": 0.4589,
+      "step": 49420
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.823761463165283,
+      "learning_rate": 9.67910150431976e-07,
+      "loss": 0.6897,
+      "step": 49421
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.244723320007324,
+      "learning_rate": 9.67754242836575e-07,
+      "loss": 0.498,
+      "step": 49422
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.369536876678467,
+      "learning_rate": 9.675983464533377e-07,
+      "loss": 0.5607,
+      "step": 49423
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.304736852645874,
+      "learning_rate": 9.67442461282701e-07,
+      "loss": 0.5385,
+      "step": 49424
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 4.001416206359863,
+      "learning_rate": 9.672865873250942e-07,
+      "loss": 0.645,
+      "step": 49425
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.9783966541290283,
+      "learning_rate": 9.671307245809535e-07,
+      "loss": 0.5055,
+      "step": 49426
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.4963936805725098,
+      "learning_rate": 9.66974873050711e-07,
+      "loss": 0.5872,
+      "step": 49427
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.45876407623291,
+      "learning_rate": 9.668190327348015e-07,
+      "loss": 0.4819,
+      "step": 49428
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.4634547233581543,
+      "learning_rate": 9.666632036336565e-07,
+      "loss": 0.6461,
+      "step": 49429
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.8040761947631836,
+      "learning_rate": 9.665073857477113e-07,
+      "loss": 0.8088,
+      "step": 49430
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.900700807571411,
+      "learning_rate": 9.663515790773963e-07,
+      "loss": 0.6792,
+      "step": 49431
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.5214176177978516,
+      "learning_rate": 9.661957836231484e-07,
+      "loss": 0.4526,
+      "step": 49432
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.9887430667877197,
+      "learning_rate": 9.660399993853975e-07,
+      "loss": 0.473,
+      "step": 49433
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.22711181640625,
+      "learning_rate": 9.658842263645785e-07,
+      "loss": 0.4575,
+      "step": 49434
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.6528780460357666,
+      "learning_rate": 9.657284645611236e-07,
+      "loss": 0.7155,
+      "step": 49435
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.295464515686035,
+      "learning_rate": 9.655727139754683e-07,
+      "loss": 0.6201,
+      "step": 49436
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.2560219764709473,
+      "learning_rate": 9.654169746080427e-07,
+      "loss": 0.5973,
+      "step": 49437
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.1761474609375,
+      "learning_rate": 9.65261246459282e-07,
+      "loss": 0.5382,
+      "step": 49438
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.905358076095581,
+      "learning_rate": 9.651055295296174e-07,
+      "loss": 0.4374,
+      "step": 49439
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.8240439891815186,
+      "learning_rate": 9.649498238194837e-07,
+      "loss": 0.3873,
+      "step": 49440
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.7688567638397217,
+      "learning_rate": 9.647941293293123e-07,
+      "loss": 0.611,
+      "step": 49441
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.6737372875213623,
+      "learning_rate": 9.646384460595377e-07,
+      "loss": 0.7235,
+      "step": 49442
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.993309736251831,
+      "learning_rate": 9.644827740105905e-07,
+      "loss": 0.5551,
+      "step": 49443
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.3597254753112793,
+      "learning_rate": 9.643271131829068e-07,
+      "loss": 0.5527,
+      "step": 49444
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.25048565864563,
+      "learning_rate": 9.641714635769156e-07,
+      "loss": 0.3407,
+      "step": 49445
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 4.646751880645752,
+      "learning_rate": 9.640158251930538e-07,
+      "loss": 0.6814,
+      "step": 49446
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.0494821071624756,
+      "learning_rate": 9.638601980317496e-07,
+      "loss": 0.4209,
+      "step": 49447
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.314393997192383,
+      "learning_rate": 9.637045820934405e-07,
+      "loss": 0.6668,
+      "step": 49448
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.432318925857544,
+      "learning_rate": 9.635489773785556e-07,
+      "loss": 0.4745,
+      "step": 49449
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 4.415089130401611,
+      "learning_rate": 9.633933838875287e-07,
+      "loss": 0.4447,
+      "step": 49450
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.330117702484131,
+      "learning_rate": 9.632378016207938e-07,
+      "loss": 0.6105,
+      "step": 49451
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.663771867752075,
+      "learning_rate": 9.630822305787813e-07,
+      "loss": 0.5438,
+      "step": 49452
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.2512154579162598,
+      "learning_rate": 9.629266707619255e-07,
+      "loss": 0.5709,
+      "step": 49453
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.3984904289245605,
+      "learning_rate": 9.627711221706576e-07,
+      "loss": 0.3424,
+      "step": 49454
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.901167154312134,
+      "learning_rate": 9.62615584805411e-07,
+      "loss": 0.5172,
+      "step": 49455
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 4.017672538757324,
+      "learning_rate": 9.624600586666194e-07,
+      "loss": 0.6521,
+      "step": 49456
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.540517807006836,
+      "learning_rate": 9.623045437547118e-07,
+      "loss": 0.5578,
+      "step": 49457
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.2400436401367188,
+      "learning_rate": 9.621490400701245e-07,
+      "loss": 0.5955,
+      "step": 49458
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.8510844707489014,
+      "learning_rate": 9.619935476132868e-07,
+      "loss": 0.6764,
+      "step": 49459
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 4.410655975341797,
+      "learning_rate": 9.618380663846326e-07,
+      "loss": 0.7233,
+      "step": 49460
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.4213409423828125,
+      "learning_rate": 9.61682596384595e-07,
+      "loss": 0.6696,
+      "step": 49461
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.620164394378662,
+      "learning_rate": 9.615271376136043e-07,
+      "loss": 0.7201,
+      "step": 49462
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.703657627105713,
+      "learning_rate": 9.613716900720948e-07,
+      "loss": 0.5682,
+      "step": 49463
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.8132119178771973,
+      "learning_rate": 9.612162537604962e-07,
+      "loss": 0.6654,
+      "step": 49464
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.8381195068359375,
+      "learning_rate": 9.61060828679244e-07,
+      "loss": 0.529,
+      "step": 49465
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.3868579864501953,
+      "learning_rate": 9.609054148287672e-07,
+      "loss": 0.6109,
+      "step": 49466
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.3198914527893066,
+      "learning_rate": 9.607500122095004e-07,
+      "loss": 0.5281,
+      "step": 49467
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.6500627994537354,
+      "learning_rate": 9.605946208218737e-07,
+      "loss": 0.5889,
+      "step": 49468
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.4012115001678467,
+      "learning_rate": 9.60439240666321e-07,
+      "loss": 0.4216,
+      "step": 49469
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 1.7841061353683472,
+      "learning_rate": 9.602838717432727e-07,
+      "loss": 0.3555,
+      "step": 49470
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.666177272796631,
+      "learning_rate": 9.60128514053163e-07,
+      "loss": 0.5196,
+      "step": 49471
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.7852587699890137,
+      "learning_rate": 9.599731675964208e-07,
+      "loss": 0.4486,
+      "step": 49472
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.632216453552246,
+      "learning_rate": 9.598178323734814e-07,
+      "loss": 0.7599,
+      "step": 49473
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.8702945709228516,
+      "learning_rate": 9.596625083847739e-07,
+      "loss": 0.5355,
+      "step": 49474
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 1.5428106784820557,
+      "learning_rate": 9.59507195630731e-07,
+      "loss": 0.4961,
+      "step": 49475
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.0404772758483887,
+      "learning_rate": 9.593518941117851e-07,
+      "loss": 0.6077,
+      "step": 49476
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.607328176498413,
+      "learning_rate": 9.591966038283696e-07,
+      "loss": 0.4673,
+      "step": 49477
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 4.8940582275390625,
+      "learning_rate": 9.590413247809134e-07,
+      "loss": 0.5043,
+      "step": 49478
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 4.763576030731201,
+      "learning_rate": 9.588860569698505e-07,
+      "loss": 0.8899,
+      "step": 49479
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.4672577381134033,
+      "learning_rate": 9.587308003956103e-07,
+      "loss": 0.8134,
+      "step": 49480
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 4.295924186706543,
+      "learning_rate": 9.585755550586267e-07,
+      "loss": 0.769,
+      "step": 49481
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.5623788833618164,
+      "learning_rate": 9.5842032095933e-07,
+      "loss": 0.5359,
+      "step": 49482
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.6472561359405518,
+      "learning_rate": 9.582650980981534e-07,
+      "loss": 0.5705,
+      "step": 49483
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.8150830268859863,
+      "learning_rate": 9.581098864755263e-07,
+      "loss": 0.4826,
+      "step": 49484
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.368140697479248,
+      "learning_rate": 9.579546860918827e-07,
+      "loss": 0.659,
+      "step": 49485
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.214362621307373,
+      "learning_rate": 9.577994969476512e-07,
+      "loss": 0.6085,
+      "step": 49486
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.6285252571105957,
+      "learning_rate": 9.576443190432672e-07,
+      "loss": 0.8895,
+      "step": 49487
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.0225884914398193,
+      "learning_rate": 9.57489152379158e-07,
+      "loss": 0.674,
+      "step": 49488
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 1.847596287727356,
+      "learning_rate": 9.573339969557587e-07,
+      "loss": 0.3601,
+      "step": 49489
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.378793716430664,
+      "learning_rate": 9.571788527734976e-07,
+      "loss": 1.0151,
+      "step": 49490
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.214841365814209,
+      "learning_rate": 9.570237198328092e-07,
+      "loss": 0.543,
+      "step": 49491
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.442486047744751,
+      "learning_rate": 9.568685981341225e-07,
+      "loss": 0.7297,
+      "step": 49492
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.4974915981292725,
+      "learning_rate": 9.5671348767787e-07,
+      "loss": 0.4006,
+      "step": 49493
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.6735687255859375,
+      "learning_rate": 9.565583884644819e-07,
+      "loss": 0.6676,
+      "step": 49494
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 4.598949909210205,
+      "learning_rate": 9.5640330049439e-07,
+      "loss": 0.4435,
+      "step": 49495
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.617655038833618,
+      "learning_rate": 9.56248223768026e-07,
+      "loss": 0.4957,
+      "step": 49496
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.6069583892822266,
+      "learning_rate": 9.560931582858217e-07,
+      "loss": 0.4377,
+      "step": 49497
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.787978172302246,
+      "learning_rate": 9.55938104048207e-07,
+      "loss": 0.6149,
+      "step": 49498
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.6759700775146484,
+      "learning_rate": 9.55783061055614e-07,
+      "loss": 0.6503,
+      "step": 49499
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.522441864013672,
+      "learning_rate": 9.55628029308472e-07,
+      "loss": 0.5574,
+      "step": 49500
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.734797716140747,
+      "learning_rate": 9.554730088072151e-07,
+      "loss": 0.4278,
+      "step": 49501
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.9115259647369385,
+      "learning_rate": 9.55317999552271e-07,
+      "loss": 0.5691,
+      "step": 49502
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.2625184059143066,
+      "learning_rate": 9.551630015440737e-07,
+      "loss": 0.6516,
+      "step": 49503
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.981139898300171,
+      "learning_rate": 9.550080147830521e-07,
+      "loss": 0.706,
+      "step": 49504
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.261009454727173,
+      "learning_rate": 9.548530392696387e-07,
+      "loss": 0.6335,
+      "step": 49505
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 1.6480830907821655,
+      "learning_rate": 9.546980750042623e-07,
+      "loss": 0.322,
+      "step": 49506
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.9593756198883057,
+      "learning_rate": 9.545431219873568e-07,
+      "loss": 0.5277,
+      "step": 49507
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.2145378589630127,
+      "learning_rate": 9.5438818021935e-07,
+      "loss": 0.276,
+      "step": 49508
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.4613022804260254,
+      "learning_rate": 9.54233249700675e-07,
+      "loss": 0.5614,
+      "step": 49509
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 1.3754217624664307,
+      "learning_rate": 9.540783304317613e-07,
+      "loss": 0.3508,
+      "step": 49510
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 5.449055194854736,
+      "learning_rate": 9.539234224130395e-07,
+      "loss": 0.6917,
+      "step": 49511
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.7358288764953613,
+      "learning_rate": 9.537685256449425e-07,
+      "loss": 0.6641,
+      "step": 49512
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.6874234676361084,
+      "learning_rate": 9.536136401278984e-07,
+      "loss": 0.4479,
+      "step": 49513
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.826432704925537,
+      "learning_rate": 9.534587658623396e-07,
+      "loss": 0.5218,
+      "step": 49514
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.5216867923736572,
+      "learning_rate": 9.533039028486952e-07,
+      "loss": 0.5911,
+      "step": 49515
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.170210361480713,
+      "learning_rate": 9.531490510873969e-07,
+      "loss": 0.599,
+      "step": 49516
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.7411630153656006,
+      "learning_rate": 9.529942105788759e-07,
+      "loss": 0.6712,
+      "step": 49517
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.9065864086151123,
+      "learning_rate": 9.528393813235609e-07,
+      "loss": 0.636,
+      "step": 49518
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.4435572624206543,
+      "learning_rate": 9.526845633218851e-07,
+      "loss": 0.5897,
+      "step": 49519
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.1092143058776855,
+      "learning_rate": 9.525297565742758e-07,
+      "loss": 0.5883,
+      "step": 49520
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.3842976093292236,
+      "learning_rate": 9.523749610811661e-07,
+      "loss": 0.5034,
+      "step": 49521
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.9208314418792725,
+      "learning_rate": 9.522201768429845e-07,
+      "loss": 0.6537,
+      "step": 49522
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 4.2266621589660645,
+      "learning_rate": 9.520654038601623e-07,
+      "loss": 0.4349,
+      "step": 49523
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 4.2648539543151855,
+      "learning_rate": 9.519106421331309e-07,
+      "loss": 0.5834,
+      "step": 49524
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.321464776992798,
+      "learning_rate": 9.517558916623182e-07,
+      "loss": 0.7518,
+      "step": 49525
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.155838966369629,
+      "learning_rate": 9.516011524481572e-07,
+      "loss": 0.6125,
+      "step": 49526
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.804659605026245,
+      "learning_rate": 9.514464244910754e-07,
+      "loss": 0.6697,
+      "step": 49527
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.8820860385894775,
+      "learning_rate": 9.512917077915057e-07,
+      "loss": 0.4767,
+      "step": 49528
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.2553443908691406,
+      "learning_rate": 9.511370023498761e-07,
+      "loss": 0.5702,
+      "step": 49529
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.6690728664398193,
+      "learning_rate": 9.509823081666187e-07,
+      "loss": 0.6527,
+      "step": 49530
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.690441370010376,
+      "learning_rate": 9.508276252421616e-07,
+      "loss": 0.5803,
+      "step": 49531
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.216003656387329,
+      "learning_rate": 9.506729535769376e-07,
+      "loss": 0.5626,
+      "step": 49532
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.9182639122009277,
+      "learning_rate": 9.505182931713735e-07,
+      "loss": 0.6055,
+      "step": 49533
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.2049405574798584,
+      "learning_rate": 9.503636440259029e-07,
+      "loss": 0.582,
+      "step": 49534
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.843938112258911,
+      "learning_rate": 9.502090061409519e-07,
+      "loss": 0.4997,
+      "step": 49535
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.4797005653381348,
+      "learning_rate": 9.500543795169531e-07,
+      "loss": 0.5233,
+      "step": 49536
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.4370434284210205,
+      "learning_rate": 9.498997641543356e-07,
+      "loss": 0.6633,
+      "step": 49537
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.67081618309021,
+      "learning_rate": 9.497451600535312e-07,
+      "loss": 0.5374,
+      "step": 49538
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 6.190176486968994,
+      "learning_rate": 9.495905672149669e-07,
+      "loss": 0.3773,
+      "step": 49539
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.1946306228637695,
+      "learning_rate": 9.49435985639075e-07,
+      "loss": 0.8106,
+      "step": 49540
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.1412129402160645,
+      "learning_rate": 9.492814153262836e-07,
+      "loss": 0.504,
+      "step": 49541
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.189668655395508,
+      "learning_rate": 9.49126856277024e-07,
+      "loss": 0.8049,
+      "step": 49542
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.694582223892212,
+      "learning_rate": 9.489723084917236e-07,
+      "loss": 0.5389,
+      "step": 49543
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 1.751431941986084,
+      "learning_rate": 9.488177719708153e-07,
+      "loss": 0.4295,
+      "step": 49544
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.2932238578796387,
+      "learning_rate": 9.486632467147255e-07,
+      "loss": 0.7875,
+      "step": 49545
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 4.847456455230713,
+      "learning_rate": 9.485087327238873e-07,
+      "loss": 0.7007,
+      "step": 49546
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.754354238510132,
+      "learning_rate": 9.48354229998727e-07,
+      "loss": 0.494,
+      "step": 49547
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.9374008178710938,
+      "learning_rate": 9.481997385396774e-07,
+      "loss": 0.7896,
+      "step": 49548
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.150904417037964,
+      "learning_rate": 9.48045258347165e-07,
+      "loss": 0.4308,
+      "step": 49549
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.341714382171631,
+      "learning_rate": 9.478907894216222e-07,
+      "loss": 0.3866,
+      "step": 49550
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.102142572402954,
+      "learning_rate": 9.477363317634753e-07,
+      "loss": 0.7483,
+      "step": 49551
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.996166229248047,
+      "learning_rate": 9.475818853731578e-07,
+      "loss": 0.5084,
+      "step": 49552
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.3845670223236084,
+      "learning_rate": 9.474274502510949e-07,
+      "loss": 0.4639,
+      "step": 49553
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.2025671005249023,
+      "learning_rate": 9.472730263977203e-07,
+      "loss": 0.4836,
+      "step": 49554
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.7835781574249268,
+      "learning_rate": 9.471186138134592e-07,
+      "loss": 0.6375,
+      "step": 49555
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.7160325050354004,
+      "learning_rate": 9.469642124987433e-07,
+      "loss": 0.6274,
+      "step": 49556
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.1689157485961914,
+      "learning_rate": 9.468098224540018e-07,
+      "loss": 0.6242,
+      "step": 49557
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.7873425483703613,
+      "learning_rate": 9.466554436796649e-07,
+      "loss": 0.6071,
+      "step": 49558
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.742550849914551,
+      "learning_rate": 9.465010761761595e-07,
+      "loss": 0.8167,
+      "step": 49559
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 1.8023751974105835,
+      "learning_rate": 9.463467199439174e-07,
+      "loss": 0.4455,
+      "step": 49560
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.520103693008423,
+      "learning_rate": 9.461923749833651e-07,
+      "loss": 0.6354,
+      "step": 49561
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.7227816581726074,
+      "learning_rate": 9.460380412949349e-07,
+      "loss": 0.5842,
+      "step": 49562
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 1.945130467414856,
+      "learning_rate": 9.458837188790526e-07,
+      "loss": 0.2218,
+      "step": 49563
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 4.30118989944458,
+      "learning_rate": 9.457294077361507e-07,
+      "loss": 0.7299,
+      "step": 49564
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.9992735385894775,
+      "learning_rate": 9.455751078666547e-07,
+      "loss": 0.6975,
+      "step": 49565
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.20058012008667,
+      "learning_rate": 9.454208192709974e-07,
+      "loss": 0.546,
+      "step": 49566
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.4313371181488037,
+      "learning_rate": 9.452665419496043e-07,
+      "loss": 0.8109,
+      "step": 49567
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.5251550674438477,
+      "learning_rate": 9.451122759029074e-07,
+      "loss": 0.5692,
+      "step": 49568
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.1549227237701416,
+      "learning_rate": 9.449580211313325e-07,
+      "loss": 0.6155,
+      "step": 49569
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.732196807861328,
+      "learning_rate": 9.44803777635312e-07,
+      "loss": 0.7303,
+      "step": 49570
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.6864001750946045,
+      "learning_rate": 9.446495454152721e-07,
+      "loss": 0.2715,
+      "step": 49571
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.130174398422241,
+      "learning_rate": 9.44495324471642e-07,
+      "loss": 0.4566,
+      "step": 49572
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.6155900955200195,
+      "learning_rate": 9.443411148048531e-07,
+      "loss": 0.4527,
+      "step": 49573
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.763160228729248,
+      "learning_rate": 9.441869164153305e-07,
+      "loss": 0.6677,
+      "step": 49574
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.5484225749969482,
+      "learning_rate": 9.440327293035062e-07,
+      "loss": 0.6992,
+      "step": 49575
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.5513620376586914,
+      "learning_rate": 9.438785534698064e-07,
+      "loss": 0.5645,
+      "step": 49576
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 7.107978820800781,
+      "learning_rate": 9.437243889146602e-07,
+      "loss": 0.489,
+      "step": 49577
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.1155693531036377,
+      "learning_rate": 9.43570235638499e-07,
+      "loss": 0.4558,
+      "step": 49578
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.616161346435547,
+      "learning_rate": 9.43416093641748e-07,
+      "loss": 0.8029,
+      "step": 49579
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.1866183280944824,
+      "learning_rate": 9.432619629248379e-07,
+      "loss": 0.5684,
+      "step": 49580
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.3979992866516113,
+      "learning_rate": 9.431078434881957e-07,
+      "loss": 0.4376,
+      "step": 49581
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 5.276735782623291,
+      "learning_rate": 9.429537353322527e-07,
+      "loss": 0.6908,
+      "step": 49582
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.6743664741516113,
+      "learning_rate": 9.427996384574334e-07,
+      "loss": 0.6229,
+      "step": 49583
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.331827163696289,
+      "learning_rate": 9.426455528641687e-07,
+      "loss": 0.6353,
+      "step": 49584
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.532911539077759,
+      "learning_rate": 9.424914785528882e-07,
+      "loss": 0.7818,
+      "step": 49585
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.0708606243133545,
+      "learning_rate": 9.423374155240178e-07,
+      "loss": 0.6155,
+      "step": 49586
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.8483681678771973,
+      "learning_rate": 9.421833637779882e-07,
+      "loss": 0.7226,
+      "step": 49587
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.351074457168579,
+      "learning_rate": 9.42029323315225e-07,
+      "loss": 0.62,
+      "step": 49588
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.380171537399292,
+      "learning_rate": 9.418752941361598e-07,
+      "loss": 0.4003,
+      "step": 49589
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.376506805419922,
+      "learning_rate": 9.417212762412175e-07,
+      "loss": 0.5,
+      "step": 49590
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.702266216278076,
+      "learning_rate": 9.415672696308292e-07,
+      "loss": 0.7305,
+      "step": 49591
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.394423723220825,
+      "learning_rate": 9.414132743054211e-07,
+      "loss": 0.5171,
+      "step": 49592
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.7729620933532715,
+      "learning_rate": 9.412592902654238e-07,
+      "loss": 0.4359,
+      "step": 49593
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.5272064208984375,
+      "learning_rate": 9.411053175112622e-07,
+      "loss": 0.7772,
+      "step": 49594
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.393791675567627,
+      "learning_rate": 9.409513560433681e-07,
+      "loss": 0.5263,
+      "step": 49595
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.4261977672576904,
+      "learning_rate": 9.407974058621661e-07,
+      "loss": 0.887,
+      "step": 49596
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.754338026046753,
+      "learning_rate": 9.406434669680859e-07,
+      "loss": 0.5089,
+      "step": 49597
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.0195932388305664,
+      "learning_rate": 9.404895393615554e-07,
+      "loss": 0.5655,
+      "step": 49598
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 4.419707298278809,
+      "learning_rate": 9.403356230430038e-07,
+      "loss": 0.7053,
+      "step": 49599
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.090223550796509,
+      "learning_rate": 9.401817180128575e-07,
+      "loss": 0.4855,
+      "step": 49600
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.502788782119751,
+      "learning_rate": 9.400278242715455e-07,
+      "loss": 0.5793,
+      "step": 49601
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.650195360183716,
+      "learning_rate": 9.398739418194946e-07,
+      "loss": 0.6432,
+      "step": 49602
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.1660637855529785,
+      "learning_rate": 9.397200706571341e-07,
+      "loss": 0.5615,
+      "step": 49603
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.4351792335510254,
+      "learning_rate": 9.395662107848902e-07,
+      "loss": 0.6244,
+      "step": 49604
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.8544578552246094,
+      "learning_rate": 9.394123622031925e-07,
+      "loss": 0.6779,
+      "step": 49605
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.429450511932373,
+      "learning_rate": 9.392585249124669e-07,
+      "loss": 0.3468,
+      "step": 49606
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.86883282661438,
+      "learning_rate": 9.391046989131436e-07,
+      "loss": 0.4107,
+      "step": 49607
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.0109481811523438,
+      "learning_rate": 9.389508842056472e-07,
+      "loss": 0.5199,
+      "step": 49608
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.3294358253479004,
+      "learning_rate": 9.387970807904084e-07,
+      "loss": 0.4033,
+      "step": 49609
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 4.314674377441406,
+      "learning_rate": 9.386432886678526e-07,
+      "loss": 0.469,
+      "step": 49610
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.458400011062622,
+      "learning_rate": 9.384895078384094e-07,
+      "loss": 0.6401,
+      "step": 49611
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.7740538120269775,
+      "learning_rate": 9.383357383025044e-07,
+      "loss": 0.5741,
+      "step": 49612
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.615910530090332,
+      "learning_rate": 9.381819800605668e-07,
+      "loss": 0.658,
+      "step": 49613
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.4225924015045166,
+      "learning_rate": 9.380282331130231e-07,
+      "loss": 0.3949,
+      "step": 49614
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 4.016534805297852,
+      "learning_rate": 9.378744974603021e-07,
+      "loss": 0.5251,
+      "step": 49615
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.704493761062622,
+      "learning_rate": 9.377207731028287e-07,
+      "loss": 0.8215,
+      "step": 49616
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 6.697678089141846,
+      "learning_rate": 9.375670600410325e-07,
+      "loss": 0.6759,
+      "step": 49617
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 9.683015823364258,
+      "learning_rate": 9.374133582753409e-07,
+      "loss": 0.5636,
+      "step": 49618
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.153212308883667,
+      "learning_rate": 9.372596678061813e-07,
+      "loss": 0.867,
+      "step": 49619
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.9087929725646973,
+      "learning_rate": 9.3710598863398e-07,
+      "loss": 0.6534,
+      "step": 49620
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.3383283615112305,
+      "learning_rate": 9.369523207591658e-07,
+      "loss": 0.4308,
+      "step": 49621
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.2411396503448486,
+      "learning_rate": 9.367986641821642e-07,
+      "loss": 0.4737,
+      "step": 49622
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.4251317977905273,
+      "learning_rate": 9.366450189034049e-07,
+      "loss": 0.7093,
+      "step": 49623
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.7930681705474854,
+      "learning_rate": 9.36491384923312e-07,
+      "loss": 0.5004,
+      "step": 49624
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 4.506557464599609,
+      "learning_rate": 9.363377622423153e-07,
+      "loss": 0.5165,
+      "step": 49625
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.3108909130096436,
+      "learning_rate": 9.361841508608399e-07,
+      "loss": 0.4,
+      "step": 49626
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.55112361907959,
+      "learning_rate": 9.360305507793155e-07,
+      "loss": 0.6719,
+      "step": 49627
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.0218145847320557,
+      "learning_rate": 9.358769619981667e-07,
+      "loss": 0.8301,
+      "step": 49628
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.099430561065674,
+      "learning_rate": 9.357233845178226e-07,
+      "loss": 0.4648,
+      "step": 49629
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.4085745811462402,
+      "learning_rate": 9.35569818338708e-07,
+      "loss": 0.5489,
+      "step": 49630
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.7647430896759033,
+      "learning_rate": 9.354162634612523e-07,
+      "loss": 0.8936,
+      "step": 49631
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 5.0938639640808105,
+      "learning_rate": 9.352627198858805e-07,
+      "loss": 0.6689,
+      "step": 49632
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.7984416484832764,
+      "learning_rate": 9.351091876130214e-07,
+      "loss": 0.6798,
+      "step": 49633
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.6801578998565674,
+      "learning_rate": 9.349556666430992e-07,
+      "loss": 0.7757,
+      "step": 49634
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.759988784790039,
+      "learning_rate": 9.34802156976543e-07,
+      "loss": 0.6773,
+      "step": 49635
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.643669843673706,
+      "learning_rate": 9.346486586137804e-07,
+      "loss": 0.5807,
+      "step": 49636
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.6322295665740967,
+      "learning_rate": 9.344951715552359e-07,
+      "loss": 0.469,
+      "step": 49637
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.3781497478485107,
+      "learning_rate": 9.343416958013368e-07,
+      "loss": 0.3831,
+      "step": 49638
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.9988350868225098,
+      "learning_rate": 9.341882313525119e-07,
+      "loss": 0.6896,
+      "step": 49639
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.7064547538757324,
+      "learning_rate": 9.340347782091852e-07,
+      "loss": 0.5286,
+      "step": 49640
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.700085401535034,
+      "learning_rate": 9.338813363717858e-07,
+      "loss": 0.5119,
+      "step": 49641
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.057910680770874,
+      "learning_rate": 9.337279058407384e-07,
+      "loss": 0.4976,
+      "step": 49642
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.036635160446167,
+      "learning_rate": 9.335744866164715e-07,
+      "loss": 0.3811,
+      "step": 49643
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.0249247550964355,
+      "learning_rate": 9.334210786994091e-07,
+      "loss": 0.4722,
+      "step": 49644
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.3925528526306152,
+      "learning_rate": 9.332676820899795e-07,
+      "loss": 0.3662,
+      "step": 49645
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.1481966972351074,
+      "learning_rate": 9.331142967886098e-07,
+      "loss": 0.5348,
+      "step": 49646
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.9989850521087646,
+      "learning_rate": 9.329609227957248e-07,
+      "loss": 0.5871,
+      "step": 49647
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.6163675785064697,
+      "learning_rate": 9.328075601117537e-07,
+      "loss": 0.5413,
+      "step": 49648
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.452735424041748,
+      "learning_rate": 9.326542087371193e-07,
+      "loss": 0.5166,
+      "step": 49649
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.3466134071350098,
+      "learning_rate": 9.325008686722514e-07,
+      "loss": 0.5497,
+      "step": 49650
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.454582691192627,
+      "learning_rate": 9.323475399175736e-07,
+      "loss": 0.4425,
+      "step": 49651
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.6438615322113037,
+      "learning_rate": 9.321942224735148e-07,
+      "loss": 0.6915,
+      "step": 49652
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 4.138458728790283,
+      "learning_rate": 9.320409163404987e-07,
+      "loss": 0.8844,
+      "step": 49653
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.666034460067749,
+      "learning_rate": 9.318876215189543e-07,
+      "loss": 0.5298,
+      "step": 49654
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.5572190284729004,
+      "learning_rate": 9.317343380093052e-07,
+      "loss": 0.8137,
+      "step": 49655
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.6307852268218994,
+      "learning_rate": 9.3158106581198e-07,
+      "loss": 0.6538,
+      "step": 49656
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.709092855453491,
+      "learning_rate": 9.314278049274022e-07,
+      "loss": 0.4268,
+      "step": 49657
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.514024496078491,
+      "learning_rate": 9.312745553560004e-07,
+      "loss": 0.6686,
+      "step": 49658
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.6938045024871826,
+      "learning_rate": 9.311213170981992e-07,
+      "loss": 0.6438,
+      "step": 49659
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.56154727935791,
+      "learning_rate": 9.309680901544272e-07,
+      "loss": 0.4331,
+      "step": 49660
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.944775342941284,
+      "learning_rate": 9.308148745251066e-07,
+      "loss": 0.8118,
+      "step": 49661
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.4168803691864014,
+      "learning_rate": 9.306616702106675e-07,
+      "loss": 0.7374,
+      "step": 49662
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.341728448867798,
+      "learning_rate": 9.305084772115319e-07,
+      "loss": 0.5671,
+      "step": 49663
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.8765084743499756,
+      "learning_rate": 9.303552955281298e-07,
+      "loss": 0.5767,
+      "step": 49664
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.1681885719299316,
+      "learning_rate": 9.302021251608834e-07,
+      "loss": 0.4928,
+      "step": 49665
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.719010353088379,
+      "learning_rate": 9.300489661102219e-07,
+      "loss": 0.6013,
+      "step": 49666
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.6845285892486572,
+      "learning_rate": 9.298958183765677e-07,
+      "loss": 0.6587,
+      "step": 49667
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.4212074279785156,
+      "learning_rate": 9.297426819603506e-07,
+      "loss": 0.447,
+      "step": 49668
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.44877028465271,
+      "learning_rate": 9.295895568619923e-07,
+      "loss": 0.443,
+      "step": 49669
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.4184820652008057,
+      "learning_rate": 9.294364430819225e-07,
+      "loss": 0.7909,
+      "step": 49670
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.7243587970733643,
+      "learning_rate": 9.292833406205637e-07,
+      "loss": 0.7863,
+      "step": 49671
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.9482898712158203,
+      "learning_rate": 9.291302494783438e-07,
+      "loss": 0.728,
+      "step": 49672
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.337228298187256,
+      "learning_rate": 9.28977169655687e-07,
+      "loss": 0.3542,
+      "step": 49673
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.514273166656494,
+      "learning_rate": 9.288241011530208e-07,
+      "loss": 0.8173,
+      "step": 49674
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 4.1092529296875,
+      "learning_rate": 9.28671043970768e-07,
+      "loss": 0.6486,
+      "step": 49675
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.5727062225341797,
+      "learning_rate": 9.285179981093573e-07,
+      "loss": 0.6249,
+      "step": 49676
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 4.22511625289917,
+      "learning_rate": 9.283649635692121e-07,
+      "loss": 0.5517,
+      "step": 49677
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.7897162437438965,
+      "learning_rate": 9.282119403507578e-07,
+      "loss": 0.5335,
+      "step": 49678
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.913788318634033,
+      "learning_rate": 9.280589284544217e-07,
+      "loss": 0.6638,
+      "step": 49679
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.456757068634033,
+      "learning_rate": 9.279059278806291e-07,
+      "loss": 0.8459,
+      "step": 49680
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.391380786895752,
+      "learning_rate": 9.277529386298034e-07,
+      "loss": 0.6799,
+      "step": 49681
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.6492433547973633,
+      "learning_rate": 9.275999607023722e-07,
+      "loss": 0.5708,
+      "step": 49682
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.295966863632202,
+      "learning_rate": 9.274469940987596e-07,
+      "loss": 0.3532,
+      "step": 49683
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.7539968490600586,
+      "learning_rate": 9.272940388193918e-07,
+      "loss": 0.7924,
+      "step": 49684
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 1.6017141342163086,
+      "learning_rate": 9.271410948646925e-07,
+      "loss": 0.3104,
+      "step": 49685
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.1713764667510986,
+      "learning_rate": 9.269881622350896e-07,
+      "loss": 0.506,
+      "step": 49686
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.8979475498199463,
+      "learning_rate": 9.268352409310056e-07,
+      "loss": 0.5224,
+      "step": 49687
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.523115634918213,
+      "learning_rate": 9.266823309528677e-07,
+      "loss": 0.5432,
+      "step": 49688
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.770367383956909,
+      "learning_rate": 9.265294323010992e-07,
+      "loss": 0.586,
+      "step": 49689
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.946552276611328,
+      "learning_rate": 9.26376544976128e-07,
+      "loss": 0.568,
+      "step": 49690
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.3287651538848877,
+      "learning_rate": 9.262236689783765e-07,
+      "loss": 0.5707,
+      "step": 49691
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.5432732105255127,
+      "learning_rate": 9.260708043082717e-07,
+      "loss": 0.8177,
+      "step": 49692
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.8359861373901367,
+      "learning_rate": 9.25917950966237e-07,
+      "loss": 0.7401,
+      "step": 49693
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 4.244421005249023,
+      "learning_rate": 9.257651089526992e-07,
+      "loss": 0.5248,
+      "step": 49694
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.8353021144866943,
+      "learning_rate": 9.25612278268081e-07,
+      "loss": 0.382,
+      "step": 49695
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.633310079574585,
+      "learning_rate": 9.25459458912809e-07,
+      "loss": 0.6778,
+      "step": 49696
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.727573871612549,
+      "learning_rate": 9.253066508873093e-07,
+      "loss": 0.6315,
+      "step": 49697
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.204498052597046,
+      "learning_rate": 9.251538541920041e-07,
+      "loss": 0.5726,
+      "step": 49698
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.6248464584350586,
+      "learning_rate": 9.250010688273192e-07,
+      "loss": 0.5044,
+      "step": 49699
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.225532293319702,
+      "learning_rate": 9.248482947936815e-07,
+      "loss": 0.517,
+      "step": 49700
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.5282397270202637,
+      "learning_rate": 9.246955320915124e-07,
+      "loss": 0.5564,
+      "step": 49701
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.8416364192962646,
+      "learning_rate": 9.245427807212404e-07,
+      "loss": 0.5615,
+      "step": 49702
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.6497504711151123,
+      "learning_rate": 9.243900406832862e-07,
+      "loss": 0.7377,
+      "step": 49703
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.670917272567749,
+      "learning_rate": 9.242373119780778e-07,
+      "loss": 0.5432,
+      "step": 49704
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 4.642180919647217,
+      "learning_rate": 9.240845946060379e-07,
+      "loss": 0.479,
+      "step": 49705
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.3235132694244385,
+      "learning_rate": 9.23931888567593e-07,
+      "loss": 0.505,
+      "step": 49706
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.8918304443359375,
+      "learning_rate": 9.237791938631652e-07,
+      "loss": 0.6401,
+      "step": 49707
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.7672150135040283,
+      "learning_rate": 9.236265104931802e-07,
+      "loss": 0.7255,
+      "step": 49708
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 4.216385841369629,
+      "learning_rate": 9.234738384580644e-07,
+      "loss": 0.6692,
+      "step": 49709
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.4443917274475098,
+      "learning_rate": 9.233211777582396e-07,
+      "loss": 0.5383,
+      "step": 49710
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.8831772804260254,
+      "learning_rate": 9.231685283941321e-07,
+      "loss": 0.4661,
+      "step": 49711
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 4.563461780548096,
+      "learning_rate": 9.230158903661651e-07,
+      "loss": 0.7531,
+      "step": 49712
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.977820634841919,
+      "learning_rate": 9.228632636747648e-07,
+      "loss": 0.4712,
+      "step": 49713
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.469762086868286,
+      "learning_rate": 9.22710648320353e-07,
+      "loss": 0.6453,
+      "step": 49714
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.3370094299316406,
+      "learning_rate": 9.225580443033566e-07,
+      "loss": 0.8119,
+      "step": 49715
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.5717647075653076,
+      "learning_rate": 9.224054516241982e-07,
+      "loss": 0.4985,
+      "step": 49716
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.2159931659698486,
+      "learning_rate": 9.222528702833034e-07,
+      "loss": 0.7408,
+      "step": 49717
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.6755759716033936,
+      "learning_rate": 9.221003002810946e-07,
+      "loss": 0.607,
+      "step": 49718
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.152330160140991,
+      "learning_rate": 9.219477416179972e-07,
+      "loss": 0.4872,
+      "step": 49719
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.830458879470825,
+      "learning_rate": 9.217951942944359e-07,
+      "loss": 0.554,
+      "step": 49720
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.106722354888916,
+      "learning_rate": 9.216426583108356e-07,
+      "loss": 0.7415,
+      "step": 49721
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.1624228954315186,
+      "learning_rate": 9.21490133667618e-07,
+      "loss": 0.4176,
+      "step": 49722
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.23663592338562,
+      "learning_rate": 9.213376203652091e-07,
+      "loss": 0.681,
+      "step": 49723
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.4173858165740967,
+      "learning_rate": 9.211851184040322e-07,
+      "loss": 0.4937,
+      "step": 49724
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 1.9108330011367798,
+      "learning_rate": 9.210326277845117e-07,
+      "loss": 0.4438,
+      "step": 49725
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.24287486076355,
+      "learning_rate": 9.208801485070706e-07,
+      "loss": 0.4675,
+      "step": 49726
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.774850368499756,
+      "learning_rate": 9.207276805721355e-07,
+      "loss": 0.3915,
+      "step": 49727
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.9895365238189697,
+      "learning_rate": 9.205752239801268e-07,
+      "loss": 0.5245,
+      "step": 49728
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.8244428634643555,
+      "learning_rate": 9.204227787314718e-07,
+      "loss": 0.6789,
+      "step": 49729
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.68064284324646,
+      "learning_rate": 9.20270344826591e-07,
+      "loss": 0.7041,
+      "step": 49730
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.8653552532196045,
+      "learning_rate": 9.201179222659118e-07,
+      "loss": 0.5427,
+      "step": 49731
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 4.749105453491211,
+      "learning_rate": 9.199655110498551e-07,
+      "loss": 0.6663,
+      "step": 49732
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.5328259468078613,
+      "learning_rate": 9.198131111788467e-07,
+      "loss": 0.5359,
+      "step": 49733
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.949488639831543,
+      "learning_rate": 9.196607226533089e-07,
+      "loss": 0.6998,
+      "step": 49734
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.3585729598999023,
+      "learning_rate": 9.19508345473667e-07,
+      "loss": 0.6403,
+      "step": 49735
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.7891178131103516,
+      "learning_rate": 9.193559796403423e-07,
+      "loss": 0.5975,
+      "step": 49736
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.588702917098999,
+      "learning_rate": 9.192036251537617e-07,
+      "loss": 0.741,
+      "step": 49737
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.1451196670532227,
+      "learning_rate": 9.190512820143455e-07,
+      "loss": 0.5853,
+      "step": 49738
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.805967092514038,
+      "learning_rate": 9.188989502225193e-07,
+      "loss": 0.6556,
+      "step": 49739
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.5612235069274902,
+      "learning_rate": 9.187466297787062e-07,
+      "loss": 0.8411,
+      "step": 49740
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.6603949069976807,
+      "learning_rate": 9.185943206833308e-07,
+      "loss": 0.6802,
+      "step": 49741
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.4072093963623047,
+      "learning_rate": 9.184420229368146e-07,
+      "loss": 0.5992,
+      "step": 49742
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.545165777206421,
+      "learning_rate": 9.182897365395832e-07,
+      "loss": 0.3747,
+      "step": 49743
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.593579053878784,
+      "learning_rate": 9.181374614920579e-07,
+      "loss": 0.8028,
+      "step": 49744
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.814887523651123,
+      "learning_rate": 9.179851977946646e-07,
+      "loss": 0.6673,
+      "step": 49745
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.3649215698242188,
+      "learning_rate": 9.178329454478241e-07,
+      "loss": 0.5497,
+      "step": 49746
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.128185749053955,
+      "learning_rate": 9.17680704451962e-07,
+      "loss": 0.7068,
+      "step": 49747
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.9259567260742188,
+      "learning_rate": 9.175284748074998e-07,
+      "loss": 0.6377,
+      "step": 49748
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.5980186462402344,
+      "learning_rate": 9.173762565148625e-07,
+      "loss": 0.4483,
+      "step": 49749
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.1347944736480713,
+      "learning_rate": 9.172240495744711e-07,
+      "loss": 0.5926,
+      "step": 49750
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.4338953495025635,
+      "learning_rate": 9.170718539867513e-07,
+      "loss": 0.5071,
+      "step": 49751
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.838409185409546,
+      "learning_rate": 9.169196697521243e-07,
+      "loss": 0.6086,
+      "step": 49752
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.4718687534332275,
+      "learning_rate": 9.16767496871015e-07,
+      "loss": 0.4846,
+      "step": 49753
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.6920948028564453,
+      "learning_rate": 9.166153353438451e-07,
+      "loss": 0.4419,
+      "step": 49754
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.3283681869506836,
+      "learning_rate": 9.164631851710387e-07,
+      "loss": 0.6184,
+      "step": 49755
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.1022398471832275,
+      "learning_rate": 9.163110463530178e-07,
+      "loss": 0.5505,
+      "step": 49756
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.7022342681884766,
+      "learning_rate": 9.16158918890206e-07,
+      "loss": 0.5718,
+      "step": 49757
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.0821690559387207,
+      "learning_rate": 9.160068027830271e-07,
+      "loss": 0.4581,
+      "step": 49758
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.835716962814331,
+      "learning_rate": 9.158546980319027e-07,
+      "loss": 0.7801,
+      "step": 49759
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.856856346130371,
+      "learning_rate": 9.157026046372563e-07,
+      "loss": 0.7207,
+      "step": 49760
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 1.7019740343093872,
+      "learning_rate": 9.155505225995121e-07,
+      "loss": 0.6767,
+      "step": 49761
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.7185184955596924,
+      "learning_rate": 9.153984519190906e-07,
+      "loss": 0.6139,
+      "step": 49762
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.297579050064087,
+      "learning_rate": 9.152463925964173e-07,
+      "loss": 0.5104,
+      "step": 49763
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.905520439147949,
+      "learning_rate": 9.150943446319122e-07,
+      "loss": 0.7418,
+      "step": 49764
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.4211435317993164,
+      "learning_rate": 9.149423080260011e-07,
+      "loss": 0.5757,
+      "step": 49765
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.5949556827545166,
+      "learning_rate": 9.147902827791034e-07,
+      "loss": 0.4108,
+      "step": 49766
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 5.330845832824707,
+      "learning_rate": 9.14638268891645e-07,
+      "loss": 0.6295,
+      "step": 49767
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.1565439701080322,
+      "learning_rate": 9.144862663640458e-07,
+      "loss": 0.7439,
+      "step": 49768
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.0135536193847656,
+      "learning_rate": 9.143342751967299e-07,
+      "loss": 0.6245,
+      "step": 49769
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.0752720832824707,
+      "learning_rate": 9.141822953901214e-07,
+      "loss": 0.5415,
+      "step": 49770
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 4.144413948059082,
+      "learning_rate": 9.140303269446394e-07,
+      "loss": 0.6166,
+      "step": 49771
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.427750587463379,
+      "learning_rate": 9.138783698607101e-07,
+      "loss": 0.4812,
+      "step": 49772
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.2608728408813477,
+      "learning_rate": 9.137264241387534e-07,
+      "loss": 0.7227,
+      "step": 49773
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.0513010025024414,
+      "learning_rate": 9.135744897791937e-07,
+      "loss": 0.4905,
+      "step": 49774
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.052030086517334,
+      "learning_rate": 9.134225667824515e-07,
+      "loss": 0.6697,
+      "step": 49775
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.619171619415283,
+      "learning_rate": 9.132706551489512e-07,
+      "loss": 0.6508,
+      "step": 49776
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.0702104568481445,
+      "learning_rate": 9.131187548791132e-07,
+      "loss": 0.2884,
+      "step": 49777
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.198410749435425,
+      "learning_rate": 9.12966865973362e-07,
+      "loss": 0.5973,
+      "step": 49778
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.2677865028381348,
+      "learning_rate": 9.12814988432118e-07,
+      "loss": 0.3729,
+      "step": 49779
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.7339305877685547,
+      "learning_rate": 9.126631222558041e-07,
+      "loss": 0.4691,
+      "step": 49780
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.2171547412872314,
+      "learning_rate": 9.125112674448433e-07,
+      "loss": 0.6115,
+      "step": 49781
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.2263662815093994,
+      "learning_rate": 9.12359423999658e-07,
+      "loss": 0.4828,
+      "step": 49782
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.0517349243164062,
+      "learning_rate": 9.12207591920669e-07,
+      "loss": 0.755,
+      "step": 49783
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.882596969604492,
+      "learning_rate": 9.120557712083012e-07,
+      "loss": 0.5086,
+      "step": 49784
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.6604185104370117,
+      "learning_rate": 9.119039618629732e-07,
+      "loss": 0.7737,
+      "step": 49785
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.204383611679077,
+      "learning_rate": 9.117521638851102e-07,
+      "loss": 0.6541,
+      "step": 49786
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.851555585861206,
+      "learning_rate": 9.116003772751314e-07,
+      "loss": 0.5717,
+      "step": 49787
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.0213279724121094,
+      "learning_rate": 9.114486020334617e-07,
+      "loss": 0.698,
+      "step": 49788
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.62251877784729,
+      "learning_rate": 9.112968381605208e-07,
+      "loss": 0.3885,
+      "step": 49789
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.9202957153320312,
+      "learning_rate": 9.111450856567327e-07,
+      "loss": 0.6062,
+      "step": 49790
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.6091954708099365,
+      "learning_rate": 9.109933445225172e-07,
+      "loss": 0.7457,
+      "step": 49791
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.020054340362549,
+      "learning_rate": 9.108416147582994e-07,
+      "loss": 0.4283,
+      "step": 49792
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.6701576709747314,
+      "learning_rate": 9.106898963644972e-07,
+      "loss": 0.6657,
+      "step": 49793
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.4542102813720703,
+      "learning_rate": 9.10538189341536e-07,
+      "loss": 0.4876,
+      "step": 49794
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.084949493408203,
+      "learning_rate": 9.103864936898349e-07,
+      "loss": 0.649,
+      "step": 49795
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 1.9488531351089478,
+      "learning_rate": 9.102348094098179e-07,
+      "loss": 0.5227,
+      "step": 49796
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.0493600368499756,
+      "learning_rate": 9.100831365019047e-07,
+      "loss": 0.7964,
+      "step": 49797
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.1490485668182373,
+      "learning_rate": 9.0993147496652e-07,
+      "loss": 0.5779,
+      "step": 49798
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.116971492767334,
+      "learning_rate": 9.097798248040817e-07,
+      "loss": 0.5822,
+      "step": 49799
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.452489137649536,
+      "learning_rate": 9.096281860150141e-07,
+      "loss": 0.7279,
+      "step": 49800
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 4.661866188049316,
+      "learning_rate": 9.094765585997378e-07,
+      "loss": 0.6386,
+      "step": 49801
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.624861001968384,
+      "learning_rate": 9.093249425586759e-07,
+      "loss": 0.4896,
+      "step": 49802
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.459085702896118,
+      "learning_rate": 9.09173337892248e-07,
+      "loss": 0.4015,
+      "step": 49803
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.670426607131958,
+      "learning_rate": 9.090217446008781e-07,
+      "loss": 0.4711,
+      "step": 49804
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.397583246231079,
+      "learning_rate": 9.088701626849844e-07,
+      "loss": 0.6147,
+      "step": 49805
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 1.9924745559692383,
+      "learning_rate": 9.08718592144992e-07,
+      "loss": 0.552,
+      "step": 49806
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.4631285667419434,
+      "learning_rate": 9.08567032981319e-07,
+      "loss": 0.5637,
+      "step": 49807
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 1.829910397529602,
+      "learning_rate": 9.084154851943899e-07,
+      "loss": 0.4481,
+      "step": 49808
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.196681499481201,
+      "learning_rate": 9.082639487846229e-07,
+      "loss": 0.6364,
+      "step": 49809
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.5828187465667725,
+      "learning_rate": 9.081124237524424e-07,
+      "loss": 0.665,
+      "step": 49810
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 4.635394096374512,
+      "learning_rate": 9.079609100982678e-07,
+      "loss": 0.5416,
+      "step": 49811
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.2829225063323975,
+      "learning_rate": 9.078094078225213e-07,
+      "loss": 0.5571,
+      "step": 49812
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.9215517044067383,
+      "learning_rate": 9.076579169256233e-07,
+      "loss": 0.7707,
+      "step": 49813
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.291842460632324,
+      "learning_rate": 9.075064374079968e-07,
+      "loss": 0.494,
+      "step": 49814
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.9643051624298096,
+      "learning_rate": 9.0735496927006e-07,
+      "loss": 0.5781,
+      "step": 49815
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.182098388671875,
+      "learning_rate": 9.07203512512238e-07,
+      "loss": 0.5075,
+      "step": 49816
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.480499744415283,
+      "learning_rate": 9.070520671349481e-07,
+      "loss": 0.5546,
+      "step": 49817
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.5139834880828857,
+      "learning_rate": 9.069006331386143e-07,
+      "loss": 0.5531,
+      "step": 49818
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 4.095538139343262,
+      "learning_rate": 9.067492105236553e-07,
+      "loss": 0.671,
+      "step": 49819
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 4.300952434539795,
+      "learning_rate": 9.065977992904934e-07,
+      "loss": 0.7802,
+      "step": 49820
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 1.7852457761764526,
+      "learning_rate": 9.0644639943955e-07,
+      "loss": 0.3879,
+      "step": 49821
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.9119350910186768,
+      "learning_rate": 9.062950109712465e-07,
+      "loss": 0.464,
+      "step": 49822
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.265990972518921,
+      "learning_rate": 9.06143633886002e-07,
+      "loss": 0.5766,
+      "step": 49823
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.072563409805298,
+      "learning_rate": 9.059922681842392e-07,
+      "loss": 0.5108,
+      "step": 49824
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.704723358154297,
+      "learning_rate": 9.058409138663776e-07,
+      "loss": 0.6202,
+      "step": 49825
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.9659855365753174,
+      "learning_rate": 9.056895709328401e-07,
+      "loss": 0.6285,
+      "step": 49826
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.401124954223633,
+      "learning_rate": 9.055382393840445e-07,
+      "loss": 0.2413,
+      "step": 49827
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.877532482147217,
+      "learning_rate": 9.053869192204145e-07,
+      "loss": 0.5659,
+      "step": 49828
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.646216869354248,
+      "learning_rate": 9.052356104423682e-07,
+      "loss": 0.5597,
+      "step": 49829
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.4769303798675537,
+      "learning_rate": 9.05084313050329e-07,
+      "loss": 0.6254,
+      "step": 49830
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.360250473022461,
+      "learning_rate": 9.049330270447155e-07,
+      "loss": 0.5353,
+      "step": 49831
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.5320487022399902,
+      "learning_rate": 9.047817524259489e-07,
+      "loss": 0.4061,
+      "step": 49832
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.6899449825286865,
+      "learning_rate": 9.046304891944513e-07,
+      "loss": 0.6289,
+      "step": 49833
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.5436103343963623,
+      "learning_rate": 9.044792373506411e-07,
+      "loss": 0.6634,
+      "step": 49834
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.7250750064849854,
+      "learning_rate": 9.043279968949414e-07,
+      "loss": 0.793,
+      "step": 49835
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.8584649562835693,
+      "learning_rate": 9.041767678277697e-07,
+      "loss": 0.834,
+      "step": 49836
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.428065299987793,
+      "learning_rate": 9.040255501495493e-07,
+      "loss": 0.5757,
+      "step": 49837
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.86696457862854,
+      "learning_rate": 9.038743438606984e-07,
+      "loss": 0.7048,
+      "step": 49838
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.092507839202881,
+      "learning_rate": 9.037231489616394e-07,
+      "loss": 0.685,
+      "step": 49839
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.3223540782928467,
+      "learning_rate": 9.035719654527909e-07,
+      "loss": 0.4708,
+      "step": 49840
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.3508076667785645,
+      "learning_rate": 9.03420793334574e-07,
+      "loss": 0.7426,
+      "step": 49841
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.7765679359436035,
+      "learning_rate": 9.032696326074097e-07,
+      "loss": 0.518,
+      "step": 49842
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.3424675464630127,
+      "learning_rate": 9.031184832717188e-07,
+      "loss": 0.8931,
+      "step": 49843
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.1442136764526367,
+      "learning_rate": 9.029673453279192e-07,
+      "loss": 0.7623,
+      "step": 49844
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.0018718242645264,
+      "learning_rate": 9.028162187764338e-07,
+      "loss": 0.4963,
+      "step": 49845
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.7779221534729004,
+      "learning_rate": 9.026651036176809e-07,
+      "loss": 0.2809,
+      "step": 49846
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.6891558170318604,
+      "learning_rate": 9.025139998520826e-07,
+      "loss": 0.652,
+      "step": 49847
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.016695022583008,
+      "learning_rate": 9.023629074800566e-07,
+      "loss": 0.5485,
+      "step": 49848
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 4.919621467590332,
+      "learning_rate": 9.022118265020258e-07,
+      "loss": 0.6083,
+      "step": 49849
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.4368443489074707,
+      "learning_rate": 9.02060756918407e-07,
+      "loss": 0.5472,
+      "step": 49850
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.3730452060699463,
+      "learning_rate": 9.019096987296238e-07,
+      "loss": 0.763,
+      "step": 49851
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.6697909832000732,
+      "learning_rate": 9.017586519360932e-07,
+      "loss": 0.607,
+      "step": 49852
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.1554622650146484,
+      "learning_rate": 9.016076165382382e-07,
+      "loss": 0.5363,
+      "step": 49853
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.537475824356079,
+      "learning_rate": 9.014565925364755e-07,
+      "loss": 0.6615,
+      "step": 49854
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 4.2412238121032715,
+      "learning_rate": 9.013055799312276e-07,
+      "loss": 0.8725,
+      "step": 49855
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.086590051651001,
+      "learning_rate": 9.011545787229126e-07,
+      "loss": 0.5478,
+      "step": 49856
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.114847183227539,
+      "learning_rate": 9.010035889119522e-07,
+      "loss": 0.4491,
+      "step": 49857
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.0951104164123535,
+      "learning_rate": 9.008526104987647e-07,
+      "loss": 0.4978,
+      "step": 49858
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 1.9846343994140625,
+      "learning_rate": 9.007016434837712e-07,
+      "loss": 0.5787,
+      "step": 49859
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.219717025756836,
+      "learning_rate": 9.005506878673897e-07,
+      "loss": 0.3142,
+      "step": 49860
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 1.7954294681549072,
+      "learning_rate": 9.003997436500412e-07,
+      "loss": 0.412,
+      "step": 49861
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.9859087467193604,
+      "learning_rate": 9.00248810832145e-07,
+      "loss": 0.6145,
+      "step": 49862
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.3918325901031494,
+      "learning_rate": 9.000978894141227e-07,
+      "loss": 0.8221,
+      "step": 49863
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.1763062477111816,
+      "learning_rate": 8.999469793963906e-07,
+      "loss": 0.5558,
+      "step": 49864
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.06191349029541,
+      "learning_rate": 8.997960807793716e-07,
+      "loss": 0.4248,
+      "step": 49865
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.3770041465759277,
+      "learning_rate": 8.99645193563482e-07,
+      "loss": 0.3968,
+      "step": 49866
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.2177295684814453,
+      "learning_rate": 8.99494317749145e-07,
+      "loss": 0.4894,
+      "step": 49867
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.0206449031829834,
+      "learning_rate": 8.993434533367768e-07,
+      "loss": 0.5595,
+      "step": 49868
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.741455316543579,
+      "learning_rate": 8.991926003267992e-07,
+      "loss": 0.5301,
+      "step": 49869
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.8229928016662598,
+      "learning_rate": 8.990417587196298e-07,
+      "loss": 0.6878,
+      "step": 49870
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.2685186862945557,
+      "learning_rate": 8.988909285156905e-07,
+      "loss": 0.5605,
+      "step": 49871
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.8984177112579346,
+      "learning_rate": 8.987401097153974e-07,
+      "loss": 0.5587,
+      "step": 49872
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.229008197784424,
+      "learning_rate": 8.985893023191732e-07,
+      "loss": 0.6254,
+      "step": 49873
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.1780073642730713,
+      "learning_rate": 8.984385063274347e-07,
+      "loss": 0.5341,
+      "step": 49874
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.0368309020996094,
+      "learning_rate": 8.982877217406033e-07,
+      "loss": 0.7758,
+      "step": 49875
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 1.9956456422805786,
+      "learning_rate": 8.981369485590958e-07,
+      "loss": 0.6707,
+      "step": 49876
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.2069101333618164,
+      "learning_rate": 8.979861867833345e-07,
+      "loss": 0.7658,
+      "step": 49877
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.179084062576294,
+      "learning_rate": 8.978354364137354e-07,
+      "loss": 0.625,
+      "step": 49878
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.9682910442352295,
+      "learning_rate": 8.976846974507203e-07,
+      "loss": 0.6687,
+      "step": 49879
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 1.7687562704086304,
+      "learning_rate": 8.975339698947067e-07,
+      "loss": 0.433,
+      "step": 49880
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.313481092453003,
+      "learning_rate": 8.973832537461135e-07,
+      "loss": 0.4871,
+      "step": 49881
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.8413023948669434,
+      "learning_rate": 8.972325490053613e-07,
+      "loss": 0.5943,
+      "step": 49882
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.2900750637054443,
+      "learning_rate": 8.970818556728694e-07,
+      "loss": 0.5446,
+      "step": 49883
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.223386764526367,
+      "learning_rate": 8.969311737490544e-07,
+      "loss": 0.6888,
+      "step": 49884
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.275442123413086,
+      "learning_rate": 8.967805032343385e-07,
+      "loss": 0.6044,
+      "step": 49885
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.1314167976379395,
+      "learning_rate": 8.966298441291371e-07,
+      "loss": 0.4367,
+      "step": 49886
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 4.026103496551514,
+      "learning_rate": 8.964791964338725e-07,
+      "loss": 0.6758,
+      "step": 49887
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.334545373916626,
+      "learning_rate": 8.963285601489613e-07,
+      "loss": 0.4451,
+      "step": 49888
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.3844895362854004,
+      "learning_rate": 8.961779352748234e-07,
+      "loss": 0.8367,
+      "step": 49889
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.1215522289276123,
+      "learning_rate": 8.960273218118765e-07,
+      "loss": 0.6131,
+      "step": 49890
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.1777191162109375,
+      "learning_rate": 8.958767197605417e-07,
+      "loss": 0.6637,
+      "step": 49891
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.9899842739105225,
+      "learning_rate": 8.95726129121235e-07,
+      "loss": 0.4747,
+      "step": 49892
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.9106202125549316,
+      "learning_rate": 8.955755498943763e-07,
+      "loss": 0.7026,
+      "step": 49893
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.102848768234253,
+      "learning_rate": 8.954249820803857e-07,
+      "loss": 0.504,
+      "step": 49894
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.425018310546875,
+      "learning_rate": 8.952744256796791e-07,
+      "loss": 0.6306,
+      "step": 49895
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.3570098876953125,
+      "learning_rate": 8.951238806926782e-07,
+      "loss": 0.442,
+      "step": 49896
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.464268922805786,
+      "learning_rate": 8.949733471197991e-07,
+      "loss": 0.4064,
+      "step": 49897
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 5.425251483917236,
+      "learning_rate": 8.948228249614621e-07,
+      "loss": 0.7932,
+      "step": 49898
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.342897653579712,
+      "learning_rate": 8.946723142180841e-07,
+      "loss": 0.5394,
+      "step": 49899
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 4.396297931671143,
+      "learning_rate": 8.945218148900858e-07,
+      "loss": 0.7944,
+      "step": 49900
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.6786627769470215,
+      "learning_rate": 8.943713269778825e-07,
+      "loss": 0.9937,
+      "step": 49901
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.5012946128845215,
+      "learning_rate": 8.94220850481895e-07,
+      "loss": 0.6188,
+      "step": 49902
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.502454996109009,
+      "learning_rate": 8.940703854025429e-07,
+      "loss": 0.6475,
+      "step": 49903
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.660658359527588,
+      "learning_rate": 8.939199317402408e-07,
+      "loss": 0.6326,
+      "step": 49904
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.0272717475891113,
+      "learning_rate": 8.937694894954097e-07,
+      "loss": 0.5528,
+      "step": 49905
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.421114921569824,
+      "learning_rate": 8.93619058668469e-07,
+      "loss": 0.4415,
+      "step": 49906
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.6355490684509277,
+      "learning_rate": 8.934686392598336e-07,
+      "loss": 0.7263,
+      "step": 49907
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.0148189067840576,
+      "learning_rate": 8.933182312699251e-07,
+      "loss": 0.7203,
+      "step": 49908
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.8498902320861816,
+      "learning_rate": 8.931678346991584e-07,
+      "loss": 0.7501,
+      "step": 49909
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.217717170715332,
+      "learning_rate": 8.930174495479554e-07,
+      "loss": 0.5807,
+      "step": 49910
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.475104331970215,
+      "learning_rate": 8.928670758167307e-07,
+      "loss": 0.5399,
+      "step": 49911
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.9429235458374023,
+      "learning_rate": 8.927167135059056e-07,
+      "loss": 0.6544,
+      "step": 49912
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.3160245418548584,
+      "learning_rate": 8.925663626158959e-07,
+      "loss": 0.6236,
+      "step": 49913
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.0752360820770264,
+      "learning_rate": 8.924160231471207e-07,
+      "loss": 0.1991,
+      "step": 49914
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.5160903930664062,
+      "learning_rate": 8.922656950999975e-07,
+      "loss": 0.6256,
+      "step": 49915
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.489612340927124,
+      "learning_rate": 8.921153784749453e-07,
+      "loss": 0.75,
+      "step": 49916
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.3897268772125244,
+      "learning_rate": 8.9196507327238e-07,
+      "loss": 0.4853,
+      "step": 49917
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.7086431980133057,
+      "learning_rate": 8.918147794927229e-07,
+      "loss": 0.5982,
+      "step": 49918
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.53041410446167,
+      "learning_rate": 8.91664497136388e-07,
+      "loss": 0.7188,
+      "step": 49919
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.051499605178833,
+      "learning_rate": 8.915142262037968e-07,
+      "loss": 0.3561,
+      "step": 49920
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 4.199395656585693,
+      "learning_rate": 8.913639666953639e-07,
+      "loss": 0.7158,
+      "step": 49921
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.6472184658050537,
+      "learning_rate": 8.912137186115088e-07,
+      "loss": 0.5523,
+      "step": 49922
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.4442875385284424,
+      "learning_rate": 8.910634819526493e-07,
+      "loss": 0.6868,
+      "step": 49923
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.1274025440216064,
+      "learning_rate": 8.909132567192042e-07,
+      "loss": 0.5336,
+      "step": 49924
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.814554452896118,
+      "learning_rate": 8.907630429115882e-07,
+      "loss": 0.9561,
+      "step": 49925
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.945774793624878,
+      "learning_rate": 8.906128405302228e-07,
+      "loss": 0.6662,
+      "step": 49926
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 4.358288764953613,
+      "learning_rate": 8.90462649575522e-07,
+      "loss": 1.0971,
+      "step": 49927
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.5144693851470947,
+      "learning_rate": 8.903124700479066e-07,
+      "loss": 0.5422,
+      "step": 49928
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.1141483783721924,
+      "learning_rate": 8.901623019477918e-07,
+      "loss": 0.5643,
+      "step": 49929
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.4018115997314453,
+      "learning_rate": 8.900121452755966e-07,
+      "loss": 0.8762,
+      "step": 49930
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.348891019821167,
+      "learning_rate": 8.898620000317371e-07,
+      "loss": 0.5662,
+      "step": 49931
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.5153067111968994,
+      "learning_rate": 8.897118662166327e-07,
+      "loss": 0.3828,
+      "step": 49932
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.7876100540161133,
+      "learning_rate": 8.895617438306991e-07,
+      "loss": 0.7258,
+      "step": 49933
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.7490880489349365,
+      "learning_rate": 8.894116328743551e-07,
+      "loss": 0.4506,
+      "step": 49934
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.322927713394165,
+      "learning_rate": 8.89261533348017e-07,
+      "loss": 0.5495,
+      "step": 49935
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.044640302658081,
+      "learning_rate": 8.891114452521033e-07,
+      "loss": 0.4319,
+      "step": 49936
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.4531586170196533,
+      "learning_rate": 8.889613685870296e-07,
+      "loss": 0.4413,
+      "step": 49937
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.3803701400756836,
+      "learning_rate": 8.888113033532153e-07,
+      "loss": 0.3906,
+      "step": 49938
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 4.315629959106445,
+      "learning_rate": 8.88661249551076e-07,
+      "loss": 0.7547,
+      "step": 49939
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.6490883827209473,
+      "learning_rate": 8.885112071810304e-07,
+      "loss": 0.7504,
+      "step": 49940
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 4.350809574127197,
+      "learning_rate": 8.88361176243494e-07,
+      "loss": 0.7523,
+      "step": 49941
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.023421049118042,
+      "learning_rate": 8.882111567388846e-07,
+      "loss": 0.4539,
+      "step": 49942
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.2858409881591797,
+      "learning_rate": 8.880611486676199e-07,
+      "loss": 0.2842,
+      "step": 49943
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.100379228591919,
+      "learning_rate": 8.879111520301182e-07,
+      "loss": 0.5525,
+      "step": 49944
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.9726898670196533,
+      "learning_rate": 8.877611668267933e-07,
+      "loss": 0.4988,
+      "step": 49945
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.112367630004883,
+      "learning_rate": 8.876111930580655e-07,
+      "loss": 0.3896,
+      "step": 49946
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.597860813140869,
+      "learning_rate": 8.874612307243496e-07,
+      "loss": 0.6257,
+      "step": 49947
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.771563768386841,
+      "learning_rate": 8.873112798260646e-07,
+      "loss": 0.5434,
+      "step": 49948
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 4.342309951782227,
+      "learning_rate": 8.871613403636248e-07,
+      "loss": 0.7849,
+      "step": 49949
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 1.7614655494689941,
+      "learning_rate": 8.870114123374501e-07,
+      "loss": 0.3824,
+      "step": 49950
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.408778429031372,
+      "learning_rate": 8.868614957479544e-07,
+      "loss": 0.6882,
+      "step": 49951
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.476351737976074,
+      "learning_rate": 8.867115905955576e-07,
+      "loss": 0.3171,
+      "step": 49952
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.319099187850952,
+      "learning_rate": 8.865616968806734e-07,
+      "loss": 0.424,
+      "step": 49953
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 4.378447532653809,
+      "learning_rate": 8.864118146037204e-07,
+      "loss": 0.5423,
+      "step": 49954
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 4.047100067138672,
+      "learning_rate": 8.862619437651165e-07,
+      "loss": 0.532,
+      "step": 49955
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.9275059700012207,
+      "learning_rate": 8.861120843652759e-07,
+      "loss": 0.5076,
+      "step": 49956
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 4.479065418243408,
+      "learning_rate": 8.859622364046172e-07,
+      "loss": 0.5462,
+      "step": 49957
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.1132917404174805,
+      "learning_rate": 8.858123998835555e-07,
+      "loss": 0.6088,
+      "step": 49958
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.461798667907715,
+      "learning_rate": 8.8566257480251e-07,
+      "loss": 0.6892,
+      "step": 49959
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.8315134048461914,
+      "learning_rate": 8.855127611618936e-07,
+      "loss": 0.8491,
+      "step": 49960
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.414163589477539,
+      "learning_rate": 8.853629589621265e-07,
+      "loss": 0.705,
+      "step": 49961
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.2266147136688232,
+      "learning_rate": 8.852131682036225e-07,
+      "loss": 0.5866,
+      "step": 49962
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.9773828983306885,
+      "learning_rate": 8.850633888867988e-07,
+      "loss": 0.8956,
+      "step": 49963
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.750990390777588,
+      "learning_rate": 8.84913621012074e-07,
+      "loss": 0.7201,
+      "step": 49964
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.208054304122925,
+      "learning_rate": 8.847638645798618e-07,
+      "loss": 0.6075,
+      "step": 49965
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.570777177810669,
+      "learning_rate": 8.846141195905788e-07,
+      "loss": 0.425,
+      "step": 49966
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.5269358158111572,
+      "learning_rate": 8.844643860446445e-07,
+      "loss": 0.3876,
+      "step": 49967
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.669428825378418,
+      "learning_rate": 8.84314663942471e-07,
+      "loss": 0.7314,
+      "step": 49968
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.442891836166382,
+      "learning_rate": 8.841649532844787e-07,
+      "loss": 0.5843,
+      "step": 49969
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.4459574222564697,
+      "learning_rate": 8.8401525407108e-07,
+      "loss": 0.6411,
+      "step": 49970
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.362950086593628,
+      "learning_rate": 8.838655663026946e-07,
+      "loss": 0.5181,
+      "step": 49971
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.864790439605713,
+      "learning_rate": 8.837158899797354e-07,
+      "loss": 0.5917,
+      "step": 49972
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.0681185722351074,
+      "learning_rate": 8.835662251026217e-07,
+      "loss": 0.6729,
+      "step": 49973
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.657712697982788,
+      "learning_rate": 8.834165716717674e-07,
+      "loss": 0.6006,
+      "step": 49974
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 1.9636365175247192,
+      "learning_rate": 8.832669296875906e-07,
+      "loss": 0.5863,
+      "step": 49975
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.4165518283843994,
+      "learning_rate": 8.83117299150505e-07,
+      "loss": 0.5674,
+      "step": 49976
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.709613084793091,
+      "learning_rate": 8.829676800609288e-07,
+      "loss": 0.4454,
+      "step": 49977
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 4.043759822845459,
+      "learning_rate": 8.828180724192764e-07,
+      "loss": 0.8051,
+      "step": 49978
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.2380003929138184,
+      "learning_rate": 8.826684762259663e-07,
+      "loss": 0.6358,
+      "step": 49979
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.4830918312072754,
+      "learning_rate": 8.825188914814109e-07,
+      "loss": 0.4741,
+      "step": 49980
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 5.139130592346191,
+      "learning_rate": 8.823693181860293e-07,
+      "loss": 0.9656,
+      "step": 49981
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.1132185459136963,
+      "learning_rate": 8.822197563402351e-07,
+      "loss": 0.4573,
+      "step": 49982
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.8937318325042725,
+      "learning_rate": 8.820702059444447e-07,
+      "loss": 0.7157,
+      "step": 49983
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.475943088531494,
+      "learning_rate": 8.819206669990748e-07,
+      "loss": 0.5798,
+      "step": 49984
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.9158830642700195,
+      "learning_rate": 8.817711395045425e-07,
+      "loss": 0.6178,
+      "step": 49985
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.4844624996185303,
+      "learning_rate": 8.816216234612601e-07,
+      "loss": 0.4739,
+      "step": 49986
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.3964617252349854,
+      "learning_rate": 8.814721188696467e-07,
+      "loss": 0.5446,
+      "step": 49987
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.2155678272247314,
+      "learning_rate": 8.813226257301156e-07,
+      "loss": 0.8475,
+      "step": 49988
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 4.162566661834717,
+      "learning_rate": 8.811731440430837e-07,
+      "loss": 0.7092,
+      "step": 49989
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.968984603881836,
+      "learning_rate": 8.810236738089661e-07,
+      "loss": 0.6768,
+      "step": 49990
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.890110492706299,
+      "learning_rate": 8.808742150281791e-07,
+      "loss": 0.7072,
+      "step": 49991
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.8313241004943848,
+      "learning_rate": 8.807247677011371e-07,
+      "loss": 0.5075,
+      "step": 49992
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.3185110092163086,
+      "learning_rate": 8.805753318282578e-07,
+      "loss": 0.442,
+      "step": 49993
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.054795026779175,
+      "learning_rate": 8.804259074099536e-07,
+      "loss": 0.6441,
+      "step": 49994
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.714174509048462,
+      "learning_rate": 8.802764944466429e-07,
+      "loss": 0.621,
+      "step": 49995
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.6386802196502686,
+      "learning_rate": 8.80127092938739e-07,
+      "loss": 0.6945,
+      "step": 49996
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.6717615127563477,
+      "learning_rate": 8.799777028866596e-07,
+      "loss": 0.7867,
+      "step": 49997
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.198235511779785,
+      "learning_rate": 8.798283242908179e-07,
+      "loss": 0.5869,
+      "step": 49998
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 5.047091007232666,
+      "learning_rate": 8.796789571516307e-07,
+      "loss": 0.6921,
+      "step": 49999
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 4.275241851806641,
+      "learning_rate": 8.795296014695121e-07,
+      "loss": 0.6315,
+      "step": 50000
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.000673294067383,
+      "learning_rate": 8.79380257244879e-07,
+      "loss": 0.5469,
+      "step": 50001
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.4100654125213623,
+      "learning_rate": 8.792309244781449e-07,
+      "loss": 0.4135,
+      "step": 50002
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.998922109603882,
+      "learning_rate": 8.790816031697258e-07,
+      "loss": 0.6708,
+      "step": 50003
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.43697452545166,
+      "learning_rate": 8.789322933200367e-07,
+      "loss": 0.5403,
+      "step": 50004
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.0923690795898438,
+      "learning_rate": 8.787829949294946e-07,
+      "loss": 0.4054,
+      "step": 50005
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.033384323120117,
+      "learning_rate": 8.786337079985119e-07,
+      "loss": 0.7602,
+      "step": 50006
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.7218270301818848,
+      "learning_rate": 8.784844325275066e-07,
+      "loss": 0.6629,
+      "step": 50007
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.6321916580200195,
+      "learning_rate": 8.7833516851689e-07,
+      "loss": 0.8314,
+      "step": 50008
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.2062816619873047,
+      "learning_rate": 8.781859159670814e-07,
+      "loss": 0.6613,
+      "step": 50009
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.322343349456787,
+      "learning_rate": 8.780366748784919e-07,
+      "loss": 0.4314,
+      "step": 50010
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.5704915523529053,
+      "learning_rate": 8.778874452515402e-07,
+      "loss": 0.6279,
+      "step": 50011
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.3998005390167236,
+      "learning_rate": 8.777382270866374e-07,
+      "loss": 0.5848,
+      "step": 50012
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.4534103870391846,
+      "learning_rate": 8.775890203842019e-07,
+      "loss": 0.4713,
+      "step": 50013
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.61624813079834,
+      "learning_rate": 8.774398251446453e-07,
+      "loss": 0.6102,
+      "step": 50014
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.5662755966186523,
+      "learning_rate": 8.772906413683863e-07,
+      "loss": 0.7367,
+      "step": 50015
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.8283724784851074,
+      "learning_rate": 8.771414690558356e-07,
+      "loss": 0.6421,
+      "step": 50016
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.381474733352661,
+      "learning_rate": 8.769923082074105e-07,
+      "loss": 0.4994,
+      "step": 50017
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.6161184310913086,
+      "learning_rate": 8.768431588235266e-07,
+      "loss": 0.5184,
+      "step": 50018
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.309276819229126,
+      "learning_rate": 8.766940209045954e-07,
+      "loss": 0.5327,
+      "step": 50019
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 1.9379829168319702,
+      "learning_rate": 8.765448944510357e-07,
+      "loss": 0.3699,
+      "step": 50020
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 4.7270612716674805,
+      "learning_rate": 8.763957794632582e-07,
+      "loss": 0.5944,
+      "step": 50021
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.049760580062866,
+      "learning_rate": 8.762466759416805e-07,
+      "loss": 0.6218,
+      "step": 50022
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.0738000869750977,
+      "learning_rate": 8.760975838867147e-07,
+      "loss": 0.5521,
+      "step": 50023
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.8449807167053223,
+      "learning_rate": 8.759485032987769e-07,
+      "loss": 0.7119,
+      "step": 50024
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.077012538909912,
+      "learning_rate": 8.757994341782821e-07,
+      "loss": 0.6292,
+      "step": 50025
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.490730047225952,
+      "learning_rate": 8.756503765256436e-07,
+      "loss": 0.4869,
+      "step": 50026
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.7591392993927,
+      "learning_rate": 8.755013303412762e-07,
+      "loss": 0.5848,
+      "step": 50027
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.573383092880249,
+      "learning_rate": 8.753522956255955e-07,
+      "loss": 0.4971,
+      "step": 50028
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.342277765274048,
+      "learning_rate": 8.752032723790138e-07,
+      "loss": 0.5361,
+      "step": 50029
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.406425714492798,
+      "learning_rate": 8.75054260601948e-07,
+      "loss": 0.5561,
+      "step": 50030
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.34592604637146,
+      "learning_rate": 8.749052602948099e-07,
+      "loss": 0.7457,
+      "step": 50031
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.4701805114746094,
+      "learning_rate": 8.74756271458016e-07,
+      "loss": 0.4695,
+      "step": 50032
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.066331148147583,
+      "learning_rate": 8.746072940919781e-07,
+      "loss": 0.6352,
+      "step": 50033
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.6508519649505615,
+      "learning_rate": 8.744583281971137e-07,
+      "loss": 0.5319,
+      "step": 50034
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.0056989192962646,
+      "learning_rate": 8.743093737738334e-07,
+      "loss": 0.4643,
+      "step": 50035
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.6396002769470215,
+      "learning_rate": 8.741604308225548e-07,
+      "loss": 0.823,
+      "step": 50036
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.9859089851379395,
+      "learning_rate": 8.740114993436893e-07,
+      "loss": 0.7157,
+      "step": 50037
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.651440382003784,
+      "learning_rate": 8.738625793376532e-07,
+      "loss": 0.5317,
+      "step": 50038
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 4.002017974853516,
+      "learning_rate": 8.737136708048588e-07,
+      "loss": 0.7866,
+      "step": 50039
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 1.9328861236572266,
+      "learning_rate": 8.735647737457215e-07,
+      "loss": 0.4629,
+      "step": 50040
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.645716905593872,
+      "learning_rate": 8.734158881606541e-07,
+      "loss": 0.5233,
+      "step": 50041
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.019483804702759,
+      "learning_rate": 8.732670140500721e-07,
+      "loss": 0.6511,
+      "step": 50042
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.6389544010162354,
+      "learning_rate": 8.731181514143877e-07,
+      "loss": 0.3631,
+      "step": 50043
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.235671281814575,
+      "learning_rate": 8.729693002540157e-07,
+      "loss": 0.8389,
+      "step": 50044
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.1789283752441406,
+      "learning_rate": 8.728204605693696e-07,
+      "loss": 0.5597,
+      "step": 50045
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.4990992546081543,
+      "learning_rate": 8.726716323608653e-07,
+      "loss": 0.572,
+      "step": 50046
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 4.365476608276367,
+      "learning_rate": 8.725228156289139e-07,
+      "loss": 0.4143,
+      "step": 50047
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.8921735286712646,
+      "learning_rate": 8.723740103739314e-07,
+      "loss": 0.6133,
+      "step": 50048
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.021211862564087,
+      "learning_rate": 8.722252165963296e-07,
+      "loss": 0.6269,
+      "step": 50049
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.5463855266571045,
+      "learning_rate": 8.720764342965244e-07,
+      "loss": 0.5845,
+      "step": 50050
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.335376024246216,
+      "learning_rate": 8.719276634749269e-07,
+      "loss": 0.4626,
+      "step": 50051
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.6135284900665283,
+      "learning_rate": 8.717789041319535e-07,
+      "loss": 0.7395,
+      "step": 50052
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.4121696949005127,
+      "learning_rate": 8.716301562680152e-07,
+      "loss": 0.6335,
+      "step": 50053
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.943828582763672,
+      "learning_rate": 8.714814198835281e-07,
+      "loss": 0.4625,
+      "step": 50054
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.445631504058838,
+      "learning_rate": 8.713326949789031e-07,
+      "loss": 0.6257,
+      "step": 50055
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.268378734588623,
+      "learning_rate": 8.71183981554557e-07,
+      "loss": 0.5516,
+      "step": 50056
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.361386299133301,
+      "learning_rate": 8.710352796108995e-07,
+      "loss": 0.7655,
+      "step": 50057
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.463785409927368,
+      "learning_rate": 8.708865891483481e-07,
+      "loss": 0.6307,
+      "step": 50058
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.739203453063965,
+      "learning_rate": 8.70737910167313e-07,
+      "loss": 0.613,
+      "step": 50059
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.355156660079956,
+      "learning_rate": 8.7058924266821e-07,
+      "loss": 0.515,
+      "step": 50060
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.8979578018188477,
+      "learning_rate": 8.704405866514504e-07,
+      "loss": 0.7807,
+      "step": 50061
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.6130878925323486,
+      "learning_rate": 8.702919421174494e-07,
+      "loss": 0.6149,
+      "step": 50062
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.7988719940185547,
+      "learning_rate": 8.701433090666184e-07,
+      "loss": 0.8738,
+      "step": 50063
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.1455695629119873,
+      "learning_rate": 8.69994687499372e-07,
+      "loss": 0.6583,
+      "step": 50064
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.9652352333068848,
+      "learning_rate": 8.698460774161227e-07,
+      "loss": 0.7286,
+      "step": 50065
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.528024435043335,
+      "learning_rate": 8.696974788172863e-07,
+      "loss": 0.5083,
+      "step": 50066
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.952342987060547,
+      "learning_rate": 8.695488917032719e-07,
+      "loss": 0.6546,
+      "step": 50067
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.293137311935425,
+      "learning_rate": 8.694003160744968e-07,
+      "loss": 0.6518,
+      "step": 50068
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.6285579204559326,
+      "learning_rate": 8.692517519313703e-07,
+      "loss": 0.506,
+      "step": 50069
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.902174234390259,
+      "learning_rate": 8.691031992743082e-07,
+      "loss": 0.6509,
+      "step": 50070
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.93784761428833,
+      "learning_rate": 8.689546581037223e-07,
+      "loss": 0.7982,
+      "step": 50071
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.3240790367126465,
+      "learning_rate": 8.688061284200266e-07,
+      "loss": 0.7519,
+      "step": 50072
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.338223457336426,
+      "learning_rate": 8.686576102236322e-07,
+      "loss": 0.6073,
+      "step": 50073
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.8385376930236816,
+      "learning_rate": 8.685091035149545e-07,
+      "loss": 0.5178,
+      "step": 50074
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.865091323852539,
+      "learning_rate": 8.683606082944046e-07,
+      "loss": 0.5839,
+      "step": 50075
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.778411626815796,
+      "learning_rate": 8.68212124562397e-07,
+      "loss": 0.5038,
+      "step": 50076
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 4.460749626159668,
+      "learning_rate": 8.680636523193425e-07,
+      "loss": 0.7313,
+      "step": 50077
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.760266065597534,
+      "learning_rate": 8.679151915656548e-07,
+      "loss": 0.5761,
+      "step": 50078
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.7866406440734863,
+      "learning_rate": 8.677667423017489e-07,
+      "loss": 0.4005,
+      "step": 50079
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.338963747024536,
+      "learning_rate": 8.67618304528034e-07,
+      "loss": 0.4871,
+      "step": 50080
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.4335134029388428,
+      "learning_rate": 8.674698782449259e-07,
+      "loss": 0.4805,
+      "step": 50081
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.019275665283203,
+      "learning_rate": 8.673214634528349e-07,
+      "loss": 0.5078,
+      "step": 50082
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.5232784748077393,
+      "learning_rate": 8.671730601521755e-07,
+      "loss": 0.831,
+      "step": 50083
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.015641450881958,
+      "learning_rate": 8.670246683433592e-07,
+      "loss": 0.7242,
+      "step": 50084
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 4.009756565093994,
+      "learning_rate": 8.668762880267984e-07,
+      "loss": 0.4922,
+      "step": 50085
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.117250680923462,
+      "learning_rate": 8.667279192029076e-07,
+      "loss": 0.5765,
+      "step": 50086
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.7442727088928223,
+      "learning_rate": 8.665795618720974e-07,
+      "loss": 0.5246,
+      "step": 50087
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.042816638946533,
+      "learning_rate": 8.664312160347821e-07,
+      "loss": 0.5196,
+      "step": 50088
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 4.116172790527344,
+      "learning_rate": 8.662828816913715e-07,
+      "loss": 0.8411,
+      "step": 50089
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.3284523487091064,
+      "learning_rate": 8.661345588422798e-07,
+      "loss": 0.6889,
+      "step": 50090
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 4.253186225891113,
+      "learning_rate": 8.659862474879204e-07,
+      "loss": 0.6301,
+      "step": 50091
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.6322309970855713,
+      "learning_rate": 8.658379476287038e-07,
+      "loss": 0.4548,
+      "step": 50092
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.0064308643341064,
+      "learning_rate": 8.656896592650443e-07,
+      "loss": 0.7634,
+      "step": 50093
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.2627196311950684,
+      "learning_rate": 8.655413823973518e-07,
+      "loss": 0.3954,
+      "step": 50094
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.3922410011291504,
+      "learning_rate": 8.65393117026041e-07,
+      "loss": 0.5539,
+      "step": 50095
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.7443289756774902,
+      "learning_rate": 8.65244863151522e-07,
+      "loss": 0.6259,
+      "step": 50096
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 4.504177570343018,
+      "learning_rate": 8.650966207742095e-07,
+      "loss": 0.6546,
+      "step": 50097
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.30568790435791,
+      "learning_rate": 8.649483898945132e-07,
+      "loss": 0.4589,
+      "step": 50098
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.6775476932525635,
+      "learning_rate": 8.648001705128473e-07,
+      "loss": 0.5695,
+      "step": 50099
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.6120591163635254,
+      "learning_rate": 8.646519626296218e-07,
+      "loss": 0.579,
+      "step": 50100
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.1187052726745605,
+      "learning_rate": 8.645037662452516e-07,
+      "loss": 0.3877,
+      "step": 50101
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.719428777694702,
+      "learning_rate": 8.643555813601457e-07,
+      "loss": 0.4747,
+      "step": 50102
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 4.091642379760742,
+      "learning_rate": 8.642074079747187e-07,
+      "loss": 0.842,
+      "step": 50103
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.156764268875122,
+      "learning_rate": 8.640592460893809e-07,
+      "loss": 0.5563,
+      "step": 50104
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.599883794784546,
+      "learning_rate": 8.639110957045449e-07,
+      "loss": 0.4792,
+      "step": 50105
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 1.9097927808761597,
+      "learning_rate": 8.637629568206224e-07,
+      "loss": 0.3432,
+      "step": 50106
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.1943471431732178,
+      "learning_rate": 8.63614829438027e-07,
+      "loss": 0.606,
+      "step": 50107
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.8606252670288086,
+      "learning_rate": 8.634667135571678e-07,
+      "loss": 0.6211,
+      "step": 50108
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.4827992916107178,
+      "learning_rate": 8.633186091784596e-07,
+      "loss": 0.6826,
+      "step": 50109
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.9815573692321777,
+      "learning_rate": 8.631705163023118e-07,
+      "loss": 0.4598,
+      "step": 50110
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.8598875999450684,
+      "learning_rate": 8.630224349291378e-07,
+      "loss": 0.6805,
+      "step": 50111
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.636991262435913,
+      "learning_rate": 8.628743650593474e-07,
+      "loss": 0.5266,
+      "step": 50112
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.6386170387268066,
+      "learning_rate": 8.62726306693355e-07,
+      "loss": 0.6604,
+      "step": 50113
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.050161123275757,
+      "learning_rate": 8.625782598315702e-07,
+      "loss": 0.489,
+      "step": 50114
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.0985677242279053,
+      "learning_rate": 8.624302244744059e-07,
+      "loss": 0.2784,
+      "step": 50115
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.503904342651367,
+      "learning_rate": 8.62282200622272e-07,
+      "loss": 0.4937,
+      "step": 50116
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 1.9786375761032104,
+      "learning_rate": 8.621341882755824e-07,
+      "loss": 0.4931,
+      "step": 50117
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.507908344268799,
+      "learning_rate": 8.619861874347468e-07,
+      "loss": 0.8487,
+      "step": 50118
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 4.200161933898926,
+      "learning_rate": 8.618381981001783e-07,
+      "loss": 0.9186,
+      "step": 50119
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.814389228820801,
+      "learning_rate": 8.616902202722865e-07,
+      "loss": 0.3751,
+      "step": 50120
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.637608528137207,
+      "learning_rate": 8.61542253951485e-07,
+      "loss": 0.5241,
+      "step": 50121
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.693265676498413,
+      "learning_rate": 8.613942991381835e-07,
+      "loss": 0.6436,
+      "step": 50122
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.912672996520996,
+      "learning_rate": 8.612463558327949e-07,
+      "loss": 0.6495,
+      "step": 50123
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 4.44477653503418,
+      "learning_rate": 8.610984240357284e-07,
+      "loss": 0.6408,
+      "step": 50124
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.4398162364959717,
+      "learning_rate": 8.60950503747397e-07,
+      "loss": 0.6097,
+      "step": 50125
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.6966781616210938,
+      "learning_rate": 8.608025949682114e-07,
+      "loss": 0.4439,
+      "step": 50126
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 4.087254047393799,
+      "learning_rate": 8.606546976985847e-07,
+      "loss": 0.4445,
+      "step": 50127
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 4.506282329559326,
+      "learning_rate": 8.605068119389254e-07,
+      "loss": 0.6182,
+      "step": 50128
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.2630391120910645,
+      "learning_rate": 8.603589376896471e-07,
+      "loss": 0.7148,
+      "step": 50129
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 5.047126770019531,
+      "learning_rate": 8.602110749511589e-07,
+      "loss": 0.745,
+      "step": 50130
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.3546998500823975,
+      "learning_rate": 8.600632237238738e-07,
+      "loss": 0.5537,
+      "step": 50131
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.2040274143218994,
+      "learning_rate": 8.599153840082009e-07,
+      "loss": 0.7486,
+      "step": 50132
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.879020929336548,
+      "learning_rate": 8.597675558045538e-07,
+      "loss": 0.6219,
+      "step": 50133
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.9208669662475586,
+      "learning_rate": 8.59619739113341e-07,
+      "loss": 0.6474,
+      "step": 50134
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.74224853515625,
+      "learning_rate": 8.594719339349755e-07,
+      "loss": 0.3616,
+      "step": 50135
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.3066115379333496,
+      "learning_rate": 8.593241402698666e-07,
+      "loss": 0.7216,
+      "step": 50136
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.182075023651123,
+      "learning_rate": 8.59176358118427e-07,
+      "loss": 0.4677,
+      "step": 50137
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.2517621517181396,
+      "learning_rate": 8.590285874810656e-07,
+      "loss": 0.6064,
+      "step": 50138
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.2428324222564697,
+      "learning_rate": 8.588808283581945e-07,
+      "loss": 0.8873,
+      "step": 50139
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.4884819984436035,
+      "learning_rate": 8.587330807502258e-07,
+      "loss": 0.4882,
+      "step": 50140
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.239199161529541,
+      "learning_rate": 8.585853446575682e-07,
+      "loss": 0.5945,
+      "step": 50141
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.6266164779663086,
+      "learning_rate": 8.584376200806344e-07,
+      "loss": 0.406,
+      "step": 50142
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.0166852474212646,
+      "learning_rate": 8.582899070198325e-07,
+      "loss": 0.5238,
+      "step": 50143
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.008488416671753,
+      "learning_rate": 8.581422054755762e-07,
+      "loss": 0.5198,
+      "step": 50144
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.466970443725586,
+      "learning_rate": 8.579945154482739e-07,
+      "loss": 0.3387,
+      "step": 50145
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.484696626663208,
+      "learning_rate": 8.578468369383369e-07,
+      "loss": 0.5457,
+      "step": 50146
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.515831470489502,
+      "learning_rate": 8.576991699461778e-07,
+      "loss": 0.407,
+      "step": 50147
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.3091580867767334,
+      "learning_rate": 8.575515144722041e-07,
+      "loss": 0.6443,
+      "step": 50148
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.646808385848999,
+      "learning_rate": 8.574038705168286e-07,
+      "loss": 0.4666,
+      "step": 50149
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.3333899974823,
+      "learning_rate": 8.572562380804605e-07,
+      "loss": 0.7397,
+      "step": 50150
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.177797794342041,
+      "learning_rate": 8.571086171635106e-07,
+      "loss": 0.716,
+      "step": 50151
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.5569076538085938,
+      "learning_rate": 8.569610077663909e-07,
+      "loss": 0.269,
+      "step": 50152
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.5409464836120605,
+      "learning_rate": 8.568134098895098e-07,
+      "loss": 0.5343,
+      "step": 50153
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.7926042079925537,
+      "learning_rate": 8.566658235332792e-07,
+      "loss": 0.5219,
+      "step": 50154
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.809938430786133,
+      "learning_rate": 8.565182486981077e-07,
+      "loss": 0.5147,
+      "step": 50155
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.1169636249542236,
+      "learning_rate": 8.563706853844084e-07,
+      "loss": 0.5896,
+      "step": 50156
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 4.9599199295043945,
+      "learning_rate": 8.562231335925886e-07,
+      "loss": 0.6906,
+      "step": 50157
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 4.793027877807617,
+      "learning_rate": 8.560755933230608e-07,
+      "loss": 0.6033,
+      "step": 50158
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.9025790691375732,
+      "learning_rate": 8.559280645762335e-07,
+      "loss": 0.7754,
+      "step": 50159
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.4251725673675537,
+      "learning_rate": 8.557805473525193e-07,
+      "loss": 0.5723,
+      "step": 50160
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.6940510272979736,
+      "learning_rate": 8.556330416523256e-07,
+      "loss": 0.6635,
+      "step": 50161
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 4.256085395812988,
+      "learning_rate": 8.554855474760648e-07,
+      "loss": 0.5339,
+      "step": 50162
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.210986614227295,
+      "learning_rate": 8.55338064824145e-07,
+      "loss": 0.6804,
+      "step": 50163
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.5462567806243896,
+      "learning_rate": 8.55190593696979e-07,
+      "loss": 0.8029,
+      "step": 50164
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.0669102668762207,
+      "learning_rate": 8.550431340949738e-07,
+      "loss": 0.3501,
+      "step": 50165
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.1663568019866943,
+      "learning_rate": 8.548956860185409e-07,
+      "loss": 0.5173,
+      "step": 50166
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.883845329284668,
+      "learning_rate": 8.547482494680904e-07,
+      "loss": 0.7518,
+      "step": 50167
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 4.347124099731445,
+      "learning_rate": 8.54600824444033e-07,
+      "loss": 0.464,
+      "step": 50168
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.1072609424591064,
+      "learning_rate": 8.544534109467767e-07,
+      "loss": 0.5059,
+      "step": 50169
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.618846893310547,
+      "learning_rate": 8.543060089767341e-07,
+      "loss": 0.5678,
+      "step": 50170
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.7873458862304688,
+      "learning_rate": 8.54158618534312e-07,
+      "loss": 0.8108,
+      "step": 50171
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.7613894939422607,
+      "learning_rate": 8.54011239619923e-07,
+      "loss": 0.5407,
+      "step": 50172
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.465277910232544,
+      "learning_rate": 8.538638722339742e-07,
+      "loss": 0.5891,
+      "step": 50173
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.3183352947235107,
+      "learning_rate": 8.537165163768785e-07,
+      "loss": 0.6881,
+      "step": 50174
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.6953561305999756,
+      "learning_rate": 8.535691720490424e-07,
+      "loss": 0.47,
+      "step": 50175
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 4.865079879760742,
+      "learning_rate": 8.534218392508781e-07,
+      "loss": 0.5062,
+      "step": 50176
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.413688898086548,
+      "learning_rate": 8.532745179827934e-07,
+      "loss": 0.435,
+      "step": 50177
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.1034016609191895,
+      "learning_rate": 8.531272082452002e-07,
+      "loss": 0.816,
+      "step": 50178
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.4496896266937256,
+      "learning_rate": 8.529799100385056e-07,
+      "loss": 0.4617,
+      "step": 50179
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.9998698234558105,
+      "learning_rate": 8.528326233631212e-07,
+      "loss": 0.4809,
+      "step": 50180
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.4141931533813477,
+      "learning_rate": 8.526853482194547e-07,
+      "loss": 0.7053,
+      "step": 50181
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.8177096843719482,
+      "learning_rate": 8.525380846079178e-07,
+      "loss": 0.5527,
+      "step": 50182
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.657744884490967,
+      "learning_rate": 8.523908325289176e-07,
+      "loss": 0.5102,
+      "step": 50183
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.2106945514678955,
+      "learning_rate": 8.522435919828664e-07,
+      "loss": 0.6886,
+      "step": 50184
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.357177495956421,
+      "learning_rate": 8.5209636297017e-07,
+      "loss": 0.5936,
+      "step": 50185
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.6660475730895996,
+      "learning_rate": 8.519491454912404e-07,
+      "loss": 0.5989,
+      "step": 50186
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.4839797019958496,
+      "learning_rate": 8.518019395464861e-07,
+      "loss": 0.6628,
+      "step": 50187
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.2941226959228516,
+      "learning_rate": 8.516547451363178e-07,
+      "loss": 0.5376,
+      "step": 50188
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.771030902862549,
+      "learning_rate": 8.515075622611424e-07,
+      "loss": 0.9925,
+      "step": 50189
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.1361451148986816,
+      "learning_rate": 8.51360390921372e-07,
+      "loss": 0.3852,
+      "step": 50190
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.6063122749328613,
+      "learning_rate": 8.512132311174126e-07,
+      "loss": 0.4146,
+      "step": 50191
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.391997814178467,
+      "learning_rate": 8.510660828496764e-07,
+      "loss": 0.5864,
+      "step": 50192
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.1883809566497803,
+      "learning_rate": 8.509189461185702e-07,
+      "loss": 0.6306,
+      "step": 50193
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.135000705718994,
+      "learning_rate": 8.507718209245047e-07,
+      "loss": 0.5235,
+      "step": 50194
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.183401584625244,
+      "learning_rate": 8.50624707267888e-07,
+      "loss": 0.7131,
+      "step": 50195
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.6577494144439697,
+      "learning_rate": 8.504776051491303e-07,
+      "loss": 0.6647,
+      "step": 50196
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.2279856204986572,
+      "learning_rate": 8.503305145686386e-07,
+      "loss": 0.5293,
+      "step": 50197
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.1801443099975586,
+      "learning_rate": 8.501834355268251e-07,
+      "loss": 0.4715,
+      "step": 50198
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.7506139278411865,
+      "learning_rate": 8.500363680240952e-07,
+      "loss": 0.6594,
+      "step": 50199
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.696289539337158,
+      "learning_rate": 8.498893120608609e-07,
+      "loss": 0.8453,
+      "step": 50200
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.5526933670043945,
+      "learning_rate": 8.497422676375283e-07,
+      "loss": 0.5995,
+      "step": 50201
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.696937084197998,
+      "learning_rate": 8.495952347545078e-07,
+      "loss": 0.9248,
+      "step": 50202
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.7309017181396484,
+      "learning_rate": 8.4944821341221e-07,
+      "loss": 0.5912,
+      "step": 50203
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.1914563179016113,
+      "learning_rate": 8.493012036110398e-07,
+      "loss": 0.5797,
+      "step": 50204
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.077726364135742,
+      "learning_rate": 8.491542053514101e-07,
+      "loss": 0.398,
+      "step": 50205
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.178126335144043,
+      "learning_rate": 8.490072186337256e-07,
+      "loss": 0.7994,
+      "step": 50206
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.599607467651367,
+      "learning_rate": 8.488602434583975e-07,
+      "loss": 0.3503,
+      "step": 50207
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.5209743976593018,
+      "learning_rate": 8.487132798258352e-07,
+      "loss": 0.5831,
+      "step": 50208
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.302095651626587,
+      "learning_rate": 8.485663277364448e-07,
+      "loss": 0.5237,
+      "step": 50209
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.4679956436157227,
+      "learning_rate": 8.484193871906377e-07,
+      "loss": 0.3864,
+      "step": 50210
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 1.9023001194000244,
+      "learning_rate": 8.482724581888197e-07,
+      "loss": 0.5643,
+      "step": 50211
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.8730080127716064,
+      "learning_rate": 8.48125540731401e-07,
+      "loss": 0.5294,
+      "step": 50212
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.7316160202026367,
+      "learning_rate": 8.479786348187907e-07,
+      "loss": 0.7473,
+      "step": 50213
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.1690621376037598,
+      "learning_rate": 8.478317404513953e-07,
+      "loss": 0.6706,
+      "step": 50214
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.588361978530884,
+      "learning_rate": 8.476848576296254e-07,
+      "loss": 0.5056,
+      "step": 50215
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.844114065170288,
+      "learning_rate": 8.475379863538874e-07,
+      "loss": 0.7249,
+      "step": 50216
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.068634510040283,
+      "learning_rate": 8.473911266245921e-07,
+      "loss": 0.4134,
+      "step": 50217
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.7722365856170654,
+      "learning_rate": 8.472442784421453e-07,
+      "loss": 0.8275,
+      "step": 50218
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.3148887157440186,
+      "learning_rate": 8.470974418069572e-07,
+      "loss": 0.4538,
+      "step": 50219
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.13576340675354,
+      "learning_rate": 8.469506167194346e-07,
+      "loss": 0.6583,
+      "step": 50220
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.8481271266937256,
+      "learning_rate": 8.46803803179988e-07,
+      "loss": 0.6486,
+      "step": 50221
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.8286240100860596,
+      "learning_rate": 8.466570011890223e-07,
+      "loss": 0.4189,
+      "step": 50222
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.4220995903015137,
+      "learning_rate": 8.465102107469497e-07,
+      "loss": 0.5498,
+      "step": 50223
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.807408094406128,
+      "learning_rate": 8.463634318541741e-07,
+      "loss": 0.5215,
+      "step": 50224
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.5221588611602783,
+      "learning_rate": 8.462166645111081e-07,
+      "loss": 0.5472,
+      "step": 50225
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.681551933288574,
+      "learning_rate": 8.460699087181557e-07,
+      "loss": 0.8717,
+      "step": 50226
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.31769061088562,
+      "learning_rate": 8.459231644757265e-07,
+      "loss": 0.624,
+      "step": 50227
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.6513593196868896,
+      "learning_rate": 8.457764317842293e-07,
+      "loss": 0.4505,
+      "step": 50228
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.443755865097046,
+      "learning_rate": 8.456297106440725e-07,
+      "loss": 0.6698,
+      "step": 50229
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.7348146438598633,
+      "learning_rate": 8.45483001055662e-07,
+      "loss": 0.4371,
+      "step": 50230
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 4.749989986419678,
+      "learning_rate": 8.453363030194083e-07,
+      "loss": 0.7484,
+      "step": 50231
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.9504706859588623,
+      "learning_rate": 8.451896165357171e-07,
+      "loss": 0.79,
+      "step": 50232
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.6398544311523438,
+      "learning_rate": 8.450429416049977e-07,
+      "loss": 0.4026,
+      "step": 50233
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.4117066860198975,
+      "learning_rate": 8.448962782276565e-07,
+      "loss": 0.5536,
+      "step": 50234
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.0816891193389893,
+      "learning_rate": 8.447496264041038e-07,
+      "loss": 0.3784,
+      "step": 50235
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.5869534015655518,
+      "learning_rate": 8.446029861347437e-07,
+      "loss": 0.5194,
+      "step": 50236
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.119479179382324,
+      "learning_rate": 8.444563574199883e-07,
+      "loss": 0.6752,
+      "step": 50237
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.7551186084747314,
+      "learning_rate": 8.443097402602412e-07,
+      "loss": 0.5535,
+      "step": 50238
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.427034378051758,
+      "learning_rate": 8.441631346559132e-07,
+      "loss": 0.6664,
+      "step": 50239
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.2714521884918213,
+      "learning_rate": 8.440165406074097e-07,
+      "loss": 0.6845,
+      "step": 50240
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.0310046672821045,
+      "learning_rate": 8.438699581151405e-07,
+      "loss": 0.6332,
+      "step": 50241
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.227004051208496,
+      "learning_rate": 8.437233871795108e-07,
+      "loss": 0.6492,
+      "step": 50242
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.65712308883667,
+      "learning_rate": 8.435768278009304e-07,
+      "loss": 0.6254,
+      "step": 50243
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.1931488513946533,
+      "learning_rate": 8.43430279979805e-07,
+      "loss": 0.5198,
+      "step": 50244
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.4641146659851074,
+      "learning_rate": 8.432837437165436e-07,
+      "loss": 0.5698,
+      "step": 50245
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 4.262625217437744,
+      "learning_rate": 8.431372190115522e-07,
+      "loss": 0.7332,
+      "step": 50246
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.0104613304138184,
+      "learning_rate": 8.429907058652387e-07,
+      "loss": 0.4476,
+      "step": 50247
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.2553138732910156,
+      "learning_rate": 8.428442042780111e-07,
+      "loss": 0.5171,
+      "step": 50248
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 5.380791664123535,
+      "learning_rate": 8.426977142502773e-07,
+      "loss": 0.7086,
+      "step": 50249
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.2115914821624756,
+      "learning_rate": 8.425512357824428e-07,
+      "loss": 0.6566,
+      "step": 50250
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.5923027992248535,
+      "learning_rate": 8.424047688749171e-07,
+      "loss": 0.4017,
+      "step": 50251
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.7706124782562256,
+      "learning_rate": 8.422583135281054e-07,
+      "loss": 0.7111,
+      "step": 50252
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.0911166667938232,
+      "learning_rate": 8.421118697424163e-07,
+      "loss": 0.5507,
+      "step": 50253
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.941901445388794,
+      "learning_rate": 8.419654375182557e-07,
+      "loss": 0.5277,
+      "step": 50254
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.749268054962158,
+      "learning_rate": 8.418190168560331e-07,
+      "loss": 0.6905,
+      "step": 50255
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.365943193435669,
+      "learning_rate": 8.416726077561526e-07,
+      "loss": 0.584,
+      "step": 50256
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.342897415161133,
+      "learning_rate": 8.415262102190241e-07,
+      "loss": 0.5852,
+      "step": 50257
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 4.622379302978516,
+      "learning_rate": 8.413798242450521e-07,
+      "loss": 0.612,
+      "step": 50258
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 4.579598903656006,
+      "learning_rate": 8.412334498346458e-07,
+      "loss": 0.6007,
+      "step": 50259
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.4571428298950195,
+      "learning_rate": 8.410870869882109e-07,
+      "loss": 0.6048,
+      "step": 50260
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.373960018157959,
+      "learning_rate": 8.409407357061555e-07,
+      "loss": 0.4911,
+      "step": 50261
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.6016056537628174,
+      "learning_rate": 8.407943959888848e-07,
+      "loss": 0.413,
+      "step": 50262
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.6558737754821777,
+      "learning_rate": 8.406480678368067e-07,
+      "loss": 0.6241,
+      "step": 50263
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 6.269711017608643,
+      "learning_rate": 8.405017512503299e-07,
+      "loss": 0.5269,
+      "step": 50264
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.8769593238830566,
+      "learning_rate": 8.403554462298579e-07,
+      "loss": 0.7736,
+      "step": 50265
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.9897267818450928,
+      "learning_rate": 8.402091527758005e-07,
+      "loss": 0.5625,
+      "step": 50266
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.000720977783203,
+      "learning_rate": 8.400628708885622e-07,
+      "loss": 0.5277,
+      "step": 50267
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.5960307121276855,
+      "learning_rate": 8.399166005685505e-07,
+      "loss": 0.6467,
+      "step": 50268
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.9498631954193115,
+      "learning_rate": 8.397703418161735e-07,
+      "loss": 0.84,
+      "step": 50269
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.5397963523864746,
+      "learning_rate": 8.396240946318351e-07,
+      "loss": 0.6224,
+      "step": 50270
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 4.002372741699219,
+      "learning_rate": 8.394778590159453e-07,
+      "loss": 0.7112,
+      "step": 50271
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.420236349105835,
+      "learning_rate": 8.393316349689079e-07,
+      "loss": 0.5636,
+      "step": 50272
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.3804855346679688,
+      "learning_rate": 8.391854224911316e-07,
+      "loss": 0.4937,
+      "step": 50273
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.5931131839752197,
+      "learning_rate": 8.390392215830206e-07,
+      "loss": 0.5296,
+      "step": 50274
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.5221946239471436,
+      "learning_rate": 8.388930322449828e-07,
+      "loss": 0.6457,
+      "step": 50275
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.8525662422180176,
+      "learning_rate": 8.387468544774263e-07,
+      "loss": 0.5388,
+      "step": 50276
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.1528244018554688,
+      "learning_rate": 8.386006882807545e-07,
+      "loss": 0.7535,
+      "step": 50277
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.679260015487671,
+      "learning_rate": 8.384545336553767e-07,
+      "loss": 0.686,
+      "step": 50278
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.055105447769165,
+      "learning_rate": 8.383083906016964e-07,
+      "loss": 0.6723,
+      "step": 50279
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.343233823776245,
+      "learning_rate": 8.381622591201227e-07,
+      "loss": 0.5691,
+      "step": 50280
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.638996124267578,
+      "learning_rate": 8.380161392110597e-07,
+      "loss": 0.5677,
+      "step": 50281
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.567697763442993,
+      "learning_rate": 8.378700308749154e-07,
+      "loss": 0.6687,
+      "step": 50282
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 4.116372585296631,
+      "learning_rate": 8.377239341120946e-07,
+      "loss": 0.6801,
+      "step": 50283
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.3077220916748047,
+      "learning_rate": 8.375778489230057e-07,
+      "loss": 0.6405,
+      "step": 50284
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.2640576362609863,
+      "learning_rate": 8.374317753080519e-07,
+      "loss": 0.6611,
+      "step": 50285
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.2709202766418457,
+      "learning_rate": 8.372857132676427e-07,
+      "loss": 0.6231,
+      "step": 50286
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.6626274585723877,
+      "learning_rate": 8.371396628021805e-07,
+      "loss": 0.6028,
+      "step": 50287
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.5642049312591553,
+      "learning_rate": 8.369936239120741e-07,
+      "loss": 0.6368,
+      "step": 50288
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 4.300511360168457,
+      "learning_rate": 8.368475965977291e-07,
+      "loss": 0.4835,
+      "step": 50289
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.9941158294677734,
+      "learning_rate": 8.367015808595519e-07,
+      "loss": 0.8889,
+      "step": 50290
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.57399582862854,
+      "learning_rate": 8.365555766979472e-07,
+      "loss": 0.617,
+      "step": 50291
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.685758113861084,
+      "learning_rate": 8.364095841133229e-07,
+      "loss": 0.5618,
+      "step": 50292
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.367068290710449,
+      "learning_rate": 8.362636031060828e-07,
+      "loss": 0.5104,
+      "step": 50293
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.4698004722595215,
+      "learning_rate": 8.361176336766353e-07,
+      "loss": 0.554,
+      "step": 50294
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.422522783279419,
+      "learning_rate": 8.359716758253828e-07,
+      "loss": 0.5669,
+      "step": 50295
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.251647710800171,
+      "learning_rate": 8.35825729552735e-07,
+      "loss": 0.5911,
+      "step": 50296
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.5538649559020996,
+      "learning_rate": 8.356797948590945e-07,
+      "loss": 0.4818,
+      "step": 50297
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.218607187271118,
+      "learning_rate": 8.355338717448697e-07,
+      "loss": 0.4881,
+      "step": 50298
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 1.9377750158309937,
+      "learning_rate": 8.353879602104642e-07,
+      "loss": 0.3628,
+      "step": 50299
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.3668460845947266,
+      "learning_rate": 8.352420602562856e-07,
+      "loss": 0.8496,
+      "step": 50300
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.9473254680633545,
+      "learning_rate": 8.350961718827378e-07,
+      "loss": 0.5774,
+      "step": 50301
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.635712146759033,
+      "learning_rate": 8.349502950902283e-07,
+      "loss": 0.4489,
+      "step": 50302
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.513378858566284,
+      "learning_rate": 8.348044298791602e-07,
+      "loss": 0.6976,
+      "step": 50303
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.2566051483154297,
+      "learning_rate": 8.346585762499426e-07,
+      "loss": 0.646,
+      "step": 50304
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 4.4049973487854,
+      "learning_rate": 8.345127342029774e-07,
+      "loss": 0.4679,
+      "step": 50305
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.1792736053466797,
+      "learning_rate": 8.343669037386731e-07,
+      "loss": 0.3817,
+      "step": 50306
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.5962064266204834,
+      "learning_rate": 8.342210848574328e-07,
+      "loss": 0.5721,
+      "step": 50307
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.5004944801330566,
+      "learning_rate": 8.340752775596633e-07,
+      "loss": 0.6731,
+      "step": 50308
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.1680004596710205,
+      "learning_rate": 8.339294818457699e-07,
+      "loss": 0.6984,
+      "step": 50309
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.862739324569702,
+      "learning_rate": 8.33783697716159e-07,
+      "loss": 0.3971,
+      "step": 50310
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.7744104862213135,
+      "learning_rate": 8.336379251712339e-07,
+      "loss": 0.7446,
+      "step": 50311
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.3283729553222656,
+      "learning_rate": 8.334921642114018e-07,
+      "loss": 0.591,
+      "step": 50312
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.9040496349334717,
+      "learning_rate": 8.333464148370663e-07,
+      "loss": 0.7845,
+      "step": 50313
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.7396957874298096,
+      "learning_rate": 8.332006770486351e-07,
+      "loss": 0.4964,
+      "step": 50314
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.6741507053375244,
+      "learning_rate": 8.330549508465102e-07,
+      "loss": 0.5255,
+      "step": 50315
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.993994951248169,
+      "learning_rate": 8.329092362310998e-07,
+      "loss": 0.6624,
+      "step": 50316
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 1.8476388454437256,
+      "learning_rate": 8.327635332028066e-07,
+      "loss": 0.2927,
+      "step": 50317
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.6599831581115723,
+      "learning_rate": 8.326178417620379e-07,
+      "loss": 0.5101,
+      "step": 50318
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.6008336544036865,
+      "learning_rate": 8.324721619091969e-07,
+      "loss": 0.7863,
+      "step": 50319
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.5047147274017334,
+      "learning_rate": 8.323264936446907e-07,
+      "loss": 0.5692,
+      "step": 50320
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.1026413440704346,
+      "learning_rate": 8.321808369689221e-07,
+      "loss": 0.4085,
+      "step": 50321
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.733121871948242,
+      "learning_rate": 8.320351918822988e-07,
+      "loss": 0.6235,
+      "step": 50322
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.6275691986083984,
+      "learning_rate": 8.318895583852227e-07,
+      "loss": 0.508,
+      "step": 50323
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.6510424613952637,
+      "learning_rate": 8.317439364781005e-07,
+      "loss": 0.7686,
+      "step": 50324
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.5858147144317627,
+      "learning_rate": 8.31598326161338e-07,
+      "loss": 0.6755,
+      "step": 50325
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.7569406032562256,
+      "learning_rate": 8.314527274353379e-07,
+      "loss": 0.7679,
+      "step": 50326
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.862630844116211,
+      "learning_rate": 8.313071403005074e-07,
+      "loss": 0.7498,
+      "step": 50327
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.8883352279663086,
+      "learning_rate": 8.31161564757249e-07,
+      "loss": 0.5995,
+      "step": 50328
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.6035733222961426,
+      "learning_rate": 8.310160008059687e-07,
+      "loss": 0.5248,
+      "step": 50329
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 5.0127339363098145,
+      "learning_rate": 8.308704484470726e-07,
+      "loss": 0.6898,
+      "step": 50330
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.040562629699707,
+      "learning_rate": 8.307249076809621e-07,
+      "loss": 0.5352,
+      "step": 50331
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.6836304664611816,
+      "learning_rate": 8.305793785080452e-07,
+      "loss": 0.49,
+      "step": 50332
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.550246477127075,
+      "learning_rate": 8.304338609287243e-07,
+      "loss": 0.583,
+      "step": 50333
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.810105085372925,
+      "learning_rate": 8.302883549434054e-07,
+      "loss": 0.5698,
+      "step": 50334
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.5314106941223145,
+      "learning_rate": 8.301428605524919e-07,
+      "loss": 0.5258,
+      "step": 50335
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.1039419174194336,
+      "learning_rate": 8.299973777563891e-07,
+      "loss": 0.7131,
+      "step": 50336
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.570901393890381,
+      "learning_rate": 8.298519065555028e-07,
+      "loss": 0.4583,
+      "step": 50337
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.4813365936279297,
+      "learning_rate": 8.297064469502341e-07,
+      "loss": 0.6953,
+      "step": 50338
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 4.720803737640381,
+      "learning_rate": 8.295609989409914e-07,
+      "loss": 0.5986,
+      "step": 50339
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 1.8876041173934937,
+      "learning_rate": 8.294155625281764e-07,
+      "loss": 0.4171,
+      "step": 50340
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.6037731170654297,
+      "learning_rate": 8.292701377121953e-07,
+      "loss": 0.5687,
+      "step": 50341
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.287325859069824,
+      "learning_rate": 8.291247244934503e-07,
+      "loss": 0.6228,
+      "step": 50342
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.5143892765045166,
+      "learning_rate": 8.289793228723481e-07,
+      "loss": 0.3851,
+      "step": 50343
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.7662277221679688,
+      "learning_rate": 8.288339328492906e-07,
+      "loss": 0.6569,
+      "step": 50344
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.5478899478912354,
+      "learning_rate": 8.286885544246847e-07,
+      "loss": 0.4501,
+      "step": 50345
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.1906824111938477,
+      "learning_rate": 8.28543187598932e-07,
+      "loss": 0.5417,
+      "step": 50346
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.4849853515625,
+      "learning_rate": 8.283978323724395e-07,
+      "loss": 0.7532,
+      "step": 50347
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.0402002334594727,
+      "learning_rate": 8.282524887456084e-07,
+      "loss": 0.3551,
+      "step": 50348
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.38131046295166,
+      "learning_rate": 8.281071567188448e-07,
+      "loss": 0.706,
+      "step": 50349
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.936415195465088,
+      "learning_rate": 8.279618362925517e-07,
+      "loss": 0.6195,
+      "step": 50350
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.4322056770324707,
+      "learning_rate": 8.278165274671357e-07,
+      "loss": 0.5331,
+      "step": 50351
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.6311001777648926,
+      "learning_rate": 8.276712302429979e-07,
+      "loss": 0.7577,
+      "step": 50352
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.4153945446014404,
+      "learning_rate": 8.275259446205442e-07,
+      "loss": 0.6556,
+      "step": 50353
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.283626079559326,
+      "learning_rate": 8.273806706001764e-07,
+      "loss": 0.7178,
+      "step": 50354
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.5136539936065674,
+      "learning_rate": 8.272354081823014e-07,
+      "loss": 0.5037,
+      "step": 50355
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.0957486629486084,
+      "learning_rate": 8.270901573673207e-07,
+      "loss": 0.7207,
+      "step": 50356
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.1097910404205322,
+      "learning_rate": 8.269449181556394e-07,
+      "loss": 0.5033,
+      "step": 50357
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.3626999855041504,
+      "learning_rate": 8.267996905476605e-07,
+      "loss": 0.5415,
+      "step": 50358
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 4.018835067749023,
+      "learning_rate": 8.266544745437893e-07,
+      "loss": 0.4468,
+      "step": 50359
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.3339786529541016,
+      "learning_rate": 8.265092701444277e-07,
+      "loss": 0.4481,
+      "step": 50360
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.216951847076416,
+      "learning_rate": 8.263640773499815e-07,
+      "loss": 0.6418,
+      "step": 50361
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 4.698602676391602,
+      "learning_rate": 8.262188961608519e-07,
+      "loss": 0.6597,
+      "step": 50362
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.3608803749084473,
+      "learning_rate": 8.260737265774454e-07,
+      "loss": 0.5617,
+      "step": 50363
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 4.119687557220459,
+      "learning_rate": 8.259285686001628e-07,
+      "loss": 0.747,
+      "step": 50364
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.619748830795288,
+      "learning_rate": 8.257834222294109e-07,
+      "loss": 0.6157,
+      "step": 50365
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.7856929302215576,
+      "learning_rate": 8.256382874655899e-07,
+      "loss": 0.6665,
+      "step": 50366
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 4.989081382751465,
+      "learning_rate": 8.254931643091063e-07,
+      "loss": 0.6022,
+      "step": 50367
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.567211151123047,
+      "learning_rate": 8.253480527603613e-07,
+      "loss": 0.4543,
+      "step": 50368
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.4383416175842285,
+      "learning_rate": 8.252029528197597e-07,
+      "loss": 0.4952,
+      "step": 50369
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.2827489376068115,
+      "learning_rate": 8.250578644877049e-07,
+      "loss": 0.7104,
+      "step": 50370
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.7289059162139893,
+      "learning_rate": 8.249127877646012e-07,
+      "loss": 0.5793,
+      "step": 50371
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.5761852264404297,
+      "learning_rate": 8.247677226508494e-07,
+      "loss": 0.562,
+      "step": 50372
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.0136590003967285,
+      "learning_rate": 8.246226691468562e-07,
+      "loss": 0.2929,
+      "step": 50373
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.3127667903900146,
+      "learning_rate": 8.244776272530219e-07,
+      "loss": 0.5504,
+      "step": 50374
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.1704633235931396,
+      "learning_rate": 8.243325969697524e-07,
+      "loss": 0.6012,
+      "step": 50375
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.6447556018829346,
+      "learning_rate": 8.241875782974485e-07,
+      "loss": 0.5852,
+      "step": 50376
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.564192771911621,
+      "learning_rate": 8.240425712365158e-07,
+      "loss": 0.4988,
+      "step": 50377
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.7913224697113037,
+      "learning_rate": 8.238975757873552e-07,
+      "loss": 0.602,
+      "step": 50378
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 9.19425106048584,
+      "learning_rate": 8.23752591950372e-07,
+      "loss": 0.6852,
+      "step": 50379
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 5.290677070617676,
+      "learning_rate": 8.236076197259673e-07,
+      "loss": 0.6657,
+      "step": 50380
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.9166603088378906,
+      "learning_rate": 8.234626591145467e-07,
+      "loss": 0.5295,
+      "step": 50381
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.471912384033203,
+      "learning_rate": 8.233177101165108e-07,
+      "loss": 0.4874,
+      "step": 50382
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.3113551139831543,
+      "learning_rate": 8.231727727322647e-07,
+      "loss": 0.4215,
+      "step": 50383
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.5263915061950684,
+      "learning_rate": 8.230278469622088e-07,
+      "loss": 0.6408,
+      "step": 50384
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.0393340587615967,
+      "learning_rate": 8.228829328067495e-07,
+      "loss": 0.8231,
+      "step": 50385
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.0990359783172607,
+      "learning_rate": 8.227380302662868e-07,
+      "loss": 0.5477,
+      "step": 50386
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.0943236351013184,
+      "learning_rate": 8.225931393412245e-07,
+      "loss": 0.6954,
+      "step": 50387
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.0240044593811035,
+      "learning_rate": 8.224482600319672e-07,
+      "loss": 0.7445,
+      "step": 50388
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.238976240158081,
+      "learning_rate": 8.223033923389151e-07,
+      "loss": 0.6091,
+      "step": 50389
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.648470640182495,
+      "learning_rate": 8.221585362624718e-07,
+      "loss": 0.5858,
+      "step": 50390
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.1014299392700195,
+      "learning_rate": 8.220136918030424e-07,
+      "loss": 0.9665,
+      "step": 50391
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.732667922973633,
+      "learning_rate": 8.218688589610264e-07,
+      "loss": 0.8688,
+      "step": 50392
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.816389799118042,
+      "learning_rate": 8.21724037736829e-07,
+      "loss": 0.6415,
+      "step": 50393
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.467371940612793,
+      "learning_rate": 8.21579228130851e-07,
+      "loss": 0.4463,
+      "step": 50394
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.769242763519287,
+      "learning_rate": 8.214344301434967e-07,
+      "loss": 0.5703,
+      "step": 50395
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.489157199859619,
+      "learning_rate": 8.212896437751672e-07,
+      "loss": 0.5765,
+      "step": 50396
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.5739967823028564,
+      "learning_rate": 8.211448690262664e-07,
+      "loss": 0.7396,
+      "step": 50397
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.500627040863037,
+      "learning_rate": 8.210001058971956e-07,
+      "loss": 0.6845,
+      "step": 50398
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.499864339828491,
+      "learning_rate": 8.208553543883574e-07,
+      "loss": 0.6546,
+      "step": 50399
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.1710643768310547,
+      "learning_rate": 8.20710614500157e-07,
+      "loss": 0.5205,
+      "step": 50400
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.0172481536865234,
+      "learning_rate": 8.20565886232993e-07,
+      "loss": 0.3923,
+      "step": 50401
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 1.6408745050430298,
+      "learning_rate": 8.204211695872704e-07,
+      "loss": 0.3787,
+      "step": 50402
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.986161708831787,
+      "learning_rate": 8.202764645633904e-07,
+      "loss": 0.5017,
+      "step": 50403
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.3926801681518555,
+      "learning_rate": 8.201317711617568e-07,
+      "loss": 0.522,
+      "step": 50404
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.222867488861084,
+      "learning_rate": 8.199870893827699e-07,
+      "loss": 0.6658,
+      "step": 50405
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.2434511184692383,
+      "learning_rate": 8.198424192268339e-07,
+      "loss": 0.54,
+      "step": 50406
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 4.253678798675537,
+      "learning_rate": 8.19697760694349e-07,
+      "loss": 0.4804,
+      "step": 50407
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 4.161564826965332,
+      "learning_rate": 8.195531137857199e-07,
+      "loss": 1.0253,
+      "step": 50408
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.887024402618408,
+      "learning_rate": 8.194084785013462e-07,
+      "loss": 0.5846,
+      "step": 50409
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.979290008544922,
+      "learning_rate": 8.19263854841632e-07,
+      "loss": 0.6291,
+      "step": 50410
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 6.131655216217041,
+      "learning_rate": 8.191192428069783e-07,
+      "loss": 0.7992,
+      "step": 50411
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.300433874130249,
+      "learning_rate": 8.189746423977896e-07,
+      "loss": 0.5514,
+      "step": 50412
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 4.229356288909912,
+      "learning_rate": 8.188300536144644e-07,
+      "loss": 0.7606,
+      "step": 50413
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.9500770568847656,
+      "learning_rate": 8.186854764574076e-07,
+      "loss": 0.5752,
+      "step": 50414
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 1.8872005939483643,
+      "learning_rate": 8.185409109270193e-07,
+      "loss": 0.4197,
+      "step": 50415
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.1915626525878906,
+      "learning_rate": 8.183963570237036e-07,
+      "loss": 1.0065,
+      "step": 50416
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 4.8575897216796875,
+      "learning_rate": 8.182518147478597e-07,
+      "loss": 0.6804,
+      "step": 50417
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.4369900226593018,
+      "learning_rate": 8.181072840998927e-07,
+      "loss": 0.3548,
+      "step": 50418
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.435105323791504,
+      "learning_rate": 8.179627650802008e-07,
+      "loss": 0.4899,
+      "step": 50419
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.5481224060058594,
+      "learning_rate": 8.178182576891896e-07,
+      "loss": 0.4055,
+      "step": 50420
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.555177688598633,
+      "learning_rate": 8.176737619272579e-07,
+      "loss": 0.5383,
+      "step": 50421
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.4125518798828125,
+      "learning_rate": 8.175292777948096e-07,
+      "loss": 0.555,
+      "step": 50422
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.6905341148376465,
+      "learning_rate": 8.173848052922445e-07,
+      "loss": 0.7077,
+      "step": 50423
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.7095248699188232,
+      "learning_rate": 8.172403444199667e-07,
+      "loss": 0.5227,
+      "step": 50424
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.953634023666382,
+      "learning_rate": 8.170958951783753e-07,
+      "loss": 0.5252,
+      "step": 50425
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.758833408355713,
+      "learning_rate": 8.16951457567875e-07,
+      "loss": 0.6793,
+      "step": 50426
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 1.893041729927063,
+      "learning_rate": 8.168070315888638e-07,
+      "loss": 0.273,
+      "step": 50427
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.524841785430908,
+      "learning_rate": 8.166626172417469e-07,
+      "loss": 0.6208,
+      "step": 50428
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.11559796333313,
+      "learning_rate": 8.165182145269224e-07,
+      "loss": 0.4454,
+      "step": 50429
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.0089900493621826,
+      "learning_rate": 8.163738234447938e-07,
+      "loss": 0.3264,
+      "step": 50430
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.501008987426758,
+      "learning_rate": 8.162294439957624e-07,
+      "loss": 0.7288,
+      "step": 50431
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.514714241027832,
+      "learning_rate": 8.160850761802308e-07,
+      "loss": 0.5216,
+      "step": 50432
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.1916816234588623,
+      "learning_rate": 8.159407199985981e-07,
+      "loss": 0.5155,
+      "step": 50433
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.931994915008545,
+      "learning_rate": 8.157963754512682e-07,
+      "loss": 0.6691,
+      "step": 50434
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 5.024991035461426,
+      "learning_rate": 8.1565204253864e-07,
+      "loss": 0.3945,
+      "step": 50435
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.7557082176208496,
+      "learning_rate": 8.155077212611168e-07,
+      "loss": 0.5592,
+      "step": 50436
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.8064987659454346,
+      "learning_rate": 8.153634116190984e-07,
+      "loss": 0.6168,
+      "step": 50437
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.9932713508605957,
+      "learning_rate": 8.152191136129878e-07,
+      "loss": 0.5525,
+      "step": 50438
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.02571964263916,
+      "learning_rate": 8.150748272431836e-07,
+      "loss": 0.8082,
+      "step": 50439
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.928175926208496,
+      "learning_rate": 8.149305525100903e-07,
+      "loss": 0.5864,
+      "step": 50440
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.2417678833007812,
+      "learning_rate": 8.147862894141057e-07,
+      "loss": 0.6898,
+      "step": 50441
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.856571674346924,
+      "learning_rate": 8.146420379556341e-07,
+      "loss": 0.6574,
+      "step": 50442
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.3483011722564697,
+      "learning_rate": 8.144977981350738e-07,
+      "loss": 0.6753,
+      "step": 50443
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.961318254470825,
+      "learning_rate": 8.143535699528288e-07,
+      "loss": 0.6416,
+      "step": 50444
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.399644136428833,
+      "learning_rate": 8.142093534092965e-07,
+      "loss": 0.4238,
+      "step": 50445
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.4812936782836914,
+      "learning_rate": 8.140651485048817e-07,
+      "loss": 0.6768,
+      "step": 50446
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.6186628341674805,
+      "learning_rate": 8.139209552399824e-07,
+      "loss": 0.7362,
+      "step": 50447
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.0838897228240967,
+      "learning_rate": 8.137767736150009e-07,
+      "loss": 0.6963,
+      "step": 50448
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.3513283729553223,
+      "learning_rate": 8.136326036303394e-07,
+      "loss": 0.3154,
+      "step": 50449
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.370227336883545,
+      "learning_rate": 8.134884452863956e-07,
+      "loss": 0.5161,
+      "step": 50450
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.4817922115325928,
+      "learning_rate": 8.133442985835721e-07,
+      "loss": 0.441,
+      "step": 50451
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.6461169719696045,
+      "learning_rate": 8.132001635222714e-07,
+      "loss": 0.4723,
+      "step": 50452
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 4.311770439147949,
+      "learning_rate": 8.130560401028914e-07,
+      "loss": 0.7343,
+      "step": 50453
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.7127301692962646,
+      "learning_rate": 8.129119283258352e-07,
+      "loss": 0.7758,
+      "step": 50454
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.6655242443084717,
+      "learning_rate": 8.127678281915008e-07,
+      "loss": 0.4762,
+      "step": 50455
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.4223556518554688,
+      "learning_rate": 8.126237397002923e-07,
+      "loss": 0.6621,
+      "step": 50456
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.3569118976593018,
+      "learning_rate": 8.124796628526071e-07,
+      "loss": 0.5451,
+      "step": 50457
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.1940746307373047,
+      "learning_rate": 8.123355976488484e-07,
+      "loss": 0.5164,
+      "step": 50458
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.517223596572876,
+      "learning_rate": 8.121915440894146e-07,
+      "loss": 0.4883,
+      "step": 50459
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.9630582332611084,
+      "learning_rate": 8.120475021747071e-07,
+      "loss": 0.4929,
+      "step": 50460
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.7103493213653564,
+      "learning_rate": 8.119034719051283e-07,
+      "loss": 0.6284,
+      "step": 50461
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.6477200984954834,
+      "learning_rate": 8.117594532810758e-07,
+      "loss": 0.7049,
+      "step": 50462
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.0541703701019287,
+      "learning_rate": 8.116154463029518e-07,
+      "loss": 0.7645,
+      "step": 50463
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.089186906814575,
+      "learning_rate": 8.114714509711557e-07,
+      "loss": 0.5065,
+      "step": 50464
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.553389072418213,
+      "learning_rate": 8.113274672860894e-07,
+      "loss": 0.6201,
+      "step": 50465
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 4.639155387878418,
+      "learning_rate": 8.111834952481512e-07,
+      "loss": 0.3425,
+      "step": 50466
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.228184700012207,
+      "learning_rate": 8.110395348577438e-07,
+      "loss": 0.3711,
+      "step": 50467
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 4.602822780609131,
+      "learning_rate": 8.108955861152651e-07,
+      "loss": 0.7511,
+      "step": 50468
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.947354793548584,
+      "learning_rate": 8.107516490211176e-07,
+      "loss": 0.5095,
+      "step": 50469
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.7699854373931885,
+      "learning_rate": 8.106077235756987e-07,
+      "loss": 0.5775,
+      "step": 50470
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.7676286697387695,
+      "learning_rate": 8.104638097794109e-07,
+      "loss": 0.7185,
+      "step": 50471
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.7330641746520996,
+      "learning_rate": 8.10319907632654e-07,
+      "loss": 0.5887,
+      "step": 50472
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.28271222114563,
+      "learning_rate": 8.101760171358286e-07,
+      "loss": 0.6126,
+      "step": 50473
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.005740165710449,
+      "learning_rate": 8.100321382893333e-07,
+      "loss": 0.6341,
+      "step": 50474
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 4.873392105102539,
+      "learning_rate": 8.098882710935701e-07,
+      "loss": 0.6033,
+      "step": 50475
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.0612123012542725,
+      "learning_rate": 8.09744415548937e-07,
+      "loss": 0.7483,
+      "step": 50476
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 4.199275493621826,
+      "learning_rate": 8.096005716558358e-07,
+      "loss": 0.293,
+      "step": 50477
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.785095691680908,
+      "learning_rate": 8.094567394146646e-07,
+      "loss": 0.5716,
+      "step": 50478
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 4.295452117919922,
+      "learning_rate": 8.093129188258259e-07,
+      "loss": 0.3566,
+      "step": 50479
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.4473814964294434,
+      "learning_rate": 8.091691098897164e-07,
+      "loss": 0.5232,
+      "step": 50480
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.444336414337158,
+      "learning_rate": 8.090253126067394e-07,
+      "loss": 0.5228,
+      "step": 50481
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.411587715148926,
+      "learning_rate": 8.088815269772915e-07,
+      "loss": 0.4209,
+      "step": 50482
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.3590104579925537,
+      "learning_rate": 8.087377530017748e-07,
+      "loss": 0.569,
+      "step": 50483
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.798719882965088,
+      "learning_rate": 8.085939906805879e-07,
+      "loss": 0.5923,
+      "step": 50484
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 4.286994934082031,
+      "learning_rate": 8.08450240014132e-07,
+      "loss": 0.5522,
+      "step": 50485
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.194990396499634,
+      "learning_rate": 8.083065010028047e-07,
+      "loss": 0.8227,
+      "step": 50486
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.0886125564575195,
+      "learning_rate": 8.081627736470077e-07,
+      "loss": 0.5979,
+      "step": 50487
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.8864026069641113,
+      "learning_rate": 8.080190579471381e-07,
+      "loss": 0.6416,
+      "step": 50488
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.2836477756500244,
+      "learning_rate": 8.078753539035988e-07,
+      "loss": 0.5301,
+      "step": 50489
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.0313451290130615,
+      "learning_rate": 8.077316615167869e-07,
+      "loss": 0.6478,
+      "step": 50490
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.650252342224121,
+      "learning_rate": 8.075879807871023e-07,
+      "loss": 0.6888,
+      "step": 50491
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.803948163986206,
+      "learning_rate": 8.074443117149456e-07,
+      "loss": 0.5159,
+      "step": 50492
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.388364315032959,
+      "learning_rate": 8.073006543007161e-07,
+      "loss": 0.3843,
+      "step": 50493
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.6179258823394775,
+      "learning_rate": 8.071570085448122e-07,
+      "loss": 0.5727,
+      "step": 50494
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.0676109790802,
+      "learning_rate": 8.07013374447635e-07,
+      "loss": 0.547,
+      "step": 50495
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.1643195152282715,
+      "learning_rate": 8.068697520095814e-07,
+      "loss": 0.9092,
+      "step": 50496
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.656848192214966,
+      "learning_rate": 8.06726141231054e-07,
+      "loss": 0.7474,
+      "step": 50497
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.083315849304199,
+      "learning_rate": 8.06582542112449e-07,
+      "loss": 0.785,
+      "step": 50498
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.249652624130249,
+      "learning_rate": 8.064389546541679e-07,
+      "loss": 0.6846,
+      "step": 50499
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.386333465576172,
+      "learning_rate": 8.06295378856608e-07,
+      "loss": 0.6898,
+      "step": 50500
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.191694974899292,
+      "learning_rate": 8.061518147201714e-07,
+      "loss": 0.5621,
+      "step": 50501
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.331937789916992,
+      "learning_rate": 8.060082622452542e-07,
+      "loss": 0.6501,
+      "step": 50502
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 1.659546971321106,
+      "learning_rate": 8.058647214322573e-07,
+      "loss": 0.3386,
+      "step": 50503
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.4217326641082764,
+      "learning_rate": 8.057211922815795e-07,
+      "loss": 0.4248,
+      "step": 50504
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.2329039573669434,
+      "learning_rate": 8.055776747936201e-07,
+      "loss": 0.5675,
+      "step": 50505
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.437499761581421,
+      "learning_rate": 8.054341689687772e-07,
+      "loss": 0.4093,
+      "step": 50506
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.2301385402679443,
+      "learning_rate": 8.052906748074513e-07,
+      "loss": 0.5808,
+      "step": 50507
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.5850718021392822,
+      "learning_rate": 8.051471923100396e-07,
+      "loss": 0.5167,
+      "step": 50508
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.623854398727417,
+      "learning_rate": 8.050037214769424e-07,
+      "loss": 0.8401,
+      "step": 50509
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.2894556522369385,
+      "learning_rate": 8.048602623085594e-07,
+      "loss": 0.5414,
+      "step": 50510
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.4299733638763428,
+      "learning_rate": 8.047168148052875e-07,
+      "loss": 0.5901,
+      "step": 50511
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.8645565509796143,
+      "learning_rate": 8.045733789675264e-07,
+      "loss": 0.6371,
+      "step": 50512
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.1980414390563965,
+      "learning_rate": 8.044299547956763e-07,
+      "loss": 0.5886,
+      "step": 50513
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.4410903453826904,
+      "learning_rate": 8.042865422901336e-07,
+      "loss": 0.5133,
+      "step": 50514
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 4.293976306915283,
+      "learning_rate": 8.041431414512995e-07,
+      "loss": 0.7154,
+      "step": 50515
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.7988710403442383,
+      "learning_rate": 8.039997522795701e-07,
+      "loss": 0.7745,
+      "step": 50516
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 4.66078519821167,
+      "learning_rate": 8.038563747753469e-07,
+      "loss": 0.4947,
+      "step": 50517
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.6011581420898438,
+      "learning_rate": 8.03713008939026e-07,
+      "loss": 0.4745,
+      "step": 50518
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 4.470354080200195,
+      "learning_rate": 8.035696547710081e-07,
+      "loss": 0.7725,
+      "step": 50519
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.4632203578948975,
+      "learning_rate": 8.034263122716906e-07,
+      "loss": 0.4234,
+      "step": 50520
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 4.607024669647217,
+      "learning_rate": 8.032829814414717e-07,
+      "loss": 0.5884,
+      "step": 50521
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.243818521499634,
+      "learning_rate": 8.031396622807524e-07,
+      "loss": 0.4913,
+      "step": 50522
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.6013307571411133,
+      "learning_rate": 8.029963547899278e-07,
+      "loss": 0.5437,
+      "step": 50523
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.8420190811157227,
+      "learning_rate": 8.028530589693995e-07,
+      "loss": 0.6619,
+      "step": 50524
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.488154172897339,
+      "learning_rate": 8.027097748195639e-07,
+      "loss": 0.4597,
+      "step": 50525
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.0328097343444824,
+      "learning_rate": 8.025665023408207e-07,
+      "loss": 0.4313,
+      "step": 50526
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.875722646713257,
+      "learning_rate": 8.024232415335664e-07,
+      "loss": 0.7404,
+      "step": 50527
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.49621844291687,
+      "learning_rate": 8.022799923982022e-07,
+      "loss": 0.5281,
+      "step": 50528
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.754314422607422,
+      "learning_rate": 8.021367549351234e-07,
+      "loss": 0.4767,
+      "step": 50529
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.488997459411621,
+      "learning_rate": 8.019935291447312e-07,
+      "loss": 0.5162,
+      "step": 50530
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 4.317864418029785,
+      "learning_rate": 8.01850315027421e-07,
+      "loss": 0.5696,
+      "step": 50531
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.010342597961426,
+      "learning_rate": 8.017071125835923e-07,
+      "loss": 0.3305,
+      "step": 50532
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.574805974960327,
+      "learning_rate": 8.015639218136434e-07,
+      "loss": 0.5703,
+      "step": 50533
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.373138904571533,
+      "learning_rate": 8.014207427179737e-07,
+      "loss": 0.5846,
+      "step": 50534
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.128885269165039,
+      "learning_rate": 8.012775752969787e-07,
+      "loss": 0.6232,
+      "step": 50535
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.400221824645996,
+      "learning_rate": 8.011344195510595e-07,
+      "loss": 0.4593,
+      "step": 50536
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.3284425735473633,
+      "learning_rate": 8.00991275480611e-07,
+      "loss": 0.6841,
+      "step": 50537
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 1.8024784326553345,
+      "learning_rate": 8.008481430860343e-07,
+      "loss": 0.3246,
+      "step": 50538
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.767207622528076,
+      "learning_rate": 8.007050223677243e-07,
+      "loss": 0.624,
+      "step": 50539
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.33357310295105,
+      "learning_rate": 8.005619133260822e-07,
+      "loss": 0.6104,
+      "step": 50540
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.3169291019439697,
+      "learning_rate": 8.004188159615028e-07,
+      "loss": 0.8302,
+      "step": 50541
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.39573073387146,
+      "learning_rate": 8.002757302743868e-07,
+      "loss": 0.6025,
+      "step": 50542
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.7915055751800537,
+      "learning_rate": 8.001326562651296e-07,
+      "loss": 0.7004,
+      "step": 50543
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.2588164806365967,
+      "learning_rate": 7.999895939341312e-07,
+      "loss": 0.424,
+      "step": 50544
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.431295156478882,
+      "learning_rate": 7.99846543281787e-07,
+      "loss": 0.5526,
+      "step": 50545
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.9249110221862793,
+      "learning_rate": 7.997035043084977e-07,
+      "loss": 0.5983,
+      "step": 50546
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 1.8900692462921143,
+      "learning_rate": 7.995604770146581e-07,
+      "loss": 0.2377,
+      "step": 50547
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.3943495750427246,
+      "learning_rate": 7.994174614006683e-07,
+      "loss": 0.6766,
+      "step": 50548
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.3222672939300537,
+      "learning_rate": 7.992744574669243e-07,
+      "loss": 0.5509,
+      "step": 50549
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.4243226051330566,
+      "learning_rate": 7.991314652138249e-07,
+      "loss": 0.6401,
+      "step": 50550
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.595367908477783,
+      "learning_rate": 7.989884846417667e-07,
+      "loss": 0.4913,
+      "step": 50551
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.5144050121307373,
+      "learning_rate": 7.988455157511477e-07,
+      "loss": 0.8305,
+      "step": 50552
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.9638783931732178,
+      "learning_rate": 7.987025585423652e-07,
+      "loss": 0.5624,
+      "step": 50553
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.0416998863220215,
+      "learning_rate": 7.985596130158185e-07,
+      "loss": 0.5822,
+      "step": 50554
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 4.545457363128662,
+      "learning_rate": 7.984166791719023e-07,
+      "loss": 0.8048,
+      "step": 50555
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.1596827507019043,
+      "learning_rate": 7.982737570110167e-07,
+      "loss": 0.4885,
+      "step": 50556
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.4072563648223877,
+      "learning_rate": 7.981308465335563e-07,
+      "loss": 0.7605,
+      "step": 50557
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.2580366134643555,
+      "learning_rate": 7.97987947739921e-07,
+      "loss": 0.3447,
+      "step": 50558
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.5744359493255615,
+      "learning_rate": 7.978450606305066e-07,
+      "loss": 0.5291,
+      "step": 50559
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.1528327465057373,
+      "learning_rate": 7.977021852057115e-07,
+      "loss": 0.6295,
+      "step": 50560
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.932971477508545,
+      "learning_rate": 7.975593214659317e-07,
+      "loss": 0.5513,
+      "step": 50561
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.6853535175323486,
+      "learning_rate": 7.97416469411566e-07,
+      "loss": 0.6811,
+      "step": 50562
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.9631009101867676,
+      "learning_rate": 7.972736290430094e-07,
+      "loss": 0.3737,
+      "step": 50563
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.67567777633667,
+      "learning_rate": 7.971308003606621e-07,
+      "loss": 0.4732,
+      "step": 50564
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.7233569622039795,
+      "learning_rate": 7.969879833649186e-07,
+      "loss": 0.38,
+      "step": 50565
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 4.569939613342285,
+      "learning_rate": 7.968451780561776e-07,
+      "loss": 0.739,
+      "step": 50566
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.172959089279175,
+      "learning_rate": 7.967023844348348e-07,
+      "loss": 0.6209,
+      "step": 50567
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 4.061857223510742,
+      "learning_rate": 7.965596025012889e-07,
+      "loss": 0.7168,
+      "step": 50568
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.2259795665740967,
+      "learning_rate": 7.964168322559351e-07,
+      "loss": 0.5112,
+      "step": 50569
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.708892345428467,
+      "learning_rate": 7.96274073699173e-07,
+      "loss": 0.632,
+      "step": 50570
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.714442014694214,
+      "learning_rate": 7.961313268313964e-07,
+      "loss": 0.4824,
+      "step": 50571
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.308441162109375,
+      "learning_rate": 7.959885916530035e-07,
+      "loss": 0.4561,
+      "step": 50572
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 4.254743576049805,
+      "learning_rate": 7.958458681643922e-07,
+      "loss": 0.6054,
+      "step": 50573
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 1.756485104560852,
+      "learning_rate": 7.957031563659595e-07,
+      "loss": 0.4381,
+      "step": 50574
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.91386079788208,
+      "learning_rate": 7.955604562580998e-07,
+      "loss": 0.451,
+      "step": 50575
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.58927059173584,
+      "learning_rate": 7.954177678412128e-07,
+      "loss": 0.3552,
+      "step": 50576
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.0499489307403564,
+      "learning_rate": 7.952750911156932e-07,
+      "loss": 0.488,
+      "step": 50577
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.7134666442871094,
+      "learning_rate": 7.951324260819387e-07,
+      "loss": 0.592,
+      "step": 50578
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.113401412963867,
+      "learning_rate": 7.949897727403455e-07,
+      "loss": 0.4737,
+      "step": 50579
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.4487175941467285,
+      "learning_rate": 7.948471310913114e-07,
+      "loss": 0.546,
+      "step": 50580
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.9049267768859863,
+      "learning_rate": 7.947045011352305e-07,
+      "loss": 0.6716,
+      "step": 50581
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.3450443744659424,
+      "learning_rate": 7.945618828725027e-07,
+      "loss": 0.6705,
+      "step": 50582
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.7939326763153076,
+      "learning_rate": 7.944192763035219e-07,
+      "loss": 0.7497,
+      "step": 50583
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.9909820556640625,
+      "learning_rate": 7.94276681428685e-07,
+      "loss": 0.4989,
+      "step": 50584
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.511601209640503,
+      "learning_rate": 7.941340982483908e-07,
+      "loss": 0.5022,
+      "step": 50585
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.4338982105255127,
+      "learning_rate": 7.939915267630327e-07,
+      "loss": 0.615,
+      "step": 50586
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.3890700340270996,
+      "learning_rate": 7.938489669730103e-07,
+      "loss": 0.4482,
+      "step": 50587
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.249901056289673,
+      "learning_rate": 7.937064188787163e-07,
+      "loss": 0.7055,
+      "step": 50588
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.1538889408111572,
+      "learning_rate": 7.935638824805509e-07,
+      "loss": 0.5132,
+      "step": 50589
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.8709566593170166,
+      "learning_rate": 7.934213577789068e-07,
+      "loss": 0.6841,
+      "step": 50590
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.8533666133880615,
+      "learning_rate": 7.932788447741835e-07,
+      "loss": 0.5318,
+      "step": 50591
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.4382619857788086,
+      "learning_rate": 7.93136343466775e-07,
+      "loss": 0.5745,
+      "step": 50592
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.110637664794922,
+      "learning_rate": 7.92993853857078e-07,
+      "loss": 0.5648,
+      "step": 50593
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.2215304374694824,
+      "learning_rate": 7.928513759454893e-07,
+      "loss": 0.3852,
+      "step": 50594
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 5.761054039001465,
+      "learning_rate": 7.927089097324059e-07,
+      "loss": 0.9322,
+      "step": 50595
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.373148202896118,
+      "learning_rate": 7.925664552182216e-07,
+      "loss": 0.5708,
+      "step": 50596
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.5348763465881348,
+      "learning_rate": 7.924240124033356e-07,
+      "loss": 0.3722,
+      "step": 50597
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 5.174375534057617,
+      "learning_rate": 7.922815812881407e-07,
+      "loss": 0.3266,
+      "step": 50598
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.720325231552124,
+      "learning_rate": 7.921391618730357e-07,
+      "loss": 0.7354,
+      "step": 50599
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.023548126220703,
+      "learning_rate": 7.919967541584139e-07,
+      "loss": 0.5985,
+      "step": 50600
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 12.642940521240234,
+      "learning_rate": 7.918543581446747e-07,
+      "loss": 0.6556,
+      "step": 50601
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.545468807220459,
+      "learning_rate": 7.917119738322104e-07,
+      "loss": 0.7687,
+      "step": 50602
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.4460322856903076,
+      "learning_rate": 7.915696012214197e-07,
+      "loss": 0.7522,
+      "step": 50603
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 5.203746318817139,
+      "learning_rate": 7.914272403126966e-07,
+      "loss": 0.6436,
+      "step": 50604
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.3203251361846924,
+      "learning_rate": 7.912848911064386e-07,
+      "loss": 0.4983,
+      "step": 50605
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.3177146911621094,
+      "learning_rate": 7.911425536030398e-07,
+      "loss": 0.5589,
+      "step": 50606
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.9005587100982666,
+      "learning_rate": 7.910002278028983e-07,
+      "loss": 0.5916,
+      "step": 50607
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.971343517303467,
+      "learning_rate": 7.908579137064066e-07,
+      "loss": 0.6219,
+      "step": 50608
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.5915369987487793,
+      "learning_rate": 7.907156113139641e-07,
+      "loss": 0.6067,
+      "step": 50609
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.3563146591186523,
+      "learning_rate": 7.905733206259631e-07,
+      "loss": 0.639,
+      "step": 50610
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.853301763534546,
+      "learning_rate": 7.904310416428018e-07,
+      "loss": 0.5616,
+      "step": 50611
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.2885594367980957,
+      "learning_rate": 7.902887743648741e-07,
+      "loss": 0.5172,
+      "step": 50612
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.990180492401123,
+      "learning_rate": 7.901465187925761e-07,
+      "loss": 0.6694,
+      "step": 50613
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 5.738933563232422,
+      "learning_rate": 7.900042749263037e-07,
+      "loss": 0.7972,
+      "step": 50614
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.4383397102355957,
+      "learning_rate": 7.898620427664533e-07,
+      "loss": 0.7913,
+      "step": 50615
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.0024096965789795,
+      "learning_rate": 7.897198223134183e-07,
+      "loss": 0.4782,
+      "step": 50616
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.42085862159729,
+      "learning_rate": 7.895776135675964e-07,
+      "loss": 0.4124,
+      "step": 50617
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.2838759422302246,
+      "learning_rate": 7.89435416529381e-07,
+      "loss": 0.7579,
+      "step": 50618
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.4178545475006104,
+      "learning_rate": 7.892932311991691e-07,
+      "loss": 0.6157,
+      "step": 50619
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.0029237270355225,
+      "learning_rate": 7.891510575773547e-07,
+      "loss": 0.6309,
+      "step": 50620
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.4792630672454834,
+      "learning_rate": 7.890088956643344e-07,
+      "loss": 0.5842,
+      "step": 50621
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 5.393438339233398,
+      "learning_rate": 7.88866745460502e-07,
+      "loss": 0.5872,
+      "step": 50622
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.108659505844116,
+      "learning_rate": 7.887246069662552e-07,
+      "loss": 0.3185,
+      "step": 50623
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.632803201675415,
+      "learning_rate": 7.885824801819858e-07,
+      "loss": 0.8746,
+      "step": 50624
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.5405399799346924,
+      "learning_rate": 7.88440365108093e-07,
+      "loss": 0.2945,
+      "step": 50625
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.426022529602051,
+      "learning_rate": 7.88298261744968e-07,
+      "loss": 0.4297,
+      "step": 50626
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.4703421592712402,
+      "learning_rate": 7.88156170093009e-07,
+      "loss": 0.5287,
+      "step": 50627
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.101382255554199,
+      "learning_rate": 7.880140901526084e-07,
+      "loss": 0.6141,
+      "step": 50628
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.160165548324585,
+      "learning_rate": 7.878720219241648e-07,
+      "loss": 0.3898,
+      "step": 50629
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.730712413787842,
+      "learning_rate": 7.877299654080694e-07,
+      "loss": 0.6379,
+      "step": 50630
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.5866620540618896,
+      "learning_rate": 7.875879206047204e-07,
+      "loss": 0.618,
+      "step": 50631
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.5526609420776367,
+      "learning_rate": 7.874458875145102e-07,
+      "loss": 0.7826,
+      "step": 50632
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.509060859680176,
+      "learning_rate": 7.873038661378346e-07,
+      "loss": 0.7373,
+      "step": 50633
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.3999102115631104,
+      "learning_rate": 7.871618564750893e-07,
+      "loss": 0.6721,
+      "step": 50634
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.256037950515747,
+      "learning_rate": 7.870198585266697e-07,
+      "loss": 0.7422,
+      "step": 50635
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.1846015453338623,
+      "learning_rate": 7.868778722929687e-07,
+      "loss": 0.3829,
+      "step": 50636
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.333656072616577,
+      "learning_rate": 7.867358977743827e-07,
+      "loss": 0.5638,
+      "step": 50637
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.9323906898498535,
+      "learning_rate": 7.865939349713053e-07,
+      "loss": 0.7477,
+      "step": 50638
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.2054717540740967,
+      "learning_rate": 7.864519838841328e-07,
+      "loss": 0.4319,
+      "step": 50639
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.43776273727417,
+      "learning_rate": 7.863100445132576e-07,
+      "loss": 0.462,
+      "step": 50640
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.8035850524902344,
+      "learning_rate": 7.861681168590767e-07,
+      "loss": 0.4122,
+      "step": 50641
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.6488819122314453,
+      "learning_rate": 7.860262009219827e-07,
+      "loss": 0.7636,
+      "step": 50642
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.406345844268799,
+      "learning_rate": 7.858842967023728e-07,
+      "loss": 0.2926,
+      "step": 50643
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.333251714706421,
+      "learning_rate": 7.857424042006384e-07,
+      "loss": 0.5394,
+      "step": 50644
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.566516876220703,
+      "learning_rate": 7.856005234171754e-07,
+      "loss": 0.4353,
+      "step": 50645
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 1.963910460472107,
+      "learning_rate": 7.854586543523801e-07,
+      "loss": 0.3149,
+      "step": 50646
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.7739744186401367,
+      "learning_rate": 7.853167970066444e-07,
+      "loss": 0.6058,
+      "step": 50647
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 25.565006256103516,
+      "learning_rate": 7.851749513803652e-07,
+      "loss": 0.5695,
+      "step": 50648
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 4.013509750366211,
+      "learning_rate": 7.85033117473934e-07,
+      "loss": 0.6752,
+      "step": 50649
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 4.248492240905762,
+      "learning_rate": 7.84891295287748e-07,
+      "loss": 0.7015,
+      "step": 50650
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.3582208156585693,
+      "learning_rate": 7.847494848221992e-07,
+      "loss": 0.585,
+      "step": 50651
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.7794477939605713,
+      "learning_rate": 7.846076860776842e-07,
+      "loss": 0.7044,
+      "step": 50652
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.1458916664123535,
+      "learning_rate": 7.84465899054595e-07,
+      "loss": 0.4943,
+      "step": 50653
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.316887378692627,
+      "learning_rate": 7.843241237533266e-07,
+      "loss": 0.2996,
+      "step": 50654
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.246398448944092,
+      "learning_rate": 7.841823601742755e-07,
+      "loss": 0.5596,
+      "step": 50655
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.529325485229492,
+      "learning_rate": 7.84040608317832e-07,
+      "loss": 0.4974,
+      "step": 50656
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.647433042526245,
+      "learning_rate": 7.838988681843923e-07,
+      "loss": 0.7421,
+      "step": 50657
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.7946064472198486,
+      "learning_rate": 7.837571397743521e-07,
+      "loss": 0.5793,
+      "step": 50658
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.69903826713562,
+      "learning_rate": 7.836154230881022e-07,
+      "loss": 0.7122,
+      "step": 50659
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.4354443550109863,
+      "learning_rate": 7.834737181260399e-07,
+      "loss": 0.6751,
+      "step": 50660
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.143465995788574,
+      "learning_rate": 7.833320248885562e-07,
+      "loss": 0.4629,
+      "step": 50661
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.0227997303009033,
+      "learning_rate": 7.831903433760479e-07,
+      "loss": 0.476,
+      "step": 50662
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.5307443141937256,
+      "learning_rate": 7.830486735889065e-07,
+      "loss": 0.3714,
+      "step": 50663
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.8357980251312256,
+      "learning_rate": 7.829070155275281e-07,
+      "loss": 0.601,
+      "step": 50664
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.1735923290252686,
+      "learning_rate": 7.827653691923043e-07,
+      "loss": 0.6627,
+      "step": 50665
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.388929605484009,
+      "learning_rate": 7.826237345836313e-07,
+      "loss": 0.3166,
+      "step": 50666
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.8678863048553467,
+      "learning_rate": 7.824821117019005e-07,
+      "loss": 0.5215,
+      "step": 50667
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.4460275173187256,
+      "learning_rate": 7.823405005475087e-07,
+      "loss": 0.7522,
+      "step": 50668
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.1412813663482666,
+      "learning_rate": 7.821989011208464e-07,
+      "loss": 0.4784,
+      "step": 50669
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.624328136444092,
+      "learning_rate": 7.820573134223102e-07,
+      "loss": 0.5633,
+      "step": 50670
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.1714441776275635,
+      "learning_rate": 7.819157374522912e-07,
+      "loss": 0.6268,
+      "step": 50671
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.897867202758789,
+      "learning_rate": 7.817741732111861e-07,
+      "loss": 0.7108,
+      "step": 50672
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.9083073139190674,
+      "learning_rate": 7.816326206993852e-07,
+      "loss": 0.561,
+      "step": 50673
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 1.8328557014465332,
+      "learning_rate": 7.814910799172843e-07,
+      "loss": 0.4511,
+      "step": 50674
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.0125038623809814,
+      "learning_rate": 7.813495508652757e-07,
+      "loss": 0.5474,
+      "step": 50675
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.8529677391052246,
+      "learning_rate": 7.81208033543755e-07,
+      "loss": 0.6118,
+      "step": 50676
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.798870086669922,
+      "learning_rate": 7.810665279531138e-07,
+      "loss": 0.9414,
+      "step": 50677
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.149970293045044,
+      "learning_rate": 7.809250340937463e-07,
+      "loss": 0.4782,
+      "step": 50678
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 5.211697578430176,
+      "learning_rate": 7.807835519660456e-07,
+      "loss": 0.9169,
+      "step": 50679
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.3097455501556396,
+      "learning_rate": 7.806420815704058e-07,
+      "loss": 0.7922,
+      "step": 50680
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.4080116748809814,
+      "learning_rate": 7.805006229072188e-07,
+      "loss": 0.5095,
+      "step": 50681
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.7910783290863037,
+      "learning_rate": 7.803591759768802e-07,
+      "loss": 0.4019,
+      "step": 50682
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.7747318744659424,
+      "learning_rate": 7.802177407797806e-07,
+      "loss": 0.8141,
+      "step": 50683
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.1443629264831543,
+      "learning_rate": 7.800763173163162e-07,
+      "loss": 0.6518,
+      "step": 50684
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.4782981872558594,
+      "learning_rate": 7.799349055868771e-07,
+      "loss": 0.6696,
+      "step": 50685
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.3967692852020264,
+      "learning_rate": 7.797935055918598e-07,
+      "loss": 0.5684,
+      "step": 50686
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.586853504180908,
+      "learning_rate": 7.796521173316546e-07,
+      "loss": 0.3482,
+      "step": 50687
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.3640410900115967,
+      "learning_rate": 7.795107408066566e-07,
+      "loss": 0.4096,
+      "step": 50688
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.3534531593322754,
+      "learning_rate": 7.793693760172578e-07,
+      "loss": 0.5833,
+      "step": 50689
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.3647243976593018,
+      "learning_rate": 7.792280229638521e-07,
+      "loss": 0.533,
+      "step": 50690
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.5270683765411377,
+      "learning_rate": 7.790866816468313e-07,
+      "loss": 0.8382,
+      "step": 50691
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.3325002193450928,
+      "learning_rate": 7.789453520665902e-07,
+      "loss": 0.7221,
+      "step": 50692
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.876000165939331,
+      "learning_rate": 7.788040342235198e-07,
+      "loss": 0.3785,
+      "step": 50693
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.407468318939209,
+      "learning_rate": 7.786627281180154e-07,
+      "loss": 0.3979,
+      "step": 50694
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 5.125589370727539,
+      "learning_rate": 7.78521433750466e-07,
+      "loss": 0.7192,
+      "step": 50695
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.079498529434204,
+      "learning_rate": 7.783801511212696e-07,
+      "loss": 0.5214,
+      "step": 50696
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 4.233657360076904,
+      "learning_rate": 7.782388802308156e-07,
+      "loss": 0.6444,
+      "step": 50697
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.7430810928344727,
+      "learning_rate": 7.780976210794989e-07,
+      "loss": 0.5838,
+      "step": 50698
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.4000442028045654,
+      "learning_rate": 7.779563736677092e-07,
+      "loss": 0.4067,
+      "step": 50699
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 7.631916046142578,
+      "learning_rate": 7.778151379958431e-07,
+      "loss": 0.6209,
+      "step": 50700
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.8091766834259033,
+      "learning_rate": 7.776739140642902e-07,
+      "loss": 0.6372,
+      "step": 50701
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.9390292167663574,
+      "learning_rate": 7.775327018734458e-07,
+      "loss": 0.5203,
+      "step": 50702
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.458414316177368,
+      "learning_rate": 7.773915014236994e-07,
+      "loss": 0.538,
+      "step": 50703
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.2812979221343994,
+      "learning_rate": 7.772503127154474e-07,
+      "loss": 0.5603,
+      "step": 50704
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.6214561462402344,
+      "learning_rate": 7.771091357490784e-07,
+      "loss": 0.7829,
+      "step": 50705
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.6652190685272217,
+      "learning_rate": 7.769679705249872e-07,
+      "loss": 0.5391,
+      "step": 50706
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.555723190307617,
+      "learning_rate": 7.768268170435678e-07,
+      "loss": 0.5121,
+      "step": 50707
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.922015905380249,
+      "learning_rate": 7.766856753052093e-07,
+      "loss": 0.5656,
+      "step": 50708
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.3439247608184814,
+      "learning_rate": 7.765445453103071e-07,
+      "loss": 0.5278,
+      "step": 50709
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.4100122451782227,
+      "learning_rate": 7.764034270592514e-07,
+      "loss": 0.4564,
+      "step": 50710
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.5256712436676025,
+      "learning_rate": 7.762623205524361e-07,
+      "loss": 0.5949,
+      "step": 50711
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.615365743637085,
+      "learning_rate": 7.761212257902523e-07,
+      "loss": 0.7845,
+      "step": 50712
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.248220443725586,
+      "learning_rate": 7.759801427730945e-07,
+      "loss": 0.7371,
+      "step": 50713
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.5688467025756836,
+      "learning_rate": 7.758390715013519e-07,
+      "loss": 0.6316,
+      "step": 50714
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.11010479927063,
+      "learning_rate": 7.756980119754187e-07,
+      "loss": 0.5854,
+      "step": 50715
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.570441961288452,
+      "learning_rate": 7.755569641956884e-07,
+      "loss": 0.5071,
+      "step": 50716
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.1400983333587646,
+      "learning_rate": 7.754159281625501e-07,
+      "loss": 0.395,
+      "step": 50717
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 4.127029895782471,
+      "learning_rate": 7.752749038763974e-07,
+      "loss": 0.682,
+      "step": 50718
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.4317336082458496,
+      "learning_rate": 7.751338913376239e-07,
+      "loss": 0.6699,
+      "step": 50719
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 5.874667644500732,
+      "learning_rate": 7.749928905466192e-07,
+      "loss": 0.6568,
+      "step": 50720
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.491788387298584,
+      "learning_rate": 7.748519015037775e-07,
+      "loss": 0.5588,
+      "step": 50721
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 4.559357166290283,
+      "learning_rate": 7.747109242094891e-07,
+      "loss": 0.6131,
+      "step": 50722
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.258761405944824,
+      "learning_rate": 7.745699586641481e-07,
+      "loss": 0.4666,
+      "step": 50723
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.136023759841919,
+      "learning_rate": 7.744290048681436e-07,
+      "loss": 0.5118,
+      "step": 50724
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 5.850556373596191,
+      "learning_rate": 7.742880628218702e-07,
+      "loss": 0.8664,
+      "step": 50725
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.3765242099761963,
+      "learning_rate": 7.741471325257177e-07,
+      "loss": 0.5351,
+      "step": 50726
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.759575128555298,
+      "learning_rate": 7.740062139800802e-07,
+      "loss": 0.5085,
+      "step": 50727
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.3227179050445557,
+      "learning_rate": 7.738653071853475e-07,
+      "loss": 0.5726,
+      "step": 50728
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.6798460483551025,
+      "learning_rate": 7.73724412141913e-07,
+      "loss": 0.5746,
+      "step": 50729
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.1691904067993164,
+      "learning_rate": 7.735835288501664e-07,
+      "loss": 0.7081,
+      "step": 50730
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.807586431503296,
+      "learning_rate": 7.73442657310502e-07,
+      "loss": 0.4242,
+      "step": 50731
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.31288480758667,
+      "learning_rate": 7.733017975233093e-07,
+      "loss": 0.9927,
+      "step": 50732
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.8723084926605225,
+      "learning_rate": 7.731609494889819e-07,
+      "loss": 0.6192,
+      "step": 50733
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.284491539001465,
+      "learning_rate": 7.730201132079096e-07,
+      "loss": 0.5391,
+      "step": 50734
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 4.166163921356201,
+      "learning_rate": 7.728792886804848e-07,
+      "loss": 0.5005,
+      "step": 50735
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.793139934539795,
+      "learning_rate": 7.727384759070989e-07,
+      "loss": 0.7018,
+      "step": 50736
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.349724292755127,
+      "learning_rate": 7.725976748881453e-07,
+      "loss": 0.5843,
+      "step": 50737
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.2994892597198486,
+      "learning_rate": 7.724568856240128e-07,
+      "loss": 0.6964,
+      "step": 50738
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.723029613494873,
+      "learning_rate": 7.723161081150948e-07,
+      "loss": 0.521,
+      "step": 50739
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.471554756164551,
+      "learning_rate": 7.721753423617811e-07,
+      "loss": 0.4867,
+      "step": 50740
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.8424787521362305,
+      "learning_rate": 7.720345883644648e-07,
+      "loss": 0.5878,
+      "step": 50741
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 4.271583557128906,
+      "learning_rate": 7.718938461235359e-07,
+      "loss": 0.4707,
+      "step": 50742
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.2512831687927246,
+      "learning_rate": 7.717531156393871e-07,
+      "loss": 0.646,
+      "step": 50743
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.595475912094116,
+      "learning_rate": 7.716123969124078e-07,
+      "loss": 0.4566,
+      "step": 50744
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.3970909118652344,
+      "learning_rate": 7.714716899429913e-07,
+      "loss": 0.4871,
+      "step": 50745
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.3500778675079346,
+      "learning_rate": 7.713309947315273e-07,
+      "loss": 0.5627,
+      "step": 50746
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 4.0395002365112305,
+      "learning_rate": 7.711903112784086e-07,
+      "loss": 0.9268,
+      "step": 50747
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.669487953186035,
+      "learning_rate": 7.710496395840244e-07,
+      "loss": 0.5133,
+      "step": 50748
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 1.871086835861206,
+      "learning_rate": 7.709089796487684e-07,
+      "loss": 0.6171,
+      "step": 50749
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 4.475836753845215,
+      "learning_rate": 7.707683314730291e-07,
+      "loss": 0.3724,
+      "step": 50750
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.6292779445648193,
+      "learning_rate": 7.706276950571995e-07,
+      "loss": 0.4776,
+      "step": 50751
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 4.381064414978027,
+      "learning_rate": 7.704870704016692e-07,
+      "loss": 0.5172,
+      "step": 50752
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.209102153778076,
+      "learning_rate": 7.703464575068309e-07,
+      "loss": 0.6473,
+      "step": 50753
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.4459588527679443,
+      "learning_rate": 7.702058563730736e-07,
+      "loss": 0.4733,
+      "step": 50754
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.4036197662353516,
+      "learning_rate": 7.700652670007902e-07,
+      "loss": 0.6955,
+      "step": 50755
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.866990804672241,
+      "learning_rate": 7.699246893903684e-07,
+      "loss": 0.5191,
+      "step": 50756
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.1736817359924316,
+      "learning_rate": 7.697841235422038e-07,
+      "loss": 0.7225,
+      "step": 50757
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.8063840866088867,
+      "learning_rate": 7.696435694566839e-07,
+      "loss": 0.6636,
+      "step": 50758
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 4.471670627593994,
+      "learning_rate": 7.695030271342013e-07,
+      "loss": 0.601,
+      "step": 50759
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.2217631340026855,
+      "learning_rate": 7.69362496575145e-07,
+      "loss": 0.5932,
+      "step": 50760
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.7306416034698486,
+      "learning_rate": 7.692219777799081e-07,
+      "loss": 0.5167,
+      "step": 50761
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 5.681035995483398,
+      "learning_rate": 7.690814707488787e-07,
+      "loss": 0.4825,
+      "step": 50762
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.360915184020996,
+      "learning_rate": 7.689409754824501e-07,
+      "loss": 0.4926,
+      "step": 50763
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.1946487426757812,
+      "learning_rate": 7.688004919810105e-07,
+      "loss": 0.5563,
+      "step": 50764
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.091996192932129,
+      "learning_rate": 7.686600202449529e-07,
+      "loss": 0.4435,
+      "step": 50765
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 4.905395984649658,
+      "learning_rate": 7.685195602746653e-07,
+      "loss": 0.7281,
+      "step": 50766
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 6.502010345458984,
+      "learning_rate": 7.683791120705409e-07,
+      "loss": 0.5261,
+      "step": 50767
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.530507802963257,
+      "learning_rate": 7.682386756329679e-07,
+      "loss": 0.7428,
+      "step": 50768
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.088456630706787,
+      "learning_rate": 7.680982509623374e-07,
+      "loss": 0.6255,
+      "step": 50769
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.739398717880249,
+      "learning_rate": 7.679578380590418e-07,
+      "loss": 0.5554,
+      "step": 50770
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.8690316677093506,
+      "learning_rate": 7.678174369234692e-07,
+      "loss": 0.4705,
+      "step": 50771
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 4.990316867828369,
+      "learning_rate": 7.676770475560119e-07,
+      "loss": 0.6513,
+      "step": 50772
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.315159559249878,
+      "learning_rate": 7.67536669957058e-07,
+      "loss": 0.6318,
+      "step": 50773
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.504974842071533,
+      "learning_rate": 7.673963041270005e-07,
+      "loss": 0.678,
+      "step": 50774
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.604053258895874,
+      "learning_rate": 7.672559500662269e-07,
+      "loss": 0.432,
+      "step": 50775
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.6144566535949707,
+      "learning_rate": 7.671156077751291e-07,
+      "loss": 0.8286,
+      "step": 50776
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 4.88271427154541,
+      "learning_rate": 7.669752772540984e-07,
+      "loss": 0.5765,
+      "step": 50777
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.299140214920044,
+      "learning_rate": 7.668349585035223e-07,
+      "loss": 0.5937,
+      "step": 50778
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.104463815689087,
+      "learning_rate": 7.666946515237922e-07,
+      "loss": 0.4659,
+      "step": 50779
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.785163640975952,
+      "learning_rate": 7.665543563153e-07,
+      "loss": 0.8267,
+      "step": 50780
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 4.161349773406982,
+      "learning_rate": 7.664140728784331e-07,
+      "loss": 0.6362,
+      "step": 50781
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.3474977016448975,
+      "learning_rate": 7.662738012135834e-07,
+      "loss": 0.679,
+      "step": 50782
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.6960031986236572,
+      "learning_rate": 7.661335413211396e-07,
+      "loss": 0.5436,
+      "step": 50783
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 1.885144829750061,
+      "learning_rate": 7.659932932014935e-07,
+      "loss": 0.5992,
+      "step": 50784
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.281947135925293,
+      "learning_rate": 7.658530568550327e-07,
+      "loss": 0.561,
+      "step": 50785
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.532665729522705,
+      "learning_rate": 7.657128322821494e-07,
+      "loss": 0.5308,
+      "step": 50786
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.9055068492889404,
+      "learning_rate": 7.655726194832314e-07,
+      "loss": 0.6897,
+      "step": 50787
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.08272647857666,
+      "learning_rate": 7.654324184586709e-07,
+      "loss": 0.3713,
+      "step": 50788
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.000966787338257,
+      "learning_rate": 7.652922292088554e-07,
+      "loss": 0.4905,
+      "step": 50789
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 4.716811656951904,
+      "learning_rate": 7.65152051734177e-07,
+      "loss": 0.3771,
+      "step": 50790
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 1.3803573846817017,
+      "learning_rate": 7.650118860350237e-07,
+      "loss": 0.2659,
+      "step": 50791
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 1.7633417844772339,
+      "learning_rate": 7.648717321117866e-07,
+      "loss": 0.3508,
+      "step": 50792
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.2839930057525635,
+      "learning_rate": 7.647315899648533e-07,
+      "loss": 0.6718,
+      "step": 50793
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.8627002239227295,
+      "learning_rate": 7.645914595946164e-07,
+      "loss": 0.6345,
+      "step": 50794
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.7193455696105957,
+      "learning_rate": 7.644513410014625e-07,
+      "loss": 0.8722,
+      "step": 50795
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.786080837249756,
+      "learning_rate": 7.643112341857828e-07,
+      "loss": 0.7511,
+      "step": 50796
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.3502821922302246,
+      "learning_rate": 7.641711391479673e-07,
+      "loss": 0.7021,
+      "step": 50797
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.7372546195983887,
+      "learning_rate": 7.640310558884056e-07,
+      "loss": 0.6394,
+      "step": 50798
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.3100574016571045,
+      "learning_rate": 7.63890984407486e-07,
+      "loss": 0.731,
+      "step": 50799
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.0357859134674072,
+      "learning_rate": 7.637509247055997e-07,
+      "loss": 0.7081,
+      "step": 50800
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.229499578475952,
+      "learning_rate": 7.636108767831341e-07,
+      "loss": 0.663,
+      "step": 50801
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.4097001552581787,
+      "learning_rate": 7.634708406404807e-07,
+      "loss": 0.5227,
+      "step": 50802
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.784396171569824,
+      "learning_rate": 7.633308162780267e-07,
+      "loss": 0.514,
+      "step": 50803
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.666292667388916,
+      "learning_rate": 7.631908036961638e-07,
+      "loss": 0.6126,
+      "step": 50804
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.4331471920013428,
+      "learning_rate": 7.630508028952788e-07,
+      "loss": 0.4832,
+      "step": 50805
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.036851406097412,
+      "learning_rate": 7.629108138757635e-07,
+      "loss": 0.6121,
+      "step": 50806
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.321474075317383,
+      "learning_rate": 7.627708366380049e-07,
+      "loss": 0.4556,
+      "step": 50807
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.498706579208374,
+      "learning_rate": 7.626308711823949e-07,
+      "loss": 0.3852,
+      "step": 50808
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.4572558403015137,
+      "learning_rate": 7.624909175093192e-07,
+      "loss": 0.3204,
+      "step": 50809
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.3625922203063965,
+      "learning_rate": 7.623509756191705e-07,
+      "loss": 0.5006,
+      "step": 50810
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.5276455879211426,
+      "learning_rate": 7.622110455123349e-07,
+      "loss": 0.6171,
+      "step": 50811
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.546640157699585,
+      "learning_rate": 7.620711271892039e-07,
+      "loss": 0.4551,
+      "step": 50812
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 4.4750590324401855,
+      "learning_rate": 7.61931220650165e-07,
+      "loss": 0.4997,
+      "step": 50813
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.594285488128662,
+      "learning_rate": 7.617913258956083e-07,
+      "loss": 0.4318,
+      "step": 50814
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 4.437295436859131,
+      "learning_rate": 7.616514429259215e-07,
+      "loss": 0.4408,
+      "step": 50815
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.0429928302764893,
+      "learning_rate": 7.61511571741495e-07,
+      "loss": 0.84,
+      "step": 50816
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.979266405105591,
+      "learning_rate": 7.61371712342715e-07,
+      "loss": 0.6648,
+      "step": 50817
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.250988006591797,
+      "learning_rate": 7.612318647299749e-07,
+      "loss": 0.5326,
+      "step": 50818
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.091667890548706,
+      "learning_rate": 7.610920289036594e-07,
+      "loss": 0.6659,
+      "step": 50819
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.6963508129119873,
+      "learning_rate": 7.60952204864161e-07,
+      "loss": 0.5866,
+      "step": 50820
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.559664011001587,
+      "learning_rate": 7.608123926118649e-07,
+      "loss": 0.4907,
+      "step": 50821
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.9097909927368164,
+      "learning_rate": 7.606725921471625e-07,
+      "loss": 0.4398,
+      "step": 50822
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.3116090297698975,
+      "learning_rate": 7.605328034704407e-07,
+      "loss": 0.3177,
+      "step": 50823
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 4.269132614135742,
+      "learning_rate": 7.603930265820902e-07,
+      "loss": 0.5607,
+      "step": 50824
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.309990882873535,
+      "learning_rate": 7.60253261482497e-07,
+      "loss": 0.3411,
+      "step": 50825
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.4498655796051025,
+      "learning_rate": 7.601135081720528e-07,
+      "loss": 0.5779,
+      "step": 50826
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 4.560762405395508,
+      "learning_rate": 7.599737666511431e-07,
+      "loss": 0.5375,
+      "step": 50827
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.296329975128174,
+      "learning_rate": 7.598340369201595e-07,
+      "loss": 0.5943,
+      "step": 50828
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.8242712020874023,
+      "learning_rate": 7.596943189794876e-07,
+      "loss": 0.625,
+      "step": 50829
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.669802665710449,
+      "learning_rate": 7.595546128295178e-07,
+      "loss": 0.5752,
+      "step": 50830
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.4357593059539795,
+      "learning_rate": 7.594149184706389e-07,
+      "loss": 0.4356,
+      "step": 50831
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.355241537094116,
+      "learning_rate": 7.592752359032373e-07,
+      "loss": 0.7036,
+      "step": 50832
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.4715638160705566,
+      "learning_rate": 7.591355651277038e-07,
+      "loss": 0.8311,
+      "step": 50833
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 1.9384442567825317,
+      "learning_rate": 7.589959061444247e-07,
+      "loss": 0.5242,
+      "step": 50834
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.5000360012054443,
+      "learning_rate": 7.588562589537906e-07,
+      "loss": 0.6827,
+      "step": 50835
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 1.842874526977539,
+      "learning_rate": 7.587166235561871e-07,
+      "loss": 0.6338,
+      "step": 50836
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.316821336746216,
+      "learning_rate": 7.585769999520038e-07,
+      "loss": 0.615,
+      "step": 50837
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.4924333095550537,
+      "learning_rate": 7.584373881416302e-07,
+      "loss": 0.6076,
+      "step": 50838
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.2654850482940674,
+      "learning_rate": 7.582977881254522e-07,
+      "loss": 0.4967,
+      "step": 50839
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 6.733548641204834,
+      "learning_rate": 7.581581999038601e-07,
+      "loss": 0.5345,
+      "step": 50840
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.778434991836548,
+      "learning_rate": 7.580186234772396e-07,
+      "loss": 0.4862,
+      "step": 50841
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.3140342235565186,
+      "learning_rate": 7.578790588459805e-07,
+      "loss": 0.8303,
+      "step": 50842
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.5992064476013184,
+      "learning_rate": 7.57739506010472e-07,
+      "loss": 0.7444,
+      "step": 50843
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.9883153438568115,
+      "learning_rate": 7.575999649710991e-07,
+      "loss": 0.4884,
+      "step": 50844
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.835263252258301,
+      "learning_rate": 7.574604357282533e-07,
+      "loss": 0.6987,
+      "step": 50845
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.525660276412964,
+      "learning_rate": 7.573209182823188e-07,
+      "loss": 0.4019,
+      "step": 50846
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.9318697452545166,
+      "learning_rate": 7.571814126336874e-07,
+      "loss": 0.4597,
+      "step": 50847
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.0008938312530518,
+      "learning_rate": 7.570419187827433e-07,
+      "loss": 0.4636,
+      "step": 50848
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.988551139831543,
+      "learning_rate": 7.569024367298778e-07,
+      "loss": 0.5388,
+      "step": 50849
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.6354503631591797,
+      "learning_rate": 7.567629664754755e-07,
+      "loss": 0.5062,
+      "step": 50850
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 4.134912490844727,
+      "learning_rate": 7.566235080199274e-07,
+      "loss": 0.6278,
+      "step": 50851
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.1611475944519043,
+      "learning_rate": 7.564840613636182e-07,
+      "loss": 0.4923,
+      "step": 50852
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.4522922039031982,
+      "learning_rate": 7.563446265069385e-07,
+      "loss": 0.4066,
+      "step": 50853
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.532151699066162,
+      "learning_rate": 7.562052034502737e-07,
+      "loss": 0.4769,
+      "step": 50854
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.6960973739624023,
+      "learning_rate": 7.560657921940135e-07,
+      "loss": 0.562,
+      "step": 50855
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.87731671333313,
+      "learning_rate": 7.559263927385435e-07,
+      "loss": 0.9404,
+      "step": 50856
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 4.301793098449707,
+      "learning_rate": 7.557870050842519e-07,
+      "loss": 0.5997,
+      "step": 50857
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.0327439308166504,
+      "learning_rate": 7.556476292315273e-07,
+      "loss": 0.7838,
+      "step": 50858
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.633044481277466,
+      "learning_rate": 7.555082651807572e-07,
+      "loss": 0.6964,
+      "step": 50859
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.598501682281494,
+      "learning_rate": 7.553689129323277e-07,
+      "loss": 0.491,
+      "step": 50860
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 1.9939810037612915,
+      "learning_rate": 7.552295724866288e-07,
+      "loss": 0.476,
+      "step": 50861
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.497156858444214,
+      "learning_rate": 7.550902438440444e-07,
+      "loss": 0.4959,
+      "step": 50862
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.1999480724334717,
+      "learning_rate": 7.549509270049655e-07,
+      "loss": 0.5306,
+      "step": 50863
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 4.655521392822266,
+      "learning_rate": 7.548116219697766e-07,
+      "loss": 0.7474,
+      "step": 50864
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 6.042613983154297,
+      "learning_rate": 7.546723287388674e-07,
+      "loss": 0.885,
+      "step": 50865
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.096341371536255,
+      "learning_rate": 7.545330473126233e-07,
+      "loss": 0.3754,
+      "step": 50866
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 1.8553977012634277,
+      "learning_rate": 7.543937776914329e-07,
+      "loss": 0.4262,
+      "step": 50867
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.8716742992401123,
+      "learning_rate": 7.542545198756824e-07,
+      "loss": 0.6415,
+      "step": 50868
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.631788730621338,
+      "learning_rate": 7.541152738657604e-07,
+      "loss": 0.5552,
+      "step": 50869
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.413243055343628,
+      "learning_rate": 7.539760396620521e-07,
+      "loss": 0.5082,
+      "step": 50870
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.4276440143585205,
+      "learning_rate": 7.538368172649474e-07,
+      "loss": 0.4417,
+      "step": 50871
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 9.900823593139648,
+      "learning_rate": 7.536976066748303e-07,
+      "loss": 0.6091,
+      "step": 50872
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.015273094177246,
+      "learning_rate": 7.535584078920904e-07,
+      "loss": 0.5194,
+      "step": 50873
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.210052251815796,
+      "learning_rate": 7.534192209171132e-07,
+      "loss": 0.5318,
+      "step": 50874
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.346243143081665,
+      "learning_rate": 7.532800457502876e-07,
+      "loss": 0.4704,
+      "step": 50875
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.724388837814331,
+      "learning_rate": 7.531408823919977e-07,
+      "loss": 0.5226,
+      "step": 50876
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.2853949069976807,
+      "learning_rate": 7.530017308426335e-07,
+      "loss": 0.4441,
+      "step": 50877
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.3256638050079346,
+      "learning_rate": 7.528625911025778e-07,
+      "loss": 0.3761,
+      "step": 50878
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 4.424319267272949,
+      "learning_rate": 7.52723463172223e-07,
+      "loss": 0.6194,
+      "step": 50879
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.25464129447937,
+      "learning_rate": 7.525843470519517e-07,
+      "loss": 0.4759,
+      "step": 50880
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.117311716079712,
+      "learning_rate": 7.524452427421535e-07,
+      "loss": 0.9567,
+      "step": 50881
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 9.425837516784668,
+      "learning_rate": 7.523061502432133e-07,
+      "loss": 0.8476,
+      "step": 50882
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.0647103786468506,
+      "learning_rate": 7.521670695555189e-07,
+      "loss": 0.3911,
+      "step": 50883
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.752509117126465,
+      "learning_rate": 7.520280006794556e-07,
+      "loss": 0.6502,
+      "step": 50884
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.94415283203125,
+      "learning_rate": 7.518889436154125e-07,
+      "loss": 0.3757,
+      "step": 50885
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.1177899837493896,
+      "learning_rate": 7.517498983637733e-07,
+      "loss": 0.6509,
+      "step": 50886
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.69637393951416,
+      "learning_rate": 7.516108649249276e-07,
+      "loss": 0.7388,
+      "step": 50887
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.8415517807006836,
+      "learning_rate": 7.514718432992596e-07,
+      "loss": 0.6441,
+      "step": 50888
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.8805959224700928,
+      "learning_rate": 7.513328334871578e-07,
+      "loss": 0.5373,
+      "step": 50889
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 1.977592945098877,
+      "learning_rate": 7.511938354890069e-07,
+      "loss": 0.4996,
+      "step": 50890
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.049471855163574,
+      "learning_rate": 7.510548493051939e-07,
+      "loss": 0.4303,
+      "step": 50891
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.889115333557129,
+      "learning_rate": 7.509158749361073e-07,
+      "loss": 0.5576,
+      "step": 50892
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 1.778870940208435,
+      "learning_rate": 7.507769123821307e-07,
+      "loss": 0.4386,
+      "step": 50893
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.9319071769714355,
+      "learning_rate": 7.506379616436526e-07,
+      "loss": 0.5203,
+      "step": 50894
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.8215529918670654,
+      "learning_rate": 7.504990227210573e-07,
+      "loss": 0.6927,
+      "step": 50895
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.5607357025146484,
+      "learning_rate": 7.503600956147339e-07,
+      "loss": 0.6235,
+      "step": 50896
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.018892288208008,
+      "learning_rate": 7.502211803250653e-07,
+      "loss": 0.6179,
+      "step": 50897
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.3998055458068848,
+      "learning_rate": 7.500822768524402e-07,
+      "loss": 0.3756,
+      "step": 50898
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 5.895844459533691,
+      "learning_rate": 7.499433851972448e-07,
+      "loss": 0.7852,
+      "step": 50899
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.2912888526916504,
+      "learning_rate": 7.49804505359864e-07,
+      "loss": 0.459,
+      "step": 50900
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.303748369216919,
+      "learning_rate": 7.496656373406857e-07,
+      "loss": 0.3422,
+      "step": 50901
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.4612278938293457,
+      "learning_rate": 7.495267811400941e-07,
+      "loss": 0.6982,
+      "step": 50902
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.0663270950317383,
+      "learning_rate": 7.493879367584761e-07,
+      "loss": 0.6635,
+      "step": 50903
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.784301519393921,
+      "learning_rate": 7.492491041962191e-07,
+      "loss": 0.6375,
+      "step": 50904
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.803652286529541,
+      "learning_rate": 7.491102834537067e-07,
+      "loss": 0.6742,
+      "step": 50905
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.2966558933258057,
+      "learning_rate": 7.489714745313276e-07,
+      "loss": 0.6565,
+      "step": 50906
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.2096338272094727,
+      "learning_rate": 7.488326774294652e-07,
+      "loss": 0.7454,
+      "step": 50907
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.632831335067749,
+      "learning_rate": 7.486938921485076e-07,
+      "loss": 0.5502,
+      "step": 50908
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.0245440006256104,
+      "learning_rate": 7.485551186888385e-07,
+      "loss": 0.8825,
+      "step": 50909
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.8429551124572754,
+      "learning_rate": 7.484163570508463e-07,
+      "loss": 0.6349,
+      "step": 50910
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.7046425342559814,
+      "learning_rate": 7.482776072349146e-07,
+      "loss": 0.7665,
+      "step": 50911
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 4.856864929199219,
+      "learning_rate": 7.48138869241431e-07,
+      "loss": 0.79,
+      "step": 50912
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.998692750930786,
+      "learning_rate": 7.48000143070779e-07,
+      "loss": 0.4476,
+      "step": 50913
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.910348653793335,
+      "learning_rate": 7.478614287233471e-07,
+      "loss": 0.4319,
+      "step": 50914
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.1978747844696045,
+      "learning_rate": 7.47722726199519e-07,
+      "loss": 0.7668,
+      "step": 50915
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.0443811416625977,
+      "learning_rate": 7.475840354996816e-07,
+      "loss": 0.4007,
+      "step": 50916
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.9061543941497803,
+      "learning_rate": 7.47445356624219e-07,
+      "loss": 0.7094,
+      "step": 50917
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.180690288543701,
+      "learning_rate": 7.473066895735181e-07,
+      "loss": 0.6071,
+      "step": 50918
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.3220577239990234,
+      "learning_rate": 7.471680343479637e-07,
+      "loss": 0.6673,
+      "step": 50919
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.6213009357452393,
+      "learning_rate": 7.470293909479437e-07,
+      "loss": 0.5837,
+      "step": 50920
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.4397132396698,
+      "learning_rate": 7.4689075937384e-07,
+      "loss": 0.4945,
+      "step": 50921
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.2435247898101807,
+      "learning_rate": 7.467521396260413e-07,
+      "loss": 0.663,
+      "step": 50922
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.172238349914551,
+      "learning_rate": 7.466135317049306e-07,
+      "loss": 0.839,
+      "step": 50923
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.7118115425109863,
+      "learning_rate": 7.464749356108952e-07,
+      "loss": 0.7542,
+      "step": 50924
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.480598211288452,
+      "learning_rate": 7.463363513443184e-07,
+      "loss": 0.6141,
+      "step": 50925
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.3773021697998047,
+      "learning_rate": 7.46197778905588e-07,
+      "loss": 0.4293,
+      "step": 50926
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.153078079223633,
+      "learning_rate": 7.460592182950865e-07,
+      "loss": 0.7812,
+      "step": 50927
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.403883934020996,
+      "learning_rate": 7.459206695132026e-07,
+      "loss": 0.5536,
+      "step": 50928
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.8724374771118164,
+      "learning_rate": 7.457821325603181e-07,
+      "loss": 0.5733,
+      "step": 50929
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.7964847087860107,
+      "learning_rate": 7.456436074368211e-07,
+      "loss": 0.7252,
+      "step": 50930
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.35347843170166,
+      "learning_rate": 7.455050941430947e-07,
+      "loss": 0.6053,
+      "step": 50931
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.406970977783203,
+      "learning_rate": 7.453665926795257e-07,
+      "loss": 0.6293,
+      "step": 50932
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.1453263759613037,
+      "learning_rate": 7.45228103046497e-07,
+      "loss": 0.3784,
+      "step": 50933
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.6790568828582764,
+      "learning_rate": 7.450896252443962e-07,
+      "loss": 0.6317,
+      "step": 50934
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.1849656105041504,
+      "learning_rate": 7.449511592736064e-07,
+      "loss": 0.5914,
+      "step": 50935
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.959202289581299,
+      "learning_rate": 7.448127051345144e-07,
+      "loss": 0.5546,
+      "step": 50936
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.449629545211792,
+      "learning_rate": 7.446742628275033e-07,
+      "loss": 0.4988,
+      "step": 50937
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 4.031574249267578,
+      "learning_rate": 7.4453583235296e-07,
+      "loss": 0.5674,
+      "step": 50938
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.497339963912964,
+      "learning_rate": 7.443974137112664e-07,
+      "loss": 0.6559,
+      "step": 50939
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.15800404548645,
+      "learning_rate": 7.442590069028116e-07,
+      "loss": 0.6437,
+      "step": 50940
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 4.14418888092041,
+      "learning_rate": 7.44120611927977e-07,
+      "loss": 0.6964,
+      "step": 50941
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 1.729914903640747,
+      "learning_rate": 7.439822287871495e-07,
+      "loss": 0.4566,
+      "step": 50942
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.5385544300079346,
+      "learning_rate": 7.438438574807122e-07,
+      "loss": 0.6261,
+      "step": 50943
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.7498748302459717,
+      "learning_rate": 7.437054980090525e-07,
+      "loss": 0.5135,
+      "step": 50944
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 5.703289031982422,
+      "learning_rate": 7.435671503725512e-07,
+      "loss": 0.5687,
+      "step": 50945
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.196641683578491,
+      "learning_rate": 7.434288145715968e-07,
+      "loss": 0.5414,
+      "step": 50946
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 4.220518112182617,
+      "learning_rate": 7.432904906065714e-07,
+      "loss": 0.6069,
+      "step": 50947
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 5.303366661071777,
+      "learning_rate": 7.431521784778611e-07,
+      "loss": 0.6635,
+      "step": 50948
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.6319427490234375,
+      "learning_rate": 7.430138781858493e-07,
+      "loss": 0.5086,
+      "step": 50949
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.0530648231506348,
+      "learning_rate": 7.42875589730922e-07,
+      "loss": 0.6533,
+      "step": 50950
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.914973735809326,
+      "learning_rate": 7.427373131134624e-07,
+      "loss": 0.5661,
+      "step": 50951
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 1.9571130275726318,
+      "learning_rate": 7.425990483338557e-07,
+      "loss": 0.4023,
+      "step": 50952
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.8811421394348145,
+      "learning_rate": 7.424607953924861e-07,
+      "loss": 0.709,
+      "step": 50953
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.930055856704712,
+      "learning_rate": 7.423225542897372e-07,
+      "loss": 0.5325,
+      "step": 50954
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.3568294048309326,
+      "learning_rate": 7.421843250259958e-07,
+      "loss": 0.9171,
+      "step": 50955
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.9530017375946045,
+      "learning_rate": 7.420461076016438e-07,
+      "loss": 0.6501,
+      "step": 50956
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.205674886703491,
+      "learning_rate": 7.419079020170677e-07,
+      "loss": 0.6162,
+      "step": 50957
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.4931342601776123,
+      "learning_rate": 7.417697082726494e-07,
+      "loss": 0.6238,
+      "step": 50958
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.512409210205078,
+      "learning_rate": 7.41631526368774e-07,
+      "loss": 0.4986,
+      "step": 50959
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.3574719429016113,
+      "learning_rate": 7.414933563058274e-07,
+      "loss": 0.6108,
+      "step": 50960
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.665403366088867,
+      "learning_rate": 7.413551980841915e-07,
+      "loss": 0.7098,
+      "step": 50961
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.323859214782715,
+      "learning_rate": 7.412170517042527e-07,
+      "loss": 0.6356,
+      "step": 50962
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.679131031036377,
+      "learning_rate": 7.410789171663929e-07,
+      "loss": 0.5257,
+      "step": 50963
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 4.460324764251709,
+      "learning_rate": 7.409407944709979e-07,
+      "loss": 0.5573,
+      "step": 50964
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.836029291152954,
+      "learning_rate": 7.4080268361845e-07,
+      "loss": 0.5859,
+      "step": 50965
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.0117478370666504,
+      "learning_rate": 7.406645846091348e-07,
+      "loss": 0.8144,
+      "step": 50966
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.0917558670043945,
+      "learning_rate": 7.405264974434362e-07,
+      "loss": 0.6422,
+      "step": 50967
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.7677693367004395,
+      "learning_rate": 7.403884221217372e-07,
+      "loss": 0.7218,
+      "step": 50968
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.933368682861328,
+      "learning_rate": 7.402503586444231e-07,
+      "loss": 0.8681,
+      "step": 50969
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.463475227355957,
+      "learning_rate": 7.40112307011876e-07,
+      "loss": 0.4906,
+      "step": 50970
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.7630348205566406,
+      "learning_rate": 7.39974267224482e-07,
+      "loss": 0.5769,
+      "step": 50971
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.9714770317077637,
+      "learning_rate": 7.398362392826224e-07,
+      "loss": 0.671,
+      "step": 50972
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.2817845344543457,
+      "learning_rate": 7.396982231866839e-07,
+      "loss": 0.8457,
+      "step": 50973
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.8871188163757324,
+      "learning_rate": 7.395602189370471e-07,
+      "loss": 0.5878,
+      "step": 50974
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.079785108566284,
+      "learning_rate": 7.394222265340989e-07,
+      "loss": 0.4451,
+      "step": 50975
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.4405460357666016,
+      "learning_rate": 7.392842459782201e-07,
+      "loss": 0.6277,
+      "step": 50976
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.6097590923309326,
+      "learning_rate": 7.391462772697966e-07,
+      "loss": 0.4915,
+      "step": 50977
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.400101661682129,
+      "learning_rate": 7.390083204092102e-07,
+      "loss": 0.5154,
+      "step": 50978
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.730623722076416,
+      "learning_rate": 7.388703753968457e-07,
+      "loss": 0.6376,
+      "step": 50979
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.5449769496917725,
+      "learning_rate": 7.387324422330861e-07,
+      "loss": 0.4386,
+      "step": 50980
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 1.75202214717865,
+      "learning_rate": 7.385945209183165e-07,
+      "loss": 0.3715,
+      "step": 50981
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.315490245819092,
+      "learning_rate": 7.384566114529179e-07,
+      "loss": 0.7274,
+      "step": 50982
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.4049108028411865,
+      "learning_rate": 7.383187138372766e-07,
+      "loss": 0.7243,
+      "step": 50983
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.397350549697876,
+      "learning_rate": 7.381808280717728e-07,
+      "loss": 0.5598,
+      "step": 50984
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.6987061500549316,
+      "learning_rate": 7.380429541567929e-07,
+      "loss": 0.5508,
+      "step": 50985
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.598583936691284,
+      "learning_rate": 7.379050920927183e-07,
+      "loss": 0.5324,
+      "step": 50986
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.8666024208068848,
+      "learning_rate": 7.377672418799336e-07,
+      "loss": 0.5469,
+      "step": 50987
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.0116443634033203,
+      "learning_rate": 7.376294035188209e-07,
+      "loss": 0.5863,
+      "step": 50988
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.125441789627075,
+      "learning_rate": 7.374915770097646e-07,
+      "loss": 0.7548,
+      "step": 50989
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.9147489070892334,
+      "learning_rate": 7.373537623531468e-07,
+      "loss": 0.6408,
+      "step": 50990
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 6.70458984375,
+      "learning_rate": 7.372159595493522e-07,
+      "loss": 0.5788,
+      "step": 50991
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 4.458709716796875,
+      "learning_rate": 7.370781685987621e-07,
+      "loss": 0.9165,
+      "step": 50992
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.283561944961548,
+      "learning_rate": 7.369403895017624e-07,
+      "loss": 0.5314,
+      "step": 50993
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.1423068046569824,
+      "learning_rate": 7.368026222587326e-07,
+      "loss": 0.7244,
+      "step": 50994
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.7918903827667236,
+      "learning_rate": 7.366648668700593e-07,
+      "loss": 0.6167,
+      "step": 50995
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.1110262870788574,
+      "learning_rate": 7.365271233361227e-07,
+      "loss": 0.4252,
+      "step": 50996
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.2057740688323975,
+      "learning_rate": 7.363893916573078e-07,
+      "loss": 0.6253,
+      "step": 50997
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.971208333969116,
+      "learning_rate": 7.362516718339957e-07,
+      "loss": 0.6157,
+      "step": 50998
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.7041499614715576,
+      "learning_rate": 7.361139638665721e-07,
+      "loss": 0.5027,
+      "step": 50999
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 4.218059539794922,
+      "learning_rate": 7.359762677554155e-07,
+      "loss": 0.6367,
+      "step": 51000
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.4419538974761963,
+      "learning_rate": 7.358385835009141e-07,
+      "loss": 0.5594,
+      "step": 51001
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.698059320449829,
+      "learning_rate": 7.357009111034469e-07,
+      "loss": 0.6104,
+      "step": 51002
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.490339756011963,
+      "learning_rate": 7.355632505633997e-07,
+      "loss": 0.6024,
+      "step": 51003
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.6071646213531494,
+      "learning_rate": 7.354256018811517e-07,
+      "loss": 0.5338,
+      "step": 51004
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.986969232559204,
+      "learning_rate": 7.352879650570893e-07,
+      "loss": 0.5138,
+      "step": 51005
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.378720998764038,
+      "learning_rate": 7.351503400915921e-07,
+      "loss": 0.5214,
+      "step": 51006
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.611631393432617,
+      "learning_rate": 7.350127269850455e-07,
+      "loss": 0.5154,
+      "step": 51007
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 4.181170463562012,
+      "learning_rate": 7.348751257378289e-07,
+      "loss": 0.7406,
+      "step": 51008
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.8992021083831787,
+      "learning_rate": 7.347375363503289e-07,
+      "loss": 0.6748,
+      "step": 51009
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.6431236267089844,
+      "learning_rate": 7.345999588229241e-07,
+      "loss": 0.6473,
+      "step": 51010
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.75187349319458,
+      "learning_rate": 7.344623931560008e-07,
+      "loss": 0.8394,
+      "step": 51011
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.5028276443481445,
+      "learning_rate": 7.34324839349938e-07,
+      "loss": 0.3934,
+      "step": 51012
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.8627593517303467,
+      "learning_rate": 7.341872974051212e-07,
+      "loss": 0.6492,
+      "step": 51013
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 1.9361169338226318,
+      "learning_rate": 7.340497673219304e-07,
+      "loss": 0.2712,
+      "step": 51014
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.5559463500976562,
+      "learning_rate": 7.339122491007494e-07,
+      "loss": 0.5779,
+      "step": 51015
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.1333720684051514,
+      "learning_rate": 7.337747427419612e-07,
+      "loss": 0.5558,
+      "step": 51016
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.8380491733551025,
+      "learning_rate": 7.336372482459458e-07,
+      "loss": 0.5037,
+      "step": 51017
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.5095348358154297,
+      "learning_rate": 7.334997656130887e-07,
+      "loss": 0.6441,
+      "step": 51018
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.471386432647705,
+      "learning_rate": 7.333622948437691e-07,
+      "loss": 0.6175,
+      "step": 51019
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.510993480682373,
+      "learning_rate": 7.332248359383703e-07,
+      "loss": 0.5687,
+      "step": 51020
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 4.024505615234375,
+      "learning_rate": 7.330873888972767e-07,
+      "loss": 0.4379,
+      "step": 51021
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.936800718307495,
+      "learning_rate": 7.329499537208673e-07,
+      "loss": 0.6049,
+      "step": 51022
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.8420052528381348,
+      "learning_rate": 7.328125304095263e-07,
+      "loss": 0.6904,
+      "step": 51023
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.4168691635131836,
+      "learning_rate": 7.326751189636344e-07,
+      "loss": 0.6583,
+      "step": 51024
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 4.946351051330566,
+      "learning_rate": 7.325377193835753e-07,
+      "loss": 0.5917,
+      "step": 51025
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.2812228202819824,
+      "learning_rate": 7.324003316697293e-07,
+      "loss": 0.603,
+      "step": 51026
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.8099937438964844,
+      "learning_rate": 7.322629558224792e-07,
+      "loss": 0.5158,
+      "step": 51027
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 1.9080384969711304,
+      "learning_rate": 7.321255918422082e-07,
+      "loss": 0.5536,
+      "step": 51028
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.8646671772003174,
+      "learning_rate": 7.31988239729296e-07,
+      "loss": 0.5026,
+      "step": 51029
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.4333255290985107,
+      "learning_rate": 7.318508994841267e-07,
+      "loss": 0.5213,
+      "step": 51030
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.182913064956665,
+      "learning_rate": 7.317135711070805e-07,
+      "loss": 0.6224,
+      "step": 51031
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.6995410919189453,
+      "learning_rate": 7.315762545985405e-07,
+      "loss": 0.7856,
+      "step": 51032
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.8561534881591797,
+      "learning_rate": 7.31438949958887e-07,
+      "loss": 0.6112,
+      "step": 51033
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.180091142654419,
+      "learning_rate": 7.31301657188504e-07,
+      "loss": 0.647,
+      "step": 51034
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.224489212036133,
+      "learning_rate": 7.311643762877707e-07,
+      "loss": 0.939,
+      "step": 51035
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.2329704761505127,
+      "learning_rate": 7.310271072570713e-07,
+      "loss": 0.4425,
+      "step": 51036
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.439948797225952,
+      "learning_rate": 7.308898500967854e-07,
+      "loss": 0.4564,
+      "step": 51037
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.8785412311553955,
+      "learning_rate": 7.307526048072966e-07,
+      "loss": 0.7501,
+      "step": 51038
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.290987491607666,
+      "learning_rate": 7.306153713889841e-07,
+      "loss": 0.5421,
+      "step": 51039
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.5495448112487793,
+      "learning_rate": 7.304781498422309e-07,
+      "loss": 0.5001,
+      "step": 51040
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 4.136617183685303,
+      "learning_rate": 7.303409401674189e-07,
+      "loss": 0.573,
+      "step": 51041
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.437241554260254,
+      "learning_rate": 7.302037423649305e-07,
+      "loss": 0.5745,
+      "step": 51042
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.7996246814727783,
+      "learning_rate": 7.300665564351445e-07,
+      "loss": 0.7184,
+      "step": 51043
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.164188861846924,
+      "learning_rate": 7.299293823784454e-07,
+      "loss": 0.6499,
+      "step": 51044
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.642565965652466,
+      "learning_rate": 7.297922201952113e-07,
+      "loss": 0.5962,
+      "step": 51045
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.836103916168213,
+      "learning_rate": 7.296550698858268e-07,
+      "loss": 0.563,
+      "step": 51046
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.8591465950012207,
+      "learning_rate": 7.295179314506712e-07,
+      "loss": 0.6864,
+      "step": 51047
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.2327725887298584,
+      "learning_rate": 7.293808048901269e-07,
+      "loss": 0.5346,
+      "step": 51048
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.2111616134643555,
+      "learning_rate": 7.29243690204574e-07,
+      "loss": 0.5508,
+      "step": 51049
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.2329227924346924,
+      "learning_rate": 7.291065873943959e-07,
+      "loss": 0.4683,
+      "step": 51050
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.6031408309936523,
+      "learning_rate": 7.289694964599709e-07,
+      "loss": 0.4743,
+      "step": 51051
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.5818095207214355,
+      "learning_rate": 7.288324174016831e-07,
+      "loss": 0.5934,
+      "step": 51052
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 4.533164978027344,
+      "learning_rate": 7.286953502199112e-07,
+      "loss": 0.5457,
+      "step": 51053
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.8168742656707764,
+      "learning_rate": 7.285582949150383e-07,
+      "loss": 0.7897,
+      "step": 51054
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.7069692611694336,
+      "learning_rate": 7.28421251487444e-07,
+      "loss": 0.5894,
+      "step": 51055
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.877490758895874,
+      "learning_rate": 7.282842199375107e-07,
+      "loss": 0.6353,
+      "step": 51056
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 4.3554463386535645,
+      "learning_rate": 7.281472002656182e-07,
+      "loss": 0.5817,
+      "step": 51057
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 6.1628618240356445,
+      "learning_rate": 7.280101924721488e-07,
+      "loss": 0.4559,
+      "step": 51058
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.6598737239837646,
+      "learning_rate": 7.278731965574814e-07,
+      "loss": 0.4497,
+      "step": 51059
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.3642077445983887,
+      "learning_rate": 7.27736212522e-07,
+      "loss": 0.6338,
+      "step": 51060
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.1286845207214355,
+      "learning_rate": 7.275992403660809e-07,
+      "loss": 0.6864,
+      "step": 51061
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 4.265109062194824,
+      "learning_rate": 7.274622800901104e-07,
+      "loss": 0.6432,
+      "step": 51062
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.38675594329834,
+      "learning_rate": 7.273253316944656e-07,
+      "loss": 0.4176,
+      "step": 51063
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.385018825531006,
+      "learning_rate": 7.271883951795299e-07,
+      "loss": 0.8905,
+      "step": 51064
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.9813168048858643,
+      "learning_rate": 7.27051470545681e-07,
+      "loss": 0.5287,
+      "step": 51065
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.6823694705963135,
+      "learning_rate": 7.269145577933028e-07,
+      "loss": 0.2656,
+      "step": 51066
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 4.4815592765808105,
+      "learning_rate": 7.267776569227731e-07,
+      "loss": 0.6392,
+      "step": 51067
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.525519371032715,
+      "learning_rate": 7.26640767934475e-07,
+      "loss": 0.5503,
+      "step": 51068
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.8422183990478516,
+      "learning_rate": 7.265038908287875e-07,
+      "loss": 0.6205,
+      "step": 51069
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.5414226055145264,
+      "learning_rate": 7.263670256060923e-07,
+      "loss": 0.4465,
+      "step": 51070
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.75066876411438,
+      "learning_rate": 7.262301722667686e-07,
+      "loss": 0.5075,
+      "step": 51071
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.6358537673950195,
+      "learning_rate": 7.26093330811199e-07,
+      "loss": 0.5907,
+      "step": 51072
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.6335256099700928,
+      "learning_rate": 7.259565012397612e-07,
+      "loss": 0.5669,
+      "step": 51073
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 8.243237495422363,
+      "learning_rate": 7.258196835528392e-07,
+      "loss": 0.5184,
+      "step": 51074
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.4380204677581787,
+      "learning_rate": 7.256828777508096e-07,
+      "loss": 0.5775,
+      "step": 51075
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.1379573345184326,
+      "learning_rate": 7.255460838340555e-07,
+      "loss": 0.565,
+      "step": 51076
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.575371026992798,
+      "learning_rate": 7.254093018029568e-07,
+      "loss": 0.673,
+      "step": 51077
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.781097412109375,
+      "learning_rate": 7.252725316578929e-07,
+      "loss": 0.5464,
+      "step": 51078
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.1191370487213135,
+      "learning_rate": 7.251357733992459e-07,
+      "loss": 0.3689,
+      "step": 51079
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.6355607509613037,
+      "learning_rate": 7.249990270273938e-07,
+      "loss": 0.4443,
+      "step": 51080
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.706723928451538,
+      "learning_rate": 7.248622925427179e-07,
+      "loss": 0.5712,
+      "step": 51081
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.3802199363708496,
+      "learning_rate": 7.24725569945599e-07,
+      "loss": 0.5432,
+      "step": 51082
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.064981698989868,
+      "learning_rate": 7.245888592364164e-07,
+      "loss": 0.8603,
+      "step": 51083
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.971635341644287,
+      "learning_rate": 7.244521604155513e-07,
+      "loss": 0.5924,
+      "step": 51084
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.918597936630249,
+      "learning_rate": 7.243154734833823e-07,
+      "loss": 0.4678,
+      "step": 51085
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.6789889335632324,
+      "learning_rate": 7.24178798440291e-07,
+      "loss": 0.5835,
+      "step": 51086
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.2308413982391357,
+      "learning_rate": 7.240421352866561e-07,
+      "loss": 0.6144,
+      "step": 51087
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.695695638656616,
+      "learning_rate": 7.239054840228577e-07,
+      "loss": 0.7846,
+      "step": 51088
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.5411338806152344,
+      "learning_rate": 7.237688446492775e-07,
+      "loss": 0.6665,
+      "step": 51089
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.3826751708984375,
+      "learning_rate": 7.236322171662929e-07,
+      "loss": 0.6539,
+      "step": 51090
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.2027101516723633,
+      "learning_rate": 7.234956015742867e-07,
+      "loss": 0.4882,
+      "step": 51091
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.4107325077056885,
+      "learning_rate": 7.233589978736361e-07,
+      "loss": 0.4737,
+      "step": 51092
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.1351215839385986,
+      "learning_rate": 7.23222406064723e-07,
+      "loss": 0.5646,
+      "step": 51093
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.645946741104126,
+      "learning_rate": 7.230858261479251e-07,
+      "loss": 0.4409,
+      "step": 51094
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.0238916873931885,
+      "learning_rate": 7.229492581236247e-07,
+      "loss": 0.4358,
+      "step": 51095
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 4.786635398864746,
+      "learning_rate": 7.228127019921987e-07,
+      "loss": 0.5139,
+      "step": 51096
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 5.144086837768555,
+      "learning_rate": 7.226761577540298e-07,
+      "loss": 0.8942,
+      "step": 51097
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.5774171352386475,
+      "learning_rate": 7.225396254094946e-07,
+      "loss": 0.6066,
+      "step": 51098
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.125117540359497,
+      "learning_rate": 7.224031049589758e-07,
+      "loss": 0.617,
+      "step": 51099
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.0784099102020264,
+      "learning_rate": 7.222665964028508e-07,
+      "loss": 0.6619,
+      "step": 51100
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.913257598876953,
+      "learning_rate": 7.221300997414993e-07,
+      "loss": 0.5457,
+      "step": 51101
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.692828893661499,
+      "learning_rate": 7.219936149753015e-07,
+      "loss": 0.5901,
+      "step": 51102
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.963604688644409,
+      "learning_rate": 7.218571421046383e-07,
+      "loss": 0.7762,
+      "step": 51103
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.016854763031006,
+      "learning_rate": 7.217206811298865e-07,
+      "loss": 0.7134,
+      "step": 51104
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.1604349613189697,
+      "learning_rate": 7.215842320514277e-07,
+      "loss": 0.4629,
+      "step": 51105
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.2400553226470947,
+      "learning_rate": 7.214477948696397e-07,
+      "loss": 0.5868,
+      "step": 51106
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.6380701065063477,
+      "learning_rate": 7.213113695849033e-07,
+      "loss": 0.8971,
+      "step": 51107
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.1342990398406982,
+      "learning_rate": 7.211749561975967e-07,
+      "loss": 0.6966,
+      "step": 51108
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.1756060123443604,
+      "learning_rate": 7.210385547081e-07,
+      "loss": 0.565,
+      "step": 51109
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.5466456413269043,
+      "learning_rate": 7.209021651167913e-07,
+      "loss": 0.6716,
+      "step": 51110
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.239258050918579,
+      "learning_rate": 7.20765787424052e-07,
+      "loss": 0.5985,
+      "step": 51111
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.8304829597473145,
+      "learning_rate": 7.206294216302589e-07,
+      "loss": 0.6253,
+      "step": 51112
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.5381407737731934,
+      "learning_rate": 7.204930677357929e-07,
+      "loss": 0.5401,
+      "step": 51113
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.877213954925537,
+      "learning_rate": 7.203567257410321e-07,
+      "loss": 0.5311,
+      "step": 51114
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.7491605281829834,
+      "learning_rate": 7.202203956463572e-07,
+      "loss": 0.4682,
+      "step": 51115
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.0657758712768555,
+      "learning_rate": 7.200840774521444e-07,
+      "loss": 0.3872,
+      "step": 51116
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.522916078567505,
+      "learning_rate": 7.199477711587761e-07,
+      "loss": 0.7058,
+      "step": 51117
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.532092332839966,
+      "learning_rate": 7.198114767666281e-07,
+      "loss": 0.6724,
+      "step": 51118
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.38702130317688,
+      "learning_rate": 7.196751942760826e-07,
+      "loss": 0.4097,
+      "step": 51119
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.1136910915374756,
+      "learning_rate": 7.195389236875156e-07,
+      "loss": 0.7248,
+      "step": 51120
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.6333651542663574,
+      "learning_rate": 7.194026650013086e-07,
+      "loss": 0.7245,
+      "step": 51121
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.8052992820739746,
+      "learning_rate": 7.192664182178366e-07,
+      "loss": 0.4676,
+      "step": 51122
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.2999324798583984,
+      "learning_rate": 7.191301833374836e-07,
+      "loss": 0.6622,
+      "step": 51123
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.3659110069274902,
+      "learning_rate": 7.189939603606249e-07,
+      "loss": 0.8465,
+      "step": 51124
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.3144266605377197,
+      "learning_rate": 7.188577492876414e-07,
+      "loss": 0.585,
+      "step": 51125
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.759763240814209,
+      "learning_rate": 7.187215501189093e-07,
+      "loss": 0.6573,
+      "step": 51126
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.0350570678710938,
+      "learning_rate": 7.1858536285481e-07,
+      "loss": 0.5767,
+      "step": 51127
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.732766628265381,
+      "learning_rate": 7.1844918749572e-07,
+      "loss": 0.5616,
+      "step": 51128
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 4.808826446533203,
+      "learning_rate": 7.183130240420199e-07,
+      "loss": 0.4811,
+      "step": 51129
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.677227020263672,
+      "learning_rate": 7.181768724940863e-07,
+      "loss": 0.6199,
+      "step": 51130
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.14227294921875,
+      "learning_rate": 7.180407328523003e-07,
+      "loss": 0.5371,
+      "step": 51131
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.4442121982574463,
+      "learning_rate": 7.179046051170374e-07,
+      "loss": 0.3681,
+      "step": 51132
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.645219564437866,
+      "learning_rate": 7.177684892886789e-07,
+      "loss": 0.6074,
+      "step": 51133
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.678849220275879,
+      "learning_rate": 7.176323853676009e-07,
+      "loss": 0.5785,
+      "step": 51134
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.568438768386841,
+      "learning_rate": 7.17496293354184e-07,
+      "loss": 0.3981,
+      "step": 51135
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.113173484802246,
+      "learning_rate": 7.17360213248805e-07,
+      "loss": 0.6375,
+      "step": 51136
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 4.172127723693848,
+      "learning_rate": 7.172241450518436e-07,
+      "loss": 0.8814,
+      "step": 51137
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.682746410369873,
+      "learning_rate": 7.170880887636766e-07,
+      "loss": 0.8126,
+      "step": 51138
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 5.979147434234619,
+      "learning_rate": 7.169520443846828e-07,
+      "loss": 0.4338,
+      "step": 51139
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.207947254180908,
+      "learning_rate": 7.168160119152423e-07,
+      "loss": 0.9246,
+      "step": 51140
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.398597002029419,
+      "learning_rate": 7.16679991355731e-07,
+      "loss": 0.5492,
+      "step": 51141
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.8715596199035645,
+      "learning_rate": 7.165439827065279e-07,
+      "loss": 0.9129,
+      "step": 51142
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.3615121841430664,
+      "learning_rate": 7.164079859680129e-07,
+      "loss": 0.5449,
+      "step": 51143
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.8961920738220215,
+      "learning_rate": 7.162720011405611e-07,
+      "loss": 0.7193,
+      "step": 51144
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.6237215995788574,
+      "learning_rate": 7.161360282245534e-07,
+      "loss": 0.5208,
+      "step": 51145
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.831153631210327,
+      "learning_rate": 7.160000672203649e-07,
+      "loss": 0.6798,
+      "step": 51146
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.907219648361206,
+      "learning_rate": 7.158641181283771e-07,
+      "loss": 0.6135,
+      "step": 51147
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.8844707012176514,
+      "learning_rate": 7.157281809489652e-07,
+      "loss": 0.3772,
+      "step": 51148
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.17846417427063,
+      "learning_rate": 7.155922556825095e-07,
+      "loss": 0.3805,
+      "step": 51149
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 6.532383441925049,
+      "learning_rate": 7.154563423293853e-07,
+      "loss": 0.3098,
+      "step": 51150
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.4805209636688232,
+      "learning_rate": 7.153204408899722e-07,
+      "loss": 0.7274,
+      "step": 51151
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.623598575592041,
+      "learning_rate": 7.151845513646488e-07,
+      "loss": 0.4027,
+      "step": 51152
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.945354700088501,
+      "learning_rate": 7.150486737537909e-07,
+      "loss": 0.7127,
+      "step": 51153
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.874966621398926,
+      "learning_rate": 7.149128080577789e-07,
+      "loss": 0.4587,
+      "step": 51154
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.0854196548461914,
+      "learning_rate": 7.147769542769878e-07,
+      "loss": 0.4344,
+      "step": 51155
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 1.784350872039795,
+      "learning_rate": 7.14641112411798e-07,
+      "loss": 0.3263,
+      "step": 51156
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.7381460666656494,
+      "learning_rate": 7.145052824625848e-07,
+      "loss": 0.5986,
+      "step": 51157
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.2728030681610107,
+      "learning_rate": 7.143694644297284e-07,
+      "loss": 0.3877,
+      "step": 51158
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.52001953125,
+      "learning_rate": 7.142336583136034e-07,
+      "loss": 0.6171,
+      "step": 51159
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.883147954940796,
+      "learning_rate": 7.140978641145907e-07,
+      "loss": 0.5706,
+      "step": 51160
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.640491247177124,
+      "learning_rate": 7.139620818330645e-07,
+      "loss": 0.5452,
+      "step": 51161
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.5044236183166504,
+      "learning_rate": 7.13826311469405e-07,
+      "loss": 0.4456,
+      "step": 51162
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.517333507537842,
+      "learning_rate": 7.136905530239885e-07,
+      "loss": 0.425,
+      "step": 51163
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.9822864532470703,
+      "learning_rate": 7.135548064971936e-07,
+      "loss": 0.5113,
+      "step": 51164
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.9440488815307617,
+      "learning_rate": 7.134190718893963e-07,
+      "loss": 0.866,
+      "step": 51165
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.796694755554199,
+      "learning_rate": 7.132833492009761e-07,
+      "loss": 0.6561,
+      "step": 51166
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.6137211322784424,
+      "learning_rate": 7.131476384323077e-07,
+      "loss": 0.888,
+      "step": 51167
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.985352039337158,
+      "learning_rate": 7.130119395837709e-07,
+      "loss": 0.4428,
+      "step": 51168
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.8410258293151855,
+      "learning_rate": 7.128762526557415e-07,
+      "loss": 0.6048,
+      "step": 51169
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.1656243801116943,
+      "learning_rate": 7.127405776485974e-07,
+      "loss": 0.3135,
+      "step": 51170
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 5.533071994781494,
+      "learning_rate": 7.126049145627151e-07,
+      "loss": 0.4209,
+      "step": 51171
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.036677598953247,
+      "learning_rate": 7.124692633984737e-07,
+      "loss": 0.3289,
+      "step": 51172
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.450273275375366,
+      "learning_rate": 7.123336241562479e-07,
+      "loss": 0.4371,
+      "step": 51173
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.730656147003174,
+      "learning_rate": 7.121979968364167e-07,
+      "loss": 0.4687,
+      "step": 51174
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.845802068710327,
+      "learning_rate": 7.120623814393562e-07,
+      "loss": 0.361,
+      "step": 51175
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.582043170928955,
+      "learning_rate": 7.119267779654443e-07,
+      "loss": 0.3876,
+      "step": 51176
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.2221949100494385,
+      "learning_rate": 7.117911864150573e-07,
+      "loss": 0.4093,
+      "step": 51177
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.667189359664917,
+      "learning_rate": 7.116556067885733e-07,
+      "loss": 0.628,
+      "step": 51178
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.3255043029785156,
+      "learning_rate": 7.115200390863674e-07,
+      "loss": 0.5176,
+      "step": 51179
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.472679376602173,
+      "learning_rate": 7.113844833088195e-07,
+      "loss": 0.4783,
+      "step": 51180
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.957374095916748,
+      "learning_rate": 7.112489394563032e-07,
+      "loss": 0.5363,
+      "step": 51181
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.70363187789917,
+      "learning_rate": 7.111134075291981e-07,
+      "loss": 0.6456,
+      "step": 51182
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.0753631591796875,
+      "learning_rate": 7.109778875278778e-07,
+      "loss": 0.6929,
+      "step": 51183
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.645479202270508,
+      "learning_rate": 7.108423794527236e-07,
+      "loss": 0.8432,
+      "step": 51184
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.0782604217529297,
+      "learning_rate": 7.107068833041087e-07,
+      "loss": 0.4065,
+      "step": 51185
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.3676841259002686,
+      "learning_rate": 7.105713990824125e-07,
+      "loss": 0.5685,
+      "step": 51186
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.3784000873565674,
+      "learning_rate": 7.10435926788009e-07,
+      "loss": 0.5656,
+      "step": 51187
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.6203079223632812,
+      "learning_rate": 7.103004664212776e-07,
+      "loss": 0.6249,
+      "step": 51188
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.6248064041137695,
+      "learning_rate": 7.101650179825926e-07,
+      "loss": 0.8029,
+      "step": 51189
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 4.362118721008301,
+      "learning_rate": 7.100295814723329e-07,
+      "loss": 0.7144,
+      "step": 51190
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.5053977966308594,
+      "learning_rate": 7.098941568908724e-07,
+      "loss": 0.6639,
+      "step": 51191
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.431114912033081,
+      "learning_rate": 7.0975874423859e-07,
+      "loss": 0.516,
+      "step": 51192
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.181331157684326,
+      "learning_rate": 7.096233435158612e-07,
+      "loss": 0.8355,
+      "step": 51193
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.1327505111694336,
+      "learning_rate": 7.094879547230632e-07,
+      "loss": 0.3563,
+      "step": 51194
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.631274461746216,
+      "learning_rate": 7.093525778605703e-07,
+      "loss": 0.5999,
+      "step": 51195
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.037099838256836,
+      "learning_rate": 7.092172129287627e-07,
+      "loss": 0.2895,
+      "step": 51196
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.7544045448303223,
+      "learning_rate": 7.090818599280131e-07,
+      "loss": 0.621,
+      "step": 51197
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.351064443588257,
+      "learning_rate": 7.089465188587003e-07,
+      "loss": 0.4655,
+      "step": 51198
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.7935051918029785,
+      "learning_rate": 7.088111897211991e-07,
+      "loss": 0.568,
+      "step": 51199
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.556278944015503,
+      "learning_rate": 7.08675872515886e-07,
+      "loss": 0.6746,
+      "step": 51200
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.396235942840576,
+      "learning_rate": 7.085405672431389e-07,
+      "loss": 0.6948,
+      "step": 51201
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.567034959793091,
+      "learning_rate": 7.08405273903332e-07,
+      "loss": 0.5577,
+      "step": 51202
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.6846747398376465,
+      "learning_rate": 7.082699924968423e-07,
+      "loss": 0.3373,
+      "step": 51203
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.4651520252227783,
+      "learning_rate": 7.081347230240471e-07,
+      "loss": 0.5087,
+      "step": 51204
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 1.9469667673110962,
+      "learning_rate": 7.079994654853207e-07,
+      "loss": 0.7049,
+      "step": 51205
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.2864301204681396,
+      "learning_rate": 7.078642198810409e-07,
+      "loss": 0.4681,
+      "step": 51206
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.4038007259368896,
+      "learning_rate": 7.077289862115816e-07,
+      "loss": 0.5075,
+      "step": 51207
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.048722267150879,
+      "learning_rate": 7.075937644773212e-07,
+      "loss": 0.528,
+      "step": 51208
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.443455934524536,
+      "learning_rate": 7.074585546786334e-07,
+      "loss": 0.4339,
+      "step": 51209
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.391350746154785,
+      "learning_rate": 7.073233568158966e-07,
+      "loss": 0.5139,
+      "step": 51210
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.6727583408355713,
+      "learning_rate": 7.07188170889484e-07,
+      "loss": 0.7635,
+      "step": 51211
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.1785051822662354,
+      "learning_rate": 7.070529968997731e-07,
+      "loss": 0.6445,
+      "step": 51212
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.5797502994537354,
+      "learning_rate": 7.069178348471411e-07,
+      "loss": 0.5681,
+      "step": 51213
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.3444976806640625,
+      "learning_rate": 7.067826847319609e-07,
+      "loss": 0.3496,
+      "step": 51214
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.3634207248687744,
+      "learning_rate": 7.066475465546113e-07,
+      "loss": 0.4444,
+      "step": 51215
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.19307017326355,
+      "learning_rate": 7.06512420315465e-07,
+      "loss": 0.4948,
+      "step": 51216
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 4.734584808349609,
+      "learning_rate": 7.063773060149004e-07,
+      "loss": 0.7305,
+      "step": 51217
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.4899628162384033,
+      "learning_rate": 7.062422036532912e-07,
+      "loss": 0.5384,
+      "step": 51218
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.206571578979492,
+      "learning_rate": 7.061071132310149e-07,
+      "loss": 0.4586,
+      "step": 51219
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.1678683757781982,
+      "learning_rate": 7.05972034748445e-07,
+      "loss": 0.8049,
+      "step": 51220
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 5.049444198608398,
+      "learning_rate": 7.058369682059596e-07,
+      "loss": 0.5635,
+      "step": 51221
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.9269514083862305,
+      "learning_rate": 7.057019136039323e-07,
+      "loss": 0.4466,
+      "step": 51222
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 4.249416828155518,
+      "learning_rate": 7.055668709427388e-07,
+      "loss": 0.6193,
+      "step": 51223
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.787423610687256,
+      "learning_rate": 7.05431840222755e-07,
+      "loss": 0.7202,
+      "step": 51224
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 4.978944301605225,
+      "learning_rate": 7.052968214443579e-07,
+      "loss": 0.9127,
+      "step": 51225
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.501086473464966,
+      "learning_rate": 7.051618146079198e-07,
+      "loss": 0.6951,
+      "step": 51226
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 4.159212112426758,
+      "learning_rate": 7.050268197138193e-07,
+      "loss": 0.8117,
+      "step": 51227
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.2005419731140137,
+      "learning_rate": 7.048918367624291e-07,
+      "loss": 0.7008,
+      "step": 51228
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.8086726665496826,
+      "learning_rate": 7.04756865754127e-07,
+      "loss": 0.4147,
+      "step": 51229
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.1266820430755615,
+      "learning_rate": 7.046219066892862e-07,
+      "loss": 0.3914,
+      "step": 51230
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.0726451873779297,
+      "learning_rate": 7.044869595682835e-07,
+      "loss": 0.67,
+      "step": 51231
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.8881585597991943,
+      "learning_rate": 7.04352024391492e-07,
+      "loss": 0.6114,
+      "step": 51232
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 4.907657146453857,
+      "learning_rate": 7.042171011592902e-07,
+      "loss": 0.7755,
+      "step": 51233
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.2948994636535645,
+      "learning_rate": 7.040821898720496e-07,
+      "loss": 0.5376,
+      "step": 51234
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.888313055038452,
+      "learning_rate": 7.039472905301487e-07,
+      "loss": 0.4732,
+      "step": 51235
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.2287425994873047,
+      "learning_rate": 7.038124031339599e-07,
+      "loss": 0.5809,
+      "step": 51236
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.6278915405273438,
+      "learning_rate": 7.036775276838603e-07,
+      "loss": 0.5544,
+      "step": 51237
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 4.473040580749512,
+      "learning_rate": 7.035426641802229e-07,
+      "loss": 0.4817,
+      "step": 51238
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 1.481747031211853,
+      "learning_rate": 7.03407812623425e-07,
+      "loss": 0.3198,
+      "step": 51239
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 4.560219764709473,
+      "learning_rate": 7.032729730138394e-07,
+      "loss": 0.4865,
+      "step": 51240
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.8494198322296143,
+      "learning_rate": 7.031381453518432e-07,
+      "loss": 0.6002,
+      "step": 51241
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.2504401206970215,
+      "learning_rate": 7.030033296378091e-07,
+      "loss": 0.5589,
+      "step": 51242
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.041980504989624,
+      "learning_rate": 7.028685258721146e-07,
+      "loss": 0.6444,
+      "step": 51243
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 4.250693321228027,
+      "learning_rate": 7.027337340551299e-07,
+      "loss": 0.4718,
+      "step": 51244
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.7374722957611084,
+      "learning_rate": 7.025989541872358e-07,
+      "loss": 0.4875,
+      "step": 51245
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.3708713054656982,
+      "learning_rate": 7.024641862688026e-07,
+      "loss": 0.6511,
+      "step": 51246
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.387833833694458,
+      "learning_rate": 7.023294303002081e-07,
+      "loss": 0.5763,
+      "step": 51247
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.002549886703491,
+      "learning_rate": 7.021946862818241e-07,
+      "loss": 0.4571,
+      "step": 51248
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.6947779655456543,
+      "learning_rate": 7.020599542140278e-07,
+      "loss": 0.5545,
+      "step": 51249
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.153688907623291,
+      "learning_rate": 7.019252340971916e-07,
+      "loss": 0.5866,
+      "step": 51250
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.600945472717285,
+      "learning_rate": 7.017905259316927e-07,
+      "loss": 0.6618,
+      "step": 51251
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.4862444400787354,
+      "learning_rate": 7.016558297179027e-07,
+      "loss": 0.3348,
+      "step": 51252
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.3445746898651123,
+      "learning_rate": 7.015211454561987e-07,
+      "loss": 0.6506,
+      "step": 51253
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.8880462646484375,
+      "learning_rate": 7.013864731469533e-07,
+      "loss": 0.7352,
+      "step": 51254
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.407153606414795,
+      "learning_rate": 7.012518127905432e-07,
+      "loss": 0.6021,
+      "step": 51255
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.226625680923462,
+      "learning_rate": 7.011171643873399e-07,
+      "loss": 0.5784,
+      "step": 51256
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.8947432041168213,
+      "learning_rate": 7.009825279377208e-07,
+      "loss": 0.7776,
+      "step": 51257
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.9724526405334473,
+      "learning_rate": 7.008479034420573e-07,
+      "loss": 0.3959,
+      "step": 51258
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.706742525100708,
+      "learning_rate": 7.007132909007264e-07,
+      "loss": 0.5856,
+      "step": 51259
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.577537775039673,
+      "learning_rate": 7.005786903141005e-07,
+      "loss": 0.6258,
+      "step": 51260
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 1.7898163795471191,
+      "learning_rate": 7.004441016825559e-07,
+      "loss": 0.2992,
+      "step": 51261
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.5996334552764893,
+      "learning_rate": 7.003095250064639e-07,
+      "loss": 0.5182,
+      "step": 51262
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 4.086456298828125,
+      "learning_rate": 7.001749602862007e-07,
+      "loss": 0.6899,
+      "step": 51263
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.8815622329711914,
+      "learning_rate": 7.000404075221407e-07,
+      "loss": 0.6619,
+      "step": 51264
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 4.7728190422058105,
+      "learning_rate": 6.99905866714658e-07,
+      "loss": 0.7855,
+      "step": 51265
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 4.982519626617432,
+      "learning_rate": 6.997713378641252e-07,
+      "loss": 0.7061,
+      "step": 51266
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.238185167312622,
+      "learning_rate": 6.996368209709187e-07,
+      "loss": 0.6769,
+      "step": 51267
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.2036819458007812,
+      "learning_rate": 6.995023160354103e-07,
+      "loss": 0.6782,
+      "step": 51268
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 5.521437168121338,
+      "learning_rate": 6.993678230579754e-07,
+      "loss": 0.7886,
+      "step": 51269
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.6688668727874756,
+      "learning_rate": 6.992333420389869e-07,
+      "loss": 0.547,
+      "step": 51270
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.457138776779175,
+      "learning_rate": 6.990988729788206e-07,
+      "loss": 0.64,
+      "step": 51271
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.0953898429870605,
+      "learning_rate": 6.98964415877848e-07,
+      "loss": 0.3747,
+      "step": 51272
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.424014091491699,
+      "learning_rate": 6.988299707364438e-07,
+      "loss": 0.5906,
+      "step": 51273
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.8694567680358887,
+      "learning_rate": 6.986955375549836e-07,
+      "loss": 0.4067,
+      "step": 51274
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.8907973766326904,
+      "learning_rate": 6.985611163338385e-07,
+      "loss": 0.7653,
+      "step": 51275
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.3531408309936523,
+      "learning_rate": 6.984267070733852e-07,
+      "loss": 0.4309,
+      "step": 51276
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.9792816638946533,
+      "learning_rate": 6.982923097739941e-07,
+      "loss": 0.8164,
+      "step": 51277
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 4.126943111419678,
+      "learning_rate": 6.981579244360421e-07,
+      "loss": 0.6475,
+      "step": 51278
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 4.030143737792969,
+      "learning_rate": 6.980235510599004e-07,
+      "loss": 0.5122,
+      "step": 51279
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.665606737136841,
+      "learning_rate": 6.978891896459444e-07,
+      "loss": 0.612,
+      "step": 51280
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.7338881492614746,
+      "learning_rate": 6.977548401945461e-07,
+      "loss": 0.6879,
+      "step": 51281
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.971592664718628,
+      "learning_rate": 6.976205027060806e-07,
+      "loss": 0.3792,
+      "step": 51282
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.314910411834717,
+      "learning_rate": 6.974861771809199e-07,
+      "loss": 0.6218,
+      "step": 51283
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.289255142211914,
+      "learning_rate": 6.973518636194388e-07,
+      "loss": 0.4654,
+      "step": 51284
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.4847335815429688,
+      "learning_rate": 6.972175620220101e-07,
+      "loss": 0.5057,
+      "step": 51285
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.889212131500244,
+      "learning_rate": 6.970832723890086e-07,
+      "loss": 0.6414,
+      "step": 51286
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.742598533630371,
+      "learning_rate": 6.969489947208052e-07,
+      "loss": 0.5921,
+      "step": 51287
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.194166898727417,
+      "learning_rate": 6.968147290177763e-07,
+      "loss": 0.7359,
+      "step": 51288
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 4.001828193664551,
+      "learning_rate": 6.966804752802919e-07,
+      "loss": 0.5544,
+      "step": 51289
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.3978350162506104,
+      "learning_rate": 6.965462335087286e-07,
+      "loss": 0.6046,
+      "step": 51290
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.261157751083374,
+      "learning_rate": 6.964120037034572e-07,
+      "loss": 0.4746,
+      "step": 51291
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.637327194213867,
+      "learning_rate": 6.962777858648523e-07,
+      "loss": 0.4555,
+      "step": 51292
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.863319158554077,
+      "learning_rate": 6.961435799932859e-07,
+      "loss": 0.4922,
+      "step": 51293
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.3885114192962646,
+      "learning_rate": 6.960093860891331e-07,
+      "loss": 0.5613,
+      "step": 51294
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.5183517932891846,
+      "learning_rate": 6.95875204152765e-07,
+      "loss": 0.7277,
+      "step": 51295
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.5153274536132812,
+      "learning_rate": 6.957410341845567e-07,
+      "loss": 0.4282,
+      "step": 51296
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.0930542945861816,
+      "learning_rate": 6.956068761848783e-07,
+      "loss": 0.4769,
+      "step": 51297
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.7206974029541016,
+      "learning_rate": 6.954727301541064e-07,
+      "loss": 0.7741,
+      "step": 51298
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.7179558277130127,
+      "learning_rate": 6.953385960926113e-07,
+      "loss": 0.6491,
+      "step": 51299
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.8854613304138184,
+      "learning_rate": 6.95204474000768e-07,
+      "loss": 0.4377,
+      "step": 51300
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.5690014362335205,
+      "learning_rate": 6.950703638789474e-07,
+      "loss": 0.678,
+      "step": 51301
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 4.924596786499023,
+      "learning_rate": 6.949362657275244e-07,
+      "loss": 0.5788,
+      "step": 51302
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.529825210571289,
+      "learning_rate": 6.948021795468702e-07,
+      "loss": 0.6176,
+      "step": 51303
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.1389577388763428,
+      "learning_rate": 6.946681053373588e-07,
+      "loss": 0.4023,
+      "step": 51304
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.8449628353118896,
+      "learning_rate": 6.945340430993608e-07,
+      "loss": 0.6361,
+      "step": 51305
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.062915325164795,
+      "learning_rate": 6.943999928332528e-07,
+      "loss": 0.3696,
+      "step": 51306
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.2572977542877197,
+      "learning_rate": 6.942659545394048e-07,
+      "loss": 0.6494,
+      "step": 51307
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 4.101670265197754,
+      "learning_rate": 6.941319282181908e-07,
+      "loss": 0.8471,
+      "step": 51308
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.0717880725860596,
+      "learning_rate": 6.93997913869982e-07,
+      "loss": 0.4386,
+      "step": 51309
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.574704647064209,
+      "learning_rate": 6.938639114951534e-07,
+      "loss": 0.7791,
+      "step": 51310
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.962399482727051,
+      "learning_rate": 6.937299210940746e-07,
+      "loss": 0.4661,
+      "step": 51311
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 1.8778588771820068,
+      "learning_rate": 6.935959426671213e-07,
+      "loss": 0.4357,
+      "step": 51312
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.479114294052124,
+      "learning_rate": 6.934619762146628e-07,
+      "loss": 0.4153,
+      "step": 51313
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.684342861175537,
+      "learning_rate": 6.933280217370747e-07,
+      "loss": 0.4821,
+      "step": 51314
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.2796411514282227,
+      "learning_rate": 6.931940792347269e-07,
+      "loss": 0.6084,
+      "step": 51315
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.5968799591064453,
+      "learning_rate": 6.930601487079941e-07,
+      "loss": 0.6596,
+      "step": 51316
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.041701078414917,
+      "learning_rate": 6.929262301572459e-07,
+      "loss": 0.5483,
+      "step": 51317
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.8359768390655518,
+      "learning_rate": 6.927923235828581e-07,
+      "loss": 0.7867,
+      "step": 51318
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.7696967124938965,
+      "learning_rate": 6.926584289852007e-07,
+      "loss": 0.6689,
+      "step": 51319
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.8658015727996826,
+      "learning_rate": 6.925245463646469e-07,
+      "loss": 0.5866,
+      "step": 51320
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.1403048038482666,
+      "learning_rate": 6.923906757215682e-07,
+      "loss": 0.6268,
+      "step": 51321
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.517669200897217,
+      "learning_rate": 6.922568170563381e-07,
+      "loss": 0.6147,
+      "step": 51322
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 4.556110858917236,
+      "learning_rate": 6.921229703693272e-07,
+      "loss": 0.6601,
+      "step": 51323
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.4031178951263428,
+      "learning_rate": 6.919891356609082e-07,
+      "loss": 0.5199,
+      "step": 51324
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.2186527252197266,
+      "learning_rate": 6.918553129314536e-07,
+      "loss": 0.5828,
+      "step": 51325
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 4.068612098693848,
+      "learning_rate": 6.917215021813373e-07,
+      "loss": 0.6393,
+      "step": 51326
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.894333600997925,
+      "learning_rate": 6.915877034109275e-07,
+      "loss": 0.417,
+      "step": 51327
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.5963046550750732,
+      "learning_rate": 6.914539166206003e-07,
+      "loss": 0.5281,
+      "step": 51328
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.394585371017456,
+      "learning_rate": 6.913201418107241e-07,
+      "loss": 0.6059,
+      "step": 51329
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.337421417236328,
+      "learning_rate": 6.911863789816736e-07,
+      "loss": 0.5738,
+      "step": 51330
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.457253932952881,
+      "learning_rate": 6.910526281338193e-07,
+      "loss": 0.5963,
+      "step": 51331
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.2564563751220703,
+      "learning_rate": 6.909188892675339e-07,
+      "loss": 0.5225,
+      "step": 51332
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.846714496612549,
+      "learning_rate": 6.907851623831879e-07,
+      "loss": 0.7042,
+      "step": 51333
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.3097035884857178,
+      "learning_rate": 6.906514474811549e-07,
+      "loss": 0.5951,
+      "step": 51334
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 1.93827223777771,
+      "learning_rate": 6.90517744561805e-07,
+      "loss": 0.5443,
+      "step": 51335
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.29587721824646,
+      "learning_rate": 6.903840536255113e-07,
+      "loss": 0.5055,
+      "step": 51336
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 4.252836227416992,
+      "learning_rate": 6.902503746726458e-07,
+      "loss": 0.6021,
+      "step": 51337
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 4.1405110359191895,
+      "learning_rate": 6.901167077035786e-07,
+      "loss": 0.7197,
+      "step": 51338
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.4904558658599854,
+      "learning_rate": 6.899830527186835e-07,
+      "loss": 0.599,
+      "step": 51339
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.5087709426879883,
+      "learning_rate": 6.898494097183295e-07,
+      "loss": 0.6609,
+      "step": 51340
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.640742301940918,
+      "learning_rate": 6.897157787028907e-07,
+      "loss": 0.7003,
+      "step": 51341
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.5182881355285645,
+      "learning_rate": 6.89582159672737e-07,
+      "loss": 0.3846,
+      "step": 51342
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.3699371814727783,
+      "learning_rate": 6.894485526282419e-07,
+      "loss": 0.416,
+      "step": 51343
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.2546498775482178,
+      "learning_rate": 6.893149575697738e-07,
+      "loss": 0.4177,
+      "step": 51344
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.563568353652954,
+      "learning_rate": 6.891813744977066e-07,
+      "loss": 0.3726,
+      "step": 51345
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.290783643722534,
+      "learning_rate": 6.890478034124104e-07,
+      "loss": 0.4894,
+      "step": 51346
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.3196089267730713,
+      "learning_rate": 6.889142443142588e-07,
+      "loss": 0.4523,
+      "step": 51347
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.4447312355041504,
+      "learning_rate": 6.887806972036209e-07,
+      "loss": 0.6507,
+      "step": 51348
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.0903334617614746,
+      "learning_rate": 6.886471620808699e-07,
+      "loss": 0.799,
+      "step": 51349
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.4054760932922363,
+      "learning_rate": 6.885136389463747e-07,
+      "loss": 0.6523,
+      "step": 51350
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.4980409145355225,
+      "learning_rate": 6.883801278005092e-07,
+      "loss": 0.6672,
+      "step": 51351
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.9769608974456787,
+      "learning_rate": 6.882466286436423e-07,
+      "loss": 0.5601,
+      "step": 51352
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.085526943206787,
+      "learning_rate": 6.881131414761471e-07,
+      "loss": 0.6268,
+      "step": 51353
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.060060501098633,
+      "learning_rate": 6.879796662983935e-07,
+      "loss": 0.7981,
+      "step": 51354
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.861780881881714,
+      "learning_rate": 6.878462031107541e-07,
+      "loss": 0.5354,
+      "step": 51355
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.86660099029541,
+      "learning_rate": 6.877127519135973e-07,
+      "loss": 0.6535,
+      "step": 51356
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.9813356399536133,
+      "learning_rate": 6.875793127072977e-07,
+      "loss": 0.7542,
+      "step": 51357
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.214088201522827,
+      "learning_rate": 6.874458854922233e-07,
+      "loss": 0.3982,
+      "step": 51358
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.3319692611694336,
+      "learning_rate": 6.873124702687472e-07,
+      "loss": 0.6499,
+      "step": 51359
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.349722146987915,
+      "learning_rate": 6.871790670372386e-07,
+      "loss": 0.5604,
+      "step": 51360
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.5797133445739746,
+      "learning_rate": 6.870456757980709e-07,
+      "loss": 0.4516,
+      "step": 51361
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 1.9631003141403198,
+      "learning_rate": 6.869122965516117e-07,
+      "loss": 0.3315,
+      "step": 51362
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.6298365592956543,
+      "learning_rate": 6.867789292982352e-07,
+      "loss": 0.4408,
+      "step": 51363
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.6110804080963135,
+      "learning_rate": 6.866455740383094e-07,
+      "loss": 0.639,
+      "step": 51364
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 1.9518274068832397,
+      "learning_rate": 6.865122307722078e-07,
+      "loss": 0.3425,
+      "step": 51365
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 4.228560924530029,
+      "learning_rate": 6.863788995002974e-07,
+      "loss": 0.6692,
+      "step": 51366
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.9488253593444824,
+      "learning_rate": 6.86245580222954e-07,
+      "loss": 0.3655,
+      "step": 51367
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.4024956226348877,
+      "learning_rate": 6.861122729405439e-07,
+      "loss": 0.4963,
+      "step": 51368
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 4.293485164642334,
+      "learning_rate": 6.859789776534415e-07,
+      "loss": 0.7503,
+      "step": 51369
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.507563591003418,
+      "learning_rate": 6.858456943620134e-07,
+      "loss": 0.3989,
+      "step": 51370
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.214872360229492,
+      "learning_rate": 6.85712423066634e-07,
+      "loss": 0.6095,
+      "step": 51371
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 6.115860462188721,
+      "learning_rate": 6.855791637676712e-07,
+      "loss": 0.69,
+      "step": 51372
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.3947901725769043,
+      "learning_rate": 6.854459164654975e-07,
+      "loss": 0.3603,
+      "step": 51373
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.8878259658813477,
+      "learning_rate": 6.85312681160481e-07,
+      "loss": 0.6109,
+      "step": 51374
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.1578524112701416,
+      "learning_rate": 6.851794578529953e-07,
+      "loss": 0.6463,
+      "step": 51375
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.420245409011841,
+      "learning_rate": 6.850462465434072e-07,
+      "loss": 0.5222,
+      "step": 51376
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.1706295013427734,
+      "learning_rate": 6.849130472320908e-07,
+      "loss": 0.4697,
+      "step": 51377
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.332047462463379,
+      "learning_rate": 6.847798599194133e-07,
+      "loss": 0.5967,
+      "step": 51378
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.437229633331299,
+      "learning_rate": 6.846466846057481e-07,
+      "loss": 0.4489,
+      "step": 51379
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.395595073699951,
+      "learning_rate": 6.845135212914628e-07,
+      "loss": 0.3727,
+      "step": 51380
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.528851270675659,
+      "learning_rate": 6.843803699769292e-07,
+      "loss": 0.5937,
+      "step": 51381
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.073260545730591,
+      "learning_rate": 6.842472306625164e-07,
+      "loss": 0.5005,
+      "step": 51382
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 4.818346977233887,
+      "learning_rate": 6.841141033485965e-07,
+      "loss": 0.5905,
+      "step": 51383
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.4023897647857666,
+      "learning_rate": 6.839809880355375e-07,
+      "loss": 0.7186,
+      "step": 51384
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.65037202835083,
+      "learning_rate": 6.838478847237101e-07,
+      "loss": 0.7813,
+      "step": 51385
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.637641668319702,
+      "learning_rate": 6.837147934134853e-07,
+      "loss": 0.2797,
+      "step": 51386
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.834561347961426,
+      "learning_rate": 6.835817141052336e-07,
+      "loss": 0.639,
+      "step": 51387
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.4333853721618652,
+      "learning_rate": 6.834486467993235e-07,
+      "loss": 0.7425,
+      "step": 51388
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.256772756576538,
+      "learning_rate": 6.833155914961265e-07,
+      "loss": 0.5156,
+      "step": 51389
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.91796612739563,
+      "learning_rate": 6.831825481960108e-07,
+      "loss": 0.54,
+      "step": 51390
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.3163652420043945,
+      "learning_rate": 6.830495168993484e-07,
+      "loss": 0.5781,
+      "step": 51391
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.7216947078704834,
+      "learning_rate": 6.829164976065067e-07,
+      "loss": 0.5285,
+      "step": 51392
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.060767889022827,
+      "learning_rate": 6.827834903178582e-07,
+      "loss": 0.7791,
+      "step": 51393
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 5.264292240142822,
+      "learning_rate": 6.826504950337709e-07,
+      "loss": 0.611,
+      "step": 51394
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.081571102142334,
+      "learning_rate": 6.825175117546156e-07,
+      "loss": 0.5807,
+      "step": 51395
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 1.9017390012741089,
+      "learning_rate": 6.82384540480761e-07,
+      "loss": 0.3515,
+      "step": 51396
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.598153591156006,
+      "learning_rate": 6.82251581212578e-07,
+      "loss": 0.5744,
+      "step": 51397
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.285583019256592,
+      "learning_rate": 6.821186339504365e-07,
+      "loss": 0.595,
+      "step": 51398
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.4808125495910645,
+      "learning_rate": 6.819856986947049e-07,
+      "loss": 0.7775,
+      "step": 51399
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.696559429168701,
+      "learning_rate": 6.818527754457544e-07,
+      "loss": 0.6053,
+      "step": 51400
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.312979221343994,
+      "learning_rate": 6.817198642039524e-07,
+      "loss": 0.795,
+      "step": 51401
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 5.4311604499816895,
+      "learning_rate": 6.81586964969671e-07,
+      "loss": 0.7447,
+      "step": 51402
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.538518190383911,
+      "learning_rate": 6.81454077743277e-07,
+      "loss": 0.3021,
+      "step": 51403
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.8304872512817383,
+      "learning_rate": 6.813212025251432e-07,
+      "loss": 0.4718,
+      "step": 51404
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.3886637687683105,
+      "learning_rate": 6.81188339315636e-07,
+      "loss": 0.5799,
+      "step": 51405
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 4.1337361335754395,
+      "learning_rate": 6.81055488115126e-07,
+      "loss": 0.6825,
+      "step": 51406
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.413283348083496,
+      "learning_rate": 6.809226489239845e-07,
+      "loss": 0.5523,
+      "step": 51407
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 4.181700706481934,
+      "learning_rate": 6.807898217425773e-07,
+      "loss": 0.7564,
+      "step": 51408
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.42700457572937,
+      "learning_rate": 6.806570065712759e-07,
+      "loss": 0.5516,
+      "step": 51409
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 1.9871724843978882,
+      "learning_rate": 6.805242034104504e-07,
+      "loss": 0.3533,
+      "step": 51410
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 1.9671132564544678,
+      "learning_rate": 6.803914122604682e-07,
+      "loss": 0.5164,
+      "step": 51411
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.055614948272705,
+      "learning_rate": 6.802586331216998e-07,
+      "loss": 0.8317,
+      "step": 51412
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.292628049850464,
+      "learning_rate": 6.801258659945131e-07,
+      "loss": 0.6282,
+      "step": 51413
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.0094525814056396,
+      "learning_rate": 6.799931108792795e-07,
+      "loss": 0.4999,
+      "step": 51414
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.4261856079101562,
+      "learning_rate": 6.798603677763655e-07,
+      "loss": 0.6306,
+      "step": 51415
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 4.1048479080200195,
+      "learning_rate": 6.797276366861421e-07,
+      "loss": 0.6314,
+      "step": 51416
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.9700474739074707,
+      "learning_rate": 6.795949176089772e-07,
+      "loss": 0.5489,
+      "step": 51417
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.302125930786133,
+      "learning_rate": 6.794622105452409e-07,
+      "loss": 0.6702,
+      "step": 51418
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.450625419616699,
+      "learning_rate": 6.793295154953012e-07,
+      "loss": 0.7226,
+      "step": 51419
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.8555750846862793,
+      "learning_rate": 6.791968324595282e-07,
+      "loss": 0.5428,
+      "step": 51420
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.4047470092773438,
+      "learning_rate": 6.790641614382893e-07,
+      "loss": 0.6496,
+      "step": 51421
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.907682418823242,
+      "learning_rate": 6.789315024319554e-07,
+      "loss": 0.6927,
+      "step": 51422
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.2717154026031494,
+      "learning_rate": 6.787988554408931e-07,
+      "loss": 0.6435,
+      "step": 51423
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.3958253860473633,
+      "learning_rate": 6.786662204654737e-07,
+      "loss": 0.473,
+      "step": 51424
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.548855781555176,
+      "learning_rate": 6.785335975060636e-07,
+      "loss": 0.2938,
+      "step": 51425
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.3571393489837646,
+      "learning_rate": 6.784009865630336e-07,
+      "loss": 0.3117,
+      "step": 51426
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.356671094894409,
+      "learning_rate": 6.782683876367496e-07,
+      "loss": 0.5207,
+      "step": 51427
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.456871747970581,
+      "learning_rate": 6.78135800727584e-07,
+      "loss": 0.8355,
+      "step": 51428
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.1238996982574463,
+      "learning_rate": 6.780032258359032e-07,
+      "loss": 0.7151,
+      "step": 51429
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.0658295154571533,
+      "learning_rate": 6.778706629620774e-07,
+      "loss": 0.5019,
+      "step": 51430
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.449596643447876,
+      "learning_rate": 6.77738112106473e-07,
+      "loss": 0.7017,
+      "step": 51431
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.1227922439575195,
+      "learning_rate": 6.776055732694603e-07,
+      "loss": 0.647,
+      "step": 51432
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.271291732788086,
+      "learning_rate": 6.774730464514068e-07,
+      "loss": 0.757,
+      "step": 51433
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.4245340824127197,
+      "learning_rate": 6.773405316526826e-07,
+      "loss": 0.7198,
+      "step": 51434
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.1810719966888428,
+      "learning_rate": 6.772080288736538e-07,
+      "loss": 0.2194,
+      "step": 51435
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.9020090103149414,
+      "learning_rate": 6.770755381146915e-07,
+      "loss": 0.5567,
+      "step": 51436
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.046370029449463,
+      "learning_rate": 6.769430593761611e-07,
+      "loss": 0.6961,
+      "step": 51437
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.128619432449341,
+      "learning_rate": 6.76810592658434e-07,
+      "loss": 0.4828,
+      "step": 51438
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.2802202701568604,
+      "learning_rate": 6.766781379618759e-07,
+      "loss": 0.5172,
+      "step": 51439
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.2398533821105957,
+      "learning_rate": 6.765456952868576e-07,
+      "loss": 0.8345,
+      "step": 51440
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.3240928649902344,
+      "learning_rate": 6.764132646337451e-07,
+      "loss": 0.5223,
+      "step": 51441
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.77770733833313,
+      "learning_rate": 6.762808460029086e-07,
+      "loss": 0.6989,
+      "step": 51442
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.1011955738067627,
+      "learning_rate": 6.761484393947143e-07,
+      "loss": 0.3592,
+      "step": 51443
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.113154411315918,
+      "learning_rate": 6.760160448095326e-07,
+      "loss": 0.5364,
+      "step": 51444
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.4759442806243896,
+      "learning_rate": 6.758836622477294e-07,
+      "loss": 0.5005,
+      "step": 51445
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.3351998329162598,
+      "learning_rate": 6.757512917096748e-07,
+      "loss": 0.4368,
+      "step": 51446
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.461195945739746,
+      "learning_rate": 6.756189331957342e-07,
+      "loss": 0.5652,
+      "step": 51447
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.512589931488037,
+      "learning_rate": 6.754865867062788e-07,
+      "loss": 0.5927,
+      "step": 51448
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.769949436187744,
+      "learning_rate": 6.753542522416745e-07,
+      "loss": 0.7766,
+      "step": 51449
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.530536651611328,
+      "learning_rate": 6.752219298022917e-07,
+      "loss": 0.4779,
+      "step": 51450
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.1059677600860596,
+      "learning_rate": 6.750896193884948e-07,
+      "loss": 0.5678,
+      "step": 51451
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.470952272415161,
+      "learning_rate": 6.749573210006544e-07,
+      "loss": 0.479,
+      "step": 51452
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.4021108150482178,
+      "learning_rate": 6.748250346391371e-07,
+      "loss": 0.5443,
+      "step": 51453
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.1700379848480225,
+      "learning_rate": 6.746927603043119e-07,
+      "loss": 0.5365,
+      "step": 51454
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.2740814685821533,
+      "learning_rate": 6.745604979965448e-07,
+      "loss": 0.5283,
+      "step": 51455
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.606837034225464,
+      "learning_rate": 6.744282477162057e-07,
+      "loss": 0.6654,
+      "step": 51456
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.9929769039154053,
+      "learning_rate": 6.742960094636597e-07,
+      "loss": 0.6153,
+      "step": 51457
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.906459331512451,
+      "learning_rate": 6.741637832392767e-07,
+      "loss": 0.7383,
+      "step": 51458
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.319990873336792,
+      "learning_rate": 6.740315690434246e-07,
+      "loss": 0.5823,
+      "step": 51459
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.1511940956115723,
+      "learning_rate": 6.738993668764688e-07,
+      "loss": 0.5635,
+      "step": 51460
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.6625802516937256,
+      "learning_rate": 6.737671767387799e-07,
+      "loss": 0.2939,
+      "step": 51461
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.6527657508850098,
+      "learning_rate": 6.736349986307222e-07,
+      "loss": 0.6941,
+      "step": 51462
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 1.97683584690094,
+      "learning_rate": 6.73502832552666e-07,
+      "loss": 0.4224,
+      "step": 51463
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.9870307445526123,
+      "learning_rate": 6.73370678504977e-07,
+      "loss": 0.5111,
+      "step": 51464
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.2991976737976074,
+      "learning_rate": 6.73238536488024e-07,
+      "loss": 0.2448,
+      "step": 51465
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 4.184596061706543,
+      "learning_rate": 6.73106406502173e-07,
+      "loss": 0.448,
+      "step": 51466
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.703059434890747,
+      "learning_rate": 6.729742885477925e-07,
+      "loss": 0.6909,
+      "step": 51467
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.377606153488159,
+      "learning_rate": 6.728421826252501e-07,
+      "loss": 0.6954,
+      "step": 51468
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.953368902206421,
+      "learning_rate": 6.727100887349114e-07,
+      "loss": 0.8357,
+      "step": 51469
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.907038450241089,
+      "learning_rate": 6.725780068771454e-07,
+      "loss": 0.4947,
+      "step": 51470
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.5615081787109375,
+      "learning_rate": 6.724459370523195e-07,
+      "loss": 0.5087,
+      "step": 51471
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.403421640396118,
+      "learning_rate": 6.723138792607997e-07,
+      "loss": 0.581,
+      "step": 51472
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.191894769668579,
+      "learning_rate": 6.721818335029546e-07,
+      "loss": 0.6244,
+      "step": 51473
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.987083673477173,
+      "learning_rate": 6.720497997791497e-07,
+      "loss": 0.5958,
+      "step": 51474
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.9051637649536133,
+      "learning_rate": 6.71917778089754e-07,
+      "loss": 0.5028,
+      "step": 51475
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.22397518157959,
+      "learning_rate": 6.717857684351326e-07,
+      "loss": 0.7386,
+      "step": 51476
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.35884952545166,
+      "learning_rate": 6.716537708156545e-07,
+      "loss": 0.4688,
+      "step": 51477
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.5241148471832275,
+      "learning_rate": 6.715217852316847e-07,
+      "loss": 0.6849,
+      "step": 51478
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.5363998413085938,
+      "learning_rate": 6.713898116835927e-07,
+      "loss": 0.7046,
+      "step": 51479
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.7200889587402344,
+      "learning_rate": 6.712578501717426e-07,
+      "loss": 0.7028,
+      "step": 51480
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.8425309658050537,
+      "learning_rate": 6.711259006965043e-07,
+      "loss": 0.5414,
+      "step": 51481
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 4.22716760635376,
+      "learning_rate": 6.709939632582418e-07,
+      "loss": 0.6135,
+      "step": 51482
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.6773769855499268,
+      "learning_rate": 6.70862037857325e-07,
+      "loss": 0.6699,
+      "step": 51483
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.7215847969055176,
+      "learning_rate": 6.707301244941178e-07,
+      "loss": 0.5721,
+      "step": 51484
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.706934928894043,
+      "learning_rate": 6.705982231689895e-07,
+      "loss": 0.6595,
+      "step": 51485
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 6.092073440551758,
+      "learning_rate": 6.704663338823042e-07,
+      "loss": 0.5643,
+      "step": 51486
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.591062068939209,
+      "learning_rate": 6.703344566344316e-07,
+      "loss": 0.527,
+      "step": 51487
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.585045337677002,
+      "learning_rate": 6.702025914257349e-07,
+      "loss": 0.7043,
+      "step": 51488
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.87387752532959,
+      "learning_rate": 6.700707382565846e-07,
+      "loss": 0.862,
+      "step": 51489
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.6427669525146484,
+      "learning_rate": 6.699388971273446e-07,
+      "loss": 0.4643,
+      "step": 51490
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.8791847229003906,
+      "learning_rate": 6.698070680383839e-07,
+      "loss": 0.5392,
+      "step": 51491
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 5.153913497924805,
+      "learning_rate": 6.696752509900661e-07,
+      "loss": 0.5522,
+      "step": 51492
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.9619853496551514,
+      "learning_rate": 6.695434459827604e-07,
+      "loss": 0.773,
+      "step": 51493
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.0890588760375977,
+      "learning_rate": 6.694116530168315e-07,
+      "loss": 0.6261,
+      "step": 51494
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.570484161376953,
+      "learning_rate": 6.692798720926469e-07,
+      "loss": 0.7254,
+      "step": 51495
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.3918135166168213,
+      "learning_rate": 6.691481032105724e-07,
+      "loss": 0.6109,
+      "step": 51496
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.802633762359619,
+      "learning_rate": 6.690163463709748e-07,
+      "loss": 0.3736,
+      "step": 51497
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.3481147289276123,
+      "learning_rate": 6.6888460157422e-07,
+      "loss": 0.601,
+      "step": 51498
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.822859048843384,
+      "learning_rate": 6.687528688206752e-07,
+      "loss": 0.586,
+      "step": 51499
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.582137107849121,
+      "learning_rate": 6.686211481107052e-07,
+      "loss": 0.4678,
+      "step": 51500
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.366945505142212,
+      "learning_rate": 6.684894394446784e-07,
+      "loss": 0.4525,
+      "step": 51501
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.9736673831939697,
+      "learning_rate": 6.683577428229587e-07,
+      "loss": 0.7287,
+      "step": 51502
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.496838092803955,
+      "learning_rate": 6.682260582459149e-07,
+      "loss": 0.5225,
+      "step": 51503
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.090601921081543,
+      "learning_rate": 6.680943857139099e-07,
+      "loss": 0.3697,
+      "step": 51504
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.83622407913208,
+      "learning_rate": 6.679627252273136e-07,
+      "loss": 0.6448,
+      "step": 51505
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.1531014442443848,
+      "learning_rate": 6.678310767864881e-07,
+      "loss": 0.717,
+      "step": 51506
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.922089099884033,
+      "learning_rate": 6.676994403918035e-07,
+      "loss": 0.5656,
+      "step": 51507
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.8792624473571777,
+      "learning_rate": 6.67567816043621e-07,
+      "loss": 0.5657,
+      "step": 51508
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.6816647052764893,
+      "learning_rate": 6.674362037423121e-07,
+      "loss": 0.5298,
+      "step": 51509
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 4.486774921417236,
+      "learning_rate": 6.673046034882386e-07,
+      "loss": 0.5252,
+      "step": 51510
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.1213488578796387,
+      "learning_rate": 6.67173015281769e-07,
+      "loss": 0.6674,
+      "step": 51511
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.6662795543670654,
+      "learning_rate": 6.670414391232676e-07,
+      "loss": 0.6628,
+      "step": 51512
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.81286358833313,
+      "learning_rate": 6.669098750131014e-07,
+      "loss": 0.488,
+      "step": 51513
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 4.198183536529541,
+      "learning_rate": 6.667783229516345e-07,
+      "loss": 0.6661,
+      "step": 51514
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.508889675140381,
+      "learning_rate": 6.666467829392353e-07,
+      "loss": 0.559,
+      "step": 51515
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.2416343688964844,
+      "learning_rate": 6.665152549762665e-07,
+      "loss": 0.7115,
+      "step": 51516
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.934859037399292,
+      "learning_rate": 6.663837390630973e-07,
+      "loss": 0.5979,
+      "step": 51517
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.7685744762420654,
+      "learning_rate": 6.662522352000894e-07,
+      "loss": 0.3592,
+      "step": 51518
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.5616469383239746,
+      "learning_rate": 6.661207433876127e-07,
+      "loss": 0.3636,
+      "step": 51519
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.706106424331665,
+      "learning_rate": 6.65989263626029e-07,
+      "loss": 0.4991,
+      "step": 51520
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 4.86422872543335,
+      "learning_rate": 6.658577959157054e-07,
+      "loss": 0.7661,
+      "step": 51521
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.57167911529541,
+      "learning_rate": 6.657263402570091e-07,
+      "loss": 0.6165,
+      "step": 51522
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.8353562355041504,
+      "learning_rate": 6.655948966503028e-07,
+      "loss": 0.5403,
+      "step": 51523
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 5.103923320770264,
+      "learning_rate": 6.654634650959551e-07,
+      "loss": 0.6829,
+      "step": 51524
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.887883424758911,
+      "learning_rate": 6.65332045594328e-07,
+      "loss": 0.5733,
+      "step": 51525
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.4514660835266113,
+      "learning_rate": 6.652006381457903e-07,
+      "loss": 0.6329,
+      "step": 51526
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.3261096477508545,
+      "learning_rate": 6.650692427507044e-07,
+      "loss": 0.1654,
+      "step": 51527
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.608438014984131,
+      "learning_rate": 6.649378594094369e-07,
+      "loss": 0.5799,
+      "step": 51528
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.886138916015625,
+      "learning_rate": 6.648064881223542e-07,
+      "loss": 0.6395,
+      "step": 51529
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.8657140731811523,
+      "learning_rate": 6.646751288898201e-07,
+      "loss": 0.6342,
+      "step": 51530
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.360281467437744,
+      "learning_rate": 6.645437817122013e-07,
+      "loss": 0.5369,
+      "step": 51531
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.5511231422424316,
+      "learning_rate": 6.64412446589861e-07,
+      "loss": 0.5604,
+      "step": 51532
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.524632453918457,
+      "learning_rate": 6.642811235231655e-07,
+      "loss": 0.444,
+      "step": 51533
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.999713182449341,
+      "learning_rate": 6.641498125124813e-07,
+      "loss": 0.7319,
+      "step": 51534
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.0862796306610107,
+      "learning_rate": 6.640185135581711e-07,
+      "loss": 0.5481,
+      "step": 51535
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.5548551082611084,
+      "learning_rate": 6.638872266606017e-07,
+      "loss": 0.5606,
+      "step": 51536
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.3504810333251953,
+      "learning_rate": 6.63755951820137e-07,
+      "loss": 0.5918,
+      "step": 51537
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 4.422217845916748,
+      "learning_rate": 6.63624689037144e-07,
+      "loss": 0.7583,
+      "step": 51538
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.761528491973877,
+      "learning_rate": 6.634934383119846e-07,
+      "loss": 0.693,
+      "step": 51539
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.567559003829956,
+      "learning_rate": 6.633621996450268e-07,
+      "loss": 0.6601,
+      "step": 51540
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 5.0383381843566895,
+      "learning_rate": 6.632309730366327e-07,
+      "loss": 0.872,
+      "step": 51541
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.5301899909973145,
+      "learning_rate": 6.630997584871701e-07,
+      "loss": 0.4785,
+      "step": 51542
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.7689578533172607,
+      "learning_rate": 6.629685559970017e-07,
+      "loss": 0.4994,
+      "step": 51543
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.510652542114258,
+      "learning_rate": 6.628373655664938e-07,
+      "loss": 0.5857,
+      "step": 51544
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.700313091278076,
+      "learning_rate": 6.627061871960088e-07,
+      "loss": 0.3619,
+      "step": 51545
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.880887746810913,
+      "learning_rate": 6.62575020885915e-07,
+      "loss": 0.7779,
+      "step": 51546
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.510265350341797,
+      "learning_rate": 6.62443866636574e-07,
+      "loss": 0.5812,
+      "step": 51547
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.4743692874908447,
+      "learning_rate": 6.623127244483524e-07,
+      "loss": 0.6756,
+      "step": 51548
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.1633760929107666,
+      "learning_rate": 6.621815943216125e-07,
+      "loss": 0.5331,
+      "step": 51549
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.6129510402679443,
+      "learning_rate": 6.620504762567226e-07,
+      "loss": 0.6796,
+      "step": 51550
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.8246922492980957,
+      "learning_rate": 6.619193702540438e-07,
+      "loss": 0.74,
+      "step": 51551
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 4.460388660430908,
+      "learning_rate": 6.617882763139439e-07,
+      "loss": 0.7054,
+      "step": 51552
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.6754310131073,
+      "learning_rate": 6.616571944367844e-07,
+      "loss": 0.7107,
+      "step": 51553
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 4.171797752380371,
+      "learning_rate": 6.615261246229321e-07,
+      "loss": 0.6625,
+      "step": 51554
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.7844817638397217,
+      "learning_rate": 6.613950668727492e-07,
+      "loss": 0.5398,
+      "step": 51555
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 5.234753608703613,
+      "learning_rate": 6.612640211866023e-07,
+      "loss": 0.6408,
+      "step": 51556
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.1706771850585938,
+      "learning_rate": 6.611329875648537e-07,
+      "loss": 0.7593,
+      "step": 51557
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.6592800617218018,
+      "learning_rate": 6.610019660078704e-07,
+      "loss": 0.5512,
+      "step": 51558
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.107891321182251,
+      "learning_rate": 6.608709565160137e-07,
+      "loss": 0.5526,
+      "step": 51559
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.7238962650299072,
+      "learning_rate": 6.607399590896507e-07,
+      "loss": 0.3589,
+      "step": 51560
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.8137564659118652,
+      "learning_rate": 6.606089737291438e-07,
+      "loss": 0.6063,
+      "step": 51561
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.7144081592559814,
+      "learning_rate": 6.604780004348582e-07,
+      "loss": 0.6203,
+      "step": 51562
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 1.742702841758728,
+      "learning_rate": 6.603470392071565e-07,
+      "loss": 0.5261,
+      "step": 51563
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.0309700965881348,
+      "learning_rate": 6.602160900464055e-07,
+      "loss": 0.4371,
+      "step": 51564
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.1770424842834473,
+      "learning_rate": 6.600851529529661e-07,
+      "loss": 0.8845,
+      "step": 51565
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.785980463027954,
+      "learning_rate": 6.599542279272059e-07,
+      "loss": 0.6779,
+      "step": 51566
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 4.5332746505737305,
+      "learning_rate": 6.598233149694861e-07,
+      "loss": 0.6942,
+      "step": 51567
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.0023372173309326,
+      "learning_rate": 6.596924140801724e-07,
+      "loss": 0.6515,
+      "step": 51568
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.150376796722412,
+      "learning_rate": 6.595615252596261e-07,
+      "loss": 0.6539,
+      "step": 51569
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.034440279006958,
+      "learning_rate": 6.594306485082158e-07,
+      "loss": 0.597,
+      "step": 51570
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 4.621651649475098,
+      "learning_rate": 6.592997838263015e-07,
+      "loss": 1.0376,
+      "step": 51571
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 4.709379196166992,
+      "learning_rate": 6.591689312142496e-07,
+      "loss": 0.695,
+      "step": 51572
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.925710678100586,
+      "learning_rate": 6.590380906724214e-07,
+      "loss": 0.4349,
+      "step": 51573
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.8838067054748535,
+      "learning_rate": 6.589072622011833e-07,
+      "loss": 0.5436,
+      "step": 51574
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.3027403354644775,
+      "learning_rate": 6.587764458008971e-07,
+      "loss": 0.4733,
+      "step": 51575
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.386683940887451,
+      "learning_rate": 6.586456414719283e-07,
+      "loss": 0.377,
+      "step": 51576
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 4.584319114685059,
+      "learning_rate": 6.585148492146387e-07,
+      "loss": 0.8287,
+      "step": 51577
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.0552215576171875,
+      "learning_rate": 6.583840690293936e-07,
+      "loss": 0.694,
+      "step": 51578
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.6182143688201904,
+      "learning_rate": 6.582533009165554e-07,
+      "loss": 0.6187,
+      "step": 51579
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.9711389541625977,
+      "learning_rate": 6.581225448764889e-07,
+      "loss": 0.5339,
+      "step": 51580
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.79461407661438,
+      "learning_rate": 6.579918009095565e-07,
+      "loss": 0.5309,
+      "step": 51581
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 1.8454580307006836,
+      "learning_rate": 6.578610690161219e-07,
+      "loss": 0.3604,
+      "step": 51582
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.615772247314453,
+      "learning_rate": 6.577303491965503e-07,
+      "loss": 0.6991,
+      "step": 51583
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.448035717010498,
+      "learning_rate": 6.575996414512032e-07,
+      "loss": 0.5079,
+      "step": 51584
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.1088356971740723,
+      "learning_rate": 6.574689457804456e-07,
+      "loss": 0.5431,
+      "step": 51585
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.1296803951263428,
+      "learning_rate": 6.573382621846391e-07,
+      "loss": 0.7386,
+      "step": 51586
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.8488452434539795,
+      "learning_rate": 6.572075906641489e-07,
+      "loss": 0.545,
+      "step": 51587
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.9288809299468994,
+      "learning_rate": 6.570769312193371e-07,
+      "loss": 0.6491,
+      "step": 51588
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.879115343093872,
+      "learning_rate": 6.56946283850567e-07,
+      "loss": 0.7828,
+      "step": 51589
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.3948588371276855,
+      "learning_rate": 6.568156485582033e-07,
+      "loss": 0.6155,
+      "step": 51590
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.9730594158172607,
+      "learning_rate": 6.566850253426071e-07,
+      "loss": 0.7459,
+      "step": 51591
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.8469138145446777,
+      "learning_rate": 6.565544142041447e-07,
+      "loss": 0.7204,
+      "step": 51592
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 4.413483619689941,
+      "learning_rate": 6.564238151431756e-07,
+      "loss": 0.6134,
+      "step": 51593
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 1.9199358224868774,
+      "learning_rate": 6.562932281600648e-07,
+      "loss": 0.4916,
+      "step": 51594
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.3213655948638916,
+      "learning_rate": 6.561626532551768e-07,
+      "loss": 0.4493,
+      "step": 51595
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.6170125007629395,
+      "learning_rate": 6.560320904288714e-07,
+      "loss": 0.281,
+      "step": 51596
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 1.8601762056350708,
+      "learning_rate": 6.559015396815155e-07,
+      "loss": 0.2939,
+      "step": 51597
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.0603837966918945,
+      "learning_rate": 6.557710010134682e-07,
+      "loss": 0.7716,
+      "step": 51598
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.7900311946868896,
+      "learning_rate": 6.556404744250961e-07,
+      "loss": 0.6366,
+      "step": 51599
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.993316650390625,
+      "learning_rate": 6.555099599167586e-07,
+      "loss": 0.6808,
+      "step": 51600
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 4.585622310638428,
+      "learning_rate": 6.553794574888222e-07,
+      "loss": 0.9228,
+      "step": 51601
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.3272602558135986,
+      "learning_rate": 6.552489671416462e-07,
+      "loss": 0.4805,
+      "step": 51602
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.659604072570801,
+      "learning_rate": 6.551184888755968e-07,
+      "loss": 0.7639,
+      "step": 51603
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.442167043685913,
+      "learning_rate": 6.549880226910343e-07,
+      "loss": 0.5092,
+      "step": 51604
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.7190160751342773,
+      "learning_rate": 6.548575685883229e-07,
+      "loss": 0.6423,
+      "step": 51605
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 1.9709378480911255,
+      "learning_rate": 6.547271265678245e-07,
+      "loss": 0.48,
+      "step": 51606
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.523897409439087,
+      "learning_rate": 6.545966966299028e-07,
+      "loss": 0.6821,
+      "step": 51607
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 6.195551872253418,
+      "learning_rate": 6.54466278774919e-07,
+      "loss": 0.6383,
+      "step": 51608
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.967471122741699,
+      "learning_rate": 6.543358730032378e-07,
+      "loss": 0.7675,
+      "step": 51609
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.4344701766967773,
+      "learning_rate": 6.54205479315218e-07,
+      "loss": 0.5558,
+      "step": 51610
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.99430513381958,
+      "learning_rate": 6.540750977112276e-07,
+      "loss": 0.6333,
+      "step": 51611
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.803842306137085,
+      "learning_rate": 6.539447281916245e-07,
+      "loss": 0.6449,
+      "step": 51612
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.567518472671509,
+      "learning_rate": 6.538143707567745e-07,
+      "loss": 0.7647,
+      "step": 51613
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.6961874961853027,
+      "learning_rate": 6.536840254070376e-07,
+      "loss": 0.5334,
+      "step": 51614
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.868908166885376,
+      "learning_rate": 6.535536921427782e-07,
+      "loss": 0.4519,
+      "step": 51615
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.801133871078491,
+      "learning_rate": 6.534233709643573e-07,
+      "loss": 0.4377,
+      "step": 51616
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.8707544803619385,
+      "learning_rate": 6.532930618721378e-07,
+      "loss": 0.5257,
+      "step": 51617
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.1701626777648926,
+      "learning_rate": 6.531627648664818e-07,
+      "loss": 0.4968,
+      "step": 51618
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.950974702835083,
+      "learning_rate": 6.530324799477522e-07,
+      "loss": 0.7496,
+      "step": 51619
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.126227378845215,
+      "learning_rate": 6.529022071163099e-07,
+      "loss": 0.5396,
+      "step": 51620
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.8503291606903076,
+      "learning_rate": 6.527719463725196e-07,
+      "loss": 0.7877,
+      "step": 51621
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.3350698947906494,
+      "learning_rate": 6.526416977167405e-07,
+      "loss": 0.6982,
+      "step": 51622
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.650784492492676,
+      "learning_rate": 6.525114611493377e-07,
+      "loss": 0.6565,
+      "step": 51623
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.5393755435943604,
+      "learning_rate": 6.523812366706705e-07,
+      "loss": 0.6711,
+      "step": 51624
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.7853832244873047,
+      "learning_rate": 6.522510242811036e-07,
+      "loss": 0.5576,
+      "step": 51625
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.17490291595459,
+      "learning_rate": 6.521208239809967e-07,
+      "loss": 0.6822,
+      "step": 51626
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.4178829193115234,
+      "learning_rate": 6.519906357707145e-07,
+      "loss": 0.432,
+      "step": 51627
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.1607158184051514,
+      "learning_rate": 6.518604596506162e-07,
+      "loss": 0.3726,
+      "step": 51628
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.2802629470825195,
+      "learning_rate": 6.517302956210664e-07,
+      "loss": 0.4417,
+      "step": 51629
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.3213858604431152,
+      "learning_rate": 6.516001436824232e-07,
+      "loss": 0.8904,
+      "step": 51630
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.5472066402435303,
+      "learning_rate": 6.514700038350535e-07,
+      "loss": 0.51,
+      "step": 51631
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.4778497219085693,
+      "learning_rate": 6.513398760793155e-07,
+      "loss": 0.595,
+      "step": 51632
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.9782257080078125,
+      "learning_rate": 6.512097604155737e-07,
+      "loss": 0.6463,
+      "step": 51633
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.7185168266296387,
+      "learning_rate": 6.510796568441868e-07,
+      "loss": 0.5517,
+      "step": 51634
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.6374828815460205,
+      "learning_rate": 6.509495653655196e-07,
+      "loss": 0.628,
+      "step": 51635
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.5133838653564453,
+      "learning_rate": 6.508194859799316e-07,
+      "loss": 0.3521,
+      "step": 51636
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.3055131435394287,
+      "learning_rate": 6.50689418687786e-07,
+      "loss": 0.5691,
+      "step": 51637
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.458397150039673,
+      "learning_rate": 6.50559363489443e-07,
+      "loss": 0.5223,
+      "step": 51638
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.9292988777160645,
+      "learning_rate": 6.504293203852663e-07,
+      "loss": 0.4721,
+      "step": 51639
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.8312718868255615,
+      "learning_rate": 6.50299289375615e-07,
+      "loss": 0.6568,
+      "step": 51640
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.175478935241699,
+      "learning_rate": 6.501692704608531e-07,
+      "loss": 0.4957,
+      "step": 51641
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.071136236190796,
+      "learning_rate": 6.500392636413399e-07,
+      "loss": 0.5783,
+      "step": 51642
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.795213222503662,
+      "learning_rate": 6.49909268917438e-07,
+      "loss": 0.3766,
+      "step": 51643
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.4882490634918213,
+      "learning_rate": 6.497792862895097e-07,
+      "loss": 0.4836,
+      "step": 51644
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.527897357940674,
+      "learning_rate": 6.496493157579148e-07,
+      "loss": 0.4505,
+      "step": 51645
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.09537935256958,
+      "learning_rate": 6.495193573230163e-07,
+      "loss": 0.6747,
+      "step": 51646
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.2245359420776367,
+      "learning_rate": 6.49389410985174e-07,
+      "loss": 0.5471,
+      "step": 51647
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.892143487930298,
+      "learning_rate": 6.492594767447508e-07,
+      "loss": 0.4725,
+      "step": 51648
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.8681774139404297,
+      "learning_rate": 6.491295546021059e-07,
+      "loss": 0.448,
+      "step": 51649
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.7494707107543945,
+      "learning_rate": 6.489996445576025e-07,
+      "loss": 0.5183,
+      "step": 51650
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 4.7683892250061035,
+      "learning_rate": 6.488697466116017e-07,
+      "loss": 0.9653,
+      "step": 51651
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.115933656692505,
+      "learning_rate": 6.487398607644629e-07,
+      "loss": 0.5857,
+      "step": 51652
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.2038893699645996,
+      "learning_rate": 6.486099870165502e-07,
+      "loss": 0.7737,
+      "step": 51653
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.3381552696228027,
+      "learning_rate": 6.484801253682216e-07,
+      "loss": 0.7777,
+      "step": 51654
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.3139467239379883,
+      "learning_rate": 6.483502758198395e-07,
+      "loss": 0.5277,
+      "step": 51655
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.4382472038269043,
+      "learning_rate": 6.482204383717666e-07,
+      "loss": 0.5066,
+      "step": 51656
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.554776668548584,
+      "learning_rate": 6.480906130243613e-07,
+      "loss": 0.5741,
+      "step": 51657
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.3215279579162598,
+      "learning_rate": 6.479607997779868e-07,
+      "loss": 0.6708,
+      "step": 51658
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.3503613471984863,
+      "learning_rate": 6.478309986330017e-07,
+      "loss": 0.3541,
+      "step": 51659
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 5.071528911590576,
+      "learning_rate": 6.477012095897695e-07,
+      "loss": 0.6293,
+      "step": 51660
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.310048818588257,
+      "learning_rate": 6.475714326486487e-07,
+      "loss": 0.5185,
+      "step": 51661
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.1306703090667725,
+      "learning_rate": 6.47441667810002e-07,
+      "loss": 0.6717,
+      "step": 51662
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.281095027923584,
+      "learning_rate": 6.473119150741891e-07,
+      "loss": 0.3702,
+      "step": 51663
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.523272752761841,
+      "learning_rate": 6.471821744415719e-07,
+      "loss": 0.647,
+      "step": 51664
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.782538890838623,
+      "learning_rate": 6.470524459125093e-07,
+      "loss": 0.8149,
+      "step": 51665
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.5506155490875244,
+      "learning_rate": 6.469227294873648e-07,
+      "loss": 0.4234,
+      "step": 51666
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.284761428833008,
+      "learning_rate": 6.467930251664961e-07,
+      "loss": 0.5661,
+      "step": 51667
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.494814872741699,
+      "learning_rate": 6.466633329502664e-07,
+      "loss": 0.3001,
+      "step": 51668
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.2287611961364746,
+      "learning_rate": 6.465336528390336e-07,
+      "loss": 0.6853,
+      "step": 51669
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 1.8168331384658813,
+      "learning_rate": 6.464039848331615e-07,
+      "loss": 0.5503,
+      "step": 51670
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.8615972995758057,
+      "learning_rate": 6.462743289330065e-07,
+      "loss": 0.5547,
+      "step": 51671
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.65151309967041,
+      "learning_rate": 6.461446851389341e-07,
+      "loss": 0.4591,
+      "step": 51672
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 7.840366840362549,
+      "learning_rate": 6.460150534513016e-07,
+      "loss": 0.4347,
+      "step": 51673
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.4593586921691895,
+      "learning_rate": 6.458854338704707e-07,
+      "loss": 0.5796,
+      "step": 51674
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.6211321353912354,
+      "learning_rate": 6.457558263968e-07,
+      "loss": 0.681,
+      "step": 51675
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.0145795345306396,
+      "learning_rate": 6.456262310306527e-07,
+      "loss": 0.7413,
+      "step": 51676
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.4128997325897217,
+      "learning_rate": 6.454966477723867e-07,
+      "loss": 0.6265,
+      "step": 51677
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.680438995361328,
+      "learning_rate": 6.453670766223641e-07,
+      "loss": 0.6308,
+      "step": 51678
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.773897409439087,
+      "learning_rate": 6.452375175809439e-07,
+      "loss": 0.5509,
+      "step": 51679
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.9133362770080566,
+      "learning_rate": 6.451079706484869e-07,
+      "loss": 0.479,
+      "step": 51680
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 5.79958438873291,
+      "learning_rate": 6.44978435825353e-07,
+      "loss": 0.655,
+      "step": 51681
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 1.8554649353027344,
+      "learning_rate": 6.448489131119035e-07,
+      "loss": 0.5073,
+      "step": 51682
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.5290138721466064,
+      "learning_rate": 6.447194025084963e-07,
+      "loss": 0.6774,
+      "step": 51683
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.293964385986328,
+      "learning_rate": 6.445899040154946e-07,
+      "loss": 0.5932,
+      "step": 51684
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 4.025113105773926,
+      "learning_rate": 6.444604176332553e-07,
+      "loss": 0.4018,
+      "step": 51685
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.4434099197387695,
+      "learning_rate": 6.443309433621408e-07,
+      "loss": 0.3442,
+      "step": 51686
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.442070960998535,
+      "learning_rate": 6.442014812025099e-07,
+      "loss": 0.5681,
+      "step": 51687
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.7552342414855957,
+      "learning_rate": 6.440720311547233e-07,
+      "loss": 0.6823,
+      "step": 51688
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.4273951053619385,
+      "learning_rate": 6.439425932191401e-07,
+      "loss": 0.6517,
+      "step": 51689
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.636746406555176,
+      "learning_rate": 6.438131673961217e-07,
+      "loss": 0.6183,
+      "step": 51690
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.5608677864074707,
+      "learning_rate": 6.436837536860246e-07,
+      "loss": 0.7265,
+      "step": 51691
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.210545063018799,
+      "learning_rate": 6.435543520892135e-07,
+      "loss": 0.4391,
+      "step": 51692
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.1834166049957275,
+      "learning_rate": 6.434249626060446e-07,
+      "loss": 0.5294,
+      "step": 51693
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.308173656463623,
+      "learning_rate": 6.4329558523688e-07,
+      "loss": 0.4746,
+      "step": 51694
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 4.035283088684082,
+      "learning_rate": 6.431662199820771e-07,
+      "loss": 0.7674,
+      "step": 51695
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.4867236614227295,
+      "learning_rate": 6.430368668419978e-07,
+      "loss": 0.6473,
+      "step": 51696
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.047208070755005,
+      "learning_rate": 6.429075258169998e-07,
+      "loss": 0.6104,
+      "step": 51697
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.1289708614349365,
+      "learning_rate": 6.427781969074448e-07,
+      "loss": 0.6343,
+      "step": 51698
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.264828681945801,
+      "learning_rate": 6.426488801136904e-07,
+      "loss": 0.7139,
+      "step": 51699
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 10.346664428710938,
+      "learning_rate": 6.425195754360986e-07,
+      "loss": 0.5603,
+      "step": 51700
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.6623952388763428,
+      "learning_rate": 6.423902828750261e-07,
+      "loss": 0.3605,
+      "step": 51701
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.6637184619903564,
+      "learning_rate": 6.422610024308351e-07,
+      "loss": 0.579,
+      "step": 51702
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.754891872406006,
+      "learning_rate": 6.421317341038824e-07,
+      "loss": 0.7175,
+      "step": 51703
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.500094413757324,
+      "learning_rate": 6.420024778945299e-07,
+      "loss": 0.5068,
+      "step": 51704
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.632700204849243,
+      "learning_rate": 6.418732338031353e-07,
+      "loss": 0.4402,
+      "step": 51705
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.3333523273468018,
+      "learning_rate": 6.417440018300586e-07,
+      "loss": 0.6296,
+      "step": 51706
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.369291305541992,
+      "learning_rate": 6.416147819756602e-07,
+      "loss": 0.5836,
+      "step": 51707
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.7919344902038574,
+      "learning_rate": 6.41485574240297e-07,
+      "loss": 0.5769,
+      "step": 51708
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.032043933868408,
+      "learning_rate": 6.41356378624331e-07,
+      "loss": 0.4537,
+      "step": 51709
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.4386398792266846,
+      "learning_rate": 6.41227195128119e-07,
+      "loss": 0.4176,
+      "step": 51710
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.143690824508667,
+      "learning_rate": 6.410980237520215e-07,
+      "loss": 0.5325,
+      "step": 51711
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.0319571495056152,
+      "learning_rate": 6.409688644963985e-07,
+      "loss": 0.51,
+      "step": 51712
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.1905577182769775,
+      "learning_rate": 6.408397173616071e-07,
+      "loss": 0.7184,
+      "step": 51713
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.4366536140441895,
+      "learning_rate": 6.407105823480081e-07,
+      "loss": 0.5107,
+      "step": 51714
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 4.1307806968688965,
+      "learning_rate": 6.405814594559595e-07,
+      "loss": 0.3625,
+      "step": 51715
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.06282114982605,
+      "learning_rate": 6.404523486858216e-07,
+      "loss": 0.2713,
+      "step": 51716
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.1010665893554688,
+      "learning_rate": 6.403232500379514e-07,
+      "loss": 0.4771,
+      "step": 51717
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.0578300952911377,
+      "learning_rate": 6.401941635127091e-07,
+      "loss": 0.9299,
+      "step": 51718
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.7543179988861084,
+      "learning_rate": 6.400650891104548e-07,
+      "loss": 0.5956,
+      "step": 51719
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.4792680740356445,
+      "learning_rate": 6.399360268315452e-07,
+      "loss": 0.6847,
+      "step": 51720
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.401794672012329,
+      "learning_rate": 6.398069766763415e-07,
+      "loss": 0.3841,
+      "step": 51721
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.1666736602783203,
+      "learning_rate": 6.396779386451995e-07,
+      "loss": 0.7711,
+      "step": 51722
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 1.843748688697815,
+      "learning_rate": 6.395489127384813e-07,
+      "loss": 0.3026,
+      "step": 51723
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.0934877395629883,
+      "learning_rate": 6.394198989565431e-07,
+      "loss": 0.5152,
+      "step": 51724
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.3899264335632324,
+      "learning_rate": 6.392908972997452e-07,
+      "loss": 0.532,
+      "step": 51725
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.382286787033081,
+      "learning_rate": 6.391619077684452e-07,
+      "loss": 0.6011,
+      "step": 51726
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.6130640506744385,
+      "learning_rate": 6.390329303630033e-07,
+      "loss": 0.5445,
+      "step": 51727
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.057692527770996,
+      "learning_rate": 6.389039650837758e-07,
+      "loss": 0.7308,
+      "step": 51728
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 4.9179182052612305,
+      "learning_rate": 6.387750119311243e-07,
+      "loss": 0.7359,
+      "step": 51729
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 1.7381529808044434,
+      "learning_rate": 6.386460709054043e-07,
+      "loss": 0.5038,
+      "step": 51730
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.971625566482544,
+      "learning_rate": 6.385171420069769e-07,
+      "loss": 0.5479,
+      "step": 51731
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.5451982021331787,
+      "learning_rate": 6.383882252361972e-07,
+      "loss": 0.4165,
+      "step": 51732
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 5.466703414916992,
+      "learning_rate": 6.382593205934279e-07,
+      "loss": 0.6396,
+      "step": 51733
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.655819892883301,
+      "learning_rate": 6.381304280790246e-07,
+      "loss": 0.481,
+      "step": 51734
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.832679271697998,
+      "learning_rate": 6.380015476933482e-07,
+      "loss": 0.5987,
+      "step": 51735
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.4040451049804688,
+      "learning_rate": 6.37872679436754e-07,
+      "loss": 0.4418,
+      "step": 51736
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.834761142730713,
+      "learning_rate": 6.377438233096028e-07,
+      "loss": 0.5536,
+      "step": 51737
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.2846245765686035,
+      "learning_rate": 6.376149793122505e-07,
+      "loss": 0.5488,
+      "step": 51738
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.185331344604492,
+      "learning_rate": 6.374861474450583e-07,
+      "loss": 0.4737,
+      "step": 51739
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.928262233734131,
+      "learning_rate": 6.373573277083816e-07,
+      "loss": 0.6997,
+      "step": 51740
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.253408670425415,
+      "learning_rate": 6.372285201025813e-07,
+      "loss": 0.5532,
+      "step": 51741
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.10629940032959,
+      "learning_rate": 6.370997246280125e-07,
+      "loss": 0.6941,
+      "step": 51742
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 1.9052071571350098,
+      "learning_rate": 6.369709412850367e-07,
+      "loss": 0.4345,
+      "step": 51743
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.6394541263580322,
+      "learning_rate": 6.368421700740085e-07,
+      "loss": 0.4736,
+      "step": 51744
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.4128379821777344,
+      "learning_rate": 6.367134109952894e-07,
+      "loss": 0.5716,
+      "step": 51745
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.46521258354187,
+      "learning_rate": 6.365846640492346e-07,
+      "loss": 0.5608,
+      "step": 51746
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.3414814472198486,
+      "learning_rate": 6.364559292362038e-07,
+      "loss": 0.4802,
+      "step": 51747
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.338921546936035,
+      "learning_rate": 6.36327206556554e-07,
+      "loss": 0.5399,
+      "step": 51748
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.5935001373291016,
+      "learning_rate": 6.361984960106443e-07,
+      "loss": 0.6009,
+      "step": 51749
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.909743547439575,
+      "learning_rate": 6.360697975988312e-07,
+      "loss": 0.6955,
+      "step": 51750
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.3894150257110596,
+      "learning_rate": 6.359411113214742e-07,
+      "loss": 0.4601,
+      "step": 51751
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.8251450061798096,
+      "learning_rate": 6.358124371789276e-07,
+      "loss": 0.7189,
+      "step": 51752
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.46217942237854,
+      "learning_rate": 6.356837751715545e-07,
+      "loss": 0.5332,
+      "step": 51753
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.6115598678588867,
+      "learning_rate": 6.355551252997083e-07,
+      "loss": 0.7266,
+      "step": 51754
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.189535140991211,
+      "learning_rate": 6.354264875637495e-07,
+      "loss": 0.6678,
+      "step": 51755
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.469541311264038,
+      "learning_rate": 6.352978619640338e-07,
+      "loss": 0.5396,
+      "step": 51756
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.154956817626953,
+      "learning_rate": 6.351692485009204e-07,
+      "loss": 0.4346,
+      "step": 51757
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.240912675857544,
+      "learning_rate": 6.350406471747656e-07,
+      "loss": 0.5656,
+      "step": 51758
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.8613526821136475,
+      "learning_rate": 6.349120579859286e-07,
+      "loss": 0.6263,
+      "step": 51759
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 4.082002639770508,
+      "learning_rate": 6.347834809347648e-07,
+      "loss": 0.6364,
+      "step": 51760
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.041264295578003,
+      "learning_rate": 6.346549160216336e-07,
+      "loss": 0.5086,
+      "step": 51761
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.126431465148926,
+      "learning_rate": 6.345263632468912e-07,
+      "loss": 0.6628,
+      "step": 51762
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 4.409718036651611,
+      "learning_rate": 6.343978226108965e-07,
+      "loss": 0.5923,
+      "step": 51763
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 4.107570648193359,
+      "learning_rate": 6.342692941140049e-07,
+      "loss": 0.4671,
+      "step": 51764
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.417454481124878,
+      "learning_rate": 6.341407777565761e-07,
+      "loss": 0.5475,
+      "step": 51765
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 5.1266679763793945,
+      "learning_rate": 6.340122735389653e-07,
+      "loss": 0.6178,
+      "step": 51766
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.5364880561828613,
+      "learning_rate": 6.338837814615312e-07,
+      "loss": 0.5849,
+      "step": 51767
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.1614301204681396,
+      "learning_rate": 6.337553015246312e-07,
+      "loss": 0.4762,
+      "step": 51768
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.2520534992218018,
+      "learning_rate": 6.336268337286211e-07,
+      "loss": 0.4709,
+      "step": 51769
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.5862319469451904,
+      "learning_rate": 6.334983780738602e-07,
+      "loss": 0.4592,
+      "step": 51770
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.2291910648345947,
+      "learning_rate": 6.333699345607037e-07,
+      "loss": 0.5881,
+      "step": 51771
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.81820011138916,
+      "learning_rate": 6.332415031895095e-07,
+      "loss": 0.5345,
+      "step": 51772
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.0103585720062256,
+      "learning_rate": 6.331130839606359e-07,
+      "loss": 0.4404,
+      "step": 51773
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.077845335006714,
+      "learning_rate": 6.32984676874438e-07,
+      "loss": 0.4945,
+      "step": 51774
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.489187002182007,
+      "learning_rate": 6.328562819312751e-07,
+      "loss": 0.299,
+      "step": 51775
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.05550479888916,
+      "learning_rate": 6.327278991315011e-07,
+      "loss": 0.3704,
+      "step": 51776
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.132585048675537,
+      "learning_rate": 6.325995284754765e-07,
+      "loss": 0.6044,
+      "step": 51777
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.85796856880188,
+      "learning_rate": 6.324711699635555e-07,
+      "loss": 0.6068,
+      "step": 51778
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.3113114833831787,
+      "learning_rate": 6.323428235960954e-07,
+      "loss": 0.5298,
+      "step": 51779
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.0785233974456787,
+      "learning_rate": 6.322144893734556e-07,
+      "loss": 0.5664,
+      "step": 51780
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.838444471359253,
+      "learning_rate": 6.320861672959894e-07,
+      "loss": 0.4848,
+      "step": 51781
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.6722829341888428,
+      "learning_rate": 6.319578573640567e-07,
+      "loss": 0.69,
+      "step": 51782
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.3309991359710693,
+      "learning_rate": 6.318295595780116e-07,
+      "loss": 0.5497,
+      "step": 51783
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 4.917741298675537,
+      "learning_rate": 6.317012739382139e-07,
+      "loss": 0.7453,
+      "step": 51784
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 4.581064224243164,
+      "learning_rate": 6.315730004450166e-07,
+      "loss": 0.5544,
+      "step": 51785
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.964397668838501,
+      "learning_rate": 6.314447390987799e-07,
+      "loss": 0.4661,
+      "step": 51786
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.149009943008423,
+      "learning_rate": 6.313164898998575e-07,
+      "loss": 0.5018,
+      "step": 51787
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.4308829307556152,
+      "learning_rate": 6.311882528486091e-07,
+      "loss": 0.6826,
+      "step": 51788
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.3588521480560303,
+      "learning_rate": 6.310600279453877e-07,
+      "loss": 0.5691,
+      "step": 51789
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.047380208969116,
+      "learning_rate": 6.30931815190553e-07,
+      "loss": 0.4708,
+      "step": 51790
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.2906625270843506,
+      "learning_rate": 6.308036145844592e-07,
+      "loss": 0.5846,
+      "step": 51791
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 11.40972900390625,
+      "learning_rate": 6.306754261274656e-07,
+      "loss": 0.5196,
+      "step": 51792
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 4.694332122802734,
+      "learning_rate": 6.305472498199239e-07,
+      "loss": 0.5502,
+      "step": 51793
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.865349769592285,
+      "learning_rate": 6.304190856621956e-07,
+      "loss": 0.5423,
+      "step": 51794
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.07653546333313,
+      "learning_rate": 6.302909336546342e-07,
+      "loss": 0.8542,
+      "step": 51795
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.5465290546417236,
+      "learning_rate": 6.301627937975979e-07,
+      "loss": 0.5465,
+      "step": 51796
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.8770155906677246,
+      "learning_rate": 6.300346660914402e-07,
+      "loss": 0.9145,
+      "step": 51797
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.1527371406555176,
+      "learning_rate": 6.29906550536521e-07,
+      "loss": 0.6514,
+      "step": 51798
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.2880072593688965,
+      "learning_rate": 6.29778447133193e-07,
+      "loss": 0.5973,
+      "step": 51799
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.515441417694092,
+      "learning_rate": 6.296503558818151e-07,
+      "loss": 0.6465,
+      "step": 51800
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.123396158218384,
+      "learning_rate": 6.295222767827413e-07,
+      "loss": 0.6033,
+      "step": 51801
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 4.3678765296936035,
+      "learning_rate": 6.293942098363304e-07,
+      "loss": 0.7008,
+      "step": 51802
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 4.475418567657471,
+      "learning_rate": 6.292661550429353e-07,
+      "loss": 0.6483,
+      "step": 51803
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.801849365234375,
+      "learning_rate": 6.29138112402915e-07,
+      "loss": 0.6961,
+      "step": 51804
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.7625856399536133,
+      "learning_rate": 6.290100819166233e-07,
+      "loss": 0.5838,
+      "step": 51805
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.279766798019409,
+      "learning_rate": 6.288820635844179e-07,
+      "loss": 0.424,
+      "step": 51806
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.393514633178711,
+      "learning_rate": 6.287540574066531e-07,
+      "loss": 0.5144,
+      "step": 51807
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.1063640117645264,
+      "learning_rate": 6.286260633836872e-07,
+      "loss": 0.6204,
+      "step": 51808
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.8099727630615234,
+      "learning_rate": 6.28498081515873e-07,
+      "loss": 0.4687,
+      "step": 51809
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.3064980506896973,
+      "learning_rate": 6.283701118035696e-07,
+      "loss": 0.6624,
+      "step": 51810
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.672245502471924,
+      "learning_rate": 6.282421542471295e-07,
+      "loss": 0.7193,
+      "step": 51811
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.814373016357422,
+      "learning_rate": 6.281142088469122e-07,
+      "loss": 0.8706,
+      "step": 51812
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.0210227966308594,
+      "learning_rate": 6.27986275603269e-07,
+      "loss": 0.4459,
+      "step": 51813
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.607224941253662,
+      "learning_rate": 6.2785835451656e-07,
+      "loss": 0.5758,
+      "step": 51814
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.5441958904266357,
+      "learning_rate": 6.277304455871386e-07,
+      "loss": 0.4998,
+      "step": 51815
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.323019027709961,
+      "learning_rate": 6.276025488153614e-07,
+      "loss": 0.5035,
+      "step": 51816
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.601950168609619,
+      "learning_rate": 6.274746642015833e-07,
+      "loss": 0.8636,
+      "step": 51817
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.6180238723754883,
+      "learning_rate": 6.273467917461607e-07,
+      "loss": 0.3815,
+      "step": 51818
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.9738495349884033,
+      "learning_rate": 6.272189314494476e-07,
+      "loss": 0.5642,
+      "step": 51819
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.394554376602173,
+      "learning_rate": 6.270910833118016e-07,
+      "loss": 0.5347,
+      "step": 51820
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.8388888835906982,
+      "learning_rate": 6.269632473335763e-07,
+      "loss": 0.8054,
+      "step": 51821
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.8081817626953125,
+      "learning_rate": 6.268354235151286e-07,
+      "loss": 0.6115,
+      "step": 51822
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.3547887802124023,
+      "learning_rate": 6.267076118568127e-07,
+      "loss": 0.6471,
+      "step": 51823
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.729383707046509,
+      "learning_rate": 6.265798123589861e-07,
+      "loss": 0.329,
+      "step": 51824
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.322565793991089,
+      "learning_rate": 6.264520250220014e-07,
+      "loss": 0.4989,
+      "step": 51825
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.698716402053833,
+      "learning_rate": 6.263242498462158e-07,
+      "loss": 0.775,
+      "step": 51826
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.5541765689849854,
+      "learning_rate": 6.261964868319836e-07,
+      "loss": 0.636,
+      "step": 51827
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.1745121479034424,
+      "learning_rate": 6.260687359796602e-07,
+      "loss": 0.4686,
+      "step": 51828
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.7209277153015137,
+      "learning_rate": 6.259409972896019e-07,
+      "loss": 0.6918,
+      "step": 51829
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.9878063201904297,
+      "learning_rate": 6.258132707621628e-07,
+      "loss": 0.5095,
+      "step": 51830
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.0040042400360107,
+      "learning_rate": 6.256855563976988e-07,
+      "loss": 0.59,
+      "step": 51831
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.2173075675964355,
+      "learning_rate": 6.25557854196564e-07,
+      "loss": 0.3819,
+      "step": 51832
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.7277815341949463,
+      "learning_rate": 6.254301641591137e-07,
+      "loss": 0.7524,
+      "step": 51833
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 4.457657337188721,
+      "learning_rate": 6.253024862857049e-07,
+      "loss": 0.413,
+      "step": 51834
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.6205108165740967,
+      "learning_rate": 6.251748205766895e-07,
+      "loss": 0.7179,
+      "step": 51835
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 1.7736573219299316,
+      "learning_rate": 6.250471670324254e-07,
+      "loss": 0.3846,
+      "step": 51836
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.904449939727783,
+      "learning_rate": 6.249195256532653e-07,
+      "loss": 0.5369,
+      "step": 51837
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 4.307457447052002,
+      "learning_rate": 6.24791896439566e-07,
+      "loss": 0.4954,
+      "step": 51838
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.6092894077301025,
+      "learning_rate": 6.246642793916802e-07,
+      "loss": 0.525,
+      "step": 51839
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.472050428390503,
+      "learning_rate": 6.245366745099646e-07,
+      "loss": 0.7351,
+      "step": 51840
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.6491200923919678,
+      "learning_rate": 6.24409081794774e-07,
+      "loss": 0.3515,
+      "step": 51841
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.3948018550872803,
+      "learning_rate": 6.242815012464615e-07,
+      "loss": 0.4196,
+      "step": 51842
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.4146831035614014,
+      "learning_rate": 6.24153932865384e-07,
+      "loss": 0.5712,
+      "step": 51843
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.678150177001953,
+      "learning_rate": 6.240263766518945e-07,
+      "loss": 0.5303,
+      "step": 51844
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.62119460105896,
+      "learning_rate": 6.238988326063488e-07,
+      "loss": 0.6636,
+      "step": 51845
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.8352761268615723,
+      "learning_rate": 6.237713007291008e-07,
+      "loss": 0.6354,
+      "step": 51846
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.1355268955230713,
+      "learning_rate": 6.23643781020506e-07,
+      "loss": 0.5624,
+      "step": 51847
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.395700693130493,
+      "learning_rate": 6.235162734809175e-07,
+      "loss": 0.579,
+      "step": 51848
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.415018081665039,
+      "learning_rate": 6.233887781106917e-07,
+      "loss": 0.5233,
+      "step": 51849
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.918515205383301,
+      "learning_rate": 6.232612949101813e-07,
+      "loss": 0.8521,
+      "step": 51850
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.901404619216919,
+      "learning_rate": 6.23133823879743e-07,
+      "loss": 0.6637,
+      "step": 51851
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.3603928089141846,
+      "learning_rate": 6.230063650197288e-07,
+      "loss": 0.6683,
+      "step": 51852
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.2405128479003906,
+      "learning_rate": 6.228789183304951e-07,
+      "loss": 0.5831,
+      "step": 51853
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.5416793823242188,
+      "learning_rate": 6.227514838123933e-07,
+      "loss": 0.5426,
+      "step": 51854
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.099928379058838,
+      "learning_rate": 6.22624061465782e-07,
+      "loss": 0.3509,
+      "step": 51855
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.2465906143188477,
+      "learning_rate": 6.224966512910119e-07,
+      "loss": 0.5259,
+      "step": 51856
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.4275097846984863,
+      "learning_rate": 6.223692532884402e-07,
+      "loss": 0.6359,
+      "step": 51857
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.7220335006713867,
+      "learning_rate": 6.222418674584186e-07,
+      "loss": 0.5879,
+      "step": 51858
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.634002208709717,
+      "learning_rate": 6.221144938013035e-07,
+      "loss": 0.5577,
+      "step": 51859
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.5623490810394287,
+      "learning_rate": 6.21987132317447e-07,
+      "loss": 0.4475,
+      "step": 51860
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.1960928440093994,
+      "learning_rate": 6.218597830072054e-07,
+      "loss": 0.6624,
+      "step": 51861
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.285079002380371,
+      "learning_rate": 6.217324458709306e-07,
+      "loss": 0.701,
+      "step": 51862
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.797060489654541,
+      "learning_rate": 6.216051209089785e-07,
+      "loss": 0.565,
+      "step": 51863
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.5225231647491455,
+      "learning_rate": 6.214778081217016e-07,
+      "loss": 0.6809,
+      "step": 51864
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.39098858833313,
+      "learning_rate": 6.213505075094556e-07,
+      "loss": 0.7824,
+      "step": 51865
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.7381534576416016,
+      "learning_rate": 6.212232190725926e-07,
+      "loss": 0.6799,
+      "step": 51866
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.794240951538086,
+      "learning_rate": 6.210959428114682e-07,
+      "loss": 0.6814,
+      "step": 51867
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.9086246490478516,
+      "learning_rate": 6.209686787264352e-07,
+      "loss": 0.6662,
+      "step": 51868
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 4.227431774139404,
+      "learning_rate": 6.208414268178486e-07,
+      "loss": 0.5039,
+      "step": 51869
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.3041961193084717,
+      "learning_rate": 6.207141870860606e-07,
+      "loss": 0.5409,
+      "step": 51870
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.797844886779785,
+      "learning_rate": 6.205869595314267e-07,
+      "loss": 0.3971,
+      "step": 51871
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.040738821029663,
+      "learning_rate": 6.204597441542987e-07,
+      "loss": 0.4089,
+      "step": 51872
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 4.5423736572265625,
+      "learning_rate": 6.203325409550331e-07,
+      "loss": 0.6574,
+      "step": 51873
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.001768112182617,
+      "learning_rate": 6.202053499339794e-07,
+      "loss": 0.5357,
+      "step": 51874
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.479782819747925,
+      "learning_rate": 6.200781710914966e-07,
+      "loss": 0.6674,
+      "step": 51875
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.4230024814605713,
+      "learning_rate": 6.199510044279344e-07,
+      "loss": 0.5038,
+      "step": 51876
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 4.203582763671875,
+      "learning_rate": 6.198238499436488e-07,
+      "loss": 0.6063,
+      "step": 51877
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.208472967147827,
+      "learning_rate": 6.19696707638991e-07,
+      "loss": 0.7293,
+      "step": 51878
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.7365612983703613,
+      "learning_rate": 6.195695775143167e-07,
+      "loss": 0.4916,
+      "step": 51879
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.894986629486084,
+      "learning_rate": 6.194424595699772e-07,
+      "loss": 0.641,
+      "step": 51880
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.9242751598358154,
+      "learning_rate": 6.193153538063285e-07,
+      "loss": 0.817,
+      "step": 51881
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.6493563652038574,
+      "learning_rate": 6.191882602237221e-07,
+      "loss": 0.4042,
+      "step": 51882
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.279416799545288,
+      "learning_rate": 6.190611788225126e-07,
+      "loss": 0.6681,
+      "step": 51883
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.292933225631714,
+      "learning_rate": 6.189341096030521e-07,
+      "loss": 0.6377,
+      "step": 51884
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 4.500417709350586,
+      "learning_rate": 6.188070525656958e-07,
+      "loss": 0.7043,
+      "step": 51885
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.775097131729126,
+      "learning_rate": 6.186800077107946e-07,
+      "loss": 0.5976,
+      "step": 51886
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.591614007949829,
+      "learning_rate": 6.185529750387037e-07,
+      "loss": 0.6166,
+      "step": 51887
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.5839977264404297,
+      "learning_rate": 6.184259545497751e-07,
+      "loss": 0.5494,
+      "step": 51888
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.3306193351745605,
+      "learning_rate": 6.182989462443634e-07,
+      "loss": 0.5356,
+      "step": 51889
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.5388729572296143,
+      "learning_rate": 6.181719501228201e-07,
+      "loss": 0.5498,
+      "step": 51890
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.259221315383911,
+      "learning_rate": 6.180449661854987e-07,
+      "loss": 0.7966,
+      "step": 51891
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.355672836303711,
+      "learning_rate": 6.17917994432754e-07,
+      "loss": 0.5349,
+      "step": 51892
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.8101792335510254,
+      "learning_rate": 6.177910348649368e-07,
+      "loss": 0.5434,
+      "step": 51893
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.5496115684509277,
+      "learning_rate": 6.176640874824013e-07,
+      "loss": 0.4507,
+      "step": 51894
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.864222526550293,
+      "learning_rate": 6.175371522855011e-07,
+      "loss": 0.5193,
+      "step": 51895
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.6745123863220215,
+      "learning_rate": 6.174102292745871e-07,
+      "loss": 0.5859,
+      "step": 51896
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.9322595596313477,
+      "learning_rate": 6.172833184500149e-07,
+      "loss": 0.6267,
+      "step": 51897
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.2508108615875244,
+      "learning_rate": 6.17156419812135e-07,
+      "loss": 0.5479,
+      "step": 51898
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.3822309970855713,
+      "learning_rate": 6.170295333613019e-07,
+      "loss": 0.5852,
+      "step": 51899
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.619706153869629,
+      "learning_rate": 6.169026590978671e-07,
+      "loss": 0.5612,
+      "step": 51900
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.0999350547790527,
+      "learning_rate": 6.167757970221849e-07,
+      "loss": 0.4519,
+      "step": 51901
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.3257007598876953,
+      "learning_rate": 6.166489471346065e-07,
+      "loss": 0.3581,
+      "step": 51902
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.5227274894714355,
+      "learning_rate": 6.16522109435485e-07,
+      "loss": 0.681,
+      "step": 51903
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.90702486038208,
+      "learning_rate": 6.163952839251747e-07,
+      "loss": 0.5071,
+      "step": 51904
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 5.546185493469238,
+      "learning_rate": 6.162684706040257e-07,
+      "loss": 0.7206,
+      "step": 51905
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.7503867149353027,
+      "learning_rate": 6.161416694723932e-07,
+      "loss": 0.5333,
+      "step": 51906
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.962782382965088,
+      "learning_rate": 6.160148805306271e-07,
+      "loss": 0.483,
+      "step": 51907
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.1215407848358154,
+      "learning_rate": 6.158881037790826e-07,
+      "loss": 0.6276,
+      "step": 51908
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.346498966217041,
+      "learning_rate": 6.1576133921811e-07,
+      "loss": 0.6403,
+      "step": 51909
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.946803092956543,
+      "learning_rate": 6.156345868480634e-07,
+      "loss": 0.5112,
+      "step": 51910
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.1786937713623047,
+      "learning_rate": 6.155078466692937e-07,
+      "loss": 0.6338,
+      "step": 51911
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.6747212409973145,
+      "learning_rate": 6.153811186821556e-07,
+      "loss": 0.4627,
+      "step": 51912
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.4025700092315674,
+      "learning_rate": 6.152544028869989e-07,
+      "loss": 0.391,
+      "step": 51913
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.4513142108917236,
+      "learning_rate": 6.151276992841781e-07,
+      "loss": 0.3745,
+      "step": 51914
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.048314332962036,
+      "learning_rate": 6.150010078740426e-07,
+      "loss": 0.4511,
+      "step": 51915
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.0823535919189453,
+      "learning_rate": 6.14874328656948e-07,
+      "loss": 0.4284,
+      "step": 51916
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.2427523136138916,
+      "learning_rate": 6.147476616332448e-07,
+      "loss": 0.4695,
+      "step": 51917
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.307553291320801,
+      "learning_rate": 6.146210068032871e-07,
+      "loss": 0.8278,
+      "step": 51918
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.265024423599243,
+      "learning_rate": 6.144943641674239e-07,
+      "loss": 0.4578,
+      "step": 51919
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 4.058133125305176,
+      "learning_rate": 6.143677337260101e-07,
+      "loss": 0.7437,
+      "step": 51920
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.5762546062469482,
+      "learning_rate": 6.142411154793959e-07,
+      "loss": 0.5247,
+      "step": 51921
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.5071980953216553,
+      "learning_rate": 6.141145094279349e-07,
+      "loss": 0.5707,
+      "step": 51922
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.199411392211914,
+      "learning_rate": 6.13987915571978e-07,
+      "loss": 0.6203,
+      "step": 51923
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 5.5058159828186035,
+      "learning_rate": 6.138613339118782e-07,
+      "loss": 0.5066,
+      "step": 51924
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.4169914722442627,
+      "learning_rate": 6.137347644479857e-07,
+      "loss": 0.7373,
+      "step": 51925
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.807410717010498,
+      "learning_rate": 6.136082071806554e-07,
+      "loss": 0.5553,
+      "step": 51926
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.3736541271209717,
+      "learning_rate": 6.134816621102363e-07,
+      "loss": 0.5739,
+      "step": 51927
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.323307514190674,
+      "learning_rate": 6.133551292370821e-07,
+      "loss": 0.5401,
+      "step": 51928
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.5115950107574463,
+      "learning_rate": 6.132286085615436e-07,
+      "loss": 0.6459,
+      "step": 51929
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.620906114578247,
+      "learning_rate": 6.131021000839732e-07,
+      "loss": 0.7273,
+      "step": 51930
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.709794521331787,
+      "learning_rate": 6.129756038047224e-07,
+      "loss": 0.5262,
+      "step": 51931
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.025538682937622,
+      "learning_rate": 6.128491197241431e-07,
+      "loss": 0.7307,
+      "step": 51932
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.4854564666748047,
+      "learning_rate": 6.127226478425868e-07,
+      "loss": 0.5809,
+      "step": 51933
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.5992562770843506,
+      "learning_rate": 6.125961881604059e-07,
+      "loss": 0.4713,
+      "step": 51934
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.5078470706939697,
+      "learning_rate": 6.124697406779495e-07,
+      "loss": 0.3649,
+      "step": 51935
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.1552391052246094,
+      "learning_rate": 6.12343305395573e-07,
+      "loss": 0.4985,
+      "step": 51936
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.176825761795044,
+      "learning_rate": 6.122168823136255e-07,
+      "loss": 0.7031,
+      "step": 51937
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.102538824081421,
+      "learning_rate": 6.120904714324593e-07,
+      "loss": 0.3826,
+      "step": 51938
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 4.251506805419922,
+      "learning_rate": 6.119640727524256e-07,
+      "loss": 0.7197,
+      "step": 51939
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.404165029525757,
+      "learning_rate": 6.118376862738767e-07,
+      "loss": 0.647,
+      "step": 51940
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.9024293422698975,
+      "learning_rate": 6.117113119971624e-07,
+      "loss": 0.658,
+      "step": 51941
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.124990940093994,
+      "learning_rate": 6.115849499226362e-07,
+      "loss": 0.4349,
+      "step": 51942
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.1825921535491943,
+      "learning_rate": 6.114586000506473e-07,
+      "loss": 0.4551,
+      "step": 51943
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.487386703491211,
+      "learning_rate": 6.113322623815488e-07,
+      "loss": 0.6047,
+      "step": 51944
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.607415199279785,
+      "learning_rate": 6.112059369156903e-07,
+      "loss": 0.5938,
+      "step": 51945
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 15.003854751586914,
+      "learning_rate": 6.11079623653425e-07,
+      "loss": 0.4677,
+      "step": 51946
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.1224963665008545,
+      "learning_rate": 6.10953322595102e-07,
+      "loss": 0.6645,
+      "step": 51947
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.583204746246338,
+      "learning_rate": 6.108270337410749e-07,
+      "loss": 0.5514,
+      "step": 51948
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.3082287311553955,
+      "learning_rate": 6.107007570916928e-07,
+      "loss": 0.6186,
+      "step": 51949
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.432910919189453,
+      "learning_rate": 6.105744926473084e-07,
+      "loss": 0.7789,
+      "step": 51950
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.131937265396118,
+      "learning_rate": 6.104482404082707e-07,
+      "loss": 0.4099,
+      "step": 51951
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.964254856109619,
+      "learning_rate": 6.103220003749328e-07,
+      "loss": 0.5902,
+      "step": 51952
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.6276838779449463,
+      "learning_rate": 6.101957725476454e-07,
+      "loss": 0.5488,
+      "step": 51953
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.4018478393554688,
+      "learning_rate": 6.100695569267584e-07,
+      "loss": 0.6445,
+      "step": 51954
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 4.282142639160156,
+      "learning_rate": 6.099433535126231e-07,
+      "loss": 0.5085,
+      "step": 51955
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.9573566913604736,
+      "learning_rate": 6.098171623055921e-07,
+      "loss": 0.516,
+      "step": 51956
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.7144932746887207,
+      "learning_rate": 6.096909833060138e-07,
+      "loss": 0.6187,
+      "step": 51957
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.9369001388549805,
+      "learning_rate": 6.095648165142414e-07,
+      "loss": 0.632,
+      "step": 51958
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.5987772941589355,
+      "learning_rate": 6.09438661930623e-07,
+      "loss": 0.6578,
+      "step": 51959
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.638624668121338,
+      "learning_rate": 6.093125195555127e-07,
+      "loss": 0.41,
+      "step": 51960
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.832815170288086,
+      "learning_rate": 6.09186389389258e-07,
+      "loss": 0.6882,
+      "step": 51961
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.0381312370300293,
+      "learning_rate": 6.090602714322124e-07,
+      "loss": 0.4714,
+      "step": 51962
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.404163122177124,
+      "learning_rate": 6.089341656847242e-07,
+      "loss": 0.5584,
+      "step": 51963
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.6866557598114014,
+      "learning_rate": 6.088080721471451e-07,
+      "loss": 0.6256,
+      "step": 51964
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.4744155406951904,
+      "learning_rate": 6.086819908198272e-07,
+      "loss": 0.5003,
+      "step": 51965
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.747832775115967,
+      "learning_rate": 6.085559217031179e-07,
+      "loss": 0.7205,
+      "step": 51966
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.637286901473999,
+      "learning_rate": 6.084298647973713e-07,
+      "loss": 0.6079,
+      "step": 51967
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.9026410579681396,
+      "learning_rate": 6.083038201029346e-07,
+      "loss": 0.5463,
+      "step": 51968
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.4554994106292725,
+      "learning_rate": 6.081777876201612e-07,
+      "loss": 0.5779,
+      "step": 51969
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.073779344558716,
+      "learning_rate": 6.080517673493985e-07,
+      "loss": 0.5388,
+      "step": 51970
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.882690191268921,
+      "learning_rate": 6.079257592909998e-07,
+      "loss": 0.5318,
+      "step": 51971
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.4950075149536133,
+      "learning_rate": 6.077997634453136e-07,
+      "loss": 0.6148,
+      "step": 51972
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.583104133605957,
+      "learning_rate": 6.076737798126919e-07,
+      "loss": 0.7767,
+      "step": 51973
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.155155658721924,
+      "learning_rate": 6.075478083934827e-07,
+      "loss": 0.5265,
+      "step": 51974
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.814387798309326,
+      "learning_rate": 6.07421849188039e-07,
+      "loss": 0.8801,
+      "step": 51975
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.3007705211639404,
+      "learning_rate": 6.072959021967073e-07,
+      "loss": 0.6658,
+      "step": 51976
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 4.45736026763916,
+      "learning_rate": 6.071699674198422e-07,
+      "loss": 0.5487,
+      "step": 51977
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.4456958770751953,
+      "learning_rate": 6.070440448577908e-07,
+      "loss": 0.7389,
+      "step": 51978
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.717623472213745,
+      "learning_rate": 6.069181345109054e-07,
+      "loss": 0.7233,
+      "step": 51979
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.134267807006836,
+      "learning_rate": 6.067922363795337e-07,
+      "loss": 0.5496,
+      "step": 51980
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.38006329536438,
+      "learning_rate": 6.06666350464028e-07,
+      "loss": 0.4201,
+      "step": 51981
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 5.4048004150390625,
+      "learning_rate": 6.06540476764737e-07,
+      "loss": 0.8036,
+      "step": 51982
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.9717867374420166,
+      "learning_rate": 6.064146152820116e-07,
+      "loss": 0.4683,
+      "step": 51983
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.628117322921753,
+      "learning_rate": 6.062887660162004e-07,
+      "loss": 0.7072,
+      "step": 51984
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.976724863052368,
+      "learning_rate": 6.061629289676557e-07,
+      "loss": 0.6956,
+      "step": 51985
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.9136834144592285,
+      "learning_rate": 6.060371041367242e-07,
+      "loss": 0.4277,
+      "step": 51986
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.844832420349121,
+      "learning_rate": 6.059112915237586e-07,
+      "loss": 0.5595,
+      "step": 51987
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.836519479751587,
+      "learning_rate": 6.057854911291073e-07,
+      "loss": 0.6603,
+      "step": 51988
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.9202880859375,
+      "learning_rate": 6.056597029531208e-07,
+      "loss": 0.7083,
+      "step": 51989
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.898625135421753,
+      "learning_rate": 6.055339269961475e-07,
+      "loss": 0.6524,
+      "step": 51990
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.565072536468506,
+      "learning_rate": 6.054081632585396e-07,
+      "loss": 0.6401,
+      "step": 51991
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.8409945964813232,
+      "learning_rate": 6.052824117406436e-07,
+      "loss": 0.5833,
+      "step": 51992
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.19712233543396,
+      "learning_rate": 6.051566724428126e-07,
+      "loss": 0.6646,
+      "step": 51993
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.8446221351623535,
+      "learning_rate": 6.050309453653935e-07,
+      "loss": 0.3999,
+      "step": 51994
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.3136990070343018,
+      "learning_rate": 6.049052305087378e-07,
+      "loss": 0.8982,
+      "step": 51995
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.665928602218628,
+      "learning_rate": 6.047795278731922e-07,
+      "loss": 0.7874,
+      "step": 51996
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.5206217765808105,
+      "learning_rate": 6.046538374591099e-07,
+      "loss": 0.4682,
+      "step": 51997
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.7881221771240234,
+      "learning_rate": 6.045281592668384e-07,
+      "loss": 0.5834,
+      "step": 51998
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 4.138326168060303,
+      "learning_rate": 6.04402493296728e-07,
+      "loss": 0.7413,
+      "step": 51999
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.9381611347198486,
+      "learning_rate": 6.042768395491266e-07,
+      "loss": 0.5694,
+      "step": 52000
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.761960744857788,
+      "learning_rate": 6.041511980243864e-07,
+      "loss": 0.5768,
+      "step": 52001
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.0249364376068115,
+      "learning_rate": 6.04025568722853e-07,
+      "loss": 0.5268,
+      "step": 52002
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.3732542991638184,
+      "learning_rate": 6.038999516448796e-07,
+      "loss": 0.5542,
+      "step": 52003
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 1.7948671579360962,
+      "learning_rate": 6.037743467908119e-07,
+      "loss": 0.2339,
+      "step": 52004
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 4.244732856750488,
+      "learning_rate": 6.036487541610025e-07,
+      "loss": 0.8172,
+      "step": 52005
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.5400636196136475,
+      "learning_rate": 6.035231737557972e-07,
+      "loss": 0.6863,
+      "step": 52006
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.593738079071045,
+      "learning_rate": 6.033976055755486e-07,
+      "loss": 0.5786,
+      "step": 52007
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.266788959503174,
+      "learning_rate": 6.032720496206035e-07,
+      "loss": 0.6096,
+      "step": 52008
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.096728801727295,
+      "learning_rate": 6.031465058913121e-07,
+      "loss": 0.8426,
+      "step": 52009
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.0182101726531982,
+      "learning_rate": 6.030209743880228e-07,
+      "loss": 0.2944,
+      "step": 52010
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.408703565597534,
+      "learning_rate": 6.028954551110855e-07,
+      "loss": 0.7396,
+      "step": 52011
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.745993137359619,
+      "learning_rate": 6.027699480608479e-07,
+      "loss": 0.5167,
+      "step": 52012
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.2816061973571777,
+      "learning_rate": 6.026444532376608e-07,
+      "loss": 0.6622,
+      "step": 52013
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.075624465942383,
+      "learning_rate": 6.025189706418716e-07,
+      "loss": 0.7902,
+      "step": 52014
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.164102077484131,
+      "learning_rate": 6.02393500273829e-07,
+      "loss": 0.3129,
+      "step": 52015
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.507768392562866,
+      "learning_rate": 6.022680421338833e-07,
+      "loss": 0.6134,
+      "step": 52016
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 5.03174352645874,
+      "learning_rate": 6.021425962223837e-07,
+      "loss": 0.642,
+      "step": 52017
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.2773942947387695,
+      "learning_rate": 6.02017162539677e-07,
+      "loss": 0.5877,
+      "step": 52018
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.8205487728118896,
+      "learning_rate": 6.018917410861141e-07,
+      "loss": 0.4555,
+      "step": 52019
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.2686855792999268,
+      "learning_rate": 6.017663318620415e-07,
+      "loss": 0.5328,
+      "step": 52020
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.804854154586792,
+      "learning_rate": 6.016409348678104e-07,
+      "loss": 0.6872,
+      "step": 52021
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.8947057723999023,
+      "learning_rate": 6.015155501037667e-07,
+      "loss": 0.8007,
+      "step": 52022
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.3214831352233887,
+      "learning_rate": 6.013901775702619e-07,
+      "loss": 0.809,
+      "step": 52023
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.6754140853881836,
+      "learning_rate": 6.012648172676422e-07,
+      "loss": 0.6916,
+      "step": 52024
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.018671989440918,
+      "learning_rate": 6.011394691962574e-07,
+      "loss": 0.5531,
+      "step": 52025
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.5946991443634033,
+      "learning_rate": 6.010141333564567e-07,
+      "loss": 0.5835,
+      "step": 52026
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 4.296830654144287,
+      "learning_rate": 6.008888097485871e-07,
+      "loss": 0.5766,
+      "step": 52027
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.90871262550354,
+      "learning_rate": 6.007634983729988e-07,
+      "loss": 0.5454,
+      "step": 52028
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.101308822631836,
+      "learning_rate": 6.006381992300375e-07,
+      "loss": 0.7534,
+      "step": 52029
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.195007562637329,
+      "learning_rate": 6.005129123200548e-07,
+      "loss": 0.5426,
+      "step": 52030
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 4.461780548095703,
+      "learning_rate": 6.003876376433964e-07,
+      "loss": 0.6657,
+      "step": 52031
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 5.092638969421387,
+      "learning_rate": 6.002623752004133e-07,
+      "loss": 0.6329,
+      "step": 52032
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.7475383281707764,
+      "learning_rate": 6.001371249914512e-07,
+      "loss": 0.691,
+      "step": 52033
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.196255683898926,
+      "learning_rate": 6.000118870168603e-07,
+      "loss": 0.5462,
+      "step": 52034
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 10.36369800567627,
+      "learning_rate": 5.998866612769872e-07,
+      "loss": 0.639,
+      "step": 52035
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.979135513305664,
+      "learning_rate": 5.997614477721819e-07,
+      "loss": 0.7765,
+      "step": 52036
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.2240140438079834,
+      "learning_rate": 5.9963624650279e-07,
+      "loss": 0.5814,
+      "step": 52037
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.907184600830078,
+      "learning_rate": 5.995110574691632e-07,
+      "loss": 0.6141,
+      "step": 52038
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.240410327911377,
+      "learning_rate": 5.993858806716463e-07,
+      "loss": 0.6309,
+      "step": 52039
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.1867330074310303,
+      "learning_rate": 5.992607161105901e-07,
+      "loss": 0.649,
+      "step": 52040
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 4.032916069030762,
+      "learning_rate": 5.9913556378634e-07,
+      "loss": 0.7401,
+      "step": 52041
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.757216453552246,
+      "learning_rate": 5.990104236992461e-07,
+      "loss": 0.6965,
+      "step": 52042
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.764948844909668,
+      "learning_rate": 5.98885295849655e-07,
+      "loss": 0.6572,
+      "step": 52043
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.539065361022949,
+      "learning_rate": 5.987601802379162e-07,
+      "loss": 0.378,
+      "step": 52044
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.8176627159118652,
+      "learning_rate": 5.986350768643756e-07,
+      "loss": 0.6617,
+      "step": 52045
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.5700125694274902,
+      "learning_rate": 5.98509985729383e-07,
+      "loss": 0.5502,
+      "step": 52046
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.571511745452881,
+      "learning_rate": 5.983849068332842e-07,
+      "loss": 0.5878,
+      "step": 52047
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.27095365524292,
+      "learning_rate": 5.982598401764289e-07,
+      "loss": 0.6458,
+      "step": 52048
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.709028959274292,
+      "learning_rate": 5.981347857591635e-07,
+      "loss": 0.506,
+      "step": 52049
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.8245136737823486,
+      "learning_rate": 5.980097435818372e-07,
+      "loss": 0.5627,
+      "step": 52050
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.3185174465179443,
+      "learning_rate": 5.978847136447957e-07,
+      "loss": 0.6089,
+      "step": 52051
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.467773675918579,
+      "learning_rate": 5.977596959483883e-07,
+      "loss": 0.4835,
+      "step": 52052
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.806793451309204,
+      "learning_rate": 5.97634690492962e-07,
+      "loss": 0.62,
+      "step": 52053
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.4567670822143555,
+      "learning_rate": 5.975096972788647e-07,
+      "loss": 0.4815,
+      "step": 52054
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.544842481613159,
+      "learning_rate": 5.973847163064428e-07,
+      "loss": 0.6907,
+      "step": 52055
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.6603894233703613,
+      "learning_rate": 5.972597475760461e-07,
+      "loss": 0.6365,
+      "step": 52056
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.3505046367645264,
+      "learning_rate": 5.971347910880182e-07,
+      "loss": 0.3939,
+      "step": 52057
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.1185238361358643,
+      "learning_rate": 5.970098468427116e-07,
+      "loss": 0.5725,
+      "step": 52058
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.5304131507873535,
+      "learning_rate": 5.968849148404699e-07,
+      "loss": 0.6109,
+      "step": 52059
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.6555423736572266,
+      "learning_rate": 5.967599950816433e-07,
+      "loss": 0.4114,
+      "step": 52060
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.329556465148926,
+      "learning_rate": 5.96635087566576e-07,
+      "loss": 0.3879,
+      "step": 52061
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.9753847122192383,
+      "learning_rate": 5.965101922956185e-07,
+      "loss": 0.4777,
+      "step": 52062
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.5390963554382324,
+      "learning_rate": 5.963853092691152e-07,
+      "loss": 0.4223,
+      "step": 52063
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.9258203506469727,
+      "learning_rate": 5.96260438487416e-07,
+      "loss": 0.4861,
+      "step": 52064
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 4.778768539428711,
+      "learning_rate": 5.961355799508655e-07,
+      "loss": 0.6705,
+      "step": 52065
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.8637404441833496,
+      "learning_rate": 5.960107336598136e-07,
+      "loss": 0.606,
+      "step": 52066
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.757690906524658,
+      "learning_rate": 5.95885899614605e-07,
+      "loss": 0.642,
+      "step": 52067
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.591085910797119,
+      "learning_rate": 5.957610778155892e-07,
+      "loss": 0.5492,
+      "step": 52068
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.5150763988494873,
+      "learning_rate": 5.956362682631106e-07,
+      "loss": 0.6074,
+      "step": 52069
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 4.440376281738281,
+      "learning_rate": 5.955114709575183e-07,
+      "loss": 0.6268,
+      "step": 52070
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 5.065555572509766,
+      "learning_rate": 5.95386685899158e-07,
+      "loss": 0.735,
+      "step": 52071
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.1313228607177734,
+      "learning_rate": 5.952619130883785e-07,
+      "loss": 0.4611,
+      "step": 52072
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.085272789001465,
+      "learning_rate": 5.951371525255246e-07,
+      "loss": 0.6961,
+      "step": 52073
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.1091575622558594,
+      "learning_rate": 5.950124042109446e-07,
+      "loss": 0.5415,
+      "step": 52074
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.2920145988464355,
+      "learning_rate": 5.948876681449845e-07,
+      "loss": 0.6078,
+      "step": 52075
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.4689886569976807,
+      "learning_rate": 5.947629443279912e-07,
+      "loss": 0.7303,
+      "step": 52076
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.5120952129364014,
+      "learning_rate": 5.946382327603123e-07,
+      "loss": 0.5527,
+      "step": 52077
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 4.9972453117370605,
+      "learning_rate": 5.945135334422947e-07,
+      "loss": 0.6363,
+      "step": 52078
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.6018457412719727,
+      "learning_rate": 5.943888463742841e-07,
+      "loss": 0.5077,
+      "step": 52079
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.7120189666748047,
+      "learning_rate": 5.942641715566289e-07,
+      "loss": 0.6443,
+      "step": 52080
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 5.051520347595215,
+      "learning_rate": 5.941395089896729e-07,
+      "loss": 0.6347,
+      "step": 52081
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.6602940559387207,
+      "learning_rate": 5.94014858673766e-07,
+      "loss": 0.5243,
+      "step": 52082
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.461792230606079,
+      "learning_rate": 5.938902206092523e-07,
+      "loss": 0.5389,
+      "step": 52083
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 4.911678791046143,
+      "learning_rate": 5.937655947964799e-07,
+      "loss": 0.461,
+      "step": 52084
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.180777072906494,
+      "learning_rate": 5.936409812357941e-07,
+      "loss": 0.4452,
+      "step": 52085
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.3393607139587402,
+      "learning_rate": 5.935163799275428e-07,
+      "loss": 0.4244,
+      "step": 52086
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.193859815597534,
+      "learning_rate": 5.93391790872071e-07,
+      "loss": 0.5412,
+      "step": 52087
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.7915875911712646,
+      "learning_rate": 5.93267214069726e-07,
+      "loss": 0.5037,
+      "step": 52088
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.447990655899048,
+      "learning_rate": 5.931426495208548e-07,
+      "loss": 0.833,
+      "step": 52089
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.1544525623321533,
+      "learning_rate": 5.93018097225802e-07,
+      "loss": 0.6482,
+      "step": 52090
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 8.658807754516602,
+      "learning_rate": 5.92893557184916e-07,
+      "loss": 0.6151,
+      "step": 52091
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.2511696815490723,
+      "learning_rate": 5.927690293985411e-07,
+      "loss": 0.6216,
+      "step": 52092
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.940244674682617,
+      "learning_rate": 5.926445138670256e-07,
+      "loss": 0.6698,
+      "step": 52093
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 5.880941867828369,
+      "learning_rate": 5.925200105907136e-07,
+      "loss": 0.7049,
+      "step": 52094
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.8690454959869385,
+      "learning_rate": 5.923955195699537e-07,
+      "loss": 0.8559,
+      "step": 52095
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.1101837158203125,
+      "learning_rate": 5.922710408050897e-07,
+      "loss": 0.5455,
+      "step": 52096
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.7933473587036133,
+      "learning_rate": 5.921465742964693e-07,
+      "loss": 0.3992,
+      "step": 52097
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 4.013016223907471,
+      "learning_rate": 5.920221200444365e-07,
+      "loss": 0.8024,
+      "step": 52098
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.0382497310638428,
+      "learning_rate": 5.918976780493408e-07,
+      "loss": 0.4932,
+      "step": 52099
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.2061848640441895,
+      "learning_rate": 5.917732483115252e-07,
+      "loss": 0.5155,
+      "step": 52100
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.5058515071868896,
+      "learning_rate": 5.916488308313378e-07,
+      "loss": 0.7185,
+      "step": 52101
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 4.094321250915527,
+      "learning_rate": 5.915244256091224e-07,
+      "loss": 0.6177,
+      "step": 52102
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.9527478218078613,
+      "learning_rate": 5.914000326452274e-07,
+      "loss": 0.549,
+      "step": 52103
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.15057110786438,
+      "learning_rate": 5.912756519399964e-07,
+      "loss": 0.5625,
+      "step": 52104
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.9523465633392334,
+      "learning_rate": 5.911512834937777e-07,
+      "loss": 0.789,
+      "step": 52105
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.721774101257324,
+      "learning_rate": 5.910269273069141e-07,
+      "loss": 0.5536,
+      "step": 52106
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.0187997817993164,
+      "learning_rate": 5.909025833797544e-07,
+      "loss": 0.475,
+      "step": 52107
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.8214235305786133,
+      "learning_rate": 5.907782517126414e-07,
+      "loss": 0.6423,
+      "step": 52108
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.1842315196990967,
+      "learning_rate": 5.90653932305924e-07,
+      "loss": 0.5188,
+      "step": 52109
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.8818938732147217,
+      "learning_rate": 5.905296251599446e-07,
+      "loss": 0.8093,
+      "step": 52110
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.794323444366455,
+      "learning_rate": 5.904053302750518e-07,
+      "loss": 0.5291,
+      "step": 52111
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.858736753463745,
+      "learning_rate": 5.902810476515891e-07,
+      "loss": 0.6201,
+      "step": 52112
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.595031976699829,
+      "learning_rate": 5.901567772899036e-07,
+      "loss": 0.6756,
+      "step": 52113
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.3252811431884766,
+      "learning_rate": 5.900325191903389e-07,
+      "loss": 0.4488,
+      "step": 52114
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.401951789855957,
+      "learning_rate": 5.899082733532429e-07,
+      "loss": 0.5908,
+      "step": 52115
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.198474884033203,
+      "learning_rate": 5.897840397789595e-07,
+      "loss": 0.4953,
+      "step": 52116
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.2026102542877197,
+      "learning_rate": 5.896598184678348e-07,
+      "loss": 0.6374,
+      "step": 52117
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.0174143314361572,
+      "learning_rate": 5.895356094202125e-07,
+      "loss": 0.3847,
+      "step": 52118
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.6507716178894043,
+      "learning_rate": 5.894114126364414e-07,
+      "loss": 0.5459,
+      "step": 52119
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.839352607727051,
+      "learning_rate": 5.892872281168638e-07,
+      "loss": 0.2799,
+      "step": 52120
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.5817854404449463,
+      "learning_rate": 5.891630558618272e-07,
+      "loss": 0.4801,
+      "step": 52121
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.290815591812134,
+      "learning_rate": 5.890388958716748e-07,
+      "loss": 0.5375,
+      "step": 52122
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.5524442195892334,
+      "learning_rate": 5.889147481467539e-07,
+      "loss": 0.6889,
+      "step": 52123
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.700932264328003,
+      "learning_rate": 5.887906126874077e-07,
+      "loss": 0.5422,
+      "step": 52124
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.4972727298736572,
+      "learning_rate": 5.886664894939831e-07,
+      "loss": 0.6897,
+      "step": 52125
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 1.7817270755767822,
+      "learning_rate": 5.885423785668237e-07,
+      "loss": 0.3794,
+      "step": 52126
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.2834742069244385,
+      "learning_rate": 5.88418279906276e-07,
+      "loss": 0.5536,
+      "step": 52127
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.944714307785034,
+      "learning_rate": 5.88294193512684e-07,
+      "loss": 0.5921,
+      "step": 52128
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.730893135070801,
+      "learning_rate": 5.881701193863936e-07,
+      "loss": 0.4651,
+      "step": 52129
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.2769508361816406,
+      "learning_rate": 5.880460575277486e-07,
+      "loss": 0.7122,
+      "step": 52130
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.4907948970794678,
+      "learning_rate": 5.879220079370962e-07,
+      "loss": 0.5525,
+      "step": 52131
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.9802162647247314,
+      "learning_rate": 5.877979706147785e-07,
+      "loss": 0.8929,
+      "step": 52132
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.1783971786499023,
+      "learning_rate": 5.876739455611424e-07,
+      "loss": 0.592,
+      "step": 52133
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 4.012660503387451,
+      "learning_rate": 5.875499327765316e-07,
+      "loss": 0.6472,
+      "step": 52134
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.0424630641937256,
+      "learning_rate": 5.87425932261293e-07,
+      "loss": 0.459,
+      "step": 52135
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.3712270259857178,
+      "learning_rate": 5.873019440157679e-07,
+      "loss": 0.8076,
+      "step": 52136
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.779099225997925,
+      "learning_rate": 5.871779680403039e-07,
+      "loss": 0.4485,
+      "step": 52137
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 4.207759857177734,
+      "learning_rate": 5.870540043352441e-07,
+      "loss": 0.5507,
+      "step": 52138
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 1.8604795932769775,
+      "learning_rate": 5.869300529009359e-07,
+      "loss": 0.4743,
+      "step": 52139
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.1896088123321533,
+      "learning_rate": 5.868061137377201e-07,
+      "loss": 0.617,
+      "step": 52140
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 4.365310192108154,
+      "learning_rate": 5.866821868459449e-07,
+      "loss": 0.654,
+      "step": 52141
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.2649343013763428,
+      "learning_rate": 5.865582722259521e-07,
+      "loss": 0.5907,
+      "step": 52142
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.026242256164551,
+      "learning_rate": 5.864343698780883e-07,
+      "loss": 0.7039,
+      "step": 52143
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.566856622695923,
+      "learning_rate": 5.863104798026964e-07,
+      "loss": 0.7454,
+      "step": 52144
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.2150862216949463,
+      "learning_rate": 5.861866020001228e-07,
+      "loss": 0.3867,
+      "step": 52145
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.224269151687622,
+      "learning_rate": 5.860627364707095e-07,
+      "loss": 0.5424,
+      "step": 52146
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.907412528991699,
+      "learning_rate": 5.859388832148033e-07,
+      "loss": 0.5539,
+      "step": 52147
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.000377893447876,
+      "learning_rate": 5.858150422327463e-07,
+      "loss": 0.6098,
+      "step": 52148
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.03645658493042,
+      "learning_rate": 5.856912135248844e-07,
+      "loss": 0.6296,
+      "step": 52149
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.688974618911743,
+      "learning_rate": 5.855673970915626e-07,
+      "loss": 0.7678,
+      "step": 52150
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.126171350479126,
+      "learning_rate": 5.854435929331226e-07,
+      "loss": 0.3618,
+      "step": 52151
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.7978227138519287,
+      "learning_rate": 5.853198010499117e-07,
+      "loss": 0.7891,
+      "step": 52152
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.865562677383423,
+      "learning_rate": 5.851960214422719e-07,
+      "loss": 0.5604,
+      "step": 52153
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.510462522506714,
+      "learning_rate": 5.850722541105491e-07,
+      "loss": 0.4161,
+      "step": 52154
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.647136688232422,
+      "learning_rate": 5.849484990550853e-07,
+      "loss": 0.4073,
+      "step": 52155
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 4.026003837585449,
+      "learning_rate": 5.848247562762266e-07,
+      "loss": 0.7564,
+      "step": 52156
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 4.613239765167236,
+      "learning_rate": 5.847010257743158e-07,
+      "loss": 0.5479,
+      "step": 52157
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.1061627864837646,
+      "learning_rate": 5.84577307549698e-07,
+      "loss": 0.5845,
+      "step": 52158
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 1.9738991260528564,
+      "learning_rate": 5.844536016027158e-07,
+      "loss": 0.4906,
+      "step": 52159
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.4448275566101074,
+      "learning_rate": 5.843299079337139e-07,
+      "loss": 0.7626,
+      "step": 52160
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.201937437057495,
+      "learning_rate": 5.842062265430365e-07,
+      "loss": 0.5817,
+      "step": 52161
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.56827974319458,
+      "learning_rate": 5.840825574310283e-07,
+      "loss": 0.624,
+      "step": 52162
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.1395280361175537,
+      "learning_rate": 5.839589005980311e-07,
+      "loss": 0.4933,
+      "step": 52163
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.8063392639160156,
+      "learning_rate": 5.838352560443916e-07,
+      "loss": 0.5657,
+      "step": 52164
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.2111408710479736,
+      "learning_rate": 5.837116237704504e-07,
+      "loss": 0.6822,
+      "step": 52165
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.6351237297058105,
+      "learning_rate": 5.835880037765535e-07,
+      "loss": 0.4732,
+      "step": 52166
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.521395444869995,
+      "learning_rate": 5.834643960630437e-07,
+      "loss": 0.3686,
+      "step": 52167
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.391909599304199,
+      "learning_rate": 5.833408006302654e-07,
+      "loss": 0.6034,
+      "step": 52168
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.5314159393310547,
+      "learning_rate": 5.832172174785611e-07,
+      "loss": 0.4209,
+      "step": 52169
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.3145689964294434,
+      "learning_rate": 5.83093646608276e-07,
+      "loss": 0.5552,
+      "step": 52170
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.107379674911499,
+      "learning_rate": 5.82970088019752e-07,
+      "loss": 0.3355,
+      "step": 52171
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.2797114849090576,
+      "learning_rate": 5.828465417133344e-07,
+      "loss": 0.4184,
+      "step": 52172
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.18841814994812,
+      "learning_rate": 5.827230076893647e-07,
+      "loss": 0.4592,
+      "step": 52173
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.2390427589416504,
+      "learning_rate": 5.825994859481887e-07,
+      "loss": 0.8141,
+      "step": 52174
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 13.233050346374512,
+      "learning_rate": 5.824759764901477e-07,
+      "loss": 0.7301,
+      "step": 52175
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.3614635467529297,
+      "learning_rate": 5.823524793155871e-07,
+      "loss": 0.7876,
+      "step": 52176
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.3048276901245117,
+      "learning_rate": 5.822289944248488e-07,
+      "loss": 0.7739,
+      "step": 52177
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.5520427227020264,
+      "learning_rate": 5.821055218182775e-07,
+      "loss": 0.5499,
+      "step": 52178
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.0540828704833984,
+      "learning_rate": 5.819820614962135e-07,
+      "loss": 0.3601,
+      "step": 52179
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 4.312763690948486,
+      "learning_rate": 5.818586134590043e-07,
+      "loss": 0.6916,
+      "step": 52180
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.7298507690429688,
+      "learning_rate": 5.817351777069908e-07,
+      "loss": 0.5482,
+      "step": 52181
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.672741651535034,
+      "learning_rate": 5.81611754240517e-07,
+      "loss": 0.5355,
+      "step": 52182
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.638164520263672,
+      "learning_rate": 5.814883430599255e-07,
+      "loss": 0.8068,
+      "step": 52183
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.778937339782715,
+      "learning_rate": 5.8136494416556e-07,
+      "loss": 0.6777,
+      "step": 52184
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.3532357215881348,
+      "learning_rate": 5.81241557557763e-07,
+      "loss": 0.3772,
+      "step": 52185
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.927809238433838,
+      "learning_rate": 5.811181832368784e-07,
+      "loss": 0.5128,
+      "step": 52186
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.288763999938965,
+      "learning_rate": 5.809948212032479e-07,
+      "loss": 0.7754,
+      "step": 52187
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.298438787460327,
+      "learning_rate": 5.808714714572161e-07,
+      "loss": 0.5613,
+      "step": 52188
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 9.085988998413086,
+      "learning_rate": 5.80748133999125e-07,
+      "loss": 0.3877,
+      "step": 52189
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.5956807136535645,
+      "learning_rate": 5.806248088293182e-07,
+      "loss": 0.4125,
+      "step": 52190
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.775050401687622,
+      "learning_rate": 5.805014959481376e-07,
+      "loss": 0.5017,
+      "step": 52191
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.740187168121338,
+      "learning_rate": 5.803781953559279e-07,
+      "loss": 0.7128,
+      "step": 52192
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.404932737350464,
+      "learning_rate": 5.802549070530294e-07,
+      "loss": 0.5548,
+      "step": 52193
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.636368989944458,
+      "learning_rate": 5.801316310397875e-07,
+      "loss": 0.4645,
+      "step": 52194
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.9074759483337402,
+      "learning_rate": 5.80008367316543e-07,
+      "loss": 0.3537,
+      "step": 52195
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.5549614429473877,
+      "learning_rate": 5.7988511588364e-07,
+      "loss": 0.4803,
+      "step": 52196
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 4.292559623718262,
+      "learning_rate": 5.797618767414204e-07,
+      "loss": 0.6777,
+      "step": 52197
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.302263021469116,
+      "learning_rate": 5.796386498902273e-07,
+      "loss": 0.5972,
+      "step": 52198
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.689711809158325,
+      "learning_rate": 5.795154353304016e-07,
+      "loss": 0.5179,
+      "step": 52199
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.399834156036377,
+      "learning_rate": 5.793922330622897e-07,
+      "loss": 0.4933,
+      "step": 52200
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.370122194290161,
+      "learning_rate": 5.792690430862308e-07,
+      "loss": 0.4436,
+      "step": 52201
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 4.280942440032959,
+      "learning_rate": 5.7914586540257e-07,
+      "loss": 0.648,
+      "step": 52202
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.1620965003967285,
+      "learning_rate": 5.790227000116472e-07,
+      "loss": 0.5473,
+      "step": 52203
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 6.349150657653809,
+      "learning_rate": 5.78899546913807e-07,
+      "loss": 0.6357,
+      "step": 52204
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.210390567779541,
+      "learning_rate": 5.787764061093903e-07,
+      "loss": 0.5015,
+      "step": 52205
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.094874143600464,
+      "learning_rate": 5.786532775987408e-07,
+      "loss": 0.6036,
+      "step": 52206
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 6.513321399688721,
+      "learning_rate": 5.785301613821998e-07,
+      "loss": 0.5124,
+      "step": 52207
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.7596521377563477,
+      "learning_rate": 5.784070574601103e-07,
+      "loss": 0.5617,
+      "step": 52208
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 1.8267450332641602,
+      "learning_rate": 5.782839658328143e-07,
+      "loss": 0.3886,
+      "step": 52209
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.734335422515869,
+      "learning_rate": 5.781608865006532e-07,
+      "loss": 0.516,
+      "step": 52210
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.6371843814849854,
+      "learning_rate": 5.780378194639718e-07,
+      "loss": 0.6471,
+      "step": 52211
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 5.260986804962158,
+      "learning_rate": 5.779147647231098e-07,
+      "loss": 0.6951,
+      "step": 52212
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 1.8214194774627686,
+      "learning_rate": 5.777917222784113e-07,
+      "loss": 0.419,
+      "step": 52213
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 4.438112258911133,
+      "learning_rate": 5.77668692130216e-07,
+      "loss": 0.6257,
+      "step": 52214
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.2130062580108643,
+      "learning_rate": 5.775456742788687e-07,
+      "loss": 0.628,
+      "step": 52215
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.0488715171813965,
+      "learning_rate": 5.774226687247098e-07,
+      "loss": 0.6308,
+      "step": 52216
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.272214412689209,
+      "learning_rate": 5.772996754680821e-07,
+      "loss": 0.736,
+      "step": 52217
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.305882453918457,
+      "learning_rate": 5.77176694509326e-07,
+      "loss": 0.5004,
+      "step": 52218
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 1.768061876296997,
+      "learning_rate": 5.770537258487857e-07,
+      "loss": 0.439,
+      "step": 52219
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.8579912185668945,
+      "learning_rate": 5.769307694868015e-07,
+      "loss": 0.8357,
+      "step": 52220
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.64914608001709,
+      "learning_rate": 5.768078254237158e-07,
+      "loss": 0.7062,
+      "step": 52221
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.2908174991607666,
+      "learning_rate": 5.766848936598707e-07,
+      "loss": 0.4104,
+      "step": 52222
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.2981934547424316,
+      "learning_rate": 5.765619741956085e-07,
+      "loss": 0.445,
+      "step": 52223
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.4507737159729004,
+      "learning_rate": 5.764390670312698e-07,
+      "loss": 0.5822,
+      "step": 52224
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.3614695072174072,
+      "learning_rate": 5.763161721671978e-07,
+      "loss": 0.731,
+      "step": 52225
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.7441229820251465,
+      "learning_rate": 5.76193289603732e-07,
+      "loss": 0.6076,
+      "step": 52226
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.2596280574798584,
+      "learning_rate": 5.76070419341217e-07,
+      "loss": 0.6455,
+      "step": 52227
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.152843952178955,
+      "learning_rate": 5.75947561379992e-07,
+      "loss": 0.4977,
+      "step": 52228
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.379281520843506,
+      "learning_rate": 5.758247157204e-07,
+      "loss": 0.9051,
+      "step": 52229
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 5.48421573638916,
+      "learning_rate": 5.757018823627814e-07,
+      "loss": 0.6169,
+      "step": 52230
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 1.962486982345581,
+      "learning_rate": 5.755790613074796e-07,
+      "loss": 0.5064,
+      "step": 52231
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.940830945968628,
+      "learning_rate": 5.754562525548335e-07,
+      "loss": 0.7097,
+      "step": 52232
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.4857659339904785,
+      "learning_rate": 5.753334561051877e-07,
+      "loss": 0.5941,
+      "step": 52233
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.247373342514038,
+      "learning_rate": 5.752106719588807e-07,
+      "loss": 0.4754,
+      "step": 52234
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.5801639556884766,
+      "learning_rate": 5.750879001162568e-07,
+      "loss": 0.5793,
+      "step": 52235
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.093632459640503,
+      "learning_rate": 5.749651405776541e-07,
+      "loss": 0.693,
+      "step": 52236
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 4.771636486053467,
+      "learning_rate": 5.748423933434171e-07,
+      "loss": 0.5553,
+      "step": 52237
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 5.588908672332764,
+      "learning_rate": 5.747196584138848e-07,
+      "loss": 0.7154,
+      "step": 52238
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.1755826473236084,
+      "learning_rate": 5.745969357894e-07,
+      "loss": 0.81,
+      "step": 52239
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.6075754165649414,
+      "learning_rate": 5.744742254703018e-07,
+      "loss": 0.6283,
+      "step": 52240
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.750094413757324,
+      "learning_rate": 5.743515274569344e-07,
+      "loss": 0.6017,
+      "step": 52241
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.052428960800171,
+      "learning_rate": 5.742288417496367e-07,
+      "loss": 0.5572,
+      "step": 52242
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.986845016479492,
+      "learning_rate": 5.74106168348752e-07,
+      "loss": 0.5629,
+      "step": 52243
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 4.1618242263793945,
+      "learning_rate": 5.739835072546184e-07,
+      "loss": 0.5689,
+      "step": 52244
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.4371535778045654,
+      "learning_rate": 5.738608584675803e-07,
+      "loss": 0.5977,
+      "step": 52245
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 4.123232364654541,
+      "learning_rate": 5.737382219879756e-07,
+      "loss": 0.507,
+      "step": 52246
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.5631070137023926,
+      "learning_rate": 5.736155978161484e-07,
+      "loss": 0.4488,
+      "step": 52247
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.2555203437805176,
+      "learning_rate": 5.734929859524363e-07,
+      "loss": 0.5954,
+      "step": 52248
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.4467825889587402,
+      "learning_rate": 5.733703863971834e-07,
+      "loss": 0.7483,
+      "step": 52249
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.678023338317871,
+      "learning_rate": 5.732477991507285e-07,
+      "loss": 0.7326,
+      "step": 52250
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.0116984844207764,
+      "learning_rate": 5.731252242134139e-07,
+      "loss": 0.7096,
+      "step": 52251
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.3278825283050537,
+      "learning_rate": 5.730026615855783e-07,
+      "loss": 0.6323,
+      "step": 52252
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.3126602172851562,
+      "learning_rate": 5.728801112675652e-07,
+      "loss": 0.4742,
+      "step": 52253
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.863132953643799,
+      "learning_rate": 5.727575732597135e-07,
+      "loss": 0.4979,
+      "step": 52254
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.9378068447113037,
+      "learning_rate": 5.726350475623649e-07,
+      "loss": 0.5703,
+      "step": 52255
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.991159677505493,
+      "learning_rate": 5.725125341758591e-07,
+      "loss": 0.5106,
+      "step": 52256
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.6515402793884277,
+      "learning_rate": 5.723900331005383e-07,
+      "loss": 0.7559,
+      "step": 52257
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.6614792346954346,
+      "learning_rate": 5.722675443367409e-07,
+      "loss": 0.4751,
+      "step": 52258
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.045153856277466,
+      "learning_rate": 5.721450678848101e-07,
+      "loss": 0.3128,
+      "step": 52259
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.4080913066864014,
+      "learning_rate": 5.720226037450832e-07,
+      "loss": 0.515,
+      "step": 52260
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.8009543418884277,
+      "learning_rate": 5.719001519179046e-07,
+      "loss": 0.6211,
+      "step": 52261
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.828462839126587,
+      "learning_rate": 5.717777124036116e-07,
+      "loss": 0.5686,
+      "step": 52262
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.009199857711792,
+      "learning_rate": 5.716552852025469e-07,
+      "loss": 0.3504,
+      "step": 52263
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.148592233657837,
+      "learning_rate": 5.715328703150496e-07,
+      "loss": 0.7333,
+      "step": 52264
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 5.666222095489502,
+      "learning_rate": 5.714104677414611e-07,
+      "loss": 0.7739,
+      "step": 52265
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 5.043419361114502,
+      "learning_rate": 5.7128807748212e-07,
+      "loss": 0.6097,
+      "step": 52266
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.6705522537231445,
+      "learning_rate": 5.711656995373687e-07,
+      "loss": 0.5964,
+      "step": 52267
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.0518076419830322,
+      "learning_rate": 5.710433339075455e-07,
+      "loss": 0.5887,
+      "step": 52268
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 6.282365322113037,
+      "learning_rate": 5.709209805929927e-07,
+      "loss": 0.6008,
+      "step": 52269
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.373924493789673,
+      "learning_rate": 5.707986395940485e-07,
+      "loss": 0.5187,
+      "step": 52270
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.079720973968506,
+      "learning_rate": 5.706763109110552e-07,
+      "loss": 0.7388,
+      "step": 52271
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.3505451679229736,
+      "learning_rate": 5.705539945443506e-07,
+      "loss": 0.55,
+      "step": 52272
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.570117712020874,
+      "learning_rate": 5.704316904942764e-07,
+      "loss": 0.5423,
+      "step": 52273
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.2614684104919434,
+      "learning_rate": 5.70309398761173e-07,
+      "loss": 0.4099,
+      "step": 52274
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.8710074424743652,
+      "learning_rate": 5.70187119345379e-07,
+      "loss": 0.6673,
+      "step": 52275
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.852369785308838,
+      "learning_rate": 5.700648522472357e-07,
+      "loss": 0.2309,
+      "step": 52276
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.7865192890167236,
+      "learning_rate": 5.699425974670819e-07,
+      "loss": 0.6012,
+      "step": 52277
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.3959076404571533,
+      "learning_rate": 5.698203550052589e-07,
+      "loss": 0.5133,
+      "step": 52278
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 4.056972026824951,
+      "learning_rate": 5.696981248621053e-07,
+      "loss": 0.8812,
+      "step": 52279
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.9872665405273438,
+      "learning_rate": 5.695759070379625e-07,
+      "loss": 0.6001,
+      "step": 52280
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.4764163494110107,
+      "learning_rate": 5.694537015331686e-07,
+      "loss": 0.5799,
+      "step": 52281
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.6764004230499268,
+      "learning_rate": 5.693315083480639e-07,
+      "loss": 0.4887,
+      "step": 52282
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 4.4676008224487305,
+      "learning_rate": 5.692093274829897e-07,
+      "loss": 0.3983,
+      "step": 52283
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 4.556665420532227,
+      "learning_rate": 5.690871589382835e-07,
+      "loss": 0.7468,
+      "step": 52284
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.7435383796691895,
+      "learning_rate": 5.689650027142857e-07,
+      "loss": 0.677,
+      "step": 52285
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.974135398864746,
+      "learning_rate": 5.688428588113376e-07,
+      "loss": 0.5303,
+      "step": 52286
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.420293092727661,
+      "learning_rate": 5.687207272297768e-07,
+      "loss": 0.4494,
+      "step": 52287
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.2949271202087402,
+      "learning_rate": 5.685986079699446e-07,
+      "loss": 0.5166,
+      "step": 52288
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.29793119430542,
+      "learning_rate": 5.684765010321786e-07,
+      "loss": 0.5276,
+      "step": 52289
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.0951685905456543,
+      "learning_rate": 5.683544064168201e-07,
+      "loss": 0.564,
+      "step": 52290
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 4.518923282623291,
+      "learning_rate": 5.682323241242072e-07,
+      "loss": 0.7145,
+      "step": 52291
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.7143595218658447,
+      "learning_rate": 5.681102541546806e-07,
+      "loss": 0.9248,
+      "step": 52292
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.9902098178863525,
+      "learning_rate": 5.679881965085781e-07,
+      "loss": 0.578,
+      "step": 52293
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.292884588241577,
+      "learning_rate": 5.678661511862415e-07,
+      "loss": 0.5297,
+      "step": 52294
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.3638856410980225,
+      "learning_rate": 5.677441181880078e-07,
+      "loss": 0.8016,
+      "step": 52295
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.088944911956787,
+      "learning_rate": 5.676220975142189e-07,
+      "loss": 0.5428,
+      "step": 52296
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.4300248622894287,
+      "learning_rate": 5.675000891652105e-07,
+      "loss": 0.4683,
+      "step": 52297
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.545318126678467,
+      "learning_rate": 5.673780931413253e-07,
+      "loss": 0.7159,
+      "step": 52298
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.717677593231201,
+      "learning_rate": 5.672561094429002e-07,
+      "loss": 0.5629,
+      "step": 52299
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.3802731037139893,
+      "learning_rate": 5.671341380702766e-07,
+      "loss": 0.5483,
+      "step": 52300
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.2416293621063232,
+      "learning_rate": 5.670121790237898e-07,
+      "loss": 0.5903,
+      "step": 52301
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.9394419193267822,
+      "learning_rate": 5.668902323037839e-07,
+      "loss": 0.6877,
+      "step": 52302
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.9055356979370117,
+      "learning_rate": 5.667682979105943e-07,
+      "loss": 0.615,
+      "step": 52303
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.1489617824554443,
+      "learning_rate": 5.666463758445629e-07,
+      "loss": 0.6879,
+      "step": 52304
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.858624219894409,
+      "learning_rate": 5.665244661060254e-07,
+      "loss": 0.3595,
+      "step": 52305
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.5788004398345947,
+      "learning_rate": 5.664025686953239e-07,
+      "loss": 0.6083,
+      "step": 52306
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.2252864837646484,
+      "learning_rate": 5.662806836127954e-07,
+      "loss": 0.632,
+      "step": 52307
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.4062962532043457,
+      "learning_rate": 5.661588108587796e-07,
+      "loss": 0.5052,
+      "step": 52308
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.019110918045044,
+      "learning_rate": 5.66036950433615e-07,
+      "loss": 0.716,
+      "step": 52309
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.5069973468780518,
+      "learning_rate": 5.659151023376408e-07,
+      "loss": 0.8368,
+      "step": 52310
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.8173937797546387,
+      "learning_rate": 5.657932665711952e-07,
+      "loss": 0.5889,
+      "step": 52311
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.9362053871154785,
+      "learning_rate": 5.656714431346183e-07,
+      "loss": 0.4863,
+      "step": 52312
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.0463483333587646,
+      "learning_rate": 5.655496320282472e-07,
+      "loss": 0.476,
+      "step": 52313
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.7123665809631348,
+      "learning_rate": 5.654278332524221e-07,
+      "loss": 0.4284,
+      "step": 52314
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.724616527557373,
+      "learning_rate": 5.6530604680748e-07,
+      "loss": 0.6907,
+      "step": 52315
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.2971749305725098,
+      "learning_rate": 5.651842726937617e-07,
+      "loss": 0.4814,
+      "step": 52316
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.833432912826538,
+      "learning_rate": 5.650625109116032e-07,
+      "loss": 0.5627,
+      "step": 52317
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.601391553878784,
+      "learning_rate": 5.649407614613462e-07,
+      "loss": 0.6941,
+      "step": 52318
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.0633299350738525,
+      "learning_rate": 5.64819024343326e-07,
+      "loss": 0.3357,
+      "step": 52319
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.556107521057129,
+      "learning_rate": 5.646972995578837e-07,
+      "loss": 0.4421,
+      "step": 52320
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.366067886352539,
+      "learning_rate": 5.645755871053548e-07,
+      "loss": 0.394,
+      "step": 52321
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.8086109161376953,
+      "learning_rate": 5.644538869860816e-07,
+      "loss": 0.8225,
+      "step": 52322
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.3221826553344727,
+      "learning_rate": 5.643321992003992e-07,
+      "loss": 0.516,
+      "step": 52323
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.646026372909546,
+      "learning_rate": 5.642105237486489e-07,
+      "loss": 0.5513,
+      "step": 52324
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 1.8901715278625488,
+      "learning_rate": 5.640888606311662e-07,
+      "loss": 0.4299,
+      "step": 52325
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.1141538619995117,
+      "learning_rate": 5.639672098482918e-07,
+      "loss": 0.5949,
+      "step": 52326
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.09458065032959,
+      "learning_rate": 5.638455714003616e-07,
+      "loss": 0.5617,
+      "step": 52327
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.478614330291748,
+      "learning_rate": 5.637239452877157e-07,
+      "loss": 0.44,
+      "step": 52328
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 4.23210334777832,
+      "learning_rate": 5.636023315106914e-07,
+      "loss": 0.7233,
+      "step": 52329
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.9130518436431885,
+      "learning_rate": 5.634807300696277e-07,
+      "loss": 0.6726,
+      "step": 52330
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 4.984939098358154,
+      "learning_rate": 5.633591409648615e-07,
+      "loss": 0.3706,
+      "step": 52331
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 1.9601106643676758,
+      "learning_rate": 5.632375641967319e-07,
+      "loss": 0.3708,
+      "step": 52332
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.166707992553711,
+      "learning_rate": 5.631159997655761e-07,
+      "loss": 0.6082,
+      "step": 52333
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.470430612564087,
+      "learning_rate": 5.62994447671732e-07,
+      "loss": 0.4199,
+      "step": 52334
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.342543125152588,
+      "learning_rate": 5.628729079155398e-07,
+      "loss": 0.5561,
+      "step": 52335
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 1.8876018524169922,
+      "learning_rate": 5.627513804973345e-07,
+      "loss": 0.4004,
+      "step": 52336
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 4.379771709442139,
+      "learning_rate": 5.626298654174566e-07,
+      "loss": 0.5844,
+      "step": 52337
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.3622965812683105,
+      "learning_rate": 5.625083626762417e-07,
+      "loss": 0.6853,
+      "step": 52338
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.5893242359161377,
+      "learning_rate": 5.6238687227403e-07,
+      "loss": 0.5403,
+      "step": 52339
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 1.848659873008728,
+      "learning_rate": 5.622653942111567e-07,
+      "loss": 0.4016,
+      "step": 52340
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 5.515697956085205,
+      "learning_rate": 5.621439284879621e-07,
+      "loss": 0.549,
+      "step": 52341
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 4.563741207122803,
+      "learning_rate": 5.620224751047814e-07,
+      "loss": 0.644,
+      "step": 52342
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.29736065864563,
+      "learning_rate": 5.61901034061954e-07,
+      "loss": 0.6274,
+      "step": 52343
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.4093544483184814,
+      "learning_rate": 5.617796053598184e-07,
+      "loss": 0.5134,
+      "step": 52344
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 1.903824806213379,
+      "learning_rate": 5.6165818899871e-07,
+      "loss": 0.233,
+      "step": 52345
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 6.4844770431518555,
+      "learning_rate": 5.615367849789671e-07,
+      "loss": 0.8176,
+      "step": 52346
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.204071044921875,
+      "learning_rate": 5.614153933009292e-07,
+      "loss": 0.6138,
+      "step": 52347
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.674081802368164,
+      "learning_rate": 5.612940139649315e-07,
+      "loss": 0.6966,
+      "step": 52348
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 1.9783217906951904,
+      "learning_rate": 5.61172646971313e-07,
+      "loss": 0.5288,
+      "step": 52349
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.0600385665893555,
+      "learning_rate": 5.610512923204097e-07,
+      "loss": 0.6569,
+      "step": 52350
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.1517748832702637,
+      "learning_rate": 5.609299500125609e-07,
+      "loss": 0.8157,
+      "step": 52351
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.0301153659820557,
+      "learning_rate": 5.608086200481022e-07,
+      "loss": 0.4216,
+      "step": 52352
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.721571922302246,
+      "learning_rate": 5.606873024273724e-07,
+      "loss": 0.6746,
+      "step": 52353
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.2430644035339355,
+      "learning_rate": 5.605659971507072e-07,
+      "loss": 0.7109,
+      "step": 52354
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.700606107711792,
+      "learning_rate": 5.604447042184463e-07,
+      "loss": 0.5045,
+      "step": 52355
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.4519898891448975,
+      "learning_rate": 5.60323423630924e-07,
+      "loss": 0.4715,
+      "step": 52356
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 7.5395708084106445,
+      "learning_rate": 5.602021553884807e-07,
+      "loss": 0.56,
+      "step": 52357
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.167757749557495,
+      "learning_rate": 5.600808994914508e-07,
+      "loss": 0.4297,
+      "step": 52358
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.7556233406066895,
+      "learning_rate": 5.599596559401738e-07,
+      "loss": 0.6327,
+      "step": 52359
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.8544678688049316,
+      "learning_rate": 5.598384247349842e-07,
+      "loss": 0.7114,
+      "step": 52360
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.6094613075256348,
+      "learning_rate": 5.597172058762224e-07,
+      "loss": 0.5706,
+      "step": 52361
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.21288800239563,
+      "learning_rate": 5.595959993642208e-07,
+      "loss": 0.4603,
+      "step": 52362
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 4.692756175994873,
+      "learning_rate": 5.594748051993221e-07,
+      "loss": 0.4152,
+      "step": 52363
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.8135154247283936,
+      "learning_rate": 5.593536233818592e-07,
+      "loss": 0.502,
+      "step": 52364
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.1086156368255615,
+      "learning_rate": 5.592324539121713e-07,
+      "loss": 0.526,
+      "step": 52365
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.3405025005340576,
+      "learning_rate": 5.591112967905937e-07,
+      "loss": 0.5626,
+      "step": 52366
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.6911604404449463,
+      "learning_rate": 5.589901520174646e-07,
+      "loss": 0.526,
+      "step": 52367
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.3336007595062256,
+      "learning_rate": 5.588690195931196e-07,
+      "loss": 0.5935,
+      "step": 52368
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.21911358833313,
+      "learning_rate": 5.587478995178969e-07,
+      "loss": 0.2937,
+      "step": 52369
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 5.473060607910156,
+      "learning_rate": 5.586267917921318e-07,
+      "loss": 0.5072,
+      "step": 52370
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 1.7591593265533447,
+      "learning_rate": 5.585056964161623e-07,
+      "loss": 0.363,
+      "step": 52371
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.6163222789764404,
+      "learning_rate": 5.583846133903242e-07,
+      "loss": 0.5834,
+      "step": 52372
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.0675852298736572,
+      "learning_rate": 5.582635427149552e-07,
+      "loss": 0.5292,
+      "step": 52373
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.334475040435791,
+      "learning_rate": 5.581424843903904e-07,
+      "loss": 0.6149,
+      "step": 52374
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.5500848293304443,
+      "learning_rate": 5.580214384169691e-07,
+      "loss": 0.6746,
+      "step": 52375
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.193547487258911,
+      "learning_rate": 5.579004047950242e-07,
+      "loss": 0.4453,
+      "step": 52376
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.6050355434417725,
+      "learning_rate": 5.577793835248957e-07,
+      "loss": 0.5053,
+      "step": 52377
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.135141134262085,
+      "learning_rate": 5.576583746069181e-07,
+      "loss": 0.7821,
+      "step": 52378
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.5311684608459473,
+      "learning_rate": 5.575373780414289e-07,
+      "loss": 0.7317,
+      "step": 52379
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.1374471187591553,
+      "learning_rate": 5.57416393828763e-07,
+      "loss": 0.5883,
+      "step": 52380
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.3201613426208496,
+      "learning_rate": 5.572954219692589e-07,
+      "loss": 0.6227,
+      "step": 52381
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.0590758323669434,
+      "learning_rate": 5.571744624632502e-07,
+      "loss": 0.49,
+      "step": 52382
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.829214572906494,
+      "learning_rate": 5.570535153110767e-07,
+      "loss": 0.4981,
+      "step": 52383
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.2127716541290283,
+      "learning_rate": 5.56932580513072e-07,
+      "loss": 0.5757,
+      "step": 52384
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.828040599822998,
+      "learning_rate": 5.568116580695748e-07,
+      "loss": 0.5938,
+      "step": 52385
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.101771354675293,
+      "learning_rate": 5.566907479809186e-07,
+      "loss": 0.8323,
+      "step": 52386
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.4230852127075195,
+      "learning_rate": 5.56569850247442e-07,
+      "loss": 0.5786,
+      "step": 52387
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.4510128498077393,
+      "learning_rate": 5.564489648694787e-07,
+      "loss": 0.5292,
+      "step": 52388
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.6921656131744385,
+      "learning_rate": 5.56328091847368e-07,
+      "loss": 0.5378,
+      "step": 52389
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.270528554916382,
+      "learning_rate": 5.562072311814425e-07,
+      "loss": 0.6438,
+      "step": 52390
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.4671709537506104,
+      "learning_rate": 5.560863828720414e-07,
+      "loss": 0.5409,
+      "step": 52391
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.5258147716522217,
+      "learning_rate": 5.559655469194985e-07,
+      "loss": 0.5978,
+      "step": 52392
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.924487352371216,
+      "learning_rate": 5.558447233241515e-07,
+      "loss": 0.4502,
+      "step": 52393
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.8897969722747803,
+      "learning_rate": 5.557239120863344e-07,
+      "loss": 0.685,
+      "step": 52394
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.0670013427734375,
+      "learning_rate": 5.556031132063843e-07,
+      "loss": 0.6098,
+      "step": 52395
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.7125580310821533,
+      "learning_rate": 5.554823266846382e-07,
+      "loss": 0.4934,
+      "step": 52396
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.323669672012329,
+      "learning_rate": 5.553615525214295e-07,
+      "loss": 0.7941,
+      "step": 52397
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.9611146450042725,
+      "learning_rate": 5.552407907170964e-07,
+      "loss": 0.735,
+      "step": 52398
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.707644462585449,
+      "learning_rate": 5.551200412719726e-07,
+      "loss": 0.5592,
+      "step": 52399
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 5.092102527618408,
+      "learning_rate": 5.549993041863961e-07,
+      "loss": 0.6517,
+      "step": 52400
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.7258777618408203,
+      "learning_rate": 5.548785794607004e-07,
+      "loss": 0.7652,
+      "step": 52401
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.317384958267212,
+      "learning_rate": 5.54757867095223e-07,
+      "loss": 0.5161,
+      "step": 52402
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.3037009239196777,
+      "learning_rate": 5.546371670902978e-07,
+      "loss": 0.5533,
+      "step": 52403
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.1955132484436035,
+      "learning_rate": 5.54516479446261e-07,
+      "loss": 0.5317,
+      "step": 52404
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.010699987411499,
+      "learning_rate": 5.543958041634501e-07,
+      "loss": 0.4958,
+      "step": 52405
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.086362361907959,
+      "learning_rate": 5.542751412421976e-07,
+      "loss": 0.7842,
+      "step": 52406
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.5268917083740234,
+      "learning_rate": 5.541544906828411e-07,
+      "loss": 0.651,
+      "step": 52407
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.4617204666137695,
+      "learning_rate": 5.540338524857164e-07,
+      "loss": 0.6755,
+      "step": 52408
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.988840341567993,
+      "learning_rate": 5.539132266511566e-07,
+      "loss": 0.502,
+      "step": 52409
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.3096373081207275,
+      "learning_rate": 5.537926131795001e-07,
+      "loss": 0.4966,
+      "step": 52410
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.8493189811706543,
+      "learning_rate": 5.536720120710792e-07,
+      "loss": 0.5567,
+      "step": 52411
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.986555814743042,
+      "learning_rate": 5.535514233262324e-07,
+      "loss": 0.3519,
+      "step": 52412
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.4266483783721924,
+      "learning_rate": 5.53430846945292e-07,
+      "loss": 0.6315,
+      "step": 52413
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 4.407377243041992,
+      "learning_rate": 5.533102829285957e-07,
+      "loss": 0.4785,
+      "step": 52414
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.180299997329712,
+      "learning_rate": 5.531897312764772e-07,
+      "loss": 0.5289,
+      "step": 52415
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.7013962268829346,
+      "learning_rate": 5.530691919892727e-07,
+      "loss": 0.5624,
+      "step": 52416
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.5209500789642334,
+      "learning_rate": 5.52948665067316e-07,
+      "loss": 0.5834,
+      "step": 52417
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.3187570571899414,
+      "learning_rate": 5.528281505109445e-07,
+      "loss": 0.301,
+      "step": 52418
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.8279178142547607,
+      "learning_rate": 5.527076483204907e-07,
+      "loss": 0.4816,
+      "step": 52419
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.8745598793029785,
+      "learning_rate": 5.525871584962922e-07,
+      "loss": 0.5187,
+      "step": 52420
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.455624580383301,
+      "learning_rate": 5.524666810386814e-07,
+      "loss": 0.6024,
+      "step": 52421
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 5.610350608825684,
+      "learning_rate": 5.523462159479959e-07,
+      "loss": 0.6581,
+      "step": 52422
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.1954753398895264,
+      "learning_rate": 5.522257632245675e-07,
+      "loss": 0.5471,
+      "step": 52423
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.7998123168945312,
+      "learning_rate": 5.521053228687346e-07,
+      "loss": 0.9052,
+      "step": 52424
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.8488895893096924,
+      "learning_rate": 5.519848948808299e-07,
+      "loss": 0.5933,
+      "step": 52425
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.308845043182373,
+      "learning_rate": 5.518644792611905e-07,
+      "loss": 0.5395,
+      "step": 52426
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.287461280822754,
+      "learning_rate": 5.517440760101478e-07,
+      "loss": 0.7781,
+      "step": 52427
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.150161027908325,
+      "learning_rate": 5.516236851280399e-07,
+      "loss": 0.5629,
+      "step": 52428
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.0729453563690186,
+      "learning_rate": 5.515033066151992e-07,
+      "loss": 0.7271,
+      "step": 52429
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.857396125793457,
+      "learning_rate": 5.513829404719623e-07,
+      "loss": 0.5668,
+      "step": 52430
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.5931169986724854,
+      "learning_rate": 5.512625866986615e-07,
+      "loss": 0.6009,
+      "step": 52431
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.88336181640625,
+      "learning_rate": 5.511422452956344e-07,
+      "loss": 0.512,
+      "step": 52432
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.635528326034546,
+      "learning_rate": 5.51021916263213e-07,
+      "loss": 0.5216,
+      "step": 52433
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.6899495124816895,
+      "learning_rate": 5.509015996017342e-07,
+      "loss": 0.5896,
+      "step": 52434
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.5066871643066406,
+      "learning_rate": 5.507812953115299e-07,
+      "loss": 0.6905,
+      "step": 52435
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.499168634414673,
+      "learning_rate": 5.506610033929377e-07,
+      "loss": 0.6397,
+      "step": 52436
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.767465114593506,
+      "learning_rate": 5.50540723846289e-07,
+      "loss": 0.4878,
+      "step": 52437
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.0008974075317383,
+      "learning_rate": 5.504204566719212e-07,
+      "loss": 0.4767,
+      "step": 52438
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.0720980167388916,
+      "learning_rate": 5.503002018701659e-07,
+      "loss": 0.7754,
+      "step": 52439
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.6531567573547363,
+      "learning_rate": 5.501799594413598e-07,
+      "loss": 0.6093,
+      "step": 52440
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.8431971073150635,
+      "learning_rate": 5.500597293858356e-07,
+      "loss": 0.7799,
+      "step": 52441
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.3327057361602783,
+      "learning_rate": 5.499395117039292e-07,
+      "loss": 0.6413,
+      "step": 52442
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 4.429494380950928,
+      "learning_rate": 5.498193063959717e-07,
+      "loss": 0.6895,
+      "step": 52443
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.9390523433685303,
+      "learning_rate": 5.49699113462302e-07,
+      "loss": 0.4679,
+      "step": 52444
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.3609888553619385,
+      "learning_rate": 5.495789329032509e-07,
+      "loss": 0.4589,
+      "step": 52445
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.750607490539551,
+      "learning_rate": 5.494587647191546e-07,
+      "loss": 0.3148,
+      "step": 52446
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.8469178676605225,
+      "learning_rate": 5.493386089103453e-07,
+      "loss": 0.6592,
+      "step": 52447
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 1.8019603490829468,
+      "learning_rate": 5.492184654771588e-07,
+      "loss": 0.4849,
+      "step": 52448
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.2813613414764404,
+      "learning_rate": 5.490983344199274e-07,
+      "loss": 0.7821,
+      "step": 52449
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.5833969116210938,
+      "learning_rate": 5.489782157389878e-07,
+      "loss": 0.6876,
+      "step": 52450
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.1096699237823486,
+      "learning_rate": 5.488581094346707e-07,
+      "loss": 0.5652,
+      "step": 52451
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.6833953857421875,
+      "learning_rate": 5.48738015507313e-07,
+      "loss": 0.5223,
+      "step": 52452
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.22210955619812,
+      "learning_rate": 5.486179339572467e-07,
+      "loss": 0.546,
+      "step": 52453
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.640826463699341,
+      "learning_rate": 5.484978647848072e-07,
+      "loss": 0.5308,
+      "step": 52454
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.7639472484588623,
+      "learning_rate": 5.483778079903262e-07,
+      "loss": 0.618,
+      "step": 52455
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.2308390140533447,
+      "learning_rate": 5.482577635741404e-07,
+      "loss": 0.5951,
+      "step": 52456
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.8634402751922607,
+      "learning_rate": 5.48137731536581e-07,
+      "loss": 0.4823,
+      "step": 52457
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.1128878593444824,
+      "learning_rate": 5.480177118779829e-07,
+      "loss": 0.6055,
+      "step": 52458
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.893688440322876,
+      "learning_rate": 5.478977045986806e-07,
+      "loss": 0.5308,
+      "step": 52459
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 4.246460914611816,
+      "learning_rate": 5.477777096990061e-07,
+      "loss": 0.6672,
+      "step": 52460
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.052447557449341,
+      "learning_rate": 5.476577271792949e-07,
+      "loss": 0.456,
+      "step": 52461
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.6126105785369873,
+      "learning_rate": 5.475377570398787e-07,
+      "loss": 0.634,
+      "step": 52462
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 4.137747287750244,
+      "learning_rate": 5.474177992810931e-07,
+      "loss": 0.4856,
+      "step": 52463
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 1.9405626058578491,
+      "learning_rate": 5.472978539032698e-07,
+      "loss": 0.3754,
+      "step": 52464
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.9186134338378906,
+      "learning_rate": 5.471779209067429e-07,
+      "loss": 0.5761,
+      "step": 52465
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.2079761028289795,
+      "learning_rate": 5.470580002918468e-07,
+      "loss": 0.8384,
+      "step": 52466
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.522782325744629,
+      "learning_rate": 5.46938092058914e-07,
+      "loss": 0.5584,
+      "step": 52467
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.029115915298462,
+      "learning_rate": 5.468181962082786e-07,
+      "loss": 0.7353,
+      "step": 52468
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.836013078689575,
+      "learning_rate": 5.46698312740273e-07,
+      "loss": 0.4717,
+      "step": 52469
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.1450355052948,
+      "learning_rate": 5.465784416552311e-07,
+      "loss": 0.5914,
+      "step": 52470
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.3788483142852783,
+      "learning_rate": 5.464585829534869e-07,
+      "loss": 0.6493,
+      "step": 52471
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.748932361602783,
+      "learning_rate": 5.463387366353723e-07,
+      "loss": 0.5404,
+      "step": 52472
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.7938649654388428,
+      "learning_rate": 5.462189027012227e-07,
+      "loss": 0.5198,
+      "step": 52473
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.3738691806793213,
+      "learning_rate": 5.460990811513684e-07,
+      "loss": 0.6913,
+      "step": 52474
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 4.821985244750977,
+      "learning_rate": 5.459792719861451e-07,
+      "loss": 0.6598,
+      "step": 52475
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.586263656616211,
+      "learning_rate": 5.458594752058844e-07,
+      "loss": 0.3495,
+      "step": 52476
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.499297857284546,
+      "learning_rate": 5.457396908109202e-07,
+      "loss": 0.5049,
+      "step": 52477
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.434800148010254,
+      "learning_rate": 5.456199188015848e-07,
+      "loss": 0.6571,
+      "step": 52478
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 1.9514118432998657,
+      "learning_rate": 5.455001591782127e-07,
+      "loss": 0.2614,
+      "step": 52479
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 4.8944549560546875,
+      "learning_rate": 5.453804119411349e-07,
+      "loss": 0.6155,
+      "step": 52480
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.630922794342041,
+      "learning_rate": 5.452606770906865e-07,
+      "loss": 0.4951,
+      "step": 52481
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.4536807537078857,
+      "learning_rate": 5.451409546271986e-07,
+      "loss": 0.6908,
+      "step": 52482
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.4058144092559814,
+      "learning_rate": 5.450212445510056e-07,
+      "loss": 0.7663,
+      "step": 52483
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.776677370071411,
+      "learning_rate": 5.449015468624375e-07,
+      "loss": 0.5808,
+      "step": 52484
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.6570236682891846,
+      "learning_rate": 5.447818615618311e-07,
+      "loss": 0.7131,
+      "step": 52485
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.964101791381836,
+      "learning_rate": 5.446621886495168e-07,
+      "loss": 0.5181,
+      "step": 52486
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.390810489654541,
+      "learning_rate": 5.445425281258288e-07,
+      "loss": 0.476,
+      "step": 52487
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.707892894744873,
+      "learning_rate": 5.444228799910983e-07,
+      "loss": 0.2522,
+      "step": 52488
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.8166959285736084,
+      "learning_rate": 5.443032442456592e-07,
+      "loss": 0.5052,
+      "step": 52489
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.98677921295166,
+      "learning_rate": 5.441836208898433e-07,
+      "loss": 0.6898,
+      "step": 52490
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.98187255859375,
+      "learning_rate": 5.440640099239836e-07,
+      "loss": 0.7075,
+      "step": 52491
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 4.24844217300415,
+      "learning_rate": 5.439444113484121e-07,
+      "loss": 0.7429,
+      "step": 52492
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 4.262761116027832,
+      "learning_rate": 5.438248251634631e-07,
+      "loss": 0.5673,
+      "step": 52493
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.593411684036255,
+      "learning_rate": 5.437052513694668e-07,
+      "loss": 0.547,
+      "step": 52494
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.351428985595703,
+      "learning_rate": 5.43585689966758e-07,
+      "loss": 0.6166,
+      "step": 52495
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.830676794052124,
+      "learning_rate": 5.434661409556669e-07,
+      "loss": 0.5432,
+      "step": 52496
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.632844924926758,
+      "learning_rate": 5.433466043365276e-07,
+      "loss": 0.9241,
+      "step": 52497
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.422264337539673,
+      "learning_rate": 5.432270801096717e-07,
+      "loss": 0.6502,
+      "step": 52498
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.622568130493164,
+      "learning_rate": 5.431075682754322e-07,
+      "loss": 0.6875,
+      "step": 52499
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.696983575820923,
+      "learning_rate": 5.429880688341399e-07,
+      "loss": 0.4043,
+      "step": 52500
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 4.378589153289795,
+      "learning_rate": 5.428685817861296e-07,
+      "loss": 0.4456,
+      "step": 52501
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.852174997329712,
+      "learning_rate": 5.427491071317303e-07,
+      "loss": 0.7538,
+      "step": 52502
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.490659236907959,
+      "learning_rate": 5.426296448712775e-07,
+      "loss": 0.5761,
+      "step": 52503
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 5.731533527374268,
+      "learning_rate": 5.425101950050998e-07,
+      "loss": 0.586,
+      "step": 52504
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.448713541030884,
+      "learning_rate": 5.423907575335336e-07,
+      "loss": 0.4622,
+      "step": 52505
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.4390196800231934,
+      "learning_rate": 5.422713324569078e-07,
+      "loss": 0.564,
+      "step": 52506
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.913236618041992,
+      "learning_rate": 5.421519197755564e-07,
+      "loss": 0.532,
+      "step": 52507
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.448148488998413,
+      "learning_rate": 5.420325194898096e-07,
+      "loss": 0.5531,
+      "step": 52508
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.6088690757751465,
+      "learning_rate": 5.419131316000014e-07,
+      "loss": 0.6754,
+      "step": 52509
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.254986524581909,
+      "learning_rate": 5.417937561064618e-07,
+      "loss": 0.4423,
+      "step": 52510
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.15033221244812,
+      "learning_rate": 5.416743930095248e-07,
+      "loss": 0.5119,
+      "step": 52511
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.8820765018463135,
+      "learning_rate": 5.4155504230952e-07,
+      "loss": 0.615,
+      "step": 52512
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.0267274379730225,
+      "learning_rate": 5.414357040067814e-07,
+      "loss": 0.5183,
+      "step": 52513
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.1432113647460938,
+      "learning_rate": 5.41316378101639e-07,
+      "loss": 0.5316,
+      "step": 52514
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.485515594482422,
+      "learning_rate": 5.411970645944265e-07,
+      "loss": 0.5133,
+      "step": 52515
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.4403388500213623,
+      "learning_rate": 5.410777634854736e-07,
+      "loss": 0.5306,
+      "step": 52516
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.6320366859436035,
+      "learning_rate": 5.409584747751139e-07,
+      "loss": 0.4672,
+      "step": 52517
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.99044132232666,
+      "learning_rate": 5.40839198463678e-07,
+      "loss": 0.5455,
+      "step": 52518
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.622738838195801,
+      "learning_rate": 5.407199345514974e-07,
+      "loss": 0.7269,
+      "step": 52519
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 4.950223922729492,
+      "learning_rate": 5.406006830389049e-07,
+      "loss": 0.682,
+      "step": 52520
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.459306478500366,
+      "learning_rate": 5.404814439262307e-07,
+      "loss": 0.3352,
+      "step": 52521
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.051443099975586,
+      "learning_rate": 5.403622172138078e-07,
+      "loss": 0.4886,
+      "step": 52522
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.291358232498169,
+      "learning_rate": 5.402430029019662e-07,
+      "loss": 0.6025,
+      "step": 52523
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.2773096561431885,
+      "learning_rate": 5.401238009910387e-07,
+      "loss": 0.688,
+      "step": 52524
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.9413485527038574,
+      "learning_rate": 5.400046114813556e-07,
+      "loss": 0.4815,
+      "step": 52525
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.6978070735931396,
+      "learning_rate": 5.39885434373249e-07,
+      "loss": 0.656,
+      "step": 52526
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 4.2904157638549805,
+      "learning_rate": 5.397662696670508e-07,
+      "loss": 0.5435,
+      "step": 52527
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 4.234829902648926,
+      "learning_rate": 5.396471173630913e-07,
+      "loss": 0.7103,
+      "step": 52528
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 4.147305965423584,
+      "learning_rate": 5.39527977461703e-07,
+      "loss": 0.653,
+      "step": 52529
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.7131154537200928,
+      "learning_rate": 5.394088499632155e-07,
+      "loss": 0.4099,
+      "step": 52530
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 5.506972312927246,
+      "learning_rate": 5.39289734867961e-07,
+      "loss": 0.5749,
+      "step": 52531
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.170074224472046,
+      "learning_rate": 5.391706321762718e-07,
+      "loss": 0.7534,
+      "step": 52532
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.7172915935516357,
+      "learning_rate": 5.390515418884767e-07,
+      "loss": 0.7186,
+      "step": 52533
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.8576767444610596,
+      "learning_rate": 5.389324640049093e-07,
+      "loss": 0.5979,
+      "step": 52534
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.7237579822540283,
+      "learning_rate": 5.388133985258987e-07,
+      "loss": 0.5757,
+      "step": 52535
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.9205150604248047,
+      "learning_rate": 5.386943454517779e-07,
+      "loss": 0.7145,
+      "step": 52536
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.148202896118164,
+      "learning_rate": 5.385753047828757e-07,
+      "loss": 0.6552,
+      "step": 52537
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.6971700191497803,
+      "learning_rate": 5.38456276519525e-07,
+      "loss": 0.6486,
+      "step": 52538
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.317949056625366,
+      "learning_rate": 5.383372606620557e-07,
+      "loss": 0.7962,
+      "step": 52539
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.5890002250671387,
+      "learning_rate": 5.382182572107996e-07,
+      "loss": 0.838,
+      "step": 52540
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 1.8144714832305908,
+      "learning_rate": 5.380992661660866e-07,
+      "loss": 0.4792,
+      "step": 52541
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 1.979752540588379,
+      "learning_rate": 5.379802875282486e-07,
+      "loss": 0.4483,
+      "step": 52542
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.7258248329162598,
+      "learning_rate": 5.378613212976152e-07,
+      "loss": 0.581,
+      "step": 52543
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.379758596420288,
+      "learning_rate": 5.377423674745186e-07,
+      "loss": 0.5065,
+      "step": 52544
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.1772725582122803,
+      "learning_rate": 5.376234260592872e-07,
+      "loss": 0.5275,
+      "step": 52545
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.9582440853118896,
+      "learning_rate": 5.375044970522553e-07,
+      "loss": 0.6589,
+      "step": 52546
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 6.156806468963623,
+      "learning_rate": 5.373855804537509e-07,
+      "loss": 0.2027,
+      "step": 52547
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.1576013565063477,
+      "learning_rate": 5.37266676264106e-07,
+      "loss": 0.8868,
+      "step": 52548
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.146268367767334,
+      "learning_rate": 5.371477844836498e-07,
+      "loss": 0.7052,
+      "step": 52549
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.873251438140869,
+      "learning_rate": 5.370289051127153e-07,
+      "loss": 0.6184,
+      "step": 52550
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.8817615509033203,
+      "learning_rate": 5.369100381516301e-07,
+      "loss": 0.4424,
+      "step": 52551
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.6006288528442383,
+      "learning_rate": 5.367911836007273e-07,
+      "loss": 0.5831,
+      "step": 52552
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.819068193435669,
+      "learning_rate": 5.366723414603353e-07,
+      "loss": 0.3463,
+      "step": 52553
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.741132974624634,
+      "learning_rate": 5.365535117307863e-07,
+      "loss": 0.8116,
+      "step": 52554
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 1.800603985786438,
+      "learning_rate": 5.364346944124088e-07,
+      "loss": 0.3201,
+      "step": 52555
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.5985705852508545,
+      "learning_rate": 5.363158895055359e-07,
+      "loss": 0.7294,
+      "step": 52556
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.567924737930298,
+      "learning_rate": 5.361970970104952e-07,
+      "loss": 0.5358,
+      "step": 52557
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.5345306396484375,
+      "learning_rate": 5.36078316927619e-07,
+      "loss": 0.5854,
+      "step": 52558
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.8720977306365967,
+      "learning_rate": 5.359595492572356e-07,
+      "loss": 0.5888,
+      "step": 52559
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.0147581100463867,
+      "learning_rate": 5.358407939996779e-07,
+      "loss": 0.6864,
+      "step": 52560
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 4.075493812561035,
+      "learning_rate": 5.357220511552735e-07,
+      "loss": 0.6559,
+      "step": 52561
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.098428249359131,
+      "learning_rate": 5.356033207243544e-07,
+      "loss": 0.4368,
+      "step": 52562
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.206951379776001,
+      "learning_rate": 5.354846027072492e-07,
+      "loss": 0.6213,
+      "step": 52563
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.9551122188568115,
+      "learning_rate": 5.353658971042902e-07,
+      "loss": 0.6403,
+      "step": 52564
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.7183492183685303,
+      "learning_rate": 5.352472039158041e-07,
+      "loss": 0.5218,
+      "step": 52565
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.8387908935546875,
+      "learning_rate": 5.351285231421244e-07,
+      "loss": 0.7372,
+      "step": 52566
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.6891067028045654,
+      "learning_rate": 5.350098547835792e-07,
+      "loss": 0.4759,
+      "step": 52567
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.2682840824127197,
+      "learning_rate": 5.348911988404998e-07,
+      "loss": 0.5921,
+      "step": 52568
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.831085681915283,
+      "learning_rate": 5.347725553132144e-07,
+      "loss": 0.3666,
+      "step": 52569
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.3491339683532715,
+      "learning_rate": 5.346539242020548e-07,
+      "loss": 0.7176,
+      "step": 52570
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.8261430263519287,
+      "learning_rate": 5.345353055073493e-07,
+      "loss": 0.7587,
+      "step": 52571
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.7285242080688477,
+      "learning_rate": 5.344166992294286e-07,
+      "loss": 0.7112,
+      "step": 52572
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 4.283848285675049,
+      "learning_rate": 5.342981053686213e-07,
+      "loss": 0.5952,
+      "step": 52573
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.5772013664245605,
+      "learning_rate": 5.341795239252596e-07,
+      "loss": 0.4468,
+      "step": 52574
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.4559590816497803,
+      "learning_rate": 5.340609548996706e-07,
+      "loss": 0.5206,
+      "step": 52575
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.3480329513549805,
+      "learning_rate": 5.339423982921854e-07,
+      "loss": 0.661,
+      "step": 52576
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.3914895057678223,
+      "learning_rate": 5.338238541031331e-07,
+      "loss": 0.4494,
+      "step": 52577
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.029042959213257,
+      "learning_rate": 5.337053223328442e-07,
+      "loss": 0.5456,
+      "step": 52578
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.189286947250366,
+      "learning_rate": 5.335868029816471e-07,
+      "loss": 0.6335,
+      "step": 52579
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.552856206893921,
+      "learning_rate": 5.334682960498727e-07,
+      "loss": 0.7452,
+      "step": 52580
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.7045538425445557,
+      "learning_rate": 5.333498015378485e-07,
+      "loss": 0.5448,
+      "step": 52581
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.151599168777466,
+      "learning_rate": 5.33231319445906e-07,
+      "loss": 0.5686,
+      "step": 52582
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.6371688842773438,
+      "learning_rate": 5.331128497743743e-07,
+      "loss": 0.7528,
+      "step": 52583
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.6732399463653564,
+      "learning_rate": 5.329943925235815e-07,
+      "loss": 0.7973,
+      "step": 52584
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.2790098190307617,
+      "learning_rate": 5.328759476938589e-07,
+      "loss": 0.6131,
+      "step": 52585
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.037294864654541,
+      "learning_rate": 5.327575152855341e-07,
+      "loss": 0.2952,
+      "step": 52586
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 5.154552459716797,
+      "learning_rate": 5.326390952989369e-07,
+      "loss": 0.5217,
+      "step": 52587
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.5849714279174805,
+      "learning_rate": 5.325206877343981e-07,
+      "loss": 0.4825,
+      "step": 52588
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.567317008972168,
+      "learning_rate": 5.324022925922445e-07,
+      "loss": 0.5654,
+      "step": 52589
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.5919644832611084,
+      "learning_rate": 5.322839098728078e-07,
+      "loss": 0.6965,
+      "step": 52590
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.6784567832946777,
+      "learning_rate": 5.321655395764147e-07,
+      "loss": 0.5653,
+      "step": 52591
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.8456015586853027,
+      "learning_rate": 5.320471817033956e-07,
+      "loss": 0.4107,
+      "step": 52592
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.004042387008667,
+      "learning_rate": 5.319288362540803e-07,
+      "loss": 0.7826,
+      "step": 52593
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.0682427883148193,
+      "learning_rate": 5.318105032287962e-07,
+      "loss": 0.7282,
+      "step": 52594
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.6854658126831055,
+      "learning_rate": 5.316921826278748e-07,
+      "loss": 0.69,
+      "step": 52595
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.4734721183776855,
+      "learning_rate": 5.315738744516419e-07,
+      "loss": 0.8717,
+      "step": 52596
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.6491212844848633,
+      "learning_rate": 5.314555787004295e-07,
+      "loss": 0.5107,
+      "step": 52597
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.672698974609375,
+      "learning_rate": 5.313372953745644e-07,
+      "loss": 0.7235,
+      "step": 52598
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.8032000064849854,
+      "learning_rate": 5.31219024474377e-07,
+      "loss": 0.4414,
+      "step": 52599
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 4.939488410949707,
+      "learning_rate": 5.311007660001943e-07,
+      "loss": 0.5728,
+      "step": 52600
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.778310537338257,
+      "learning_rate": 5.309825199523472e-07,
+      "loss": 0.6284,
+      "step": 52601
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.5285236835479736,
+      "learning_rate": 5.30864286331163e-07,
+      "loss": 0.4197,
+      "step": 52602
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.9096791744232178,
+      "learning_rate": 5.307460651369717e-07,
+      "loss": 0.6855,
+      "step": 52603
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.906233787536621,
+      "learning_rate": 5.306278563701007e-07,
+      "loss": 0.6355,
+      "step": 52604
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.590439796447754,
+      "learning_rate": 5.305096600308801e-07,
+      "loss": 0.5285,
+      "step": 52605
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.353403329849243,
+      "learning_rate": 5.30391476119636e-07,
+      "loss": 0.7371,
+      "step": 52606
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.755784034729004,
+      "learning_rate": 5.302733046367009e-07,
+      "loss": 0.4221,
+      "step": 52607
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.815640926361084,
+      "learning_rate": 5.301551455824e-07,
+      "loss": 0.5623,
+      "step": 52608
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.754521369934082,
+      "learning_rate": 5.300369989570647e-07,
+      "loss": 0.5526,
+      "step": 52609
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.9692206382751465,
+      "learning_rate": 5.299188647610204e-07,
+      "loss": 0.6883,
+      "step": 52610
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.7632079124450684,
+      "learning_rate": 5.298007429945989e-07,
+      "loss": 0.9702,
+      "step": 52611
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.611290454864502,
+      "learning_rate": 5.296826336581256e-07,
+      "loss": 0.4153,
+      "step": 52612
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 4.0181097984313965,
+      "learning_rate": 5.295645367519308e-07,
+      "loss": 0.633,
+      "step": 52613
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.4357802867889404,
+      "learning_rate": 5.294464522763421e-07,
+      "loss": 0.5845,
+      "step": 52614
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.5176379680633545,
+      "learning_rate": 5.29328380231689e-07,
+      "loss": 0.3577,
+      "step": 52615
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.178818702697754,
+      "learning_rate": 5.292103206182974e-07,
+      "loss": 0.4559,
+      "step": 52616
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.3261892795562744,
+      "learning_rate": 5.290922734364989e-07,
+      "loss": 0.683,
+      "step": 52617
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.5001108646392822,
+      "learning_rate": 5.289742386866181e-07,
+      "loss": 0.369,
+      "step": 52618
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.510528326034546,
+      "learning_rate": 5.288562163689864e-07,
+      "loss": 0.6737,
+      "step": 52619
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.6260323524475098,
+      "learning_rate": 5.287382064839303e-07,
+      "loss": 0.732,
+      "step": 52620
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.520209550857544,
+      "learning_rate": 5.286202090317783e-07,
+      "loss": 0.7643,
+      "step": 52621
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.750556230545044,
+      "learning_rate": 5.285022240128579e-07,
+      "loss": 0.4047,
+      "step": 52622
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.643425941467285,
+      "learning_rate": 5.283842514274989e-07,
+      "loss": 0.3941,
+      "step": 52623
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.622157573699951,
+      "learning_rate": 5.282662912760266e-07,
+      "loss": 0.6128,
+      "step": 52624
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.2693285942077637,
+      "learning_rate": 5.281483435587725e-07,
+      "loss": 0.6672,
+      "step": 52625
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.894746780395508,
+      "learning_rate": 5.280304082760602e-07,
+      "loss": 0.7278,
+      "step": 52626
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.8929543495178223,
+      "learning_rate": 5.279124854282219e-07,
+      "loss": 0.6355,
+      "step": 52627
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.84855055809021,
+      "learning_rate": 5.277945750155827e-07,
+      "loss": 0.6905,
+      "step": 52628
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.74410343170166,
+      "learning_rate": 5.276766770384728e-07,
+      "loss": 0.4912,
+      "step": 52629
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.5730884075164795,
+      "learning_rate": 5.275587914972175e-07,
+      "loss": 0.6064,
+      "step": 52630
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.2970335483551025,
+      "learning_rate": 5.274409183921469e-07,
+      "loss": 0.5304,
+      "step": 52631
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.8245034217834473,
+      "learning_rate": 5.273230577235866e-07,
+      "loss": 0.5708,
+      "step": 52632
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.778531551361084,
+      "learning_rate": 5.272052094918662e-07,
+      "loss": 0.4725,
+      "step": 52633
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.1722371578216553,
+      "learning_rate": 5.270873736973114e-07,
+      "loss": 0.6443,
+      "step": 52634
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.3059580326080322,
+      "learning_rate": 5.269695503402523e-07,
+      "loss": 0.5341,
+      "step": 52635
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.1690549850463867,
+      "learning_rate": 5.268517394210143e-07,
+      "loss": 0.5823,
+      "step": 52636
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.665391683578491,
+      "learning_rate": 5.267339409399269e-07,
+      "loss": 0.6473,
+      "step": 52637
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.755225419998169,
+      "learning_rate": 5.266161548973159e-07,
+      "loss": 0.4669,
+      "step": 52638
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.452613115310669,
+      "learning_rate": 5.264983812935104e-07,
+      "loss": 0.6029,
+      "step": 52639
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.426602840423584,
+      "learning_rate": 5.263806201288358e-07,
+      "loss": 0.6574,
+      "step": 52640
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.3904993534088135,
+      "learning_rate": 5.262628714036223e-07,
+      "loss": 0.5177,
+      "step": 52641
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.3641955852508545,
+      "learning_rate": 5.261451351181945e-07,
+      "loss": 0.4509,
+      "step": 52642
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.5938706398010254,
+      "learning_rate": 5.260274112728813e-07,
+      "loss": 0.5155,
+      "step": 52643
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.905388355255127,
+      "learning_rate": 5.259096998680107e-07,
+      "loss": 0.6173,
+      "step": 52644
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.8749701976776123,
+      "learning_rate": 5.257920009039086e-07,
+      "loss": 0.5268,
+      "step": 52645
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.1934330463409424,
+      "learning_rate": 5.256743143809034e-07,
+      "loss": 0.5313,
+      "step": 52646
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.931511163711548,
+      "learning_rate": 5.255566402993212e-07,
+      "loss": 0.598,
+      "step": 52647
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.335298776626587,
+      "learning_rate": 5.254389786594899e-07,
+      "loss": 0.7745,
+      "step": 52648
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 4.282130718231201,
+      "learning_rate": 5.253213294617371e-07,
+      "loss": 0.4748,
+      "step": 52649
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.8830137252807617,
+      "learning_rate": 5.252036927063886e-07,
+      "loss": 0.3215,
+      "step": 52650
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.876410722732544,
+      "learning_rate": 5.250860683937737e-07,
+      "loss": 0.7676,
+      "step": 52651
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.102224826812744,
+      "learning_rate": 5.249684565242169e-07,
+      "loss": 0.4164,
+      "step": 52652
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.6088972091674805,
+      "learning_rate": 5.248508570980471e-07,
+      "loss": 0.6041,
+      "step": 52653
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.428818941116333,
+      "learning_rate": 5.247332701155899e-07,
+      "loss": 0.6133,
+      "step": 52654
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 1.980103611946106,
+      "learning_rate": 5.246156955771731e-07,
+      "loss": 0.4394,
+      "step": 52655
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.7984464168548584,
+      "learning_rate": 5.244981334831245e-07,
+      "loss": 0.4563,
+      "step": 52656
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.735973834991455,
+      "learning_rate": 5.24380583833769e-07,
+      "loss": 0.3612,
+      "step": 52657
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.5596330165863037,
+      "learning_rate": 5.242630466294352e-07,
+      "loss": 0.6371,
+      "step": 52658
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.9412877559661865,
+      "learning_rate": 5.241455218704483e-07,
+      "loss": 0.578,
+      "step": 52659
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.72713565826416,
+      "learning_rate": 5.240280095571377e-07,
+      "loss": 0.639,
+      "step": 52660
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.7274019718170166,
+      "learning_rate": 5.239105096898267e-07,
+      "loss": 0.5501,
+      "step": 52661
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 1.849073052406311,
+      "learning_rate": 5.237930222688447e-07,
+      "loss": 0.5451,
+      "step": 52662
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.234614133834839,
+      "learning_rate": 5.23675547294517e-07,
+      "loss": 0.6557,
+      "step": 52663
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.892894983291626,
+      "learning_rate": 5.235580847671718e-07,
+      "loss": 0.6168,
+      "step": 52664
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.188448429107666,
+      "learning_rate": 5.234406346871335e-07,
+      "loss": 0.5856,
+      "step": 52665
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.4309568405151367,
+      "learning_rate": 5.233231970547303e-07,
+      "loss": 0.5484,
+      "step": 52666
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.6405014991760254,
+      "learning_rate": 5.232057718702871e-07,
+      "loss": 0.6025,
+      "step": 52667
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 4.859018802642822,
+      "learning_rate": 5.230883591341329e-07,
+      "loss": 0.7062,
+      "step": 52668
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.020312547683716,
+      "learning_rate": 5.229709588465925e-07,
+      "loss": 0.5965,
+      "step": 52669
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.558406114578247,
+      "learning_rate": 5.228535710079935e-07,
+      "loss": 0.642,
+      "step": 52670
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.4212422370910645,
+      "learning_rate": 5.227361956186605e-07,
+      "loss": 0.5902,
+      "step": 52671
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.102813482284546,
+      "learning_rate": 5.226188326789217e-07,
+      "loss": 0.465,
+      "step": 52672
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.0001492500305176,
+      "learning_rate": 5.225014821891016e-07,
+      "loss": 0.5369,
+      "step": 52673
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.971461296081543,
+      "learning_rate": 5.223841441495292e-07,
+      "loss": 0.6499,
+      "step": 52674
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.294027090072632,
+      "learning_rate": 5.222668185605278e-07,
+      "loss": 0.4919,
+      "step": 52675
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.60373592376709,
+      "learning_rate": 5.221495054224257e-07,
+      "loss": 0.7403,
+      "step": 52676
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.4937973022460938,
+      "learning_rate": 5.22032204735548e-07,
+      "loss": 0.3176,
+      "step": 52677
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.4211530685424805,
+      "learning_rate": 5.219149165002219e-07,
+      "loss": 0.743,
+      "step": 52678
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.9366300106048584,
+      "learning_rate": 5.217976407167718e-07,
+      "loss": 0.6429,
+      "step": 52679
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.3408262729644775,
+      "learning_rate": 5.216803773855256e-07,
+      "loss": 0.6688,
+      "step": 52680
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.5253489017486572,
+      "learning_rate": 5.215631265068083e-07,
+      "loss": 0.4679,
+      "step": 52681
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.6015424728393555,
+      "learning_rate": 5.21445888080947e-07,
+      "loss": 0.5042,
+      "step": 52682
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.94805908203125,
+      "learning_rate": 5.213286621082658e-07,
+      "loss": 0.4325,
+      "step": 52683
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.408533811569214,
+      "learning_rate": 5.212114485890934e-07,
+      "loss": 0.5677,
+      "step": 52684
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 7.4522223472595215,
+      "learning_rate": 5.210942475237524e-07,
+      "loss": 0.731,
+      "step": 52685
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.3927865028381348,
+      "learning_rate": 5.209770589125718e-07,
+      "loss": 0.6368,
+      "step": 52686
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.602015256881714,
+      "learning_rate": 5.208598827558742e-07,
+      "loss": 0.55,
+      "step": 52687
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.043605089187622,
+      "learning_rate": 5.207427190539894e-07,
+      "loss": 0.6483,
+      "step": 52688
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.114835023880005,
+      "learning_rate": 5.206255678072402e-07,
+      "loss": 0.6122,
+      "step": 52689
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.6323466300964355,
+      "learning_rate": 5.20508429015954e-07,
+      "loss": 0.4107,
+      "step": 52690
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.261601686477661,
+      "learning_rate": 5.203913026804553e-07,
+      "loss": 0.526,
+      "step": 52691
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.1844043731689453,
+      "learning_rate": 5.202741888010709e-07,
+      "loss": 0.7548,
+      "step": 52692
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.343751907348633,
+      "learning_rate": 5.201570873781247e-07,
+      "loss": 0.615,
+      "step": 52693
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.796412229537964,
+      "learning_rate": 5.200399984119448e-07,
+      "loss": 0.5667,
+      "step": 52694
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 5.722910404205322,
+      "learning_rate": 5.19922921902854e-07,
+      "loss": 0.3592,
+      "step": 52695
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 4.8320488929748535,
+      "learning_rate": 5.198058578511805e-07,
+      "loss": 0.6766,
+      "step": 52696
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 1.971079707145691,
+      "learning_rate": 5.196888062572475e-07,
+      "loss": 0.5395,
+      "step": 52697
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 4.978667259216309,
+      "learning_rate": 5.195717671213829e-07,
+      "loss": 0.6125,
+      "step": 52698
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.699047803878784,
+      "learning_rate": 5.194547404439093e-07,
+      "loss": 0.3794,
+      "step": 52699
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.838834524154663,
+      "learning_rate": 5.193377262251553e-07,
+      "loss": 0.7025,
+      "step": 52700
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.8156590461730957,
+      "learning_rate": 5.19220724465443e-07,
+      "loss": 0.5948,
+      "step": 52701
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 4.266236782073975,
+      "learning_rate": 5.191037351651007e-07,
+      "loss": 0.7614,
+      "step": 52702
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 5.173490047454834,
+      "learning_rate": 5.189867583244513e-07,
+      "loss": 0.4201,
+      "step": 52703
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 1.9816640615463257,
+      "learning_rate": 5.188697939438209e-07,
+      "loss": 0.3663,
+      "step": 52704
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.8388681411743164,
+      "learning_rate": 5.18752842023536e-07,
+      "loss": 0.4282,
+      "step": 52705
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.2431905269622803,
+      "learning_rate": 5.186359025639193e-07,
+      "loss": 0.4041,
+      "step": 52706
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.362070322036743,
+      "learning_rate": 5.185189755652992e-07,
+      "loss": 0.6236,
+      "step": 52707
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.9700636863708496,
+      "learning_rate": 5.184020610279978e-07,
+      "loss": 0.5879,
+      "step": 52708
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.446747303009033,
+      "learning_rate": 5.182851589523408e-07,
+      "loss": 0.6577,
+      "step": 52709
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.6329429149627686,
+      "learning_rate": 5.181682693386553e-07,
+      "loss": 0.5398,
+      "step": 52710
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 1.7621588706970215,
+      "learning_rate": 5.180513921872643e-07,
+      "loss": 0.4247,
+      "step": 52711
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.3081977367401123,
+      "learning_rate": 5.179345274984938e-07,
+      "loss": 0.7535,
+      "step": 52712
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.4724669456481934,
+      "learning_rate": 5.178176752726677e-07,
+      "loss": 0.3539,
+      "step": 52713
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.9408295154571533,
+      "learning_rate": 5.177008355101127e-07,
+      "loss": 0.7809,
+      "step": 52714
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.6050374507904053,
+      "learning_rate": 5.175840082111511e-07,
+      "loss": 0.6462,
+      "step": 52715
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 4.213013648986816,
+      "learning_rate": 5.174671933761094e-07,
+      "loss": 0.7054,
+      "step": 52716
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.6740942001342773,
+      "learning_rate": 5.173503910053135e-07,
+      "loss": 0.5534,
+      "step": 52717
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.5811080932617188,
+      "learning_rate": 5.172336010990858e-07,
+      "loss": 0.5005,
+      "step": 52718
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.4018728733062744,
+      "learning_rate": 5.171168236577534e-07,
+      "loss": 0.4317,
+      "step": 52719
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.8533520698547363,
+      "learning_rate": 5.170000586816382e-07,
+      "loss": 0.5765,
+      "step": 52720
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.068020820617676,
+      "learning_rate": 5.168833061710676e-07,
+      "loss": 0.6815,
+      "step": 52721
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.0773136615753174,
+      "learning_rate": 5.167665661263643e-07,
+      "loss": 0.4344,
+      "step": 52722
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.4746899604797363,
+      "learning_rate": 5.166498385478552e-07,
+      "loss": 0.6321,
+      "step": 52723
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 1.9880813360214233,
+      "learning_rate": 5.165331234358617e-07,
+      "loss": 0.2738,
+      "step": 52724
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.621178388595581,
+      "learning_rate": 5.164164207907113e-07,
+      "loss": 0.7093,
+      "step": 52725
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.8525197505950928,
+      "learning_rate": 5.16299730612726e-07,
+      "loss": 0.5193,
+      "step": 52726
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.5314314365386963,
+      "learning_rate": 5.161830529022327e-07,
+      "loss": 0.6442,
+      "step": 52727
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.5282204151153564,
+      "learning_rate": 5.160663876595529e-07,
+      "loss": 0.5891,
+      "step": 52728
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.031405210494995,
+      "learning_rate": 5.159497348850145e-07,
+      "loss": 0.51,
+      "step": 52729
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.3891398906707764,
+      "learning_rate": 5.158330945789392e-07,
+      "loss": 0.7265,
+      "step": 52730
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 4.977895736694336,
+      "learning_rate": 5.157164667416531e-07,
+      "loss": 0.6069,
+      "step": 52731
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.214725971221924,
+      "learning_rate": 5.155998513734789e-07,
+      "loss": 0.6786,
+      "step": 52732
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 5.302492618560791,
+      "learning_rate": 5.154832484747424e-07,
+      "loss": 0.7347,
+      "step": 52733
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 6.352992534637451,
+      "learning_rate": 5.153666580457662e-07,
+      "loss": 0.8381,
+      "step": 52734
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.6260123252868652,
+      "learning_rate": 5.152500800868765e-07,
+      "loss": 0.475,
+      "step": 52735
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.960388660430908,
+      "learning_rate": 5.151335145983949e-07,
+      "loss": 0.5035,
+      "step": 52736
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.4590070247650146,
+      "learning_rate": 5.150169615806477e-07,
+      "loss": 0.5744,
+      "step": 52737
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.3543248176574707,
+      "learning_rate": 5.149004210339581e-07,
+      "loss": 0.8286,
+      "step": 52738
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.074781894683838,
+      "learning_rate": 5.147838929586508e-07,
+      "loss": 0.4258,
+      "step": 52739
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 5.245769500732422,
+      "learning_rate": 5.146673773550481e-07,
+      "loss": 0.5587,
+      "step": 52740
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.184722900390625,
+      "learning_rate": 5.145508742234762e-07,
+      "loss": 0.5104,
+      "step": 52741
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.7275235652923584,
+      "learning_rate": 5.144343835642573e-07,
+      "loss": 0.6795,
+      "step": 52742
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.5422921180725098,
+      "learning_rate": 5.14317905377717e-07,
+      "loss": 0.358,
+      "step": 52743
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.06901216506958,
+      "learning_rate": 5.14201439664177e-07,
+      "loss": 0.5824,
+      "step": 52744
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.7621238231658936,
+      "learning_rate": 5.140849864239639e-07,
+      "loss": 0.3757,
+      "step": 52745
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.209016799926758,
+      "learning_rate": 5.139685456573984e-07,
+      "loss": 0.63,
+      "step": 52746
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.9753174781799316,
+      "learning_rate": 5.13852117364807e-07,
+      "loss": 0.6042,
+      "step": 52747
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.6552977561950684,
+      "learning_rate": 5.137357015465105e-07,
+      "loss": 0.5923,
+      "step": 52748
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.607806444168091,
+      "learning_rate": 5.136192982028365e-07,
+      "loss": 0.4252,
+      "step": 52749
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.6270344257354736,
+      "learning_rate": 5.135029073341053e-07,
+      "loss": 0.5527,
+      "step": 52750
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.8680784702301025,
+      "learning_rate": 5.133865289406431e-07,
+      "loss": 0.6537,
+      "step": 52751
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 4.026362895965576,
+      "learning_rate": 5.13270163022771e-07,
+      "loss": 0.5172,
+      "step": 52752
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 1.9248439073562622,
+      "learning_rate": 5.131538095808147e-07,
+      "loss": 0.369,
+      "step": 52753
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.0907557010650635,
+      "learning_rate": 5.130374686150963e-07,
+      "loss": 0.3787,
+      "step": 52754
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.1801486015319824,
+      "learning_rate": 5.129211401259404e-07,
+      "loss": 0.5538,
+      "step": 52755
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.6546432971954346,
+      "learning_rate": 5.128048241136696e-07,
+      "loss": 0.4937,
+      "step": 52756
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.8077096939086914,
+      "learning_rate": 5.12688520578608e-07,
+      "loss": 0.5892,
+      "step": 52757
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.7356369495391846,
+      "learning_rate": 5.125722295210778e-07,
+      "loss": 0.595,
+      "step": 52758
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.4744770526885986,
+      "learning_rate": 5.12455950941404e-07,
+      "loss": 0.8378,
+      "step": 52759
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.748448371887207,
+      "learning_rate": 5.123396848399087e-07,
+      "loss": 0.4387,
+      "step": 52760
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.544180154800415,
+      "learning_rate": 5.122234312169161e-07,
+      "loss": 0.5909,
+      "step": 52761
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.87935209274292,
+      "learning_rate": 5.121071900727481e-07,
+      "loss": 0.4975,
+      "step": 52762
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.151379108428955,
+      "learning_rate": 5.119909614077295e-07,
+      "loss": 0.5865,
+      "step": 52763
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.3853211402893066,
+      "learning_rate": 5.118747452221822e-07,
+      "loss": 0.6865,
+      "step": 52764
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.7908928394317627,
+      "learning_rate": 5.117585415164311e-07,
+      "loss": 0.566,
+      "step": 52765
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.580881357192993,
+      "learning_rate": 5.116423502907969e-07,
+      "loss": 0.8017,
+      "step": 52766
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.8479576110839844,
+      "learning_rate": 5.115261715456038e-07,
+      "loss": 0.5088,
+      "step": 52767
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.1847176551818848,
+      "learning_rate": 5.11410005281176e-07,
+      "loss": 0.4544,
+      "step": 52768
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 5.988442897796631,
+      "learning_rate": 5.112938514978349e-07,
+      "loss": 0.473,
+      "step": 52769
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.8335375785827637,
+      "learning_rate": 5.111777101959042e-07,
+      "loss": 0.5544,
+      "step": 52770
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.7765305042266846,
+      "learning_rate": 5.110615813757069e-07,
+      "loss": 0.7581,
+      "step": 52771
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.4013071060180664,
+      "learning_rate": 5.109454650375656e-07,
+      "loss": 0.5454,
+      "step": 52772
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.7013773918151855,
+      "learning_rate": 5.108293611818038e-07,
+      "loss": 0.5927,
+      "step": 52773
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.2339727878570557,
+      "learning_rate": 5.10713269808743e-07,
+      "loss": 0.3956,
+      "step": 52774
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.673292636871338,
+      "learning_rate": 5.105971909187079e-07,
+      "loss": 0.5447,
+      "step": 52775
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.1910383701324463,
+      "learning_rate": 5.104811245120195e-07,
+      "loss": 0.4566,
+      "step": 52776
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.6119089126586914,
+      "learning_rate": 5.103650705890012e-07,
+      "loss": 0.4718,
+      "step": 52777
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.327510356903076,
+      "learning_rate": 5.102490291499762e-07,
+      "loss": 0.4544,
+      "step": 52778
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.8684136867523193,
+      "learning_rate": 5.101330001952665e-07,
+      "loss": 0.4856,
+      "step": 52779
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.2888782024383545,
+      "learning_rate": 5.100169837251956e-07,
+      "loss": 0.492,
+      "step": 52780
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 1.600577473640442,
+      "learning_rate": 5.099009797400845e-07,
+      "loss": 0.2968,
+      "step": 52781
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 5.575084686279297,
+      "learning_rate": 5.097849882402578e-07,
+      "loss": 0.4962,
+      "step": 52782
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 4.265759468078613,
+      "learning_rate": 5.096690092260359e-07,
+      "loss": 0.447,
+      "step": 52783
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 6.344744682312012,
+      "learning_rate": 5.095530426977436e-07,
+      "loss": 0.7362,
+      "step": 52784
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.4200220108032227,
+      "learning_rate": 5.094370886557009e-07,
+      "loss": 0.7511,
+      "step": 52785
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.9422619342803955,
+      "learning_rate": 5.093211471002324e-07,
+      "loss": 0.4374,
+      "step": 52786
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.692967176437378,
+      "learning_rate": 5.092052180316587e-07,
+      "loss": 0.5638,
+      "step": 52787
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.5844826698303223,
+      "learning_rate": 5.09089301450304e-07,
+      "loss": 0.774,
+      "step": 52788
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.075521230697632,
+      "learning_rate": 5.089733973564876e-07,
+      "loss": 0.4304,
+      "step": 52789
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.1935060024261475,
+      "learning_rate": 5.088575057505357e-07,
+      "loss": 0.5396,
+      "step": 52790
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.8805267810821533,
+      "learning_rate": 5.087416266327677e-07,
+      "loss": 0.5464,
+      "step": 52791
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.360469341278076,
+      "learning_rate": 5.086257600035077e-07,
+      "loss": 0.6473,
+      "step": 52792
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 6.024907112121582,
+      "learning_rate": 5.08509905863076e-07,
+      "loss": 0.6575,
+      "step": 52793
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.7898850440979004,
+      "learning_rate": 5.083940642117968e-07,
+      "loss": 0.6377,
+      "step": 52794
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.7893707752227783,
+      "learning_rate": 5.082782350499899e-07,
+      "loss": 0.5748,
+      "step": 52795
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.3199217319488525,
+      "learning_rate": 5.081624183779799e-07,
+      "loss": 0.26,
+      "step": 52796
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 4.415125846862793,
+      "learning_rate": 5.080466141960866e-07,
+      "loss": 0.6596,
+      "step": 52797
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.44779634475708,
+      "learning_rate": 5.079308225046337e-07,
+      "loss": 0.5977,
+      "step": 52798
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.6652512550354004,
+      "learning_rate": 5.078150433039413e-07,
+      "loss": 0.6077,
+      "step": 52799
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.932403802871704,
+      "learning_rate": 5.07699276594334e-07,
+      "loss": 0.5377,
+      "step": 52800
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 4.103146076202393,
+      "learning_rate": 5.075835223761306e-07,
+      "loss": 0.6609,
+      "step": 52801
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.8991549015045166,
+      "learning_rate": 5.074677806496553e-07,
+      "loss": 0.5488,
+      "step": 52802
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 4.222811698913574,
+      "learning_rate": 5.073520514152286e-07,
+      "loss": 0.595,
+      "step": 52803
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.137286424636841,
+      "learning_rate": 5.07236334673174e-07,
+      "loss": 0.5606,
+      "step": 52804
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 1.8807575702667236,
+      "learning_rate": 5.071206304238108e-07,
+      "loss": 0.4902,
+      "step": 52805
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.2228710651397705,
+      "learning_rate": 5.07004938667463e-07,
+      "loss": 0.5613,
+      "step": 52806
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 4.3941826820373535,
+      "learning_rate": 5.068892594044511e-07,
+      "loss": 0.7977,
+      "step": 52807
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 10.116150856018066,
+      "learning_rate": 5.067735926350969e-07,
+      "loss": 0.4613,
+      "step": 52808
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.0128989219665527,
+      "learning_rate": 5.066579383597209e-07,
+      "loss": 0.6626,
+      "step": 52809
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.1318230628967285,
+      "learning_rate": 5.065422965786476e-07,
+      "loss": 0.8803,
+      "step": 52810
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.6324195861816406,
+      "learning_rate": 5.064266672921964e-07,
+      "loss": 0.6319,
+      "step": 52811
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.144366502761841,
+      "learning_rate": 5.063110505006896e-07,
+      "loss": 0.4835,
+      "step": 52812
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.1189870834350586,
+      "learning_rate": 5.061954462044477e-07,
+      "loss": 0.6723,
+      "step": 52813
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.5951874256134033,
+      "learning_rate": 5.060798544037943e-07,
+      "loss": 0.558,
+      "step": 52814
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.3881752490997314,
+      "learning_rate": 5.059642750990479e-07,
+      "loss": 0.4367,
+      "step": 52815
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 4.024661064147949,
+      "learning_rate": 5.058487082905323e-07,
+      "loss": 0.4628,
+      "step": 52816
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.833963632583618,
+      "learning_rate": 5.057331539785671e-07,
+      "loss": 0.5512,
+      "step": 52817
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.9264607429504395,
+      "learning_rate": 5.056176121634748e-07,
+      "loss": 0.5935,
+      "step": 52818
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.005070447921753,
+      "learning_rate": 5.055020828455758e-07,
+      "loss": 0.5423,
+      "step": 52819
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.846848487854004,
+      "learning_rate": 5.053865660251928e-07,
+      "loss": 0.6416,
+      "step": 52820
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.5017080307006836,
+      "learning_rate": 5.052710617026452e-07,
+      "loss": 0.548,
+      "step": 52821
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 1.8570247888565063,
+      "learning_rate": 5.051555698782557e-07,
+      "loss": 0.4884,
+      "step": 52822
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.2051563262939453,
+      "learning_rate": 5.050400905523439e-07,
+      "loss": 0.4065,
+      "step": 52823
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.3503823280334473,
+      "learning_rate": 5.049246237252326e-07,
+      "loss": 0.5024,
+      "step": 52824
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.469381093978882,
+      "learning_rate": 5.048091693972412e-07,
+      "loss": 0.5094,
+      "step": 52825
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.9888391494750977,
+      "learning_rate": 5.046937275686925e-07,
+      "loss": 0.603,
+      "step": 52826
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.810967206954956,
+      "learning_rate": 5.045782982399055e-07,
+      "loss": 0.6001,
+      "step": 52827
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.733222723007202,
+      "learning_rate": 5.044628814112023e-07,
+      "loss": 0.5689,
+      "step": 52828
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.7897567749023438,
+      "learning_rate": 5.043474770829049e-07,
+      "loss": 0.2385,
+      "step": 52829
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.280613899230957,
+      "learning_rate": 5.042320852553323e-07,
+      "loss": 0.3464,
+      "step": 52830
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.9675228595733643,
+      "learning_rate": 5.041167059288055e-07,
+      "loss": 0.8609,
+      "step": 52831
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.3689608573913574,
+      "learning_rate": 5.040013391036475e-07,
+      "loss": 0.4544,
+      "step": 52832
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.013516426086426,
+      "learning_rate": 5.038859847801758e-07,
+      "loss": 0.5334,
+      "step": 52833
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 4.114360809326172,
+      "learning_rate": 5.037706429587147e-07,
+      "loss": 0.7632,
+      "step": 52834
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.0417213439941406,
+      "learning_rate": 5.036553136395816e-07,
+      "loss": 0.6566,
+      "step": 52835
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.6995160579681396,
+      "learning_rate": 5.035399968230998e-07,
+      "loss": 0.5998,
+      "step": 52836
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.734531879425049,
+      "learning_rate": 5.034246925095882e-07,
+      "loss": 0.7836,
+      "step": 52837
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.2768070697784424,
+      "learning_rate": 5.033094006993689e-07,
+      "loss": 0.2261,
+      "step": 52838
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.0110347270965576,
+      "learning_rate": 5.031941213927605e-07,
+      "loss": 0.3389,
+      "step": 52839
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.088409423828125,
+      "learning_rate": 5.030788545900844e-07,
+      "loss": 0.729,
+      "step": 52840
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.91270112991333,
+      "learning_rate": 5.029636002916626e-07,
+      "loss": 0.4553,
+      "step": 52841
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.212574005126953,
+      "learning_rate": 5.028483584978139e-07,
+      "loss": 0.4312,
+      "step": 52842
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.3292179107666016,
+      "learning_rate": 5.0273312920886e-07,
+      "loss": 0.513,
+      "step": 52843
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.0813987255096436,
+      "learning_rate": 5.026179124251196e-07,
+      "loss": 0.6511,
+      "step": 52844
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.7362513542175293,
+      "learning_rate": 5.025027081469147e-07,
+      "loss": 0.8303,
+      "step": 52845
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.5547659397125244,
+      "learning_rate": 5.02387516374564e-07,
+      "loss": 0.5474,
+      "step": 52846
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.5654804706573486,
+      "learning_rate": 5.022723371083899e-07,
+      "loss": 0.5246,
+      "step": 52847
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.8532826900482178,
+      "learning_rate": 5.021571703487104e-07,
+      "loss": 0.7468,
+      "step": 52848
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.385408878326416,
+      "learning_rate": 5.020420160958483e-07,
+      "loss": 0.6119,
+      "step": 52849
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.5751616954803467,
+      "learning_rate": 5.01926874350121e-07,
+      "loss": 0.4856,
+      "step": 52850
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.098829507827759,
+      "learning_rate": 5.018117451118504e-07,
+      "loss": 0.5511,
+      "step": 52851
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.824993133544922,
+      "learning_rate": 5.016966283813562e-07,
+      "loss": 0.4929,
+      "step": 52852
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.922245502471924,
+      "learning_rate": 5.015815241589595e-07,
+      "loss": 0.6414,
+      "step": 52853
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.832850456237793,
+      "learning_rate": 5.01466432444978e-07,
+      "loss": 0.7545,
+      "step": 52854
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.1271779537200928,
+      "learning_rate": 5.013513532397346e-07,
+      "loss": 0.442,
+      "step": 52855
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 1.7112430334091187,
+      "learning_rate": 5.012362865435472e-07,
+      "loss": 0.2535,
+      "step": 52856
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.505524158477783,
+      "learning_rate": 5.011212323567366e-07,
+      "loss": 0.5424,
+      "step": 52857
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.58599853515625,
+      "learning_rate": 5.010061906796221e-07,
+      "loss": 0.6021,
+      "step": 52858
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.3248298168182373,
+      "learning_rate": 5.008911615125251e-07,
+      "loss": 0.5634,
+      "step": 52859
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.4194159507751465,
+      "learning_rate": 5.00776144855763e-07,
+      "loss": 0.6887,
+      "step": 52860
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.3856372833251953,
+      "learning_rate": 5.006611407096579e-07,
+      "loss": 0.4539,
+      "step": 52861
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.2889597415924072,
+      "learning_rate": 5.005461490745278e-07,
+      "loss": 0.4494,
+      "step": 52862
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.100121021270752,
+      "learning_rate": 5.004311699506947e-07,
+      "loss": 0.7015,
+      "step": 52863
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.241229772567749,
+      "learning_rate": 5.003162033384756e-07,
+      "loss": 0.447,
+      "step": 52864
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.377915143966675,
+      "learning_rate": 5.002012492381925e-07,
+      "loss": 0.4163,
+      "step": 52865
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.2930943965911865,
+      "learning_rate": 5.000863076501633e-07,
+      "loss": 0.6642,
+      "step": 52866
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 1.7934547662734985,
+      "learning_rate": 4.99971378574709e-07,
+      "loss": 0.4564,
+      "step": 52867
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.0053274631500244,
+      "learning_rate": 4.998564620121476e-07,
+      "loss": 0.5875,
+      "step": 52868
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.9426283836364746,
+      "learning_rate": 4.997415579628012e-07,
+      "loss": 0.6594,
+      "step": 52869
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.507150888442993,
+      "learning_rate": 4.99626666426985e-07,
+      "loss": 0.7297,
+      "step": 52870
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.0921854972839355,
+      "learning_rate": 4.995117874050237e-07,
+      "loss": 0.5964,
+      "step": 52871
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.8782479763031006,
+      "learning_rate": 4.993969208972327e-07,
+      "loss": 0.6041,
+      "step": 52872
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.6430728435516357,
+      "learning_rate": 4.992820669039339e-07,
+      "loss": 0.5751,
+      "step": 52873
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.740849018096924,
+      "learning_rate": 4.991672254254448e-07,
+      "loss": 0.4266,
+      "step": 52874
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.434720754623413,
+      "learning_rate": 4.990523964620869e-07,
+      "loss": 0.5248,
+      "step": 52875
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.7281360626220703,
+      "learning_rate": 4.989375800141772e-07,
+      "loss": 0.533,
+      "step": 52876
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.9184932708740234,
+      "learning_rate": 4.988227760820363e-07,
+      "loss": 0.56,
+      "step": 52877
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.786487579345703,
+      "learning_rate": 4.987079846659826e-07,
+      "loss": 0.7435,
+      "step": 52878
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 4.297891616821289,
+      "learning_rate": 4.985932057663368e-07,
+      "loss": 0.6485,
+      "step": 52879
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.5255377292633057,
+      "learning_rate": 4.984784393834158e-07,
+      "loss": 0.5031,
+      "step": 52880
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 4.1276750564575195,
+      "learning_rate": 4.983636855175411e-07,
+      "loss": 0.6447,
+      "step": 52881
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.2317466735839844,
+      "learning_rate": 4.982489441690297e-07,
+      "loss": 0.6382,
+      "step": 52882
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.507819652557373,
+      "learning_rate": 4.981342153382024e-07,
+      "loss": 0.6876,
+      "step": 52883
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.6409997940063477,
+      "learning_rate": 4.980194990253762e-07,
+      "loss": 0.5532,
+      "step": 52884
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.2436680793762207,
+      "learning_rate": 4.97904795230873e-07,
+      "loss": 0.4533,
+      "step": 52885
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.577761173248291,
+      "learning_rate": 4.977901039550082e-07,
+      "loss": 0.5354,
+      "step": 52886
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.9070332050323486,
+      "learning_rate": 4.976754251981042e-07,
+      "loss": 0.6316,
+      "step": 52887
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.5911996364593506,
+      "learning_rate": 4.97560758960477e-07,
+      "loss": 0.5496,
+      "step": 52888
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.3319146633148193,
+      "learning_rate": 4.974461052424467e-07,
+      "loss": 0.7078,
+      "step": 52889
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.2667360305786133,
+      "learning_rate": 4.973314640443327e-07,
+      "loss": 0.6749,
+      "step": 52890
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.5980117321014404,
+      "learning_rate": 4.972168353664531e-07,
+      "loss": 0.5085,
+      "step": 52891
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.0989086627960205,
+      "learning_rate": 4.971022192091257e-07,
+      "loss": 0.5264,
+      "step": 52892
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.8475136756896973,
+      "learning_rate": 4.969876155726716e-07,
+      "loss": 0.7826,
+      "step": 52893
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.6711204051971436,
+      "learning_rate": 4.968730244574071e-07,
+      "loss": 0.815,
+      "step": 52894
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.651942729949951,
+      "learning_rate": 4.967584458636532e-07,
+      "loss": 0.6043,
+      "step": 52895
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.6198978424072266,
+      "learning_rate": 4.966438797917256e-07,
+      "loss": 0.7417,
+      "step": 52896
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.7822513580322266,
+      "learning_rate": 4.965293262419451e-07,
+      "loss": 0.3957,
+      "step": 52897
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.7373385429382324,
+      "learning_rate": 4.964147852146294e-07,
+      "loss": 0.572,
+      "step": 52898
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.927316427230835,
+      "learning_rate": 4.963002567100971e-07,
+      "loss": 0.5403,
+      "step": 52899
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.8556110858917236,
+      "learning_rate": 4.961857407286663e-07,
+      "loss": 0.3662,
+      "step": 52900
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.37355375289917,
+      "learning_rate": 4.960712372706561e-07,
+      "loss": 0.4739,
+      "step": 52901
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.631807565689087,
+      "learning_rate": 4.959567463363852e-07,
+      "loss": 0.6407,
+      "step": 52902
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.881805419921875,
+      "learning_rate": 4.958422679261698e-07,
+      "loss": 0.4629,
+      "step": 52903
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.795973062515259,
+      "learning_rate": 4.957278020403311e-07,
+      "loss": 0.6812,
+      "step": 52904
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.9617888927459717,
+      "learning_rate": 4.956133486791853e-07,
+      "loss": 0.5066,
+      "step": 52905
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 1.5421618223190308,
+      "learning_rate": 4.954989078430522e-07,
+      "loss": 0.4123,
+      "step": 52906
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.570016384124756,
+      "learning_rate": 4.953844795322483e-07,
+      "loss": 0.7173,
+      "step": 52907
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 4.3484206199646,
+      "learning_rate": 4.952700637470936e-07,
+      "loss": 0.75,
+      "step": 52908
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.454176902770996,
+      "learning_rate": 4.951556604879049e-07,
+      "loss": 0.7367,
+      "step": 52909
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.5131497383117676,
+      "learning_rate": 4.950412697550011e-07,
+      "loss": 0.5403,
+      "step": 52910
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 4.660999774932861,
+      "learning_rate": 4.949268915486988e-07,
+      "loss": 0.6771,
+      "step": 52911
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.4849634170532227,
+      "learning_rate": 4.948125258693176e-07,
+      "loss": 0.567,
+      "step": 52912
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.2369513511657715,
+      "learning_rate": 4.946981727171751e-07,
+      "loss": 0.5804,
+      "step": 52913
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.1547458171844482,
+      "learning_rate": 4.9458383209259e-07,
+      "loss": 0.4842,
+      "step": 52914
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.384127378463745,
+      "learning_rate": 4.944695039958786e-07,
+      "loss": 0.5168,
+      "step": 52915
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.274420976638794,
+      "learning_rate": 4.943551884273607e-07,
+      "loss": 0.4629,
+      "step": 52916
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.663818120956421,
+      "learning_rate": 4.94240885387352e-07,
+      "loss": 0.6965,
+      "step": 52917
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 1.9221090078353882,
+      "learning_rate": 4.941265948761731e-07,
+      "loss": 0.439,
+      "step": 52918
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 13.834433555603027,
+      "learning_rate": 4.940123168941385e-07,
+      "loss": 0.5269,
+      "step": 52919
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.9576053619384766,
+      "learning_rate": 4.938980514415693e-07,
+      "loss": 0.41,
+      "step": 52920
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.5993599891662598,
+      "learning_rate": 4.937837985187799e-07,
+      "loss": 0.5551,
+      "step": 52921
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.8125216960906982,
+      "learning_rate": 4.936695581260908e-07,
+      "loss": 0.6139,
+      "step": 52922
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.8852553367614746,
+      "learning_rate": 4.935553302638179e-07,
+      "loss": 0.5713,
+      "step": 52923
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.0614802837371826,
+      "learning_rate": 4.934411149322804e-07,
+      "loss": 0.2737,
+      "step": 52924
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 8.319494247436523,
+      "learning_rate": 4.933269121317941e-07,
+      "loss": 0.4804,
+      "step": 52925
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.5820107460021973,
+      "learning_rate": 4.932127218626782e-07,
+      "loss": 0.4822,
+      "step": 52926
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.19496750831604,
+      "learning_rate": 4.930985441252483e-07,
+      "loss": 0.6659,
+      "step": 52927
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.0838942527770996,
+      "learning_rate": 4.929843789198241e-07,
+      "loss": 0.5912,
+      "step": 52928
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.894691228866577,
+      "learning_rate": 4.928702262467211e-07,
+      "loss": 0.6172,
+      "step": 52929
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.508823871612549,
+      "learning_rate": 4.927560861062586e-07,
+      "loss": 0.6151,
+      "step": 52930
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.1315832138061523,
+      "learning_rate": 4.926419584987513e-07,
+      "loss": 0.8269,
+      "step": 52931
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.7262473106384277,
+      "learning_rate": 4.925278434245195e-07,
+      "loss": 0.4656,
+      "step": 52932
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.240778923034668,
+      "learning_rate": 4.924137408838786e-07,
+      "loss": 0.7121,
+      "step": 52933
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.0672945976257324,
+      "learning_rate": 4.922996508771478e-07,
+      "loss": 0.6519,
+      "step": 52934
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.916297197341919,
+      "learning_rate": 4.921855734046416e-07,
+      "loss": 0.5311,
+      "step": 52935
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.0393669605255127,
+      "learning_rate": 4.920715084666799e-07,
+      "loss": 0.6671,
+      "step": 52936
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.38728666305542,
+      "learning_rate": 4.919574560635776e-07,
+      "loss": 0.4514,
+      "step": 52937
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.2545418739318848,
+      "learning_rate": 4.918434161956536e-07,
+      "loss": 0.6633,
+      "step": 52938
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.5228657722473145,
+      "learning_rate": 4.917293888632235e-07,
+      "loss": 0.512,
+      "step": 52939
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.3062331676483154,
+      "learning_rate": 4.916153740666063e-07,
+      "loss": 0.3467,
+      "step": 52940
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.6273982524871826,
+      "learning_rate": 4.915013718061163e-07,
+      "loss": 0.5259,
+      "step": 52941
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.081697702407837,
+      "learning_rate": 4.913873820820736e-07,
+      "loss": 0.4903,
+      "step": 52942
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 6.783285140991211,
+      "learning_rate": 4.912734048947926e-07,
+      "loss": 0.6925,
+      "step": 52943
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.171168088912964,
+      "learning_rate": 4.911594402445919e-07,
+      "loss": 0.6806,
+      "step": 52944
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.731550931930542,
+      "learning_rate": 4.910454881317867e-07,
+      "loss": 0.7415,
+      "step": 52945
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.0932681560516357,
+      "learning_rate": 4.909315485566962e-07,
+      "loss": 0.5682,
+      "step": 52946
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.221503734588623,
+      "learning_rate": 4.908176215196347e-07,
+      "loss": 0.6949,
+      "step": 52947
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.455352544784546,
+      "learning_rate": 4.907037070209214e-07,
+      "loss": 0.7075,
+      "step": 52948
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.0644867420196533,
+      "learning_rate": 4.905898050608709e-07,
+      "loss": 0.5684,
+      "step": 52949
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.598562240600586,
+      "learning_rate": 4.90475915639802e-07,
+      "loss": 0.5225,
+      "step": 52950
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.2748968601226807,
+      "learning_rate": 4.903620387580293e-07,
+      "loss": 0.849,
+      "step": 52951
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.0264968872070312,
+      "learning_rate": 4.902481744158699e-07,
+      "loss": 0.7042,
+      "step": 52952
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.1904006004333496,
+      "learning_rate": 4.901343226136412e-07,
+      "loss": 0.5302,
+      "step": 52953
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.40732741355896,
+      "learning_rate": 4.900204833516608e-07,
+      "loss": 0.5282,
+      "step": 52954
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.6498703956604004,
+      "learning_rate": 4.89906656630243e-07,
+      "loss": 0.6763,
+      "step": 52955
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.588597536087036,
+      "learning_rate": 4.897928424497056e-07,
+      "loss": 0.7623,
+      "step": 52956
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.551964521408081,
+      "learning_rate": 4.896790408103646e-07,
+      "loss": 0.6161,
+      "step": 52957
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.2011804580688477,
+      "learning_rate": 4.895652517125371e-07,
+      "loss": 0.712,
+      "step": 52958
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.5148134231567383,
+      "learning_rate": 4.894514751565377e-07,
+      "loss": 0.7287,
+      "step": 52959
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.9233345985412598,
+      "learning_rate": 4.893377111426856e-07,
+      "loss": 0.4443,
+      "step": 52960
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.1720404624938965,
+      "learning_rate": 4.892239596712945e-07,
+      "loss": 0.672,
+      "step": 52961
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.4584381580352783,
+      "learning_rate": 4.891102207426818e-07,
+      "loss": 0.6133,
+      "step": 52962
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.8901920318603516,
+      "learning_rate": 4.889964943571646e-07,
+      "loss": 0.8017,
+      "step": 52963
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.798957347869873,
+      "learning_rate": 4.88882780515057e-07,
+      "loss": 0.6649,
+      "step": 52964
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.447185754776001,
+      "learning_rate": 4.887690792166777e-07,
+      "loss": 0.4416,
+      "step": 52965
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.298412561416626,
+      "learning_rate": 4.886553904623409e-07,
+      "loss": 0.7122,
+      "step": 52966
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.3039348125457764,
+      "learning_rate": 4.885417142523641e-07,
+      "loss": 0.6222,
+      "step": 52967
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.8861746788024902,
+      "learning_rate": 4.88428050587062e-07,
+      "loss": 0.7994,
+      "step": 52968
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.0228075981140137,
+      "learning_rate": 4.883143994667527e-07,
+      "loss": 0.555,
+      "step": 52969
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.3464715480804443,
+      "learning_rate": 4.882007608917493e-07,
+      "loss": 0.4857,
+      "step": 52970
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.724376678466797,
+      "learning_rate": 4.880871348623712e-07,
+      "loss": 0.4237,
+      "step": 52971
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.9002280235290527,
+      "learning_rate": 4.879735213789311e-07,
+      "loss": 0.4194,
+      "step": 52972
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.7330877780914307,
+      "learning_rate": 4.878599204417461e-07,
+      "loss": 0.4688,
+      "step": 52973
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.966362476348877,
+      "learning_rate": 4.877463320511328e-07,
+      "loss": 0.697,
+      "step": 52974
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.094473123550415,
+      "learning_rate": 4.876327562074074e-07,
+      "loss": 0.613,
+      "step": 52975
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 1.8007599115371704,
+      "learning_rate": 4.875191929108841e-07,
+      "loss": 0.3289,
+      "step": 52976
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.4579153060913086,
+      "learning_rate": 4.874056421618806e-07,
+      "loss": 0.66,
+      "step": 52977
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.533237934112549,
+      "learning_rate": 4.872921039607104e-07,
+      "loss": 0.7878,
+      "step": 52978
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.684932231903076,
+      "learning_rate": 4.871785783076916e-07,
+      "loss": 0.4995,
+      "step": 52979
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 4.3088698387146,
+      "learning_rate": 4.870650652031372e-07,
+      "loss": 0.6354,
+      "step": 52980
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 4.952099800109863,
+      "learning_rate": 4.869515646473655e-07,
+      "loss": 0.6117,
+      "step": 52981
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.1184146404266357,
+      "learning_rate": 4.868380766406899e-07,
+      "loss": 0.6232,
+      "step": 52982
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 4.32078742980957,
+      "learning_rate": 4.867246011834275e-07,
+      "loss": 0.4217,
+      "step": 52983
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.7427589893341064,
+      "learning_rate": 4.866111382758931e-07,
+      "loss": 0.5868,
+      "step": 52984
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.096269369125366,
+      "learning_rate": 4.86497687918403e-07,
+      "loss": 0.4445,
+      "step": 52985
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.7603209018707275,
+      "learning_rate": 4.863842501112709e-07,
+      "loss": 0.5987,
+      "step": 52986
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.090440273284912,
+      "learning_rate": 4.862708248548143e-07,
+      "loss": 0.5256,
+      "step": 52987
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.476470947265625,
+      "learning_rate": 4.861574121493467e-07,
+      "loss": 0.5456,
+      "step": 52988
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 5.480643272399902,
+      "learning_rate": 4.860440119951853e-07,
+      "loss": 0.5995,
+      "step": 52989
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 5.367238998413086,
+      "learning_rate": 4.859306243926442e-07,
+      "loss": 0.5351,
+      "step": 52990
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.778646469116211,
+      "learning_rate": 4.858172493420393e-07,
+      "loss": 0.7261,
+      "step": 52991
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.067847967147827,
+      "learning_rate": 4.857038868436842e-07,
+      "loss": 0.4216,
+      "step": 52992
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.6129393577575684,
+      "learning_rate": 4.855905368978969e-07,
+      "loss": 0.4355,
+      "step": 52993
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.22760009765625,
+      "learning_rate": 4.854771995049906e-07,
+      "loss": 0.48,
+      "step": 52994
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 1.8988724946975708,
+      "learning_rate": 4.853638746652817e-07,
+      "loss": 0.5061,
+      "step": 52995
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.744274854660034,
+      "learning_rate": 4.852505623790837e-07,
+      "loss": 0.6734,
+      "step": 52996
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.842087745666504,
+      "learning_rate": 4.851372626467138e-07,
+      "loss": 0.4374,
+      "step": 52997
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.525575876235962,
+      "learning_rate": 4.850239754684849e-07,
+      "loss": 0.5585,
+      "step": 52998
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.9755001068115234,
+      "learning_rate": 4.849107008447135e-07,
+      "loss": 0.8392,
+      "step": 52999
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.2462258338928223,
+      "learning_rate": 4.847974387757137e-07,
+      "loss": 0.5999,
+      "step": 53000
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.980076789855957,
+      "learning_rate": 4.846841892618015e-07,
+      "loss": 0.6073,
+      "step": 53001
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.302461862564087,
+      "learning_rate": 4.845709523032899e-07,
+      "loss": 0.6071,
+      "step": 53002
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.6583809852600098,
+      "learning_rate": 4.844577279004963e-07,
+      "loss": 0.5737,
+      "step": 53003
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.2992045879364014,
+      "learning_rate": 4.84344516053733e-07,
+      "loss": 0.5038,
+      "step": 53004
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.0637502670288086,
+      "learning_rate": 4.842313167633172e-07,
+      "loss": 0.6591,
+      "step": 53005
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.790315628051758,
+      "learning_rate": 4.841181300295611e-07,
+      "loss": 0.532,
+      "step": 53006
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.353438377380371,
+      "learning_rate": 4.840049558527821e-07,
+      "loss": 0.7026,
+      "step": 53007
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.3635053634643555,
+      "learning_rate": 4.838917942332922e-07,
+      "loss": 0.5006,
+      "step": 53008
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.623581886291504,
+      "learning_rate": 4.837786451714088e-07,
+      "loss": 0.6416,
+      "step": 53009
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.6629393100738525,
+      "learning_rate": 4.836655086674441e-07,
+      "loss": 0.5953,
+      "step": 53010
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.7071585655212402,
+      "learning_rate": 4.835523847217149e-07,
+      "loss": 0.6657,
+      "step": 53011
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.678065776824951,
+      "learning_rate": 4.834392733345333e-07,
+      "loss": 0.6305,
+      "step": 53012
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.654783248901367,
+      "learning_rate": 4.833261745062157e-07,
+      "loss": 0.3917,
+      "step": 53013
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.015080451965332,
+      "learning_rate": 4.832130882370756e-07,
+      "loss": 0.8705,
+      "step": 53014
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.782797336578369,
+      "learning_rate": 4.831000145274289e-07,
+      "loss": 0.6729,
+      "step": 53015
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.6615118980407715,
+      "learning_rate": 4.829869533775877e-07,
+      "loss": 0.5339,
+      "step": 53016
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.0116610527038574,
+      "learning_rate": 4.828739047878695e-07,
+      "loss": 0.3139,
+      "step": 53017
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.6184020042419434,
+      "learning_rate": 4.82760868758585e-07,
+      "loss": 0.3985,
+      "step": 53018
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 4.294189929962158,
+      "learning_rate": 4.826478452900518e-07,
+      "loss": 1.0309,
+      "step": 53019
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.5985753536224365,
+      "learning_rate": 4.825348343825815e-07,
+      "loss": 0.9213,
+      "step": 53020
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.5488409996032715,
+      "learning_rate": 4.824218360364902e-07,
+      "loss": 0.6801,
+      "step": 53021
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.3173961639404297,
+      "learning_rate": 4.823088502520911e-07,
+      "loss": 0.5811,
+      "step": 53022
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.664093255996704,
+      "learning_rate": 4.821958770296992e-07,
+      "loss": 0.618,
+      "step": 53023
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.8338205814361572,
+      "learning_rate": 4.820829163696278e-07,
+      "loss": 0.7912,
+      "step": 53024
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.2362420558929443,
+      "learning_rate": 4.819699682721907e-07,
+      "loss": 0.4384,
+      "step": 53025
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.2184505462646484,
+      "learning_rate": 4.818570327377043e-07,
+      "loss": 0.6181,
+      "step": 53026
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.773547410964966,
+      "learning_rate": 4.817441097664794e-07,
+      "loss": 0.6583,
+      "step": 53027
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.3309597969055176,
+      "learning_rate": 4.816311993588324e-07,
+      "loss": 0.6464,
+      "step": 53028
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.0165905952453613,
+      "learning_rate": 4.815183015150759e-07,
+      "loss": 0.7062,
+      "step": 53029
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.9600958824157715,
+      "learning_rate": 4.814054162355253e-07,
+      "loss": 0.7128,
+      "step": 53030
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.366201877593994,
+      "learning_rate": 4.812925435204923e-07,
+      "loss": 0.7687,
+      "step": 53031
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.6205906867980957,
+      "learning_rate": 4.811796833702931e-07,
+      "loss": 0.7697,
+      "step": 53032
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.5071816444396973,
+      "learning_rate": 4.810668357852394e-07,
+      "loss": 0.5831,
+      "step": 53033
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.6928043365478516,
+      "learning_rate": 4.80954000765646e-07,
+      "loss": 0.6671,
+      "step": 53034
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.588560104370117,
+      "learning_rate": 4.808411783118278e-07,
+      "loss": 0.4946,
+      "step": 53035
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 4.407763481140137,
+      "learning_rate": 4.80728368424096e-07,
+      "loss": 0.6895,
+      "step": 53036
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.330540657043457,
+      "learning_rate": 4.80615571102766e-07,
+      "loss": 0.5854,
+      "step": 53037
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.8463103771209717,
+      "learning_rate": 4.805027863481515e-07,
+      "loss": 0.5992,
+      "step": 53038
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 5.471492767333984,
+      "learning_rate": 4.803900141605656e-07,
+      "loss": 0.5895,
+      "step": 53039
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.818563461303711,
+      "learning_rate": 4.802772545403223e-07,
+      "loss": 0.6442,
+      "step": 53040
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.321894645690918,
+      "learning_rate": 4.801645074877342e-07,
+      "loss": 0.6703,
+      "step": 53041
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.2722699642181396,
+      "learning_rate": 4.800517730031163e-07,
+      "loss": 0.5938,
+      "step": 53042
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.3536713123321533,
+      "learning_rate": 4.799390510867802e-07,
+      "loss": 0.7173,
+      "step": 53043
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.064134359359741,
+      "learning_rate": 4.798263417390414e-07,
+      "loss": 0.6092,
+      "step": 53044
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.405864715576172,
+      "learning_rate": 4.797136449602114e-07,
+      "loss": 0.6632,
+      "step": 53045
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.9785685539245605,
+      "learning_rate": 4.796009607506047e-07,
+      "loss": 0.585,
+      "step": 53046
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.8118741512298584,
+      "learning_rate": 4.79488289110534e-07,
+      "loss": 0.696,
+      "step": 53047
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.7786853313446045,
+      "learning_rate": 4.79375630040314e-07,
+      "loss": 0.6673,
+      "step": 53048
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.4130072593688965,
+      "learning_rate": 4.792629835402557e-07,
+      "loss": 0.6672,
+      "step": 53049
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.7299106121063232,
+      "learning_rate": 4.791503496106748e-07,
+      "loss": 0.3954,
+      "step": 53050
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.2258713245391846,
+      "learning_rate": 4.790377282518816e-07,
+      "loss": 0.4908,
+      "step": 53051
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.830695629119873,
+      "learning_rate": 4.789251194641925e-07,
+      "loss": 0.7178,
+      "step": 53052
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.5237560272216797,
+      "learning_rate": 4.788125232479168e-07,
+      "loss": 0.552,
+      "step": 53053
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.0835461616516113,
+      "learning_rate": 4.786999396033715e-07,
+      "loss": 0.6137,
+      "step": 53054
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.5503690242767334,
+      "learning_rate": 4.785873685308673e-07,
+      "loss": 0.4502,
+      "step": 53055
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.563133955001831,
+      "learning_rate": 4.784748100307191e-07,
+      "loss": 0.6341,
+      "step": 53056
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.9822890758514404,
+      "learning_rate": 4.783622641032371e-07,
+      "loss": 0.4641,
+      "step": 53057
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 5.719451427459717,
+      "learning_rate": 4.782497307487371e-07,
+      "loss": 0.7332,
+      "step": 53058
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.6438465118408203,
+      "learning_rate": 4.781372099675296e-07,
+      "loss": 0.6532,
+      "step": 53059
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.43717098236084,
+      "learning_rate": 4.780247017599298e-07,
+      "loss": 0.5535,
+      "step": 53060
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.8458728790283203,
+      "learning_rate": 4.779122061262481e-07,
+      "loss": 0.337,
+      "step": 53061
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.0069265365600586,
+      "learning_rate": 4.777997230667997e-07,
+      "loss": 0.4795,
+      "step": 53062
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.4054386615753174,
+      "learning_rate": 4.776872525818955e-07,
+      "loss": 0.5098,
+      "step": 53063
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.671692132949829,
+      "learning_rate": 4.775747946718495e-07,
+      "loss": 0.4897,
+      "step": 53064
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.7585418224334717,
+      "learning_rate": 4.774623493369729e-07,
+      "loss": 0.5718,
+      "step": 53065
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.964812755584717,
+      "learning_rate": 4.773499165775802e-07,
+      "loss": 0.4366,
+      "step": 53066
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.810556173324585,
+      "learning_rate": 4.772374963939825e-07,
+      "loss": 0.4297,
+      "step": 53067
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.740760087966919,
+      "learning_rate": 4.771250887864937e-07,
+      "loss": 0.5899,
+      "step": 53068
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.4084839820861816,
+      "learning_rate": 4.770126937554248e-07,
+      "loss": 0.6447,
+      "step": 53069
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.6234230995178223,
+      "learning_rate": 4.769003113010906e-07,
+      "loss": 0.7353,
+      "step": 53070
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 4.570064544677734,
+      "learning_rate": 4.7678794142380083e-07,
+      "loss": 0.6155,
+      "step": 53071
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.7586350440979004,
+      "learning_rate": 4.766755841238707e-07,
+      "loss": 0.4777,
+      "step": 53072
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 4.154148101806641,
+      "learning_rate": 4.765632394016101e-07,
+      "loss": 0.5952,
+      "step": 53073
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.0873677730560303,
+      "learning_rate": 4.7645090725733254e-07,
+      "loss": 0.77,
+      "step": 53074
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.7908101081848145,
+      "learning_rate": 4.7633858769135067e-07,
+      "loss": 0.5669,
+      "step": 53075
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.056117296218872,
+      "learning_rate": 4.762262807039775e-07,
+      "loss": 0.7926,
+      "step": 53076
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.6845862865448,
+      "learning_rate": 4.761139862955233e-07,
+      "loss": 0.7392,
+      "step": 53077
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 4.545118808746338,
+      "learning_rate": 4.760017044663029e-07,
+      "loss": 0.6686,
+      "step": 53078
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.9164724349975586,
+      "learning_rate": 4.7588943521662547e-07,
+      "loss": 0.533,
+      "step": 53079
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.7864387035369873,
+      "learning_rate": 4.7577717854680574e-07,
+      "loss": 0.4631,
+      "step": 53080
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.5335328578948975,
+      "learning_rate": 4.75664934457154e-07,
+      "loss": 0.433,
+      "step": 53081
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.399836301803589,
+      "learning_rate": 4.755527029479845e-07,
+      "loss": 0.5479,
+      "step": 53082
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 5.342588424682617,
+      "learning_rate": 4.754404840196064e-07,
+      "loss": 0.5278,
+      "step": 53083
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.280707836151123,
+      "learning_rate": 4.75328277672335e-07,
+      "loss": 0.4848,
+      "step": 53084
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 4.238003253936768,
+      "learning_rate": 4.7521608390647954e-07,
+      "loss": 0.5752,
+      "step": 53085
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.5224859714508057,
+      "learning_rate": 4.7510390272235306e-07,
+      "loss": 0.5707,
+      "step": 53086
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.187584161758423,
+      "learning_rate": 4.7499173412026813e-07,
+      "loss": 0.6964,
+      "step": 53087
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.077573537826538,
+      "learning_rate": 4.7487957810053554e-07,
+      "loss": 0.4677,
+      "step": 53088
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.828904151916504,
+      "learning_rate": 4.7476743466346896e-07,
+      "loss": 0.7828,
+      "step": 53089
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 1.725658893585205,
+      "learning_rate": 4.746553038093776e-07,
+      "loss": 0.4699,
+      "step": 53090
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 4.6392502784729,
+      "learning_rate": 4.745431855385757e-07,
+      "loss": 0.7016,
+      "step": 53091
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.9512124061584473,
+      "learning_rate": 4.744310798513729e-07,
+      "loss": 0.4039,
+      "step": 53092
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.3727550506591797,
+      "learning_rate": 4.743189867480824e-07,
+      "loss": 0.47,
+      "step": 53093
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.7469124794006348,
+      "learning_rate": 4.74206906229015e-07,
+      "loss": 0.4521,
+      "step": 53094
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.671050548553467,
+      "learning_rate": 4.7409483829448324e-07,
+      "loss": 0.4645,
+      "step": 53095
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.417477607727051,
+      "learning_rate": 4.739827829447985e-07,
+      "loss": 0.6231,
+      "step": 53096
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.764861822128296,
+      "learning_rate": 4.7387074018027167e-07,
+      "loss": 0.6764,
+      "step": 53097
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.2411320209503174,
+      "learning_rate": 4.7375871000121423e-07,
+      "loss": 0.6024,
+      "step": 53098
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.216059923171997,
+      "learning_rate": 4.736466924079397e-07,
+      "loss": 0.4864,
+      "step": 53099
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.491757392883301,
+      "learning_rate": 4.735346874007574e-07,
+      "loss": 0.4495,
+      "step": 53100
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.516979694366455,
+      "learning_rate": 4.7342269497997974e-07,
+      "loss": 0.659,
+      "step": 53101
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.6968398094177246,
+      "learning_rate": 4.733107151459171e-07,
+      "loss": 0.7602,
+      "step": 53102
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.440127372741699,
+      "learning_rate": 4.731987478988831e-07,
+      "loss": 0.6125,
+      "step": 53103
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.486839771270752,
+      "learning_rate": 4.730867932391858e-07,
+      "loss": 0.6713,
+      "step": 53104
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.2531919479370117,
+      "learning_rate": 4.7297485116714004e-07,
+      "loss": 0.7649,
+      "step": 53105
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.4412078857421875,
+      "learning_rate": 4.728629216830538e-07,
+      "loss": 0.7239,
+      "step": 53106
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.838087797164917,
+      "learning_rate": 4.7275100478724135e-07,
+      "loss": 0.5711,
+      "step": 53107
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.698124647140503,
+      "learning_rate": 4.726391004800113e-07,
+      "loss": 0.5739,
+      "step": 53108
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.9644272327423096,
+      "learning_rate": 4.725272087616767e-07,
+      "loss": 0.5435,
+      "step": 53109
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.232253313064575,
+      "learning_rate": 4.724153296325468e-07,
+      "loss": 0.5013,
+      "step": 53110
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.831622362136841,
+      "learning_rate": 4.7230346309293516e-07,
+      "loss": 0.8202,
+      "step": 53111
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.5619020462036133,
+      "learning_rate": 4.7219160914315054e-07,
+      "loss": 0.4276,
+      "step": 53112
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.0859720706939697,
+      "learning_rate": 4.720797677835054e-07,
+      "loss": 0.3834,
+      "step": 53113
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.424215793609619,
+      "learning_rate": 4.719679390143084e-07,
+      "loss": 0.6358,
+      "step": 53114
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.9913313388824463,
+      "learning_rate": 4.718561228358748e-07,
+      "loss": 0.6798,
+      "step": 53115
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 4.220822811126709,
+      "learning_rate": 4.717443192485116e-07,
+      "loss": 0.4036,
+      "step": 53116
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.47808575630188,
+      "learning_rate": 4.716325282525319e-07,
+      "loss": 0.6717,
+      "step": 53117
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.0535314083099365,
+      "learning_rate": 4.7152074984824547e-07,
+      "loss": 0.6953,
+      "step": 53118
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.9856717586517334,
+      "learning_rate": 4.7140898403596367e-07,
+      "loss": 0.7318,
+      "step": 53119
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.096949338912964,
+      "learning_rate": 4.7129723081599634e-07,
+      "loss": 0.4018,
+      "step": 53120
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.0046656131744385,
+      "learning_rate": 4.711854901886559e-07,
+      "loss": 0.3683,
+      "step": 53121
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.264878511428833,
+      "learning_rate": 4.710737621542516e-07,
+      "loss": 0.6822,
+      "step": 53122
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 4.077401161193848,
+      "learning_rate": 4.7096204671309484e-07,
+      "loss": 0.6363,
+      "step": 53123
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.0779433250427246,
+      "learning_rate": 4.7085034386549544e-07,
+      "loss": 0.5564,
+      "step": 53124
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.174436092376709,
+      "learning_rate": 4.707386536117653e-07,
+      "loss": 0.3767,
+      "step": 53125
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.7292263507843018,
+      "learning_rate": 4.7062697595221365e-07,
+      "loss": 0.4126,
+      "step": 53126
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.67655611038208,
+      "learning_rate": 4.705153108871524e-07,
+      "loss": 0.4553,
+      "step": 53127
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.4983725547790527,
+      "learning_rate": 4.704036584168903e-07,
+      "loss": 0.7406,
+      "step": 53128
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.434875249862671,
+      "learning_rate": 4.702920185417398e-07,
+      "loss": 0.7087,
+      "step": 53129
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.7047505378723145,
+      "learning_rate": 4.70180391262009e-07,
+      "loss": 0.7267,
+      "step": 53130
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.21832013130188,
+      "learning_rate": 4.70068776578011e-07,
+      "loss": 0.769,
+      "step": 53131
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.9315028190612793,
+      "learning_rate": 4.6995717449005385e-07,
+      "loss": 0.6121,
+      "step": 53132
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.491114854812622,
+      "learning_rate": 4.6984558499844957e-07,
+      "loss": 0.3407,
+      "step": 53133
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.9947967529296875,
+      "learning_rate": 4.697340081035068e-07,
+      "loss": 0.4337,
+      "step": 53134
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.343442678451538,
+      "learning_rate": 4.696224438055369e-07,
+      "loss": 0.5177,
+      "step": 53135
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.602099657058716,
+      "learning_rate": 4.695108921048497e-07,
+      "loss": 0.7358,
+      "step": 53136
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.0628340244293213,
+      "learning_rate": 4.693993530017565e-07,
+      "loss": 0.6074,
+      "step": 53137
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 4.223957061767578,
+      "learning_rate": 4.692878264965656e-07,
+      "loss": 0.4782,
+      "step": 53138
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 14.369372367858887,
+      "learning_rate": 4.6917631258958876e-07,
+      "loss": 0.6354,
+      "step": 53139
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.3741519451141357,
+      "learning_rate": 4.6906481128113413e-07,
+      "loss": 0.4135,
+      "step": 53140
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.304135799407959,
+      "learning_rate": 4.6895332257151426e-07,
+      "loss": 0.575,
+      "step": 53141
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.029686212539673,
+      "learning_rate": 4.6884184646103726e-07,
+      "loss": 0.4711,
+      "step": 53142
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.1377198696136475,
+      "learning_rate": 4.6873038295001394e-07,
+      "loss": 0.5022,
+      "step": 53143
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.878171920776367,
+      "learning_rate": 4.686189320387535e-07,
+      "loss": 0.5267,
+      "step": 53144
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.9559452533721924,
+      "learning_rate": 4.6850749372756685e-07,
+      "loss": 0.3502,
+      "step": 53145
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.407424211502075,
+      "learning_rate": 4.6839606801676264e-07,
+      "loss": 0.598,
+      "step": 53146
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.432114362716675,
+      "learning_rate": 4.682846549066522e-07,
+      "loss": 0.6235,
+      "step": 53147
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.780268430709839,
+      "learning_rate": 4.681732543975437e-07,
+      "loss": 0.8004,
+      "step": 53148
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.730574131011963,
+      "learning_rate": 4.6806186648974737e-07,
+      "loss": 0.8464,
+      "step": 53149
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.365661144256592,
+      "learning_rate": 4.6795049118357474e-07,
+      "loss": 0.6973,
+      "step": 53150
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.842235565185547,
+      "learning_rate": 4.6783912847933266e-07,
+      "loss": 0.5443,
+      "step": 53151
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 4.067227363586426,
+      "learning_rate": 4.6772777837733376e-07,
+      "loss": 0.8805,
+      "step": 53152
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 4.060845375061035,
+      "learning_rate": 4.676164408778844e-07,
+      "loss": 0.9787,
+      "step": 53153
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.8511831760406494,
+      "learning_rate": 4.675051159812971e-07,
+      "loss": 0.5527,
+      "step": 53154
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.3090641498565674,
+      "learning_rate": 4.6739380368787946e-07,
+      "loss": 0.6345,
+      "step": 53155
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.640186071395874,
+      "learning_rate": 4.672825039979412e-07,
+      "loss": 0.495,
+      "step": 53156
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.493152379989624,
+      "learning_rate": 4.671712169117937e-07,
+      "loss": 0.3736,
+      "step": 53157
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.77829647064209,
+      "learning_rate": 4.6705994242974397e-07,
+      "loss": 0.4584,
+      "step": 53158
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 4.855897903442383,
+      "learning_rate": 4.669486805521023e-07,
+      "loss": 0.5844,
+      "step": 53159
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.0509705543518066,
+      "learning_rate": 4.66837431279179e-07,
+      "loss": 0.6872,
+      "step": 53160
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.983924388885498,
+      "learning_rate": 4.667261946112822e-07,
+      "loss": 0.6079,
+      "step": 53161
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.8525471687316895,
+      "learning_rate": 4.666149705487222e-07,
+      "loss": 0.4847,
+      "step": 53162
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 4.2996673583984375,
+      "learning_rate": 4.66503759091807e-07,
+      "loss": 0.6074,
+      "step": 53163
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.8724801540374756,
+      "learning_rate": 4.663925602408476e-07,
+      "loss": 0.7135,
+      "step": 53164
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.308518886566162,
+      "learning_rate": 4.662813739961508e-07,
+      "loss": 0.4898,
+      "step": 53165
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.636897087097168,
+      "learning_rate": 4.6617020035802874e-07,
+      "loss": 0.4344,
+      "step": 53166
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.120760679244995,
+      "learning_rate": 4.660590393267872e-07,
+      "loss": 0.5736,
+      "step": 53167
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 4.809042930603027,
+      "learning_rate": 4.6594789090273816e-07,
+      "loss": 0.7507,
+      "step": 53168
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.8448069095611572,
+      "learning_rate": 4.658367550861892e-07,
+      "loss": 0.7147,
+      "step": 53169
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.014012098312378,
+      "learning_rate": 4.6572563187745e-07,
+      "loss": 0.7031,
+      "step": 53170
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.923988103866577,
+      "learning_rate": 4.656145212768287e-07,
+      "loss": 0.3036,
+      "step": 53171
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.8208868503570557,
+      "learning_rate": 4.655034232846356e-07,
+      "loss": 0.6876,
+      "step": 53172
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.118603467941284,
+      "learning_rate": 4.6539233790117764e-07,
+      "loss": 0.5521,
+      "step": 53173
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 5.314394474029541,
+      "learning_rate": 4.6528126512676576e-07,
+      "loss": 0.4677,
+      "step": 53174
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.533165216445923,
+      "learning_rate": 4.651702049617063e-07,
+      "loss": 0.5022,
+      "step": 53175
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.8554956912994385,
+      "learning_rate": 4.650591574063118e-07,
+      "loss": 0.7128,
+      "step": 53176
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.0835697650909424,
+      "learning_rate": 4.6494812246088763e-07,
+      "loss": 0.7087,
+      "step": 53177
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.6468284130096436,
+      "learning_rate": 4.648371001257451e-07,
+      "loss": 0.6912,
+      "step": 53178
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.3237147331237793,
+      "learning_rate": 4.6472609040119076e-07,
+      "loss": 0.4891,
+      "step": 53179
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.1970202922821045,
+      "learning_rate": 4.6461509328753484e-07,
+      "loss": 0.7466,
+      "step": 53180
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.7716073989868164,
+      "learning_rate": 4.645041087850849e-07,
+      "loss": 0.6025,
+      "step": 53181
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 4.4618916511535645,
+      "learning_rate": 4.643931368941501e-07,
+      "loss": 0.5128,
+      "step": 53182
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 18.898252487182617,
+      "learning_rate": 4.642821776150386e-07,
+      "loss": 0.5854,
+      "step": 53183
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.8186326026916504,
+      "learning_rate": 4.6417123094806007e-07,
+      "loss": 0.4223,
+      "step": 53184
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.4365386962890625,
+      "learning_rate": 4.640602968935215e-07,
+      "loss": 0.5883,
+      "step": 53185
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.749215602874756,
+      "learning_rate": 4.639493754517327e-07,
+      "loss": 0.5403,
+      "step": 53186
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.6048943996429443,
+      "learning_rate": 4.6383846662300056e-07,
+      "loss": 0.562,
+      "step": 53187
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 5.066439151763916,
+      "learning_rate": 4.637275704076355e-07,
+      "loss": 0.6402,
+      "step": 53188
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.0810365676879883,
+      "learning_rate": 4.636166868059433e-07,
+      "loss": 0.6317,
+      "step": 53189
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.647756814956665,
+      "learning_rate": 4.6350581581823484e-07,
+      "loss": 0.5989,
+      "step": 53190
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 4.667864799499512,
+      "learning_rate": 4.6339495744481655e-07,
+      "loss": 0.4621,
+      "step": 53191
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.488271951675415,
+      "learning_rate": 4.632841116859982e-07,
+      "loss": 0.4475,
+      "step": 53192
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.669842481613159,
+      "learning_rate": 4.631732785420867e-07,
+      "loss": 0.786,
+      "step": 53193
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.629817008972168,
+      "learning_rate": 4.630624580133913e-07,
+      "loss": 0.5136,
+      "step": 53194
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.523576021194458,
+      "learning_rate": 4.6295165010021845e-07,
+      "loss": 0.5587,
+      "step": 53195
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.740983247756958,
+      "learning_rate": 4.6284085480287787e-07,
+      "loss": 0.4826,
+      "step": 53196
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.141047716140747,
+      "learning_rate": 4.627300721216771e-07,
+      "loss": 0.4563,
+      "step": 53197
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.7913050651550293,
+      "learning_rate": 4.626193020569253e-07,
+      "loss": 0.5284,
+      "step": 53198
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.979734420776367,
+      "learning_rate": 4.625085446089289e-07,
+      "loss": 0.5616,
+      "step": 53199
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.335192918777466,
+      "learning_rate": 4.623977997779966e-07,
+      "loss": 0.6153,
+      "step": 53200
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.808274269104004,
+      "learning_rate": 4.6228706756443586e-07,
+      "loss": 0.5195,
+      "step": 53201
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.9648067951202393,
+      "learning_rate": 4.621763479685554e-07,
+      "loss": 0.5398,
+      "step": 53202
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.4184391498565674,
+      "learning_rate": 4.6206564099066207e-07,
+      "loss": 0.5256,
+      "step": 53203
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.873095750808716,
+      "learning_rate": 4.619549466310652e-07,
+      "loss": 0.7383,
+      "step": 53204
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.1279420852661133,
+      "learning_rate": 4.618442648900706e-07,
+      "loss": 0.6413,
+      "step": 53205
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.7531187534332275,
+      "learning_rate": 4.6173359576798805e-07,
+      "loss": 0.5752,
+      "step": 53206
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.6475601196289062,
+      "learning_rate": 4.616229392651228e-07,
+      "loss": 0.5851,
+      "step": 53207
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 4.143253326416016,
+      "learning_rate": 4.615122953817858e-07,
+      "loss": 0.6321,
+      "step": 53208
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.444692850112915,
+      "learning_rate": 4.614016641182817e-07,
+      "loss": 0.5366,
+      "step": 53209
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.4962778091430664,
+      "learning_rate": 4.612910454749192e-07,
+      "loss": 0.4202,
+      "step": 53210
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.00823712348938,
+      "learning_rate": 4.6118043945200696e-07,
+      "loss": 0.6264,
+      "step": 53211
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.4903371334075928,
+      "learning_rate": 4.610698460498508e-07,
+      "loss": 0.5197,
+      "step": 53212
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.6395132541656494,
+      "learning_rate": 4.6095926526876e-07,
+      "loss": 0.4999,
+      "step": 53213
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.9729225635528564,
+      "learning_rate": 4.6084869710904034e-07,
+      "loss": 0.3492,
+      "step": 53214
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.220499277114868,
+      "learning_rate": 4.6073814157100107e-07,
+      "loss": 0.2656,
+      "step": 53215
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.122352361679077,
+      "learning_rate": 4.6062759865494746e-07,
+      "loss": 0.6446,
+      "step": 53216
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.9491500854492188,
+      "learning_rate": 4.605170683611876e-07,
+      "loss": 0.4687,
+      "step": 53217
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 4.730080604553223,
+      "learning_rate": 4.604065506900307e-07,
+      "loss": 0.5478,
+      "step": 53218
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.929008722305298,
+      "learning_rate": 4.602960456417815e-07,
+      "loss": 0.3828,
+      "step": 53219
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 4.265282154083252,
+      "learning_rate": 4.6018555321674986e-07,
+      "loss": 0.6627,
+      "step": 53220
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.6201274394989014,
+      "learning_rate": 4.6007507341523985e-07,
+      "loss": 0.5009,
+      "step": 53221
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.9106035232543945,
+      "learning_rate": 4.599646062375607e-07,
+      "loss": 0.5208,
+      "step": 53222
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.4352352619171143,
+      "learning_rate": 4.598541516840199e-07,
+      "loss": 0.9313,
+      "step": 53223
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.685269594192505,
+      "learning_rate": 4.5974370975492343e-07,
+      "loss": 0.8072,
+      "step": 53224
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.8628647327423096,
+      "learning_rate": 4.5963328045057985e-07,
+      "loss": 0.7094,
+      "step": 53225
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.530357837677002,
+      "learning_rate": 4.5952286377129395e-07,
+      "loss": 0.7832,
+      "step": 53226
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.758950710296631,
+      "learning_rate": 4.594124597173749e-07,
+      "loss": 0.624,
+      "step": 53227
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.562854051589966,
+      "learning_rate": 4.593020682891286e-07,
+      "loss": 0.6135,
+      "step": 53228
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.161302089691162,
+      "learning_rate": 4.5919168948686253e-07,
+      "loss": 0.4495,
+      "step": 53229
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.681330680847168,
+      "learning_rate": 4.590813233108826e-07,
+      "loss": 0.6626,
+      "step": 53230
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.3088741302490234,
+      "learning_rate": 4.5897096976149745e-07,
+      "loss": 0.6412,
+      "step": 53231
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.456219434738159,
+      "learning_rate": 4.5886062883901183e-07,
+      "loss": 0.4872,
+      "step": 53232
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.5867881774902344,
+      "learning_rate": 4.5875030054373494e-07,
+      "loss": 0.4225,
+      "step": 53233
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.289663553237915,
+      "learning_rate": 4.5863998487597096e-07,
+      "loss": 0.6902,
+      "step": 53234
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.546457290649414,
+      "learning_rate": 4.5852968183602906e-07,
+      "loss": 0.5642,
+      "step": 53235
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.312734603881836,
+      "learning_rate": 4.58419391424213e-07,
+      "loss": 0.63,
+      "step": 53236
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.900919198989868,
+      "learning_rate": 4.58309113640833e-07,
+      "loss": 0.6762,
+      "step": 53237
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.8911120891571045,
+      "learning_rate": 4.581988484861932e-07,
+      "loss": 0.4243,
+      "step": 53238
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.244736909866333,
+      "learning_rate": 4.5808859596060183e-07,
+      "loss": 0.6712,
+      "step": 53239
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.5134029388427734,
+      "learning_rate": 4.579783560643641e-07,
+      "loss": 0.4775,
+      "step": 53240
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.5551934242248535,
+      "learning_rate": 4.578681287977876e-07,
+      "loss": 0.5732,
+      "step": 53241
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.4733633995056152,
+      "learning_rate": 4.5775791416117766e-07,
+      "loss": 0.478,
+      "step": 53242
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.269885540008545,
+      "learning_rate": 4.5764771215484174e-07,
+      "loss": 0.5299,
+      "step": 53243
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.3726861476898193,
+      "learning_rate": 4.575375227790857e-07,
+      "loss": 0.3828,
+      "step": 53244
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.134936571121216,
+      "learning_rate": 4.574273460342166e-07,
+      "loss": 0.6581,
+      "step": 53245
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.38618803024292,
+      "learning_rate": 4.5731718192053964e-07,
+      "loss": 0.6618,
+      "step": 53246
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 5.705967903137207,
+      "learning_rate": 4.5720703043836303e-07,
+      "loss": 0.8435,
+      "step": 53247
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.394176959991455,
+      "learning_rate": 4.570968915879909e-07,
+      "loss": 0.5741,
+      "step": 53248
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.410797595977783,
+      "learning_rate": 4.5698676536973076e-07,
+      "loss": 0.7343,
+      "step": 53249
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.001396894454956,
+      "learning_rate": 4.568766517838885e-07,
+      "loss": 0.6022,
+      "step": 53250
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.916365385055542,
+      "learning_rate": 4.5676655083077114e-07,
+      "loss": 0.5375,
+      "step": 53251
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.0848641395568848,
+      "learning_rate": 4.566564625106828e-07,
+      "loss": 0.4722,
+      "step": 53252
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.2167294025421143,
+      "learning_rate": 4.565463868239317e-07,
+      "loss": 0.4729,
+      "step": 53253
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.6009769439697266,
+      "learning_rate": 4.5643632377082246e-07,
+      "loss": 0.5294,
+      "step": 53254
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.3331329822540283,
+      "learning_rate": 4.5632627335166266e-07,
+      "loss": 0.4583,
+      "step": 53255
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.701237201690674,
+      "learning_rate": 4.562162355667565e-07,
+      "loss": 0.4699,
+      "step": 53256
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.26259708404541,
+      "learning_rate": 4.56106210416411e-07,
+      "loss": 0.5449,
+      "step": 53257
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.2581591606140137,
+      "learning_rate": 4.5599619790093143e-07,
+      "loss": 0.6096,
+      "step": 53258
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 6.486936569213867,
+      "learning_rate": 4.5588619802062527e-07,
+      "loss": 0.6903,
+      "step": 53259
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.412227153778076,
+      "learning_rate": 4.557762107757968e-07,
+      "loss": 0.6832,
+      "step": 53260
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.7395215034484863,
+      "learning_rate": 4.55666236166753e-07,
+      "loss": 0.6556,
+      "step": 53261
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.3485054969787598,
+      "learning_rate": 4.55556274193798e-07,
+      "loss": 0.5085,
+      "step": 53262
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.997396230697632,
+      "learning_rate": 4.5544632485723993e-07,
+      "loss": 0.6961,
+      "step": 53263
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.7500107288360596,
+      "learning_rate": 4.553363881573819e-07,
+      "loss": 0.6387,
+      "step": 53264
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.922219753265381,
+      "learning_rate": 4.5522646409453255e-07,
+      "loss": 0.356,
+      "step": 53265
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.9438998699188232,
+      "learning_rate": 4.5511655266899435e-07,
+      "loss": 0.6534,
+      "step": 53266
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.2139530181884766,
+      "learning_rate": 4.5500665388107544e-07,
+      "loss": 0.795,
+      "step": 53267
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.741400718688965,
+      "learning_rate": 4.548967677310795e-07,
+      "loss": 0.6473,
+      "step": 53268
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.452655553817749,
+      "learning_rate": 4.54786894219314e-07,
+      "loss": 0.6933,
+      "step": 53269
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.415719985961914,
+      "learning_rate": 4.546770333460826e-07,
+      "loss": 0.6087,
+      "step": 53270
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.154266834259033,
+      "learning_rate": 4.5456718511169117e-07,
+      "loss": 0.5973,
+      "step": 53271
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.303162097930908,
+      "learning_rate": 4.544573495164473e-07,
+      "loss": 0.8069,
+      "step": 53272
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 4.214244842529297,
+      "learning_rate": 4.543475265606534e-07,
+      "loss": 0.8098,
+      "step": 53273
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 10.2516508102417,
+      "learning_rate": 4.542377162446171e-07,
+      "loss": 0.5645,
+      "step": 53274
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.6138176918029785,
+      "learning_rate": 4.5412791856864204e-07,
+      "loss": 0.6321,
+      "step": 53275
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 5.455537796020508,
+      "learning_rate": 4.540181335330351e-07,
+      "loss": 0.6094,
+      "step": 53276
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.032125234603882,
+      "learning_rate": 4.539083611381001e-07,
+      "loss": 0.4346,
+      "step": 53277
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.3857316970825195,
+      "learning_rate": 4.537986013841428e-07,
+      "loss": 0.5951,
+      "step": 53278
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.4537763595581055,
+      "learning_rate": 4.5368885427146957e-07,
+      "loss": 0.391,
+      "step": 53279
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.393627166748047,
+      "learning_rate": 4.535791198003836e-07,
+      "loss": 0.6368,
+      "step": 53280
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.890495538711548,
+      "learning_rate": 4.534693979711918e-07,
+      "loss": 0.6106,
+      "step": 53281
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.736635208129883,
+      "learning_rate": 4.5335968878419735e-07,
+      "loss": 0.4241,
+      "step": 53282
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 1.7587343454360962,
+      "learning_rate": 4.5324999223970655e-07,
+      "loss": 0.3513,
+      "step": 53283
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.7263314723968506,
+      "learning_rate": 4.531403083380254e-07,
+      "loss": 0.4979,
+      "step": 53284
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 8.140987396240234,
+      "learning_rate": 4.5303063707945625e-07,
+      "loss": 0.4231,
+      "step": 53285
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.560666084289551,
+      "learning_rate": 4.5292097846430684e-07,
+      "loss": 0.4958,
+      "step": 53286
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 6.410791873931885,
+      "learning_rate": 4.528113324928801e-07,
+      "loss": 0.76,
+      "step": 53287
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.8387842178344727,
+      "learning_rate": 4.5270169916548254e-07,
+      "loss": 0.4378,
+      "step": 53288
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.607891798019409,
+      "learning_rate": 4.525920784824167e-07,
+      "loss": 0.6458,
+      "step": 53289
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.3014750480651855,
+      "learning_rate": 4.5248247044399005e-07,
+      "loss": 0.5498,
+      "step": 53290
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.2089083194732666,
+      "learning_rate": 4.523728750505052e-07,
+      "loss": 0.5532,
+      "step": 53291
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.700925827026367,
+      "learning_rate": 4.522632923022691e-07,
+      "loss": 0.51,
+      "step": 53292
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.9729225635528564,
+      "learning_rate": 4.5215372219958366e-07,
+      "loss": 0.4402,
+      "step": 53293
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.236130952835083,
+      "learning_rate": 4.520441647427559e-07,
+      "loss": 0.5757,
+      "step": 53294
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.5302951335906982,
+      "learning_rate": 4.51934619932089e-07,
+      "loss": 0.5022,
+      "step": 53295
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.3747074604034424,
+      "learning_rate": 4.5182508776788927e-07,
+      "loss": 0.4828,
+      "step": 53296
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 4.8895487785339355,
+      "learning_rate": 4.5171556825045815e-07,
+      "loss": 0.7037,
+      "step": 53297
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.323483467102051,
+      "learning_rate": 4.5160606138010377e-07,
+      "loss": 0.65,
+      "step": 53298
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.6863787174224854,
+      "learning_rate": 4.5149656715712866e-07,
+      "loss": 0.6545,
+      "step": 53299
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.4479496479034424,
+      "learning_rate": 4.513870855818381e-07,
+      "loss": 0.8176,
+      "step": 53300
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.2248904705047607,
+      "learning_rate": 4.5127761665453575e-07,
+      "loss": 0.6534,
+      "step": 53301
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.062486171722412,
+      "learning_rate": 4.511681603755269e-07,
+      "loss": 0.7975,
+      "step": 53302
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.983891487121582,
+      "learning_rate": 4.510587167451147e-07,
+      "loss": 0.7121,
+      "step": 53303
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.783952474594116,
+      "learning_rate": 4.5094928576360444e-07,
+      "loss": 0.7286,
+      "step": 53304
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.1752684116363525,
+      "learning_rate": 4.5083986743129973e-07,
+      "loss": 0.5468,
+      "step": 53305
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.6163549423217773,
+      "learning_rate": 4.5073046174850585e-07,
+      "loss": 0.5997,
+      "step": 53306
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.009570598602295,
+      "learning_rate": 4.50621068715526e-07,
+      "loss": 0.8645,
+      "step": 53307
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.725083589553833,
+      "learning_rate": 4.505116883326649e-07,
+      "loss": 0.5289,
+      "step": 53308
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.1511318683624268,
+      "learning_rate": 4.5040232060022614e-07,
+      "loss": 0.5404,
+      "step": 53309
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.804856061935425,
+      "learning_rate": 4.5029296551851455e-07,
+      "loss": 0.5057,
+      "step": 53310
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.4696149826049805,
+      "learning_rate": 4.501836230878337e-07,
+      "loss": 0.6116,
+      "step": 53311
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.262437343597412,
+      "learning_rate": 4.500742933084884e-07,
+      "loss": 0.5776,
+      "step": 53312
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.80470871925354,
+      "learning_rate": 4.499649761807806e-07,
+      "loss": 0.5711,
+      "step": 53313
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.5413877964019775,
+      "learning_rate": 4.4985567170501734e-07,
+      "loss": 0.3946,
+      "step": 53314
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.9597959518432617,
+      "learning_rate": 4.4974637988149994e-07,
+      "loss": 0.6819,
+      "step": 53315
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.631340503692627,
+      "learning_rate": 4.496371007105338e-07,
+      "loss": 0.513,
+      "step": 53316
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 4.245666027069092,
+      "learning_rate": 4.4952783419242143e-07,
+      "loss": 0.5925,
+      "step": 53317
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.5540287494659424,
+      "learning_rate": 4.4941858032746754e-07,
+      "loss": 0.5738,
+      "step": 53318
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.157588481903076,
+      "learning_rate": 4.493093391159764e-07,
+      "loss": 0.5571,
+      "step": 53319
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.993943452835083,
+      "learning_rate": 4.492001105582516e-07,
+      "loss": 0.7304,
+      "step": 53320
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.7023239135742188,
+      "learning_rate": 4.4909089465459576e-07,
+      "loss": 0.8485,
+      "step": 53321
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.097795009613037,
+      "learning_rate": 4.4898169140531467e-07,
+      "loss": 0.4848,
+      "step": 53322
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 4.955067157745361,
+      "learning_rate": 4.488725008107092e-07,
+      "loss": 0.6259,
+      "step": 53323
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.820936679840088,
+      "learning_rate": 4.487633228710853e-07,
+      "loss": 0.5436,
+      "step": 53324
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.7972590923309326,
+      "learning_rate": 4.486541575867448e-07,
+      "loss": 0.7614,
+      "step": 53325
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 4.476130962371826,
+      "learning_rate": 4.485450049579931e-07,
+      "loss": 0.654,
+      "step": 53326
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.181749105453491,
+      "learning_rate": 4.4843586498513224e-07,
+      "loss": 0.3544,
+      "step": 53327
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.7264482975006104,
+      "learning_rate": 4.4832673766846693e-07,
+      "loss": 0.4635,
+      "step": 53328
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.738311529159546,
+      "learning_rate": 4.482176230082985e-07,
+      "loss": 0.4171,
+      "step": 53329
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 15.724356651306152,
+      "learning_rate": 4.4810852100493295e-07,
+      "loss": 0.7116,
+      "step": 53330
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.1843554973602295,
+      "learning_rate": 4.479994316586711e-07,
+      "loss": 0.5076,
+      "step": 53331
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.2292685508728027,
+      "learning_rate": 4.4789035496981936e-07,
+      "loss": 0.7831,
+      "step": 53332
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.4743762016296387,
+      "learning_rate": 4.477812909386775e-07,
+      "loss": 0.5896,
+      "step": 53333
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.9940013885498047,
+      "learning_rate": 4.476722395655514e-07,
+      "loss": 0.6493,
+      "step": 53334
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.6352624893188477,
+      "learning_rate": 4.4756320085074357e-07,
+      "loss": 0.5795,
+      "step": 53335
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.446032762527466,
+      "learning_rate": 4.4745417479455656e-07,
+      "loss": 0.5707,
+      "step": 53336
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.530975580215454,
+      "learning_rate": 4.473451613972951e-07,
+      "loss": 0.3942,
+      "step": 53337
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.4309232234954834,
+      "learning_rate": 4.472361606592601e-07,
+      "loss": 0.7676,
+      "step": 53338
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 6.615821838378906,
+      "learning_rate": 4.4712717258075635e-07,
+      "loss": 0.4063,
+      "step": 53339
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.3345417976379395,
+      "learning_rate": 4.470181971620868e-07,
+      "loss": 0.5327,
+      "step": 53340
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.8904507160186768,
+      "learning_rate": 4.4690923440355307e-07,
+      "loss": 0.6111,
+      "step": 53341
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.682410717010498,
+      "learning_rate": 4.468002843054603e-07,
+      "loss": 0.4484,
+      "step": 53342
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 4.579816818237305,
+      "learning_rate": 4.4669134686810945e-07,
+      "loss": 0.7951,
+      "step": 53343
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.5653939247131348,
+      "learning_rate": 4.4658242209180415e-07,
+      "loss": 0.4852,
+      "step": 53344
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.3825488090515137,
+      "learning_rate": 4.46473509976848e-07,
+      "loss": 0.7535,
+      "step": 53345
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.8958964347839355,
+      "learning_rate": 4.463646105235425e-07,
+      "loss": 0.5576,
+      "step": 53346
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.9315457344055176,
+      "learning_rate": 4.462557237321924e-07,
+      "loss": 0.4976,
+      "step": 53347
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.5122361183166504,
+      "learning_rate": 4.4614684960309795e-07,
+      "loss": 0.6982,
+      "step": 53348
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.9576497077941895,
+      "learning_rate": 4.460379881365645e-07,
+      "loss": 0.4126,
+      "step": 53349
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.4522995948791504,
+      "learning_rate": 4.459291393328924e-07,
+      "loss": 0.6053,
+      "step": 53350
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.269355535507202,
+      "learning_rate": 4.458203031923858e-07,
+      "loss": 0.5887,
+      "step": 53351
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.543848991394043,
+      "learning_rate": 4.457114797153467e-07,
+      "loss": 0.6165,
+      "step": 53352
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.5898776054382324,
+      "learning_rate": 4.4560266890207813e-07,
+      "loss": 0.8612,
+      "step": 53353
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.187138319015503,
+      "learning_rate": 4.454938707528822e-07,
+      "loss": 0.5553,
+      "step": 53354
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.180182456970215,
+      "learning_rate": 4.453850852680625e-07,
+      "loss": 0.4049,
+      "step": 53355
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.9410316944122314,
+      "learning_rate": 4.4527631244791927e-07,
+      "loss": 1.0552,
+      "step": 53356
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.1667938232421875,
+      "learning_rate": 4.451675522927573e-07,
+      "loss": 0.6424,
+      "step": 53357
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.518548011779785,
+      "learning_rate": 4.450588048028764e-07,
+      "loss": 0.7639,
+      "step": 53358
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.7602128982543945,
+      "learning_rate": 4.4495006997858305e-07,
+      "loss": 0.5142,
+      "step": 53359
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 4.118893623352051,
+      "learning_rate": 4.448413478201757e-07,
+      "loss": 0.531,
+      "step": 53360
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.5777816772460938,
+      "learning_rate": 4.447326383279593e-07,
+      "loss": 0.6775,
+      "step": 53361
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.962559700012207,
+      "learning_rate": 4.4462394150223354e-07,
+      "loss": 0.7136,
+      "step": 53362
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.9130122661590576,
+      "learning_rate": 4.445152573433037e-07,
+      "loss": 0.6612,
+      "step": 53363
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.8577284812927246,
+      "learning_rate": 4.44406585851469e-07,
+      "loss": 0.6974,
+      "step": 53364
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.288296937942505,
+      "learning_rate": 4.4429792702703425e-07,
+      "loss": 0.4133,
+      "step": 53365
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.640287160873413,
+      "learning_rate": 4.4418928087029966e-07,
+      "loss": 0.5718,
+      "step": 53366
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.3283255100250244,
+      "learning_rate": 4.4408064738156845e-07,
+      "loss": 0.5634,
+      "step": 53367
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.25113844871521,
+      "learning_rate": 4.4397202656114134e-07,
+      "loss": 0.5108,
+      "step": 53368
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.880166530609131,
+      "learning_rate": 4.4386341840932265e-07,
+      "loss": 0.6962,
+      "step": 53369
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.5508875846862793,
+      "learning_rate": 4.437548229264121e-07,
+      "loss": 0.5355,
+      "step": 53370
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.8192429542541504,
+      "learning_rate": 4.436462401127128e-07,
+      "loss": 0.7281,
+      "step": 53371
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.473881959915161,
+      "learning_rate": 4.435376699685262e-07,
+      "loss": 0.8024,
+      "step": 53372
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.9241950511932373,
+      "learning_rate": 4.434291124941553e-07,
+      "loss": 0.6305,
+      "step": 53373
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.7916946411132812,
+      "learning_rate": 4.4332056768989994e-07,
+      "loss": 0.4568,
+      "step": 53374
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.0237364768981934,
+      "learning_rate": 4.432120355560643e-07,
+      "loss": 0.7069,
+      "step": 53375
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.016911745071411,
+      "learning_rate": 4.431035160929481e-07,
+      "loss": 0.4759,
+      "step": 53376
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.584182024002075,
+      "learning_rate": 4.429950093008545e-07,
+      "loss": 0.5338,
+      "step": 53377
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.6511130332946777,
+      "learning_rate": 4.428865151800843e-07,
+      "loss": 0.5942,
+      "step": 53378
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 1.947243094444275,
+      "learning_rate": 4.42778033730939e-07,
+      "loss": 0.3542,
+      "step": 53379
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.0254039764404297,
+      "learning_rate": 4.42669564953721e-07,
+      "loss": 0.5959,
+      "step": 53380
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.6721725463867188,
+      "learning_rate": 4.4256110884873306e-07,
+      "loss": 0.6129,
+      "step": 53381
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.244023561477661,
+      "learning_rate": 4.424526654162742e-07,
+      "loss": 0.7038,
+      "step": 53382
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.4102323055267334,
+      "learning_rate": 4.4234423465664756e-07,
+      "loss": 0.6814,
+      "step": 53383
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.1023714542388916,
+      "learning_rate": 4.42235816570154e-07,
+      "loss": 0.6354,
+      "step": 53384
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.789928674697876,
+      "learning_rate": 4.421274111570961e-07,
+      "loss": 0.8834,
+      "step": 53385
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.2080042362213135,
+      "learning_rate": 4.420190184177736e-07,
+      "loss": 0.617,
+      "step": 53386
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.603790283203125,
+      "learning_rate": 4.419106383524896e-07,
+      "loss": 0.585,
+      "step": 53387
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.2135252952575684,
+      "learning_rate": 4.4180227096154325e-07,
+      "loss": 0.5847,
+      "step": 53388
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.3162693977355957,
+      "learning_rate": 4.4169391624523885e-07,
+      "loss": 0.7236,
+      "step": 53389
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.697028160095215,
+      "learning_rate": 4.415855742038749e-07,
+      "loss": 0.544,
+      "step": 53390
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.313896417617798,
+      "learning_rate": 4.4147724483775467e-07,
+      "loss": 0.3788,
+      "step": 53391
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.00258731842041,
+      "learning_rate": 4.413689281471778e-07,
+      "loss": 0.6511,
+      "step": 53392
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.6317272186279297,
+      "learning_rate": 4.4126062413244685e-07,
+      "loss": 0.631,
+      "step": 53393
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.8305604457855225,
+      "learning_rate": 4.411523327938616e-07,
+      "loss": 0.5854,
+      "step": 53394
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.492370843887329,
+      "learning_rate": 4.41044054131724e-07,
+      "loss": 0.4714,
+      "step": 53395
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.9379847049713135,
+      "learning_rate": 4.409357881463361e-07,
+      "loss": 0.5639,
+      "step": 53396
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.255340814590454,
+      "learning_rate": 4.4082753483799644e-07,
+      "loss": 0.8963,
+      "step": 53397
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 4.4077301025390625,
+      "learning_rate": 4.407192942070088e-07,
+      "loss": 0.5441,
+      "step": 53398
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.8338096141815186,
+      "learning_rate": 4.406110662536722e-07,
+      "loss": 0.6756,
+      "step": 53399
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.136289358139038,
+      "learning_rate": 4.405028509782877e-07,
+      "loss": 0.6464,
+      "step": 53400
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.8754310607910156,
+      "learning_rate": 4.4039464838115833e-07,
+      "loss": 0.7071,
+      "step": 53401
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.487550735473633,
+      "learning_rate": 4.402864584625821e-07,
+      "loss": 0.5203,
+      "step": 53402
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.0012099742889404,
+      "learning_rate": 4.4017828122286165e-07,
+      "loss": 0.7055,
+      "step": 53403
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 6.546829700469971,
+      "learning_rate": 4.400701166622967e-07,
+      "loss": 0.5508,
+      "step": 53404
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.1686434745788574,
+      "learning_rate": 4.399619647811898e-07,
+      "loss": 0.341,
+      "step": 53405
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.6980302333831787,
+      "learning_rate": 4.39853825579839e-07,
+      "loss": 0.6951,
+      "step": 53406
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.6905572414398193,
+      "learning_rate": 4.3974569905854694e-07,
+      "loss": 0.6935,
+      "step": 53407
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.5931880474090576,
+      "learning_rate": 4.3963758521761437e-07,
+      "loss": 0.4808,
+      "step": 53408
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.403639078140259,
+      "learning_rate": 4.3952948405734053e-07,
+      "loss": 0.6079,
+      "step": 53409
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.4927258491516113,
+      "learning_rate": 4.3942139557802744e-07,
+      "loss": 0.5573,
+      "step": 53410
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.9033761024475098,
+      "learning_rate": 4.3931331977997425e-07,
+      "loss": 0.5455,
+      "step": 53411
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.2457244396209717,
+      "learning_rate": 4.392052566634836e-07,
+      "loss": 0.6982,
+      "step": 53412
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.6487467288970947,
+      "learning_rate": 4.390972062288529e-07,
+      "loss": 0.6943,
+      "step": 53413
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.2156858444213867,
+      "learning_rate": 4.389891684763858e-07,
+      "loss": 0.6306,
+      "step": 53414
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.527070999145508,
+      "learning_rate": 4.3888114340638e-07,
+      "loss": 0.5566,
+      "step": 53415
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.34381103515625,
+      "learning_rate": 4.387731310191384e-07,
+      "loss": 0.7219,
+      "step": 53416
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.286104440689087,
+      "learning_rate": 4.386651313149587e-07,
+      "loss": 0.3856,
+      "step": 53417
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.547194719314575,
+      "learning_rate": 4.385571442941439e-07,
+      "loss": 0.6319,
+      "step": 53418
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 4.11116886138916,
+      "learning_rate": 4.384491699569909e-07,
+      "loss": 0.4862,
+      "step": 53419
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.4731228351593018,
+      "learning_rate": 4.3834120830380354e-07,
+      "loss": 0.3153,
+      "step": 53420
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.7648603916168213,
+      "learning_rate": 4.382332593348798e-07,
+      "loss": 0.743,
+      "step": 53421
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.6704885959625244,
+      "learning_rate": 4.381253230505217e-07,
+      "loss": 0.6659,
+      "step": 53422
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.2418131828308105,
+      "learning_rate": 4.380173994510267e-07,
+      "loss": 0.5637,
+      "step": 53423
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.291844606399536,
+      "learning_rate": 4.3790948853669747e-07,
+      "loss": 0.437,
+      "step": 53424
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.8833425045013428,
+      "learning_rate": 4.3780159030783196e-07,
+      "loss": 0.6104,
+      "step": 53425
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.450267791748047,
+      "learning_rate": 4.3769370476473225e-07,
+      "loss": 0.7336,
+      "step": 53426
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.0841944217681885,
+      "learning_rate": 4.3758583190769645e-07,
+      "loss": 0.435,
+      "step": 53427
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 1.8920916318893433,
+      "learning_rate": 4.374779717370259e-07,
+      "loss": 0.4943,
+      "step": 53428
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.254936933517456,
+      "learning_rate": 4.3737012425301925e-07,
+      "loss": 0.6483,
+      "step": 53429
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.2782092094421387,
+      "learning_rate": 4.3726228945597804e-07,
+      "loss": 0.4664,
+      "step": 53430
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.4144749641418457,
+      "learning_rate": 4.371544673462003e-07,
+      "loss": 0.7771,
+      "step": 53431
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.877342700958252,
+      "learning_rate": 4.370466579239874e-07,
+      "loss": 0.7126,
+      "step": 53432
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.0055453777313232,
+      "learning_rate": 4.36938861189638e-07,
+      "loss": 0.5682,
+      "step": 53433
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.1396429538726807,
+      "learning_rate": 4.368310771434525e-07,
+      "loss": 0.3589,
+      "step": 53434
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 4.399777889251709,
+      "learning_rate": 4.367233057857301e-07,
+      "loss": 0.5757,
+      "step": 53435
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.448706865310669,
+      "learning_rate": 4.36615547116771e-07,
+      "loss": 0.5343,
+      "step": 53436
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.103224515914917,
+      "learning_rate": 4.3650780113687396e-07,
+      "loss": 0.5376,
+      "step": 53437
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 4.585581302642822,
+      "learning_rate": 4.364000678463404e-07,
+      "loss": 0.6621,
+      "step": 53438
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.2430732250213623,
+      "learning_rate": 4.362923472454672e-07,
+      "loss": 0.4342,
+      "step": 53439
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.263988733291626,
+      "learning_rate": 4.361846393345559e-07,
+      "loss": 0.572,
+      "step": 53440
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.83343768119812,
+      "learning_rate": 4.3607694411390513e-07,
+      "loss": 0.5081,
+      "step": 53441
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.5661754608154297,
+      "learning_rate": 4.359692615838157e-07,
+      "loss": 0.3735,
+      "step": 53442
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.0109238624572754,
+      "learning_rate": 4.358615917445852e-07,
+      "loss": 0.8084,
+      "step": 53443
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.606401205062866,
+      "learning_rate": 4.357539345965145e-07,
+      "loss": 0.7617,
+      "step": 53444
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 5.464874267578125,
+      "learning_rate": 4.356462901399017e-07,
+      "loss": 0.4611,
+      "step": 53445
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 5.263369560241699,
+      "learning_rate": 4.355386583750476e-07,
+      "loss": 0.5807,
+      "step": 53446
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.4050543308258057,
+      "learning_rate": 4.354310393022498e-07,
+      "loss": 0.6244,
+      "step": 53447
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.2716610431671143,
+      "learning_rate": 4.3532343292180925e-07,
+      "loss": 0.8658,
+      "step": 53448
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.190286159515381,
+      "learning_rate": 4.3521583923402276e-07,
+      "loss": 0.5151,
+      "step": 53449
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.538276195526123,
+      "learning_rate": 4.3510825823919243e-07,
+      "loss": 0.6409,
+      "step": 53450
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.4053027629852295,
+      "learning_rate": 4.3500068993761526e-07,
+      "loss": 0.6289,
+      "step": 53451
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.383988380432129,
+      "learning_rate": 4.3489313432959146e-07,
+      "loss": 0.577,
+      "step": 53452
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 4.751138210296631,
+      "learning_rate": 4.347855914154192e-07,
+      "loss": 0.8092,
+      "step": 53453
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.7135560512542725,
+      "learning_rate": 4.346780611953988e-07,
+      "loss": 0.4628,
+      "step": 53454
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.268787384033203,
+      "learning_rate": 4.3457054366982774e-07,
+      "loss": 0.4302,
+      "step": 53455
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.000112533569336,
+      "learning_rate": 4.344630388390059e-07,
+      "loss": 0.7972,
+      "step": 53456
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.5685296058654785,
+      "learning_rate": 4.3435554670323235e-07,
+      "loss": 0.5543,
+      "step": 53457
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.968007802963257,
+      "learning_rate": 4.342480672628052e-07,
+      "loss": 0.7067,
+      "step": 53458
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.7693655490875244,
+      "learning_rate": 4.3414060051802486e-07,
+      "loss": 0.5872,
+      "step": 53459
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.9992141723632812,
+      "learning_rate": 4.340331464691883e-07,
+      "loss": 0.5499,
+      "step": 53460
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.657395601272583,
+      "learning_rate": 4.339257051165946e-07,
+      "loss": 0.5748,
+      "step": 53461
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.8185904026031494,
+      "learning_rate": 4.338182764605442e-07,
+      "loss": 0.691,
+      "step": 53462
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.058216094970703,
+      "learning_rate": 4.337108605013335e-07,
+      "loss": 0.7303,
+      "step": 53463
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 6.508783340454102,
+      "learning_rate": 4.336034572392633e-07,
+      "loss": 0.5023,
+      "step": 53464
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.151425838470459,
+      "learning_rate": 4.3349606667463075e-07,
+      "loss": 0.5036,
+      "step": 53465
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.9569051265716553,
+      "learning_rate": 4.3338868880773543e-07,
+      "loss": 0.5654,
+      "step": 53466
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.346559524536133,
+      "learning_rate": 4.33281323638875e-07,
+      "loss": 0.6179,
+      "step": 53467
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.838038206100464,
+      "learning_rate": 4.33173971168348e-07,
+      "loss": 0.3849,
+      "step": 53468
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.322922468185425,
+      "learning_rate": 4.330666313964549e-07,
+      "loss": 0.3814,
+      "step": 53469
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.010373830795288,
+      "learning_rate": 4.329593043234914e-07,
+      "loss": 0.5542,
+      "step": 53470
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.951101064682007,
+      "learning_rate": 4.3285198994975796e-07,
+      "loss": 0.708,
+      "step": 53471
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.376037836074829,
+      "learning_rate": 4.3274468827555147e-07,
+      "loss": 0.2465,
+      "step": 53472
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.6621408462524414,
+      "learning_rate": 4.3263739930117233e-07,
+      "loss": 0.7612,
+      "step": 53473
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 10.243341445922852,
+      "learning_rate": 4.325301230269163e-07,
+      "loss": 0.4926,
+      "step": 53474
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.5111024379730225,
+      "learning_rate": 4.324228594530838e-07,
+      "loss": 0.8137,
+      "step": 53475
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.5207700729370117,
+      "learning_rate": 4.3231560857997177e-07,
+      "loss": 0.5268,
+      "step": 53476
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.0937349796295166,
+      "learning_rate": 4.3220837040787944e-07,
+      "loss": 0.3841,
+      "step": 53477
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.4053966999053955,
+      "learning_rate": 4.3210114493710376e-07,
+      "loss": 0.516,
+      "step": 53478
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.3908865451812744,
+      "learning_rate": 4.3199393216794505e-07,
+      "loss": 0.7135,
+      "step": 53479
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.64505934715271,
+      "learning_rate": 4.3188673210069754e-07,
+      "loss": 0.6471,
+      "step": 53480
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.4349303245544434,
+      "learning_rate": 4.3177954473566374e-07,
+      "loss": 0.602,
+      "step": 53481
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.2713465690612793,
+      "learning_rate": 4.31672370073139e-07,
+      "loss": 0.6438,
+      "step": 53482
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.500096082687378,
+      "learning_rate": 4.315652081134225e-07,
+      "loss": 0.4461,
+      "step": 53483
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.0573570728302,
+      "learning_rate": 4.314580588568112e-07,
+      "loss": 0.6898,
+      "step": 53484
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.2065277099609375,
+      "learning_rate": 4.313509223036044e-07,
+      "loss": 0.5647,
+      "step": 53485
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.1504271030426025,
+      "learning_rate": 4.312437984540985e-07,
+      "loss": 0.4622,
+      "step": 53486
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.199944257736206,
+      "learning_rate": 4.3113668730859314e-07,
+      "loss": 0.6101,
+      "step": 53487
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.9823014736175537,
+      "learning_rate": 4.3102958886738376e-07,
+      "loss": 0.4662,
+      "step": 53488
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.6986660957336426,
+      "learning_rate": 4.309225031307712e-07,
+      "loss": 0.4087,
+      "step": 53489
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.2450788021087646,
+      "learning_rate": 4.308154300990497e-07,
+      "loss": 0.4287,
+      "step": 53490
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.3493754863739014,
+      "learning_rate": 4.307083697725206e-07,
+      "loss": 0.696,
+      "step": 53491
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 4.102476119995117,
+      "learning_rate": 4.306013221514782e-07,
+      "loss": 0.7474,
+      "step": 53492
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.0328261852264404,
+      "learning_rate": 4.304942872362228e-07,
+      "loss": 0.6551,
+      "step": 53493
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.287790298461914,
+      "learning_rate": 4.3038726502705075e-07,
+      "loss": 0.5986,
+      "step": 53494
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 4.9996490478515625,
+      "learning_rate": 4.3028025552426023e-07,
+      "loss": 0.4455,
+      "step": 53495
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.875807046890259,
+      "learning_rate": 4.3017325872814764e-07,
+      "loss": 0.6011,
+      "step": 53496
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.7969160079956055,
+      "learning_rate": 4.3006627463901216e-07,
+      "loss": 0.6104,
+      "step": 53497
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 4.42070198059082,
+      "learning_rate": 4.2995930325714976e-07,
+      "loss": 0.6935,
+      "step": 53498
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 4.309429168701172,
+      "learning_rate": 4.2985234458285897e-07,
+      "loss": 0.5694,
+      "step": 53499
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 4.240737438201904,
+      "learning_rate": 4.2974539861643625e-07,
+      "loss": 0.7448,
+      "step": 53500
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.650240659713745,
+      "learning_rate": 4.2963846535817976e-07,
+      "loss": 0.6496,
+      "step": 53501
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.9440457820892334,
+      "learning_rate": 4.2953154480838643e-07,
+      "loss": 0.8754,
+      "step": 53502
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.3169026374816895,
+      "learning_rate": 4.2942463696735437e-07,
+      "loss": 0.5212,
+      "step": 53503
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.519639253616333,
+      "learning_rate": 4.2931774183537946e-07,
+      "loss": 0.549,
+      "step": 53504
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.903998851776123,
+      "learning_rate": 4.2921085941276086e-07,
+      "loss": 0.3506,
+      "step": 53505
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.8878889083862305,
+      "learning_rate": 4.291039896997934e-07,
+      "loss": 0.4161,
+      "step": 53506
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.6449880599975586,
+      "learning_rate": 4.289971326967768e-07,
+      "loss": 0.7507,
+      "step": 53507
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 5.0565690994262695,
+      "learning_rate": 4.2889028840400524e-07,
+      "loss": 0.5594,
+      "step": 53508
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.6923439502716064,
+      "learning_rate": 4.287834568217786e-07,
+      "loss": 0.4721,
+      "step": 53509
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.066498041152954,
+      "learning_rate": 4.2867663795039206e-07,
+      "loss": 0.6587,
+      "step": 53510
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.7600009441375732,
+      "learning_rate": 4.285698317901438e-07,
+      "loss": 0.5685,
+      "step": 53511
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.204185724258423,
+      "learning_rate": 4.2846303834133016e-07,
+      "loss": 0.5054,
+      "step": 53512
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.4352123737335205,
+      "learning_rate": 4.2835625760424824e-07,
+      "loss": 0.402,
+      "step": 53513
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.64524245262146,
+      "learning_rate": 4.28249489579195e-07,
+      "loss": 0.422,
+      "step": 53514
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.801449775695801,
+      "learning_rate": 4.281427342664679e-07,
+      "loss": 0.4355,
+      "step": 53515
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.564300537109375,
+      "learning_rate": 4.280359916663618e-07,
+      "loss": 0.5349,
+      "step": 53516
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.8523054122924805,
+      "learning_rate": 4.2792926177917636e-07,
+      "loss": 0.6571,
+      "step": 53517
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.1395721435546875,
+      "learning_rate": 4.278225446052059e-07,
+      "loss": 0.3708,
+      "step": 53518
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.3262712955474854,
+      "learning_rate": 4.277158401447479e-07,
+      "loss": 0.5837,
+      "step": 53519
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.3898584842681885,
+      "learning_rate": 4.2760914839810044e-07,
+      "loss": 0.5783,
+      "step": 53520
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.8003861904144287,
+      "learning_rate": 4.275024693655583e-07,
+      "loss": 0.4436,
+      "step": 53521
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.558230400085449,
+      "learning_rate": 4.27395803047419e-07,
+      "loss": 0.6352,
+      "step": 53522
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.280698776245117,
+      "learning_rate": 4.2728914944397957e-07,
+      "loss": 0.6485,
+      "step": 53523
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 4.685089111328125,
+      "learning_rate": 4.2718250855553533e-07,
+      "loss": 0.9622,
+      "step": 53524
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.6257705688476562,
+      "learning_rate": 4.2707588038238436e-07,
+      "loss": 0.7607,
+      "step": 53525
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.5800843238830566,
+      "learning_rate": 4.269692649248214e-07,
+      "loss": 0.6091,
+      "step": 53526
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.383580446243286,
+      "learning_rate": 4.2686266218314453e-07,
+      "loss": 0.5854,
+      "step": 53527
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.8572773933410645,
+      "learning_rate": 4.267560721576491e-07,
+      "loss": 0.6023,
+      "step": 53528
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.6323349475860596,
+      "learning_rate": 4.266494948486316e-07,
+      "loss": 0.5601,
+      "step": 53529
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.792844295501709,
+      "learning_rate": 4.265429302563895e-07,
+      "loss": 0.579,
+      "step": 53530
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 4.798114776611328,
+      "learning_rate": 4.2643637838121696e-07,
+      "loss": 0.6496,
+      "step": 53531
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.9476027488708496,
+      "learning_rate": 4.2632983922341275e-07,
+      "loss": 0.553,
+      "step": 53532
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.584744691848755,
+      "learning_rate": 4.26223312783271e-07,
+      "loss": 0.4774,
+      "step": 53533
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.9603183269500732,
+      "learning_rate": 4.261167990610898e-07,
+      "loss": 0.372,
+      "step": 53534
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.9445960521698,
+      "learning_rate": 4.26010298057164e-07,
+      "loss": 0.742,
+      "step": 53535
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.768882989883423,
+      "learning_rate": 4.2590380977179045e-07,
+      "loss": 0.4035,
+      "step": 53536
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.468416213989258,
+      "learning_rate": 4.25797334205264e-07,
+      "loss": 0.4824,
+      "step": 53537
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.8769800662994385,
+      "learning_rate": 4.2569087135788225e-07,
+      "loss": 0.3965,
+      "step": 53538
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.5010886192321777,
+      "learning_rate": 4.2558442122994035e-07,
+      "loss": 0.5103,
+      "step": 53539
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 1.9748302698135376,
+      "learning_rate": 4.254779838217349e-07,
+      "loss": 0.5278,
+      "step": 53540
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.802938461303711,
+      "learning_rate": 4.2537155913356e-07,
+      "loss": 0.6186,
+      "step": 53541
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.740569591522217,
+      "learning_rate": 4.2526514716571487e-07,
+      "loss": 0.5426,
+      "step": 53542
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.5334818363189697,
+      "learning_rate": 4.2515874791849265e-07,
+      "loss": 0.4519,
+      "step": 53543
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.8045215606689453,
+      "learning_rate": 4.250523613921914e-07,
+      "loss": 0.6566,
+      "step": 53544
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.36281156539917,
+      "learning_rate": 4.249459875871048e-07,
+      "loss": 0.5675,
+      "step": 53545
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.0514159202575684,
+      "learning_rate": 4.2483962650352986e-07,
+      "loss": 0.417,
+      "step": 53546
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.2040371894836426,
+      "learning_rate": 4.247332781417618e-07,
+      "loss": 0.3642,
+      "step": 53547
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.3480489253997803,
+      "learning_rate": 4.2462694250209713e-07,
+      "loss": 0.5436,
+      "step": 53548
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.0252201557159424,
+      "learning_rate": 4.2452061958483005e-07,
+      "loss": 0.2867,
+      "step": 53549
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.644446611404419,
+      "learning_rate": 4.244143093902581e-07,
+      "loss": 0.5895,
+      "step": 53550
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.8524036407470703,
+      "learning_rate": 4.243080119186749e-07,
+      "loss": 0.4784,
+      "step": 53551
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.0612173080444336,
+      "learning_rate": 4.24201727170378e-07,
+      "loss": 0.4788,
+      "step": 53552
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.8731589317321777,
+      "learning_rate": 4.240954551456611e-07,
+      "loss": 0.5648,
+      "step": 53553
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.5491573810577393,
+      "learning_rate": 4.239891958448211e-07,
+      "loss": 0.7198,
+      "step": 53554
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.4226460456848145,
+      "learning_rate": 4.238829492681523e-07,
+      "loss": 0.5573,
+      "step": 53555
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 4.797149181365967,
+      "learning_rate": 4.237767154159522e-07,
+      "loss": 0.6256,
+      "step": 53556
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.3147614002227783,
+      "learning_rate": 4.2367049428851337e-07,
+      "loss": 0.7134,
+      "step": 53557
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.3534016609191895,
+      "learning_rate": 4.235642858861333e-07,
+      "loss": 0.6926,
+      "step": 53558
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.5086171627044678,
+      "learning_rate": 4.2345809020910566e-07,
+      "loss": 0.3989,
+      "step": 53559
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.313728094100952,
+      "learning_rate": 4.23351907257728e-07,
+      "loss": 0.6555,
+      "step": 53560
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.4769232273101807,
+      "learning_rate": 4.232457370322929e-07,
+      "loss": 0.6656,
+      "step": 53561
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.3290293216705322,
+      "learning_rate": 4.2313957953309726e-07,
+      "loss": 0.7716,
+      "step": 53562
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 4.187580108642578,
+      "learning_rate": 4.230334347604359e-07,
+      "loss": 0.5258,
+      "step": 53563
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.96946382522583,
+      "learning_rate": 4.229273027146047e-07,
+      "loss": 0.6494,
+      "step": 53564
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.8710618019104004,
+      "learning_rate": 4.228211833958967e-07,
+      "loss": 0.3274,
+      "step": 53565
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.181874990463257,
+      "learning_rate": 4.227150768046101e-07,
+      "loss": 0.4519,
+      "step": 53566
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.7711238861083984,
+      "learning_rate": 4.226089829410368e-07,
+      "loss": 0.5359,
+      "step": 53567
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.397753953933716,
+      "learning_rate": 4.2250290180547435e-07,
+      "loss": 0.7184,
+      "step": 53568
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.389653205871582,
+      "learning_rate": 4.2239683339821536e-07,
+      "loss": 0.7228,
+      "step": 53569
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.6904287338256836,
+      "learning_rate": 4.222907777195573e-07,
+      "loss": 0.6407,
+      "step": 53570
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.0276854038238525,
+      "learning_rate": 4.221847347697922e-07,
+      "loss": 0.3738,
+      "step": 53571
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.6476566791534424,
+      "learning_rate": 4.2207870454921806e-07,
+      "loss": 0.725,
+      "step": 53572
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 5.2884392738342285,
+      "learning_rate": 4.2197268705812644e-07,
+      "loss": 0.6461,
+      "step": 53573
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.8750462532043457,
+      "learning_rate": 4.2186668229681537e-07,
+      "loss": 0.5967,
+      "step": 53574
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.5437276363372803,
+      "learning_rate": 4.2176069026557686e-07,
+      "loss": 0.5422,
+      "step": 53575
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.181218147277832,
+      "learning_rate": 4.216547109647079e-07,
+      "loss": 0.6949,
+      "step": 53576
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.4852728843688965,
+      "learning_rate": 4.21548744394501e-07,
+      "loss": 0.7112,
+      "step": 53577
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.494448184967041,
+      "learning_rate": 4.214427905552532e-07,
+      "loss": 0.548,
+      "step": 53578
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.447051763534546,
+      "learning_rate": 4.213368494472564e-07,
+      "loss": 0.5339,
+      "step": 53579
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.4672231674194336,
+      "learning_rate": 4.212309210708071e-07,
+      "loss": 0.2889,
+      "step": 53580
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.85137677192688,
+      "learning_rate": 4.2112500542620003e-07,
+      "loss": 0.7141,
+      "step": 53581
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.6405062675476074,
+      "learning_rate": 4.210191025137278e-07,
+      "loss": 0.4795,
+      "step": 53582
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.578340530395508,
+      "learning_rate": 4.2091321233368676e-07,
+      "loss": 0.7495,
+      "step": 53583
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.528393030166626,
+      "learning_rate": 4.2080733488637115e-07,
+      "loss": 0.539,
+      "step": 53584
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.963613510131836,
+      "learning_rate": 4.207014701720741e-07,
+      "loss": 0.6718,
+      "step": 53585
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.7671282291412354,
+      "learning_rate": 4.2059561819109194e-07,
+      "loss": 0.4827,
+      "step": 53586
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.974459409713745,
+      "learning_rate": 4.204897789437168e-07,
+      "loss": 0.6504,
+      "step": 53587
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 5.67002534866333,
+      "learning_rate": 4.20383952430245e-07,
+      "loss": 0.5463,
+      "step": 53588
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 4.008198261260986,
+      "learning_rate": 4.202781386509691e-07,
+      "loss": 0.6206,
+      "step": 53589
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 4.721677780151367,
+      "learning_rate": 4.2017233760618447e-07,
+      "loss": 0.5552,
+      "step": 53590
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.490633964538574,
+      "learning_rate": 4.200665492961847e-07,
+      "loss": 0.546,
+      "step": 53591
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.880127429962158,
+      "learning_rate": 4.1996077372126354e-07,
+      "loss": 0.6077,
+      "step": 53592
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.3801465034484863,
+      "learning_rate": 4.198550108817173e-07,
+      "loss": 0.4234,
+      "step": 53593
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.173712730407715,
+      "learning_rate": 4.1974926077783697e-07,
+      "loss": 0.613,
+      "step": 53594
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.1370363235473633,
+      "learning_rate": 4.196435234099194e-07,
+      "loss": 0.6369,
+      "step": 53595
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.8429806232452393,
+      "learning_rate": 4.1953779877825675e-07,
+      "loss": 0.5355,
+      "step": 53596
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.519892930984497,
+      "learning_rate": 4.1943208688314417e-07,
+      "loss": 0.6757,
+      "step": 53597
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.9319546222686768,
+      "learning_rate": 4.1932638772487434e-07,
+      "loss": 0.4541,
+      "step": 53598
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 4.407080173492432,
+      "learning_rate": 4.192207013037425e-07,
+      "loss": 0.3492,
+      "step": 53599
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.9866037368774414,
+      "learning_rate": 4.1911502762004063e-07,
+      "loss": 0.4042,
+      "step": 53600
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.4461400508880615,
+      "learning_rate": 4.1900936667406524e-07,
+      "loss": 0.4331,
+      "step": 53601
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.1818766593933105,
+      "learning_rate": 4.189037184661077e-07,
+      "loss": 0.3194,
+      "step": 53602
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 5.150166988372803,
+      "learning_rate": 4.187980829964633e-07,
+      "loss": 0.5758,
+      "step": 53603
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 6.643280029296875,
+      "learning_rate": 4.1869246026542463e-07,
+      "loss": 0.5699,
+      "step": 53604
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 4.287621021270752,
+      "learning_rate": 4.1858685027328703e-07,
+      "loss": 0.4584,
+      "step": 53605
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.2601749897003174,
+      "learning_rate": 4.1848125302034247e-07,
+      "loss": 0.5077,
+      "step": 53606
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.339597225189209,
+      "learning_rate": 4.1837566850688625e-07,
+      "loss": 0.6008,
+      "step": 53607
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 4.271677494049072,
+      "learning_rate": 4.1827009673320985e-07,
+      "loss": 0.8267,
+      "step": 53608
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 4.411992073059082,
+      "learning_rate": 4.1816453769960853e-07,
+      "loss": 0.6601,
+      "step": 53609
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.1902477741241455,
+      "learning_rate": 4.180589914063748e-07,
+      "loss": 0.6846,
+      "step": 53610
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 4.580879211425781,
+      "learning_rate": 4.179534578538036e-07,
+      "loss": 0.6411,
+      "step": 53611
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.0788450241088867,
+      "learning_rate": 4.178479370421862e-07,
+      "loss": 0.7628,
+      "step": 53612
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.525587797164917,
+      "learning_rate": 4.1774242897181795e-07,
+      "loss": 0.4977,
+      "step": 53613
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.838343858718872,
+      "learning_rate": 4.1763693364299083e-07,
+      "loss": 0.6008,
+      "step": 53614
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.3013792037963867,
+      "learning_rate": 4.175314510559997e-07,
+      "loss": 0.7301,
+      "step": 53615
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.2107388973236084,
+      "learning_rate": 4.1742598121113586e-07,
+      "loss": 0.631,
+      "step": 53616
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.335061550140381,
+      "learning_rate": 4.1732052410869474e-07,
+      "loss": 0.6028,
+      "step": 53617
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.0272130966186523,
+      "learning_rate": 4.1721507974896767e-07,
+      "loss": 0.7292,
+      "step": 53618
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.1587893962860107,
+      "learning_rate": 4.1710964813224953e-07,
+      "loss": 0.5463,
+      "step": 53619
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.9692718982696533,
+      "learning_rate": 4.1700422925883165e-07,
+      "loss": 0.6173,
+      "step": 53620
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.789210319519043,
+      "learning_rate": 4.168988231290089e-07,
+      "loss": 0.7092,
+      "step": 53621
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.383162260055542,
+      "learning_rate": 4.167934297430731e-07,
+      "loss": 0.5491,
+      "step": 53622
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.864900827407837,
+      "learning_rate": 4.166880491013181e-07,
+      "loss": 0.5841,
+      "step": 53623
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.025404691696167,
+      "learning_rate": 4.1658268120403634e-07,
+      "loss": 0.5559,
+      "step": 53624
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.2639718055725098,
+      "learning_rate": 4.16477326051522e-07,
+      "loss": 0.516,
+      "step": 53625
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.301614761352539,
+      "learning_rate": 4.163719836440666e-07,
+      "loss": 0.6798,
+      "step": 53626
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.479931592941284,
+      "learning_rate": 4.1626665398196433e-07,
+      "loss": 0.648,
+      "step": 53627
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.717216730117798,
+      "learning_rate": 4.1616133706550653e-07,
+      "loss": 0.5634,
+      "step": 53628
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.495898485183716,
+      "learning_rate": 4.1605603289498755e-07,
+      "loss": 0.4156,
+      "step": 53629
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.5228374004364014,
+      "learning_rate": 4.1595074147069923e-07,
+      "loss": 0.5678,
+      "step": 53630
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.3718318939208984,
+      "learning_rate": 4.158454627929348e-07,
+      "loss": 0.4308,
+      "step": 53631
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 4.922532081604004,
+      "learning_rate": 4.1574019686198674e-07,
+      "loss": 0.7402,
+      "step": 53632
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.290238380432129,
+      "learning_rate": 4.1563494367814814e-07,
+      "loss": 0.3876,
+      "step": 53633
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.4601476192474365,
+      "learning_rate": 4.1552970324171107e-07,
+      "loss": 0.6601,
+      "step": 53634
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.70735239982605,
+      "learning_rate": 4.154244755529691e-07,
+      "loss": 0.5568,
+      "step": 53635
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.157837152481079,
+      "learning_rate": 4.153192606122136e-07,
+      "loss": 0.512,
+      "step": 53636
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.0519516468048096,
+      "learning_rate": 4.1521405841973836e-07,
+      "loss": 0.7554,
+      "step": 53637
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.2189137935638428,
+      "learning_rate": 4.151088689758348e-07,
+      "loss": 0.5975,
+      "step": 53638
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.5936272144317627,
+      "learning_rate": 4.150036922807965e-07,
+      "loss": 0.8127,
+      "step": 53639
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.443141222000122,
+      "learning_rate": 4.148985283349144e-07,
+      "loss": 0.4831,
+      "step": 53640
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.5726096630096436,
+      "learning_rate": 4.147933771384821e-07,
+      "loss": 0.6253,
+      "step": 53641
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.6578586101531982,
+      "learning_rate": 4.1468823869179275e-07,
+      "loss": 0.481,
+      "step": 53642
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.732177734375,
+      "learning_rate": 4.1458311299513667e-07,
+      "loss": 0.5356,
+      "step": 53643
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 4.081407070159912,
+      "learning_rate": 4.1447800004880687e-07,
+      "loss": 0.7746,
+      "step": 53644
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.9166533946990967,
+      "learning_rate": 4.143728998530977e-07,
+      "loss": 0.6552,
+      "step": 53645
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.881232500076294,
+      "learning_rate": 4.142678124082983e-07,
+      "loss": 0.5493,
+      "step": 53646
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.8136000633239746,
+      "learning_rate": 4.141627377147028e-07,
+      "loss": 0.4876,
+      "step": 53647
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 4.700672626495361,
+      "learning_rate": 4.1405767577260223e-07,
+      "loss": 0.6761,
+      "step": 53648
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.3542611598968506,
+      "learning_rate": 4.1395262658229017e-07,
+      "loss": 0.4342,
+      "step": 53649
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 4.06215763092041,
+      "learning_rate": 4.138475901440575e-07,
+      "loss": 0.4786,
+      "step": 53650
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.660613536834717,
+      "learning_rate": 4.137425664581968e-07,
+      "loss": 0.49,
+      "step": 53651
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.5811009407043457,
+      "learning_rate": 4.136375555249994e-07,
+      "loss": 0.4511,
+      "step": 53652
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.7405776977539062,
+      "learning_rate": 4.1353255734475796e-07,
+      "loss": 0.8191,
+      "step": 53653
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.4816863536834717,
+      "learning_rate": 4.134275719177655e-07,
+      "loss": 0.4729,
+      "step": 53654
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.4011008739471436,
+      "learning_rate": 4.133225992443113e-07,
+      "loss": 0.4851,
+      "step": 53655
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.260734796524048,
+      "learning_rate": 4.1321763932469007e-07,
+      "loss": 0.8299,
+      "step": 53656
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.4813969135284424,
+      "learning_rate": 4.131126921591916e-07,
+      "loss": 0.3915,
+      "step": 53657
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 4.813287734985352,
+      "learning_rate": 4.1300775774810896e-07,
+      "loss": 0.7153,
+      "step": 53658
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.9979305267333984,
+      "learning_rate": 4.129028360917325e-07,
+      "loss": 0.3984,
+      "step": 53659
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.4800291061401367,
+      "learning_rate": 4.127979271903559e-07,
+      "loss": 0.5922,
+      "step": 53660
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 5.840725421905518,
+      "learning_rate": 4.1269303104426897e-07,
+      "loss": 0.4231,
+      "step": 53661
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.0218005180358887,
+      "learning_rate": 4.125881476537652e-07,
+      "loss": 0.8006,
+      "step": 53662
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.8103854656219482,
+      "learning_rate": 4.124832770191339e-07,
+      "loss": 0.4067,
+      "step": 53663
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.708681106567383,
+      "learning_rate": 4.123784191406688e-07,
+      "loss": 0.5058,
+      "step": 53664
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.309201955795288,
+      "learning_rate": 4.1227357401866063e-07,
+      "loss": 0.6005,
+      "step": 53665
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.0581398010253906,
+      "learning_rate": 4.121687416534015e-07,
+      "loss": 0.631,
+      "step": 53666
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.331394672393799,
+      "learning_rate": 4.120639220451822e-07,
+      "loss": 0.3653,
+      "step": 53667
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.3570332527160645,
+      "learning_rate": 4.1195911519429476e-07,
+      "loss": 0.5378,
+      "step": 53668
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 6.792757511138916,
+      "learning_rate": 4.118543211010301e-07,
+      "loss": 0.6137,
+      "step": 53669
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.037898540496826,
+      "learning_rate": 4.1174953976568013e-07,
+      "loss": 0.4444,
+      "step": 53670
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.7823328971862793,
+      "learning_rate": 4.116447711885352e-07,
+      "loss": 0.5056,
+      "step": 53671
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.5961899757385254,
+      "learning_rate": 4.1154001536988785e-07,
+      "loss": 0.5226,
+      "step": 53672
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.5209929943084717,
+      "learning_rate": 4.1143527231002844e-07,
+      "loss": 0.5431,
+      "step": 53673
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.6688055992126465,
+      "learning_rate": 4.1133054200924947e-07,
+      "loss": 0.6156,
+      "step": 53674
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.112205743789673,
+      "learning_rate": 4.1122582446784023e-07,
+      "loss": 0.7462,
+      "step": 53675
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.909654378890991,
+      "learning_rate": 4.111211196860937e-07,
+      "loss": 0.92,
+      "step": 53676
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.7419660091400146,
+      "learning_rate": 4.110164276642997e-07,
+      "loss": 0.7535,
+      "step": 53677
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.705073833465576,
+      "learning_rate": 4.1091174840275083e-07,
+      "loss": 0.6115,
+      "step": 53678
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.534715175628662,
+      "learning_rate": 4.1080708190173625e-07,
+      "loss": 0.4926,
+      "step": 53679
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 11.04839038848877,
+      "learning_rate": 4.1070242816154905e-07,
+      "loss": 0.5753,
+      "step": 53680
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.2879040241241455,
+      "learning_rate": 4.1059778718247843e-07,
+      "loss": 0.6963,
+      "step": 53681
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.907280206680298,
+      "learning_rate": 4.1049315896481646e-07,
+      "loss": 0.4521,
+      "step": 53682
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.563215970993042,
+      "learning_rate": 4.1038854350885337e-07,
+      "loss": 0.3762,
+      "step": 53683
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.275517463684082,
+      "learning_rate": 4.102839408148801e-07,
+      "loss": 0.6568,
+      "step": 53684
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.0374529361724854,
+      "learning_rate": 4.1017935088318803e-07,
+      "loss": 0.359,
+      "step": 53685
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.7751359939575195,
+      "learning_rate": 4.1007477371406923e-07,
+      "loss": 0.6136,
+      "step": 53686
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.4462661743164062,
+      "learning_rate": 4.0997020930781117e-07,
+      "loss": 0.6358,
+      "step": 53687
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 1.5432010889053345,
+      "learning_rate": 4.0986565766470807e-07,
+      "loss": 0.3839,
+      "step": 53688
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.8290224075317383,
+      "learning_rate": 4.097611187850481e-07,
+      "loss": 0.4619,
+      "step": 53689
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.9182584285736084,
+      "learning_rate": 4.0965659266912374e-07,
+      "loss": 0.5156,
+      "step": 53690
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.313079595565796,
+      "learning_rate": 4.095520793172236e-07,
+      "loss": 0.4307,
+      "step": 53691
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.6827094554901123,
+      "learning_rate": 4.094475787296409e-07,
+      "loss": 0.6111,
+      "step": 53692
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 4.582637786865234,
+      "learning_rate": 4.0934309090666367e-07,
+      "loss": 0.6103,
+      "step": 53693
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.648123025894165,
+      "learning_rate": 4.092386158485845e-07,
+      "loss": 0.4714,
+      "step": 53694
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.7507643699645996,
+      "learning_rate": 4.09134153555692e-07,
+      "loss": 0.4841,
+      "step": 53695
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.1070752143859863,
+      "learning_rate": 4.0902970402827866e-07,
+      "loss": 0.5513,
+      "step": 53696
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.474080801010132,
+      "learning_rate": 4.089252672666333e-07,
+      "loss": 0.9052,
+      "step": 53697
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.790674924850464,
+      "learning_rate": 4.088208432710472e-07,
+      "loss": 0.5137,
+      "step": 53698
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.5700154304504395,
+      "learning_rate": 4.0871643204180965e-07,
+      "loss": 0.8598,
+      "step": 53699
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.1195263862609863,
+      "learning_rate": 4.086120335792132e-07,
+      "loss": 0.6508,
+      "step": 53700
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.3013906478881836,
+      "learning_rate": 4.0850764788354535e-07,
+      "loss": 0.5339,
+      "step": 53701
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.3471689224243164,
+      "learning_rate": 4.084032749550987e-07,
+      "loss": 0.7686,
+      "step": 53702
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.773763656616211,
+      "learning_rate": 4.0829891479416184e-07,
+      "loss": 0.4676,
+      "step": 53703
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.767688512802124,
+      "learning_rate": 4.0819456740102516e-07,
+      "loss": 0.5263,
+      "step": 53704
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.4787886142730713,
+      "learning_rate": 4.080902327759795e-07,
+      "loss": 0.4328,
+      "step": 53705
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.431253433227539,
+      "learning_rate": 4.079859109193157e-07,
+      "loss": 0.4774,
+      "step": 53706
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.053021192550659,
+      "learning_rate": 4.07881601831322e-07,
+      "loss": 0.6578,
+      "step": 53707
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.776810646057129,
+      "learning_rate": 4.077773055122902e-07,
+      "loss": 0.7752,
+      "step": 53708
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.8015317916870117,
+      "learning_rate": 4.076730219625086e-07,
+      "loss": 0.6793,
+      "step": 53709
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.5264577865600586,
+      "learning_rate": 4.075687511822685e-07,
+      "loss": 0.5663,
+      "step": 53710
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.292858600616455,
+      "learning_rate": 4.074644931718591e-07,
+      "loss": 0.5421,
+      "step": 53711
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.4140465259552,
+      "learning_rate": 4.073602479315708e-07,
+      "loss": 0.4007,
+      "step": 53712
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.9922690391540527,
+      "learning_rate": 4.072560154616928e-07,
+      "loss": 0.5387,
+      "step": 53713
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.7732229232788086,
+      "learning_rate": 4.071517957625148e-07,
+      "loss": 0.7325,
+      "step": 53714
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.4922144412994385,
+      "learning_rate": 4.070475888343284e-07,
+      "loss": 0.6482,
+      "step": 53715
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.9292378425598145,
+      "learning_rate": 4.0694339467742095e-07,
+      "loss": 0.5548,
+      "step": 53716
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.0961952209472656,
+      "learning_rate": 4.0683921329208455e-07,
+      "loss": 0.5416,
+      "step": 53717
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.401754379272461,
+      "learning_rate": 4.067350446786067e-07,
+      "loss": 0.5791,
+      "step": 53718
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.8914082050323486,
+      "learning_rate": 4.0663088883727886e-07,
+      "loss": 0.6996,
+      "step": 53719
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.2916905879974365,
+      "learning_rate": 4.0652674576838856e-07,
+      "loss": 0.4835,
+      "step": 53720
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 4.718326091766357,
+      "learning_rate": 4.064226154722278e-07,
+      "loss": 0.5903,
+      "step": 53721
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.9613828659057617,
+      "learning_rate": 4.0631849794908417e-07,
+      "loss": 0.7194,
+      "step": 53722
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.9066572189331055,
+      "learning_rate": 4.06214393199249e-07,
+      "loss": 0.5947,
+      "step": 53723
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.642618179321289,
+      "learning_rate": 4.0611030122300944e-07,
+      "loss": 0.8208,
+      "step": 53724
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.3027892112731934,
+      "learning_rate": 4.060062220206568e-07,
+      "loss": 0.5688,
+      "step": 53725
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.8253426551818848,
+      "learning_rate": 4.0590215559247915e-07,
+      "loss": 0.5635,
+      "step": 53726
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.6217916011810303,
+      "learning_rate": 4.0579810193876803e-07,
+      "loss": 0.5401,
+      "step": 53727
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.243323802947998,
+      "learning_rate": 4.0569406105981036e-07,
+      "loss": 0.6207,
+      "step": 53728
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.2371561527252197,
+      "learning_rate": 4.0559003295589704e-07,
+      "loss": 0.7678,
+      "step": 53729
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.850818634033203,
+      "learning_rate": 4.054860176273162e-07,
+      "loss": 0.5065,
+      "step": 53730
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.2831127643585205,
+      "learning_rate": 4.0538201507435816e-07,
+      "loss": 0.5097,
+      "step": 53731
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.3879501819610596,
+      "learning_rate": 4.0527802529731097e-07,
+      "loss": 0.4943,
+      "step": 53732
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.1584200859069824,
+      "learning_rate": 4.05174048296465e-07,
+      "loss": 0.3576,
+      "step": 53733
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.5219459533691406,
+      "learning_rate": 4.0507008407210834e-07,
+      "loss": 0.7213,
+      "step": 53734
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.2190120220184326,
+      "learning_rate": 4.049661326245308e-07,
+      "loss": 0.6692,
+      "step": 53735
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.628514051437378,
+      "learning_rate": 4.0486219395402036e-07,
+      "loss": 0.7046,
+      "step": 53736
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.4512906074523926,
+      "learning_rate": 4.0475826806086805e-07,
+      "loss": 0.5731,
+      "step": 53737
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 4.4306511878967285,
+      "learning_rate": 4.046543549453602e-07,
+      "loss": 0.4899,
+      "step": 53738
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 4.1788482666015625,
+      "learning_rate": 4.045504546077883e-07,
+      "loss": 0.5068,
+      "step": 53739
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.0807437896728516,
+      "learning_rate": 4.044465670484393e-07,
+      "loss": 0.6393,
+      "step": 53740
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.1299800872802734,
+      "learning_rate": 4.043426922676036e-07,
+      "loss": 0.8383,
+      "step": 53741
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 4.7515549659729,
+      "learning_rate": 4.042388302655681e-07,
+      "loss": 0.9233,
+      "step": 53742
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.8879761695861816,
+      "learning_rate": 4.0413498104262426e-07,
+      "loss": 0.6526,
+      "step": 53743
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.033756971359253,
+      "learning_rate": 4.0403114459905804e-07,
+      "loss": 0.5446,
+      "step": 53744
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.7493207454681396,
+      "learning_rate": 4.0392732093515964e-07,
+      "loss": 0.5761,
+      "step": 53745
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.9078540802001953,
+      "learning_rate": 4.038235100512178e-07,
+      "loss": 0.5655,
+      "step": 53746
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.9341275691986084,
+      "learning_rate": 4.037197119475217e-07,
+      "loss": 0.4818,
+      "step": 53747
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.66215181350708,
+      "learning_rate": 4.036159266243583e-07,
+      "loss": 0.5327,
+      "step": 53748
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.3164236545562744,
+      "learning_rate": 4.035121540820186e-07,
+      "loss": 0.3663,
+      "step": 53749
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 1.9117969274520874,
+      "learning_rate": 4.0340839432078835e-07,
+      "loss": 0.3894,
+      "step": 53750
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.658271312713623,
+      "learning_rate": 4.0330464734095854e-07,
+      "loss": 0.6016,
+      "step": 53751
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.318392038345337,
+      "learning_rate": 4.032009131428155e-07,
+      "loss": 0.7499,
+      "step": 53752
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.365708827972412,
+      "learning_rate": 4.0309719172664964e-07,
+      "loss": 0.6365,
+      "step": 53753
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.3886351585388184,
+      "learning_rate": 4.029934830927473e-07,
+      "loss": 0.4925,
+      "step": 53754
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.049597978591919,
+      "learning_rate": 4.0288978724139946e-07,
+      "loss": 0.6118,
+      "step": 53755
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.558104991912842,
+      "learning_rate": 4.027861041728914e-07,
+      "loss": 0.4998,
+      "step": 53756
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.8446497917175293,
+      "learning_rate": 4.026824338875146e-07,
+      "loss": 0.8317,
+      "step": 53757
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.7818355560302734,
+      "learning_rate": 4.025787763855549e-07,
+      "loss": 0.5167,
+      "step": 53758
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.698820114135742,
+      "learning_rate": 4.0247513166730146e-07,
+      "loss": 0.4126,
+      "step": 53759
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.161818504333496,
+      "learning_rate": 4.0237149973304244e-07,
+      "loss": 0.4179,
+      "step": 53760
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.690100908279419,
+      "learning_rate": 4.0226788058306656e-07,
+      "loss": 0.5602,
+      "step": 53761
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.920456886291504,
+      "learning_rate": 4.0216427421766014e-07,
+      "loss": 0.5604,
+      "step": 53762
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.2672250270843506,
+      "learning_rate": 4.020606806371136e-07,
+      "loss": 0.6355,
+      "step": 53763
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.2032177448272705,
+      "learning_rate": 4.019570998417127e-07,
+      "loss": 0.6076,
+      "step": 53764
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.9047582149505615,
+      "learning_rate": 4.0185353183174735e-07,
+      "loss": 0.686,
+      "step": 53765
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.479092836380005,
+      "learning_rate": 4.017499766075045e-07,
+      "loss": 0.5207,
+      "step": 53766
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.3386378288269043,
+      "learning_rate": 4.0164643416927274e-07,
+      "loss": 0.618,
+      "step": 53767
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.0166308879852295,
+      "learning_rate": 4.015429045173397e-07,
+      "loss": 0.8794,
+      "step": 53768
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.159205436706543,
+      "learning_rate": 4.014393876519934e-07,
+      "loss": 0.8353,
+      "step": 53769
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.345128059387207,
+      "learning_rate": 4.0133588357352094e-07,
+      "loss": 0.4716,
+      "step": 53770
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.765615701675415,
+      "learning_rate": 4.0123239228221147e-07,
+      "loss": 0.3864,
+      "step": 53771
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.0842604637145996,
+      "learning_rate": 4.011289137783514e-07,
+      "loss": 0.4891,
+      "step": 53772
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.9464755058288574,
+      "learning_rate": 4.0102544806222944e-07,
+      "loss": 0.6054,
+      "step": 53773
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.344132900238037,
+      "learning_rate": 4.00921995134132e-07,
+      "loss": 0.6538,
+      "step": 53774
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.2659170627593994,
+      "learning_rate": 4.0081855499434887e-07,
+      "loss": 0.4437,
+      "step": 53775
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.0681962966918945,
+      "learning_rate": 4.0071512764316536e-07,
+      "loss": 0.4534,
+      "step": 53776
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.098529815673828,
+      "learning_rate": 4.006117130808701e-07,
+      "loss": 0.4792,
+      "step": 53777
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.8630664348602295,
+      "learning_rate": 4.005083113077518e-07,
+      "loss": 0.8084,
+      "step": 53778
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.9761295318603516,
+      "learning_rate": 4.004049223240963e-07,
+      "loss": 0.4866,
+      "step": 53779
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.085088014602661,
+      "learning_rate": 4.0030154613019177e-07,
+      "loss": 0.5532,
+      "step": 53780
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.023207187652588,
+      "learning_rate": 4.0019818272632516e-07,
+      "loss": 0.5563,
+      "step": 53781
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.799560070037842,
+      "learning_rate": 4.000948321127851e-07,
+      "loss": 0.3944,
+      "step": 53782
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.6501238346099854,
+      "learning_rate": 3.9999149428985694e-07,
+      "loss": 0.7407,
+      "step": 53783
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 4.00204610824585,
+      "learning_rate": 3.9988816925783047e-07,
+      "loss": 0.4716,
+      "step": 53784
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.074091672897339,
+      "learning_rate": 3.99784857016991e-07,
+      "loss": 0.54,
+      "step": 53785
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.749457836151123,
+      "learning_rate": 3.996815575676266e-07,
+      "loss": 0.5735,
+      "step": 53786
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.1780972480773926,
+      "learning_rate": 3.995782709100249e-07,
+      "loss": 0.6359,
+      "step": 53787
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.4416942596435547,
+      "learning_rate": 3.9947499704447223e-07,
+      "loss": 0.4733,
+      "step": 53788
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.730452060699463,
+      "learning_rate": 3.993717359712557e-07,
+      "loss": 0.4098,
+      "step": 53789
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.838951587677002,
+      "learning_rate": 3.9926848769066385e-07,
+      "loss": 0.5589,
+      "step": 53790
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.6314754486083984,
+      "learning_rate": 3.9916525220298263e-07,
+      "loss": 0.3345,
+      "step": 53791
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.4916982650756836,
+      "learning_rate": 3.990620295084996e-07,
+      "loss": 0.5019,
+      "step": 53792
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.8857574462890625,
+      "learning_rate": 3.9895881960750116e-07,
+      "loss": 0.7228,
+      "step": 53793
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.944589853286743,
+      "learning_rate": 3.988556225002749e-07,
+      "loss": 0.5619,
+      "step": 53794
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.7064836025238037,
+      "learning_rate": 3.987524381871072e-07,
+      "loss": 0.4059,
+      "step": 53795
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.6690940856933594,
+      "learning_rate": 3.986492666682856e-07,
+      "loss": 0.5539,
+      "step": 53796
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.892690420150757,
+      "learning_rate": 3.985461079440961e-07,
+      "loss": 0.6127,
+      "step": 53797
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.949679136276245,
+      "learning_rate": 3.984429620148272e-07,
+      "loss": 0.6393,
+      "step": 53798
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.5829479694366455,
+      "learning_rate": 3.9833982888076383e-07,
+      "loss": 0.8218,
+      "step": 53799
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.29561710357666,
+      "learning_rate": 3.9823670854219396e-07,
+      "loss": 0.5991,
+      "step": 53800
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 4.662498950958252,
+      "learning_rate": 3.9813360099940355e-07,
+      "loss": 0.5737,
+      "step": 53801
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 4.094954967498779,
+      "learning_rate": 3.9803050625268015e-07,
+      "loss": 0.7054,
+      "step": 53802
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.978276014328003,
+      "learning_rate": 3.979274243023096e-07,
+      "loss": 0.4337,
+      "step": 53803
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 4.2277607917785645,
+      "learning_rate": 3.9782435514857945e-07,
+      "loss": 0.7322,
+      "step": 53804
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 4.251157283782959,
+      "learning_rate": 3.97721298791775e-07,
+      "loss": 0.297,
+      "step": 53805
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.8073456287384033,
+      "learning_rate": 3.9761825523218333e-07,
+      "loss": 0.4852,
+      "step": 53806
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.897709608078003,
+      "learning_rate": 3.975152244700914e-07,
+      "loss": 0.5873,
+      "step": 53807
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.636894941329956,
+      "learning_rate": 3.9741220650578615e-07,
+      "loss": 0.6452,
+      "step": 53808
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.8657922744750977,
+      "learning_rate": 3.9730920133955297e-07,
+      "loss": 0.5635,
+      "step": 53809
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.0414164066314697,
+      "learning_rate": 3.972062089716794e-07,
+      "loss": 0.6423,
+      "step": 53810
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.2014758586883545,
+      "learning_rate": 3.971032294024496e-07,
+      "loss": 0.5704,
+      "step": 53811
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.5862319469451904,
+      "learning_rate": 3.9700026263215283e-07,
+      "loss": 0.493,
+      "step": 53812
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.4936487674713135,
+      "learning_rate": 3.9689730866107336e-07,
+      "loss": 0.6489,
+      "step": 53813
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.5772972106933594,
+      "learning_rate": 3.967943674894986e-07,
+      "loss": 0.6054,
+      "step": 53814
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.203399658203125,
+      "learning_rate": 3.9669143911771404e-07,
+      "loss": 0.5586,
+      "step": 53815
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.350006341934204,
+      "learning_rate": 3.965885235460065e-07,
+      "loss": 0.4236,
+      "step": 53816
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 5.034145832061768,
+      "learning_rate": 3.964856207746609e-07,
+      "loss": 0.6334,
+      "step": 53817
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.283787250518799,
+      "learning_rate": 3.963827308039653e-07,
+      "loss": 0.5552,
+      "step": 53818
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.1314871311187744,
+      "learning_rate": 3.9627985363420384e-07,
+      "loss": 0.5763,
+      "step": 53819
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.0160017013549805,
+      "learning_rate": 3.961769892656642e-07,
+      "loss": 0.628,
+      "step": 53820
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.8441073894500732,
+      "learning_rate": 3.96074137698631e-07,
+      "loss": 0.6522,
+      "step": 53821
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.210101842880249,
+      "learning_rate": 3.9597129893339194e-07,
+      "loss": 0.7914,
+      "step": 53822
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.4236361980438232,
+      "learning_rate": 3.958684729702311e-07,
+      "loss": 0.5681,
+      "step": 53823
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.5010671615600586,
+      "learning_rate": 3.9576565980943615e-07,
+      "loss": 0.5189,
+      "step": 53824
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.7789807319641113,
+      "learning_rate": 3.9566285945129125e-07,
+      "loss": 0.4776,
+      "step": 53825
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.1479811668395996,
+      "learning_rate": 3.955600718960834e-07,
+      "loss": 0.4411,
+      "step": 53826
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 1.9047855138778687,
+      "learning_rate": 3.954572971440973e-07,
+      "loss": 0.3907,
+      "step": 53827
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.744798183441162,
+      "learning_rate": 3.953545351956212e-07,
+      "loss": 0.4437,
+      "step": 53828
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.976055383682251,
+      "learning_rate": 3.9525178605093804e-07,
+      "loss": 0.6759,
+      "step": 53829
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 1.9951421022415161,
+      "learning_rate": 3.951490497103355e-07,
+      "loss": 0.4863,
+      "step": 53830
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.0863897800445557,
+      "learning_rate": 3.9504632617409777e-07,
+      "loss": 0.4414,
+      "step": 53831
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.165203332901001,
+      "learning_rate": 3.949436154425118e-07,
+      "loss": 0.5876,
+      "step": 53832
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 4.294848918914795,
+      "learning_rate": 3.9484091751586186e-07,
+      "loss": 0.7768,
+      "step": 53833
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 1.6919528245925903,
+      "learning_rate": 3.9473823239443486e-07,
+      "loss": 0.2514,
+      "step": 53834
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.8309848308563232,
+      "learning_rate": 3.946355600785151e-07,
+      "loss": 0.5336,
+      "step": 53835
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 5.181920051574707,
+      "learning_rate": 3.94532900568389e-07,
+      "loss": 0.6817,
+      "step": 53836
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.75937819480896,
+      "learning_rate": 3.944302538643413e-07,
+      "loss": 0.3687,
+      "step": 53837
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.904114246368408,
+      "learning_rate": 3.9432761996665735e-07,
+      "loss": 0.5539,
+      "step": 53838
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.9840619564056396,
+      "learning_rate": 3.9422499887562414e-07,
+      "loss": 0.6718,
+      "step": 53839
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 5.682886123657227,
+      "learning_rate": 3.941223905915248e-07,
+      "loss": 0.5348,
+      "step": 53840
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.09112286567688,
+      "learning_rate": 3.9401979511464684e-07,
+      "loss": 0.6843,
+      "step": 53841
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.1880252361297607,
+      "learning_rate": 3.939172124452734e-07,
+      "loss": 0.4919,
+      "step": 53842
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 4.13908576965332,
+      "learning_rate": 3.938146425836914e-07,
+      "loss": 0.6824,
+      "step": 53843
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.47062349319458,
+      "learning_rate": 3.937120855301846e-07,
+      "loss": 0.6841,
+      "step": 53844
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.089444160461426,
+      "learning_rate": 3.9360954128503995e-07,
+      "loss": 0.5928,
+      "step": 53845
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.6842217445373535,
+      "learning_rate": 3.935070098485405e-07,
+      "loss": 0.5501,
+      "step": 53846
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 4.835790157318115,
+      "learning_rate": 3.9340449122097226e-07,
+      "loss": 0.6634,
+      "step": 53847
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 4.661413192749023,
+      "learning_rate": 3.9330198540262156e-07,
+      "loss": 0.8317,
+      "step": 53848
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 4.053213596343994,
+      "learning_rate": 3.9319949239377155e-07,
+      "loss": 0.6349,
+      "step": 53849
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.839860200881958,
+      "learning_rate": 3.930970121947081e-07,
+      "loss": 0.5911,
+      "step": 53850
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.988415479660034,
+      "learning_rate": 3.929945448057165e-07,
+      "loss": 0.5018,
+      "step": 53851
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.706488847732544,
+      "learning_rate": 3.928920902270811e-07,
+      "loss": 0.5579,
+      "step": 53852
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.9291024208068848,
+      "learning_rate": 3.9278964845908705e-07,
+      "loss": 0.5596,
+      "step": 53853
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.7670116424560547,
+      "learning_rate": 3.926872195020187e-07,
+      "loss": 0.4656,
+      "step": 53854
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.3644959926605225,
+      "learning_rate": 3.925848033561619e-07,
+      "loss": 0.6672,
+      "step": 53855
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.321573495864868,
+      "learning_rate": 3.924824000217997e-07,
+      "loss": 0.4511,
+      "step": 53856
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.3716917037963867,
+      "learning_rate": 3.9238000949921915e-07,
+      "loss": 0.543,
+      "step": 53857
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.2534914016723633,
+      "learning_rate": 3.922776317887028e-07,
+      "loss": 0.4917,
+      "step": 53858
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.2765533924102783,
+      "learning_rate": 3.921752668905371e-07,
+      "loss": 0.5907,
+      "step": 53859
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.8806679248809814,
+      "learning_rate": 3.920729148050051e-07,
+      "loss": 0.7068,
+      "step": 53860
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.9692459106445312,
+      "learning_rate": 3.919705755323927e-07,
+      "loss": 0.6766,
+      "step": 53861
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.8198721408843994,
+      "learning_rate": 3.9186824907298303e-07,
+      "loss": 0.5186,
+      "step": 53862
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.284309148788452,
+      "learning_rate": 3.917659354270631e-07,
+      "loss": 0.5541,
+      "step": 53863
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.203104019165039,
+      "learning_rate": 3.916636345949143e-07,
+      "loss": 0.5125,
+      "step": 53864
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 4.240694522857666,
+      "learning_rate": 3.9156134657682367e-07,
+      "loss": 0.7257,
+      "step": 53865
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.89693546295166,
+      "learning_rate": 3.914590713730737e-07,
+      "loss": 0.6771,
+      "step": 53866
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.16536021232605,
+      "learning_rate": 3.913568089839498e-07,
+      "loss": 0.4743,
+      "step": 53867
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.810420036315918,
+      "learning_rate": 3.9125455940973555e-07,
+      "loss": 0.6401,
+      "step": 53868
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.702059745788574,
+      "learning_rate": 3.9115232265071746e-07,
+      "loss": 0.5648,
+      "step": 53869
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.4530858993530273,
+      "learning_rate": 3.9105009870717694e-07,
+      "loss": 0.5821,
+      "step": 53870
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.436467409133911,
+      "learning_rate": 3.909478875794009e-07,
+      "loss": 0.4295,
+      "step": 53871
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.5513787269592285,
+      "learning_rate": 3.9084568926767095e-07,
+      "loss": 0.5721,
+      "step": 53872
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.5682387351989746,
+      "learning_rate": 3.907435037722734e-07,
+      "loss": 0.8029,
+      "step": 53873
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 10.994007110595703,
+      "learning_rate": 3.906413310934909e-07,
+      "loss": 0.3971,
+      "step": 53874
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.9381704330444336,
+      "learning_rate": 3.905391712316087e-07,
+      "loss": 0.5421,
+      "step": 53875
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.250462055206299,
+      "learning_rate": 3.904370241869093e-07,
+      "loss": 0.5722,
+      "step": 53876
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.6819584369659424,
+      "learning_rate": 3.9033488995967925e-07,
+      "loss": 0.6093,
+      "step": 53877
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.9972083568573,
+      "learning_rate": 3.902327685501994e-07,
+      "loss": 0.695,
+      "step": 53878
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.09104323387146,
+      "learning_rate": 3.9013065995875677e-07,
+      "loss": 0.6212,
+      "step": 53879
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.887683629989624,
+      "learning_rate": 3.900285641856333e-07,
+      "loss": 0.7523,
+      "step": 53880
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.07580304145813,
+      "learning_rate": 3.899264812311138e-07,
+      "loss": 0.5403,
+      "step": 53881
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.714291572570801,
+      "learning_rate": 3.898244110954807e-07,
+      "loss": 0.6594,
+      "step": 53882
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.797475814819336,
+      "learning_rate": 3.897223537790201e-07,
+      "loss": 0.8174,
+      "step": 53883
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.2338578701019287,
+      "learning_rate": 3.896203092820139e-07,
+      "loss": 0.5354,
+      "step": 53884
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.865588665008545,
+      "learning_rate": 3.895182776047474e-07,
+      "loss": 0.3906,
+      "step": 53885
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.5467336177825928,
+      "learning_rate": 3.8941625874750253e-07,
+      "loss": 0.6175,
+      "step": 53886
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.600412607192993,
+      "learning_rate": 3.8931425271056366e-07,
+      "loss": 0.6463,
+      "step": 53887
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.4399852752685547,
+      "learning_rate": 3.8921225949421436e-07,
+      "loss": 0.5297,
+      "step": 53888
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.9495344161987305,
+      "learning_rate": 3.8911027909874e-07,
+      "loss": 0.6314,
+      "step": 53889
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.633362293243408,
+      "learning_rate": 3.890083115244214e-07,
+      "loss": 0.619,
+      "step": 53890
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 6.10296106338501,
+      "learning_rate": 3.8890635677154455e-07,
+      "loss": 0.5519,
+      "step": 53891
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.464200258255005,
+      "learning_rate": 3.888044148403908e-07,
+      "loss": 0.4116,
+      "step": 53892
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.2447941303253174,
+      "learning_rate": 3.887024857312455e-07,
+      "loss": 0.5411,
+      "step": 53893
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.8199238777160645,
+      "learning_rate": 3.8860056944439004e-07,
+      "loss": 0.5753,
+      "step": 53894
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.872602701187134,
+      "learning_rate": 3.8849866598010986e-07,
+      "loss": 0.7064,
+      "step": 53895
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.277712345123291,
+      "learning_rate": 3.8839677533868693e-07,
+      "loss": 0.5109,
+      "step": 53896
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.5872483253479004,
+      "learning_rate": 3.8829489752040536e-07,
+      "loss": 0.6611,
+      "step": 53897
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.533301591873169,
+      "learning_rate": 3.881930325255473e-07,
+      "loss": 0.554,
+      "step": 53898
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.8600690364837646,
+      "learning_rate": 3.8809118035439797e-07,
+      "loss": 0.6362,
+      "step": 53899
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.818544626235962,
+      "learning_rate": 3.879893410072383e-07,
+      "loss": 0.793,
+      "step": 53900
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.0111083984375,
+      "learning_rate": 3.87887514484353e-07,
+      "loss": 0.3677,
+      "step": 53901
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.5041189193725586,
+      "learning_rate": 3.8778570078602527e-07,
+      "loss": 0.7681,
+      "step": 53902
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.479412078857422,
+      "learning_rate": 3.8768389991253654e-07,
+      "loss": 0.7482,
+      "step": 53903
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.6892764568328857,
+      "learning_rate": 3.875821118641726e-07,
+      "loss": 0.4122,
+      "step": 53904
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.397737503051758,
+      "learning_rate": 3.8748033664121387e-07,
+      "loss": 0.6722,
+      "step": 53905
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 4.8047027587890625,
+      "learning_rate": 3.873785742439451e-07,
+      "loss": 0.6355,
+      "step": 53906
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.585376501083374,
+      "learning_rate": 3.872768246726477e-07,
+      "loss": 0.6305,
+      "step": 53907
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.7147252559661865,
+      "learning_rate": 3.871750879276054e-07,
+      "loss": 0.5942,
+      "step": 53908
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.539020538330078,
+      "learning_rate": 3.8707336400910233e-07,
+      "loss": 0.5509,
+      "step": 53909
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 4.151694297790527,
+      "learning_rate": 3.8697165291741947e-07,
+      "loss": 0.7255,
+      "step": 53910
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 4.081179618835449,
+      "learning_rate": 3.868699546528398e-07,
+      "loss": 0.5917,
+      "step": 53911
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.2360119819641113,
+      "learning_rate": 3.8676826921564827e-07,
+      "loss": 0.5668,
+      "step": 53912
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.715726375579834,
+      "learning_rate": 3.866665966061245e-07,
+      "loss": 0.6303,
+      "step": 53913
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.0362236499786377,
+      "learning_rate": 3.8656493682455386e-07,
+      "loss": 0.4944,
+      "step": 53914
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 10.7015962600708,
+      "learning_rate": 3.864632898712173e-07,
+      "loss": 0.5186,
+      "step": 53915
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.2810792922973633,
+      "learning_rate": 3.863616557463984e-07,
+      "loss": 0.6063,
+      "step": 53916
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.540297031402588,
+      "learning_rate": 3.862600344503792e-07,
+      "loss": 0.4331,
+      "step": 53917
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.973954677581787,
+      "learning_rate": 3.8615842598344275e-07,
+      "loss": 0.6794,
+      "step": 53918
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.487464666366577,
+      "learning_rate": 3.860568303458706e-07,
+      "loss": 0.6335,
+      "step": 53919
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.105952024459839,
+      "learning_rate": 3.859552475379469e-07,
+      "loss": 0.7702,
+      "step": 53920
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.1981587409973145,
+      "learning_rate": 3.858536775599525e-07,
+      "loss": 0.615,
+      "step": 53921
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.6037583351135254,
+      "learning_rate": 3.8575212041217114e-07,
+      "loss": 0.7478,
+      "step": 53922
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.8976027965545654,
+      "learning_rate": 3.856505760948842e-07,
+      "loss": 0.8027,
+      "step": 53923
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.3045716285705566,
+      "learning_rate": 3.8554904460837483e-07,
+      "loss": 0.5562,
+      "step": 53924
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.4299488067626953,
+      "learning_rate": 3.854475259529245e-07,
+      "loss": 0.6777,
+      "step": 53925
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 4.817792892456055,
+      "learning_rate": 3.8534602012881614e-07,
+      "loss": 0.4579,
+      "step": 53926
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 1.890123963356018,
+      "learning_rate": 3.852445271363314e-07,
+      "loss": 0.3465,
+      "step": 53927
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.856274127960205,
+      "learning_rate": 3.851430469757528e-07,
+      "loss": 0.6966,
+      "step": 53928
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.077312707901001,
+      "learning_rate": 3.850415796473628e-07,
+      "loss": 0.4584,
+      "step": 53929
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.4789819717407227,
+      "learning_rate": 3.8494012515144397e-07,
+      "loss": 0.3623,
+      "step": 53930
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 4.209479808807373,
+      "learning_rate": 3.8483868348827726e-07,
+      "loss": 0.4092,
+      "step": 53931
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.5276386737823486,
+      "learning_rate": 3.847372546581457e-07,
+      "loss": 0.5458,
+      "step": 53932
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.677025556564331,
+      "learning_rate": 3.846358386613308e-07,
+      "loss": 0.5499,
+      "step": 53933
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 1.6669175624847412,
+      "learning_rate": 3.845344354981145e-07,
+      "loss": 0.37,
+      "step": 53934
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.340588092803955,
+      "learning_rate": 3.844330451687789e-07,
+      "loss": 0.6941,
+      "step": 53935
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 5.263334274291992,
+      "learning_rate": 3.843316676736064e-07,
+      "loss": 0.3865,
+      "step": 53936
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 4.611542701721191,
+      "learning_rate": 3.84230303012878e-07,
+      "loss": 0.7187,
+      "step": 53937
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.741640567779541,
+      "learning_rate": 3.8412895118687674e-07,
+      "loss": 0.6314,
+      "step": 53938
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.3378899097442627,
+      "learning_rate": 3.8402761219588246e-07,
+      "loss": 0.5238,
+      "step": 53939
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.939110279083252,
+      "learning_rate": 3.839262860401793e-07,
+      "loss": 0.5554,
+      "step": 53940
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.653606414794922,
+      "learning_rate": 3.838249727200477e-07,
+      "loss": 0.6751,
+      "step": 53941
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 4.752566814422607,
+      "learning_rate": 3.837236722357696e-07,
+      "loss": 0.6831,
+      "step": 53942
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.705803632736206,
+      "learning_rate": 3.8362238458762636e-07,
+      "loss": 0.7537,
+      "step": 53943
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.006028652191162,
+      "learning_rate": 3.8352110977590064e-07,
+      "loss": 0.3876,
+      "step": 53944
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 1.9539769887924194,
+      "learning_rate": 3.8341984780087214e-07,
+      "loss": 0.4592,
+      "step": 53945
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.45697283744812,
+      "learning_rate": 3.8331859866282463e-07,
+      "loss": 0.4617,
+      "step": 53946
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.8306784629821777,
+      "learning_rate": 3.832173623620378e-07,
+      "loss": 0.6855,
+      "step": 53947
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.8862714767456055,
+      "learning_rate": 3.831161388987942e-07,
+      "loss": 0.5367,
+      "step": 53948
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 9.60889720916748,
+      "learning_rate": 3.830149282733753e-07,
+      "loss": 0.622,
+      "step": 53949
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.095576763153076,
+      "learning_rate": 3.829137304860625e-07,
+      "loss": 0.7222,
+      "step": 53950
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 4.503000736236572,
+      "learning_rate": 3.828125455371368e-07,
+      "loss": 0.5387,
+      "step": 53951
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.2509348392486572,
+      "learning_rate": 3.8271137342688e-07,
+      "loss": 0.3437,
+      "step": 53952
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.668677568435669,
+      "learning_rate": 3.8261021415557263e-07,
+      "loss": 0.5409,
+      "step": 53953
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.7053046226501465,
+      "learning_rate": 3.825090677234977e-07,
+      "loss": 0.476,
+      "step": 53954
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 1.7555315494537354,
+      "learning_rate": 3.824079341309345e-07,
+      "loss": 0.335,
+      "step": 53955
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.6435844898223877,
+      "learning_rate": 3.8230681337816545e-07,
+      "loss": 0.4336,
+      "step": 53956
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.6565322875976562,
+      "learning_rate": 3.822057054654704e-07,
+      "loss": 0.5959,
+      "step": 53957
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.49967622756958,
+      "learning_rate": 3.821046103931325e-07,
+      "loss": 0.618,
+      "step": 53958
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.576303720474243,
+      "learning_rate": 3.820035281614309e-07,
+      "loss": 0.6172,
+      "step": 53959
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.1414175033569336,
+      "learning_rate": 3.819024587706488e-07,
+      "loss": 0.6874,
+      "step": 53960
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.434765577316284,
+      "learning_rate": 3.8180140222106474e-07,
+      "loss": 0.487,
+      "step": 53961
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.417590856552124,
+      "learning_rate": 3.8170035851296084e-07,
+      "loss": 0.413,
+      "step": 53962
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.428914785385132,
+      "learning_rate": 3.81599327646619e-07,
+      "loss": 0.6854,
+      "step": 53963
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 4.595257759094238,
+      "learning_rate": 3.8149830962231847e-07,
+      "loss": 0.5708,
+      "step": 53964
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.865919589996338,
+      "learning_rate": 3.813973044403424e-07,
+      "loss": 0.4836,
+      "step": 53965
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.3047165870666504,
+      "learning_rate": 3.8129631210096884e-07,
+      "loss": 0.5494,
+      "step": 53966
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.397569179534912,
+      "learning_rate": 3.8119533260448147e-07,
+      "loss": 0.6671,
+      "step": 53967
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.439263105392456,
+      "learning_rate": 3.8109436595115843e-07,
+      "loss": 0.7843,
+      "step": 53968
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 4.700729846954346,
+      "learning_rate": 3.8099341214128114e-07,
+      "loss": 0.7319,
+      "step": 53969
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.4708547592163086,
+      "learning_rate": 3.808924711751327e-07,
+      "loss": 0.5054,
+      "step": 53970
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 6.621177673339844,
+      "learning_rate": 3.807915430529907e-07,
+      "loss": 0.6356,
+      "step": 53971
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.725877046585083,
+      "learning_rate": 3.8069062777513764e-07,
+      "loss": 0.5607,
+      "step": 53972
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.186046838760376,
+      "learning_rate": 3.805897253418528e-07,
+      "loss": 0.4373,
+      "step": 53973
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 4.85042667388916,
+      "learning_rate": 3.8048883575341754e-07,
+      "loss": 0.7358,
+      "step": 53974
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.5518486499786377,
+      "learning_rate": 3.803879590101128e-07,
+      "loss": 0.4323,
+      "step": 53975
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.211562395095825,
+      "learning_rate": 3.802870951122184e-07,
+      "loss": 0.6195,
+      "step": 53976
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.9720022678375244,
+      "learning_rate": 3.8018624406001514e-07,
+      "loss": 0.7685,
+      "step": 53977
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.355605363845825,
+      "learning_rate": 3.8008540585378287e-07,
+      "loss": 0.5217,
+      "step": 53978
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.7731549739837646,
+      "learning_rate": 3.799845804938035e-07,
+      "loss": 0.5906,
+      "step": 53979
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.0886049270629883,
+      "learning_rate": 3.798837679803552e-07,
+      "loss": 0.6418,
+      "step": 53980
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.078092575073242,
+      "learning_rate": 3.7978296831372e-07,
+      "loss": 0.4176,
+      "step": 53981
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.8768398761749268,
+      "learning_rate": 3.79682181494177e-07,
+      "loss": 0.8897,
+      "step": 53982
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.0183773040771484,
+      "learning_rate": 3.795814075220078e-07,
+      "loss": 0.4101,
+      "step": 53983
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.776394844055176,
+      "learning_rate": 3.794806463974915e-07,
+      "loss": 0.4581,
+      "step": 53984
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.0560860633850098,
+      "learning_rate": 3.7937989812090903e-07,
+      "loss": 0.7661,
+      "step": 53985
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.8085999488830566,
+      "learning_rate": 3.792791626925391e-07,
+      "loss": 0.7451,
+      "step": 53986
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.3761887550354004,
+      "learning_rate": 3.7917844011266416e-07,
+      "loss": 0.6071,
+      "step": 53987
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.286595106124878,
+      "learning_rate": 3.790777303815618e-07,
+      "loss": 0.5847,
+      "step": 53988
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.316377639770508,
+      "learning_rate": 3.7897703349951355e-07,
+      "loss": 0.5249,
+      "step": 53989
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.7261855602264404,
+      "learning_rate": 3.7887634946679854e-07,
+      "loss": 0.5544,
+      "step": 53990
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.30698561668396,
+      "learning_rate": 3.7877567828369876e-07,
+      "loss": 0.3402,
+      "step": 53991
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.1078450679779053,
+      "learning_rate": 3.786750199504913e-07,
+      "loss": 0.421,
+      "step": 53992
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.639108419418335,
+      "learning_rate": 3.785743744674586e-07,
+      "loss": 0.6424,
+      "step": 53993
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 5.196949005126953,
+      "learning_rate": 3.784737418348783e-07,
+      "loss": 0.474,
+      "step": 53994
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.795630693435669,
+      "learning_rate": 3.783731220530323e-07,
+      "loss": 0.7014,
+      "step": 53995
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.5878965854644775,
+      "learning_rate": 3.782725151221989e-07,
+      "loss": 0.5815,
+      "step": 53996
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.85585880279541,
+      "learning_rate": 3.781719210426582e-07,
+      "loss": 0.4426,
+      "step": 53997
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.279331684112549,
+      "learning_rate": 3.7807133981468956e-07,
+      "loss": 0.5889,
+      "step": 53998
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.384347438812256,
+      "learning_rate": 3.779707714385744e-07,
+      "loss": 0.3892,
+      "step": 53999
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.9539899826049805,
+      "learning_rate": 3.778702159145897e-07,
+      "loss": 0.7662,
+      "step": 54000
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.2245795726776123,
+      "learning_rate": 3.777696732430175e-07,
+      "loss": 0.5463,
+      "step": 54001
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.81771183013916,
+      "learning_rate": 3.776691434241353e-07,
+      "loss": 0.3733,
+      "step": 54002
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.7752811908721924,
+      "learning_rate": 3.7756862645822455e-07,
+      "loss": 0.416,
+      "step": 54003
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.2426834106445312,
+      "learning_rate": 3.774681223455634e-07,
+      "loss": 0.4054,
+      "step": 54004
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.3745310306549072,
+      "learning_rate": 3.773676310864322e-07,
+      "loss": 0.6625,
+      "step": 54005
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.7853169441223145,
+      "learning_rate": 3.772671526811089e-07,
+      "loss": 0.6281,
+      "step": 54006
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.342771530151367,
+      "learning_rate": 3.771666871298751e-07,
+      "loss": 0.5133,
+      "step": 54007
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.5826759338378906,
+      "learning_rate": 3.770662344330084e-07,
+      "loss": 0.6788,
+      "step": 54008
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.0300991535186768,
+      "learning_rate": 3.769657945907884e-07,
+      "loss": 0.7239,
+      "step": 54009
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.591560125350952,
+      "learning_rate": 3.7686536760349445e-07,
+      "loss": 0.5276,
+      "step": 54010
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.909132242202759,
+      "learning_rate": 3.7676495347140683e-07,
+      "loss": 0.4889,
+      "step": 54011
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.717435598373413,
+      "learning_rate": 3.766645521948037e-07,
+      "loss": 0.5352,
+      "step": 54012
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.940486192703247,
+      "learning_rate": 3.7656416377396534e-07,
+      "loss": 0.4181,
+      "step": 54013
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.7206995487213135,
+      "learning_rate": 3.7646378820916875e-07,
+      "loss": 0.8315,
+      "step": 54014
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.901536226272583,
+      "learning_rate": 3.7636342550069536e-07,
+      "loss": 0.5375,
+      "step": 54015
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.4250781536102295,
+      "learning_rate": 3.7626307564882225e-07,
+      "loss": 0.3894,
+      "step": 54016
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 5.848911762237549,
+      "learning_rate": 3.761627386538308e-07,
+      "loss": 0.5543,
+      "step": 54017
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.9049739837646484,
+      "learning_rate": 3.7606241451599745e-07,
+      "loss": 0.53,
+      "step": 54018
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.2854366302490234,
+      "learning_rate": 3.759621032356031e-07,
+      "loss": 0.5918,
+      "step": 54019
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.5435690879821777,
+      "learning_rate": 3.758618048129253e-07,
+      "loss": 0.6671,
+      "step": 54020
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 4.565596103668213,
+      "learning_rate": 3.757615192482444e-07,
+      "loss": 0.6183,
+      "step": 54021
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.117872714996338,
+      "learning_rate": 3.7566124654183743e-07,
+      "loss": 0.6279,
+      "step": 54022
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.7217116355895996,
+      "learning_rate": 3.755609866939841e-07,
+      "loss": 0.7435,
+      "step": 54023
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.532514810562134,
+      "learning_rate": 3.754607397049642e-07,
+      "loss": 0.8954,
+      "step": 54024
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.904576539993286,
+      "learning_rate": 3.753605055750553e-07,
+      "loss": 0.8313,
+      "step": 54025
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.0203349590301514,
+      "learning_rate": 3.752602843045366e-07,
+      "loss": 0.5279,
+      "step": 54026
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.5778579711914062,
+      "learning_rate": 3.751600758936863e-07,
+      "loss": 0.7959,
+      "step": 54027
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.3775484561920166,
+      "learning_rate": 3.750598803427835e-07,
+      "loss": 0.6285,
+      "step": 54028
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.9570372104644775,
+      "learning_rate": 3.749596976521064e-07,
+      "loss": 0.6322,
+      "step": 54029
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.149466037750244,
+      "learning_rate": 3.748595278219336e-07,
+      "loss": 0.6472,
+      "step": 54030
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.1945712566375732,
+      "learning_rate": 3.747593708525443e-07,
+      "loss": 0.5564,
+      "step": 54031
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.2511584758758545,
+      "learning_rate": 3.7465922674421616e-07,
+      "loss": 0.6344,
+      "step": 54032
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.263490915298462,
+      "learning_rate": 3.745590954972289e-07,
+      "loss": 0.5536,
+      "step": 54033
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.004584312438965,
+      "learning_rate": 3.74458977111859e-07,
+      "loss": 0.6561,
+      "step": 54034
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.5770657062530518,
+      "learning_rate": 3.7435887158838615e-07,
+      "loss": 0.669,
+      "step": 54035
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.3757643699645996,
+      "learning_rate": 3.742587789270891e-07,
+      "loss": 0.6205,
+      "step": 54036
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.3910810947418213,
+      "learning_rate": 3.7415869912824423e-07,
+      "loss": 0.6143,
+      "step": 54037
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 4.727572441101074,
+      "learning_rate": 3.7405863219213245e-07,
+      "loss": 0.5956,
+      "step": 54038
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 1.7169828414916992,
+      "learning_rate": 3.7395857811903023e-07,
+      "loss": 0.445,
+      "step": 54039
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 4.35481071472168,
+      "learning_rate": 3.738585369092168e-07,
+      "loss": 0.7462,
+      "step": 54040
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 4.239257335662842,
+      "learning_rate": 3.7375850856296856e-07,
+      "loss": 0.4921,
+      "step": 54041
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.8581912517547607,
+      "learning_rate": 3.7365849308056645e-07,
+      "loss": 0.7595,
+      "step": 54042
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.883262872695923,
+      "learning_rate": 3.7355849046228575e-07,
+      "loss": 0.398,
+      "step": 54043
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.332447052001953,
+      "learning_rate": 3.734585007084063e-07,
+      "loss": 0.5465,
+      "step": 54044
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.5733890533447266,
+      "learning_rate": 3.73358523819205e-07,
+      "loss": 0.7134,
+      "step": 54045
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.110382080078125,
+      "learning_rate": 3.732585597949617e-07,
+      "loss": 0.6271,
+      "step": 54046
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.938072681427002,
+      "learning_rate": 3.731586086359518e-07,
+      "loss": 0.6005,
+      "step": 54047
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.5633068084716797,
+      "learning_rate": 3.7305867034245603e-07,
+      "loss": 0.6381,
+      "step": 54048
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.5570597648620605,
+      "learning_rate": 3.729587449147493e-07,
+      "loss": 0.7044,
+      "step": 54049
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.307218074798584,
+      "learning_rate": 3.7285883235311127e-07,
+      "loss": 0.4387,
+      "step": 54050
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.8158483505249023,
+      "learning_rate": 3.727589326578196e-07,
+      "loss": 0.5673,
+      "step": 54051
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.8158962726593018,
+      "learning_rate": 3.726590458291524e-07,
+      "loss": 0.3429,
+      "step": 54052
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.754331350326538,
+      "learning_rate": 3.72559171867386e-07,
+      "loss": 0.522,
+      "step": 54053
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.852344036102295,
+      "learning_rate": 3.724593107728003e-07,
+      "loss": 0.6311,
+      "step": 54054
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.3771767616271973,
+      "learning_rate": 3.723594625456711e-07,
+      "loss": 0.7326,
+      "step": 54055
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.941864013671875,
+      "learning_rate": 3.7225962718627715e-07,
+      "loss": 0.7193,
+      "step": 54056
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.68190336227417,
+      "learning_rate": 3.7215980469489487e-07,
+      "loss": 0.5627,
+      "step": 54057
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.4500207901000977,
+      "learning_rate": 3.720599950718029e-07,
+      "loss": 0.7432,
+      "step": 54058
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.549121856689453,
+      "learning_rate": 3.7196019831727825e-07,
+      "loss": 0.5326,
+      "step": 54059
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.3244826793670654,
+      "learning_rate": 3.7186041443159904e-07,
+      "loss": 0.5701,
+      "step": 54060
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.8323962688446045,
+      "learning_rate": 3.717606434150417e-07,
+      "loss": 0.4313,
+      "step": 54061
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.0766725540161133,
+      "learning_rate": 3.7166088526788493e-07,
+      "loss": 0.5691,
+      "step": 54062
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.607574701309204,
+      "learning_rate": 3.715611399904045e-07,
+      "loss": 0.5902,
+      "step": 54063
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.9874563217163086,
+      "learning_rate": 3.7146140758287983e-07,
+      "loss": 0.5032,
+      "step": 54064
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.7593672275543213,
+      "learning_rate": 3.7136168804558614e-07,
+      "loss": 0.8755,
+      "step": 54065
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.790754556655884,
+      "learning_rate": 3.7126198137880207e-07,
+      "loss": 0.6849,
+      "step": 54066
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.4289326667785645,
+      "learning_rate": 3.711622875828041e-07,
+      "loss": 0.6333,
+      "step": 54067
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 4.206671714782715,
+      "learning_rate": 3.710626066578704e-07,
+      "loss": 0.5456,
+      "step": 54068
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.7331795692443848,
+      "learning_rate": 3.709629386042768e-07,
+      "loss": 0.5902,
+      "step": 54069
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.2574150562286377,
+      "learning_rate": 3.708632834223014e-07,
+      "loss": 0.8026,
+      "step": 54070
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.5186541080474854,
+      "learning_rate": 3.7076364111222063e-07,
+      "loss": 0.7512,
+      "step": 54071
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.544262170791626,
+      "learning_rate": 3.706640116743132e-07,
+      "loss": 0.4923,
+      "step": 54072
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.4468166828155518,
+      "learning_rate": 3.7056439510885444e-07,
+      "loss": 0.5973,
+      "step": 54073
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.871584177017212,
+      "learning_rate": 3.704647914161219e-07,
+      "loss": 0.6784,
+      "step": 54074
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.9302287101745605,
+      "learning_rate": 3.7036520059639257e-07,
+      "loss": 0.634,
+      "step": 54075
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.7685000896453857,
+      "learning_rate": 3.7026562264994346e-07,
+      "loss": 0.6277,
+      "step": 54076
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.98799204826355,
+      "learning_rate": 3.701660575770505e-07,
+      "loss": 0.545,
+      "step": 54077
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.5041134357452393,
+      "learning_rate": 3.700665053779928e-07,
+      "loss": 0.5542,
+      "step": 54078
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.440943956375122,
+      "learning_rate": 3.6996696605304414e-07,
+      "loss": 0.4566,
+      "step": 54079
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 1.9375972747802734,
+      "learning_rate": 3.6986743960248426e-07,
+      "loss": 0.3513,
+      "step": 54080
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.484797954559326,
+      "learning_rate": 3.697679260265874e-07,
+      "loss": 0.7795,
+      "step": 54081
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.0798301696777344,
+      "learning_rate": 3.696684253256327e-07,
+      "loss": 0.7181,
+      "step": 54082
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 1.7082351446151733,
+      "learning_rate": 3.6956893749989455e-07,
+      "loss": 0.3302,
+      "step": 54083
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.753155469894409,
+      "learning_rate": 3.694694625496509e-07,
+      "loss": 0.2701,
+      "step": 54084
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.439199924468994,
+      "learning_rate": 3.693700004751777e-07,
+      "loss": 0.2926,
+      "step": 54085
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.352229356765747,
+      "learning_rate": 3.6927055127675193e-07,
+      "loss": 0.6076,
+      "step": 54086
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.927536725997925,
+      "learning_rate": 3.6917111495465063e-07,
+      "loss": 0.7701,
+      "step": 54087
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.9771406650543213,
+      "learning_rate": 3.6907169150914854e-07,
+      "loss": 0.756,
+      "step": 54088
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.3990061283111572,
+      "learning_rate": 3.6897228094052496e-07,
+      "loss": 0.4109,
+      "step": 54089
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.514024257659912,
+      "learning_rate": 3.6887288324905344e-07,
+      "loss": 0.5012,
+      "step": 54090
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.9232418537139893,
+      "learning_rate": 3.687734984350111e-07,
+      "loss": 0.6439,
+      "step": 54091
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.5959372520446777,
+      "learning_rate": 3.6867412649867595e-07,
+      "loss": 0.6137,
+      "step": 54092
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.454254388809204,
+      "learning_rate": 3.6857476744032283e-07,
+      "loss": 0.6435,
+      "step": 54093
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.241844654083252,
+      "learning_rate": 3.6847542126022815e-07,
+      "loss": 0.4034,
+      "step": 54094
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.495692014694214,
+      "learning_rate": 3.683760879586684e-07,
+      "loss": 0.5013,
+      "step": 54095
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 1.9828417301177979,
+      "learning_rate": 3.682767675359189e-07,
+      "loss": 0.3343,
+      "step": 54096
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.9207019805908203,
+      "learning_rate": 3.681774599922583e-07,
+      "loss": 0.6042,
+      "step": 54097
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.0026538372039795,
+      "learning_rate": 3.680781653279597e-07,
+      "loss": 0.5379,
+      "step": 54098
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.540416955947876,
+      "learning_rate": 3.6797888354330124e-07,
+      "loss": 0.4774,
+      "step": 54099
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.565535068511963,
+      "learning_rate": 3.678796146385577e-07,
+      "loss": 0.4557,
+      "step": 54100
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.9261512756347656,
+      "learning_rate": 3.677803586140066e-07,
+      "loss": 0.5428,
+      "step": 54101
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.9552395343780518,
+      "learning_rate": 3.676811154699222e-07,
+      "loss": 0.736,
+      "step": 54102
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.1293439865112305,
+      "learning_rate": 3.675818852065821e-07,
+      "loss": 0.7848,
+      "step": 54103
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.9712274074554443,
+      "learning_rate": 3.6748266782426046e-07,
+      "loss": 0.2927,
+      "step": 54104
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 1.9209177494049072,
+      "learning_rate": 3.6738346332323536e-07,
+      "loss": 0.3878,
+      "step": 54105
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.491253137588501,
+      "learning_rate": 3.6728427170378054e-07,
+      "loss": 0.4705,
+      "step": 54106
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.5366857051849365,
+      "learning_rate": 3.67185092966173e-07,
+      "loss": 0.4098,
+      "step": 54107
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.689119815826416,
+      "learning_rate": 3.67085927110688e-07,
+      "loss": 0.5298,
+      "step": 54108
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.621345043182373,
+      "learning_rate": 3.6698677413760207e-07,
+      "loss": 0.4944,
+      "step": 54109
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.2572264671325684,
+      "learning_rate": 3.668876340471894e-07,
+      "loss": 0.4372,
+      "step": 54110
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.111154079437256,
+      "learning_rate": 3.66788506839727e-07,
+      "loss": 0.7165,
+      "step": 54111
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 1.720266342163086,
+      "learning_rate": 3.666893925154896e-07,
+      "loss": 0.468,
+      "step": 54112
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.9458634853363037,
+      "learning_rate": 3.6659029107475484e-07,
+      "loss": 0.6476,
+      "step": 54113
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 4.069878101348877,
+      "learning_rate": 3.664912025177952e-07,
+      "loss": 0.4785,
+      "step": 54114
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.164595603942871,
+      "learning_rate": 3.6639212684488945e-07,
+      "loss": 0.5547,
+      "step": 54115
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.118201971054077,
+      "learning_rate": 3.6629306405631004e-07,
+      "loss": 0.5577,
+      "step": 54116
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.3978636264801025,
+      "learning_rate": 3.6619401415233456e-07,
+      "loss": 0.7101,
+      "step": 54117
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.8385403156280518,
+      "learning_rate": 3.660949771332367e-07,
+      "loss": 0.6371,
+      "step": 54118
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.8040566444396973,
+      "learning_rate": 3.65995952999294e-07,
+      "loss": 0.502,
+      "step": 54119
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.5616366863250732,
+      "learning_rate": 3.658969417507796e-07,
+      "loss": 0.5015,
+      "step": 54120
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 6.01997709274292,
+      "learning_rate": 3.657979433879705e-07,
+      "loss": 0.5062,
+      "step": 54121
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 4.660482406616211,
+      "learning_rate": 3.656989579111409e-07,
+      "loss": 0.893,
+      "step": 54122
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.4925458431243896,
+      "learning_rate": 3.655999853205672e-07,
+      "loss": 0.5961,
+      "step": 54123
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.0839993953704834,
+      "learning_rate": 3.6550102561652256e-07,
+      "loss": 0.5503,
+      "step": 54124
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 4.8611297607421875,
+      "learning_rate": 3.654020787992851e-07,
+      "loss": 0.6582,
+      "step": 54125
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.0545554161071777,
+      "learning_rate": 3.653031448691269e-07,
+      "loss": 0.3704,
+      "step": 54126
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.057997226715088,
+      "learning_rate": 3.6520422382632535e-07,
+      "loss": 0.6455,
+      "step": 54127
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.5572474002838135,
+      "learning_rate": 3.651053156711537e-07,
+      "loss": 0.7536,
+      "step": 54128
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.806936025619507,
+      "learning_rate": 3.650064204038889e-07,
+      "loss": 0.7949,
+      "step": 54129
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 4.38273286819458,
+      "learning_rate": 3.6490753802480406e-07,
+      "loss": 0.3698,
+      "step": 54130
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.5629959106445312,
+      "learning_rate": 3.6480866853417506e-07,
+      "loss": 0.5683,
+      "step": 54131
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.172262191772461,
+      "learning_rate": 3.6470981193227675e-07,
+      "loss": 0.4441,
+      "step": 54132
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.880049228668213,
+      "learning_rate": 3.6461096821938434e-07,
+      "loss": 0.6473,
+      "step": 54133
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.9391579627990723,
+      "learning_rate": 3.6451213739577217e-07,
+      "loss": 0.6003,
+      "step": 54134
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.9123103618621826,
+      "learning_rate": 3.644133194617161e-07,
+      "loss": 0.4292,
+      "step": 54135
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.9178154468536377,
+      "learning_rate": 3.643145144174892e-07,
+      "loss": 0.6842,
+      "step": 54136
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.696117401123047,
+      "learning_rate": 3.6421572226336743e-07,
+      "loss": 0.4976,
+      "step": 54137
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.138645648956299,
+      "learning_rate": 3.6411694299962496e-07,
+      "loss": 0.6843,
+      "step": 54138
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.2883403301239014,
+      "learning_rate": 3.6401817662653714e-07,
+      "loss": 0.7466,
+      "step": 54139
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 5.084110736846924,
+      "learning_rate": 3.639194231443771e-07,
+      "loss": 0.6082,
+      "step": 54140
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.0884158611297607,
+      "learning_rate": 3.638206825534213e-07,
+      "loss": 0.5316,
+      "step": 54141
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.8019015789031982,
+      "learning_rate": 3.637219548539428e-07,
+      "loss": 0.3927,
+      "step": 54142
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.6977860927581787,
+      "learning_rate": 3.6362324004621753e-07,
+      "loss": 0.5394,
+      "step": 54143
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.806260108947754,
+      "learning_rate": 3.6352453813051805e-07,
+      "loss": 0.6244,
+      "step": 54144
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 4.465907573699951,
+      "learning_rate": 3.634258491071213e-07,
+      "loss": 0.4943,
+      "step": 54145
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.868926525115967,
+      "learning_rate": 3.6332717297629936e-07,
+      "loss": 0.4952,
+      "step": 54146
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.1610419750213623,
+      "learning_rate": 3.632285097383276e-07,
+      "loss": 0.5466,
+      "step": 54147
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.806338310241699,
+      "learning_rate": 3.6312985939348123e-07,
+      "loss": 0.6104,
+      "step": 54148
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 1.9138559103012085,
+      "learning_rate": 3.630312219420329e-07,
+      "loss": 0.4264,
+      "step": 54149
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 4.747113227844238,
+      "learning_rate": 3.6293259738425843e-07,
+      "loss": 0.6556,
+      "step": 54150
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.1447250843048096,
+      "learning_rate": 3.628339857204305e-07,
+      "loss": 0.6174,
+      "step": 54151
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 4.845737934112549,
+      "learning_rate": 3.627353869508238e-07,
+      "loss": 0.5725,
+      "step": 54152
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.862795352935791,
+      "learning_rate": 3.6263680107571417e-07,
+      "loss": 0.6425,
+      "step": 54153
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.9765560626983643,
+      "learning_rate": 3.625382280953732e-07,
+      "loss": 0.7419,
+      "step": 54154
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.400279998779297,
+      "learning_rate": 3.624396680100772e-07,
+      "loss": 0.3281,
+      "step": 54155
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 4.6989336013793945,
+      "learning_rate": 3.623411208200989e-07,
+      "loss": 0.5489,
+      "step": 54156
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.233682155609131,
+      "learning_rate": 3.6224258652571286e-07,
+      "loss": 0.7578,
+      "step": 54157
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.4100067615509033,
+      "learning_rate": 3.621440651271918e-07,
+      "loss": 0.5493,
+      "step": 54158
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.251152515411377,
+      "learning_rate": 3.62045556624811e-07,
+      "loss": 0.6957,
+      "step": 54159
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.868229866027832,
+      "learning_rate": 3.6194706101884523e-07,
+      "loss": 0.4836,
+      "step": 54160
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.689060688018799,
+      "learning_rate": 3.61848578309566e-07,
+      "loss": 0.5001,
+      "step": 54161
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 1.7074037790298462,
+      "learning_rate": 3.617501084972497e-07,
+      "loss": 0.4535,
+      "step": 54162
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.3406081199645996,
+      "learning_rate": 3.616516515821672e-07,
+      "loss": 0.7736,
+      "step": 54163
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.068979263305664,
+      "learning_rate": 3.6155320756459557e-07,
+      "loss": 0.7491,
+      "step": 54164
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.7942757606506348,
+      "learning_rate": 3.6145477644480565e-07,
+      "loss": 0.6798,
+      "step": 54165
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.662782907485962,
+      "learning_rate": 3.613563582230728e-07,
+      "loss": 0.6374,
+      "step": 54166
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 4.2774553298950195,
+      "learning_rate": 3.6125795289966956e-07,
+      "loss": 0.4946,
+      "step": 54167
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 4.68558931350708,
+      "learning_rate": 3.611595604748713e-07,
+      "loss": 0.581,
+      "step": 54168
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.6385176181793213,
+      "learning_rate": 3.6106118094895004e-07,
+      "loss": 0.2362,
+      "step": 54169
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.64665150642395,
+      "learning_rate": 3.609628143221805e-07,
+      "loss": 0.7653,
+      "step": 54170
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.731984853744507,
+      "learning_rate": 3.608644605948342e-07,
+      "loss": 0.7275,
+      "step": 54171
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.6796090602874756,
+      "learning_rate": 3.6076611976718635e-07,
+      "loss": 0.4878,
+      "step": 54172
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 4.201244831085205,
+      "learning_rate": 3.606677918395102e-07,
+      "loss": 0.6287,
+      "step": 54173
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.348036050796509,
+      "learning_rate": 3.605694768120793e-07,
+      "loss": 0.3891,
+      "step": 54174
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.379716634750366,
+      "learning_rate": 3.6047117468516635e-07,
+      "loss": 0.7556,
+      "step": 54175
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.8605258464813232,
+      "learning_rate": 3.6037288545904604e-07,
+      "loss": 0.5722,
+      "step": 54176
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.209935426712036,
+      "learning_rate": 3.6027460913398926e-07,
+      "loss": 0.6148,
+      "step": 54177
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.5890324115753174,
+      "learning_rate": 3.601763457102719e-07,
+      "loss": 0.5996,
+      "step": 54178
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.3885257244110107,
+      "learning_rate": 3.600780951881655e-07,
+      "loss": 0.4282,
+      "step": 54179
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.005852222442627,
+      "learning_rate": 3.599798575679442e-07,
+      "loss": 0.5735,
+      "step": 54180
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.0844547748565674,
+      "learning_rate": 3.598816328498794e-07,
+      "loss": 0.303,
+      "step": 54181
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.157688617706299,
+      "learning_rate": 3.597834210342471e-07,
+      "loss": 0.6917,
+      "step": 54182
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.178497314453125,
+      "learning_rate": 3.596852221213182e-07,
+      "loss": 0.447,
+      "step": 54183
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.697683095932007,
+      "learning_rate": 3.595870361113668e-07,
+      "loss": 0.7691,
+      "step": 54184
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.334510564804077,
+      "learning_rate": 3.5948886300466445e-07,
+      "loss": 0.6206,
+      "step": 54185
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.781667470932007,
+      "learning_rate": 3.5939070280148645e-07,
+      "loss": 0.6602,
+      "step": 54186
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.4213907718658447,
+      "learning_rate": 3.5929255550210317e-07,
+      "loss": 0.8344,
+      "step": 54187
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.111002206802368,
+      "learning_rate": 3.591944211067905e-07,
+      "loss": 0.7667,
+      "step": 54188
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.6794748306274414,
+      "learning_rate": 3.5909629961581815e-07,
+      "loss": 0.5358,
+      "step": 54189
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.331738233566284,
+      "learning_rate": 3.5899819102946153e-07,
+      "loss": 0.5872,
+      "step": 54190
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.4910569190979004,
+      "learning_rate": 3.589000953479915e-07,
+      "loss": 0.6749,
+      "step": 54191
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.142958879470825,
+      "learning_rate": 3.588020125716818e-07,
+      "loss": 0.3701,
+      "step": 54192
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.390625,
+      "learning_rate": 3.587039427008049e-07,
+      "loss": 0.4645,
+      "step": 54193
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.5773863792419434,
+      "learning_rate": 3.5860588573563446e-07,
+      "loss": 0.533,
+      "step": 54194
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.8090453147888184,
+      "learning_rate": 3.5850784167644147e-07,
+      "loss": 0.4169,
+      "step": 54195
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.677318811416626,
+      "learning_rate": 3.5840981052350066e-07,
+      "loss": 0.7733,
+      "step": 54196
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.7418181896209717,
+      "learning_rate": 3.5831179227708234e-07,
+      "loss": 0.6409,
+      "step": 54197
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.031097412109375,
+      "learning_rate": 3.582137869374608e-07,
+      "loss": 0.6822,
+      "step": 54198
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.110957145690918,
+      "learning_rate": 3.581157945049074e-07,
+      "loss": 0.5966,
+      "step": 54199
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.4800336360931396,
+      "learning_rate": 3.5801781497969534e-07,
+      "loss": 0.6236,
+      "step": 54200
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.854637861251831,
+      "learning_rate": 3.579198483620966e-07,
+      "loss": 0.6699,
+      "step": 54201
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.7894437313079834,
+      "learning_rate": 3.5782189465238425e-07,
+      "loss": 0.4833,
+      "step": 54202
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.8441555500030518,
+      "learning_rate": 3.5772395385082924e-07,
+      "loss": 0.8162,
+      "step": 54203
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.3671343326568604,
+      "learning_rate": 3.5762602595770637e-07,
+      "loss": 0.5805,
+      "step": 54204
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.4098150730133057,
+      "learning_rate": 3.5752811097328477e-07,
+      "loss": 0.7548,
+      "step": 54205
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.7069602012634277,
+      "learning_rate": 3.5743020889783984e-07,
+      "loss": 0.583,
+      "step": 54206
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 4.395603179931641,
+      "learning_rate": 3.573323197316414e-07,
+      "loss": 0.803,
+      "step": 54207
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.4542500972747803,
+      "learning_rate": 3.572344434749625e-07,
+      "loss": 0.6794,
+      "step": 54208
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.7430577278137207,
+      "learning_rate": 3.5713658012807627e-07,
+      "loss": 0.6486,
+      "step": 54209
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.137049913406372,
+      "learning_rate": 3.570387296912531e-07,
+      "loss": 0.7471,
+      "step": 54210
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.04489803314209,
+      "learning_rate": 3.5694089216476657e-07,
+      "loss": 0.5578,
+      "step": 54211
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.5393967628479004,
+      "learning_rate": 3.5684306754888764e-07,
+      "loss": 0.5215,
+      "step": 54212
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.1569983959198,
+      "learning_rate": 3.567452558438883e-07,
+      "loss": 0.3663,
+      "step": 54213
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.0002171993255615,
+      "learning_rate": 3.566474570500422e-07,
+      "loss": 0.8317,
+      "step": 54214
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.7821905612945557,
+      "learning_rate": 3.5654967116761916e-07,
+      "loss": 0.5682,
+      "step": 54215
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 1.8355052471160889,
+      "learning_rate": 3.5645189819689286e-07,
+      "loss": 0.4746,
+      "step": 54216
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.704072952270508,
+      "learning_rate": 3.5635413813813356e-07,
+      "loss": 0.5031,
+      "step": 54217
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.5112829208374023,
+      "learning_rate": 3.5625639099161445e-07,
+      "loss": 0.5579,
+      "step": 54218
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.096520185470581,
+      "learning_rate": 3.561586567576064e-07,
+      "loss": 0.6724,
+      "step": 54219
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.813004970550537,
+      "learning_rate": 3.5606093543638087e-07,
+      "loss": 0.5958,
+      "step": 54220
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 5.095491886138916,
+      "learning_rate": 3.559632270282115e-07,
+      "loss": 0.7,
+      "step": 54221
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.292712688446045,
+      "learning_rate": 3.5586553153336755e-07,
+      "loss": 0.8685,
+      "step": 54222
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.0196540355682373,
+      "learning_rate": 3.557678489521227e-07,
+      "loss": 0.6034,
+      "step": 54223
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.253195285797119,
+      "learning_rate": 3.5567017928474725e-07,
+      "loss": 0.5627,
+      "step": 54224
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.3508493900299072,
+      "learning_rate": 3.5557252253151375e-07,
+      "loss": 0.6575,
+      "step": 54225
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.7505569458007812,
+      "learning_rate": 3.554748786926926e-07,
+      "loss": 0.6663,
+      "step": 54226
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.3951170444488525,
+      "learning_rate": 3.553772477685563e-07,
+      "loss": 0.3958,
+      "step": 54227
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.3147521018981934,
+      "learning_rate": 3.552796297593752e-07,
+      "loss": 0.5829,
+      "step": 54228
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 1.9505001306533813,
+      "learning_rate": 3.551820246654225e-07,
+      "loss": 0.4934,
+      "step": 54229
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.9947001934051514,
+      "learning_rate": 3.550844324869679e-07,
+      "loss": 0.5639,
+      "step": 54230
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.1788389682769775,
+      "learning_rate": 3.549868532242845e-07,
+      "loss": 0.5743,
+      "step": 54231
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 4.317789077758789,
+      "learning_rate": 3.5488928687764104e-07,
+      "loss": 0.5589,
+      "step": 54232
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.052523136138916,
+      "learning_rate": 3.547917334473111e-07,
+      "loss": 0.4558,
+      "step": 54233
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.135324239730835,
+      "learning_rate": 3.546941929335651e-07,
+      "loss": 0.5559,
+      "step": 54234
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 5.5469818115234375,
+      "learning_rate": 3.5459666533667504e-07,
+      "loss": 0.9779,
+      "step": 54235
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 4.9000043869018555,
+      "learning_rate": 3.544991506569101e-07,
+      "loss": 0.5431,
+      "step": 54236
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.1137709617614746,
+      "learning_rate": 3.5440164889454455e-07,
+      "loss": 0.5151,
+      "step": 54237
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 4.901977062225342,
+      "learning_rate": 3.5430416004984645e-07,
+      "loss": 0.7188,
+      "step": 54238
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.362980604171753,
+      "learning_rate": 3.54206684123089e-07,
+      "loss": 0.6406,
+      "step": 54239
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.291229486465454,
+      "learning_rate": 3.5410922111454137e-07,
+      "loss": 0.5166,
+      "step": 54240
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 4.394410133361816,
+      "learning_rate": 3.540117710244767e-07,
+      "loss": 0.8199,
+      "step": 54241
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 4.99462366104126,
+      "learning_rate": 3.539143338531642e-07,
+      "loss": 0.5256,
+      "step": 54242
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.1185007095336914,
+      "learning_rate": 3.538169096008759e-07,
+      "loss": 0.4054,
+      "step": 54243
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 4.475495338439941,
+      "learning_rate": 3.5371949826788153e-07,
+      "loss": 0.4893,
+      "step": 54244
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.1943628787994385,
+      "learning_rate": 3.536220998544537e-07,
+      "loss": 0.6671,
+      "step": 54245
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.8731067180633545,
+      "learning_rate": 3.5352471436086113e-07,
+      "loss": 0.6182,
+      "step": 54246
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.603618860244751,
+      "learning_rate": 3.534273417873768e-07,
+      "loss": 0.5428,
+      "step": 54247
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.632103443145752,
+      "learning_rate": 3.533299821342695e-07,
+      "loss": 0.6458,
+      "step": 54248
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.414832353591919,
+      "learning_rate": 3.5323263540181175e-07,
+      "loss": 0.6728,
+      "step": 54249
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.158715009689331,
+      "learning_rate": 3.5313530159027275e-07,
+      "loss": 0.6711,
+      "step": 54250
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.2507221698760986,
+      "learning_rate": 3.53037980699924e-07,
+      "loss": 0.5607,
+      "step": 54251
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.519937038421631,
+      "learning_rate": 3.5294067273103526e-07,
+      "loss": 0.6758,
+      "step": 54252
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.4730491638183594,
+      "learning_rate": 3.5284337768387746e-07,
+      "loss": 0.7421,
+      "step": 54253
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.233596086502075,
+      "learning_rate": 3.52746095558722e-07,
+      "loss": 0.5166,
+      "step": 54254
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.7293310165405273,
+      "learning_rate": 3.5264882635583865e-07,
+      "loss": 0.4685,
+      "step": 54255
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.050045967102051,
+      "learning_rate": 3.525515700754978e-07,
+      "loss": 0.4258,
+      "step": 54256
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.6416754722595215,
+      "learning_rate": 3.524543267179709e-07,
+      "loss": 0.5825,
+      "step": 54257
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.9573974609375,
+      "learning_rate": 3.5235709628352656e-07,
+      "loss": 0.4812,
+      "step": 54258
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.673241138458252,
+      "learning_rate": 3.5225987877243686e-07,
+      "loss": 0.4751,
+      "step": 54259
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.9499459266662598,
+      "learning_rate": 3.521626741849704e-07,
+      "loss": 0.6752,
+      "step": 54260
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.485703706741333,
+      "learning_rate": 3.5206548252139925e-07,
+      "loss": 0.5293,
+      "step": 54261
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 4.277674674987793,
+      "learning_rate": 3.5196830378199207e-07,
+      "loss": 0.5721,
+      "step": 54262
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.8249828815460205,
+      "learning_rate": 3.5187113796702086e-07,
+      "loss": 0.7179,
+      "step": 54263
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.929206609725952,
+      "learning_rate": 3.5177398507675377e-07,
+      "loss": 0.6348,
+      "step": 54264
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 5.40234899520874,
+      "learning_rate": 3.5167684511146273e-07,
+      "loss": 0.553,
+      "step": 54265
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.0162177085876465,
+      "learning_rate": 3.5157971807141643e-07,
+      "loss": 0.4,
+      "step": 54266
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.50346302986145,
+      "learning_rate": 3.514826039568864e-07,
+      "loss": 0.4937,
+      "step": 54267
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.1102426052093506,
+      "learning_rate": 3.513855027681412e-07,
+      "loss": 0.4084,
+      "step": 54268
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.5409979820251465,
+      "learning_rate": 3.512884145054518e-07,
+      "loss": 0.4804,
+      "step": 54269
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.238717555999756,
+      "learning_rate": 3.511913391690874e-07,
+      "loss": 0.8491,
+      "step": 54270
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.515425682067871,
+      "learning_rate": 3.510942767593184e-07,
+      "loss": 0.3371,
+      "step": 54271
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.3550713062286377,
+      "learning_rate": 3.509972272764156e-07,
+      "loss": 0.5082,
+      "step": 54272
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.9303717613220215,
+      "learning_rate": 3.5090019072064716e-07,
+      "loss": 0.8542,
+      "step": 54273
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.119138240814209,
+      "learning_rate": 3.5080316709228344e-07,
+      "loss": 0.5823,
+      "step": 54274
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.5228614807128906,
+      "learning_rate": 3.507061563915953e-07,
+      "loss": 0.6872,
+      "step": 54275
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.135406255722046,
+      "learning_rate": 3.506091586188515e-07,
+      "loss": 0.774,
+      "step": 54276
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.176544666290283,
+      "learning_rate": 3.5051217377432224e-07,
+      "loss": 0.6331,
+      "step": 54277
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.6247849464416504,
+      "learning_rate": 3.504152018582757e-07,
+      "loss": 0.4868,
+      "step": 54278
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.947572708129883,
+      "learning_rate": 3.503182428709839e-07,
+      "loss": 0.571,
+      "step": 54279
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.835875988006592,
+      "learning_rate": 3.5022129681271447e-07,
+      "loss": 0.5889,
+      "step": 54280
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.6131324768066406,
+      "learning_rate": 3.501243636837376e-07,
+      "loss": 0.4174,
+      "step": 54281
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.4389944076538086,
+      "learning_rate": 3.500274434843243e-07,
+      "loss": 0.7707,
+      "step": 54282
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.3277580738067627,
+      "learning_rate": 3.499305362147415e-07,
+      "loss": 0.5344,
+      "step": 54283
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.6886603832244873,
+      "learning_rate": 3.498336418752607e-07,
+      "loss": 0.6938,
+      "step": 54284
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.2794687747955322,
+      "learning_rate": 3.4973676046615003e-07,
+      "loss": 0.497,
+      "step": 54285
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.2925264835357666,
+      "learning_rate": 3.496398919876798e-07,
+      "loss": 0.4577,
+      "step": 54286
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.797046184539795,
+      "learning_rate": 3.495430364401187e-07,
+      "loss": 0.4993,
+      "step": 54287
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.2640082836151123,
+      "learning_rate": 3.494461938237365e-07,
+      "loss": 0.3417,
+      "step": 54288
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.5445005893707275,
+      "learning_rate": 3.493493641388018e-07,
+      "loss": 0.6651,
+      "step": 54289
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.500401258468628,
+      "learning_rate": 3.4925254738558566e-07,
+      "loss": 0.5524,
+      "step": 54290
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.4594225883483887,
+      "learning_rate": 3.4915574356435444e-07,
+      "loss": 0.6974,
+      "step": 54291
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.604802131652832,
+      "learning_rate": 3.4905895267538017e-07,
+      "loss": 0.6616,
+      "step": 54292
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 4.426314353942871,
+      "learning_rate": 3.4896217471892924e-07,
+      "loss": 0.4729,
+      "step": 54293
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.026237726211548,
+      "learning_rate": 3.488654096952726e-07,
+      "loss": 0.5512,
+      "step": 54294
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.143843173980713,
+      "learning_rate": 3.4876865760467894e-07,
+      "loss": 0.4973,
+      "step": 54295
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.73783016204834,
+      "learning_rate": 3.48671918447418e-07,
+      "loss": 0.7684,
+      "step": 54296
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.5857436656951904,
+      "learning_rate": 3.485751922237568e-07,
+      "loss": 0.8762,
+      "step": 54297
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.332706928253174,
+      "learning_rate": 3.484784789339668e-07,
+      "loss": 0.489,
+      "step": 54298
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.1105124950408936,
+      "learning_rate": 3.48381778578315e-07,
+      "loss": 0.5689,
+      "step": 54299
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.3517398834228516,
+      "learning_rate": 3.4828509115707123e-07,
+      "loss": 0.4631,
+      "step": 54300
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 4.5008931159973145,
+      "learning_rate": 3.481884166705035e-07,
+      "loss": 0.7343,
+      "step": 54301
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.3162529468536377,
+      "learning_rate": 3.480917551188823e-07,
+      "loss": 0.4285,
+      "step": 54302
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 4.029056072235107,
+      "learning_rate": 3.4799510650247395e-07,
+      "loss": 0.3455,
+      "step": 54303
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.524519443511963,
+      "learning_rate": 3.4789847082154937e-07,
+      "loss": 0.446,
+      "step": 54304
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.6500418186187744,
+      "learning_rate": 3.478018480763762e-07,
+      "loss": 0.4078,
+      "step": 54305
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.656669855117798,
+      "learning_rate": 3.4770523826722356e-07,
+      "loss": 0.4204,
+      "step": 54306
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.471374034881592,
+      "learning_rate": 3.4760864139435913e-07,
+      "loss": 0.5018,
+      "step": 54307
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.759812355041504,
+      "learning_rate": 3.4751205745805315e-07,
+      "loss": 0.7441,
+      "step": 54308
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.7919023036956787,
+      "learning_rate": 3.474154864585727e-07,
+      "loss": 0.6213,
+      "step": 54309
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 5.3584184646606445,
+      "learning_rate": 3.4731892839618753e-07,
+      "loss": 0.585,
+      "step": 54310
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.4429237842559814,
+      "learning_rate": 3.4722238327116465e-07,
+      "loss": 0.5119,
+      "step": 54311
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.467909336090088,
+      "learning_rate": 3.4712585108377383e-07,
+      "loss": 0.7537,
+      "step": 54312
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.855994462966919,
+      "learning_rate": 3.470293318342827e-07,
+      "loss": 0.6144,
+      "step": 54313
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 4.559184551239014,
+      "learning_rate": 3.469328255229598e-07,
+      "loss": 0.4795,
+      "step": 54314
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 4.9959940910339355,
+      "learning_rate": 3.468363321500734e-07,
+      "loss": 0.4641,
+      "step": 54315
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.231581926345825,
+      "learning_rate": 3.4673985171589317e-07,
+      "loss": 0.6666,
+      "step": 54316
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.8580431938171387,
+      "learning_rate": 3.466433842206851e-07,
+      "loss": 0.6212,
+      "step": 54317
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.80403995513916,
+      "learning_rate": 3.4654692966472003e-07,
+      "loss": 0.5381,
+      "step": 54318
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.24615740776062,
+      "learning_rate": 3.464504880482633e-07,
+      "loss": 0.5228,
+      "step": 54319
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.626661777496338,
+      "learning_rate": 3.463540593715853e-07,
+      "loss": 0.778,
+      "step": 54320
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.2655909061431885,
+      "learning_rate": 3.4625764363495295e-07,
+      "loss": 0.5047,
+      "step": 54321
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.0749399662017822,
+      "learning_rate": 3.4616124083863554e-07,
+      "loss": 0.558,
+      "step": 54322
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.8295013904571533,
+      "learning_rate": 3.4606485098289897e-07,
+      "loss": 0.4542,
+      "step": 54323
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.3503148555755615,
+      "learning_rate": 3.4596847406801414e-07,
+      "loss": 0.6661,
+      "step": 54324
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.815335750579834,
+      "learning_rate": 3.458721100942464e-07,
+      "loss": 0.5098,
+      "step": 54325
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.106630325317383,
+      "learning_rate": 3.45775759061866e-07,
+      "loss": 0.5327,
+      "step": 54326
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.4901962280273438,
+      "learning_rate": 3.456794209711384e-07,
+      "loss": 0.3928,
+      "step": 54327
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.7110435962677,
+      "learning_rate": 3.4558309582233384e-07,
+      "loss": 0.5842,
+      "step": 54328
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.566579818725586,
+      "learning_rate": 3.454867836157183e-07,
+      "loss": 0.7103,
+      "step": 54329
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.7695560455322266,
+      "learning_rate": 3.45390484351561e-07,
+      "loss": 0.6274,
+      "step": 54330
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.9967305660247803,
+      "learning_rate": 3.452941980301289e-07,
+      "loss": 0.54,
+      "step": 54331
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 5.5581560134887695,
+      "learning_rate": 3.4519792465168955e-07,
+      "loss": 0.5591,
+      "step": 54332
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 4.013919830322266,
+      "learning_rate": 3.451016642165117e-07,
+      "loss": 0.927,
+      "step": 54333
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.3404908180236816,
+      "learning_rate": 3.4500541672486177e-07,
+      "loss": 0.4638,
+      "step": 54334
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.882499933242798,
+      "learning_rate": 3.4490918217700784e-07,
+      "loss": 0.4264,
+      "step": 54335
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.731435775756836,
+      "learning_rate": 3.4481296057321813e-07,
+      "loss": 0.6779,
+      "step": 54336
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.7486884593963623,
+      "learning_rate": 3.4471675191375955e-07,
+      "loss": 0.5695,
+      "step": 54337
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 5.400201797485352,
+      "learning_rate": 3.446205561989002e-07,
+      "loss": 0.5867,
+      "step": 54338
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.4098174571990967,
+      "learning_rate": 3.4452437342890667e-07,
+      "loss": 0.7806,
+      "step": 54339
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 4.538758277893066,
+      "learning_rate": 3.4442820360404695e-07,
+      "loss": 0.5969,
+      "step": 54340
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 7.836548805236816,
+      "learning_rate": 3.443320467245881e-07,
+      "loss": 0.6793,
+      "step": 54341
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.0628457069396973,
+      "learning_rate": 3.442359027907988e-07,
+      "loss": 0.5121,
+      "step": 54342
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.584245204925537,
+      "learning_rate": 3.441397718029443e-07,
+      "loss": 0.654,
+      "step": 54343
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.5316874980926514,
+      "learning_rate": 3.440436537612929e-07,
+      "loss": 0.6221,
+      "step": 54344
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.402061939239502,
+      "learning_rate": 3.439475486661126e-07,
+      "loss": 0.4844,
+      "step": 54345
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.03650164604187,
+      "learning_rate": 3.4385145651766925e-07,
+      "loss": 0.8104,
+      "step": 54346
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 4.148941516876221,
+      "learning_rate": 3.4375537731623167e-07,
+      "loss": 0.6994,
+      "step": 54347
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.894390821456909,
+      "learning_rate": 3.4365931106206563e-07,
+      "loss": 0.5153,
+      "step": 54348
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 4.965895175933838,
+      "learning_rate": 3.435632577554393e-07,
+      "loss": 0.7391,
+      "step": 54349
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.866187572479248,
+      "learning_rate": 3.4346721739661805e-07,
+      "loss": 0.4426,
+      "step": 54350
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 4.6490092277526855,
+      "learning_rate": 3.433711899858716e-07,
+      "loss": 0.7021,
+      "step": 54351
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.5794875621795654,
+      "learning_rate": 3.4327517552346425e-07,
+      "loss": 0.5685,
+      "step": 54352
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.0304012298583984,
+      "learning_rate": 3.431791740096646e-07,
+      "loss": 0.4336,
+      "step": 54353
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 4.156428813934326,
+      "learning_rate": 3.430831854447386e-07,
+      "loss": 0.6269,
+      "step": 54354
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.735948085784912,
+      "learning_rate": 3.4298720982895437e-07,
+      "loss": 0.4258,
+      "step": 54355
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.4878005981445312,
+      "learning_rate": 3.4289124716257727e-07,
+      "loss": 0.4549,
+      "step": 54356
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 4.45217227935791,
+      "learning_rate": 3.4279529744587646e-07,
+      "loss": 0.5849,
+      "step": 54357
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 1.637387752532959,
+      "learning_rate": 3.4269936067911626e-07,
+      "loss": 0.3578,
+      "step": 54358
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.834383010864258,
+      "learning_rate": 3.4260343686256524e-07,
+      "loss": 0.6234,
+      "step": 54359
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.2332136631011963,
+      "learning_rate": 3.425075259964888e-07,
+      "loss": 0.686,
+      "step": 54360
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.436527729034424,
+      "learning_rate": 3.424116280811546e-07,
+      "loss": 0.7218,
+      "step": 54361
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.8027894496917725,
+      "learning_rate": 3.4231574311682837e-07,
+      "loss": 0.5455,
+      "step": 54362
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.821096420288086,
+      "learning_rate": 3.4221987110377775e-07,
+      "loss": 0.5386,
+      "step": 54363
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.341493606567383,
+      "learning_rate": 3.4212401204226863e-07,
+      "loss": 0.3657,
+      "step": 54364
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.826303243637085,
+      "learning_rate": 3.4202816593256804e-07,
+      "loss": 0.7195,
+      "step": 54365
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.8850772380828857,
+      "learning_rate": 3.419323327749413e-07,
+      "loss": 0.6422,
+      "step": 54366
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.554121971130371,
+      "learning_rate": 3.4183651256965713e-07,
+      "loss": 0.3777,
+      "step": 54367
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.859492301940918,
+      "learning_rate": 3.417407053169797e-07,
+      "loss": 0.5288,
+      "step": 54368
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 4.43788480758667,
+      "learning_rate": 3.4164491101717657e-07,
+      "loss": 0.7321,
+      "step": 54369
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.174400568008423,
+      "learning_rate": 3.4154912967051366e-07,
+      "loss": 0.3992,
+      "step": 54370
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.785693407058716,
+      "learning_rate": 3.41453361277258e-07,
+      "loss": 0.5202,
+      "step": 54371
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.4742376804351807,
+      "learning_rate": 3.4135760583767496e-07,
+      "loss": 0.5581,
+      "step": 54372
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.8758890628814697,
+      "learning_rate": 3.4126186335203204e-07,
+      "loss": 0.5184,
+      "step": 54373
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.534623622894287,
+      "learning_rate": 3.411661338205935e-07,
+      "loss": 0.6694,
+      "step": 54374
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.8277053833007812,
+      "learning_rate": 3.410704172436269e-07,
+      "loss": 0.7902,
+      "step": 54375
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.5483038425445557,
+      "learning_rate": 3.409747136213981e-07,
+      "loss": 0.5706,
+      "step": 54376
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.6184463500976562,
+      "learning_rate": 3.4087902295417416e-07,
+      "loss": 0.5944,
+      "step": 54377
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.7691314220428467,
+      "learning_rate": 3.4078334524221936e-07,
+      "loss": 0.5536,
+      "step": 54378
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 4.268438816070557,
+      "learning_rate": 3.406876804858017e-07,
+      "loss": 0.5095,
+      "step": 54379
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 4.218228816986084,
+      "learning_rate": 3.405920286851855e-07,
+      "loss": 0.5974,
+      "step": 54380
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.2112011909484863,
+      "learning_rate": 3.4049638984063826e-07,
+      "loss": 0.4664,
+      "step": 54381
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.1087539196014404,
+      "learning_rate": 3.404007639524237e-07,
+      "loss": 0.5779,
+      "step": 54382
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.577730178833008,
+      "learning_rate": 3.4030515102081053e-07,
+      "loss": 0.4434,
+      "step": 54383
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.689457893371582,
+      "learning_rate": 3.4020955104606235e-07,
+      "loss": 0.5773,
+      "step": 54384
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.827622413635254,
+      "learning_rate": 3.401139640284462e-07,
+      "loss": 0.3877,
+      "step": 54385
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.1308908462524414,
+      "learning_rate": 3.400183899682269e-07,
+      "loss": 0.3718,
+      "step": 54386
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.555314064025879,
+      "learning_rate": 3.399228288656714e-07,
+      "loss": 0.5472,
+      "step": 54387
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.368067741394043,
+      "learning_rate": 3.3982728072104397e-07,
+      "loss": 0.4258,
+      "step": 54388
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 4.700398921966553,
+      "learning_rate": 3.3973174553461217e-07,
+      "loss": 0.5506,
+      "step": 54389
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.8379757404327393,
+      "learning_rate": 3.3963622330664027e-07,
+      "loss": 0.4763,
+      "step": 54390
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.3340518474578857,
+      "learning_rate": 3.395407140373941e-07,
+      "loss": 0.6991,
+      "step": 54391
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.445998191833496,
+      "learning_rate": 3.3944521772713903e-07,
+      "loss": 0.5039,
+      "step": 54392
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.8923192024230957,
+      "learning_rate": 3.3934973437614093e-07,
+      "loss": 0.5803,
+      "step": 54393
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.2414963245391846,
+      "learning_rate": 3.3925426398466577e-07,
+      "loss": 0.6203,
+      "step": 54394
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.8348748683929443,
+      "learning_rate": 3.391588065529783e-07,
+      "loss": 0.619,
+      "step": 54395
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.4111852645874023,
+      "learning_rate": 3.3906336208134384e-07,
+      "loss": 0.5588,
+      "step": 54396
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 1.9536991119384766,
+      "learning_rate": 3.3896793057002885e-07,
+      "loss": 0.528,
+      "step": 54397
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.0294346809387207,
+      "learning_rate": 3.388725120192976e-07,
+      "loss": 0.5342,
+      "step": 54398
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.3375868797302246,
+      "learning_rate": 3.3877710642941654e-07,
+      "loss": 0.4446,
+      "step": 54399
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.380892038345337,
+      "learning_rate": 3.386817138006487e-07,
+      "loss": 0.5517,
+      "step": 54400
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.0071120262145996,
+      "learning_rate": 3.385863341332624e-07,
+      "loss": 0.5666,
+      "step": 54401
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.333812713623047,
+      "learning_rate": 3.3849096742752053e-07,
+      "loss": 0.7377,
+      "step": 54402
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.5513710975646973,
+      "learning_rate": 3.3839561368368914e-07,
+      "loss": 0.6494,
+      "step": 54403
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.903876781463623,
+      "learning_rate": 3.3830027290203295e-07,
+      "loss": 0.5746,
+      "step": 54404
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.303948402404785,
+      "learning_rate": 3.382049450828173e-07,
+      "loss": 0.5676,
+      "step": 54405
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.6560404300689697,
+      "learning_rate": 3.381096302263082e-07,
+      "loss": 0.4465,
+      "step": 54406
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 4.2493896484375,
+      "learning_rate": 3.3801432833276915e-07,
+      "loss": 0.7828,
+      "step": 54407
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.9101409912109375,
+      "learning_rate": 3.379190394024667e-07,
+      "loss": 0.5922,
+      "step": 54408
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.1406502723693848,
+      "learning_rate": 3.378237634356635e-07,
+      "loss": 0.7226,
+      "step": 54409
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.2058866024017334,
+      "learning_rate": 3.377285004326275e-07,
+      "loss": 0.54,
+      "step": 54410
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.1511189937591553,
+      "learning_rate": 3.3763325039362084e-07,
+      "loss": 0.7202,
+      "step": 54411
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.075029134750366,
+      "learning_rate": 3.3753801331891043e-07,
+      "loss": 0.6463,
+      "step": 54412
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.4576973915100098,
+      "learning_rate": 3.374427892087595e-07,
+      "loss": 0.6909,
+      "step": 54413
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.559617042541504,
+      "learning_rate": 3.3734757806343386e-07,
+      "loss": 0.6868,
+      "step": 54414
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.1756327152252197,
+      "learning_rate": 3.372523798831978e-07,
+      "loss": 0.7356,
+      "step": 54415
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.0370283126831055,
+      "learning_rate": 3.3715719466831553e-07,
+      "loss": 0.5507,
+      "step": 54416
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.658276319503784,
+      "learning_rate": 3.3706202241905293e-07,
+      "loss": 0.6276,
+      "step": 54417
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.412607431411743,
+      "learning_rate": 3.369668631356743e-07,
+      "loss": 0.6088,
+      "step": 54418
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.12640643119812,
+      "learning_rate": 3.368717168184438e-07,
+      "loss": 0.4299,
+      "step": 54419
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.717108726501465,
+      "learning_rate": 3.3677658346762674e-07,
+      "loss": 0.6641,
+      "step": 54420
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.4443488121032715,
+      "learning_rate": 3.3668146308348635e-07,
+      "loss": 0.7673,
+      "step": 54421
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.7947704792022705,
+      "learning_rate": 3.365863556662885e-07,
+      "loss": 0.6779,
+      "step": 54422
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.856799602508545,
+      "learning_rate": 3.3649126121629625e-07,
+      "loss": 0.5678,
+      "step": 54423
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.292835235595703,
+      "learning_rate": 3.3639617973377615e-07,
+      "loss": 0.432,
+      "step": 54424
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 4.482272148132324,
+      "learning_rate": 3.3630111121898954e-07,
+      "loss": 0.472,
+      "step": 54425
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 4.612947463989258,
+      "learning_rate": 3.362060556722041e-07,
+      "loss": 0.8212,
+      "step": 54426
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.1076016426086426,
+      "learning_rate": 3.3611101309368124e-07,
+      "loss": 0.3699,
+      "step": 54427
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.102292776107788,
+      "learning_rate": 3.360159834836879e-07,
+      "loss": 0.6674,
+      "step": 54428
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.388122797012329,
+      "learning_rate": 3.3592096684248567e-07,
+      "loss": 0.7674,
+      "step": 54429
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.612971544265747,
+      "learning_rate": 3.358259631703409e-07,
+      "loss": 0.4602,
+      "step": 54430
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.1236536502838135,
+      "learning_rate": 3.3573097246751563e-07,
+      "loss": 0.5808,
+      "step": 54431
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.5266716480255127,
+      "learning_rate": 3.356359947342769e-07,
+      "loss": 0.6529,
+      "step": 54432
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 4.014199733734131,
+      "learning_rate": 3.3554102997088555e-07,
+      "loss": 0.7154,
+      "step": 54433
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.721386671066284,
+      "learning_rate": 3.3544607817760864e-07,
+      "loss": 0.6095,
+      "step": 54434
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.8940916061401367,
+      "learning_rate": 3.3535113935470817e-07,
+      "loss": 0.7254,
+      "step": 54435
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.2871854305267334,
+      "learning_rate": 3.3525621350244843e-07,
+      "loss": 0.6262,
+      "step": 54436
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 4.281846046447754,
+      "learning_rate": 3.351613006210935e-07,
+      "loss": 0.8198,
+      "step": 54437
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.6299877166748047,
+      "learning_rate": 3.3506640071090836e-07,
+      "loss": 0.5832,
+      "step": 54438
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.405186414718628,
+      "learning_rate": 3.349715137721554e-07,
+      "loss": 0.4113,
+      "step": 54439
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.4153053760528564,
+      "learning_rate": 3.348766398051001e-07,
+      "loss": 0.4016,
+      "step": 54440
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.7622907161712646,
+      "learning_rate": 3.347817788100038e-07,
+      "loss": 0.347,
+      "step": 54441
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.8547022342681885,
+      "learning_rate": 3.346869307871331e-07,
+      "loss": 0.5198,
+      "step": 54442
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.6657049655914307,
+      "learning_rate": 3.345920957367499e-07,
+      "loss": 0.6848,
+      "step": 54443
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 4.384881973266602,
+      "learning_rate": 3.344972736591184e-07,
+      "loss": 0.7155,
+      "step": 54444
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.7806079387664795,
+      "learning_rate": 3.344024645545019e-07,
+      "loss": 0.5045,
+      "step": 54445
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.707418918609619,
+      "learning_rate": 3.34307668423165e-07,
+      "loss": 0.6934,
+      "step": 54446
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.714427947998047,
+      "learning_rate": 3.3421288526536977e-07,
+      "loss": 0.6136,
+      "step": 54447
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 4.961358547210693,
+      "learning_rate": 3.3411811508138157e-07,
+      "loss": 0.5563,
+      "step": 54448
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.820204973220825,
+      "learning_rate": 3.3402335787146246e-07,
+      "loss": 0.5347,
+      "step": 54449
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.583163261413574,
+      "learning_rate": 3.3392861363587713e-07,
+      "loss": 0.4206,
+      "step": 54450
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.666085720062256,
+      "learning_rate": 3.338338823748877e-07,
+      "loss": 0.5743,
+      "step": 54451
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.419667959213257,
+      "learning_rate": 3.3373916408875884e-07,
+      "loss": 0.5469,
+      "step": 54452
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.53979754447937,
+      "learning_rate": 3.3364445877775263e-07,
+      "loss": 0.4987,
+      "step": 54453
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 4.536853790283203,
+      "learning_rate": 3.335497664421339e-07,
+      "loss": 0.6974,
+      "step": 54454
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.908670663833618,
+      "learning_rate": 3.33455087082164e-07,
+      "loss": 0.8222,
+      "step": 54455
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.601909637451172,
+      "learning_rate": 3.3336042069810783e-07,
+      "loss": 0.6701,
+      "step": 54456
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 1.7438409328460693,
+      "learning_rate": 3.332657672902284e-07,
+      "loss": 0.399,
+      "step": 54457
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.3195440769195557,
+      "learning_rate": 3.3317112685878894e-07,
+      "loss": 0.5178,
+      "step": 54458
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.7355470657348633,
+      "learning_rate": 3.3307649940405194e-07,
+      "loss": 0.5257,
+      "step": 54459
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.816150188446045,
+      "learning_rate": 3.3298188492628115e-07,
+      "loss": 0.5826,
+      "step": 54460
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.7378718852996826,
+      "learning_rate": 3.3288728342573904e-07,
+      "loss": 0.5885,
+      "step": 54461
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.6486854553222656,
+      "learning_rate": 3.3279269490268993e-07,
+      "loss": 0.6911,
+      "step": 54462
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.1390841007232666,
+      "learning_rate": 3.3269811935739524e-07,
+      "loss": 0.3946,
+      "step": 54463
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.541039228439331,
+      "learning_rate": 3.3260355679011925e-07,
+      "loss": 0.6148,
+      "step": 54464
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.9094276428222656,
+      "learning_rate": 3.325090072011233e-07,
+      "loss": 0.4914,
+      "step": 54465
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.8510613441467285,
+      "learning_rate": 3.3241447059067235e-07,
+      "loss": 0.5672,
+      "step": 54466
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.581198215484619,
+      "learning_rate": 3.323199469590277e-07,
+      "loss": 0.4747,
+      "step": 54467
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.1363365650177,
+      "learning_rate": 3.3222543630645257e-07,
+      "loss": 0.5656,
+      "step": 54468
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.3769023418426514,
+      "learning_rate": 3.3213093863321056e-07,
+      "loss": 0.5511,
+      "step": 54469
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.7723031044006348,
+      "learning_rate": 3.320364539395632e-07,
+      "loss": 0.6939,
+      "step": 54470
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.9269940853118896,
+      "learning_rate": 3.3194198222577413e-07,
+      "loss": 0.6908,
+      "step": 54471
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.22607684135437,
+      "learning_rate": 3.3184752349210535e-07,
+      "loss": 0.6068,
+      "step": 54472
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.146486759185791,
+      "learning_rate": 3.317530777388206e-07,
+      "loss": 0.5655,
+      "step": 54473
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.3667964935302734,
+      "learning_rate": 3.3165864496618073e-07,
+      "loss": 0.6693,
+      "step": 54474
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 1.9237159490585327,
+      "learning_rate": 3.315642251744505e-07,
+      "loss": 0.3798,
+      "step": 54475
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.6312620639801025,
+      "learning_rate": 3.3146981836389035e-07,
+      "loss": 0.7042,
+      "step": 54476
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.2615890502929688,
+      "learning_rate": 3.313754245347639e-07,
+      "loss": 0.5632,
+      "step": 54477
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.8696773052215576,
+      "learning_rate": 3.3128104368733374e-07,
+      "loss": 0.7718,
+      "step": 54478
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 1.847652792930603,
+      "learning_rate": 3.3118667582186236e-07,
+      "loss": 0.3694,
+      "step": 54479
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.9771344661712646,
+      "learning_rate": 3.310923209386113e-07,
+      "loss": 0.7421,
+      "step": 54480
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.512510061264038,
+      "learning_rate": 3.3099797903784423e-07,
+      "loss": 0.4621,
+      "step": 54481
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.1768147945404053,
+      "learning_rate": 3.3090365011982206e-07,
+      "loss": 0.5768,
+      "step": 54482
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 5.0892815589904785,
+      "learning_rate": 3.308093341848079e-07,
+      "loss": 0.5826,
+      "step": 54483
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 4.3136820793151855,
+      "learning_rate": 3.307150312330637e-07,
+      "loss": 0.6058,
+      "step": 54484
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.390639543533325,
+      "learning_rate": 3.3062074126485266e-07,
+      "loss": 0.6032,
+      "step": 54485
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.317470073699951,
+      "learning_rate": 3.305264642804351e-07,
+      "loss": 0.458,
+      "step": 54486
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.157839775085449,
+      "learning_rate": 3.3043220028007527e-07,
+      "loss": 0.294,
+      "step": 54487
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.928807497024536,
+      "learning_rate": 3.3033794926403293e-07,
+      "loss": 0.3763,
+      "step": 54488
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 4.062666416168213,
+      "learning_rate": 3.3024371123257295e-07,
+      "loss": 0.7002,
+      "step": 54489
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.863750457763672,
+      "learning_rate": 3.3014948618595445e-07,
+      "loss": 0.6905,
+      "step": 54490
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.3327596187591553,
+      "learning_rate": 3.3005527412444226e-07,
+      "loss": 0.3348,
+      "step": 54491
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 4.627323627471924,
+      "learning_rate": 3.2996107504829565e-07,
+      "loss": 0.6779,
+      "step": 54492
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.225931406021118,
+      "learning_rate": 3.2986688895777887e-07,
+      "loss": 0.8331,
+      "step": 54493
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 1.983359456062317,
+      "learning_rate": 3.297727158531522e-07,
+      "loss": 0.4089,
+      "step": 54494
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.2557990550994873,
+      "learning_rate": 3.2967855573467823e-07,
+      "loss": 0.4186,
+      "step": 54495
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 1.9944785833358765,
+      "learning_rate": 3.2958440860261843e-07,
+      "loss": 0.5033,
+      "step": 54496
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.794235944747925,
+      "learning_rate": 3.294902744572348e-07,
+      "loss": 0.5322,
+      "step": 54497
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.347503185272217,
+      "learning_rate": 3.293961532987888e-07,
+      "loss": 0.4548,
+      "step": 54498
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 4.793853759765625,
+      "learning_rate": 3.293020451275436e-07,
+      "loss": 0.6401,
+      "step": 54499
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.603156566619873,
+      "learning_rate": 3.292079499437589e-07,
+      "loss": 0.5285,
+      "step": 54500
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.040661096572876,
+      "learning_rate": 3.291138677476979e-07,
+      "loss": 0.5206,
+      "step": 54501
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.2503323554992676,
+      "learning_rate": 3.290197985396204e-07,
+      "loss": 0.4015,
+      "step": 54502
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.5169873237609863,
+      "learning_rate": 3.2892574231979e-07,
+      "loss": 0.5363,
+      "step": 54503
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.970168352127075,
+      "learning_rate": 3.288316990884666e-07,
+      "loss": 0.4916,
+      "step": 54504
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.443448305130005,
+      "learning_rate": 3.287376688459132e-07,
+      "loss": 0.6388,
+      "step": 54505
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.8344011306762695,
+      "learning_rate": 3.2864365159238966e-07,
+      "loss": 0.6229,
+      "step": 54506
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.4705092906951904,
+      "learning_rate": 3.285496473281591e-07,
+      "loss": 0.6125,
+      "step": 54507
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.744560956954956,
+      "learning_rate": 3.2845565605348074e-07,
+      "loss": 0.7349,
+      "step": 54508
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 4.488748073577881,
+      "learning_rate": 3.283616777686183e-07,
+      "loss": 0.5571,
+      "step": 54509
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.839615821838379,
+      "learning_rate": 3.2826771247383094e-07,
+      "loss": 0.527,
+      "step": 54510
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 4.187271595001221,
+      "learning_rate": 3.281737601693824e-07,
+      "loss": 0.5878,
+      "step": 54511
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.656916856765747,
+      "learning_rate": 3.280798208555314e-07,
+      "loss": 0.4952,
+      "step": 54512
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.3475255966186523,
+      "learning_rate": 3.2798589453254094e-07,
+      "loss": 0.5158,
+      "step": 54513
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 4.0529465675354,
+      "learning_rate": 3.278919812006709e-07,
+      "loss": 0.5923,
+      "step": 54514
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.134246349334717,
+      "learning_rate": 3.2779808086018386e-07,
+      "loss": 0.5683,
+      "step": 54515
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.074995279312134,
+      "learning_rate": 3.27704193511339e-07,
+      "loss": 0.5058,
+      "step": 54516
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.0003445148468018,
+      "learning_rate": 3.276103191543989e-07,
+      "loss": 0.6321,
+      "step": 54517
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 4.48151969909668,
+      "learning_rate": 3.2751645778962394e-07,
+      "loss": 0.5769,
+      "step": 54518
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.257781505584717,
+      "learning_rate": 3.274226094172761e-07,
+      "loss": 0.5829,
+      "step": 54519
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.366485834121704,
+      "learning_rate": 3.273287740376152e-07,
+      "loss": 0.6879,
+      "step": 54520
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.189514398574829,
+      "learning_rate": 3.272349516509032e-07,
+      "loss": 0.6979,
+      "step": 54521
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.2940542697906494,
+      "learning_rate": 3.2714114225739946e-07,
+      "loss": 0.4829,
+      "step": 54522
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 4.83405065536499,
+      "learning_rate": 3.2704734585736644e-07,
+      "loss": 0.7662,
+      "step": 54523
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.638002634048462,
+      "learning_rate": 3.2695356245106336e-07,
+      "loss": 0.5094,
+      "step": 54524
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.85949444770813,
+      "learning_rate": 3.2685979203875286e-07,
+      "loss": 0.8901,
+      "step": 54525
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.205233335494995,
+      "learning_rate": 3.267660346206936e-07,
+      "loss": 0.5323,
+      "step": 54526
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.271158218383789,
+      "learning_rate": 3.266722901971481e-07,
+      "loss": 0.6814,
+      "step": 54527
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.6613590717315674,
+      "learning_rate": 3.265785587683762e-07,
+      "loss": 0.5751,
+      "step": 54528
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.8547534942626953,
+      "learning_rate": 3.2648484033463765e-07,
+      "loss": 0.6536,
+      "step": 54529
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.8548378944396973,
+      "learning_rate": 3.263911348961957e-07,
+      "loss": 0.3675,
+      "step": 54530
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.0913946628570557,
+      "learning_rate": 3.262974424533077e-07,
+      "loss": 0.617,
+      "step": 54531
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.880380868911743,
+      "learning_rate": 3.26203763006237e-07,
+      "loss": 0.6209,
+      "step": 54532
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 23.846454620361328,
+      "learning_rate": 3.261100965552422e-07,
+      "loss": 0.5175,
+      "step": 54533
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.0673892498016357,
+      "learning_rate": 3.2601644310058465e-07,
+      "loss": 0.4465,
+      "step": 54534
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 4.073832035064697,
+      "learning_rate": 3.259228026425237e-07,
+      "loss": 0.577,
+      "step": 54535
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 4.183732032775879,
+      "learning_rate": 3.258291751813214e-07,
+      "loss": 0.6633,
+      "step": 54536
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.355428457260132,
+      "learning_rate": 3.257355607172363e-07,
+      "loss": 0.6099,
+      "step": 54537
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 4.118654727935791,
+      "learning_rate": 3.2564195925053e-07,
+      "loss": 0.5048,
+      "step": 54538
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.4999122619628906,
+      "learning_rate": 3.255483707814627e-07,
+      "loss": 0.7679,
+      "step": 54539
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.8473613262176514,
+      "learning_rate": 3.2545479531029376e-07,
+      "loss": 0.7489,
+      "step": 54540
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 9.289323806762695,
+      "learning_rate": 3.253612328372835e-07,
+      "loss": 0.6286,
+      "step": 54541
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.131721019744873,
+      "learning_rate": 3.2526768336269333e-07,
+      "loss": 0.7179,
+      "step": 54542
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.4061872959136963,
+      "learning_rate": 3.251741468867825e-07,
+      "loss": 0.3927,
+      "step": 54543
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.4748876094818115,
+      "learning_rate": 3.250806234098114e-07,
+      "loss": 0.452,
+      "step": 54544
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.615586996078491,
+      "learning_rate": 3.2498711293203864e-07,
+      "loss": 0.4428,
+      "step": 54545
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.052030563354492,
+      "learning_rate": 3.248936154537269e-07,
+      "loss": 0.6297,
+      "step": 54546
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.392595052719116,
+      "learning_rate": 3.2480013097513365e-07,
+      "loss": 0.5819,
+      "step": 54547
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.0107333660125732,
+      "learning_rate": 3.2470665949652034e-07,
+      "loss": 0.6119,
+      "step": 54548
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.912799119949341,
+      "learning_rate": 3.2461320101814573e-07,
+      "loss": 0.6853,
+      "step": 54549
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.311284303665161,
+      "learning_rate": 3.2451975554027124e-07,
+      "loss": 0.6862,
+      "step": 54550
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.117757558822632,
+      "learning_rate": 3.2442632306315504e-07,
+      "loss": 0.596,
+      "step": 54551
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.9096598625183105,
+      "learning_rate": 3.2433290358705904e-07,
+      "loss": 0.7307,
+      "step": 54552
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 7.329758167266846,
+      "learning_rate": 3.2423949711224036e-07,
+      "loss": 0.4172,
+      "step": 54553
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.778270959854126,
+      "learning_rate": 3.2414610363896094e-07,
+      "loss": 0.5166,
+      "step": 54554
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.109511137008667,
+      "learning_rate": 3.240527231674784e-07,
+      "loss": 0.7571,
+      "step": 54555
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.606527090072632,
+      "learning_rate": 3.239593556980547e-07,
+      "loss": 0.7936,
+      "step": 54556
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.6473429203033447,
+      "learning_rate": 3.2386600123094803e-07,
+      "loss": 0.3586,
+      "step": 54557
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.2004308700561523,
+      "learning_rate": 3.237726597664176e-07,
+      "loss": 0.6551,
+      "step": 54558
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.5967366695404053,
+      "learning_rate": 3.236793313047243e-07,
+      "loss": 0.5183,
+      "step": 54559
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.796591281890869,
+      "learning_rate": 3.2358601584612736e-07,
+      "loss": 0.5266,
+      "step": 54560
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.590244770050049,
+      "learning_rate": 3.2349271339088494e-07,
+      "loss": 0.5316,
+      "step": 54561
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.195915699005127,
+      "learning_rate": 3.233994239392585e-07,
+      "loss": 0.5075,
+      "step": 54562
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.1006805896759033,
+      "learning_rate": 3.233061474915056e-07,
+      "loss": 0.6541,
+      "step": 54563
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.2824840545654297,
+      "learning_rate": 3.232128840478871e-07,
+      "loss": 0.4299,
+      "step": 54564
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.7460014820098877,
+      "learning_rate": 3.231196336086606e-07,
+      "loss": 0.642,
+      "step": 54565
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.945218563079834,
+      "learning_rate": 3.2302639617408703e-07,
+      "loss": 0.567,
+      "step": 54566
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.3762919902801514,
+      "learning_rate": 3.229331717444245e-07,
+      "loss": 0.6659,
+      "step": 54567
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.884052276611328,
+      "learning_rate": 3.2283996031993394e-07,
+      "loss": 0.5273,
+      "step": 54568
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 6.930631160736084,
+      "learning_rate": 3.2274676190087227e-07,
+      "loss": 0.5435,
+      "step": 54569
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.214545965194702,
+      "learning_rate": 3.226535764875005e-07,
+      "loss": 0.7204,
+      "step": 54570
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.4281973838806152,
+      "learning_rate": 3.2256040408007615e-07,
+      "loss": 0.4846,
+      "step": 54571
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.194209575653076,
+      "learning_rate": 3.224672446788596e-07,
+      "loss": 0.5937,
+      "step": 54572
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.2395107746124268,
+      "learning_rate": 3.223740982841089e-07,
+      "loss": 0.5914,
+      "step": 54573
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.6410365104675293,
+      "learning_rate": 3.222809648960845e-07,
+      "loss": 0.6288,
+      "step": 54574
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 4.756432056427002,
+      "learning_rate": 3.221878445150434e-07,
+      "loss": 0.6056,
+      "step": 54575
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.870018243789673,
+      "learning_rate": 3.2209473714124637e-07,
+      "loss": 0.5228,
+      "step": 54576
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.281859874725342,
+      "learning_rate": 3.2200164277495063e-07,
+      "loss": 0.6394,
+      "step": 54577
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.9497897624969482,
+      "learning_rate": 3.219085614164158e-07,
+      "loss": 0.6603,
+      "step": 54578
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 5.7507548332214355,
+      "learning_rate": 3.218154930659012e-07,
+      "loss": 0.3692,
+      "step": 54579
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.2399742603302,
+      "learning_rate": 3.217224377236661e-07,
+      "loss": 0.4925,
+      "step": 54580
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.826516628265381,
+      "learning_rate": 3.2162939538996685e-07,
+      "loss": 0.7453,
+      "step": 54581
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.728694438934326,
+      "learning_rate": 3.21536366065065e-07,
+      "loss": 0.5136,
+      "step": 54582
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 9.9461669921875,
+      "learning_rate": 3.2144334974921755e-07,
+      "loss": 0.7332,
+      "step": 54583
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 4.1183762550354,
+      "learning_rate": 3.213503464426837e-07,
+      "loss": 0.7232,
+      "step": 54584
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.8127081394195557,
+      "learning_rate": 3.2125735614572106e-07,
+      "loss": 0.5427,
+      "step": 54585
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.9083094596862793,
+      "learning_rate": 3.2116437885858997e-07,
+      "loss": 0.5486,
+      "step": 54586
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.8931381702423096,
+      "learning_rate": 3.2107141458154746e-07,
+      "loss": 0.4336,
+      "step": 54587
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.27949857711792,
+      "learning_rate": 3.209784633148527e-07,
+      "loss": 0.7436,
+      "step": 54588
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.3515560626983643,
+      "learning_rate": 3.208855250587639e-07,
+      "loss": 0.5681,
+      "step": 54589
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 4.44568395614624,
+      "learning_rate": 3.2079259981353915e-07,
+      "loss": 0.7519,
+      "step": 54590
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.166931390762329,
+      "learning_rate": 3.206996875794388e-07,
+      "loss": 0.7384,
+      "step": 54591
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.7118887901306152,
+      "learning_rate": 3.206067883567182e-07,
+      "loss": 0.659,
+      "step": 54592
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.1374714374542236,
+      "learning_rate": 3.205139021456383e-07,
+      "loss": 0.5934,
+      "step": 54593
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.803058624267578,
+      "learning_rate": 3.2042102894645543e-07,
+      "loss": 0.52,
+      "step": 54594
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.6992228031158447,
+      "learning_rate": 3.203281687594295e-07,
+      "loss": 0.5559,
+      "step": 54595
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 4.148427486419678,
+      "learning_rate": 3.2023532158481694e-07,
+      "loss": 0.7337,
+      "step": 54596
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.704209089279175,
+      "learning_rate": 3.201424874228776e-07,
+      "loss": 0.6419,
+      "step": 54597
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.078443765640259,
+      "learning_rate": 3.2004966627386837e-07,
+      "loss": 0.6809,
+      "step": 54598
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.353304862976074,
+      "learning_rate": 3.19956858138048e-07,
+      "loss": 0.4922,
+      "step": 54599
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.895960807800293,
+      "learning_rate": 3.1986406301567464e-07,
+      "loss": 0.5259,
+      "step": 54600
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.36590313911438,
+      "learning_rate": 3.1977128090700584e-07,
+      "loss": 0.589,
+      "step": 54601
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.148590564727783,
+      "learning_rate": 3.196785118122997e-07,
+      "loss": 0.5715,
+      "step": 54602
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.5988683700561523,
+      "learning_rate": 3.1958575573181494e-07,
+      "loss": 0.5775,
+      "step": 54603
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.409759521484375,
+      "learning_rate": 3.19493012665808e-07,
+      "loss": 0.586,
+      "step": 54604
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.058671236038208,
+      "learning_rate": 3.194002826145387e-07,
+      "loss": 0.667,
+      "step": 54605
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.3081092834472656,
+      "learning_rate": 3.193075655782629e-07,
+      "loss": 0.518,
+      "step": 54606
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.389056921005249,
+      "learning_rate": 3.1921486155724044e-07,
+      "loss": 0.4316,
+      "step": 54607
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.186061382293701,
+      "learning_rate": 3.1912217055172666e-07,
+      "loss": 0.6451,
+      "step": 54608
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.4226796627044678,
+      "learning_rate": 3.190294925619819e-07,
+      "loss": 0.5151,
+      "step": 54609
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.162351131439209,
+      "learning_rate": 3.189368275882615e-07,
+      "loss": 0.3924,
+      "step": 54610
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 10.342907905578613,
+      "learning_rate": 3.1884417563082527e-07,
+      "loss": 0.5657,
+      "step": 54611
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.8408613204956055,
+      "learning_rate": 3.1875153668992854e-07,
+      "loss": 0.7229,
+      "step": 54612
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.60579776763916,
+      "learning_rate": 3.186589107658311e-07,
+      "loss": 0.5022,
+      "step": 54613
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.2365403175354004,
+      "learning_rate": 3.185662978587889e-07,
+      "loss": 0.5795,
+      "step": 54614
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.974459409713745,
+      "learning_rate": 3.1847369796906116e-07,
+      "loss": 0.7876,
+      "step": 54615
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.0882747173309326,
+      "learning_rate": 3.183811110969032e-07,
+      "loss": 0.6468,
+      "step": 54616
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.1420657634735107,
+      "learning_rate": 3.182885372425748e-07,
+      "loss": 0.6447,
+      "step": 54617
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.9839026927948,
+      "learning_rate": 3.181959764063308e-07,
+      "loss": 0.54,
+      "step": 54618
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.4805996417999268,
+      "learning_rate": 3.1810342858843047e-07,
+      "loss": 0.424,
+      "step": 54619
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 4.477192401885986,
+      "learning_rate": 3.180108937891302e-07,
+      "loss": 0.6621,
+      "step": 54620
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.3683457374572754,
+      "learning_rate": 3.179183720086887e-07,
+      "loss": 0.6736,
+      "step": 54621
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.860039710998535,
+      "learning_rate": 3.178258632473613e-07,
+      "loss": 0.4584,
+      "step": 54622
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.7375519275665283,
+      "learning_rate": 3.177333675054073e-07,
+      "loss": 0.7358,
+      "step": 54623
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.683544397354126,
+      "learning_rate": 3.1764088478308144e-07,
+      "loss": 0.3835,
+      "step": 54624
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.3946659564971924,
+      "learning_rate": 3.1754841508064347e-07,
+      "loss": 0.4601,
+      "step": 54625
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.177640438079834,
+      "learning_rate": 3.174559583983483e-07,
+      "loss": 0.5757,
+      "step": 54626
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.7848289012908936,
+      "learning_rate": 3.173635147364551e-07,
+      "loss": 0.4845,
+      "step": 54627
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.1733829975128174,
+      "learning_rate": 3.172710840952187e-07,
+      "loss": 0.628,
+      "step": 54628
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.010716438293457,
+      "learning_rate": 3.1717866647489825e-07,
+      "loss": 0.5548,
+      "step": 54629
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.98467755317688,
+      "learning_rate": 3.1708626187574867e-07,
+      "loss": 0.3947,
+      "step": 54630
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 4.059182167053223,
+      "learning_rate": 3.1699387029802906e-07,
+      "loss": 0.6386,
+      "step": 54631
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.1963648796081543,
+      "learning_rate": 3.169014917419938e-07,
+      "loss": 0.3544,
+      "step": 54632
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.797678232192993,
+      "learning_rate": 3.168091262079026e-07,
+      "loss": 0.5367,
+      "step": 54633
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.1538100242614746,
+      "learning_rate": 3.167167736960097e-07,
+      "loss": 0.6595,
+      "step": 54634
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 8.452201843261719,
+      "learning_rate": 3.1662443420657384e-07,
+      "loss": 0.6066,
+      "step": 54635
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 4.389010429382324,
+      "learning_rate": 3.165321077398503e-07,
+      "loss": 0.5634,
+      "step": 54636
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.022167205810547,
+      "learning_rate": 3.164397942960973e-07,
+      "loss": 0.5071,
+      "step": 54637
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.742002010345459,
+      "learning_rate": 3.163474938755701e-07,
+      "loss": 0.679,
+      "step": 54638
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.5966899394989014,
+      "learning_rate": 3.162552064785257e-07,
+      "loss": 0.6899,
+      "step": 54639
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.1718568801879883,
+      "learning_rate": 3.161629321052212e-07,
+      "loss": 0.4577,
+      "step": 54640
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.9005885124206543,
+      "learning_rate": 3.160706707559136e-07,
+      "loss": 0.6037,
+      "step": 54641
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.525587320327759,
+      "learning_rate": 3.159784224308582e-07,
+      "loss": 0.7459,
+      "step": 54642
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.0736403465270996,
+      "learning_rate": 3.1588618713031315e-07,
+      "loss": 0.5694,
+      "step": 54643
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.3230063915252686,
+      "learning_rate": 3.157939648545327e-07,
+      "loss": 0.5067,
+      "step": 54644
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.4312057495117188,
+      "learning_rate": 3.157017556037756e-07,
+      "loss": 0.6075,
+      "step": 54645
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.8442533016204834,
+      "learning_rate": 3.1560955937829597e-07,
+      "loss": 0.4393,
+      "step": 54646
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.624584674835205,
+      "learning_rate": 3.1551737617835197e-07,
+      "loss": 0.5992,
+      "step": 54647
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.151238203048706,
+      "learning_rate": 3.15425206004199e-07,
+      "loss": 0.5963,
+      "step": 54648
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.041530132293701,
+      "learning_rate": 3.153330488560946e-07,
+      "loss": 0.6325,
+      "step": 54649
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.227276563644409,
+      "learning_rate": 3.1524090473429303e-07,
+      "loss": 0.5891,
+      "step": 54650
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.7604260444641113,
+      "learning_rate": 3.1514877363905185e-07,
+      "loss": 0.5506,
+      "step": 54651
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.9658007621765137,
+      "learning_rate": 3.1505665557062694e-07,
+      "loss": 0.7652,
+      "step": 54652
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.838927745819092,
+      "learning_rate": 3.149645505292742e-07,
+      "loss": 0.6653,
+      "step": 54653
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.9794952869415283,
+      "learning_rate": 3.148724585152507e-07,
+      "loss": 0.5748,
+      "step": 54654
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.809579372406006,
+      "learning_rate": 3.147803795288107e-07,
+      "loss": 0.5481,
+      "step": 54655
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.256533145904541,
+      "learning_rate": 3.146883135702128e-07,
+      "loss": 0.5694,
+      "step": 54656
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.4014170169830322,
+      "learning_rate": 3.1459626063971016e-07,
+      "loss": 0.4269,
+      "step": 54657
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.017033338546753,
+      "learning_rate": 3.1450422073756095e-07,
+      "loss": 0.5674,
+      "step": 54658
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.704570770263672,
+      "learning_rate": 3.144121938640199e-07,
+      "loss": 0.637,
+      "step": 54659
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.313305616378784,
+      "learning_rate": 3.1432018001934294e-07,
+      "loss": 0.5341,
+      "step": 54660
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.3096113204956055,
+      "learning_rate": 3.142281792037871e-07,
+      "loss": 0.4919,
+      "step": 54661
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.1864588260650635,
+      "learning_rate": 3.1413619141760665e-07,
+      "loss": 0.4115,
+      "step": 54662
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.525097131729126,
+      "learning_rate": 3.14044216661058e-07,
+      "loss": 0.6438,
+      "step": 54663
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.1583456993103027,
+      "learning_rate": 3.139522549343982e-07,
+      "loss": 0.7966,
+      "step": 54664
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.029369831085205,
+      "learning_rate": 3.1386030623788034e-07,
+      "loss": 0.7211,
+      "step": 54665
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.7910733222961426,
+      "learning_rate": 3.137683705717626e-07,
+      "loss": 0.5909,
+      "step": 54666
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.3311655521392822,
+      "learning_rate": 3.136764479362986e-07,
+      "loss": 0.4225,
+      "step": 54667
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.3229150772094727,
+      "learning_rate": 3.1358453833174594e-07,
+      "loss": 0.436,
+      "step": 54668
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.7377867698669434,
+      "learning_rate": 3.134926417583578e-07,
+      "loss": 0.5541,
+      "step": 54669
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.7360658645629883,
+      "learning_rate": 3.134007582163923e-07,
+      "loss": 0.6059,
+      "step": 54670
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 16.084491729736328,
+      "learning_rate": 3.13308887706103e-07,
+      "loss": 0.7689,
+      "step": 54671
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.4247243404388428,
+      "learning_rate": 3.1321703022774595e-07,
+      "loss": 0.4576,
+      "step": 54672
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.67126727104187,
+      "learning_rate": 3.1312518578157647e-07,
+      "loss": 0.4183,
+      "step": 54673
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.2165491580963135,
+      "learning_rate": 3.13033354367851e-07,
+      "loss": 0.4767,
+      "step": 54674
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.089081287384033,
+      "learning_rate": 3.129415359868232e-07,
+      "loss": 0.7282,
+      "step": 54675
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.892840623855591,
+      "learning_rate": 3.128497306387496e-07,
+      "loss": 0.368,
+      "step": 54676
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.152343988418579,
+      "learning_rate": 3.1275793832388437e-07,
+      "loss": 0.5958,
+      "step": 54677
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.6876399517059326,
+      "learning_rate": 3.126661590424845e-07,
+      "loss": 0.6615,
+      "step": 54678
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 4.502122402191162,
+      "learning_rate": 3.125743927948027e-07,
+      "loss": 0.5356,
+      "step": 54679
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 5.443615913391113,
+      "learning_rate": 3.1248263958109647e-07,
+      "loss": 0.973,
+      "step": 54680
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.1093497276306152,
+      "learning_rate": 3.123908994016195e-07,
+      "loss": 0.5869,
+      "step": 54681
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.2644948959350586,
+      "learning_rate": 3.122991722566282e-07,
+      "loss": 0.7592,
+      "step": 54682
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.4918274879455566,
+      "learning_rate": 3.1220745814637577e-07,
+      "loss": 0.622,
+      "step": 54683
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.0671839714050293,
+      "learning_rate": 3.121157570711197e-07,
+      "loss": 0.4538,
+      "step": 54684
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 4.41851806640625,
+      "learning_rate": 3.120240690311127e-07,
+      "loss": 0.5497,
+      "step": 54685
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.920353412628174,
+      "learning_rate": 3.1193239402661114e-07,
+      "loss": 0.4963,
+      "step": 54686
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 4.4449028968811035,
+      "learning_rate": 3.118407320578687e-07,
+      "loss": 0.2237,
+      "step": 54687
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.0912911891937256,
+      "learning_rate": 3.1174908312514184e-07,
+      "loss": 0.6418,
+      "step": 54688
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.091779947280884,
+      "learning_rate": 3.1165744722868373e-07,
+      "loss": 0.5576,
+      "step": 54689
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 4.304824352264404,
+      "learning_rate": 3.1156582436875026e-07,
+      "loss": 0.595,
+      "step": 54690
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.2491376399993896,
+      "learning_rate": 3.114742145455957e-07,
+      "loss": 0.3503,
+      "step": 54691
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.082162857055664,
+      "learning_rate": 3.113826177594759e-07,
+      "loss": 0.6447,
+      "step": 54692
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.523514986038208,
+      "learning_rate": 3.112910340106434e-07,
+      "loss": 0.5546,
+      "step": 54693
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.912763833999634,
+      "learning_rate": 3.111994632993548e-07,
+      "loss": 0.3024,
+      "step": 54694
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.4466233253479004,
+      "learning_rate": 3.111079056258637e-07,
+      "loss": 0.5445,
+      "step": 54695
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.4007558822631836,
+      "learning_rate": 3.1101636099042543e-07,
+      "loss": 0.5517,
+      "step": 54696
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 4.026549339294434,
+      "learning_rate": 3.1092482939329373e-07,
+      "loss": 0.7145,
+      "step": 54697
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.810279369354248,
+      "learning_rate": 3.108333108347239e-07,
+      "loss": 0.9038,
+      "step": 54698
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.591468095779419,
+      "learning_rate": 3.1074180531496965e-07,
+      "loss": 0.4777,
+      "step": 54699
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.303176164627075,
+      "learning_rate": 3.1065031283428527e-07,
+      "loss": 0.6043,
+      "step": 54700
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.835308790206909,
+      "learning_rate": 3.1055883339292604e-07,
+      "loss": 0.717,
+      "step": 54701
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.533970355987549,
+      "learning_rate": 3.104673669911468e-07,
+      "loss": 0.5532,
+      "step": 54702
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.522425651550293,
+      "learning_rate": 3.103759136292006e-07,
+      "loss": 0.4421,
+      "step": 54703
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 4.519903659820557,
+      "learning_rate": 3.1028447330734233e-07,
+      "loss": 0.9711,
+      "step": 54704
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.7309939861297607,
+      "learning_rate": 3.1019304602582565e-07,
+      "loss": 0.6287,
+      "step": 54705
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 4.8355207443237305,
+      "learning_rate": 3.1010163178490593e-07,
+      "loss": 0.4923,
+      "step": 54706
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.625516414642334,
+      "learning_rate": 3.100102305848363e-07,
+      "loss": 0.5193,
+      "step": 54707
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.061113119125366,
+      "learning_rate": 3.0991884242587143e-07,
+      "loss": 0.5428,
+      "step": 54708
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.8256430625915527,
+      "learning_rate": 3.098274673082652e-07,
+      "loss": 0.4602,
+      "step": 54709
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.6312074661254883,
+      "learning_rate": 3.097361052322723e-07,
+      "loss": 0.6288,
+      "step": 54710
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.244298219680786,
+      "learning_rate": 3.0964475619814526e-07,
+      "loss": 0.2923,
+      "step": 54711
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.114400625228882,
+      "learning_rate": 3.0955342020614064e-07,
+      "loss": 0.674,
+      "step": 54712
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.6642184257507324,
+      "learning_rate": 3.094620972565093e-07,
+      "loss": 0.501,
+      "step": 54713
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.4911246299743652,
+      "learning_rate": 3.093707873495072e-07,
+      "loss": 0.7508,
+      "step": 54714
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.2912039756774902,
+      "learning_rate": 3.092794904853885e-07,
+      "loss": 0.5427,
+      "step": 54715
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.9464282989501953,
+      "learning_rate": 3.091882066644053e-07,
+      "loss": 0.634,
+      "step": 54716
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.609691858291626,
+      "learning_rate": 3.0909693588681346e-07,
+      "loss": 0.4553,
+      "step": 54717
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.625788450241089,
+      "learning_rate": 3.0900567815286497e-07,
+      "loss": 0.4995,
+      "step": 54718
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.5535244941711426,
+      "learning_rate": 3.089144334628152e-07,
+      "loss": 0.8105,
+      "step": 54719
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.7483301162719727,
+      "learning_rate": 3.0882320181691627e-07,
+      "loss": 0.593,
+      "step": 54720
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.4472899436950684,
+      "learning_rate": 3.0873198321542285e-07,
+      "loss": 0.4703,
+      "step": 54721
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.6457736492156982,
+      "learning_rate": 3.086407776585887e-07,
+      "loss": 0.631,
+      "step": 54722
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.164273500442505,
+      "learning_rate": 3.085495851466669e-07,
+      "loss": 0.5638,
+      "step": 54723
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.832170009613037,
+      "learning_rate": 3.084584056799117e-07,
+      "loss": 0.4938,
+      "step": 54724
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.9971487522125244,
+      "learning_rate": 3.083672392585757e-07,
+      "loss": 0.5546,
+      "step": 54725
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.750910520553589,
+      "learning_rate": 3.0827608588291257e-07,
+      "loss": 0.5423,
+      "step": 54726
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 4.9508538246154785,
+      "learning_rate": 3.0818494555317656e-07,
+      "loss": 0.6899,
+      "step": 54727
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.0410053730010986,
+      "learning_rate": 3.0809381826962026e-07,
+      "loss": 0.339,
+      "step": 54728
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.183032989501953,
+      "learning_rate": 3.0800270403249845e-07,
+      "loss": 0.3727,
+      "step": 54729
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 4.904811382293701,
+      "learning_rate": 3.0791160284206255e-07,
+      "loss": 0.7309,
+      "step": 54730
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.180697441101074,
+      "learning_rate": 3.078205146985674e-07,
+      "loss": 0.8644,
+      "step": 54731
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.807684898376465,
+      "learning_rate": 3.0772943960226456e-07,
+      "loss": 0.6707,
+      "step": 54732
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.7903099060058594,
+      "learning_rate": 3.0763837755340973e-07,
+      "loss": 0.6176,
+      "step": 54733
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.5236940383911133,
+      "learning_rate": 3.07547328552254e-07,
+      "loss": 0.5459,
+      "step": 54734
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.998215675354004,
+      "learning_rate": 3.0745629259905153e-07,
+      "loss": 0.6354,
+      "step": 54735
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.3402762413024902,
+      "learning_rate": 3.073652696940549e-07,
+      "loss": 0.6633,
+      "step": 54736
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.1954197883605957,
+      "learning_rate": 3.0727425983751835e-07,
+      "loss": 0.539,
+      "step": 54737
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.3171987533569336,
+      "learning_rate": 3.071832630296934e-07,
+      "loss": 0.6597,
+      "step": 54738
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.105468273162842,
+      "learning_rate": 3.0709227927083417e-07,
+      "loss": 0.5323,
+      "step": 54739
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.782275438308716,
+      "learning_rate": 3.070013085611928e-07,
+      "loss": 0.7574,
+      "step": 54740
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.865891218185425,
+      "learning_rate": 3.0691035090102297e-07,
+      "loss": 0.5396,
+      "step": 54741
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.6713528633117676,
+      "learning_rate": 3.0681940629057715e-07,
+      "loss": 0.6489,
+      "step": 54742
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.5888257026672363,
+      "learning_rate": 3.067284747301091e-07,
+      "loss": 0.4974,
+      "step": 54743
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.0668084621429443,
+      "learning_rate": 3.066375562198709e-07,
+      "loss": 0.4024,
+      "step": 54744
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.5376737117767334,
+      "learning_rate": 3.065466507601156e-07,
+      "loss": 0.3265,
+      "step": 54745
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.7331769466400146,
+      "learning_rate": 3.064557583510952e-07,
+      "loss": 0.5844,
+      "step": 54746
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.665130853652954,
+      "learning_rate": 3.06364878993064e-07,
+      "loss": 0.6149,
+      "step": 54747
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.2315616607666016,
+      "learning_rate": 3.0627401268627287e-07,
+      "loss": 0.5433,
+      "step": 54748
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.453801393508911,
+      "learning_rate": 3.061831594309761e-07,
+      "loss": 0.7429,
+      "step": 54749
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.6541638374328613,
+      "learning_rate": 3.060923192274251e-07,
+      "loss": 0.6105,
+      "step": 54750
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.078882932662964,
+      "learning_rate": 3.060014920758736e-07,
+      "loss": 0.3467,
+      "step": 54751
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.4473519325256348,
+      "learning_rate": 3.05910677976573e-07,
+      "loss": 0.5613,
+      "step": 54752
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.8346686363220215,
+      "learning_rate": 3.058198769297771e-07,
+      "loss": 0.8412,
+      "step": 54753
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.1904537677764893,
+      "learning_rate": 3.057290889357367e-07,
+      "loss": 0.4598,
+      "step": 54754
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.4792280197143555,
+      "learning_rate": 3.0563831399470557e-07,
+      "loss": 0.2507,
+      "step": 54755
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.934011936187744,
+      "learning_rate": 3.0554755210693563e-07,
+      "loss": 0.5791,
+      "step": 54756
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.0546810626983643,
+      "learning_rate": 3.054568032726801e-07,
+      "loss": 0.4815,
+      "step": 54757
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.348090410232544,
+      "learning_rate": 3.053660674921893e-07,
+      "loss": 0.4798,
+      "step": 54758
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.538480043411255,
+      "learning_rate": 3.052753447657181e-07,
+      "loss": 0.5484,
+      "step": 54759
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.7538349628448486,
+      "learning_rate": 3.0518463509351616e-07,
+      "loss": 0.325,
+      "step": 54760
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.6695289611816406,
+      "learning_rate": 3.050939384758378e-07,
+      "loss": 0.5545,
+      "step": 54761
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 9.282108306884766,
+      "learning_rate": 3.050032549129339e-07,
+      "loss": 0.6068,
+      "step": 54762
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.76692533493042,
+      "learning_rate": 3.049125844050582e-07,
+      "loss": 0.4666,
+      "step": 54763
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.766073226928711,
+      "learning_rate": 3.048219269524605e-07,
+      "loss": 0.4944,
+      "step": 54764
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.419917583465576,
+      "learning_rate": 3.0473128255539553e-07,
+      "loss": 0.4677,
+      "step": 54765
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.9907801151275635,
+      "learning_rate": 3.0464065121411313e-07,
+      "loss": 0.8278,
+      "step": 54766
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.200049877166748,
+      "learning_rate": 3.045500329288664e-07,
+      "loss": 0.7644,
+      "step": 54767
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.1216790676116943,
+      "learning_rate": 3.0445942769990687e-07,
+      "loss": 0.4101,
+      "step": 54768
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.321338653564453,
+      "learning_rate": 3.043688355274871e-07,
+      "loss": 0.6849,
+      "step": 54769
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 5.0371479988098145,
+      "learning_rate": 3.042782564118579e-07,
+      "loss": 0.6527,
+      "step": 54770
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.478351593017578,
+      "learning_rate": 3.041876903532731e-07,
+      "loss": 0.531,
+      "step": 54771
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.7856361865997314,
+      "learning_rate": 3.0409713735198186e-07,
+      "loss": 0.5361,
+      "step": 54772
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.778615951538086,
+      "learning_rate": 3.0400659740823846e-07,
+      "loss": 0.8306,
+      "step": 54773
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.6097609996795654,
+      "learning_rate": 3.0391607052229213e-07,
+      "loss": 0.6677,
+      "step": 54774
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.8719840049743652,
+      "learning_rate": 3.038255566943965e-07,
+      "loss": 0.64,
+      "step": 54775
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.2611796855926514,
+      "learning_rate": 3.037350559248037e-07,
+      "loss": 0.6831,
+      "step": 54776
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.2260169982910156,
+      "learning_rate": 3.036445682137634e-07,
+      "loss": 0.5971,
+      "step": 54777
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 4.594561576843262,
+      "learning_rate": 3.0355409356152944e-07,
+      "loss": 0.5259,
+      "step": 54778
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.9535207748413086,
+      "learning_rate": 3.034636319683509e-07,
+      "loss": 0.4903,
+      "step": 54779
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.4192633628845215,
+      "learning_rate": 3.0337318343448163e-07,
+      "loss": 0.3206,
+      "step": 54780
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.4516561031341553,
+      "learning_rate": 3.032827479601713e-07,
+      "loss": 0.3612,
+      "step": 54781
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.7939858436584473,
+      "learning_rate": 3.0319232554567257e-07,
+      "loss": 0.592,
+      "step": 54782
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.402837038040161,
+      "learning_rate": 3.031019161912369e-07,
+      "loss": 0.4066,
+      "step": 54783
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.7648885250091553,
+      "learning_rate": 3.0301151989711453e-07,
+      "loss": 0.6259,
+      "step": 54784
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.4845142364501953,
+      "learning_rate": 3.0292113666355874e-07,
+      "loss": 0.7053,
+      "step": 54785
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 4.078545570373535,
+      "learning_rate": 3.028307664908187e-07,
+      "loss": 0.539,
+      "step": 54786
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.9495816230773926,
+      "learning_rate": 3.0274040937914637e-07,
+      "loss": 0.5322,
+      "step": 54787
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.5773141384124756,
+      "learning_rate": 3.0265006532879447e-07,
+      "loss": 0.3453,
+      "step": 54788
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.247058629989624,
+      "learning_rate": 3.0255973434001275e-07,
+      "loss": 0.4146,
+      "step": 54789
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.109921932220459,
+      "learning_rate": 3.0246941641305263e-07,
+      "loss": 0.7506,
+      "step": 54790
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.2674386501312256,
+      "learning_rate": 3.0237911154816504e-07,
+      "loss": 0.7529,
+      "step": 54791
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.8986246585845947,
+      "learning_rate": 3.0228881974560196e-07,
+      "loss": 0.5913,
+      "step": 54792
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.450582504272461,
+      "learning_rate": 3.021985410056133e-07,
+      "loss": 0.5439,
+      "step": 54793
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.3056936264038086,
+      "learning_rate": 3.02108275328451e-07,
+      "loss": 0.2992,
+      "step": 54794
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.4722139835357666,
+      "learning_rate": 3.0201802271436543e-07,
+      "loss": 0.6269,
+      "step": 54795
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.5658934116363525,
+      "learning_rate": 3.01927783163608e-07,
+      "loss": 0.7861,
+      "step": 54796
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.342968225479126,
+      "learning_rate": 3.018375566764292e-07,
+      "loss": 0.6224,
+      "step": 54797
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.592867612838745,
+      "learning_rate": 3.0174734325308096e-07,
+      "loss": 0.8383,
+      "step": 54798
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 4.41869592666626,
+      "learning_rate": 3.0165714289381254e-07,
+      "loss": 0.5216,
+      "step": 54799
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 4.1870598793029785,
+      "learning_rate": 3.0156695559887594e-07,
+      "loss": 0.45,
+      "step": 54800
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.2942087650299072,
+      "learning_rate": 3.0147678136852153e-07,
+      "loss": 0.5928,
+      "step": 54801
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.165201187133789,
+      "learning_rate": 3.013866202029997e-07,
+      "loss": 0.6641,
+      "step": 54802
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.7333202362060547,
+      "learning_rate": 3.012964721025613e-07,
+      "loss": 0.5037,
+      "step": 54803
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.623464584350586,
+      "learning_rate": 3.012063370674584e-07,
+      "loss": 0.7457,
+      "step": 54804
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.3441381454467773,
+      "learning_rate": 3.011162150979391e-07,
+      "loss": 0.4544,
+      "step": 54805
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.989814519882202,
+      "learning_rate": 3.010261061942565e-07,
+      "loss": 0.5623,
+      "step": 54806
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.0825953483581543,
+      "learning_rate": 3.0093601035665945e-07,
+      "loss": 0.4726,
+      "step": 54807
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.5028884410858154,
+      "learning_rate": 3.0084592758539976e-07,
+      "loss": 0.4104,
+      "step": 54808
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.7010128498077393,
+      "learning_rate": 3.0075585788072623e-07,
+      "loss": 0.6202,
+      "step": 54809
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.6356892585754395,
+      "learning_rate": 3.006658012428915e-07,
+      "loss": 0.4655,
+      "step": 54810
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.2420973777770996,
+      "learning_rate": 3.0057575767214354e-07,
+      "loss": 0.583,
+      "step": 54811
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.119906425476074,
+      "learning_rate": 3.004857271687345e-07,
+      "loss": 0.4496,
+      "step": 54812
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 4.141378879547119,
+      "learning_rate": 3.003957097329141e-07,
+      "loss": 0.6545,
+      "step": 54813
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.816417694091797,
+      "learning_rate": 3.0030570536493277e-07,
+      "loss": 0.6023,
+      "step": 54814
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.9140501022338867,
+      "learning_rate": 3.002157140650408e-07,
+      "loss": 0.5456,
+      "step": 54815
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.591886281967163,
+      "learning_rate": 3.001257358334886e-07,
+      "loss": 0.416,
+      "step": 54816
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.2170677185058594,
+      "learning_rate": 3.0003577067052535e-07,
+      "loss": 0.5575,
+      "step": 54817
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.49245548248291,
+      "learning_rate": 2.999458185764026e-07,
+      "loss": 0.2393,
+      "step": 54818
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.9683303833007812,
+      "learning_rate": 2.998558795513695e-07,
+      "loss": 0.5272,
+      "step": 54819
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.5846424102783203,
+      "learning_rate": 2.9976595359567653e-07,
+      "loss": 0.6208,
+      "step": 54820
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 5.433877468109131,
+      "learning_rate": 2.996760407095739e-07,
+      "loss": 0.5377,
+      "step": 54821
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.0643014907836914,
+      "learning_rate": 2.9958614089331093e-07,
+      "loss": 0.6588,
+      "step": 54822
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.20572829246521,
+      "learning_rate": 2.99496254147138e-07,
+      "loss": 0.7103,
+      "step": 54823
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.708897829055786,
+      "learning_rate": 2.9940638047130597e-07,
+      "loss": 0.6906,
+      "step": 54824
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 5.551976203918457,
+      "learning_rate": 2.99316519866063e-07,
+      "loss": 0.6054,
+      "step": 54825
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 5.34713888168335,
+      "learning_rate": 2.992266723316606e-07,
+      "loss": 0.4346,
+      "step": 54826
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.432570457458496,
+      "learning_rate": 2.9913683786834734e-07,
+      "loss": 0.4945,
+      "step": 54827
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.1389005184173584,
+      "learning_rate": 2.9904701647637424e-07,
+      "loss": 0.6755,
+      "step": 54828
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.8579578399658203,
+      "learning_rate": 2.9895720815598997e-07,
+      "loss": 0.5915,
+      "step": 54829
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.5450668334960938,
+      "learning_rate": 2.9886741290744483e-07,
+      "loss": 0.6867,
+      "step": 54830
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.7465765476226807,
+      "learning_rate": 2.9877763073098807e-07,
+      "loss": 0.4865,
+      "step": 54831
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.1563239097595215,
+      "learning_rate": 2.9868786162687014e-07,
+      "loss": 0.5242,
+      "step": 54832
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.1748857498168945,
+      "learning_rate": 2.9859810559533906e-07,
+      "loss": 0.6345,
+      "step": 54833
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.460662841796875,
+      "learning_rate": 2.9850836263664694e-07,
+      "loss": 0.4446,
+      "step": 54834
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.9560325145721436,
+      "learning_rate": 2.984186327510402e-07,
+      "loss": 0.5648,
+      "step": 54835
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 6.319894790649414,
+      "learning_rate": 2.9832891593877147e-07,
+      "loss": 0.6131,
+      "step": 54836
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.4890193939208984,
+      "learning_rate": 2.9823921220008766e-07,
+      "loss": 0.8966,
+      "step": 54837
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.573598623275757,
+      "learning_rate": 2.9814952153523924e-07,
+      "loss": 0.3252,
+      "step": 54838
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.7636702060699463,
+      "learning_rate": 2.980598439444765e-07,
+      "loss": 0.7013,
+      "step": 54839
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 4.01878547668457,
+      "learning_rate": 2.9797017942804707e-07,
+      "loss": 0.7637,
+      "step": 54840
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.5425469875335693,
+      "learning_rate": 2.978805279862024e-07,
+      "loss": 0.4016,
+      "step": 54841
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.2091867923736572,
+      "learning_rate": 2.9779088961918947e-07,
+      "loss": 0.5906,
+      "step": 54842
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.989575147628784,
+      "learning_rate": 2.977012643272581e-07,
+      "loss": 0.6636,
+      "step": 54843
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.41007661819458,
+      "learning_rate": 2.9761165211065923e-07,
+      "loss": 0.3415,
+      "step": 54844
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.1989381313323975,
+      "learning_rate": 2.9752205296963987e-07,
+      "loss": 0.5011,
+      "step": 54845
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.8731465339660645,
+      "learning_rate": 2.974324669044509e-07,
+      "loss": 0.797,
+      "step": 54846
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 5.298654079437256,
+      "learning_rate": 2.9734289391534e-07,
+      "loss": 0.3713,
+      "step": 54847
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.9200026988983154,
+      "learning_rate": 2.9725333400255684e-07,
+      "loss": 0.7421,
+      "step": 54848
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.167996406555176,
+      "learning_rate": 2.971637871663513e-07,
+      "loss": 0.4921,
+      "step": 54849
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.0945615768432617,
+      "learning_rate": 2.970742534069704e-07,
+      "loss": 0.4799,
+      "step": 54850
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.76918625831604,
+      "learning_rate": 2.9698473272466555e-07,
+      "loss": 0.5706,
+      "step": 54851
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.5742743015289307,
+      "learning_rate": 2.968952251196833e-07,
+      "loss": 0.4497,
+      "step": 54852
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.441239356994629,
+      "learning_rate": 2.968057305922745e-07,
+      "loss": 0.7673,
+      "step": 54853
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.9233999252319336,
+      "learning_rate": 2.9671624914268617e-07,
+      "loss": 0.3885,
+      "step": 54854
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 9.203202247619629,
+      "learning_rate": 2.966267807711687e-07,
+      "loss": 0.459,
+      "step": 54855
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 4.51809024810791,
+      "learning_rate": 2.9653732547796965e-07,
+      "loss": 0.7596,
+      "step": 54856
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 1.72620689868927,
+      "learning_rate": 2.964478832633388e-07,
+      "loss": 0.3372,
+      "step": 54857
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.8946597576141357,
+      "learning_rate": 2.963584541275244e-07,
+      "loss": 0.2574,
+      "step": 54858
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 4.807754039764404,
+      "learning_rate": 2.96269038070775e-07,
+      "loss": 0.5593,
+      "step": 54859
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.237278699874878,
+      "learning_rate": 2.9617963509333937e-07,
+      "loss": 0.7658,
+      "step": 54860
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.3325295448303223,
+      "learning_rate": 2.9609024519546613e-07,
+      "loss": 0.5378,
+      "step": 54861
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.4413042068481445,
+      "learning_rate": 2.960008683774035e-07,
+      "loss": 0.6202,
+      "step": 54862
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 4.291924953460693,
+      "learning_rate": 2.959115046394001e-07,
+      "loss": 0.5613,
+      "step": 54863
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.584665060043335,
+      "learning_rate": 2.9582215398170466e-07,
+      "loss": 0.6465,
+      "step": 54864
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.6449315547943115,
+      "learning_rate": 2.957328164045664e-07,
+      "loss": 0.755,
+      "step": 54865
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 4.219904899597168,
+      "learning_rate": 2.956434919082324e-07,
+      "loss": 0.6516,
+      "step": 54866
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.1422502994537354,
+      "learning_rate": 2.9555418049295183e-07,
+      "loss": 0.606,
+      "step": 54867
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.6617698669433594,
+      "learning_rate": 2.954648821589723e-07,
+      "loss": 0.7443,
+      "step": 54868
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.476527690887451,
+      "learning_rate": 2.9537559690654305e-07,
+      "loss": 0.5057,
+      "step": 54869
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.4078638553619385,
+      "learning_rate": 2.952863247359111e-07,
+      "loss": 0.5718,
+      "step": 54870
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.313481330871582,
+      "learning_rate": 2.9519706564732687e-07,
+      "loss": 0.7137,
+      "step": 54871
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.0170273780822754,
+      "learning_rate": 2.951078196410356e-07,
+      "loss": 0.4727,
+      "step": 54872
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.6695003509521484,
+      "learning_rate": 2.950185867172883e-07,
+      "loss": 0.4172,
+      "step": 54873
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.347886085510254,
+      "learning_rate": 2.949293668763309e-07,
+      "loss": 0.5001,
+      "step": 54874
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.2234842777252197,
+      "learning_rate": 2.9484016011841257e-07,
+      "loss": 0.4036,
+      "step": 54875
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.6348891258239746,
+      "learning_rate": 2.9475096644378086e-07,
+      "loss": 0.5653,
+      "step": 54876
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.1542088985443115,
+      "learning_rate": 2.9466178585268513e-07,
+      "loss": 0.4879,
+      "step": 54877
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.429105520248413,
+      "learning_rate": 2.945726183453712e-07,
+      "loss": 0.6851,
+      "step": 54878
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.35378098487854,
+      "learning_rate": 2.9448346392208895e-07,
+      "loss": 0.6399,
+      "step": 54879
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 5.347625255584717,
+      "learning_rate": 2.9439432258308476e-07,
+      "loss": 0.4914,
+      "step": 54880
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.579474449157715,
+      "learning_rate": 2.9430519432860794e-07,
+      "loss": 0.4182,
+      "step": 54881
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.2382359504699707,
+      "learning_rate": 2.9421607915890493e-07,
+      "loss": 0.446,
+      "step": 54882
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 4.143705368041992,
+      "learning_rate": 2.941269770742239e-07,
+      "loss": 0.4506,
+      "step": 54883
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 4.496916770935059,
+      "learning_rate": 2.9403788807481293e-07,
+      "loss": 0.6673,
+      "step": 54884
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.0772693157196045,
+      "learning_rate": 2.9394881216092074e-07,
+      "loss": 0.4807,
+      "step": 54885
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.12833571434021,
+      "learning_rate": 2.9385974933279274e-07,
+      "loss": 0.4817,
+      "step": 54886
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.830637216567993,
+      "learning_rate": 2.937706995906792e-07,
+      "loss": 0.8251,
+      "step": 54887
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.9437618255615234,
+      "learning_rate": 2.936816629348249e-07,
+      "loss": 0.5049,
+      "step": 54888
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.227217435836792,
+      "learning_rate": 2.9359263936547975e-07,
+      "loss": 0.2163,
+      "step": 54889
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.4203546047210693,
+      "learning_rate": 2.935036288828902e-07,
+      "loss": 0.5338,
+      "step": 54890
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.848717451095581,
+      "learning_rate": 2.934146314873043e-07,
+      "loss": 0.7145,
+      "step": 54891
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.9254209995269775,
+      "learning_rate": 2.933256471789686e-07,
+      "loss": 0.6104,
+      "step": 54892
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.032133102416992,
+      "learning_rate": 2.9323667595813175e-07,
+      "loss": 0.6415,
+      "step": 54893
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.648385524749756,
+      "learning_rate": 2.931477178250397e-07,
+      "loss": 0.8136,
+      "step": 54894
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.800729513168335,
+      "learning_rate": 2.9305877277994164e-07,
+      "loss": 0.4786,
+      "step": 54895
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.7281723022460938,
+      "learning_rate": 2.92969840823083e-07,
+      "loss": 0.6126,
+      "step": 54896
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.2189853191375732,
+      "learning_rate": 2.928809219547124e-07,
+      "loss": 0.4177,
+      "step": 54897
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 4.309010028839111,
+      "learning_rate": 2.9279201617507633e-07,
+      "loss": 0.7199,
+      "step": 54898
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.8665997982025146,
+      "learning_rate": 2.9270312348442187e-07,
+      "loss": 0.4352,
+      "step": 54899
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.560709238052368,
+      "learning_rate": 2.9261424388299766e-07,
+      "loss": 0.6157,
+      "step": 54900
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.7818355560302734,
+      "learning_rate": 2.9252537737104903e-07,
+      "loss": 0.4288,
+      "step": 54901
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.8413586616516113,
+      "learning_rate": 2.9243652394882474e-07,
+      "loss": 0.6389,
+      "step": 54902
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.783424139022827,
+      "learning_rate": 2.9234768361657006e-07,
+      "loss": 0.6202,
+      "step": 54903
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 4.017490863800049,
+      "learning_rate": 2.922588563745332e-07,
+      "loss": 0.6711,
+      "step": 54904
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 1.9738205671310425,
+      "learning_rate": 2.921700422229612e-07,
+      "loss": 0.4213,
+      "step": 54905
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.4479219913482666,
+      "learning_rate": 2.920812411620999e-07,
+      "loss": 0.6371,
+      "step": 54906
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.7035884857177734,
+      "learning_rate": 2.919924531921986e-07,
+      "loss": 0.6827,
+      "step": 54907
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.220194101333618,
+      "learning_rate": 2.91903678313501e-07,
+      "loss": 0.4854,
+      "step": 54908
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.725602149963379,
+      "learning_rate": 2.918149165262568e-07,
+      "loss": 0.5432,
+      "step": 54909
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 4.223390579223633,
+      "learning_rate": 2.9172616783071094e-07,
+      "loss": 0.5384,
+      "step": 54910
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.3240013122558594,
+      "learning_rate": 2.916374322271104e-07,
+      "loss": 0.7546,
+      "step": 54911
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.8566904067993164,
+      "learning_rate": 2.9154870971570326e-07,
+      "loss": 0.5916,
+      "step": 54912
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 4.110382556915283,
+      "learning_rate": 2.9146000029673496e-07,
+      "loss": 0.6307,
+      "step": 54913
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.2327041625976562,
+      "learning_rate": 2.91371303970453e-07,
+      "loss": 0.6399,
+      "step": 54914
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.300705909729004,
+      "learning_rate": 2.9128262073710276e-07,
+      "loss": 0.5969,
+      "step": 54915
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 1.8723849058151245,
+      "learning_rate": 2.911939505969324e-07,
+      "loss": 0.4088,
+      "step": 54916
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.6571362018585205,
+      "learning_rate": 2.911052935501868e-07,
+      "loss": 0.5725,
+      "step": 54917
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.7270772457122803,
+      "learning_rate": 2.910166495971145e-07,
+      "loss": 0.5718,
+      "step": 54918
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.5897624492645264,
+      "learning_rate": 2.9092801873795986e-07,
+      "loss": 0.3485,
+      "step": 54919
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 5.423129081726074,
+      "learning_rate": 2.9083940097297094e-07,
+      "loss": 0.4494,
+      "step": 54920
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.84732985496521,
+      "learning_rate": 2.907507963023931e-07,
+      "loss": 0.759,
+      "step": 54921
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.780317544937134,
+      "learning_rate": 2.90662204726474e-07,
+      "loss": 0.57,
+      "step": 54922
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.7621641159057617,
+      "learning_rate": 2.9057362624545835e-07,
+      "loss": 0.5675,
+      "step": 54923
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.8899924755096436,
+      "learning_rate": 2.9048506085959273e-07,
+      "loss": 0.6236,
+      "step": 54924
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.396088123321533,
+      "learning_rate": 2.903965085691246e-07,
+      "loss": 0.7732,
+      "step": 54925
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 4.380699157714844,
+      "learning_rate": 2.9030796937429994e-07,
+      "loss": 0.5862,
+      "step": 54926
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.899770498275757,
+      "learning_rate": 2.902194432753641e-07,
+      "loss": 0.7444,
+      "step": 54927
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 4.298923969268799,
+      "learning_rate": 2.9013093027256414e-07,
+      "loss": 0.763,
+      "step": 54928
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.515150308609009,
+      "learning_rate": 2.9004243036614476e-07,
+      "loss": 0.7419,
+      "step": 54929
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.8156392574310303,
+      "learning_rate": 2.899539435563542e-07,
+      "loss": 0.9393,
+      "step": 54930
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 4.274327278137207,
+      "learning_rate": 2.898654698434361e-07,
+      "loss": 0.6327,
+      "step": 54931
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.2612171173095703,
+      "learning_rate": 2.897770092276386e-07,
+      "loss": 0.4845,
+      "step": 54932
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.3563904762268066,
+      "learning_rate": 2.89688561709206e-07,
+      "loss": 0.3898,
+      "step": 54933
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.5157275199890137,
+      "learning_rate": 2.8960012728838585e-07,
+      "loss": 0.6051,
+      "step": 54934
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.0930709838867188,
+      "learning_rate": 2.895117059654223e-07,
+      "loss": 0.4819,
+      "step": 54935
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.9152674674987793,
+      "learning_rate": 2.894232977405631e-07,
+      "loss": 0.4866,
+      "step": 54936
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.0024845600128174,
+      "learning_rate": 2.893349026140524e-07,
+      "loss": 0.4326,
+      "step": 54937
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.8818345069885254,
+      "learning_rate": 2.892465205861372e-07,
+      "loss": 0.5918,
+      "step": 54938
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.837757110595703,
+      "learning_rate": 2.8915815165706187e-07,
+      "loss": 0.5102,
+      "step": 54939
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.300485134124756,
+      "learning_rate": 2.8906979582707386e-07,
+      "loss": 0.379,
+      "step": 54940
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 4.797301769256592,
+      "learning_rate": 2.889814530964169e-07,
+      "loss": 0.4938,
+      "step": 54941
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.0635673999786377,
+      "learning_rate": 2.8889312346533914e-07,
+      "loss": 0.6725,
+      "step": 54942
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.488651990890503,
+      "learning_rate": 2.8880480693408375e-07,
+      "loss": 0.5694,
+      "step": 54943
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.395939588546753,
+      "learning_rate": 2.8871650350289715e-07,
+      "loss": 0.8547,
+      "step": 54944
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.740299701690674,
+      "learning_rate": 2.886282131720253e-07,
+      "loss": 0.4258,
+      "step": 54945
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.967294931411743,
+      "learning_rate": 2.88539935941714e-07,
+      "loss": 0.6078,
+      "step": 54946
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.1353261470794678,
+      "learning_rate": 2.8845167181220766e-07,
+      "loss": 0.7157,
+      "step": 54947
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 5.077091217041016,
+      "learning_rate": 2.883634207837527e-07,
+      "loss": 0.4575,
+      "step": 54948
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.8315799236297607,
+      "learning_rate": 2.8827518285659386e-07,
+      "loss": 0.5513,
+      "step": 54949
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.639660596847534,
+      "learning_rate": 2.8818695803097707e-07,
+      "loss": 0.6499,
+      "step": 54950
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.887857437133789,
+      "learning_rate": 2.8809874630714605e-07,
+      "loss": 0.4633,
+      "step": 54951
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.419712543487549,
+      "learning_rate": 2.88010547685349e-07,
+      "loss": 0.5464,
+      "step": 54952
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.003605365753174,
+      "learning_rate": 2.879223621658278e-07,
+      "loss": 0.7357,
+      "step": 54953
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.557096242904663,
+      "learning_rate": 2.878341897488307e-07,
+      "loss": 0.4924,
+      "step": 54954
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.870950698852539,
+      "learning_rate": 2.877460304346008e-07,
+      "loss": 0.7953,
+      "step": 54955
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 4.791754245758057,
+      "learning_rate": 2.8765788422338405e-07,
+      "loss": 0.6982,
+      "step": 54956
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.8121981620788574,
+      "learning_rate": 2.875697511154252e-07,
+      "loss": 0.7052,
+      "step": 54957
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.333003520965576,
+      "learning_rate": 2.874816311109702e-07,
+      "loss": 0.605,
+      "step": 54958
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.01663875579834,
+      "learning_rate": 2.873935242102627e-07,
+      "loss": 0.5613,
+      "step": 54959
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.920153856277466,
+      "learning_rate": 2.8730543041354863e-07,
+      "loss": 0.5941,
+      "step": 54960
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.378868579864502,
+      "learning_rate": 2.8721734972107287e-07,
+      "loss": 0.6685,
+      "step": 54961
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.9379355907440186,
+      "learning_rate": 2.871292821330801e-07,
+      "loss": 0.445,
+      "step": 54962
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.5487070083618164,
+      "learning_rate": 2.870412276498158e-07,
+      "loss": 0.4976,
+      "step": 54963
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.3709819316864014,
+      "learning_rate": 2.869531862715236e-07,
+      "loss": 0.3949,
+      "step": 54964
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 5.431694507598877,
+      "learning_rate": 2.868651579984488e-07,
+      "loss": 0.635,
+      "step": 54965
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.588230848312378,
+      "learning_rate": 2.8677714283083745e-07,
+      "loss": 0.4468,
+      "step": 54966
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.4305636882781982,
+      "learning_rate": 2.8668914076893263e-07,
+      "loss": 0.4871,
+      "step": 54967
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.5781381130218506,
+      "learning_rate": 2.866011518129802e-07,
+      "loss": 0.6081,
+      "step": 54968
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.922311782836914,
+      "learning_rate": 2.865131759632234e-07,
+      "loss": 0.5136,
+      "step": 54969
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.605563163757324,
+      "learning_rate": 2.8642521321990855e-07,
+      "loss": 0.4761,
+      "step": 54970
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.6490800380706787,
+      "learning_rate": 2.8633726358327894e-07,
+      "loss": 0.7138,
+      "step": 54971
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.3902440071105957,
+      "learning_rate": 2.8624932705357934e-07,
+      "loss": 0.504,
+      "step": 54972
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.5098013877868652,
+      "learning_rate": 2.861614036310556e-07,
+      "loss": 0.3303,
+      "step": 54973
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.5047812461853027,
+      "learning_rate": 2.8607349331595037e-07,
+      "loss": 0.6838,
+      "step": 54974
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 6.88240909576416,
+      "learning_rate": 2.859855961085095e-07,
+      "loss": 0.6316,
+      "step": 54975
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.5047218799591064,
+      "learning_rate": 2.858977120089762e-07,
+      "loss": 0.6133,
+      "step": 54976
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.56711483001709,
+      "learning_rate": 2.858098410175958e-07,
+      "loss": 0.7634,
+      "step": 54977
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.4228360652923584,
+      "learning_rate": 2.8572198313461195e-07,
+      "loss": 0.5853,
+      "step": 54978
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.8678159713745117,
+      "learning_rate": 2.8563413836026956e-07,
+      "loss": 0.7151,
+      "step": 54979
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.0788164138793945,
+      "learning_rate": 2.855463066948122e-07,
+      "loss": 0.5902,
+      "step": 54980
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.281909465789795,
+      "learning_rate": 2.8545848813848474e-07,
+      "loss": 0.6432,
+      "step": 54981
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.641388177871704,
+      "learning_rate": 2.853706826915309e-07,
+      "loss": 0.6988,
+      "step": 54982
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.799109697341919,
+      "learning_rate": 2.8528289035419543e-07,
+      "loss": 0.6612,
+      "step": 54983
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.9671530723571777,
+      "learning_rate": 2.851951111267215e-07,
+      "loss": 0.5525,
+      "step": 54984
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.431020498275757,
+      "learning_rate": 2.851073450093539e-07,
+      "loss": 0.8551,
+      "step": 54985
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.102104902267456,
+      "learning_rate": 2.850195920023363e-07,
+      "loss": 0.6917,
+      "step": 54986
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.343546152114868,
+      "learning_rate": 2.849318521059141e-07,
+      "loss": 0.7947,
+      "step": 54987
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 5.133305549621582,
+      "learning_rate": 2.848441253203288e-07,
+      "loss": 0.8307,
+      "step": 54988
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 4.381138324737549,
+      "learning_rate": 2.847564116458268e-07,
+      "loss": 0.6834,
+      "step": 54989
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.8837780952453613,
+      "learning_rate": 2.846687110826496e-07,
+      "loss": 0.5548,
+      "step": 54990
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.6031134128570557,
+      "learning_rate": 2.8458102363104365e-07,
+      "loss": 0.7103,
+      "step": 54991
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.542052745819092,
+      "learning_rate": 2.844933492912505e-07,
+      "loss": 0.7247,
+      "step": 54992
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.210418701171875,
+      "learning_rate": 2.844056880635154e-07,
+      "loss": 0.5444,
+      "step": 54993
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 4.420046329498291,
+      "learning_rate": 2.8431803994808104e-07,
+      "loss": 0.6218,
+      "step": 54994
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 4.3379621505737305,
+      "learning_rate": 2.842304049451927e-07,
+      "loss": 0.7001,
+      "step": 54995
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.1961185932159424,
+      "learning_rate": 2.841427830550919e-07,
+      "loss": 0.4748,
+      "step": 54996
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.0512170791625977,
+      "learning_rate": 2.84055174278024e-07,
+      "loss": 0.4755,
+      "step": 54997
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.4233553409576416,
+      "learning_rate": 2.839675786142315e-07,
+      "loss": 0.4678,
+      "step": 54998
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.5899503231048584,
+      "learning_rate": 2.838799960639593e-07,
+      "loss": 0.3112,
+      "step": 54999
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.8331165313720703,
+      "learning_rate": 2.8379242662744944e-07,
+      "loss": 0.4867,
+      "step": 55000
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.492180347442627,
+      "learning_rate": 2.8370487030494665e-07,
+      "loss": 0.4226,
+      "step": 55001
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.285736560821533,
+      "learning_rate": 2.8361732709669356e-07,
+      "loss": 0.7854,
+      "step": 55002
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.97527813911438,
+      "learning_rate": 2.835297970029338e-07,
+      "loss": 0.6225,
+      "step": 55003
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.589688301086426,
+      "learning_rate": 2.834422800239106e-07,
+      "loss": 0.7927,
+      "step": 55004
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.857154130935669,
+      "learning_rate": 2.8335477615986764e-07,
+      "loss": 0.7408,
+      "step": 55005
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.6331443786621094,
+      "learning_rate": 2.8326728541104796e-07,
+      "loss": 0.5243,
+      "step": 55006
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.6681528091430664,
+      "learning_rate": 2.8317980777769596e-07,
+      "loss": 0.9016,
+      "step": 55007
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.680331230163574,
+      "learning_rate": 2.830923432600535e-07,
+      "loss": 0.5001,
+      "step": 55008
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.7883572578430176,
+      "learning_rate": 2.8300489185836444e-07,
+      "loss": 0.4199,
+      "step": 55009
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.076092004776001,
+      "learning_rate": 2.8291745357287127e-07,
+      "loss": 0.7065,
+      "step": 55010
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.1098673343658447,
+      "learning_rate": 2.8283002840381824e-07,
+      "loss": 0.2971,
+      "step": 55011
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 5.029606819152832,
+      "learning_rate": 2.8274261635144685e-07,
+      "loss": 0.4519,
+      "step": 55012
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.561173915863037,
+      "learning_rate": 2.826552174160024e-07,
+      "loss": 0.6287,
+      "step": 55013
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.8701131343841553,
+      "learning_rate": 2.8256783159772536e-07,
+      "loss": 0.4499,
+      "step": 55014
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.210206985473633,
+      "learning_rate": 2.8248045889686103e-07,
+      "loss": 0.5724,
+      "step": 55015
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.056873083114624,
+      "learning_rate": 2.823930993136503e-07,
+      "loss": 0.4827,
+      "step": 55016
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 5.3497748374938965,
+      "learning_rate": 2.8230575284833805e-07,
+      "loss": 0.7125,
+      "step": 55017
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.3202126026153564,
+      "learning_rate": 2.822184195011657e-07,
+      "loss": 0.4904,
+      "step": 55018
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.4764621257781982,
+      "learning_rate": 2.821310992723769e-07,
+      "loss": 0.708,
+      "step": 55019
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.205172538757324,
+      "learning_rate": 2.820437921622138e-07,
+      "loss": 0.4136,
+      "step": 55020
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.8045783042907715,
+      "learning_rate": 2.8195649817092e-07,
+      "loss": 0.5603,
+      "step": 55021
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.60426926612854,
+      "learning_rate": 2.8186921729873697e-07,
+      "loss": 0.6541,
+      "step": 55022
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.441803455352783,
+      "learning_rate": 2.8178194954590845e-07,
+      "loss": 0.7176,
+      "step": 55023
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.8988375663757324,
+      "learning_rate": 2.8169469491267753e-07,
+      "loss": 0.9152,
+      "step": 55024
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.6528313159942627,
+      "learning_rate": 2.816074533992852e-07,
+      "loss": 0.3371,
+      "step": 55025
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.421720504760742,
+      "learning_rate": 2.815202250059751e-07,
+      "loss": 0.7124,
+      "step": 55026
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 4.448538780212402,
+      "learning_rate": 2.8143300973299093e-07,
+      "loss": 0.518,
+      "step": 55027
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.2322487831115723,
+      "learning_rate": 2.8134580758057253e-07,
+      "loss": 0.4042,
+      "step": 55028
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.526623010635376,
+      "learning_rate": 2.8125861854896463e-07,
+      "loss": 0.6361,
+      "step": 55029
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.1417417526245117,
+      "learning_rate": 2.811714426384082e-07,
+      "loss": 0.7528,
+      "step": 55030
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.8080451488494873,
+      "learning_rate": 2.8108427984914697e-07,
+      "loss": 0.4927,
+      "step": 55031
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.1146109104156494,
+      "learning_rate": 2.809971301814218e-07,
+      "loss": 0.6861,
+      "step": 55032
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.0891993045806885,
+      "learning_rate": 2.80909993635477e-07,
+      "loss": 0.5774,
+      "step": 55033
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.8305766582489014,
+      "learning_rate": 2.808228702115523e-07,
+      "loss": 0.4777,
+      "step": 55034
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.7610256671905518,
+      "learning_rate": 2.80735759909892e-07,
+      "loss": 0.6766,
+      "step": 55035
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.837329626083374,
+      "learning_rate": 2.806486627307381e-07,
+      "loss": 0.4967,
+      "step": 55036
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.5680649280548096,
+      "learning_rate": 2.8056157867433153e-07,
+      "loss": 0.4587,
+      "step": 55037
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.0835094451904297,
+      "learning_rate": 2.80474507740916e-07,
+      "loss": 0.6572,
+      "step": 55038
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 4.152533054351807,
+      "learning_rate": 2.803874499307324e-07,
+      "loss": 0.5693,
+      "step": 55039
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.759533405303955,
+      "learning_rate": 2.803004052440239e-07,
+      "loss": 0.6607,
+      "step": 55040
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.0834696292877197,
+      "learning_rate": 2.8021337368103085e-07,
+      "loss": 0.6814,
+      "step": 55041
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.1010029315948486,
+      "learning_rate": 2.801263552419975e-07,
+      "loss": 0.5585,
+      "step": 55042
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.118081569671631,
+      "learning_rate": 2.800393499271636e-07,
+      "loss": 0.517,
+      "step": 55043
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 4.190207481384277,
+      "learning_rate": 2.7995235773677355e-07,
+      "loss": 0.6831,
+      "step": 55044
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.6914641857147217,
+      "learning_rate": 2.7986537867106645e-07,
+      "loss": 0.8131,
+      "step": 55045
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.5444648265838623,
+      "learning_rate": 2.797784127302855e-07,
+      "loss": 0.6336,
+      "step": 55046
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 4.76259708404541,
+      "learning_rate": 2.7969145991467275e-07,
+      "loss": 0.4513,
+      "step": 55047
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.5918288230895996,
+      "learning_rate": 2.796045202244707e-07,
+      "loss": 0.4722,
+      "step": 55048
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.926804542541504,
+      "learning_rate": 2.7951759365991925e-07,
+      "loss": 0.551,
+      "step": 55049
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.808049201965332,
+      "learning_rate": 2.794306802212615e-07,
+      "loss": 0.6332,
+      "step": 55050
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 4.039093971252441,
+      "learning_rate": 2.793437799087384e-07,
+      "loss": 0.6669,
+      "step": 55051
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.8034026622772217,
+      "learning_rate": 2.792568927225925e-07,
+      "loss": 0.6159,
+      "step": 55052
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.0690553188323975,
+      "learning_rate": 2.7917001866306416e-07,
+      "loss": 0.5182,
+      "step": 55053
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.854712963104248,
+      "learning_rate": 2.79083157730396e-07,
+      "loss": 0.8125,
+      "step": 55054
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.772271156311035,
+      "learning_rate": 2.7899630992482827e-07,
+      "loss": 0.6253,
+      "step": 55055
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.908006429672241,
+      "learning_rate": 2.7890947524660375e-07,
+      "loss": 0.5115,
+      "step": 55056
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.6810734272003174,
+      "learning_rate": 2.788226536959632e-07,
+      "loss": 0.8086,
+      "step": 55057
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.264411449432373,
+      "learning_rate": 2.787358452731487e-07,
+      "loss": 0.467,
+      "step": 55058
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 4.2376933097839355,
+      "learning_rate": 2.7864904997840057e-07,
+      "loss": 0.8108,
+      "step": 55059
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.9291419982910156,
+      "learning_rate": 2.785622678119615e-07,
+      "loss": 0.5229,
+      "step": 55060
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.3202812671661377,
+      "learning_rate": 2.7847549877407177e-07,
+      "loss": 0.8513,
+      "step": 55061
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.4282283782958984,
+      "learning_rate": 2.783887428649729e-07,
+      "loss": 0.6625,
+      "step": 55062
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 4.390693187713623,
+      "learning_rate": 2.7830200008490574e-07,
+      "loss": 0.6075,
+      "step": 55063
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.2250101566314697,
+      "learning_rate": 2.782152704341129e-07,
+      "loss": 0.7571,
+      "step": 55064
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.5274083614349365,
+      "learning_rate": 2.7812855391283365e-07,
+      "loss": 0.452,
+      "step": 55065
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.351691484451294,
+      "learning_rate": 2.7804185052131003e-07,
+      "loss": 0.6111,
+      "step": 55066
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.7612059116363525,
+      "learning_rate": 2.779551602597835e-07,
+      "loss": 0.7092,
+      "step": 55067
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.293677568435669,
+      "learning_rate": 2.77868483128495e-07,
+      "loss": 0.4245,
+      "step": 55068
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.9417097568511963,
+      "learning_rate": 2.777818191276849e-07,
+      "loss": 0.4448,
+      "step": 55069
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.773279905319214,
+      "learning_rate": 2.776951682575951e-07,
+      "loss": 0.6205,
+      "step": 55070
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.927304744720459,
+      "learning_rate": 2.77608530518465e-07,
+      "loss": 0.4275,
+      "step": 55071
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.9022023677825928,
+      "learning_rate": 2.7752190591053775e-07,
+      "loss": 0.6425,
+      "step": 55072
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.4668335914611816,
+      "learning_rate": 2.774352944340519e-07,
+      "loss": 0.6721,
+      "step": 55073
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.052891254425049,
+      "learning_rate": 2.7734869608925076e-07,
+      "loss": 0.508,
+      "step": 55074
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 4.924718379974365,
+      "learning_rate": 2.7726211087637233e-07,
+      "loss": 0.6733,
+      "step": 55075
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.5874505043029785,
+      "learning_rate": 2.7717553879565983e-07,
+      "loss": 0.6136,
+      "step": 55076
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.851609706878662,
+      "learning_rate": 2.7708897984735196e-07,
+      "loss": 0.4617,
+      "step": 55077
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.010695219039917,
+      "learning_rate": 2.7700243403169126e-07,
+      "loss": 0.5881,
+      "step": 55078
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.647552490234375,
+      "learning_rate": 2.76915901348917e-07,
+      "loss": 0.5377,
+      "step": 55079
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.4897751808166504,
+      "learning_rate": 2.7682938179927067e-07,
+      "loss": 0.5538,
+      "step": 55080
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.7368323802948,
+      "learning_rate": 2.7674287538299206e-07,
+      "loss": 0.4924,
+      "step": 55081
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.0082006454467773,
+      "learning_rate": 2.766563821003232e-07,
+      "loss": 0.5427,
+      "step": 55082
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 1.8876160383224487,
+      "learning_rate": 2.7656990195150224e-07,
+      "loss": 0.3232,
+      "step": 55083
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.1491963863372803,
+      "learning_rate": 2.7648343493677123e-07,
+      "loss": 0.6357,
+      "step": 55084
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 4.033743858337402,
+      "learning_rate": 2.763969810563705e-07,
+      "loss": 0.7088,
+      "step": 55085
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.997786521911621,
+      "learning_rate": 2.763105403105404e-07,
+      "loss": 0.7842,
+      "step": 55086
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.841632127761841,
+      "learning_rate": 2.762241126995202e-07,
+      "loss": 0.6341,
+      "step": 55087
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.8095645904541016,
+      "learning_rate": 2.7613769822355253e-07,
+      "loss": 0.6454,
+      "step": 55088
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.821546792984009,
+      "learning_rate": 2.7605129688287544e-07,
+      "loss": 0.6153,
+      "step": 55089
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.1907896995544434,
+      "learning_rate": 2.759649086777311e-07,
+      "loss": 0.4382,
+      "step": 55090
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.7049264907836914,
+      "learning_rate": 2.7587853360835746e-07,
+      "loss": 0.9173,
+      "step": 55091
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.5657434463500977,
+      "learning_rate": 2.7579217167499674e-07,
+      "loss": 0.723,
+      "step": 55092
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.7499704360961914,
+      "learning_rate": 2.7570582287788804e-07,
+      "loss": 0.5959,
+      "step": 55093
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.23477840423584,
+      "learning_rate": 2.756194872172718e-07,
+      "loss": 0.3705,
+      "step": 55094
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.1666505336761475,
+      "learning_rate": 2.7553316469338733e-07,
+      "loss": 0.4508,
+      "step": 55095
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.1323227882385254,
+      "learning_rate": 2.754468553064754e-07,
+      "loss": 0.8617,
+      "step": 55096
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.8139140605926514,
+      "learning_rate": 2.7536055905677706e-07,
+      "loss": 0.4053,
+      "step": 55097
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.5112011432647705,
+      "learning_rate": 2.752742759445298e-07,
+      "loss": 0.5865,
+      "step": 55098
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.6657345294952393,
+      "learning_rate": 2.7518800596997574e-07,
+      "loss": 0.5498,
+      "step": 55099
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.0739405155181885,
+      "learning_rate": 2.7510174913335355e-07,
+      "loss": 0.5171,
+      "step": 55100
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.6675920486450195,
+      "learning_rate": 2.750155054349035e-07,
+      "loss": 0.5658,
+      "step": 55101
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.381948232650757,
+      "learning_rate": 2.7492927487486497e-07,
+      "loss": 0.5775,
+      "step": 55102
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 5.557335376739502,
+      "learning_rate": 2.7484305745347885e-07,
+      "loss": 0.4887,
+      "step": 55103
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.6216392517089844,
+      "learning_rate": 2.747568531709832e-07,
+      "loss": 0.4824,
+      "step": 55104
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.549238920211792,
+      "learning_rate": 2.746706620276191e-07,
+      "loss": 0.4801,
+      "step": 55105
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.4111320972442627,
+      "learning_rate": 2.745844840236256e-07,
+      "loss": 0.7245,
+      "step": 55106
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.117265462875366,
+      "learning_rate": 2.7449831915924205e-07,
+      "loss": 0.7081,
+      "step": 55107
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 4.328470230102539,
+      "learning_rate": 2.744121674347083e-07,
+      "loss": 0.723,
+      "step": 55108
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.024839162826538,
+      "learning_rate": 2.7432602885026526e-07,
+      "loss": 0.6958,
+      "step": 55109
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.6254067420959473,
+      "learning_rate": 2.7423990340615046e-07,
+      "loss": 0.5745,
+      "step": 55110
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.9777655601501465,
+      "learning_rate": 2.7415379110260485e-07,
+      "loss": 0.4055,
+      "step": 55111
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 11.723753929138184,
+      "learning_rate": 2.740676919398666e-07,
+      "loss": 0.779,
+      "step": 55112
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.4548656940460205,
+      "learning_rate": 2.73981605918176e-07,
+      "loss": 0.6235,
+      "step": 55113
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.785646677017212,
+      "learning_rate": 2.738955330377718e-07,
+      "loss": 0.8108,
+      "step": 55114
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.0137948989868164,
+      "learning_rate": 2.738094732988944e-07,
+      "loss": 0.474,
+      "step": 55115
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.582693099975586,
+      "learning_rate": 2.7372342670178133e-07,
+      "loss": 0.6562,
+      "step": 55116
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.2557053565979004,
+      "learning_rate": 2.73637393246674e-07,
+      "loss": 0.774,
+      "step": 55117
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.113616466522217,
+      "learning_rate": 2.7355137293380963e-07,
+      "loss": 0.6751,
+      "step": 55118
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.9698734283447266,
+      "learning_rate": 2.73465365763429e-07,
+      "loss": 0.7859,
+      "step": 55119
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.9420039653778076,
+      "learning_rate": 2.733793717357702e-07,
+      "loss": 0.5316,
+      "step": 55120
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.3090710639953613,
+      "learning_rate": 2.732933908510732e-07,
+      "loss": 0.4549,
+      "step": 55121
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 8.888047218322754,
+      "learning_rate": 2.732074231095755e-07,
+      "loss": 0.4494,
+      "step": 55122
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 5.169941425323486,
+      "learning_rate": 2.7312146851151857e-07,
+      "loss": 0.6844,
+      "step": 55123
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.172145366668701,
+      "learning_rate": 2.730355270571394e-07,
+      "loss": 0.4176,
+      "step": 55124
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.7428717613220215,
+      "learning_rate": 2.729495987466779e-07,
+      "loss": 0.6453,
+      "step": 55125
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.3163704872131348,
+      "learning_rate": 2.728636835803722e-07,
+      "loss": 0.6245,
+      "step": 55126
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.803293466567993,
+      "learning_rate": 2.7277778155846204e-07,
+      "loss": 0.4606,
+      "step": 55127
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.177821159362793,
+      "learning_rate": 2.7269189268118614e-07,
+      "loss": 0.5203,
+      "step": 55128
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.3479957580566406,
+      "learning_rate": 2.7260601694878374e-07,
+      "loss": 0.7578,
+      "step": 55129
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.6300771236419678,
+      "learning_rate": 2.7252015436149194e-07,
+      "loss": 0.7315,
+      "step": 55130
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.7888131141662598,
+      "learning_rate": 2.724343049195516e-07,
+      "loss": 0.6551,
+      "step": 55131
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.5162088871002197,
+      "learning_rate": 2.723484686232003e-07,
+      "loss": 0.8221,
+      "step": 55132
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.5110104084014893,
+      "learning_rate": 2.722626454726768e-07,
+      "loss": 0.7103,
+      "step": 55133
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 4.323699951171875,
+      "learning_rate": 2.721768354682197e-07,
+      "loss": 0.5876,
+      "step": 55134
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 4.137643814086914,
+      "learning_rate": 2.7209103861006834e-07,
+      "loss": 0.6504,
+      "step": 55135
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.265826463699341,
+      "learning_rate": 2.7200525489845976e-07,
+      "loss": 0.5236,
+      "step": 55136
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 4.343760967254639,
+      "learning_rate": 2.7191948433363426e-07,
+      "loss": 0.562,
+      "step": 55137
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.431382179260254,
+      "learning_rate": 2.718337269158289e-07,
+      "loss": 0.4973,
+      "step": 55138
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.6226985454559326,
+      "learning_rate": 2.717479826452829e-07,
+      "loss": 0.6515,
+      "step": 55139
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.555361747741699,
+      "learning_rate": 2.7166225152223446e-07,
+      "loss": 0.5688,
+      "step": 55140
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 4.324984073638916,
+      "learning_rate": 2.7157653354692226e-07,
+      "loss": 0.669,
+      "step": 55141
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.0954623222351074,
+      "learning_rate": 2.7149082871958386e-07,
+      "loss": 0.6243,
+      "step": 55142
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.73933482170105,
+      "learning_rate": 2.714051370404591e-07,
+      "loss": 0.5291,
+      "step": 55143
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.5257458686828613,
+      "learning_rate": 2.7131945850978447e-07,
+      "loss": 0.6419,
+      "step": 55144
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.5282764434814453,
+      "learning_rate": 2.712337931277986e-07,
+      "loss": 0.6375,
+      "step": 55145
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 4.233903408050537,
+      "learning_rate": 2.711481408947414e-07,
+      "loss": 0.6065,
+      "step": 55146
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.7507550716400146,
+      "learning_rate": 2.710625018108487e-07,
+      "loss": 0.6827,
+      "step": 55147
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.4354612827301025,
+      "learning_rate": 2.7097687587635977e-07,
+      "loss": 0.4605,
+      "step": 55148
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.024040699005127,
+      "learning_rate": 2.708912630915134e-07,
+      "loss": 0.8943,
+      "step": 55149
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.4111828804016113,
+      "learning_rate": 2.708056634565459e-07,
+      "loss": 0.5801,
+      "step": 55150
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.414102554321289,
+      "learning_rate": 2.7072007697169723e-07,
+      "loss": 0.4998,
+      "step": 55151
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.3900296688079834,
+      "learning_rate": 2.706345036372038e-07,
+      "loss": 0.6546,
+      "step": 55152
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 4.1099042892456055,
+      "learning_rate": 2.7054894345330427e-07,
+      "loss": 0.535,
+      "step": 55153
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.0711867809295654,
+      "learning_rate": 2.704633964202363e-07,
+      "loss": 0.496,
+      "step": 55154
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.6002368927001953,
+      "learning_rate": 2.703778625382386e-07,
+      "loss": 0.7468,
+      "step": 55155
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.222266674041748,
+      "learning_rate": 2.7029234180754707e-07,
+      "loss": 0.6831,
+      "step": 55156
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.507174015045166,
+      "learning_rate": 2.702068342284014e-07,
+      "loss": 0.3368,
+      "step": 55157
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.6973929405212402,
+      "learning_rate": 2.701213398010388e-07,
+      "loss": 0.4397,
+      "step": 55158
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.5474233627319336,
+      "learning_rate": 2.700358585256968e-07,
+      "loss": 0.3166,
+      "step": 55159
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.147543430328369,
+      "learning_rate": 2.699503904026135e-07,
+      "loss": 0.4944,
+      "step": 55160
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.578108787536621,
+      "learning_rate": 2.6986493543202594e-07,
+      "loss": 0.4105,
+      "step": 55161
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.9342892169952393,
+      "learning_rate": 2.6977949361417235e-07,
+      "loss": 0.5915,
+      "step": 55162
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.729678153991699,
+      "learning_rate": 2.696940649492891e-07,
+      "loss": 0.604,
+      "step": 55163
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.3710873126983643,
+      "learning_rate": 2.69608649437616e-07,
+      "loss": 0.6069,
+      "step": 55164
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.6529414653778076,
+      "learning_rate": 2.695232470793879e-07,
+      "loss": 0.6205,
+      "step": 55165
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.488745927810669,
+      "learning_rate": 2.6943785787484466e-07,
+      "loss": 0.4189,
+      "step": 55166
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.9834253787994385,
+      "learning_rate": 2.6935248182422157e-07,
+      "loss": 0.4889,
+      "step": 55167
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.337752103805542,
+      "learning_rate": 2.6926711892775736e-07,
+      "loss": 0.5889,
+      "step": 55168
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.6600565910339355,
+      "learning_rate": 2.6918176918568904e-07,
+      "loss": 0.5098,
+      "step": 55169
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 4.92645263671875,
+      "learning_rate": 2.690964325982548e-07,
+      "loss": 0.6405,
+      "step": 55170
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.5193638801574707,
+      "learning_rate": 2.690111091656905e-07,
+      "loss": 0.5622,
+      "step": 55171
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.449955940246582,
+      "learning_rate": 2.6892579888823487e-07,
+      "loss": 0.4881,
+      "step": 55172
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.4387543201446533,
+      "learning_rate": 2.688405017661233e-07,
+      "loss": 0.5368,
+      "step": 55173
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 4.126628398895264,
+      "learning_rate": 2.6875521779959444e-07,
+      "loss": 0.5481,
+      "step": 55174
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.820146083831787,
+      "learning_rate": 2.686699469888848e-07,
+      "loss": 0.6081,
+      "step": 55175
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.3490004539489746,
+      "learning_rate": 2.68584689334232e-07,
+      "loss": 0.4593,
+      "step": 55176
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.373713970184326,
+      "learning_rate": 2.6849944483587186e-07,
+      "loss": 0.4409,
+      "step": 55177
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.348628282546997,
+      "learning_rate": 2.684142134940437e-07,
+      "loss": 0.7066,
+      "step": 55178
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.7954206466674805,
+      "learning_rate": 2.683289953089818e-07,
+      "loss": 0.6503,
+      "step": 55179
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.9474289417266846,
+      "learning_rate": 2.682437902809254e-07,
+      "loss": 0.6091,
+      "step": 55180
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.8481438159942627,
+      "learning_rate": 2.6815859841010925e-07,
+      "loss": 0.5278,
+      "step": 55181
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.2664058208465576,
+      "learning_rate": 2.680734196967727e-07,
+      "loss": 0.5533,
+      "step": 55182
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.63757586479187,
+      "learning_rate": 2.679882541411505e-07,
+      "loss": 0.5613,
+      "step": 55183
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.5558993816375732,
+      "learning_rate": 2.6790310174348135e-07,
+      "loss": 0.5414,
+      "step": 55184
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.1828978061676025,
+      "learning_rate": 2.6781796250400005e-07,
+      "loss": 0.5934,
+      "step": 55185
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 4.405135631561279,
+      "learning_rate": 2.6773283642294477e-07,
+      "loss": 0.3937,
+      "step": 55186
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 5.953634262084961,
+      "learning_rate": 2.676477235005509e-07,
+      "loss": 0.5904,
+      "step": 55187
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.640108108520508,
+      "learning_rate": 2.6756262373705655e-07,
+      "loss": 0.4351,
+      "step": 55188
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.295687675476074,
+      "learning_rate": 2.674775371326971e-07,
+      "loss": 0.6742,
+      "step": 55189
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.826402425765991,
+      "learning_rate": 2.673924636877112e-07,
+      "loss": 0.6459,
+      "step": 55190
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.1017744541168213,
+      "learning_rate": 2.6730740340233263e-07,
+      "loss": 0.5079,
+      "step": 55191
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.921471118927002,
+      "learning_rate": 2.6722235627680005e-07,
+      "loss": 0.5979,
+      "step": 55192
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.411616563796997,
+      "learning_rate": 2.671373223113488e-07,
+      "loss": 0.743,
+      "step": 55193
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.1915969848632812,
+      "learning_rate": 2.67052301506216e-07,
+      "loss": 0.5739,
+      "step": 55194
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.1049249172210693,
+      "learning_rate": 2.669672938616374e-07,
+      "loss": 0.5893,
+      "step": 55195
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.628817081451416,
+      "learning_rate": 2.668822993778503e-07,
+      "loss": 0.8082,
+      "step": 55196
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.7320494651794434,
+      "learning_rate": 2.667973180550892e-07,
+      "loss": 0.6986,
+      "step": 55197
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.7563352584838867,
+      "learning_rate": 2.6671234989359306e-07,
+      "loss": 0.6607,
+      "step": 55198
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.7304933071136475,
+      "learning_rate": 2.66627394893596e-07,
+      "loss": 0.638,
+      "step": 55199
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.7983036041259766,
+      "learning_rate": 2.6654245305533566e-07,
+      "loss": 0.647,
+      "step": 55200
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.673771381378174,
+      "learning_rate": 2.664575243790468e-07,
+      "loss": 0.5025,
+      "step": 55201
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.5477328300476074,
+      "learning_rate": 2.663726088649665e-07,
+      "loss": 0.5485,
+      "step": 55202
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.7883427143096924,
+      "learning_rate": 2.662877065133307e-07,
+      "loss": 0.5639,
+      "step": 55203
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 20.25000762939453,
+      "learning_rate": 2.662028173243758e-07,
+      "loss": 0.3139,
+      "step": 55204
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.8633222579956055,
+      "learning_rate": 2.6611794129833723e-07,
+      "loss": 0.5518,
+      "step": 55205
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.6986429691314697,
+      "learning_rate": 2.66033078435452e-07,
+      "loss": 0.4728,
+      "step": 55206
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.528846502304077,
+      "learning_rate": 2.6594822873595437e-07,
+      "loss": 0.816,
+      "step": 55207
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.155088186264038,
+      "learning_rate": 2.658633922000814e-07,
+      "loss": 0.4689,
+      "step": 55208
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.454271078109741,
+      "learning_rate": 2.6577856882806897e-07,
+      "loss": 0.5584,
+      "step": 55209
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.922956943511963,
+      "learning_rate": 2.6569375862015355e-07,
+      "loss": 0.5262,
+      "step": 55210
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.5215628147125244,
+      "learning_rate": 2.6560896157656945e-07,
+      "loss": 0.5736,
+      "step": 55211
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.6364500522613525,
+      "learning_rate": 2.655241776975542e-07,
+      "loss": 0.5626,
+      "step": 55212
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.592031240463257,
+      "learning_rate": 2.654394069833416e-07,
+      "loss": 0.6194,
+      "step": 55213
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.637148380279541,
+      "learning_rate": 2.6535464943416967e-07,
+      "loss": 0.4798,
+      "step": 55214
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.248647689819336,
+      "learning_rate": 2.6526990505027164e-07,
+      "loss": 0.5336,
+      "step": 55215
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.6736855506896973,
+      "learning_rate": 2.65185173831885e-07,
+      "loss": 0.7183,
+      "step": 55216
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 4.231584548950195,
+      "learning_rate": 2.6510045577924413e-07,
+      "loss": 0.5793,
+      "step": 55217
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.4006757736206055,
+      "learning_rate": 2.65015750892586e-07,
+      "loss": 0.6842,
+      "step": 55218
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.1519787311553955,
+      "learning_rate": 2.6493105917214437e-07,
+      "loss": 0.6992,
+      "step": 55219
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.588071823120117,
+      "learning_rate": 2.6484638061815616e-07,
+      "loss": 0.543,
+      "step": 55220
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.8211865425109863,
+      "learning_rate": 2.6476171523085626e-07,
+      "loss": 0.5628,
+      "step": 55221
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.5729591846466064,
+      "learning_rate": 2.6467706301048e-07,
+      "loss": 0.8102,
+      "step": 55222
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.161761999130249,
+      "learning_rate": 2.6459242395726334e-07,
+      "loss": 0.6312,
+      "step": 55223
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 4.174306392669678,
+      "learning_rate": 2.645077980714411e-07,
+      "loss": 0.7338,
+      "step": 55224
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.7940587997436523,
+      "learning_rate": 2.644231853532486e-07,
+      "loss": 0.4807,
+      "step": 55225
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.9013960361480713,
+      "learning_rate": 2.6433858580292127e-07,
+      "loss": 0.7592,
+      "step": 55226
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.430232286453247,
+      "learning_rate": 2.642539994206944e-07,
+      "loss": 0.5527,
+      "step": 55227
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.5351922512054443,
+      "learning_rate": 2.641694262068029e-07,
+      "loss": 0.6843,
+      "step": 55228
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.694108247756958,
+      "learning_rate": 2.64084866161482e-07,
+      "loss": 0.7201,
+      "step": 55229
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.4564201831817627,
+      "learning_rate": 2.6400031928496663e-07,
+      "loss": 0.4756,
+      "step": 55230
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.0985798835754395,
+      "learning_rate": 2.639157855774932e-07,
+      "loss": 0.5271,
+      "step": 55231
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.377739667892456,
+      "learning_rate": 2.638312650392955e-07,
+      "loss": 0.6327,
+      "step": 55232
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 1.9332730770111084,
+      "learning_rate": 2.6374675767060875e-07,
+      "loss": 0.4774,
+      "step": 55233
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.218442440032959,
+      "learning_rate": 2.636622634716679e-07,
+      "loss": 0.7249,
+      "step": 55234
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 6.6748762130737305,
+      "learning_rate": 2.6357778244270826e-07,
+      "loss": 0.6561,
+      "step": 55235
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.691025733947754,
+      "learning_rate": 2.634933145839641e-07,
+      "loss": 0.598,
+      "step": 55236
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.382436990737915,
+      "learning_rate": 2.634088598956719e-07,
+      "loss": 0.5233,
+      "step": 55237
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.333568572998047,
+      "learning_rate": 2.6332441837806367e-07,
+      "loss": 0.4828,
+      "step": 55238
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.87434458732605,
+      "learning_rate": 2.63239990031377e-07,
+      "loss": 0.5454,
+      "step": 55239
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.1634228229522705,
+      "learning_rate": 2.6315557485584455e-07,
+      "loss": 0.6011,
+      "step": 55240
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.4642343521118164,
+      "learning_rate": 2.630711728517032e-07,
+      "loss": 0.6035,
+      "step": 55241
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.1885147094726562,
+      "learning_rate": 2.629867840191852e-07,
+      "loss": 0.5393,
+      "step": 55242
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.4429781436920166,
+      "learning_rate": 2.6290240835852745e-07,
+      "loss": 0.8375,
+      "step": 55243
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.250420093536377,
+      "learning_rate": 2.628180458699625e-07,
+      "loss": 0.5797,
+      "step": 55244
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.6472320556640625,
+      "learning_rate": 2.627336965537269e-07,
+      "loss": 0.439,
+      "step": 55245
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.522942066192627,
+      "learning_rate": 2.6264936041005383e-07,
+      "loss": 0.5888,
+      "step": 55246
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.8553454875946045,
+      "learning_rate": 2.625650374391786e-07,
+      "loss": 0.657,
+      "step": 55247
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.8263964653015137,
+      "learning_rate": 2.624807276413344e-07,
+      "loss": 0.887,
+      "step": 55248
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.504216194152832,
+      "learning_rate": 2.6239643101675706e-07,
+      "loss": 0.2851,
+      "step": 55249
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.0598502159118652,
+      "learning_rate": 2.6231214756568034e-07,
+      "loss": 0.6252,
+      "step": 55250
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 5.36441707611084,
+      "learning_rate": 2.622278772883391e-07,
+      "loss": 0.8118,
+      "step": 55251
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.2761545181274414,
+      "learning_rate": 2.6214362018496695e-07,
+      "loss": 0.5757,
+      "step": 55252
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.9746363162994385,
+      "learning_rate": 2.6205937625579936e-07,
+      "loss": 0.5851,
+      "step": 55253
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.847275733947754,
+      "learning_rate": 2.6197514550106884e-07,
+      "loss": 0.6313,
+      "step": 55254
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.9152233600616455,
+      "learning_rate": 2.618909279210113e-07,
+      "loss": 0.7084,
+      "step": 55255
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.037567615509033,
+      "learning_rate": 2.6180672351585943e-07,
+      "loss": 0.6306,
+      "step": 55256
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.9324026107788086,
+      "learning_rate": 2.6172253228584907e-07,
+      "loss": 0.698,
+      "step": 55257
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.9422855377197266,
+      "learning_rate": 2.616383542312123e-07,
+      "loss": 0.5082,
+      "step": 55258
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.488300085067749,
+      "learning_rate": 2.6155418935218503e-07,
+      "loss": 0.5305,
+      "step": 55259
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.557112216949463,
+      "learning_rate": 2.614700376489998e-07,
+      "loss": 0.4505,
+      "step": 55260
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.725520133972168,
+      "learning_rate": 2.6138589912189205e-07,
+      "loss": 0.6607,
+      "step": 55261
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.8832156658172607,
+      "learning_rate": 2.6130177377109434e-07,
+      "loss": 0.5692,
+      "step": 55262
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.0712406635284424,
+      "learning_rate": 2.61217661596842e-07,
+      "loss": 0.6089,
+      "step": 55263
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 6.3205647468566895,
+      "learning_rate": 2.611335625993672e-07,
+      "loss": 0.7473,
+      "step": 55264
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.494237184524536,
+      "learning_rate": 2.610494767789057e-07,
+      "loss": 0.3273,
+      "step": 55265
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.9647908210754395,
+      "learning_rate": 2.609654041356896e-07,
+      "loss": 0.6317,
+      "step": 55266
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.069941759109497,
+      "learning_rate": 2.608813446699537e-07,
+      "loss": 0.7175,
+      "step": 55267
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.9013760089874268,
+      "learning_rate": 2.6079729838193123e-07,
+      "loss": 0.4128,
+      "step": 55268
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.7856273651123047,
+      "learning_rate": 2.607132652718564e-07,
+      "loss": 0.607,
+      "step": 55269
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.379514455795288,
+      "learning_rate": 2.606292453399617e-07,
+      "loss": 0.4703,
+      "step": 55270
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.448988437652588,
+      "learning_rate": 2.6054523858648315e-07,
+      "loss": 0.7801,
+      "step": 55271
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.3925836086273193,
+      "learning_rate": 2.604612450116517e-07,
+      "loss": 0.6613,
+      "step": 55272
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.141613006591797,
+      "learning_rate": 2.6037726461570325e-07,
+      "loss": 0.5212,
+      "step": 55273
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.3278396129608154,
+      "learning_rate": 2.602932973988692e-07,
+      "loss": 0.6225,
+      "step": 55274
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.7803525924682617,
+      "learning_rate": 2.6020934336138446e-07,
+      "loss": 0.5356,
+      "step": 55275
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.191480875015259,
+      "learning_rate": 2.601254025034811e-07,
+      "loss": 0.5741,
+      "step": 55276
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.4071803092956543,
+      "learning_rate": 2.600414748253943e-07,
+      "loss": 0.7778,
+      "step": 55277
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.8667051792144775,
+      "learning_rate": 2.599575603273563e-07,
+      "loss": 0.6022,
+      "step": 55278
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.199986219406128,
+      "learning_rate": 2.598736590096007e-07,
+      "loss": 0.6287,
+      "step": 55279
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.9323432445526123,
+      "learning_rate": 2.597897708723607e-07,
+      "loss": 0.5422,
+      "step": 55280
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.790778398513794,
+      "learning_rate": 2.597058959158688e-07,
+      "loss": 0.5676,
+      "step": 55281
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 1.9884182214736938,
+      "learning_rate": 2.5962203414036046e-07,
+      "loss": 0.695,
+      "step": 55282
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.0339395999908447,
+      "learning_rate": 2.5953818554606656e-07,
+      "loss": 0.5169,
+      "step": 55283
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.2922120094299316,
+      "learning_rate": 2.594543501332214e-07,
+      "loss": 0.6421,
+      "step": 55284
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.911640167236328,
+      "learning_rate": 2.593705279020575e-07,
+      "loss": 0.5078,
+      "step": 55285
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.5477099418640137,
+      "learning_rate": 2.5928671885280923e-07,
+      "loss": 0.634,
+      "step": 55286
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.2865428924560547,
+      "learning_rate": 2.592029229857079e-07,
+      "loss": 0.7132,
+      "step": 55287
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.6517539024353027,
+      "learning_rate": 2.5911914030098793e-07,
+      "loss": 0.665,
+      "step": 55288
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 4.299777507781982,
+      "learning_rate": 2.590353707988808e-07,
+      "loss": 0.8762,
+      "step": 55289
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.467559814453125,
+      "learning_rate": 2.589516144796206e-07,
+      "loss": 0.4481,
+      "step": 55290
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.3808016777038574,
+      "learning_rate": 2.5886787134344004e-07,
+      "loss": 0.7206,
+      "step": 55291
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.7003462314605713,
+      "learning_rate": 2.5878414139057174e-07,
+      "loss": 0.5416,
+      "step": 55292
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.5554563999176025,
+      "learning_rate": 2.587004246212482e-07,
+      "loss": 0.7247,
+      "step": 55293
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.687915086746216,
+      "learning_rate": 2.586167210357038e-07,
+      "loss": 0.7511,
+      "step": 55294
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.584202527999878,
+      "learning_rate": 2.5853303063416934e-07,
+      "loss": 0.5401,
+      "step": 55295
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.480020046234131,
+      "learning_rate": 2.584493534168786e-07,
+      "loss": 0.5172,
+      "step": 55296
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.416374921798706,
+      "learning_rate": 2.5836568938406357e-07,
+      "loss": 0.5639,
+      "step": 55297
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.6110830307006836,
+      "learning_rate": 2.58282038535958e-07,
+      "loss": 0.5946,
+      "step": 55298
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.2629125118255615,
+      "learning_rate": 2.581984008727934e-07,
+      "loss": 0.635,
+      "step": 55299
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.2153546810150146,
+      "learning_rate": 2.5811477639480273e-07,
+      "loss": 0.7423,
+      "step": 55300
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.1428914070129395,
+      "learning_rate": 2.580311651022183e-07,
+      "loss": 0.6997,
+      "step": 55301
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.9295554161071777,
+      "learning_rate": 2.579475669952736e-07,
+      "loss": 0.567,
+      "step": 55302
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.466897964477539,
+      "learning_rate": 2.578639820741996e-07,
+      "loss": 0.3444,
+      "step": 55303
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.77126145362854,
+      "learning_rate": 2.5778041033923005e-07,
+      "loss": 0.4945,
+      "step": 55304
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.68005108833313,
+      "learning_rate": 2.576968517905959e-07,
+      "loss": 0.818,
+      "step": 55305
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.2322235107421875,
+      "learning_rate": 2.5761330642853077e-07,
+      "loss": 0.5626,
+      "step": 55306
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.7682769298553467,
+      "learning_rate": 2.575297742532662e-07,
+      "loss": 0.6153,
+      "step": 55307
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.985349416732788,
+      "learning_rate": 2.574462552650353e-07,
+      "loss": 0.5885,
+      "step": 55308
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.821152448654175,
+      "learning_rate": 2.57362749464069e-07,
+      "loss": 0.6223,
+      "step": 55309
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.4347739219665527,
+      "learning_rate": 2.5727925685060053e-07,
+      "loss": 0.4401,
+      "step": 55310
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.319258213043213,
+      "learning_rate": 2.571957774248612e-07,
+      "loss": 0.2918,
+      "step": 55311
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.262336492538452,
+      "learning_rate": 2.5711231118708493e-07,
+      "loss": 0.6462,
+      "step": 55312
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 4.055539608001709,
+      "learning_rate": 2.570288581375019e-07,
+      "loss": 0.5828,
+      "step": 55313
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.8981595039367676,
+      "learning_rate": 2.5694541827634533e-07,
+      "loss": 0.8164,
+      "step": 55314
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.0711240768432617,
+      "learning_rate": 2.568619916038462e-07,
+      "loss": 0.7611,
+      "step": 55315
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.1204330921173096,
+      "learning_rate": 2.567785781202375e-07,
+      "loss": 0.75,
+      "step": 55316
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.4091951847076416,
+      "learning_rate": 2.566951778257504e-07,
+      "loss": 0.3907,
+      "step": 55317
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.74123215675354,
+      "learning_rate": 2.5661179072061783e-07,
+      "loss": 0.5513,
+      "step": 55318
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.4037892818450928,
+      "learning_rate": 2.5652841680506967e-07,
+      "loss": 0.5989,
+      "step": 55319
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.6716253757476807,
+      "learning_rate": 2.564450560793402e-07,
+      "loss": 0.5194,
+      "step": 55320
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.7304675579071045,
+      "learning_rate": 2.5636170854365926e-07,
+      "loss": 0.5082,
+      "step": 55321
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.9524054527282715,
+      "learning_rate": 2.562783741982605e-07,
+      "loss": 0.5914,
+      "step": 55322
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.7513911724090576,
+      "learning_rate": 2.561950530433732e-07,
+      "loss": 0.755,
+      "step": 55323
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.972029685974121,
+      "learning_rate": 2.5611174507923163e-07,
+      "loss": 0.6586,
+      "step": 55324
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 4.589922904968262,
+      "learning_rate": 2.5602845030606507e-07,
+      "loss": 0.4131,
+      "step": 55325
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 6.808872699737549,
+      "learning_rate": 2.5594516872410714e-07,
+      "loss": 0.5513,
+      "step": 55326
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.5478482246398926,
+      "learning_rate": 2.558619003335877e-07,
+      "loss": 0.5283,
+      "step": 55327
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.4700770378112793,
+      "learning_rate": 2.5577864513473995e-07,
+      "loss": 0.631,
+      "step": 55328
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 7.013462066650391,
+      "learning_rate": 2.556954031277942e-07,
+      "loss": 0.6998,
+      "step": 55329
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.5647664070129395,
+      "learning_rate": 2.556121743129825e-07,
+      "loss": 0.5386,
+      "step": 55330
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.6815381050109863,
+      "learning_rate": 2.555289586905346e-07,
+      "loss": 0.5626,
+      "step": 55331
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.648932456970215,
+      "learning_rate": 2.554457562606849e-07,
+      "loss": 0.4777,
+      "step": 55332
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 4.422593116760254,
+      "learning_rate": 2.553625670236626e-07,
+      "loss": 0.5275,
+      "step": 55333
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 4.148910045623779,
+      "learning_rate": 2.5527939097970023e-07,
+      "loss": 0.615,
+      "step": 55334
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.6982779502868652,
+      "learning_rate": 2.5519622812902767e-07,
+      "loss": 0.5876,
+      "step": 55335
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.9117183685302734,
+      "learning_rate": 2.5511307847187804e-07,
+      "loss": 0.4392,
+      "step": 55336
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.233151435852051,
+      "learning_rate": 2.550299420084801e-07,
+      "loss": 0.4066,
+      "step": 55337
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.103656530380249,
+      "learning_rate": 2.5494681873906747e-07,
+      "loss": 0.3444,
+      "step": 55338
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.4210715293884277,
+      "learning_rate": 2.5486370866387e-07,
+      "loss": 0.6758,
+      "step": 55339
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.507739543914795,
+      "learning_rate": 2.547806117831192e-07,
+      "loss": 0.4813,
+      "step": 55340
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 4.016443729400635,
+      "learning_rate": 2.5469752809704483e-07,
+      "loss": 0.6227,
+      "step": 55341
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.57757830619812,
+      "learning_rate": 2.5461445760587954e-07,
+      "loss": 0.5591,
+      "step": 55342
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.486142158508301,
+      "learning_rate": 2.545314003098542e-07,
+      "loss": 0.6583,
+      "step": 55343
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.1042566299438477,
+      "learning_rate": 2.5444835620919874e-07,
+      "loss": 0.6641,
+      "step": 55344
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.4926347732543945,
+      "learning_rate": 2.543653253041456e-07,
+      "loss": 0.7107,
+      "step": 55345
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.953474760055542,
+      "learning_rate": 2.5428230759492357e-07,
+      "loss": 0.5372,
+      "step": 55346
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.9241607189178467,
+      "learning_rate": 2.541993030817658e-07,
+      "loss": 0.5194,
+      "step": 55347
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.827974557876587,
+      "learning_rate": 2.5411631176490094e-07,
+      "loss": 0.4449,
+      "step": 55348
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.4626400470733643,
+      "learning_rate": 2.540333336445616e-07,
+      "loss": 0.7089,
+      "step": 55349
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.8907010555267334,
+      "learning_rate": 2.539503687209771e-07,
+      "loss": 0.5683,
+      "step": 55350
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.271939277648926,
+      "learning_rate": 2.538674169943789e-07,
+      "loss": 0.6261,
+      "step": 55351
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.6264286041259766,
+      "learning_rate": 2.537844784649973e-07,
+      "loss": 0.616,
+      "step": 55352
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.432847023010254,
+      "learning_rate": 2.5370155313306334e-07,
+      "loss": 0.8135,
+      "step": 55353
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.97585391998291,
+      "learning_rate": 2.5361864099880675e-07,
+      "loss": 0.6851,
+      "step": 55354
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.5838277339935303,
+      "learning_rate": 2.5353574206245957e-07,
+      "loss": 0.5845,
+      "step": 55355
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.8886969089508057,
+      "learning_rate": 2.534528563242511e-07,
+      "loss": 0.5956,
+      "step": 55356
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.90027117729187,
+      "learning_rate": 2.5336998378441227e-07,
+      "loss": 0.6006,
+      "step": 55357
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.111833333969116,
+      "learning_rate": 2.5328712444317284e-07,
+      "loss": 0.5908,
+      "step": 55358
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 4.110607624053955,
+      "learning_rate": 2.532042783007649e-07,
+      "loss": 0.5789,
+      "step": 55359
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.948568344116211,
+      "learning_rate": 2.5312144535741654e-07,
+      "loss": 0.4259,
+      "step": 55360
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.110703945159912,
+      "learning_rate": 2.530386256133599e-07,
+      "loss": 0.8419,
+      "step": 55361
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.9711380004882812,
+      "learning_rate": 2.5295581906882416e-07,
+      "loss": 0.6372,
+      "step": 55362
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.7693464756011963,
+      "learning_rate": 2.5287302572404024e-07,
+      "loss": 0.5917,
+      "step": 55363
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.1096484661102295,
+      "learning_rate": 2.5279024557923805e-07,
+      "loss": 0.6267,
+      "step": 55364
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.6110241413116455,
+      "learning_rate": 2.5270747863464785e-07,
+      "loss": 0.4991,
+      "step": 55365
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.17293381690979,
+      "learning_rate": 2.526247248904995e-07,
+      "loss": 0.5787,
+      "step": 55366
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 4.094547271728516,
+      "learning_rate": 2.5254198434702395e-07,
+      "loss": 0.8742,
+      "step": 55367
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.050027847290039,
+      "learning_rate": 2.5245925700444986e-07,
+      "loss": 0.3678,
+      "step": 55368
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.8478338718414307,
+      "learning_rate": 2.523765428630093e-07,
+      "loss": 0.4579,
+      "step": 55369
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.0462820529937744,
+      "learning_rate": 2.5229384192292985e-07,
+      "loss": 0.7885,
+      "step": 55370
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 6.27404260635376,
+      "learning_rate": 2.5221115418444297e-07,
+      "loss": 0.5207,
+      "step": 55371
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.2877373695373535,
+      "learning_rate": 2.521284796477785e-07,
+      "loss": 0.4735,
+      "step": 55372
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.8920845985412598,
+      "learning_rate": 2.520458183131663e-07,
+      "loss": 0.6042,
+      "step": 55373
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.403672218322754,
+      "learning_rate": 2.519631701808356e-07,
+      "loss": 0.4985,
+      "step": 55374
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.1942501068115234,
+      "learning_rate": 2.518805352510179e-07,
+      "loss": 0.7523,
+      "step": 55375
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.847827672958374,
+      "learning_rate": 2.5179791352394014e-07,
+      "loss": 0.6479,
+      "step": 55376
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 1.8512227535247803,
+      "learning_rate": 2.51715304999835e-07,
+      "loss": 0.4347,
+      "step": 55377
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.859393358230591,
+      "learning_rate": 2.5163270967893006e-07,
+      "loss": 0.5433,
+      "step": 55378
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.7913405895233154,
+      "learning_rate": 2.5155012756145625e-07,
+      "loss": 0.5938,
+      "step": 55379
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.5436642169952393,
+      "learning_rate": 2.514675586476423e-07,
+      "loss": 0.5946,
+      "step": 55380
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.9541964530944824,
+      "learning_rate": 2.513850029377185e-07,
+      "loss": 0.5802,
+      "step": 55381
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.4882254600524902,
+      "learning_rate": 2.5130246043191365e-07,
+      "loss": 0.6396,
+      "step": 55382
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.999587297439575,
+      "learning_rate": 2.512199311304586e-07,
+      "loss": 0.5775,
+      "step": 55383
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.2598178386688232,
+      "learning_rate": 2.5113741503358105e-07,
+      "loss": 0.5695,
+      "step": 55384
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.2513339519500732,
+      "learning_rate": 2.5105491214151236e-07,
+      "loss": 0.4214,
+      "step": 55385
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.20296573638916,
+      "learning_rate": 2.5097242245448016e-07,
+      "loss": 0.5912,
+      "step": 55386
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.195434808731079,
+      "learning_rate": 2.5088994597271544e-07,
+      "loss": 0.4806,
+      "step": 55387
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 1.6015782356262207,
+      "learning_rate": 2.508074826964457e-07,
+      "loss": 0.4085,
+      "step": 55388
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.7069950103759766,
+      "learning_rate": 2.50725032625902e-07,
+      "loss": 0.3944,
+      "step": 55389
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.641113758087158,
+      "learning_rate": 2.5064259576131233e-07,
+      "loss": 0.2653,
+      "step": 55390
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.266406774520874,
+      "learning_rate": 2.505601721029072e-07,
+      "loss": 0.5065,
+      "step": 55391
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.2503232955932617,
+      "learning_rate": 2.504777616509135e-07,
+      "loss": 0.4651,
+      "step": 55392
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.5909433364868164,
+      "learning_rate": 2.5039536440556344e-07,
+      "loss": 0.5576,
+      "step": 55393
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.6103360652923584,
+      "learning_rate": 2.5031298036708394e-07,
+      "loss": 0.5196,
+      "step": 55394
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.7949845790863037,
+      "learning_rate": 2.5023060953570543e-07,
+      "loss": 0.3548,
+      "step": 55395
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.714806079864502,
+      "learning_rate": 2.501482519116555e-07,
+      "loss": 0.5429,
+      "step": 55396
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.168069362640381,
+      "learning_rate": 2.5006590749516444e-07,
+      "loss": 0.6055,
+      "step": 55397
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.766462802886963,
+      "learning_rate": 2.4998357628646053e-07,
+      "loss": 0.3221,
+      "step": 55398
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.2403106689453125,
+      "learning_rate": 2.499012582857735e-07,
+      "loss": 0.3896,
+      "step": 55399
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.118321180343628,
+      "learning_rate": 2.4981895349333095e-07,
+      "loss": 0.7312,
+      "step": 55400
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 4.393922805786133,
+      "learning_rate": 2.4973666190936274e-07,
+      "loss": 0.4578,
+      "step": 55401
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.5975260734558105,
+      "learning_rate": 2.49654383534097e-07,
+      "loss": 0.6116,
+      "step": 55402
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.89337420463562,
+      "learning_rate": 2.4957211836776353e-07,
+      "loss": 0.5918,
+      "step": 55403
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.0438344478607178,
+      "learning_rate": 2.4948986641058994e-07,
+      "loss": 0.6135,
+      "step": 55404
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.9679226875305176,
+      "learning_rate": 2.494076276628049e-07,
+      "loss": 0.5941,
+      "step": 55405
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.3051202297210693,
+      "learning_rate": 2.4932540212463885e-07,
+      "loss": 0.4596,
+      "step": 55406
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.3613014221191406,
+      "learning_rate": 2.4924318979631823e-07,
+      "loss": 0.7571,
+      "step": 55407
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.269958972930908,
+      "learning_rate": 2.49160990678074e-07,
+      "loss": 0.4294,
+      "step": 55408
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 4.188235759735107,
+      "learning_rate": 2.49078804770132e-07,
+      "loss": 0.726,
+      "step": 55409
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.543142080307007,
+      "learning_rate": 2.4899663207272273e-07,
+      "loss": 0.4166,
+      "step": 55410
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.3417985439300537,
+      "learning_rate": 2.489144725860737e-07,
+      "loss": 0.5671,
+      "step": 55411
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 4.967053413391113,
+      "learning_rate": 2.4883232631041365e-07,
+      "loss": 0.5013,
+      "step": 55412
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.4892783164978027,
+      "learning_rate": 2.487501932459713e-07,
+      "loss": 0.6797,
+      "step": 55413
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.506107807159424,
+      "learning_rate": 2.486680733929747e-07,
+      "loss": 0.5953,
+      "step": 55414
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.6433374881744385,
+      "learning_rate": 2.485859667516521e-07,
+      "loss": 0.5258,
+      "step": 55415
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.4707038402557373,
+      "learning_rate": 2.4850387332223283e-07,
+      "loss": 0.4368,
+      "step": 55416
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.9531075954437256,
+      "learning_rate": 2.484217931049437e-07,
+      "loss": 0.639,
+      "step": 55417
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 4.67140531539917,
+      "learning_rate": 2.4833972610001365e-07,
+      "loss": 0.5465,
+      "step": 55418
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.536045551300049,
+      "learning_rate": 2.4825767230767074e-07,
+      "loss": 0.6029,
+      "step": 55419
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.538604974746704,
+      "learning_rate": 2.4817563172814364e-07,
+      "loss": 0.7898,
+      "step": 55420
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.078963279724121,
+      "learning_rate": 2.480936043616594e-07,
+      "loss": 0.5341,
+      "step": 55421
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.032043933868408,
+      "learning_rate": 2.4801159020844735e-07,
+      "loss": 0.332,
+      "step": 55422
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.8434741497039795,
+      "learning_rate": 2.4792958926873445e-07,
+      "loss": 0.5272,
+      "step": 55423
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.334851026535034,
+      "learning_rate": 2.4784760154275e-07,
+      "loss": 0.5458,
+      "step": 55424
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.2918105125427246,
+      "learning_rate": 2.4776562703071995e-07,
+      "loss": 0.5556,
+      "step": 55425
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.6229920387268066,
+      "learning_rate": 2.476836657328746e-07,
+      "loss": 0.6434,
+      "step": 55426
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.6782376766204834,
+      "learning_rate": 2.4760171764943996e-07,
+      "loss": 0.6601,
+      "step": 55427
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.921578884124756,
+      "learning_rate": 2.4751978278064524e-07,
+      "loss": 0.4642,
+      "step": 55428
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.2542574405670166,
+      "learning_rate": 2.47437861126717e-07,
+      "loss": 0.4934,
+      "step": 55429
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.7572038173675537,
+      "learning_rate": 2.473559526878844e-07,
+      "loss": 0.6402,
+      "step": 55430
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.9399328231811523,
+      "learning_rate": 2.472740574643739e-07,
+      "loss": 0.4681,
+      "step": 55431
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.5587220191955566,
+      "learning_rate": 2.4719217545641384e-07,
+      "loss": 0.6937,
+      "step": 55432
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 4.461311340332031,
+      "learning_rate": 2.4711030666423164e-07,
+      "loss": 0.5461,
+      "step": 55433
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.3436570167541504,
+      "learning_rate": 2.4702845108805606e-07,
+      "loss": 0.3985,
+      "step": 55434
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 5.916017055511475,
+      "learning_rate": 2.469466087281136e-07,
+      "loss": 0.7038,
+      "step": 55435
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.4330034255981445,
+      "learning_rate": 2.468647795846324e-07,
+      "loss": 0.3031,
+      "step": 55436
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.374180555343628,
+      "learning_rate": 2.467829636578389e-07,
+      "loss": 0.5207,
+      "step": 55437
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.6701912879943848,
+      "learning_rate": 2.4670116094796193e-07,
+      "loss": 0.5149,
+      "step": 55438
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.861422538757324,
+      "learning_rate": 2.4661937145522787e-07,
+      "loss": 0.4535,
+      "step": 55439
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 5.992692470550537,
+      "learning_rate": 2.4653759517986543e-07,
+      "loss": 0.7629,
+      "step": 55440
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.749410629272461,
+      "learning_rate": 2.4645583212210056e-07,
+      "loss": 0.4847,
+      "step": 55441
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.36594820022583,
+      "learning_rate": 2.4637408228216254e-07,
+      "loss": 0.6606,
+      "step": 55442
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.5117812156677246,
+      "learning_rate": 2.4629234566027616e-07,
+      "loss": 0.5504,
+      "step": 55443
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.7317702770233154,
+      "learning_rate": 2.462106222566707e-07,
+      "loss": 0.5352,
+      "step": 55444
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.3719444274902344,
+      "learning_rate": 2.461289120715721e-07,
+      "loss": 0.6355,
+      "step": 55445
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.7275118827819824,
+      "learning_rate": 2.4604721510520845e-07,
+      "loss": 0.6074,
+      "step": 55446
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.7947704792022705,
+      "learning_rate": 2.459655313578063e-07,
+      "loss": 0.5989,
+      "step": 55447
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.105821132659912,
+      "learning_rate": 2.458838608295938e-07,
+      "loss": 0.6442,
+      "step": 55448
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.1648778915405273,
+      "learning_rate": 2.4580220352079687e-07,
+      "loss": 0.5707,
+      "step": 55449
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.585402488708496,
+      "learning_rate": 2.457205594316431e-07,
+      "loss": 0.4494,
+      "step": 55450
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.183867931365967,
+      "learning_rate": 2.456389285623595e-07,
+      "loss": 0.3756,
+      "step": 55451
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.227266311645508,
+      "learning_rate": 2.4555731091317316e-07,
+      "loss": 0.4624,
+      "step": 55452
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.1249513626098633,
+      "learning_rate": 2.4547570648430997e-07,
+      "loss": 0.6929,
+      "step": 55453
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.9473438262939453,
+      "learning_rate": 2.4539411527599865e-07,
+      "loss": 0.4895,
+      "step": 55454
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.0586493015289307,
+      "learning_rate": 2.453125372884646e-07,
+      "loss": 0.7579,
+      "step": 55455
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.990203857421875,
+      "learning_rate": 2.4523097252193594e-07,
+      "loss": 0.4411,
+      "step": 55456
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.5610437393188477,
+      "learning_rate": 2.4514942097663805e-07,
+      "loss": 0.7044,
+      "step": 55457
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.354872226715088,
+      "learning_rate": 2.4506788265279967e-07,
+      "loss": 0.7594,
+      "step": 55458
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 4.3966522216796875,
+      "learning_rate": 2.4498635755064505e-07,
+      "loss": 0.5843,
+      "step": 55459
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 4.606323719024658,
+      "learning_rate": 2.4490484567040284e-07,
+      "loss": 0.6461,
+      "step": 55460
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 4.167840003967285,
+      "learning_rate": 2.448233470122985e-07,
+      "loss": 0.5553,
+      "step": 55461
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.3974826335906982,
+      "learning_rate": 2.4474186157655953e-07,
+      "loss": 0.7431,
+      "step": 55462
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.7811269760131836,
+      "learning_rate": 2.446603893634114e-07,
+      "loss": 0.6423,
+      "step": 55463
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.4763851165771484,
+      "learning_rate": 2.4457893037308225e-07,
+      "loss": 0.5756,
+      "step": 55464
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.615664005279541,
+      "learning_rate": 2.4449748460579683e-07,
+      "loss": 0.5551,
+      "step": 55465
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.116868734359741,
+      "learning_rate": 2.444160520617822e-07,
+      "loss": 0.379,
+      "step": 55466
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.4728190898895264,
+      "learning_rate": 2.44334632741266e-07,
+      "loss": 0.3669,
+      "step": 55467
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.856096029281616,
+      "learning_rate": 2.4425322664447305e-07,
+      "loss": 0.7465,
+      "step": 55468
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.1623270511627197,
+      "learning_rate": 2.441718337716309e-07,
+      "loss": 0.722,
+      "step": 55469
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.3630809783935547,
+      "learning_rate": 2.4409045412296437e-07,
+      "loss": 0.5088,
+      "step": 55470
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.113966226577759,
+      "learning_rate": 2.440090876987017e-07,
+      "loss": 0.7986,
+      "step": 55471
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.6621148586273193,
+      "learning_rate": 2.439277344990676e-07,
+      "loss": 0.6739,
+      "step": 55472
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 1.9729002714157104,
+      "learning_rate": 2.438463945242886e-07,
+      "loss": 0.5142,
+      "step": 55473
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.5731875896453857,
+      "learning_rate": 2.4376506777459174e-07,
+      "loss": 0.5111,
+      "step": 55474
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.675899028778076,
+      "learning_rate": 2.4368375425020187e-07,
+      "loss": 0.4498,
+      "step": 55475
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.031891345977783,
+      "learning_rate": 2.436024539513465e-07,
+      "loss": 0.5817,
+      "step": 55476
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.737685203552246,
+      "learning_rate": 2.4352116687825e-07,
+      "loss": 0.5082,
+      "step": 55477
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.4968974590301514,
+      "learning_rate": 2.4343989303113936e-07,
+      "loss": 0.3066,
+      "step": 55478
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.1396923065185547,
+      "learning_rate": 2.43358632410241e-07,
+      "loss": 0.641,
+      "step": 55479
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.791032075881958,
+      "learning_rate": 2.4327738501578047e-07,
+      "loss": 0.6018,
+      "step": 55480
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 4.692330360412598,
+      "learning_rate": 2.4319615084798355e-07,
+      "loss": 0.7566,
+      "step": 55481
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.712567090988159,
+      "learning_rate": 2.431149299070756e-07,
+      "loss": 0.8309,
+      "step": 55482
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.2026255130767822,
+      "learning_rate": 2.4303372219328437e-07,
+      "loss": 0.4445,
+      "step": 55483
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.492732048034668,
+      "learning_rate": 2.4295252770683284e-07,
+      "loss": 0.8374,
+      "step": 55484
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 1.994476318359375,
+      "learning_rate": 2.428713464479493e-07,
+      "loss": 0.3539,
+      "step": 55485
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.2293190956115723,
+      "learning_rate": 2.427901784168579e-07,
+      "loss": 0.6344,
+      "step": 55486
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.0978822708129883,
+      "learning_rate": 2.427090236137858e-07,
+      "loss": 0.4836,
+      "step": 55487
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.011780261993408,
+      "learning_rate": 2.4262788203895715e-07,
+      "loss": 0.3726,
+      "step": 55488
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.033592462539673,
+      "learning_rate": 2.425467536925985e-07,
+      "loss": 0.5531,
+      "step": 55489
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.5218517780303955,
+      "learning_rate": 2.4246563857493466e-07,
+      "loss": 0.6766,
+      "step": 55490
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.28690242767334,
+      "learning_rate": 2.423845366861921e-07,
+      "loss": 0.5796,
+      "step": 55491
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.1469898223876953,
+      "learning_rate": 2.423034480265957e-07,
+      "loss": 0.3702,
+      "step": 55492
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.0130321979522705,
+      "learning_rate": 2.4222237259637073e-07,
+      "loss": 0.6223,
+      "step": 55493
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.326359510421753,
+      "learning_rate": 2.421413103957432e-07,
+      "loss": 0.6906,
+      "step": 55494
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.830888509750366,
+      "learning_rate": 2.42060261424939e-07,
+      "loss": 0.5968,
+      "step": 55495
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.749586343765259,
+      "learning_rate": 2.4197922568418186e-07,
+      "loss": 0.6036,
+      "step": 55496
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 1.8413848876953125,
+      "learning_rate": 2.4189820317369936e-07,
+      "loss": 0.3427,
+      "step": 55497
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.383183240890503,
+      "learning_rate": 2.4181719389371416e-07,
+      "loss": 0.5295,
+      "step": 55498
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 4.0412397384643555,
+      "learning_rate": 2.4173619784445433e-07,
+      "loss": 0.8413,
+      "step": 55499
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.4657740592956543,
+      "learning_rate": 2.416552150261425e-07,
+      "loss": 0.5183,
+      "step": 55500
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 4.772900581359863,
+      "learning_rate": 2.4157424543900575e-07,
+      "loss": 0.6143,
+      "step": 55501
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.030313014984131,
+      "learning_rate": 2.414932890832677e-07,
+      "loss": 0.7739,
+      "step": 55502
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.9952540397644043,
+      "learning_rate": 2.41412345959155e-07,
+      "loss": 0.5983,
+      "step": 55503
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.7424814701080322,
+      "learning_rate": 2.413314160668917e-07,
+      "loss": 0.717,
+      "step": 55504
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.2571604251861572,
+      "learning_rate": 2.412504994067033e-07,
+      "loss": 0.4465,
+      "step": 55505
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.388380527496338,
+      "learning_rate": 2.4116959597881406e-07,
+      "loss": 0.6256,
+      "step": 55506
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.0705509185791016,
+      "learning_rate": 2.410887057834499e-07,
+      "loss": 0.6952,
+      "step": 55507
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.9195337295532227,
+      "learning_rate": 2.410078288208351e-07,
+      "loss": 0.4879,
+      "step": 55508
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.6295342445373535,
+      "learning_rate": 2.4092696509119493e-07,
+      "loss": 0.4302,
+      "step": 55509
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.596738338470459,
+      "learning_rate": 2.4084611459475385e-07,
+      "loss": 0.74,
+      "step": 55510
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 4.310971260070801,
+      "learning_rate": 2.4076527733173704e-07,
+      "loss": 0.4654,
+      "step": 55511
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.6380817890167236,
+      "learning_rate": 2.406844533023689e-07,
+      "loss": 0.6732,
+      "step": 55512
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.5957319736480713,
+      "learning_rate": 2.406036425068753e-07,
+      "loss": 0.6885,
+      "step": 55513
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.6202502250671387,
+      "learning_rate": 2.4052284494547827e-07,
+      "loss": 0.698,
+      "step": 55514
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.7062368392944336,
+      "learning_rate": 2.404420606184055e-07,
+      "loss": 0.6365,
+      "step": 55515
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.958249807357788,
+      "learning_rate": 2.4036128952588e-07,
+      "loss": 0.7423,
+      "step": 55516
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.7258920669555664,
+      "learning_rate": 2.4028053166812783e-07,
+      "loss": 0.829,
+      "step": 55517
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.4398882389068604,
+      "learning_rate": 2.4019978704537096e-07,
+      "loss": 0.5147,
+      "step": 55518
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.9619760513305664,
+      "learning_rate": 2.4011905565783646e-07,
+      "loss": 0.664,
+      "step": 55519
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.337045907974243,
+      "learning_rate": 2.4003833750574746e-07,
+      "loss": 0.4839,
+      "step": 55520
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.5042269229888916,
+      "learning_rate": 2.399576325893288e-07,
+      "loss": 0.3482,
+      "step": 55521
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 1.888747215270996,
+      "learning_rate": 2.3987694090880475e-07,
+      "loss": 0.3144,
+      "step": 55522
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 4.119846343994141,
+      "learning_rate": 2.397962624644001e-07,
+      "loss": 0.6803,
+      "step": 55523
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.123677968978882,
+      "learning_rate": 2.397155972563381e-07,
+      "loss": 0.5544,
+      "step": 55524
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.9561216831207275,
+      "learning_rate": 2.3963494528484455e-07,
+      "loss": 0.6091,
+      "step": 55525
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.720351219177246,
+      "learning_rate": 2.3955430655014213e-07,
+      "loss": 0.5489,
+      "step": 55526
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.6403186321258545,
+      "learning_rate": 2.3947368105245614e-07,
+      "loss": 0.5109,
+      "step": 55527
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 4.7398810386657715,
+      "learning_rate": 2.3939306879201094e-07,
+      "loss": 0.6338,
+      "step": 55528
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.2320334911346436,
+      "learning_rate": 2.393124697690297e-07,
+      "loss": 0.5634,
+      "step": 55529
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.39168119430542,
+      "learning_rate": 2.3923188398373767e-07,
+      "loss": 0.6156,
+      "step": 55530
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.343104600906372,
+      "learning_rate": 2.391513114363575e-07,
+      "loss": 0.5453,
+      "step": 55531
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.6723103523254395,
+      "learning_rate": 2.3907075212711515e-07,
+      "loss": 0.5653,
+      "step": 55532
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.9683878421783447,
+      "learning_rate": 2.3899020605623267e-07,
+      "loss": 0.5948,
+      "step": 55533
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.5291380882263184,
+      "learning_rate": 2.3890967322393486e-07,
+      "loss": 0.4747,
+      "step": 55534
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.8934807777404785,
+      "learning_rate": 2.3882915363044654e-07,
+      "loss": 0.5934,
+      "step": 55535
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.828601360321045,
+      "learning_rate": 2.387486472759898e-07,
+      "loss": 0.8638,
+      "step": 55536
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.366398811340332,
+      "learning_rate": 2.386681541607905e-07,
+      "loss": 0.4453,
+      "step": 55537
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.8815109729766846,
+      "learning_rate": 2.385876742850707e-07,
+      "loss": 0.416,
+      "step": 55538
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.7137701511383057,
+      "learning_rate": 2.385072076490547e-07,
+      "loss": 0.6898,
+      "step": 55539
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.0196316242218018,
+      "learning_rate": 2.3842675425296727e-07,
+      "loss": 0.5968,
+      "step": 55540
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.6876657009124756,
+      "learning_rate": 2.3834631409703047e-07,
+      "loss": 0.4884,
+      "step": 55541
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.1425302028656006,
+      "learning_rate": 2.382658871814697e-07,
+      "loss": 0.5266,
+      "step": 55542
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.155712366104126,
+      "learning_rate": 2.38185473506507e-07,
+      "loss": 0.7054,
+      "step": 55543
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.318087577819824,
+      "learning_rate": 2.3810507307236719e-07,
+      "loss": 0.5309,
+      "step": 55544
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.343174934387207,
+      "learning_rate": 2.3802468587927286e-07,
+      "loss": 0.4866,
+      "step": 55545
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.744577169418335,
+      "learning_rate": 2.3794431192744882e-07,
+      "loss": 0.7719,
+      "step": 55546
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.8151473999023438,
+      "learning_rate": 2.378639512171166e-07,
+      "loss": 0.7395,
+      "step": 55547
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.4981601238250732,
+      "learning_rate": 2.3778360374850207e-07,
+      "loss": 0.3282,
+      "step": 55548
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 4.263980388641357,
+      "learning_rate": 2.377032695218262e-07,
+      "loss": 0.6272,
+      "step": 55549
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.810863494873047,
+      "learning_rate": 2.3762294853731438e-07,
+      "loss": 0.7041,
+      "step": 55550
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.8925704956054688,
+      "learning_rate": 2.3754264079518862e-07,
+      "loss": 0.6273,
+      "step": 55551
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.560969114303589,
+      "learning_rate": 2.374623462956732e-07,
+      "loss": 0.603,
+      "step": 55552
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 4.299163341522217,
+      "learning_rate": 2.373820650389902e-07,
+      "loss": 0.6968,
+      "step": 55553
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.9049580097198486,
+      "learning_rate": 2.3730179702536326e-07,
+      "loss": 0.6345,
+      "step": 55554
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.4912402629852295,
+      "learning_rate": 2.3722154225501614e-07,
+      "loss": 0.5443,
+      "step": 55555
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.214005708694458,
+      "learning_rate": 2.3714130072817254e-07,
+      "loss": 0.6645,
+      "step": 55556
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.766443967819214,
+      "learning_rate": 2.3706107244505393e-07,
+      "loss": 0.7234,
+      "step": 55557
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.5264687538146973,
+      "learning_rate": 2.3698085740588517e-07,
+      "loss": 0.7014,
+      "step": 55558
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.821611166000366,
+      "learning_rate": 2.369006556108877e-07,
+      "loss": 0.4499,
+      "step": 55559
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 6.6730828285217285,
+      "learning_rate": 2.3682046706028582e-07,
+      "loss": 0.5247,
+      "step": 55560
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.5947184562683105,
+      "learning_rate": 2.36740291754301e-07,
+      "loss": 0.644,
+      "step": 55561
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 4.323366165161133,
+      "learning_rate": 2.366601296931581e-07,
+      "loss": 0.6837,
+      "step": 55562
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.896974802017212,
+      "learning_rate": 2.36579980877078e-07,
+      "loss": 0.6827,
+      "step": 55563
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.3538448810577393,
+      "learning_rate": 2.36499845306285e-07,
+      "loss": 0.6054,
+      "step": 55564
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.2491469383239746,
+      "learning_rate": 2.3641972298100113e-07,
+      "loss": 0.6239,
+      "step": 55565
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.8074512481689453,
+      "learning_rate": 2.363396139014501e-07,
+      "loss": 0.5351,
+      "step": 55566
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 6.054932594299316,
+      "learning_rate": 2.3625951806785395e-07,
+      "loss": 0.6746,
+      "step": 55567
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.176497220993042,
+      "learning_rate": 2.3617943548043587e-07,
+      "loss": 0.7795,
+      "step": 55568
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.4851231575012207,
+      "learning_rate": 2.3609936613941731e-07,
+      "loss": 0.7753,
+      "step": 55569
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.5212955474853516,
+      "learning_rate": 2.3601931004502253e-07,
+      "loss": 0.5265,
+      "step": 55570
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.5419840812683105,
+      "learning_rate": 2.3593926719747307e-07,
+      "loss": 0.5956,
+      "step": 55571
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.2271766662597656,
+      "learning_rate": 2.3585923759699203e-07,
+      "loss": 0.7132,
+      "step": 55572
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.062832832336426,
+      "learning_rate": 2.3577922124380148e-07,
+      "loss": 0.5257,
+      "step": 55573
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.4236862659454346,
+      "learning_rate": 2.3569921813812457e-07,
+      "loss": 0.5423,
+      "step": 55574
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 4.893705368041992,
+      "learning_rate": 2.3561922828018224e-07,
+      "loss": 0.6603,
+      "step": 55575
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.2448699474334717,
+      "learning_rate": 2.3553925167019932e-07,
+      "loss": 0.5016,
+      "step": 55576
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.841017961502075,
+      "learning_rate": 2.354592883083967e-07,
+      "loss": 0.7835,
+      "step": 55577
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.117039680480957,
+      "learning_rate": 2.3537933819499703e-07,
+      "loss": 0.538,
+      "step": 55578
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.429952621459961,
+      "learning_rate": 2.3529940133022233e-07,
+      "loss": 0.5624,
+      "step": 55579
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.883469820022583,
+      "learning_rate": 2.3521947771429576e-07,
+      "loss": 0.3825,
+      "step": 55580
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.6475048065185547,
+      "learning_rate": 2.351395673474377e-07,
+      "loss": 0.3658,
+      "step": 55581
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.7825920581817627,
+      "learning_rate": 2.3505967022987242e-07,
+      "loss": 0.4363,
+      "step": 55582
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.5625646114349365,
+      "learning_rate": 2.3497978636182084e-07,
+      "loss": 0.5666,
+      "step": 55583
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.8286497592926025,
+      "learning_rate": 2.3489991574350612e-07,
+      "loss": 0.5551,
+      "step": 55584
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 4.355452537536621,
+      "learning_rate": 2.3482005837514865e-07,
+      "loss": 0.5197,
+      "step": 55585
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.4720511436462402,
+      "learning_rate": 2.3474021425697212e-07,
+      "loss": 0.5952,
+      "step": 55586
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.782038688659668,
+      "learning_rate": 2.3466038338919806e-07,
+      "loss": 0.4093,
+      "step": 55587
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 1.8508270978927612,
+      "learning_rate": 2.3458056577204847e-07,
+      "loss": 0.3078,
+      "step": 55588
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.9871673583984375,
+      "learning_rate": 2.345007614057443e-07,
+      "loss": 0.5193,
+      "step": 55589
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.3275249004364014,
+      "learning_rate": 2.344209702905087e-07,
+      "loss": 0.4479,
+      "step": 55590
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.5894224643707275,
+      "learning_rate": 2.3434119242656373e-07,
+      "loss": 0.4943,
+      "step": 55591
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.49444580078125,
+      "learning_rate": 2.3426142781412974e-07,
+      "loss": 0.3584,
+      "step": 55592
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.8051838874816895,
+      "learning_rate": 2.3418167645342993e-07,
+      "loss": 0.7081,
+      "step": 55593
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.8804664611816406,
+      "learning_rate": 2.341019383446852e-07,
+      "loss": 0.6096,
+      "step": 55594
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 4.515739440917969,
+      "learning_rate": 2.3402221348811815e-07,
+      "loss": 0.5669,
+      "step": 55595
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.6023359298706055,
+      "learning_rate": 2.3394250188394972e-07,
+      "loss": 0.6764,
+      "step": 55596
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.568915605545044,
+      "learning_rate": 2.3386280353240143e-07,
+      "loss": 0.4086,
+      "step": 55597
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.4126689434051514,
+      "learning_rate": 2.3378311843369638e-07,
+      "loss": 0.5929,
+      "step": 55598
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.8760955333709717,
+      "learning_rate": 2.3370344658805388e-07,
+      "loss": 0.778,
+      "step": 55599
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.9999191761016846,
+      "learning_rate": 2.3362378799569652e-07,
+      "loss": 0.4905,
+      "step": 55600
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.7081551551818848,
+      "learning_rate": 2.335441426568469e-07,
+      "loss": 0.6722,
+      "step": 55601
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.279313087463379,
+      "learning_rate": 2.3346451057172427e-07,
+      "loss": 0.628,
+      "step": 55602
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.698744773864746,
+      "learning_rate": 2.3338489174055235e-07,
+      "loss": 0.777,
+      "step": 55603
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.806990146636963,
+      "learning_rate": 2.3330528616355096e-07,
+      "loss": 0.4703,
+      "step": 55604
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.618823766708374,
+      "learning_rate": 2.3322569384094217e-07,
+      "loss": 0.3838,
+      "step": 55605
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.728543281555176,
+      "learning_rate": 2.3314611477294634e-07,
+      "loss": 0.68,
+      "step": 55606
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.5836703777313232,
+      "learning_rate": 2.3306654895978664e-07,
+      "loss": 0.4857,
+      "step": 55607
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.9993886947631836,
+      "learning_rate": 2.3298699640168177e-07,
+      "loss": 0.605,
+      "step": 55608
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.560563087463379,
+      "learning_rate": 2.3290745709885541e-07,
+      "loss": 0.9339,
+      "step": 55609
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.643364906311035,
+      "learning_rate": 2.328279310515269e-07,
+      "loss": 0.3877,
+      "step": 55610
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.450887441635132,
+      "learning_rate": 2.3274841825991878e-07,
+      "loss": 0.5914,
+      "step": 55611
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 6.624809265136719,
+      "learning_rate": 2.3266891872425091e-07,
+      "loss": 0.5974,
+      "step": 55612
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.848881959915161,
+      "learning_rate": 2.325894324447453e-07,
+      "loss": 0.5556,
+      "step": 55613
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 5.28510046005249,
+      "learning_rate": 2.325099594216218e-07,
+      "loss": 0.6854,
+      "step": 55614
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.9160659313201904,
+      "learning_rate": 2.3243049965510244e-07,
+      "loss": 0.441,
+      "step": 55615
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.7645320892333984,
+      "learning_rate": 2.3235105314540763e-07,
+      "loss": 0.7616,
+      "step": 55616
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.7376327514648438,
+      "learning_rate": 2.3227161989275936e-07,
+      "loss": 0.5008,
+      "step": 55617
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.8066844940185547,
+      "learning_rate": 2.3219219989737751e-07,
+      "loss": 0.5307,
+      "step": 55618
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 4.748955726623535,
+      "learning_rate": 2.32112793159483e-07,
+      "loss": 0.6314,
+      "step": 55619
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.3635716438293457,
+      "learning_rate": 2.3203339967929617e-07,
+      "loss": 0.669,
+      "step": 55620
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.2366130352020264,
+      "learning_rate": 2.319540194570391e-07,
+      "loss": 0.5939,
+      "step": 55621
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.5036346912384033,
+      "learning_rate": 2.3187465249293163e-07,
+      "loss": 0.6644,
+      "step": 55622
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.2516133785247803,
+      "learning_rate": 2.3179529878719464e-07,
+      "loss": 0.6276,
+      "step": 55623
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.104259490966797,
+      "learning_rate": 2.31715958340048e-07,
+      "loss": 0.6242,
+      "step": 55624
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.5606601238250732,
+      "learning_rate": 2.3163663115171375e-07,
+      "loss": 0.5521,
+      "step": 55625
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.983139991760254,
+      "learning_rate": 2.3155731722241116e-07,
+      "loss": 0.6651,
+      "step": 55626
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.561114549636841,
+      "learning_rate": 2.3147801655236224e-07,
+      "loss": 0.4368,
+      "step": 55627
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.3378846645355225,
+      "learning_rate": 2.3139872914178573e-07,
+      "loss": 0.6381,
+      "step": 55628
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.6238603591918945,
+      "learning_rate": 2.313194549909037e-07,
+      "loss": 0.5087,
+      "step": 55629
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.3589868545532227,
+      "learning_rate": 2.3124019409993537e-07,
+      "loss": 0.5845,
+      "step": 55630
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.708484172821045,
+      "learning_rate": 2.3116094646910226e-07,
+      "loss": 0.5095,
+      "step": 55631
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.273627758026123,
+      "learning_rate": 2.3108171209862306e-07,
+      "loss": 0.6477,
+      "step": 55632
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.1169233322143555,
+      "learning_rate": 2.310024909887204e-07,
+      "loss": 0.4224,
+      "step": 55633
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.2799570560455322,
+      "learning_rate": 2.3092328313961242e-07,
+      "loss": 0.6107,
+      "step": 55634
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.477701425552368,
+      "learning_rate": 2.3084408855152062e-07,
+      "loss": 0.542,
+      "step": 55635
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.5539207458496094,
+      "learning_rate": 2.307649072246637e-07,
+      "loss": 0.5925,
+      "step": 55636
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.1591901779174805,
+      "learning_rate": 2.3068573915926428e-07,
+      "loss": 0.4765,
+      "step": 55637
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.1603357791900635,
+      "learning_rate": 2.3060658435554107e-07,
+      "loss": 0.3943,
+      "step": 55638
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 4.1126627922058105,
+      "learning_rate": 2.3052744281371443e-07,
+      "loss": 0.5029,
+      "step": 55639
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.692868947982788,
+      "learning_rate": 2.3044831453400364e-07,
+      "loss": 0.6749,
+      "step": 55640
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.5975558757781982,
+      "learning_rate": 2.3036919951663017e-07,
+      "loss": 0.833,
+      "step": 55641
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.765477418899536,
+      "learning_rate": 2.302900977618122e-07,
+      "loss": 0.6633,
+      "step": 55642
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.869908094406128,
+      "learning_rate": 2.302110092697718e-07,
+      "loss": 0.5024,
+      "step": 55643
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.313502550125122,
+      "learning_rate": 2.3013193404072655e-07,
+      "loss": 0.528,
+      "step": 55644
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.435875654220581,
+      "learning_rate": 2.30052872074899e-07,
+      "loss": 0.5396,
+      "step": 55645
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.082568883895874,
+      "learning_rate": 2.2997382337250628e-07,
+      "loss": 0.6311,
+      "step": 55646
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 4.589494228363037,
+      "learning_rate": 2.298947879337704e-07,
+      "loss": 0.726,
+      "step": 55647
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.963148355484009,
+      "learning_rate": 2.298157657589095e-07,
+      "loss": 0.5931,
+      "step": 55648
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.321000337600708,
+      "learning_rate": 2.2973675684814455e-07,
+      "loss": 0.5006,
+      "step": 55649
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.0536649227142334,
+      "learning_rate": 2.2965776120169425e-07,
+      "loss": 0.6175,
+      "step": 55650
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.585055112838745,
+      "learning_rate": 2.295787788197784e-07,
+      "loss": 0.6407,
+      "step": 55651
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.071861505508423,
+      "learning_rate": 2.29499809702618e-07,
+      "loss": 0.5656,
+      "step": 55652
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.7257235050201416,
+      "learning_rate": 2.2942085385043055e-07,
+      "loss": 0.6667,
+      "step": 55653
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.036404609680176,
+      "learning_rate": 2.2934191126343708e-07,
+      "loss": 0.4384,
+      "step": 55654
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.9366865158081055,
+      "learning_rate": 2.2926298194185626e-07,
+      "loss": 0.5239,
+      "step": 55655
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.0044543743133545,
+      "learning_rate": 2.2918406588590792e-07,
+      "loss": 0.4329,
+      "step": 55656
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.6619856357574463,
+      "learning_rate": 2.2910516309581244e-07,
+      "loss": 0.5193,
+      "step": 55657
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.394137144088745,
+      "learning_rate": 2.290262735717874e-07,
+      "loss": 0.5225,
+      "step": 55658
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.1082732677459717,
+      "learning_rate": 2.289473973140538e-07,
+      "loss": 0.5503,
+      "step": 55659
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.0320065021514893,
+      "learning_rate": 2.2886853432282918e-07,
+      "loss": 0.3971,
+      "step": 55660
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.402747392654419,
+      "learning_rate": 2.287896845983345e-07,
+      "loss": 0.5605,
+      "step": 55661
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 5.000169277191162,
+      "learning_rate": 2.2871084814078793e-07,
+      "loss": 0.6996,
+      "step": 55662
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 24.18380355834961,
+      "learning_rate": 2.2863202495040926e-07,
+      "loss": 0.6481,
+      "step": 55663
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.6384799480438232,
+      "learning_rate": 2.2855321502741833e-07,
+      "loss": 0.6033,
+      "step": 55664
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.608630418777466,
+      "learning_rate": 2.2847441837203277e-07,
+      "loss": 0.5277,
+      "step": 55665
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.985989570617676,
+      "learning_rate": 2.2839563498447238e-07,
+      "loss": 0.6602,
+      "step": 55666
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.2551138401031494,
+      "learning_rate": 2.283168648649564e-07,
+      "loss": 0.566,
+      "step": 55667
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.1062381267547607,
+      "learning_rate": 2.2823810801370418e-07,
+      "loss": 0.5567,
+      "step": 55668
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.4402761459350586,
+      "learning_rate": 2.2815936443093322e-07,
+      "loss": 0.5335,
+      "step": 55669
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 4.451872825622559,
+      "learning_rate": 2.2808063411686454e-07,
+      "loss": 0.7823,
+      "step": 55670
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.9169063568115234,
+      "learning_rate": 2.2800191707171515e-07,
+      "loss": 0.4425,
+      "step": 55671
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 4.685306072235107,
+      "learning_rate": 2.2792321329570545e-07,
+      "loss": 0.9186,
+      "step": 55672
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.8225395679473877,
+      "learning_rate": 2.27844522789053e-07,
+      "loss": 0.42,
+      "step": 55673
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 5.266274452209473,
+      "learning_rate": 2.2776584555197767e-07,
+      "loss": 0.4823,
+      "step": 55674
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.735119581222534,
+      "learning_rate": 2.276871815846976e-07,
+      "loss": 0.6003,
+      "step": 55675
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.4050540924072266,
+      "learning_rate": 2.2760853088743152e-07,
+      "loss": 0.9117,
+      "step": 55676
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.7216176986694336,
+      "learning_rate": 2.2752989346039812e-07,
+      "loss": 0.3913,
+      "step": 55677
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.098504066467285,
+      "learning_rate": 2.2745126930381723e-07,
+      "loss": 0.5827,
+      "step": 55678
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.7364747524261475,
+      "learning_rate": 2.273726584179059e-07,
+      "loss": 0.6829,
+      "step": 55679
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 4.054515838623047,
+      "learning_rate": 2.2729406080288395e-07,
+      "loss": 0.6745,
+      "step": 55680
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.776564836502075,
+      "learning_rate": 2.2721547645896847e-07,
+      "loss": 0.5879,
+      "step": 55681
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.329833507537842,
+      "learning_rate": 2.2713690538637977e-07,
+      "loss": 0.8791,
+      "step": 55682
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 4.013696670532227,
+      "learning_rate": 2.270583475853344e-07,
+      "loss": 0.7458,
+      "step": 55683
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.8753821849823,
+      "learning_rate": 2.269798030560527e-07,
+      "loss": 0.7233,
+      "step": 55684
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 6.702651500701904,
+      "learning_rate": 2.269012717987512e-07,
+      "loss": 0.7034,
+      "step": 55685
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 5.378214359283447,
+      "learning_rate": 2.2682275381365026e-07,
+      "loss": 0.4839,
+      "step": 55686
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.1250033378601074,
+      "learning_rate": 2.2674424910096637e-07,
+      "loss": 0.7058,
+      "step": 55687
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.775909185409546,
+      "learning_rate": 2.2666575766091937e-07,
+      "loss": 0.5535,
+      "step": 55688
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.594634532928467,
+      "learning_rate": 2.2658727949372573e-07,
+      "loss": 0.4042,
+      "step": 55689
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 4.400479793548584,
+      "learning_rate": 2.2650881459960583e-07,
+      "loss": 0.4532,
+      "step": 55690
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.405611038208008,
+      "learning_rate": 2.2643036297877562e-07,
+      "loss": 0.5503,
+      "step": 55691
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.439499616622925,
+      "learning_rate": 2.263519246314555e-07,
+      "loss": 0.7534,
+      "step": 55692
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.7709410190582275,
+      "learning_rate": 2.2627349955786137e-07,
+      "loss": 0.691,
+      "step": 55693
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.525979995727539,
+      "learning_rate": 2.261950877582131e-07,
+      "loss": 0.6251,
+      "step": 55694
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.521483898162842,
+      "learning_rate": 2.2611668923272712e-07,
+      "loss": 0.6809,
+      "step": 55695
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 1.8859115839004517,
+      "learning_rate": 2.260383039816233e-07,
+      "loss": 0.2593,
+      "step": 55696
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.2156341075897217,
+      "learning_rate": 2.2595993200511757e-07,
+      "loss": 0.5377,
+      "step": 55697
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.6144261360168457,
+      "learning_rate": 2.2588157330342975e-07,
+      "loss": 0.5603,
+      "step": 55698
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.3042237758636475,
+      "learning_rate": 2.2580322787677577e-07,
+      "loss": 0.5997,
+      "step": 55699
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.564547061920166,
+      "learning_rate": 2.2572489572537605e-07,
+      "loss": 0.6731,
+      "step": 55700
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 5.354365825653076,
+      "learning_rate": 2.256465768494459e-07,
+      "loss": 0.6919,
+      "step": 55701
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.0304019451141357,
+      "learning_rate": 2.2556827124920467e-07,
+      "loss": 0.6574,
+      "step": 55702
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.5055081844329834,
+      "learning_rate": 2.2548997892486878e-07,
+      "loss": 0.3526,
+      "step": 55703
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.537651300430298,
+      "learning_rate": 2.2541169987665755e-07,
+      "loss": 0.4731,
+      "step": 55704
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.118880271911621,
+      "learning_rate": 2.2533343410478746e-07,
+      "loss": 0.3547,
+      "step": 55705
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.0197019577026367,
+      "learning_rate": 2.252551816094767e-07,
+      "loss": 0.6859,
+      "step": 55706
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.936405897140503,
+      "learning_rate": 2.2517694239094168e-07,
+      "loss": 0.716,
+      "step": 55707
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.5612263679504395,
+      "learning_rate": 2.2509871644940228e-07,
+      "loss": 0.5864,
+      "step": 55708
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.4084067344665527,
+      "learning_rate": 2.2502050378507334e-07,
+      "loss": 0.415,
+      "step": 55709
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 6.9987874031066895,
+      "learning_rate": 2.2494230439817467e-07,
+      "loss": 0.6551,
+      "step": 55710
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.409055471420288,
+      "learning_rate": 2.2486411828892164e-07,
+      "loss": 0.7399,
+      "step": 55711
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.1870129108428955,
+      "learning_rate": 2.2478594545753297e-07,
+      "loss": 0.6374,
+      "step": 55712
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.863337755203247,
+      "learning_rate": 2.2470778590422627e-07,
+      "loss": 0.7929,
+      "step": 55713
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.920783281326294,
+      "learning_rate": 2.24629639629218e-07,
+      "loss": 0.5588,
+      "step": 55714
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 5.068840026855469,
+      "learning_rate": 2.245515066327264e-07,
+      "loss": 0.4933,
+      "step": 55715
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.0384361743927,
+      "learning_rate": 2.2447338691496734e-07,
+      "loss": 0.6174,
+      "step": 55716
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.294341564178467,
+      "learning_rate": 2.2439528047615843e-07,
+      "loss": 0.6355,
+      "step": 55717
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 5.377964496612549,
+      "learning_rate": 2.2431718731651842e-07,
+      "loss": 0.5801,
+      "step": 55718
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.474069118499756,
+      "learning_rate": 2.2423910743626264e-07,
+      "loss": 0.6565,
+      "step": 55719
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.6141085624694824,
+      "learning_rate": 2.241610408356093e-07,
+      "loss": 0.7983,
+      "step": 55720
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.2948057651519775,
+      "learning_rate": 2.240829875147743e-07,
+      "loss": 0.5426,
+      "step": 55721
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.9430088996887207,
+      "learning_rate": 2.240049474739764e-07,
+      "loss": 0.783,
+      "step": 55722
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.21832275390625,
+      "learning_rate": 2.239269207134309e-07,
+      "loss": 0.6401,
+      "step": 55723
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.066981315612793,
+      "learning_rate": 2.238489072333555e-07,
+      "loss": 0.6766,
+      "step": 55724
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.3360226154327393,
+      "learning_rate": 2.2377090703396775e-07,
+      "loss": 0.6185,
+      "step": 55725
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.6591031551361084,
+      "learning_rate": 2.2369292011548306e-07,
+      "loss": 0.695,
+      "step": 55726
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.674525022506714,
+      "learning_rate": 2.2361494647811955e-07,
+      "loss": 0.4597,
+      "step": 55727
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.924588203430176,
+      "learning_rate": 2.2353698612209318e-07,
+      "loss": 0.6394,
+      "step": 55728
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 4.127934455871582,
+      "learning_rate": 2.2345903904762212e-07,
+      "loss": 0.544,
+      "step": 55729
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 4.788629055023193,
+      "learning_rate": 2.2338110525492117e-07,
+      "loss": 0.6915,
+      "step": 55730
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.1566905975341797,
+      "learning_rate": 2.2330318474420852e-07,
+      "loss": 0.6883,
+      "step": 55731
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.0329747200012207,
+      "learning_rate": 2.2322527751570012e-07,
+      "loss": 0.6526,
+      "step": 55732
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.852062702178955,
+      "learning_rate": 2.2314738356961297e-07,
+      "loss": 0.5427,
+      "step": 55733
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.1283209323883057,
+      "learning_rate": 2.2306950290616304e-07,
+      "loss": 0.6115,
+      "step": 55734
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.9171180725097656,
+      "learning_rate": 2.2299163552556847e-07,
+      "loss": 0.635,
+      "step": 55735
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.4390017986297607,
+      "learning_rate": 2.2291378142804355e-07,
+      "loss": 0.4138,
+      "step": 55736
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.962703227996826,
+      "learning_rate": 2.2283594061380586e-07,
+      "loss": 0.739,
+      "step": 55737
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.5711634159088135,
+      "learning_rate": 2.2275811308307194e-07,
+      "loss": 0.487,
+      "step": 55738
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 5.087020397186279,
+      "learning_rate": 2.2268029883605823e-07,
+      "loss": 0.4504,
+      "step": 55739
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.13713002204895,
+      "learning_rate": 2.2260249787298127e-07,
+      "loss": 0.7641,
+      "step": 55740
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.6150999069213867,
+      "learning_rate": 2.2252471019405696e-07,
+      "loss": 0.5827,
+      "step": 55741
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.709876298904419,
+      "learning_rate": 2.2244693579950128e-07,
+      "loss": 0.4818,
+      "step": 55742
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 4.301429271697998,
+      "learning_rate": 2.223691746895318e-07,
+      "loss": 0.7783,
+      "step": 55743
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.79506254196167,
+      "learning_rate": 2.2229142686436335e-07,
+      "loss": 0.4788,
+      "step": 55744
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.300396203994751,
+      "learning_rate": 2.2221369232421298e-07,
+      "loss": 0.5889,
+      "step": 55745
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.287794589996338,
+      "learning_rate": 2.2213597106929608e-07,
+      "loss": 0.5174,
+      "step": 55746
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.1372592449188232,
+      "learning_rate": 2.2205826309983026e-07,
+      "loss": 0.7813,
+      "step": 55747
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.516022205352783,
+      "learning_rate": 2.219805684160292e-07,
+      "loss": 0.6242,
+      "step": 55748
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.191955327987671,
+      "learning_rate": 2.2190288701811114e-07,
+      "loss": 0.4187,
+      "step": 55749
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.609635829925537,
+      "learning_rate": 2.2182521890629083e-07,
+      "loss": 0.5384,
+      "step": 55750
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.769667625427246,
+      "learning_rate": 2.2174756408078536e-07,
+      "loss": 0.4239,
+      "step": 55751
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.30885910987854,
+      "learning_rate": 2.2166992254180897e-07,
+      "loss": 0.3671,
+      "step": 55752
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.2745440006256104,
+      "learning_rate": 2.2159229428957928e-07,
+      "loss": 0.603,
+      "step": 55753
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 4.072448253631592,
+      "learning_rate": 2.2151467932431058e-07,
+      "loss": 0.6832,
+      "step": 55754
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 5.291050910949707,
+      "learning_rate": 2.2143707764622045e-07,
+      "loss": 0.5039,
+      "step": 55755
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.2677876949310303,
+      "learning_rate": 2.213594892555232e-07,
+      "loss": 0.5202,
+      "step": 55756
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.8158328533172607,
+      "learning_rate": 2.2128191415243527e-07,
+      "loss": 0.3511,
+      "step": 55757
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.2414331436157227,
+      "learning_rate": 2.2120435233717097e-07,
+      "loss": 0.6551,
+      "step": 55758
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.565643787384033,
+      "learning_rate": 2.2112680380994843e-07,
+      "loss": 0.471,
+      "step": 55759
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.9941418170928955,
+      "learning_rate": 2.2104926857098197e-07,
+      "loss": 0.6472,
+      "step": 55760
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.0076234340667725,
+      "learning_rate": 2.2097174662048748e-07,
+      "loss": 0.6525,
+      "step": 55761
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.357604503631592,
+      "learning_rate": 2.208942379586798e-07,
+      "loss": 0.5053,
+      "step": 55762
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 1.5979210138320923,
+      "learning_rate": 2.2081674258577546e-07,
+      "loss": 0.3488,
+      "step": 55763
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 5.305222034454346,
+      "learning_rate": 2.207392605019887e-07,
+      "loss": 0.5602,
+      "step": 55764
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.906320810317993,
+      "learning_rate": 2.2066179170753655e-07,
+      "loss": 0.4194,
+      "step": 55765
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.867540121078491,
+      "learning_rate": 2.2058433620263277e-07,
+      "loss": 0.5753,
+      "step": 55766
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.019017457962036,
+      "learning_rate": 2.205068939874938e-07,
+      "loss": 0.6834,
+      "step": 55767
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.299039363861084,
+      "learning_rate": 2.2042946506233454e-07,
+      "loss": 0.5719,
+      "step": 55768
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.745439291000366,
+      "learning_rate": 2.2035204942737143e-07,
+      "loss": 0.6585,
+      "step": 55769
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.077573537826538,
+      "learning_rate": 2.2027464708281766e-07,
+      "loss": 0.6877,
+      "step": 55770
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.9836387634277344,
+      "learning_rate": 2.2019725802889026e-07,
+      "loss": 0.5839,
+      "step": 55771
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.8246514797210693,
+      "learning_rate": 2.201198822658035e-07,
+      "loss": 0.5521,
+      "step": 55772
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.1068828105926514,
+      "learning_rate": 2.2004251979377277e-07,
+      "loss": 0.7091,
+      "step": 55773
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.086811065673828,
+      "learning_rate": 2.1996517061301292e-07,
+      "loss": 0.4551,
+      "step": 55774
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.5935158729553223,
+      "learning_rate": 2.198878347237393e-07,
+      "loss": 0.6717,
+      "step": 55775
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.9162116050720215,
+      "learning_rate": 2.1981051212616734e-07,
+      "loss": 0.6532,
+      "step": 55776
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.7095675468444824,
+      "learning_rate": 2.1973320282051125e-07,
+      "loss": 0.4974,
+      "step": 55777
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.8359384536743164,
+      "learning_rate": 2.1965590680698644e-07,
+      "loss": 0.5371,
+      "step": 55778
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.6702797412872314,
+      "learning_rate": 2.1957862408580832e-07,
+      "loss": 0.4286,
+      "step": 55779
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 4.209223747253418,
+      "learning_rate": 2.195013546571906e-07,
+      "loss": 0.599,
+      "step": 55780
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.6336333751678467,
+      "learning_rate": 2.1942409852134915e-07,
+      "loss": 0.4742,
+      "step": 55781
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.7897493839263916,
+      "learning_rate": 2.1934685567849834e-07,
+      "loss": 0.5598,
+      "step": 55782
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.7443857192993164,
+      "learning_rate": 2.192696261288535e-07,
+      "loss": 0.4856,
+      "step": 55783
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.8826634883880615,
+      "learning_rate": 2.1919240987262835e-07,
+      "loss": 0.5254,
+      "step": 55784
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.8610377311706543,
+      "learning_rate": 2.191152069100383e-07,
+      "loss": 0.562,
+      "step": 55785
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.593177556991577,
+      "learning_rate": 2.1903801724129757e-07,
+      "loss": 0.5602,
+      "step": 55786
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.107991933822632,
+      "learning_rate": 2.1896084086662162e-07,
+      "loss": 0.6506,
+      "step": 55787
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.291602849960327,
+      "learning_rate": 2.1888367778622465e-07,
+      "loss": 0.6282,
+      "step": 55788
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.6367838382720947,
+      "learning_rate": 2.1880652800032043e-07,
+      "loss": 0.5978,
+      "step": 55789
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.579334259033203,
+      "learning_rate": 2.1872939150912486e-07,
+      "loss": 0.6624,
+      "step": 55790
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.565675735473633,
+      "learning_rate": 2.1865226831285168e-07,
+      "loss": 0.569,
+      "step": 55791
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.880571126937866,
+      "learning_rate": 2.1857515841171516e-07,
+      "loss": 0.6582,
+      "step": 55792
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.0736756324768066,
+      "learning_rate": 2.1849806180593013e-07,
+      "loss": 0.6492,
+      "step": 55793
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.356555223464966,
+      "learning_rate": 2.1842097849571086e-07,
+      "loss": 0.5281,
+      "step": 55794
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.486100912094116,
+      "learning_rate": 2.1834390848127162e-07,
+      "loss": 0.4468,
+      "step": 55795
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 1.815261721611023,
+      "learning_rate": 2.182668517628267e-07,
+      "loss": 0.4857,
+      "step": 55796
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.963907480239868,
+      "learning_rate": 2.181898083405898e-07,
+      "loss": 0.646,
+      "step": 55797
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.3577663898468018,
+      "learning_rate": 2.1811277821477627e-07,
+      "loss": 0.4686,
+      "step": 55798
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.978715658187866,
+      "learning_rate": 2.1803576138559935e-07,
+      "loss": 0.4565,
+      "step": 55799
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.6376070976257324,
+      "learning_rate": 2.1795875785327435e-07,
+      "loss": 0.5102,
+      "step": 55800
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.074512004852295,
+      "learning_rate": 2.1788176761801395e-07,
+      "loss": 0.8355,
+      "step": 55801
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 4.239931106567383,
+      "learning_rate": 2.1780479068003346e-07,
+      "loss": 0.76,
+      "step": 55802
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.685307025909424,
+      "learning_rate": 2.177278270395461e-07,
+      "loss": 0.5355,
+      "step": 55803
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.2143328189849854,
+      "learning_rate": 2.1765087669676665e-07,
+      "loss": 0.7646,
+      "step": 55804
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.886861562728882,
+      "learning_rate": 2.1757393965190777e-07,
+      "loss": 0.7707,
+      "step": 55805
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.0942482948303223,
+      "learning_rate": 2.1749701590518534e-07,
+      "loss": 0.6155,
+      "step": 55806
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.2946786880493164,
+      "learning_rate": 2.174201054568109e-07,
+      "loss": 0.6364,
+      "step": 55807
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 4.201108455657959,
+      "learning_rate": 2.173432083070004e-07,
+      "loss": 0.665,
+      "step": 55808
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.149873971939087,
+      "learning_rate": 2.172663244559664e-07,
+      "loss": 0.5929,
+      "step": 55809
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.9517178535461426,
+      "learning_rate": 2.1718945390392376e-07,
+      "loss": 0.7595,
+      "step": 55810
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.2855656147003174,
+      "learning_rate": 2.1711259665108454e-07,
+      "loss": 0.6214,
+      "step": 55811
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.1622722148895264,
+      "learning_rate": 2.1703575269766463e-07,
+      "loss": 0.5602,
+      "step": 55812
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.229207992553711,
+      "learning_rate": 2.1695892204387558e-07,
+      "loss": 0.5759,
+      "step": 55813
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.916696071624756,
+      "learning_rate": 2.1688210468993277e-07,
+      "loss": 0.5593,
+      "step": 55814
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 4.888401985168457,
+      "learning_rate": 2.168053006360482e-07,
+      "loss": 0.8011,
+      "step": 55815
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.2989754676818848,
+      "learning_rate": 2.167285098824373e-07,
+      "loss": 0.6613,
+      "step": 55816
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.8071000576019287,
+      "learning_rate": 2.166517324293116e-07,
+      "loss": 0.5459,
+      "step": 55817
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.0367581844329834,
+      "learning_rate": 2.1657496827688584e-07,
+      "loss": 0.3711,
+      "step": 55818
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.845273971557617,
+      "learning_rate": 2.164982174253727e-07,
+      "loss": 0.4636,
+      "step": 55819
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.913975715637207,
+      "learning_rate": 2.1642147987498696e-07,
+      "loss": 0.453,
+      "step": 55820
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.897520065307617,
+      "learning_rate": 2.1634475562594015e-07,
+      "loss": 0.5342,
+      "step": 55821
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.7775659561157227,
+      "learning_rate": 2.1626804467844765e-07,
+      "loss": 0.3605,
+      "step": 55822
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.1071994304656982,
+      "learning_rate": 2.1619134703272093e-07,
+      "loss": 0.6064,
+      "step": 55823
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 5.327869892120361,
+      "learning_rate": 2.1611466268897484e-07,
+      "loss": 0.6447,
+      "step": 55824
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.0678882598876953,
+      "learning_rate": 2.1603799164742088e-07,
+      "loss": 0.808,
+      "step": 55825
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.2915499210357666,
+      "learning_rate": 2.159613339082739e-07,
+      "loss": 0.5467,
+      "step": 55826
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 4.125021457672119,
+      "learning_rate": 2.1588468947174535e-07,
+      "loss": 0.6863,
+      "step": 55827
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.254911184310913,
+      "learning_rate": 2.1580805833805008e-07,
+      "loss": 0.5317,
+      "step": 55828
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 5.299112319946289,
+      "learning_rate": 2.1573144050740013e-07,
+      "loss": 0.3224,
+      "step": 55829
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 4.912501335144043,
+      "learning_rate": 2.1565483598000925e-07,
+      "loss": 0.4986,
+      "step": 55830
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.4519550800323486,
+      "learning_rate": 2.1557824475608947e-07,
+      "loss": 0.55,
+      "step": 55831
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.8203608989715576,
+      "learning_rate": 2.1550166683585505e-07,
+      "loss": 0.7009,
+      "step": 55832
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.2857186794281006,
+      "learning_rate": 2.1542510221951696e-07,
+      "loss": 0.4716,
+      "step": 55833
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.4355685710906982,
+      "learning_rate": 2.1534855090729057e-07,
+      "loss": 0.3613,
+      "step": 55834
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.707610845565796,
+      "learning_rate": 2.1527201289938627e-07,
+      "loss": 0.7315,
+      "step": 55835
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.9750306606292725,
+      "learning_rate": 2.1519548819601887e-07,
+      "loss": 0.6412,
+      "step": 55836
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 4.25416898727417,
+      "learning_rate": 2.1511897679740047e-07,
+      "loss": 0.4791,
+      "step": 55837
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.0770599842071533,
+      "learning_rate": 2.1504247870374307e-07,
+      "loss": 0.6154,
+      "step": 55838
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.8258731365203857,
+      "learning_rate": 2.1496599391526041e-07,
+      "loss": 0.5555,
+      "step": 55839
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.337204694747925,
+      "learning_rate": 2.148895224321651e-07,
+      "loss": 0.3922,
+      "step": 55840
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.7192070484161377,
+      "learning_rate": 2.148130642546692e-07,
+      "loss": 0.741,
+      "step": 55841
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.8322997093200684,
+      "learning_rate": 2.1473661938298583e-07,
+      "loss": 0.3823,
+      "step": 55842
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.4208593368530273,
+      "learning_rate": 2.146601878173271e-07,
+      "loss": 0.516,
+      "step": 55843
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.6286232471466064,
+      "learning_rate": 2.1458376955790616e-07,
+      "loss": 0.6143,
+      "step": 55844
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 4.049505233764648,
+      "learning_rate": 2.145073646049345e-07,
+      "loss": 0.8797,
+      "step": 55845
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.298150062561035,
+      "learning_rate": 2.144309729586258e-07,
+      "loss": 0.5969,
+      "step": 55846
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.179178476333618,
+      "learning_rate": 2.1435459461919106e-07,
+      "loss": 0.6296,
+      "step": 55847
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.9695801734924316,
+      "learning_rate": 2.1427822958684342e-07,
+      "loss": 0.6026,
+      "step": 55848
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.1060497760772705,
+      "learning_rate": 2.1420187786179602e-07,
+      "loss": 0.2985,
+      "step": 55849
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.9704651832580566,
+      "learning_rate": 2.1412553944425985e-07,
+      "loss": 0.9892,
+      "step": 55850
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.9416263103485107,
+      "learning_rate": 2.1404921433444857e-07,
+      "loss": 0.4778,
+      "step": 55851
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.957195281982422,
+      "learning_rate": 2.1397290253257262e-07,
+      "loss": 0.855,
+      "step": 55852
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.3592875003814697,
+      "learning_rate": 2.1389660403884572e-07,
+      "loss": 0.6109,
+      "step": 55853
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 4.246585369110107,
+      "learning_rate": 2.1382031885347877e-07,
+      "loss": 0.5029,
+      "step": 55854
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.4558486938476562,
+      "learning_rate": 2.137440469766855e-07,
+      "loss": 0.3788,
+      "step": 55855
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.8770053386688232,
+      "learning_rate": 2.136677884086763e-07,
+      "loss": 0.7009,
+      "step": 55856
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.0257046222686768,
+      "learning_rate": 2.135915431496649e-07,
+      "loss": 0.3793,
+      "step": 55857
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.423757314682007,
+      "learning_rate": 2.1351531119986114e-07,
+      "loss": 0.39,
+      "step": 55858
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 1.9597207307815552,
+      "learning_rate": 2.134390925594787e-07,
+      "loss": 0.3928,
+      "step": 55859
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.4913394451141357,
+      "learning_rate": 2.1336288722872856e-07,
+      "loss": 0.5035,
+      "step": 55860
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 4.808740139007568,
+      "learning_rate": 2.132866952078244e-07,
+      "loss": 0.6456,
+      "step": 55861
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.8907387256622314,
+      "learning_rate": 2.1321051649697555e-07,
+      "loss": 0.6219,
+      "step": 55862
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.408094644546509,
+      "learning_rate": 2.1313435109639625e-07,
+      "loss": 0.318,
+      "step": 55863
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.167947292327881,
+      "learning_rate": 2.1305819900629632e-07,
+      "loss": 0.8107,
+      "step": 55864
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.9369754791259766,
+      "learning_rate": 2.1298206022688895e-07,
+      "loss": 0.5406,
+      "step": 55865
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.2786760330200195,
+      "learning_rate": 2.129059347583845e-07,
+      "loss": 0.4897,
+      "step": 55866
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.2540528774261475,
+      "learning_rate": 2.1282982260099616e-07,
+      "loss": 0.3468,
+      "step": 55867
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.2019927501678467,
+      "learning_rate": 2.1275372375493376e-07,
+      "loss": 0.6072,
+      "step": 55868
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.024796962738037,
+      "learning_rate": 2.12677638220411e-07,
+      "loss": 0.6351,
+      "step": 55869
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.593855857849121,
+      "learning_rate": 2.1260156599763715e-07,
+      "loss": 0.4425,
+      "step": 55870
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.6262247562408447,
+      "learning_rate": 2.1252550708682595e-07,
+      "loss": 0.495,
+      "step": 55871
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 5.278550624847412,
+      "learning_rate": 2.1244946148818723e-07,
+      "loss": 0.4267,
+      "step": 55872
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.972789764404297,
+      "learning_rate": 2.123734292019336e-07,
+      "loss": 0.727,
+      "step": 55873
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.777306079864502,
+      "learning_rate": 2.1229741022827545e-07,
+      "loss": 0.4917,
+      "step": 55874
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.1619303226470947,
+      "learning_rate": 2.1222140456742535e-07,
+      "loss": 0.5637,
+      "step": 55875
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.6052324771881104,
+      "learning_rate": 2.1214541221959373e-07,
+      "loss": 0.7739,
+      "step": 55876
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.733093023300171,
+      "learning_rate": 2.1206943318499208e-07,
+      "loss": 0.5952,
+      "step": 55877
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.6186041831970215,
+      "learning_rate": 2.119934674638313e-07,
+      "loss": 0.5121,
+      "step": 55878
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.529843330383301,
+      "learning_rate": 2.1191751505632408e-07,
+      "loss": 0.5028,
+      "step": 55879
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 5.073066711425781,
+      "learning_rate": 2.1184157596267906e-07,
+      "loss": 0.7131,
+      "step": 55880
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.118000030517578,
+      "learning_rate": 2.1176565018311001e-07,
+      "loss": 0.4155,
+      "step": 55881
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.7801694869995117,
+      "learning_rate": 2.1168973771782674e-07,
+      "loss": 0.8001,
+      "step": 55882
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.6717281341552734,
+      "learning_rate": 2.1161383856704075e-07,
+      "loss": 0.3516,
+      "step": 55883
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.666404962539673,
+      "learning_rate": 2.1153795273096244e-07,
+      "loss": 0.6341,
+      "step": 55884
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.324763059616089,
+      "learning_rate": 2.1146208020980441e-07,
+      "loss": 0.6161,
+      "step": 55885
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.885571002960205,
+      "learning_rate": 2.1138622100377538e-07,
+      "loss": 0.4853,
+      "step": 55886
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.280841827392578,
+      "learning_rate": 2.1131037511308794e-07,
+      "loss": 0.6599,
+      "step": 55887
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.3342671394348145,
+      "learning_rate": 2.1123454253795195e-07,
+      "loss": 0.6755,
+      "step": 55888
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.6115331649780273,
+      "learning_rate": 2.1115872327857944e-07,
+      "loss": 0.4255,
+      "step": 55889
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 4.447134017944336,
+      "learning_rate": 2.1108291733518028e-07,
+      "loss": 0.5813,
+      "step": 55890
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.686235189437866,
+      "learning_rate": 2.1100712470796536e-07,
+      "loss": 0.7399,
+      "step": 55891
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.1438329219818115,
+      "learning_rate": 2.1093134539714567e-07,
+      "loss": 0.7113,
+      "step": 55892
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.7579638957977295,
+      "learning_rate": 2.1085557940293267e-07,
+      "loss": 0.6582,
+      "step": 55893
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.4593138694763184,
+      "learning_rate": 2.107798267255351e-07,
+      "loss": 0.632,
+      "step": 55894
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.8906891345977783,
+      "learning_rate": 2.1070408736516557e-07,
+      "loss": 0.4802,
+      "step": 55895
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.906653642654419,
+      "learning_rate": 2.1062836132203333e-07,
+      "loss": 0.7224,
+      "step": 55896
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.163069248199463,
+      "learning_rate": 2.1055264859634938e-07,
+      "loss": 0.412,
+      "step": 55897
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.770939588546753,
+      "learning_rate": 2.1047694918832516e-07,
+      "loss": 0.4501,
+      "step": 55898
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.9740285873413086,
+      "learning_rate": 2.1040126309816943e-07,
+      "loss": 0.5105,
+      "step": 55899
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.2858288288116455,
+      "learning_rate": 2.1032559032609368e-07,
+      "loss": 0.5725,
+      "step": 55900
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.9707632064819336,
+      "learning_rate": 2.1024993087230884e-07,
+      "loss": 0.6574,
+      "step": 55901
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.3118574619293213,
+      "learning_rate": 2.1017428473702416e-07,
+      "loss": 0.4699,
+      "step": 55902
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.329840898513794,
+      "learning_rate": 2.1009865192045065e-07,
+      "loss": 0.7125,
+      "step": 55903
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.6920394897460938,
+      "learning_rate": 2.100230324227981e-07,
+      "loss": 0.5806,
+      "step": 55904
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.3259572982788086,
+      "learning_rate": 2.09947426244278e-07,
+      "loss": 0.6628,
+      "step": 55905
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 4.5660624504089355,
+      "learning_rate": 2.098718333850991e-07,
+      "loss": 0.5121,
+      "step": 55906
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.562072992324829,
+      "learning_rate": 2.0979625384547231e-07,
+      "loss": 0.7055,
+      "step": 55907
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.81402587890625,
+      "learning_rate": 2.0972068762560748e-07,
+      "loss": 0.6873,
+      "step": 55908
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.94956374168396,
+      "learning_rate": 2.0964513472571502e-07,
+      "loss": 0.9019,
+      "step": 55909
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.1386146545410156,
+      "learning_rate": 2.0956959514600527e-07,
+      "loss": 0.5164,
+      "step": 55910
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 8.739718437194824,
+      "learning_rate": 2.0949406888668755e-07,
+      "loss": 0.69,
+      "step": 55911
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 4.682912349700928,
+      "learning_rate": 2.0941855594797278e-07,
+      "loss": 0.3508,
+      "step": 55912
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.5483031272888184,
+      "learning_rate": 2.0934305633006967e-07,
+      "loss": 0.7281,
+      "step": 55913
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.684968948364258,
+      "learning_rate": 2.0926757003318976e-07,
+      "loss": 0.4684,
+      "step": 55914
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 5.034478664398193,
+      "learning_rate": 2.0919209705754116e-07,
+      "loss": 0.8248,
+      "step": 55915
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 4.713203430175781,
+      "learning_rate": 2.0911663740333542e-07,
+      "loss": 0.8372,
+      "step": 55916
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 5.030672073364258,
+      "learning_rate": 2.0904119107078125e-07,
+      "loss": 0.6462,
+      "step": 55917
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.6541197299957275,
+      "learning_rate": 2.08965758060089e-07,
+      "loss": 0.4699,
+      "step": 55918
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.8781964778900146,
+      "learning_rate": 2.08890338371468e-07,
+      "loss": 0.5817,
+      "step": 55919
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.714083194732666,
+      "learning_rate": 2.0881493200512804e-07,
+      "loss": 0.4708,
+      "step": 55920
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.5961766242980957,
+      "learning_rate": 2.0873953896127842e-07,
+      "loss": 0.5425,
+      "step": 55921
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.0864925384521484,
+      "learning_rate": 2.0866415924013063e-07,
+      "loss": 0.3745,
+      "step": 55922
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.151116132736206,
+      "learning_rate": 2.085887928418917e-07,
+      "loss": 0.4228,
+      "step": 55923
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.2961697578430176,
+      "learning_rate": 2.0851343976677373e-07,
+      "loss": 0.8617,
+      "step": 55924
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.4010865688323975,
+      "learning_rate": 2.0843810001498377e-07,
+      "loss": 0.5628,
+      "step": 55925
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.494478940963745,
+      "learning_rate": 2.0836277358673329e-07,
+      "loss": 0.6199,
+      "step": 55926
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.0560898780822754,
+      "learning_rate": 2.0828746048223049e-07,
+      "loss": 0.3805,
+      "step": 55927
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.6364824771881104,
+      "learning_rate": 2.082121607016857e-07,
+      "loss": 0.492,
+      "step": 55928
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 4.202820301055908,
+      "learning_rate": 2.0813687424530715e-07,
+      "loss": 0.5735,
+      "step": 55929
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.8125109672546387,
+      "learning_rate": 2.0806160111330575e-07,
+      "loss": 0.7581,
+      "step": 55930
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.482355833053589,
+      "learning_rate": 2.0798634130588913e-07,
+      "loss": 0.445,
+      "step": 55931
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.8438291549682617,
+      "learning_rate": 2.079110948232682e-07,
+      "loss": 0.6735,
+      "step": 55932
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.7559163570404053,
+      "learning_rate": 2.0783586166565062e-07,
+      "loss": 0.5187,
+      "step": 55933
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.8994827270507812,
+      "learning_rate": 2.0776064183324674e-07,
+      "loss": 0.5041,
+      "step": 55934
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.8028512001037598,
+      "learning_rate": 2.0768543532626472e-07,
+      "loss": 0.5297,
+      "step": 55935
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.702363967895508,
+      "learning_rate": 2.0761024214491555e-07,
+      "loss": 0.7591,
+      "step": 55936
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.687526226043701,
+      "learning_rate": 2.0753506228940567e-07,
+      "loss": 0.4764,
+      "step": 55937
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.50715708732605,
+      "learning_rate": 2.0745989575994607e-07,
+      "loss": 0.502,
+      "step": 55938
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 4.878893852233887,
+      "learning_rate": 2.073847425567449e-07,
+      "loss": 0.8813,
+      "step": 55939
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.4551687240600586,
+      "learning_rate": 2.0730960268001198e-07,
+      "loss": 0.6004,
+      "step": 55940
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.243922472000122,
+      "learning_rate": 2.072344761299544e-07,
+      "loss": 0.3895,
+      "step": 55941
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 4.8044023513793945,
+      "learning_rate": 2.0715936290678363e-07,
+      "loss": 0.4877,
+      "step": 55942
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 7.482945919036865,
+      "learning_rate": 2.0708426301070616e-07,
+      "loss": 0.5125,
+      "step": 55943
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.433622121810913,
+      "learning_rate": 2.0700917644193296e-07,
+      "loss": 0.5551,
+      "step": 55944
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.1366770267486572,
+      "learning_rate": 2.069341032006711e-07,
+      "loss": 0.86,
+      "step": 55945
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.4713118076324463,
+      "learning_rate": 2.0685904328713036e-07,
+      "loss": 0.7065,
+      "step": 55946
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.896730422973633,
+      "learning_rate": 2.0678399670151895e-07,
+      "loss": 0.5184,
+      "step": 55947
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.9808764457702637,
+      "learning_rate": 2.0670896344404556e-07,
+      "loss": 0.6126,
+      "step": 55948
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.5750341415405273,
+      "learning_rate": 2.0663394351491895e-07,
+      "loss": 0.5233,
+      "step": 55949
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.4241182804107666,
+      "learning_rate": 2.0655893691434782e-07,
+      "loss": 0.4806,
+      "step": 55950
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.9790682792663574,
+      "learning_rate": 2.064839436425403e-07,
+      "loss": 0.5634,
+      "step": 55951
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.9324581623077393,
+      "learning_rate": 2.0640896369970576e-07,
+      "loss": 0.6997,
+      "step": 55952
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.668034553527832,
+      "learning_rate": 2.0633399708605173e-07,
+      "loss": 0.5992,
+      "step": 55953
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.664088010787964,
+      "learning_rate": 2.062590438017875e-07,
+      "loss": 0.606,
+      "step": 55954
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.2862868309020996,
+      "learning_rate": 2.0618410384712074e-07,
+      "loss": 0.5731,
+      "step": 55955
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.4327359199523926,
+      "learning_rate": 2.0610917722226064e-07,
+      "loss": 0.6343,
+      "step": 55956
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.9397499561309814,
+      "learning_rate": 2.0603426392741431e-07,
+      "loss": 0.5065,
+      "step": 55957
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 6.027556896209717,
+      "learning_rate": 2.0595936396279158e-07,
+      "loss": 0.7126,
+      "step": 55958
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.341784715652466,
+      "learning_rate": 2.058844773285995e-07,
+      "loss": 0.2323,
+      "step": 55959
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.943760395050049,
+      "learning_rate": 2.058096040250468e-07,
+      "loss": 0.3371,
+      "step": 55960
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.5348050594329834,
+      "learning_rate": 2.0573474405234161e-07,
+      "loss": 0.6403,
+      "step": 55961
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.36142635345459,
+      "learning_rate": 2.0565989741069269e-07,
+      "loss": 0.5886,
+      "step": 55962
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.5609254837036133,
+      "learning_rate": 2.0558506410030653e-07,
+      "loss": 0.6256,
+      "step": 55963
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.344558000564575,
+      "learning_rate": 2.0551024412139354e-07,
+      "loss": 0.5699,
+      "step": 55964
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.0156826972961426,
+      "learning_rate": 2.054354374741596e-07,
+      "loss": 0.5413,
+      "step": 55965
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.4669110774993896,
+      "learning_rate": 2.0536064415881406e-07,
+      "loss": 0.5549,
+      "step": 55966
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.7371864318847656,
+      "learning_rate": 2.052858641755645e-07,
+      "loss": 0.64,
+      "step": 55967
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.4854791164398193,
+      "learning_rate": 2.0521109752461854e-07,
+      "loss": 0.5219,
+      "step": 55968
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.8582394123077393,
+      "learning_rate": 2.0513634420618433e-07,
+      "loss": 0.4362,
+      "step": 55969
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.347886800765991,
+      "learning_rate": 2.0506160422047006e-07,
+      "loss": 0.4273,
+      "step": 55970
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.5539822578430176,
+      "learning_rate": 2.0498687756768276e-07,
+      "loss": 0.767,
+      "step": 55971
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.718017578125,
+      "learning_rate": 2.0491216424803117e-07,
+      "loss": 0.645,
+      "step": 55972
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 5.662323474884033,
+      "learning_rate": 2.0483746426172234e-07,
+      "loss": 0.5312,
+      "step": 55973
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.676711320877075,
+      "learning_rate": 2.047627776089639e-07,
+      "loss": 0.5312,
+      "step": 55974
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.705231189727783,
+      "learning_rate": 2.0468810428996454e-07,
+      "loss": 0.7632,
+      "step": 55975
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.978316068649292,
+      "learning_rate": 2.046134443049308e-07,
+      "loss": 0.7335,
+      "step": 55976
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.9337992668151855,
+      "learning_rate": 2.0453879765407136e-07,
+      "loss": 0.4809,
+      "step": 55977
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 1.9141019582748413,
+      "learning_rate": 2.044641643375922e-07,
+      "loss": 0.4323,
+      "step": 55978
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.54872727394104,
+      "learning_rate": 2.0438954435570202e-07,
+      "loss": 0.3951,
+      "step": 55979
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.0619564056396484,
+      "learning_rate": 2.0431493770860788e-07,
+      "loss": 0.5261,
+      "step": 55980
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.161785364151001,
+      "learning_rate": 2.0424034439651742e-07,
+      "loss": 0.5931,
+      "step": 55981
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.697033643722534,
+      "learning_rate": 2.0416576441963764e-07,
+      "loss": 0.5125,
+      "step": 55982
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.121565103530884,
+      "learning_rate": 2.0409119777817732e-07,
+      "loss": 0.5516,
+      "step": 55983
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.004378080368042,
+      "learning_rate": 2.040166444723418e-07,
+      "loss": 0.5329,
+      "step": 55984
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.0885002613067627,
+      "learning_rate": 2.0394210450234043e-07,
+      "loss": 0.4758,
+      "step": 55985
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.9840965270996094,
+      "learning_rate": 2.0386757786837853e-07,
+      "loss": 0.5424,
+      "step": 55986
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.4377825260162354,
+      "learning_rate": 2.0379306457066483e-07,
+      "loss": 0.2535,
+      "step": 55987
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.053645610809326,
+      "learning_rate": 2.0371856460940532e-07,
+      "loss": 0.6442,
+      "step": 55988
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.1653218269348145,
+      "learning_rate": 2.036440779848087e-07,
+      "loss": 0.5356,
+      "step": 55989
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 1.9616713523864746,
+      "learning_rate": 2.035696046970803e-07,
+      "loss": 0.3401,
+      "step": 55990
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.923156976699829,
+      "learning_rate": 2.0349514474642839e-07,
+      "loss": 0.7311,
+      "step": 55991
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.6950066089630127,
+      "learning_rate": 2.0342069813305942e-07,
+      "loss": 0.6301,
+      "step": 55992
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 1.9130325317382812,
+      "learning_rate": 2.0334626485718102e-07,
+      "loss": 0.5281,
+      "step": 55993
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.0680410861968994,
+      "learning_rate": 2.0327184491899966e-07,
+      "loss": 0.6015,
+      "step": 55994
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.6965482234954834,
+      "learning_rate": 2.031974383187224e-07,
+      "loss": 0.5192,
+      "step": 55995
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.718737840652466,
+      "learning_rate": 2.0312304505655577e-07,
+      "loss": 0.6102,
+      "step": 55996
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.876988410949707,
+      "learning_rate": 2.030486651327074e-07,
+      "loss": 0.7725,
+      "step": 55997
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.7669992446899414,
+      "learning_rate": 2.029742985473837e-07,
+      "loss": 0.7009,
+      "step": 55998
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.3914103507995605,
+      "learning_rate": 2.028999453007918e-07,
+      "loss": 0.5762,
+      "step": 55999
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.9452264308929443,
+      "learning_rate": 2.0282560539313712e-07,
+      "loss": 0.4635,
+      "step": 56000
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.103466272354126,
+      "learning_rate": 2.027512788246283e-07,
+      "loss": 0.4772,
+      "step": 56001
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 5.3799262046813965,
+      "learning_rate": 2.0267696559546968e-07,
+      "loss": 0.5741,
+      "step": 56002
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.269221544265747,
+      "learning_rate": 2.0260266570587105e-07,
+      "loss": 0.4385,
+      "step": 56003
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.6502652168273926,
+      "learning_rate": 2.0252837915603617e-07,
+      "loss": 0.6958,
+      "step": 56004
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.0179176330566406,
+      "learning_rate": 2.024541059461732e-07,
+      "loss": 0.594,
+      "step": 56005
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.69313907623291,
+      "learning_rate": 2.0237984607648754e-07,
+      "loss": 0.5745,
+      "step": 56006
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.004962205886841,
+      "learning_rate": 2.0230559954718732e-07,
+      "loss": 0.5033,
+      "step": 56007
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.9599008560180664,
+      "learning_rate": 2.022313663584774e-07,
+      "loss": 0.5027,
+      "step": 56008
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 4.348636150360107,
+      "learning_rate": 2.0215714651056483e-07,
+      "loss": 0.5608,
+      "step": 56009
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.540496349334717,
+      "learning_rate": 2.0208294000365557e-07,
+      "loss": 0.4477,
+      "step": 56010
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.4627575874328613,
+      "learning_rate": 2.0200874683795723e-07,
+      "loss": 0.4407,
+      "step": 56011
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.4703474044799805,
+      "learning_rate": 2.0193456701367408e-07,
+      "loss": 0.5127,
+      "step": 56012
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.560234308242798,
+      "learning_rate": 2.018604005310143e-07,
+      "loss": 0.4708,
+      "step": 56013
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.2622241973876953,
+      "learning_rate": 2.017862473901827e-07,
+      "loss": 0.4332,
+      "step": 56014
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.6316637992858887,
+      "learning_rate": 2.0171210759138693e-07,
+      "loss": 0.4159,
+      "step": 56015
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.538637638092041,
+      "learning_rate": 2.016379811348318e-07,
+      "loss": 0.6032,
+      "step": 56016
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.0655291080474854,
+      "learning_rate": 2.0156386802072437e-07,
+      "loss": 0.2768,
+      "step": 56017
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.5846025943756104,
+      "learning_rate": 2.0148976824926948e-07,
+      "loss": 0.4638,
+      "step": 56018
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.544828176498413,
+      "learning_rate": 2.0141568182067473e-07,
+      "loss": 0.5394,
+      "step": 56019
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.6795666217803955,
+      "learning_rate": 2.013416087351444e-07,
+      "loss": 0.7852,
+      "step": 56020
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.218144416809082,
+      "learning_rate": 2.0126754899288615e-07,
+      "loss": 0.8072,
+      "step": 56021
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.5388176441192627,
+      "learning_rate": 2.0119350259410474e-07,
+      "loss": 0.8835,
+      "step": 56022
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.8606057167053223,
+      "learning_rate": 2.011194695390073e-07,
+      "loss": 0.5674,
+      "step": 56023
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.9426815509796143,
+      "learning_rate": 2.0104544982779805e-07,
+      "loss": 0.4462,
+      "step": 56024
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.8613646030426025,
+      "learning_rate": 2.0097144346068465e-07,
+      "loss": 0.5878,
+      "step": 56025
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.321983575820923,
+      "learning_rate": 2.008974504378708e-07,
+      "loss": 0.3695,
+      "step": 56026
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.119875907897949,
+      "learning_rate": 2.0082347075956465e-07,
+      "loss": 0.5177,
+      "step": 56027
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.640597343444824,
+      "learning_rate": 2.0074950442596942e-07,
+      "loss": 0.4137,
+      "step": 56028
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.045863151550293,
+      "learning_rate": 2.0067555143729267e-07,
+      "loss": 0.4708,
+      "step": 56029
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.2189722061157227,
+      "learning_rate": 2.0060161179373928e-07,
+      "loss": 0.7568,
+      "step": 56030
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.941190719604492,
+      "learning_rate": 2.0052768549551515e-07,
+      "loss": 1.1557,
+      "step": 56031
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.907531976699829,
+      "learning_rate": 2.004537725428246e-07,
+      "loss": 0.5161,
+      "step": 56032
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.4123971462249756,
+      "learning_rate": 2.003798729358747e-07,
+      "loss": 0.4799,
+      "step": 56033
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 5.764449119567871,
+      "learning_rate": 2.0030598667487077e-07,
+      "loss": 0.8263,
+      "step": 56034
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.529597043991089,
+      "learning_rate": 2.0023211376001718e-07,
+      "loss": 0.4204,
+      "step": 56035
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.57189679145813,
+      "learning_rate": 2.001582541915209e-07,
+      "loss": 0.5465,
+      "step": 56036
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.7815892696380615,
+      "learning_rate": 2.0008440796958572e-07,
+      "loss": 0.6071,
+      "step": 56037
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.881884813308716,
+      "learning_rate": 2.0001057509441812e-07,
+      "loss": 0.5634,
+      "step": 56038
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.722384452819824,
+      "learning_rate": 1.9993675556622293e-07,
+      "loss": 0.6004,
+      "step": 56039
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.3969266414642334,
+      "learning_rate": 1.9986294938520557e-07,
+      "loss": 0.6909,
+      "step": 56040
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.766878128051758,
+      "learning_rate": 1.9978915655157084e-07,
+      "loss": 0.7235,
+      "step": 56041
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.1892178058624268,
+      "learning_rate": 1.9971537706552414e-07,
+      "loss": 0.677,
+      "step": 56042
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.1127243041992188,
+      "learning_rate": 1.9964161092727142e-07,
+      "loss": 0.4739,
+      "step": 56043
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 4.058660984039307,
+      "learning_rate": 1.995678581370164e-07,
+      "loss": 0.7075,
+      "step": 56044
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.8809120655059814,
+      "learning_rate": 1.9949411869496505e-07,
+      "loss": 0.4809,
+      "step": 56045
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.4987964630126953,
+      "learning_rate": 1.9942039260132273e-07,
+      "loss": 0.7425,
+      "step": 56046
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.2260568141937256,
+      "learning_rate": 1.9934667985629374e-07,
+      "loss": 0.8004,
+      "step": 56047
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.8570358753204346,
+      "learning_rate": 1.992729804600835e-07,
+      "loss": 0.5508,
+      "step": 56048
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 6.711363315582275,
+      "learning_rate": 1.991992944128962e-07,
+      "loss": 0.3791,
+      "step": 56049
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.621720552444458,
+      "learning_rate": 1.9912562171493732e-07,
+      "loss": 0.7065,
+      "step": 56050
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.4162795543670654,
+      "learning_rate": 1.9905196236641168e-07,
+      "loss": 0.3241,
+      "step": 56051
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.707561492919922,
+      "learning_rate": 1.9897831636752464e-07,
+      "loss": 0.5532,
+      "step": 56052
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.275871753692627,
+      "learning_rate": 1.9890468371847937e-07,
+      "loss": 0.6083,
+      "step": 56053
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.6381728649139404,
+      "learning_rate": 1.9883106441948296e-07,
+      "loss": 0.6655,
+      "step": 56054
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.310652732849121,
+      "learning_rate": 1.9875745847073745e-07,
+      "loss": 0.3205,
+      "step": 56055
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 4.171785354614258,
+      "learning_rate": 1.986838658724499e-07,
+      "loss": 0.5559,
+      "step": 56056
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.1928560733795166,
+      "learning_rate": 1.9861028662482297e-07,
+      "loss": 0.5947,
+      "step": 56057
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.704737424850464,
+      "learning_rate": 1.9853672072806308e-07,
+      "loss": 0.9788,
+      "step": 56058
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.889624834060669,
+      "learning_rate": 1.9846316818237344e-07,
+      "loss": 0.5178,
+      "step": 56059
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.772644519805908,
+      "learning_rate": 1.9838962898795945e-07,
+      "loss": 0.5016,
+      "step": 56060
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 4.277254581451416,
+      "learning_rate": 1.9831610314502426e-07,
+      "loss": 0.7828,
+      "step": 56061
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.4394240379333496,
+      "learning_rate": 1.982425906537744e-07,
+      "loss": 0.6782,
+      "step": 56062
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.4317209720611572,
+      "learning_rate": 1.9816909151441133e-07,
+      "loss": 0.7186,
+      "step": 56063
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 1.7328171730041504,
+      "learning_rate": 1.980956057271427e-07,
+      "loss": 0.2808,
+      "step": 56064
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 1.8758982419967651,
+      "learning_rate": 1.9802213329217057e-07,
+      "loss": 0.2179,
+      "step": 56065
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.5720560550689697,
+      "learning_rate": 1.9794867420970088e-07,
+      "loss": 0.7392,
+      "step": 56066
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.7362594604492188,
+      "learning_rate": 1.9787522847993678e-07,
+      "loss": 0.5129,
+      "step": 56067
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 4.917361259460449,
+      "learning_rate": 1.978017961030826e-07,
+      "loss": 0.677,
+      "step": 56068
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.0571653842926025,
+      "learning_rate": 1.9772837707934256e-07,
+      "loss": 0.4551,
+      "step": 56069
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.816617012023926,
+      "learning_rate": 1.9765497140892155e-07,
+      "loss": 0.4613,
+      "step": 56070
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.9386425018310547,
+      "learning_rate": 1.9758157909202215e-07,
+      "loss": 0.4191,
+      "step": 56071
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.447673797607422,
+      "learning_rate": 1.975082001288503e-07,
+      "loss": 0.6388,
+      "step": 56072
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 5.495072364807129,
+      "learning_rate": 1.974348345196081e-07,
+      "loss": 0.4146,
+      "step": 56073
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 5.294228553771973,
+      "learning_rate": 1.9736148226450146e-07,
+      "loss": 0.6847,
+      "step": 56074
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.748739004135132,
+      "learning_rate": 1.9728814336373247e-07,
+      "loss": 0.497,
+      "step": 56075
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.425090789794922,
+      "learning_rate": 1.9721481781750705e-07,
+      "loss": 0.4449,
+      "step": 56076
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.658761978149414,
+      "learning_rate": 1.971415056260273e-07,
+      "loss": 0.4306,
+      "step": 56077
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.052018404006958,
+      "learning_rate": 1.97068206789498e-07,
+      "loss": 0.4369,
+      "step": 56078
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.5901730060577393,
+      "learning_rate": 1.9699492130812293e-07,
+      "loss": 0.7872,
+      "step": 56079
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 5.62396764755249,
+      "learning_rate": 1.9692164918210578e-07,
+      "loss": 0.6321,
+      "step": 56080
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 8.756369590759277,
+      "learning_rate": 1.9684839041165026e-07,
+      "loss": 0.4927,
+      "step": 56081
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.446857452392578,
+      "learning_rate": 1.9677514499696017e-07,
+      "loss": 0.6472,
+      "step": 56082
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.6211767196655273,
+      "learning_rate": 1.9670191293823805e-07,
+      "loss": 0.4134,
+      "step": 56083
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 4.78054141998291,
+      "learning_rate": 1.966286942356893e-07,
+      "loss": 0.5181,
+      "step": 56084
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 4.151687145233154,
+      "learning_rate": 1.965554888895166e-07,
+      "loss": 0.6104,
+      "step": 56085
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 5.212197780609131,
+      "learning_rate": 1.9648229689992416e-07,
+      "loss": 0.5213,
+      "step": 56086
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.1970958709716797,
+      "learning_rate": 1.9640911826711462e-07,
+      "loss": 0.468,
+      "step": 56087
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.9381353855133057,
+      "learning_rate": 1.9633595299129227e-07,
+      "loss": 0.6659,
+      "step": 56088
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.401728630065918,
+      "learning_rate": 1.962628010726597e-07,
+      "loss": 0.5649,
+      "step": 56089
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.3544325828552246,
+      "learning_rate": 1.9618966251142124e-07,
+      "loss": 0.6203,
+      "step": 56090
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.0061919689178467,
+      "learning_rate": 1.9611653730777892e-07,
+      "loss": 0.7364,
+      "step": 56091
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.514266014099121,
+      "learning_rate": 1.960434254619381e-07,
+      "loss": 0.4555,
+      "step": 56092
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.660670042037964,
+      "learning_rate": 1.959703269740998e-07,
+      "loss": 0.5914,
+      "step": 56093
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.603668212890625,
+      "learning_rate": 1.958972418444688e-07,
+      "loss": 0.6309,
+      "step": 56094
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.378392457962036,
+      "learning_rate": 1.958241700732483e-07,
+      "loss": 0.7534,
+      "step": 56095
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.6238207817077637,
+      "learning_rate": 1.9575111166064033e-07,
+      "loss": 0.5983,
+      "step": 56096
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.736254930496216,
+      "learning_rate": 1.9567806660684973e-07,
+      "loss": 0.6331,
+      "step": 56097
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 4.155040740966797,
+      "learning_rate": 1.9560503491207806e-07,
+      "loss": 0.687,
+      "step": 56098
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 4.549703598022461,
+      "learning_rate": 1.9553201657652955e-07,
+      "loss": 0.69,
+      "step": 56099
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.711595296859741,
+      "learning_rate": 1.954590116004057e-07,
+      "loss": 0.5181,
+      "step": 56100
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 5.069757461547852,
+      "learning_rate": 1.9538601998391137e-07,
+      "loss": 0.6301,
+      "step": 56101
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.9025094509124756,
+      "learning_rate": 1.9531304172724863e-07,
+      "loss": 0.5289,
+      "step": 56102
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.521812915802002,
+      "learning_rate": 1.9524007683061952e-07,
+      "loss": 0.6273,
+      "step": 56103
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 1.9008445739746094,
+      "learning_rate": 1.9516712529422943e-07,
+      "loss": 0.5062,
+      "step": 56104
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.0477819442749023,
+      "learning_rate": 1.950941871182782e-07,
+      "loss": 0.8494,
+      "step": 56105
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.628561019897461,
+      "learning_rate": 1.950212623029707e-07,
+      "loss": 0.6703,
+      "step": 56106
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.8824236392974854,
+      "learning_rate": 1.9494835084850893e-07,
+      "loss": 0.5949,
+      "step": 56107
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.249720811843872,
+      "learning_rate": 1.9487545275509612e-07,
+      "loss": 0.5262,
+      "step": 56108
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.4972267150878906,
+      "learning_rate": 1.9480256802293429e-07,
+      "loss": 0.4516,
+      "step": 56109
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.330618143081665,
+      "learning_rate": 1.9472969665222663e-07,
+      "loss": 0.5258,
+      "step": 56110
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.8258190155029297,
+      "learning_rate": 1.9465683864317574e-07,
+      "loss": 0.5015,
+      "step": 56111
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.5279488563537598,
+      "learning_rate": 1.945839939959837e-07,
+      "loss": 0.7265,
+      "step": 56112
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.478513479232788,
+      "learning_rate": 1.9451116271085368e-07,
+      "loss": 0.5794,
+      "step": 56113
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.5206758975982666,
+      "learning_rate": 1.9443834478798772e-07,
+      "loss": 0.6395,
+      "step": 56114
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.870262861251831,
+      "learning_rate": 1.9436554022758902e-07,
+      "loss": 0.5987,
+      "step": 56115
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.4356637001037598,
+      "learning_rate": 1.9429274902985907e-07,
+      "loss": 0.5589,
+      "step": 56116
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.0792229175567627,
+      "learning_rate": 1.9421997119500102e-07,
+      "loss": 0.3377,
+      "step": 56117
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.4844045639038086,
+      "learning_rate": 1.941472067232164e-07,
+      "loss": 0.4745,
+      "step": 56118
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.734565019607544,
+      "learning_rate": 1.9407445561470838e-07,
+      "loss": 0.6143,
+      "step": 56119
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.008491039276123,
+      "learning_rate": 1.9400171786967903e-07,
+      "loss": 0.6219,
+      "step": 56120
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 7.729611396789551,
+      "learning_rate": 1.9392899348833095e-07,
+      "loss": 0.6851,
+      "step": 56121
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.9221222400665283,
+      "learning_rate": 1.938562824708651e-07,
+      "loss": 0.5748,
+      "step": 56122
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.969942569732666,
+      "learning_rate": 1.9378358481748517e-07,
+      "loss": 0.5578,
+      "step": 56123
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.615326404571533,
+      "learning_rate": 1.9371090052839158e-07,
+      "loss": 0.5027,
+      "step": 56124
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.309654474258423,
+      "learning_rate": 1.9363822960378863e-07,
+      "loss": 0.6234,
+      "step": 56125
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.3650147914886475,
+      "learning_rate": 1.9356557204387616e-07,
+      "loss": 0.6225,
+      "step": 56126
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.78114914894104,
+      "learning_rate": 1.9349292784885843e-07,
+      "loss": 0.5506,
+      "step": 56127
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.18021297454834,
+      "learning_rate": 1.9342029701893583e-07,
+      "loss": 0.5963,
+      "step": 56128
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.167097568511963,
+      "learning_rate": 1.93347679554311e-07,
+      "loss": 0.7089,
+      "step": 56129
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.077739715576172,
+      "learning_rate": 1.9327507545518542e-07,
+      "loss": 0.7083,
+      "step": 56130
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.8527417182922363,
+      "learning_rate": 1.9320248472176172e-07,
+      "loss": 0.4562,
+      "step": 56131
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.151672840118408,
+      "learning_rate": 1.9312990735424087e-07,
+      "loss": 0.3787,
+      "step": 56132
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.6619763374328613,
+      "learning_rate": 1.9305734335282544e-07,
+      "loss": 0.5913,
+      "step": 56133
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.8363704681396484,
+      "learning_rate": 1.9298479271771586e-07,
+      "loss": 0.5172,
+      "step": 56134
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.566727638244629,
+      "learning_rate": 1.9291225544911585e-07,
+      "loss": 0.6933,
+      "step": 56135
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.9134411811828613,
+      "learning_rate": 1.9283973154722524e-07,
+      "loss": 0.6458,
+      "step": 56136
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 4.132694244384766,
+      "learning_rate": 1.927672210122472e-07,
+      "loss": 0.5524,
+      "step": 56137
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.6169323921203613,
+      "learning_rate": 1.9269472384438215e-07,
+      "loss": 0.3396,
+      "step": 56138
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.6450350284576416,
+      "learning_rate": 1.9262224004383267e-07,
+      "loss": 0.7987,
+      "step": 56139
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.7930397987365723,
+      "learning_rate": 1.9254976961079974e-07,
+      "loss": 0.5979,
+      "step": 56140
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 4.432360649108887,
+      "learning_rate": 1.9247731254548485e-07,
+      "loss": 0.5619,
+      "step": 56141
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 1.7902188301086426,
+      "learning_rate": 1.9240486884808952e-07,
+      "loss": 0.303,
+      "step": 56142
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.778451442718506,
+      "learning_rate": 1.9233243851881579e-07,
+      "loss": 0.6308,
+      "step": 56143
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.726308822631836,
+      "learning_rate": 1.922600215578635e-07,
+      "loss": 0.5375,
+      "step": 56144
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.1418135166168213,
+      "learning_rate": 1.9218761796543583e-07,
+      "loss": 0.6712,
+      "step": 56145
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.783464193344116,
+      "learning_rate": 1.921152277417332e-07,
+      "loss": 0.4525,
+      "step": 56146
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.7451889514923096,
+      "learning_rate": 1.9204285088695761e-07,
+      "loss": 0.6331,
+      "step": 56147
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.58717679977417,
+      "learning_rate": 1.9197048740130897e-07,
+      "loss": 0.2874,
+      "step": 56148
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.6041407585144043,
+      "learning_rate": 1.9189813728498986e-07,
+      "loss": 0.5038,
+      "step": 56149
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.5703084468841553,
+      "learning_rate": 1.918258005382001e-07,
+      "loss": 0.7168,
+      "step": 56150
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.987917423248291,
+      "learning_rate": 1.9175347716114235e-07,
+      "loss": 0.6637,
+      "step": 56151
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.8426952362060547,
+      "learning_rate": 1.916811671540164e-07,
+      "loss": 0.71,
+      "step": 56152
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.714015483856201,
+      "learning_rate": 1.9160887051702492e-07,
+      "loss": 0.4288,
+      "step": 56153
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.8840043544769287,
+      "learning_rate": 1.9153658725036662e-07,
+      "loss": 0.6381,
+      "step": 56154
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.503606081008911,
+      "learning_rate": 1.9146431735424465e-07,
+      "loss": 0.4463,
+      "step": 56155
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.7081081867218018,
+      "learning_rate": 1.9139206082885885e-07,
+      "loss": 0.4718,
+      "step": 56156
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.869436502456665,
+      "learning_rate": 1.9131981767440966e-07,
+      "loss": 0.4861,
+      "step": 56157
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.7771143913269043,
+      "learning_rate": 1.9124758789109965e-07,
+      "loss": 0.6969,
+      "step": 56158
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.6989572048187256,
+      "learning_rate": 1.9117537147912812e-07,
+      "loss": 0.3479,
+      "step": 56159
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.7076056003570557,
+      "learning_rate": 1.9110316843869714e-07,
+      "loss": 0.613,
+      "step": 56160
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.955432891845703,
+      "learning_rate": 1.9103097877000654e-07,
+      "loss": 0.6615,
+      "step": 56161
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.745142698287964,
+      "learning_rate": 1.9095880247325727e-07,
+      "loss": 0.5749,
+      "step": 56162
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.360896348953247,
+      "learning_rate": 1.9088663954864973e-07,
+      "loss": 0.6759,
+      "step": 56163
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.1963553428649902,
+      "learning_rate": 1.9081448999638484e-07,
+      "loss": 0.5602,
+      "step": 56164
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.3511128425598145,
+      "learning_rate": 1.9074235381666362e-07,
+      "loss": 0.7819,
+      "step": 56165
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.352604627609253,
+      "learning_rate": 1.906702310096864e-07,
+      "loss": 0.3805,
+      "step": 56166
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.569307804107666,
+      "learning_rate": 1.9059812157565305e-07,
+      "loss": 0.5034,
+      "step": 56167
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.365746021270752,
+      "learning_rate": 1.9052602551476562e-07,
+      "loss": 0.6371,
+      "step": 56168
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.9528605937957764,
+      "learning_rate": 1.9045394282722285e-07,
+      "loss": 0.6528,
+      "step": 56169
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.8862111568450928,
+      "learning_rate": 1.903818735132268e-07,
+      "loss": 0.7157,
+      "step": 56170
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.3290770053863525,
+      "learning_rate": 1.9030981757297616e-07,
+      "loss": 0.6543,
+      "step": 56171
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.9586985111236572,
+      "learning_rate": 1.9023777500667252e-07,
+      "loss": 0.5108,
+      "step": 56172
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.191260576248169,
+      "learning_rate": 1.9016574581451563e-07,
+      "loss": 0.458,
+      "step": 56173
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.040436029434204,
+      "learning_rate": 1.9009372999670706e-07,
+      "loss": 0.3115,
+      "step": 56174
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.3442623615264893,
+      "learning_rate": 1.900217275534444e-07,
+      "loss": 0.5806,
+      "step": 56175
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.4874143600463867,
+      "learning_rate": 1.8994973848493082e-07,
+      "loss": 0.7363,
+      "step": 56176
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.494717597961426,
+      "learning_rate": 1.898777627913645e-07,
+      "loss": 0.5918,
+      "step": 56177
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.2258336544036865,
+      "learning_rate": 1.898058004729464e-07,
+      "loss": 0.6241,
+      "step": 56178
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.5193705558776855,
+      "learning_rate": 1.8973385152987577e-07,
+      "loss": 0.5295,
+      "step": 56179
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.2937402725219727,
+      "learning_rate": 1.8966191596235416e-07,
+      "loss": 0.5704,
+      "step": 56180
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.149315118789673,
+      "learning_rate": 1.895899937705803e-07,
+      "loss": 0.4377,
+      "step": 56181
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.8711788654327393,
+      "learning_rate": 1.895180849547551e-07,
+      "loss": 0.6465,
+      "step": 56182
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.5189590454101562,
+      "learning_rate": 1.894461895150773e-07,
+      "loss": 0.5884,
+      "step": 56183
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.805004358291626,
+      "learning_rate": 1.8937430745174844e-07,
+      "loss": 0.3967,
+      "step": 56184
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.8248987197875977,
+      "learning_rate": 1.893024387649667e-07,
+      "loss": 0.534,
+      "step": 56185
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.8871068954467773,
+      "learning_rate": 1.892305834549335e-07,
+      "loss": 0.6399,
+      "step": 56186
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.620737075805664,
+      "learning_rate": 1.891587415218471e-07,
+      "loss": 0.5806,
+      "step": 56187
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.724302291870117,
+      "learning_rate": 1.8908691296590897e-07,
+      "loss": 0.3517,
+      "step": 56188
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.74219012260437,
+      "learning_rate": 1.8901509778731674e-07,
+      "loss": 0.4085,
+      "step": 56189
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.3689634799957275,
+      "learning_rate": 1.8894329598627247e-07,
+      "loss": 0.582,
+      "step": 56190
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 4.474024295806885,
+      "learning_rate": 1.888715075629738e-07,
+      "loss": 0.4705,
+      "step": 56191
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.020005226135254,
+      "learning_rate": 1.8879973251762218e-07,
+      "loss": 0.5994,
+      "step": 56192
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.6104648113250732,
+      "learning_rate": 1.8872797085041527e-07,
+      "loss": 0.5422,
+      "step": 56193
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.538361072540283,
+      "learning_rate": 1.8865622256155402e-07,
+      "loss": 0.4919,
+      "step": 56194
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 4.584949016571045,
+      "learning_rate": 1.8858448765123716e-07,
+      "loss": 0.434,
+      "step": 56195
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 4.083983898162842,
+      "learning_rate": 1.8851276611966507e-07,
+      "loss": 0.4344,
+      "step": 56196
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.3535704612731934,
+      "learning_rate": 1.8844105796703537e-07,
+      "loss": 0.5984,
+      "step": 56197
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.7047595977783203,
+      "learning_rate": 1.883693631935496e-07,
+      "loss": 0.4763,
+      "step": 56198
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.5538463592529297,
+      "learning_rate": 1.882976817994059e-07,
+      "loss": 0.7431,
+      "step": 56199
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.8952410221099854,
+      "learning_rate": 1.8822601378480409e-07,
+      "loss": 0.3903,
+      "step": 56200
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.619842767715454,
+      "learning_rate": 1.881543591499424e-07,
+      "loss": 0.6048,
+      "step": 56201
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.689587354660034,
+      "learning_rate": 1.8808271789502174e-07,
+      "loss": 0.39,
+      "step": 56202
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.7852602005004883,
+      "learning_rate": 1.8801109002023976e-07,
+      "loss": 0.5372,
+      "step": 56203
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.5799994468688965,
+      "learning_rate": 1.8793947552579684e-07,
+      "loss": 0.4719,
+      "step": 56204
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.3074097633361816,
+      "learning_rate": 1.878678744118906e-07,
+      "loss": 0.4358,
+      "step": 56205
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.705932855606079,
+      "learning_rate": 1.8779628667872253e-07,
+      "loss": 0.6343,
+      "step": 56206
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.4458889961242676,
+      "learning_rate": 1.8772471232648915e-07,
+      "loss": 0.3675,
+      "step": 56207
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.889345407485962,
+      "learning_rate": 1.8765315135539142e-07,
+      "loss": 0.5534,
+      "step": 56208
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.541839361190796,
+      "learning_rate": 1.875816037656275e-07,
+      "loss": 0.5629,
+      "step": 56209
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.932110548019409,
+      "learning_rate": 1.8751006955739615e-07,
+      "loss": 0.5919,
+      "step": 56210
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.5317516326904297,
+      "learning_rate": 1.8743854873089606e-07,
+      "loss": 0.5135,
+      "step": 56211
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.8914921283721924,
+      "learning_rate": 1.8736704128632766e-07,
+      "loss": 0.4268,
+      "step": 56212
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 5.998629093170166,
+      "learning_rate": 1.8729554722388744e-07,
+      "loss": 0.513,
+      "step": 56213
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.3376972675323486,
+      "learning_rate": 1.8722406654377634e-07,
+      "loss": 0.5301,
+      "step": 56214
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 7.006629943847656,
+      "learning_rate": 1.8715259924619146e-07,
+      "loss": 0.6975,
+      "step": 56215
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.8757033348083496,
+      "learning_rate": 1.8708114533133314e-07,
+      "loss": 0.6788,
+      "step": 56216
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.5705344676971436,
+      "learning_rate": 1.870097047993985e-07,
+      "loss": 0.6172,
+      "step": 56217
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.3349530696868896,
+      "learning_rate": 1.8693827765058682e-07,
+      "loss": 0.513,
+      "step": 56218
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.2431910037994385,
+      "learning_rate": 1.8686686388509734e-07,
+      "loss": 0.4574,
+      "step": 56219
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.8233838081359863,
+      "learning_rate": 1.867954635031277e-07,
+      "loss": 0.4253,
+      "step": 56220
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.309142589569092,
+      "learning_rate": 1.8672407650487723e-07,
+      "loss": 0.6463,
+      "step": 56221
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.7847015857696533,
+      "learning_rate": 1.8665270289054294e-07,
+      "loss": 0.4758,
+      "step": 56222
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.396710157394409,
+      "learning_rate": 1.8658134266032578e-07,
+      "loss": 0.4608,
+      "step": 56223
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.768333673477173,
+      "learning_rate": 1.8650999581442176e-07,
+      "loss": 0.6268,
+      "step": 56224
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.5783133506774902,
+      "learning_rate": 1.864386623530301e-07,
+      "loss": 0.6652,
+      "step": 56225
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.2998220920562744,
+      "learning_rate": 1.8636734227635012e-07,
+      "loss": 0.3908,
+      "step": 56226
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.082155704498291,
+      "learning_rate": 1.8629603558457887e-07,
+      "loss": 0.4591,
+      "step": 56227
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.077427625656128,
+      "learning_rate": 1.8622474227791564e-07,
+      "loss": 0.629,
+      "step": 56228
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.8525314331054688,
+      "learning_rate": 1.861534623565575e-07,
+      "loss": 0.4649,
+      "step": 56229
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.915562868118286,
+      "learning_rate": 1.860821958207032e-07,
+      "loss": 0.8883,
+      "step": 56230
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.1551976203918457,
+      "learning_rate": 1.8601094267055142e-07,
+      "loss": 0.5323,
+      "step": 56231
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.3737926483154297,
+      "learning_rate": 1.8593970290629925e-07,
+      "loss": 0.5855,
+      "step": 56232
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.879767656326294,
+      "learning_rate": 1.85868476528146e-07,
+      "loss": 0.6095,
+      "step": 56233
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.1322412490844727,
+      "learning_rate": 1.8579726353628867e-07,
+      "loss": 0.5727,
+      "step": 56234
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.530100107192993,
+      "learning_rate": 1.8572606393092663e-07,
+      "loss": 0.617,
+      "step": 56235
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.591144561767578,
+      "learning_rate": 1.8565487771225577e-07,
+      "loss": 0.67,
+      "step": 56236
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.536888599395752,
+      "learning_rate": 1.8558370488047594e-07,
+      "loss": 0.6253,
+      "step": 56237
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.7392020225524902,
+      "learning_rate": 1.8551254543578367e-07,
+      "loss": 0.3893,
+      "step": 56238
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.656513214111328,
+      "learning_rate": 1.8544139937837824e-07,
+      "loss": 0.5471,
+      "step": 56239
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.318843126296997,
+      "learning_rate": 1.8537026670845614e-07,
+      "loss": 0.643,
+      "step": 56240
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.162139415740967,
+      "learning_rate": 1.8529914742621613e-07,
+      "loss": 0.7564,
+      "step": 56241
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.152050256729126,
+      "learning_rate": 1.8522804153185526e-07,
+      "loss": 0.5343,
+      "step": 56242
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.74821400642395,
+      "learning_rate": 1.8515694902557168e-07,
+      "loss": 0.6681,
+      "step": 56243
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 6.347071647644043,
+      "learning_rate": 1.8508586990756305e-07,
+      "loss": 0.8402,
+      "step": 56244
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.202155590057373,
+      "learning_rate": 1.8501480417802697e-07,
+      "loss": 0.3624,
+      "step": 56245
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.490307569503784,
+      "learning_rate": 1.8494375183715995e-07,
+      "loss": 0.5403,
+      "step": 56246
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.0448386669158936,
+      "learning_rate": 1.848727128851613e-07,
+      "loss": 0.534,
+      "step": 56247
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.456508159637451,
+      "learning_rate": 1.8480168732222803e-07,
+      "loss": 0.4844,
+      "step": 56248
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.155442953109741,
+      "learning_rate": 1.8473067514855724e-07,
+      "loss": 0.5368,
+      "step": 56249
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.2880308628082275,
+      "learning_rate": 1.8465967636434657e-07,
+      "loss": 0.4282,
+      "step": 56250
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.804628849029541,
+      "learning_rate": 1.8458869096979415e-07,
+      "loss": 0.4932,
+      "step": 56251
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 7.290932655334473,
+      "learning_rate": 1.8451771896509597e-07,
+      "loss": 0.6635,
+      "step": 56252
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.138005256652832,
+      "learning_rate": 1.8444676035045074e-07,
+      "loss": 0.532,
+      "step": 56253
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.8267550468444824,
+      "learning_rate": 1.843758151260544e-07,
+      "loss": 0.2718,
+      "step": 56254
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.576280355453491,
+      "learning_rate": 1.8430488329210517e-07,
+      "loss": 0.6431,
+      "step": 56255
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.5552926063537598,
+      "learning_rate": 1.8423396484880008e-07,
+      "loss": 0.3376,
+      "step": 56256
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.7561981678009033,
+      "learning_rate": 1.8416305979633676e-07,
+      "loss": 0.5805,
+      "step": 56257
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.6518778800964355,
+      "learning_rate": 1.8409216813491115e-07,
+      "loss": 0.5372,
+      "step": 56258
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.79365873336792,
+      "learning_rate": 1.8402128986472145e-07,
+      "loss": 0.6426,
+      "step": 56259
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.1184659004211426,
+      "learning_rate": 1.8395042498596417e-07,
+      "loss": 0.6172,
+      "step": 56260
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.05802845954895,
+      "learning_rate": 1.8387957349883744e-07,
+      "loss": 0.5008,
+      "step": 56261
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.3403968811035156,
+      "learning_rate": 1.8380873540353673e-07,
+      "loss": 0.5814,
+      "step": 56262
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.805986166000366,
+      "learning_rate": 1.8373791070026016e-07,
+      "loss": 0.5531,
+      "step": 56263
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.0161588191986084,
+      "learning_rate": 1.8366709938920369e-07,
+      "loss": 0.6117,
+      "step": 56264
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.4164857864379883,
+      "learning_rate": 1.835963014705655e-07,
+      "loss": 0.6664,
+      "step": 56265
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.7158727645874023,
+      "learning_rate": 1.8352551694454046e-07,
+      "loss": 0.6723,
+      "step": 56266
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.2584140300750732,
+      "learning_rate": 1.8345474581132783e-07,
+      "loss": 0.5764,
+      "step": 56267
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.0291266441345215,
+      "learning_rate": 1.83383988071123e-07,
+      "loss": 0.4781,
+      "step": 56268
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.514885425567627,
+      "learning_rate": 1.8331324372412307e-07,
+      "loss": 0.5562,
+      "step": 56269
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 5.936831951141357,
+      "learning_rate": 1.8324251277052395e-07,
+      "loss": 0.434,
+      "step": 56270
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.4194493293762207,
+      "learning_rate": 1.8317179521052386e-07,
+      "loss": 0.4081,
+      "step": 56271
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.5392587184906006,
+      "learning_rate": 1.8310109104431817e-07,
+      "loss": 0.6682,
+      "step": 56272
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.8270628452301025,
+      "learning_rate": 1.830304002721045e-07,
+      "loss": 0.6007,
+      "step": 56273
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.178356885910034,
+      "learning_rate": 1.829597228940777e-07,
+      "loss": 0.7151,
+      "step": 56274
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.653977632522583,
+      "learning_rate": 1.828890589104365e-07,
+      "loss": 0.6335,
+      "step": 56275
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.171738862991333,
+      "learning_rate": 1.8281840832137575e-07,
+      "loss": 0.6556,
+      "step": 56276
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 4.029463768005371,
+      "learning_rate": 1.827477711270925e-07,
+      "loss": 0.6768,
+      "step": 56277
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.0366499423980713,
+      "learning_rate": 1.8267714732778274e-07,
+      "loss": 0.4091,
+      "step": 56278
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.567551374435425,
+      "learning_rate": 1.826065369236435e-07,
+      "loss": 0.584,
+      "step": 56279
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.1046533584594727,
+      "learning_rate": 1.825359399148713e-07,
+      "loss": 0.4681,
+      "step": 56280
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 4.063736438751221,
+      "learning_rate": 1.824653563016615e-07,
+      "loss": 0.7863,
+      "step": 56281
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.94522762298584,
+      "learning_rate": 1.8239478608421125e-07,
+      "loss": 0.5502,
+      "step": 56282
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.2317347526550293,
+      "learning_rate": 1.8232422926271587e-07,
+      "loss": 0.4426,
+      "step": 56283
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.0349771976470947,
+      "learning_rate": 1.8225368583737246e-07,
+      "loss": 0.5906,
+      "step": 56284
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.326714038848877,
+      "learning_rate": 1.8218315580837642e-07,
+      "loss": 0.5331,
+      "step": 56285
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.999183177947998,
+      "learning_rate": 1.8211263917592427e-07,
+      "loss": 0.4018,
+      "step": 56286
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.613968849182129,
+      "learning_rate": 1.8204213594021247e-07,
+      "loss": 0.6098,
+      "step": 56287
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.091226577758789,
+      "learning_rate": 1.8197164610143592e-07,
+      "loss": 0.7674,
+      "step": 56288
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.050417900085449,
+      "learning_rate": 1.819011696597922e-07,
+      "loss": 0.6035,
+      "step": 56289
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.588042974472046,
+      "learning_rate": 1.8183070661547563e-07,
+      "loss": 0.6225,
+      "step": 56290
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.2712085247039795,
+      "learning_rate": 1.8176025696868327e-07,
+      "loss": 0.5412,
+      "step": 56291
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.322084426879883,
+      "learning_rate": 1.8168982071961104e-07,
+      "loss": 0.5433,
+      "step": 56292
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.3939361572265625,
+      "learning_rate": 1.8161939786845385e-07,
+      "loss": 0.7283,
+      "step": 56293
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.767774820327759,
+      "learning_rate": 1.815489884154087e-07,
+      "loss": 0.539,
+      "step": 56294
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.0290026664733887,
+      "learning_rate": 1.8147859236067045e-07,
+      "loss": 0.6504,
+      "step": 56295
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.7021942138671875,
+      "learning_rate": 1.8140820970443508e-07,
+      "loss": 0.5646,
+      "step": 56296
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 5.0031609535217285,
+      "learning_rate": 1.813378404468985e-07,
+      "loss": 0.4259,
+      "step": 56297
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.4946303367614746,
+      "learning_rate": 1.8126748458825672e-07,
+      "loss": 0.6028,
+      "step": 56298
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.6718924045562744,
+      "learning_rate": 1.8119714212870397e-07,
+      "loss": 0.6457,
+      "step": 56299
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.9742043018341064,
+      "learning_rate": 1.8112681306843795e-07,
+      "loss": 0.7176,
+      "step": 56300
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.8986623287200928,
+      "learning_rate": 1.810564974076523e-07,
+      "loss": 0.4274,
+      "step": 56301
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.8132483959198,
+      "learning_rate": 1.8098619514654358e-07,
+      "loss": 0.5199,
+      "step": 56302
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.1674866676330566,
+      "learning_rate": 1.8091590628530664e-07,
+      "loss": 0.5588,
+      "step": 56303
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.584247350692749,
+      "learning_rate": 1.8084563082413798e-07,
+      "loss": 0.6084,
+      "step": 56304
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.095808267593384,
+      "learning_rate": 1.807753687632313e-07,
+      "loss": 0.8148,
+      "step": 56305
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.2625081539154053,
+      "learning_rate": 1.8070512010278429e-07,
+      "loss": 0.4642,
+      "step": 56306
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.178781747817993,
+      "learning_rate": 1.8063488484298953e-07,
+      "loss": 0.3694,
+      "step": 56307
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 5.4578962326049805,
+      "learning_rate": 1.8056466298404462e-07,
+      "loss": 0.5257,
+      "step": 56308
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.0945394039154053,
+      "learning_rate": 1.8049445452614334e-07,
+      "loss": 0.4561,
+      "step": 56309
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 4.369399547576904,
+      "learning_rate": 1.8042425946948272e-07,
+      "loss": 0.5696,
+      "step": 56310
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.187626838684082,
+      "learning_rate": 1.8035407781425596e-07,
+      "loss": 0.6038,
+      "step": 56311
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.727975368499756,
+      "learning_rate": 1.8028390956065954e-07,
+      "loss": 0.5985,
+      "step": 56312
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.639418840408325,
+      "learning_rate": 1.8021375470888726e-07,
+      "loss": 0.4864,
+      "step": 56313
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.707969903945923,
+      "learning_rate": 1.801436132591361e-07,
+      "loss": 0.5142,
+      "step": 56314
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.856262445449829,
+      "learning_rate": 1.800734852115993e-07,
+      "loss": 0.5161,
+      "step": 56315
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.644017457962036,
+      "learning_rate": 1.8000337056647276e-07,
+      "loss": 0.6366,
+      "step": 56316
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.5789592266082764,
+      "learning_rate": 1.7993326932395084e-07,
+      "loss": 0.5332,
+      "step": 56317
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.956207036972046,
+      "learning_rate": 1.7986318148422943e-07,
+      "loss": 0.5134,
+      "step": 56318
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.942974328994751,
+      "learning_rate": 1.7979310704750231e-07,
+      "loss": 0.4983,
+      "step": 56319
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.495798349380493,
+      "learning_rate": 1.7972304601396539e-07,
+      "loss": 0.7822,
+      "step": 56320
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.40891170501709,
+      "learning_rate": 1.7965299838381246e-07,
+      "loss": 0.617,
+      "step": 56321
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 4.755918502807617,
+      "learning_rate": 1.7958296415723941e-07,
+      "loss": 0.418,
+      "step": 56322
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.825612783432007,
+      "learning_rate": 1.7951294333443948e-07,
+      "loss": 0.4515,
+      "step": 56323
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.831041097640991,
+      "learning_rate": 1.7944293591560913e-07,
+      "loss": 0.5246,
+      "step": 56324
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.320988893508911,
+      "learning_rate": 1.7937294190094156e-07,
+      "loss": 0.6142,
+      "step": 56325
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.9022982120513916,
+      "learning_rate": 1.7930296129063275e-07,
+      "loss": 0.6916,
+      "step": 56326
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.908331871032715,
+      "learning_rate": 1.7923299408487527e-07,
+      "loss": 0.5264,
+      "step": 56327
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.4111623764038086,
+      "learning_rate": 1.7916304028386566e-07,
+      "loss": 0.6228,
+      "step": 56328
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.761488676071167,
+      "learning_rate": 1.7909309988779767e-07,
+      "loss": 0.4099,
+      "step": 56329
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.511430025100708,
+      "learning_rate": 1.790231728968661e-07,
+      "loss": 0.6757,
+      "step": 56330
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 4.3446807861328125,
+      "learning_rate": 1.7895325931126472e-07,
+      "loss": 0.6478,
+      "step": 56331
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.8589093685150146,
+      "learning_rate": 1.7888335913118892e-07,
+      "loss": 0.7504,
+      "step": 56332
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.116215944290161,
+      "learning_rate": 1.7881347235683134e-07,
+      "loss": 0.5353,
+      "step": 56333
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 1.9434324502944946,
+      "learning_rate": 1.7874359898838845e-07,
+      "loss": 0.4956,
+      "step": 56334
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.01936411857605,
+      "learning_rate": 1.786737390260529e-07,
+      "loss": 0.4803,
+      "step": 56335
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.2193713188171387,
+      "learning_rate": 1.7860389247002007e-07,
+      "loss": 0.5584,
+      "step": 56336
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.1549222469329834,
+      "learning_rate": 1.7853405932048318e-07,
+      "loss": 0.3806,
+      "step": 56337
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 1.8056609630584717,
+      "learning_rate": 1.7846423957763758e-07,
+      "loss": 0.4825,
+      "step": 56338
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.297013521194458,
+      "learning_rate": 1.7839443324167648e-07,
+      "loss": 0.7899,
+      "step": 56339
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.7667126655578613,
+      "learning_rate": 1.7832464031279417e-07,
+      "loss": 0.3497,
+      "step": 56340
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.9622278213500977,
+      "learning_rate": 1.7825486079118436e-07,
+      "loss": 0.767,
+      "step": 56341
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.6381032466888428,
+      "learning_rate": 1.7818509467704193e-07,
+      "loss": 0.6332,
+      "step": 56342
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.6629467010498047,
+      "learning_rate": 1.7811534197056058e-07,
+      "loss": 0.5596,
+      "step": 56343
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.307633876800537,
+      "learning_rate": 1.7804560267193405e-07,
+      "loss": 0.4761,
+      "step": 56344
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.008223533630371,
+      "learning_rate": 1.7797587678135664e-07,
+      "loss": 0.8548,
+      "step": 56345
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.6098642349243164,
+      "learning_rate": 1.7790616429902097e-07,
+      "loss": 0.8892,
+      "step": 56346
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 4.054758548736572,
+      "learning_rate": 1.7783646522512243e-07,
+      "loss": 0.5907,
+      "step": 56347
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.0272762775421143,
+      "learning_rate": 1.7776677955985476e-07,
+      "loss": 0.3353,
+      "step": 56348
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.5079267024993896,
+      "learning_rate": 1.7769710730341006e-07,
+      "loss": 0.6943,
+      "step": 56349
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 4.102001667022705,
+      "learning_rate": 1.7762744845598424e-07,
+      "loss": 0.4914,
+      "step": 56350
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 4.045434951782227,
+      "learning_rate": 1.7755780301776994e-07,
+      "loss": 0.8862,
+      "step": 56351
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.835920572280884,
+      "learning_rate": 1.7748817098896088e-07,
+      "loss": 0.4943,
+      "step": 56352
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 1.9078067541122437,
+      "learning_rate": 1.774185523697497e-07,
+      "loss": 0.2839,
+      "step": 56353
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.518026113510132,
+      "learning_rate": 1.773489471603318e-07,
+      "loss": 0.4823,
+      "step": 56354
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.081698179244995,
+      "learning_rate": 1.772793553608998e-07,
+      "loss": 0.5006,
+      "step": 56355
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.089914321899414,
+      "learning_rate": 1.7720977697164686e-07,
+      "loss": 0.6093,
+      "step": 56356
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.806917905807495,
+      "learning_rate": 1.7714021199276789e-07,
+      "loss": 0.4812,
+      "step": 56357
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.057255268096924,
+      "learning_rate": 1.7707066042445432e-07,
+      "loss": 0.736,
+      "step": 56358
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.395512342453003,
+      "learning_rate": 1.7700112226690104e-07,
+      "loss": 0.533,
+      "step": 56359
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.330106735229492,
+      "learning_rate": 1.7693159752030008e-07,
+      "loss": 0.5091,
+      "step": 56360
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.6071629524230957,
+      "learning_rate": 1.768620861848469e-07,
+      "loss": 0.4538,
+      "step": 56361
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.6555888652801514,
+      "learning_rate": 1.767925882607324e-07,
+      "loss": 0.3261,
+      "step": 56362
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.4262137413024902,
+      "learning_rate": 1.767231037481515e-07,
+      "loss": 0.468,
+      "step": 56363
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.930997371673584,
+      "learning_rate": 1.7665363264729618e-07,
+      "loss": 0.7161,
+      "step": 56364
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.2624874114990234,
+      "learning_rate": 1.7658417495836133e-07,
+      "loss": 0.7406,
+      "step": 56365
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.9335649013519287,
+      "learning_rate": 1.765147306815379e-07,
+      "loss": 0.7753,
+      "step": 56366
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.0829737186431885,
+      "learning_rate": 1.7644529981702073e-07,
+      "loss": 0.5025,
+      "step": 56367
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.7162163257598877,
+      "learning_rate": 1.7637588236500136e-07,
+      "loss": 0.6854,
+      "step": 56368
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.2852518558502197,
+      "learning_rate": 1.763064783256746e-07,
+      "loss": 0.4289,
+      "step": 56369
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.782055377960205,
+      "learning_rate": 1.76237087699232e-07,
+      "loss": 0.7362,
+      "step": 56370
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.333749771118164,
+      "learning_rate": 1.7616771048586779e-07,
+      "loss": 0.5172,
+      "step": 56371
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.285667657852173,
+      "learning_rate": 1.7609834668577297e-07,
+      "loss": 0.6109,
+      "step": 56372
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.848153829574585,
+      "learning_rate": 1.7602899629914295e-07,
+      "loss": 0.4631,
+      "step": 56373
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.752383232116699,
+      "learning_rate": 1.759596593261681e-07,
+      "loss": 0.5218,
+      "step": 56374
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.6198911666870117,
+      "learning_rate": 1.7589033576704272e-07,
+      "loss": 0.4379,
+      "step": 56375
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.1061110496520996,
+      "learning_rate": 1.758210256219589e-07,
+      "loss": 0.4171,
+      "step": 56376
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.1258742809295654,
+      "learning_rate": 1.7575172889111036e-07,
+      "loss": 0.5558,
+      "step": 56377
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.0080573558807373,
+      "learning_rate": 1.7568244557468806e-07,
+      "loss": 0.3631,
+      "step": 56378
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.642066240310669,
+      "learning_rate": 1.7561317567288684e-07,
+      "loss": 0.8149,
+      "step": 56379
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.5783286094665527,
+      "learning_rate": 1.7554391918589707e-07,
+      "loss": 0.5116,
+      "step": 56380
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.7008848190307617,
+      "learning_rate": 1.754746761139131e-07,
+      "loss": 0.6282,
+      "step": 56381
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.3147222995758057,
+      "learning_rate": 1.7540544645712587e-07,
+      "loss": 0.5157,
+      "step": 56382
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 4.017062664031982,
+      "learning_rate": 1.7533623021572965e-07,
+      "loss": 0.6319,
+      "step": 56383
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.1825084686279297,
+      "learning_rate": 1.7526702738991541e-07,
+      "loss": 0.7977,
+      "step": 56384
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.9846479892730713,
+      "learning_rate": 1.7519783797987688e-07,
+      "loss": 0.7485,
+      "step": 56385
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.4733691215515137,
+      "learning_rate": 1.7512866198580502e-07,
+      "loss": 0.6359,
+      "step": 56386
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.9587907791137695,
+      "learning_rate": 1.7505949940789302e-07,
+      "loss": 0.5815,
+      "step": 56387
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.4170169830322266,
+      "learning_rate": 1.7499035024633238e-07,
+      "loss": 0.5511,
+      "step": 56388
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.596040964126587,
+      "learning_rate": 1.749212145013174e-07,
+      "loss": 0.7187,
+      "step": 56389
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 4.554486274719238,
+      "learning_rate": 1.7485209217303845e-07,
+      "loss": 0.6912,
+      "step": 56390
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.1059212684631348,
+      "learning_rate": 1.7478298326168874e-07,
+      "loss": 0.6249,
+      "step": 56391
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 8.911134719848633,
+      "learning_rate": 1.7471388776745923e-07,
+      "loss": 0.809,
+      "step": 56392
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.9457337856292725,
+      "learning_rate": 1.7464480569054366e-07,
+      "loss": 0.4914,
+      "step": 56393
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.876568078994751,
+      "learning_rate": 1.745757370311324e-07,
+      "loss": 0.6138,
+      "step": 56394
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.8690876960754395,
+      "learning_rate": 1.7450668178941922e-07,
+      "loss": 0.4676,
+      "step": 56395
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.4795587062835693,
+      "learning_rate": 1.7443763996559448e-07,
+      "loss": 0.4443,
+      "step": 56396
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.9133546352386475,
+      "learning_rate": 1.743686115598514e-07,
+      "loss": 0.5104,
+      "step": 56397
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.789762020111084,
+      "learning_rate": 1.742995965723815e-07,
+      "loss": 0.4388,
+      "step": 56398
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.2430009841918945,
+      "learning_rate": 1.742305950033768e-07,
+      "loss": 0.3541,
+      "step": 56399
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.9830174446105957,
+      "learning_rate": 1.7416160685302885e-07,
+      "loss": 0.7245,
+      "step": 56400
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.5429766178131104,
+      "learning_rate": 1.740926321215297e-07,
+      "loss": 0.7419,
+      "step": 56401
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.9592573642730713,
+      "learning_rate": 1.740236708090709e-07,
+      "loss": 0.6702,
+      "step": 56402
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.137789011001587,
+      "learning_rate": 1.7395472291584447e-07,
+      "loss": 0.6287,
+      "step": 56403
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 5.176107883453369,
+      "learning_rate": 1.738857884420425e-07,
+      "loss": 0.5751,
+      "step": 56404
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 4.51318883895874,
+      "learning_rate": 1.738168673878554e-07,
+      "loss": 0.5651,
+      "step": 56405
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.329270839691162,
+      "learning_rate": 1.7374795975347637e-07,
+      "loss": 0.582,
+      "step": 56406
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.169187307357788,
+      "learning_rate": 1.736790655390963e-07,
+      "loss": 0.6035,
+      "step": 56407
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.8869166374206543,
+      "learning_rate": 1.736101847449062e-07,
+      "loss": 0.5874,
+      "step": 56408
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.5248498916625977,
+      "learning_rate": 1.735413173710987e-07,
+      "loss": 0.6354,
+      "step": 56409
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.401780605316162,
+      "learning_rate": 1.7347246341786472e-07,
+      "loss": 0.526,
+      "step": 56410
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.123330593109131,
+      "learning_rate": 1.734036228853958e-07,
+      "loss": 0.6432,
+      "step": 56411
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.5172810554504395,
+      "learning_rate": 1.733347957738829e-07,
+      "loss": 0.545,
+      "step": 56412
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.5775434970855713,
+      "learning_rate": 1.7326598208351864e-07,
+      "loss": 0.5767,
+      "step": 56413
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.704935073852539,
+      "learning_rate": 1.7319718181449286e-07,
+      "loss": 0.586,
+      "step": 56414
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.9879910945892334,
+      "learning_rate": 1.7312839496699707e-07,
+      "loss": 0.477,
+      "step": 56415
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.5537426471710205,
+      "learning_rate": 1.7305962154122448e-07,
+      "loss": 0.4419,
+      "step": 56416
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.2750606536865234,
+      "learning_rate": 1.7299086153736378e-07,
+      "loss": 0.4634,
+      "step": 56417
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 5.370748043060303,
+      "learning_rate": 1.7292211495560762e-07,
+      "loss": 0.7529,
+      "step": 56418
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.428464651107788,
+      "learning_rate": 1.728533817961464e-07,
+      "loss": 0.4444,
+      "step": 56419
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.0073108673095703,
+      "learning_rate": 1.7278466205917223e-07,
+      "loss": 0.5229,
+      "step": 56420
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.321772575378418,
+      "learning_rate": 1.7271595574487542e-07,
+      "loss": 0.6507,
+      "step": 56421
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.5985655784606934,
+      "learning_rate": 1.7264726285344702e-07,
+      "loss": 0.5527,
+      "step": 56422
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.1879687309265137,
+      "learning_rate": 1.7257858338507848e-07,
+      "loss": 0.4192,
+      "step": 56423
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.3524811267852783,
+      "learning_rate": 1.7250991733996026e-07,
+      "loss": 0.532,
+      "step": 56424
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 4.501449108123779,
+      "learning_rate": 1.7244126471828326e-07,
+      "loss": 0.823,
+      "step": 56425
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.6548914909362793,
+      "learning_rate": 1.7237262552023958e-07,
+      "loss": 0.8077,
+      "step": 56426
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.583930730819702,
+      "learning_rate": 1.7230399974601853e-07,
+      "loss": 0.5789,
+      "step": 56427
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.6677563190460205,
+      "learning_rate": 1.7223538739581157e-07,
+      "loss": 0.482,
+      "step": 56428
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 4.198051929473877,
+      "learning_rate": 1.721667884698086e-07,
+      "loss": 0.4183,
+      "step": 56429
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 1.766563892364502,
+      "learning_rate": 1.7209820296820279e-07,
+      "loss": 0.2637,
+      "step": 56430
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.330348253250122,
+      "learning_rate": 1.720296308911823e-07,
+      "loss": 0.6354,
+      "step": 56431
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.115323543548584,
+      "learning_rate": 1.7196107223893977e-07,
+      "loss": 0.2941,
+      "step": 56432
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.488626718521118,
+      "learning_rate": 1.7189252701166393e-07,
+      "loss": 0.5015,
+      "step": 56433
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.471008539199829,
+      "learning_rate": 1.7182399520954684e-07,
+      "loss": 0.3788,
+      "step": 56434
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.2248001098632812,
+      "learning_rate": 1.7175547683277839e-07,
+      "loss": 0.3884,
+      "step": 56435
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.991607427597046,
+      "learning_rate": 1.7168697188154948e-07,
+      "loss": 0.7133,
+      "step": 56436
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.25854754447937,
+      "learning_rate": 1.7161848035604944e-07,
+      "loss": 0.4791,
+      "step": 56437
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.705453872680664,
+      "learning_rate": 1.715500022564709e-07,
+      "loss": 1.0078,
+      "step": 56438
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.8338305950164795,
+      "learning_rate": 1.7148153758300202e-07,
+      "loss": 0.6289,
+      "step": 56439
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.873979330062866,
+      "learning_rate": 1.7141308633583543e-07,
+      "loss": 0.3948,
+      "step": 56440
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.9063825607299805,
+      "learning_rate": 1.7134464851515874e-07,
+      "loss": 0.7011,
+      "step": 56441
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.9259843826293945,
+      "learning_rate": 1.7127622412116517e-07,
+      "loss": 0.5514,
+      "step": 56442
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.345970392227173,
+      "learning_rate": 1.7120781315404233e-07,
+      "loss": 0.505,
+      "step": 56443
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.3976216316223145,
+      "learning_rate": 1.711394156139823e-07,
+      "loss": 0.4675,
+      "step": 56444
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.1132898330688477,
+      "learning_rate": 1.7107103150117432e-07,
+      "loss": 0.3804,
+      "step": 56445
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.327927827835083,
+      "learning_rate": 1.710026608158094e-07,
+      "loss": 0.5548,
+      "step": 56446
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.812464952468872,
+      "learning_rate": 1.709343035580763e-07,
+      "loss": 0.6251,
+      "step": 56447
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.000201940536499,
+      "learning_rate": 1.7086595972816643e-07,
+      "loss": 0.5351,
+      "step": 56448
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.813826322555542,
+      "learning_rate": 1.707976293262681e-07,
+      "loss": 0.7713,
+      "step": 56449
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 4.233452796936035,
+      "learning_rate": 1.7072931235257385e-07,
+      "loss": 0.6732,
+      "step": 56450
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.1399495601654053,
+      "learning_rate": 1.7066100880727187e-07,
+      "loss": 0.52,
+      "step": 56451
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.3936846256256104,
+      "learning_rate": 1.7059271869055317e-07,
+      "loss": 0.4525,
+      "step": 56452
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.903632164001465,
+      "learning_rate": 1.7052444200260588e-07,
+      "loss": 0.3201,
+      "step": 56453
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.0421674251556396,
+      "learning_rate": 1.7045617874362207e-07,
+      "loss": 0.5936,
+      "step": 56454
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.080073118209839,
+      "learning_rate": 1.7038792891378942e-07,
+      "loss": 0.6071,
+      "step": 56455
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.854295015335083,
+      "learning_rate": 1.7031969251329937e-07,
+      "loss": 0.4108,
+      "step": 56456
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.8026719093322754,
+      "learning_rate": 1.7025146954234073e-07,
+      "loss": 0.7486,
+      "step": 56457
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.6720147132873535,
+      "learning_rate": 1.7018326000110387e-07,
+      "loss": 0.7449,
+      "step": 56458
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.127899408340454,
+      "learning_rate": 1.701150638897775e-07,
+      "loss": 0.3938,
+      "step": 56459
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.9281022548675537,
+      "learning_rate": 1.7004688120855206e-07,
+      "loss": 0.6082,
+      "step": 56460
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.261958599090576,
+      "learning_rate": 1.6997871195761683e-07,
+      "loss": 0.6112,
+      "step": 56461
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.0079896450042725,
+      "learning_rate": 1.6991055613716168e-07,
+      "loss": 0.6317,
+      "step": 56462
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.5516929626464844,
+      "learning_rate": 1.6984241374737588e-07,
+      "loss": 0.6973,
+      "step": 56463
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.6079342365264893,
+      "learning_rate": 1.6977428478844815e-07,
+      "loss": 0.4625,
+      "step": 56464
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.630711793899536,
+      "learning_rate": 1.697061692605695e-07,
+      "loss": 0.4504,
+      "step": 56465
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.8527777194976807,
+      "learning_rate": 1.6963806716392806e-07,
+      "loss": 0.6262,
+      "step": 56466
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.0671749114990234,
+      "learning_rate": 1.6956997849871426e-07,
+      "loss": 0.5586,
+      "step": 56467
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.4264039993286133,
+      "learning_rate": 1.6950190326511628e-07,
+      "loss": 0.4981,
+      "step": 56468
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.212463140487671,
+      "learning_rate": 1.69433841463324e-07,
+      "loss": 0.5458,
+      "step": 56469
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.7212607860565186,
+      "learning_rate": 1.6936579309352718e-07,
+      "loss": 0.6462,
+      "step": 56470
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.589453935623169,
+      "learning_rate": 1.692977581559141e-07,
+      "loss": 0.6176,
+      "step": 56471
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.179448366165161,
+      "learning_rate": 1.6922973665067455e-07,
+      "loss": 0.3682,
+      "step": 56472
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.797985553741455,
+      "learning_rate": 1.6916172857799674e-07,
+      "loss": 1.0078,
+      "step": 56473
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 1.9871833324432373,
+      "learning_rate": 1.690937339380716e-07,
+      "loss": 0.4386,
+      "step": 56474
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.7816123962402344,
+      "learning_rate": 1.6902575273108623e-07,
+      "loss": 0.3568,
+      "step": 56475
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.875535726547241,
+      "learning_rate": 1.6895778495723049e-07,
+      "loss": 0.8751,
+      "step": 56476
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.409921884536743,
+      "learning_rate": 1.6888983061669416e-07,
+      "loss": 0.5783,
+      "step": 56477
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.4289863109588623,
+      "learning_rate": 1.688218897096644e-07,
+      "loss": 0.506,
+      "step": 56478
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.7055256366729736,
+      "learning_rate": 1.6875396223633212e-07,
+      "loss": 0.7285,
+      "step": 56479
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.859093427658081,
+      "learning_rate": 1.6868604819688439e-07,
+      "loss": 0.6522,
+      "step": 56480
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.9841532707214355,
+      "learning_rate": 1.6861814759151218e-07,
+      "loss": 0.5255,
+      "step": 56481
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.0149998664855957,
+      "learning_rate": 1.6855026042040146e-07,
+      "loss": 0.8845,
+      "step": 56482
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.229198694229126,
+      "learning_rate": 1.6848238668374372e-07,
+      "loss": 0.5848,
+      "step": 56483
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.8256125450134277,
+      "learning_rate": 1.6841452638172606e-07,
+      "loss": 0.6263,
+      "step": 56484
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.4281704425811768,
+      "learning_rate": 1.6834667951453775e-07,
+      "loss": 0.5476,
+      "step": 56485
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.790952205657959,
+      "learning_rate": 1.6827884608236754e-07,
+      "loss": 0.4806,
+      "step": 56486
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.0139822959899902,
+      "learning_rate": 1.6821102608540362e-07,
+      "loss": 0.393,
+      "step": 56487
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.7255728244781494,
+      "learning_rate": 1.6814321952383471e-07,
+      "loss": 0.4106,
+      "step": 56488
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.3080012798309326,
+      "learning_rate": 1.6807542639785012e-07,
+      "loss": 0.4579,
+      "step": 56489
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.9469776153564453,
+      "learning_rate": 1.680076467076358e-07,
+      "loss": 0.6673,
+      "step": 56490
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.308774709701538,
+      "learning_rate": 1.6793988045338383e-07,
+      "loss": 0.6217,
+      "step": 56491
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.04439115524292,
+      "learning_rate": 1.678721276352807e-07,
+      "loss": 0.4695,
+      "step": 56492
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.00736403465271,
+      "learning_rate": 1.678043882535152e-07,
+      "loss": 0.4378,
+      "step": 56493
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.160963773727417,
+      "learning_rate": 1.677366623082749e-07,
+      "loss": 0.665,
+      "step": 56494
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 4.740776538848877,
+      "learning_rate": 1.6766894979974912e-07,
+      "loss": 0.3455,
+      "step": 56495
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.6197948455810547,
+      "learning_rate": 1.676012507281255e-07,
+      "loss": 0.5518,
+      "step": 56496
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.989725112915039,
+      "learning_rate": 1.6753356509359275e-07,
+      "loss": 0.5849,
+      "step": 56497
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.179584264755249,
+      "learning_rate": 1.6746589289633853e-07,
+      "loss": 0.5724,
+      "step": 56498
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.02131986618042,
+      "learning_rate": 1.673982341365521e-07,
+      "loss": 0.4632,
+      "step": 56499
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.1598544120788574,
+      "learning_rate": 1.6733058881442e-07,
+      "loss": 0.5289,
+      "step": 56500
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.3762083053588867,
+      "learning_rate": 1.6726295693013205e-07,
+      "loss": 0.6392,
+      "step": 56501
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.7536990642547607,
+      "learning_rate": 1.6719533848387482e-07,
+      "loss": 0.5382,
+      "step": 56502
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.297942638397217,
+      "learning_rate": 1.67127733475837e-07,
+      "loss": 0.4633,
+      "step": 56503
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.141773223876953,
+      "learning_rate": 1.6706014190620624e-07,
+      "loss": 0.7267,
+      "step": 56504
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.7910993099212646,
+      "learning_rate": 1.669925637751718e-07,
+      "loss": 0.7976,
+      "step": 56505
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.8858964443206787,
+      "learning_rate": 1.669249990829197e-07,
+      "loss": 0.6005,
+      "step": 56506
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.694673776626587,
+      "learning_rate": 1.6685744782963865e-07,
+      "loss": 0.555,
+      "step": 56507
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.777656316757202,
+      "learning_rate": 1.6678991001551681e-07,
+      "loss": 0.821,
+      "step": 56508
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 1.871476173400879,
+      "learning_rate": 1.6672238564074183e-07,
+      "loss": 0.381,
+      "step": 56509
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 4.035617828369141,
+      "learning_rate": 1.6665487470550023e-07,
+      "loss": 0.6338,
+      "step": 56510
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.6407363414764404,
+      "learning_rate": 1.6658737720998185e-07,
+      "loss": 0.5737,
+      "step": 56511
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.518289804458618,
+      "learning_rate": 1.6651989315437267e-07,
+      "loss": 0.6016,
+      "step": 56512
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.8355679512023926,
+      "learning_rate": 1.6645242253886196e-07,
+      "loss": 0.6283,
+      "step": 56513
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.9140820503234863,
+      "learning_rate": 1.6638496536363513e-07,
+      "loss": 0.7051,
+      "step": 56514
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.152803421020508,
+      "learning_rate": 1.6631752162888205e-07,
+      "loss": 0.7154,
+      "step": 56515
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.2757444381713867,
+      "learning_rate": 1.6625009133478864e-07,
+      "loss": 0.7113,
+      "step": 56516
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.4025518894195557,
+      "learning_rate": 1.661826744815437e-07,
+      "loss": 0.4751,
+      "step": 56517
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 4.361596584320068,
+      "learning_rate": 1.6611527106933312e-07,
+      "loss": 0.7163,
+      "step": 56518
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.6248433589935303,
+      "learning_rate": 1.6604788109834568e-07,
+      "loss": 0.8274,
+      "step": 56519
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.482523202896118,
+      "learning_rate": 1.6598050456876792e-07,
+      "loss": 0.4023,
+      "step": 56520
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.8880317211151123,
+      "learning_rate": 1.6591314148078796e-07,
+      "loss": 0.4595,
+      "step": 56521
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.8190996646881104,
+      "learning_rate": 1.6584579183459183e-07,
+      "loss": 0.5969,
+      "step": 56522
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.007795810699463,
+      "learning_rate": 1.6577845563036878e-07,
+      "loss": 0.6662,
+      "step": 56523
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.162177324295044,
+      "learning_rate": 1.6571113286830364e-07,
+      "loss": 0.6131,
+      "step": 56524
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.9397945404052734,
+      "learning_rate": 1.6564382354858578e-07,
+      "loss": 0.6166,
+      "step": 56525
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.9015586376190186,
+      "learning_rate": 1.655765276714011e-07,
+      "loss": 0.6787,
+      "step": 56526
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.5959410667419434,
+      "learning_rate": 1.6550924523693668e-07,
+      "loss": 0.6474,
+      "step": 56527
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 1.9565836191177368,
+      "learning_rate": 1.6544197624538073e-07,
+      "loss": 0.5754,
+      "step": 56528
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.274336099624634,
+      "learning_rate": 1.6537472069691917e-07,
+      "loss": 0.7018,
+      "step": 56529
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.091179132461548,
+      "learning_rate": 1.6530747859173968e-07,
+      "loss": 0.3592,
+      "step": 56530
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.6180431842803955,
+      "learning_rate": 1.6524024993002875e-07,
+      "loss": 0.4766,
+      "step": 56531
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.5710854530334473,
+      "learning_rate": 1.6517303471197342e-07,
+      "loss": 0.4711,
+      "step": 56532
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.021056652069092,
+      "learning_rate": 1.651058329377614e-07,
+      "loss": 0.5422,
+      "step": 56533
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 6.290925979614258,
+      "learning_rate": 1.65038644607578e-07,
+      "loss": 0.5443,
+      "step": 56534
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.9409053325653076,
+      "learning_rate": 1.6497146972161205e-07,
+      "loss": 0.6077,
+      "step": 56535
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.262141466140747,
+      "learning_rate": 1.6490430828004776e-07,
+      "loss": 0.6013,
+      "step": 56536
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.8947181701660156,
+      "learning_rate": 1.648371602830745e-07,
+      "loss": 0.6354,
+      "step": 56537
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.2676279544830322,
+      "learning_rate": 1.647700257308771e-07,
+      "loss": 0.5542,
+      "step": 56538
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.3656015396118164,
+      "learning_rate": 1.6470290462364313e-07,
+      "loss": 0.5928,
+      "step": 56539
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.673144578933716,
+      "learning_rate": 1.6463579696155918e-07,
+      "loss": 0.5371,
+      "step": 56540
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.9213345050811768,
+      "learning_rate": 1.6456870274481119e-07,
+      "loss": 0.8045,
+      "step": 56541
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.0756571292877197,
+      "learning_rate": 1.6450162197358732e-07,
+      "loss": 0.5724,
+      "step": 56542
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 4.0709228515625,
+      "learning_rate": 1.644345546480719e-07,
+      "loss": 0.4841,
+      "step": 56543
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 4.002711296081543,
+      "learning_rate": 1.6436750076845364e-07,
+      "loss": 0.6694,
+      "step": 56544
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 4.263081073760986,
+      "learning_rate": 1.6430046033491686e-07,
+      "loss": 0.8469,
+      "step": 56545
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.291557550430298,
+      "learning_rate": 1.6423343334764974e-07,
+      "loss": 0.3345,
+      "step": 56546
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.2869632244110107,
+      "learning_rate": 1.6416641980683713e-07,
+      "loss": 0.8072,
+      "step": 56547
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.202444314956665,
+      "learning_rate": 1.640994197126672e-07,
+      "loss": 0.4678,
+      "step": 56548
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.296285629272461,
+      "learning_rate": 1.6403243306532424e-07,
+      "loss": 0.3799,
+      "step": 56549
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.8128604888916016,
+      "learning_rate": 1.639654598649959e-07,
+      "loss": 0.5225,
+      "step": 56550
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.153519868850708,
+      "learning_rate": 1.6389850011186758e-07,
+      "loss": 0.5615,
+      "step": 56551
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.2229065895080566,
+      "learning_rate": 1.6383155380612637e-07,
+      "loss": 0.574,
+      "step": 56552
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.2508416175842285,
+      "learning_rate": 1.6376462094795765e-07,
+      "loss": 0.5167,
+      "step": 56553
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 4.170883655548096,
+      "learning_rate": 1.6369770153754848e-07,
+      "loss": 0.9871,
+      "step": 56554
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.8314361572265625,
+      "learning_rate": 1.6363079557508321e-07,
+      "loss": 0.342,
+      "step": 56555
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.834829807281494,
+      "learning_rate": 1.6356390306074998e-07,
+      "loss": 0.4885,
+      "step": 56556
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.0381414890289307,
+      "learning_rate": 1.634970239947331e-07,
+      "loss": 0.4875,
+      "step": 56557
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.5132713317871094,
+      "learning_rate": 1.6343015837721966e-07,
+      "loss": 0.453,
+      "step": 56558
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 4.638520240783691,
+      "learning_rate": 1.6336330620839502e-07,
+      "loss": 0.5358,
+      "step": 56559
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.1544973850250244,
+      "learning_rate": 1.6329646748844518e-07,
+      "loss": 0.5717,
+      "step": 56560
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 6.569893836975098,
+      "learning_rate": 1.6322964221755555e-07,
+      "loss": 0.6143,
+      "step": 56561
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.8821191787719727,
+      "learning_rate": 1.631628303959132e-07,
+      "loss": 0.3471,
+      "step": 56562
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.9872958660125732,
+      "learning_rate": 1.630960320237024e-07,
+      "loss": 0.6183,
+      "step": 56563
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.0696544647216797,
+      "learning_rate": 1.6302924710111024e-07,
+      "loss": 0.5813,
+      "step": 56564
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.4338178634643555,
+      "learning_rate": 1.6296247562832102e-07,
+      "loss": 0.2748,
+      "step": 56565
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.820037364959717,
+      "learning_rate": 1.6289571760552182e-07,
+      "loss": 0.5262,
+      "step": 56566
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.6607489585876465,
+      "learning_rate": 1.628289730328969e-07,
+      "loss": 0.5596,
+      "step": 56567
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.4510350227355957,
+      "learning_rate": 1.6276224191063338e-07,
+      "loss": 0.3553,
+      "step": 56568
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.6876957416534424,
+      "learning_rate": 1.62695524238915e-07,
+      "loss": 0.5667,
+      "step": 56569
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.799670457839966,
+      "learning_rate": 1.6262882001792933e-07,
+      "loss": 0.5402,
+      "step": 56570
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.530465841293335,
+      "learning_rate": 1.6256212924785964e-07,
+      "loss": 0.6875,
+      "step": 56571
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.9756996631622314,
+      "learning_rate": 1.624954519288935e-07,
+      "loss": 0.6799,
+      "step": 56572
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.668215274810791,
+      "learning_rate": 1.6242878806121465e-07,
+      "loss": 0.4786,
+      "step": 56573
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 6.28393030166626,
+      "learning_rate": 1.6236213764500964e-07,
+      "loss": 0.6527,
+      "step": 56574
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 5.976486682891846,
+      "learning_rate": 1.6229550068046328e-07,
+      "loss": 0.5402,
+      "step": 56575
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.4867260456085205,
+      "learning_rate": 1.6222887716776104e-07,
+      "loss": 0.5339,
+      "step": 56576
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.896108627319336,
+      "learning_rate": 1.6216226710708716e-07,
+      "loss": 0.668,
+      "step": 56577
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 4.695390701293945,
+      "learning_rate": 1.6209567049862873e-07,
+      "loss": 0.6975,
+      "step": 56578
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.244414806365967,
+      "learning_rate": 1.620290873425695e-07,
+      "loss": 0.5958,
+      "step": 56579
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 5.0579514503479,
+      "learning_rate": 1.6196251763909544e-07,
+      "loss": 0.3406,
+      "step": 56580
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.7031307220458984,
+      "learning_rate": 1.6189596138839026e-07,
+      "loss": 0.7561,
+      "step": 56581
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.9494776725769043,
+      "learning_rate": 1.6182941859064104e-07,
+      "loss": 0.6014,
+      "step": 56582
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.946911573410034,
+      "learning_rate": 1.61762889246031e-07,
+      "loss": 0.647,
+      "step": 56583
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.1594929695129395,
+      "learning_rate": 1.6169637335474663e-07,
+      "loss": 0.81,
+      "step": 56584
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.261730670928955,
+      "learning_rate": 1.616298709169717e-07,
+      "loss": 0.558,
+      "step": 56585
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.0234735012054443,
+      "learning_rate": 1.615633819328921e-07,
+      "loss": 0.5071,
+      "step": 56586
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.5174999237060547,
+      "learning_rate": 1.6149690640269112e-07,
+      "loss": 0.789,
+      "step": 56587
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.0571208000183105,
+      "learning_rate": 1.614304443265552e-07,
+      "loss": 0.4744,
+      "step": 56588
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.5750060081481934,
+      "learning_rate": 1.613639957046692e-07,
+      "loss": 0.7131,
+      "step": 56589
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.7538793087005615,
+      "learning_rate": 1.612975605372169e-07,
+      "loss": 0.3199,
+      "step": 56590
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.5947248935699463,
+      "learning_rate": 1.6123113882438312e-07,
+      "loss": 0.322,
+      "step": 56591
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 4.1359477043151855,
+      "learning_rate": 1.6116473056635384e-07,
+      "loss": 0.5442,
+      "step": 56592
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.9278719425201416,
+      "learning_rate": 1.6109833576331224e-07,
+      "loss": 0.6873,
+      "step": 56593
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.732938051223755,
+      "learning_rate": 1.6103195441544372e-07,
+      "loss": 0.5182,
+      "step": 56594
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.581763982772827,
+      "learning_rate": 1.6096558652293204e-07,
+      "loss": 0.5963,
+      "step": 56595
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.1863250732421875,
+      "learning_rate": 1.6089923208596314e-07,
+      "loss": 0.5432,
+      "step": 56596
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.3753981590270996,
+      "learning_rate": 1.6083289110471966e-07,
+      "loss": 0.5422,
+      "step": 56597
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.5757436752319336,
+      "learning_rate": 1.6076656357938814e-07,
+      "loss": 0.4966,
+      "step": 56598
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.783670663833618,
+      "learning_rate": 1.6070024951015117e-07,
+      "loss": 0.6931,
+      "step": 56599
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.087205171585083,
+      "learning_rate": 1.606339488971942e-07,
+      "loss": 0.3305,
+      "step": 56600
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.202585458755493,
+      "learning_rate": 1.6056766174070204e-07,
+      "loss": 0.5357,
+      "step": 56601
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 1.9757853746414185,
+      "learning_rate": 1.6050138804085734e-07,
+      "loss": 0.3655,
+      "step": 56602
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.937203884124756,
+      "learning_rate": 1.604351277978461e-07,
+      "loss": 0.5749,
+      "step": 56603
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.0539462566375732,
+      "learning_rate": 1.6036888101185144e-07,
+      "loss": 0.5003,
+      "step": 56604
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.9544265270233154,
+      "learning_rate": 1.6030264768305827e-07,
+      "loss": 0.6018,
+      "step": 56605
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 4.184601783752441,
+      "learning_rate": 1.6023642781164973e-07,
+      "loss": 0.7594,
+      "step": 56606
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.503875970840454,
+      "learning_rate": 1.601702213978118e-07,
+      "loss": 0.5221,
+      "step": 56607
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 4.950857162475586,
+      "learning_rate": 1.6010402844172655e-07,
+      "loss": 0.5276,
+      "step": 56608
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.460693597793579,
+      "learning_rate": 1.6003784894357943e-07,
+      "loss": 0.5906,
+      "step": 56609
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.0290181636810303,
+      "learning_rate": 1.5997168290355359e-07,
+      "loss": 0.632,
+      "step": 56610
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.881115674972534,
+      "learning_rate": 1.5990553032183387e-07,
+      "loss": 0.6017,
+      "step": 56611
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.879861354827881,
+      "learning_rate": 1.5983939119860235e-07,
+      "loss": 0.5995,
+      "step": 56612
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.6953415870666504,
+      "learning_rate": 1.5977326553404558e-07,
+      "loss": 0.5554,
+      "step": 56613
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 4.539408206939697,
+      "learning_rate": 1.5970715332834564e-07,
+      "loss": 0.7348,
+      "step": 56614
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.984708070755005,
+      "learning_rate": 1.5964105458168733e-07,
+      "loss": 0.6516,
+      "step": 56615
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.1230719089508057,
+      "learning_rate": 1.5957496929425386e-07,
+      "loss": 0.5066,
+      "step": 56616
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 4.703561305999756,
+      "learning_rate": 1.5950889746622954e-07,
+      "loss": 0.6911,
+      "step": 56617
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.5896315574645996,
+      "learning_rate": 1.59442839097797e-07,
+      "loss": 0.6868,
+      "step": 56618
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.879481077194214,
+      "learning_rate": 1.5937679418914109e-07,
+      "loss": 0.5691,
+      "step": 56619
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.602050304412842,
+      "learning_rate": 1.5931076274044444e-07,
+      "loss": 0.3865,
+      "step": 56620
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.1637399196624756,
+      "learning_rate": 1.592447447518919e-07,
+      "loss": 0.3547,
+      "step": 56621
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.4124722480773926,
+      "learning_rate": 1.5917874022366552e-07,
+      "loss": 0.4945,
+      "step": 56622
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.096485137939453,
+      "learning_rate": 1.5911274915595022e-07,
+      "loss": 0.628,
+      "step": 56623
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.2343554496765137,
+      "learning_rate": 1.5904677154892857e-07,
+      "loss": 0.5243,
+      "step": 56624
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.632826328277588,
+      "learning_rate": 1.5898080740278487e-07,
+      "loss": 0.5671,
+      "step": 56625
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.4160068035125732,
+      "learning_rate": 1.589148567177018e-07,
+      "loss": 0.56,
+      "step": 56626
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.304889678955078,
+      "learning_rate": 1.5884891949386306e-07,
+      "loss": 0.4877,
+      "step": 56627
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.2657206058502197,
+      "learning_rate": 1.5878299573145184e-07,
+      "loss": 0.502,
+      "step": 56628
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.1316351890563965,
+      "learning_rate": 1.587170854306519e-07,
+      "loss": 0.6348,
+      "step": 56629
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.6537272930145264,
+      "learning_rate": 1.5865118859164585e-07,
+      "loss": 0.5672,
+      "step": 56630
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.108689546585083,
+      "learning_rate": 1.5858530521461746e-07,
+      "loss": 0.5278,
+      "step": 56631
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.917816400527954,
+      "learning_rate": 1.5851943529974877e-07,
+      "loss": 0.4994,
+      "step": 56632
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.4378132820129395,
+      "learning_rate": 1.5845357884722467e-07,
+      "loss": 0.7152,
+      "step": 56633
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.276909351348877,
+      "learning_rate": 1.5838773585722722e-07,
+      "loss": 0.6844,
+      "step": 56634
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.3314757347106934,
+      "learning_rate": 1.5832190632994015e-07,
+      "loss": 0.4674,
+      "step": 56635
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.736276149749756,
+      "learning_rate": 1.5825609026554556e-07,
+      "loss": 0.8823,
+      "step": 56636
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.5322158336639404,
+      "learning_rate": 1.5819028766422772e-07,
+      "loss": 0.7298,
+      "step": 56637
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.8174245357513428,
+      "learning_rate": 1.5812449852616818e-07,
+      "loss": 0.38,
+      "step": 56638
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.5886237621307373,
+      "learning_rate": 1.580587228515512e-07,
+      "loss": 0.5928,
+      "step": 56639
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 7.270091533660889,
+      "learning_rate": 1.5799296064055892e-07,
+      "loss": 0.6388,
+      "step": 56640
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.7364089488983154,
+      "learning_rate": 1.5792721189337444e-07,
+      "loss": 0.7813,
+      "step": 56641
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 4.481265544891357,
+      "learning_rate": 1.5786147661017992e-07,
+      "loss": 0.5565,
+      "step": 56642
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.573936700820923,
+      "learning_rate": 1.577957547911596e-07,
+      "loss": 0.529,
+      "step": 56643
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.7670326232910156,
+      "learning_rate": 1.5773004643649448e-07,
+      "loss": 0.5874,
+      "step": 56644
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.1373515129089355,
+      "learning_rate": 1.576643515463694e-07,
+      "loss": 0.5123,
+      "step": 56645
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 4.735020637512207,
+      "learning_rate": 1.575986701209642e-07,
+      "loss": 0.7772,
+      "step": 56646
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.205928087234497,
+      "learning_rate": 1.575330021604643e-07,
+      "loss": 0.4648,
+      "step": 56647
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.5348384380340576,
+      "learning_rate": 1.5746734766505068e-07,
+      "loss": 0.506,
+      "step": 56648
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.3369429111480713,
+      "learning_rate": 1.574017066349065e-07,
+      "loss": 0.3525,
+      "step": 56649
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.8158175945281982,
+      "learning_rate": 1.573360790702133e-07,
+      "loss": 0.3354,
+      "step": 56650
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.3197824954986572,
+      "learning_rate": 1.5727046497115485e-07,
+      "loss": 0.5696,
+      "step": 56651
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.2966129779815674,
+      "learning_rate": 1.5720486433791315e-07,
+      "loss": 0.4703,
+      "step": 56652
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.0021212100982666,
+      "learning_rate": 1.571392771706709e-07,
+      "loss": 0.5002,
+      "step": 56653
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.681957960128784,
+      "learning_rate": 1.570737034696096e-07,
+      "loss": 0.6284,
+      "step": 56654
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.5291881561279297,
+      "learning_rate": 1.5700814323491243e-07,
+      "loss": 0.6251,
+      "step": 56655
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.0324838161468506,
+      "learning_rate": 1.5694259646676147e-07,
+      "loss": 0.3597,
+      "step": 56656
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.0871706008911133,
+      "learning_rate": 1.568770631653388e-07,
+      "loss": 0.5749,
+      "step": 56657
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.519441843032837,
+      "learning_rate": 1.5681154333082648e-07,
+      "loss": 0.67,
+      "step": 56658
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.4739303588867188,
+      "learning_rate": 1.5674603696340772e-07,
+      "loss": 0.5317,
+      "step": 56659
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.1100332736968994,
+      "learning_rate": 1.5668054406326294e-07,
+      "loss": 0.595,
+      "step": 56660
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.593902587890625,
+      "learning_rate": 1.566150646305753e-07,
+      "loss": 0.581,
+      "step": 56661
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.7017662525177,
+      "learning_rate": 1.5654959866552743e-07,
+      "loss": 0.514,
+      "step": 56662
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.585806369781494,
+      "learning_rate": 1.5648414616830033e-07,
+      "loss": 0.7862,
+      "step": 56663
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 4.206542491912842,
+      "learning_rate": 1.5641870713907658e-07,
+      "loss": 0.8107,
+      "step": 56664
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 4.661118984222412,
+      "learning_rate": 1.5635328157803774e-07,
+      "loss": 0.8424,
+      "step": 56665
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.8735549449920654,
+      "learning_rate": 1.5628786948536645e-07,
+      "loss": 0.6052,
+      "step": 56666
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.686196804046631,
+      "learning_rate": 1.5622247086124364e-07,
+      "loss": 0.4867,
+      "step": 56667
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.436933994293213,
+      "learning_rate": 1.5615708570585196e-07,
+      "loss": 0.5284,
+      "step": 56668
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.485368251800537,
+      "learning_rate": 1.560917140193724e-07,
+      "loss": 0.5698,
+      "step": 56669
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.497598171234131,
+      "learning_rate": 1.5602635580198754e-07,
+      "loss": 0.5675,
+      "step": 56670
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.6438751220703125,
+      "learning_rate": 1.559610110538784e-07,
+      "loss": 0.3985,
+      "step": 56671
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 4.18350076675415,
+      "learning_rate": 1.5589567977522812e-07,
+      "loss": 0.5327,
+      "step": 56672
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.582122325897217,
+      "learning_rate": 1.5583036196621548e-07,
+      "loss": 0.5812,
+      "step": 56673
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.596198081970215,
+      "learning_rate": 1.557650576270253e-07,
+      "loss": 0.4613,
+      "step": 56674
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.5714216232299805,
+      "learning_rate": 1.556997667578375e-07,
+      "loss": 0.5401,
+      "step": 56675
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.22670841217041,
+      "learning_rate": 1.5563448935883408e-07,
+      "loss": 0.6179,
+      "step": 56676
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.397263765335083,
+      "learning_rate": 1.5556922543019605e-07,
+      "loss": 0.801,
+      "step": 56677
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 1.7728638648986816,
+      "learning_rate": 1.5550397497210602e-07,
+      "loss": 0.4476,
+      "step": 56678
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.375483512878418,
+      "learning_rate": 1.5543873798474385e-07,
+      "loss": 0.3087,
+      "step": 56679
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.8718020915985107,
+      "learning_rate": 1.5537351446829218e-07,
+      "loss": 0.6889,
+      "step": 56680
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 4.906932830810547,
+      "learning_rate": 1.5530830442293143e-07,
+      "loss": 0.7318,
+      "step": 56681
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.718435764312744,
+      "learning_rate": 1.5524310784884422e-07,
+      "loss": 0.8398,
+      "step": 56682
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.2430315017700195,
+      "learning_rate": 1.5517792474620985e-07,
+      "loss": 0.5984,
+      "step": 56683
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.2086477279663086,
+      "learning_rate": 1.5511275511521208e-07,
+      "loss": 0.4077,
+      "step": 56684
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 4.803215026855469,
+      "learning_rate": 1.5504759895602962e-07,
+      "loss": 0.6081,
+      "step": 56685
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.999713659286499,
+      "learning_rate": 1.5498245626884568e-07,
+      "loss": 0.6231,
+      "step": 56686
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.702934980392456,
+      "learning_rate": 1.549173270538401e-07,
+      "loss": 0.5815,
+      "step": 56687
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.620579481124878,
+      "learning_rate": 1.54852211311195e-07,
+      "loss": 0.5923,
+      "step": 56688
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 5.727742671966553,
+      "learning_rate": 1.5478710904109017e-07,
+      "loss": 0.7836,
+      "step": 56689
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.237485408782959,
+      "learning_rate": 1.5472202024370776e-07,
+      "loss": 0.546,
+      "step": 56690
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.0779168605804443,
+      "learning_rate": 1.5465694491922812e-07,
+      "loss": 0.564,
+      "step": 56691
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.0428597927093506,
+      "learning_rate": 1.545918830678328e-07,
+      "loss": 0.5936,
+      "step": 56692
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.030480146408081,
+      "learning_rate": 1.545268346897011e-07,
+      "loss": 0.3721,
+      "step": 56693
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.4999806880950928,
+      "learning_rate": 1.544617997850162e-07,
+      "loss": 0.3674,
+      "step": 56694
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.9895260334014893,
+      "learning_rate": 1.5439677835395738e-07,
+      "loss": 0.3883,
+      "step": 56695
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.0679168701171875,
+      "learning_rate": 1.5433177039670676e-07,
+      "loss": 0.4721,
+      "step": 56696
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.151193380355835,
+      "learning_rate": 1.5426677591344308e-07,
+      "loss": 0.7234,
+      "step": 56697
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.501258611679077,
+      "learning_rate": 1.542017949043495e-07,
+      "loss": 0.5646,
+      "step": 56698
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.7192611694335938,
+      "learning_rate": 1.5413682736960424e-07,
+      "loss": 0.4797,
+      "step": 56699
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.452507972717285,
+      "learning_rate": 1.5407187330939045e-07,
+      "loss": 0.6431,
+      "step": 56700
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.433865547180176,
+      "learning_rate": 1.5400693272388635e-07,
+      "loss": 0.7214,
+      "step": 56701
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.794618606567383,
+      "learning_rate": 1.53942005613274e-07,
+      "loss": 0.4631,
+      "step": 56702
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.7950901985168457,
+      "learning_rate": 1.5387709197773326e-07,
+      "loss": 0.5605,
+      "step": 56703
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.6482741832733154,
+      "learning_rate": 1.5381219181744567e-07,
+      "loss": 0.5779,
+      "step": 56704
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.8893089294433594,
+      "learning_rate": 1.5374730513259052e-07,
+      "loss": 0.5759,
+      "step": 56705
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.702646493911743,
+      "learning_rate": 1.5368243192334875e-07,
+      "loss": 0.6487,
+      "step": 56706
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.213707208633423,
+      "learning_rate": 1.5361757218990026e-07,
+      "loss": 0.541,
+      "step": 56707
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 4.00986385345459,
+      "learning_rate": 1.5355272593242653e-07,
+      "loss": 0.4711,
+      "step": 56708
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 4.719272613525391,
+      "learning_rate": 1.5348789315110635e-07,
+      "loss": 0.5572,
+      "step": 56709
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.67464542388916,
+      "learning_rate": 1.5342307384612176e-07,
+      "loss": 0.4842,
+      "step": 56710
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.3427112102508545,
+      "learning_rate": 1.5335826801765096e-07,
+      "loss": 0.4651,
+      "step": 56711
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.667727470397949,
+      "learning_rate": 1.532934756658755e-07,
+      "loss": 0.7093,
+      "step": 56712
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.350637435913086,
+      "learning_rate": 1.5322869679097518e-07,
+      "loss": 0.3212,
+      "step": 56713
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.8003392219543457,
+      "learning_rate": 1.5316393139313101e-07,
+      "loss": 0.6034,
+      "step": 56714
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.476144790649414,
+      "learning_rate": 1.5309917947252117e-07,
+      "loss": 0.5752,
+      "step": 56715
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.736055374145508,
+      "learning_rate": 1.5303444102932775e-07,
+      "loss": 0.3957,
+      "step": 56716
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 4.327470302581787,
+      "learning_rate": 1.5296971606372946e-07,
+      "loss": 0.7088,
+      "step": 56717
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.696938991546631,
+      "learning_rate": 1.5290500457590675e-07,
+      "loss": 0.5185,
+      "step": 56718
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.6922011375427246,
+      "learning_rate": 1.5284030656603944e-07,
+      "loss": 0.643,
+      "step": 56719
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.604402542114258,
+      "learning_rate": 1.527756220343074e-07,
+      "loss": 0.3212,
+      "step": 56720
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 1.9219013452529907,
+      "learning_rate": 1.5271095098089052e-07,
+      "loss": 0.3509,
+      "step": 56721
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.805537462234497,
+      "learning_rate": 1.5264629340596916e-07,
+      "loss": 0.9372,
+      "step": 56722
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.2120654582977295,
+      "learning_rate": 1.5258164930972152e-07,
+      "loss": 0.5698,
+      "step": 56723
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.042762279510498,
+      "learning_rate": 1.5251701869232916e-07,
+      "loss": 0.4933,
+      "step": 56724
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.1095314025878906,
+      "learning_rate": 1.5245240155397135e-07,
+      "loss": 0.4108,
+      "step": 56725
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.7888503074645996,
+      "learning_rate": 1.5238779789482684e-07,
+      "loss": 0.6841,
+      "step": 56726
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.70245361328125,
+      "learning_rate": 1.5232320771507603e-07,
+      "loss": 0.6453,
+      "step": 56727
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.1981399059295654,
+      "learning_rate": 1.5225863101489825e-07,
+      "loss": 0.7353,
+      "step": 56728
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.0854673385620117,
+      "learning_rate": 1.521940677944739e-07,
+      "loss": 0.6301,
+      "step": 56729
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 4.645040035247803,
+      "learning_rate": 1.5212951805398112e-07,
+      "loss": 0.6172,
+      "step": 56730
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.415512800216675,
+      "learning_rate": 1.5206498179360096e-07,
+      "loss": 0.4818,
+      "step": 56731
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.676666498184204,
+      "learning_rate": 1.52000459013511e-07,
+      "loss": 0.3626,
+      "step": 56732
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 4.155804634094238,
+      "learning_rate": 1.5193594971389225e-07,
+      "loss": 0.5313,
+      "step": 56733
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 4.204125881195068,
+      "learning_rate": 1.5187145389492287e-07,
+      "loss": 0.7749,
+      "step": 56734
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.4975714683532715,
+      "learning_rate": 1.5180697155678327e-07,
+      "loss": 0.4037,
+      "step": 56735
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.941885471343994,
+      "learning_rate": 1.5174250269965218e-07,
+      "loss": 0.8008,
+      "step": 56736
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.6877570152282715,
+      "learning_rate": 1.516780473237095e-07,
+      "loss": 0.6143,
+      "step": 56737
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.2143747806549072,
+      "learning_rate": 1.5161360542913338e-07,
+      "loss": 0.661,
+      "step": 56738
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.1912717819213867,
+      "learning_rate": 1.5154917701610372e-07,
+      "loss": 0.8115,
+      "step": 56739
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.1752407550811768,
+      "learning_rate": 1.5148476208479924e-07,
+      "loss": 0.6024,
+      "step": 56740
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.7011289596557617,
+      "learning_rate": 1.5142036063540033e-07,
+      "loss": 0.7959,
+      "step": 56741
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 4.074750900268555,
+      "learning_rate": 1.513559726680841e-07,
+      "loss": 0.7058,
+      "step": 56742
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 4.97252893447876,
+      "learning_rate": 1.5129159818303096e-07,
+      "loss": 0.8438,
+      "step": 56743
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.154930591583252,
+      "learning_rate": 1.512272371804191e-07,
+      "loss": 0.6559,
+      "step": 56744
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.6872663497924805,
+      "learning_rate": 1.5116288966042892e-07,
+      "loss": 0.567,
+      "step": 56745
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.0294740200042725,
+      "learning_rate": 1.510985556232375e-07,
+      "loss": 0.5604,
+      "step": 56746
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.902268409729004,
+      "learning_rate": 1.510342350690247e-07,
+      "loss": 0.6561,
+      "step": 56747
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.9937782287597656,
+      "learning_rate": 1.5096992799796928e-07,
+      "loss": 0.6198,
+      "step": 56748
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 4.297765254974365,
+      "learning_rate": 1.5090563441024996e-07,
+      "loss": 0.5637,
+      "step": 56749
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.711488723754883,
+      "learning_rate": 1.508413543060455e-07,
+      "loss": 0.3813,
+      "step": 56750
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.5437920093536377,
+      "learning_rate": 1.5077708768553524e-07,
+      "loss": 0.6707,
+      "step": 56751
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.6886537075042725,
+      "learning_rate": 1.5071283454889618e-07,
+      "loss": 0.4925,
+      "step": 56752
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.973177671432495,
+      "learning_rate": 1.5064859489630935e-07,
+      "loss": 0.6602,
+      "step": 56753
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.5026304721832275,
+      "learning_rate": 1.5058436872795068e-07,
+      "loss": 0.7162,
+      "step": 56754
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.2563955783843994,
+      "learning_rate": 1.5052015604400173e-07,
+      "loss": 0.7969,
+      "step": 56755
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.511763095855713,
+      "learning_rate": 1.5045595684463842e-07,
+      "loss": 0.6759,
+      "step": 56756
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.7885143756866455,
+      "learning_rate": 1.5039177113004177e-07,
+      "loss": 0.77,
+      "step": 56757
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 6.051385879516602,
+      "learning_rate": 1.5032759890038773e-07,
+      "loss": 0.6037,
+      "step": 56758
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.2631266117095947,
+      "learning_rate": 1.502634401558567e-07,
+      "loss": 0.2886,
+      "step": 56759
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 1.9939005374908447,
+      "learning_rate": 1.5019929489662576e-07,
+      "loss": 0.5392,
+      "step": 56760
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.802603006362915,
+      "learning_rate": 1.5013516312287425e-07,
+      "loss": 0.4261,
+      "step": 56761
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.7944283485412598,
+      "learning_rate": 1.5007104483477974e-07,
+      "loss": 0.4731,
+      "step": 56762
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.4548134803771973,
+      "learning_rate": 1.5000694003252103e-07,
+      "loss": 0.6253,
+      "step": 56763
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 4.272271633148193,
+      "learning_rate": 1.4994284871627574e-07,
+      "loss": 0.6478,
+      "step": 56764
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.227182149887085,
+      "learning_rate": 1.4987877088622315e-07,
+      "loss": 0.5524,
+      "step": 56765
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.1392011642456055,
+      "learning_rate": 1.498147065425398e-07,
+      "loss": 0.8834,
+      "step": 56766
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.873223304748535,
+      "learning_rate": 1.4975065568540614e-07,
+      "loss": 0.7154,
+      "step": 56767
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.413635492324829,
+      "learning_rate": 1.4968661831499808e-07,
+      "loss": 0.3915,
+      "step": 56768
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 4.3272318840026855,
+      "learning_rate": 1.4962259443149497e-07,
+      "loss": 0.7805,
+      "step": 56769
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 4.674956321716309,
+      "learning_rate": 1.4955858403507383e-07,
+      "loss": 0.6995,
+      "step": 56770
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.0396196842193604,
+      "learning_rate": 1.4949458712591403e-07,
+      "loss": 0.3561,
+      "step": 56771
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.4609663486480713,
+      "learning_rate": 1.4943060370419204e-07,
+      "loss": 0.6695,
+      "step": 56772
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.4794819355010986,
+      "learning_rate": 1.4936663377008666e-07,
+      "loss": 0.5014,
+      "step": 56773
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.646230936050415,
+      "learning_rate": 1.4930267732377545e-07,
+      "loss": 0.4389,
+      "step": 56774
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 5.572635173797607,
+      "learning_rate": 1.4923873436543724e-07,
+      "loss": 0.68,
+      "step": 56775
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 1.9239025115966797,
+      "learning_rate": 1.4917480489524793e-07,
+      "loss": 0.4057,
+      "step": 56776
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.683119297027588,
+      "learning_rate": 1.4911088891338687e-07,
+      "loss": 0.4228,
+      "step": 56777
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 1.9716074466705322,
+      "learning_rate": 1.4904698642003112e-07,
+      "loss": 0.3467,
+      "step": 56778
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.9927711486816406,
+      "learning_rate": 1.4898309741535887e-07,
+      "loss": 0.5933,
+      "step": 56779
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.854727268218994,
+      "learning_rate": 1.4891922189954665e-07,
+      "loss": 0.5238,
+      "step": 56780
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.6967933177948,
+      "learning_rate": 1.4885535987277377e-07,
+      "loss": 0.6332,
+      "step": 56781
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.8493762016296387,
+      "learning_rate": 1.487915113352162e-07,
+      "loss": 0.5645,
+      "step": 56782
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.763692855834961,
+      "learning_rate": 1.487276762870521e-07,
+      "loss": 0.5188,
+      "step": 56783
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.3109476566314697,
+      "learning_rate": 1.4866385472845912e-07,
+      "loss": 0.6911,
+      "step": 56784
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 5.057995796203613,
+      "learning_rate": 1.4860004665961436e-07,
+      "loss": 0.6941,
+      "step": 56785
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.647960901260376,
+      "learning_rate": 1.48536252080696e-07,
+      "loss": 0.4841,
+      "step": 56786
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.2396578788757324,
+      "learning_rate": 1.4847247099188055e-07,
+      "loss": 0.3202,
+      "step": 56787
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.7992165088653564,
+      "learning_rate": 1.4840870339334624e-07,
+      "loss": 0.5219,
+      "step": 56788
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.0665712356567383,
+      "learning_rate": 1.48344949285269e-07,
+      "loss": 0.553,
+      "step": 56789
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.3006675243377686,
+      "learning_rate": 1.4828120866782814e-07,
+      "loss": 0.4079,
+      "step": 56790
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.1983461380004883,
+      "learning_rate": 1.4821748154119853e-07,
+      "loss": 0.5577,
+      "step": 56791
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.20872163772583,
+      "learning_rate": 1.4815376790555948e-07,
+      "loss": 0.2653,
+      "step": 56792
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.2831027507781982,
+      "learning_rate": 1.4809006776108693e-07,
+      "loss": 0.6983,
+      "step": 56793
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.1734628677368164,
+      "learning_rate": 1.4802638110795852e-07,
+      "loss": 0.2041,
+      "step": 56794
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 4.157134056091309,
+      "learning_rate": 1.4796270794635082e-07,
+      "loss": 0.9377,
+      "step": 56795
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.256253242492676,
+      "learning_rate": 1.4789904827644086e-07,
+      "loss": 0.3867,
+      "step": 56796
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.0464282035827637,
+      "learning_rate": 1.478354020984063e-07,
+      "loss": 0.5417,
+      "step": 56797
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.777060031890869,
+      "learning_rate": 1.4777176941242478e-07,
+      "loss": 0.4552,
+      "step": 56798
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 4.75602388381958,
+      "learning_rate": 1.4770815021867114e-07,
+      "loss": 0.6854,
+      "step": 56799
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.066453218460083,
+      "learning_rate": 1.4764454451732414e-07,
+      "loss": 0.5285,
+      "step": 56800
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 4.392385959625244,
+      "learning_rate": 1.475809523085592e-07,
+      "loss": 0.773,
+      "step": 56801
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 1.989492654800415,
+      "learning_rate": 1.4751737359255446e-07,
+      "loss": 0.3091,
+      "step": 56802
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.7352356910705566,
+      "learning_rate": 1.4745380836948597e-07,
+      "loss": 0.7703,
+      "step": 56803
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.3590614795684814,
+      "learning_rate": 1.4739025663953076e-07,
+      "loss": 0.5025,
+      "step": 56804
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.391446828842163,
+      "learning_rate": 1.4732671840286482e-07,
+      "loss": 0.4132,
+      "step": 56805
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.787569761276245,
+      "learning_rate": 1.4726319365966635e-07,
+      "loss": 0.621,
+      "step": 56806
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.356987953186035,
+      "learning_rate": 1.4719968241011017e-07,
+      "loss": 0.5044,
+      "step": 56807
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.894860029220581,
+      "learning_rate": 1.4713618465437452e-07,
+      "loss": 0.5476,
+      "step": 56808
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.434871196746826,
+      "learning_rate": 1.4707270039263477e-07,
+      "loss": 0.5056,
+      "step": 56809
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.7020082473754883,
+      "learning_rate": 1.4700922962506857e-07,
+      "loss": 0.5132,
+      "step": 56810
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.9342148303985596,
+      "learning_rate": 1.469457723518508e-07,
+      "loss": 0.578,
+      "step": 56811
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.627434730529785,
+      "learning_rate": 1.4688232857315964e-07,
+      "loss": 0.6692,
+      "step": 56812
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.1334166526794434,
+      "learning_rate": 1.4681889828916994e-07,
+      "loss": 0.5672,
+      "step": 56813
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.091768264770508,
+      "learning_rate": 1.4675548150005936e-07,
+      "loss": 0.367,
+      "step": 56814
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 13.443726539611816,
+      "learning_rate": 1.4669207820600272e-07,
+      "loss": 0.5175,
+      "step": 56815
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.3616747856140137,
+      "learning_rate": 1.4662868840717826e-07,
+      "loss": 0.4394,
+      "step": 56816
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.6311984062194824,
+      "learning_rate": 1.4656531210376136e-07,
+      "loss": 0.574,
+      "step": 56817
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.594597578048706,
+      "learning_rate": 1.4650194929592798e-07,
+      "loss": 0.5797,
+      "step": 56818
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.7999353408813477,
+      "learning_rate": 1.464385999838547e-07,
+      "loss": 0.7742,
+      "step": 56819
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.7715394496917725,
+      "learning_rate": 1.463752641677174e-07,
+      "loss": 0.6875,
+      "step": 56820
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.324889659881592,
+      "learning_rate": 1.4631194184769214e-07,
+      "loss": 0.49,
+      "step": 56821
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.169895887374878,
+      "learning_rate": 1.462486330239554e-07,
+      "loss": 0.703,
+      "step": 56822
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.6332027912139893,
+      "learning_rate": 1.4618533769668263e-07,
+      "loss": 0.5478,
+      "step": 56823
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 1.7527430057525635,
+      "learning_rate": 1.4612205586605032e-07,
+      "loss": 0.2147,
+      "step": 56824
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.5725882053375244,
+      "learning_rate": 1.460587875322339e-07,
+      "loss": 0.4663,
+      "step": 56825
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.8306632041931152,
+      "learning_rate": 1.4599553269541045e-07,
+      "loss": 0.3985,
+      "step": 56826
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.075056314468384,
+      "learning_rate": 1.4593229135575427e-07,
+      "loss": 0.3858,
+      "step": 56827
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.779029607772827,
+      "learning_rate": 1.4586906351344298e-07,
+      "loss": 0.5352,
+      "step": 56828
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.2319657802581787,
+      "learning_rate": 1.4580584916865036e-07,
+      "loss": 0.3739,
+      "step": 56829
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.0478672981262207,
+      "learning_rate": 1.4574264832155404e-07,
+      "loss": 0.4096,
+      "step": 56830
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 4.749795436859131,
+      "learning_rate": 1.4567946097232833e-07,
+      "loss": 0.5187,
+      "step": 56831
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.9692580699920654,
+      "learning_rate": 1.4561628712115028e-07,
+      "loss": 0.6877,
+      "step": 56832
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 4.176361560821533,
+      "learning_rate": 1.4555312676819423e-07,
+      "loss": 0.7001,
+      "step": 56833
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.7517504692077637,
+      "learning_rate": 1.4548997991363667e-07,
+      "loss": 0.5002,
+      "step": 56834
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.722041606903076,
+      "learning_rate": 1.4542684655765194e-07,
+      "loss": 0.6377,
+      "step": 56835
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 4.653817176818848,
+      "learning_rate": 1.453637267004182e-07,
+      "loss": 0.5259,
+      "step": 56836
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.994030475616455,
+      "learning_rate": 1.4530062034210812e-07,
+      "loss": 0.6878,
+      "step": 56837
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.258418321609497,
+      "learning_rate": 1.4523752748289933e-07,
+      "loss": 0.7018,
+      "step": 56838
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.1703875064849854,
+      "learning_rate": 1.451744481229661e-07,
+      "loss": 0.5039,
+      "step": 56839
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.1127960681915283,
+      "learning_rate": 1.4511138226248388e-07,
+      "loss": 0.659,
+      "step": 56840
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 4.888186454772949,
+      "learning_rate": 1.450483299016281e-07,
+      "loss": 0.8641,
+      "step": 56841
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.608929395675659,
+      "learning_rate": 1.4498529104057467e-07,
+      "loss": 0.516,
+      "step": 56842
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.286649465560913,
+      "learning_rate": 1.4492226567949797e-07,
+      "loss": 0.4766,
+      "step": 56843
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.6042542457580566,
+      "learning_rate": 1.4485925381857392e-07,
+      "loss": 0.4903,
+      "step": 56844
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.546541690826416,
+      "learning_rate": 1.4479625545797737e-07,
+      "loss": 0.3524,
+      "step": 56845
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.352006673812866,
+      "learning_rate": 1.4473327059788377e-07,
+      "loss": 0.4812,
+      "step": 56846
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.6837422847747803,
+      "learning_rate": 1.4467029923846798e-07,
+      "loss": 0.516,
+      "step": 56847
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 4.078876495361328,
+      "learning_rate": 1.446073413799054e-07,
+      "loss": 0.8005,
+      "step": 56848
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.58713698387146,
+      "learning_rate": 1.445443970223709e-07,
+      "loss": 0.5522,
+      "step": 56849
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.3502445220947266,
+      "learning_rate": 1.4448146616603932e-07,
+      "loss": 0.7156,
+      "step": 56850
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.1907694339752197,
+      "learning_rate": 1.4441854881108608e-07,
+      "loss": 0.419,
+      "step": 56851
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.763190269470215,
+      "learning_rate": 1.4435564495768605e-07,
+      "loss": 0.52,
+      "step": 56852
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.248917579650879,
+      "learning_rate": 1.442927546060141e-07,
+      "loss": 0.4723,
+      "step": 56853
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.2588186264038086,
+      "learning_rate": 1.4422987775624453e-07,
+      "loss": 0.3968,
+      "step": 56854
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.811002731323242,
+      "learning_rate": 1.4416701440855273e-07,
+      "loss": 0.7128,
+      "step": 56855
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.5920114517211914,
+      "learning_rate": 1.441041645631136e-07,
+      "loss": 0.6852,
+      "step": 56856
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.204286575317383,
+      "learning_rate": 1.4404132822010142e-07,
+      "loss": 0.4468,
+      "step": 56857
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.5984678268432617,
+      "learning_rate": 1.4397850537969104e-07,
+      "loss": 0.5264,
+      "step": 56858
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.321744918823242,
+      "learning_rate": 1.4391569604205845e-07,
+      "loss": 0.4547,
+      "step": 56859
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.19372296333313,
+      "learning_rate": 1.4385290020737573e-07,
+      "loss": 0.4747,
+      "step": 56860
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.1455721855163574,
+      "learning_rate": 1.4379011787582055e-07,
+      "loss": 0.4271,
+      "step": 56861
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.055948257446289,
+      "learning_rate": 1.4372734904756436e-07,
+      "loss": 0.4437,
+      "step": 56862
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 4.07484245300293,
+      "learning_rate": 1.436645937227843e-07,
+      "loss": 0.5664,
+      "step": 56863
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.4708476066589355,
+      "learning_rate": 1.43601851901653e-07,
+      "loss": 0.6426,
+      "step": 56864
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.588670015335083,
+      "learning_rate": 1.435391235843464e-07,
+      "loss": 0.5194,
+      "step": 56865
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 4.558998107910156,
+      "learning_rate": 1.4347640877103774e-07,
+      "loss": 0.5309,
+      "step": 56866
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 4.088741302490234,
+      "learning_rate": 1.4341370746190243e-07,
+      "loss": 0.6077,
+      "step": 56867
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 5.536384105682373,
+      "learning_rate": 1.4335101965711362e-07,
+      "loss": 0.7192,
+      "step": 56868
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.616453170776367,
+      "learning_rate": 1.4328834535684676e-07,
+      "loss": 0.6668,
+      "step": 56869
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 4.585185527801514,
+      "learning_rate": 1.4322568456127506e-07,
+      "loss": 0.8466,
+      "step": 56870
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.5118050575256348,
+      "learning_rate": 1.4316303727057446e-07,
+      "loss": 0.4171,
+      "step": 56871
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.5300726890563965,
+      "learning_rate": 1.4310040348491706e-07,
+      "loss": 0.5531,
+      "step": 56872
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.437591552734375,
+      "learning_rate": 1.4303778320447825e-07,
+      "loss": 0.6975,
+      "step": 56873
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.548614263534546,
+      "learning_rate": 1.429751764294318e-07,
+      "loss": 0.4646,
+      "step": 56874
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.1286120414733887,
+      "learning_rate": 1.4291258315995203e-07,
+      "loss": 0.6449,
+      "step": 56875
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.3460516929626465,
+      "learning_rate": 1.4285000339621212e-07,
+      "loss": 0.4328,
+      "step": 56876
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 6.587860107421875,
+      "learning_rate": 1.4278743713838805e-07,
+      "loss": 0.8443,
+      "step": 56877
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.5425643920898438,
+      "learning_rate": 1.4272488438665188e-07,
+      "loss": 0.491,
+      "step": 56878
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.8591067790985107,
+      "learning_rate": 1.426623451411785e-07,
+      "loss": 0.3587,
+      "step": 56879
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.9312539100646973,
+      "learning_rate": 1.4259981940214107e-07,
+      "loss": 0.7127,
+      "step": 56880
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.5493404865264893,
+      "learning_rate": 1.4253730716971504e-07,
+      "loss": 0.6047,
+      "step": 56881
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.484437942504883,
+      "learning_rate": 1.424748084440719e-07,
+      "loss": 0.4631,
+      "step": 56882
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.4326300621032715,
+      "learning_rate": 1.4241232322538713e-07,
+      "loss": 0.5208,
+      "step": 56883
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.4035494327545166,
+      "learning_rate": 1.4234985151383384e-07,
+      "loss": 0.5155,
+      "step": 56884
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.3913655281066895,
+      "learning_rate": 1.4228739330958642e-07,
+      "loss": 0.6525,
+      "step": 56885
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.8684260845184326,
+      "learning_rate": 1.4222494861281745e-07,
+      "loss": 0.5843,
+      "step": 56886
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.6415538787841797,
+      "learning_rate": 1.4216251742370178e-07,
+      "loss": 0.3862,
+      "step": 56887
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.290942907333374,
+      "learning_rate": 1.4210009974241156e-07,
+      "loss": 0.3508,
+      "step": 56888
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.9107437133789062,
+      "learning_rate": 1.4203769556912162e-07,
+      "loss": 0.5394,
+      "step": 56889
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.500871419906616,
+      "learning_rate": 1.4197530490400459e-07,
+      "loss": 0.6243,
+      "step": 56890
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.1064608097076416,
+      "learning_rate": 1.419129277472353e-07,
+      "loss": 0.6458,
+      "step": 56891
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.3899354934692383,
+      "learning_rate": 1.4185056409898535e-07,
+      "loss": 0.4918,
+      "step": 56892
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.3331828117370605,
+      "learning_rate": 1.4178821395942954e-07,
+      "loss": 0.3569,
+      "step": 56893
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.371997594833374,
+      "learning_rate": 1.4172587732874054e-07,
+      "loss": 0.6216,
+      "step": 56894
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.526730537414551,
+      "learning_rate": 1.416635542070921e-07,
+      "loss": 0.7717,
+      "step": 56895
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.934221029281616,
+      "learning_rate": 1.416012445946563e-07,
+      "loss": 0.5801,
+      "step": 56896
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.318650007247925,
+      "learning_rate": 1.4153894849160853e-07,
+      "loss": 0.7221,
+      "step": 56897
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.372375965118408,
+      "learning_rate": 1.414766658981209e-07,
+      "loss": 0.5442,
+      "step": 56898
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.7822065353393555,
+      "learning_rate": 1.4141439681436665e-07,
+      "loss": 0.5326,
+      "step": 56899
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 4.374467372894287,
+      "learning_rate": 1.4135214124051833e-07,
+      "loss": 0.6268,
+      "step": 56900
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.351181745529175,
+      "learning_rate": 1.4128989917675028e-07,
+      "loss": 0.617,
+      "step": 56901
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.900570869445801,
+      "learning_rate": 1.412276706232346e-07,
+      "loss": 0.6662,
+      "step": 56902
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.731034278869629,
+      "learning_rate": 1.4116545558014506e-07,
+      "loss": 0.9421,
+      "step": 56903
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.129380226135254,
+      "learning_rate": 1.411032540476537e-07,
+      "loss": 0.3833,
+      "step": 56904
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 4.238342761993408,
+      "learning_rate": 1.4104106602593427e-07,
+      "loss": 0.7636,
+      "step": 56905
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.111569404602051,
+      "learning_rate": 1.4097889151515886e-07,
+      "loss": 0.5433,
+      "step": 56906
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.0604050159454346,
+      "learning_rate": 1.409167305155018e-07,
+      "loss": 0.6335,
+      "step": 56907
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.534724712371826,
+      "learning_rate": 1.4085458302713462e-07,
+      "loss": 0.8276,
+      "step": 56908
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.259112596511841,
+      "learning_rate": 1.4079244905022992e-07,
+      "loss": 0.5287,
+      "step": 56909
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.722395420074463,
+      "learning_rate": 1.4073032858496205e-07,
+      "loss": 0.6623,
+      "step": 56910
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 4.3271284103393555,
+      "learning_rate": 1.406682216315025e-07,
+      "loss": 0.7005,
+      "step": 56911
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.5417418479919434,
+      "learning_rate": 1.406061281900245e-07,
+      "loss": 0.6972,
+      "step": 56912
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.0352895259857178,
+      "learning_rate": 1.4054404826070012e-07,
+      "loss": 0.4747,
+      "step": 56913
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.6274213790893555,
+      "learning_rate": 1.4048198184370253e-07,
+      "loss": 0.5135,
+      "step": 56914
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.132577896118164,
+      "learning_rate": 1.404199289392044e-07,
+      "loss": 0.7418,
+      "step": 56915
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 4.471600532531738,
+      "learning_rate": 1.4035788954737782e-07,
+      "loss": 0.6761,
+      "step": 56916
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.5612704753875732,
+      "learning_rate": 1.4029586366839487e-07,
+      "loss": 0.8285,
+      "step": 56917
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.9892783164978027,
+      "learning_rate": 1.4023385130242927e-07,
+      "loss": 0.5174,
+      "step": 56918
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.898374080657959,
+      "learning_rate": 1.4017185244965314e-07,
+      "loss": 0.6497,
+      "step": 56919
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.0569727420806885,
+      "learning_rate": 1.4010986711023744e-07,
+      "loss": 0.5241,
+      "step": 56920
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.386505365371704,
+      "learning_rate": 1.4004789528435648e-07,
+      "loss": 0.6723,
+      "step": 56921
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.887482166290283,
+      "learning_rate": 1.3998593697218177e-07,
+      "loss": 0.5936,
+      "step": 56922
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 4.77940034866333,
+      "learning_rate": 1.3992399217388485e-07,
+      "loss": 0.3885,
+      "step": 56923
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.4641647338867188,
+      "learning_rate": 1.398620608896395e-07,
+      "loss": 0.5503,
+      "step": 56924
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.1699740886688232,
+      "learning_rate": 1.3980014311961665e-07,
+      "loss": 0.6024,
+      "step": 56925
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.820512533187866,
+      "learning_rate": 1.397382388639895e-07,
+      "loss": 0.5635,
+      "step": 56926
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.1946451663970947,
+      "learning_rate": 1.3967634812292908e-07,
+      "loss": 0.3759,
+      "step": 56927
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.559746265411377,
+      "learning_rate": 1.3961447089660852e-07,
+      "loss": 0.6764,
+      "step": 56928
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.9754042625427246,
+      "learning_rate": 1.3955260718519881e-07,
+      "loss": 0.7804,
+      "step": 56929
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.126349687576294,
+      "learning_rate": 1.3949075698887314e-07,
+      "loss": 0.5086,
+      "step": 56930
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.3852877616882324,
+      "learning_rate": 1.3942892030780198e-07,
+      "loss": 0.6625,
+      "step": 56931
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.7441368103027344,
+      "learning_rate": 1.39367097142159e-07,
+      "loss": 0.512,
+      "step": 56932
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.3073105812072754,
+      "learning_rate": 1.393052874921147e-07,
+      "loss": 0.3546,
+      "step": 56933
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 7.888769149780273,
+      "learning_rate": 1.392434913578422e-07,
+      "loss": 0.6613,
+      "step": 56934
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 4.078393936157227,
+      "learning_rate": 1.3918170873951197e-07,
+      "loss": 0.4685,
+      "step": 56935
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.020740270614624,
+      "learning_rate": 1.391199396372972e-07,
+      "loss": 0.6412,
+      "step": 56936
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.12823224067688,
+      "learning_rate": 1.3905818405136774e-07,
+      "loss": 0.5612,
+      "step": 56937
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 4.447317600250244,
+      "learning_rate": 1.3899644198189787e-07,
+      "loss": 0.5911,
+      "step": 56938
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.1931581497192383,
+      "learning_rate": 1.3893471342905695e-07,
+      "loss": 0.3047,
+      "step": 56939
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.100546360015869,
+      "learning_rate": 1.3887299839301815e-07,
+      "loss": 0.9688,
+      "step": 56940
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 4.838685989379883,
+      "learning_rate": 1.3881129687395245e-07,
+      "loss": 0.6824,
+      "step": 56941
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.504432201385498,
+      "learning_rate": 1.3874960887203138e-07,
+      "loss": 0.5686,
+      "step": 56942
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.829472064971924,
+      "learning_rate": 1.3868793438742644e-07,
+      "loss": 0.5318,
+      "step": 56943
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.4038009643554688,
+      "learning_rate": 1.3862627342030976e-07,
+      "loss": 0.4197,
+      "step": 56944
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.829721212387085,
+      "learning_rate": 1.385646259708523e-07,
+      "loss": 0.5811,
+      "step": 56945
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.3789143562316895,
+      "learning_rate": 1.3850299203922501e-07,
+      "loss": 0.4884,
+      "step": 56946
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.4551198482513428,
+      "learning_rate": 1.3844137162560002e-07,
+      "loss": 0.5491,
+      "step": 56947
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 4.473392486572266,
+      "learning_rate": 1.3837976473014824e-07,
+      "loss": 0.5221,
+      "step": 56948
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.904984712600708,
+      "learning_rate": 1.3831817135304126e-07,
+      "loss": 0.6445,
+      "step": 56949
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.035189151763916,
+      "learning_rate": 1.3825659149445058e-07,
+      "loss": 0.5751,
+      "step": 56950
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 4.170634746551514,
+      "learning_rate": 1.3819502515454607e-07,
+      "loss": 0.6803,
+      "step": 56951
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.0876591205596924,
+      "learning_rate": 1.3813347233350093e-07,
+      "loss": 0.6518,
+      "step": 56952
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 6.618950843811035,
+      "learning_rate": 1.3807193303148447e-07,
+      "loss": 0.4903,
+      "step": 56953
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.1137828826904297,
+      "learning_rate": 1.380104072486693e-07,
+      "loss": 0.7079,
+      "step": 56954
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.438947916030884,
+      "learning_rate": 1.3794889498522534e-07,
+      "loss": 0.5686,
+      "step": 56955
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.3773343563079834,
+      "learning_rate": 1.3788739624132464e-07,
+      "loss": 0.6065,
+      "step": 56956
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.0296714305877686,
+      "learning_rate": 1.3782591101713704e-07,
+      "loss": 0.6205,
+      "step": 56957
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.2674636840820312,
+      "learning_rate": 1.3776443931283467e-07,
+      "loss": 0.6178,
+      "step": 56958
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.227185010910034,
+      "learning_rate": 1.3770298112858792e-07,
+      "loss": 0.3168,
+      "step": 56959
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.028820753097534,
+      "learning_rate": 1.3764153646456779e-07,
+      "loss": 0.4944,
+      "step": 56960
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.756746768951416,
+      "learning_rate": 1.3758010532094468e-07,
+      "loss": 0.6757,
+      "step": 56961
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.4693603515625,
+      "learning_rate": 1.3751868769789012e-07,
+      "loss": 0.7143,
+      "step": 56962
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.8203115463256836,
+      "learning_rate": 1.3745728359557452e-07,
+      "loss": 0.7043,
+      "step": 56963
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.9072723388671875,
+      "learning_rate": 1.3739589301416888e-07,
+      "loss": 0.6038,
+      "step": 56964
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 4.779133319854736,
+      "learning_rate": 1.3733451595384305e-07,
+      "loss": 0.64,
+      "step": 56965
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.565547227859497,
+      "learning_rate": 1.3727315241476912e-07,
+      "loss": 0.6431,
+      "step": 56966
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.5486676692962646,
+      "learning_rate": 1.372118023971164e-07,
+      "loss": 0.5229,
+      "step": 56967
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.8355584144592285,
+      "learning_rate": 1.3715046590105642e-07,
+      "loss": 0.5553,
+      "step": 56968
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.8040084838867188,
+      "learning_rate": 1.37089142926759e-07,
+      "loss": 0.682,
+      "step": 56969
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.1730527877807617,
+      "learning_rate": 1.3702783347439462e-07,
+      "loss": 0.5129,
+      "step": 56970
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.8126912117004395,
+      "learning_rate": 1.369665375441348e-07,
+      "loss": 0.6674,
+      "step": 56971
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.217848539352417,
+      "learning_rate": 1.3690525513614884e-07,
+      "loss": 0.5027,
+      "step": 56972
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.9763877391815186,
+      "learning_rate": 1.3684398625060823e-07,
+      "loss": 0.5261,
+      "step": 56973
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.4169535636901855,
+      "learning_rate": 1.3678273088768234e-07,
+      "loss": 0.5182,
+      "step": 56974
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.2872679233551025,
+      "learning_rate": 1.3672148904754213e-07,
+      "loss": 0.5507,
+      "step": 56975
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.475249767303467,
+      "learning_rate": 1.3666026073035688e-07,
+      "loss": 0.4951,
+      "step": 56976
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.1212806701660156,
+      "learning_rate": 1.3659904593629815e-07,
+      "loss": 0.5088,
+      "step": 56977
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.7354438304901123,
+      "learning_rate": 1.3653784466553577e-07,
+      "loss": 0.7171,
+      "step": 56978
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.5970723628997803,
+      "learning_rate": 1.3647665691823907e-07,
+      "loss": 0.6384,
+      "step": 56979
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.9269838333129883,
+      "learning_rate": 1.364154826945796e-07,
+      "loss": 0.6932,
+      "step": 56980
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.2266762256622314,
+      "learning_rate": 1.363543219947261e-07,
+      "loss": 0.8036,
+      "step": 56981
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.730229377746582,
+      "learning_rate": 1.3629317481884952e-07,
+      "loss": 0.8569,
+      "step": 56982
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.750786542892456,
+      "learning_rate": 1.3623204116711975e-07,
+      "loss": 0.6044,
+      "step": 56983
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.5432393550872803,
+      "learning_rate": 1.3617092103970664e-07,
+      "loss": 0.7742,
+      "step": 56984
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.968827486038208,
+      "learning_rate": 1.3610981443678006e-07,
+      "loss": 0.5298,
+      "step": 56985
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 1.9973859786987305,
+      "learning_rate": 1.3604872135851045e-07,
+      "loss": 0.3101,
+      "step": 56986
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 5.757291316986084,
+      "learning_rate": 1.3598764180506708e-07,
+      "loss": 0.6343,
+      "step": 56987
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.9020168781280518,
+      "learning_rate": 1.3592657577661927e-07,
+      "loss": 0.4973,
+      "step": 56988
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.7261931896209717,
+      "learning_rate": 1.3586552327333856e-07,
+      "loss": 0.4936,
+      "step": 56989
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.882260799407959,
+      "learning_rate": 1.3580448429539316e-07,
+      "loss": 0.6268,
+      "step": 56990
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.1156296730041504,
+      "learning_rate": 1.3574345884295347e-07,
+      "loss": 0.4589,
+      "step": 56991
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.9299912452697754,
+      "learning_rate": 1.356824469161888e-07,
+      "loss": 0.5922,
+      "step": 56992
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.6759579181671143,
+      "learning_rate": 1.356214485152696e-07,
+      "loss": 0.4609,
+      "step": 56993
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.342421770095825,
+      "learning_rate": 1.3556046364036402e-07,
+      "loss": 0.6784,
+      "step": 56994
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.7801098823547363,
+      "learning_rate": 1.354994922916436e-07,
+      "loss": 0.5876,
+      "step": 56995
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.9181482791900635,
+      "learning_rate": 1.3543853446927603e-07,
+      "loss": 0.4437,
+      "step": 56996
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.433762788772583,
+      "learning_rate": 1.3537759017343167e-07,
+      "loss": 0.5399,
+      "step": 56997
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 10.897089004516602,
+      "learning_rate": 1.353166594042793e-07,
+      "loss": 0.5533,
+      "step": 56998
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.7006072998046875,
+      "learning_rate": 1.3525574216198988e-07,
+      "loss": 0.4498,
+      "step": 56999
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.5109410285949707,
+      "learning_rate": 1.3519483844673164e-07,
+      "loss": 0.6942,
+      "step": 57000
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 4.005553722381592,
+      "learning_rate": 1.3513394825867443e-07,
+      "loss": 0.5366,
+      "step": 57001
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.942309856414795,
+      "learning_rate": 1.3507307159798644e-07,
+      "loss": 0.4749,
+      "step": 57002
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.6773221492767334,
+      "learning_rate": 1.3501220846483864e-07,
+      "loss": 0.6517,
+      "step": 57003
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 4.414402484893799,
+      "learning_rate": 1.3495135885939926e-07,
+      "loss": 0.4635,
+      "step": 57004
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.6000406742095947,
+      "learning_rate": 1.3489052278183756e-07,
+      "loss": 0.488,
+      "step": 57005
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.0698025226593018,
+      "learning_rate": 1.348297002323229e-07,
+      "loss": 0.5282,
+      "step": 57006
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.532177448272705,
+      "learning_rate": 1.3476889121102456e-07,
+      "loss": 0.4572,
+      "step": 57007
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.7476561069488525,
+      "learning_rate": 1.347080957181107e-07,
+      "loss": 0.6192,
+      "step": 57008
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.0317513942718506,
+      "learning_rate": 1.346473137537513e-07,
+      "loss": 0.5668,
+      "step": 57009
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.871795892715454,
+      "learning_rate": 1.3458654531811498e-07,
+      "loss": 0.5697,
+      "step": 57010
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.881969690322876,
+      "learning_rate": 1.345257904113717e-07,
+      "loss": 0.4273,
+      "step": 57011
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.52050518989563,
+      "learning_rate": 1.344650490336885e-07,
+      "loss": 0.5702,
+      "step": 57012
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.6282413005828857,
+      "learning_rate": 1.3440432118523583e-07,
+      "loss": 0.4783,
+      "step": 57013
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 4.277461528778076,
+      "learning_rate": 1.3434360686618188e-07,
+      "loss": 0.4848,
+      "step": 57014
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 5.328364849090576,
+      "learning_rate": 1.342829060766959e-07,
+      "loss": 0.7303,
+      "step": 57015
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.5652523040771484,
+      "learning_rate": 1.342222188169462e-07,
+      "loss": 0.6757,
+      "step": 57016
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.0877842903137207,
+      "learning_rate": 1.3416154508710198e-07,
+      "loss": 0.6738,
+      "step": 57017
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.2649788856506348,
+      "learning_rate": 1.3410088488733098e-07,
+      "loss": 0.6201,
+      "step": 57018
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.1213526725769043,
+      "learning_rate": 1.34040238217803e-07,
+      "loss": 0.7933,
+      "step": 57019
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.4068267345428467,
+      "learning_rate": 1.3397960507868625e-07,
+      "loss": 0.6441,
+      "step": 57020
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.8226680755615234,
+      "learning_rate": 1.3391898547015003e-07,
+      "loss": 0.7058,
+      "step": 57021
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.472987651824951,
+      "learning_rate": 1.3385837939236147e-07,
+      "loss": 0.7166,
+      "step": 57022
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.672659158706665,
+      "learning_rate": 1.3379778684549038e-07,
+      "loss": 0.5665,
+      "step": 57023
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 5.128246307373047,
+      "learning_rate": 1.3373720782970446e-07,
+      "loss": 0.4982,
+      "step": 57024
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.9758076667785645,
+      "learning_rate": 1.3367664234517242e-07,
+      "loss": 0.5123,
+      "step": 57025
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.9866743087768555,
+      "learning_rate": 1.3361609039206247e-07,
+      "loss": 0.5907,
+      "step": 57026
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.1496217250823975,
+      "learning_rate": 1.3355555197054392e-07,
+      "loss": 0.5075,
+      "step": 57027
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.9331061840057373,
+      "learning_rate": 1.3349502708078332e-07,
+      "loss": 0.5132,
+      "step": 57028
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.685591459274292,
+      "learning_rate": 1.3343451572295108e-07,
+      "loss": 0.6906,
+      "step": 57029
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.158731698989868,
+      "learning_rate": 1.3337401789721372e-07,
+      "loss": 0.2676,
+      "step": 57030
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.9930474758148193,
+      "learning_rate": 1.3331353360374e-07,
+      "loss": 0.5097,
+      "step": 57031
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.464430093765259,
+      "learning_rate": 1.3325306284269868e-07,
+      "loss": 0.6056,
+      "step": 57032
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.260983467102051,
+      "learning_rate": 1.331926056142574e-07,
+      "loss": 0.4305,
+      "step": 57033
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.8428006172180176,
+      "learning_rate": 1.3313216191858437e-07,
+      "loss": 0.6127,
+      "step": 57034
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.977290391921997,
+      "learning_rate": 1.330717317558472e-07,
+      "loss": 0.6486,
+      "step": 57035
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.5213520526885986,
+      "learning_rate": 1.3301131512621524e-07,
+      "loss": 0.7375,
+      "step": 57036
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.8639075756073,
+      "learning_rate": 1.3295091202985445e-07,
+      "loss": 0.7028,
+      "step": 57037
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.1848902702331543,
+      "learning_rate": 1.3289052246693468e-07,
+      "loss": 0.642,
+      "step": 57038
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 4.79368257522583,
+      "learning_rate": 1.3283014643762305e-07,
+      "loss": 0.3707,
+      "step": 57039
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.4884345531463623,
+      "learning_rate": 1.3276978394208717e-07,
+      "loss": 0.6667,
+      "step": 57040
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.367610454559326,
+      "learning_rate": 1.327094349804958e-07,
+      "loss": 0.4545,
+      "step": 57041
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.7487199306488037,
+      "learning_rate": 1.3264909955301553e-07,
+      "loss": 0.5401,
+      "step": 57042
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.1087727546691895,
+      "learning_rate": 1.3258877765981448e-07,
+      "loss": 0.6134,
+      "step": 57043
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 4.166691780090332,
+      "learning_rate": 1.3252846930106144e-07,
+      "loss": 0.6869,
+      "step": 57044
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.4877681732177734,
+      "learning_rate": 1.3246817447692295e-07,
+      "loss": 0.565,
+      "step": 57045
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.620612621307373,
+      "learning_rate": 1.3240789318756776e-07,
+      "loss": 0.6199,
+      "step": 57046
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.9948060512542725,
+      "learning_rate": 1.3234762543316183e-07,
+      "loss": 0.4749,
+      "step": 57047
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.3437211513519287,
+      "learning_rate": 1.3228737121387446e-07,
+      "loss": 0.7227,
+      "step": 57048
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.6736857891082764,
+      "learning_rate": 1.3222713052987168e-07,
+      "loss": 0.7003,
+      "step": 57049
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.8396072387695312,
+      "learning_rate": 1.321669033813222e-07,
+      "loss": 0.6462,
+      "step": 57050
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.929847002029419,
+      "learning_rate": 1.3210668976839257e-07,
+      "loss": 0.5997,
+      "step": 57051
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 4.095726490020752,
+      "learning_rate": 1.3204648969125156e-07,
+      "loss": 0.7099,
+      "step": 57052
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.207160234451294,
+      "learning_rate": 1.319863031500651e-07,
+      "loss": 0.4402,
+      "step": 57053
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.4144139289855957,
+      "learning_rate": 1.3192613014500143e-07,
+      "loss": 0.5037,
+      "step": 57054
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 1.8772834539413452,
+      "learning_rate": 1.3186597067622708e-07,
+      "loss": 0.31,
+      "step": 57055
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.871588706970215,
+      "learning_rate": 1.3180582474391024e-07,
+      "loss": 0.6578,
+      "step": 57056
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.2623789310455322,
+      "learning_rate": 1.3174569234821743e-07,
+      "loss": 0.7481,
+      "step": 57057
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.3733348846435547,
+      "learning_rate": 1.3168557348931631e-07,
+      "loss": 0.489,
+      "step": 57058
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.088470935821533,
+      "learning_rate": 1.3162546816737344e-07,
+      "loss": 0.5516,
+      "step": 57059
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.152578592300415,
+      "learning_rate": 1.3156537638255697e-07,
+      "loss": 0.4085,
+      "step": 57060
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.27091908454895,
+      "learning_rate": 1.315052981350329e-07,
+      "loss": 0.5741,
+      "step": 57061
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.690927267074585,
+      "learning_rate": 1.3144523342496995e-07,
+      "loss": 0.6847,
+      "step": 57062
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 5.108325958251953,
+      "learning_rate": 1.3138518225253306e-07,
+      "loss": 0.7793,
+      "step": 57063
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.3150827884674072,
+      "learning_rate": 1.3132514461789038e-07,
+      "loss": 0.618,
+      "step": 57064
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.7797796726226807,
+      "learning_rate": 1.3126512052120843e-07,
+      "loss": 0.8209,
+      "step": 57065
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 4.188164710998535,
+      "learning_rate": 1.312051099626549e-07,
+      "loss": 0.7251,
+      "step": 57066
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.9135966300964355,
+      "learning_rate": 1.3114511294239518e-07,
+      "loss": 0.6937,
+      "step": 57067
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.3848373889923096,
+      "learning_rate": 1.3108512946059804e-07,
+      "loss": 0.6,
+      "step": 57068
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.5577738285064697,
+      "learning_rate": 1.3102515951742833e-07,
+      "loss": 0.6203,
+      "step": 57069
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.4713547229766846,
+      "learning_rate": 1.3096520311305429e-07,
+      "loss": 0.5608,
+      "step": 57070
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.7129881381988525,
+      "learning_rate": 1.3090526024764126e-07,
+      "loss": 0.6064,
+      "step": 57071
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.508962392807007,
+      "learning_rate": 1.3084533092135753e-07,
+      "loss": 0.3725,
+      "step": 57072
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.8013861179351807,
+      "learning_rate": 1.3078541513436848e-07,
+      "loss": 0.5077,
+      "step": 57073
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.10597562789917,
+      "learning_rate": 1.3072551288684175e-07,
+      "loss": 0.7192,
+      "step": 57074
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.769580364227295,
+      "learning_rate": 1.3066562417894225e-07,
+      "loss": 0.62,
+      "step": 57075
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.4235174655914307,
+      "learning_rate": 1.3060574901083867e-07,
+      "loss": 0.5593,
+      "step": 57076
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.007159471511841,
+      "learning_rate": 1.3054588738269536e-07,
+      "loss": 0.5362,
+      "step": 57077
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.255255699157715,
+      "learning_rate": 1.3048603929468052e-07,
+      "loss": 0.418,
+      "step": 57078
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.6917850971221924,
+      "learning_rate": 1.30426204746959e-07,
+      "loss": 0.6132,
+      "step": 57079
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 5.381771564483643,
+      "learning_rate": 1.3036638373969844e-07,
+      "loss": 0.693,
+      "step": 57080
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.093527317047119,
+      "learning_rate": 1.3030657627306487e-07,
+      "loss": 0.6972,
+      "step": 57081
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.2041702270507812,
+      "learning_rate": 1.3024678234722476e-07,
+      "loss": 0.5948,
+      "step": 57082
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.1384737491607666,
+      "learning_rate": 1.3018700196234357e-07,
+      "loss": 0.7226,
+      "step": 57083
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.252234935760498,
+      "learning_rate": 1.3012723511858894e-07,
+      "loss": 0.4853,
+      "step": 57084
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.565065383911133,
+      "learning_rate": 1.3006748181612516e-07,
+      "loss": 0.6512,
+      "step": 57085
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.342026948928833,
+      "learning_rate": 1.300077420551199e-07,
+      "loss": 0.4696,
+      "step": 57086
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.5241029262542725,
+      "learning_rate": 1.2994801583573858e-07,
+      "loss": 0.5151,
+      "step": 57087
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.454071283340454,
+      "learning_rate": 1.2988830315814771e-07,
+      "loss": 0.7907,
+      "step": 57088
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.64316987991333,
+      "learning_rate": 1.2982860402251274e-07,
+      "loss": 0.586,
+      "step": 57089
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.5198116302490234,
+      "learning_rate": 1.2976891842900074e-07,
+      "loss": 0.4436,
+      "step": 57090
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.561737060546875,
+      "learning_rate": 1.2970924637777604e-07,
+      "loss": 0.6223,
+      "step": 57091
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.956294298171997,
+      "learning_rate": 1.296495878690063e-07,
+      "loss": 0.5016,
+      "step": 57092
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.8155717849731445,
+      "learning_rate": 1.2958994290285577e-07,
+      "loss": 0.6066,
+      "step": 57093
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.7179489135742188,
+      "learning_rate": 1.2953031147949102e-07,
+      "loss": 0.5239,
+      "step": 57094
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.205134153366089,
+      "learning_rate": 1.2947069359907915e-07,
+      "loss": 0.5239,
+      "step": 57095
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 4.081915855407715,
+      "learning_rate": 1.2941108926178337e-07,
+      "loss": 0.7631,
+      "step": 57096
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.4794304370880127,
+      "learning_rate": 1.2935149846777184e-07,
+      "loss": 0.6331,
+      "step": 57097
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.4226274490356445,
+      "learning_rate": 1.2929192121720836e-07,
+      "loss": 0.7644,
+      "step": 57098
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.8994405269622803,
+      "learning_rate": 1.2923235751026054e-07,
+      "loss": 0.5749,
+      "step": 57099
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 4.037416458129883,
+      "learning_rate": 1.2917280734709158e-07,
+      "loss": 0.4414,
+      "step": 57100
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.2076852321624756,
+      "learning_rate": 1.2911327072786918e-07,
+      "loss": 0.5363,
+      "step": 57101
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.816901206970215,
+      "learning_rate": 1.2905374765275813e-07,
+      "loss": 0.6464,
+      "step": 57102
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 4.52471399307251,
+      "learning_rate": 1.2899423812192336e-07,
+      "loss": 0.2895,
+      "step": 57103
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 6.327198028564453,
+      "learning_rate": 1.2893474213553136e-07,
+      "loss": 0.6104,
+      "step": 57104
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.8620996475219727,
+      "learning_rate": 1.2887525969374704e-07,
+      "loss": 0.6447,
+      "step": 57105
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.8783118724823,
+      "learning_rate": 1.2881579079673577e-07,
+      "loss": 0.5808,
+      "step": 57106
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.3933193683624268,
+      "learning_rate": 1.2875633544466304e-07,
+      "loss": 0.6169,
+      "step": 57107
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 4.068149566650391,
+      "learning_rate": 1.2869689363769422e-07,
+      "loss": 0.6567,
+      "step": 57108
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.1763432025909424,
+      "learning_rate": 1.2863746537599476e-07,
+      "loss": 0.5604,
+      "step": 57109
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.3106958866119385,
+      "learning_rate": 1.2857805065972895e-07,
+      "loss": 0.4436,
+      "step": 57110
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.722916841506958,
+      "learning_rate": 1.2851864948906334e-07,
+      "loss": 0.5171,
+      "step": 57111
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.162905216217041,
+      "learning_rate": 1.284592618641617e-07,
+      "loss": 0.3768,
+      "step": 57112
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.6600868701934814,
+      "learning_rate": 1.2839988778519107e-07,
+      "loss": 0.5226,
+      "step": 57113
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.902571439743042,
+      "learning_rate": 1.2834052725231415e-07,
+      "loss": 0.7571,
+      "step": 57114
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 4.615223407745361,
+      "learning_rate": 1.282811802656986e-07,
+      "loss": 0.597,
+      "step": 57115
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 4.032228946685791,
+      "learning_rate": 1.2822184682550697e-07,
+      "loss": 0.5644,
+      "step": 57116
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.074816942214966,
+      "learning_rate": 1.281625269319059e-07,
+      "loss": 0.7158,
+      "step": 57117
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.740391254425049,
+      "learning_rate": 1.2810322058505963e-07,
+      "loss": 0.4808,
+      "step": 57118
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.5163862705230713,
+      "learning_rate": 1.2804392778513364e-07,
+      "loss": 0.7083,
+      "step": 57119
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.6479339599609375,
+      "learning_rate": 1.2798464853229108e-07,
+      "loss": 0.6543,
+      "step": 57120
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.9485690593719482,
+      "learning_rate": 1.2792538282669964e-07,
+      "loss": 0.5099,
+      "step": 57121
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.062056064605713,
+      "learning_rate": 1.2786613066852193e-07,
+      "loss": 0.526,
+      "step": 57122
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.383812665939331,
+      "learning_rate": 1.278068920579234e-07,
+      "loss": 0.6985,
+      "step": 57123
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.7454354763031006,
+      "learning_rate": 1.2774766699506891e-07,
+      "loss": 0.6825,
+      "step": 57124
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.581698417663574,
+      "learning_rate": 1.2768845548012277e-07,
+      "loss": 0.9431,
+      "step": 57125
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.4188270568847656,
+      "learning_rate": 1.2762925751324984e-07,
+      "loss": 0.4749,
+      "step": 57126
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.4992926120758057,
+      "learning_rate": 1.27570073094615e-07,
+      "loss": 0.7436,
+      "step": 57127
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 4.9769158363342285,
+      "learning_rate": 1.2751090222438257e-07,
+      "loss": 0.5924,
+      "step": 57128
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.8913753032684326,
+      "learning_rate": 1.2745174490271683e-07,
+      "loss": 0.7619,
+      "step": 57129
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 6.086330890655518,
+      "learning_rate": 1.2739260112978214e-07,
+      "loss": 0.7388,
+      "step": 57130
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.563976287841797,
+      "learning_rate": 1.2733347090574443e-07,
+      "loss": 0.7311,
+      "step": 57131
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.3931186199188232,
+      "learning_rate": 1.2727435423076583e-07,
+      "loss": 0.6193,
+      "step": 57132
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.004409074783325,
+      "learning_rate": 1.2721525110501231e-07,
+      "loss": 0.6201,
+      "step": 57133
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.61316180229187,
+      "learning_rate": 1.2715616152864761e-07,
+      "loss": 0.4957,
+      "step": 57134
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.3988020420074463,
+      "learning_rate": 1.2709708550183664e-07,
+      "loss": 0.4839,
+      "step": 57135
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.0074210166931152,
+      "learning_rate": 1.270380230247431e-07,
+      "loss": 0.3624,
+      "step": 57136
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.785160541534424,
+      "learning_rate": 1.2697897409753134e-07,
+      "loss": 0.609,
+      "step": 57137
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.5171573162078857,
+      "learning_rate": 1.2691993872036511e-07,
+      "loss": 0.6256,
+      "step": 57138
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.2501065731048584,
+      "learning_rate": 1.2686091689340985e-07,
+      "loss": 0.4583,
+      "step": 57139
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.669874429702759,
+      "learning_rate": 1.2680190861682761e-07,
+      "loss": 0.3574,
+      "step": 57140
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.8082330226898193,
+      "learning_rate": 1.2674291389078497e-07,
+      "loss": 0.6603,
+      "step": 57141
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.8738977909088135,
+      "learning_rate": 1.2668393271544399e-07,
+      "loss": 0.5785,
+      "step": 57142
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.799708843231201,
+      "learning_rate": 1.2662496509096955e-07,
+      "loss": 0.6001,
+      "step": 57143
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.2999954223632812,
+      "learning_rate": 1.265660110175254e-07,
+      "loss": 0.6845,
+      "step": 57144
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.133180618286133,
+      "learning_rate": 1.2650707049527644e-07,
+      "loss": 0.4033,
+      "step": 57145
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.979715347290039,
+      "learning_rate": 1.2644814352438471e-07,
+      "loss": 0.6393,
+      "step": 57146
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.830416679382324,
+      "learning_rate": 1.263892301050157e-07,
+      "loss": 0.5699,
+      "step": 57147
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.399325370788574,
+      "learning_rate": 1.2633033023733143e-07,
+      "loss": 0.8858,
+      "step": 57148
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.047247886657715,
+      "learning_rate": 1.2627144392149792e-07,
+      "loss": 0.3484,
+      "step": 57149
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.7331161499023438,
+      "learning_rate": 1.2621257115767727e-07,
+      "loss": 0.4985,
+      "step": 57150
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.770503282546997,
+      "learning_rate": 1.2615371194603375e-07,
+      "loss": 0.4268,
+      "step": 57151
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.183398962020874,
+      "learning_rate": 1.2609486628673117e-07,
+      "loss": 0.6686,
+      "step": 57152
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 4.135092258453369,
+      "learning_rate": 1.2603603417993272e-07,
+      "loss": 0.6211,
+      "step": 57153
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 4.330853462219238,
+      "learning_rate": 1.259772156258021e-07,
+      "loss": 0.7465,
+      "step": 57154
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.7514548301696777,
+      "learning_rate": 1.2591841062450316e-07,
+      "loss": 0.6012,
+      "step": 57155
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.6480343341827393,
+      "learning_rate": 1.258596191761996e-07,
+      "loss": 0.5404,
+      "step": 57156
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.3008806705474854,
+      "learning_rate": 1.2580084128105406e-07,
+      "loss": 0.8039,
+      "step": 57157
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.4352006912231445,
+      "learning_rate": 1.257420769392309e-07,
+      "loss": 0.5518,
+      "step": 57158
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.1568655967712402,
+      "learning_rate": 1.2568332615089274e-07,
+      "loss": 0.5807,
+      "step": 57159
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.503695487976074,
+      "learning_rate": 1.2562458891620333e-07,
+      "loss": 0.5695,
+      "step": 57160
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.42008638381958,
+      "learning_rate": 1.2556586523532588e-07,
+      "loss": 0.6687,
+      "step": 57161
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.0226643085479736,
+      "learning_rate": 1.255071551084236e-07,
+      "loss": 0.3785,
+      "step": 57162
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.7140352725982666,
+      "learning_rate": 1.2544845853566024e-07,
+      "loss": 0.3931,
+      "step": 57163
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 4.59293270111084,
+      "learning_rate": 1.2538977551719844e-07,
+      "loss": 0.6474,
+      "step": 57164
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 4.13621187210083,
+      "learning_rate": 1.2533110605320197e-07,
+      "loss": 0.7852,
+      "step": 57165
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.702110528945923,
+      "learning_rate": 1.252724501438335e-07,
+      "loss": 0.5016,
+      "step": 57166
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 4.30046272277832,
+      "learning_rate": 1.2521380778925563e-07,
+      "loss": 0.6477,
+      "step": 57167
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.536221981048584,
+      "learning_rate": 1.2515517898963269e-07,
+      "loss": 0.5239,
+      "step": 57168
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.691450834274292,
+      "learning_rate": 1.2509656374512679e-07,
+      "loss": 0.6596,
+      "step": 57169
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.734550714492798,
+      "learning_rate": 1.250379620559017e-07,
+      "loss": 0.473,
+      "step": 57170
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 7.285667419433594,
+      "learning_rate": 1.249793739221189e-07,
+      "loss": 0.5812,
+      "step": 57171
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 4.772488117218018,
+      "learning_rate": 1.249207993439433e-07,
+      "loss": 0.6881,
+      "step": 57172
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.751239061355591,
+      "learning_rate": 1.248622383215359e-07,
+      "loss": 0.5991,
+      "step": 57173
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.8916852474212646,
+      "learning_rate": 1.2480369085506094e-07,
+      "loss": 0.4957,
+      "step": 57174
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.658468008041382,
+      "learning_rate": 1.2474515694468002e-07,
+      "loss": 0.4244,
+      "step": 57175
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.5897178649902344,
+      "learning_rate": 1.2468663659055692e-07,
+      "loss": 0.4444,
+      "step": 57176
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.0654613971710205,
+      "learning_rate": 1.2462812979285366e-07,
+      "loss": 0.3651,
+      "step": 57177
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.486008644104004,
+      "learning_rate": 1.2456963655173348e-07,
+      "loss": 0.5564,
+      "step": 57178
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.9091641902923584,
+      "learning_rate": 1.245111568673585e-07,
+      "loss": 0.6665,
+      "step": 57179
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.107126235961914,
+      "learning_rate": 1.2445269073989185e-07,
+      "loss": 0.4798,
+      "step": 57180
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.4026474952697754,
+      "learning_rate": 1.2439423816949514e-07,
+      "loss": 0.7271,
+      "step": 57181
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.3924741744995117,
+      "learning_rate": 1.243357991563321e-07,
+      "loss": 0.5891,
+      "step": 57182
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.681577205657959,
+      "learning_rate": 1.242773737005648e-07,
+      "loss": 0.3275,
+      "step": 57183
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.9258334636688232,
+      "learning_rate": 1.2421896180235593e-07,
+      "loss": 0.4767,
+      "step": 57184
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.550503730773926,
+      "learning_rate": 1.24160563461867e-07,
+      "loss": 0.5435,
+      "step": 57185
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.9549925327301025,
+      "learning_rate": 1.2410217867926179e-07,
+      "loss": 0.5193,
+      "step": 57186
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.0245800018310547,
+      "learning_rate": 1.240438074547007e-07,
+      "loss": 0.6092,
+      "step": 57187
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.9681129455566406,
+      "learning_rate": 1.239854497883486e-07,
+      "loss": 0.6261,
+      "step": 57188
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.2674782276153564,
+      "learning_rate": 1.2392710568036538e-07,
+      "loss": 0.4435,
+      "step": 57189
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.4520134925842285,
+      "learning_rate": 1.2386877513091478e-07,
+      "loss": 0.5811,
+      "step": 57190
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.99589204788208,
+      "learning_rate": 1.238104581401578e-07,
+      "loss": 0.5247,
+      "step": 57191
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.232276439666748,
+      "learning_rate": 1.2375215470825818e-07,
+      "loss": 0.4541,
+      "step": 57192
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.245727300643921,
+      "learning_rate": 1.2369386483537638e-07,
+      "loss": 0.5129,
+      "step": 57193
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.850867509841919,
+      "learning_rate": 1.2363558852167613e-07,
+      "loss": 0.7178,
+      "step": 57194
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.7018401622772217,
+      "learning_rate": 1.2357732576731784e-07,
+      "loss": 0.739,
+      "step": 57195
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.773834228515625,
+      "learning_rate": 1.235190765724642e-07,
+      "loss": 0.3486,
+      "step": 57196
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.732729196548462,
+      "learning_rate": 1.234608409372773e-07,
+      "loss": 0.5057,
+      "step": 57197
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.046335220336914,
+      "learning_rate": 1.2340261886191974e-07,
+      "loss": 0.3444,
+      "step": 57198
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.035806894302368,
+      "learning_rate": 1.23344410346552e-07,
+      "loss": 0.7518,
+      "step": 57199
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.797075033187866,
+      "learning_rate": 1.2328621539133668e-07,
+      "loss": 0.5669,
+      "step": 57200
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.623481512069702,
+      "learning_rate": 1.232280339964348e-07,
+      "loss": 0.4792,
+      "step": 57201
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.558391809463501,
+      "learning_rate": 1.2316986616201009e-07,
+      "loss": 0.6547,
+      "step": 57202
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.1009328365325928,
+      "learning_rate": 1.2311171188822247e-07,
+      "loss": 0.6558,
+      "step": 57203
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.7017600536346436,
+      "learning_rate": 1.2305357117523454e-07,
+      "loss": 0.4617,
+      "step": 57204
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.659830331802368,
+      "learning_rate": 1.229954440232073e-07,
+      "loss": 0.5206,
+      "step": 57205
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.305931806564331,
+      "learning_rate": 1.2293733043230337e-07,
+      "loss": 0.4384,
+      "step": 57206
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.8868637084960938,
+      "learning_rate": 1.2287923040268323e-07,
+      "loss": 0.3149,
+      "step": 57207
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 1.7584608793258667,
+      "learning_rate": 1.228211439345095e-07,
+      "loss": 0.346,
+      "step": 57208
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.9931883811950684,
+      "learning_rate": 1.227630710279426e-07,
+      "loss": 0.6219,
+      "step": 57209
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 1.8964468240737915,
+      "learning_rate": 1.2270501168314464e-07,
+      "loss": 0.3928,
+      "step": 57210
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.6527047157287598,
+      "learning_rate": 1.2264696590027713e-07,
+      "loss": 0.7287,
+      "step": 57211
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.627291440963745,
+      "learning_rate": 1.225889336795011e-07,
+      "loss": 0.485,
+      "step": 57212
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.619507312774658,
+      "learning_rate": 1.2253091502097802e-07,
+      "loss": 0.7172,
+      "step": 57213
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.9794363975524902,
+      "learning_rate": 1.2247290992487004e-07,
+      "loss": 0.7832,
+      "step": 57214
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.4828693866729736,
+      "learning_rate": 1.2241491839133703e-07,
+      "loss": 0.7047,
+      "step": 57215
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.7198941707611084,
+      "learning_rate": 1.2235694042054101e-07,
+      "loss": 0.5012,
+      "step": 57216
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.1524534225463867,
+      "learning_rate": 1.2229897601264307e-07,
+      "loss": 0.6724,
+      "step": 57217
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.0646822452545166,
+      "learning_rate": 1.2224102516780467e-07,
+      "loss": 0.6462,
+      "step": 57218
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.723374366760254,
+      "learning_rate": 1.2218308788618683e-07,
+      "loss": 0.5835,
+      "step": 57219
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.9324541091918945,
+      "learning_rate": 1.2212516416795052e-07,
+      "loss": 0.4996,
+      "step": 57220
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.13942289352417,
+      "learning_rate": 1.220672540132567e-07,
+      "loss": 0.6913,
+      "step": 57221
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.194352865219116,
+      "learning_rate": 1.2200935742226637e-07,
+      "loss": 0.5765,
+      "step": 57222
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.3418779373168945,
+      "learning_rate": 1.2195147439514056e-07,
+      "loss": 0.5516,
+      "step": 57223
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.031982898712158,
+      "learning_rate": 1.2189360493204127e-07,
+      "loss": 0.6043,
+      "step": 57224
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.685938596725464,
+      "learning_rate": 1.2183574903312735e-07,
+      "loss": 0.5548,
+      "step": 57225
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.4745607376098633,
+      "learning_rate": 1.2177790669856194e-07,
+      "loss": 0.6917,
+      "step": 57226
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.9918015003204346,
+      "learning_rate": 1.2172007792850383e-07,
+      "loss": 0.5491,
+      "step": 57227
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 4.305046558380127,
+      "learning_rate": 1.2166226272311453e-07,
+      "loss": 0.8537,
+      "step": 57228
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.6608457565307617,
+      "learning_rate": 1.2160446108255563e-07,
+      "loss": 0.7697,
+      "step": 57229
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.352839231491089,
+      "learning_rate": 1.2154667300698696e-07,
+      "loss": 0.4192,
+      "step": 57230
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.059021472930908,
+      "learning_rate": 1.2148889849657008e-07,
+      "loss": 0.6092,
+      "step": 57231
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.2973272800445557,
+      "learning_rate": 1.2143113755146429e-07,
+      "loss": 0.4683,
+      "step": 57232
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.8151731491088867,
+      "learning_rate": 1.213733901718317e-07,
+      "loss": 0.7166,
+      "step": 57233
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.5195224285125732,
+      "learning_rate": 1.213156563578316e-07,
+      "loss": 0.3957,
+      "step": 57234
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 4.316628932952881,
+      "learning_rate": 1.2125793610962556e-07,
+      "loss": 0.5954,
+      "step": 57235
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.7402539253234863,
+      "learning_rate": 1.2120022942737287e-07,
+      "loss": 0.5886,
+      "step": 57236
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 1.8407707214355469,
+      "learning_rate": 1.2114253631123508e-07,
+      "loss": 0.3342,
+      "step": 57237
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.4262917041778564,
+      "learning_rate": 1.2108485676137204e-07,
+      "loss": 0.6045,
+      "step": 57238
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.229552984237671,
+      "learning_rate": 1.2102719077794477e-07,
+      "loss": 0.6777,
+      "step": 57239
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.892296314239502,
+      "learning_rate": 1.209695383611126e-07,
+      "loss": 0.3761,
+      "step": 57240
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.3174262046813965,
+      "learning_rate": 1.20911899511037e-07,
+      "loss": 0.7368,
+      "step": 57241
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.438436269760132,
+      "learning_rate": 1.2085427422787676e-07,
+      "loss": 0.696,
+      "step": 57242
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.5735366344451904,
+      "learning_rate": 1.2079666251179344e-07,
+      "loss": 0.7233,
+      "step": 57243
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.4310402870178223,
+      "learning_rate": 1.2073906436294692e-07,
+      "loss": 0.5095,
+      "step": 57244
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.321624517440796,
+      "learning_rate": 1.2068147978149757e-07,
+      "loss": 0.7193,
+      "step": 57245
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.4443747997283936,
+      "learning_rate": 1.2062390876760478e-07,
+      "loss": 0.4749,
+      "step": 57246
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.28592586517334,
+      "learning_rate": 1.2056635132142947e-07,
+      "loss": 0.4406,
+      "step": 57247
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.6794676780700684,
+      "learning_rate": 1.2050880744313098e-07,
+      "loss": 0.4354,
+      "step": 57248
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.3847835063934326,
+      "learning_rate": 1.2045127713286974e-07,
+      "loss": 0.4157,
+      "step": 57249
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.527768135070801,
+      "learning_rate": 1.2039376039080508e-07,
+      "loss": 0.6856,
+      "step": 57250
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.7840075492858887,
+      "learning_rate": 1.2033625721709796e-07,
+      "loss": 0.4382,
+      "step": 57251
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 5.7220988273620605,
+      "learning_rate": 1.202787676119077e-07,
+      "loss": 0.5339,
+      "step": 57252
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.8397834300994873,
+      "learning_rate": 1.202212915753942e-07,
+      "loss": 0.4916,
+      "step": 57253
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 4.12620735168457,
+      "learning_rate": 1.2016382910771673e-07,
+      "loss": 0.5225,
+      "step": 57254
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.1375293731689453,
+      "learning_rate": 1.201063802090363e-07,
+      "loss": 0.5992,
+      "step": 57255
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 4.627770900726318,
+      "learning_rate": 1.2004894487951168e-07,
+      "loss": 0.5007,
+      "step": 57256
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.438957691192627,
+      "learning_rate": 1.199915231193033e-07,
+      "loss": 0.7047,
+      "step": 57257
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.5675811767578125,
+      "learning_rate": 1.1993411492856988e-07,
+      "loss": 0.6746,
+      "step": 57258
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.7651047706604004,
+      "learning_rate": 1.1987672030747187e-07,
+      "loss": 0.4104,
+      "step": 57259
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.8436055183410645,
+      "learning_rate": 1.1981933925616808e-07,
+      "loss": 0.4896,
+      "step": 57260
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.743565082550049,
+      "learning_rate": 1.1976197177481885e-07,
+      "loss": 0.6753,
+      "step": 57261
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.5662646293640137,
+      "learning_rate": 1.19704617863583e-07,
+      "loss": 0.5835,
+      "step": 57262
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.2505943775177,
+      "learning_rate": 1.1964727752262095e-07,
+      "loss": 0.4364,
+      "step": 57263
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.1371421813964844,
+      "learning_rate": 1.1958995075209146e-07,
+      "loss": 0.5298,
+      "step": 57264
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.64500093460083,
+      "learning_rate": 1.1953263755215438e-07,
+      "loss": 0.7623,
+      "step": 57265
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.7167296409606934,
+      "learning_rate": 1.1947533792296795e-07,
+      "loss": 0.4915,
+      "step": 57266
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 6.528163433074951,
+      "learning_rate": 1.1941805186469313e-07,
+      "loss": 0.5309,
+      "step": 57267
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.8540523052215576,
+      "learning_rate": 1.1936077937748815e-07,
+      "loss": 0.6505,
+      "step": 57268
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.5439720153808594,
+      "learning_rate": 1.1930352046151284e-07,
+      "loss": 0.6053,
+      "step": 57269
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.233574390411377,
+      "learning_rate": 1.1924627511692543e-07,
+      "loss": 0.4777,
+      "step": 57270
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.7092225551605225,
+      "learning_rate": 1.1918904334388637e-07,
+      "loss": 0.5915,
+      "step": 57271
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.403247833251953,
+      "learning_rate": 1.1913182514255329e-07,
+      "loss": 0.6551,
+      "step": 57272
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.3184316158294678,
+      "learning_rate": 1.1907462051308716e-07,
+      "loss": 0.7844,
+      "step": 57273
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.9394965171813965,
+      "learning_rate": 1.1901742945564565e-07,
+      "loss": 0.679,
+      "step": 57274
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.7841265201568604,
+      "learning_rate": 1.1896025197038863e-07,
+      "loss": 0.5951,
+      "step": 57275
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.2143049240112305,
+      "learning_rate": 1.1890308805747374e-07,
+      "loss": 0.5031,
+      "step": 57276
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.8944461345672607,
+      "learning_rate": 1.1884593771706199e-07,
+      "loss": 0.6669,
+      "step": 57277
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.430469512939453,
+      "learning_rate": 1.1878880094931045e-07,
+      "loss": 0.6911,
+      "step": 57278
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.1630702018737793,
+      "learning_rate": 1.1873167775437844e-07,
+      "loss": 0.5011,
+      "step": 57279
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.078559398651123,
+      "learning_rate": 1.1867456813242584e-07,
+      "loss": 0.5595,
+      "step": 57280
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.607111692428589,
+      "learning_rate": 1.1861747208360974e-07,
+      "loss": 0.4886,
+      "step": 57281
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.0451488494873047,
+      "learning_rate": 1.1856038960809113e-07,
+      "loss": 0.621,
+      "step": 57282
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.5747275352478027,
+      "learning_rate": 1.1850332070602599e-07,
+      "loss": 0.5443,
+      "step": 57283
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.9949309825897217,
+      "learning_rate": 1.1844626537757531e-07,
+      "loss": 0.5603,
+      "step": 57284
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.954594612121582,
+      "learning_rate": 1.1838922362289673e-07,
+      "loss": 0.6011,
+      "step": 57285
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.9542088508605957,
+      "learning_rate": 1.1833219544214902e-07,
+      "loss": 0.5563,
+      "step": 57286
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.5218677520751953,
+      "learning_rate": 1.1827518083549094e-07,
+      "loss": 0.5565,
+      "step": 57287
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.292965888977051,
+      "learning_rate": 1.182181798030807e-07,
+      "loss": 0.5004,
+      "step": 57288
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.34586763381958,
+      "learning_rate": 1.1816119234507706e-07,
+      "loss": 0.3948,
+      "step": 57289
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.573359727859497,
+      "learning_rate": 1.1810421846163822e-07,
+      "loss": 0.6793,
+      "step": 57290
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 1.957296371459961,
+      "learning_rate": 1.1804725815292295e-07,
+      "loss": 0.4628,
+      "step": 57291
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.1066110134124756,
+      "learning_rate": 1.1799031141908946e-07,
+      "loss": 0.5172,
+      "step": 57292
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.8605031967163086,
+      "learning_rate": 1.1793337826029538e-07,
+      "loss": 0.6192,
+      "step": 57293
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.867595672607422,
+      "learning_rate": 1.1787645867670061e-07,
+      "loss": 0.6284,
+      "step": 57294
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.017655849456787,
+      "learning_rate": 1.1781955266846224e-07,
+      "loss": 0.401,
+      "step": 57295
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.5588314533233643,
+      "learning_rate": 1.1776266023573846e-07,
+      "loss": 0.5475,
+      "step": 57296
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.1984364986419678,
+      "learning_rate": 1.1770578137868804e-07,
+      "loss": 0.6188,
+      "step": 57297
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.6331264972686768,
+      "learning_rate": 1.176489160974692e-07,
+      "loss": 0.5988,
+      "step": 57298
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.363471508026123,
+      "learning_rate": 1.1759206439223902e-07,
+      "loss": 0.466,
+      "step": 57299
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.6875576972961426,
+      "learning_rate": 1.1753522626315739e-07,
+      "loss": 0.9495,
+      "step": 57300
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.9803552627563477,
+      "learning_rate": 1.1747840171038028e-07,
+      "loss": 0.5921,
+      "step": 57301
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 5.132303237915039,
+      "learning_rate": 1.1742159073406701e-07,
+      "loss": 0.4189,
+      "step": 57302
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.8754453659057617,
+      "learning_rate": 1.1736479333437467e-07,
+      "loss": 0.5158,
+      "step": 57303
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.383045196533203,
+      "learning_rate": 1.173080095114626e-07,
+      "loss": 0.3877,
+      "step": 57304
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.334301710128784,
+      "learning_rate": 1.1725123926548787e-07,
+      "loss": 0.7298,
+      "step": 57305
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.517580270767212,
+      "learning_rate": 1.1719448259660814e-07,
+      "loss": 0.6619,
+      "step": 57306
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.745633363723755,
+      "learning_rate": 1.1713773950498109e-07,
+      "loss": 0.4162,
+      "step": 57307
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.5963873863220215,
+      "learning_rate": 1.1708100999076543e-07,
+      "loss": 0.5579,
+      "step": 57308
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.484600305557251,
+      "learning_rate": 1.1702429405411775e-07,
+      "loss": 0.3979,
+      "step": 57309
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.607452154159546,
+      "learning_rate": 1.1696759169519678e-07,
+      "loss": 0.6033,
+      "step": 57310
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.185840368270874,
+      "learning_rate": 1.1691090291415907e-07,
+      "loss": 0.6872,
+      "step": 57311
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.6045777797698975,
+      "learning_rate": 1.168542277111634e-07,
+      "loss": 0.19,
+      "step": 57312
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.630561351776123,
+      "learning_rate": 1.1679756608636683e-07,
+      "loss": 0.5183,
+      "step": 57313
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 5.565749645233154,
+      "learning_rate": 1.1674091803992705e-07,
+      "loss": 0.7389,
+      "step": 57314
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 4.658322334289551,
+      "learning_rate": 1.1668428357200112e-07,
+      "loss": 0.9301,
+      "step": 57315
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.633932590484619,
+      "learning_rate": 1.1662766268274728e-07,
+      "loss": 0.5621,
+      "step": 57316
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.513216733932495,
+      "learning_rate": 1.1657105537232205e-07,
+      "loss": 0.7272,
+      "step": 57317
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.8853375911712646,
+      "learning_rate": 1.1651446164088364e-07,
+      "loss": 0.612,
+      "step": 57318
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.645113468170166,
+      "learning_rate": 1.164578814885886e-07,
+      "loss": 0.5575,
+      "step": 57319
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.5190019607543945,
+      "learning_rate": 1.164013149155957e-07,
+      "loss": 0.552,
+      "step": 57320
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.7375552654266357,
+      "learning_rate": 1.1634476192206034e-07,
+      "loss": 0.5445,
+      "step": 57321
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.037836790084839,
+      "learning_rate": 1.1628822250814187e-07,
+      "loss": 0.7561,
+      "step": 57322
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.777686357498169,
+      "learning_rate": 1.1623169667399514e-07,
+      "loss": 0.6373,
+      "step": 57323
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 5.614842891693115,
+      "learning_rate": 1.1617518441977893e-07,
+      "loss": 0.6541,
+      "step": 57324
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.8443479537963867,
+      "learning_rate": 1.1611868574565033e-07,
+      "loss": 0.8868,
+      "step": 57325
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 4.051880836486816,
+      "learning_rate": 1.1606220065176588e-07,
+      "loss": 0.5095,
+      "step": 57326
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.7880759239196777,
+      "learning_rate": 1.1600572913828323e-07,
+      "loss": 0.618,
+      "step": 57327
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.946113348007202,
+      "learning_rate": 1.1594927120535892e-07,
+      "loss": 0.5516,
+      "step": 57328
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 4.397619247436523,
+      "learning_rate": 1.1589282685315007e-07,
+      "loss": 0.9113,
+      "step": 57329
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 5.7194294929504395,
+      "learning_rate": 1.1583639608181373e-07,
+      "loss": 0.7656,
+      "step": 57330
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 4.521806716918945,
+      "learning_rate": 1.1577997889150594e-07,
+      "loss": 0.6179,
+      "step": 57331
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.028275966644287,
+      "learning_rate": 1.1572357528238542e-07,
+      "loss": 0.5978,
+      "step": 57332
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.0532283782958984,
+      "learning_rate": 1.1566718525460763e-07,
+      "loss": 0.6408,
+      "step": 57333
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.183347702026367,
+      "learning_rate": 1.1561080880832965e-07,
+      "loss": 0.7163,
+      "step": 57334
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.3575809001922607,
+      "learning_rate": 1.1555444594370801e-07,
+      "loss": 0.4773,
+      "step": 57335
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.376481771469116,
+      "learning_rate": 1.1549809666089984e-07,
+      "loss": 0.4645,
+      "step": 57336
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 5.577714443206787,
+      "learning_rate": 1.1544176096006166e-07,
+      "loss": 0.6778,
+      "step": 57337
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.8003900051116943,
+      "learning_rate": 1.1538543884135056e-07,
+      "loss": 0.5148,
+      "step": 57338
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.6999120712280273,
+      "learning_rate": 1.1532913030492198e-07,
+      "loss": 0.4949,
+      "step": 57339
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.060088634490967,
+      "learning_rate": 1.1527283535093358e-07,
+      "loss": 0.3794,
+      "step": 57340
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 4.303103446960449,
+      "learning_rate": 1.152165539795419e-07,
+      "loss": 0.6998,
+      "step": 57341
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.6068928241729736,
+      "learning_rate": 1.1516028619090292e-07,
+      "loss": 0.6301,
+      "step": 57342
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.9185032844543457,
+      "learning_rate": 1.1510403198517317e-07,
+      "loss": 0.2973,
+      "step": 57343
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.9853875637054443,
+      "learning_rate": 1.1504779136250921e-07,
+      "loss": 0.4968,
+      "step": 57344
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 1.9491541385650635,
+      "learning_rate": 1.1499156432306702e-07,
+      "loss": 0.5166,
+      "step": 57345
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.454739809036255,
+      "learning_rate": 1.1493535086700425e-07,
+      "loss": 0.7355,
+      "step": 57346
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.938990831375122,
+      "learning_rate": 1.1487915099447523e-07,
+      "loss": 0.4396,
+      "step": 57347
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 4.29060173034668,
+      "learning_rate": 1.1482296470563814e-07,
+      "loss": 0.4415,
+      "step": 57348
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.818376302719116,
+      "learning_rate": 1.1476679200064789e-07,
+      "loss": 0.4185,
+      "step": 57349
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 1.9758446216583252,
+      "learning_rate": 1.1471063287966156e-07,
+      "loss": 0.4794,
+      "step": 57350
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 4.984532356262207,
+      "learning_rate": 1.1465448734283403e-07,
+      "loss": 0.533,
+      "step": 57351
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.2423150539398193,
+      "learning_rate": 1.1459835539032293e-07,
+      "loss": 0.8191,
+      "step": 57352
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.409327507019043,
+      "learning_rate": 1.1454223702228374e-07,
+      "loss": 0.7056,
+      "step": 57353
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.619948148727417,
+      "learning_rate": 1.1448613223887184e-07,
+      "loss": 0.4393,
+      "step": 57354
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.214252233505249,
+      "learning_rate": 1.1443004104024491e-07,
+      "loss": 0.6849,
+      "step": 57355
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.255506753921509,
+      "learning_rate": 1.1437396342655727e-07,
+      "loss": 0.6005,
+      "step": 57356
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.49881649017334,
+      "learning_rate": 1.1431789939796544e-07,
+      "loss": 0.5854,
+      "step": 57357
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.5144498348236084,
+      "learning_rate": 1.1426184895462488e-07,
+      "loss": 0.4985,
+      "step": 57358
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.8174681663513184,
+      "learning_rate": 1.1420581209669268e-07,
+      "loss": 0.6664,
+      "step": 57359
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.0828588008880615,
+      "learning_rate": 1.1414978882432315e-07,
+      "loss": 0.7601,
+      "step": 57360
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.395244598388672,
+      "learning_rate": 1.140937791376734e-07,
+      "loss": 0.4125,
+      "step": 57361
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.8415188789367676,
+      "learning_rate": 1.1403778303689828e-07,
+      "loss": 1.0373,
+      "step": 57362
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 4.3320183753967285,
+      "learning_rate": 1.139818005221538e-07,
+      "loss": 0.707,
+      "step": 57363
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.139163017272949,
+      "learning_rate": 1.1392583159359483e-07,
+      "loss": 0.5463,
+      "step": 57364
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.4807164669036865,
+      "learning_rate": 1.1386987625137902e-07,
+      "loss": 0.5943,
+      "step": 57365
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.6243858337402344,
+      "learning_rate": 1.1381393449566014e-07,
+      "loss": 0.7174,
+      "step": 57366
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.6162662506103516,
+      "learning_rate": 1.1375800632659417e-07,
+      "loss": 0.3833,
+      "step": 57367
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.978398084640503,
+      "learning_rate": 1.1370209174433655e-07,
+      "loss": 0.5831,
+      "step": 57368
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.631685495376587,
+      "learning_rate": 1.1364619074904382e-07,
+      "loss": 0.4307,
+      "step": 57369
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.7956020832061768,
+      "learning_rate": 1.1359030334086974e-07,
+      "loss": 0.4923,
+      "step": 57370
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 5.414146900177002,
+      "learning_rate": 1.135344295199714e-07,
+      "loss": 0.5185,
+      "step": 57371
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.188817024230957,
+      "learning_rate": 1.1347856928650259e-07,
+      "loss": 0.4705,
+      "step": 57372
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.4266302585601807,
+      "learning_rate": 1.1342272264061927e-07,
+      "loss": 0.6116,
+      "step": 57373
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.916734457015991,
+      "learning_rate": 1.1336688958247688e-07,
+      "loss": 0.5218,
+      "step": 57374
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.7211716175079346,
+      "learning_rate": 1.1331107011223142e-07,
+      "loss": 0.5582,
+      "step": 57375
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.610572576522827,
+      "learning_rate": 1.1325526423003608e-07,
+      "loss": 0.7775,
+      "step": 57376
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.7200498580932617,
+      "learning_rate": 1.1319947193604797e-07,
+      "loss": 0.6618,
+      "step": 57377
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 4.43357515335083,
+      "learning_rate": 1.131436932304214e-07,
+      "loss": 0.4896,
+      "step": 57378
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.948625087738037,
+      "learning_rate": 1.1308792811331127e-07,
+      "loss": 0.6676,
+      "step": 57379
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.2280075550079346,
+      "learning_rate": 1.1303217658487297e-07,
+      "loss": 0.4425,
+      "step": 57380
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.1731553077697754,
+      "learning_rate": 1.1297643864526198e-07,
+      "loss": 0.6109,
+      "step": 57381
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 5.4766035079956055,
+      "learning_rate": 1.1292071429463202e-07,
+      "loss": 0.6506,
+      "step": 57382
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.076578378677368,
+      "learning_rate": 1.1286500353313967e-07,
+      "loss": 0.41,
+      "step": 57383
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.4144840240478516,
+      "learning_rate": 1.1280930636093812e-07,
+      "loss": 0.7788,
+      "step": 57384
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.027289867401123,
+      "learning_rate": 1.1275362277818335e-07,
+      "loss": 0.4858,
+      "step": 57385
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.9524641036987305,
+      "learning_rate": 1.1269795278503026e-07,
+      "loss": 0.7885,
+      "step": 57386
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.9291114807128906,
+      "learning_rate": 1.1264229638163315e-07,
+      "loss": 0.6463,
+      "step": 57387
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.1209518909454346,
+      "learning_rate": 1.1258665356814691e-07,
+      "loss": 0.7002,
+      "step": 57388
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.5156242847442627,
+      "learning_rate": 1.1253102434472696e-07,
+      "loss": 0.6289,
+      "step": 57389
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.8386640548706055,
+      "learning_rate": 1.1247540871152651e-07,
+      "loss": 0.5735,
+      "step": 57390
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.583475112915039,
+      "learning_rate": 1.1241980666870156e-07,
+      "loss": 0.5228,
+      "step": 57391
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.5160470008850098,
+      "learning_rate": 1.1236421821640586e-07,
+      "loss": 0.8291,
+      "step": 57392
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 4.009751796722412,
+      "learning_rate": 1.1230864335479486e-07,
+      "loss": 0.8547,
+      "step": 57393
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.7863292694091797,
+      "learning_rate": 1.1225308208402175e-07,
+      "loss": 0.5136,
+      "step": 57394
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.624340772628784,
+      "learning_rate": 1.1219753440424253e-07,
+      "loss": 0.4465,
+      "step": 57395
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.880035638809204,
+      "learning_rate": 1.121420003156104e-07,
+      "loss": 0.3403,
+      "step": 57396
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.877490997314453,
+      "learning_rate": 1.120864798182808e-07,
+      "loss": 0.6741,
+      "step": 57397
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.3259496688842773,
+      "learning_rate": 1.1203097291240695e-07,
+      "loss": 0.562,
+      "step": 57398
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.6404330730438232,
+      "learning_rate": 1.1197547959814481e-07,
+      "loss": 0.3433,
+      "step": 57399
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 4.189505100250244,
+      "learning_rate": 1.1191999987564705e-07,
+      "loss": 0.6738,
+      "step": 57400
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 4.061244964599609,
+      "learning_rate": 1.1186453374506911e-07,
+      "loss": 0.5934,
+      "step": 57401
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 4.369009017944336,
+      "learning_rate": 1.118090812065642e-07,
+      "loss": 0.6891,
+      "step": 57402
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.88977313041687,
+      "learning_rate": 1.1175364226028773e-07,
+      "loss": 0.5807,
+      "step": 57403
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.7339425086975098,
+      "learning_rate": 1.1169821690639292e-07,
+      "loss": 0.4987,
+      "step": 57404
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.3571293354034424,
+      "learning_rate": 1.1164280514503411e-07,
+      "loss": 0.6413,
+      "step": 57405
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.7961905002593994,
+      "learning_rate": 1.1158740697636505e-07,
+      "loss": 0.4368,
+      "step": 57406
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.144627332687378,
+      "learning_rate": 1.1153202240054118e-07,
+      "loss": 0.6395,
+      "step": 57407
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.482771635055542,
+      "learning_rate": 1.1147665141771458e-07,
+      "loss": 0.5215,
+      "step": 57408
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.8527913093566895,
+      "learning_rate": 1.1142129402804069e-07,
+      "loss": 0.6592,
+      "step": 57409
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 4.674037456512451,
+      "learning_rate": 1.1136595023167218e-07,
+      "loss": 0.5911,
+      "step": 57410
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.5039448738098145,
+      "learning_rate": 1.1131062002876447e-07,
+      "loss": 0.6763,
+      "step": 57411
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.231297492980957,
+      "learning_rate": 1.1125530341947022e-07,
+      "loss": 0.4869,
+      "step": 57412
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.0492637157440186,
+      "learning_rate": 1.1120000040394319e-07,
+      "loss": 0.5841,
+      "step": 57413
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.101797342300415,
+      "learning_rate": 1.1114471098233826e-07,
+      "loss": 0.6137,
+      "step": 57414
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.4808621406555176,
+      "learning_rate": 1.1108943515480808e-07,
+      "loss": 0.4618,
+      "step": 57415
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.35090708732605,
+      "learning_rate": 1.1103417292150697e-07,
+      "loss": 0.775,
+      "step": 57416
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 7.5581793785095215,
+      "learning_rate": 1.1097892428258816e-07,
+      "loss": 0.3737,
+      "step": 57417
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.330322265625,
+      "learning_rate": 1.1092368923820596e-07,
+      "loss": 0.6587,
+      "step": 57418
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.391040563583374,
+      "learning_rate": 1.1086846778851301e-07,
+      "loss": 0.6119,
+      "step": 57419
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.5238287448883057,
+      "learning_rate": 1.1081325993366365e-07,
+      "loss": 0.5668,
+      "step": 57420
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.7685163021087646,
+      "learning_rate": 1.1075806567381109e-07,
+      "loss": 0.4292,
+      "step": 57421
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.3189802169799805,
+      "learning_rate": 1.1070288500910853e-07,
+      "loss": 0.6656,
+      "step": 57422
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.843531608581543,
+      "learning_rate": 1.1064771793970974e-07,
+      "loss": 0.4748,
+      "step": 57423
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.1487021446228027,
+      "learning_rate": 1.105925644657685e-07,
+      "loss": 0.3377,
+      "step": 57424
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.174562931060791,
+      "learning_rate": 1.1053742458743632e-07,
+      "loss": 0.3897,
+      "step": 57425
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.46268630027771,
+      "learning_rate": 1.1048229830486923e-07,
+      "loss": 0.4953,
+      "step": 57426
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.513094425201416,
+      "learning_rate": 1.1042718561821875e-07,
+      "loss": 0.6335,
+      "step": 57427
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.3212172985076904,
+      "learning_rate": 1.103720865276392e-07,
+      "loss": 0.546,
+      "step": 57428
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.26814603805542,
+      "learning_rate": 1.1031700103328213e-07,
+      "loss": 0.3623,
+      "step": 57429
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.1166012287139893,
+      "learning_rate": 1.1026192913530298e-07,
+      "loss": 0.8358,
+      "step": 57430
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.309737205505371,
+      "learning_rate": 1.1020687083385272e-07,
+      "loss": 0.4651,
+      "step": 57431
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.1532962322235107,
+      "learning_rate": 1.1015182612908626e-07,
+      "loss": 0.4872,
+      "step": 57432
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.5257177352905273,
+      "learning_rate": 1.1009679502115456e-07,
+      "loss": 0.4792,
+      "step": 57433
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.6352357864379883,
+      "learning_rate": 1.1004177751021306e-07,
+      "loss": 0.6602,
+      "step": 57434
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.5656535625457764,
+      "learning_rate": 1.0998677359641275e-07,
+      "loss": 0.6109,
+      "step": 57435
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 4.137993335723877,
+      "learning_rate": 1.0993178327990794e-07,
+      "loss": 0.5006,
+      "step": 57436
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.1515085697174072,
+      "learning_rate": 1.0987680656085075e-07,
+      "loss": 0.8258,
+      "step": 57437
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.021958589553833,
+      "learning_rate": 1.098218434393944e-07,
+      "loss": 0.7458,
+      "step": 57438
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 4.246359348297119,
+      "learning_rate": 1.0976689391569151e-07,
+      "loss": 0.6676,
+      "step": 57439
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.5804529190063477,
+      "learning_rate": 1.0971195798989475e-07,
+      "loss": 0.4664,
+      "step": 57440
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.6258583068847656,
+      "learning_rate": 1.0965703566215735e-07,
+      "loss": 0.3651,
+      "step": 57441
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.8110594749450684,
+      "learning_rate": 1.0960212693263195e-07,
+      "loss": 0.51,
+      "step": 57442
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.4093239307403564,
+      "learning_rate": 1.0954723180147065e-07,
+      "loss": 0.4468,
+      "step": 57443
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.1465659141540527,
+      "learning_rate": 1.0949235026882666e-07,
+      "loss": 0.4226,
+      "step": 57444
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.3295319080352783,
+      "learning_rate": 1.0943748233485152e-07,
+      "loss": 0.6327,
+      "step": 57445
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.3020384311676025,
+      "learning_rate": 1.0938262799970012e-07,
+      "loss": 0.3856,
+      "step": 57446
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.1441526412963867,
+      "learning_rate": 1.0932778726352289e-07,
+      "loss": 0.6514,
+      "step": 57447
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.835939407348633,
+      "learning_rate": 1.0927296012647303e-07,
+      "loss": 0.678,
+      "step": 57448
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 4.482985973358154,
+      "learning_rate": 1.0921814658870322e-07,
+      "loss": 0.5828,
+      "step": 57449
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.322894334793091,
+      "learning_rate": 1.0916334665036554e-07,
+      "loss": 0.8203,
+      "step": 57450
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.2328073978424072,
+      "learning_rate": 1.091085603116121e-07,
+      "loss": 0.5781,
+      "step": 57451
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.692748546600342,
+      "learning_rate": 1.090537875725961e-07,
+      "loss": 0.5079,
+      "step": 57452
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.3887624740600586,
+      "learning_rate": 1.0899902843346855e-07,
+      "loss": 0.4035,
+      "step": 57453
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.679283618927002,
+      "learning_rate": 1.0894428289438319e-07,
+      "loss": 0.5959,
+      "step": 57454
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.4371137619018555,
+      "learning_rate": 1.0888955095549159e-07,
+      "loss": 0.4943,
+      "step": 57455
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.696108818054199,
+      "learning_rate": 1.0883483261694582e-07,
+      "loss": 0.5583,
+      "step": 57456
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.7907776832580566,
+      "learning_rate": 1.0878012787889746e-07,
+      "loss": 0.5861,
+      "step": 57457
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 4.174241542816162,
+      "learning_rate": 1.0872543674150026e-07,
+      "loss": 1.0239,
+      "step": 57458
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.7756237983703613,
+      "learning_rate": 1.0867075920490466e-07,
+      "loss": 0.5488,
+      "step": 57459
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.5713398456573486,
+      "learning_rate": 1.0861609526926332e-07,
+      "loss": 0.6292,
+      "step": 57460
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.586566686630249,
+      "learning_rate": 1.085614449347283e-07,
+      "loss": 0.464,
+      "step": 57461
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.205624580383301,
+      "learning_rate": 1.0850680820145232e-07,
+      "loss": 0.4927,
+      "step": 57462
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.4123167991638184,
+      "learning_rate": 1.0845218506958522e-07,
+      "loss": 0.4908,
+      "step": 57463
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.7625181674957275,
+      "learning_rate": 1.0839757553928077e-07,
+      "loss": 0.4539,
+      "step": 57464
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 5.1586480140686035,
+      "learning_rate": 1.0834297961069051e-07,
+      "loss": 0.6086,
+      "step": 57465
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.185725688934326,
+      "learning_rate": 1.0828839728396545e-07,
+      "loss": 0.5148,
+      "step": 57466
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.8356850147247314,
+      "learning_rate": 1.0823382855925824e-07,
+      "loss": 0.6941,
+      "step": 57467
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.9781107902526855,
+      "learning_rate": 1.0817927343672041e-07,
+      "loss": 0.4407,
+      "step": 57468
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.7360308170318604,
+      "learning_rate": 1.0812473191650296e-07,
+      "loss": 0.6213,
+      "step": 57469
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 5.685962677001953,
+      "learning_rate": 1.0807020399875857e-07,
+      "loss": 0.5629,
+      "step": 57470
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.1992886066436768,
+      "learning_rate": 1.0801568968363818e-07,
+      "loss": 0.4344,
+      "step": 57471
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.2866249084472656,
+      "learning_rate": 1.0796118897129337e-07,
+      "loss": 0.4979,
+      "step": 57472
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 1.9286309480667114,
+      "learning_rate": 1.0790670186187624e-07,
+      "loss": 0.4317,
+      "step": 57473
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 4.006227016448975,
+      "learning_rate": 1.0785222835553776e-07,
+      "loss": 0.6282,
+      "step": 57474
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 4.509486675262451,
+      "learning_rate": 1.0779776845242951e-07,
+      "loss": 0.6881,
+      "step": 57475
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 10.924530982971191,
+      "learning_rate": 1.07743322152703e-07,
+      "loss": 0.6563,
+      "step": 57476
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.33542799949646,
+      "learning_rate": 1.0768888945651035e-07,
+      "loss": 0.4989,
+      "step": 57477
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.835374116897583,
+      "learning_rate": 1.0763447036400143e-07,
+      "loss": 0.4794,
+      "step": 57478
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.8834228515625,
+      "learning_rate": 1.075800648753289e-07,
+      "loss": 0.4761,
+      "step": 57479
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.7082746028900146,
+      "learning_rate": 1.0752567299064265e-07,
+      "loss": 0.6049,
+      "step": 57480
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.2390851974487305,
+      "learning_rate": 1.0747129471009532e-07,
+      "loss": 0.5026,
+      "step": 57481
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.899925947189331,
+      "learning_rate": 1.0741693003383735e-07,
+      "loss": 0.5019,
+      "step": 57482
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.2029061317443848,
+      "learning_rate": 1.0736257896202029e-07,
+      "loss": 0.4418,
+      "step": 57483
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.307737112045288,
+      "learning_rate": 1.0730824149479457e-07,
+      "loss": 0.5701,
+      "step": 57484
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.96567440032959,
+      "learning_rate": 1.0725391763231285e-07,
+      "loss": 0.374,
+      "step": 57485
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.432241201400757,
+      "learning_rate": 1.071996073747239e-07,
+      "loss": 0.4774,
+      "step": 57486
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.6565349102020264,
+      "learning_rate": 1.0714531072218037e-07,
+      "loss": 0.638,
+      "step": 57487
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 5.250286102294922,
+      "learning_rate": 1.070910276748327e-07,
+      "loss": 0.5063,
+      "step": 57488
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.8672397136688232,
+      "learning_rate": 1.0703675823283243e-07,
+      "loss": 0.5197,
+      "step": 57489
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.109494924545288,
+      "learning_rate": 1.0698250239632946e-07,
+      "loss": 0.4938,
+      "step": 57490
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.572786808013916,
+      "learning_rate": 1.0692826016547586e-07,
+      "loss": 0.4295,
+      "step": 57491
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.6018505096435547,
+      "learning_rate": 1.0687403154042097e-07,
+      "loss": 0.3506,
+      "step": 57492
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.8569843769073486,
+      "learning_rate": 1.0681981652131746e-07,
+      "loss": 0.2857,
+      "step": 57493
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.576618194580078,
+      "learning_rate": 1.0676561510831407e-07,
+      "loss": 0.6444,
+      "step": 57494
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.084902048110962,
+      "learning_rate": 1.0671142730156292e-07,
+      "loss": 0.6216,
+      "step": 57495
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.137011766433716,
+      "learning_rate": 1.0665725310121389e-07,
+      "loss": 0.4198,
+      "step": 57496
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.40817928314209,
+      "learning_rate": 1.0660309250741851e-07,
+      "loss": 0.5357,
+      "step": 57497
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.7413032054901123,
+      "learning_rate": 1.0654894552032613e-07,
+      "loss": 0.5442,
+      "step": 57498
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.7164106369018555,
+      "learning_rate": 1.0649481214008883e-07,
+      "loss": 0.6702,
+      "step": 57499
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.690575361251831,
+      "learning_rate": 1.064406923668554e-07,
+      "loss": 0.6222,
+      "step": 57500
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.750992774963379,
+      "learning_rate": 1.063865862007779e-07,
+      "loss": 0.56,
+      "step": 57501
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.884481906890869,
+      "learning_rate": 1.063324936420057e-07,
+      "loss": 0.5268,
+      "step": 57502
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.6732890605926514,
+      "learning_rate": 1.0627841469068978e-07,
+      "loss": 0.679,
+      "step": 57503
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.982952117919922,
+      "learning_rate": 1.0622434934698e-07,
+      "loss": 0.6727,
+      "step": 57504
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 5.2461628913879395,
+      "learning_rate": 1.0617029761102792e-07,
+      "loss": 0.7037,
+      "step": 57505
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.0963501930236816,
+      "learning_rate": 1.0611625948298177e-07,
+      "loss": 0.6878,
+      "step": 57506
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.999720811843872,
+      "learning_rate": 1.0606223496299417e-07,
+      "loss": 0.4939,
+      "step": 57507
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.0902390480041504,
+      "learning_rate": 1.0600822405121336e-07,
+      "loss": 0.4569,
+      "step": 57508
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 4.7985429763793945,
+      "learning_rate": 1.0595422674779088e-07,
+      "loss": 0.6216,
+      "step": 57509
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.231356143951416,
+      "learning_rate": 1.0590024305287605e-07,
+      "loss": 0.3218,
+      "step": 57510
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.2236640453338623,
+      "learning_rate": 1.0584627296661931e-07,
+      "loss": 0.468,
+      "step": 57511
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.0132553577423096,
+      "learning_rate": 1.0579231648917055e-07,
+      "loss": 0.4855,
+      "step": 57512
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.6559386253356934,
+      "learning_rate": 1.0573837362068074e-07,
+      "loss": 0.4524,
+      "step": 57513
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.546066999435425,
+      "learning_rate": 1.056844443612981e-07,
+      "loss": 0.5653,
+      "step": 57514
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.0354409217834473,
+      "learning_rate": 1.0563052871117418e-07,
+      "loss": 0.4669,
+      "step": 57515
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.4500935077667236,
+      "learning_rate": 1.0557662667045776e-07,
+      "loss": 0.6252,
+      "step": 57516
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.176124095916748,
+      "learning_rate": 1.0552273823929982e-07,
+      "loss": 0.5134,
+      "step": 57517
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 4.091145992279053,
+      "learning_rate": 1.0546886341784912e-07,
+      "loss": 0.511,
+      "step": 57518
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 4.190977573394775,
+      "learning_rate": 1.0541500220625667e-07,
+      "loss": 0.7395,
+      "step": 57519
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.4760522842407227,
+      "learning_rate": 1.0536115460467067e-07,
+      "loss": 0.719,
+      "step": 57520
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.6598944664001465,
+      "learning_rate": 1.0530732061324267e-07,
+      "loss": 0.5573,
+      "step": 57521
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.760439872741699,
+      "learning_rate": 1.0525350023212089e-07,
+      "loss": 0.6855,
+      "step": 57522
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.8322951793670654,
+      "learning_rate": 1.0519969346145575e-07,
+      "loss": 0.5771,
+      "step": 57523
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.770620107650757,
+      "learning_rate": 1.0514590030139604e-07,
+      "loss": 0.5455,
+      "step": 57524
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.3716585636138916,
+      "learning_rate": 1.0509212075209218e-07,
+      "loss": 0.3987,
+      "step": 57525
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.7836313247680664,
+      "learning_rate": 1.0503835481369351e-07,
+      "loss": 0.6594,
+      "step": 57526
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.87600040435791,
+      "learning_rate": 1.0498460248634935e-07,
+      "loss": 0.5986,
+      "step": 57527
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.6108686923980713,
+      "learning_rate": 1.0493086377020955e-07,
+      "loss": 0.7744,
+      "step": 57528
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.8800137042999268,
+      "learning_rate": 1.0487713866542349e-07,
+      "loss": 0.3762,
+      "step": 57529
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.592273712158203,
+      "learning_rate": 1.048234271721399e-07,
+      "loss": 0.431,
+      "step": 57530
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.8882811069488525,
+      "learning_rate": 1.0476972929050866e-07,
+      "loss": 0.4973,
+      "step": 57531
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.7560410499572754,
+      "learning_rate": 1.0471604502067912e-07,
+      "loss": 0.5633,
+      "step": 57532
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.4324870109558105,
+      "learning_rate": 1.0466237436280058e-07,
+      "loss": 0.7564,
+      "step": 57533
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 4.061153411865234,
+      "learning_rate": 1.0460871731702183e-07,
+      "loss": 0.7567,
+      "step": 57534
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.932468891143799,
+      "learning_rate": 1.0455507388349273e-07,
+      "loss": 0.6477,
+      "step": 57535
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.4398748874664307,
+      "learning_rate": 1.0450144406236207e-07,
+      "loss": 0.6917,
+      "step": 57536
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.5010440349578857,
+      "learning_rate": 1.044478278537786e-07,
+      "loss": 0.5546,
+      "step": 57537
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.501434564590454,
+      "learning_rate": 1.0439422525789223e-07,
+      "loss": 0.6362,
+      "step": 57538
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.5394041538238525,
+      "learning_rate": 1.0434063627485114e-07,
+      "loss": 0.5408,
+      "step": 57539
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.248946189880371,
+      "learning_rate": 1.0428706090480523e-07,
+      "loss": 0.6458,
+      "step": 57540
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.4071521759033203,
+      "learning_rate": 1.042334991479027e-07,
+      "loss": 0.5514,
+      "step": 57541
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.9173836708068848,
+      "learning_rate": 1.0417995100429346e-07,
+      "loss": 0.52,
+      "step": 57542
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.4606666564941406,
+      "learning_rate": 1.0412641647412458e-07,
+      "loss": 0.3247,
+      "step": 57543
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.7574880123138428,
+      "learning_rate": 1.0407289555754707e-07,
+      "loss": 0.5147,
+      "step": 57544
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.5628323554992676,
+      "learning_rate": 1.0401938825470858e-07,
+      "loss": 0.4828,
+      "step": 57545
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.3215901851654053,
+      "learning_rate": 1.039658945657579e-07,
+      "loss": 0.59,
+      "step": 57546
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 5.569374084472656,
+      "learning_rate": 1.0391241449084432e-07,
+      "loss": 0.5453,
+      "step": 57547
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 4.674160480499268,
+      "learning_rate": 1.0385894803011554e-07,
+      "loss": 0.6245,
+      "step": 57548
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.0463011264801025,
+      "learning_rate": 1.0380549518372086e-07,
+      "loss": 0.4749,
+      "step": 57549
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.371258497238159,
+      "learning_rate": 1.0375205595180959e-07,
+      "loss": 0.6778,
+      "step": 57550
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.486661434173584,
+      "learning_rate": 1.0369863033452943e-07,
+      "loss": 0.4853,
+      "step": 57551
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.9197943210601807,
+      "learning_rate": 1.0364521833202967e-07,
+      "loss": 0.4867,
+      "step": 57552
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.101727247238159,
+      "learning_rate": 1.0359181994445744e-07,
+      "loss": 0.5261,
+      "step": 57553
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.0550711154937744,
+      "learning_rate": 1.0353843517196316e-07,
+      "loss": 0.6839,
+      "step": 57554
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.596831798553467,
+      "learning_rate": 1.0348506401469339e-07,
+      "loss": 0.6081,
+      "step": 57555
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.2231180667877197,
+      "learning_rate": 1.0343170647279799e-07,
+      "loss": 0.6682,
+      "step": 57556
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.08705997467041,
+      "learning_rate": 1.0337836254642408e-07,
+      "loss": 0.5773,
+      "step": 57557
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.439229965209961,
+      "learning_rate": 1.0332503223572099e-07,
+      "loss": 0.3908,
+      "step": 57558
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 4.144199848175049,
+      "learning_rate": 1.0327171554083637e-07,
+      "loss": 0.7559,
+      "step": 57559
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.6776084899902344,
+      "learning_rate": 1.0321841246191899e-07,
+      "loss": 0.4227,
+      "step": 57560
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 4.292359352111816,
+      "learning_rate": 1.0316512299911707e-07,
+      "loss": 0.6925,
+      "step": 57561
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.7449047565460205,
+      "learning_rate": 1.0311184715257827e-07,
+      "loss": 0.7406,
+      "step": 57562
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.5691635608673096,
+      "learning_rate": 1.030585849224508e-07,
+      "loss": 0.6587,
+      "step": 57563
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 4.124330997467041,
+      "learning_rate": 1.0300533630888343e-07,
+      "loss": 0.4915,
+      "step": 57564
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.9967377185821533,
+      "learning_rate": 1.0295210131202326e-07,
+      "loss": 0.621,
+      "step": 57565
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.0115859508514404,
+      "learning_rate": 1.0289887993201964e-07,
+      "loss": 0.7307,
+      "step": 57566
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.3940742015838623,
+      "learning_rate": 1.0284567216901852e-07,
+      "loss": 0.6133,
+      "step": 57567
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.1429038047790527,
+      "learning_rate": 1.0279247802316983e-07,
+      "loss": 0.8215,
+      "step": 57568
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.6714656352996826,
+      "learning_rate": 1.0273929749462008e-07,
+      "loss": 0.5865,
+      "step": 57569
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.105283737182617,
+      "learning_rate": 1.0268613058351862e-07,
+      "loss": 0.7596,
+      "step": 57570
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.4012043476104736,
+      "learning_rate": 1.0263297729001198e-07,
+      "loss": 0.5368,
+      "step": 57571
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.5842270851135254,
+      "learning_rate": 1.0257983761424894e-07,
+      "loss": 0.6891,
+      "step": 57572
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.9288852214813232,
+      "learning_rate": 1.0252671155637605e-07,
+      "loss": 0.5859,
+      "step": 57573
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 5.65631628036499,
+      "learning_rate": 1.0247359911654264e-07,
+      "loss": 0.6541,
+      "step": 57574
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.345698833465576,
+      "learning_rate": 1.0242050029489469e-07,
+      "loss": 0.4035,
+      "step": 57575
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.8803110122680664,
+      "learning_rate": 1.0236741509158099e-07,
+      "loss": 0.5102,
+      "step": 57576
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.26767635345459,
+      "learning_rate": 1.0231434350674863e-07,
+      "loss": 0.4332,
+      "step": 57577
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.856424331665039,
+      "learning_rate": 1.0226128554054582e-07,
+      "loss": 0.6172,
+      "step": 57578
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.8437254428863525,
+      "learning_rate": 1.0220824119311912e-07,
+      "loss": 0.6658,
+      "step": 57579
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.928821325302124,
+      "learning_rate": 1.0215521046461729e-07,
+      "loss": 0.6088,
+      "step": 57580
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 5.99737024307251,
+      "learning_rate": 1.0210219335518633e-07,
+      "loss": 0.6171,
+      "step": 57581
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.7998881340026855,
+      "learning_rate": 1.0204918986497447e-07,
+      "loss": 0.7078,
+      "step": 57582
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.123239517211914,
+      "learning_rate": 1.019961999941288e-07,
+      "loss": 0.7624,
+      "step": 57583
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.9865574836730957,
+      "learning_rate": 1.0194322374279753e-07,
+      "loss": 0.6543,
+      "step": 57584
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.201380491256714,
+      "learning_rate": 1.0189026111112665e-07,
+      "loss": 0.4674,
+      "step": 57585
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.6515305042266846,
+      "learning_rate": 1.018373120992644e-07,
+      "loss": 0.6552,
+      "step": 57586
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.4369730949401855,
+      "learning_rate": 1.0178437670735786e-07,
+      "loss": 0.636,
+      "step": 57587
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.344444513320923,
+      "learning_rate": 1.0173145493555359e-07,
+      "loss": 0.641,
+      "step": 57588
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.181180953979492,
+      "learning_rate": 1.0167854678399924e-07,
+      "loss": 0.4124,
+      "step": 57589
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 4.236358165740967,
+      "learning_rate": 1.0162565225284194e-07,
+      "loss": 0.7424,
+      "step": 57590
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.6984829902648926,
+      "learning_rate": 1.0157277134222876e-07,
+      "loss": 0.7219,
+      "step": 57591
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.6116998195648193,
+      "learning_rate": 1.0151990405230739e-07,
+      "loss": 0.2872,
+      "step": 57592
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 4.4407477378845215,
+      "learning_rate": 1.0146705038322324e-07,
+      "loss": 0.5606,
+      "step": 57593
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.8955931663513184,
+      "learning_rate": 1.0141421033512511e-07,
+      "loss": 0.6362,
+      "step": 57594
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 6.273453235626221,
+      "learning_rate": 1.0136138390815787e-07,
+      "loss": 0.648,
+      "step": 57595
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.5302953720092773,
+      "learning_rate": 1.0130857110247028e-07,
+      "loss": 0.4263,
+      "step": 57596
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.915240526199341,
+      "learning_rate": 1.0125577191820835e-07,
+      "loss": 0.5401,
+      "step": 57597
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.6584551334381104,
+      "learning_rate": 1.0120298635551861e-07,
+      "loss": 0.593,
+      "step": 57598
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.3564841747283936,
+      "learning_rate": 1.0115021441454875e-07,
+      "loss": 0.4443,
+      "step": 57599
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.4101760387420654,
+      "learning_rate": 1.0109745609544474e-07,
+      "loss": 0.6306,
+      "step": 57600
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.840883255004883,
+      "learning_rate": 1.0104471139835426e-07,
+      "loss": 0.5317,
+      "step": 57601
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.233673095703125,
+      "learning_rate": 1.0099198032342217e-07,
+      "loss": 0.6053,
+      "step": 57602
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 4.256737232208252,
+      "learning_rate": 1.009392628707967e-07,
+      "loss": 0.6102,
+      "step": 57603
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.231250524520874,
+      "learning_rate": 1.0088655904062383e-07,
+      "loss": 0.3912,
+      "step": 57604
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.1351184844970703,
+      "learning_rate": 1.0083386883305013e-07,
+      "loss": 0.5145,
+      "step": 57605
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.5749053955078125,
+      "learning_rate": 1.0078119224822213e-07,
+      "loss": 0.5656,
+      "step": 57606
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.120586633682251,
+      "learning_rate": 1.0072852928628696e-07,
+      "loss": 0.5746,
+      "step": 57607
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.802689790725708,
+      "learning_rate": 1.0067587994738948e-07,
+      "loss": 0.6182,
+      "step": 57608
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.474647045135498,
+      "learning_rate": 1.0062324423167735e-07,
+      "loss": 0.7035,
+      "step": 57609
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.415153741836548,
+      "learning_rate": 1.0057062213929658e-07,
+      "loss": 0.4018,
+      "step": 57610
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.812284231185913,
+      "learning_rate": 1.0051801367039427e-07,
+      "loss": 0.4781,
+      "step": 57611
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.2666409015655518,
+      "learning_rate": 1.004654188251153e-07,
+      "loss": 0.7586,
+      "step": 57612
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.760850429534912,
+      "learning_rate": 1.0041283760360676e-07,
+      "loss": 0.474,
+      "step": 57613
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.3763303756713867,
+      "learning_rate": 1.0036027000601467e-07,
+      "loss": 0.5757,
+      "step": 57614
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.3667478561401367,
+      "learning_rate": 1.0030771603248501e-07,
+      "loss": 0.6036,
+      "step": 57615
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.3235104084014893,
+      "learning_rate": 1.0025517568316434e-07,
+      "loss": 0.6233,
+      "step": 57616
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.043362617492676,
+      "learning_rate": 1.0020264895819864e-07,
+      "loss": 0.4948,
+      "step": 57617
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.151188850402832,
+      "learning_rate": 1.0015013585773392e-07,
+      "loss": 0.5739,
+      "step": 57618
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.405001640319824,
+      "learning_rate": 1.0009763638191616e-07,
+      "loss": 0.408,
+      "step": 57619
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.4630770683288574,
+      "learning_rate": 1.0004515053089081e-07,
+      "loss": 0.6051,
+      "step": 57620
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.030552387237549,
+      "learning_rate": 9.999267830480496e-08,
+      "loss": 0.5134,
+      "step": 57621
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.9052178859710693,
+      "learning_rate": 9.994021970380352e-08,
+      "loss": 0.7202,
+      "step": 57622
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.5914523601531982,
+      "learning_rate": 9.988777472803301e-08,
+      "loss": 0.5353,
+      "step": 57623
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.988481044769287,
+      "learning_rate": 9.983534337763834e-08,
+      "loss": 0.7865,
+      "step": 57624
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.721494436264038,
+      "learning_rate": 9.97829256527666e-08,
+      "loss": 0.6054,
+      "step": 57625
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.700636625289917,
+      "learning_rate": 9.973052155356266e-08,
+      "loss": 0.9383,
+      "step": 57626
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.362197160720825,
+      "learning_rate": 9.967813108017254e-08,
+      "loss": 0.6338,
+      "step": 57627
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.487659454345703,
+      "learning_rate": 9.962575423274112e-08,
+      "loss": 0.6876,
+      "step": 57628
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.0554473400115967,
+      "learning_rate": 9.957339101141549e-08,
+      "loss": 0.6677,
+      "step": 57629
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.1543731689453125,
+      "learning_rate": 9.952104141634055e-08,
+      "loss": 0.446,
+      "step": 57630
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.658385992050171,
+      "learning_rate": 9.946870544766173e-08,
+      "loss": 0.6024,
+      "step": 57631
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.656865358352661,
+      "learning_rate": 9.941638310552448e-08,
+      "loss": 0.625,
+      "step": 57632
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.520434856414795,
+      "learning_rate": 9.936407439007479e-08,
+      "loss": 0.5546,
+      "step": 57633
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 4.178755760192871,
+      "learning_rate": 9.931177930145752e-08,
+      "loss": 0.6799,
+      "step": 57634
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.459616184234619,
+      "learning_rate": 9.92594978398187e-08,
+      "loss": 0.6699,
+      "step": 57635
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.081763744354248,
+      "learning_rate": 9.920723000530263e-08,
+      "loss": 0.4313,
+      "step": 57636
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.8771181106567383,
+      "learning_rate": 9.915497579805644e-08,
+      "loss": 0.635,
+      "step": 57637
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.592853307723999,
+      "learning_rate": 9.910273521822334e-08,
+      "loss": 0.6214,
+      "step": 57638
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 4.055034160614014,
+      "learning_rate": 9.905050826595042e-08,
+      "loss": 0.4607,
+      "step": 57639
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.5901153087615967,
+      "learning_rate": 9.899829494138146e-08,
+      "loss": 0.6431,
+      "step": 57640
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.7162303924560547,
+      "learning_rate": 9.894609524466248e-08,
+      "loss": 0.5053,
+      "step": 57641
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.0652458667755127,
+      "learning_rate": 9.889390917593832e-08,
+      "loss": 0.4144,
+      "step": 57642
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.469555139541626,
+      "learning_rate": 9.884173673535502e-08,
+      "loss": 0.704,
+      "step": 57643
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.2293930053710938,
+      "learning_rate": 9.878957792305577e-08,
+      "loss": 0.7478,
+      "step": 57644
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 18.486082077026367,
+      "learning_rate": 9.873743273918712e-08,
+      "loss": 0.551,
+      "step": 57645
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 4.432354927062988,
+      "learning_rate": 9.868530118389341e-08,
+      "loss": 0.69,
+      "step": 57646
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.8253445625305176,
+      "learning_rate": 9.863318325732062e-08,
+      "loss": 0.444,
+      "step": 57647
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 4.683812618255615,
+      "learning_rate": 9.8581078959612e-08,
+      "loss": 0.7829,
+      "step": 57648
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.407137393951416,
+      "learning_rate": 9.852898829091295e-08,
+      "loss": 0.529,
+      "step": 57649
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.3974826335906982,
+      "learning_rate": 9.847691125136894e-08,
+      "loss": 0.4539,
+      "step": 57650
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.708961248397827,
+      "learning_rate": 9.842484784112538e-08,
+      "loss": 0.7823,
+      "step": 57651
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.052436113357544,
+      "learning_rate": 9.837279806032496e-08,
+      "loss": 0.5351,
+      "step": 57652
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.8648664951324463,
+      "learning_rate": 9.832076190911478e-08,
+      "loss": 0.5323,
+      "step": 57653
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.520179510116577,
+      "learning_rate": 9.826873938763748e-08,
+      "loss": 0.4452,
+      "step": 57654
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.2441234588623047,
+      "learning_rate": 9.821673049603908e-08,
+      "loss": 0.6207,
+      "step": 57655
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.874929904937744,
+      "learning_rate": 9.81647352344628e-08,
+      "loss": 0.6566,
+      "step": 57656
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.5788791179656982,
+      "learning_rate": 9.811275360305517e-08,
+      "loss": 0.5447,
+      "step": 57657
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.286825656890869,
+      "learning_rate": 9.806078560195886e-08,
+      "loss": 0.6,
+      "step": 57658
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.0643560886383057,
+      "learning_rate": 9.800883123131988e-08,
+      "loss": 0.6808,
+      "step": 57659
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.4345712661743164,
+      "learning_rate": 9.795689049128199e-08,
+      "loss": 0.5013,
+      "step": 57660
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 5.3591508865356445,
+      "learning_rate": 9.790496338198896e-08,
+      "loss": 0.7226,
+      "step": 57661
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.0147337913513184,
+      "learning_rate": 9.78530499035868e-08,
+      "loss": 0.5415,
+      "step": 57662
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 4.345408916473389,
+      "learning_rate": 9.780115005621815e-08,
+      "loss": 0.7923,
+      "step": 57663
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 4.215707302093506,
+      "learning_rate": 9.774926384002847e-08,
+      "loss": 0.513,
+      "step": 57664
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.524264097213745,
+      "learning_rate": 9.769739125516154e-08,
+      "loss": 0.4447,
+      "step": 57665
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 4.677182674407959,
+      "learning_rate": 9.764553230176166e-08,
+      "loss": 0.597,
+      "step": 57666
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 5.243930339813232,
+      "learning_rate": 9.759368697997317e-08,
+      "loss": 0.6514,
+      "step": 57667
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 4.051185131072998,
+      "learning_rate": 9.754185528994043e-08,
+      "loss": 0.6773,
+      "step": 57668
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.947901487350464,
+      "learning_rate": 9.749003723180662e-08,
+      "loss": 0.6924,
+      "step": 57669
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.372843027114868,
+      "learning_rate": 9.743823280571662e-08,
+      "loss": 0.5326,
+      "step": 57670
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.423412322998047,
+      "learning_rate": 9.73864420118148e-08,
+      "loss": 0.4413,
+      "step": 57671
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.7594947814941406,
+      "learning_rate": 9.733466485024434e-08,
+      "loss": 0.5468,
+      "step": 57672
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.0823826789855957,
+      "learning_rate": 9.728290132114959e-08,
+      "loss": 0.7538,
+      "step": 57673
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.567178249359131,
+      "learning_rate": 9.723115142467487e-08,
+      "loss": 0.7209,
+      "step": 57674
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.29138445854187,
+      "learning_rate": 9.717941516096341e-08,
+      "loss": 0.5612,
+      "step": 57675
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.8562166690826416,
+      "learning_rate": 9.712769253016008e-08,
+      "loss": 0.5583,
+      "step": 57676
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 4.395280361175537,
+      "learning_rate": 9.707598353240699e-08,
+      "loss": 0.6518,
+      "step": 57677
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.439319133758545,
+      "learning_rate": 9.702428816784959e-08,
+      "loss": 0.5933,
+      "step": 57678
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.865151882171631,
+      "learning_rate": 9.697260643663053e-08,
+      "loss": 0.6651,
+      "step": 57679
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.603975772857666,
+      "learning_rate": 9.692093833889416e-08,
+      "loss": 0.5869,
+      "step": 57680
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 4.159378528594971,
+      "learning_rate": 9.686928387478367e-08,
+      "loss": 0.5876,
+      "step": 57681
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.648357391357422,
+      "learning_rate": 9.681764304444341e-08,
+      "loss": 0.6192,
+      "step": 57682
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.0835466384887695,
+      "learning_rate": 9.676601584801603e-08,
+      "loss": 0.2435,
+      "step": 57683
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.959996461868286,
+      "learning_rate": 9.671440228564588e-08,
+      "loss": 0.7996,
+      "step": 57684
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.9443986415863037,
+      "learning_rate": 9.666280235747561e-08,
+      "loss": 0.647,
+      "step": 57685
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.7643649578094482,
+      "learning_rate": 9.66112160636501e-08,
+      "loss": 0.5456,
+      "step": 57686
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.7262661457061768,
+      "learning_rate": 9.655964340431146e-08,
+      "loss": 0.4708,
+      "step": 57687
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.373638391494751,
+      "learning_rate": 9.650808437960346e-08,
+      "loss": 0.5457,
+      "step": 57688
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.8081936836242676,
+      "learning_rate": 9.645653898966933e-08,
+      "loss": 0.6165,
+      "step": 57689
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.8658297061920166,
+      "learning_rate": 9.64050072346534e-08,
+      "loss": 0.5033,
+      "step": 57690
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.6794376373291016,
+      "learning_rate": 9.635348911469778e-08,
+      "loss": 0.594,
+      "step": 57691
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.0727317333221436,
+      "learning_rate": 9.630198462994622e-08,
+      "loss": 0.401,
+      "step": 57692
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.5038645267486572,
+      "learning_rate": 9.625049378054197e-08,
+      "loss": 0.4846,
+      "step": 57693
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.041379451751709,
+      "learning_rate": 9.619901656662823e-08,
+      "loss": 0.6036,
+      "step": 57694
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.5668554306030273,
+      "learning_rate": 9.614755298834766e-08,
+      "loss": 0.3822,
+      "step": 57695
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.4237101078033447,
+      "learning_rate": 9.609610304584405e-08,
+      "loss": 0.6635,
+      "step": 57696
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.6337549686431885,
+      "learning_rate": 9.604466673926005e-08,
+      "loss": 0.4311,
+      "step": 57697
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.0493385791778564,
+      "learning_rate": 9.599324406873888e-08,
+      "loss": 0.3912,
+      "step": 57698
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.6425352096557617,
+      "learning_rate": 9.594183503442323e-08,
+      "loss": 0.5967,
+      "step": 57699
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.143573760986328,
+      "learning_rate": 9.589043963645628e-08,
+      "loss": 0.4575,
+      "step": 57700
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.1364784240722656,
+      "learning_rate": 9.583905787498072e-08,
+      "loss": 0.5461,
+      "step": 57701
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.775725841522217,
+      "learning_rate": 9.578768975013974e-08,
+      "loss": 0.5732,
+      "step": 57702
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 5.018921375274658,
+      "learning_rate": 9.573633526207605e-08,
+      "loss": 0.5018,
+      "step": 57703
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 4.642227649688721,
+      "learning_rate": 9.568499441093282e-08,
+      "loss": 0.5824,
+      "step": 57704
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.3453445434570312,
+      "learning_rate": 9.563366719685163e-08,
+      "loss": 0.7237,
+      "step": 57705
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.785658121109009,
+      "learning_rate": 9.55823536199768e-08,
+      "loss": 0.556,
+      "step": 57706
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.311492443084717,
+      "learning_rate": 9.553105368044935e-08,
+      "loss": 0.5476,
+      "step": 57707
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.2050793170928955,
+      "learning_rate": 9.547976737841358e-08,
+      "loss": 0.4659,
+      "step": 57708
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.7065088748931885,
+      "learning_rate": 9.542849471401106e-08,
+      "loss": 0.6185,
+      "step": 57709
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.139180898666382,
+      "learning_rate": 9.537723568738444e-08,
+      "loss": 0.6534,
+      "step": 57710
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.9124016761779785,
+      "learning_rate": 9.532599029867584e-08,
+      "loss": 0.6706,
+      "step": 57711
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.4791650772094727,
+      "learning_rate": 9.527475854802958e-08,
+      "loss": 0.6899,
+      "step": 57712
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.0006885528564453,
+      "learning_rate": 9.522354043558613e-08,
+      "loss": 0.5306,
+      "step": 57713
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.0378711223602295,
+      "learning_rate": 9.517233596148867e-08,
+      "loss": 0.5783,
+      "step": 57714
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.761600971221924,
+      "learning_rate": 9.512114512587933e-08,
+      "loss": 0.4894,
+      "step": 57715
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.4256417751312256,
+      "learning_rate": 9.506996792890133e-08,
+      "loss": 0.5669,
+      "step": 57716
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 10.302275657653809,
+      "learning_rate": 9.501880437069566e-08,
+      "loss": 0.5456,
+      "step": 57717
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.4007656574249268,
+      "learning_rate": 9.49676544514061e-08,
+      "loss": 0.8142,
+      "step": 57718
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.7900233268737793,
+      "learning_rate": 9.491651817117309e-08,
+      "loss": 0.616,
+      "step": 57719
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.1272380352020264,
+      "learning_rate": 9.48653955301404e-08,
+      "loss": 0.5354,
+      "step": 57720
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.660937786102295,
+      "learning_rate": 9.481428652844904e-08,
+      "loss": 0.7212,
+      "step": 57721
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.387542486190796,
+      "learning_rate": 9.476319116624167e-08,
+      "loss": 0.5439,
+      "step": 57722
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.8677518367767334,
+      "learning_rate": 9.471210944366094e-08,
+      "loss": 0.7648,
+      "step": 57723
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 4.614992618560791,
+      "learning_rate": 9.466104136084786e-08,
+      "loss": 0.6372,
+      "step": 57724
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.4523487091064453,
+      "learning_rate": 9.460998691794509e-08,
+      "loss": 0.2532,
+      "step": 57725
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.2921369075775146,
+      "learning_rate": 9.455894611509364e-08,
+      "loss": 0.4638,
+      "step": 57726
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.862104654312134,
+      "learning_rate": 9.450791895243727e-08,
+      "loss": 0.5465,
+      "step": 57727
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.698540210723877,
+      "learning_rate": 9.445690543011588e-08,
+      "loss": 0.6854,
+      "step": 57728
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.6431779861450195,
+      "learning_rate": 9.440590554827211e-08,
+      "loss": 0.5474,
+      "step": 57729
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.782508134841919,
+      "learning_rate": 9.43549193070481e-08,
+      "loss": 0.7282,
+      "step": 57730
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.651484251022339,
+      "learning_rate": 9.430394670658538e-08,
+      "loss": 0.4765,
+      "step": 57731
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.7464349269866943,
+      "learning_rate": 9.425298774702551e-08,
+      "loss": 0.5151,
+      "step": 57732
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.6938507556915283,
+      "learning_rate": 9.420204242851061e-08,
+      "loss": 0.6657,
+      "step": 57733
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 1.9227991104125977,
+      "learning_rate": 9.415111075118166e-08,
+      "loss": 0.2655,
+      "step": 57734
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.282336473464966,
+      "learning_rate": 9.410019271518133e-08,
+      "loss": 0.5156,
+      "step": 57735
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.4042956829071045,
+      "learning_rate": 9.404928832065008e-08,
+      "loss": 0.35,
+      "step": 57736
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.32722806930542,
+      "learning_rate": 9.399839756772999e-08,
+      "loss": 0.6387,
+      "step": 57737
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.412336826324463,
+      "learning_rate": 9.394752045656264e-08,
+      "loss": 0.5089,
+      "step": 57738
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.9609005451202393,
+      "learning_rate": 9.389665698728955e-08,
+      "loss": 0.5252,
+      "step": 57739
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 4.235916614532471,
+      "learning_rate": 9.384580716005177e-08,
+      "loss": 0.5871,
+      "step": 57740
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 4.143280506134033,
+      "learning_rate": 9.379497097499135e-08,
+      "loss": 0.6847,
+      "step": 57741
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.2721614837646484,
+      "learning_rate": 9.374414843224878e-08,
+      "loss": 0.7815,
+      "step": 57742
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.1450605392456055,
+      "learning_rate": 9.369333953196557e-08,
+      "loss": 0.767,
+      "step": 57743
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.3694100379943848,
+      "learning_rate": 9.364254427428332e-08,
+      "loss": 0.6987,
+      "step": 57744
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.212524652481079,
+      "learning_rate": 9.359176265934411e-08,
+      "loss": 0.5113,
+      "step": 57745
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.803117275238037,
+      "learning_rate": 9.354099468728728e-08,
+      "loss": 0.5675,
+      "step": 57746
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.748509168624878,
+      "learning_rate": 9.349024035825549e-08,
+      "loss": 0.8193,
+      "step": 57747
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.936406135559082,
+      "learning_rate": 9.343949967238863e-08,
+      "loss": 0.4708,
+      "step": 57748
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.676083564758301,
+      "learning_rate": 9.338877262982881e-08,
+      "loss": 0.5822,
+      "step": 57749
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.7482693195343018,
+      "learning_rate": 9.333805923071648e-08,
+      "loss": 0.5759,
+      "step": 57750
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.382251739501953,
+      "learning_rate": 9.328735947519318e-08,
+      "loss": 0.5186,
+      "step": 57751
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.5501368045806885,
+      "learning_rate": 9.323667336339992e-08,
+      "loss": 0.3969,
+      "step": 57752
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.957669973373413,
+      "learning_rate": 9.318600089547714e-08,
+      "loss": 0.6144,
+      "step": 57753
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.419879674911499,
+      "learning_rate": 9.313534207156583e-08,
+      "loss": 0.5615,
+      "step": 57754
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.2715377807617188,
+      "learning_rate": 9.308469689180754e-08,
+      "loss": 0.7322,
+      "step": 57755
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.5173561573028564,
+      "learning_rate": 9.303406535634163e-08,
+      "loss": 0.4276,
+      "step": 57756
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.2881741523742676,
+      "learning_rate": 9.298344746531073e-08,
+      "loss": 0.7767,
+      "step": 57757
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.677372694015503,
+      "learning_rate": 9.293284321885365e-08,
+      "loss": 0.6006,
+      "step": 57758
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.389235019683838,
+      "learning_rate": 9.288225261711304e-08,
+      "loss": 0.4947,
+      "step": 57759
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.5555105209350586,
+      "learning_rate": 9.283167566022821e-08,
+      "loss": 0.5247,
+      "step": 57760
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 4.049221992492676,
+      "learning_rate": 9.278111234834019e-08,
+      "loss": 0.7706,
+      "step": 57761
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.6350417137145996,
+      "learning_rate": 9.273056268158942e-08,
+      "loss": 0.5627,
+      "step": 57762
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.3354339599609375,
+      "learning_rate": 9.268002666011744e-08,
+      "loss": 0.563,
+      "step": 57763
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.8215484619140625,
+      "learning_rate": 9.262950428406304e-08,
+      "loss": 0.4946,
+      "step": 57764
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.6351318359375,
+      "learning_rate": 9.25789955535683e-08,
+      "loss": 0.5922,
+      "step": 57765
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.980363130569458,
+      "learning_rate": 9.25285004687726e-08,
+      "loss": 0.472,
+      "step": 57766
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.6164090633392334,
+      "learning_rate": 9.247801902981746e-08,
+      "loss": 0.4798,
+      "step": 57767
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.9868316650390625,
+      "learning_rate": 9.242755123684166e-08,
+      "loss": 0.505,
+      "step": 57768
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.674849510192871,
+      "learning_rate": 9.237709708998677e-08,
+      "loss": 0.6648,
+      "step": 57769
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.8390283584594727,
+      "learning_rate": 9.232665658939266e-08,
+      "loss": 0.6937,
+      "step": 57770
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 4.488449573516846,
+      "learning_rate": 9.227622973519978e-08,
+      "loss": 0.4486,
+      "step": 57771
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.646374464035034,
+      "learning_rate": 9.222581652754803e-08,
+      "loss": 0.555,
+      "step": 57772
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.1280248165130615,
+      "learning_rate": 9.217541696657783e-08,
+      "loss": 0.5503,
+      "step": 57773
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.5063204765319824,
+      "learning_rate": 9.212503105242965e-08,
+      "loss": 0.5933,
+      "step": 57774
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.236664295196533,
+      "learning_rate": 9.207465878524279e-08,
+      "loss": 0.5479,
+      "step": 57775
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.203692674636841,
+      "learning_rate": 9.202430016515773e-08,
+      "loss": 0.5142,
+      "step": 57776
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.2995128631591797,
+      "learning_rate": 9.19739551923149e-08,
+      "loss": 0.3974,
+      "step": 57777
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.7373929023742676,
+      "learning_rate": 9.19236238668536e-08,
+      "loss": 0.5955,
+      "step": 57778
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.856112480163574,
+      "learning_rate": 9.187330618891487e-08,
+      "loss": 0.8136,
+      "step": 57779
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.673870801925659,
+      "learning_rate": 9.182300215863693e-08,
+      "loss": 0.5227,
+      "step": 57780
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 4.490900039672852,
+      "learning_rate": 9.177271177616076e-08,
+      "loss": 0.7269,
+      "step": 57781
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.7999165058135986,
+      "learning_rate": 9.172243504162625e-08,
+      "loss": 0.7251,
+      "step": 57782
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.095020294189453,
+      "learning_rate": 9.167217195517276e-08,
+      "loss": 0.1911,
+      "step": 57783
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.0113461017608643,
+      "learning_rate": 9.162192251694014e-08,
+      "loss": 0.5407,
+      "step": 57784
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 4.477488994598389,
+      "learning_rate": 9.157168672706829e-08,
+      "loss": 0.7188,
+      "step": 57785
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.308954954147339,
+      "learning_rate": 9.152146458569711e-08,
+      "loss": 0.4266,
+      "step": 57786
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.7308411598205566,
+      "learning_rate": 9.147125609296592e-08,
+      "loss": 0.5617,
+      "step": 57787
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.8562841415405273,
+      "learning_rate": 9.142106124901462e-08,
+      "loss": 0.6448,
+      "step": 57788
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.064723491668701,
+      "learning_rate": 9.137088005398198e-08,
+      "loss": 0.4558,
+      "step": 57789
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.4438295364379883,
+      "learning_rate": 9.132071250800845e-08,
+      "loss": 0.53,
+      "step": 57790
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.295562267303467,
+      "learning_rate": 9.12705586112328e-08,
+      "loss": 0.5089,
+      "step": 57791
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.228938102722168,
+      "learning_rate": 9.122041836379548e-08,
+      "loss": 0.5106,
+      "step": 57792
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.151883840560913,
+      "learning_rate": 9.117029176583526e-08,
+      "loss": 0.4644,
+      "step": 57793
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 4.412519931793213,
+      "learning_rate": 9.112017881749091e-08,
+      "loss": 0.7029,
+      "step": 57794
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.214350938796997,
+      "learning_rate": 9.10700795189029e-08,
+      "loss": 0.4887,
+      "step": 57795
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.6567046642303467,
+      "learning_rate": 9.101999387020998e-08,
+      "loss": 0.7448,
+      "step": 57796
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.3522257804870605,
+      "learning_rate": 9.096992187155151e-08,
+      "loss": 0.5251,
+      "step": 57797
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.2673959732055664,
+      "learning_rate": 9.091986352306682e-08,
+      "loss": 0.5105,
+      "step": 57798
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.2435827255249023,
+      "learning_rate": 9.086981882489465e-08,
+      "loss": 0.5174,
+      "step": 57799
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 4.14560604095459,
+      "learning_rate": 9.081978777717493e-08,
+      "loss": 0.6517,
+      "step": 57800
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 12.582282066345215,
+      "learning_rate": 9.076977038004642e-08,
+      "loss": 0.5725,
+      "step": 57801
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.2497711181640625,
+      "learning_rate": 9.07197666336479e-08,
+      "loss": 0.5035,
+      "step": 57802
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.163292169570923,
+      "learning_rate": 9.06697765381187e-08,
+      "loss": 0.5838,
+      "step": 57803
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.460197925567627,
+      "learning_rate": 9.061980009359816e-08,
+      "loss": 0.6689,
+      "step": 57804
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 4.0674309730529785,
+      "learning_rate": 9.056983730022395e-08,
+      "loss": 0.6701,
+      "step": 57805
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.802283763885498,
+      "learning_rate": 9.051988815813706e-08,
+      "loss": 0.5124,
+      "step": 57806
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.1076481342315674,
+      "learning_rate": 9.046995266747405e-08,
+      "loss": 0.6771,
+      "step": 57807
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.6437430381774902,
+      "learning_rate": 9.042003082837591e-08,
+      "loss": 0.5161,
+      "step": 57808
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.509920358657837,
+      "learning_rate": 9.037012264097977e-08,
+      "loss": 0.436,
+      "step": 57809
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.6198744773864746,
+      "learning_rate": 9.032022810542552e-08,
+      "loss": 0.623,
+      "step": 57810
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.4937937259674072,
+      "learning_rate": 9.02703472218508e-08,
+      "loss": 0.7159,
+      "step": 57811
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.0950183868408203,
+      "learning_rate": 9.022047999039607e-08,
+      "loss": 0.4846,
+      "step": 57812
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 4.770118713378906,
+      "learning_rate": 9.017062641119844e-08,
+      "loss": 0.8946,
+      "step": 57813
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.1019585132598877,
+      "learning_rate": 9.012078648439727e-08,
+      "loss": 0.587,
+      "step": 57814
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.454728364944458,
+      "learning_rate": 9.007096021013073e-08,
+      "loss": 0.4054,
+      "step": 57815
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.0046637058258057,
+      "learning_rate": 9.002114758853764e-08,
+      "loss": 0.6715,
+      "step": 57816
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.271089553833008,
+      "learning_rate": 8.99713486197562e-08,
+      "loss": 0.7049,
+      "step": 57817
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 5.707869529724121,
+      "learning_rate": 8.992156330392576e-08,
+      "loss": 0.5533,
+      "step": 57818
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.554647207260132,
+      "learning_rate": 8.987179164118287e-08,
+      "loss": 0.5262,
+      "step": 57819
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.8506252765655518,
+      "learning_rate": 8.982203363166853e-08,
+      "loss": 0.6724,
+      "step": 57820
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.417983293533325,
+      "learning_rate": 8.977228927551873e-08,
+      "loss": 0.7935,
+      "step": 57821
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.603930711746216,
+      "learning_rate": 8.972255857287337e-08,
+      "loss": 0.5606,
+      "step": 57822
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.6572585105895996,
+      "learning_rate": 8.967284152386957e-08,
+      "loss": 0.6959,
+      "step": 57823
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.8783230781555176,
+      "learning_rate": 8.962313812864664e-08,
+      "loss": 0.5727,
+      "step": 57824
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.543097734451294,
+      "learning_rate": 8.957344838734228e-08,
+      "loss": 0.7203,
+      "step": 57825
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.5988965034484863,
+      "learning_rate": 8.952377230009468e-08,
+      "loss": 0.5017,
+      "step": 57826
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.597195625305176,
+      "learning_rate": 8.947410986704153e-08,
+      "loss": 0.5432,
+      "step": 57827
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.6945247650146484,
+      "learning_rate": 8.942446108832159e-08,
+      "loss": 0.4459,
+      "step": 57828
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 4.565770149230957,
+      "learning_rate": 8.937482596407254e-08,
+      "loss": 0.4688,
+      "step": 57829
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.9926228523254395,
+      "learning_rate": 8.932520449443316e-08,
+      "loss": 0.7085,
+      "step": 57830
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.3680522441864014,
+      "learning_rate": 8.927559667954e-08,
+      "loss": 0.587,
+      "step": 57831
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 4.471436023712158,
+      "learning_rate": 8.922600251953184e-08,
+      "loss": 0.6148,
+      "step": 57832
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.2167062759399414,
+      "learning_rate": 8.91764220145469e-08,
+      "loss": 0.519,
+      "step": 57833
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.491950511932373,
+      "learning_rate": 8.912685516472286e-08,
+      "loss": 0.4905,
+      "step": 57834
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.8622734546661377,
+      "learning_rate": 8.907730197019682e-08,
+      "loss": 0.7294,
+      "step": 57835
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.7310283184051514,
+      "learning_rate": 8.902776243110812e-08,
+      "loss": 0.5096,
+      "step": 57836
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.6924102306365967,
+      "learning_rate": 8.897823654759219e-08,
+      "loss": 0.659,
+      "step": 57837
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.7338802814483643,
+      "learning_rate": 8.89287243197895e-08,
+      "loss": 0.4813,
+      "step": 57838
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.9987809658050537,
+      "learning_rate": 8.887922574783491e-08,
+      "loss": 0.6184,
+      "step": 57839
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.7187845706939697,
+      "learning_rate": 8.882974083186835e-08,
+      "loss": 0.2907,
+      "step": 57840
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 5.170485973358154,
+      "learning_rate": 8.878026957202635e-08,
+      "loss": 0.634,
+      "step": 57841
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 4.195148468017578,
+      "learning_rate": 8.873081196844658e-08,
+      "loss": 0.5119,
+      "step": 57842
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.8759100437164307,
+      "learning_rate": 8.86813680212667e-08,
+      "loss": 0.4524,
+      "step": 57843
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 4.041312217712402,
+      "learning_rate": 8.863193773062384e-08,
+      "loss": 0.6805,
+      "step": 57844
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.053986072540283,
+      "learning_rate": 8.858252109665566e-08,
+      "loss": 0.3197,
+      "step": 57845
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.3661551475524902,
+      "learning_rate": 8.853311811949983e-08,
+      "loss": 0.7792,
+      "step": 57846
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.6533539295196533,
+      "learning_rate": 8.8483728799294e-08,
+      "loss": 0.5085,
+      "step": 57847
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.759289503097534,
+      "learning_rate": 8.843435313617421e-08,
+      "loss": 0.491,
+      "step": 57848
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 5.231639385223389,
+      "learning_rate": 8.838499113027921e-08,
+      "loss": 0.632,
+      "step": 57849
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.934910774230957,
+      "learning_rate": 8.8335642781745e-08,
+      "loss": 0.7865,
+      "step": 57850
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.666093111038208,
+      "learning_rate": 8.828630809070981e-08,
+      "loss": 0.4614,
+      "step": 57851
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.175992250442505,
+      "learning_rate": 8.823698705731021e-08,
+      "loss": 0.7484,
+      "step": 57852
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.000383138656616,
+      "learning_rate": 8.818767968168385e-08,
+      "loss": 0.4379,
+      "step": 57853
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.9034392833709717,
+      "learning_rate": 8.813838596396784e-08,
+      "loss": 0.552,
+      "step": 57854
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.115962028503418,
+      "learning_rate": 8.80891059042982e-08,
+      "loss": 0.6752,
+      "step": 57855
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.3594772815704346,
+      "learning_rate": 8.803983950281314e-08,
+      "loss": 0.4969,
+      "step": 57856
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 5.018993377685547,
+      "learning_rate": 8.799058675964866e-08,
+      "loss": 0.6135,
+      "step": 57857
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.230865001678467,
+      "learning_rate": 8.7941347674943e-08,
+      "loss": 0.5441,
+      "step": 57858
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.9010841846466064,
+      "learning_rate": 8.789212224883215e-08,
+      "loss": 0.5339,
+      "step": 57859
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.814673662185669,
+      "learning_rate": 8.784291048145322e-08,
+      "loss": 0.6335,
+      "step": 57860
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.556990385055542,
+      "learning_rate": 8.779371237294277e-08,
+      "loss": 0.7926,
+      "step": 57861
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.836374521255493,
+      "learning_rate": 8.77445279234379e-08,
+      "loss": 0.4412,
+      "step": 57862
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.6720972061157227,
+      "learning_rate": 8.769535713307575e-08,
+      "loss": 0.5192,
+      "step": 57863
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.862060070037842,
+      "learning_rate": 8.764620000199231e-08,
+      "loss": 0.515,
+      "step": 57864
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.687530517578125,
+      "learning_rate": 8.759705653032469e-08,
+      "loss": 0.4004,
+      "step": 57865
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.855264663696289,
+      "learning_rate": 8.754792671820889e-08,
+      "loss": 0.6829,
+      "step": 57866
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.5495219230651855,
+      "learning_rate": 8.749881056578257e-08,
+      "loss": 0.5018,
+      "step": 57867
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 6.13986349105835,
+      "learning_rate": 8.744970807318176e-08,
+      "loss": 0.7029,
+      "step": 57868
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.8938522338867188,
+      "learning_rate": 8.7400619240543e-08,
+      "loss": 0.5432,
+      "step": 57869
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.7337136268615723,
+      "learning_rate": 8.735154406800284e-08,
+      "loss": 0.4885,
+      "step": 57870
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.19753098487854,
+      "learning_rate": 8.730248255569784e-08,
+      "loss": 0.5356,
+      "step": 57871
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.8382456302642822,
+      "learning_rate": 8.725343470376346e-08,
+      "loss": 0.7851,
+      "step": 57872
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.6784167289733887,
+      "learning_rate": 8.720440051233792e-08,
+      "loss": 0.5875,
+      "step": 57873
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.519071102142334,
+      "learning_rate": 8.71553799815561e-08,
+      "loss": 0.5112,
+      "step": 57874
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.470292329788208,
+      "learning_rate": 8.710637311155512e-08,
+      "loss": 0.5774,
+      "step": 57875
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.248599052429199,
+      "learning_rate": 8.705737990247043e-08,
+      "loss": 0.518,
+      "step": 57876
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.740307092666626,
+      "learning_rate": 8.700840035443914e-08,
+      "loss": 0.3944,
+      "step": 57877
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.7862143516540527,
+      "learning_rate": 8.69594344675967e-08,
+      "loss": 0.5484,
+      "step": 57878
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.52744197845459,
+      "learning_rate": 8.69104822420802e-08,
+      "loss": 0.3714,
+      "step": 57879
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.1844749450683594,
+      "learning_rate": 8.686154367802458e-08,
+      "loss": 0.5978,
+      "step": 57880
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.4713189601898193,
+      "learning_rate": 8.68126187755669e-08,
+      "loss": 0.5487,
+      "step": 57881
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.1932108402252197,
+      "learning_rate": 8.676370753484264e-08,
+      "loss": 0.6316,
+      "step": 57882
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.6159331798553467,
+      "learning_rate": 8.67148099559878e-08,
+      "loss": 0.5891,
+      "step": 57883
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.744267463684082,
+      "learning_rate": 8.666592603913837e-08,
+      "loss": 0.5476,
+      "step": 57884
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.6374475955963135,
+      "learning_rate": 8.661705578443091e-08,
+      "loss": 0.4692,
+      "step": 57885
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.0533804893493652,
+      "learning_rate": 8.656819919200088e-08,
+      "loss": 0.4475,
+      "step": 57886
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.6948535442352295,
+      "learning_rate": 8.651935626198371e-08,
+      "loss": 0.3758,
+      "step": 57887
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.9351646900177,
+      "learning_rate": 8.647052699451541e-08,
+      "loss": 0.5723,
+      "step": 57888
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.4315898418426514,
+      "learning_rate": 8.642171138973254e-08,
+      "loss": 0.53,
+      "step": 57889
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.9235293865203857,
+      "learning_rate": 8.637290944776944e-08,
+      "loss": 0.6261,
+      "step": 57890
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.2699241638183594,
+      "learning_rate": 8.632412116876376e-08,
+      "loss": 0.6481,
+      "step": 57891
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 4.219994068145752,
+      "learning_rate": 8.627534655284875e-08,
+      "loss": 0.6453,
+      "step": 57892
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.753758430480957,
+      "learning_rate": 8.62265856001615e-08,
+      "loss": 0.6879,
+      "step": 57893
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 4.085086822509766,
+      "learning_rate": 8.617783831083748e-08,
+      "loss": 0.6432,
+      "step": 57894
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.8671746253967285,
+      "learning_rate": 8.612910468501268e-08,
+      "loss": 0.43,
+      "step": 57895
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.0109317302703857,
+      "learning_rate": 8.608038472282143e-08,
+      "loss": 0.5651,
+      "step": 57896
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.004544734954834,
+      "learning_rate": 8.60316784244003e-08,
+      "loss": 0.4436,
+      "step": 57897
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.1796875,
+      "learning_rate": 8.598298578988417e-08,
+      "loss": 0.6132,
+      "step": 57898
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.3105194568634033,
+      "learning_rate": 8.593430681940851e-08,
+      "loss": 0.3739,
+      "step": 57899
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.0587422847747803,
+      "learning_rate": 8.588564151310818e-08,
+      "loss": 0.6405,
+      "step": 57900
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 4.408985137939453,
+      "learning_rate": 8.583698987111977e-08,
+      "loss": 0.6935,
+      "step": 57901
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 38.760128021240234,
+      "learning_rate": 8.578835189357704e-08,
+      "loss": 0.4601,
+      "step": 57902
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 4.747323036193848,
+      "learning_rate": 8.573972758061655e-08,
+      "loss": 0.6289,
+      "step": 57903
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.000673532485962,
+      "learning_rate": 8.569111693237265e-08,
+      "loss": 0.5216,
+      "step": 57904
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.308580160140991,
+      "learning_rate": 8.564251994898131e-08,
+      "loss": 0.6474,
+      "step": 57905
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.5613884925842285,
+      "learning_rate": 8.559393663057692e-08,
+      "loss": 0.3638,
+      "step": 57906
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 1.8774514198303223,
+      "learning_rate": 8.554536697729433e-08,
+      "loss": 0.5014,
+      "step": 57907
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 4.4124274253845215,
+      "learning_rate": 8.549681098926954e-08,
+      "loss": 0.5937,
+      "step": 57908
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.602168560028076,
+      "learning_rate": 8.544826866663692e-08,
+      "loss": 0.7073,
+      "step": 57909
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.115238904953003,
+      "learning_rate": 8.539974000953188e-08,
+      "loss": 0.5648,
+      "step": 57910
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.2802908420562744,
+      "learning_rate": 8.535122501808879e-08,
+      "loss": 0.5977,
+      "step": 57911
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.907693862915039,
+      "learning_rate": 8.530272369244363e-08,
+      "loss": 0.4698,
+      "step": 57912
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.418442964553833,
+      "learning_rate": 8.525423603272964e-08,
+      "loss": 0.2977,
+      "step": 57913
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.9153716564178467,
+      "learning_rate": 8.520576203908281e-08,
+      "loss": 0.4461,
+      "step": 57914
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.9858641624450684,
+      "learning_rate": 8.51573017116375e-08,
+      "loss": 0.6962,
+      "step": 57915
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 4.383421897888184,
+      "learning_rate": 8.510885505052857e-08,
+      "loss": 0.473,
+      "step": 57916
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 5.350240230560303,
+      "learning_rate": 8.506042205589093e-08,
+      "loss": 0.4268,
+      "step": 57917
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 4.1877970695495605,
+      "learning_rate": 8.501200272785892e-08,
+      "loss": 0.7587,
+      "step": 57918
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 1.9888659715652466,
+      "learning_rate": 8.496359706656743e-08,
+      "loss": 0.3554,
+      "step": 57919
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.4108409881591797,
+      "learning_rate": 8.491520507215134e-08,
+      "loss": 0.6188,
+      "step": 57920
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.8087639808654785,
+      "learning_rate": 8.486682674474445e-08,
+      "loss": 0.4619,
+      "step": 57921
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.052001476287842,
+      "learning_rate": 8.481846208448163e-08,
+      "loss": 0.6542,
+      "step": 57922
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.5184216499328613,
+      "learning_rate": 8.477011109149724e-08,
+      "loss": 0.5811,
+      "step": 57923
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.4353203773498535,
+      "learning_rate": 8.472177376592672e-08,
+      "loss": 0.6695,
+      "step": 57924
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.157343626022339,
+      "learning_rate": 8.467345010790273e-08,
+      "loss": 0.6948,
+      "step": 57925
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.3945422172546387,
+      "learning_rate": 8.462514011756073e-08,
+      "loss": 0.5193,
+      "step": 57926
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.9961931705474854,
+      "learning_rate": 8.457684379503505e-08,
+      "loss": 0.7902,
+      "step": 57927
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.562070369720459,
+      "learning_rate": 8.452856114046004e-08,
+      "loss": 0.6541,
+      "step": 57928
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.6456053256988525,
+      "learning_rate": 8.448029215396891e-08,
+      "loss": 0.6053,
+      "step": 57929
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.6378259658813477,
+      "learning_rate": 8.443203683569767e-08,
+      "loss": 0.6887,
+      "step": 57930
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.26259183883667,
+      "learning_rate": 8.438379518577899e-08,
+      "loss": 0.5106,
+      "step": 57931
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 4.231701374053955,
+      "learning_rate": 8.433556720434777e-08,
+      "loss": 0.4695,
+      "step": 57932
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.4142792224884033,
+      "learning_rate": 8.428735289153722e-08,
+      "loss": 0.7732,
+      "step": 57933
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 4.023147106170654,
+      "learning_rate": 8.42391522474828e-08,
+      "loss": 0.7584,
+      "step": 57934
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.8431549072265625,
+      "learning_rate": 8.419096527231718e-08,
+      "loss": 0.6015,
+      "step": 57935
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.6409475803375244,
+      "learning_rate": 8.41427919661758e-08,
+      "loss": 0.5521,
+      "step": 57936
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 4.079041957855225,
+      "learning_rate": 8.409463232919135e-08,
+      "loss": 0.5289,
+      "step": 57937
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.0186684131622314,
+      "learning_rate": 8.404648636149815e-08,
+      "loss": 0.4063,
+      "step": 57938
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.7272942066192627,
+      "learning_rate": 8.399835406323054e-08,
+      "loss": 0.5344,
+      "step": 57939
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.2071821689605713,
+      "learning_rate": 8.395023543452175e-08,
+      "loss": 0.4515,
+      "step": 57940
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.762361526489258,
+      "learning_rate": 8.3902130475505e-08,
+      "loss": 0.2939,
+      "step": 57941
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.305030107498169,
+      "learning_rate": 8.385403918631574e-08,
+      "loss": 0.6229,
+      "step": 57942
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.3116133213043213,
+      "learning_rate": 8.380596156708665e-08,
+      "loss": 0.76,
+      "step": 57943
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.0730154514312744,
+      "learning_rate": 8.37578976179515e-08,
+      "loss": 0.7889,
+      "step": 57944
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.175495147705078,
+      "learning_rate": 8.370984733904353e-08,
+      "loss": 0.4495,
+      "step": 57945
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.293860673904419,
+      "learning_rate": 8.366181073049707e-08,
+      "loss": 0.6556,
+      "step": 57946
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 6.212005138397217,
+      "learning_rate": 8.361378779244533e-08,
+      "loss": 0.6049,
+      "step": 57947
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.1429834365844727,
+      "learning_rate": 8.356577852502212e-08,
+      "loss": 0.6772,
+      "step": 57948
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.2733113765716553,
+      "learning_rate": 8.351778292836065e-08,
+      "loss": 0.4702,
+      "step": 57949
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 5.349039554595947,
+      "learning_rate": 8.346980100259473e-08,
+      "loss": 0.6573,
+      "step": 57950
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.9835379123687744,
+      "learning_rate": 8.342183274785697e-08,
+      "loss": 0.547,
+      "step": 57951
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.6264545917510986,
+      "learning_rate": 8.337387816428177e-08,
+      "loss": 0.6068,
+      "step": 57952
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.738281726837158,
+      "learning_rate": 8.332593725200122e-08,
+      "loss": 0.7018,
+      "step": 57953
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.372157335281372,
+      "learning_rate": 8.32780100111502e-08,
+      "loss": 0.4824,
+      "step": 57954
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.183180570602417,
+      "learning_rate": 8.323009644186087e-08,
+      "loss": 0.4571,
+      "step": 57955
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.1152844429016113,
+      "learning_rate": 8.318219654426696e-08,
+      "loss": 0.3167,
+      "step": 57956
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.7436106204986572,
+      "learning_rate": 8.313431031850116e-08,
+      "loss": 0.597,
+      "step": 57957
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.5941712856292725,
+      "learning_rate": 8.308643776469727e-08,
+      "loss": 0.7558,
+      "step": 57958
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.726294755935669,
+      "learning_rate": 8.303857888298795e-08,
+      "loss": 0.5188,
+      "step": 57959
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.1918997764587402,
+      "learning_rate": 8.29907336735064e-08,
+      "loss": 0.5079,
+      "step": 57960
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 4.445743560791016,
+      "learning_rate": 8.294290213638589e-08,
+      "loss": 0.6216,
+      "step": 57961
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.2786736488342285,
+      "learning_rate": 8.289508427175908e-08,
+      "loss": 0.5874,
+      "step": 57962
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.9285900592803955,
+      "learning_rate": 8.284728007975862e-08,
+      "loss": 0.4967,
+      "step": 57963
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.6070728302001953,
+      "learning_rate": 8.279948956051886e-08,
+      "loss": 0.3566,
+      "step": 57964
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.6953282356262207,
+      "learning_rate": 8.275171271417081e-08,
+      "loss": 0.7775,
+      "step": 57965
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 4.2581467628479,
+      "learning_rate": 8.270394954084881e-08,
+      "loss": 0.7518,
+      "step": 57966
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.3546056747436523,
+      "learning_rate": 8.265620004068442e-08,
+      "loss": 0.5131,
+      "step": 57967
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 4.850883483886719,
+      "learning_rate": 8.260846421381141e-08,
+      "loss": 0.4164,
+      "step": 57968
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.047682285308838,
+      "learning_rate": 8.256074206036191e-08,
+      "loss": 0.4417,
+      "step": 57969
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.098405122756958,
+      "learning_rate": 8.251303358046914e-08,
+      "loss": 0.6509,
+      "step": 57970
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.305551290512085,
+      "learning_rate": 8.246533877426576e-08,
+      "loss": 0.5021,
+      "step": 57971
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.1103906631469727,
+      "learning_rate": 8.241765764188336e-08,
+      "loss": 0.3197,
+      "step": 57972
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.1054189205169678,
+      "learning_rate": 8.236999018345571e-08,
+      "loss": 0.5792,
+      "step": 57973
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.19789981842041,
+      "learning_rate": 8.232233639911491e-08,
+      "loss": 0.5271,
+      "step": 57974
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.0631332397460938,
+      "learning_rate": 8.227469628899365e-08,
+      "loss": 0.5498,
+      "step": 57975
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.547459840774536,
+      "learning_rate": 8.222706985322404e-08,
+      "loss": 0.4883,
+      "step": 57976
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.2111940383911133,
+      "learning_rate": 8.217945709193876e-08,
+      "loss": 0.6336,
+      "step": 57977
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 4.183290958404541,
+      "learning_rate": 8.213185800527046e-08,
+      "loss": 0.7508,
+      "step": 57978
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.416440725326538,
+      "learning_rate": 8.208427259335073e-08,
+      "loss": 0.5687,
+      "step": 57979
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 4.594489574432373,
+      "learning_rate": 8.203670085631221e-08,
+      "loss": 0.5516,
+      "step": 57980
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.227935552597046,
+      "learning_rate": 8.19891427942876e-08,
+      "loss": 0.8143,
+      "step": 57981
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.5909688472747803,
+      "learning_rate": 8.194159840740901e-08,
+      "loss": 0.6796,
+      "step": 57982
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.101400375366211,
+      "learning_rate": 8.189406769580854e-08,
+      "loss": 0.5667,
+      "step": 57983
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.397965669631958,
+      "learning_rate": 8.184655065961778e-08,
+      "loss": 0.4972,
+      "step": 57984
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 6.178415298461914,
+      "learning_rate": 8.179904729896993e-08,
+      "loss": 0.548,
+      "step": 57985
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 4.128243923187256,
+      "learning_rate": 8.1751557613996e-08,
+      "loss": 0.6384,
+      "step": 57986
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 4.433526515960693,
+      "learning_rate": 8.170408160482923e-08,
+      "loss": 0.8859,
+      "step": 57987
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.312898635864258,
+      "learning_rate": 8.165661927160006e-08,
+      "loss": 0.4782,
+      "step": 57988
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.0928566455841064,
+      "learning_rate": 8.160917061444229e-08,
+      "loss": 0.4519,
+      "step": 57989
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.4897894859313965,
+      "learning_rate": 8.156173563348635e-08,
+      "loss": 0.4764,
+      "step": 57990
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.131845235824585,
+      "learning_rate": 8.151431432886548e-08,
+      "loss": 0.6275,
+      "step": 57991
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.316962480545044,
+      "learning_rate": 8.146690670071011e-08,
+      "loss": 0.4536,
+      "step": 57992
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.494983434677124,
+      "learning_rate": 8.141951274915294e-08,
+      "loss": 0.6135,
+      "step": 57993
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.450657606124878,
+      "learning_rate": 8.137213247432496e-08,
+      "loss": 0.4267,
+      "step": 57994
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.025322914123535,
+      "learning_rate": 8.13247658763594e-08,
+      "loss": 0.6238,
+      "step": 57995
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.095858335494995,
+      "learning_rate": 8.127741295538672e-08,
+      "loss": 0.6069,
+      "step": 57996
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.136855125427246,
+      "learning_rate": 8.123007371153901e-08,
+      "loss": 0.5154,
+      "step": 57997
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.6429243087768555,
+      "learning_rate": 8.118274814494787e-08,
+      "loss": 0.756,
+      "step": 57998
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.8140900135040283,
+      "learning_rate": 8.113543625574483e-08,
+      "loss": 0.5037,
+      "step": 57999
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 5.273845195770264,
+      "learning_rate": 8.108813804406146e-08,
+      "loss": 0.4078,
+      "step": 58000
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.528456687927246,
+      "learning_rate": 8.104085351002988e-08,
+      "loss": 0.5917,
+      "step": 58001
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.1939570903778076,
+      "learning_rate": 8.099358265378e-08,
+      "loss": 0.4732,
+      "step": 58002
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 4.711791515350342,
+      "learning_rate": 8.094632547544501e-08,
+      "loss": 0.5939,
+      "step": 58003
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.133127212524414,
+      "learning_rate": 8.08990819751554e-08,
+      "loss": 0.7255,
+      "step": 58004
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 5.177954196929932,
+      "learning_rate": 8.08518521530427e-08,
+      "loss": 0.7665,
+      "step": 58005
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.7470431327819824,
+      "learning_rate": 8.080463600923793e-08,
+      "loss": 0.6281,
+      "step": 58006
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.719846487045288,
+      "learning_rate": 8.075743354387322e-08,
+      "loss": 0.5557,
+      "step": 58007
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.905430793762207,
+      "learning_rate": 8.0710244757079e-08,
+      "loss": 0.6521,
+      "step": 58008
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.5179603099823,
+      "learning_rate": 8.066306964898685e-08,
+      "loss": 0.4046,
+      "step": 58009
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.463534355163574,
+      "learning_rate": 8.061590821972775e-08,
+      "loss": 0.6134,
+      "step": 58010
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.403177261352539,
+      "learning_rate": 8.05687604694333e-08,
+      "loss": 0.4968,
+      "step": 58011
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.1293070316314697,
+      "learning_rate": 8.052162639823391e-08,
+      "loss": 0.7716,
+      "step": 58012
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.138312578201294,
+      "learning_rate": 8.047450600626117e-08,
+      "loss": 0.4386,
+      "step": 58013
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.0158562660217285,
+      "learning_rate": 8.042739929364607e-08,
+      "loss": 0.5092,
+      "step": 58014
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 4.064672946929932,
+      "learning_rate": 8.038030626051907e-08,
+      "loss": 0.589,
+      "step": 58015
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.9338085651397705,
+      "learning_rate": 8.033322690701117e-08,
+      "loss": 0.6093,
+      "step": 58016
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.2068049907684326,
+      "learning_rate": 8.02861612332545e-08,
+      "loss": 0.5494,
+      "step": 58017
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.9964230060577393,
+      "learning_rate": 8.023910923937839e-08,
+      "loss": 0.4998,
+      "step": 58018
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.628772020339966,
+      "learning_rate": 8.019207092551495e-08,
+      "loss": 0.435,
+      "step": 58019
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.53835391998291,
+      "learning_rate": 8.014504629179409e-08,
+      "loss": 0.7555,
+      "step": 58020
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.6762218475341797,
+      "learning_rate": 8.009803533834681e-08,
+      "loss": 0.4708,
+      "step": 58021
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 5.022742748260498,
+      "learning_rate": 8.005103806530357e-08,
+      "loss": 0.7137,
+      "step": 58022
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.977266550064087,
+      "learning_rate": 8.000405447279591e-08,
+      "loss": 0.6844,
+      "step": 58023
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.777097702026367,
+      "learning_rate": 7.995708456095374e-08,
+      "loss": 0.5614,
+      "step": 58024
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.2414262294769287,
+      "learning_rate": 7.991012832990752e-08,
+      "loss": 0.6628,
+      "step": 58025
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.2656702995300293,
+      "learning_rate": 7.986318577978825e-08,
+      "loss": 0.6535,
+      "step": 58026
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.6798923015594482,
+      "learning_rate": 7.981625691072636e-08,
+      "loss": 0.6933,
+      "step": 58027
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 4.456367492675781,
+      "learning_rate": 7.976934172285233e-08,
+      "loss": 0.7962,
+      "step": 58028
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.52484130859375,
+      "learning_rate": 7.972244021629715e-08,
+      "loss": 0.5618,
+      "step": 58029
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.9024834632873535,
+      "learning_rate": 7.967555239118963e-08,
+      "loss": 0.5259,
+      "step": 58030
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.840492010116577,
+      "learning_rate": 7.962867824766185e-08,
+      "loss": 0.5076,
+      "step": 58031
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.5752384662628174,
+      "learning_rate": 7.958181778584317e-08,
+      "loss": 0.4971,
+      "step": 58032
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.5513484477996826,
+      "learning_rate": 7.953497100586461e-08,
+      "loss": 0.7402,
+      "step": 58033
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.651024580001831,
+      "learning_rate": 7.948813790785604e-08,
+      "loss": 0.9587,
+      "step": 58034
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.656229019165039,
+      "learning_rate": 7.944131849194736e-08,
+      "loss": 0.6986,
+      "step": 58035
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 4.240996360778809,
+      "learning_rate": 7.939451275826904e-08,
+      "loss": 0.4473,
+      "step": 58036
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.649360418319702,
+      "learning_rate": 7.934772070695152e-08,
+      "loss": 0.6779,
+      "step": 58037
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.00312876701355,
+      "learning_rate": 7.930094233812469e-08,
+      "loss": 0.4793,
+      "step": 58038
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 9.107254028320312,
+      "learning_rate": 7.925417765191845e-08,
+      "loss": 0.587,
+      "step": 58039
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.6841742992401123,
+      "learning_rate": 7.920742664846271e-08,
+      "loss": 0.3172,
+      "step": 58040
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.0391316413879395,
+      "learning_rate": 7.916068932788845e-08,
+      "loss": 0.4118,
+      "step": 58041
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.7383902072906494,
+      "learning_rate": 7.911396569032448e-08,
+      "loss": 0.4843,
+      "step": 58042
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.875478744506836,
+      "learning_rate": 7.906725573590123e-08,
+      "loss": 0.6358,
+      "step": 58043
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.304779291152954,
+      "learning_rate": 7.902055946474863e-08,
+      "loss": 0.5933,
+      "step": 58044
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.9599575996398926,
+      "learning_rate": 7.897387687699599e-08,
+      "loss": 0.4784,
+      "step": 58045
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.4686455726623535,
+      "learning_rate": 7.892720797277431e-08,
+      "loss": 0.5491,
+      "step": 58046
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.513930320739746,
+      "learning_rate": 7.888055275221185e-08,
+      "loss": 0.6969,
+      "step": 58047
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.533618688583374,
+      "learning_rate": 7.88339112154396e-08,
+      "loss": 0.4646,
+      "step": 58048
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.363896369934082,
+      "learning_rate": 7.878728336258635e-08,
+      "loss": 0.5047,
+      "step": 58049
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.936384439468384,
+      "learning_rate": 7.874066919378254e-08,
+      "loss": 0.7121,
+      "step": 58050
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.712855100631714,
+      "learning_rate": 7.869406870915697e-08,
+      "loss": 0.6333,
+      "step": 58051
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 1.9125583171844482,
+      "learning_rate": 7.864748190884008e-08,
+      "loss": 0.4219,
+      "step": 58052
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.5579514503479004,
+      "learning_rate": 7.860090879296067e-08,
+      "loss": 0.5118,
+      "step": 58053
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.560336112976074,
+      "learning_rate": 7.855434936164863e-08,
+      "loss": 0.4927,
+      "step": 58054
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.6968507766723633,
+      "learning_rate": 7.850780361503274e-08,
+      "loss": 0.6727,
+      "step": 58055
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.5284454822540283,
+      "learning_rate": 7.846127155324401e-08,
+      "loss": 0.7111,
+      "step": 58056
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.8606069087982178,
+      "learning_rate": 7.841475317641012e-08,
+      "loss": 0.6298,
+      "step": 58057
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.460594654083252,
+      "learning_rate": 7.836824848466151e-08,
+      "loss": 0.6604,
+      "step": 58058
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 7.3509039878845215,
+      "learning_rate": 7.832175747812697e-08,
+      "loss": 0.6524,
+      "step": 58059
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.1221060752868652,
+      "learning_rate": 7.827528015693586e-08,
+      "loss": 0.6174,
+      "step": 58060
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.135624885559082,
+      "learning_rate": 7.82288165212175e-08,
+      "loss": 0.6087,
+      "step": 58061
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.862416982650757,
+      "learning_rate": 7.818236657110123e-08,
+      "loss": 0.6739,
+      "step": 58062
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.916846513748169,
+      "learning_rate": 7.813593030671585e-08,
+      "loss": 0.6926,
+      "step": 58063
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.929398536682129,
+      "learning_rate": 7.808950772819124e-08,
+      "loss": 0.7684,
+      "step": 58064
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.2776427268981934,
+      "learning_rate": 7.804309883565509e-08,
+      "loss": 0.7242,
+      "step": 58065
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 4.646359920501709,
+      "learning_rate": 7.799670362923784e-08,
+      "loss": 0.6929,
+      "step": 58066
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.5061583518981934,
+      "learning_rate": 7.795032210906772e-08,
+      "loss": 0.5995,
+      "step": 58067
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.4320833683013916,
+      "learning_rate": 7.790395427527409e-08,
+      "loss": 0.593,
+      "step": 58068
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.362511396408081,
+      "learning_rate": 7.785760012798516e-08,
+      "loss": 0.5603,
+      "step": 58069
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.7288591861724854,
+      "learning_rate": 7.781125966733083e-08,
+      "loss": 0.3611,
+      "step": 58070
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 4.315158367156982,
+      "learning_rate": 7.776493289343934e-08,
+      "loss": 0.6336,
+      "step": 58071
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.1084940433502197,
+      "learning_rate": 7.771861980644002e-08,
+      "loss": 0.4873,
+      "step": 58072
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 4.644721984863281,
+      "learning_rate": 7.767232040646056e-08,
+      "loss": 0.7086,
+      "step": 58073
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.735769748687744,
+      "learning_rate": 7.762603469363139e-08,
+      "loss": 0.6514,
+      "step": 58074
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.5800514221191406,
+      "learning_rate": 7.757976266807966e-08,
+      "loss": 0.6076,
+      "step": 58075
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.4553444385528564,
+      "learning_rate": 7.753350432993412e-08,
+      "loss": 0.5386,
+      "step": 58076
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.6027817726135254,
+      "learning_rate": 7.748725967932468e-08,
+      "loss": 0.5669,
+      "step": 58077
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.655026912689209,
+      "learning_rate": 7.744102871637904e-08,
+      "loss": 0.41,
+      "step": 58078
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.12321400642395,
+      "learning_rate": 7.73948114412254e-08,
+      "loss": 0.4413,
+      "step": 58079
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 4.016123294830322,
+      "learning_rate": 7.734860785399311e-08,
+      "loss": 0.7195,
+      "step": 58080
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.5304036140441895,
+      "learning_rate": 7.730241795481042e-08,
+      "loss": 0.6263,
+      "step": 58081
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.7237255573272705,
+      "learning_rate": 7.725624174380553e-08,
+      "loss": 0.7642,
+      "step": 58082
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.4078426361083984,
+      "learning_rate": 7.721007922110668e-08,
+      "loss": 0.585,
+      "step": 58083
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.3640177249908447,
+      "learning_rate": 7.716393038684322e-08,
+      "loss": 0.694,
+      "step": 58084
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.4140942096710205,
+      "learning_rate": 7.711779524114172e-08,
+      "loss": 0.6568,
+      "step": 58085
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.371615171432495,
+      "learning_rate": 7.707167378413205e-08,
+      "loss": 0.4066,
+      "step": 58086
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.785512924194336,
+      "learning_rate": 7.70255660159419e-08,
+      "loss": 0.3758,
+      "step": 58087
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.703381299972534,
+      "learning_rate": 7.697947193669953e-08,
+      "loss": 0.5218,
+      "step": 58088
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.200854778289795,
+      "learning_rate": 7.693339154653257e-08,
+      "loss": 0.4854,
+      "step": 58089
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 5.040499687194824,
+      "learning_rate": 7.688732484557038e-08,
+      "loss": 0.7619,
+      "step": 58090
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.9090616703033447,
+      "learning_rate": 7.684127183393952e-08,
+      "loss": 0.4866,
+      "step": 58091
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.8879754543304443,
+      "learning_rate": 7.679523251176879e-08,
+      "loss": 0.5974,
+      "step": 58092
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.4530484676361084,
+      "learning_rate": 7.674920687918697e-08,
+      "loss": 0.3648,
+      "step": 58093
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.7909491062164307,
+      "learning_rate": 7.670319493632061e-08,
+      "loss": 0.5211,
+      "step": 58094
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.69537353515625,
+      "learning_rate": 7.665719668329908e-08,
+      "loss": 0.6312,
+      "step": 58095
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.278217077255249,
+      "learning_rate": 7.661121212024891e-08,
+      "loss": 0.5283,
+      "step": 58096
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.5526065826416016,
+      "learning_rate": 7.656524124729892e-08,
+      "loss": 0.6041,
+      "step": 58097
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.5933010578155518,
+      "learning_rate": 7.651928406457676e-08,
+      "loss": 0.5573,
+      "step": 58098
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.98775315284729,
+      "learning_rate": 7.647334057220956e-08,
+      "loss": 0.5642,
+      "step": 58099
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.1623306274414062,
+      "learning_rate": 7.642741077032611e-08,
+      "loss": 0.7196,
+      "step": 58100
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.007042169570923,
+      "learning_rate": 7.638149465905354e-08,
+      "loss": 0.5785,
+      "step": 58101
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.6551504135131836,
+      "learning_rate": 7.63355922385195e-08,
+      "loss": 0.5547,
+      "step": 58102
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.972445011138916,
+      "learning_rate": 7.628970350885168e-08,
+      "loss": 0.7031,
+      "step": 58103
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.2815675735473633,
+      "learning_rate": 7.624382847017775e-08,
+      "loss": 0.4359,
+      "step": 58104
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.411829710006714,
+      "learning_rate": 7.619796712262539e-08,
+      "loss": 0.6055,
+      "step": 58105
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.2594990730285645,
+      "learning_rate": 7.615211946632173e-08,
+      "loss": 0.4991,
+      "step": 58106
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 1.9829397201538086,
+      "learning_rate": 7.610628550139498e-08,
+      "loss": 0.396,
+      "step": 58107
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 4.0455169677734375,
+      "learning_rate": 7.606046522797173e-08,
+      "loss": 0.6186,
+      "step": 58108
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 5.964128494262695,
+      "learning_rate": 7.601465864617962e-08,
+      "loss": 0.4543,
+      "step": 58109
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.668023109436035,
+      "learning_rate": 7.596886575614638e-08,
+      "loss": 0.7322,
+      "step": 58110
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.312838077545166,
+      "learning_rate": 7.592308655799962e-08,
+      "loss": 0.4997,
+      "step": 58111
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.60903000831604,
+      "learning_rate": 7.58773210518654e-08,
+      "loss": 0.678,
+      "step": 58112
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.9926562309265137,
+      "learning_rate": 7.58315692378725e-08,
+      "loss": 0.5475,
+      "step": 58113
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.4040637016296387,
+      "learning_rate": 7.578583111614635e-08,
+      "loss": 0.3449,
+      "step": 58114
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 4.5835747718811035,
+      "learning_rate": 7.574010668681631e-08,
+      "loss": 0.4979,
+      "step": 58115
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.36548113822937,
+      "learning_rate": 7.569439595000726e-08,
+      "loss": 0.5494,
+      "step": 58116
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.678929090499878,
+      "learning_rate": 7.564869890584802e-08,
+      "loss": 0.5673,
+      "step": 58117
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.1576170921325684,
+      "learning_rate": 7.560301555446458e-08,
+      "loss": 0.5983,
+      "step": 58118
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.796557664871216,
+      "learning_rate": 7.555734589598518e-08,
+      "loss": 0.5535,
+      "step": 58119
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 4.1769866943359375,
+      "learning_rate": 7.551168993053526e-08,
+      "loss": 0.4375,
+      "step": 58120
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.478898048400879,
+      "learning_rate": 7.546604765824362e-08,
+      "loss": 0.4628,
+      "step": 58121
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.6701571941375732,
+      "learning_rate": 7.542041907923514e-08,
+      "loss": 0.7306,
+      "step": 58122
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.553881883621216,
+      "learning_rate": 7.537480419363807e-08,
+      "loss": 0.6474,
+      "step": 58123
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.9432461261749268,
+      "learning_rate": 7.532920300157898e-08,
+      "loss": 0.5386,
+      "step": 58124
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.4853334426879883,
+      "learning_rate": 7.528361550318442e-08,
+      "loss": 0.7374,
+      "step": 58125
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.3727304935455322,
+      "learning_rate": 7.523804169858095e-08,
+      "loss": 0.6854,
+      "step": 58126
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.540339946746826,
+      "learning_rate": 7.519248158789627e-08,
+      "loss": 0.6146,
+      "step": 58127
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.024172067642212,
+      "learning_rate": 7.514693517125581e-08,
+      "loss": 0.8133,
+      "step": 58128
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.348630905151367,
+      "learning_rate": 7.510140244878783e-08,
+      "loss": 0.5467,
+      "step": 58129
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.3929443359375,
+      "learning_rate": 7.505588342061665e-08,
+      "loss": 0.5484,
+      "step": 58130
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.6921253204345703,
+      "learning_rate": 7.501037808687107e-08,
+      "loss": 0.5779,
+      "step": 58131
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.532001495361328,
+      "learning_rate": 7.496488644767596e-08,
+      "loss": 0.6429,
+      "step": 58132
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.0664215087890625,
+      "learning_rate": 7.49194085031596e-08,
+      "loss": 0.4724,
+      "step": 58133
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.1711082458496094,
+      "learning_rate": 7.48739442534463e-08,
+      "loss": 0.4437,
+      "step": 58134
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.5470175743103027,
+      "learning_rate": 7.482849369866429e-08,
+      "loss": 0.6413,
+      "step": 58135
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.392274856567383,
+      "learning_rate": 7.47830568389385e-08,
+      "loss": 0.6874,
+      "step": 58136
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.3964452743530273,
+      "learning_rate": 7.473763367439656e-08,
+      "loss": 0.5975,
+      "step": 58137
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.6566197872161865,
+      "learning_rate": 7.469222420516398e-08,
+      "loss": 0.4708,
+      "step": 58138
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.713595390319824,
+      "learning_rate": 7.464682843136728e-08,
+      "loss": 0.5247,
+      "step": 58139
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.0503363609313965,
+      "learning_rate": 7.460144635313249e-08,
+      "loss": 0.4169,
+      "step": 58140
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.3369314670562744,
+      "learning_rate": 7.455607797058673e-08,
+      "loss": 0.6294,
+      "step": 58141
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.827045202255249,
+      "learning_rate": 7.451072328385433e-08,
+      "loss": 0.5346,
+      "step": 58142
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.1547305583953857,
+      "learning_rate": 7.446538229306355e-08,
+      "loss": 0.7353,
+      "step": 58143
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.949505567550659,
+      "learning_rate": 7.44200549983387e-08,
+      "loss": 0.6464,
+      "step": 58144
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.5732357501983643,
+      "learning_rate": 7.437474139980694e-08,
+      "loss": 0.4882,
+      "step": 58145
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.05329966545105,
+      "learning_rate": 7.432944149759313e-08,
+      "loss": 0.5596,
+      "step": 58146
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.9840612411499023,
+      "learning_rate": 7.428415529182498e-08,
+      "loss": 0.6681,
+      "step": 58147
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.277034282684326,
+      "learning_rate": 7.423888278262681e-08,
+      "loss": 0.5841,
+      "step": 58148
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.531709909439087,
+      "learning_rate": 7.41936239701252e-08,
+      "loss": 0.5743,
+      "step": 58149
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.9222288131713867,
+      "learning_rate": 7.414837885444559e-08,
+      "loss": 0.6156,
+      "step": 58150
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 4.709177494049072,
+      "learning_rate": 7.410314743571456e-08,
+      "loss": 0.5874,
+      "step": 58151
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.2766175270080566,
+      "learning_rate": 7.405792971405646e-08,
+      "loss": 0.6284,
+      "step": 58152
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.562244176864624,
+      "learning_rate": 7.401272568959894e-08,
+      "loss": 0.476,
+      "step": 58153
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.4329309463500977,
+      "learning_rate": 7.396753536246637e-08,
+      "loss": 0.5971,
+      "step": 58154
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.7456183433532715,
+      "learning_rate": 7.39223587327842e-08,
+      "loss": 0.7483,
+      "step": 58155
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.3183326721191406,
+      "learning_rate": 7.387719580067953e-08,
+      "loss": 0.7608,
+      "step": 58156
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.031442165374756,
+      "learning_rate": 7.383204656627673e-08,
+      "loss": 0.5881,
+      "step": 58157
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 9.901435852050781,
+      "learning_rate": 7.378691102970126e-08,
+      "loss": 0.8529,
+      "step": 58158
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.5547127723693848,
+      "learning_rate": 7.374178919107911e-08,
+      "loss": 0.5927,
+      "step": 58159
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.46028995513916,
+      "learning_rate": 7.369668105053573e-08,
+      "loss": 0.5971,
+      "step": 58160
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.6117916107177734,
+      "learning_rate": 7.36515866081966e-08,
+      "loss": 0.612,
+      "step": 58161
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.324634552001953,
+      "learning_rate": 7.36065058641866e-08,
+      "loss": 0.7208,
+      "step": 58162
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 4.090143203735352,
+      "learning_rate": 7.356143881863175e-08,
+      "loss": 0.6882,
+      "step": 58163
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.839597225189209,
+      "learning_rate": 7.351638547165696e-08,
+      "loss": 0.6018,
+      "step": 58164
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.8271687030792236,
+      "learning_rate": 7.347134582338767e-08,
+      "loss": 0.6251,
+      "step": 58165
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.0588254928588867,
+      "learning_rate": 7.342631987394877e-08,
+      "loss": 0.6012,
+      "step": 58166
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 5.053055763244629,
+      "learning_rate": 7.338130762346629e-08,
+      "loss": 0.4863,
+      "step": 58167
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 4.400653839111328,
+      "learning_rate": 7.333630907206457e-08,
+      "loss": 0.6222,
+      "step": 58168
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.864344596862793,
+      "learning_rate": 7.329132421986851e-08,
+      "loss": 0.575,
+      "step": 58169
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.4523439407348633,
+      "learning_rate": 7.324635306700468e-08,
+      "loss": 0.6481,
+      "step": 58170
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.958770751953125,
+      "learning_rate": 7.320139561359629e-08,
+      "loss": 0.5715,
+      "step": 58171
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.157966375350952,
+      "learning_rate": 7.315645185976939e-08,
+      "loss": 0.623,
+      "step": 58172
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.322591543197632,
+      "learning_rate": 7.311152180564885e-08,
+      "loss": 0.551,
+      "step": 58173
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.2579541206359863,
+      "learning_rate": 7.306660545135957e-08,
+      "loss": 0.5638,
+      "step": 58174
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.1189260482788086,
+      "learning_rate": 7.302170279702647e-08,
+      "loss": 0.5665,
+      "step": 58175
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 4.013131141662598,
+      "learning_rate": 7.297681384277444e-08,
+      "loss": 0.7375,
+      "step": 58176
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.76326322555542,
+      "learning_rate": 7.293193858872727e-08,
+      "loss": 0.7074,
+      "step": 58177
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 4.388077259063721,
+      "learning_rate": 7.288707703501152e-08,
+      "loss": 0.6295,
+      "step": 58178
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.776374101638794,
+      "learning_rate": 7.284222918175044e-08,
+      "loss": 0.5751,
+      "step": 58179
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.1061511039733887,
+      "learning_rate": 7.279739502907002e-08,
+      "loss": 0.51,
+      "step": 58180
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.5353481769561768,
+      "learning_rate": 7.275257457709406e-08,
+      "loss": 0.5529,
+      "step": 58181
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.792093276977539,
+      "learning_rate": 7.270776782594746e-08,
+      "loss": 0.4749,
+      "step": 58182
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.9335432052612305,
+      "learning_rate": 7.266297477575457e-08,
+      "loss": 0.3959,
+      "step": 58183
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.4874942302703857,
+      "learning_rate": 7.261819542664028e-08,
+      "loss": 0.6273,
+      "step": 58184
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.0968563556671143,
+      "learning_rate": 7.257342977872894e-08,
+      "loss": 0.5285,
+      "step": 58185
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.4654579162597656,
+      "learning_rate": 7.25286778321449e-08,
+      "loss": 0.5902,
+      "step": 58186
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.4983761310577393,
+      "learning_rate": 7.248393958701306e-08,
+      "loss": 0.4593,
+      "step": 58187
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.148723602294922,
+      "learning_rate": 7.243921504345719e-08,
+      "loss": 0.4449,
+      "step": 58188
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 5.960835933685303,
+      "learning_rate": 7.239450420160221e-08,
+      "loss": 0.7209,
+      "step": 58189
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.015885353088379,
+      "learning_rate": 7.234980706157191e-08,
+      "loss": 0.7369,
+      "step": 58190
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.91803240776062,
+      "learning_rate": 7.230512362349062e-08,
+      "loss": 0.6332,
+      "step": 58191
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 4.080280303955078,
+      "learning_rate": 7.226045388748381e-08,
+      "loss": 0.7189,
+      "step": 58192
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.216127872467041,
+      "learning_rate": 7.221579785367361e-08,
+      "loss": 0.6246,
+      "step": 58193
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.1560559272766113,
+      "learning_rate": 7.2171155522186e-08,
+      "loss": 0.6418,
+      "step": 58194
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 4.260299205780029,
+      "learning_rate": 7.212652689314425e-08,
+      "loss": 0.7732,
+      "step": 58195
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.733870506286621,
+      "learning_rate": 7.208191196667269e-08,
+      "loss": 0.3897,
+      "step": 58196
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.3464853763580322,
+      "learning_rate": 7.203731074289455e-08,
+      "loss": 0.4492,
+      "step": 58197
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 4.5485615730285645,
+      "learning_rate": 7.199272322193529e-08,
+      "loss": 0.5251,
+      "step": 58198
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 6.9344916343688965,
+      "learning_rate": 7.194814940391759e-08,
+      "loss": 0.5991,
+      "step": 58199
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 5.95155143737793,
+      "learning_rate": 7.19035892889669e-08,
+      "loss": 0.3045,
+      "step": 58200
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.9670300483703613,
+      "learning_rate": 7.185904287720535e-08,
+      "loss": 0.3994,
+      "step": 58201
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.0921268463134766,
+      "learning_rate": 7.18145101687584e-08,
+      "loss": 0.4311,
+      "step": 58202
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.6587817668914795,
+      "learning_rate": 7.176999116374872e-08,
+      "loss": 0.6542,
+      "step": 58203
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 4.4149322509765625,
+      "learning_rate": 7.172548586230066e-08,
+      "loss": 0.6969,
+      "step": 58204
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.807213306427002,
+      "learning_rate": 7.168099426453801e-08,
+      "loss": 0.4828,
+      "step": 58205
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 4.233402252197266,
+      "learning_rate": 7.1636516370584e-08,
+      "loss": 0.5473,
+      "step": 58206
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.6216511726379395,
+      "learning_rate": 7.159205218056298e-08,
+      "loss": 0.6192,
+      "step": 58207
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.4407334327697754,
+      "learning_rate": 7.154760169459818e-08,
+      "loss": 0.3864,
+      "step": 58208
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.894071578979492,
+      "learning_rate": 7.150316491281284e-08,
+      "loss": 0.7287,
+      "step": 58209
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.870518922805786,
+      "learning_rate": 7.14587418353313e-08,
+      "loss": 0.5146,
+      "step": 58210
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.2853682041168213,
+      "learning_rate": 7.14143324622768e-08,
+      "loss": 0.6069,
+      "step": 58211
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.7838950157165527,
+      "learning_rate": 7.136993679377258e-08,
+      "loss": 0.513,
+      "step": 58212
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.068403959274292,
+      "learning_rate": 7.132555482994241e-08,
+      "loss": 0.6902,
+      "step": 58213
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.158055067062378,
+      "learning_rate": 7.128118657090954e-08,
+      "loss": 0.5233,
+      "step": 58214
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.856638193130493,
+      "learning_rate": 7.123683201679721e-08,
+      "loss": 0.5567,
+      "step": 58215
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.8293371200561523,
+      "learning_rate": 7.119249116772864e-08,
+      "loss": 0.6127,
+      "step": 58216
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.9640395641326904,
+      "learning_rate": 7.114816402382763e-08,
+      "loss": 0.5301,
+      "step": 58217
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.1872153282165527,
+      "learning_rate": 7.11038505852174e-08,
+      "loss": 0.5884,
+      "step": 58218
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.7925949096679688,
+      "learning_rate": 7.10595508520201e-08,
+      "loss": 0.6149,
+      "step": 58219
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 4.874464511871338,
+      "learning_rate": 7.101526482436117e-08,
+      "loss": 0.6071,
+      "step": 58220
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.6077377796173096,
+      "learning_rate": 7.097099250236106e-08,
+      "loss": 0.6345,
+      "step": 58221
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.422222852706909,
+      "learning_rate": 7.092673388614524e-08,
+      "loss": 0.4853,
+      "step": 58222
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.418842077255249,
+      "learning_rate": 7.088248897583472e-08,
+      "loss": 0.6499,
+      "step": 58223
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 4.277017116546631,
+      "learning_rate": 7.083825777155439e-08,
+      "loss": 0.7902,
+      "step": 58224
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.272397041320801,
+      "learning_rate": 7.079404027342584e-08,
+      "loss": 0.4562,
+      "step": 58225
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.5463707447052,
+      "learning_rate": 7.074983648157285e-08,
+      "loss": 0.4477,
+      "step": 58226
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.2903172969818115,
+      "learning_rate": 7.070564639611755e-08,
+      "loss": 0.7019,
+      "step": 58227
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 4.254712104797363,
+      "learning_rate": 7.066147001718316e-08,
+      "loss": 0.7513,
+      "step": 58228
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 13.731742858886719,
+      "learning_rate": 7.061730734489348e-08,
+      "loss": 0.5711,
+      "step": 58229
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.6985890865325928,
+      "learning_rate": 7.057315837936951e-08,
+      "loss": 0.5831,
+      "step": 58230
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.52565336227417,
+      "learning_rate": 7.052902312073617e-08,
+      "loss": 0.2803,
+      "step": 58231
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 4.878519058227539,
+      "learning_rate": 7.048490156911392e-08,
+      "loss": 0.4839,
+      "step": 58232
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.9185822010040283,
+      "learning_rate": 7.044079372462708e-08,
+      "loss": 0.668,
+      "step": 58233
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.686067581176758,
+      "learning_rate": 7.039669958739781e-08,
+      "loss": 0.6382,
+      "step": 58234
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.7837932109832764,
+      "learning_rate": 7.035261915754821e-08,
+      "loss": 0.5557,
+      "step": 58235
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.706676483154297,
+      "learning_rate": 7.030855243520152e-08,
+      "loss": 0.7342,
+      "step": 58236
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.7120630741119385,
+      "learning_rate": 7.026449942048041e-08,
+      "loss": 0.6609,
+      "step": 58237
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.862732172012329,
+      "learning_rate": 7.022046011350647e-08,
+      "loss": 0.5168,
+      "step": 58238
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.1804120540618896,
+      "learning_rate": 7.017643451440237e-08,
+      "loss": 0.4209,
+      "step": 58239
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 5.07828426361084,
+      "learning_rate": 7.013242262329134e-08,
+      "loss": 0.6918,
+      "step": 58240
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.575516700744629,
+      "learning_rate": 7.008842444029551e-08,
+      "loss": 0.6834,
+      "step": 58241
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 6.222311496734619,
+      "learning_rate": 7.004443996553645e-08,
+      "loss": 0.8389,
+      "step": 58242
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.8071212768554688,
+      "learning_rate": 7.000046919913739e-08,
+      "loss": 0.6881,
+      "step": 58243
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.373880624771118,
+      "learning_rate": 6.99565121412199e-08,
+      "loss": 0.5561,
+      "step": 58244
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.2633919715881348,
+      "learning_rate": 6.991256879190722e-08,
+      "loss": 0.513,
+      "step": 58245
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.557283878326416,
+      "learning_rate": 6.98686391513198e-08,
+      "loss": 0.555,
+      "step": 58246
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.413004159927368,
+      "learning_rate": 6.982472321958201e-08,
+      "loss": 0.4706,
+      "step": 58247
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.045797348022461,
+      "learning_rate": 6.978082099681372e-08,
+      "loss": 0.6589,
+      "step": 58248
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.411054849624634,
+      "learning_rate": 6.973693248313873e-08,
+      "loss": 0.5886,
+      "step": 58249
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.982358932495117,
+      "learning_rate": 6.969305767867806e-08,
+      "loss": 0.465,
+      "step": 58250
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.7773611545562744,
+      "learning_rate": 6.96491965835544e-08,
+      "loss": 0.5808,
+      "step": 58251
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.440070629119873,
+      "learning_rate": 6.960534919788875e-08,
+      "loss": 0.5055,
+      "step": 58252
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.356562614440918,
+      "learning_rate": 6.956151552180435e-08,
+      "loss": 0.6668,
+      "step": 58253
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.172334671020508,
+      "learning_rate": 6.951769555542165e-08,
+      "loss": 0.6431,
+      "step": 58254
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.990157127380371,
+      "learning_rate": 6.94738892988639e-08,
+      "loss": 0.5087,
+      "step": 58255
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.41776967048645,
+      "learning_rate": 6.94300967522521e-08,
+      "loss": 0.6437,
+      "step": 58256
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.5518970489501953,
+      "learning_rate": 6.938631791570838e-08,
+      "loss": 0.7606,
+      "step": 58257
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.675168514251709,
+      "learning_rate": 6.934255278935375e-08,
+      "loss": 0.5948,
+      "step": 58258
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.953042984008789,
+      "learning_rate": 6.929880137331035e-08,
+      "loss": 0.6164,
+      "step": 58259
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.3160386085510254,
+      "learning_rate": 6.925506366769974e-08,
+      "loss": 0.4871,
+      "step": 58260
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.5860090255737305,
+      "learning_rate": 6.92113396726446e-08,
+      "loss": 0.6502,
+      "step": 58261
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.4350016117095947,
+      "learning_rate": 6.916762938826482e-08,
+      "loss": 0.6482,
+      "step": 58262
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.1513919830322266,
+      "learning_rate": 6.91239328146831e-08,
+      "loss": 0.4812,
+      "step": 58263
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.2824952602386475,
+      "learning_rate": 6.908024995202045e-08,
+      "loss": 0.5341,
+      "step": 58264
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.897078514099121,
+      "learning_rate": 6.903658080039899e-08,
+      "loss": 0.671,
+      "step": 58265
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.667323589324951,
+      "learning_rate": 6.899292535993862e-08,
+      "loss": 0.4086,
+      "step": 58266
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 4.60669469833374,
+      "learning_rate": 6.894928363076259e-08,
+      "loss": 0.7976,
+      "step": 58267
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.235476016998291,
+      "learning_rate": 6.890565561299078e-08,
+      "loss": 0.464,
+      "step": 58268
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.982247829437256,
+      "learning_rate": 6.88620413067459e-08,
+      "loss": 0.4989,
+      "step": 58269
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.760526418685913,
+      "learning_rate": 6.881844071214783e-08,
+      "loss": 0.7169,
+      "step": 58270
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.144686698913574,
+      "learning_rate": 6.87748538293187e-08,
+      "loss": 0.7301,
+      "step": 58271
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.3317017555236816,
+      "learning_rate": 6.873128065837897e-08,
+      "loss": 0.6887,
+      "step": 58272
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.2955751419067383,
+      "learning_rate": 6.868772119945078e-08,
+      "loss": 0.6559,
+      "step": 58273
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.381964921951294,
+      "learning_rate": 6.8644175452654e-08,
+      "loss": 0.4924,
+      "step": 58274
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.462136745452881,
+      "learning_rate": 6.860064341811134e-08,
+      "loss": 0.6247,
+      "step": 58275
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.7861711978912354,
+      "learning_rate": 6.855712509594215e-08,
+      "loss": 0.6319,
+      "step": 58276
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.8474934101104736,
+      "learning_rate": 6.851362048626852e-08,
+      "loss": 0.4402,
+      "step": 58277
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.897932291030884,
+      "learning_rate": 6.84701295892115e-08,
+      "loss": 0.6667,
+      "step": 58278
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.668304681777954,
+      "learning_rate": 6.842665240489099e-08,
+      "loss": 0.9417,
+      "step": 58279
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.6606812477111816,
+      "learning_rate": 6.83831889334291e-08,
+      "loss": 0.6282,
+      "step": 58280
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.059816837310791,
+      "learning_rate": 6.83397391749463e-08,
+      "loss": 0.3909,
+      "step": 58281
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.89164400100708,
+      "learning_rate": 6.829630312956248e-08,
+      "loss": 0.7025,
+      "step": 58282
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.608706474304199,
+      "learning_rate": 6.825288079739978e-08,
+      "loss": 0.7771,
+      "step": 58283
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.683171510696411,
+      "learning_rate": 6.82094721785781e-08,
+      "loss": 0.6893,
+      "step": 58284
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 4.68237829208374,
+      "learning_rate": 6.816607727321845e-08,
+      "loss": 0.6303,
+      "step": 58285
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 4.112144470214844,
+      "learning_rate": 6.812269608144128e-08,
+      "loss": 0.6291,
+      "step": 58286
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.354435920715332,
+      "learning_rate": 6.807932860336764e-08,
+      "loss": 0.5625,
+      "step": 58287
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 4.588364601135254,
+      "learning_rate": 6.803597483911794e-08,
+      "loss": 0.7033,
+      "step": 58288
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.5730271339416504,
+      "learning_rate": 6.799263478881213e-08,
+      "loss": 0.464,
+      "step": 58289
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.769824504852295,
+      "learning_rate": 6.794930845257176e-08,
+      "loss": 0.4187,
+      "step": 58290
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.632207155227661,
+      "learning_rate": 6.790599583051616e-08,
+      "loss": 0.8439,
+      "step": 58291
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 4.001253128051758,
+      "learning_rate": 6.786269692276693e-08,
+      "loss": 0.3586,
+      "step": 58292
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.9898414611816406,
+      "learning_rate": 6.781941172944396e-08,
+      "loss": 0.3485,
+      "step": 58293
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.4744925498962402,
+      "learning_rate": 6.77761402506677e-08,
+      "loss": 0.55,
+      "step": 58294
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.9733505249023438,
+      "learning_rate": 6.773288248655752e-08,
+      "loss": 0.5286,
+      "step": 58295
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.7511186599731445,
+      "learning_rate": 6.768963843723553e-08,
+      "loss": 0.6743,
+      "step": 58296
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.358631134033203,
+      "learning_rate": 6.764640810281997e-08,
+      "loss": 0.609,
+      "step": 58297
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.9183685779571533,
+      "learning_rate": 6.760319148343297e-08,
+      "loss": 0.615,
+      "step": 58298
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 5.530845642089844,
+      "learning_rate": 6.755998857919277e-08,
+      "loss": 0.9489,
+      "step": 58299
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.4776697158813477,
+      "learning_rate": 6.751679939022094e-08,
+      "loss": 0.6843,
+      "step": 58300
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.760856866836548,
+      "learning_rate": 6.747362391663737e-08,
+      "loss": 0.4734,
+      "step": 58301
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.1220526695251465,
+      "learning_rate": 6.743046215856197e-08,
+      "loss": 0.4918,
+      "step": 58302
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.735222339630127,
+      "learning_rate": 6.73873141161141e-08,
+      "loss": 0.6739,
+      "step": 58303
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 8.27047348022461,
+      "learning_rate": 6.734417978941532e-08,
+      "loss": 0.4589,
+      "step": 58304
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.6643502712249756,
+      "learning_rate": 6.730105917858332e-08,
+      "loss": 0.4667,
+      "step": 58305
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.9284067153930664,
+      "learning_rate": 6.725795228374021e-08,
+      "loss": 0.7103,
+      "step": 58306
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.952265977859497,
+      "learning_rate": 6.721485910500425e-08,
+      "loss": 0.4183,
+      "step": 58307
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 4.639678478240967,
+      "learning_rate": 6.717177964249644e-08,
+      "loss": 0.4616,
+      "step": 58308
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 4.223237037658691,
+      "learning_rate": 6.712871389633557e-08,
+      "loss": 0.5325,
+      "step": 58309
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.2762842178344727,
+      "learning_rate": 6.708566186664267e-08,
+      "loss": 0.5164,
+      "step": 58310
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.0416274070739746,
+      "learning_rate": 6.704262355353542e-08,
+      "loss": 0.4797,
+      "step": 58311
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.1012635231018066,
+      "learning_rate": 6.699959895713593e-08,
+      "loss": 0.6564,
+      "step": 58312
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.6131505966186523,
+      "learning_rate": 6.695658807756189e-08,
+      "loss": 0.7557,
+      "step": 58313
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.630310535430908,
+      "learning_rate": 6.691359091493376e-08,
+      "loss": 0.6038,
+      "step": 58314
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.970280885696411,
+      "learning_rate": 6.68706074693709e-08,
+      "loss": 0.7228,
+      "step": 58315
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.060675621032715,
+      "learning_rate": 6.682763774099266e-08,
+      "loss": 0.5324,
+      "step": 58316
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 4.660686492919922,
+      "learning_rate": 6.678468172991892e-08,
+      "loss": 0.5618,
+      "step": 58317
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 4.779169082641602,
+      "learning_rate": 6.674173943626961e-08,
+      "loss": 0.7982,
+      "step": 58318
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 4.416950225830078,
+      "learning_rate": 6.669881086016239e-08,
+      "loss": 0.5701,
+      "step": 58319
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 5.2654314041137695,
+      "learning_rate": 6.665589600171829e-08,
+      "loss": 0.5421,
+      "step": 58320
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.4937241077423096,
+      "learning_rate": 6.661299486105555e-08,
+      "loss": 0.7924,
+      "step": 58321
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.023616313934326,
+      "learning_rate": 6.657010743829462e-08,
+      "loss": 0.824,
+      "step": 58322
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.7144436836242676,
+      "learning_rate": 6.65272337335543e-08,
+      "loss": 0.3989,
+      "step": 58323
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.543670177459717,
+      "learning_rate": 6.648437374695338e-08,
+      "loss": 0.6514,
+      "step": 58324
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.2101190090179443,
+      "learning_rate": 6.644152747861066e-08,
+      "loss": 0.451,
+      "step": 58325
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.8475770950317383,
+      "learning_rate": 6.63986949286466e-08,
+      "loss": 0.5361,
+      "step": 58326
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.042005777359009,
+      "learning_rate": 6.635587609717942e-08,
+      "loss": 0.6866,
+      "step": 58327
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.3625009059906006,
+      "learning_rate": 6.631307098432849e-08,
+      "loss": 0.4279,
+      "step": 58328
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.2858963012695312,
+      "learning_rate": 6.627027959021259e-08,
+      "loss": 0.5246,
+      "step": 58329
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.960167169570923,
+      "learning_rate": 6.622750191495109e-08,
+      "loss": 0.6692,
+      "step": 58330
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.4255645275115967,
+      "learning_rate": 6.618473795866221e-08,
+      "loss": 0.5122,
+      "step": 58331
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.675445318222046,
+      "learning_rate": 6.614198772146529e-08,
+      "loss": 0.6301,
+      "step": 58332
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.255645751953125,
+      "learning_rate": 6.60992512034797e-08,
+      "loss": 0.4317,
+      "step": 58333
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.3531243801116943,
+      "learning_rate": 6.605652840482369e-08,
+      "loss": 0.415,
+      "step": 58334
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.892019510269165,
+      "learning_rate": 6.601381932561601e-08,
+      "loss": 0.628,
+      "step": 58335
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.7392578125,
+      "learning_rate": 6.59711239659755e-08,
+      "loss": 0.5481,
+      "step": 58336
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.899768352508545,
+      "learning_rate": 6.592844232602091e-08,
+      "loss": 0.5665,
+      "step": 58337
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.672196626663208,
+      "learning_rate": 6.588577440587163e-08,
+      "loss": 0.7346,
+      "step": 58338
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.4860942363739014,
+      "learning_rate": 6.584312020564476e-08,
+      "loss": 0.586,
+      "step": 58339
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.311631679534912,
+      "learning_rate": 6.580047972546022e-08,
+      "loss": 0.5346,
+      "step": 58340
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 5.019178867340088,
+      "learning_rate": 6.575785296543569e-08,
+      "loss": 0.6644,
+      "step": 58341
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.3580470085144043,
+      "learning_rate": 6.571523992569107e-08,
+      "loss": 0.4798,
+      "step": 58342
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.0547735691070557,
+      "learning_rate": 6.567264060634348e-08,
+      "loss": 0.547,
+      "step": 58343
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.994389057159424,
+      "learning_rate": 6.563005500751174e-08,
+      "loss": 0.6133,
+      "step": 58344
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.696833372116089,
+      "learning_rate": 6.558748312931462e-08,
+      "loss": 0.6624,
+      "step": 58345
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.198613166809082,
+      "learning_rate": 6.554492497186981e-08,
+      "loss": 0.4869,
+      "step": 58346
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 4.405813217163086,
+      "learning_rate": 6.550238053529612e-08,
+      "loss": 0.5494,
+      "step": 58347
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.3200249671936035,
+      "learning_rate": 6.545984981971232e-08,
+      "loss": 0.7169,
+      "step": 58348
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.450974702835083,
+      "learning_rate": 6.54173328252361e-08,
+      "loss": 0.6274,
+      "step": 58349
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.793393850326538,
+      "learning_rate": 6.537482955198516e-08,
+      "loss": 0.539,
+      "step": 58350
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.119497060775757,
+      "learning_rate": 6.533234000007882e-08,
+      "loss": 0.6619,
+      "step": 58351
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.671841621398926,
+      "learning_rate": 6.528986416963424e-08,
+      "loss": 0.8034,
+      "step": 58352
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.8506085872650146,
+      "learning_rate": 6.524740206077018e-08,
+      "loss": 0.7008,
+      "step": 58353
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.3217499256134033,
+      "learning_rate": 6.520495367360435e-08,
+      "loss": 0.6529,
+      "step": 58354
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.336754560470581,
+      "learning_rate": 6.516251900825554e-08,
+      "loss": 0.3728,
+      "step": 58355
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.25767183303833,
+      "learning_rate": 6.512009806484032e-08,
+      "loss": 0.3575,
+      "step": 58356
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.8172571659088135,
+      "learning_rate": 6.507769084347804e-08,
+      "loss": 0.6804,
+      "step": 58357
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.2775440216064453,
+      "learning_rate": 6.503529734428582e-08,
+      "loss": 0.8991,
+      "step": 58358
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.700178623199463,
+      "learning_rate": 6.49929175673819e-08,
+      "loss": 0.5727,
+      "step": 58359
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.7970235347747803,
+      "learning_rate": 6.495055151288344e-08,
+      "loss": 0.656,
+      "step": 58360
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.7692739963531494,
+      "learning_rate": 6.490819918090919e-08,
+      "loss": 0.5271,
+      "step": 58361
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 6.932567596435547,
+      "learning_rate": 6.486586057157629e-08,
+      "loss": 0.5094,
+      "step": 58362
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.4817140102386475,
+      "learning_rate": 6.482353568500299e-08,
+      "loss": 0.4557,
+      "step": 58363
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.497178077697754,
+      "learning_rate": 6.478122452130642e-08,
+      "loss": 0.5388,
+      "step": 58364
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.0044169425964355,
+      "learning_rate": 6.47389270806048e-08,
+      "loss": 0.6036,
+      "step": 58365
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.2606515884399414,
+      "learning_rate": 6.469664336301528e-08,
+      "loss": 0.2416,
+      "step": 58366
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.800471782684326,
+      "learning_rate": 6.465437336865555e-08,
+      "loss": 0.5757,
+      "step": 58367
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.423543930053711,
+      "learning_rate": 6.461211709764326e-08,
+      "loss": 0.502,
+      "step": 58368
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.6482276916503906,
+      "learning_rate": 6.45698745500961e-08,
+      "loss": 0.6298,
+      "step": 58369
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.850008487701416,
+      "learning_rate": 6.452764572613068e-08,
+      "loss": 0.8304,
+      "step": 58370
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.826394557952881,
+      "learning_rate": 6.44854306258652e-08,
+      "loss": 0.7405,
+      "step": 58371
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.648681402206421,
+      "learning_rate": 6.444322924941682e-08,
+      "loss": 0.5085,
+      "step": 58372
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.483612298965454,
+      "learning_rate": 6.440104159690319e-08,
+      "loss": 0.5773,
+      "step": 58373
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 6.285324573516846,
+      "learning_rate": 6.435886766844146e-08,
+      "loss": 0.8008,
+      "step": 58374
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 7.215214729309082,
+      "learning_rate": 6.431670746414875e-08,
+      "loss": 0.5621,
+      "step": 58375
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.0010640621185303,
+      "learning_rate": 6.427456098414164e-08,
+      "loss": 0.5154,
+      "step": 58376
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.838557481765747,
+      "learning_rate": 6.423242822853892e-08,
+      "loss": 0.651,
+      "step": 58377
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.9508752822875977,
+      "learning_rate": 6.419030919745606e-08,
+      "loss": 0.5457,
+      "step": 58378
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 4.051480293273926,
+      "learning_rate": 6.414820389101129e-08,
+      "loss": 0.7443,
+      "step": 58379
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.2005531787872314,
+      "learning_rate": 6.410611230932117e-08,
+      "loss": 0.4832,
+      "step": 58380
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.453052043914795,
+      "learning_rate": 6.406403445250287e-08,
+      "loss": 0.6099,
+      "step": 58381
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.9079701900482178,
+      "learning_rate": 6.402197032067348e-08,
+      "loss": 0.4999,
+      "step": 58382
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.5249075889587402,
+      "learning_rate": 6.39799199139507e-08,
+      "loss": 0.4251,
+      "step": 58383
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.5976476669311523,
+      "learning_rate": 6.393788323244998e-08,
+      "loss": 0.5913,
+      "step": 58384
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.8020966053009033,
+      "learning_rate": 6.389586027628902e-08,
+      "loss": 0.5004,
+      "step": 58385
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 4.920810222625732,
+      "learning_rate": 6.385385104558439e-08,
+      "loss": 0.4886,
+      "step": 58386
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 4.2633466720581055,
+      "learning_rate": 6.381185554045377e-08,
+      "loss": 0.5919,
+      "step": 58387
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.8782973289489746,
+      "learning_rate": 6.376987376101206e-08,
+      "loss": 0.5557,
+      "step": 58388
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.0980522632598877,
+      "learning_rate": 6.372790570737809e-08,
+      "loss": 0.7139,
+      "step": 58389
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.820192575454712,
+      "learning_rate": 6.368595137966727e-08,
+      "loss": 0.4955,
+      "step": 58390
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.3037936687469482,
+      "learning_rate": 6.364401077799731e-08,
+      "loss": 0.9158,
+      "step": 58391
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.6520566940307617,
+      "learning_rate": 6.360208390248312e-08,
+      "loss": 0.3436,
+      "step": 58392
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.4780259132385254,
+      "learning_rate": 6.35601707532435e-08,
+      "loss": 0.7161,
+      "step": 58393
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.346555233001709,
+      "learning_rate": 6.351827133039279e-08,
+      "loss": 0.4029,
+      "step": 58394
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 11.390917778015137,
+      "learning_rate": 6.347638563404922e-08,
+      "loss": 0.5492,
+      "step": 58395
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.987212896347046,
+      "learning_rate": 6.343451366432773e-08,
+      "loss": 0.52,
+      "step": 58396
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.4356367588043213,
+      "learning_rate": 6.339265542134654e-08,
+      "loss": 0.6273,
+      "step": 58397
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.3670923709869385,
+      "learning_rate": 6.335081090521999e-08,
+      "loss": 0.2708,
+      "step": 58398
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 4.238781452178955,
+      "learning_rate": 6.330898011606634e-08,
+      "loss": 0.7651,
+      "step": 58399
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.220547914505005,
+      "learning_rate": 6.326716305400105e-08,
+      "loss": 0.6599,
+      "step": 58400
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 4.000099182128906,
+      "learning_rate": 6.322535971914012e-08,
+      "loss": 0.5542,
+      "step": 58401
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.3024497032165527,
+      "learning_rate": 6.318357011160014e-08,
+      "loss": 0.5824,
+      "step": 58402
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.033578395843506,
+      "learning_rate": 6.314179423149714e-08,
+      "loss": 0.3903,
+      "step": 58403
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 4.154600620269775,
+      "learning_rate": 6.310003207894767e-08,
+      "loss": 0.5191,
+      "step": 58404
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.6277782917022705,
+      "learning_rate": 6.305828365406774e-08,
+      "loss": 0.528,
+      "step": 58405
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.905048370361328,
+      "learning_rate": 6.301654895697285e-08,
+      "loss": 0.9431,
+      "step": 58406
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 6.7720842361450195,
+      "learning_rate": 6.29748279877801e-08,
+      "loss": 0.5333,
+      "step": 58407
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.592834234237671,
+      "learning_rate": 6.29331207466044e-08,
+      "loss": 0.5217,
+      "step": 58408
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.7520577907562256,
+      "learning_rate": 6.28914272335629e-08,
+      "loss": 0.561,
+      "step": 58409
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.393825054168701,
+      "learning_rate": 6.284974744876993e-08,
+      "loss": 0.6235,
+      "step": 58410
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.669858694076538,
+      "learning_rate": 6.280808139234318e-08,
+      "loss": 0.5931,
+      "step": 58411
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.601921796798706,
+      "learning_rate": 6.2766429064397e-08,
+      "loss": 0.5225,
+      "step": 58412
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 6.924752235412598,
+      "learning_rate": 6.272479046504797e-08,
+      "loss": 0.7532,
+      "step": 58413
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.231170654296875,
+      "learning_rate": 6.268316559441212e-08,
+      "loss": 0.4572,
+      "step": 58414
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.2770333290100098,
+      "learning_rate": 6.264155445260434e-08,
+      "loss": 0.7022,
+      "step": 58415
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.6507396697998047,
+      "learning_rate": 6.25999570397412e-08,
+      "loss": 0.5949,
+      "step": 58416
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.680366277694702,
+      "learning_rate": 6.255837335593762e-08,
+      "loss": 0.7409,
+      "step": 58417
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.6977267265319824,
+      "learning_rate": 6.251680340131017e-08,
+      "loss": 0.4577,
+      "step": 58418
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 1.8498280048370361,
+      "learning_rate": 6.24752471759732e-08,
+      "loss": 0.4066,
+      "step": 58419
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 9.19238567352295,
+      "learning_rate": 6.243370468004384e-08,
+      "loss": 0.5088,
+      "step": 58420
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.48268985748291,
+      "learning_rate": 6.239217591363588e-08,
+      "loss": 0.7343,
+      "step": 58421
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.2181832790374756,
+      "learning_rate": 6.235066087686537e-08,
+      "loss": 0.5206,
+      "step": 58422
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 1.8385512828826904,
+      "learning_rate": 6.230915956984884e-08,
+      "loss": 0.3546,
+      "step": 58423
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.268336057662964,
+      "learning_rate": 6.226767199270012e-08,
+      "loss": 0.6595,
+      "step": 58424
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.787109851837158,
+      "learning_rate": 6.222619814553577e-08,
+      "loss": 0.3805,
+      "step": 58425
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.3620641231536865,
+      "learning_rate": 6.21847380284707e-08,
+      "loss": 0.5989,
+      "step": 58426
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.673724412918091,
+      "learning_rate": 6.214329164161925e-08,
+      "loss": 0.5245,
+      "step": 58427
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.8375720977783203,
+      "learning_rate": 6.210185898509857e-08,
+      "loss": 0.4505,
+      "step": 58428
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.519592523574829,
+      "learning_rate": 6.206044005902246e-08,
+      "loss": 0.5703,
+      "step": 58429
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.4237782955169678,
+      "learning_rate": 6.201903486350636e-08,
+      "loss": 0.4727,
+      "step": 58430
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.5752665996551514,
+      "learning_rate": 6.197764339866575e-08,
+      "loss": 0.5608,
+      "step": 58431
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.6668574810028076,
+      "learning_rate": 6.193626566461552e-08,
+      "loss": 0.6693,
+      "step": 58432
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.2665562629699707,
+      "learning_rate": 6.189490166147005e-08,
+      "loss": 0.4443,
+      "step": 58433
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.0501205921173096,
+      "learning_rate": 6.18535513893459e-08,
+      "loss": 0.431,
+      "step": 58434
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.450178384780884,
+      "learning_rate": 6.181221484835687e-08,
+      "loss": 0.6705,
+      "step": 58435
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.0911784172058105,
+      "learning_rate": 6.177089203861786e-08,
+      "loss": 0.618,
+      "step": 58436
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.3481814861297607,
+      "learning_rate": 6.172958296024434e-08,
+      "loss": 0.461,
+      "step": 58437
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 4.657694339752197,
+      "learning_rate": 6.168828761335122e-08,
+      "loss": 0.7736,
+      "step": 58438
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.6919801235198975,
+      "learning_rate": 6.164700599805229e-08,
+      "loss": 0.7918,
+      "step": 58439
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 4.477050304412842,
+      "learning_rate": 6.160573811446358e-08,
+      "loss": 0.5399,
+      "step": 58440
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.1147243976593018,
+      "learning_rate": 6.156448396269943e-08,
+      "loss": 0.46,
+      "step": 58441
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.4343857765197754,
+      "learning_rate": 6.15232435428742e-08,
+      "loss": 0.3587,
+      "step": 58442
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.5662450790405273,
+      "learning_rate": 6.14820168551028e-08,
+      "loss": 0.4753,
+      "step": 58443
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.274519920349121,
+      "learning_rate": 6.144080389950014e-08,
+      "loss": 0.8293,
+      "step": 58444
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 4.328380107879639,
+      "learning_rate": 6.139960467618e-08,
+      "loss": 0.4181,
+      "step": 58445
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 4.443139553070068,
+      "learning_rate": 6.135841918525842e-08,
+      "loss": 0.8887,
+      "step": 58446
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.678126811981201,
+      "learning_rate": 6.131724742684808e-08,
+      "loss": 0.2942,
+      "step": 58447
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.6841516494750977,
+      "learning_rate": 6.127608940106555e-08,
+      "loss": 0.5187,
+      "step": 58448
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.704329013824463,
+      "learning_rate": 6.123494510802296e-08,
+      "loss": 0.5714,
+      "step": 58449
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.61698842048645,
+      "learning_rate": 6.11938145478369e-08,
+      "loss": 0.5179,
+      "step": 58450
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.942143201828003,
+      "learning_rate": 6.115269772062005e-08,
+      "loss": 0.553,
+      "step": 58451
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.289945363998413,
+      "learning_rate": 6.111159462648786e-08,
+      "loss": 0.5614,
+      "step": 58452
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.918968915939331,
+      "learning_rate": 6.107050526555359e-08,
+      "loss": 0.3902,
+      "step": 58453
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.8169944286346436,
+      "learning_rate": 6.102942963793268e-08,
+      "loss": 0.5813,
+      "step": 58454
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 4.169754505157471,
+      "learning_rate": 6.09883677437384e-08,
+      "loss": 0.8416,
+      "step": 58455
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.106584072113037,
+      "learning_rate": 6.094731958308508e-08,
+      "loss": 0.4429,
+      "step": 58456
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 5.543837547302246,
+      "learning_rate": 6.090628515608765e-08,
+      "loss": 0.3939,
+      "step": 58457
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.3066461086273193,
+      "learning_rate": 6.086526446285934e-08,
+      "loss": 0.5458,
+      "step": 58458
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.7508914470672607,
+      "learning_rate": 6.082425750351395e-08,
+      "loss": 0.4191,
+      "step": 58459
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.048560619354248,
+      "learning_rate": 6.078326427816694e-08,
+      "loss": 0.5862,
+      "step": 58460
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 4.968349933624268,
+      "learning_rate": 6.074228478693101e-08,
+      "loss": 0.383,
+      "step": 58461
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.195422887802124,
+      "learning_rate": 6.070131902991993e-08,
+      "loss": 0.5658,
+      "step": 58462
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.8040976524353027,
+      "learning_rate": 6.066036700724864e-08,
+      "loss": 0.7276,
+      "step": 58463
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.4141039848327637,
+      "learning_rate": 6.061942871903093e-08,
+      "loss": 0.5284,
+      "step": 58464
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.3025877475738525,
+      "learning_rate": 6.057850416538002e-08,
+      "loss": 0.5842,
+      "step": 58465
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.5743749141693115,
+      "learning_rate": 6.053759334640974e-08,
+      "loss": 0.7448,
+      "step": 58466
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.459393262863159,
+      "learning_rate": 6.049669626223387e-08,
+      "loss": 0.57,
+      "step": 58467
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.1544041633605957,
+      "learning_rate": 6.045581291296676e-08,
+      "loss": 0.5486,
+      "step": 58468
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.640815496444702,
+      "learning_rate": 6.041494329872166e-08,
+      "loss": 0.6141,
+      "step": 58469
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.1702799797058105,
+      "learning_rate": 6.037408741961237e-08,
+      "loss": 0.4803,
+      "step": 58470
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.5501129627227783,
+      "learning_rate": 6.033324527575157e-08,
+      "loss": 0.6514,
+      "step": 58471
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.0186104774475098,
+      "learning_rate": 6.029241686725417e-08,
+      "loss": 0.4533,
+      "step": 58472
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.3539164066314697,
+      "learning_rate": 6.025160219423287e-08,
+      "loss": 0.4831,
+      "step": 58473
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.040665864944458,
+      "learning_rate": 6.021080125680145e-08,
+      "loss": 0.5491,
+      "step": 58474
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 4.559281349182129,
+      "learning_rate": 6.017001405507317e-08,
+      "loss": 0.6029,
+      "step": 58475
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.058178663253784,
+      "learning_rate": 6.01292405891618e-08,
+      "loss": 0.5389,
+      "step": 58476
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.138361692428589,
+      "learning_rate": 6.008848085918063e-08,
+      "loss": 0.5501,
+      "step": 58477
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.195650577545166,
+      "learning_rate": 6.004773486524229e-08,
+      "loss": 0.474,
+      "step": 58478
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.467785120010376,
+      "learning_rate": 6.000700260746118e-08,
+      "loss": 0.4954,
+      "step": 58479
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.2591288089752197,
+      "learning_rate": 5.99662840859494e-08,
+      "loss": 0.5439,
+      "step": 58480
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 4.1134772300720215,
+      "learning_rate": 5.992557930082133e-08,
+      "loss": 0.6362,
+      "step": 58481
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.617063522338867,
+      "learning_rate": 5.988488825218963e-08,
+      "loss": 0.6555,
+      "step": 58482
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.558023691177368,
+      "learning_rate": 5.984421094016701e-08,
+      "loss": 0.7226,
+      "step": 58483
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.5803961753845215,
+      "learning_rate": 5.980354736486781e-08,
+      "loss": 0.556,
+      "step": 58484
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.3241000175476074,
+      "learning_rate": 5.976289752640363e-08,
+      "loss": 0.5924,
+      "step": 58485
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 4.323370456695557,
+      "learning_rate": 5.972226142488823e-08,
+      "loss": 0.6497,
+      "step": 58486
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.5703465938568115,
+      "learning_rate": 5.968163906043489e-08,
+      "loss": 0.7288,
+      "step": 58487
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 5.471929550170898,
+      "learning_rate": 5.964103043315572e-08,
+      "loss": 0.7438,
+      "step": 58488
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.6711723804473877,
+      "learning_rate": 5.960043554316508e-08,
+      "loss": 0.557,
+      "step": 58489
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.26912784576416,
+      "learning_rate": 5.955985439057399e-08,
+      "loss": 0.5287,
+      "step": 58490
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 5.345161437988281,
+      "learning_rate": 5.9519286975496807e-08,
+      "loss": 0.6135,
+      "step": 58491
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.9161975383758545,
+      "learning_rate": 5.9478733298045096e-08,
+      "loss": 0.4397,
+      "step": 58492
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.9686062335968018,
+      "learning_rate": 5.9438193358332674e-08,
+      "loss": 0.4944,
+      "step": 58493
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.0888898372650146,
+      "learning_rate": 5.939766715647166e-08,
+      "loss": 0.2596,
+      "step": 58494
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.61433744430542,
+      "learning_rate": 5.93571546925753e-08,
+      "loss": 0.5545,
+      "step": 58495
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.3057539463043213,
+      "learning_rate": 5.9316655966755734e-08,
+      "loss": 0.478,
+      "step": 58496
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.0114634037017822,
+      "learning_rate": 5.927617097912563e-08,
+      "loss": 0.5347,
+      "step": 58497
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.8688888549804688,
+      "learning_rate": 5.92356997297977e-08,
+      "loss": 0.6015,
+      "step": 58498
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.3008530139923096,
+      "learning_rate": 5.9195242218884616e-08,
+      "loss": 0.5548,
+      "step": 58499
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.6325390338897705,
+      "learning_rate": 5.915479844649852e-08,
+      "loss": 0.5845,
+      "step": 58500
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 4.079750061035156,
+      "learning_rate": 5.911436841275209e-08,
+      "loss": 0.7151,
+      "step": 58501
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 4.036159992218018,
+      "learning_rate": 5.907395211775746e-08,
+      "loss": 0.4874,
+      "step": 58502
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.7065606117248535,
+      "learning_rate": 5.903354956162732e-08,
+      "loss": 0.5879,
+      "step": 58503
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.663102865219116,
+      "learning_rate": 5.8993160744474366e-08,
+      "loss": 0.7536,
+      "step": 58504
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.049757719039917,
+      "learning_rate": 5.895278566641016e-08,
+      "loss": 0.4557,
+      "step": 58505
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.324357271194458,
+      "learning_rate": 5.8912424327547404e-08,
+      "loss": 0.5629,
+      "step": 58506
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.721923828125,
+      "learning_rate": 5.887207672799822e-08,
+      "loss": 0.7204,
+      "step": 58507
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.5759453773498535,
+      "learning_rate": 5.883174286787474e-08,
+      "loss": 0.6112,
+      "step": 58508
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.4812214374542236,
+      "learning_rate": 5.879142274728911e-08,
+      "loss": 0.6572,
+      "step": 58509
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.776803970336914,
+      "learning_rate": 5.875111636635345e-08,
+      "loss": 0.6254,
+      "step": 58510
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.293010950088501,
+      "learning_rate": 5.8710823725180444e-08,
+      "loss": 0.4019,
+      "step": 58511
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.832617998123169,
+      "learning_rate": 5.8670544823881125e-08,
+      "loss": 0.5663,
+      "step": 58512
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.6626389026641846,
+      "learning_rate": 5.863027966256818e-08,
+      "loss": 0.6374,
+      "step": 58513
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 6.179531574249268,
+      "learning_rate": 5.859002824135263e-08,
+      "loss": 0.5556,
+      "step": 58514
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.3138623237609863,
+      "learning_rate": 5.8549790560348264e-08,
+      "loss": 0.5706,
+      "step": 58515
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.9477193355560303,
+      "learning_rate": 5.8509566619665005e-08,
+      "loss": 0.5714,
+      "step": 58516
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 4.331811904907227,
+      "learning_rate": 5.846935641941554e-08,
+      "loss": 0.4483,
+      "step": 58517
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.622783660888672,
+      "learning_rate": 5.842915995971199e-08,
+      "loss": 0.536,
+      "step": 58518
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.4277091026306152,
+      "learning_rate": 5.8388977240665946e-08,
+      "loss": 0.468,
+      "step": 58519
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 1.8651776313781738,
+      "learning_rate": 5.834880826238842e-08,
+      "loss": 0.4426,
+      "step": 58520
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.158848285675049,
+      "learning_rate": 5.830865302499211e-08,
+      "loss": 0.4509,
+      "step": 58521
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.560354232788086,
+      "learning_rate": 5.8268511528588036e-08,
+      "loss": 0.504,
+      "step": 58522
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.8488287925720215,
+      "learning_rate": 5.822838377328777e-08,
+      "loss": 0.5639,
+      "step": 58523
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.929333209991455,
+      "learning_rate": 5.818826975920344e-08,
+      "loss": 0.5594,
+      "step": 58524
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.532139301300049,
+      "learning_rate": 5.814816948644608e-08,
+      "loss": 0.3617,
+      "step": 58525
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.8669772148132324,
+      "learning_rate": 5.810808295512782e-08,
+      "loss": 0.8054,
+      "step": 58526
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 4.211967468261719,
+      "learning_rate": 5.8068010165359675e-08,
+      "loss": 0.6034,
+      "step": 58527
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.067126989364624,
+      "learning_rate": 5.8027951117252676e-08,
+      "loss": 0.6923,
+      "step": 58528
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.534351110458374,
+      "learning_rate": 5.798790581091895e-08,
+      "loss": 0.6989,
+      "step": 58529
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.532400131225586,
+      "learning_rate": 5.7947874246469525e-08,
+      "loss": 0.4243,
+      "step": 58530
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.564892292022705,
+      "learning_rate": 5.790785642401598e-08,
+      "loss": 0.5614,
+      "step": 58531
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.386350393295288,
+      "learning_rate": 5.786785234366876e-08,
+      "loss": 0.5285,
+      "step": 58532
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.612976551055908,
+      "learning_rate": 5.782786200554003e-08,
+      "loss": 0.6116,
+      "step": 58533
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.270047187805176,
+      "learning_rate": 5.778788540974023e-08,
+      "loss": 0.5564,
+      "step": 58534
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 4.266953945159912,
+      "learning_rate": 5.774792255638151e-08,
+      "loss": 0.8027,
+      "step": 58535
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.3925163745880127,
+      "learning_rate": 5.770797344557377e-08,
+      "loss": 0.6283,
+      "step": 58536
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.729973077774048,
+      "learning_rate": 5.766803807742916e-08,
+      "loss": 0.5008,
+      "step": 58537
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.956345558166504,
+      "learning_rate": 5.762811645205868e-08,
+      "loss": 0.5612,
+      "step": 58538
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.5134639739990234,
+      "learning_rate": 5.7588208569572257e-08,
+      "loss": 0.7012,
+      "step": 58539
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.1449520587921143,
+      "learning_rate": 5.754831443008202e-08,
+      "loss": 0.5489,
+      "step": 58540
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.5857486724853516,
+      "learning_rate": 5.750843403369788e-08,
+      "loss": 0.5752,
+      "step": 58541
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.5375192165374756,
+      "learning_rate": 5.746856738053197e-08,
+      "loss": 0.5016,
+      "step": 58542
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.5171701908111572,
+      "learning_rate": 5.742871447069365e-08,
+      "loss": 0.6615,
+      "step": 58543
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.587782859802246,
+      "learning_rate": 5.7388875304295046e-08,
+      "loss": 0.6088,
+      "step": 58544
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.8286240100860596,
+      "learning_rate": 5.734904988144663e-08,
+      "loss": 0.6933,
+      "step": 58545
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.101822853088379,
+      "learning_rate": 5.730923820225831e-08,
+      "loss": 0.5132,
+      "step": 58546
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.1471588611602783,
+      "learning_rate": 5.726944026684167e-08,
+      "loss": 0.5782,
+      "step": 58547
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.403787612915039,
+      "learning_rate": 5.722965607530773e-08,
+      "loss": 0.4491,
+      "step": 58548
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.570362091064453,
+      "learning_rate": 5.718988562776584e-08,
+      "loss": 0.7295,
+      "step": 58549
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.066408157348633,
+      "learning_rate": 5.715012892432759e-08,
+      "loss": 0.4832,
+      "step": 58550
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.6036531925201416,
+      "learning_rate": 5.7110385965102875e-08,
+      "loss": 0.3379,
+      "step": 58551
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 5.270400047302246,
+      "learning_rate": 5.707065675020329e-08,
+      "loss": 0.4513,
+      "step": 58552
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.9822049140930176,
+      "learning_rate": 5.703094127973763e-08,
+      "loss": 0.6229,
+      "step": 58553
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 4.21885347366333,
+      "learning_rate": 5.6991239553818024e-08,
+      "loss": 0.5158,
+      "step": 58554
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.611313581466675,
+      "learning_rate": 5.6951551572553276e-08,
+      "loss": 0.5455,
+      "step": 58555
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 4.376176357269287,
+      "learning_rate": 5.691187733605552e-08,
+      "loss": 0.6886,
+      "step": 58556
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 4.187076091766357,
+      "learning_rate": 5.6872216844433003e-08,
+      "loss": 0.5876,
+      "step": 58557
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.847639799118042,
+      "learning_rate": 5.6832570097797854e-08,
+      "loss": 0.6683,
+      "step": 58558
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.958813428878784,
+      "learning_rate": 5.679293709625944e-08,
+      "loss": 0.6629,
+      "step": 58559
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.075525999069214,
+      "learning_rate": 5.675331783992821e-08,
+      "loss": 0.6106,
+      "step": 58560
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.8482284545898438,
+      "learning_rate": 5.671371232891409e-08,
+      "loss": 0.5768,
+      "step": 58561
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.532259464263916,
+      "learning_rate": 5.6674120563327554e-08,
+      "loss": 0.498,
+      "step": 58562
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 4.141707897186279,
+      "learning_rate": 5.6634542543277936e-08,
+      "loss": 0.6193,
+      "step": 58563
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.2574963569641113,
+      "learning_rate": 5.6594978268875724e-08,
+      "loss": 0.4898,
+      "step": 58564
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.959472417831421,
+      "learning_rate": 5.655542774023137e-08,
+      "loss": 0.6516,
+      "step": 58565
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 1.8623093366622925,
+      "learning_rate": 5.65158909574548e-08,
+      "loss": 0.517,
+      "step": 58566
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.7422003746032715,
+      "learning_rate": 5.647636792065536e-08,
+      "loss": 0.533,
+      "step": 58567
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.7118794918060303,
+      "learning_rate": 5.6436858629943524e-08,
+      "loss": 0.6433,
+      "step": 58568
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 4.748680114746094,
+      "learning_rate": 5.6397363085428645e-08,
+      "loss": 0.5476,
+      "step": 58569
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.2238070964813232,
+      "learning_rate": 5.635788128722064e-08,
+      "loss": 0.6447,
+      "step": 58570
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.4853217601776123,
+      "learning_rate": 5.631841323542942e-08,
+      "loss": 0.9581,
+      "step": 58571
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 4.316132545471191,
+      "learning_rate": 5.6278958930165436e-08,
+      "loss": 0.7413,
+      "step": 58572
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.868206739425659,
+      "learning_rate": 5.623951837153696e-08,
+      "loss": 0.686,
+      "step": 58573
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.475844383239746,
+      "learning_rate": 5.6200091559655e-08,
+      "loss": 0.5475,
+      "step": 58574
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.111121416091919,
+      "learning_rate": 5.616067849462781e-08,
+      "loss": 0.5336,
+      "step": 58575
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.0490028858184814,
+      "learning_rate": 5.61212791765664e-08,
+      "loss": 0.4905,
+      "step": 58576
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.0394906997680664,
+      "learning_rate": 5.608189360557958e-08,
+      "loss": 0.6248,
+      "step": 58577
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 5.844193935394287,
+      "learning_rate": 5.604252178177727e-08,
+      "loss": 0.7953,
+      "step": 58578
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.2476649284362793,
+      "learning_rate": 5.6003163705268257e-08,
+      "loss": 0.5319,
+      "step": 58579
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.4222235679626465,
+      "learning_rate": 5.5963819376162464e-08,
+      "loss": 0.4286,
+      "step": 58580
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.770894765853882,
+      "learning_rate": 5.5924488794569244e-08,
+      "loss": 0.6767,
+      "step": 58581
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.0918734073638916,
+      "learning_rate": 5.588517196059795e-08,
+      "loss": 0.5021,
+      "step": 58582
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.850085973739624,
+      "learning_rate": 5.584586887435739e-08,
+      "loss": 0.6,
+      "step": 58583
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.0542197227478027,
+      "learning_rate": 5.580657953595747e-08,
+      "loss": 0.5907,
+      "step": 58584
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.572774887084961,
+      "learning_rate": 5.5767303945507555e-08,
+      "loss": 0.6229,
+      "step": 58585
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.568311929702759,
+      "learning_rate": 5.5728042103117e-08,
+      "loss": 0.6967,
+      "step": 58586
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.46319580078125,
+      "learning_rate": 5.568879400889404e-08,
+      "loss": 0.4512,
+      "step": 58587
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.275444269180298,
+      "learning_rate": 5.564955966294861e-08,
+      "loss": 0.5112,
+      "step": 58588
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.491257905960083,
+      "learning_rate": 5.5610339065389486e-08,
+      "loss": 0.5371,
+      "step": 58589
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.823781728744507,
+      "learning_rate": 5.557113221632604e-08,
+      "loss": 0.6037,
+      "step": 58590
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.834360122680664,
+      "learning_rate": 5.553193911586652e-08,
+      "loss": 0.4627,
+      "step": 58591
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.250549554824829,
+      "learning_rate": 5.5492759764121384e-08,
+      "loss": 0.3959,
+      "step": 58592
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.625861883163452,
+      "learning_rate": 5.545359416119778e-08,
+      "loss": 0.4788,
+      "step": 58593
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.2235727310180664,
+      "learning_rate": 5.5414442307205606e-08,
+      "loss": 0.4254,
+      "step": 58594
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.667003870010376,
+      "learning_rate": 5.5375304202253676e-08,
+      "loss": 0.6089,
+      "step": 58595
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.8648762702941895,
+      "learning_rate": 5.5336179846450785e-08,
+      "loss": 0.3566,
+      "step": 58596
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 5.468733787536621,
+      "learning_rate": 5.5297069239905745e-08,
+      "loss": 0.5848,
+      "step": 58597
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 7.786631107330322,
+      "learning_rate": 5.525797238272679e-08,
+      "loss": 0.649,
+      "step": 58598
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.0206141471862793,
+      "learning_rate": 5.521888927502328e-08,
+      "loss": 0.6621,
+      "step": 58599
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.466834783554077,
+      "learning_rate": 5.517981991690402e-08,
+      "loss": 0.5109,
+      "step": 58600
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.823540449142456,
+      "learning_rate": 5.514076430847726e-08,
+      "loss": 0.526,
+      "step": 58601
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 4.401412487030029,
+      "learning_rate": 5.5101722449851234e-08,
+      "loss": 0.5182,
+      "step": 58602
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.5607211589813232,
+      "learning_rate": 5.5062694341135316e-08,
+      "loss": 0.4158,
+      "step": 58603
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.1218695640563965,
+      "learning_rate": 5.502367998243718e-08,
+      "loss": 0.5751,
+      "step": 58604
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.940629720687866,
+      "learning_rate": 5.4984679373865645e-08,
+      "loss": 0.6816,
+      "step": 58605
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.239354133605957,
+      "learning_rate": 5.4945692515530055e-08,
+      "loss": 0.7175,
+      "step": 58606
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.2649574279785156,
+      "learning_rate": 5.490671940753756e-08,
+      "loss": 0.5901,
+      "step": 58607
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 5.440943717956543,
+      "learning_rate": 5.486776004999694e-08,
+      "loss": 0.6287,
+      "step": 58608
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 4.038450717926025,
+      "learning_rate": 5.482881444301702e-08,
+      "loss": 0.4604,
+      "step": 58609
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 4.818476676940918,
+      "learning_rate": 5.4789882586704926e-08,
+      "loss": 0.4787,
+      "step": 58610
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.5723838806152344,
+      "learning_rate": 5.4750964481170565e-08,
+      "loss": 0.5982,
+      "step": 58611
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.7331862449645996,
+      "learning_rate": 5.471206012652053e-08,
+      "loss": 0.5764,
+      "step": 58612
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.9010374546051025,
+      "learning_rate": 5.467316952286417e-08,
+      "loss": 0.6256,
+      "step": 58613
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 6.607819080352783,
+      "learning_rate": 5.463429267030862e-08,
+      "loss": 0.5107,
+      "step": 58614
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.024872064590454,
+      "learning_rate": 5.4595429568962686e-08,
+      "loss": 0.4302,
+      "step": 58615
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.943089723587036,
+      "learning_rate": 5.455658021893462e-08,
+      "loss": 0.6935,
+      "step": 58616
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.66274356842041,
+      "learning_rate": 5.45177446203321e-08,
+      "loss": 0.47,
+      "step": 58617
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.655773401260376,
+      "learning_rate": 5.447892277326283e-08,
+      "loss": 0.3921,
+      "step": 58618
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.6736106872558594,
+      "learning_rate": 5.444011467783505e-08,
+      "loss": 0.741,
+      "step": 58619
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.0438733100891113,
+      "learning_rate": 5.440132033415646e-08,
+      "loss": 0.5674,
+      "step": 58620
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.5322937965393066,
+      "learning_rate": 5.4362539742335854e-08,
+      "loss": 0.626,
+      "step": 58621
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.3655171394348145,
+      "learning_rate": 5.432377290247981e-08,
+      "loss": 0.4582,
+      "step": 58622
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.492725372314453,
+      "learning_rate": 5.428501981469658e-08,
+      "loss": 0.3987,
+      "step": 58623
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.8680357933044434,
+      "learning_rate": 5.424628047909441e-08,
+      "loss": 0.6343,
+      "step": 58624
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.5506744384765625,
+      "learning_rate": 5.4207554895779893e-08,
+      "loss": 0.4117,
+      "step": 58625
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.075150728225708,
+      "learning_rate": 5.4168843064861806e-08,
+      "loss": 0.4501,
+      "step": 58626
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.842139720916748,
+      "learning_rate": 5.413014498644786e-08,
+      "loss": 0.5763,
+      "step": 58627
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.6112887859344482,
+      "learning_rate": 5.4091460660644636e-08,
+      "loss": 0.4736,
+      "step": 58628
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 4.046167850494385,
+      "learning_rate": 5.4052790087560926e-08,
+      "loss": 0.6779,
+      "step": 58629
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 6.565440654754639,
+      "learning_rate": 5.401413326730276e-08,
+      "loss": 0.5719,
+      "step": 58630
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.472496509552002,
+      "learning_rate": 5.39754901999795e-08,
+      "loss": 0.3102,
+      "step": 58631
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.154458999633789,
+      "learning_rate": 5.393686088569661e-08,
+      "loss": 0.5511,
+      "step": 58632
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 5.199941635131836,
+      "learning_rate": 5.3898245324562894e-08,
+      "loss": 0.5713,
+      "step": 58633
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.9866530895233154,
+      "learning_rate": 5.385964351668549e-08,
+      "loss": 0.6381,
+      "step": 58634
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.213670253753662,
+      "learning_rate": 5.3821055462171534e-08,
+      "loss": 0.5776,
+      "step": 58635
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.115320920944214,
+      "learning_rate": 5.378248116112816e-08,
+      "loss": 0.5579,
+      "step": 58636
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.387714385986328,
+      "learning_rate": 5.3743920613662515e-08,
+      "loss": 0.3602,
+      "step": 58637
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.4018850326538086,
+      "learning_rate": 5.370537381988228e-08,
+      "loss": 0.5404,
+      "step": 58638
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 5.132506370544434,
+      "learning_rate": 5.3666840779895146e-08,
+      "loss": 0.5227,
+      "step": 58639
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.494229793548584,
+      "learning_rate": 5.362832149380659e-08,
+      "loss": 0.5821,
+      "step": 58640
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.3457610607147217,
+      "learning_rate": 5.358981596172541e-08,
+      "loss": 0.4053,
+      "step": 58641
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.5003035068511963,
+      "learning_rate": 5.3551324183757635e-08,
+      "loss": 0.7299,
+      "step": 58642
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.3038995265960693,
+      "learning_rate": 5.351284616001096e-08,
+      "loss": 0.4967,
+      "step": 58643
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.470097541809082,
+      "learning_rate": 5.347438189059195e-08,
+      "loss": 0.7344,
+      "step": 58644
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.68424391746521,
+      "learning_rate": 5.343593137560721e-08,
+      "loss": 0.4411,
+      "step": 58645
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.551433563232422,
+      "learning_rate": 5.339749461516441e-08,
+      "loss": 0.3742,
+      "step": 58646
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.763047218322754,
+      "learning_rate": 5.335907160937015e-08,
+      "loss": 0.6543,
+      "step": 58647
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.474781036376953,
+      "learning_rate": 5.3320662358331554e-08,
+      "loss": 0.82,
+      "step": 58648
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.048153877258301,
+      "learning_rate": 5.3282266862154655e-08,
+      "loss": 0.6815,
+      "step": 58649
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.836104393005371,
+      "learning_rate": 5.3243885120947136e-08,
+      "loss": 0.5252,
+      "step": 58650
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 4.349075794219971,
+      "learning_rate": 5.320551713481503e-08,
+      "loss": 0.6873,
+      "step": 58651
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.7257370948791504,
+      "learning_rate": 5.316716290386548e-08,
+      "loss": 0.4818,
+      "step": 58652
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.271523952484131,
+      "learning_rate": 5.31288224282045e-08,
+      "loss": 0.9219,
+      "step": 58653
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.9265167713165283,
+      "learning_rate": 5.3090495707939784e-08,
+      "loss": 0.6572,
+      "step": 58654
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 4.012546062469482,
+      "learning_rate": 5.3052182743176804e-08,
+      "loss": 0.6186,
+      "step": 58655
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 7.051413059234619,
+      "learning_rate": 5.3013883534022145e-08,
+      "loss": 0.5539,
+      "step": 58656
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.2001070976257324,
+      "learning_rate": 5.29755980805835e-08,
+      "loss": 0.2912,
+      "step": 58657
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.515803337097168,
+      "learning_rate": 5.293732638296578e-08,
+      "loss": 0.535,
+      "step": 58658
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 4.8709187507629395,
+      "learning_rate": 5.289906844127668e-08,
+      "loss": 0.4516,
+      "step": 58659
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 4.463720321655273,
+      "learning_rate": 5.2860824255622225e-08,
+      "loss": 0.7544,
+      "step": 58660
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 4.897171974182129,
+      "learning_rate": 5.282259382610788e-08,
+      "loss": 0.7708,
+      "step": 58661
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 12.637346267700195,
+      "learning_rate": 5.278437715284135e-08,
+      "loss": 0.5009,
+      "step": 58662
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.7825400829315186,
+      "learning_rate": 5.274617423592754e-08,
+      "loss": 0.7452,
+      "step": 58663
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 4.983597278594971,
+      "learning_rate": 5.270798507547359e-08,
+      "loss": 0.4179,
+      "step": 58664
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.4284963607788086,
+      "learning_rate": 5.266980967158553e-08,
+      "loss": 0.5213,
+      "step": 58665
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 4.428764343261719,
+      "learning_rate": 5.2631648024369374e-08,
+      "loss": 0.6297,
+      "step": 58666
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.676867723464966,
+      "learning_rate": 5.259350013393116e-08,
+      "loss": 0.5679,
+      "step": 58667
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.7108168601989746,
+      "learning_rate": 5.255536600037692e-08,
+      "loss": 0.5911,
+      "step": 58668
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 4.678099155426025,
+      "learning_rate": 5.2517245623813774e-08,
+      "loss": 0.5088,
+      "step": 58669
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.8794329166412354,
+      "learning_rate": 5.2479139004345535e-08,
+      "loss": 0.6617,
+      "step": 58670
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.453883647918701,
+      "learning_rate": 5.24410461420799e-08,
+      "loss": 0.8025,
+      "step": 58671
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 4.025121688842773,
+      "learning_rate": 5.240296703712289e-08,
+      "loss": 0.7662,
+      "step": 58672
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.741593360900879,
+      "learning_rate": 5.236490168957942e-08,
+      "loss": 0.4406,
+      "step": 58673
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.2435898780822754,
+      "learning_rate": 5.2326850099555516e-08,
+      "loss": 0.5364,
+      "step": 58674
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.403047800064087,
+      "learning_rate": 5.228881226715721e-08,
+      "loss": 0.6686,
+      "step": 58675
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.6493258476257324,
+      "learning_rate": 5.225078819249052e-08,
+      "loss": 0.7377,
+      "step": 58676
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.3011984825134277,
+      "learning_rate": 5.2212777875660925e-08,
+      "loss": 0.5673,
+      "step": 58677
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.740379810333252,
+      "learning_rate": 5.217478131677445e-08,
+      "loss": 0.6596,
+      "step": 58678
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.340485095977783,
+      "learning_rate": 5.213679851593545e-08,
+      "loss": 0.5933,
+      "step": 58679
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.46974515914917,
+      "learning_rate": 5.209882947325107e-08,
+      "loss": 0.584,
+      "step": 58680
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.119051218032837,
+      "learning_rate": 5.2060874188826214e-08,
+      "loss": 0.6519,
+      "step": 58681
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.92081880569458,
+      "learning_rate": 5.2022932662766925e-08,
+      "loss": 0.5796,
+      "step": 58682
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.5160341262817383,
+      "learning_rate": 5.19850048951781e-08,
+      "loss": 0.4971,
+      "step": 58683
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 4.076722145080566,
+      "learning_rate": 5.194709088616523e-08,
+      "loss": 0.6353,
+      "step": 58684
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 4.421329975128174,
+      "learning_rate": 5.190919063583377e-08,
+      "loss": 0.6885,
+      "step": 58685
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.7565596103668213,
+      "learning_rate": 5.18713041442892e-08,
+      "loss": 0.4605,
+      "step": 58686
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.523500680923462,
+      "learning_rate": 5.183343141163699e-08,
+      "loss": 0.6359,
+      "step": 58687
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.8222126960754395,
+      "learning_rate": 5.1795572437983166e-08,
+      "loss": 0.4129,
+      "step": 58688
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.965372085571289,
+      "learning_rate": 5.1757727223430974e-08,
+      "loss": 0.3934,
+      "step": 58689
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.4364545345306396,
+      "learning_rate": 5.171989576808811e-08,
+      "loss": 0.6997,
+      "step": 58690
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.0192854404449463,
+      "learning_rate": 5.168207807205783e-08,
+      "loss": 0.7521,
+      "step": 58691
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 4.643139362335205,
+      "learning_rate": 5.164427413544615e-08,
+      "loss": 0.8361,
+      "step": 58692
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.265193223953247,
+      "learning_rate": 5.160648395835799e-08,
+      "loss": 0.386,
+      "step": 58693
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 4.439743995666504,
+      "learning_rate": 5.156870754089882e-08,
+      "loss": 0.7483,
+      "step": 58694
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.52103328704834,
+      "learning_rate": 5.153094488317245e-08,
+      "loss": 0.3988,
+      "step": 58695
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.9475462436676025,
+      "learning_rate": 5.149319598528602e-08,
+      "loss": 0.6545,
+      "step": 58696
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.414433240890503,
+      "learning_rate": 5.145546084734221e-08,
+      "loss": 0.5422,
+      "step": 58697
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.532989263534546,
+      "learning_rate": 5.141773946944761e-08,
+      "loss": 0.4847,
+      "step": 58698
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.498000383377075,
+      "learning_rate": 5.1380031851706035e-08,
+      "loss": 0.5241,
+      "step": 58699
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 4.238192081451416,
+      "learning_rate": 5.13423379942235e-08,
+      "loss": 0.5099,
+      "step": 58700
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.984572649002075,
+      "learning_rate": 5.1304657897103263e-08,
+      "loss": 0.6334,
+      "step": 58701
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.6131768226623535,
+      "learning_rate": 5.1266991560451895e-08,
+      "loss": 0.6293,
+      "step": 58702
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.0639901161193848,
+      "learning_rate": 5.122933898437266e-08,
+      "loss": 0.3411,
+      "step": 58703
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.6792171001434326,
+      "learning_rate": 5.1191700168970457e-08,
+      "loss": 0.5518,
+      "step": 58704
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.591780662536621,
+      "learning_rate": 5.1154075114350775e-08,
+      "loss": 0.6514,
+      "step": 58705
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.1814632415771484,
+      "learning_rate": 5.1116463820617415e-08,
+      "loss": 0.3932,
+      "step": 58706
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 1.8114086389541626,
+      "learning_rate": 5.1078866287875285e-08,
+      "loss": 0.2874,
+      "step": 58707
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.460066318511963,
+      "learning_rate": 5.104128251622931e-08,
+      "loss": 0.5942,
+      "step": 58708
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.880120038986206,
+      "learning_rate": 5.100371250578329e-08,
+      "loss": 0.6863,
+      "step": 58709
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.4026293754577637,
+      "learning_rate": 5.096615625664214e-08,
+      "loss": 0.6881,
+      "step": 58710
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.906662940979004,
+      "learning_rate": 5.0928613768909674e-08,
+      "loss": 0.5266,
+      "step": 58711
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.1812243461608887,
+      "learning_rate": 5.089108504269136e-08,
+      "loss": 0.5152,
+      "step": 58712
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.6964080333709717,
+      "learning_rate": 5.0853570078091e-08,
+      "loss": 0.4366,
+      "step": 58713
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.000426769256592,
+      "learning_rate": 5.08160688752124e-08,
+      "loss": 0.3865,
+      "step": 58714
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.802960157394409,
+      "learning_rate": 5.077858143416048e-08,
+      "loss": 0.517,
+      "step": 58715
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.862942695617676,
+      "learning_rate": 5.07411077550396e-08,
+      "loss": 0.5651,
+      "step": 58716
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 5.342236042022705,
+      "learning_rate": 5.070364783795356e-08,
+      "loss": 0.7605,
+      "step": 58717
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.309487819671631,
+      "learning_rate": 5.0666201683006175e-08,
+      "loss": 0.5216,
+      "step": 58718
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.2556612491607666,
+      "learning_rate": 5.0628769290302357e-08,
+      "loss": 0.4387,
+      "step": 58719
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.7064731121063232,
+      "learning_rate": 5.0591350659945913e-08,
+      "loss": 0.5963,
+      "step": 58720
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.7463886737823486,
+      "learning_rate": 5.0553945792040646e-08,
+      "loss": 0.4326,
+      "step": 58721
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.7177934646606445,
+      "learning_rate": 5.051655468669092e-08,
+      "loss": 0.6886,
+      "step": 58722
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.6281495094299316,
+      "learning_rate": 5.047917734400054e-08,
+      "loss": 0.4286,
+      "step": 58723
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.595219850540161,
+      "learning_rate": 5.044181376407331e-08,
+      "loss": 0.5716,
+      "step": 58724
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.3139538764953613,
+      "learning_rate": 5.040446394701359e-08,
+      "loss": 0.6263,
+      "step": 58725
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.5389699935913086,
+      "learning_rate": 5.036712789292464e-08,
+      "loss": 0.5998,
+      "step": 58726
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.0256731510162354,
+      "learning_rate": 5.032980560191025e-08,
+      "loss": 0.5982,
+      "step": 58727
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.855436086654663,
+      "learning_rate": 5.02924970740748e-08,
+      "loss": 0.3624,
+      "step": 58728
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.9412763118743896,
+      "learning_rate": 5.025520230952152e-08,
+      "loss": 0.5738,
+      "step": 58729
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.6234891414642334,
+      "learning_rate": 5.021792130835479e-08,
+      "loss": 0.679,
+      "step": 58730
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.903165102005005,
+      "learning_rate": 5.018065407067729e-08,
+      "loss": 0.7271,
+      "step": 58731
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.206831932067871,
+      "learning_rate": 5.0143400596592836e-08,
+      "loss": 0.5945,
+      "step": 58732
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.9736549854278564,
+      "learning_rate": 5.0106160886206345e-08,
+      "loss": 0.5557,
+      "step": 58733
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.2599408626556396,
+      "learning_rate": 5.0068934939619394e-08,
+      "loss": 0.5083,
+      "step": 58734
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.5086636543273926,
+      "learning_rate": 5.003172275693691e-08,
+      "loss": 0.5642,
+      "step": 58735
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.5280585289001465,
+      "learning_rate": 4.999452433826213e-08,
+      "loss": 0.4485,
+      "step": 58736
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.601191997528076,
+      "learning_rate": 4.995733968369776e-08,
+      "loss": 0.7549,
+      "step": 58737
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.091174602508545,
+      "learning_rate": 4.992016879334816e-08,
+      "loss": 0.5268,
+      "step": 58738
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.498347043991089,
+      "learning_rate": 4.988301166731602e-08,
+      "loss": 0.548,
+      "step": 58739
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.702418565750122,
+      "learning_rate": 4.984586830570459e-08,
+      "loss": 0.8416,
+      "step": 58740
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 4.990204811096191,
+      "learning_rate": 4.980873870861769e-08,
+      "loss": 0.8412,
+      "step": 58741
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.3412320613861084,
+      "learning_rate": 4.9771622876158555e-08,
+      "loss": 0.7215,
+      "step": 58742
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.0216240882873535,
+      "learning_rate": 4.9734520808429885e-08,
+      "loss": 0.826,
+      "step": 58743
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 10.435587882995605,
+      "learning_rate": 4.9697432505534936e-08,
+      "loss": 0.5818,
+      "step": 58744
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.463470697402954,
+      "learning_rate": 4.966035796757751e-08,
+      "loss": 0.5812,
+      "step": 58745
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.8823490142822266,
+      "learning_rate": 4.9623297194659744e-08,
+      "loss": 0.511,
+      "step": 58746
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.643710136413574,
+      "learning_rate": 4.95862501868849e-08,
+      "loss": 0.4861,
+      "step": 58747
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.744063377380371,
+      "learning_rate": 4.954921694435622e-08,
+      "loss": 0.4229,
+      "step": 58748
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 4.648381233215332,
+      "learning_rate": 4.95121974671775e-08,
+      "loss": 0.8333,
+      "step": 58749
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.6774940490722656,
+      "learning_rate": 4.9475191755450345e-08,
+      "loss": 0.5809,
+      "step": 58750
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.6172170639038086,
+      "learning_rate": 4.9438199809278e-08,
+      "loss": 0.6221,
+      "step": 58751
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.38370680809021,
+      "learning_rate": 4.940122162876371e-08,
+      "loss": 0.5484,
+      "step": 58752
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.5845143795013428,
+      "learning_rate": 4.936425721401017e-08,
+      "loss": 0.4387,
+      "step": 58753
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.716386556625366,
+      "learning_rate": 4.932730656512008e-08,
+      "loss": 0.5922,
+      "step": 58754
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.092496633529663,
+      "learning_rate": 4.9290369682196135e-08,
+      "loss": 0.5534,
+      "step": 58755
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.6673426628112793,
+      "learning_rate": 4.925344656534104e-08,
+      "loss": 0.4936,
+      "step": 58756
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.5627381801605225,
+      "learning_rate": 4.921653721465747e-08,
+      "loss": 0.7272,
+      "step": 58757
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 4.206989765167236,
+      "learning_rate": 4.9179641630248135e-08,
+      "loss": 0.6616,
+      "step": 58758
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.5460028648376465,
+      "learning_rate": 4.914275981221572e-08,
+      "loss": 0.5386,
+      "step": 58759
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.9454164505004883,
+      "learning_rate": 4.9105891760662385e-08,
+      "loss": 0.5116,
+      "step": 58760
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 4.248807430267334,
+      "learning_rate": 4.9069037475691363e-08,
+      "loss": 0.6763,
+      "step": 58761
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.0373785495758057,
+      "learning_rate": 4.903219695740424e-08,
+      "loss": 0.5793,
+      "step": 58762
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.7207045555114746,
+      "learning_rate": 4.899537020590428e-08,
+      "loss": 0.3153,
+      "step": 58763
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.931992530822754,
+      "learning_rate": 4.895855722129306e-08,
+      "loss": 0.4806,
+      "step": 58764
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.6655800342559814,
+      "learning_rate": 4.892175800367382e-08,
+      "loss": 0.5389,
+      "step": 58765
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.01507568359375,
+      "learning_rate": 4.888497255314817e-08,
+      "loss": 0.4987,
+      "step": 58766
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.982692241668701,
+      "learning_rate": 4.884820086981879e-08,
+      "loss": 0.7273,
+      "step": 58767
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.186105966567993,
+      "learning_rate": 4.881144295378781e-08,
+      "loss": 0.4607,
+      "step": 58768
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.013993263244629,
+      "learning_rate": 4.877469880515739e-08,
+      "loss": 0.603,
+      "step": 58769
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.3534135818481445,
+      "learning_rate": 4.873796842402967e-08,
+      "loss": 0.5326,
+      "step": 58770
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.2519919872283936,
+      "learning_rate": 4.8701251810507336e-08,
+      "loss": 0.4436,
+      "step": 58771
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.4266135692596436,
+      "learning_rate": 4.8664548964691416e-08,
+      "loss": 0.4116,
+      "step": 58772
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.2689146995544434,
+      "learning_rate": 4.862785988668517e-08,
+      "loss": 0.6238,
+      "step": 58773
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.750070571899414,
+      "learning_rate": 4.859118457658962e-08,
+      "loss": 0.5109,
+      "step": 58774
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.9615745544433594,
+      "learning_rate": 4.855452303450747e-08,
+      "loss": 0.687,
+      "step": 58775
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.15130615234375,
+      "learning_rate": 4.8517875260539746e-08,
+      "loss": 0.4658,
+      "step": 58776
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.9609644412994385,
+      "learning_rate": 4.8481241254789704e-08,
+      "loss": 0.5958,
+      "step": 58777
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.8616111278533936,
+      "learning_rate": 4.8444621017357806e-08,
+      "loss": 0.5724,
+      "step": 58778
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.4207026958465576,
+      "learning_rate": 4.8408014548347316e-08,
+      "loss": 0.5199,
+      "step": 58779
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.3604471683502197,
+      "learning_rate": 4.8371421847858145e-08,
+      "loss": 0.6563,
+      "step": 58780
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.82122802734375,
+      "learning_rate": 4.8334842915994106e-08,
+      "loss": 0.5461,
+      "step": 58781
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 1.9109373092651367,
+      "learning_rate": 4.829827775285512e-08,
+      "loss": 0.5344,
+      "step": 58782
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.0366461277008057,
+      "learning_rate": 4.826172635854387e-08,
+      "loss": 0.5179,
+      "step": 58783
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.8785037994384766,
+      "learning_rate": 4.822518873316251e-08,
+      "loss": 0.4698,
+      "step": 58784
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.738710641860962,
+      "learning_rate": 4.8188664876810954e-08,
+      "loss": 0.7172,
+      "step": 58785
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.402005195617676,
+      "learning_rate": 4.8152154789592456e-08,
+      "loss": 0.5861,
+      "step": 58786
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.6330313682556152,
+      "learning_rate": 4.8115658471607486e-08,
+      "loss": 0.5888,
+      "step": 58787
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.2855257987976074,
+      "learning_rate": 4.8079175922957635e-08,
+      "loss": 0.8134,
+      "step": 58788
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.704291582107544,
+      "learning_rate": 4.8042707143745036e-08,
+      "loss": 0.502,
+      "step": 58789
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 4.5029754638671875,
+      "learning_rate": 4.8006252134070176e-08,
+      "loss": 0.8305,
+      "step": 58790
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.2222719192504883,
+      "learning_rate": 4.796981089403519e-08,
+      "loss": 0.4943,
+      "step": 58791
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.800360679626465,
+      "learning_rate": 4.79333834237411e-08,
+      "loss": 0.6067,
+      "step": 58792
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.5245251655578613,
+      "learning_rate": 4.7896969723288947e-08,
+      "loss": 0.6363,
+      "step": 58793
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.827869176864624,
+      "learning_rate": 4.786056979278031e-08,
+      "loss": 0.5059,
+      "step": 58794
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.194622755050659,
+      "learning_rate": 4.7824183632316224e-08,
+      "loss": 0.5003,
+      "step": 58795
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.426023244857788,
+      "learning_rate": 4.778781124199827e-08,
+      "loss": 0.6139,
+      "step": 58796
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.744769811630249,
+      "learning_rate": 4.775145262192693e-08,
+      "loss": 0.4361,
+      "step": 58797
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.467444658279419,
+      "learning_rate": 4.7715107772203785e-08,
+      "loss": 0.403,
+      "step": 58798
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 4.871857166290283,
+      "learning_rate": 4.767877669292986e-08,
+      "loss": 0.5704,
+      "step": 58799
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.4090240001678467,
+      "learning_rate": 4.76424593842062e-08,
+      "loss": 0.5315,
+      "step": 58800
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.2951319217681885,
+      "learning_rate": 4.760615584613326e-08,
+      "loss": 0.613,
+      "step": 58801
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.961759567260742,
+      "learning_rate": 4.756986607881264e-08,
+      "loss": 0.5932,
+      "step": 58802
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.4841415882110596,
+      "learning_rate": 4.753359008234481e-08,
+      "loss": 0.3828,
+      "step": 58803
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.693444013595581,
+      "learning_rate": 4.7497327856831365e-08,
+      "loss": 0.5316,
+      "step": 58804
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.9434468746185303,
+      "learning_rate": 4.7461079402372214e-08,
+      "loss": 0.5841,
+      "step": 58805
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.1624791622161865,
+      "learning_rate": 4.7424844719068384e-08,
+      "loss": 0.3735,
+      "step": 58806
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.498351812362671,
+      "learning_rate": 4.7388623807020915e-08,
+      "loss": 0.3985,
+      "step": 58807
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.5325427055358887,
+      "learning_rate": 4.735241666633028e-08,
+      "loss": 0.5632,
+      "step": 58808
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.9069771766662598,
+      "learning_rate": 4.731622329709751e-08,
+      "loss": 0.5871,
+      "step": 58809
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.933215379714966,
+      "learning_rate": 4.7280043699422515e-08,
+      "loss": 0.5596,
+      "step": 58810
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.1037755012512207,
+      "learning_rate": 4.724387787340689e-08,
+      "loss": 0.4601,
+      "step": 58811
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.0957210063934326,
+      "learning_rate": 4.7207725819150565e-08,
+      "loss": 0.6117,
+      "step": 58812
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 5.244986534118652,
+      "learning_rate": 4.717158753675399e-08,
+      "loss": 0.4509,
+      "step": 58813
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.4929308891296387,
+      "learning_rate": 4.713546302631822e-08,
+      "loss": 0.3949,
+      "step": 58814
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 4.930666446685791,
+      "learning_rate": 4.7099352287943156e-08,
+      "loss": 0.4925,
+      "step": 58815
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.753032684326172,
+      "learning_rate": 4.706325532172984e-08,
+      "loss": 0.6441,
+      "step": 58816
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.845928907394409,
+      "learning_rate": 4.702717212777763e-08,
+      "loss": 0.5671,
+      "step": 58817
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.8799853324890137,
+      "learning_rate": 4.699110270618756e-08,
+      "loss": 0.5194,
+      "step": 58818
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.5021514892578125,
+      "learning_rate": 4.695504705705956e-08,
+      "loss": 0.8577,
+      "step": 58819
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.7823431491851807,
+      "learning_rate": 4.691900518049464e-08,
+      "loss": 0.5735,
+      "step": 58820
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.1495800018310547,
+      "learning_rate": 4.688297707659217e-08,
+      "loss": 0.5912,
+      "step": 58821
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.324187755584717,
+      "learning_rate": 4.684696274545319e-08,
+      "loss": 0.7495,
+      "step": 58822
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.578136444091797,
+      "learning_rate": 4.681096218717651e-08,
+      "loss": 0.6218,
+      "step": 58823
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.6608498096466064,
+      "learning_rate": 4.6774975401863706e-08,
+      "loss": 0.4547,
+      "step": 58824
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.7059552669525146,
+      "learning_rate": 4.6739002389613596e-08,
+      "loss": 0.4843,
+      "step": 58825
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.5626156330108643,
+      "learning_rate": 4.6703043150527206e-08,
+      "loss": 0.5557,
+      "step": 58826
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.488593101501465,
+      "learning_rate": 4.66670976847039e-08,
+      "loss": 0.5514,
+      "step": 58827
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.304187536239624,
+      "learning_rate": 4.6631165992243596e-08,
+      "loss": 0.3904,
+      "step": 58828
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.3394079208374023,
+      "learning_rate": 4.659524807324678e-08,
+      "loss": 0.4614,
+      "step": 58829
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.68078875541687,
+      "learning_rate": 4.6559343927812806e-08,
+      "loss": 0.6096,
+      "step": 58830
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.1571269035339355,
+      "learning_rate": 4.6523453556041595e-08,
+      "loss": 0.2404,
+      "step": 58831
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.874880313873291,
+      "learning_rate": 4.6487576958033633e-08,
+      "loss": 0.8459,
+      "step": 58832
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.8319358825683594,
+      "learning_rate": 4.645171413388772e-08,
+      "loss": 0.5593,
+      "step": 58833
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.667431592941284,
+      "learning_rate": 4.641586508370377e-08,
+      "loss": 0.5091,
+      "step": 58834
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.6687841415405273,
+      "learning_rate": 4.6380029807581714e-08,
+      "loss": 0.3013,
+      "step": 58835
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 4.1611104011535645,
+      "learning_rate": 4.634420830562147e-08,
+      "loss": 0.8437,
+      "step": 58836
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.378983974456787,
+      "learning_rate": 4.6308400577921853e-08,
+      "loss": 0.4347,
+      "step": 58837
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.05564546585083,
+      "learning_rate": 4.627260662458333e-08,
+      "loss": 0.6137,
+      "step": 58838
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.9429447650909424,
+      "learning_rate": 4.623682644570471e-08,
+      "loss": 0.5636,
+      "step": 58839
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.2293946743011475,
+      "learning_rate": 4.6201060041385915e-08,
+      "loss": 0.507,
+      "step": 58840
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.649043560028076,
+      "learning_rate": 4.616530741172631e-08,
+      "loss": 0.5783,
+      "step": 58841
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.4100594520568848,
+      "learning_rate": 4.612956855682526e-08,
+      "loss": 0.52,
+      "step": 58842
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.901179552078247,
+      "learning_rate": 4.6093843476781587e-08,
+      "loss": 0.4703,
+      "step": 58843
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 4.835853099822998,
+      "learning_rate": 4.605813217169519e-08,
+      "loss": 0.741,
+      "step": 58844
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.2593889236450195,
+      "learning_rate": 4.6022434641666004e-08,
+      "loss": 0.4288,
+      "step": 58845
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.1795506477355957,
+      "learning_rate": 4.598675088679172e-08,
+      "loss": 0.5527,
+      "step": 58846
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.645125389099121,
+      "learning_rate": 4.5951080907173376e-08,
+      "loss": 0.9966,
+      "step": 58847
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.981879949569702,
+      "learning_rate": 4.5915424702908664e-08,
+      "loss": 0.5602,
+      "step": 58848
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 5.967657566070557,
+      "learning_rate": 4.587978227409751e-08,
+      "loss": 0.5631,
+      "step": 58849
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.5975537300109863,
+      "learning_rate": 4.584415362083872e-08,
+      "loss": 0.3249,
+      "step": 58850
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.5265846252441406,
+      "learning_rate": 4.5808538743231655e-08,
+      "loss": 0.5014,
+      "step": 58851
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.668023109436035,
+      "learning_rate": 4.577293764137514e-08,
+      "loss": 0.6826,
+      "step": 58852
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.4161276817321777,
+      "learning_rate": 4.573735031536797e-08,
+      "loss": 0.7757,
+      "step": 58853
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.756669521331787,
+      "learning_rate": 4.5701776765310066e-08,
+      "loss": 0.3653,
+      "step": 58854
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.7970762252807617,
+      "learning_rate": 4.566621699129858e-08,
+      "loss": 0.4402,
+      "step": 58855
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.5849249362945557,
+      "learning_rate": 4.563067099343399e-08,
+      "loss": 0.6462,
+      "step": 58856
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.8379669189453125,
+      "learning_rate": 4.559513877181454e-08,
+      "loss": 0.6503,
+      "step": 58857
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.6782023906707764,
+      "learning_rate": 4.555962032653849e-08,
+      "loss": 0.6398,
+      "step": 58858
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.4929888248443604,
+      "learning_rate": 4.5524115657705756e-08,
+      "loss": 0.6024,
+      "step": 58859
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.5616886615753174,
+      "learning_rate": 4.548862476541405e-08,
+      "loss": 0.5308,
+      "step": 58860
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.8791282176971436,
+      "learning_rate": 4.5453147649763276e-08,
+      "loss": 0.7662,
+      "step": 58861
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 4.204098701477051,
+      "learning_rate": 4.5417684310850585e-08,
+      "loss": 0.4611,
+      "step": 58862
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.691030263900757,
+      "learning_rate": 4.538223474877534e-08,
+      "loss": 0.6846,
+      "step": 58863
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 4.104915142059326,
+      "learning_rate": 4.5346798963635805e-08,
+      "loss": 0.7953,
+      "step": 58864
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.992420196533203,
+      "learning_rate": 4.531137695553134e-08,
+      "loss": 0.586,
+      "step": 58865
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.2019340991973877,
+      "learning_rate": 4.527596872455964e-08,
+      "loss": 0.4077,
+      "step": 58866
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.853456497192383,
+      "learning_rate": 4.524057427081951e-08,
+      "loss": 0.681,
+      "step": 58867
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.512834072113037,
+      "learning_rate": 4.520519359440923e-08,
+      "loss": 0.7842,
+      "step": 58868
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.0479464530944824,
+      "learning_rate": 4.5169826695427024e-08,
+      "loss": 0.5177,
+      "step": 58869
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.2863922119140625,
+      "learning_rate": 4.513447357397116e-08,
+      "loss": 0.8792,
+      "step": 58870
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.4155333042144775,
+      "learning_rate": 4.509913423014045e-08,
+      "loss": 0.5868,
+      "step": 58871
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.1256167888641357,
+      "learning_rate": 4.506380866403315e-08,
+      "loss": 0.5558,
+      "step": 58872
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.660753011703491,
+      "learning_rate": 4.502849687574695e-08,
+      "loss": 0.7196,
+      "step": 58873
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.4331648349761963,
+      "learning_rate": 4.499319886538012e-08,
+      "loss": 0.4333,
+      "step": 58874
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 4.583682060241699,
+      "learning_rate": 4.495791463303145e-08,
+      "loss": 1.1117,
+      "step": 58875
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.8849313259124756,
+      "learning_rate": 4.492264417879866e-08,
+      "loss": 0.6132,
+      "step": 58876
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.6557722091674805,
+      "learning_rate": 4.488738750277943e-08,
+      "loss": 0.5541,
+      "step": 58877
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.4669687747955322,
+      "learning_rate": 4.485214460507259e-08,
+      "loss": 0.7863,
+      "step": 58878
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 4.742055892944336,
+      "learning_rate": 4.481691548577527e-08,
+      "loss": 0.4807,
+      "step": 58879
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.298916816711426,
+      "learning_rate": 4.478170014498573e-08,
+      "loss": 0.7231,
+      "step": 58880
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.884796619415283,
+      "learning_rate": 4.4746498582802775e-08,
+      "loss": 0.6185,
+      "step": 58881
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.4572813510894775,
+      "learning_rate": 4.4711310799323005e-08,
+      "loss": 0.5737,
+      "step": 58882
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.049771785736084,
+      "learning_rate": 4.467613679464467e-08,
+      "loss": 0.4231,
+      "step": 58883
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.7016854286193848,
+      "learning_rate": 4.464097656886601e-08,
+      "loss": 0.6424,
+      "step": 58884
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.9687790870666504,
+      "learning_rate": 4.460583012208419e-08,
+      "loss": 0.7908,
+      "step": 58885
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.603764772415161,
+      "learning_rate": 4.457069745439746e-08,
+      "loss": 0.6117,
+      "step": 58886
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.349984645843506,
+      "learning_rate": 4.453557856590296e-08,
+      "loss": 0.6097,
+      "step": 58887
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 4.525740623474121,
+      "learning_rate": 4.4500473456698946e-08,
+      "loss": 0.4737,
+      "step": 58888
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.751946210861206,
+      "learning_rate": 4.446538212688256e-08,
+      "loss": 0.3623,
+      "step": 58889
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.528369903564453,
+      "learning_rate": 4.443030457655151e-08,
+      "loss": 0.4266,
+      "step": 58890
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.5397748947143555,
+      "learning_rate": 4.4395240805803485e-08,
+      "loss": 0.8823,
+      "step": 58891
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.919994592666626,
+      "learning_rate": 4.436019081473619e-08,
+      "loss": 0.7574,
+      "step": 58892
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 1.8307232856750488,
+      "learning_rate": 4.432515460344622e-08,
+      "loss": 0.3938,
+      "step": 58893
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.36755633354187,
+      "learning_rate": 4.429013217203182e-08,
+      "loss": 0.7267,
+      "step": 58894
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.4762368202209473,
+      "learning_rate": 4.4255123520590696e-08,
+      "loss": 0.7956,
+      "step": 58895
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.5822606086730957,
+      "learning_rate": 4.4220128649218874e-08,
+      "loss": 0.5628,
+      "step": 58896
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.8035693168640137,
+      "learning_rate": 4.418514755801462e-08,
+      "loss": 0.5553,
+      "step": 58897
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.624255895614624,
+      "learning_rate": 4.4150180247074514e-08,
+      "loss": 0.881,
+      "step": 58898
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.593423843383789,
+      "learning_rate": 4.411522671649682e-08,
+      "loss": 0.5505,
+      "step": 58899
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.5729987621307373,
+      "learning_rate": 4.4080286966378116e-08,
+      "loss": 0.6349,
+      "step": 58900
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.5289418697357178,
+      "learning_rate": 4.4045360996815554e-08,
+      "loss": 0.4463,
+      "step": 58901
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.1621832847595215,
+      "learning_rate": 4.4010448807906284e-08,
+      "loss": 0.4622,
+      "step": 58902
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 4.733559608459473,
+      "learning_rate": 4.397555039974744e-08,
+      "loss": 0.3557,
+      "step": 58903
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.090498924255371,
+      "learning_rate": 4.394066577243561e-08,
+      "loss": 0.424,
+      "step": 58904
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 5.261594295501709,
+      "learning_rate": 4.390579492606906e-08,
+      "loss": 0.6167,
+      "step": 58905
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.128227949142456,
+      "learning_rate": 4.3870937860743253e-08,
+      "loss": 0.546,
+      "step": 58906
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.6497766971588135,
+      "learning_rate": 4.3836094576555356e-08,
+      "loss": 0.7027,
+      "step": 58907
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.5287671089172363,
+      "learning_rate": 4.380126507360361e-08,
+      "loss": 0.7704,
+      "step": 58908
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.681734323501587,
+      "learning_rate": 4.3766449351983486e-08,
+      "loss": 0.7921,
+      "step": 58909
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.921553373336792,
+      "learning_rate": 4.373164741179214e-08,
+      "loss": 0.6632,
+      "step": 58910
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.5003440380096436,
+      "learning_rate": 4.369685925312672e-08,
+      "loss": 0.7536,
+      "step": 58911
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.557872772216797,
+      "learning_rate": 4.366208487608326e-08,
+      "loss": 0.6055,
+      "step": 58912
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.4718050956726074,
+      "learning_rate": 4.3627324280758886e-08,
+      "loss": 0.4996,
+      "step": 58913
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.3641788959503174,
+      "learning_rate": 4.359257746725021e-08,
+      "loss": 0.6137,
+      "step": 58914
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 4.746817111968994,
+      "learning_rate": 4.3557844435654365e-08,
+      "loss": 0.7622,
+      "step": 58915
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 1.9595695734024048,
+      "learning_rate": 4.352312518606683e-08,
+      "loss": 0.3772,
+      "step": 58916
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.932809352874756,
+      "learning_rate": 4.348841971858531e-08,
+      "loss": 0.6716,
+      "step": 58917
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.3462533950805664,
+      "learning_rate": 4.3453728033305275e-08,
+      "loss": 0.4272,
+      "step": 58918
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.9242448806762695,
+      "learning_rate": 4.341905013032388e-08,
+      "loss": 0.4593,
+      "step": 58919
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.519784450531006,
+      "learning_rate": 4.338438600973716e-08,
+      "loss": 0.5506,
+      "step": 58920
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.681218385696411,
+      "learning_rate": 4.334973567164169e-08,
+      "loss": 0.5233,
+      "step": 58921
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.3859429359436035,
+      "learning_rate": 4.3315099116134076e-08,
+      "loss": 0.4266,
+      "step": 58922
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.5954227447509766,
+      "learning_rate": 4.328047634331034e-08,
+      "loss": 0.8313,
+      "step": 58923
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.8059420585632324,
+      "learning_rate": 4.324586735326652e-08,
+      "loss": 0.7694,
+      "step": 58924
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.300999402999878,
+      "learning_rate": 4.321127214609922e-08,
+      "loss": 0.5651,
+      "step": 58925
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.234179973602295,
+      "learning_rate": 4.3176690721904446e-08,
+      "loss": 0.3302,
+      "step": 58926
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.2346081733703613,
+      "learning_rate": 4.3142123080778256e-08,
+      "loss": 0.5047,
+      "step": 58927
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.36085844039917,
+      "learning_rate": 4.310756922281778e-08,
+      "loss": 0.6092,
+      "step": 58928
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.926176071166992,
+      "learning_rate": 4.307302914811739e-08,
+      "loss": 0.6468,
+      "step": 58929
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.1973013877868652,
+      "learning_rate": 4.3038502856773686e-08,
+      "loss": 0.5409,
+      "step": 58930
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.8473405838012695,
+      "learning_rate": 4.300399034888381e-08,
+      "loss": 0.7984,
+      "step": 58931
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.377469539642334,
+      "learning_rate": 4.296949162454267e-08,
+      "loss": 0.4347,
+      "step": 58932
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.980544328689575,
+      "learning_rate": 4.2935006683846315e-08,
+      "loss": 0.5924,
+      "step": 58933
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 4.240532398223877,
+      "learning_rate": 4.290053552689077e-08,
+      "loss": 0.8111,
+      "step": 58934
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 4.577244281768799,
+      "learning_rate": 4.286607815377208e-08,
+      "loss": 0.6727,
+      "step": 58935
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.4552390575408936,
+      "learning_rate": 4.283163456458572e-08,
+      "loss": 0.3642,
+      "step": 58936
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 4.780491828918457,
+      "learning_rate": 4.279720475942717e-08,
+      "loss": 0.5793,
+      "step": 58937
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.844717264175415,
+      "learning_rate": 4.276278873839357e-08,
+      "loss": 0.5821,
+      "step": 58938
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.2809958457946777,
+      "learning_rate": 4.2728386501578733e-08,
+      "loss": 0.6083,
+      "step": 58939
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.3524415493011475,
+      "learning_rate": 4.269399804907981e-08,
+      "loss": 0.4482,
+      "step": 58940
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.786921739578247,
+      "learning_rate": 4.2659623380991166e-08,
+      "loss": 0.5265,
+      "step": 58941
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.2973523139953613,
+      "learning_rate": 4.2625262497409395e-08,
+      "loss": 0.5394,
+      "step": 58942
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.603191614151001,
+      "learning_rate": 4.259091539842997e-08,
+      "loss": 0.4619,
+      "step": 58943
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.5853781700134277,
+      "learning_rate": 4.2556582084147814e-08,
+      "loss": 0.5091,
+      "step": 58944
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.723646879196167,
+      "learning_rate": 4.252226255465841e-08,
+      "loss": 0.5511,
+      "step": 58945
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.1652631759643555,
+      "learning_rate": 4.248795681005835e-08,
+      "loss": 0.6842,
+      "step": 58946
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.8114066123962402,
+      "learning_rate": 4.2453664850441444e-08,
+      "loss": 0.3323,
+      "step": 58947
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.4261527061462402,
+      "learning_rate": 4.2419386675903726e-08,
+      "loss": 0.4653,
+      "step": 58948
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.081219434738159,
+      "learning_rate": 4.238512228654068e-08,
+      "loss": 0.7759,
+      "step": 58949
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.893827438354492,
+      "learning_rate": 4.2350871682447784e-08,
+      "loss": 0.4387,
+      "step": 58950
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.3207895755767822,
+      "learning_rate": 4.2316634863719394e-08,
+      "loss": 0.4967,
+      "step": 58951
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.5881168842315674,
+      "learning_rate": 4.2282411830451566e-08,
+      "loss": 0.4786,
+      "step": 58952
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.791123390197754,
+      "learning_rate": 4.2248202582739206e-08,
+      "loss": 0.429,
+      "step": 58953
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 4.142944812774658,
+      "learning_rate": 4.221400712067725e-08,
+      "loss": 0.794,
+      "step": 58954
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.763416290283203,
+      "learning_rate": 4.217982544436061e-08,
+      "loss": 0.7837,
+      "step": 58955
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 4.04090690612793,
+      "learning_rate": 4.214565755388533e-08,
+      "loss": 0.8154,
+      "step": 58956
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.894550323486328,
+      "learning_rate": 4.2111503449345225e-08,
+      "loss": 0.693,
+      "step": 58957
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.585083484649658,
+      "learning_rate": 4.2077363130835766e-08,
+      "loss": 0.4926,
+      "step": 58958
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 4.147609233856201,
+      "learning_rate": 4.204323659845189e-08,
+      "loss": 0.6097,
+      "step": 58959
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.2665603160858154,
+      "learning_rate": 4.2009123852289057e-08,
+      "loss": 0.6216,
+      "step": 58960
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.416675090789795,
+      "learning_rate": 4.197502489244054e-08,
+      "loss": 0.5839,
+      "step": 58961
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.9835379123687744,
+      "learning_rate": 4.1940939719002927e-08,
+      "loss": 0.3693,
+      "step": 58962
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.419147253036499,
+      "learning_rate": 4.1906868332070026e-08,
+      "loss": 0.554,
+      "step": 58963
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.431058406829834,
+      "learning_rate": 4.187281073173677e-08,
+      "loss": 0.3903,
+      "step": 58964
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 4.225691795349121,
+      "learning_rate": 4.183876691809807e-08,
+      "loss": 0.5236,
+      "step": 58965
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 4.102914810180664,
+      "learning_rate": 4.180473689124831e-08,
+      "loss": 0.5165,
+      "step": 58966
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.5261151790618896,
+      "learning_rate": 4.177072065128185e-08,
+      "loss": 0.6732,
+      "step": 58967
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.5010955333709717,
+      "learning_rate": 4.173671819829361e-08,
+      "loss": 0.8877,
+      "step": 58968
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 5.091678142547607,
+      "learning_rate": 4.170272953237853e-08,
+      "loss": 0.5868,
+      "step": 58969
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.858400583267212,
+      "learning_rate": 4.1668754653630406e-08,
+      "loss": 0.7975,
+      "step": 58970
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.160672426223755,
+      "learning_rate": 4.1634793562144174e-08,
+      "loss": 0.6654,
+      "step": 58971
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.6526169776916504,
+      "learning_rate": 4.16008462580142e-08,
+      "loss": 0.6926,
+      "step": 58972
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.0312647819519043,
+      "learning_rate": 4.156691274133484e-08,
+      "loss": 0.6689,
+      "step": 58973
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.2815146446228027,
+      "learning_rate": 4.1532993012200484e-08,
+      "loss": 0.6687,
+      "step": 58974
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.7830042839050293,
+      "learning_rate": 4.1499087070704935e-08,
+      "loss": 0.5903,
+      "step": 58975
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.162883996963501,
+      "learning_rate": 4.1465194916943675e-08,
+      "loss": 0.4464,
+      "step": 58976
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.92033052444458,
+      "learning_rate": 4.14313165510094e-08,
+      "loss": 0.4105,
+      "step": 58977
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.484659433364868,
+      "learning_rate": 4.1397451972997604e-08,
+      "loss": 0.486,
+      "step": 58978
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.663384437561035,
+      "learning_rate": 4.136360118300209e-08,
+      "loss": 0.9467,
+      "step": 58979
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.7289419174194336,
+      "learning_rate": 4.1329764181116116e-08,
+      "loss": 0.5905,
+      "step": 58980
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 4.135831356048584,
+      "learning_rate": 4.129594096743517e-08,
+      "loss": 0.5815,
+      "step": 58981
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.702422618865967,
+      "learning_rate": 4.1262131542052496e-08,
+      "loss": 0.5317,
+      "step": 58982
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.6187448501586914,
+      "learning_rate": 4.122833590506248e-08,
+      "loss": 0.3871,
+      "step": 58983
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.1699106693267822,
+      "learning_rate": 4.1194554056558365e-08,
+      "loss": 0.6318,
+      "step": 58984
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 1.9980298280715942,
+      "learning_rate": 4.116078599663509e-08,
+      "loss": 0.4242,
+      "step": 58985
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.3330562114715576,
+      "learning_rate": 4.11270317253859e-08,
+      "loss": 0.7228,
+      "step": 58986
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 4.300914764404297,
+      "learning_rate": 4.109329124290462e-08,
+      "loss": 0.4956,
+      "step": 58987
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.7964773178100586,
+      "learning_rate": 4.1059564549285056e-08,
+      "loss": 0.488,
+      "step": 58988
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.91580867767334,
+      "learning_rate": 4.102585164462158e-08,
+      "loss": 0.6499,
+      "step": 58989
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.824709177017212,
+      "learning_rate": 4.0992152529007455e-08,
+      "loss": 0.4645,
+      "step": 58990
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.333108901977539,
+      "learning_rate": 4.095846720253649e-08,
+      "loss": 0.771,
+      "step": 58991
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.056121587753296,
+      "learning_rate": 4.0924795665301944e-08,
+      "loss": 0.4371,
+      "step": 58992
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.1461946964263916,
+      "learning_rate": 4.0891137917398184e-08,
+      "loss": 0.6346,
+      "step": 58993
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.505167245864868,
+      "learning_rate": 4.085749395891847e-08,
+      "loss": 0.6685,
+      "step": 58994
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.247708797454834,
+      "learning_rate": 4.0823863789956064e-08,
+      "loss": 0.3966,
+      "step": 58995
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.9177539348602295,
+      "learning_rate": 4.0790247410605335e-08,
+      "loss": 0.3936,
+      "step": 58996
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 1.7457530498504639,
+      "learning_rate": 4.0756644820958426e-08,
+      "loss": 0.2954,
+      "step": 58997
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.663501262664795,
+      "learning_rate": 4.072305602110971e-08,
+      "loss": 0.6905,
+      "step": 58998
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.8381154537200928,
+      "learning_rate": 4.0689481011153e-08,
+      "loss": 0.4419,
+      "step": 58999
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.488755702972412,
+      "learning_rate": 4.065591979118044e-08,
+      "loss": 0.6659,
+      "step": 59000
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.741279363632202,
+      "learning_rate": 4.062237236128641e-08,
+      "loss": 0.6563,
+      "step": 59001
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 4.046363353729248,
+      "learning_rate": 4.058883872156305e-08,
+      "loss": 0.5825,
+      "step": 59002
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 4.845088481903076,
+      "learning_rate": 4.055531887210473e-08,
+      "loss": 0.7664,
+      "step": 59003
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.5952579975128174,
+      "learning_rate": 4.0521812813004156e-08,
+      "loss": 0.459,
+      "step": 59004
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.6687538623809814,
+      "learning_rate": 4.0488320544354585e-08,
+      "loss": 0.5165,
+      "step": 59005
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.6868278980255127,
+      "learning_rate": 4.0454842066249276e-08,
+      "loss": 0.7172,
+      "step": 59006
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.6465346813201904,
+      "learning_rate": 4.042137737878094e-08,
+      "loss": 0.5675,
+      "step": 59007
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.7083449363708496,
+      "learning_rate": 4.038792648204282e-08,
+      "loss": 0.4776,
+      "step": 59008
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.9503276348114014,
+      "learning_rate": 4.035448937612818e-08,
+      "loss": 0.7387,
+      "step": 59009
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.321967124938965,
+      "learning_rate": 4.032106606112918e-08,
+      "loss": 0.5375,
+      "step": 59010
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.4374842643737793,
+      "learning_rate": 4.0287656537139617e-08,
+      "loss": 0.5357,
+      "step": 59011
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.413522243499756,
+      "learning_rate": 4.025426080425221e-08,
+      "loss": 0.4882,
+      "step": 59012
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.587552309036255,
+      "learning_rate": 4.022087886256021e-08,
+      "loss": 0.7089,
+      "step": 59013
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.7132534980773926,
+      "learning_rate": 4.018751071215521e-08,
+      "loss": 0.5682,
+      "step": 59014
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.3521840572357178,
+      "learning_rate": 4.0154156353131576e-08,
+      "loss": 0.5634,
+      "step": 59015
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.076404333114624,
+      "learning_rate": 4.0120815785580915e-08,
+      "loss": 0.3933,
+      "step": 59016
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 4.072781085968018,
+      "learning_rate": 4.008748900959592e-08,
+      "loss": 0.5052,
+      "step": 59017
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.3498623371124268,
+      "learning_rate": 4.0054176025269844e-08,
+      "loss": 0.6216,
+      "step": 59018
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.729036331176758,
+      "learning_rate": 4.00208768326954e-08,
+      "loss": 0.4689,
+      "step": 59019
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.740049123764038,
+      "learning_rate": 3.998759143196418e-08,
+      "loss": 0.5337,
+      "step": 59020
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.8083746433258057,
+      "learning_rate": 3.995431982317e-08,
+      "loss": 0.4733,
+      "step": 59021
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.5966858863830566,
+      "learning_rate": 3.992106200640389e-08,
+      "loss": 0.6949,
+      "step": 59022
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.8741002082824707,
+      "learning_rate": 3.9887817981760225e-08,
+      "loss": 0.5401,
+      "step": 59023
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.6877481937408447,
+      "learning_rate": 3.9854587749330045e-08,
+      "loss": 0.5906,
+      "step": 59024
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.1075193881988525,
+      "learning_rate": 3.982137130920605e-08,
+      "loss": 0.493,
+      "step": 59025
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.431488513946533,
+      "learning_rate": 3.978816866148094e-08,
+      "loss": 0.4871,
+      "step": 59026
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 4.674104690551758,
+      "learning_rate": 3.9754979806246317e-08,
+      "loss": 0.5696,
+      "step": 59027
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.3042309284210205,
+      "learning_rate": 3.9721804743595435e-08,
+      "loss": 0.4837,
+      "step": 59028
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.5966310501098633,
+      "learning_rate": 3.968864347361989e-08,
+      "loss": 0.5369,
+      "step": 59029
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.399902820587158,
+      "learning_rate": 3.965549599641183e-08,
+      "loss": 0.6494,
+      "step": 59030
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.7708065509796143,
+      "learning_rate": 3.9622362312063956e-08,
+      "loss": 0.362,
+      "step": 59031
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.43205189704895,
+      "learning_rate": 3.9589242420667864e-08,
+      "loss": 0.5075,
+      "step": 59032
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.5128395557403564,
+      "learning_rate": 3.955613632231625e-08,
+      "loss": 0.4981,
+      "step": 59033
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.829585075378418,
+      "learning_rate": 3.952304401710016e-08,
+      "loss": 0.6041,
+      "step": 59034
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.5653674602508545,
+      "learning_rate": 3.948996550511286e-08,
+      "loss": 0.5018,
+      "step": 59035
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.1889138221740723,
+      "learning_rate": 3.945690078644537e-08,
+      "loss": 0.3387,
+      "step": 59036
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.83890438079834,
+      "learning_rate": 3.942384986119041e-08,
+      "loss": 0.5413,
+      "step": 59037
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.291933059692383,
+      "learning_rate": 3.939081272943901e-08,
+      "loss": 0.3229,
+      "step": 59038
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 5.416863441467285,
+      "learning_rate": 3.9357789391283875e-08,
+      "loss": 0.7072,
+      "step": 59039
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.074928045272827,
+      "learning_rate": 3.932477984681604e-08,
+      "loss": 0.6373,
+      "step": 59040
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.188918113708496,
+      "learning_rate": 3.929178409612766e-08,
+      "loss": 0.5024,
+      "step": 59041
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.7811994552612305,
+      "learning_rate": 3.925880213931033e-08,
+      "loss": 0.4963,
+      "step": 59042
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.9120168685913086,
+      "learning_rate": 3.9225833976456187e-08,
+      "loss": 0.5993,
+      "step": 59043
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.7353198528289795,
+      "learning_rate": 3.9192879607656834e-08,
+      "loss": 0.567,
+      "step": 59044
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.644242763519287,
+      "learning_rate": 3.9159939033003856e-08,
+      "loss": 0.4614,
+      "step": 59045
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.8655502796173096,
+      "learning_rate": 3.91270122525883e-08,
+      "loss": 0.4294,
+      "step": 59046
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.9188005924224854,
+      "learning_rate": 3.9094099266502315e-08,
+      "loss": 0.8774,
+      "step": 59047
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.602147340774536,
+      "learning_rate": 3.906120007483749e-08,
+      "loss": 0.5493,
+      "step": 59048
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.4064509868621826,
+      "learning_rate": 3.902831467768431e-08,
+      "loss": 0.6814,
+      "step": 59049
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.5138213634490967,
+      "learning_rate": 3.899544307513603e-08,
+      "loss": 0.7957,
+      "step": 59050
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 4.616320610046387,
+      "learning_rate": 3.896258526728203e-08,
+      "loss": 0.672,
+      "step": 59051
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.5688588619232178,
+      "learning_rate": 3.8929741254215e-08,
+      "loss": 0.5011,
+      "step": 59052
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.674575090408325,
+      "learning_rate": 3.8896911036025445e-08,
+      "loss": 0.8088,
+      "step": 59053
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.21710467338562,
+      "learning_rate": 3.886409461280605e-08,
+      "loss": 0.6016,
+      "step": 59054
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.8213281631469727,
+      "learning_rate": 3.883129198464619e-08,
+      "loss": 0.8275,
+      "step": 59055
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.545222520828247,
+      "learning_rate": 3.879850315163913e-08,
+      "loss": 0.6244,
+      "step": 59056
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.703909158706665,
+      "learning_rate": 3.876572811387369e-08,
+      "loss": 0.7178,
+      "step": 59057
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.969351291656494,
+      "learning_rate": 3.873296687144312e-08,
+      "loss": 0.5075,
+      "step": 59058
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.5910043716430664,
+      "learning_rate": 3.870021942443736e-08,
+      "loss": 0.4648,
+      "step": 59059
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.0141518115997314,
+      "learning_rate": 3.866748577294743e-08,
+      "loss": 0.8253,
+      "step": 59060
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 5.362110614776611,
+      "learning_rate": 3.863476591706494e-08,
+      "loss": 0.7515,
+      "step": 59061
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.106997489929199,
+      "learning_rate": 3.860205985688092e-08,
+      "loss": 0.5502,
+      "step": 59062
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.7065069675445557,
+      "learning_rate": 3.8569367592485307e-08,
+      "loss": 0.491,
+      "step": 59063
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 4.120459079742432,
+      "learning_rate": 3.853668912397024e-08,
+      "loss": 0.4374,
+      "step": 59064
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.227491855621338,
+      "learning_rate": 3.8504024451425646e-08,
+      "loss": 0.5306,
+      "step": 59065
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.20595121383667,
+      "learning_rate": 3.8471373574943125e-08,
+      "loss": 0.5216,
+      "step": 59066
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.846989393234253,
+      "learning_rate": 3.843873649461316e-08,
+      "loss": 0.3994,
+      "step": 59067
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.3934555053710938,
+      "learning_rate": 3.840611321052623e-08,
+      "loss": 0.5077,
+      "step": 59068
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.930692434310913,
+      "learning_rate": 3.837350372277282e-08,
+      "loss": 0.4853,
+      "step": 59069
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.0705084800720215,
+      "learning_rate": 3.8340908031444526e-08,
+      "loss": 0.4738,
+      "step": 59070
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 4.513277530670166,
+      "learning_rate": 3.830832613663127e-08,
+      "loss": 0.6883,
+      "step": 59071
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.939145565032959,
+      "learning_rate": 3.8275758038424095e-08,
+      "loss": 0.658,
+      "step": 59072
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.1366653442382812,
+      "learning_rate": 3.824320373691293e-08,
+      "loss": 0.4007,
+      "step": 59073
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.7567315101623535,
+      "learning_rate": 3.82106632321888e-08,
+      "loss": 0.6802,
+      "step": 59074
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 5.326651573181152,
+      "learning_rate": 3.817813652434221e-08,
+      "loss": 0.7047,
+      "step": 59075
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 1.9614630937576294,
+      "learning_rate": 3.814562361346364e-08,
+      "loss": 0.2561,
+      "step": 59076
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.4447195529937744,
+      "learning_rate": 3.8113124499642996e-08,
+      "loss": 0.4256,
+      "step": 59077
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.60615611076355,
+      "learning_rate": 3.8080639182971335e-08,
+      "loss": 0.6749,
+      "step": 59078
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.635389566421509,
+      "learning_rate": 3.804816766353858e-08,
+      "loss": 0.4091,
+      "step": 59079
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 5.293529510498047,
+      "learning_rate": 3.801570994143522e-08,
+      "loss": 0.6586,
+      "step": 59080
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 4.150028705596924,
+      "learning_rate": 3.798326601675118e-08,
+      "loss": 0.6496,
+      "step": 59081
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.498506784439087,
+      "learning_rate": 3.7950835889576934e-08,
+      "loss": 0.4764,
+      "step": 59082
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.72805118560791,
+      "learning_rate": 3.791841956000242e-08,
+      "loss": 0.4237,
+      "step": 59083
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.5911221504211426,
+      "learning_rate": 3.7886017028118114e-08,
+      "loss": 0.63,
+      "step": 59084
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.926284074783325,
+      "learning_rate": 3.78536282940134e-08,
+      "loss": 0.4153,
+      "step": 59085
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.6642251014709473,
+      "learning_rate": 3.7821253357779864e-08,
+      "loss": 0.6468,
+      "step": 59086
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.676065683364868,
+      "learning_rate": 3.778889221950577e-08,
+      "loss": 0.4875,
+      "step": 59087
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.2904088497161865,
+      "learning_rate": 3.775654487928215e-08,
+      "loss": 0.4204,
+      "step": 59088
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.56644606590271,
+      "learning_rate": 3.772421133719839e-08,
+      "loss": 0.6019,
+      "step": 59089
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.240912675857544,
+      "learning_rate": 3.7691891593344964e-08,
+      "loss": 0.6233,
+      "step": 59090
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.2740535736083984,
+      "learning_rate": 3.76595856478118e-08,
+      "loss": 0.7363,
+      "step": 59091
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.5070154666900635,
+      "learning_rate": 3.762729350068772e-08,
+      "loss": 0.6189,
+      "step": 59092
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.4323806762695312,
+      "learning_rate": 3.759501515206321e-08,
+      "loss": 0.3372,
+      "step": 59093
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.74595308303833,
+      "learning_rate": 3.7562750602028743e-08,
+      "loss": 0.6629,
+      "step": 59094
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.3934378623962402,
+      "learning_rate": 3.7530499850672584e-08,
+      "loss": 0.381,
+      "step": 59095
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.1244077682495117,
+      "learning_rate": 3.7498262898085225e-08,
+      "loss": 0.7722,
+      "step": 59096
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.254625082015991,
+      "learning_rate": 3.746603974435659e-08,
+      "loss": 0.697,
+      "step": 59097
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 4.062146186828613,
+      "learning_rate": 3.743383038957549e-08,
+      "loss": 0.6351,
+      "step": 59098
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.492724895477295,
+      "learning_rate": 3.740163483383241e-08,
+      "loss": 0.6684,
+      "step": 59099
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.7315280437469482,
+      "learning_rate": 3.736945307721562e-08,
+      "loss": 0.4779,
+      "step": 59100
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.685506820678711,
+      "learning_rate": 3.73372851198156e-08,
+      "loss": 0.1801,
+      "step": 59101
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.2340376377105713,
+      "learning_rate": 3.730513096172228e-08,
+      "loss": 0.5417,
+      "step": 59102
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.4778859615325928,
+      "learning_rate": 3.7272990603023364e-08,
+      "loss": 0.5989,
+      "step": 59103
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 4.774454116821289,
+      "learning_rate": 3.724086404380933e-08,
+      "loss": 0.5105,
+      "step": 59104
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.1951494216918945,
+      "learning_rate": 3.7208751284169564e-08,
+      "loss": 0.4561,
+      "step": 59105
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 5.598423480987549,
+      "learning_rate": 3.717665232419343e-08,
+      "loss": 0.7275,
+      "step": 59106
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.6718339920043945,
+      "learning_rate": 3.714456716396975e-08,
+      "loss": 0.7359,
+      "step": 59107
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.527848720550537,
+      "learning_rate": 3.7112495803587334e-08,
+      "loss": 0.5884,
+      "step": 59108
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.9993574619293213,
+      "learning_rate": 3.708043824313667e-08,
+      "loss": 0.7102,
+      "step": 59109
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 4.2989325523376465,
+      "learning_rate": 3.704839448270548e-08,
+      "loss": 0.5308,
+      "step": 59110
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.4502928256988525,
+      "learning_rate": 3.701636452238422e-08,
+      "loss": 0.5349,
+      "step": 59111
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.526381254196167,
+      "learning_rate": 3.698434836226061e-08,
+      "loss": 0.6356,
+      "step": 59112
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 4.689813613891602,
+      "learning_rate": 3.695234600242459e-08,
+      "loss": 0.6258,
+      "step": 59113
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.320249557495117,
+      "learning_rate": 3.692035744296496e-08,
+      "loss": 0.5758,
+      "step": 59114
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.0331356525421143,
+      "learning_rate": 3.68883826839711e-08,
+      "loss": 0.4273,
+      "step": 59115
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.5628721714019775,
+      "learning_rate": 3.6856421725530724e-08,
+      "loss": 0.4715,
+      "step": 59116
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.73412823677063,
+      "learning_rate": 3.682447456773375e-08,
+      "loss": 0.5827,
+      "step": 59117
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.4659831523895264,
+      "learning_rate": 3.679254121066844e-08,
+      "loss": 0.5743,
+      "step": 59118
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.9186906814575195,
+      "learning_rate": 3.6760621654424176e-08,
+      "loss": 0.547,
+      "step": 59119
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.740379571914673,
+      "learning_rate": 3.6728715899088663e-08,
+      "loss": 0.5947,
+      "step": 59120
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.8915021419525146,
+      "learning_rate": 3.6696823944751824e-08,
+      "loss": 0.6126,
+      "step": 59121
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.692002534866333,
+      "learning_rate": 3.666494579150193e-08,
+      "loss": 0.6025,
+      "step": 59122
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.981477975845337,
+      "learning_rate": 3.663308143942723e-08,
+      "loss": 0.7322,
+      "step": 59123
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.8153042793273926,
+      "learning_rate": 3.660123088861656e-08,
+      "loss": 0.4954,
+      "step": 59124
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.3658974170684814,
+      "learning_rate": 3.6569394139159276e-08,
+      "loss": 0.5564,
+      "step": 59125
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.3297576904296875,
+      "learning_rate": 3.653757119114198e-08,
+      "loss": 0.3591,
+      "step": 59126
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 4.734982013702393,
+      "learning_rate": 3.6505762044655166e-08,
+      "loss": 0.7632,
+      "step": 59127
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.1496362686157227,
+      "learning_rate": 3.6473966699786535e-08,
+      "loss": 0.7737,
+      "step": 59128
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.622804880142212,
+      "learning_rate": 3.644218515662379e-08,
+      "loss": 0.4508,
+      "step": 59129
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.8470470905303955,
+      "learning_rate": 3.6410417415256305e-08,
+      "loss": 0.5246,
+      "step": 59130
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.879363536834717,
+      "learning_rate": 3.6378663475772344e-08,
+      "loss": 0.7322,
+      "step": 59131
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.9112889766693115,
+      "learning_rate": 3.634692333825907e-08,
+      "loss": 0.8201,
+      "step": 59132
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 5.130229949951172,
+      "learning_rate": 3.631519700280639e-08,
+      "loss": 0.5304,
+      "step": 59133
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.0394415855407715,
+      "learning_rate": 3.6283484469500916e-08,
+      "loss": 0.4459,
+      "step": 59134
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.3048460483551025,
+      "learning_rate": 3.625178573843258e-08,
+      "loss": 0.4554,
+      "step": 59135
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 4.165968418121338,
+      "learning_rate": 3.622010080968796e-08,
+      "loss": 0.6321,
+      "step": 59136
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.1305723190307617,
+      "learning_rate": 3.6188429683355894e-08,
+      "loss": 0.6751,
+      "step": 59137
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.10427188873291,
+      "learning_rate": 3.6156772359524085e-08,
+      "loss": 0.4265,
+      "step": 59138
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.8195436000823975,
+      "learning_rate": 3.6125128838281345e-08,
+      "loss": 0.5255,
+      "step": 59139
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.8349602222442627,
+      "learning_rate": 3.609349911971427e-08,
+      "loss": 0.6285,
+      "step": 59140
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.9295709133148193,
+      "learning_rate": 3.6061883203912794e-08,
+      "loss": 0.5131,
+      "step": 59141
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.0987915992736816,
+      "learning_rate": 3.603028109096296e-08,
+      "loss": 0.6141,
+      "step": 59142
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.0135602951049805,
+      "learning_rate": 3.599869278095358e-08,
+      "loss": 0.6668,
+      "step": 59143
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.176192045211792,
+      "learning_rate": 3.596711827397181e-08,
+      "loss": 0.618,
+      "step": 59144
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.4937171936035156,
+      "learning_rate": 3.593555757010647e-08,
+      "loss": 0.5738,
+      "step": 59145
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.9753336906433105,
+      "learning_rate": 3.590401066944416e-08,
+      "loss": 0.4536,
+      "step": 59146
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.432880163192749,
+      "learning_rate": 3.5872477572074236e-08,
+      "loss": 0.4696,
+      "step": 59147
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 1.4923430681228638,
+      "learning_rate": 3.5840958278082203e-08,
+      "loss": 0.1922,
+      "step": 59148
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.9380664825439453,
+      "learning_rate": 3.580945278755743e-08,
+      "loss": 0.5586,
+      "step": 59149
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.715378999710083,
+      "learning_rate": 3.5777961100587064e-08,
+      "loss": 0.7651,
+      "step": 59150
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.3336427211761475,
+      "learning_rate": 3.5746483217258266e-08,
+      "loss": 0.6151,
+      "step": 59151
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.6840505599975586,
+      "learning_rate": 3.571501913765874e-08,
+      "loss": 0.9432,
+      "step": 59152
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.422422170639038,
+      "learning_rate": 3.5683568861876186e-08,
+      "loss": 0.664,
+      "step": 59153
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.6264986991882324,
+      "learning_rate": 3.565213238999832e-08,
+      "loss": 0.6121,
+      "step": 59154
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.836211919784546,
+      "learning_rate": 3.5620709722111736e-08,
+      "loss": 0.6378,
+      "step": 59155
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.407597303390503,
+      "learning_rate": 3.55893008583047e-08,
+      "loss": 0.6692,
+      "step": 59156
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.921279191970825,
+      "learning_rate": 3.5557905798663805e-08,
+      "loss": 0.7662,
+      "step": 59157
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.8772501945495605,
+      "learning_rate": 3.5526524543276766e-08,
+      "loss": 0.5705,
+      "step": 59158
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.637382984161377,
+      "learning_rate": 3.5495157092230726e-08,
+      "loss": 0.6702,
+      "step": 59159
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 1.9151062965393066,
+      "learning_rate": 3.546380344561229e-08,
+      "loss": 0.3982,
+      "step": 59160
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.855729818344116,
+      "learning_rate": 3.543246360351027e-08,
+      "loss": 0.4753,
+      "step": 59161
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.8452048301696777,
+      "learning_rate": 3.540113756601016e-08,
+      "loss": 0.6284,
+      "step": 59162
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.603384256362915,
+      "learning_rate": 3.5369825333200215e-08,
+      "loss": 0.5342,
+      "step": 59163
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.739851713180542,
+      "learning_rate": 3.5338526905166483e-08,
+      "loss": 0.4909,
+      "step": 59164
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.231415271759033,
+      "learning_rate": 3.530724228199667e-08,
+      "loss": 0.9258,
+      "step": 59165
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.6725375652313232,
+      "learning_rate": 3.527597146377737e-08,
+      "loss": 0.8173,
+      "step": 59166
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.9113428592681885,
+      "learning_rate": 3.5244714450595743e-08,
+      "loss": 0.653,
+      "step": 59167
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 4.241074562072754,
+      "learning_rate": 3.521347124253949e-08,
+      "loss": 0.6166,
+      "step": 59168
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.944709062576294,
+      "learning_rate": 3.51822418396941e-08,
+      "loss": 0.6213,
+      "step": 59169
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.3329055309295654,
+      "learning_rate": 3.515102624214728e-08,
+      "loss": 0.4935,
+      "step": 59170
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.7080905437469482,
+      "learning_rate": 3.511982444998507e-08,
+      "loss": 0.3688,
+      "step": 59171
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.7071897983551025,
+      "learning_rate": 3.5088636463295186e-08,
+      "loss": 0.5621,
+      "step": 59172
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.162904977798462,
+      "learning_rate": 3.505746228216422e-08,
+      "loss": 0.6404,
+      "step": 59173
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.763991117477417,
+      "learning_rate": 3.5026301906677664e-08,
+      "loss": 0.6516,
+      "step": 59174
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.248155117034912,
+      "learning_rate": 3.499515533692377e-08,
+      "loss": 0.5347,
+      "step": 59175
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.3429224491119385,
+      "learning_rate": 3.496402257298803e-08,
+      "loss": 0.5823,
+      "step": 59176
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 4.0349578857421875,
+      "learning_rate": 3.4932903614957604e-08,
+      "loss": 0.7597,
+      "step": 59177
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.6442999839782715,
+      "learning_rate": 3.4901798462919075e-08,
+      "loss": 0.575,
+      "step": 59178
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.9698171615600586,
+      "learning_rate": 3.487070711695795e-08,
+      "loss": 0.5733,
+      "step": 59179
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.143394708633423,
+      "learning_rate": 3.4839629577161915e-08,
+      "loss": 0.4059,
+      "step": 59180
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.3933727741241455,
+      "learning_rate": 3.4808565843617023e-08,
+      "loss": 0.4906,
+      "step": 59181
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.9992923736572266,
+      "learning_rate": 3.477751591640932e-08,
+      "loss": 0.5458,
+      "step": 59182
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.621455192565918,
+      "learning_rate": 3.4746479795624846e-08,
+      "loss": 0.5296,
+      "step": 59183
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.45281720161438,
+      "learning_rate": 3.47154574813513e-08,
+      "loss": 0.4513,
+      "step": 59184
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 4.249275207519531,
+      "learning_rate": 3.468444897367307e-08,
+      "loss": 0.4717,
+      "step": 59185
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.215550184249878,
+      "learning_rate": 3.465345427267786e-08,
+      "loss": 0.4862,
+      "step": 59186
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 5.734969139099121,
+      "learning_rate": 3.4622473378451706e-08,
+      "loss": 0.4958,
+      "step": 59187
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 4.077587604522705,
+      "learning_rate": 3.4591506291080104e-08,
+      "loss": 0.8488,
+      "step": 59188
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.2622156143188477,
+      "learning_rate": 3.456055301064909e-08,
+      "loss": 0.6348,
+      "step": 59189
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.0985610485076904,
+      "learning_rate": 3.452961353724526e-08,
+      "loss": 0.2933,
+      "step": 59190
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.3367910385131836,
+      "learning_rate": 3.449868787095412e-08,
+      "loss": 0.5499,
+      "step": 59191
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.74033522605896,
+      "learning_rate": 3.44677760118628e-08,
+      "loss": 0.4787,
+      "step": 59192
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.1410293579101562,
+      "learning_rate": 3.443687796005568e-08,
+      "loss": 0.6825,
+      "step": 59193
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.924992561340332,
+      "learning_rate": 3.4405993715619926e-08,
+      "loss": 0.5869,
+      "step": 59194
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.4573440551757812,
+      "learning_rate": 3.437512327864101e-08,
+      "loss": 0.7021,
+      "step": 59195
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.063229560852051,
+      "learning_rate": 3.434426664920443e-08,
+      "loss": 0.6237,
+      "step": 59196
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.519975185394287,
+      "learning_rate": 3.431342382739622e-08,
+      "loss": 0.5323,
+      "step": 59197
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.2574586868286133,
+      "learning_rate": 3.428259481330243e-08,
+      "loss": 0.8256,
+      "step": 59198
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.3845291137695312,
+      "learning_rate": 3.425177960700854e-08,
+      "loss": 0.5233,
+      "step": 59199
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.695481061935425,
+      "learning_rate": 3.422097820860004e-08,
+      "loss": 0.4362,
+      "step": 59200
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.348937511444092,
+      "learning_rate": 3.4190190618162975e-08,
+      "loss": 0.6055,
+      "step": 59201
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 4.951026439666748,
+      "learning_rate": 3.415941683578283e-08,
+      "loss": 0.8674,
+      "step": 59202
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.367515802383423,
+      "learning_rate": 3.4128656861545094e-08,
+      "loss": 0.6863,
+      "step": 59203
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.0857410430908203,
+      "learning_rate": 3.409791069553525e-08,
+      "loss": 0.5783,
+      "step": 59204
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.4656288623809814,
+      "learning_rate": 3.406717833783824e-08,
+      "loss": 0.6687,
+      "step": 59205
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.1463868618011475,
+      "learning_rate": 3.40364597885412e-08,
+      "loss": 0.7542,
+      "step": 59206
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.183126449584961,
+      "learning_rate": 3.4005755047727406e-08,
+      "loss": 0.5226,
+      "step": 59207
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 5.146200656890869,
+      "learning_rate": 3.3975064115484014e-08,
+      "loss": 0.9775,
+      "step": 59208
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 1.9866575002670288,
+      "learning_rate": 3.3944386991895394e-08,
+      "loss": 0.4686,
+      "step": 59209
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.6178152561187744,
+      "learning_rate": 3.391372367704759e-08,
+      "loss": 0.5626,
+      "step": 59210
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.3936288356781006,
+      "learning_rate": 3.388307417102499e-08,
+      "loss": 0.4748,
+      "step": 59211
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.3370933532714844,
+      "learning_rate": 3.3852438473913066e-08,
+      "loss": 0.655,
+      "step": 59212
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.6781396865844727,
+      "learning_rate": 3.382181658579731e-08,
+      "loss": 0.4388,
+      "step": 59213
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.1662158966064453,
+      "learning_rate": 3.3791208506762655e-08,
+      "loss": 0.6755,
+      "step": 59214
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.6553285121917725,
+      "learning_rate": 3.376061423689403e-08,
+      "loss": 0.5755,
+      "step": 59215
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.884164333343506,
+      "learning_rate": 3.373003377627693e-08,
+      "loss": 0.5631,
+      "step": 59216
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.7908997535705566,
+      "learning_rate": 3.369946712499628e-08,
+      "loss": 0.4738,
+      "step": 59217
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.5839734077453613,
+      "learning_rate": 3.366891428313701e-08,
+      "loss": 0.5969,
+      "step": 59218
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.4578967094421387,
+      "learning_rate": 3.363837525078406e-08,
+      "loss": 0.4753,
+      "step": 59219
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.820949077606201,
+      "learning_rate": 3.360785002802236e-08,
+      "loss": 0.5687,
+      "step": 59220
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.076868772506714,
+      "learning_rate": 3.357733861493684e-08,
+      "loss": 0.4564,
+      "step": 59221
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.1782939434051514,
+      "learning_rate": 3.354684101161188e-08,
+      "loss": 0.561,
+      "step": 59222
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.991117477416992,
+      "learning_rate": 3.351635721813295e-08,
+      "loss": 0.6358,
+      "step": 59223
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.719269275665283,
+      "learning_rate": 3.3485887234585014e-08,
+      "loss": 0.6957,
+      "step": 59224
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.34248423576355,
+      "learning_rate": 3.3455431061051315e-08,
+      "loss": 0.5653,
+      "step": 59225
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.0647194385528564,
+      "learning_rate": 3.342498869761845e-08,
+      "loss": 0.4937,
+      "step": 59226
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.5438318252563477,
+      "learning_rate": 3.3394560144369706e-08,
+      "loss": 0.5596,
+      "step": 59227
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.5390055179595947,
+      "learning_rate": 3.336414540138999e-08,
+      "loss": 0.6259,
+      "step": 59228
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.144089937210083,
+      "learning_rate": 3.3333744468764805e-08,
+      "loss": 0.808,
+      "step": 59229
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.4714245796203613,
+      "learning_rate": 3.3303357346577416e-08,
+      "loss": 0.8504,
+      "step": 59230
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 12.338814735412598,
+      "learning_rate": 3.327298403491275e-08,
+      "loss": 0.6952,
+      "step": 59231
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.4506540298461914,
+      "learning_rate": 3.324262453385518e-08,
+      "loss": 0.4414,
+      "step": 59232
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 5.329126834869385,
+      "learning_rate": 3.321227884348965e-08,
+      "loss": 0.6679,
+      "step": 59233
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.388363838195801,
+      "learning_rate": 3.318194696389998e-08,
+      "loss": 0.5257,
+      "step": 59234
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.7405664920806885,
+      "learning_rate": 3.315162889517054e-08,
+      "loss": 0.8585,
+      "step": 59235
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.0965399742126465,
+      "learning_rate": 3.3121324637386264e-08,
+      "loss": 0.4029,
+      "step": 59236
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.648571729660034,
+      "learning_rate": 3.309103419063098e-08,
+      "loss": 0.6395,
+      "step": 59237
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.678920030593872,
+      "learning_rate": 3.30607575549885e-08,
+      "loss": 0.6382,
+      "step": 59238
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.9171156883239746,
+      "learning_rate": 3.30304947305432e-08,
+      "loss": 0.6008,
+      "step": 59239
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.207122564315796,
+      "learning_rate": 3.300024571738003e-08,
+      "loss": 0.5994,
+      "step": 59240
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.422150135040283,
+      "learning_rate": 3.297001051558224e-08,
+      "loss": 0.6655,
+      "step": 59241
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.750039577484131,
+      "learning_rate": 3.293978912523421e-08,
+      "loss": 0.7066,
+      "step": 59242
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.5116214752197266,
+      "learning_rate": 3.290958154641977e-08,
+      "loss": 0.6653,
+      "step": 59243
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.4284098148345947,
+      "learning_rate": 3.287938777922328e-08,
+      "loss": 0.4007,
+      "step": 59244
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.9884033203125,
+      "learning_rate": 3.284920782372802e-08,
+      "loss": 0.6546,
+      "step": 59245
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 4.167926788330078,
+      "learning_rate": 3.281904168001893e-08,
+      "loss": 0.7179,
+      "step": 59246
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.90348482131958,
+      "learning_rate": 3.2788889348179257e-08,
+      "loss": 0.661,
+      "step": 59247
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.044198989868164,
+      "learning_rate": 3.275875082829227e-08,
+      "loss": 0.5384,
+      "step": 59248
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.682353973388672,
+      "learning_rate": 3.2728626120443474e-08,
+      "loss": 0.5639,
+      "step": 59249
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.136662483215332,
+      "learning_rate": 3.269851522471501e-08,
+      "loss": 0.4468,
+      "step": 59250
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.43196964263916,
+      "learning_rate": 3.266841814119126e-08,
+      "loss": 0.5137,
+      "step": 59251
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.0649867057800293,
+      "learning_rate": 3.263833486995549e-08,
+      "loss": 0.7063,
+      "step": 59252
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.516568422317505,
+      "learning_rate": 3.260826541109208e-08,
+      "loss": 0.5543,
+      "step": 59253
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.1667966842651367,
+      "learning_rate": 3.2578209764684284e-08,
+      "loss": 0.4846,
+      "step": 59254
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.012967109680176,
+      "learning_rate": 3.254816793081539e-08,
+      "loss": 0.8631,
+      "step": 59255
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.34513258934021,
+      "learning_rate": 3.25181399095692e-08,
+      "loss": 0.4902,
+      "step": 59256
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.6845436096191406,
+      "learning_rate": 3.248812570102955e-08,
+      "loss": 0.4806,
+      "step": 59257
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.2267627716064453,
+      "learning_rate": 3.245812530527914e-08,
+      "loss": 0.4264,
+      "step": 59258
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.8129220008850098,
+      "learning_rate": 3.242813872240236e-08,
+      "loss": 0.5385,
+      "step": 59259
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.6519405841827393,
+      "learning_rate": 3.2398165952481356e-08,
+      "loss": 0.3908,
+      "step": 59260
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.810279607772827,
+      "learning_rate": 3.23682069956005e-08,
+      "loss": 0.5965,
+      "step": 59261
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.2446110248565674,
+      "learning_rate": 3.233826185184252e-08,
+      "loss": 0.5551,
+      "step": 59262
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.6781322956085205,
+      "learning_rate": 3.2308330521291234e-08,
+      "loss": 0.5472,
+      "step": 59263
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.6853954792022705,
+      "learning_rate": 3.227841300402934e-08,
+      "loss": 0.4309,
+      "step": 59264
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.62577223777771,
+      "learning_rate": 3.224850930014067e-08,
+      "loss": 0.6324,
+      "step": 59265
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.7289533615112305,
+      "learning_rate": 3.221861940970739e-08,
+      "loss": 0.5344,
+      "step": 59266
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.002335548400879,
+      "learning_rate": 3.21887433328133e-08,
+      "loss": 0.5744,
+      "step": 59267
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.627112865447998,
+      "learning_rate": 3.215888106954112e-08,
+      "loss": 0.5679,
+      "step": 59268
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.9479053020477295,
+      "learning_rate": 3.212903261997469e-08,
+      "loss": 0.7402,
+      "step": 59269
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.8906960487365723,
+      "learning_rate": 3.209919798419558e-08,
+      "loss": 0.6337,
+      "step": 59270
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.3285720348358154,
+      "learning_rate": 3.2069377162288196e-08,
+      "loss": 0.4558,
+      "step": 59271
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.173832893371582,
+      "learning_rate": 3.2039570154334673e-08,
+      "loss": 0.6366,
+      "step": 59272
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.544760227203369,
+      "learning_rate": 3.2009776960418294e-08,
+      "loss": 0.566,
+      "step": 59273
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.1477341651916504,
+      "learning_rate": 3.19799975806212e-08,
+      "loss": 0.6408,
+      "step": 59274
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.1205623149871826,
+      "learning_rate": 3.1950232015027224e-08,
+      "loss": 0.3543,
+      "step": 59275
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.520616054534912,
+      "learning_rate": 3.192048026371797e-08,
+      "loss": 0.4489,
+      "step": 59276
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.9360899925231934,
+      "learning_rate": 3.1890742326777246e-08,
+      "loss": 0.4297,
+      "step": 59277
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.633300542831421,
+      "learning_rate": 3.1861018204287217e-08,
+      "loss": 0.3719,
+      "step": 59278
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.338192939758301,
+      "learning_rate": 3.18313078963306e-08,
+      "loss": 0.5911,
+      "step": 59279
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.930906057357788,
+      "learning_rate": 3.180161140298954e-08,
+      "loss": 0.5209,
+      "step": 59280
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.1733198165893555,
+      "learning_rate": 3.1771928724347866e-08,
+      "loss": 0.697,
+      "step": 59281
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.8555350303649902,
+      "learning_rate": 3.1742259860486623e-08,
+      "loss": 0.464,
+      "step": 59282
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.1353352069854736,
+      "learning_rate": 3.1712604811489076e-08,
+      "loss": 0.4006,
+      "step": 59283
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.0146753787994385,
+      "learning_rate": 3.168296357743794e-08,
+      "loss": 0.6143,
+      "step": 59284
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.4200737476348877,
+      "learning_rate": 3.165333615841537e-08,
+      "loss": 0.5089,
+      "step": 59285
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.805150032043457,
+      "learning_rate": 3.162372255450297e-08,
+      "loss": 0.5947,
+      "step": 59286
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.12545108795166,
+      "learning_rate": 3.159412276578455e-08,
+      "loss": 0.8791,
+      "step": 59287
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.656216859817505,
+      "learning_rate": 3.156453679234117e-08,
+      "loss": 0.5838,
+      "step": 59288
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.9410436153411865,
+      "learning_rate": 3.153496463425609e-08,
+      "loss": 0.7877,
+      "step": 59289
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.5164289474487305,
+      "learning_rate": 3.150540629161092e-08,
+      "loss": 0.5823,
+      "step": 59290
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.0767822265625,
+      "learning_rate": 3.14758617644878e-08,
+      "loss": 0.6542,
+      "step": 59291
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.155973434448242,
+      "learning_rate": 3.1446331052968906e-08,
+      "loss": 0.6868,
+      "step": 59292
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.9607064723968506,
+      "learning_rate": 3.141681415713693e-08,
+      "loss": 0.6465,
+      "step": 59293
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.8934738636016846,
+      "learning_rate": 3.138731107707349e-08,
+      "loss": 0.4349,
+      "step": 59294
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.475557327270508,
+      "learning_rate": 3.135782181286018e-08,
+      "loss": 0.6241,
+      "step": 59295
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.281531810760498,
+      "learning_rate": 3.132834636457971e-08,
+      "loss": 0.7399,
+      "step": 59296
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.91642427444458,
+      "learning_rate": 3.129888473231368e-08,
+      "loss": 0.6341,
+      "step": 59297
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 6.132034778594971,
+      "learning_rate": 3.126943691614426e-08,
+      "loss": 0.6246,
+      "step": 59298
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.813185214996338,
+      "learning_rate": 3.1240002916153034e-08,
+      "loss": 0.6323,
+      "step": 59299
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.5032105445861816,
+      "learning_rate": 3.1210582732422726e-08,
+      "loss": 0.5849,
+      "step": 59300
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.6921260356903076,
+      "learning_rate": 3.1181176365033815e-08,
+      "loss": 0.5541,
+      "step": 59301
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.1036911010742188,
+      "learning_rate": 3.1151783814069026e-08,
+      "loss": 0.6656,
+      "step": 59302
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.2623987197875977,
+      "learning_rate": 3.1122405079609396e-08,
+      "loss": 0.5207,
+      "step": 59303
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.344918727874756,
+      "learning_rate": 3.109304016173709e-08,
+      "loss": 0.5023,
+      "step": 59304
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.2331960201263428,
+      "learning_rate": 3.1063689060533696e-08,
+      "loss": 0.6784,
+      "step": 59305
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.112170457839966,
+      "learning_rate": 3.103435177608083e-08,
+      "loss": 0.516,
+      "step": 59306
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.9179043769836426,
+      "learning_rate": 3.100502830845953e-08,
+      "loss": 0.3482,
+      "step": 59307
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.8405518531799316,
+      "learning_rate": 3.0975718657752505e-08,
+      "loss": 0.8373,
+      "step": 59308
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.872053384780884,
+      "learning_rate": 3.094642282404026e-08,
+      "loss": 0.6397,
+      "step": 59309
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.9531466960906982,
+      "learning_rate": 3.091714080740493e-08,
+      "loss": 0.5373,
+      "step": 59310
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 4.628238677978516,
+      "learning_rate": 3.0887872607927025e-08,
+      "loss": 0.656,
+      "step": 59311
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 5.152879238128662,
+      "learning_rate": 3.0858618225688695e-08,
+      "loss": 0.5729,
+      "step": 59312
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.674858808517456,
+      "learning_rate": 3.0829377660770987e-08,
+      "loss": 0.7417,
+      "step": 59313
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.538830518722534,
+      "learning_rate": 3.08001509132555e-08,
+      "loss": 0.6376,
+      "step": 59314
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 4.083517074584961,
+      "learning_rate": 3.0770937983222724e-08,
+      "loss": 0.6859,
+      "step": 59315
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.4946935176849365,
+      "learning_rate": 3.074173887075482e-08,
+      "loss": 0.5981,
+      "step": 59316
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.9375648498535156,
+      "learning_rate": 3.071255357593284e-08,
+      "loss": 0.503,
+      "step": 59317
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.8379855155944824,
+      "learning_rate": 3.068338209883781e-08,
+      "loss": 0.6247,
+      "step": 59318
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 1.7767553329467773,
+      "learning_rate": 3.065422443955024e-08,
+      "loss": 0.413,
+      "step": 59319
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 4.703735828399658,
+      "learning_rate": 3.062508059815172e-08,
+      "loss": 0.6673,
+      "step": 59320
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.6070287227630615,
+      "learning_rate": 3.059595057472387e-08,
+      "loss": 0.8632,
+      "step": 59321
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.7156260013580322,
+      "learning_rate": 3.05668343693466e-08,
+      "loss": 0.5446,
+      "step": 59322
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.813795566558838,
+      "learning_rate": 3.0537731982101525e-08,
+      "loss": 0.5237,
+      "step": 59323
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 4.013780117034912,
+      "learning_rate": 3.050864341306914e-08,
+      "loss": 0.6314,
+      "step": 59324
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.7782442569732666,
+      "learning_rate": 3.047956866233104e-08,
+      "loss": 0.3855,
+      "step": 59325
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.993701934814453,
+      "learning_rate": 3.045050772996772e-08,
+      "loss": 0.6959,
+      "step": 59326
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 4.038589000701904,
+      "learning_rate": 3.0421460616059106e-08,
+      "loss": 0.6517,
+      "step": 59327
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.3182766437530518,
+      "learning_rate": 3.0392427320687923e-08,
+      "loss": 0.6147,
+      "step": 59328
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.505237579345703,
+      "learning_rate": 3.036340784393299e-08,
+      "loss": 0.5898,
+      "step": 59329
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.280663251876831,
+      "learning_rate": 3.033440218587591e-08,
+      "loss": 0.533,
+      "step": 59330
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.8610212802886963,
+      "learning_rate": 3.030541034659662e-08,
+      "loss": 0.5064,
+      "step": 59331
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.5215539932250977,
+      "learning_rate": 3.027643232617727e-08,
+      "loss": 0.6954,
+      "step": 59332
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.4909305572509766,
+      "learning_rate": 3.02474681246967e-08,
+      "loss": 0.4708,
+      "step": 59333
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 4.6770806312561035,
+      "learning_rate": 3.02185177422365e-08,
+      "loss": 0.7772,
+      "step": 59334
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.639414072036743,
+      "learning_rate": 3.0189581178877155e-08,
+      "loss": 0.4189,
+      "step": 59335
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 4.496167182922363,
+      "learning_rate": 3.016065843469862e-08,
+      "loss": 0.3927,
+      "step": 59336
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 4.199945449829102,
+      "learning_rate": 3.013174950978137e-08,
+      "loss": 0.6385,
+      "step": 59337
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.4975051879882812,
+      "learning_rate": 3.0102854404205904e-08,
+      "loss": 0.4733,
+      "step": 59338
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 4.240503787994385,
+      "learning_rate": 3.007397311805271e-08,
+      "loss": 0.5394,
+      "step": 59339
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 10.51459789276123,
+      "learning_rate": 3.004510565140228e-08,
+      "loss": 0.7282,
+      "step": 59340
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.9971678256988525,
+      "learning_rate": 3.0016252004334555e-08,
+      "loss": 0.5126,
+      "step": 59341
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.64528751373291,
+      "learning_rate": 2.998741217693002e-08,
+      "loss": 0.4819,
+      "step": 59342
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.357105016708374,
+      "learning_rate": 2.995858616926806e-08,
+      "loss": 0.5794,
+      "step": 59343
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 5.825497150421143,
+      "learning_rate": 2.992977398143026e-08,
+      "loss": 0.6934,
+      "step": 59344
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.8816909790039062,
+      "learning_rate": 2.9900975613495475e-08,
+      "loss": 0.5308,
+      "step": 59345
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.7429585456848145,
+      "learning_rate": 2.987219106554417e-08,
+      "loss": 0.3315,
+      "step": 59346
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 4.046050071716309,
+      "learning_rate": 2.9843420337656303e-08,
+      "loss": 0.6131,
+      "step": 59347
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 4.012269973754883,
+      "learning_rate": 2.981466342991235e-08,
+      "loss": 0.4098,
+      "step": 59348
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.3444371223449707,
+      "learning_rate": 2.9785920342391695e-08,
+      "loss": 0.5085,
+      "step": 59349
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.070103406906128,
+      "learning_rate": 2.975719107517483e-08,
+      "loss": 0.5667,
+      "step": 59350
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.9057564735412598,
+      "learning_rate": 2.9728475628341136e-08,
+      "loss": 0.3756,
+      "step": 59351
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.9763638973236084,
+      "learning_rate": 2.9699774001970548e-08,
+      "loss": 0.6568,
+      "step": 59352
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.491168737411499,
+      "learning_rate": 2.967108619614356e-08,
+      "loss": 0.784,
+      "step": 59353
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.3407421112060547,
+      "learning_rate": 2.9642412210938442e-08,
+      "loss": 0.4435,
+      "step": 59354
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 4.042481422424316,
+      "learning_rate": 2.9613752046436795e-08,
+      "loss": 0.6924,
+      "step": 59355
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.102670431137085,
+      "learning_rate": 2.958510570271633e-08,
+      "loss": 0.3375,
+      "step": 59356
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.4721949100494385,
+      "learning_rate": 2.95564731798581e-08,
+      "loss": 0.5703,
+      "step": 59357
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.8854761123657227,
+      "learning_rate": 2.9527854477941485e-08,
+      "loss": 0.5738,
+      "step": 59358
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.564845085144043,
+      "learning_rate": 2.9499249597045865e-08,
+      "loss": 0.4544,
+      "step": 59359
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.789808750152588,
+      "learning_rate": 2.9470658537251174e-08,
+      "loss": 0.657,
+      "step": 59360
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.002375602722168,
+      "learning_rate": 2.9442081298636238e-08,
+      "loss": 0.7205,
+      "step": 59361
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.0279502868652344,
+      "learning_rate": 2.9413517881280997e-08,
+      "loss": 0.5414,
+      "step": 59362
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.7387502193450928,
+      "learning_rate": 2.9384968285264826e-08,
+      "loss": 0.7019,
+      "step": 59363
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.7714335918426514,
+      "learning_rate": 2.9356432510666556e-08,
+      "loss": 0.6765,
+      "step": 59364
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 4.322212219238281,
+      "learning_rate": 2.9327910557566674e-08,
+      "loss": 0.7036,
+      "step": 59365
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.969815492630005,
+      "learning_rate": 2.92994024260429e-08,
+      "loss": 0.4221,
+      "step": 59366
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.1537346839904785,
+      "learning_rate": 2.9270908116176278e-08,
+      "loss": 0.5242,
+      "step": 59367
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.3225786685943604,
+      "learning_rate": 2.9242427628045078e-08,
+      "loss": 0.4219,
+      "step": 59368
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.158928155899048,
+      "learning_rate": 2.9213960961728127e-08,
+      "loss": 0.5143,
+      "step": 59369
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.722057342529297,
+      "learning_rate": 2.918550811730536e-08,
+      "loss": 0.6585,
+      "step": 59370
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.7853634357452393,
+      "learning_rate": 2.9157069094855606e-08,
+      "loss": 0.5348,
+      "step": 59371
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 5.559404373168945,
+      "learning_rate": 2.9128643894457688e-08,
+      "loss": 0.5069,
+      "step": 59372
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.9729292392730713,
+      "learning_rate": 2.910023251619154e-08,
+      "loss": 0.6003,
+      "step": 59373
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.9283595085144043,
+      "learning_rate": 2.907183496013488e-08,
+      "loss": 0.3791,
+      "step": 59374
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 6.279928207397461,
+      "learning_rate": 2.9043451226367648e-08,
+      "loss": 0.6411,
+      "step": 59375
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.0795063972473145,
+      "learning_rate": 2.9015081314967552e-08,
+      "loss": 0.6776,
+      "step": 59376
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.7881455421447754,
+      "learning_rate": 2.898672522601509e-08,
+      "loss": 0.6102,
+      "step": 59377
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.861947536468506,
+      "learning_rate": 2.8958382959587972e-08,
+      "loss": 0.4208,
+      "step": 59378
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.3833653926849365,
+      "learning_rate": 2.8930054515766136e-08,
+      "loss": 0.6255,
+      "step": 59379
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.0355234146118164,
+      "learning_rate": 2.8901739894626745e-08,
+      "loss": 0.467,
+      "step": 59380
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.1610889434814453,
+      "learning_rate": 2.887343909625029e-08,
+      "loss": 0.6042,
+      "step": 59381
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.670039653778076,
+      "learning_rate": 2.8845152120713928e-08,
+      "loss": 0.7087,
+      "step": 59382
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.1401898860931396,
+      "learning_rate": 2.881687896809704e-08,
+      "loss": 0.5189,
+      "step": 59383
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 4.485258102416992,
+      "learning_rate": 2.8788619638478455e-08,
+      "loss": 0.8253,
+      "step": 59384
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.187901258468628,
+      "learning_rate": 2.876037413193644e-08,
+      "loss": 0.4248,
+      "step": 59385
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 4.327296257019043,
+      "learning_rate": 2.8732142448549273e-08,
+      "loss": 0.3619,
+      "step": 59386
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.686096429824829,
+      "learning_rate": 2.870392458839577e-08,
+      "loss": 0.7523,
+      "step": 59387
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.8011393547058105,
+      "learning_rate": 2.867572055155421e-08,
+      "loss": 0.576,
+      "step": 59388
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.1951839923858643,
+      "learning_rate": 2.8647530338103413e-08,
+      "loss": 0.9165,
+      "step": 59389
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.5838844776153564,
+      "learning_rate": 2.8619353948121654e-08,
+      "loss": 0.5213,
+      "step": 59390
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.6633753776550293,
+      "learning_rate": 2.85911913816872e-08,
+      "loss": 0.6991,
+      "step": 59391
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.20257568359375,
+      "learning_rate": 2.856304263887777e-08,
+      "loss": 0.6754,
+      "step": 59392
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.2689125537872314,
+      "learning_rate": 2.8534907719772742e-08,
+      "loss": 0.4992,
+      "step": 59393
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.2809102535247803,
+      "learning_rate": 2.850678662444928e-08,
+      "loss": 0.5548,
+      "step": 59394
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 1.8397613763809204,
+      "learning_rate": 2.847867935298676e-08,
+      "loss": 0.1879,
+      "step": 59395
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.1936795711517334,
+      "learning_rate": 2.8450585905462347e-08,
+      "loss": 0.4776,
+      "step": 59396
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 5.865474224090576,
+      "learning_rate": 2.842250628195431e-08,
+      "loss": 0.6016,
+      "step": 59397
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.8047471046447754,
+      "learning_rate": 2.8394440482540364e-08,
+      "loss": 0.7424,
+      "step": 59398
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.47076678276062,
+      "learning_rate": 2.8366388507299893e-08,
+      "loss": 0.6732,
+      "step": 59399
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.21225643157959,
+      "learning_rate": 2.8338350356310052e-08,
+      "loss": 0.5911,
+      "step": 59400
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 4.496381759643555,
+      "learning_rate": 2.831032602964856e-08,
+      "loss": 0.7088,
+      "step": 59401
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 8.944236755371094,
+      "learning_rate": 2.828231552739369e-08,
+      "loss": 0.4566,
+      "step": 59402
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.6263999938964844,
+      "learning_rate": 2.8254318849623152e-08,
+      "loss": 0.6704,
+      "step": 59403
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.462470293045044,
+      "learning_rate": 2.8226335996415223e-08,
+      "loss": 0.2396,
+      "step": 59404
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.935577154159546,
+      "learning_rate": 2.8198366967847056e-08,
+      "loss": 0.6453,
+      "step": 59405
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.9249231815338135,
+      "learning_rate": 2.817041176399693e-08,
+      "loss": 0.5475,
+      "step": 59406
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.516597032546997,
+      "learning_rate": 2.8142470384942556e-08,
+      "loss": 0.6741,
+      "step": 59407
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.3879594802856445,
+      "learning_rate": 2.8114542830761094e-08,
+      "loss": 0.4859,
+      "step": 59408
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.077928066253662,
+      "learning_rate": 2.8086629101530815e-08,
+      "loss": 0.492,
+      "step": 59409
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.2913432121276855,
+      "learning_rate": 2.805872919732888e-08,
+      "loss": 0.7121,
+      "step": 59410
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.2873470783233643,
+      "learning_rate": 2.803084311823301e-08,
+      "loss": 0.7375,
+      "step": 59411
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.950200080871582,
+      "learning_rate": 2.800297086432091e-08,
+      "loss": 0.7884,
+      "step": 59412
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 4.383046627044678,
+      "learning_rate": 2.7975112435670305e-08,
+      "loss": 0.7443,
+      "step": 59413
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.7862229347229004,
+      "learning_rate": 2.794726783235835e-08,
+      "loss": 0.6544,
+      "step": 59414
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.6827428340911865,
+      "learning_rate": 2.7919437054462207e-08,
+      "loss": 0.5905,
+      "step": 59415
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 4.9070725440979,
+      "learning_rate": 2.7891620102059593e-08,
+      "loss": 0.8927,
+      "step": 59416
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.6641626358032227,
+      "learning_rate": 2.7863816975227663e-08,
+      "loss": 0.6472,
+      "step": 59417
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.1194403171539307,
+      "learning_rate": 2.7836027674043588e-08,
+      "loss": 0.6089,
+      "step": 59418
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.1816070079803467,
+      "learning_rate": 2.7808252198585628e-08,
+      "loss": 0.4999,
+      "step": 59419
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.5383212566375732,
+      "learning_rate": 2.7780490548929838e-08,
+      "loss": 0.6102,
+      "step": 59420
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.51865291595459,
+      "learning_rate": 2.7752742725153937e-08,
+      "loss": 0.5881,
+      "step": 59421
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.0858218669891357,
+      "learning_rate": 2.772500872733508e-08,
+      "loss": 0.5997,
+      "step": 59422
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 4.017990589141846,
+      "learning_rate": 2.7697288555550427e-08,
+      "loss": 0.6824,
+      "step": 59423
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.1026313304901123,
+      "learning_rate": 2.7669582209876588e-08,
+      "loss": 0.6551,
+      "step": 59424
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.5526139736175537,
+      "learning_rate": 2.7641889690391278e-08,
+      "loss": 0.602,
+      "step": 59425
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.939438581466675,
+      "learning_rate": 2.7614210997171098e-08,
+      "loss": 0.3961,
+      "step": 59426
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 4.054930686950684,
+      "learning_rate": 2.7586546130293214e-08,
+      "loss": 0.6199,
+      "step": 59427
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.889350414276123,
+      "learning_rate": 2.7558895089834225e-08,
+      "loss": 0.5562,
+      "step": 59428
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.3272511959075928,
+      "learning_rate": 2.75312578758713e-08,
+      "loss": 0.6162,
+      "step": 59429
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.76299786567688,
+      "learning_rate": 2.750363448848159e-08,
+      "loss": 0.7373,
+      "step": 59430
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.876035690307617,
+      "learning_rate": 2.7476024927740597e-08,
+      "loss": 0.6686,
+      "step": 59431
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 4.510998725891113,
+      "learning_rate": 2.7448429193727145e-08,
+      "loss": 0.5656,
+      "step": 59432
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 1.8225940465927124,
+      "learning_rate": 2.7420847286516172e-08,
+      "loss": 0.3631,
+      "step": 59433
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.5361242294311523,
+      "learning_rate": 2.7393279206185397e-08,
+      "loss": 0.7863,
+      "step": 59434
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.8472237586975098,
+      "learning_rate": 2.7365724952810868e-08,
+      "loss": 0.5778,
+      "step": 59435
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 7.07488489151001,
+      "learning_rate": 2.7338184526469747e-08,
+      "loss": 0.6468,
+      "step": 59436
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.6178908348083496,
+      "learning_rate": 2.7310657927237526e-08,
+      "loss": 0.4565,
+      "step": 59437
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.0563840866088867,
+      "learning_rate": 2.728314515519248e-08,
+      "loss": 0.5594,
+      "step": 59438
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.080031394958496,
+      "learning_rate": 2.7255646210409548e-08,
+      "loss": 0.7985,
+      "step": 59439
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.3658862113952637,
+      "learning_rate": 2.7228161092966444e-08,
+      "loss": 0.5707,
+      "step": 59440
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.7752845287323,
+      "learning_rate": 2.7200689802938662e-08,
+      "loss": 0.3977,
+      "step": 59441
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.0926835536956787,
+      "learning_rate": 2.717323234040281e-08,
+      "loss": 0.4809,
+      "step": 59442
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.142657518386841,
+      "learning_rate": 2.7145788705435494e-08,
+      "loss": 0.6308,
+      "step": 59443
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.595526933670044,
+      "learning_rate": 2.7118358898112763e-08,
+      "loss": 0.4954,
+      "step": 59444
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.6226558685302734,
+      "learning_rate": 2.709094291851122e-08,
+      "loss": 0.5238,
+      "step": 59445
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.377319574356079,
+      "learning_rate": 2.7063540766706363e-08,
+      "loss": 0.5229,
+      "step": 59446
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.9245996475219727,
+      "learning_rate": 2.7036152442775353e-08,
+      "loss": 0.5838,
+      "step": 59447
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.0214967727661133,
+      "learning_rate": 2.7008777946793684e-08,
+      "loss": 0.6401,
+      "step": 59448
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.261329174041748,
+      "learning_rate": 2.698141727883796e-08,
+      "loss": 0.5843,
+      "step": 59449
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.7851104736328125,
+      "learning_rate": 2.6954070438983683e-08,
+      "loss": 0.5485,
+      "step": 59450
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.4666476249694824,
+      "learning_rate": 2.6926737427307448e-08,
+      "loss": 0.594,
+      "step": 59451
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.5696041584014893,
+      "learning_rate": 2.6899418243884756e-08,
+      "loss": 0.45,
+      "step": 59452
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 4.435314655303955,
+      "learning_rate": 2.6872112888791658e-08,
+      "loss": 0.6566,
+      "step": 59453
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.533620834350586,
+      "learning_rate": 2.6844821362104758e-08,
+      "loss": 0.6085,
+      "step": 59454
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.836158037185669,
+      "learning_rate": 2.6817543663898993e-08,
+      "loss": 0.4147,
+      "step": 59455
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.7630155086517334,
+      "learning_rate": 2.6790279794250973e-08,
+      "loss": 0.8452,
+      "step": 59456
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 5.760504245758057,
+      "learning_rate": 2.676302975323619e-08,
+      "loss": 0.5985,
+      "step": 59457
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.596557855606079,
+      "learning_rate": 2.673579354092959e-08,
+      "loss": 0.6224,
+      "step": 59458
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.2363815307617188,
+      "learning_rate": 2.6708571157408325e-08,
+      "loss": 0.7206,
+      "step": 59459
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 4.063207149505615,
+      "learning_rate": 2.6681362602747896e-08,
+      "loss": 0.4094,
+      "step": 59460
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.597818613052368,
+      "learning_rate": 2.6654167877022687e-08,
+      "loss": 0.3959,
+      "step": 59461
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.253427028656006,
+      "learning_rate": 2.6626986980309856e-08,
+      "loss": 0.4717,
+      "step": 59462
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.196080446243286,
+      "learning_rate": 2.6599819912684345e-08,
+      "loss": 0.7578,
+      "step": 59463
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.6874616146087646,
+      "learning_rate": 2.6572666674221648e-08,
+      "loss": 0.4339,
+      "step": 59464
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.936471939086914,
+      "learning_rate": 2.6545527264996706e-08,
+      "loss": 0.5251,
+      "step": 59465
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.590977430343628,
+      "learning_rate": 2.6518401685086126e-08,
+      "loss": 0.5113,
+      "step": 59466
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.164036512374878,
+      "learning_rate": 2.649128993456429e-08,
+      "loss": 0.4821,
+      "step": 59467
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.4492781162261963,
+      "learning_rate": 2.646419201350725e-08,
+      "loss": 0.6574,
+      "step": 59468
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.2438929080963135,
+      "learning_rate": 2.64371079219905e-08,
+      "loss": 0.4419,
+      "step": 59469
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 4.004599094390869,
+      "learning_rate": 2.6410037660088426e-08,
+      "loss": 0.8693,
+      "step": 59470
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.1832163333892822,
+      "learning_rate": 2.638298122787708e-08,
+      "loss": 0.8965,
+      "step": 59471
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 6.155000686645508,
+      "learning_rate": 2.6355938625431954e-08,
+      "loss": 0.6447,
+      "step": 59472
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.29453444480896,
+      "learning_rate": 2.6328909852827435e-08,
+      "loss": 0.6589,
+      "step": 59473
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.671192169189453,
+      "learning_rate": 2.6301894910139015e-08,
+      "loss": 0.4841,
+      "step": 59474
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.74985671043396,
+      "learning_rate": 2.6274893797441637e-08,
+      "loss": 0.6195,
+      "step": 59475
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.5427117347717285,
+      "learning_rate": 2.6247906514810794e-08,
+      "loss": 0.4499,
+      "step": 59476
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.41542911529541,
+      "learning_rate": 2.622093306232143e-08,
+      "loss": 0.477,
+      "step": 59477
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.6361331939697266,
+      "learning_rate": 2.619397344004848e-08,
+      "loss": 0.6389,
+      "step": 59478
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.801921844482422,
+      "learning_rate": 2.6167027648066335e-08,
+      "loss": 0.5475,
+      "step": 59479
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.000967264175415,
+      "learning_rate": 2.614009568645104e-08,
+      "loss": 0.5661,
+      "step": 59480
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.9437930583953857,
+      "learning_rate": 2.6113177555276426e-08,
+      "loss": 0.587,
+      "step": 59481
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 4.2369794845581055,
+      "learning_rate": 2.6086273254618544e-08,
+      "loss": 0.6271,
+      "step": 59482
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.76304292678833,
+      "learning_rate": 2.605938278455067e-08,
+      "loss": 0.5821,
+      "step": 59483
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.9664154052734375,
+      "learning_rate": 2.6032506145148294e-08,
+      "loss": 0.3225,
+      "step": 59484
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.556091547012329,
+      "learning_rate": 2.600564333648692e-08,
+      "loss": 0.5858,
+      "step": 59485
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.44520902633667,
+      "learning_rate": 2.597879435864037e-08,
+      "loss": 0.4834,
+      "step": 59486
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.068467617034912,
+      "learning_rate": 2.595195921168303e-08,
+      "loss": 0.6272,
+      "step": 59487
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 6.204939842224121,
+      "learning_rate": 2.59251378956904e-08,
+      "loss": 0.503,
+      "step": 59488
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.5891387462615967,
+      "learning_rate": 2.589833041073686e-08,
+      "loss": 0.2547,
+      "step": 59489
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.692950487136841,
+      "learning_rate": 2.5871536756896243e-08,
+      "loss": 0.6742,
+      "step": 59490
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.934865713119507,
+      "learning_rate": 2.5844756934243485e-08,
+      "loss": 0.7861,
+      "step": 59491
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.385993719100952,
+      "learning_rate": 2.5817990942853533e-08,
+      "loss": 0.5352,
+      "step": 59492
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.199404001235962,
+      "learning_rate": 2.579123878280021e-08,
+      "loss": 0.6141,
+      "step": 59493
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 1.8975028991699219,
+      "learning_rate": 2.5764500454157903e-08,
+      "loss": 0.3916,
+      "step": 59494
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.191988706588745,
+      "learning_rate": 2.5737775957000998e-08,
+      "loss": 0.2542,
+      "step": 59495
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.233227014541626,
+      "learning_rate": 2.5711065291404436e-08,
+      "loss": 0.5626,
+      "step": 59496
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 5.532283306121826,
+      "learning_rate": 2.5684368457441488e-08,
+      "loss": 0.634,
+      "step": 59497
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.4210965633392334,
+      "learning_rate": 2.5657685455187098e-08,
+      "loss": 0.5725,
+      "step": 59498
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.932074785232544,
+      "learning_rate": 2.5631016284715092e-08,
+      "loss": 0.6649,
+      "step": 59499
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.5608677864074707,
+      "learning_rate": 2.560436094609986e-08,
+      "loss": 0.782,
+      "step": 59500
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.4238693714141846,
+      "learning_rate": 2.5577719439415784e-08,
+      "loss": 0.4708,
+      "step": 59501
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 4.7568840980529785,
+      "learning_rate": 2.5551091764736137e-08,
+      "loss": 0.5411,
+      "step": 59502
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.10764479637146,
+      "learning_rate": 2.552447792213586e-08,
+      "loss": 0.351,
+      "step": 59503
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.076582193374634,
+      "learning_rate": 2.5497877911688228e-08,
+      "loss": 0.5687,
+      "step": 59504
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.317638397216797,
+      "learning_rate": 2.547129173346763e-08,
+      "loss": 0.433,
+      "step": 59505
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.0813910961151123,
+      "learning_rate": 2.5444719387547888e-08,
+      "loss": 0.473,
+      "step": 59506
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 6.480198860168457,
+      "learning_rate": 2.5418160874002838e-08,
+      "loss": 0.465,
+      "step": 59507
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.413062334060669,
+      "learning_rate": 2.539161619290631e-08,
+      "loss": 0.4233,
+      "step": 59508
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.417026996612549,
+      "learning_rate": 2.536508534433213e-08,
+      "loss": 0.4889,
+      "step": 59509
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.9231374263763428,
+      "learning_rate": 2.533856832835413e-08,
+      "loss": 0.6306,
+      "step": 59510
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.7874035835266113,
+      "learning_rate": 2.531206514504614e-08,
+      "loss": 0.7171,
+      "step": 59511
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.303884744644165,
+      "learning_rate": 2.5285575794481987e-08,
+      "loss": 0.5557,
+      "step": 59512
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.0467095375061035,
+      "learning_rate": 2.5259100276734948e-08,
+      "loss": 0.4776,
+      "step": 59513
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 4.4697160720825195,
+      "learning_rate": 2.52326385918783e-08,
+      "loss": 0.4321,
+      "step": 59514
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 5.716630935668945,
+      "learning_rate": 2.5206190739986424e-08,
+      "loss": 0.4507,
+      "step": 59515
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.379099130630493,
+      "learning_rate": 2.5179756721132597e-08,
+      "loss": 0.512,
+      "step": 59516
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 10.525379180908203,
+      "learning_rate": 2.5153336535390095e-08,
+      "loss": 0.6291,
+      "step": 59517
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.262359857559204,
+      "learning_rate": 2.512693018283274e-08,
+      "loss": 0.4707,
+      "step": 59518
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.0682690143585205,
+      "learning_rate": 2.5100537663533264e-08,
+      "loss": 0.4871,
+      "step": 59519
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.9820473194122314,
+      "learning_rate": 2.5074158977566042e-08,
+      "loss": 0.5076,
+      "step": 59520
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.4106428623199463,
+      "learning_rate": 2.50477941250038e-08,
+      "loss": 0.6446,
+      "step": 59521
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 4.2626471519470215,
+      "learning_rate": 2.5021443105919808e-08,
+      "loss": 0.5481,
+      "step": 59522
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.5285820960998535,
+      "learning_rate": 2.4995105920387897e-08,
+      "loss": 0.6942,
+      "step": 59523
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.4512555599212646,
+      "learning_rate": 2.4968782568480786e-08,
+      "loss": 0.3724,
+      "step": 59524
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.2480955123901367,
+      "learning_rate": 2.4942473050271754e-08,
+      "loss": 0.5995,
+      "step": 59525
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.4381918907165527,
+      "learning_rate": 2.4916177365833516e-08,
+      "loss": 0.5101,
+      "step": 59526
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 4.2632575035095215,
+      "learning_rate": 2.488989551524046e-08,
+      "loss": 0.5783,
+      "step": 59527
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.324920892715454,
+      "learning_rate": 2.4863627498564192e-08,
+      "loss": 0.8046,
+      "step": 59528
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.730727195739746,
+      "learning_rate": 2.48373733158791e-08,
+      "loss": 0.5974,
+      "step": 59529
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.9121108055114746,
+      "learning_rate": 2.4811132967257345e-08,
+      "loss": 0.4786,
+      "step": 59530
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.8652853965759277,
+      "learning_rate": 2.4784906452772207e-08,
+      "loss": 0.4439,
+      "step": 59531
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.304839611053467,
+      "learning_rate": 2.4758693772495846e-08,
+      "loss": 0.7314,
+      "step": 59532
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.369647979736328,
+      "learning_rate": 2.473249492650265e-08,
+      "loss": 0.5126,
+      "step": 59533
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.876011371612549,
+      "learning_rate": 2.470630991486367e-08,
+      "loss": 0.6244,
+      "step": 59534
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.221524477005005,
+      "learning_rate": 2.468013873765329e-08,
+      "loss": 0.3437,
+      "step": 59535
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.6152589321136475,
+      "learning_rate": 2.465398139494368e-08,
+      "loss": 0.4433,
+      "step": 59536
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.5385870933532715,
+      "learning_rate": 2.4627837886807005e-08,
+      "loss": 0.6008,
+      "step": 59537
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.9468445777893066,
+      "learning_rate": 2.460170821331709e-08,
+      "loss": 0.4454,
+      "step": 59538
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.5482327938079834,
+      "learning_rate": 2.4575592374545542e-08,
+      "loss": 0.5976,
+      "step": 59539
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.5071449279785156,
+      "learning_rate": 2.4549490370565643e-08,
+      "loss": 0.3575,
+      "step": 59540
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.8210575580596924,
+      "learning_rate": 2.4523402201450107e-08,
+      "loss": 0.5004,
+      "step": 59541
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.1508371829986572,
+      "learning_rate": 2.449732786727055e-08,
+      "loss": 0.3528,
+      "step": 59542
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.6876726150512695,
+      "learning_rate": 2.4471267368100793e-08,
+      "loss": 0.5506,
+      "step": 59543
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 4.168923854827881,
+      "learning_rate": 2.4445220704011896e-08,
+      "loss": 0.8024,
+      "step": 59544
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.0527749061584473,
+      "learning_rate": 2.441918787507713e-08,
+      "loss": 0.5094,
+      "step": 59545
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.614018440246582,
+      "learning_rate": 2.4393168881368668e-08,
+      "loss": 0.5665,
+      "step": 59546
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.914036750793457,
+      "learning_rate": 2.4367163722958666e-08,
+      "loss": 0.561,
+      "step": 59547
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.8387351036071777,
+      "learning_rate": 2.43411723999204e-08,
+      "loss": 0.5624,
+      "step": 59548
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.2939724922180176,
+      "learning_rate": 2.4315194912324926e-08,
+      "loss": 0.5049,
+      "step": 59549
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.5801496505737305,
+      "learning_rate": 2.428923126024496e-08,
+      "loss": 0.7142,
+      "step": 59550
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.58335280418396,
+      "learning_rate": 2.4263281443752674e-08,
+      "loss": 0.6408,
+      "step": 59551
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.049302577972412,
+      "learning_rate": 2.4237345462920226e-08,
+      "loss": 0.4933,
+      "step": 59552
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.871887445449829,
+      "learning_rate": 2.4211423317819782e-08,
+      "loss": 0.8064,
+      "step": 59553
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.349405527114868,
+      "learning_rate": 2.4185515008523507e-08,
+      "loss": 0.6208,
+      "step": 59554
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.4961354732513428,
+      "learning_rate": 2.4159620535103566e-08,
+      "loss": 0.4568,
+      "step": 59555
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 4.115774154663086,
+      "learning_rate": 2.413373989763157e-08,
+      "loss": 0.5666,
+      "step": 59556
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 4.174051284790039,
+      "learning_rate": 2.4107873096179124e-08,
+      "loss": 0.6148,
+      "step": 59557
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.7723138332366943,
+      "learning_rate": 2.4082020130819507e-08,
+      "loss": 0.5564,
+      "step": 59558
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.006091356277466,
+      "learning_rate": 2.4056181001623215e-08,
+      "loss": 0.4449,
+      "step": 59559
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.5539135932922363,
+      "learning_rate": 2.403035570866241e-08,
+      "loss": 0.724,
+      "step": 59560
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 4.924902439117432,
+      "learning_rate": 2.4004544252009265e-08,
+      "loss": 0.7094,
+      "step": 59561
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.5855472087860107,
+      "learning_rate": 2.397874663173594e-08,
+      "loss": 0.5054,
+      "step": 59562
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.277923345565796,
+      "learning_rate": 2.3952962847913485e-08,
+      "loss": 0.4374,
+      "step": 59563
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.838923931121826,
+      "learning_rate": 2.3927192900613517e-08,
+      "loss": 0.6137,
+      "step": 59564
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.3794872760772705,
+      "learning_rate": 2.3901436789908195e-08,
+      "loss": 0.5468,
+      "step": 59565
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 4.594961166381836,
+      "learning_rate": 2.387569451586913e-08,
+      "loss": 0.4863,
+      "step": 59566
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 4.373183250427246,
+      "learning_rate": 2.3849966078566823e-08,
+      "loss": 0.4617,
+      "step": 59567
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.9030373096466064,
+      "learning_rate": 2.3824251478074543e-08,
+      "loss": 0.7249,
+      "step": 59568
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.8839235305786133,
+      "learning_rate": 2.379855071446224e-08,
+      "loss": 0.587,
+      "step": 59569
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.89658260345459,
+      "learning_rate": 2.3772863787802636e-08,
+      "loss": 0.6102,
+      "step": 59570
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.6093907356262207,
+      "learning_rate": 2.3747190698166224e-08,
+      "loss": 0.7246,
+      "step": 59571
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.3305459022521973,
+      "learning_rate": 2.3721531445624614e-08,
+      "loss": 0.5648,
+      "step": 59572
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.4026317596435547,
+      "learning_rate": 2.3695886030249415e-08,
+      "loss": 0.47,
+      "step": 59573
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.2708234786987305,
+      "learning_rate": 2.3670254452111685e-08,
+      "loss": 0.519,
+      "step": 59574
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.259045124053955,
+      "learning_rate": 2.364463671128303e-08,
+      "loss": 0.4811,
+      "step": 59575
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.5511295795440674,
+      "learning_rate": 2.3619032807834507e-08,
+      "loss": 0.4865,
+      "step": 59576
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.9563751220703125,
+      "learning_rate": 2.3593442741836614e-08,
+      "loss": 0.5767,
+      "step": 59577
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.7831850051879883,
+      "learning_rate": 2.356786651336207e-08,
+      "loss": 0.6171,
+      "step": 59578
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 4.450468063354492,
+      "learning_rate": 2.3542304122480264e-08,
+      "loss": 0.4606,
+      "step": 59579
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.6048007011413574,
+      "learning_rate": 2.351675556926336e-08,
+      "loss": 0.505,
+      "step": 59580
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 5.0251688957214355,
+      "learning_rate": 2.3491220853782416e-08,
+      "loss": 0.61,
+      "step": 59581
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.560997724533081,
+      "learning_rate": 2.3465699976107924e-08,
+      "loss": 0.4207,
+      "step": 59582
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.1814849376678467,
+      "learning_rate": 2.3440192936310946e-08,
+      "loss": 0.6523,
+      "step": 59583
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.7437570095062256,
+      "learning_rate": 2.341469973446253e-08,
+      "loss": 0.3472,
+      "step": 59584
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.5525295734405518,
+      "learning_rate": 2.3389220370633738e-08,
+      "loss": 0.5725,
+      "step": 59585
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.839320659637451,
+      "learning_rate": 2.336375484489506e-08,
+      "loss": 0.7317,
+      "step": 59586
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 4.352101802825928,
+      "learning_rate": 2.3338303157317554e-08,
+      "loss": 0.6866,
+      "step": 59587
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.3564345836639404,
+      "learning_rate": 2.3312865307972278e-08,
+      "loss": 0.61,
+      "step": 59588
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 4.66206169128418,
+      "learning_rate": 2.328744129692917e-08,
+      "loss": 0.361,
+      "step": 59589
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.649599313735962,
+      "learning_rate": 2.3262031124259843e-08,
+      "loss": 0.5685,
+      "step": 59590
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.1782195568084717,
+      "learning_rate": 2.323663479003424e-08,
+      "loss": 0.3787,
+      "step": 59591
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 6.310906410217285,
+      "learning_rate": 2.3211252294322862e-08,
+      "loss": 0.5205,
+      "step": 59592
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.71561598777771,
+      "learning_rate": 2.318588363719676e-08,
+      "loss": 0.5297,
+      "step": 59593
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.5165042877197266,
+      "learning_rate": 2.3160528818726992e-08,
+      "loss": 0.7359,
+      "step": 59594
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.3729522228240967,
+      "learning_rate": 2.3135187838982943e-08,
+      "loss": 0.5882,
+      "step": 59595
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.1464760303497314,
+      "learning_rate": 2.3109860698035115e-08,
+      "loss": 0.7774,
+      "step": 59596
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.6109933853149414,
+      "learning_rate": 2.3084547395955116e-08,
+      "loss": 0.5055,
+      "step": 59597
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.6878058910369873,
+      "learning_rate": 2.3059247932812335e-08,
+      "loss": 0.2875,
+      "step": 59598
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.0587825775146484,
+      "learning_rate": 2.3033962308677273e-08,
+      "loss": 0.5989,
+      "step": 59599
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.3210127353668213,
+      "learning_rate": 2.3008690523620424e-08,
+      "loss": 0.6664,
+      "step": 59600
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.983602523803711,
+      "learning_rate": 2.298343257771174e-08,
+      "loss": 0.4561,
+      "step": 59601
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 4.073328018188477,
+      "learning_rate": 2.295818847102227e-08,
+      "loss": 0.7501,
+      "step": 59602
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.1362709999084473,
+      "learning_rate": 2.2932958203621404e-08,
+      "loss": 0.7953,
+      "step": 59603
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.342865228652954,
+      "learning_rate": 2.2907741775579083e-08,
+      "loss": 0.5042,
+      "step": 59604
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 4.095401287078857,
+      "learning_rate": 2.288253918696637e-08,
+      "loss": 0.6982,
+      "step": 59605
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.3247599601745605,
+      "learning_rate": 2.2857350437852642e-08,
+      "loss": 0.7207,
+      "step": 59606
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.5689806938171387,
+      "learning_rate": 2.283217552830841e-08,
+      "loss": 0.6025,
+      "step": 59607
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.9344329833984375,
+      "learning_rate": 2.2807014458403052e-08,
+      "loss": 0.639,
+      "step": 59608
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.3635034561157227,
+      "learning_rate": 2.2781867228207076e-08,
+      "loss": 0.5811,
+      "step": 59609
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.6526124477386475,
+      "learning_rate": 2.275673383778987e-08,
+      "loss": 0.6591,
+      "step": 59610
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 4.022793292999268,
+      "learning_rate": 2.273161428722248e-08,
+      "loss": 0.492,
+      "step": 59611
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.7551615238189697,
+      "learning_rate": 2.2706508576573194e-08,
+      "loss": 0.5353,
+      "step": 59612
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.5284583568573,
+      "learning_rate": 2.2681416705913063e-08,
+      "loss": 0.7294,
+      "step": 59613
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 5.139306545257568,
+      "learning_rate": 2.2656338675310917e-08,
+      "loss": 0.4649,
+      "step": 59614
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.9301464557647705,
+      "learning_rate": 2.2631274484837262e-08,
+      "loss": 0.4197,
+      "step": 59615
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.8026297092437744,
+      "learning_rate": 2.2606224134561483e-08,
+      "loss": 0.7402,
+      "step": 59616
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.0876200199127197,
+      "learning_rate": 2.258118762455297e-08,
+      "loss": 0.539,
+      "step": 59617
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.7835354804992676,
+      "learning_rate": 2.2556164954881665e-08,
+      "loss": 0.5407,
+      "step": 59618
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.639235258102417,
+      "learning_rate": 2.253115612561696e-08,
+      "loss": 0.6201,
+      "step": 59619
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.0795786380767822,
+      "learning_rate": 2.2506161136828797e-08,
+      "loss": 0.4448,
+      "step": 59620
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.3739187717437744,
+      "learning_rate": 2.2481179988586012e-08,
+      "loss": 0.4843,
+      "step": 59621
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.824925184249878,
+      "learning_rate": 2.2456212680958543e-08,
+      "loss": 0.6507,
+      "step": 59622
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.580965518951416,
+      "learning_rate": 2.2431259214016345e-08,
+      "loss": 0.4919,
+      "step": 59623
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.8960845470428467,
+      "learning_rate": 2.2406319587827685e-08,
+      "loss": 0.6069,
+      "step": 59624
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.3674445152282715,
+      "learning_rate": 2.2381393802462514e-08,
+      "loss": 0.8932,
+      "step": 59625
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.8340344429016113,
+      "learning_rate": 2.2356481857989665e-08,
+      "loss": 0.7743,
+      "step": 59626
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.27876877784729,
+      "learning_rate": 2.233158375447908e-08,
+      "loss": 0.5355,
+      "step": 59627
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.001891851425171,
+      "learning_rate": 2.2306699491999596e-08,
+      "loss": 0.4438,
+      "step": 59628
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.1568262577056885,
+      "learning_rate": 2.2281829070620597e-08,
+      "loss": 0.6476,
+      "step": 59629
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.256066083908081,
+      "learning_rate": 2.2256972490410924e-08,
+      "loss": 0.5759,
+      "step": 59630
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.269243001937866,
+      "learning_rate": 2.2232129751439957e-08,
+      "loss": 0.7918,
+      "step": 59631
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.5427019596099854,
+      "learning_rate": 2.220730085377709e-08,
+      "loss": 0.4212,
+      "step": 59632
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.1711771488189697,
+      "learning_rate": 2.2182485797490604e-08,
+      "loss": 0.462,
+      "step": 59633
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.5082762241363525,
+      "learning_rate": 2.2157684582649884e-08,
+      "loss": 0.6271,
+      "step": 59634
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.5763747692108154,
+      "learning_rate": 2.213289720932432e-08,
+      "loss": 0.9058,
+      "step": 59635
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.191561698913574,
+      "learning_rate": 2.2108123677582194e-08,
+      "loss": 0.4673,
+      "step": 59636
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.309860944747925,
+      "learning_rate": 2.2083363987492334e-08,
+      "loss": 0.581,
+      "step": 59637
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.233091115951538,
+      "learning_rate": 2.2058618139124132e-08,
+      "loss": 0.3944,
+      "step": 59638
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.6982553005218506,
+      "learning_rate": 2.2033886132545867e-08,
+      "loss": 0.3395,
+      "step": 59639
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.440319299697876,
+      "learning_rate": 2.200916796782693e-08,
+      "loss": 0.5893,
+      "step": 59640
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.7671780586242676,
+      "learning_rate": 2.1984463645035593e-08,
+      "loss": 0.8168,
+      "step": 59641
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.358527898788452,
+      "learning_rate": 2.1959773164240694e-08,
+      "loss": 0.5049,
+      "step": 59642
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.783395290374756,
+      "learning_rate": 2.1935096525510512e-08,
+      "loss": 0.5178,
+      "step": 59643
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.777637004852295,
+      "learning_rate": 2.191043372891444e-08,
+      "loss": 0.493,
+      "step": 59644
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.3669159412384033,
+      "learning_rate": 2.1885784774520193e-08,
+      "loss": 0.4078,
+      "step": 59645
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.182408094406128,
+      "learning_rate": 2.1861149662397163e-08,
+      "loss": 0.5801,
+      "step": 59646
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.773298740386963,
+      "learning_rate": 2.1836528392613076e-08,
+      "loss": 0.7469,
+      "step": 59647
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.471219062805176,
+      "learning_rate": 2.1811920965236767e-08,
+      "loss": 0.5557,
+      "step": 59648
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.6909050941467285,
+      "learning_rate": 2.1787327380337063e-08,
+      "loss": 0.4234,
+      "step": 59649
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.151913642883301,
+      "learning_rate": 2.1762747637981695e-08,
+      "loss": 0.5271,
+      "step": 59650
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.655890703201294,
+      "learning_rate": 2.173818173823894e-08,
+      "loss": 0.5471,
+      "step": 59651
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.6017754077911377,
+      "learning_rate": 2.171362968117763e-08,
+      "loss": 0.5436,
+      "step": 59652
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.4655611515045166,
+      "learning_rate": 2.168909146686604e-08,
+      "loss": 0.787,
+      "step": 59653
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.781235933303833,
+      "learning_rate": 2.1664567095371903e-08,
+      "loss": 0.5634,
+      "step": 59654
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.2526700496673584,
+      "learning_rate": 2.1640056566764044e-08,
+      "loss": 0.4747,
+      "step": 59655
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.224637746810913,
+      "learning_rate": 2.1615559881109637e-08,
+      "loss": 0.6186,
+      "step": 59656
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.237983465194702,
+      "learning_rate": 2.1591077038477513e-08,
+      "loss": 0.5304,
+      "step": 59657
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.780505895614624,
+      "learning_rate": 2.1566608038935955e-08,
+      "loss": 0.8342,
+      "step": 59658
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.415418863296509,
+      "learning_rate": 2.154215288255268e-08,
+      "loss": 0.7416,
+      "step": 59659
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.694291114807129,
+      "learning_rate": 2.1517711569395418e-08,
+      "loss": 0.5973,
+      "step": 59660
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.358180284500122,
+      "learning_rate": 2.149328409953244e-08,
+      "loss": 0.69,
+      "step": 59661
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.085411787033081,
+      "learning_rate": 2.1468870473032035e-08,
+      "loss": 0.4828,
+      "step": 59662
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 4.373777866363525,
+      "learning_rate": 2.1444470689961362e-08,
+      "loss": 0.5575,
+      "step": 59663
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.288801431655884,
+      "learning_rate": 2.1420084750388704e-08,
+      "loss": 0.5973,
+      "step": 59664
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.390193462371826,
+      "learning_rate": 2.1395712654381785e-08,
+      "loss": 0.6452,
+      "step": 59665
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.1737277507781982,
+      "learning_rate": 2.1371354402008325e-08,
+      "loss": 0.4557,
+      "step": 59666
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.9579970836639404,
+      "learning_rate": 2.1347009993336053e-08,
+      "loss": 0.7962,
+      "step": 59667
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.3883328437805176,
+      "learning_rate": 2.132267942843269e-08,
+      "loss": 0.5481,
+      "step": 59668
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 4.380234718322754,
+      "learning_rate": 2.1298362707365962e-08,
+      "loss": 0.4132,
+      "step": 59669
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 5.6771650314331055,
+      "learning_rate": 2.1274059830203585e-08,
+      "loss": 0.5769,
+      "step": 59670
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.602938175201416,
+      "learning_rate": 2.1249770797012738e-08,
+      "loss": 0.6313,
+      "step": 59671
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 4.335402488708496,
+      "learning_rate": 2.122549560786169e-08,
+      "loss": 0.4334,
+      "step": 59672
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.5149905681610107,
+      "learning_rate": 2.120123426281706e-08,
+      "loss": 0.5091,
+      "step": 59673
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.412027597427368,
+      "learning_rate": 2.1176986761946572e-08,
+      "loss": 0.3979,
+      "step": 59674
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.138176679611206,
+      "learning_rate": 2.1152753105317947e-08,
+      "loss": 0.4846,
+      "step": 59675
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.881763458251953,
+      "learning_rate": 2.1128533292998355e-08,
+      "loss": 0.4855,
+      "step": 59676
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.572234869003296,
+      "learning_rate": 2.110432732505496e-08,
+      "loss": 0.4848,
+      "step": 59677
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.6682839393615723,
+      "learning_rate": 2.1080135201555498e-08,
+      "loss": 0.5737,
+      "step": 59678
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.1619462966918945,
+      "learning_rate": 2.1055956922567123e-08,
+      "loss": 0.645,
+      "step": 59679
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.230977773666382,
+      "learning_rate": 2.103179248815701e-08,
+      "loss": 0.4213,
+      "step": 59680
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 6.5307793617248535,
+      "learning_rate": 2.1007641898392327e-08,
+      "loss": 0.8725,
+      "step": 59681
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.7568206787109375,
+      "learning_rate": 2.098350515333969e-08,
+      "loss": 0.6967,
+      "step": 59682
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.7308349609375,
+      "learning_rate": 2.0959382253067374e-08,
+      "loss": 0.4892,
+      "step": 59683
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.292100667953491,
+      "learning_rate": 2.0935273197641436e-08,
+      "loss": 0.4058,
+      "step": 59684
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.3832521438598633,
+      "learning_rate": 2.0911177987129604e-08,
+      "loss": 0.5503,
+      "step": 59685
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.9038708209991455,
+      "learning_rate": 2.088709662159849e-08,
+      "loss": 0.5484,
+      "step": 59686
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.8922080993652344,
+      "learning_rate": 2.0863029101114706e-08,
+      "loss": 0.7132,
+      "step": 59687
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 9.252827644348145,
+      "learning_rate": 2.0838975425745978e-08,
+      "loss": 0.6654,
+      "step": 59688
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.9433348178863525,
+      "learning_rate": 2.0814935595558918e-08,
+      "loss": 0.5526,
+      "step": 59689
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.5788588523864746,
+      "learning_rate": 2.0790909610620137e-08,
+      "loss": 0.6267,
+      "step": 59690
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.127065896987915,
+      "learning_rate": 2.0766897470996806e-08,
+      "loss": 0.818,
+      "step": 59691
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.5785200595855713,
+      "learning_rate": 2.0742899176755537e-08,
+      "loss": 0.4629,
+      "step": 59692
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.6856446266174316,
+      "learning_rate": 2.071891472796239e-08,
+      "loss": 0.4239,
+      "step": 59693
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.6023950576782227,
+      "learning_rate": 2.069494412468509e-08,
+      "loss": 0.6835,
+      "step": 59694
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.393876075744629,
+      "learning_rate": 2.0670987366989693e-08,
+      "loss": 0.5254,
+      "step": 59695
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.9884204864501953,
+      "learning_rate": 2.0647044454943365e-08,
+      "loss": 0.3828,
+      "step": 59696
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.6012449264526367,
+      "learning_rate": 2.062311538861217e-08,
+      "loss": 0.4852,
+      "step": 59697
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.1977062225341797,
+      "learning_rate": 2.0599200168062718e-08,
+      "loss": 0.8189,
+      "step": 59698
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.9948437213897705,
+      "learning_rate": 2.0575298793361065e-08,
+      "loss": 0.6357,
+      "step": 59699
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.4977564811706543,
+      "learning_rate": 2.055141126457494e-08,
+      "loss": 0.6802,
+      "step": 59700
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.9300999641418457,
+      "learning_rate": 2.0527537581769285e-08,
+      "loss": 0.7722,
+      "step": 59701
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.340653896331787,
+      "learning_rate": 2.0503677745011828e-08,
+      "loss": 0.4581,
+      "step": 59702
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.339027166366577,
+      "learning_rate": 2.0479831754367517e-08,
+      "loss": 0.5083,
+      "step": 59703
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.2429561614990234,
+      "learning_rate": 2.0455999609904076e-08,
+      "loss": 0.4474,
+      "step": 59704
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.8626880645751953,
+      "learning_rate": 2.0432181311687004e-08,
+      "loss": 0.5473,
+      "step": 59705
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.9132840633392334,
+      "learning_rate": 2.0408376859782365e-08,
+      "loss": 0.5476,
+      "step": 59706
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.9188201427459717,
+      "learning_rate": 2.0384586254256766e-08,
+      "loss": 0.6339,
+      "step": 59707
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.5776207447052,
+      "learning_rate": 2.0360809495176826e-08,
+      "loss": 0.7424,
+      "step": 59708
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.590808153152466,
+      "learning_rate": 2.033704658260749e-08,
+      "loss": 0.43,
+      "step": 59709
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.4016287326812744,
+      "learning_rate": 2.031329751661537e-08,
+      "loss": 0.738,
+      "step": 59710
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 4.028809547424316,
+      "learning_rate": 2.0289562297266528e-08,
+      "loss": 0.78,
+      "step": 59711
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.3380401134490967,
+      "learning_rate": 2.026584092462702e-08,
+      "loss": 0.6336,
+      "step": 59712
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.2494730949401855,
+      "learning_rate": 2.0242133398762908e-08,
+      "loss": 0.4415,
+      "step": 59713
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.5968668460845947,
+      "learning_rate": 2.021843971973969e-08,
+      "loss": 0.5829,
+      "step": 59714
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.3710458278656006,
+      "learning_rate": 2.0194759887623983e-08,
+      "loss": 0.7201,
+      "step": 59715
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.748180627822876,
+      "learning_rate": 2.0171093902480732e-08,
+      "loss": 0.5413,
+      "step": 59716
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.2980024814605713,
+      "learning_rate": 2.014744176437655e-08,
+      "loss": 0.5161,
+      "step": 59717
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.977478504180908,
+      "learning_rate": 2.0123803473376392e-08,
+      "loss": 0.5059,
+      "step": 59718
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 4.218637943267822,
+      "learning_rate": 2.0100179029546864e-08,
+      "loss": 0.5541,
+      "step": 59719
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.59694242477417,
+      "learning_rate": 2.0076568432952913e-08,
+      "loss": 0.7065,
+      "step": 59720
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.7814223766326904,
+      "learning_rate": 2.0052971683660606e-08,
+      "loss": 0.6459,
+      "step": 59721
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.3707387447357178,
+      "learning_rate": 2.0029388781735437e-08,
+      "loss": 0.6301,
+      "step": 59722
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.82049298286438,
+      "learning_rate": 2.0005819727242913e-08,
+      "loss": 0.6548,
+      "step": 59723
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 4.410541534423828,
+      "learning_rate": 1.9982264520249097e-08,
+      "loss": 0.58,
+      "step": 59724
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.5700910091400146,
+      "learning_rate": 1.995872316081837e-08,
+      "loss": 0.6703,
+      "step": 59725
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.934980869293213,
+      "learning_rate": 1.993519564901736e-08,
+      "loss": 0.5743,
+      "step": 59726
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.9809834957122803,
+      "learning_rate": 1.9911681984910447e-08,
+      "loss": 0.6251,
+      "step": 59727
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.020476818084717,
+      "learning_rate": 1.9888182168564252e-08,
+      "loss": 0.6872,
+      "step": 59728
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.6674587726593018,
+      "learning_rate": 1.9864696200043164e-08,
+      "loss": 0.6183,
+      "step": 59729
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.7353265285491943,
+      "learning_rate": 1.984122407941269e-08,
+      "loss": 0.667,
+      "step": 59730
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.4853515625,
+      "learning_rate": 1.9817765806738333e-08,
+      "loss": 0.726,
+      "step": 59731
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.0568602085113525,
+      "learning_rate": 1.9794321382085036e-08,
+      "loss": 0.3627,
+      "step": 59732
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.168888568878174,
+      "learning_rate": 1.9770890805517752e-08,
+      "loss": 0.4175,
+      "step": 59733
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.8459553718566895,
+      "learning_rate": 1.9747474077102534e-08,
+      "loss": 0.4388,
+      "step": 59734
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.7447593212127686,
+      "learning_rate": 1.972407119690378e-08,
+      "loss": 0.5937,
+      "step": 59735
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.962245464324951,
+      "learning_rate": 1.970068216498644e-08,
+      "loss": 0.6492,
+      "step": 59736
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.373412609100342,
+      "learning_rate": 1.9677306981416567e-08,
+      "loss": 0.5773,
+      "step": 59737
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.511674165725708,
+      "learning_rate": 1.9653945646257443e-08,
+      "loss": 0.5439,
+      "step": 59738
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.563443183898926,
+      "learning_rate": 1.9630598159575687e-08,
+      "loss": 0.5255,
+      "step": 59739
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.1190590858459473,
+      "learning_rate": 1.9607264521435133e-08,
+      "loss": 0.6363,
+      "step": 59740
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.6631457805633545,
+      "learning_rate": 1.9583944731901837e-08,
+      "loss": 0.8428,
+      "step": 59741
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.9719231128692627,
+      "learning_rate": 1.956063879103909e-08,
+      "loss": 0.4659,
+      "step": 59742
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.766260862350464,
+      "learning_rate": 1.953734669891294e-08,
+      "loss": 0.5449,
+      "step": 59743
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.0710015296936035,
+      "learning_rate": 1.9514068455587233e-08,
+      "loss": 0.4845,
+      "step": 59744
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 8.258831024169922,
+      "learning_rate": 1.9490804061127465e-08,
+      "loss": 0.5062,
+      "step": 59745
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.0216217041015625,
+      "learning_rate": 1.9467553515598037e-08,
+      "loss": 0.5081,
+      "step": 59746
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.0989532470703125,
+      "learning_rate": 1.944431681906389e-08,
+      "loss": 0.5226,
+      "step": 59747
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 1.7541248798370361,
+      "learning_rate": 1.9421093971588866e-08,
+      "loss": 0.3951,
+      "step": 59748
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.2651405334472656,
+      "learning_rate": 1.9397884973238466e-08,
+      "loss": 0.3994,
+      "step": 59749
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.7029731273651123,
+      "learning_rate": 1.9374689824075975e-08,
+      "loss": 0.7737,
+      "step": 59750
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.4366652965545654,
+      "learning_rate": 1.9351508524167452e-08,
+      "loss": 0.4535,
+      "step": 59751
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.5240566730499268,
+      "learning_rate": 1.9328341073576173e-08,
+      "loss": 0.4379,
+      "step": 59752
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.220665454864502,
+      "learning_rate": 1.930518747236765e-08,
+      "loss": 0.4787,
+      "step": 59753
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.968742847442627,
+      "learning_rate": 1.9282047720604603e-08,
+      "loss": 0.7076,
+      "step": 59754
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.207420587539673,
+      "learning_rate": 1.9258921818353095e-08,
+      "loss": 0.5117,
+      "step": 59755
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.505528450012207,
+      "learning_rate": 1.9235809765676405e-08,
+      "loss": 0.6062,
+      "step": 59756
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.5839645862579346,
+      "learning_rate": 1.9212711562639486e-08,
+      "loss": 0.5618,
+      "step": 59757
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.5730667114257812,
+      "learning_rate": 1.9189627209305615e-08,
+      "loss": 0.5877,
+      "step": 59758
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 5.531312942504883,
+      "learning_rate": 1.9166556705739748e-08,
+      "loss": 0.651,
+      "step": 59759
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.1466808319091797,
+      "learning_rate": 1.9143500052005714e-08,
+      "loss": 0.7415,
+      "step": 59760
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.019054412841797,
+      "learning_rate": 1.912045724816791e-08,
+      "loss": 0.6889,
+      "step": 59761
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.6565074920654297,
+      "learning_rate": 1.9097428294290175e-08,
+      "loss": 0.4036,
+      "step": 59762
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.711674451828003,
+      "learning_rate": 1.9074413190436348e-08,
+      "loss": 0.3678,
+      "step": 59763
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.0780487060546875,
+      "learning_rate": 1.9051411936670816e-08,
+      "loss": 0.886,
+      "step": 59764
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.7517800331115723,
+      "learning_rate": 1.902842453305742e-08,
+      "loss": 0.5176,
+      "step": 59765
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.235942840576172,
+      "learning_rate": 1.9005450979660557e-08,
+      "loss": 0.4853,
+      "step": 59766
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.4137959480285645,
+      "learning_rate": 1.8982491276542948e-08,
+      "loss": 0.442,
+      "step": 59767
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 4.565796375274658,
+      "learning_rate": 1.8959545423769544e-08,
+      "loss": 0.7831,
+      "step": 59768
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.5979480743408203,
+      "learning_rate": 1.8936613421403628e-08,
+      "loss": 0.4717,
+      "step": 59769
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.0239906311035156,
+      "learning_rate": 1.8913695269508482e-08,
+      "loss": 0.4756,
+      "step": 59770
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.281949758529663,
+      "learning_rate": 1.8890790968149053e-08,
+      "loss": 0.698,
+      "step": 59771
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 8.619589805603027,
+      "learning_rate": 1.886790051738807e-08,
+      "loss": 0.7242,
+      "step": 59772
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.196082353591919,
+      "learning_rate": 1.8845023917289928e-08,
+      "loss": 0.4123,
+      "step": 59773
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.103896379470825,
+      "learning_rate": 1.8822161167917353e-08,
+      "loss": 0.776,
+      "step": 59774
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.982239007949829,
+      "learning_rate": 1.8799312269334182e-08,
+      "loss": 0.579,
+      "step": 59775
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.9708516597747803,
+      "learning_rate": 1.877647722160425e-08,
+      "loss": 0.5978,
+      "step": 59776
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 14.891746520996094,
+      "learning_rate": 1.87536560247914e-08,
+      "loss": 0.8343,
+      "step": 59777
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.4975736141204834,
+      "learning_rate": 1.873084867895836e-08,
+      "loss": 0.4409,
+      "step": 59778
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.449812412261963,
+      "learning_rate": 1.8708055184168404e-08,
+      "loss": 0.7787,
+      "step": 59779
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.8793089389801025,
+      "learning_rate": 1.868527554048538e-08,
+      "loss": 0.6602,
+      "step": 59780
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.195525646209717,
+      "learning_rate": 1.8662509747973124e-08,
+      "loss": 0.7063,
+      "step": 59781
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.3851332664489746,
+      "learning_rate": 1.8639757806693803e-08,
+      "loss": 0.5769,
+      "step": 59782
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.830829381942749,
+      "learning_rate": 1.8617019716711816e-08,
+      "loss": 0.5729,
+      "step": 59783
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 6.215850830078125,
+      "learning_rate": 1.859429547808933e-08,
+      "loss": 0.5483,
+      "step": 59784
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.8205549716949463,
+      "learning_rate": 1.8571585090890188e-08,
+      "loss": 0.6402,
+      "step": 59785
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.445939302444458,
+      "learning_rate": 1.8548888555177115e-08,
+      "loss": 0.4093,
+      "step": 59786
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.966569662094116,
+      "learning_rate": 1.8526205871013948e-08,
+      "loss": 0.7296,
+      "step": 59787
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.1407053470611572,
+      "learning_rate": 1.8503537038463415e-08,
+      "loss": 0.4249,
+      "step": 59788
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.440293550491333,
+      "learning_rate": 1.8480882057588245e-08,
+      "loss": 0.5389,
+      "step": 59789
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.3910250663757324,
+      "learning_rate": 1.8458240928451166e-08,
+      "loss": 0.606,
+      "step": 59790
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.7139017581939697,
+      "learning_rate": 1.8435613651116013e-08,
+      "loss": 0.7544,
+      "step": 59791
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.3736066818237305,
+      "learning_rate": 1.8413000225645515e-08,
+      "loss": 0.6402,
+      "step": 59792
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.109487295150757,
+      "learning_rate": 1.8390400652101847e-08,
+      "loss": 0.5768,
+      "step": 59793
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.197208881378174,
+      "learning_rate": 1.8367814930548288e-08,
+      "loss": 0.5536,
+      "step": 59794
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.530092477798462,
+      "learning_rate": 1.8345243061048124e-08,
+      "loss": 0.4851,
+      "step": 59795
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.338576078414917,
+      "learning_rate": 1.8322685043663523e-08,
+      "loss": 0.4842,
+      "step": 59796
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 5.010489463806152,
+      "learning_rate": 1.8300140878457218e-08,
+      "loss": 0.5701,
+      "step": 59797
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.8097310066223145,
+      "learning_rate": 1.8277610565491932e-08,
+      "loss": 0.3655,
+      "step": 59798
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.6720831394195557,
+      "learning_rate": 1.8255094104830396e-08,
+      "loss": 0.4098,
+      "step": 59799
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.4648680686950684,
+      "learning_rate": 1.8232591496535333e-08,
+      "loss": 0.7051,
+      "step": 59800
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.0882561206817627,
+      "learning_rate": 1.8210102740669477e-08,
+      "loss": 0.4817,
+      "step": 59801
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.43757700920105,
+      "learning_rate": 1.8187627837294997e-08,
+      "loss": 0.4953,
+      "step": 59802
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 4.267410755157471,
+      "learning_rate": 1.816516678647462e-08,
+      "loss": 0.2232,
+      "step": 59803
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.2927463054656982,
+      "learning_rate": 1.8142719588269963e-08,
+      "loss": 0.4466,
+      "step": 59804
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.483349084854126,
+      "learning_rate": 1.8120286242744867e-08,
+      "loss": 0.4779,
+      "step": 59805
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 4.022871971130371,
+      "learning_rate": 1.8097866749960945e-08,
+      "loss": 0.6411,
+      "step": 59806
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.8698253631591797,
+      "learning_rate": 1.8075461109980375e-08,
+      "loss": 0.4892,
+      "step": 59807
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.36053204536438,
+      "learning_rate": 1.805306932286588e-08,
+      "loss": 0.5103,
+      "step": 59808
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 5.53071928024292,
+      "learning_rate": 1.803069138867908e-08,
+      "loss": 0.4285,
+      "step": 59809
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.672168016433716,
+      "learning_rate": 1.8008327307483253e-08,
+      "loss": 0.4944,
+      "step": 59810
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.9187676906585693,
+      "learning_rate": 1.7985977079339468e-08,
+      "loss": 0.525,
+      "step": 59811
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 4.077964782714844,
+      "learning_rate": 1.796364070431045e-08,
+      "loss": 0.6531,
+      "step": 59812
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.786529302597046,
+      "learning_rate": 1.794131818245837e-08,
+      "loss": 0.5877,
+      "step": 59813
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.285959243774414,
+      "learning_rate": 1.7919009513845397e-08,
+      "loss": 0.7376,
+      "step": 59814
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 4.43513298034668,
+      "learning_rate": 1.7896714698532603e-08,
+      "loss": 0.4681,
+      "step": 59815
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.9665307998657227,
+      "learning_rate": 1.787443373658382e-08,
+      "loss": 0.6511,
+      "step": 59816
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.256122589111328,
+      "learning_rate": 1.7852166628059e-08,
+      "loss": 0.453,
+      "step": 59817
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.311070203781128,
+      "learning_rate": 1.7829913373021422e-08,
+      "loss": 0.6674,
+      "step": 59818
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.9447906017303467,
+      "learning_rate": 1.7807673971532157e-08,
+      "loss": 0.672,
+      "step": 59819
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.0086214542388916,
+      "learning_rate": 1.7785448423653373e-08,
+      "loss": 0.7038,
+      "step": 59820
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.540814161300659,
+      "learning_rate": 1.7763236729446686e-08,
+      "loss": 0.463,
+      "step": 59821
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 4.304352760314941,
+      "learning_rate": 1.7741038888974828e-08,
+      "loss": 0.5801,
+      "step": 59822
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.2117302417755127,
+      "learning_rate": 1.7718854902298298e-08,
+      "loss": 0.6074,
+      "step": 59823
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.003777503967285,
+      "learning_rate": 1.769668476947872e-08,
+      "loss": 0.7781,
+      "step": 59824
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.3716609477996826,
+      "learning_rate": 1.7674528490578824e-08,
+      "loss": 0.5093,
+      "step": 59825
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.99182391166687,
+      "learning_rate": 1.7652386065659667e-08,
+      "loss": 0.6221,
+      "step": 59826
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.6032955646514893,
+      "learning_rate": 1.763025749478231e-08,
+      "loss": 0.6036,
+      "step": 59827
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.437734603881836,
+      "learning_rate": 1.7608142778009483e-08,
+      "loss": 0.6393,
+      "step": 59828
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.995753288269043,
+      "learning_rate": 1.7586041915401697e-08,
+      "loss": 0.7306,
+      "step": 59829
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.1163930892944336,
+      "learning_rate": 1.7563954907020565e-08,
+      "loss": 0.8238,
+      "step": 59830
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 1.5911948680877686,
+      "learning_rate": 1.7541881752927704e-08,
+      "loss": 0.3911,
+      "step": 59831
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.7393600940704346,
+      "learning_rate": 1.7519822453184177e-08,
+      "loss": 0.8941,
+      "step": 59832
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.5444276332855225,
+      "learning_rate": 1.7497777007851602e-08,
+      "loss": 0.6264,
+      "step": 59833
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.9236502647399902,
+      "learning_rate": 1.7475745416991595e-08,
+      "loss": 0.5285,
+      "step": 59834
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.8470773696899414,
+      "learning_rate": 1.7453727680664666e-08,
+      "loss": 0.3857,
+      "step": 59835
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 4.42864990234375,
+      "learning_rate": 1.7431723798932432e-08,
+      "loss": 0.5884,
+      "step": 59836
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.0935568809509277,
+      "learning_rate": 1.7409733771855953e-08,
+      "loss": 0.5869,
+      "step": 59837
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.4795515537261963,
+      "learning_rate": 1.7387757599496845e-08,
+      "loss": 0.775,
+      "step": 59838
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.443577289581299,
+      "learning_rate": 1.736579528191562e-08,
+      "loss": 0.604,
+      "step": 59839
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.699171304702759,
+      "learning_rate": 1.734384681917389e-08,
+      "loss": 0.5456,
+      "step": 59840
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.4727470874786377,
+      "learning_rate": 1.732191221133217e-08,
+      "loss": 0.4496,
+      "step": 59841
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.407471179962158,
+      "learning_rate": 1.7299991458451514e-08,
+      "loss": 0.7053,
+      "step": 59842
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.758028507232666,
+      "learning_rate": 1.7278084560592988e-08,
+      "loss": 0.3229,
+      "step": 59843
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 4.551173686981201,
+      "learning_rate": 1.7256191517817655e-08,
+      "loss": 0.7049,
+      "step": 59844
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 4.489675521850586,
+      "learning_rate": 1.723431233018602e-08,
+      "loss": 0.5937,
+      "step": 59845
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.6180787086486816,
+      "learning_rate": 1.7212446997759703e-08,
+      "loss": 0.5037,
+      "step": 59846
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.9833571910858154,
+      "learning_rate": 1.7190595520598653e-08,
+      "loss": 0.6359,
+      "step": 59847
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.174050807952881,
+      "learning_rate": 1.7168757898764487e-08,
+      "loss": 0.4989,
+      "step": 59848
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.8056981563568115,
+      "learning_rate": 1.7146934132316605e-08,
+      "loss": 0.4647,
+      "step": 59849
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.67460298538208,
+      "learning_rate": 1.7125124221317178e-08,
+      "loss": 0.5577,
+      "step": 59850
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.9427716732025146,
+      "learning_rate": 1.7103328165825605e-08,
+      "loss": 0.7229,
+      "step": 59851
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 4.343594074249268,
+      "learning_rate": 1.7081545965902945e-08,
+      "loss": 0.6053,
+      "step": 59852
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.686875581741333,
+      "learning_rate": 1.7059777621610263e-08,
+      "loss": 0.7066,
+      "step": 59853
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.168318748474121,
+      "learning_rate": 1.703802313300751e-08,
+      "loss": 0.6547,
+      "step": 59854
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.414165735244751,
+      "learning_rate": 1.701628250015519e-08,
+      "loss": 0.6271,
+      "step": 59855
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.722475051879883,
+      "learning_rate": 1.699455572311437e-08,
+      "loss": 0.4004,
+      "step": 59856
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 7.5919060707092285,
+      "learning_rate": 1.6972842801945e-08,
+      "loss": 0.6901,
+      "step": 59857
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 4.483240127563477,
+      "learning_rate": 1.695114373670703e-08,
+      "loss": 0.6378,
+      "step": 59858
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 5.178640365600586,
+      "learning_rate": 1.6929458527461527e-08,
+      "loss": 0.6408,
+      "step": 59859
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 4.747476577758789,
+      "learning_rate": 1.6907787174268443e-08,
+      "loss": 0.7088,
+      "step": 59860
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.416106939315796,
+      "learning_rate": 1.6886129677188278e-08,
+      "loss": 0.6872,
+      "step": 59861
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 1.8918125629425049,
+      "learning_rate": 1.6864486036280437e-08,
+      "loss": 0.2966,
+      "step": 59862
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.6255011558532715,
+      "learning_rate": 1.6842856251606532e-08,
+      "loss": 0.7277,
+      "step": 59863
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.6777966022491455,
+      "learning_rate": 1.6821240323225406e-08,
+      "loss": 0.5449,
+      "step": 59864
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.749150514602661,
+      "learning_rate": 1.679963825119757e-08,
+      "loss": 0.5505,
+      "step": 59865
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.3265554904937744,
+      "learning_rate": 1.6778050035583526e-08,
+      "loss": 0.6183,
+      "step": 59866
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 4.562371253967285,
+      "learning_rate": 1.6756475676442673e-08,
+      "loss": 0.6589,
+      "step": 59867
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.3585011959075928,
+      "learning_rate": 1.673491517383552e-08,
+      "loss": 0.4295,
+      "step": 59868
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.707059621810913,
+      "learning_rate": 1.6713368527822017e-08,
+      "loss": 0.5557,
+      "step": 59869
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.263744831085205,
+      "learning_rate": 1.669183573846156e-08,
+      "loss": 0.644,
+      "step": 59870
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.4859466552734375,
+      "learning_rate": 1.6670316805814102e-08,
+      "loss": 0.6527,
+      "step": 59871
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.298628091812134,
+      "learning_rate": 1.664881172994015e-08,
+      "loss": 0.4424,
+      "step": 59872
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 4.010233402252197,
+      "learning_rate": 1.6627320510899102e-08,
+      "loss": 0.7077,
+      "step": 59873
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.7158262729644775,
+      "learning_rate": 1.6605843148750357e-08,
+      "loss": 0.5099,
+      "step": 59874
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.562941789627075,
+      "learning_rate": 1.6584379643553862e-08,
+      "loss": 0.6133,
+      "step": 59875
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 4.097439289093018,
+      "learning_rate": 1.656292999536957e-08,
+      "loss": 0.5392,
+      "step": 59876
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.5047948360443115,
+      "learning_rate": 1.654149420425688e-08,
+      "loss": 0.5661,
+      "step": 59877
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.401414632797241,
+      "learning_rate": 1.652007227027519e-08,
+      "loss": 0.6612,
+      "step": 59878
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.2710633277893066,
+      "learning_rate": 1.6498664193485003e-08,
+      "loss": 0.3738,
+      "step": 59879
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.6309590339660645,
+      "learning_rate": 1.647726997394461e-08,
+      "loss": 0.4965,
+      "step": 59880
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.6671805381774902,
+      "learning_rate": 1.645588961171396e-08,
+      "loss": 0.3565,
+      "step": 59881
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.3422956466674805,
+      "learning_rate": 1.6434523106853006e-08,
+      "loss": 0.6948,
+      "step": 59882
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.464956521987915,
+      "learning_rate": 1.641317045942059e-08,
+      "loss": 0.497,
+      "step": 59883
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.8583271503448486,
+      "learning_rate": 1.6391831669476112e-08,
+      "loss": 0.349,
+      "step": 59884
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.5858378410339355,
+      "learning_rate": 1.637050673707896e-08,
+      "loss": 0.634,
+      "step": 59885
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 4.910968780517578,
+      "learning_rate": 1.6349195662289098e-08,
+      "loss": 0.5784,
+      "step": 59886
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 1.7184504270553589,
+      "learning_rate": 1.6327898445164802e-08,
+      "loss": 0.3386,
+      "step": 59887
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.989558696746826,
+      "learning_rate": 1.6306615085765475e-08,
+      "loss": 0.6138,
+      "step": 59888
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.8622257709503174,
+      "learning_rate": 1.6285345584151068e-08,
+      "loss": 0.5476,
+      "step": 59889
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.6870882511138916,
+      "learning_rate": 1.6264089940379867e-08,
+      "loss": 0.5772,
+      "step": 59890
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.5014164447784424,
+      "learning_rate": 1.624284815451127e-08,
+      "loss": 0.4561,
+      "step": 59891
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.3679254055023193,
+      "learning_rate": 1.6221620226604116e-08,
+      "loss": 0.6475,
+      "step": 59892
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.0244083404541016,
+      "learning_rate": 1.6200406156718363e-08,
+      "loss": 0.6388,
+      "step": 59893
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 5.253472805023193,
+      "learning_rate": 1.6179205944911735e-08,
+      "loss": 0.5563,
+      "step": 59894
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.169649839401245,
+      "learning_rate": 1.6158019591243636e-08,
+      "loss": 0.6765,
+      "step": 59895
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.559647560119629,
+      "learning_rate": 1.6136847095773455e-08,
+      "loss": 0.6937,
+      "step": 59896
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.8715567588806152,
+      "learning_rate": 1.6115688458560043e-08,
+      "loss": 0.4471,
+      "step": 59897
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 4.162070274353027,
+      "learning_rate": 1.6094543679661122e-08,
+      "loss": 0.5498,
+      "step": 59898
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.908263921737671,
+      "learning_rate": 1.6073412759136653e-08,
+      "loss": 0.6053,
+      "step": 59899
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.9440081119537354,
+      "learning_rate": 1.6052295697044917e-08,
+      "loss": 0.6616,
+      "step": 59900
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.9108705520629883,
+      "learning_rate": 1.603119249344476e-08,
+      "loss": 0.5442,
+      "step": 59901
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.0061161518096924,
+      "learning_rate": 1.6010103148395016e-08,
+      "loss": 0.5913,
+      "step": 59902
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.7717323303222656,
+      "learning_rate": 1.5989027661953982e-08,
+      "loss": 0.6639,
+      "step": 59903
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.804469585418701,
+      "learning_rate": 1.5967966034179937e-08,
+      "loss": 0.4154,
+      "step": 59904
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.0918381214141846,
+      "learning_rate": 1.5946918265132282e-08,
+      "loss": 0.6266,
+      "step": 59905
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.396540641784668,
+      "learning_rate": 1.5925884354869303e-08,
+      "loss": 0.5778,
+      "step": 59906
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.3279612064361572,
+      "learning_rate": 1.5904864303449284e-08,
+      "loss": 0.4928,
+      "step": 59907
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.7838613986968994,
+      "learning_rate": 1.5883858110930518e-08,
+      "loss": 0.4849,
+      "step": 59908
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.61842679977417,
+      "learning_rate": 1.5862865777372393e-08,
+      "loss": 0.6301,
+      "step": 59909
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.3716812133789062,
+      "learning_rate": 1.5841887302831538e-08,
+      "loss": 0.5114,
+      "step": 59910
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 4.405546188354492,
+      "learning_rate": 1.58209226873679e-08,
+      "loss": 0.7259,
+      "step": 59911
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.9299750328063965,
+      "learning_rate": 1.5799971931039216e-08,
+      "loss": 0.6316,
+      "step": 59912
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.312711000442505,
+      "learning_rate": 1.5779035033903767e-08,
+      "loss": 0.5514,
+      "step": 59913
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.4917755126953125,
+      "learning_rate": 1.5758111996019286e-08,
+      "loss": 0.4855,
+      "step": 59914
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.0065174102783203,
+      "learning_rate": 1.573720281744462e-08,
+      "loss": 0.5085,
+      "step": 59915
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.989192008972168,
+      "learning_rate": 1.5716307498236938e-08,
+      "loss": 0.4793,
+      "step": 59916
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 4.7276411056518555,
+      "learning_rate": 1.5695426038455643e-08,
+      "loss": 0.663,
+      "step": 59917
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.449920415878296,
+      "learning_rate": 1.567455843815846e-08,
+      "loss": 0.5887,
+      "step": 59918
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.415581226348877,
+      "learning_rate": 1.5653704697402573e-08,
+      "loss": 0.6128,
+      "step": 59919
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.2791972160339355,
+      "learning_rate": 1.5632864816247374e-08,
+      "loss": 0.7656,
+      "step": 59920
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.1063249111175537,
+      "learning_rate": 1.5612038794749485e-08,
+      "loss": 0.4711,
+      "step": 59921
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.401137590408325,
+      "learning_rate": 1.5591226632967194e-08,
+      "loss": 0.7852,
+      "step": 59922
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.2115085124969482,
+      "learning_rate": 1.5570428330958785e-08,
+      "loss": 0.4985,
+      "step": 59923
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.3526172637939453,
+      "learning_rate": 1.554964388878144e-08,
+      "loss": 0.6564,
+      "step": 59924
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.5589210987091064,
+      "learning_rate": 1.552887330649344e-08,
+      "loss": 0.6117,
+      "step": 59925
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.968524217605591,
+      "learning_rate": 1.5508116584152523e-08,
+      "loss": 0.5857,
+      "step": 59926
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.0724053382873535,
+      "learning_rate": 1.5487373721816412e-08,
+      "loss": 0.5756,
+      "step": 59927
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.816521644592285,
+      "learning_rate": 1.546664471954229e-08,
+      "loss": 0.4412,
+      "step": 59928
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 5.593622207641602,
+      "learning_rate": 1.544592957738844e-08,
+      "loss": 1.0021,
+      "step": 59929
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.971622943878174,
+      "learning_rate": 1.5425228295412598e-08,
+      "loss": 0.5438,
+      "step": 59930
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.928647994995117,
+      "learning_rate": 1.540454087367138e-08,
+      "loss": 0.4721,
+      "step": 59931
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.866872549057007,
+      "learning_rate": 1.538386731222308e-08,
+      "loss": 0.5679,
+      "step": 59932
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.7334680557250977,
+      "learning_rate": 1.5363207611124865e-08,
+      "loss": 0.8358,
+      "step": 59933
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 4.933405876159668,
+      "learning_rate": 1.5342561770434474e-08,
+      "loss": 0.5391,
+      "step": 59934
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.6126372814178467,
+      "learning_rate": 1.532192979020852e-08,
+      "loss": 0.6719,
+      "step": 59935
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.7523353099823,
+      "learning_rate": 1.5301311670505857e-08,
+      "loss": 0.6596,
+      "step": 59936
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.808680772781372,
+      "learning_rate": 1.5280707411382544e-08,
+      "loss": 0.7519,
+      "step": 59937
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.044297456741333,
+      "learning_rate": 1.5260117012896315e-08,
+      "loss": 0.4522,
+      "step": 59938
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.0743613243103027,
+      "learning_rate": 1.5239540475103786e-08,
+      "loss": 0.4437,
+      "step": 59939
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 4.133922576904297,
+      "learning_rate": 1.5218977798063807e-08,
+      "loss": 0.7594,
+      "step": 59940
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.0858561992645264,
+      "learning_rate": 1.519842898183188e-08,
+      "loss": 0.6486,
+      "step": 59941
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.7455966472625732,
+      "learning_rate": 1.5177894026465744e-08,
+      "loss": 0.774,
+      "step": 59942
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.1590797901153564,
+      "learning_rate": 1.5157372932022573e-08,
+      "loss": 0.4552,
+      "step": 59943
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 4.556820392608643,
+      "learning_rate": 1.5136865698559545e-08,
+      "loss": 0.6412,
+      "step": 59944
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.608628749847412,
+      "learning_rate": 1.5116372326133278e-08,
+      "loss": 0.4867,
+      "step": 59945
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.2800416946411133,
+      "learning_rate": 1.509589281480095e-08,
+      "loss": 0.6304,
+      "step": 59946
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.2096240520477295,
+      "learning_rate": 1.507542716461974e-08,
+      "loss": 0.6707,
+      "step": 59947
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 6.260316371917725,
+      "learning_rate": 1.5054975375646817e-08,
+      "loss": 0.7521,
+      "step": 59948
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.083064317703247,
+      "learning_rate": 1.5034537447937704e-08,
+      "loss": 0.5623,
+      "step": 59949
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.801114082336426,
+      "learning_rate": 1.5014113381550677e-08,
+      "loss": 0.5537,
+      "step": 59950
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.214582681655884,
+      "learning_rate": 1.4993703176541806e-08,
+      "loss": 0.7913,
+      "step": 59951
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.3964486122131348,
+      "learning_rate": 1.4973306832968272e-08,
+      "loss": 0.519,
+      "step": 59952
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.575291156768799,
+      "learning_rate": 1.4952924350886688e-08,
+      "loss": 0.5888,
+      "step": 59953
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.262465715408325,
+      "learning_rate": 1.4932555730353126e-08,
+      "loss": 0.4241,
+      "step": 59954
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.130858898162842,
+      "learning_rate": 1.491220097142476e-08,
+      "loss": 0.4413,
+      "step": 59955
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.6018941402435303,
+      "learning_rate": 1.4891860074158215e-08,
+      "loss": 0.4021,
+      "step": 59956
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 1.8658783435821533,
+      "learning_rate": 1.4871533038610108e-08,
+      "loss": 0.4282,
+      "step": 59957
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.15000057220459,
+      "learning_rate": 1.4851219864836508e-08,
+      "loss": 0.6671,
+      "step": 59958
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.1487207412719727,
+      "learning_rate": 1.4830920552894035e-08,
+      "loss": 0.4371,
+      "step": 59959
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.5714473724365234,
+      "learning_rate": 1.4810635102839865e-08,
+      "loss": 0.6196,
+      "step": 59960
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.5009477138519287,
+      "learning_rate": 1.4790363514728955e-08,
+      "loss": 0.5585,
+      "step": 59961
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.8775038719177246,
+      "learning_rate": 1.4770105788619593e-08,
+      "loss": 0.6081,
+      "step": 59962
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.7352359294891357,
+      "learning_rate": 1.4749861924566178e-08,
+      "loss": 0.308,
+      "step": 59963
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.760970115661621,
+      "learning_rate": 1.4729631922626442e-08,
+      "loss": 0.604,
+      "step": 59964
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.1411373615264893,
+      "learning_rate": 1.4709415782855896e-08,
+      "loss": 0.4431,
+      "step": 59965
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 4.2999162673950195,
+      "learning_rate": 1.4689213505310607e-08,
+      "loss": 0.4847,
+      "step": 59966
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.685936450958252,
+      "learning_rate": 1.4669025090047196e-08,
+      "loss": 0.4154,
+      "step": 59967
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 4.149756908416748,
+      "learning_rate": 1.4648850537122284e-08,
+      "loss": 0.5957,
+      "step": 59968
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.8855793476104736,
+      "learning_rate": 1.4628689846590827e-08,
+      "loss": 0.5575,
+      "step": 59969
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 5.6838178634643555,
+      "learning_rate": 1.4608543018509447e-08,
+      "loss": 0.428,
+      "step": 59970
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 4.115393161773682,
+      "learning_rate": 1.458841005293421e-08,
+      "loss": 0.6642,
+      "step": 59971
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.454197645187378,
+      "learning_rate": 1.4568290949921182e-08,
+      "loss": 0.8502,
+      "step": 59972
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.7512638568878174,
+      "learning_rate": 1.4548185709525875e-08,
+      "loss": 0.3585,
+      "step": 59973
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.276395559310913,
+      "learning_rate": 1.4528094331804909e-08,
+      "loss": 0.5801,
+      "step": 59974
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.373863935470581,
+      "learning_rate": 1.4508016816813242e-08,
+      "loss": 0.4687,
+      "step": 59975
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.8270130157470703,
+      "learning_rate": 1.4487953164607493e-08,
+      "loss": 0.549,
+      "step": 59976
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 4.817200183868408,
+      "learning_rate": 1.446790337524262e-08,
+      "loss": 0.4404,
+      "step": 59977
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 4.535511016845703,
+      "learning_rate": 1.4447867448775243e-08,
+      "loss": 0.8158,
+      "step": 59978
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.659670352935791,
+      "learning_rate": 1.4427845385260875e-08,
+      "loss": 0.525,
+      "step": 59979
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.2062652111053467,
+      "learning_rate": 1.4407837184754469e-08,
+      "loss": 0.5147,
+      "step": 59980
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.5923802852630615,
+      "learning_rate": 1.4387842847312649e-08,
+      "loss": 0.5258,
+      "step": 59981
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 5.131900310516357,
+      "learning_rate": 1.4367862372990926e-08,
+      "loss": 0.4005,
+      "step": 59982
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.920971393585205,
+      "learning_rate": 1.4347895761844255e-08,
+      "loss": 0.7253,
+      "step": 59983
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.7452118396759033,
+      "learning_rate": 1.4327943013928148e-08,
+      "loss": 0.6634,
+      "step": 59984
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.629495859146118,
+      "learning_rate": 1.4308004129298669e-08,
+      "loss": 0.5168,
+      "step": 59985
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.914097785949707,
+      "learning_rate": 1.4288079108010223e-08,
+      "loss": 0.6245,
+      "step": 59986
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.4629123210906982,
+      "learning_rate": 1.4268167950119427e-08,
+      "loss": 0.4762,
+      "step": 59987
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.4757611751556396,
+      "learning_rate": 1.4248270655681795e-08,
+      "loss": 0.558,
+      "step": 59988
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.89623761177063,
+      "learning_rate": 1.4228387224751172e-08,
+      "loss": 0.3154,
+      "step": 59989
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 4.290214538574219,
+      "learning_rate": 1.420851765738418e-08,
+      "loss": 0.6593,
+      "step": 59990
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.457428216934204,
+      "learning_rate": 1.4188661953635218e-08,
+      "loss": 0.728,
+      "step": 59991
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.757594585418701,
+      "learning_rate": 1.41688201135598e-08,
+      "loss": 0.7576,
+      "step": 59992
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.571789503097534,
+      "learning_rate": 1.4148992137213435e-08,
+      "loss": 0.6973,
+      "step": 59993
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.654945135116577,
+      "learning_rate": 1.412917802465108e-08,
+      "loss": 0.6261,
+      "step": 59994
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.080850839614868,
+      "learning_rate": 1.4109377775927691e-08,
+      "loss": 0.633,
+      "step": 59995
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.5255661010742188,
+      "learning_rate": 1.4089591391098223e-08,
+      "loss": 0.478,
+      "step": 59996
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 4.595333576202393,
+      "learning_rate": 1.4069818870217633e-08,
+      "loss": 0.4751,
+      "step": 59997
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.4488885402679443,
+      "learning_rate": 1.4050060213341432e-08,
+      "loss": 0.4609,
+      "step": 59998
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.1933257579803467,
+      "learning_rate": 1.403031542052402e-08,
+      "loss": 0.4484,
+      "step": 59999
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.5753540992736816,
+      "learning_rate": 1.4010584491820911e-08,
+      "loss": 0.8942,
+      "step": 60000
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 4.297773838043213,
+      "learning_rate": 1.3990867427286503e-08,
+      "loss": 0.5359,
+      "step": 60001
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.0323095321655273,
+      "learning_rate": 1.3971164226975198e-08,
+      "loss": 0.2887,
+      "step": 60002
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 4.750086784362793,
+      "learning_rate": 1.3951474890942507e-08,
+      "loss": 0.5709,
+      "step": 60003
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.5053892135620117,
+      "learning_rate": 1.3931799419242831e-08,
+      "loss": 0.843,
+      "step": 60004
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.2253589630126953,
+      "learning_rate": 1.391213781193168e-08,
+      "loss": 0.4917,
+      "step": 60005
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.7252602577209473,
+      "learning_rate": 1.3892490069062347e-08,
+      "loss": 0.4564,
+      "step": 60006
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.8467235565185547,
+      "learning_rate": 1.3872856190690342e-08,
+      "loss": 0.4756,
+      "step": 60007
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 4.843059539794922,
+      "learning_rate": 1.3853236176870066e-08,
+      "loss": 0.6912,
+      "step": 60008
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.6195216178894043,
+      "learning_rate": 1.383363002765592e-08,
+      "loss": 0.4562,
+      "step": 60009
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.0177559852600098,
+      "learning_rate": 1.3814037743102859e-08,
+      "loss": 0.6893,
+      "step": 60010
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.571051836013794,
+      "learning_rate": 1.3794459323264731e-08,
+      "loss": 0.5604,
+      "step": 60011
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.05212140083313,
+      "learning_rate": 1.3774894768196489e-08,
+      "loss": 0.7791,
+      "step": 60012
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 10.631122589111328,
+      "learning_rate": 1.3755344077953093e-08,
+      "loss": 0.657,
+      "step": 60013
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.3576347827911377,
+      "learning_rate": 1.3735807252587274e-08,
+      "loss": 0.5414,
+      "step": 60014
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 4.020802021026611,
+      "learning_rate": 1.3716284292154547e-08,
+      "loss": 0.7601,
+      "step": 60015
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.9181222915649414,
+      "learning_rate": 1.3696775196708756e-08,
+      "loss": 0.6306,
+      "step": 60016
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.7356436252593994,
+      "learning_rate": 1.3677279966304858e-08,
+      "loss": 0.5866,
+      "step": 60017
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 4.653815269470215,
+      "learning_rate": 1.3657798600996141e-08,
+      "loss": 0.7689,
+      "step": 60018
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.6363720893859863,
+      "learning_rate": 1.3638331100837566e-08,
+      "loss": 0.4863,
+      "step": 60019
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.76448392868042,
+      "learning_rate": 1.3618877465882418e-08,
+      "loss": 0.6361,
+      "step": 60020
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.554180145263672,
+      "learning_rate": 1.3599437696185103e-08,
+      "loss": 0.5965,
+      "step": 60021
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.5158205032348633,
+      "learning_rate": 1.3580011791800018e-08,
+      "loss": 0.3918,
+      "step": 60022
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.3331375122070312,
+      "learning_rate": 1.3560599752781011e-08,
+      "loss": 0.4641,
+      "step": 60023
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.6525309085845947,
+      "learning_rate": 1.3541201579181929e-08,
+      "loss": 0.6872,
+      "step": 60024
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.778550386428833,
+      "learning_rate": 1.3521817271057169e-08,
+      "loss": 0.7655,
+      "step": 60025
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 4.65944766998291,
+      "learning_rate": 1.3502446828460025e-08,
+      "loss": 0.8924,
+      "step": 60026
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.128540277481079,
+      "learning_rate": 1.3483090251444342e-08,
+      "loss": 0.4797,
+      "step": 60027
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.2612557411193848,
+      "learning_rate": 1.3463747540063965e-08,
+      "loss": 0.6587,
+      "step": 60028
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.4616994857788086,
+      "learning_rate": 1.344441869437385e-08,
+      "loss": 0.6962,
+      "step": 60029
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.5679972171783447,
+      "learning_rate": 1.342510371442618e-08,
+      "loss": 0.4124,
+      "step": 60030
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.0282115936279297,
+      "learning_rate": 1.3405802600275353e-08,
+      "loss": 0.6073,
+      "step": 60031
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.459556818008423,
+      "learning_rate": 1.338651535197466e-08,
+      "loss": 0.3278,
+      "step": 60032
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.38065242767334,
+      "learning_rate": 1.3367241969578503e-08,
+      "loss": 0.6256,
+      "step": 60033
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.926819086074829,
+      "learning_rate": 1.3347982453139619e-08,
+      "loss": 0.8713,
+      "step": 60034
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 18.621463775634766,
+      "learning_rate": 1.3328736802712405e-08,
+      "loss": 0.4178,
+      "step": 60035
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.442232847213745,
+      "learning_rate": 1.33095050183496e-08,
+      "loss": 0.3964,
+      "step": 60036
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.628052234649658,
+      "learning_rate": 1.3290287100105048e-08,
+      "loss": 0.4713,
+      "step": 60037
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.5420737266540527,
+      "learning_rate": 1.3271083048032597e-08,
+      "loss": 0.5957,
+      "step": 60038
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.504448652267456,
+      "learning_rate": 1.3251892862184423e-08,
+      "loss": 0.4459,
+      "step": 60039
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 4.883889198303223,
+      "learning_rate": 1.3232716542615487e-08,
+      "loss": 0.885,
+      "step": 60040
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.09596586227417,
+      "learning_rate": 1.3213554089377411e-08,
+      "loss": 0.3704,
+      "step": 60041
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.230071544647217,
+      "learning_rate": 1.3194405502525153e-08,
+      "loss": 0.3505,
+      "step": 60042
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 4.327718734741211,
+      "learning_rate": 1.3175270782110339e-08,
+      "loss": 0.6072,
+      "step": 60043
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.766957998275757,
+      "learning_rate": 1.3156149928187922e-08,
+      "loss": 0.4764,
+      "step": 60044
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.3765196800231934,
+      "learning_rate": 1.313704294080953e-08,
+      "loss": 0.6277,
+      "step": 60045
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.374420166015625,
+      "learning_rate": 1.3117949820029008e-08,
+      "loss": 0.5582,
+      "step": 60046
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 1.9721721410751343,
+      "learning_rate": 1.3098870565899091e-08,
+      "loss": 0.3667,
+      "step": 60047
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.1293017864227295,
+      "learning_rate": 1.307980517847307e-08,
+      "loss": 0.81,
+      "step": 60048
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.5526654720306396,
+      "learning_rate": 1.3060753657804792e-08,
+      "loss": 0.6097,
+      "step": 60049
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.784532070159912,
+      "learning_rate": 1.3041716003945326e-08,
+      "loss": 0.5725,
+      "step": 60050
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.683863639831543,
+      "learning_rate": 1.3022692216949073e-08,
+      "loss": 0.4868,
+      "step": 60051
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.1801743507385254,
+      "learning_rate": 1.3003682296868769e-08,
+      "loss": 0.4552,
+      "step": 60052
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.3429067134857178,
+      "learning_rate": 1.2984686243757149e-08,
+      "loss": 0.2011,
+      "step": 60053
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.5753982067108154,
+      "learning_rate": 1.296570405766695e-08,
+      "loss": 0.6343,
+      "step": 60054
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.299628496170044,
+      "learning_rate": 1.2946735738650352e-08,
+      "loss": 0.5594,
+      "step": 60055
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 4.0041303634643555,
+      "learning_rate": 1.2927781286761199e-08,
+      "loss": 0.8588,
+      "step": 60056
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.793335199356079,
+      "learning_rate": 1.2908840702051673e-08,
+      "loss": 0.6542,
+      "step": 60057
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.7954163551330566,
+      "learning_rate": 1.288991398457451e-08,
+      "loss": 0.5772,
+      "step": 60058
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.093820810317993,
+      "learning_rate": 1.2871001134381888e-08,
+      "loss": 0.4246,
+      "step": 60059
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.5148916244506836,
+      "learning_rate": 1.2852102151527101e-08,
+      "loss": 0.4526,
+      "step": 60060
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.591352939605713,
+      "learning_rate": 1.2833217036062328e-08,
+      "loss": 0.4987,
+      "step": 60061
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 4.348520755767822,
+      "learning_rate": 1.2814345788040306e-08,
+      "loss": 0.5604,
+      "step": 60062
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.6419310569763184,
+      "learning_rate": 1.2795488407513212e-08,
+      "loss": 0.6845,
+      "step": 60063
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 4.224991798400879,
+      "learning_rate": 1.277664489453323e-08,
+      "loss": 0.5946,
+      "step": 60064
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.5818018913269043,
+      "learning_rate": 1.275781524915365e-08,
+      "loss": 0.6406,
+      "step": 60065
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.851724147796631,
+      "learning_rate": 1.2738999471426095e-08,
+      "loss": 0.4549,
+      "step": 60066
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.725202798843384,
+      "learning_rate": 1.2720197561403302e-08,
+      "loss": 0.5858,
+      "step": 60067
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 4.978499889373779,
+      "learning_rate": 1.2701409519136898e-08,
+      "loss": 0.648,
+      "step": 60068
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.066216468811035,
+      "learning_rate": 1.2682635344680172e-08,
+      "loss": 0.497,
+      "step": 60069
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.3808910846710205,
+      "learning_rate": 1.266387503808475e-08,
+      "loss": 0.5938,
+      "step": 60070
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 4.620325565338135,
+      "learning_rate": 1.2645128599402256e-08,
+      "loss": 0.7451,
+      "step": 60071
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.8917362689971924,
+      "learning_rate": 1.2626396028685428e-08,
+      "loss": 0.4601,
+      "step": 60072
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.7642805576324463,
+      "learning_rate": 1.2607677325986446e-08,
+      "loss": 0.3843,
+      "step": 60073
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.289802312850952,
+      "learning_rate": 1.2588972491357487e-08,
+      "loss": 0.6534,
+      "step": 60074
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 4.284840106964111,
+      "learning_rate": 1.2570281524850181e-08,
+      "loss": 0.5642,
+      "step": 60075
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.853651762008667,
+      "learning_rate": 1.2551604426516151e-08,
+      "loss": 0.7306,
+      "step": 60076
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.5899786949157715,
+      "learning_rate": 1.2532941196408132e-08,
+      "loss": 0.7643,
+      "step": 60077
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.476701021194458,
+      "learning_rate": 1.2514291834577752e-08,
+      "loss": 0.4893,
+      "step": 60078
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.4565577507019043,
+      "learning_rate": 1.2495656341076634e-08,
+      "loss": 0.8441,
+      "step": 60079
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 4.477560043334961,
+      "learning_rate": 1.2477034715956404e-08,
+      "loss": 0.4896,
+      "step": 60080
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 4.069235324859619,
+      "learning_rate": 1.2458426959269799e-08,
+      "loss": 0.9229,
+      "step": 60081
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.392409563064575,
+      "learning_rate": 1.243983307106733e-08,
+      "loss": 0.8233,
+      "step": 60082
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.2693710327148438,
+      "learning_rate": 1.2421253051401738e-08,
+      "loss": 0.4389,
+      "step": 60083
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.8788809776306152,
+      "learning_rate": 1.240268690032409e-08,
+      "loss": 0.5579,
+      "step": 60084
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.5396010875701904,
+      "learning_rate": 1.2384134617886012e-08,
+      "loss": 0.6945,
+      "step": 60085
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.714109182357788,
+      "learning_rate": 1.2365596204139685e-08,
+      "loss": 0.5387,
+      "step": 60086
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.8290350437164307,
+      "learning_rate": 1.2347071659135623e-08,
+      "loss": 0.7085,
+      "step": 60087
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.8202013969421387,
+      "learning_rate": 1.2328560982926563e-08,
+      "loss": 0.7009,
+      "step": 60088
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.1862337589263916,
+      "learning_rate": 1.2310064175562464e-08,
+      "loss": 0.7014,
+      "step": 60089
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.967374324798584,
+      "learning_rate": 1.2291581237096618e-08,
+      "loss": 0.5882,
+      "step": 60090
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.3273158073425293,
+      "learning_rate": 1.2273112167578983e-08,
+      "loss": 0.5916,
+      "step": 60091
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.6469838619232178,
+      "learning_rate": 1.2254656967061184e-08,
+      "loss": 0.4597,
+      "step": 60092
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.384241819381714,
+      "learning_rate": 1.2236215635594851e-08,
+      "loss": 0.5106,
+      "step": 60093
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.242659568786621,
+      "learning_rate": 1.2217788173231604e-08,
+      "loss": 0.3961,
+      "step": 60094
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.5982580184936523,
+      "learning_rate": 1.2199374580021961e-08,
+      "loss": 0.4686,
+      "step": 60095
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.757110834121704,
+      "learning_rate": 1.2180974856017546e-08,
+      "loss": 0.5242,
+      "step": 60096
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.3514907360076904,
+      "learning_rate": 1.2162589001268876e-08,
+      "loss": 0.4191,
+      "step": 60097
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.7666444778442383,
+      "learning_rate": 1.2144217015827576e-08,
+      "loss": 0.5799,
+      "step": 60098
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.0849382877349854,
+      "learning_rate": 1.2125858899745268e-08,
+      "loss": 0.7469,
+      "step": 60099
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.9302785396575928,
+      "learning_rate": 1.2107514653071917e-08,
+      "loss": 0.5552,
+      "step": 60100
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.721364974975586,
+      "learning_rate": 1.2089184275859145e-08,
+      "loss": 0.5944,
+      "step": 60101
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 4.133647918701172,
+      "learning_rate": 1.2070867768158023e-08,
+      "loss": 0.5832,
+      "step": 60102
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.838548183441162,
+      "learning_rate": 1.2052565130019623e-08,
+      "loss": 0.9565,
+      "step": 60103
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.1710398197174072,
+      "learning_rate": 1.2034276361493901e-08,
+      "loss": 0.533,
+      "step": 60104
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.251878261566162,
+      "learning_rate": 1.2016001462632487e-08,
+      "loss": 0.4796,
+      "step": 60105
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.635793685913086,
+      "learning_rate": 1.199774043348645e-08,
+      "loss": 0.5486,
+      "step": 60106
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.3678572177886963,
+      "learning_rate": 1.1979493274105747e-08,
+      "loss": 0.5173,
+      "step": 60107
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.0807647705078125,
+      "learning_rate": 1.196125998454145e-08,
+      "loss": 0.408,
+      "step": 60108
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.8906824588775635,
+      "learning_rate": 1.1943040564844077e-08,
+      "loss": 0.6697,
+      "step": 60109
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.240584373474121,
+      "learning_rate": 1.192483501506525e-08,
+      "loss": 0.7555,
+      "step": 60110
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.766019821166992,
+      "learning_rate": 1.1906643335254376e-08,
+      "loss": 0.6633,
+      "step": 60111
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 5.3069915771484375,
+      "learning_rate": 1.1888465525463077e-08,
+      "loss": 0.691,
+      "step": 60112
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.473998785018921,
+      "learning_rate": 1.1870301585740763e-08,
+      "loss": 0.7946,
+      "step": 60113
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.210608959197998,
+      "learning_rate": 1.18521515161385e-08,
+      "loss": 0.4719,
+      "step": 60114
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.225205421447754,
+      "learning_rate": 1.183401531670736e-08,
+      "loss": 0.4688,
+      "step": 60115
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.0050482749938965,
+      "learning_rate": 1.1815892987496747e-08,
+      "loss": 0.5219,
+      "step": 60116
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.8152401447296143,
+      "learning_rate": 1.1797784528557732e-08,
+      "loss": 0.3146,
+      "step": 60117
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.2929861545562744,
+      "learning_rate": 1.1779689939940276e-08,
+      "loss": 0.7439,
+      "step": 60118
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.8233282566070557,
+      "learning_rate": 1.1761609221695446e-08,
+      "loss": 0.9137,
+      "step": 60119
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.858696222305298,
+      "learning_rate": 1.1743542373872096e-08,
+      "loss": 0.5689,
+      "step": 60120
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 5.533843040466309,
+      "learning_rate": 1.1725489396521849e-08,
+      "loss": 0.7313,
+      "step": 60121
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.7845888137817383,
+      "learning_rate": 1.1707450289694666e-08,
+      "loss": 0.4942,
+      "step": 60122
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.640500068664551,
+      "learning_rate": 1.1689425053439952e-08,
+      "loss": 0.8652,
+      "step": 60123
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.5337533950805664,
+      "learning_rate": 1.167141368780822e-08,
+      "loss": 0.3881,
+      "step": 60124
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 4.148382186889648,
+      "learning_rate": 1.1653416192849431e-08,
+      "loss": 0.6938,
+      "step": 60125
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.8804099559783936,
+      "learning_rate": 1.1635432568614103e-08,
+      "loss": 0.5787,
+      "step": 60126
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 4.7218217849731445,
+      "learning_rate": 1.1617462815152191e-08,
+      "loss": 0.6418,
+      "step": 60127
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.694117784500122,
+      "learning_rate": 1.1599506932513105e-08,
+      "loss": 0.7345,
+      "step": 60128
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 4.641429424285889,
+      "learning_rate": 1.1581564920747357e-08,
+      "loss": 0.434,
+      "step": 60129
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.9539811611175537,
+      "learning_rate": 1.1563636779904352e-08,
+      "loss": 0.7757,
+      "step": 60130
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.2458949089050293,
+      "learning_rate": 1.1545722510034052e-08,
+      "loss": 0.483,
+      "step": 60131
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.6804656982421875,
+      "learning_rate": 1.1527822111186415e-08,
+      "loss": 0.6342,
+      "step": 60132
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 4.165462970733643,
+      "learning_rate": 1.1509935583411402e-08,
+      "loss": 0.7316,
+      "step": 60133
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.6062815189361572,
+      "learning_rate": 1.1492062926758417e-08,
+      "loss": 0.6633,
+      "step": 60134
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.6607720851898193,
+      "learning_rate": 1.1474204141277422e-08,
+      "loss": 0.559,
+      "step": 60135
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.689493417739868,
+      "learning_rate": 1.1456359227017822e-08,
+      "loss": 0.6374,
+      "step": 60136
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 4.196478366851807,
+      "learning_rate": 1.143852818402902e-08,
+      "loss": 0.5703,
+      "step": 60137
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.9941282272338867,
+      "learning_rate": 1.1420711012360974e-08,
+      "loss": 0.543,
+      "step": 60138
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 5.907413482666016,
+      "learning_rate": 1.1402907712063648e-08,
+      "loss": 0.5646,
+      "step": 60139
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.9718916416168213,
+      "learning_rate": 1.1385118283185892e-08,
+      "loss": 0.4045,
+      "step": 60140
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.499066114425659,
+      "learning_rate": 1.1367342725777109e-08,
+      "loss": 0.62,
+      "step": 60141
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 4.709290981292725,
+      "learning_rate": 1.1349581039887259e-08,
+      "loss": 0.6214,
+      "step": 60142
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.1948182582855225,
+      "learning_rate": 1.1331833225565191e-08,
+      "loss": 0.514,
+      "step": 60143
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.6934449672698975,
+      "learning_rate": 1.1314099282860313e-08,
+      "loss": 0.5742,
+      "step": 60144
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.7642552852630615,
+      "learning_rate": 1.1296379211822584e-08,
+      "loss": 0.6006,
+      "step": 60145
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.2151079177856445,
+      "learning_rate": 1.1278673012500851e-08,
+      "loss": 0.2959,
+      "step": 60146
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 4.106210231781006,
+      "learning_rate": 1.1260980684943967e-08,
+      "loss": 0.7938,
+      "step": 60147
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.9305930137634277,
+      "learning_rate": 1.1243302229201336e-08,
+      "loss": 0.6976,
+      "step": 60148
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.706068277359009,
+      "learning_rate": 1.1225637645322918e-08,
+      "loss": 0.4244,
+      "step": 60149
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 4.489842414855957,
+      "learning_rate": 1.1207986933356451e-08,
+      "loss": 0.4035,
+      "step": 60150
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.705780267715454,
+      "learning_rate": 1.1190350093351899e-08,
+      "loss": 0.682,
+      "step": 60151
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.491886854171753,
+      "learning_rate": 1.1172727125358108e-08,
+      "loss": 0.753,
+      "step": 60152
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 4.284393310546875,
+      "learning_rate": 1.1155118029424483e-08,
+      "loss": 0.5519,
+      "step": 60153
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.603665590286255,
+      "learning_rate": 1.1137522805599322e-08,
+      "loss": 0.7819,
+      "step": 60154
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.1748321056365967,
+      "learning_rate": 1.1119941453932026e-08,
+      "loss": 0.7447,
+      "step": 60155
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.291038751602173,
+      "learning_rate": 1.1102373974470893e-08,
+      "loss": 0.3916,
+      "step": 60156
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 1.800761342048645,
+      "learning_rate": 1.1084820367265325e-08,
+      "loss": 0.3462,
+      "step": 60157
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.463101387023926,
+      "learning_rate": 1.1067280632364175e-08,
+      "loss": 0.433,
+      "step": 60158
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.000807285308838,
+      "learning_rate": 1.104975476981629e-08,
+      "loss": 0.533,
+      "step": 60159
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.5558876991271973,
+      "learning_rate": 1.1032242779669411e-08,
+      "loss": 0.5314,
+      "step": 60160
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.6699137687683105,
+      "learning_rate": 1.1014744661973497e-08,
+      "loss": 0.613,
+      "step": 60161
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.4387173652648926,
+      "learning_rate": 1.099726041677629e-08,
+      "loss": 0.8557,
+      "step": 60162
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.6662991046905518,
+      "learning_rate": 1.0979790044127193e-08,
+      "loss": 0.7225,
+      "step": 60163
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.7344539165496826,
+      "learning_rate": 1.0962333544073945e-08,
+      "loss": 0.3648,
+      "step": 60164
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.6262362003326416,
+      "learning_rate": 1.0944890916665952e-08,
+      "loss": 0.5318,
+      "step": 60165
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 4.839778900146484,
+      "learning_rate": 1.0927462161950952e-08,
+      "loss": 0.8193,
+      "step": 60166
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 4.1104512214660645,
+      "learning_rate": 1.0910047279977797e-08,
+      "loss": 0.5685,
+      "step": 60167
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.155488967895508,
+      "learning_rate": 1.0892646270794782e-08,
+      "loss": 0.5624,
+      "step": 60168
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 4.028041362762451,
+      "learning_rate": 1.0875259134450755e-08,
+      "loss": 0.6106,
+      "step": 60169
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.793491840362549,
+      "learning_rate": 1.0857885870992902e-08,
+      "loss": 0.5814,
+      "step": 60170
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 4.486532211303711,
+      "learning_rate": 1.0840526480471181e-08,
+      "loss": 0.778,
+      "step": 60171
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.025573492050171,
+      "learning_rate": 1.0823180962932222e-08,
+      "loss": 0.9143,
+      "step": 60172
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.516348123550415,
+      "learning_rate": 1.0805849318425432e-08,
+      "loss": 0.3038,
+      "step": 60173
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.178554058074951,
+      "learning_rate": 1.0788531546998549e-08,
+      "loss": 0.5511,
+      "step": 60174
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.7475225925445557,
+      "learning_rate": 1.0771227648699311e-08,
+      "loss": 0.5965,
+      "step": 60175
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.6325159072875977,
+      "learning_rate": 1.0753937623577127e-08,
+      "loss": 0.6103,
+      "step": 60176
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.831115961074829,
+      "learning_rate": 1.0736661471678621e-08,
+      "loss": 0.465,
+      "step": 60177
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.6777453422546387,
+      "learning_rate": 1.0719399193052649e-08,
+      "loss": 0.6398,
+      "step": 60178
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.191913366317749,
+      "learning_rate": 1.070215078774639e-08,
+      "loss": 0.4329,
+      "step": 60179
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.4809587001800537,
+      "learning_rate": 1.0684916255809253e-08,
+      "loss": 0.5692,
+      "step": 60180
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.693694591522217,
+      "learning_rate": 1.0667695597287864e-08,
+      "loss": 0.4624,
+      "step": 60181
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 7.234747886657715,
+      "learning_rate": 1.0650488812231074e-08,
+      "loss": 0.4842,
+      "step": 60182
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.5251972675323486,
+      "learning_rate": 1.0633295900685514e-08,
+      "loss": 0.4422,
+      "step": 60183
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.949341058731079,
+      "learning_rate": 1.0616116862700032e-08,
+      "loss": 0.5999,
+      "step": 60184
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.245004177093506,
+      "learning_rate": 1.0598951698322369e-08,
+      "loss": 0.5372,
+      "step": 60185
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.0454092025756836,
+      "learning_rate": 1.0581800407599708e-08,
+      "loss": 0.5419,
+      "step": 60186
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.8198256492614746,
+      "learning_rate": 1.0564662990579788e-08,
+      "loss": 0.6619,
+      "step": 60187
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.1756958961486816,
+      "learning_rate": 1.0547539447310906e-08,
+      "loss": 0.6267,
+      "step": 60188
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 4.121792316436768,
+      "learning_rate": 1.053042977783969e-08,
+      "loss": 0.7831,
+      "step": 60189
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.1821236610412598,
+      "learning_rate": 1.0513333982214436e-08,
+      "loss": 0.5492,
+      "step": 60190
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.8002493381500244,
+      "learning_rate": 1.0496252060482326e-08,
+      "loss": 0.6288,
+      "step": 60191
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.780032157897949,
+      "learning_rate": 1.0479184012691101e-08,
+      "loss": 0.9159,
+      "step": 60192
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.405802011489868,
+      "learning_rate": 1.0462129838888502e-08,
+      "loss": 0.5435,
+      "step": 60193
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 4.748373985290527,
+      "learning_rate": 1.0445089539121711e-08,
+      "loss": 0.4852,
+      "step": 60194
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.5138115882873535,
+      "learning_rate": 1.0428063113437358e-08,
+      "loss": 0.6824,
+      "step": 60195
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.0762791633605957,
+      "learning_rate": 1.041105056188374e-08,
+      "loss": 0.3968,
+      "step": 60196
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.2584798336029053,
+      "learning_rate": 1.0394051884508039e-08,
+      "loss": 0.4996,
+      "step": 60197
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 4.688658714294434,
+      "learning_rate": 1.0377067081356883e-08,
+      "loss": 0.8186,
+      "step": 60198
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.9034857749938965,
+      "learning_rate": 1.0360096152478016e-08,
+      "loss": 0.3308,
+      "step": 60199
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.7024776935577393,
+      "learning_rate": 1.0343139097919174e-08,
+      "loss": 0.3407,
+      "step": 60200
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 4.323941230773926,
+      "learning_rate": 1.0326195917726433e-08,
+      "loss": 0.6977,
+      "step": 60201
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.8732423782348633,
+      "learning_rate": 1.0309266611947533e-08,
+      "loss": 0.4911,
+      "step": 60202
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 6.713048458099365,
+      "learning_rate": 1.0292351180629101e-08,
+      "loss": 0.3819,
+      "step": 60203
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 4.8108720779418945,
+      "learning_rate": 1.0275449623818324e-08,
+      "loss": 0.6989,
+      "step": 60204
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.1579978466033936,
+      "learning_rate": 1.025856194156294e-08,
+      "loss": 0.483,
+      "step": 60205
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.3540308475494385,
+      "learning_rate": 1.0241688133909022e-08,
+      "loss": 0.6,
+      "step": 60206
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.36474347114563,
+      "learning_rate": 1.0224828200903758e-08,
+      "loss": 0.6999,
+      "step": 60207
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.3244569301605225,
+      "learning_rate": 1.0207982142593775e-08,
+      "loss": 0.4488,
+      "step": 60208
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.4532675743103027,
+      "learning_rate": 1.0191149959026258e-08,
+      "loss": 0.429,
+      "step": 60209
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.4379775524139404,
+      "learning_rate": 1.0174331650248392e-08,
+      "loss": 0.4177,
+      "step": 60210
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 4.889495849609375,
+      "learning_rate": 1.0157527216305696e-08,
+      "loss": 0.4663,
+      "step": 60211
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.837477922439575,
+      "learning_rate": 1.0140736657246464e-08,
+      "loss": 0.4046,
+      "step": 60212
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.471712827682495,
+      "learning_rate": 1.012395997311566e-08,
+      "loss": 0.4714,
+      "step": 60213
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.657498359680176,
+      "learning_rate": 1.010719716396158e-08,
+      "loss": 0.4807,
+      "step": 60214
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.5203442573547363,
+      "learning_rate": 1.0090448229830297e-08,
+      "loss": 0.437,
+      "step": 60215
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.9344074726104736,
+      "learning_rate": 1.0073713170767885e-08,
+      "loss": 0.4399,
+      "step": 60216
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.6465585231781006,
+      "learning_rate": 1.0056991986820974e-08,
+      "loss": 0.8288,
+      "step": 60217
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 4.60861349105835,
+      "learning_rate": 1.0040284678036749e-08,
+      "loss": 0.5701,
+      "step": 60218
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.8409578800201416,
+      "learning_rate": 1.0023591244461284e-08,
+      "loss": 1.0408,
+      "step": 60219
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.596940517425537,
+      "learning_rate": 1.0006911686140652e-08,
+      "loss": 0.6085,
+      "step": 60220
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.053288221359253,
+      "learning_rate": 9.990246003121485e-09,
+      "loss": 0.6344,
+      "step": 60221
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.513123035430908,
+      "learning_rate": 9.973594195449853e-09,
+      "loss": 0.4849,
+      "step": 60222
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 4.680856704711914,
+      "learning_rate": 9.956956263172945e-09,
+      "loss": 0.7139,
+      "step": 60223
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 5.487244606018066,
+      "learning_rate": 9.940332206336278e-09,
+      "loss": 0.5586,
+      "step": 60224
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.9364137649536133,
+      "learning_rate": 9.923722024985928e-09,
+      "loss": 0.3585,
+      "step": 60225
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.049027442932129,
+      "learning_rate": 9.90712571916852e-09,
+      "loss": 0.4231,
+      "step": 60226
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.3626527786254883,
+      "learning_rate": 9.890543288930132e-09,
+      "loss": 0.4361,
+      "step": 60227
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 4.890777111053467,
+      "learning_rate": 9.873974734316838e-09,
+      "loss": 0.6117,
+      "step": 60228
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.5626609325408936,
+      "learning_rate": 9.857420055374711e-09,
+      "loss": 0.7044,
+      "step": 60229
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.9292757511138916,
+      "learning_rate": 9.84087925214927e-09,
+      "loss": 0.7313,
+      "step": 60230
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.371335983276367,
+      "learning_rate": 9.824352324687702e-09,
+      "loss": 0.7316,
+      "step": 60231
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.798820734024048,
+      "learning_rate": 9.807839273034968e-09,
+      "loss": 0.3884,
+      "step": 60232
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.093486785888672,
+      "learning_rate": 9.791340097237145e-09,
+      "loss": 0.6943,
+      "step": 60233
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.3544232845306396,
+      "learning_rate": 9.774854797340305e-09,
+      "loss": 0.8267,
+      "step": 60234
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.4288296699523926,
+      "learning_rate": 9.758383373389967e-09,
+      "loss": 0.5117,
+      "step": 60235
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.5343832969665527,
+      "learning_rate": 9.741925825432763e-09,
+      "loss": 0.6935,
+      "step": 60236
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.465475082397461,
+      "learning_rate": 9.725482153513099e-09,
+      "loss": 0.3391,
+      "step": 60237
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.3775322437286377,
+      "learning_rate": 9.709052357678162e-09,
+      "loss": 0.5434,
+      "step": 60238
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.3256947994232178,
+      "learning_rate": 9.692636437972913e-09,
+      "loss": 0.4043,
+      "step": 60239
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.0796549320220947,
+      "learning_rate": 9.67623439444232e-09,
+      "loss": 0.3848,
+      "step": 60240
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.1820597648620605,
+      "learning_rate": 9.659846227133563e-09,
+      "loss": 0.4952,
+      "step": 60241
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.7865779399871826,
+      "learning_rate": 9.643471936090498e-09,
+      "loss": 0.5148,
+      "step": 60242
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.7588694095611572,
+      "learning_rate": 9.627111521360311e-09,
+      "loss": 0.6146,
+      "step": 60243
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.30220365524292,
+      "learning_rate": 9.610764982987408e-09,
+      "loss": 0.601,
+      "step": 60244
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.944293737411499,
+      "learning_rate": 9.594432321017311e-09,
+      "loss": 0.4659,
+      "step": 60245
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.8584237098693848,
+      "learning_rate": 9.578113535496091e-09,
+      "loss": 0.6896,
+      "step": 60246
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.5999321937561035,
+      "learning_rate": 9.561808626468162e-09,
+      "loss": 0.5081,
+      "step": 60247
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 5.039510726928711,
+      "learning_rate": 9.545517593979593e-09,
+      "loss": 0.4867,
+      "step": 60248
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.5777618885040283,
+      "learning_rate": 9.529240438075904e-09,
+      "loss": 0.4112,
+      "step": 60249
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.6114308834075928,
+      "learning_rate": 9.512977158800951e-09,
+      "loss": 0.6384,
+      "step": 60250
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.187906503677368,
+      "learning_rate": 9.496727756201918e-09,
+      "loss": 0.5035,
+      "step": 60251
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 4.683199405670166,
+      "learning_rate": 9.480492230322103e-09,
+      "loss": 0.5159,
+      "step": 60252
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.7553067207336426,
+      "learning_rate": 9.464270581208135e-09,
+      "loss": 0.6607,
+      "step": 60253
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.356794595718384,
+      "learning_rate": 9.448062808904424e-09,
+      "loss": 0.6421,
+      "step": 60254
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 4.869533538818359,
+      "learning_rate": 9.431868913456487e-09,
+      "loss": 0.402,
+      "step": 60255
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.2375411987304688,
+      "learning_rate": 9.415688894908182e-09,
+      "loss": 0.6202,
+      "step": 60256
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.9853029251098633,
+      "learning_rate": 9.399522753306134e-09,
+      "loss": 0.4967,
+      "step": 60257
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.8376355171203613,
+      "learning_rate": 9.383370488694199e-09,
+      "loss": 0.6467,
+      "step": 60258
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.450765609741211,
+      "learning_rate": 9.36723210111734e-09,
+      "loss": 0.6166,
+      "step": 60259
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.818207263946533,
+      "learning_rate": 9.351107590621077e-09,
+      "loss": 0.5585,
+      "step": 60260
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.5249032974243164,
+      "learning_rate": 9.33499695724982e-09,
+      "loss": 0.4989,
+      "step": 60261
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.759799003601074,
+      "learning_rate": 9.318900201047975e-09,
+      "loss": 0.5075,
+      "step": 60262
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.3669753074645996,
+      "learning_rate": 9.302817322061064e-09,
+      "loss": 0.6239,
+      "step": 60263
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.9816315174102783,
+      "learning_rate": 9.286748320333494e-09,
+      "loss": 0.6776,
+      "step": 60264
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.259756565093994,
+      "learning_rate": 9.270693195909675e-09,
+      "loss": 0.4219,
+      "step": 60265
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.979191303253174,
+      "learning_rate": 9.25465194883457e-09,
+      "loss": 0.6334,
+      "step": 60266
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.7953200340270996,
+      "learning_rate": 9.238624579153143e-09,
+      "loss": 0.5966,
+      "step": 60267
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.636030435562134,
+      "learning_rate": 9.222611086908695e-09,
+      "loss": 0.5268,
+      "step": 60268
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.565532684326172,
+      "learning_rate": 9.206611472147298e-09,
+      "loss": 0.5686,
+      "step": 60269
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.728947162628174,
+      "learning_rate": 9.19062573491225e-09,
+      "loss": 0.5793,
+      "step": 60270
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.0782463550567627,
+      "learning_rate": 9.174653875248518e-09,
+      "loss": 0.7717,
+      "step": 60271
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.4745051860809326,
+      "learning_rate": 9.15869589320051e-09,
+      "loss": 0.6757,
+      "step": 60272
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.7402756214141846,
+      "learning_rate": 9.142751788812632e-09,
+      "loss": 0.531,
+      "step": 60273
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.2067103385925293,
+      "learning_rate": 9.126821562129295e-09,
+      "loss": 0.8487,
+      "step": 60274
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.653932571411133,
+      "learning_rate": 9.110905213194354e-09,
+      "loss": 0.5874,
+      "step": 60275
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.683546543121338,
+      "learning_rate": 9.095002742052773e-09,
+      "loss": 0.5619,
+      "step": 60276
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.623807668685913,
+      "learning_rate": 9.079114148747848e-09,
+      "loss": 0.5356,
+      "step": 60277
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.5737922191619873,
+      "learning_rate": 9.063239433324544e-09,
+      "loss": 0.4865,
+      "step": 60278
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.1268088817596436,
+      "learning_rate": 9.047378595826716e-09,
+      "loss": 0.842,
+      "step": 60279
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.559804677963257,
+      "learning_rate": 9.031531636298218e-09,
+      "loss": 0.5679,
+      "step": 60280
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 4.149755954742432,
+      "learning_rate": 9.015698554784014e-09,
+      "loss": 0.5633,
+      "step": 60281
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 5.28482723236084,
+      "learning_rate": 8.999879351326846e-09,
+      "loss": 0.4182,
+      "step": 60282
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.4315617084503174,
+      "learning_rate": 8.984074025971678e-09,
+      "loss": 0.4614,
+      "step": 60283
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.995483160018921,
+      "learning_rate": 8.96828257876181e-09,
+      "loss": 0.612,
+      "step": 60284
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.757516384124756,
+      "learning_rate": 8.952505009741652e-09,
+      "loss": 0.412,
+      "step": 60285
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 6.4242939949035645,
+      "learning_rate": 8.936741318955055e-09,
+      "loss": 0.4189,
+      "step": 60286
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.7484030723571777,
+      "learning_rate": 8.92099150644532e-09,
+      "loss": 0.5301,
+      "step": 60287
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 4.786787033081055,
+      "learning_rate": 8.905255572256855e-09,
+      "loss": 0.523,
+      "step": 60288
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.4896020889282227,
+      "learning_rate": 8.889533516432958e-09,
+      "loss": 0.6669,
+      "step": 60289
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.208915948867798,
+      "learning_rate": 8.873825339018038e-09,
+      "loss": 0.5603,
+      "step": 60290
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.4231791496276855,
+      "learning_rate": 8.85813104005484e-09,
+      "loss": 0.7392,
+      "step": 60291
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.1283323764801025,
+      "learning_rate": 8.842450619587773e-09,
+      "loss": 0.8165,
+      "step": 60292
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.9919216632843018,
+      "learning_rate": 8.826784077659578e-09,
+      "loss": 0.4936,
+      "step": 60293
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.571715831756592,
+      "learning_rate": 8.811131414314666e-09,
+      "loss": 0.3729,
+      "step": 60294
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.0787322521209717,
+      "learning_rate": 8.795492629596336e-09,
+      "loss": 0.4609,
+      "step": 60295
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 5.656162261962891,
+      "learning_rate": 8.779867723547886e-09,
+      "loss": 0.5011,
+      "step": 60296
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.224616765975952,
+      "learning_rate": 8.76425669621317e-09,
+      "loss": 0.3845,
+      "step": 60297
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.8177716732025146,
+      "learning_rate": 8.74865954763493e-09,
+      "loss": 0.6929,
+      "step": 60298
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 4.533016204833984,
+      "learning_rate": 8.733076277857023e-09,
+      "loss": 0.5996,
+      "step": 60299
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.9102845191955566,
+      "learning_rate": 8.717506886922743e-09,
+      "loss": 0.61,
+      "step": 60300
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.3973710536956787,
+      "learning_rate": 8.701951374875395e-09,
+      "loss": 0.4924,
+      "step": 60301
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.6368637084960938,
+      "learning_rate": 8.686409741757718e-09,
+      "loss": 0.7219,
+      "step": 60302
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.525972366333008,
+      "learning_rate": 8.670881987613567e-09,
+      "loss": 0.6243,
+      "step": 60303
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.3077328205108643,
+      "learning_rate": 8.65536811248624e-09,
+      "loss": 0.3444,
+      "step": 60304
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.187636375427246,
+      "learning_rate": 8.639868116417926e-09,
+      "loss": 0.8329,
+      "step": 60305
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.6224913597106934,
+      "learning_rate": 8.624381999453036e-09,
+      "loss": 0.5946,
+      "step": 60306
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.863288640975952,
+      "learning_rate": 8.6089097616332e-09,
+      "loss": 0.5248,
+      "step": 60307
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 4.21610689163208,
+      "learning_rate": 8.593451403002829e-09,
+      "loss": 0.3858,
+      "step": 60308
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.995758533477783,
+      "learning_rate": 8.578006923603554e-09,
+      "loss": 0.5476,
+      "step": 60309
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 4.8164191246032715,
+      "learning_rate": 8.562576323479787e-09,
+      "loss": 0.7076,
+      "step": 60310
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.4026904106140137,
+      "learning_rate": 8.547159602673161e-09,
+      "loss": 0.4057,
+      "step": 60311
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.3782856464385986,
+      "learning_rate": 8.531756761226973e-09,
+      "loss": 0.4552,
+      "step": 60312
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.741503953933716,
+      "learning_rate": 8.516367799184522e-09,
+      "loss": 0.7146,
+      "step": 60313
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 4.025196075439453,
+      "learning_rate": 8.500992716587998e-09,
+      "loss": 0.4628,
+      "step": 60314
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 6.115457534790039,
+      "learning_rate": 8.485631513480141e-09,
+      "loss": 0.5256,
+      "step": 60315
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.381639242172241,
+      "learning_rate": 8.470284189904254e-09,
+      "loss": 0.7087,
+      "step": 60316
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.35524845123291,
+      "learning_rate": 8.454950745901969e-09,
+      "loss": 0.4309,
+      "step": 60317
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.0074024200439453,
+      "learning_rate": 8.439631181517139e-09,
+      "loss": 0.5572,
+      "step": 60318
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 4.580905437469482,
+      "learning_rate": 8.424325496791951e-09,
+      "loss": 0.656,
+      "step": 60319
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.0989632606506348,
+      "learning_rate": 8.409033691768599e-09,
+      "loss": 0.5889,
+      "step": 60320
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 1.7715849876403809,
+      "learning_rate": 8.393755766489819e-09,
+      "loss": 0.3468,
+      "step": 60321
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.4123375415802,
+      "learning_rate": 8.37849172099836e-09,
+      "loss": 0.4767,
+      "step": 60322
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.39129638671875,
+      "learning_rate": 8.363241555335855e-09,
+      "loss": 0.7274,
+      "step": 60323
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.1014981269836426,
+      "learning_rate": 8.3480052695456e-09,
+      "loss": 0.4299,
+      "step": 60324
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.9083919525146484,
+      "learning_rate": 8.332782863669785e-09,
+      "loss": 0.6078,
+      "step": 60325
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 4.855077266693115,
+      "learning_rate": 8.317574337750045e-09,
+      "loss": 0.4671,
+      "step": 60326
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.861366033554077,
+      "learning_rate": 8.302379691829676e-09,
+      "loss": 0.6034,
+      "step": 60327
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.9019064903259277,
+      "learning_rate": 8.287198925950312e-09,
+      "loss": 0.5558,
+      "step": 60328
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.3854002952575684,
+      "learning_rate": 8.272032040154699e-09,
+      "loss": 0.4388,
+      "step": 60329
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.010521173477173,
+      "learning_rate": 8.256879034483911e-09,
+      "loss": 0.8033,
+      "step": 60330
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.3387815952301025,
+      "learning_rate": 8.24173990898125e-09,
+      "loss": 0.5988,
+      "step": 60331
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.3735859394073486,
+      "learning_rate": 8.226614663688349e-09,
+      "loss": 0.5869,
+      "step": 60332
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.6006877422332764,
+      "learning_rate": 8.211503298646838e-09,
+      "loss": 0.7837,
+      "step": 60333
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.6032161712646484,
+      "learning_rate": 8.196405813900022e-09,
+      "loss": 0.551,
+      "step": 60334
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 1.9031000137329102,
+      "learning_rate": 8.181322209488417e-09,
+      "loss": 0.3843,
+      "step": 60335
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.1102325916290283,
+      "learning_rate": 8.166252485454773e-09,
+      "loss": 0.7019,
+      "step": 60336
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.145625352859497,
+      "learning_rate": 8.151196641840165e-09,
+      "loss": 0.466,
+      "step": 60337
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.5230250358581543,
+      "learning_rate": 8.13615467868789e-09,
+      "loss": 0.6424,
+      "step": 60338
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.888295888900757,
+      "learning_rate": 8.121126596038476e-09,
+      "loss": 0.6267,
+      "step": 60339
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.191235065460205,
+      "learning_rate": 8.106112393934663e-09,
+      "loss": 0.739,
+      "step": 60340
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.8724489212036133,
+      "learning_rate": 8.091112072417529e-09,
+      "loss": 0.5765,
+      "step": 60341
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 1.9958852529525757,
+      "learning_rate": 8.076125631529264e-09,
+      "loss": 0.5089,
+      "step": 60342
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.896817684173584,
+      "learning_rate": 8.061153071310945e-09,
+      "loss": 0.7438,
+      "step": 60343
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.7704246044158936,
+      "learning_rate": 8.046194391804207e-09,
+      "loss": 0.5451,
+      "step": 60344
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.1063811779022217,
+      "learning_rate": 8.031249593051793e-09,
+      "loss": 0.5811,
+      "step": 60345
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.229473114013672,
+      "learning_rate": 8.01631867509367e-09,
+      "loss": 0.3881,
+      "step": 60346
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.845785140991211,
+      "learning_rate": 8.001401637972584e-09,
+      "loss": 0.8613,
+      "step": 60347
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.8520171642303467,
+      "learning_rate": 7.98649848172961e-09,
+      "loss": 0.3354,
+      "step": 60348
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.3742353916168213,
+      "learning_rate": 7.971609206405829e-09,
+      "loss": 0.5661,
+      "step": 60349
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.079265832901001,
+      "learning_rate": 7.956733812042872e-09,
+      "loss": 0.3577,
+      "step": 60350
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 33.3842887878418,
+      "learning_rate": 7.941872298682374e-09,
+      "loss": 0.5723,
+      "step": 60351
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.1875665187835693,
+      "learning_rate": 7.927024666365412e-09,
+      "loss": 0.8411,
+      "step": 60352
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 4.838754177093506,
+      "learning_rate": 7.912190915133621e-09,
+      "loss": 0.9069,
+      "step": 60353
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.10231876373291,
+      "learning_rate": 7.897371045027524e-09,
+      "loss": 0.9905,
+      "step": 60354
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.476320266723633,
+      "learning_rate": 7.882565056088754e-09,
+      "loss": 0.6295,
+      "step": 60355
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.240987539291382,
+      "learning_rate": 7.867772948358942e-09,
+      "loss": 0.4116,
+      "step": 60356
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.357935667037964,
+      "learning_rate": 7.852994721878615e-09,
+      "loss": 0.7268,
+      "step": 60357
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 4.210336208343506,
+      "learning_rate": 7.838230376688849e-09,
+      "loss": 0.4898,
+      "step": 60358
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.652103900909424,
+      "learning_rate": 7.823479912830722e-09,
+      "loss": 0.6113,
+      "step": 60359
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.51578950881958,
+      "learning_rate": 7.808743330345314e-09,
+      "loss": 0.548,
+      "step": 60360
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.7658321857452393,
+      "learning_rate": 7.794020629274258e-09,
+      "loss": 0.4517,
+      "step": 60361
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.412123918533325,
+      "learning_rate": 7.779311809657519e-09,
+      "loss": 0.5821,
+      "step": 60362
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.990511655807495,
+      "learning_rate": 7.76461687153618e-09,
+      "loss": 0.5686,
+      "step": 60363
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.7715258598327637,
+      "learning_rate": 7.74993581495187e-09,
+      "loss": 0.7263,
+      "step": 60364
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 1.8333866596221924,
+      "learning_rate": 7.735268639944559e-09,
+      "loss": 0.4618,
+      "step": 60365
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.3303167819976807,
+      "learning_rate": 7.720615346555326e-09,
+      "loss": 0.5903,
+      "step": 60366
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 4.260060787200928,
+      "learning_rate": 7.705975934824694e-09,
+      "loss": 0.549,
+      "step": 60367
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.8489885330200195,
+      "learning_rate": 7.69135040479374e-09,
+      "loss": 0.5963,
+      "step": 60368
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.621824026107788,
+      "learning_rate": 7.676738756502988e-09,
+      "loss": 0.542,
+      "step": 60369
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.1293957233428955,
+      "learning_rate": 7.66214098999296e-09,
+      "loss": 0.7772,
+      "step": 60370
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.4567325115203857,
+      "learning_rate": 7.647557105304737e-09,
+      "loss": 0.6354,
+      "step": 60371
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.2010977268218994,
+      "learning_rate": 7.632987102478285e-09,
+      "loss": 0.7042,
+      "step": 60372
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.0989303588867188,
+      "learning_rate": 7.618430981554126e-09,
+      "loss": 0.456,
+      "step": 60373
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 4.170089244842529,
+      "learning_rate": 7.603888742572785e-09,
+      "loss": 0.6093,
+      "step": 60374
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.4034383296966553,
+      "learning_rate": 7.58936038557534e-09,
+      "loss": 0.7205,
+      "step": 60375
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 1.5921365022659302,
+      "learning_rate": 7.574845910601203e-09,
+      "loss": 0.1553,
+      "step": 60376
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.428049325942993,
+      "learning_rate": 7.560345317691453e-09,
+      "loss": 0.6821,
+      "step": 60377
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.223714590072632,
+      "learning_rate": 7.545858606886058e-09,
+      "loss": 0.7022,
+      "step": 60378
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.8296899795532227,
+      "learning_rate": 7.531385778225541e-09,
+      "loss": 0.6933,
+      "step": 60379
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.4863381385803223,
+      "learning_rate": 7.51692683174987e-09,
+      "loss": 0.5627,
+      "step": 60380
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.442897319793701,
+      "learning_rate": 7.50248176749957e-09,
+      "loss": 0.485,
+      "step": 60381
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.6682119369506836,
+      "learning_rate": 7.488050585514605e-09,
+      "loss": 0.5997,
+      "step": 60382
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.6655356884002686,
+      "learning_rate": 7.473633285834947e-09,
+      "loss": 0.4779,
+      "step": 60383
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.800920009613037,
+      "learning_rate": 7.459229868501117e-09,
+      "loss": 0.5515,
+      "step": 60384
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.618405818939209,
+      "learning_rate": 7.444840333553083e-09,
+      "loss": 0.6862,
+      "step": 60385
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.6076340675354004,
+      "learning_rate": 7.430464681030258e-09,
+      "loss": 0.4762,
+      "step": 60386
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.7175257205963135,
+      "learning_rate": 7.416102910973166e-09,
+      "loss": 0.473,
+      "step": 60387
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.7796506881713867,
+      "learning_rate": 7.401755023421775e-09,
+      "loss": 0.8946,
+      "step": 60388
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.956360340118408,
+      "learning_rate": 7.387421018416052e-09,
+      "loss": 0.4816,
+      "step": 60389
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.1708014011383057,
+      "learning_rate": 7.373100895994856e-09,
+      "loss": 0.342,
+      "step": 60390
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.832808494567871,
+      "learning_rate": 7.3587946561998194e-09,
+      "loss": 0.6751,
+      "step": 60391
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.6856796741485596,
+      "learning_rate": 7.34450229906869e-09,
+      "loss": 0.4828,
+      "step": 60392
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.5352444648742676,
+      "learning_rate": 7.330223824643101e-09,
+      "loss": 0.7482,
+      "step": 60393
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.382683038711548,
+      "learning_rate": 7.315959232960801e-09,
+      "loss": 0.5272,
+      "step": 60394
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 4.085788726806641,
+      "learning_rate": 7.301708524063422e-09,
+      "loss": 0.6107,
+      "step": 60395
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.7024240493774414,
+      "learning_rate": 7.287471697989268e-09,
+      "loss": 0.6838,
+      "step": 60396
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 4.583050727844238,
+      "learning_rate": 7.273248754778306e-09,
+      "loss": 0.6391,
+      "step": 60397
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.642120599746704,
+      "learning_rate": 7.2590396944705045e-09,
+      "loss": 0.5622,
+      "step": 60398
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.0843276977539062,
+      "learning_rate": 7.244844517104721e-09,
+      "loss": 0.4686,
+      "step": 60399
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.704066753387451,
+      "learning_rate": 7.230663222720924e-09,
+      "loss": 0.4919,
+      "step": 60400
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.969510793685913,
+      "learning_rate": 7.216495811357971e-09,
+      "loss": 0.7779,
+      "step": 60401
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.0371477603912354,
+      "learning_rate": 7.20234228305583e-09,
+      "loss": 0.4868,
+      "step": 60402
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.5182628631591797,
+      "learning_rate": 7.188202637853914e-09,
+      "loss": 0.3644,
+      "step": 60403
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.264742612838745,
+      "learning_rate": 7.174076875790525e-09,
+      "loss": 0.6247,
+      "step": 60404
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.7386844158172607,
+      "learning_rate": 7.159964996906188e-09,
+      "loss": 0.603,
+      "step": 60405
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.514920473098755,
+      "learning_rate": 7.145867001239204e-09,
+      "loss": 0.4547,
+      "step": 60406
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 5.476569175720215,
+      "learning_rate": 7.131782888829541e-09,
+      "loss": 0.5498,
+      "step": 60407
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.8608851432800293,
+      "learning_rate": 7.117712659716058e-09,
+      "loss": 0.5937,
+      "step": 60408
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 4.418087482452393,
+      "learning_rate": 7.103656313938168e-09,
+      "loss": 0.7368,
+      "step": 60409
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.178597927093506,
+      "learning_rate": 7.089613851534171e-09,
+      "loss": 0.6992,
+      "step": 60410
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.6897761821746826,
+      "learning_rate": 7.075585272543483e-09,
+      "loss": 0.5617,
+      "step": 60411
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.1651668548583984,
+      "learning_rate": 7.0615705770049615e-09,
+      "loss": 0.5116,
+      "step": 60412
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.760313034057617,
+      "learning_rate": 7.047569764958573e-09,
+      "loss": 0.7137,
+      "step": 60413
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.7037529945373535,
+      "learning_rate": 7.033582836442065e-09,
+      "loss": 0.5468,
+      "step": 60414
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.892672300338745,
+      "learning_rate": 7.019609791494853e-09,
+      "loss": 0.471,
+      "step": 60415
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.279545307159424,
+      "learning_rate": 7.005650630155236e-09,
+      "loss": 0.628,
+      "step": 60416
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.883640766143799,
+      "learning_rate": 6.991705352463185e-09,
+      "loss": 0.5311,
+      "step": 60417
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.9652369022369385,
+      "learning_rate": 6.9777739584564465e-09,
+      "loss": 0.473,
+      "step": 60418
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.890629768371582,
+      "learning_rate": 6.963856448173878e-09,
+      "loss": 0.5947,
+      "step": 60419
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.262899398803711,
+      "learning_rate": 6.949952821654893e-09,
+      "loss": 0.5415,
+      "step": 60420
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.3499374389648438,
+      "learning_rate": 6.936063078937238e-09,
+      "loss": 0.5249,
+      "step": 60421
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.5432143211364746,
+      "learning_rate": 6.922187220060328e-09,
+      "loss": 0.7973,
+      "step": 60422
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.1911964416503906,
+      "learning_rate": 6.908325245061909e-09,
+      "loss": 0.7958,
+      "step": 60423
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.555225372314453,
+      "learning_rate": 6.8944771539813935e-09,
+      "loss": 0.486,
+      "step": 60424
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.609241247177124,
+      "learning_rate": 6.880642946856531e-09,
+      "loss": 0.7165,
+      "step": 60425
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.4278438091278076,
+      "learning_rate": 6.866822623726177e-09,
+      "loss": 0.5214,
+      "step": 60426
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.5083606243133545,
+      "learning_rate": 6.853016184629191e-09,
+      "loss": 0.8272,
+      "step": 60427
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.647974967956543,
+      "learning_rate": 6.8392236296027644e-09,
+      "loss": 0.728,
+      "step": 60428
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.1650185585021973,
+      "learning_rate": 6.8254449586863116e-09,
+      "loss": 0.4333,
+      "step": 60429
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.3504068851470947,
+      "learning_rate": 6.811680171918134e-09,
+      "loss": 0.6211,
+      "step": 60430
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.7641403675079346,
+      "learning_rate": 6.7979292693359785e-09,
+      "loss": 0.6447,
+      "step": 60431
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.9619221687316895,
+      "learning_rate": 6.78419225097815e-09,
+      "loss": 0.4932,
+      "step": 60432
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.3743138313293457,
+      "learning_rate": 6.770469116882949e-09,
+      "loss": 0.6915,
+      "step": 60433
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.818425416946411,
+      "learning_rate": 6.7567598670886805e-09,
+      "loss": 0.421,
+      "step": 60434
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.778696298599243,
+      "learning_rate": 6.74306450163309e-09,
+      "loss": 0.5909,
+      "step": 60435
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.919908285140991,
+      "learning_rate": 6.729383020555036e-09,
+      "loss": 0.4048,
+      "step": 60436
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 4.820588111877441,
+      "learning_rate": 6.715715423891711e-09,
+      "loss": 0.4493,
+      "step": 60437
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.910884380340576,
+      "learning_rate": 6.702061711681418e-09,
+      "loss": 0.4869,
+      "step": 60438
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.5351486206054688,
+      "learning_rate": 6.688421883962459e-09,
+      "loss": 0.808,
+      "step": 60439
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.8810698986053467,
+      "learning_rate": 6.674795940772027e-09,
+      "loss": 0.5268,
+      "step": 60440
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.5252671241760254,
+      "learning_rate": 6.661183882148426e-09,
+      "loss": 0.7325,
+      "step": 60441
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.1504547595977783,
+      "learning_rate": 6.647585708129956e-09,
+      "loss": 0.6297,
+      "step": 60442
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.3032522201538086,
+      "learning_rate": 6.63400141875381e-09,
+      "loss": 0.8257,
+      "step": 60443
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.5514631271362305,
+      "learning_rate": 6.620431014058293e-09,
+      "loss": 0.5393,
+      "step": 60444
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 4.240435600280762,
+      "learning_rate": 6.60687449408004e-09,
+      "loss": 1.0032,
+      "step": 60445
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.734118938446045,
+      "learning_rate": 6.59333185885791e-09,
+      "loss": 0.5857,
+      "step": 60446
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.6970388889312744,
+      "learning_rate": 6.579803108429095e-09,
+      "loss": 0.4986,
+      "step": 60447
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.657269239425659,
+      "learning_rate": 6.5662882428313426e-09,
+      "loss": 0.6133,
+      "step": 60448
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 4.385453701019287,
+      "learning_rate": 6.5527872621024e-09,
+      "loss": 0.4837,
+      "step": 60449
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.3357346057891846,
+      "learning_rate": 6.539300166279461e-09,
+      "loss": 0.7029,
+      "step": 60450
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 4.118470191955566,
+      "learning_rate": 6.525826955399717e-09,
+      "loss": 0.7663,
+      "step": 60451
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.5420217514038086,
+      "learning_rate": 6.5123676295014706e-09,
+      "loss": 0.7633,
+      "step": 60452
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.23162579536438,
+      "learning_rate": 6.49892218862136e-09,
+      "loss": 0.5382,
+      "step": 60453
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.53586483001709,
+      "learning_rate": 6.485490632797686e-09,
+      "loss": 0.6019,
+      "step": 60454
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.7765510082244873,
+      "learning_rate": 6.472072962066534e-09,
+      "loss": 0.5986,
+      "step": 60455
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.817526340484619,
+      "learning_rate": 6.4586691764662034e-09,
+      "loss": 0.5812,
+      "step": 60456
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.638927936553955,
+      "learning_rate": 6.445279276033889e-09,
+      "loss": 0.4587,
+      "step": 60457
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.6807901859283447,
+      "learning_rate": 6.431903260806782e-09,
+      "loss": 0.4814,
+      "step": 60458
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 5.193289756774902,
+      "learning_rate": 6.418541130821521e-09,
+      "loss": 0.5382,
+      "step": 60459
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 4.336882591247559,
+      "learning_rate": 6.4051928861158516e-09,
+      "loss": 0.542,
+      "step": 60460
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.0845038890838623,
+      "learning_rate": 6.391858526726413e-09,
+      "loss": 0.6789,
+      "step": 60461
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.2303311824798584,
+      "learning_rate": 6.378538052690952e-09,
+      "loss": 0.6993,
+      "step": 60462
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.70241641998291,
+      "learning_rate": 6.365231464046106e-09,
+      "loss": 0.6124,
+      "step": 60463
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.59301495552063,
+      "learning_rate": 6.351938760828513e-09,
+      "loss": 0.4871,
+      "step": 60464
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.8433761596679688,
+      "learning_rate": 6.33865994307592e-09,
+      "loss": 0.5364,
+      "step": 60465
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.934009075164795,
+      "learning_rate": 6.325395010824964e-09,
+      "loss": 0.6679,
+      "step": 60466
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.4454805850982666,
+      "learning_rate": 6.312143964111728e-09,
+      "loss": 0.588,
+      "step": 60467
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.1907877922058105,
+      "learning_rate": 6.298906802974514e-09,
+      "loss": 0.3848,
+      "step": 60468
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.0644946098327637,
+      "learning_rate": 6.28568352744885e-09,
+      "loss": 0.5382,
+      "step": 60469
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.193995475769043,
+      "learning_rate": 6.272474137572482e-09,
+      "loss": 0.5626,
+      "step": 60470
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.790151596069336,
+      "learning_rate": 6.2592786333809384e-09,
+      "loss": 0.6481,
+      "step": 60471
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.567108392715454,
+      "learning_rate": 6.246097014911967e-09,
+      "loss": 0.631,
+      "step": 60472
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.4642348289489746,
+      "learning_rate": 6.232929282202205e-09,
+      "loss": 0.6299,
+      "step": 60473
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.906860828399658,
+      "learning_rate": 6.219775435287733e-09,
+      "loss": 0.5469,
+      "step": 60474
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.505859851837158,
+      "learning_rate": 6.206635474205192e-09,
+      "loss": 0.4921,
+      "step": 60475
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 4.525510311126709,
+      "learning_rate": 6.193509398991771e-09,
+      "loss": 0.5411,
+      "step": 60476
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.625335216522217,
+      "learning_rate": 6.1803972096829976e-09,
+      "loss": 0.4616,
+      "step": 60477
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.3517069816589355,
+      "learning_rate": 6.167298906316066e-09,
+      "loss": 0.4697,
+      "step": 60478
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.860405921936035,
+      "learning_rate": 6.154214488927057e-09,
+      "loss": 0.7609,
+      "step": 60479
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.6275012493133545,
+      "learning_rate": 6.141143957552054e-09,
+      "loss": 0.558,
+      "step": 60480
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 4.446737766265869,
+      "learning_rate": 6.128087312228248e-09,
+      "loss": 0.491,
+      "step": 60481
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 4.342499732971191,
+      "learning_rate": 6.115044552991723e-09,
+      "loss": 0.3956,
+      "step": 60482
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.172477960586548,
+      "learning_rate": 6.102015679878004e-09,
+      "loss": 0.7884,
+      "step": 60483
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.21942400932312,
+      "learning_rate": 6.08900069292373e-09,
+      "loss": 0.4368,
+      "step": 60484
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.811276435852051,
+      "learning_rate": 6.075999592165538e-09,
+      "loss": 0.4512,
+      "step": 60485
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.3829219341278076,
+      "learning_rate": 6.063012377638955e-09,
+      "loss": 0.4519,
+      "step": 60486
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 4.229171276092529,
+      "learning_rate": 6.050039049380618e-09,
+      "loss": 0.635,
+      "step": 60487
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.9738781452178955,
+      "learning_rate": 6.037079607426611e-09,
+      "loss": 0.594,
+      "step": 60488
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 6.665803909301758,
+      "learning_rate": 6.024134051812458e-09,
+      "loss": 0.5837,
+      "step": 60489
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.9800524711608887,
+      "learning_rate": 6.0112023825742435e-09,
+      "loss": 0.3755,
+      "step": 60490
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.1873319149017334,
+      "learning_rate": 5.99828459974805e-09,
+      "loss": 0.4996,
+      "step": 60491
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 4.703763961791992,
+      "learning_rate": 5.985380703369959e-09,
+      "loss": 0.5699,
+      "step": 60492
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 4.921908378601074,
+      "learning_rate": 5.972490693475497e-09,
+      "loss": 0.7146,
+      "step": 60493
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.4372758865356445,
+      "learning_rate": 5.959614570101302e-09,
+      "loss": 0.6349,
+      "step": 60494
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.797203540802002,
+      "learning_rate": 5.946752333281791e-09,
+      "loss": 0.7407,
+      "step": 60495
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.433184862136841,
+      "learning_rate": 5.933903983054157e-09,
+      "loss": 0.4333,
+      "step": 60496
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 5.096050262451172,
+      "learning_rate": 5.921069519453371e-09,
+      "loss": 0.6649,
+      "step": 60497
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.5510096549987793,
+      "learning_rate": 5.90824894251496e-09,
+      "loss": 0.6671,
+      "step": 60498
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.593212366104126,
+      "learning_rate": 5.895442252275563e-09,
+      "loss": 0.4764,
+      "step": 60499
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.376946210861206,
+      "learning_rate": 5.882649448769595e-09,
+      "loss": 0.6086,
+      "step": 60500
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.3469502925872803,
+      "learning_rate": 5.86987053203314e-09,
+      "loss": 0.6151,
+      "step": 60501
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.825556993484497,
+      "learning_rate": 5.857105502101723e-09,
+      "loss": 0.4368,
+      "step": 60502
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 5.020337104797363,
+      "learning_rate": 5.8443543590114286e-09,
+      "loss": 0.6177,
+      "step": 60503
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.650672435760498,
+      "learning_rate": 5.831617102796672e-09,
+      "loss": 0.6769,
+      "step": 60504
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.2152514457702637,
+      "learning_rate": 5.8188937334929805e-09,
+      "loss": 0.3355,
+      "step": 60505
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.9696273803710938,
+      "learning_rate": 5.806184251136438e-09,
+      "loss": 0.6001,
+      "step": 60506
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.203321933746338,
+      "learning_rate": 5.793488655762014e-09,
+      "loss": 0.4975,
+      "step": 60507
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.0544538497924805,
+      "learning_rate": 5.780806947405237e-09,
+      "loss": 0.5952,
+      "step": 60508
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.093311309814453,
+      "learning_rate": 5.76813912610108e-09,
+      "loss": 0.3483,
+      "step": 60509
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.400061845779419,
+      "learning_rate": 5.755485191884514e-09,
+      "loss": 0.3963,
+      "step": 60510
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.5388383865356445,
+      "learning_rate": 5.742845144791065e-09,
+      "loss": 0.7066,
+      "step": 60511
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.842545747756958,
+      "learning_rate": 5.730218984856262e-09,
+      "loss": 0.5268,
+      "step": 60512
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.8053064346313477,
+      "learning_rate": 5.717606712114521e-09,
+      "loss": 0.6464,
+      "step": 60513
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.839792251586914,
+      "learning_rate": 5.705008326601369e-09,
+      "loss": 0.464,
+      "step": 60514
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.5091981887817383,
+      "learning_rate": 5.6924238283517785e-09,
+      "loss": 0.5505,
+      "step": 60515
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.7716238498687744,
+      "learning_rate": 5.679853217400167e-09,
+      "loss": 0.5959,
+      "step": 60516
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.1723315715789795,
+      "learning_rate": 5.667296493782615e-09,
+      "loss": 0.4229,
+      "step": 60517
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.1090571880340576,
+      "learning_rate": 5.654753657532985e-09,
+      "loss": 0.5689,
+      "step": 60518
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 4.011288166046143,
+      "learning_rate": 5.642224708686806e-09,
+      "loss": 0.5601,
+      "step": 60519
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 4.238708019256592,
+      "learning_rate": 5.629709647278492e-09,
+      "loss": 0.5996,
+      "step": 60520
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.095944881439209,
+      "learning_rate": 5.6172084733430164e-09,
+      "loss": 0.6752,
+      "step": 60521
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.7684621810913086,
+      "learning_rate": 5.604721186915352e-09,
+      "loss": 0.6598,
+      "step": 60522
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.2792234420776367,
+      "learning_rate": 5.592247788029914e-09,
+      "loss": 0.3546,
+      "step": 60523
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.0542402267456055,
+      "learning_rate": 5.5797882767211206e-09,
+      "loss": 0.6235,
+      "step": 60524
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.0936691761016846,
+      "learning_rate": 5.5673426530244986e-09,
+      "loss": 0.559,
+      "step": 60525
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 4.242058753967285,
+      "learning_rate": 5.5549109169739104e-09,
+      "loss": 0.6504,
+      "step": 60526
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.788968324661255,
+      "learning_rate": 5.542493068604327e-09,
+      "loss": 0.4971,
+      "step": 60527
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.014893054962158,
+      "learning_rate": 5.53008910794961e-09,
+      "loss": 0.663,
+      "step": 60528
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 4.114935398101807,
+      "learning_rate": 5.517699035045288e-09,
+      "loss": 0.5452,
+      "step": 60529
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.840803623199463,
+      "learning_rate": 5.505322849925221e-09,
+      "loss": 0.5873,
+      "step": 60530
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.6666290760040283,
+      "learning_rate": 5.492960552623272e-09,
+      "loss": 0.3982,
+      "step": 60531
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.2471580505371094,
+      "learning_rate": 5.480612143174968e-09,
+      "loss": 0.5102,
+      "step": 60532
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.561109781265259,
+      "learning_rate": 5.468277621614171e-09,
+      "loss": 0.8065,
+      "step": 60533
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.4576473236083984,
+      "learning_rate": 5.4559569879747424e-09,
+      "loss": 0.5444,
+      "step": 60534
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.4509999752044678,
+      "learning_rate": 5.4436502422916536e-09,
+      "loss": 0.4353,
+      "step": 60535
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.458442211151123,
+      "learning_rate": 5.431357384598213e-09,
+      "loss": 0.6177,
+      "step": 60536
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.846205711364746,
+      "learning_rate": 5.419078414929946e-09,
+      "loss": 0.77,
+      "step": 60537
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.205629825592041,
+      "learning_rate": 5.406813333319605e-09,
+      "loss": 0.5226,
+      "step": 60538
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.8369455337524414,
+      "learning_rate": 5.394562139802162e-09,
+      "loss": 0.4762,
+      "step": 60539
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.2103495597839355,
+      "learning_rate": 5.3823248344109235e-09,
+      "loss": 0.5892,
+      "step": 60540
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.1756253242492676,
+      "learning_rate": 5.370101417180862e-09,
+      "loss": 0.598,
+      "step": 60541
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.5580883026123047,
+      "learning_rate": 5.357891888145839e-09,
+      "loss": 0.5878,
+      "step": 60542
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.9502673149108887,
+      "learning_rate": 5.3456962473391605e-09,
+      "loss": 0.6656,
+      "step": 60543
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.2603366374969482,
+      "learning_rate": 5.33351449479469e-09,
+      "loss": 0.6253,
+      "step": 60544
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.5042197704315186,
+      "learning_rate": 5.321346630547397e-09,
+      "loss": 0.4091,
+      "step": 60545
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.887702703475952,
+      "learning_rate": 5.309192654630035e-09,
+      "loss": 0.5208,
+      "step": 60546
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.0468497276306152,
+      "learning_rate": 5.297052567076466e-09,
+      "loss": 0.3968,
+      "step": 60547
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.866460084915161,
+      "learning_rate": 5.284926367921106e-09,
+      "loss": 0.6406,
+      "step": 60548
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.8906972408294678,
+      "learning_rate": 5.272814057197262e-09,
+      "loss": 0.597,
+      "step": 60549
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.0376198291778564,
+      "learning_rate": 5.260715634938241e-09,
+      "loss": 0.5492,
+      "step": 60550
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.1785097122192383,
+      "learning_rate": 5.248631101178459e-09,
+      "loss": 0.5864,
+      "step": 60551
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.1196131706237793,
+      "learning_rate": 5.2365604559512226e-09,
+      "loss": 0.7221,
+      "step": 60552
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.9107649326324463,
+      "learning_rate": 5.22450369928984e-09,
+      "loss": 0.3172,
+      "step": 60553
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.1769802570343018,
+      "learning_rate": 5.212460831228172e-09,
+      "loss": 0.4833,
+      "step": 60554
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.316631555557251,
+      "learning_rate": 5.200431851799525e-09,
+      "loss": 0.5028,
+      "step": 60555
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.3220555782318115,
+      "learning_rate": 5.188416761037207e-09,
+      "loss": 0.6801,
+      "step": 60556
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 4.171575546264648,
+      "learning_rate": 5.176415558975079e-09,
+      "loss": 0.6471,
+      "step": 60557
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 4.301151275634766,
+      "learning_rate": 5.164428245646447e-09,
+      "loss": 0.9211,
+      "step": 60558
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.5609450340270996,
+      "learning_rate": 5.152454821084063e-09,
+      "loss": 0.5844,
+      "step": 60559
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.772690534591675,
+      "learning_rate": 5.140495285321234e-09,
+      "loss": 0.5723,
+      "step": 60560
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.806000232696533,
+      "learning_rate": 5.1285496383923775e-09,
+      "loss": 0.8245,
+      "step": 60561
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 4.52339506149292,
+      "learning_rate": 5.116617880329133e-09,
+      "loss": 0.5438,
+      "step": 60562
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.5604538917541504,
+      "learning_rate": 5.1047000111659196e-09,
+      "loss": 0.4609,
+      "step": 60563
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.8254268169403076,
+      "learning_rate": 5.092796030935487e-09,
+      "loss": 0.598,
+      "step": 60564
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.857769727706909,
+      "learning_rate": 5.080905939670589e-09,
+      "loss": 0.6554,
+      "step": 60565
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 4.409609317779541,
+      "learning_rate": 5.069029737404529e-09,
+      "loss": 0.6466,
+      "step": 60566
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.5746047496795654,
+      "learning_rate": 5.057167424170617e-09,
+      "loss": 0.4417,
+      "step": 60567
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.603294849395752,
+      "learning_rate": 5.045319000001603e-09,
+      "loss": 0.5742,
+      "step": 60568
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.731654644012451,
+      "learning_rate": 5.033484464930239e-09,
+      "loss": 0.445,
+      "step": 60569
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.8816165924072266,
+      "learning_rate": 5.021663818989275e-09,
+      "loss": 0.6397,
+      "step": 60570
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.5169241428375244,
+      "learning_rate": 5.00985706221202e-09,
+      "loss": 0.5065,
+      "step": 60571
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.767368793487549,
+      "learning_rate": 4.99806419463178e-09,
+      "loss": 0.6834,
+      "step": 60572
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.5094006061553955,
+      "learning_rate": 4.986285216280196e-09,
+      "loss": 0.4656,
+      "step": 60573
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.3909149169921875,
+      "learning_rate": 4.974520127190574e-09,
+      "loss": 0.6733,
+      "step": 60574
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.183400869369507,
+      "learning_rate": 4.962768927395667e-09,
+      "loss": 0.5822,
+      "step": 60575
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.632838010787964,
+      "learning_rate": 4.951031616928226e-09,
+      "loss": 0.5669,
+      "step": 60576
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 4.135024547576904,
+      "learning_rate": 4.939308195821002e-09,
+      "loss": 0.6085,
+      "step": 60577
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.7838327884674072,
+      "learning_rate": 4.927598664106192e-09,
+      "loss": 0.5121,
+      "step": 60578
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.970189094543457,
+      "learning_rate": 4.915903021816548e-09,
+      "loss": 0.5392,
+      "step": 60579
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.7815423011779785,
+      "learning_rate": 4.904221268984266e-09,
+      "loss": 0.5692,
+      "step": 60580
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.8377180099487305,
+      "learning_rate": 4.8925534056426525e-09,
+      "loss": 0.5689,
+      "step": 60581
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.5621378421783447,
+      "learning_rate": 4.880899431823349e-09,
+      "loss": 0.4604,
+      "step": 60582
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.0431673526763916,
+      "learning_rate": 4.869259347559108e-09,
+      "loss": 0.5616,
+      "step": 60583
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.1266894340515137,
+      "learning_rate": 4.857633152882124e-09,
+      "loss": 0.6044,
+      "step": 60584
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.6165049076080322,
+      "learning_rate": 4.846020847824595e-09,
+      "loss": 0.5034,
+      "step": 60585
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.2580783367156982,
+      "learning_rate": 4.834422432419827e-09,
+      "loss": 0.6946,
+      "step": 60586
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.7608695030212402,
+      "learning_rate": 4.822837906698907e-09,
+      "loss": 0.5188,
+      "step": 60587
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.051821708679199,
+      "learning_rate": 4.811267270694031e-09,
+      "loss": 0.3971,
+      "step": 60588
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.609992027282715,
+      "learning_rate": 4.7997105244379505e-09,
+      "loss": 0.4301,
+      "step": 60589
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.400954008102417,
+      "learning_rate": 4.7881676679628615e-09,
+      "loss": 0.4385,
+      "step": 60590
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.7158377170562744,
+      "learning_rate": 4.776638701300962e-09,
+      "loss": 0.5851,
+      "step": 60591
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 5.061398029327393,
+      "learning_rate": 4.7651236244833365e-09,
+      "loss": 0.5324,
+      "step": 60592
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.5656166076660156,
+      "learning_rate": 4.753622437542737e-09,
+      "loss": 0.6425,
+      "step": 60593
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.321894645690918,
+      "learning_rate": 4.742135140511361e-09,
+      "loss": 0.6829,
+      "step": 60594
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.5108656883239746,
+      "learning_rate": 4.73066173342085e-09,
+      "loss": 0.6383,
+      "step": 60595
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.6680209636688232,
+      "learning_rate": 4.719202216302843e-09,
+      "loss": 0.5945,
+      "step": 60596
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 5.391095161437988,
+      "learning_rate": 4.7077565891895386e-09,
+      "loss": 0.6024,
+      "step": 60597
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.651555299758911,
+      "learning_rate": 4.696324852112577e-09,
+      "loss": 0.433,
+      "step": 60598
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.409834384918213,
+      "learning_rate": 4.684907005104156e-09,
+      "loss": 0.6407,
+      "step": 60599
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.094871759414673,
+      "learning_rate": 4.6735030481953605e-09,
+      "loss": 0.522,
+      "step": 60600
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.550906181335449,
+      "learning_rate": 4.662112981418387e-09,
+      "loss": 0.6475,
+      "step": 60601
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.5263314247131348,
+      "learning_rate": 4.650736804804878e-09,
+      "loss": 0.6828,
+      "step": 60602
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.1666147708892822,
+      "learning_rate": 4.639374518386474e-09,
+      "loss": 0.5628,
+      "step": 60603
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.424838066101074,
+      "learning_rate": 4.628026122194818e-09,
+      "loss": 0.5981,
+      "step": 60604
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.7681925296783447,
+      "learning_rate": 4.616691616260993e-09,
+      "loss": 0.6606,
+      "step": 60605
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.289059638977051,
+      "learning_rate": 4.605371000616643e-09,
+      "loss": 0.5378,
+      "step": 60606
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 4.044445991516113,
+      "learning_rate": 4.594064275293408e-09,
+      "loss": 0.6431,
+      "step": 60607
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.0571470260620117,
+      "learning_rate": 4.582771440323486e-09,
+      "loss": 0.43,
+      "step": 60608
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.30560564994812,
+      "learning_rate": 4.571492495736851e-09,
+      "loss": 0.4042,
+      "step": 60609
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.1101677417755127,
+      "learning_rate": 4.560227441565701e-09,
+      "loss": 0.6446,
+      "step": 60610
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 4.271983623504639,
+      "learning_rate": 4.548976277841677e-09,
+      "loss": 0.6562,
+      "step": 60611
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.8505444526672363,
+      "learning_rate": 4.537739004594754e-09,
+      "loss": 0.5679,
+      "step": 60612
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.2463648319244385,
+      "learning_rate": 4.526515621857685e-09,
+      "loss": 0.5388,
+      "step": 60613
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.9812984466552734,
+      "learning_rate": 4.515306129661001e-09,
+      "loss": 0.7866,
+      "step": 60614
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.5839414596557617,
+      "learning_rate": 4.504110528035788e-09,
+      "loss": 0.6167,
+      "step": 60615
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.303826332092285,
+      "learning_rate": 4.492928817013131e-09,
+      "loss": 0.5216,
+      "step": 60616
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.487330675125122,
+      "learning_rate": 4.481760996624118e-09,
+      "loss": 0.4303,
+      "step": 60617
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.718942880630493,
+      "learning_rate": 4.47060706690039e-09,
+      "loss": 0.718,
+      "step": 60618
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 1.8166980743408203,
+      "learning_rate": 4.459467027872477e-09,
+      "loss": 0.3703,
+      "step": 60619
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.7627766132354736,
+      "learning_rate": 4.448340879571467e-09,
+      "loss": 0.5941,
+      "step": 60620
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.731473684310913,
+      "learning_rate": 4.4372286220278896e-09,
+      "loss": 0.6102,
+      "step": 60621
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.929765224456787,
+      "learning_rate": 4.426130255273387e-09,
+      "loss": 0.5075,
+      "step": 60622
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 4.029892444610596,
+      "learning_rate": 4.415045779338489e-09,
+      "loss": 0.8866,
+      "step": 60623
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.359189748764038,
+      "learning_rate": 4.403975194253729e-09,
+      "loss": 0.4634,
+      "step": 60624
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.2619881629943848,
+      "learning_rate": 4.392918500050191e-09,
+      "loss": 0.4435,
+      "step": 60625
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.642448902130127,
+      "learning_rate": 4.381875696758408e-09,
+      "loss": 0.4871,
+      "step": 60626
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.6080880165100098,
+      "learning_rate": 4.370846784408911e-09,
+      "loss": 0.5649,
+      "step": 60627
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 4.053861618041992,
+      "learning_rate": 4.35983176303334e-09,
+      "loss": 0.7542,
+      "step": 60628
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 4.277231216430664,
+      "learning_rate": 4.348830632661671e-09,
+      "loss": 0.4879,
+      "step": 60629
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.5498385429382324,
+      "learning_rate": 4.337843393323882e-09,
+      "loss": 0.5624,
+      "step": 60630
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.0663111209869385,
+      "learning_rate": 4.3268700450516124e-09,
+      "loss": 0.5438,
+      "step": 60631
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.4351847171783447,
+      "learning_rate": 4.315910587874838e-09,
+      "loss": 0.6663,
+      "step": 60632
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.290890693664551,
+      "learning_rate": 4.3049650218240925e-09,
+      "loss": 0.3848,
+      "step": 60633
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.1850948333740234,
+      "learning_rate": 4.294033346929349e-09,
+      "loss": 0.5668,
+      "step": 60634
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.322524070739746,
+      "learning_rate": 4.283115563222251e-09,
+      "loss": 0.708,
+      "step": 60635
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.929842233657837,
+      "learning_rate": 4.272211670731663e-09,
+      "loss": 0.3126,
+      "step": 60636
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.38662052154541,
+      "learning_rate": 4.2613216694892264e-09,
+      "loss": 0.7491,
+      "step": 60637
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.978849172592163,
+      "learning_rate": 4.250445559523808e-09,
+      "loss": 0.6805,
+      "step": 60638
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.8826568126678467,
+      "learning_rate": 4.239583340867048e-09,
+      "loss": 0.5389,
+      "step": 60639
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.7139511108398438,
+      "learning_rate": 4.228735013548368e-09,
+      "loss": 0.5594,
+      "step": 60640
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.6949877738952637,
+      "learning_rate": 4.2179005775982995e-09,
+      "loss": 0.6217,
+      "step": 60641
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.801732301712036,
+      "learning_rate": 4.2070800330468175e-09,
+      "loss": 0.7017,
+      "step": 60642
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.3923513889312744,
+      "learning_rate": 4.196273379923898e-09,
+      "loss": 0.7319,
+      "step": 60643
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.3963992595672607,
+      "learning_rate": 4.185480618259518e-09,
+      "loss": 0.5262,
+      "step": 60644
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.619922399520874,
+      "learning_rate": 4.1747017480842086e-09,
+      "loss": 0.6272,
+      "step": 60645
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.6721529960632324,
+      "learning_rate": 4.163936769426835e-09,
+      "loss": 0.6397,
+      "step": 60646
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.823162078857422,
+      "learning_rate": 4.153185682319039e-09,
+      "loss": 0.7284,
+      "step": 60647
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 4.075070381164551,
+      "learning_rate": 4.1424484867891305e-09,
+      "loss": 0.7364,
+      "step": 60648
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.6197633743286133,
+      "learning_rate": 4.1317251828676406e-09,
+      "loss": 0.6269,
+      "step": 60649
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 6.248808860778809,
+      "learning_rate": 4.121015770584547e-09,
+      "loss": 0.5869,
+      "step": 60650
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 4.075695514678955,
+      "learning_rate": 4.110320249969268e-09,
+      "loss": 0.6554,
+      "step": 60651
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.3084867000579834,
+      "learning_rate": 4.099638621051227e-09,
+      "loss": 0.3448,
+      "step": 60652
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.0662925243377686,
+      "learning_rate": 4.0889708838609544e-09,
+      "loss": 0.7457,
+      "step": 60653
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.692826271057129,
+      "learning_rate": 4.07831703842787e-09,
+      "loss": 0.4554,
+      "step": 60654
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.1066946983337402,
+      "learning_rate": 4.067677084781396e-09,
+      "loss": 0.5545,
+      "step": 60655
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.7809815406799316,
+      "learning_rate": 4.057051022951508e-09,
+      "loss": 0.5802,
+      "step": 60656
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.9791362285614014,
+      "learning_rate": 4.046438852967072e-09,
+      "loss": 0.4461,
+      "step": 60657
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.0962483882904053,
+      "learning_rate": 4.035840574858063e-09,
+      "loss": 0.4257,
+      "step": 60658
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.229196071624756,
+      "learning_rate": 4.025256188653903e-09,
+      "loss": 0.6441,
+      "step": 60659
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 4.448100566864014,
+      "learning_rate": 4.0146856943834575e-09,
+      "loss": 0.5982,
+      "step": 60660
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.494062662124634,
+      "learning_rate": 4.004129092077258e-09,
+      "loss": 0.5577,
+      "step": 60661
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.7248551845550537,
+      "learning_rate": 3.9935863817636145e-09,
+      "loss": 0.7314,
+      "step": 60662
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.7141056060791016,
+      "learning_rate": 3.983057563472503e-09,
+      "loss": 0.5768,
+      "step": 60663
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.6415657997131348,
+      "learning_rate": 3.972542637233345e-09,
+      "loss": 0.6046,
+      "step": 60664
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.594390392303467,
+      "learning_rate": 3.962041603074451e-09,
+      "loss": 0.9361,
+      "step": 60665
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.8203160762786865,
+      "learning_rate": 3.951554461025797e-09,
+      "loss": 0.5334,
+      "step": 60666
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.7196640968322754,
+      "learning_rate": 3.941081211116249e-09,
+      "loss": 0.4686,
+      "step": 60667
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 4.412988662719727,
+      "learning_rate": 3.930621853374672e-09,
+      "loss": 0.6268,
+      "step": 60668
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.6632425785064697,
+      "learning_rate": 3.920176387830488e-09,
+      "loss": 0.5067,
+      "step": 60669
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.941361427307129,
+      "learning_rate": 3.909744814513117e-09,
+      "loss": 0.4681,
+      "step": 60670
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.661672353744507,
+      "learning_rate": 3.8993271334508696e-09,
+      "loss": 0.6955,
+      "step": 60671
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.2968127727508545,
+      "learning_rate": 3.8889233446737225e-09,
+      "loss": 0.7273,
+      "step": 60672
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.6451759338378906,
+      "learning_rate": 3.8785334482088766e-09,
+      "loss": 0.5567,
+      "step": 60673
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.491461753845215,
+      "learning_rate": 3.868157444086862e-09,
+      "loss": 0.4144,
+      "step": 60674
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 5.639628887176514,
+      "learning_rate": 3.85779533233599e-09,
+      "loss": 0.5127,
+      "step": 60675
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.566087245941162,
+      "learning_rate": 3.84744711298457e-09,
+      "loss": 0.6262,
+      "step": 60676
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.2391324043273926,
+      "learning_rate": 3.837112786062025e-09,
+      "loss": 0.5939,
+      "step": 60677
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.9885661602020264,
+      "learning_rate": 3.826792351596665e-09,
+      "loss": 0.5394,
+      "step": 60678
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.5154857635498047,
+      "learning_rate": 3.816485809617354e-09,
+      "loss": 0.6747,
+      "step": 60679
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.4613265991210938,
+      "learning_rate": 3.80619316015296e-09,
+      "loss": 0.65,
+      "step": 60680
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.2369368076324463,
+      "learning_rate": 3.795914403231793e-09,
+      "loss": 0.685,
+      "step": 60681
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.0530335903167725,
+      "learning_rate": 3.785649538882719e-09,
+      "loss": 0.5932,
+      "step": 60682
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 4.602506160736084,
+      "learning_rate": 3.7753985671340475e-09,
+      "loss": 0.6777,
+      "step": 60683
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.573798418045044,
+      "learning_rate": 3.765161488014646e-09,
+      "loss": 0.6023,
+      "step": 60684
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.6034436225891113,
+      "learning_rate": 3.754938301552269e-09,
+      "loss": 0.7272,
+      "step": 60685
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.689481735229492,
+      "learning_rate": 3.744729007776338e-09,
+      "loss": 0.6722,
+      "step": 60686
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.7601191997528076,
+      "learning_rate": 3.734533606714052e-09,
+      "loss": 0.5799,
+      "step": 60687
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.8743231296539307,
+      "learning_rate": 3.7243520983948346e-09,
+      "loss": 0.6002,
+      "step": 60688
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 4.6617302894592285,
+      "learning_rate": 3.714184482846439e-09,
+      "loss": 0.6355,
+      "step": 60689
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.5124258995056152,
+      "learning_rate": 3.704030760097177e-09,
+      "loss": 0.4176,
+      "step": 60690
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 4.1322126388549805,
+      "learning_rate": 3.6938909301753587e-09,
+      "loss": 0.6023,
+      "step": 60691
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.5953454971313477,
+      "learning_rate": 3.6837649931092956e-09,
+      "loss": 0.6043,
+      "step": 60692
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 4.377191066741943,
+      "learning_rate": 3.6736529489267424e-09,
+      "loss": 0.5317,
+      "step": 60693
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.593989610671997,
+      "learning_rate": 3.6635547976560106e-09,
+      "loss": 0.6174,
+      "step": 60694
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.725315809249878,
+      "learning_rate": 3.65347053932541e-09,
+      "loss": 0.7416,
+      "step": 60695
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.4642927646636963,
+      "learning_rate": 3.643400173963252e-09,
+      "loss": 0.6055,
+      "step": 60696
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.686739444732666,
+      "learning_rate": 3.6333437015967366e-09,
+      "loss": 0.6374,
+      "step": 60697
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.216864585876465,
+      "learning_rate": 3.623301122254175e-09,
+      "loss": 0.4565,
+      "step": 60698
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.1964476108551025,
+      "learning_rate": 3.613272435963322e-09,
+      "loss": 0.5917,
+      "step": 60699
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.295372724533081,
+      "learning_rate": 3.603257642752489e-09,
+      "loss": 0.3559,
+      "step": 60700
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.774030923843384,
+      "learning_rate": 3.5932567426494314e-09,
+      "loss": 0.4686,
+      "step": 60701
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.2291250228881836,
+      "learning_rate": 3.5832697356819045e-09,
+      "loss": 0.4922,
+      "step": 60702
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.0151264667510986,
+      "learning_rate": 3.5732966218771093e-09,
+      "loss": 0.6133,
+      "step": 60703
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.8383967876434326,
+      "learning_rate": 3.563337401263911e-09,
+      "loss": 0.6091,
+      "step": 60704
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.8034791946411133,
+      "learning_rate": 3.5533920738689555e-09,
+      "loss": 0.6163,
+      "step": 60705
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.9074764251708984,
+      "learning_rate": 3.5434606397205528e-09,
+      "loss": 0.5798,
+      "step": 60706
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 4.240560054779053,
+      "learning_rate": 3.533543098845904e-09,
+      "loss": 0.5907,
+      "step": 60707
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.135345935821533,
+      "learning_rate": 3.5236394512727645e-09,
+      "loss": 0.6264,
+      "step": 60708
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.6602277755737305,
+      "learning_rate": 3.5137496970288898e-09,
+      "loss": 0.5012,
+      "step": 60709
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.695119619369507,
+      "learning_rate": 3.5038738361414803e-09,
+      "loss": 0.4965,
+      "step": 60710
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.9981908798217773,
+      "learning_rate": 3.494011868637737e-09,
+      "loss": 0.5216,
+      "step": 60711
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.380915880203247,
+      "learning_rate": 3.4841637945454144e-09,
+      "loss": 0.594,
+      "step": 60712
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.9901485443115234,
+      "learning_rate": 3.474329613892269e-09,
+      "loss": 0.6197,
+      "step": 60713
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.895134449005127,
+      "learning_rate": 3.4645093267049457e-09,
+      "loss": 0.6597,
+      "step": 60714
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.604875326156616,
+      "learning_rate": 3.4547029330112e-09,
+      "loss": 0.4927,
+      "step": 60715
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.5620689392089844,
+      "learning_rate": 3.444910432838233e-09,
+      "loss": 0.4073,
+      "step": 60716
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 4.250800132751465,
+      "learning_rate": 3.435131826213245e-09,
+      "loss": 0.7642,
+      "step": 60717
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.390713691711426,
+      "learning_rate": 3.425367113163436e-09,
+      "loss": 0.5874,
+      "step": 60718
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.7831132411956787,
+      "learning_rate": 3.4156162937154514e-09,
+      "loss": 0.51,
+      "step": 60719
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.8184351921081543,
+      "learning_rate": 3.405879367897602e-09,
+      "loss": 0.4145,
+      "step": 60720
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.1343352794647217,
+      "learning_rate": 3.3961563357354234e-09,
+      "loss": 0.5806,
+      "step": 60721
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.5381462574005127,
+      "learning_rate": 3.3864471972577804e-09,
+      "loss": 0.5577,
+      "step": 60722
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.126824378967285,
+      "learning_rate": 3.376751952489654e-09,
+      "loss": 0.6161,
+      "step": 60723
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.769749879837036,
+      "learning_rate": 3.3670706014593545e-09,
+      "loss": 0.6886,
+      "step": 60724
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 1.9740022420883179,
+      "learning_rate": 3.3574031441935273e-09,
+      "loss": 0.2727,
+      "step": 60725
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.6823015213012695,
+      "learning_rate": 3.347749580719373e-09,
+      "loss": 0.2627,
+      "step": 60726
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.4864914417266846,
+      "learning_rate": 3.3381099110629812e-09,
+      "loss": 0.7289,
+      "step": 60727
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.0164546966552734,
+      "learning_rate": 3.328484135251553e-09,
+      "loss": 0.5577,
+      "step": 60728
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.205169439315796,
+      "learning_rate": 3.3188722533117334e-09,
+      "loss": 0.5653,
+      "step": 60729
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 4.320944786071777,
+      "learning_rate": 3.3092742652701683e-09,
+      "loss": 0.7023,
+      "step": 60730
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 1.9607619047164917,
+      "learning_rate": 3.2996901711540575e-09,
+      "loss": 0.3627,
+      "step": 60731
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.8692901134490967,
+      "learning_rate": 3.2901199709894915e-09,
+      "loss": 0.6681,
+      "step": 60732
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.466097354888916,
+      "learning_rate": 3.280563664803116e-09,
+      "loss": 0.4206,
+      "step": 60733
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.447338581085205,
+      "learning_rate": 3.2710212526215758e-09,
+      "loss": 0.6914,
+      "step": 60734
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.9634900093078613,
+      "learning_rate": 3.2614927344720714e-09,
+      "loss": 0.6203,
+      "step": 60735
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.8667497634887695,
+      "learning_rate": 3.2519781103801385e-09,
+      "loss": 0.5836,
+      "step": 60736
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.307281970977783,
+      "learning_rate": 3.242477380372977e-09,
+      "loss": 0.456,
+      "step": 60737
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 5.673224449157715,
+      "learning_rate": 3.2329905444766774e-09,
+      "loss": 0.5871,
+      "step": 60738
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.096034526824951,
+      "learning_rate": 3.2235176027173298e-09,
+      "loss": 0.6627,
+      "step": 60739
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.7603142261505127,
+      "learning_rate": 3.2140585551215797e-09,
+      "loss": 0.455,
+      "step": 60740
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 4.280810832977295,
+      "learning_rate": 3.204613401715517e-09,
+      "loss": 0.7837,
+      "step": 60741
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.536518096923828,
+      "learning_rate": 3.195182142525788e-09,
+      "loss": 0.8098,
+      "step": 60742
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.251077651977539,
+      "learning_rate": 3.185764777578482e-09,
+      "loss": 0.4979,
+      "step": 60743
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.2869527339935303,
+      "learning_rate": 3.176361306899689e-09,
+      "loss": 0.6148,
+      "step": 60744
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.2650327682495117,
+      "learning_rate": 3.1669717305155e-09,
+      "loss": 0.5742,
+      "step": 60745
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 4.064792156219482,
+      "learning_rate": 3.157596048452005e-09,
+      "loss": 0.7137,
+      "step": 60746
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.708021640777588,
+      "learning_rate": 3.1482342607352944e-09,
+      "loss": 0.4436,
+      "step": 60747
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.6808950901031494,
+      "learning_rate": 3.1388863673920135e-09,
+      "loss": 0.4787,
+      "step": 60748
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.8911311626434326,
+      "learning_rate": 3.129552368447142e-09,
+      "loss": 0.7174,
+      "step": 60749
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.6228067874908447,
+      "learning_rate": 3.120232263927325e-09,
+      "loss": 0.4974,
+      "step": 60750
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.5857911109924316,
+      "learning_rate": 3.1109260538580987e-09,
+      "loss": 0.6851,
+      "step": 60751
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.9668161869049072,
+      "learning_rate": 3.1016337382655527e-09,
+      "loss": 0.8421,
+      "step": 60752
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 5.695356845855713,
+      "learning_rate": 3.0923553171757768e-09,
+      "loss": 0.51,
+      "step": 60753
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.071442127227783,
+      "learning_rate": 3.083090790613752e-09,
+      "loss": 0.5804,
+      "step": 60754
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.2591640949249268,
+      "learning_rate": 3.073840158606123e-09,
+      "loss": 0.512,
+      "step": 60755
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 4.082330226898193,
+      "learning_rate": 3.064603421178425e-09,
+      "loss": 0.5582,
+      "step": 60756
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.563051223754883,
+      "learning_rate": 3.0553805783556377e-09,
+      "loss": 0.4763,
+      "step": 60757
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.521909475326538,
+      "learning_rate": 3.046171630164407e-09,
+      "loss": 0.4696,
+      "step": 60758
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.1283211708068848,
+      "learning_rate": 3.036976576629713e-09,
+      "loss": 0.623,
+      "step": 60759
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.9623231887817383,
+      "learning_rate": 3.0277954177770908e-09,
+      "loss": 0.4998,
+      "step": 60760
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.7538676261901855,
+      "learning_rate": 3.01862815363263e-09,
+      "loss": 0.6086,
+      "step": 60761
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.72670841217041,
+      "learning_rate": 3.0094747842213113e-09,
+      "loss": 0.4293,
+      "step": 60762
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.953186273574829,
+      "learning_rate": 3.0003353095686696e-09,
+      "loss": 0.7426,
+      "step": 60763
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.5664114952087402,
+      "learning_rate": 2.9912097297002396e-09,
+      "loss": 0.6762,
+      "step": 60764
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.8453564643859863,
+      "learning_rate": 2.982098044641002e-09,
+      "loss": 0.6109,
+      "step": 60765
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.444546937942505,
+      "learning_rate": 2.9730002544170465e-09,
+      "loss": 0.5662,
+      "step": 60766
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.665914535522461,
+      "learning_rate": 2.9639163590527985e-09,
+      "loss": 0.5363,
+      "step": 60767
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.608990430831909,
+      "learning_rate": 2.954846358574348e-09,
+      "loss": 0.5879,
+      "step": 60768
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 4.550878047943115,
+      "learning_rate": 2.945790253006675e-09,
+      "loss": 0.4719,
+      "step": 60769
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.801079034805298,
+      "learning_rate": 2.9367480423747596e-09,
+      "loss": 0.7767,
+      "step": 60770
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.4446654319763184,
+      "learning_rate": 2.9277197267035816e-09,
+      "loss": 0.7984,
+      "step": 60771
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.895218849182129,
+      "learning_rate": 2.9187053060181214e-09,
+      "loss": 0.4103,
+      "step": 60772
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.0732810497283936,
+      "learning_rate": 2.909704780344469e-09,
+      "loss": 0.5584,
+      "step": 60773
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.788792848587036,
+      "learning_rate": 2.900718149706494e-09,
+      "loss": 0.8492,
+      "step": 60774
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.697913408279419,
+      "learning_rate": 2.891745414130287e-09,
+      "loss": 0.6338,
+      "step": 60775
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 7.576393127441406,
+      "learning_rate": 2.8827865736397175e-09,
+      "loss": 0.8728,
+      "step": 60776
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.1551880836486816,
+      "learning_rate": 2.873841628260321e-09,
+      "loss": 0.5163,
+      "step": 60777
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.3402016162872314,
+      "learning_rate": 2.864910578016522e-09,
+      "loss": 0.7367,
+      "step": 60778
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.182013750076294,
+      "learning_rate": 2.855993422933856e-09,
+      "loss": 0.5636,
+      "step": 60779
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.9795899391174316,
+      "learning_rate": 2.8470901630361923e-09,
+      "loss": 0.6068,
+      "step": 60780
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.6809420585632324,
+      "learning_rate": 2.838200798349067e-09,
+      "loss": 0.644,
+      "step": 60781
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.3421835899353027,
+      "learning_rate": 2.8293253288969036e-09,
+      "loss": 0.7079,
+      "step": 60782
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.3364906311035156,
+      "learning_rate": 2.820463754704128e-09,
+      "loss": 0.7108,
+      "step": 60783
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.7609825134277344,
+      "learning_rate": 2.8116160757962753e-09,
+      "loss": 0.626,
+      "step": 60784
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 4.410417079925537,
+      "learning_rate": 2.8027822921966598e-09,
+      "loss": 0.614,
+      "step": 60785
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.1097304821014404,
+      "learning_rate": 2.7939624039308165e-09,
+      "loss": 0.5831,
+      "step": 60786
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.7035202980041504,
+      "learning_rate": 2.785156411023171e-09,
+      "loss": 0.4752,
+      "step": 60787
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.5286638736724854,
+      "learning_rate": 2.7763643134975927e-09,
+      "loss": 0.7772,
+      "step": 60788
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.378871202468872,
+      "learning_rate": 2.7675861113796167e-09,
+      "loss": 0.622,
+      "step": 60789
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.324192762374878,
+      "learning_rate": 2.7588218046925575e-09,
+      "loss": 0.5126,
+      "step": 60790
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.3066647052764893,
+      "learning_rate": 2.75007139346084e-09,
+      "loss": 0.3974,
+      "step": 60791
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 4.143410682678223,
+      "learning_rate": 2.741334877709445e-09,
+      "loss": 0.5676,
+      "step": 60792
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.6142733097076416,
+      "learning_rate": 2.7326122574627966e-09,
+      "loss": 0.5238,
+      "step": 60793
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 9.481524467468262,
+      "learning_rate": 2.72390353274421e-09,
+      "loss": 0.4744,
+      "step": 60794
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.820010185241699,
+      "learning_rate": 2.7152087035786643e-09,
+      "loss": 0.5578,
+      "step": 60795
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.8402011394500732,
+      "learning_rate": 2.7065277699900304e-09,
+      "loss": 0.5368,
+      "step": 60796
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.9259209632873535,
+      "learning_rate": 2.6978607320027327e-09,
+      "loss": 0.6784,
+      "step": 60797
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 4.109570503234863,
+      "learning_rate": 2.689207589640086e-09,
+      "loss": 0.6747,
+      "step": 60798
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 4.751652717590332,
+      "learning_rate": 2.68056834292707e-09,
+      "loss": 0.7371,
+      "step": 60799
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 8.338821411132812,
+      "learning_rate": 2.671942991887555e-09,
+      "loss": 0.3126,
+      "step": 60800
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.3176093101501465,
+      "learning_rate": 2.6633315365448554e-09,
+      "loss": 0.4688,
+      "step": 60801
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.5892651081085205,
+      "learning_rate": 2.654733976923951e-09,
+      "loss": 0.5344,
+      "step": 60802
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.5055198669433594,
+      "learning_rate": 2.6461503130476017e-09,
+      "loss": 0.6354,
+      "step": 60803
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 4.724593639373779,
+      "learning_rate": 2.6375805449407877e-09,
+      "loss": 0.5455,
+      "step": 60804
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.9350156784057617,
+      "learning_rate": 2.629024672626268e-09,
+      "loss": 0.4286,
+      "step": 60805
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.9232754707336426,
+      "learning_rate": 2.620482696129023e-09,
+      "loss": 0.4194,
+      "step": 60806
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 4.0640034675598145,
+      "learning_rate": 2.611954615471812e-09,
+      "loss": 0.7764,
+      "step": 60807
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.9059700965881348,
+      "learning_rate": 2.6034404306785056e-09,
+      "loss": 0.7308,
+      "step": 60808
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.231811285018921,
+      "learning_rate": 2.5949401417729727e-09,
+      "loss": 0.662,
+      "step": 60809
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.5963892936706543,
+      "learning_rate": 2.586453748779083e-09,
+      "loss": 0.3416,
+      "step": 60810
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.959869384765625,
+      "learning_rate": 2.577981251720152e-09,
+      "loss": 0.5689,
+      "step": 60811
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.740736484527588,
+      "learning_rate": 2.569522650620049e-09,
+      "loss": 0.6817,
+      "step": 60812
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.49178147315979,
+      "learning_rate": 2.5610779455015335e-09,
+      "loss": 0.4957,
+      "step": 60813
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.2731544971466064,
+      "learning_rate": 2.5526471363884752e-09,
+      "loss": 0.5328,
+      "step": 60814
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.6835873126983643,
+      "learning_rate": 2.5442302233047443e-09,
+      "loss": 0.7813,
+      "step": 60815
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 4.872591495513916,
+      "learning_rate": 2.5358272062736557e-09,
+      "loss": 0.6847,
+      "step": 60816
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.3525302410125732,
+      "learning_rate": 2.527438085317968e-09,
+      "loss": 0.5492,
+      "step": 60817
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.122133255004883,
+      "learning_rate": 2.519062860461552e-09,
+      "loss": 0.5025,
+      "step": 60818
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.6472818851470947,
+      "learning_rate": 2.5107015317271664e-09,
+      "loss": 0.6177,
+      "step": 60819
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 4.017016887664795,
+      "learning_rate": 2.502354099138682e-09,
+      "loss": 0.5442,
+      "step": 60820
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.752882480621338,
+      "learning_rate": 2.4940205627194126e-09,
+      "loss": 0.5435,
+      "step": 60821
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.153075695037842,
+      "learning_rate": 2.485700922491563e-09,
+      "loss": 0.4905,
+      "step": 60822
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.9602291584014893,
+      "learning_rate": 2.477395178479003e-09,
+      "loss": 0.3765,
+      "step": 60823
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.420710563659668,
+      "learning_rate": 2.4691033307050472e-09,
+      "loss": 0.4838,
+      "step": 60824
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.160529136657715,
+      "learning_rate": 2.4608253791919e-09,
+      "loss": 0.6402,
+      "step": 60825
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.3396129608154297,
+      "learning_rate": 2.4525613239634315e-09,
+      "loss": 0.6069,
+      "step": 60826
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.0075840950012207,
+      "learning_rate": 2.444311165042401e-09,
+      "loss": 0.7008,
+      "step": 60827
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 4.260237693786621,
+      "learning_rate": 2.4360749024510132e-09,
+      "loss": 0.5214,
+      "step": 60828
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.8036365509033203,
+      "learning_rate": 2.4278525362131376e-09,
+      "loss": 0.5242,
+      "step": 60829
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.796447515487671,
+      "learning_rate": 2.419644066351534e-09,
+      "loss": 0.7816,
+      "step": 60830
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.77091121673584,
+      "learning_rate": 2.4114494928884067e-09,
+      "loss": 0.5145,
+      "step": 60831
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.45794677734375,
+      "learning_rate": 2.403268815847071e-09,
+      "loss": 0.5577,
+      "step": 60832
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.2342453002929688,
+      "learning_rate": 2.39510203524973e-09,
+      "loss": 0.6029,
+      "step": 60833
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.6759724617004395,
+      "learning_rate": 2.3869491511196994e-09,
+      "loss": 0.547,
+      "step": 60834
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.11978816986084,
+      "learning_rate": 2.378810163479739e-09,
+      "loss": 0.6329,
+      "step": 60835
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.7287206649780273,
+      "learning_rate": 2.3706850723520525e-09,
+      "loss": 0.5497,
+      "step": 60836
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.1330134868621826,
+      "learning_rate": 2.3625738777588447e-09,
+      "loss": 0.7531,
+      "step": 60837
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.0444164276123047,
+      "learning_rate": 2.3544765797234305e-09,
+      "loss": 0.3812,
+      "step": 60838
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.329153060913086,
+      "learning_rate": 2.346393178268014e-09,
+      "loss": 0.6434,
+      "step": 60839
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.687659502029419,
+      "learning_rate": 2.3383236734153546e-09,
+      "loss": 0.4887,
+      "step": 60840
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.0724942684173584,
+      "learning_rate": 2.3302680651876575e-09,
+      "loss": 0.2348,
+      "step": 60841
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.1417651176452637,
+      "learning_rate": 2.3222263536071264e-09,
+      "loss": 0.5968,
+      "step": 60842
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.688206911087036,
+      "learning_rate": 2.3141985386965216e-09,
+      "loss": 0.3764,
+      "step": 60843
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 4.608914852142334,
+      "learning_rate": 2.306184620477492e-09,
+      "loss": 0.6798,
+      "step": 60844
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.1792867183685303,
+      "learning_rate": 2.298184598973352e-09,
+      "loss": 0.421,
+      "step": 60845
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.56514048576355,
+      "learning_rate": 2.2901984742051962e-09,
+      "loss": 0.3868,
+      "step": 60846
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 4.534711837768555,
+      "learning_rate": 2.2822262461963397e-09,
+      "loss": 0.7019,
+      "step": 60847
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.255808115005493,
+      "learning_rate": 2.2742679149678758e-09,
+      "loss": 0.4954,
+      "step": 60848
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.9905307292938232,
+      "learning_rate": 2.26632348054312e-09,
+      "loss": 0.4756,
+      "step": 60849
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.727541923522949,
+      "learning_rate": 2.258392942943166e-09,
+      "loss": 0.4584,
+      "step": 60850
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.710637092590332,
+      "learning_rate": 2.2504763021902187e-09,
+      "loss": 0.7308,
+      "step": 60851
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.604703903198242,
+      "learning_rate": 2.2425735583070373e-09,
+      "loss": 0.7483,
+      "step": 60852
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 1.632059931755066,
+      "learning_rate": 2.2346847113147164e-09,
+      "loss": 0.349,
+      "step": 60853
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.7707581520080566,
+      "learning_rate": 2.22680976123546e-09,
+      "loss": 0.4542,
+      "step": 60854
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.0238335132598877,
+      "learning_rate": 2.218948708091473e-09,
+      "loss": 0.5557,
+      "step": 60855
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.633282423019409,
+      "learning_rate": 2.2111015519044043e-09,
+      "loss": 0.4905,
+      "step": 60856
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.2629144191741943,
+      "learning_rate": 2.2032682926959036e-09,
+      "loss": 0.4261,
+      "step": 60857
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.585232734680176,
+      "learning_rate": 2.1954489304876204e-09,
+      "loss": 0.4818,
+      "step": 60858
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.416245937347412,
+      "learning_rate": 2.1876434653017585e-09,
+      "loss": 0.6818,
+      "step": 60859
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.649935245513916,
+      "learning_rate": 2.179851897159968e-09,
+      "loss": 0.633,
+      "step": 60860
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.708652973175049,
+      "learning_rate": 2.172074226083898e-09,
+      "loss": 0.605,
+      "step": 60861
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.728278398513794,
+      "learning_rate": 2.1643104520946423e-09,
+      "loss": 0.497,
+      "step": 60862
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.58774995803833,
+      "learning_rate": 2.156560575214406e-09,
+      "loss": 0.5783,
+      "step": 60863
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 7.632461071014404,
+      "learning_rate": 2.1488245954648378e-09,
+      "loss": 0.5204,
+      "step": 60864
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.909183979034424,
+      "learning_rate": 2.1411025128670326e-09,
+      "loss": 0.4692,
+      "step": 60865
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.6229090690612793,
+      "learning_rate": 2.1333943274426395e-09,
+      "loss": 0.5648,
+      "step": 60866
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.19547963142395,
+      "learning_rate": 2.1257000392127523e-09,
+      "loss": 0.5806,
+      "step": 60867
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 4.542627811431885,
+      "learning_rate": 2.118019648199021e-09,
+      "loss": 0.5921,
+      "step": 60868
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.626911163330078,
+      "learning_rate": 2.110353154423095e-09,
+      "loss": 0.8882,
+      "step": 60869
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.001941442489624,
+      "learning_rate": 2.1027005579055127e-09,
+      "loss": 0.3985,
+      "step": 60870
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.675548791885376,
+      "learning_rate": 2.095061858668479e-09,
+      "loss": 0.6376,
+      "step": 60871
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.6509177684783936,
+      "learning_rate": 2.0874370567330883e-09,
+      "loss": 0.2846,
+      "step": 60872
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.033322811126709,
+      "learning_rate": 2.0798261521198793e-09,
+      "loss": 0.4846,
+      "step": 60873
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.128502607345581,
+      "learning_rate": 2.0722291448505018e-09,
+      "loss": 0.2931,
+      "step": 60874
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.8972299098968506,
+      "learning_rate": 2.0646460349460493e-09,
+      "loss": 0.4799,
+      "step": 60875
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.1286697387695312,
+      "learning_rate": 2.057076822427617e-09,
+      "loss": 0.3733,
+      "step": 60876
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.7349374294281006,
+      "learning_rate": 2.0495215073157436e-09,
+      "loss": 0.4637,
+      "step": 60877
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.152226448059082,
+      "learning_rate": 2.0419800896320785e-09,
+      "loss": 0.5483,
+      "step": 60878
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 4.031077861785889,
+      "learning_rate": 2.0344525693977155e-09,
+      "loss": 0.5961,
+      "step": 60879
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.2032902240753174,
+      "learning_rate": 2.0269389466331947e-09,
+      "loss": 0.5765,
+      "step": 60880
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.04394793510437,
+      "learning_rate": 2.019439221359054e-09,
+      "loss": 0.6344,
+      "step": 60881
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.272249698638916,
+      "learning_rate": 2.011953393596944e-09,
+      "loss": 0.6112,
+      "step": 60882
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.202193260192871,
+      "learning_rate": 2.0044814633674026e-09,
+      "loss": 0.3714,
+      "step": 60883
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 6.544267654418945,
+      "learning_rate": 1.9970234306904145e-09,
+      "loss": 0.5653,
+      "step": 60884
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.402637481689453,
+      "learning_rate": 1.9895792955881845e-09,
+      "loss": 0.816,
+      "step": 60885
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.7885215282440186,
+      "learning_rate": 1.9821490580801404e-09,
+      "loss": 0.5377,
+      "step": 60886
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.6586644649505615,
+      "learning_rate": 1.9747327181879326e-09,
+      "loss": 0.7153,
+      "step": 60887
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 4.151577949523926,
+      "learning_rate": 1.96733027593099e-09,
+      "loss": 0.514,
+      "step": 60888
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.51664137840271,
+      "learning_rate": 1.959941731330961e-09,
+      "loss": 0.5306,
+      "step": 60889
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.019838809967041,
+      "learning_rate": 1.9525670844083857e-09,
+      "loss": 0.4217,
+      "step": 60890
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.5366907119750977,
+      "learning_rate": 1.9452063351832474e-09,
+      "loss": 0.9008,
+      "step": 60891
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.459218740463257,
+      "learning_rate": 1.9378594836755307e-09,
+      "loss": 0.5882,
+      "step": 60892
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.5110490322113037,
+      "learning_rate": 1.9305265299068843e-09,
+      "loss": 0.6376,
+      "step": 60893
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.983025074005127,
+      "learning_rate": 1.9232074738972926e-09,
+      "loss": 0.5972,
+      "step": 60894
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.4515929222106934,
+      "learning_rate": 1.91590231566674e-09,
+      "loss": 0.5824,
+      "step": 60895
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.7837445735931396,
+      "learning_rate": 1.908611055235765e-09,
+      "loss": 0.4633,
+      "step": 60896
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.7126336097717285,
+      "learning_rate": 1.901333692624352e-09,
+      "loss": 0.6348,
+      "step": 60897
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.281634569168091,
+      "learning_rate": 1.8940702278530397e-09,
+      "loss": 0.4408,
+      "step": 60898
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.1564478874206543,
+      "learning_rate": 1.886820660942368e-09,
+      "loss": 0.6146,
+      "step": 60899
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.450639247894287,
+      "learning_rate": 1.879584991911765e-09,
+      "loss": 0.7876,
+      "step": 60900
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.920706272125244,
+      "learning_rate": 1.87236322078177e-09,
+      "loss": 0.3898,
+      "step": 60901
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.7639360427856445,
+      "learning_rate": 1.865155347572367e-09,
+      "loss": 0.667,
+      "step": 60902
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.5273327827453613,
+      "learning_rate": 1.8579613723035406e-09,
+      "loss": 0.536,
+      "step": 60903
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.108888626098633,
+      "learning_rate": 1.8507812949958293e-09,
+      "loss": 0.6366,
+      "step": 60904
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.0042827129364014,
+      "learning_rate": 1.8436151156681071e-09,
+      "loss": 0.6971,
+      "step": 60905
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.835907459259033,
+      "learning_rate": 1.836462834341468e-09,
+      "loss": 0.6646,
+      "step": 60906
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.3740718364715576,
+      "learning_rate": 1.8293244510347862e-09,
+      "loss": 0.7819,
+      "step": 60907
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 4.360135555267334,
+      "learning_rate": 1.8221999657686006e-09,
+      "loss": 0.6433,
+      "step": 60908
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 7.109294414520264,
+      "learning_rate": 1.8150893785628954e-09,
+      "loss": 0.5751,
+      "step": 60909
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.746999502182007,
+      "learning_rate": 1.807992689436544e-09,
+      "loss": 0.5663,
+      "step": 60910
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.994192600250244,
+      "learning_rate": 1.800909898410086e-09,
+      "loss": 0.6021,
+      "step": 60911
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.572484016418457,
+      "learning_rate": 1.79384100550295e-09,
+      "loss": 0.4102,
+      "step": 60912
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 4.667359828948975,
+      "learning_rate": 1.7867860107345647e-09,
+      "loss": 0.7021,
+      "step": 60913
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.4391303062438965,
+      "learning_rate": 1.7797449141249146e-09,
+      "loss": 0.4857,
+      "step": 60914
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 1.9548393487930298,
+      "learning_rate": 1.7727177156934283e-09,
+      "loss": 0.3674,
+      "step": 60915
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.7924184799194336,
+      "learning_rate": 1.765704415459535e-09,
+      "loss": 0.5532,
+      "step": 60916
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.734302043914795,
+      "learning_rate": 1.7587050134426632e-09,
+      "loss": 0.5605,
+      "step": 60917
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.2643494606018066,
+      "learning_rate": 1.751719509662797e-09,
+      "loss": 0.5864,
+      "step": 60918
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.5116724967956543,
+      "learning_rate": 1.7447479041388106e-09,
+      "loss": 0.3923,
+      "step": 60919
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.0079967975616455,
+      "learning_rate": 1.7377901968906873e-09,
+      "loss": 0.6112,
+      "step": 60920
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.71773099899292,
+      "learning_rate": 1.7308463879367466e-09,
+      "loss": 0.9171,
+      "step": 60921
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.5435874462127686,
+      "learning_rate": 1.723916477297527e-09,
+      "loss": 0.7094,
+      "step": 60922
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.30918288230896,
+      "learning_rate": 1.7170004649913473e-09,
+      "loss": 0.4712,
+      "step": 60923
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 4.181159019470215,
+      "learning_rate": 1.7100983510376367e-09,
+      "loss": 0.7914,
+      "step": 60924
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.869515895843506,
+      "learning_rate": 1.703210135455824e-09,
+      "loss": 0.5844,
+      "step": 60925
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.7940726280212402,
+      "learning_rate": 1.6963358182653377e-09,
+      "loss": 0.6368,
+      "step": 60926
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 6.395294189453125,
+      "learning_rate": 1.689475399484497e-09,
+      "loss": 0.4371,
+      "step": 60927
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.8986356258392334,
+      "learning_rate": 1.6826288791332857e-09,
+      "loss": 0.398,
+      "step": 60928
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.443800926208496,
+      "learning_rate": 1.6757962572300224e-09,
+      "loss": 0.5346,
+      "step": 60929
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.961944580078125,
+      "learning_rate": 1.668977533793581e-09,
+      "loss": 0.3513,
+      "step": 60930
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.7314794063568115,
+      "learning_rate": 1.6621727088439453e-09,
+      "loss": 0.4933,
+      "step": 60931
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.7142834663391113,
+      "learning_rate": 1.6553817823988794e-09,
+      "loss": 0.644,
+      "step": 60932
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 4.125654220581055,
+      "learning_rate": 1.6486047544783667e-09,
+      "loss": 0.9599,
+      "step": 60933
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.7276360988616943,
+      "learning_rate": 1.6418416251001713e-09,
+      "loss": 0.7953,
+      "step": 60934
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.123572826385498,
+      "learning_rate": 1.6350923942837216e-09,
+      "loss": 0.8622,
+      "step": 60935
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.9355881214141846,
+      "learning_rate": 1.6283570620473366e-09,
+      "loss": 0.7012,
+      "step": 60936
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.888270139694214,
+      "learning_rate": 1.621635628410445e-09,
+      "loss": 0.4514,
+      "step": 60937
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.745652914047241,
+      "learning_rate": 1.6149280933913658e-09,
+      "loss": 0.6971,
+      "step": 60938
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.0360820293426514,
+      "learning_rate": 1.6082344570084173e-09,
+      "loss": 0.3231,
+      "step": 60939
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.0218942165374756,
+      "learning_rate": 1.6015547192810287e-09,
+      "loss": 0.6702,
+      "step": 60940
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 5.4830241203308105,
+      "learning_rate": 1.5948888802269635e-09,
+      "loss": 0.753,
+      "step": 60941
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.8289196491241455,
+      "learning_rate": 1.5882369398650954e-09,
+      "loss": 0.5508,
+      "step": 60942
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.0943524837493896,
+      "learning_rate": 1.581598898213743e-09,
+      "loss": 0.7146,
+      "step": 60943
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.3470566272735596,
+      "learning_rate": 1.5749747552917805e-09,
+      "loss": 0.6154,
+      "step": 60944
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.024594306945801,
+      "learning_rate": 1.568364511116971e-09,
+      "loss": 0.4511,
+      "step": 60945
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.5272586345672607,
+      "learning_rate": 1.5617681657081885e-09,
+      "loss": 0.479,
+      "step": 60946
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.233219861984253,
+      "learning_rate": 1.5551857190837515e-09,
+      "loss": 0.3237,
+      "step": 60947
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.4943249225616455,
+      "learning_rate": 1.548617171261979e-09,
+      "loss": 0.569,
+      "step": 60948
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.411524534225464,
+      "learning_rate": 1.5420625222611896e-09,
+      "loss": 0.3983,
+      "step": 60949
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.3820273876190186,
+      "learning_rate": 1.5355217720991467e-09,
+      "loss": 0.5511,
+      "step": 60950
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 4.462559700012207,
+      "learning_rate": 1.528994920794724e-09,
+      "loss": 0.5437,
+      "step": 60951
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.5034215450286865,
+      "learning_rate": 1.5224819683651304e-09,
+      "loss": 0.5809,
+      "step": 60952
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.6616759300231934,
+      "learning_rate": 1.5159829148297944e-09,
+      "loss": 0.7256,
+      "step": 60953
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.8675925731658936,
+      "learning_rate": 1.5094977602053695e-09,
+      "loss": 0.803,
+      "step": 60954
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.2363455295562744,
+      "learning_rate": 1.5030265045107295e-09,
+      "loss": 0.633,
+      "step": 60955
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.92392635345459,
+      "learning_rate": 1.4965691477636379e-09,
+      "loss": 0.5387,
+      "step": 60956
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.318480968475342,
+      "learning_rate": 1.4901256899824135e-09,
+      "loss": 0.5154,
+      "step": 60957
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.265981435775757,
+      "learning_rate": 1.4836961311848197e-09,
+      "loss": 0.4882,
+      "step": 60958
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.8581697940826416,
+      "learning_rate": 1.477280471388065e-09,
+      "loss": 0.4661,
+      "step": 60959
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.60988712310791,
+      "learning_rate": 1.4708787106104682e-09,
+      "loss": 0.4539,
+      "step": 60960
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.3954710960388184,
+      "learning_rate": 1.464490848870348e-09,
+      "loss": 0.5344,
+      "step": 60961
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.653416633605957,
+      "learning_rate": 1.4581168861843576e-09,
+      "loss": 0.5736,
+      "step": 60962
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.7408607006073,
+      "learning_rate": 1.4517568225713708e-09,
+      "loss": 0.8113,
+      "step": 60963
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.185913324356079,
+      "learning_rate": 1.445410658048596e-09,
+      "loss": 0.6992,
+      "step": 60964
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.385681390762329,
+      "learning_rate": 1.4390783926332418e-09,
+      "loss": 0.4596,
+      "step": 60965
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.867988109588623,
+      "learning_rate": 1.432760026343627e-09,
+      "loss": 0.508,
+      "step": 60966
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.8458480834960938,
+      "learning_rate": 1.4264555591964047e-09,
+      "loss": 0.8269,
+      "step": 60967
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.1070778369903564,
+      "learning_rate": 1.4201649912104486e-09,
+      "loss": 0.5943,
+      "step": 60968
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 21.732202529907227,
+      "learning_rate": 1.4138883224018574e-09,
+      "loss": 0.47,
+      "step": 60969
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.012639045715332,
+      "learning_rate": 1.4076255527895043e-09,
+      "loss": 0.5108,
+      "step": 60970
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.261824369430542,
+      "learning_rate": 1.401376682389488e-09,
+      "loss": 0.4511,
+      "step": 60971
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.731315851211548,
+      "learning_rate": 1.395141711219572e-09,
+      "loss": 0.5929,
+      "step": 60972
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.5764501094818115,
+      "learning_rate": 1.3889206392975197e-09,
+      "loss": 0.7076,
+      "step": 60973
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.021909236907959,
+      "learning_rate": 1.3827134666405396e-09,
+      "loss": 0.5237,
+      "step": 60974
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.7301785945892334,
+      "learning_rate": 1.376520193265285e-09,
+      "loss": 0.426,
+      "step": 60975
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 5.598278999328613,
+      "learning_rate": 1.3703408191900748e-09,
+      "loss": 0.7875,
+      "step": 60976
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 4.412508487701416,
+      "learning_rate": 1.3641753444310068e-09,
+      "loss": 0.394,
+      "step": 60977
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.4518747329711914,
+      "learning_rate": 1.35802376900529e-09,
+      "loss": 0.5373,
+      "step": 60978
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.7193679809570312,
+      "learning_rate": 1.3518860929306876e-09,
+      "loss": 0.6801,
+      "step": 60979
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.0051426887512207,
+      "learning_rate": 1.3457623162238532e-09,
+      "loss": 0.6549,
+      "step": 60980
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.9632914066314697,
+      "learning_rate": 1.339652438901995e-09,
+      "loss": 0.6233,
+      "step": 60981
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.196880340576172,
+      "learning_rate": 1.3335564609823215e-09,
+      "loss": 0.7456,
+      "step": 60982
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.479933023452759,
+      "learning_rate": 1.327474382480931e-09,
+      "loss": 0.5966,
+      "step": 60983
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.1054673194885254,
+      "learning_rate": 1.3214062034155873e-09,
+      "loss": 0.5801,
+      "step": 60984
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.1318519115448,
+      "learning_rate": 1.3153519238029433e-09,
+      "loss": 0.4812,
+      "step": 60985
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.2087700366973877,
+      "learning_rate": 1.3093115436596527e-09,
+      "loss": 0.5318,
+      "step": 60986
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.2100601196289062,
+      "learning_rate": 1.3032850630023685e-09,
+      "loss": 0.305,
+      "step": 60987
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.9815750122070312,
+      "learning_rate": 1.2972724818488546e-09,
+      "loss": 0.5706,
+      "step": 60988
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 4.156719207763672,
+      "learning_rate": 1.291273800214099e-09,
+      "loss": 0.5712,
+      "step": 60989
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.784902334213257,
+      "learning_rate": 1.2852890181164202e-09,
+      "loss": 0.5649,
+      "step": 60990
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.419924736022949,
+      "learning_rate": 1.2793181355713613e-09,
+      "loss": 0.3704,
+      "step": 60991
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 7.9490203857421875,
+      "learning_rate": 1.273361152596131e-09,
+      "loss": 0.8122,
+      "step": 60992
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.313737392425537,
+      "learning_rate": 1.2674180692073823e-09,
+      "loss": 0.5404,
+      "step": 60993
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.661078453063965,
+      "learning_rate": 1.2614888854212137e-09,
+      "loss": 0.7017,
+      "step": 60994
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 5.104494094848633,
+      "learning_rate": 1.2555736012542785e-09,
+      "loss": 0.5265,
+      "step": 60995
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.7424967288970947,
+      "learning_rate": 1.249672216722675e-09,
+      "loss": 0.4767,
+      "step": 60996
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.4044103622436523,
+      "learning_rate": 1.2437847318436115e-09,
+      "loss": 0.5852,
+      "step": 60997
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.3034846782684326,
+      "learning_rate": 1.237911146632631e-09,
+      "loss": 0.4143,
+      "step": 60998
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.2820870876312256,
+      "learning_rate": 1.2320514611069424e-09,
+      "loss": 0.4393,
+      "step": 60999
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.31449818611145,
+      "learning_rate": 1.2262056752820884e-09,
+      "loss": 0.5388,
+      "step": 61000
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.1725525856018066,
+      "learning_rate": 1.2203737891741674e-09,
+      "loss": 0.6809,
+      "step": 61001
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.4273388385772705,
+      "learning_rate": 1.2145558028003878e-09,
+      "loss": 0.5362,
+      "step": 61002
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.8094849586486816,
+      "learning_rate": 1.2087517161762929e-09,
+      "loss": 0.3738,
+      "step": 61003
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.0894155502319336,
+      "learning_rate": 1.2029615293174258e-09,
+      "loss": 0.5475,
+      "step": 61004
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 4.276566505432129,
+      "learning_rate": 1.1971852422409946e-09,
+      "loss": 0.4923,
+      "step": 61005
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.1731903553009033,
+      "learning_rate": 1.1914228549625428e-09,
+      "loss": 0.5817,
+      "step": 61006
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.5259621143341064,
+      "learning_rate": 1.1856743674981686e-09,
+      "loss": 0.7865,
+      "step": 61007
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.4615259170532227,
+      "learning_rate": 1.17993977986397e-09,
+      "loss": 0.3681,
+      "step": 61008
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 4.381690502166748,
+      "learning_rate": 1.1742190920760455e-09,
+      "loss": 0.6328,
+      "step": 61009
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.7041587829589844,
+      "learning_rate": 1.1685123041493828e-09,
+      "loss": 0.5757,
+      "step": 61010
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.971585273742676,
+      "learning_rate": 1.1628194161006357e-09,
+      "loss": 0.4905,
+      "step": 61011
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 4.418673038482666,
+      "learning_rate": 1.157140427945902e-09,
+      "loss": 0.6709,
+      "step": 61012
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.3593430519104004,
+      "learning_rate": 1.15147533970017e-09,
+      "loss": 0.4281,
+      "step": 61013
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.9573757648468018,
+      "learning_rate": 1.1458241513795377e-09,
+      "loss": 0.4015,
+      "step": 61014
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.7786638736724854,
+      "learning_rate": 1.1401868629995482e-09,
+      "loss": 0.778,
+      "step": 61015
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.540468215942383,
+      "learning_rate": 1.1345634745763e-09,
+      "loss": 0.4304,
+      "step": 61016
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.4094114303588867,
+      "learning_rate": 1.1289539861253362e-09,
+      "loss": 0.4664,
+      "step": 61017
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.611607789993286,
+      "learning_rate": 1.1233583976621999e-09,
+      "loss": 0.8171,
+      "step": 61018
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.025535821914673,
+      "learning_rate": 1.1177767092024338e-09,
+      "loss": 0.5082,
+      "step": 61019
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.1862847805023193,
+      "learning_rate": 1.1122089207610264e-09,
+      "loss": 0.7847,
+      "step": 61020
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.1846742630004883,
+      "learning_rate": 1.106655032354631e-09,
+      "loss": 0.6978,
+      "step": 61021
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.1273155212402344,
+      "learning_rate": 1.1011150439976804e-09,
+      "loss": 0.5268,
+      "step": 61022
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.002058506011963,
+      "learning_rate": 1.0955889557057176e-09,
+      "loss": 0.55,
+      "step": 61023
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.2917473316192627,
+      "learning_rate": 1.0900767674942858e-09,
+      "loss": 0.3751,
+      "step": 61024
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.3552865982055664,
+      "learning_rate": 1.0845784793789282e-09,
+      "loss": 0.585,
+      "step": 61025
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.548288106918335,
+      "learning_rate": 1.0790940913746328e-09,
+      "loss": 0.5585,
+      "step": 61026
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.546355724334717,
+      "learning_rate": 1.0736236034963876e-09,
+      "loss": 0.5704,
+      "step": 61027
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 4.0754523277282715,
+      "learning_rate": 1.0681670157602907e-09,
+      "loss": 0.5513,
+      "step": 61028
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.50014066696167,
+      "learning_rate": 1.0627243281807754e-09,
+      "loss": 0.4347,
+      "step": 61029
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.6458182334899902,
+      "learning_rate": 1.0572955407733843e-09,
+      "loss": 0.4604,
+      "step": 61030
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.957221269607544,
+      "learning_rate": 1.0518806535525506e-09,
+      "loss": 0.5937,
+      "step": 61031
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.5986459255218506,
+      "learning_rate": 1.0464796665343724e-09,
+      "loss": 0.4587,
+      "step": 61032
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.756779432296753,
+      "learning_rate": 1.0410925797327277e-09,
+      "loss": 0.5036,
+      "step": 61033
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 5.880678176879883,
+      "learning_rate": 1.0357193931631593e-09,
+      "loss": 0.5661,
+      "step": 61034
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.628843307495117,
+      "learning_rate": 1.0303601068412107e-09,
+      "loss": 0.7589,
+      "step": 61035
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.6698830127716064,
+      "learning_rate": 1.0250147207807593e-09,
+      "loss": 0.4642,
+      "step": 61036
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.8163864612579346,
+      "learning_rate": 1.0196832349967934e-09,
+      "loss": 0.5583,
+      "step": 61037
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.511042833328247,
+      "learning_rate": 1.014365649504856e-09,
+      "loss": 0.6747,
+      "step": 61038
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.373371124267578,
+      "learning_rate": 1.00906196431938e-09,
+      "loss": 0.8243,
+      "step": 61039
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.7292404174804688,
+      "learning_rate": 1.0037721794547984e-09,
+      "loss": 0.6624,
+      "step": 61040
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.9828615188598633,
+      "learning_rate": 9.98496294926099e-10,
+      "loss": 0.5599,
+      "step": 61041
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.5052623748779297,
+      "learning_rate": 9.932343107477149e-10,
+      "loss": 0.4993,
+      "step": 61042
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.5028116703033447,
+      "learning_rate": 9.87986226934634e-10,
+      "loss": 0.5204,
+      "step": 61043
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.7634334564208984,
+      "learning_rate": 9.827520435018444e-10,
+      "loss": 0.4534,
+      "step": 61044
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.0606095790863037,
+      "learning_rate": 9.775317604626688e-10,
+      "loss": 0.609,
+      "step": 61045
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.8054137229919434,
+      "learning_rate": 9.723253778326503e-10,
+      "loss": 0.6539,
+      "step": 61046
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 5.281545639038086,
+      "learning_rate": 9.671328956256665e-10,
+      "loss": 0.549,
+      "step": 61047
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 4.160970211029053,
+      "learning_rate": 9.619543138561505e-10,
+      "loss": 0.6618,
+      "step": 61048
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.0217018127441406,
+      "learning_rate": 9.567896325390902e-10,
+      "loss": 0.4522,
+      "step": 61049
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 4.33492374420166,
+      "learning_rate": 9.516388516889186e-10,
+      "loss": 0.8135,
+      "step": 61050
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 4.242478370666504,
+      "learning_rate": 9.465019713189582e-10,
+      "loss": 0.7967,
+      "step": 61051
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.192291736602783,
+      "learning_rate": 9.413789914441973e-10,
+      "loss": 0.4,
+      "step": 61052
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.364213228225708,
+      "learning_rate": 9.362699120790686e-10,
+      "loss": 0.4427,
+      "step": 61053
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.3581714630126953,
+      "learning_rate": 9.311747332368948e-10,
+      "loss": 0.475,
+      "step": 61054
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.6665406227111816,
+      "learning_rate": 9.26093454933219e-10,
+      "loss": 0.5284,
+      "step": 61055
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.8856265544891357,
+      "learning_rate": 9.210260771808088e-10,
+      "loss": 0.4461,
+      "step": 61056
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.615025520324707,
+      "learning_rate": 9.159725999946523e-10,
+      "loss": 0.4608,
+      "step": 61057
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.0980329513549805,
+      "learning_rate": 9.10933023388072e-10,
+      "loss": 0.5985,
+      "step": 61058
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.991328001022339,
+      "learning_rate": 9.059073473755009e-10,
+      "loss": 0.7569,
+      "step": 61059
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.0429821014404297,
+      "learning_rate": 9.008955719713719e-10,
+      "loss": 0.5213,
+      "step": 61060
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.767162561416626,
+      "learning_rate": 8.958976971890077e-10,
+      "loss": 0.5696,
+      "step": 61061
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.0337677001953125,
+      "learning_rate": 8.909137230422859e-10,
+      "loss": 0.4616,
+      "step": 61062
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.941507339477539,
+      "learning_rate": 8.859436495450846e-10,
+      "loss": 0.6074,
+      "step": 61063
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.2300424575805664,
+      "learning_rate": 8.809874767118365e-10,
+      "loss": 0.6037,
+      "step": 61064
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.037186861038208,
+      "learning_rate": 8.760452045553092e-10,
+      "loss": 0.7413,
+      "step": 61065
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.040865659713745,
+      "learning_rate": 8.711168330899356e-10,
+      "loss": 0.7786,
+      "step": 61066
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.888065814971924,
+      "learning_rate": 8.662023623295934e-10,
+      "loss": 0.5995,
+      "step": 61067
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 7.105446815490723,
+      "learning_rate": 8.613017922870504e-10,
+      "loss": 0.4928,
+      "step": 61068
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.9306886196136475,
+      "learning_rate": 8.564151229772943e-10,
+      "loss": 0.5506,
+      "step": 61069
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.8482959270477295,
+      "learning_rate": 8.515423544125379e-10,
+      "loss": 0.668,
+      "step": 61070
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.671710252761841,
+      "learning_rate": 8.466834866072138e-10,
+      "loss": 0.4216,
+      "step": 61071
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.756704092025757,
+      "learning_rate": 8.418385195740896e-10,
+      "loss": 0.7124,
+      "step": 61072
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.0492305755615234,
+      "learning_rate": 8.370074533275985e-10,
+      "loss": 0.633,
+      "step": 61073
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.2750093936920166,
+      "learning_rate": 8.321902878805077e-10,
+      "loss": 0.4828,
+      "step": 61074
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.40810489654541,
+      "learning_rate": 8.273870232461401e-10,
+      "loss": 0.402,
+      "step": 61075
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.0720789432525635,
+      "learning_rate": 8.225976594383734e-10,
+      "loss": 0.727,
+      "step": 61076
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.1352407932281494,
+      "learning_rate": 8.17822196469975e-10,
+      "loss": 0.5548,
+      "step": 61077
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.304144859313965,
+      "learning_rate": 8.13060634354823e-10,
+      "loss": 0.3,
+      "step": 61078
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 4.099022388458252,
+      "learning_rate": 8.083129731051298e-10,
+      "loss": 0.6699,
+      "step": 61079
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.589728355407715,
+      "learning_rate": 8.035792127353281e-10,
+      "loss": 0.5341,
+      "step": 61080
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.279168128967285,
+      "learning_rate": 7.988593532581856e-10,
+      "loss": 0.4629,
+      "step": 61081
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.4093990325927734,
+      "learning_rate": 7.941533946864699e-10,
+      "loss": 0.5958,
+      "step": 61082
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.22594952583313,
+      "learning_rate": 7.894613370335036e-10,
+      "loss": 0.4107,
+      "step": 61083
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.4671223163604736,
+      "learning_rate": 7.847831803120543e-10,
+      "loss": 0.7123,
+      "step": 61084
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.7215676307678223,
+      "learning_rate": 7.801189245354446e-10,
+      "loss": 0.7408,
+      "step": 61085
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.0671284198760986,
+      "learning_rate": 7.754685697164421e-10,
+      "loss": 0.5015,
+      "step": 61086
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.1298232078552246,
+      "learning_rate": 7.708321158683696e-10,
+      "loss": 0.8081,
+      "step": 61087
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 7.754110813140869,
+      "learning_rate": 7.662095630039945e-10,
+      "loss": 0.4519,
+      "step": 61088
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.9103784561157227,
+      "learning_rate": 7.616009111360844e-10,
+      "loss": 0.6029,
+      "step": 61089
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.352450370788574,
+      "learning_rate": 7.570061602768519e-10,
+      "loss": 0.6029,
+      "step": 61090
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.834961414337158,
+      "learning_rate": 7.524253104401746e-10,
+      "loss": 0.5421,
+      "step": 61091
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.4807040691375732,
+      "learning_rate": 7.4785836163771e-10,
+      "loss": 0.6545,
+      "step": 61092
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 4.016885757446289,
+      "learning_rate": 7.433053138833358e-10,
+      "loss": 0.6191,
+      "step": 61093
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.5939888954162598,
+      "learning_rate": 7.387661671887091e-10,
+      "loss": 0.5828,
+      "step": 61094
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.163203001022339,
+      "learning_rate": 7.34240921566598e-10,
+      "loss": 0.7109,
+      "step": 61095
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.590397834777832,
+      "learning_rate": 7.297295770303247e-10,
+      "loss": 0.4166,
+      "step": 61096
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.281893730163574,
+      "learning_rate": 7.252321335915468e-10,
+      "loss": 0.6294,
+      "step": 61097
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.9554696083068848,
+      "learning_rate": 7.207485912630318e-10,
+      "loss": 0.6805,
+      "step": 61098
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 10.034076690673828,
+      "learning_rate": 7.162789500575474e-10,
+      "loss": 0.5927,
+      "step": 61099
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.445134401321411,
+      "learning_rate": 7.118232099873057e-10,
+      "loss": 0.7208,
+      "step": 61100
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.301859140396118,
+      "learning_rate": 7.073813710645194e-10,
+      "loss": 0.5344,
+      "step": 61101
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.0512278079986572,
+      "learning_rate": 7.029534333019561e-10,
+      "loss": 0.4423,
+      "step": 61102
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.403278350830078,
+      "learning_rate": 6.985393967118281e-10,
+      "loss": 0.3428,
+      "step": 61103
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.90393328666687,
+      "learning_rate": 6.941392613057929e-10,
+      "loss": 0.4558,
+      "step": 61104
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.6691765785217285,
+      "learning_rate": 6.897530270971731e-10,
+      "loss": 0.631,
+      "step": 61105
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.857107400894165,
+      "learning_rate": 6.853806940970709e-10,
+      "loss": 0.5751,
+      "step": 61106
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.4884352684020996,
+      "learning_rate": 6.810222623182538e-10,
+      "loss": 0.5031,
+      "step": 61107
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 4.176062107086182,
+      "learning_rate": 6.766777317729345e-10,
+      "loss": 0.4976,
+      "step": 61108
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.4871859550476074,
+      "learning_rate": 6.723471024727702e-10,
+      "loss": 0.4934,
+      "step": 61109
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 4.452648639678955,
+      "learning_rate": 6.680303744299732e-10,
+      "loss": 0.6021,
+      "step": 61110
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.42325496673584,
+      "learning_rate": 6.637275476567561e-10,
+      "loss": 0.5989,
+      "step": 61111
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.8454957008361816,
+      "learning_rate": 6.594386221647764e-10,
+      "loss": 0.6312,
+      "step": 61112
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.828061580657959,
+      "learning_rate": 6.551635979662463e-10,
+      "loss": 0.5305,
+      "step": 61113
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.5566561222076416,
+      "learning_rate": 6.509024750728232e-10,
+      "loss": 0.4272,
+      "step": 61114
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.9344394207000732,
+      "learning_rate": 6.466552534967197e-10,
+      "loss": 0.3513,
+      "step": 61115
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.0907607078552246,
+      "learning_rate": 6.424219332490378e-10,
+      "loss": 0.5801,
+      "step": 61116
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.1575469970703125,
+      "learning_rate": 6.382025143419901e-10,
+      "loss": 0.6133,
+      "step": 61117
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.9143078327178955,
+      "learning_rate": 6.33996996787789e-10,
+      "loss": 0.5334,
+      "step": 61118
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.0541152954101562,
+      "learning_rate": 6.298053805969817e-10,
+      "loss": 0.6657,
+      "step": 61119
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.4507858753204346,
+      "learning_rate": 6.256276657823357e-10,
+      "loss": 0.4536,
+      "step": 61120
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 4.185442924499512,
+      "learning_rate": 6.214638523549532e-10,
+      "loss": 0.6112,
+      "step": 61121
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.951770067214966,
+      "learning_rate": 6.173139403264916e-10,
+      "loss": 0.6499,
+      "step": 61122
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.0046627521514893,
+      "learning_rate": 6.131779297080531e-10,
+      "loss": 0.6305,
+      "step": 61123
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.683056116104126,
+      "learning_rate": 6.090558205118502e-10,
+      "loss": 0.5039,
+      "step": 61124
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.1728408336639404,
+      "learning_rate": 6.049476127495402e-10,
+      "loss": 0.5144,
+      "step": 61125
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.2193334102630615,
+      "learning_rate": 6.008533064311151e-10,
+      "loss": 0.4506,
+      "step": 61126
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 6.022141456604004,
+      "learning_rate": 5.967729015693425e-10,
+      "loss": 0.6525,
+      "step": 61127
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.439112901687622,
+      "learning_rate": 5.927063981753245e-10,
+      "loss": 0.6168,
+      "step": 61128
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.6655139923095703,
+      "learning_rate": 5.886537962601635e-10,
+      "loss": 0.7266,
+      "step": 61129
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.2712419033050537,
+      "learning_rate": 5.846150958349617e-10,
+      "loss": 0.6867,
+      "step": 61130
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.1917481422424316,
+      "learning_rate": 5.805902969108213e-10,
+      "loss": 0.6409,
+      "step": 61131
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.2079317569732666,
+      "learning_rate": 5.765793994999546e-10,
+      "loss": 0.4453,
+      "step": 61132
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.0474891662597656,
+      "learning_rate": 5.725824036123539e-10,
+      "loss": 0.4428,
+      "step": 61133
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.5429506301879883,
+      "learning_rate": 5.685993092596765e-10,
+      "loss": 0.4437,
+      "step": 61134
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.4952008724212646,
+      "learning_rate": 5.646301164524692e-10,
+      "loss": 0.5032,
+      "step": 61135
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.65101957321167,
+      "learning_rate": 5.606748252029448e-10,
+      "loss": 0.4923,
+      "step": 61136
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 4.405506134033203,
+      "learning_rate": 5.5673343552054e-10,
+      "loss": 0.5142,
+      "step": 61137
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.494661569595337,
+      "learning_rate": 5.528059474174674e-10,
+      "loss": 0.7514,
+      "step": 61138
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.2692956924438477,
+      "learning_rate": 5.48892360904274e-10,
+      "loss": 0.8967,
+      "step": 61139
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.320603847503662,
+      "learning_rate": 5.44992675991507e-10,
+      "loss": 0.5251,
+      "step": 61140
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.4584953784942627,
+      "learning_rate": 5.411068926908236e-10,
+      "loss": 0.5378,
+      "step": 61141
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 4.493257999420166,
+      "learning_rate": 5.372350110116609e-10,
+      "loss": 0.6313,
+      "step": 61142
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.0946667194366455,
+      "learning_rate": 5.333770309662312e-10,
+      "loss": 0.682,
+      "step": 61143
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.544539451599121,
+      "learning_rate": 5.295329525645265e-10,
+      "loss": 0.6542,
+      "step": 61144
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.685589075088501,
+      "learning_rate": 5.257027758170941e-10,
+      "loss": 0.5789,
+      "step": 61145
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.88063645362854,
+      "learning_rate": 5.21886500735036e-10,
+      "loss": 0.743,
+      "step": 61146
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.5175530910491943,
+      "learning_rate": 5.180841273288994e-10,
+      "loss": 0.5252,
+      "step": 61147
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.1055071353912354,
+      "learning_rate": 5.142956556092316e-10,
+      "loss": 0.4515,
+      "step": 61148
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.631319522857666,
+      "learning_rate": 5.105210855860244e-10,
+      "loss": 0.3563,
+      "step": 61149
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.985433578491211,
+      "learning_rate": 5.067604172703799e-10,
+      "loss": 0.5998,
+      "step": 61150
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.6647756099700928,
+      "learning_rate": 5.030136506722905e-10,
+      "loss": 0.5895,
+      "step": 61151
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.769324541091919,
+      "learning_rate": 4.992807858028581e-10,
+      "loss": 0.542,
+      "step": 61152
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.5603175163269043,
+      "learning_rate": 4.955618226720749e-10,
+      "loss": 0.5934,
+      "step": 61153
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.7735023498535156,
+      "learning_rate": 4.918567612899328e-10,
+      "loss": 0.6313,
+      "step": 61154
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.7965524196624756,
+      "learning_rate": 4.88165601667534e-10,
+      "loss": 0.3309,
+      "step": 61155
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.388784408569336,
+      "learning_rate": 4.844883438143155e-10,
+      "loss": 0.9133,
+      "step": 61156
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.2685375213623047,
+      "learning_rate": 4.808249877413795e-10,
+      "loss": 0.6201,
+      "step": 61157
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.0409862995147705,
+      "learning_rate": 4.771755334576078e-10,
+      "loss": 0.6887,
+      "step": 61158
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.530165195465088,
+      "learning_rate": 4.735399809746577e-10,
+      "loss": 0.6477,
+      "step": 61159
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.356351137161255,
+      "learning_rate": 4.699183303014109e-10,
+      "loss": 0.5986,
+      "step": 61160
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.434617042541504,
+      "learning_rate": 4.663105814489699e-10,
+      "loss": 0.4494,
+      "step": 61161
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 4.94893217086792,
+      "learning_rate": 4.6271673442677135e-10,
+      "loss": 0.6385,
+      "step": 61162
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.442854881286621,
+      "learning_rate": 4.591367892442522e-10,
+      "loss": 0.6575,
+      "step": 61163
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.869774103164673,
+      "learning_rate": 4.555707459125147e-10,
+      "loss": 0.467,
+      "step": 61164
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.2367846965789795,
+      "learning_rate": 4.520186044409958e-10,
+      "loss": 0.6902,
+      "step": 61165
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.298858165740967,
+      "learning_rate": 4.484803648396874e-10,
+      "loss": 0.5328,
+      "step": 61166
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.717681646347046,
+      "learning_rate": 4.449560271180264e-10,
+      "loss": 0.6344,
+      "step": 61167
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.639331102371216,
+      "learning_rate": 4.4144559128600494e-10,
+      "loss": 0.4952,
+      "step": 61168
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 4.011984348297119,
+      "learning_rate": 4.3794905735361494e-10,
+      "loss": 0.611,
+      "step": 61169
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.320183277130127,
+      "learning_rate": 4.3446642533029327e-10,
+      "loss": 0.7737,
+      "step": 61170
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.7897088527679443,
+      "learning_rate": 4.3099769522547685e-10,
+      "loss": 0.6211,
+      "step": 61171
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.2067296504974365,
+      "learning_rate": 4.275428670497128e-10,
+      "loss": 0.5792,
+      "step": 61172
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.415544271469116,
+      "learning_rate": 4.2410194081188296e-10,
+      "loss": 0.4587,
+      "step": 61173
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.6066527366638184,
+      "learning_rate": 4.206749165219792e-10,
+      "loss": 0.5947,
+      "step": 61174
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.2272849082946777,
+      "learning_rate": 4.1726179418888344e-10,
+      "loss": 0.5517,
+      "step": 61175
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.6337497234344482,
+      "learning_rate": 4.1386257382258765e-10,
+      "loss": 0.6516,
+      "step": 61176
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.1524906158447266,
+      "learning_rate": 4.104772554325287e-10,
+      "loss": 0.5051,
+      "step": 61177
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.2413666248321533,
+      "learning_rate": 4.071058390281435e-10,
+      "loss": 0.5934,
+      "step": 61178
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 1.8868167400360107,
+      "learning_rate": 4.0374832461886894e-10,
+      "loss": 0.3135,
+      "step": 61179
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.2033934593200684,
+      "learning_rate": 4.004047122135868e-10,
+      "loss": 0.3867,
+      "step": 61180
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.9368200302124023,
+      "learning_rate": 3.97075001821734e-10,
+      "loss": 0.5353,
+      "step": 61181
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.4126617908477783,
+      "learning_rate": 3.937591934533025e-10,
+      "loss": 0.5738,
+      "step": 61182
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 4.8990092277526855,
+      "learning_rate": 3.9045728711661903e-10,
+      "loss": 0.7669,
+      "step": 61183
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.7896292209625244,
+      "learning_rate": 3.8716928282112043e-10,
+      "loss": 0.5154,
+      "step": 61184
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.3500733375549316,
+      "learning_rate": 3.838951805756885e-10,
+      "loss": 0.6015,
+      "step": 61185
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.489363193511963,
+      "learning_rate": 3.806349803903153e-10,
+      "loss": 0.7663,
+      "step": 61186
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.598322868347168,
+      "learning_rate": 3.773886822733275e-10,
+      "loss": 0.7116,
+      "step": 61187
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 4.496837615966797,
+      "learning_rate": 3.741562862336068e-10,
+      "loss": 0.5607,
+      "step": 61188
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.4858922958374023,
+      "learning_rate": 3.7093779228059014e-10,
+      "loss": 0.5422,
+      "step": 61189
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.4046642780303955,
+      "learning_rate": 3.677332004231593e-10,
+      "loss": 0.6452,
+      "step": 61190
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 13.77392578125,
+      "learning_rate": 3.645425106701961e-10,
+      "loss": 0.7661,
+      "step": 61191
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 4.204793453216553,
+      "learning_rate": 3.6136572303058227e-10,
+      "loss": 0.4612,
+      "step": 61192
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.2614731788635254,
+      "learning_rate": 3.582028375131996e-10,
+      "loss": 0.6707,
+      "step": 61193
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.476212978363037,
+      "learning_rate": 3.5505385412637484e-10,
+      "loss": 0.5714,
+      "step": 61194
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.825291872024536,
+      "learning_rate": 3.5191877287954476e-10,
+      "loss": 0.4638,
+      "step": 61195
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.865675210952759,
+      "learning_rate": 3.4879759378103616e-10,
+      "loss": 0.6388,
+      "step": 61196
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.334043264389038,
+      "learning_rate": 3.456903168397308e-10,
+      "loss": 0.3618,
+      "step": 61197
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 4.48281717300415,
+      "learning_rate": 3.425969420639552e-10,
+      "loss": 0.7056,
+      "step": 61198
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.0387299060821533,
+      "learning_rate": 3.3951746946259134e-10,
+      "loss": 0.5556,
+      "step": 61199
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 4.46421480178833,
+      "learning_rate": 3.364518990439658e-10,
+      "loss": 0.7057,
+      "step": 61200
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.71891188621521,
+      "learning_rate": 3.334002308169604e-10,
+      "loss": 0.6464,
+      "step": 61201
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.260080337524414,
+      "learning_rate": 3.3036246478990176e-10,
+      "loss": 0.3646,
+      "step": 61202
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 5.870881080627441,
+      "learning_rate": 3.273386009711166e-10,
+      "loss": 0.5279,
+      "step": 61203
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.8842852115631104,
+      "learning_rate": 3.2432863936893155e-10,
+      "loss": 0.3701,
+      "step": 61204
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.3946022987365723,
+      "learning_rate": 3.213325799916733e-10,
+      "loss": 0.4929,
+      "step": 61205
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.4070253372192383,
+      "learning_rate": 3.183504228482237e-10,
+      "loss": 0.658,
+      "step": 61206
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.5124289989471436,
+      "learning_rate": 3.1538216794635425e-10,
+      "loss": 0.3971,
+      "step": 61207
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 4.5508809089660645,
+      "learning_rate": 3.124278152949467e-10,
+      "loss": 0.6083,
+      "step": 61208
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.8136918544769287,
+      "learning_rate": 3.0948736490121755e-10,
+      "loss": 0.5952,
+      "step": 61209
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.0616586208343506,
+      "learning_rate": 3.065608167740486e-10,
+      "loss": 0.5945,
+      "step": 61210
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.728184700012207,
+      "learning_rate": 3.0364817092121135e-10,
+      "loss": 0.4499,
+      "step": 61211
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.150838851928711,
+      "learning_rate": 3.0074942735103253e-10,
+      "loss": 0.545,
+      "step": 61212
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.9660024642944336,
+      "learning_rate": 2.9786458607183877e-10,
+      "loss": 0.5563,
+      "step": 61213
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.124985694885254,
+      "learning_rate": 2.9499364709140167e-10,
+      "loss": 0.5701,
+      "step": 61214
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.6264243125915527,
+      "learning_rate": 2.9213661041749277e-10,
+      "loss": 0.6257,
+      "step": 61215
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.162102460861206,
+      "learning_rate": 2.892934760578836e-10,
+      "loss": 0.5072,
+      "step": 61216
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.308650255203247,
+      "learning_rate": 2.864642440209009e-10,
+      "loss": 0.6331,
+      "step": 61217
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.009748697280884,
+      "learning_rate": 2.836489143148713e-10,
+      "loss": 0.5965,
+      "step": 61218
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.7491817474365234,
+      "learning_rate": 2.8084748694645615e-10,
+      "loss": 0.8005,
+      "step": 61219
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.7871010303497314,
+      "learning_rate": 2.7805996192453723e-10,
+      "loss": 0.4817,
+      "step": 61220
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.7059273719787598,
+      "learning_rate": 2.75286339256331e-10,
+      "loss": 0.6119,
+      "step": 61221
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.654733419418335,
+      "learning_rate": 2.7252661894960897e-10,
+      "loss": 0.5059,
+      "step": 61222
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.118518352508545,
+      "learning_rate": 2.697808010115877e-10,
+      "loss": 0.6733,
+      "step": 61223
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.8696746826171875,
+      "learning_rate": 2.6704888545059373e-10,
+      "loss": 0.4859,
+      "step": 61224
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.69313645362854,
+      "learning_rate": 2.643308722743987e-10,
+      "loss": 0.5696,
+      "step": 61225
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 4.515815734863281,
+      "learning_rate": 2.6162676148966393e-10,
+      "loss": 0.5878,
+      "step": 61226
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.2034504413604736,
+      "learning_rate": 2.589365531047161e-10,
+      "loss": 0.5909,
+      "step": 61227
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.2186102867126465,
+      "learning_rate": 2.562602471267717e-10,
+      "loss": 0.4337,
+      "step": 61228
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.6027443408966064,
+      "learning_rate": 2.5359784356304707e-10,
+      "loss": 0.6839,
+      "step": 61229
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.3163676261901855,
+      "learning_rate": 2.5094934242075875e-10,
+      "loss": 0.4449,
+      "step": 61230
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.721237897872925,
+      "learning_rate": 2.483147437082334e-10,
+      "loss": 0.4275,
+      "step": 61231
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.742465019226074,
+      "learning_rate": 2.456940474321323e-10,
+      "loss": 0.6176,
+      "step": 61232
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.8985369205474854,
+      "learning_rate": 2.430872535991169e-10,
+      "loss": 0.6876,
+      "step": 61233
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 5.429381370544434,
+      "learning_rate": 2.404943622180689e-10,
+      "loss": 0.6426,
+      "step": 61234
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.7358477115631104,
+      "learning_rate": 2.379153732945394e-10,
+      "loss": 0.4368,
+      "step": 61235
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.116586923599243,
+      "learning_rate": 2.353502868368551e-10,
+      "loss": 0.7372,
+      "step": 61236
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.5685317516326904,
+      "learning_rate": 2.3279910285167738e-10,
+      "loss": 0.5995,
+      "step": 61237
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.4503064155578613,
+      "learning_rate": 2.3026182134566756e-10,
+      "loss": 0.3683,
+      "step": 61238
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.1267573833465576,
+      "learning_rate": 2.2773844232659714e-10,
+      "loss": 0.7608,
+      "step": 61239
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.730466604232788,
+      "learning_rate": 2.2522896580168264e-10,
+      "loss": 0.5238,
+      "step": 61240
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.8586361408233643,
+      "learning_rate": 2.2273339177703024e-10,
+      "loss": 0.6178,
+      "step": 61241
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.2507965564727783,
+      "learning_rate": 2.202517202598564e-10,
+      "loss": 0.5676,
+      "step": 61242
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.3156797885894775,
+      "learning_rate": 2.1778395125737762e-10,
+      "loss": 0.4468,
+      "step": 61243
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.7699379920959473,
+      "learning_rate": 2.1533008477625518e-10,
+      "loss": 0.6043,
+      "step": 61244
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.546631097793579,
+      "learning_rate": 2.1289012082370552e-10,
+      "loss": 0.5485,
+      "step": 61245
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.9314026832580566,
+      "learning_rate": 2.1046405940583493e-10,
+      "loss": 0.9485,
+      "step": 61246
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.0006659030914307,
+      "learning_rate": 2.0805190052930469e-10,
+      "loss": 0.4354,
+      "step": 61247
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.6999731063842773,
+      "learning_rate": 2.0565364420188637e-10,
+      "loss": 0.3325,
+      "step": 61248
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 4.294594764709473,
+      "learning_rate": 2.0326929042913112e-10,
+      "loss": 0.6972,
+      "step": 61249
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.864656925201416,
+      "learning_rate": 2.0089883921825536e-10,
+      "loss": 0.6357,
+      "step": 61250
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.315532684326172,
+      "learning_rate": 1.9854229057592045e-10,
+      "loss": 0.6354,
+      "step": 61251
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.6077187061309814,
+      "learning_rate": 1.961996445082326e-10,
+      "loss": 0.6566,
+      "step": 61252
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.9777991771698,
+      "learning_rate": 1.9387090102185312e-10,
+      "loss": 0.6121,
+      "step": 61253
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.40010404586792,
+      "learning_rate": 1.915560601234434e-10,
+      "loss": 0.5766,
+      "step": 61254
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.022050619125366,
+      "learning_rate": 1.8925512181910965e-10,
+      "loss": 0.7946,
+      "step": 61255
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 1.8326536417007446,
+      "learning_rate": 1.8696808611606832e-10,
+      "loss": 0.4288,
+      "step": 61256
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.0065038204193115,
+      "learning_rate": 1.846949530198705e-10,
+      "loss": 0.5235,
+      "step": 61257
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.401761531829834,
+      "learning_rate": 1.8243572253662245e-10,
+      "loss": 0.8387,
+      "step": 61258
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.0465521812438965,
+      "learning_rate": 1.801903946735406e-10,
+      "loss": 0.4828,
+      "step": 61259
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.8972461223602295,
+      "learning_rate": 1.7795896943617608e-10,
+      "loss": 0.6226,
+      "step": 61260
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 6.526818752288818,
+      "learning_rate": 1.7574144683119022e-10,
+      "loss": 0.6258,
+      "step": 61261
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.2582404613494873,
+      "learning_rate": 1.735378268641341e-10,
+      "loss": 0.5423,
+      "step": 61262
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.692704916000366,
+      "learning_rate": 1.7134810954166914e-10,
+      "loss": 0.4534,
+      "step": 61263
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 1.836508870124817,
+      "learning_rate": 1.691722948699015e-10,
+      "loss": 0.4356,
+      "step": 61264
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.3122143745422363,
+      "learning_rate": 1.6701038285438232e-10,
+      "loss": 0.4119,
+      "step": 61265
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.0643205642700195,
+      "learning_rate": 1.6486237350121782e-10,
+      "loss": 0.8611,
+      "step": 61266
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.065966844558716,
+      "learning_rate": 1.6272826681706932e-10,
+      "loss": 0.3779,
+      "step": 61267
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.4888265132904053,
+      "learning_rate": 1.6060806280748798e-10,
+      "loss": 0.5272,
+      "step": 61268
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.3317794799804688,
+      "learning_rate": 1.5850176147746977e-10,
+      "loss": 0.5734,
+      "step": 61269
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.0170204639434814,
+      "learning_rate": 1.5640936283423115e-10,
+      "loss": 0.4621,
+      "step": 61270
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.562737226486206,
+      "learning_rate": 1.5433086688332322e-10,
+      "loss": 0.7424,
+      "step": 61271
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.141571283340454,
+      "learning_rate": 1.52266273629742e-10,
+      "loss": 0.8076,
+      "step": 61272
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.2869813442230225,
+      "learning_rate": 1.5021558308014883e-10,
+      "loss": 0.4527,
+      "step": 61273
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.399698257446289,
+      "learning_rate": 1.481787952395397e-10,
+      "loss": 0.5825,
+      "step": 61274
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.586320638656616,
+      "learning_rate": 1.4615591011346575e-10,
+      "loss": 0.6356,
+      "step": 61275
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.3471221923828125,
+      "learning_rate": 1.441469277085883e-10,
+      "loss": 0.4719,
+      "step": 61276
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.86712384223938,
+      "learning_rate": 1.4215184802990333e-10,
+      "loss": 0.588,
+      "step": 61277
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.282986879348755,
+      "learning_rate": 1.4017067108240688e-10,
+      "loss": 0.5789,
+      "step": 61278
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 5.320700645446777,
+      "learning_rate": 1.3820339687220518e-10,
+      "loss": 0.7062,
+      "step": 61279
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.9238102436065674,
+      "learning_rate": 1.3625002540484934e-10,
+      "loss": 0.7098,
+      "step": 61280
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.1616322994232178,
+      "learning_rate": 1.3431055668533533e-10,
+      "loss": 0.7629,
+      "step": 61281
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 4.13045072555542,
+      "learning_rate": 1.323849907197694e-10,
+      "loss": 0.4236,
+      "step": 61282
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 4.365956783294678,
+      "learning_rate": 1.3047332751259246e-10,
+      "loss": 0.8899,
+      "step": 61283
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 7.60333251953125,
+      "learning_rate": 1.285755670699107e-10,
+      "loss": 0.686,
+      "step": 61284
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.0607144832611084,
+      "learning_rate": 1.2669170939672014e-10,
+      "loss": 0.7158,
+      "step": 61285
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.6916208267211914,
+      "learning_rate": 1.248217544980168e-10,
+      "loss": 0.694,
+      "step": 61286
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 5.81589937210083,
+      "learning_rate": 1.229657023787967e-10,
+      "loss": 0.3487,
+      "step": 61287
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.751331329345703,
+      "learning_rate": 1.21123553045166e-10,
+      "loss": 0.6168,
+      "step": 61288
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.494481086730957,
+      "learning_rate": 1.1929530650156563e-10,
+      "loss": 0.6674,
+      "step": 61289
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.409247636795044,
+      "learning_rate": 1.174809627529916e-10,
+      "loss": 0.6129,
+      "step": 61290
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 4.041528701782227,
+      "learning_rate": 1.15680521804995e-10,
+      "loss": 0.5356,
+      "step": 61291
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.5353901386260986,
+      "learning_rate": 1.1389398366201676e-10,
+      "loss": 0.6042,
+      "step": 61292
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.792747974395752,
+      "learning_rate": 1.1212134832960798e-10,
+      "loss": 0.4723,
+      "step": 61293
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 4.59293794631958,
+      "learning_rate": 1.1036261581220953e-10,
+      "loss": 0.5128,
+      "step": 61294
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.471750497817993,
+      "learning_rate": 1.0861778611481744e-10,
+      "loss": 0.5449,
+      "step": 61295
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.4051246643066406,
+      "learning_rate": 1.068868592424277e-10,
+      "loss": 0.5524,
+      "step": 61296
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.082719326019287,
+      "learning_rate": 1.0516983520003632e-10,
+      "loss": 0.4695,
+      "step": 61297
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 4.730565547943115,
+      "learning_rate": 1.0346671399152907e-10,
+      "loss": 0.4641,
+      "step": 61298
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.4739956855773926,
+      "learning_rate": 1.017774956230122e-10,
+      "loss": 0.7214,
+      "step": 61299
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.3574206829071045,
+      "learning_rate": 1.0010218009781636e-10,
+      "loss": 0.5033,
+      "step": 61300
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 4.486630916595459,
+      "learning_rate": 9.844076742149267e-11,
+      "loss": 0.6958,
+      "step": 61301
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.544049024581909,
+      "learning_rate": 9.679325759848201e-11,
+      "loss": 0.6392,
+      "step": 61302
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.902857780456543,
+      "learning_rate": 9.51596506332253e-11,
+      "loss": 0.5544,
+      "step": 61303
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.104740619659424,
+      "learning_rate": 9.353994653071851e-11,
+      "loss": 0.6247,
+      "step": 61304
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.5824689865112305,
+      "learning_rate": 9.193414529429235e-11,
+      "loss": 0.9373,
+      "step": 61305
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.1651761531829834,
+      "learning_rate": 9.0342246930053e-11,
+      "loss": 0.4481,
+      "step": 61306
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.70953631401062,
+      "learning_rate": 8.876425144077605e-11,
+      "loss": 0.7706,
+      "step": 61307
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 6.033905982971191,
+      "learning_rate": 8.720015883201261e-11,
+      "loss": 0.4351,
+      "step": 61308
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.409464120864868,
+      "learning_rate": 8.564996910764844e-11,
+      "loss": 0.4974,
+      "step": 61309
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.718829393386841,
+      "learning_rate": 8.411368227212447e-11,
+      "loss": 0.7333,
+      "step": 61310
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 6.858797073364258,
+      "learning_rate": 8.259129832988155e-11,
+      "loss": 0.5107,
+      "step": 61311
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.439424753189087,
+      "learning_rate": 8.108281728425038e-11,
+      "loss": 0.3301,
+      "step": 61312
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.2381792068481445,
+      "learning_rate": 7.958823914078206e-11,
+      "loss": 0.6146,
+      "step": 61313
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.580695867538452,
+      "learning_rate": 7.810756390280727e-11,
+      "loss": 0.5658,
+      "step": 61314
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.6015450954437256,
+      "learning_rate": 7.664079157476689e-11,
+      "loss": 0.5165,
+      "step": 61315
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 5.427521705627441,
+      "learning_rate": 7.51879221599916e-11,
+      "loss": 0.6088,
+      "step": 61316
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.2556116580963135,
+      "learning_rate": 7.37489556634774e-11,
+      "loss": 0.8047,
+      "step": 61317
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 6.870641708374023,
+      "learning_rate": 7.232389208911006e-11,
+      "loss": 0.3801,
+      "step": 61318
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.2965526580810547,
+      "learning_rate": 7.091273144022026e-11,
+      "loss": 0.6138,
+      "step": 61319
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.3849663734436035,
+      "learning_rate": 6.951547372124889e-11,
+      "loss": 0.7078,
+      "step": 61320
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.5490329265594482,
+      "learning_rate": 6.813211893608174e-11,
+      "loss": 0.6954,
+      "step": 61321
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.512450933456421,
+      "learning_rate": 6.676266708804947e-11,
+      "loss": 0.5018,
+      "step": 61322
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.6582553386688232,
+      "learning_rate": 6.540711818159295e-11,
+      "loss": 0.6951,
+      "step": 61323
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.6278576850891113,
+      "learning_rate": 6.406547222004289e-11,
+      "loss": 0.7747,
+      "step": 61324
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.0418834686279297,
+      "learning_rate": 6.273772920728505e-11,
+      "loss": 0.75,
+      "step": 61325
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.1603779792785645,
+      "learning_rate": 6.142388914720521e-11,
+      "loss": 0.5233,
+      "step": 61326
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.7150089740753174,
+      "learning_rate": 6.012395204368915e-11,
+      "loss": 0.5381,
+      "step": 61327
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.5082619190216064,
+      "learning_rate": 5.883791789951243e-11,
+      "loss": 0.6237,
+      "step": 61328
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.821680784225464,
+      "learning_rate": 5.7565786718560837e-11,
+      "loss": 0.7223,
+      "step": 61329
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.7446155548095703,
+      "learning_rate": 5.630755850472014e-11,
+      "loss": 0.6526,
+      "step": 61330
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.408752202987671,
+      "learning_rate": 5.50632332607659e-11,
+      "loss": 0.4853,
+      "step": 61331
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.4425201416015625,
+      "learning_rate": 5.383281099113902e-11,
+      "loss": 0.3929,
+      "step": 61332
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.4058687686920166,
+      "learning_rate": 5.261629169861504e-11,
+      "loss": 0.8215,
+      "step": 61333
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 5.853254795074463,
+      "learning_rate": 5.141367538707975e-11,
+      "loss": 0.4354,
+      "step": 61334
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.250643014907837,
+      "learning_rate": 5.022496205930871e-11,
+      "loss": 0.5295,
+      "step": 61335
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.7909469604492188,
+      "learning_rate": 4.905015171863259e-11,
+      "loss": 0.4391,
+      "step": 61336
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.347775936126709,
+      "learning_rate": 4.788924436893716e-11,
+      "loss": 0.8962,
+      "step": 61337
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.229151725769043,
+      "learning_rate": 4.6742240012442876e-11,
+      "loss": 0.4046,
+      "step": 61338
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.838212490081787,
+      "learning_rate": 4.5609138653590626e-11,
+      "loss": 0.555,
+      "step": 61339
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.8038547039031982,
+      "learning_rate": 4.448994029460085e-11,
+      "loss": 0.6539,
+      "step": 61340
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.4757840633392334,
+      "learning_rate": 4.338464493880423e-11,
+      "loss": 0.6063,
+      "step": 61341
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.4892401695251465,
+      "learning_rate": 4.229325258953143e-11,
+      "loss": 0.7448,
+      "step": 61342
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.262970209121704,
+      "learning_rate": 4.1215763249002895e-11,
+      "loss": 0.6831,
+      "step": 61343
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 4.6779327392578125,
+      "learning_rate": 4.0152176921659516e-11,
+      "loss": 0.7757,
+      "step": 61344
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.9157419204711914,
+      "learning_rate": 3.9102493609166626e-11,
+      "loss": 0.4744,
+      "step": 61345
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 4.663172721862793,
+      "learning_rate": 3.8066713314854894e-11,
+      "loss": 0.605,
+      "step": 61346
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.408677339553833,
+      "learning_rate": 3.7044836042055e-11,
+      "loss": 0.6293,
+      "step": 61347
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.27150297164917,
+      "learning_rate": 3.6036861792432264e-11,
+      "loss": 0.5768,
+      "step": 61348
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.7119266986846924,
+      "learning_rate": 3.5042790570427587e-11,
+      "loss": 0.5037,
+      "step": 61349
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 1.7436089515686035,
+      "learning_rate": 3.40626223777063e-11,
+      "loss": 0.4014,
+      "step": 61350
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 6.704439640045166,
+      "learning_rate": 3.3096357217043964e-11,
+      "loss": 0.4761,
+      "step": 61351
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.6898627281188965,
+      "learning_rate": 3.214399509121613e-11,
+      "loss": 0.8568,
+      "step": 61352
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.0384538173675537,
+      "learning_rate": 3.120553600299836e-11,
+      "loss": 0.5585,
+      "step": 61353
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.4390950202941895,
+      "learning_rate": 3.028097995516621e-11,
+      "loss": 0.5265,
+      "step": 61354
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 4.372436046600342,
+      "learning_rate": 2.9370326949940134e-11,
+      "loss": 0.7345,
+      "step": 61355
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.043269634246826,
+      "learning_rate": 2.8473576990095675e-11,
+      "loss": 0.5067,
+      "step": 61356
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.5912539958953857,
+      "learning_rate": 2.759073007785329e-11,
+      "loss": 0.6205,
+      "step": 61357
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 4.129412651062012,
+      "learning_rate": 2.6721786215988532e-11,
+      "loss": 0.7917,
+      "step": 61358
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.503272771835327,
+      "learning_rate": 2.586674540672185e-11,
+      "loss": 0.4184,
+      "step": 61359
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.6585350036621094,
+      "learning_rate": 2.50256076528288e-11,
+      "loss": 0.5701,
+      "step": 61360
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 4.762491226196289,
+      "learning_rate": 2.4198372955974713e-11,
+      "loss": 0.5923,
+      "step": 61361
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.978303909301758,
+      "learning_rate": 2.338504131838004e-11,
+      "loss": 0.532,
+      "step": 61362
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 5.835544586181641,
+      "learning_rate": 2.2585612743375452e-11,
+      "loss": 0.7153,
+      "step": 61363
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 4.254486083984375,
+      "learning_rate": 2.1800087232626278e-11,
+      "loss": 0.9629,
+      "step": 61364
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.4215734004974365,
+      "learning_rate": 2.1028464787797855e-11,
+      "loss": 0.7583,
+      "step": 61365
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.5408434867858887,
+      "learning_rate": 2.0270745411665737e-11,
+      "loss": 0.6306,
+      "step": 61366
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.032701253890991,
+      "learning_rate": 1.9526929105895265e-11,
+      "loss": 0.6816,
+      "step": 61367
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.3843183517456055,
+      "learning_rate": 1.879701587270688e-11,
+      "loss": 0.6188,
+      "step": 61368
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.512024402618408,
+      "learning_rate": 1.8081005714321032e-11,
+      "loss": 0.5513,
+      "step": 61369
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.1745498180389404,
+      "learning_rate": 1.7378898632958165e-11,
+      "loss": 0.5487,
+      "step": 61370
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.6282317638397217,
+      "learning_rate": 1.66906946297285e-11,
+      "loss": 0.5622,
+      "step": 61371
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.4895644187927246,
+      "learning_rate": 1.6016393707407595e-11,
+      "loss": 0.6907,
+      "step": 61372
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.830347776412964,
+      "learning_rate": 1.535599586766079e-11,
+      "loss": 0.6352,
+      "step": 61373
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.4764461517333984,
+      "learning_rate": 1.47095011115983e-11,
+      "loss": 0.4664,
+      "step": 61374
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.534872055053711,
+      "learning_rate": 1.4076909441995689e-11,
+      "loss": 0.3788,
+      "step": 61375
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.6149444580078125,
+      "learning_rate": 1.3458220859963177e-11,
+      "loss": 0.5839,
+      "step": 61376
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.5577704906463623,
+      "learning_rate": 1.2853435367721212e-11,
+      "loss": 0.6901,
+      "step": 61377
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.2455365657806396,
+      "learning_rate": 1.2262552966935126e-11,
+      "loss": 0.6727,
+      "step": 61378
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.830624580383301,
+      "learning_rate": 1.1685573658715144e-11,
+      "loss": 0.5154,
+      "step": 61379
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.3824057579040527,
+      "learning_rate": 1.11224974447266e-11,
+      "loss": 0.5055,
+      "step": 61380
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.5025928020477295,
+      "learning_rate": 1.0573324326634826e-11,
+      "loss": 0.5009,
+      "step": 61381
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.83866548538208,
+      "learning_rate": 1.0038054306660272e-11,
+      "loss": 0.788,
+      "step": 61382
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.8813178539276123,
+      "learning_rate": 9.516687385358047e-12,
+      "loss": 0.2836,
+      "step": 61383
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.6671574115753174,
+      "learning_rate": 9.009223564948599e-12,
+      "loss": 0.5706,
+      "step": 61384
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.175276756286621,
+      "learning_rate": 8.515662845987039e-12,
+      "loss": 0.5041,
+      "step": 61385
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.3915867805480957,
+      "learning_rate": 8.036005230693811e-12,
+      "loss": 0.4792,
+      "step": 61386
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 4.985440254211426,
+      "learning_rate": 7.570250719624028e-12,
+      "loss": 0.6865,
+      "step": 61387
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.52408504486084,
+      "learning_rate": 7.1183993144430255e-12,
+      "loss": 0.5369,
+      "step": 61388
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.7455248832702637,
+      "learning_rate": 6.680451016816136e-12,
+      "loss": 0.5631,
+      "step": 61389
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 4.028948783874512,
+      "learning_rate": 6.256405827298473e-12,
+      "loss": 0.5173,
+      "step": 61390
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.5502734184265137,
+      "learning_rate": 5.84626374755537e-12,
+      "loss": 0.533,
+      "step": 61391
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.9243733882904053,
+      "learning_rate": 5.450024778141938e-12,
+      "loss": 0.3683,
+      "step": 61392
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.605046033859253,
+      "learning_rate": 5.067688921278624e-12,
+      "loss": 0.5278,
+      "step": 61393
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.7528438568115234,
+      "learning_rate": 4.6992561764103164e-12,
+      "loss": 0.6025,
+      "step": 61394
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.4877378940582275,
+      "learning_rate": 4.34472654575746e-12,
+      "loss": 0.4137,
+      "step": 61395
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 4.768109321594238,
+      "learning_rate": 4.00410003043028e-12,
+      "loss": 0.5117,
+      "step": 61396
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.1275434494018555,
+      "learning_rate": 3.677376630428775e-12,
+      "loss": 0.6348,
+      "step": 61397
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 7.860161304473877,
+      "learning_rate": 3.3645563468631682e-12,
+      "loss": 0.5853,
+      "step": 61398
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 4.514647960662842,
+      "learning_rate": 3.065639181398794e-12,
+      "loss": 0.5654,
+      "step": 61399
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.671342372894287,
+      "learning_rate": 2.7806251345907642e-12,
+      "loss": 0.4733,
+      "step": 61400
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.830066204071045,
+      "learning_rate": 2.5095142064390787e-12,
+      "loss": 0.4934,
+      "step": 61401
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.494065761566162,
+      "learning_rate": 2.252306398609072e-12,
+      "loss": 0.5063,
+      "step": 61402
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.93573260307312,
+      "learning_rate": 2.0090017111007442e-12,
+      "loss": 0.5892,
+      "step": 61403
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.6837823390960693,
+      "learning_rate": 1.7796001455794298e-12,
+      "loss": 0.5286,
+      "step": 61404
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.993518352508545,
+      "learning_rate": 1.5641017014900172e-12,
+      "loss": 0.6486,
+      "step": 61405
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 4.090750217437744,
+      "learning_rate": 1.3625063804978411e-12,
+      "loss": 0.6478,
+      "step": 61406
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.3651251792907715,
+      "learning_rate": 1.1748141820477898e-12,
+      "loss": 0.7425,
+      "step": 61407
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.0970749855041504,
+      "learning_rate": 1.001025107805198e-12,
+      "loss": 0.4574,
+      "step": 61408
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.384104013442993,
+      "learning_rate": 8.411391577700657e-13,
+      "loss": 0.5635,
+      "step": 61409
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.636643171310425,
+      "learning_rate": 6.951563319423926e-13,
+      "loss": 0.7098,
+      "step": 61410
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.3556206226348877,
+      "learning_rate": 5.630766314324021e-13,
+      "loss": 0.5579,
+      "step": 61411
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.79532790184021,
+      "learning_rate": 4.44900056240094e-13,
+      "loss": 0.5652,
+      "step": 61412
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.831293821334839,
+      "learning_rate": 3.4062660692057993e-13,
+      "loss": 0.3811,
+      "step": 61413
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.283438205718994,
+      "learning_rate": 2.502562834738598e-13,
+      "loss": 0.4612,
+      "step": 61414
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.111842632293701,
+      "learning_rate": 1.737890858999336e-13,
+      "loss": 0.6834,
+      "step": 61415
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.7786877155303955,
+      "learning_rate": 1.1122501530902441e-13,
+      "loss": 0.6494,
+      "step": 61416
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.1657426357269287,
+      "learning_rate": 6.25640711460207e-14,
+      "loss": 0.8077,
+      "step": 61417
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.6316778659820557,
+      "learning_rate": 2.780625396603398e-14,
+      "loss": 0.5077,
+      "step": 61418
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.4698894023895264,
+      "learning_rate": 6.951563213952739e-15,
+      "loss": 0.4161,
+      "step": 61419
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.435880184173584,
+      "learning_rate": 0.0,
+      "loss": 0.7015,
+      "step": 61420
+    },
+    {
+      "epoch": 1.0,
+      "step": 61420,
+      "total_flos": 1.105198174690712e+19,
+      "train_loss": 0.6322650376320351,
+      "train_runtime": 169038.1881,
+      "train_samples_per_second": 5.814,
+      "train_steps_per_second": 0.363
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 61420,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 1000,
+  "total_flos": 1.105198174690712e+19,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}